Studies in Surface Science and Catalysis 71 CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL II
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis Advisory Editors: 6.Delmon and J.T. Yates Vol. 71
CATALYSIS AND AUTOMOTIVE POLLUTION CONTROL II Proceedingsof the Second InternationalSymposium (CAPoC 2). Brussels, Belgium, September 10-13,1990
Editor
A. Crucq Unite de Recherche sur la Catalyse, Universite Libre de Bruxelles, Brussels, Belgium
ELSEVIER
Amsterdam
- Oxford - New York - Tokyo
1991
ELSEVIER SCIENCE PUBLISHERS B.V. Sara Burgerhartstraat 25 P.O. Box211,1000AEAmsterdam,The Netherlands Distributors forthe Unitedstates and Canada: ELSEVIER SCIENCE PUBLISHING COMPANY, INC. 655 Avenue of the Americas NewYork, NY 10010, USA
Library of C o n g r e s s Cataloging-in-Publication
Data
1990 Brussels, Belgium) C A P O C (2nd C a t a l y s i s a n d a u t o m o t i v e p o l l u t i o n c o n r r o l I1 proceedings 3f t h e S e c o n d I n t e r n a t i o n a l S y m p o s i u m ( C A P o C 2). B r u s s e l s . B e l g i u m . S e p t e m b e r 10-13. 1990 / e d i t o r , A. C r u c q . p. c m . -- ( S t u d i e s in s u r f a c e s c i e n c e a n d c a t a l y s i s 71) I n c l u d e s b i b l i o g r a p h i c a l r e f e r e n c e s a n d index. I S B N 0-444-88787-3 1. Automobiles--Catalytic converters--Congresses. 2. C a t a l y s i s - 2 o n g r e s s e s . 3. A u t o m o b i l e s - - M o t o r s - - E x h a u s t gas--Congresses. 4. C e r i u m o x i d e s - - C o n g r e s s e s . I. C r u c q . A . (And?&,. 193311. T i t l e . 111. S e r i e s . TL214.P6C3E8 1990 629.25'28--dc20 9 1-30962
.
CIP
ISBN: 0-444-88787-3
0 Elsevier Science Publishers B.V., 1991 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 Publishers B.V./ Academic Publishing Division, P.O. Box 330,1000 AH Amsterdam,The Netherlands. Special regulations for readers in the USA - This publication has been registered with the Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained from theCCCaboutconditions underwhich photocopiesof partsofthis publication may be made in the USA. All other copyright questions, including photocopying outside of the USA, should be referred to the Publisher. No responsibility is assumed by the Publisher for any injury and/or damage t o 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. Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this publication does not constitute a guarantee or endorsement of the quality or value of such product or of the claims made of it by its manufacturer. This book is printed on acid-free paper. Printed in The Netherlands
V
CONTENTS
- Foreword - Acknowledgments - Financial Support - Organizing and Scientific Committee
X XI1 XI11 XIV
- Scientific Papers GENERALINTRODUCTION TO THE PROBLEM OF EXHAUST GAS POLLUTION
Opening address E. Deworme
1
Energy and Air Pollutants in Belgium. The Contribution of Automotive Trafic since 1980 W.J. Hecq (General lecture)
5
Environmentalist Views on Automotive Exhaust Gases K. Taschner (General lecture)
17
The Evolution of the EEC Regulations on Vehicle Emissions H. Henssler (General lecture)
35
The Technological Development of the European Vehicles and the Environment C. Cucchi and M. Hublin (General lecture)
41
Industrial Application of Catalysts for Offgas Treatment T. Schmidt (General lecture)
55
Characterization of Exhaust Emissions from Two Heavy Duty Vehicles Fueled with Eight Different Diesel Fuels K.E. Egeback, G. Mason, U. Rannug and R. Westerholm
-
75
Fuels Evolution in Europe and Consequences on Exhaust Catalysis J.C. Guibet (General lecture)
93
Supply and Demand of Precious Metals for Automotive and Other Uses M.C.F. Steel (General lecture)
105
The Value of Spent Exhaust Catalysts W. Fierain (General lecture)
115
VJ TECHNICAL PAPERS
Role of CeOz in automotive exhaust catalysts
Cerium Oxide Stabilization :Physical Property and Three-way Activity Considerations J.E. Kubsh, J.S. Rieck and N.D. Spencer
125
The Role of Ceria in Three-way Catalysts A.F. Diwell, R.R. Rajaram, H.A. Shaw andT.J. Truex
-
139
Characterization of Oxidation Catalysts by CO-TPR and Selective Carbon-Carbon Bond Rupture Ch. Serre, F. Garin, G . Belot, G. Maire and R. Roche
-
153
Design and Performance Evaluation of Automotive Emission Control Catalysts R.G. Silver, J.C. Summers and W.B. Williamson
167
Influence of Rh and Ce02 Addition on the Activity and Selectivity of a PtIA1203 Catalyst in the CO+NO and CO+NO+O2 Reactions G. Leclercq, C. Dathy, G. Mabilon and L. Leclercq
181
Influence of Water in the Activity of Catalytic Converters M. Weibel, F. Garin, P. Bernhardt, G. Maire and M. Prigent
195
The Role of the Oxygen Vacancies at the Support in the CO Oxidation on RhlCe02 and RhlTi02 Autocatalysts G. Munuera, A. Fernandez and A.R. Gonzalez-Elipe
207
Physico-Chemical Properties of Ce-Containing Three-Way Catalysts and the Effect of Ce on Catalyst Activity J.G. Nunan, H.J. Robota, M. J. Cohn and S.A.Bradley
22 1
Studies to the Functioning of Automotive Exhaust Catalysts Using In-situ Positron Emission Tomography K.A. Vonkeman, G. Jonkers and R.A. van Santen
239
TPD and XPS Studies of CO and NO on Highly Dispersed Pt+Rh Automotive Exhaust Catalysts : Evidence for Noble Metal-Ceria Interaction P. Loof, B. Kasemo, L. Bjomkvist, S. Andersson and A. Frestad
253
VII
Laser Raman Characterizationof Sulface Phase Precious Metal Oxides Formed on Ce02 L.L. Murrell, S.J. Tauster and D.R. Anderson
275
Depollution of Diesel engines Diesel Emission Control E.S. Lox, B.H. Engler and E. Koberstein (General lecture) - 291 Kinetics of Soot Oxidation on Potassium-Copper-Vanadium Catalyst P. Ciambelli, P. Parrella and S. Vaccaro
323
Catalystsfor Diesel Powered Vehicles D. J. Ball and R.G. Stack
337
Use of base metals in exhaust catalysis Catalytic Automotive Pollution Control without Noble Metals S. Stegenga, N. Dekker, J. Bijsterbosch, F. Kapteijn, J. Moulijn, G. Belot and R. Roche
353
Three-way Activity and Sulfur Tolerance of Single Phase Perovskites D. Jovanovic, V. Dondur, A. Terlecki-Baricevic and B. Grbic
371
Fundamental studies on model catalysts NO Reduction and Adsorption Intermediates on Pt-Rh Alloy Catalysts L. Heezen, V.N. Kilian,R.F. van Slooten, R.M. Wolf and B.E. Nieuwenhuys
381
A Comparative Kinetic Study of the CO-02 Reaction over Pt-Rh ( I l l ) , (100),(410) and (210) Single Crystal Sulfaces J. Siera, R. van Silfhout, F. Rutten and B.E. Nieuwenhuys -
395
Molecular Beam Studies of CO Oxidation on Rh(l10) M. Bowker, Q. Guo, P.D.A. Pudney and R.W. Joyner
409
-
VIII
Unregulated emissions Role of Sulphate Decomposition in the Emission and Control of Hydrogen Sulphide from Autocatalysts A.F. Diwell, S.E. Golunski, J.R. Taylor and T.J. Truex
-
The Effect of Aging on Nitrous Oxide N 2 0 Formation by Automotive Three-way Catalysts M. Prigent, G. de Soete and R. Dozikre
417
425
Theoretical approach to the catalytic converter Flow, Heat and Mass Transfer in a Monolithic Catalytic Converter D. Schweich (General lecture)
437
A New Theoretical Approach to Catalytic Converters J.P. Leclerc. D. Schweich and J. Villermaux
465
Supports for Automotive Catalysts New developments in Catalytic Converter Durability S.T. Gulati (General lecture) Study of Platinum Deposit on Ferritic Stainless Steel Conversion Coatings L. Aries, P. de Veyrac, M. Mantel and J.P. Traverse
48 1
~
509
Monolith Substrate Efsects on Catalyst Light-Off T.S. Jasper, K. Robinson, D. Anderton and D.H. Cuttler -
523
Comparison of Metal Foil and Ceramic Monolithic Automotive Catalytic Converters G. L. Vaneman
537
Sols as Precursors to Transitional Aluminas and these Aluminas as Host Supports for Ce02 and Zr02 Micro Domains L. L. Murrell and S.J. Tauster
547
Preparation of Thermostable High-Sugace-Area Aluminas and Properties of the Alumina-Supported Pt Catalysts F. Mizukami, K. Maeda, M. Watanabe, K. Masuda, T. Sano and K. Kuno
557
IX
Ageing of Automotive Catalysts Kinetics of the Physico-Chemical and Catalytic Activity Evolution of a Pt-Rh Catalyst in Automotive Exhaust Gas G. Mabilon, D. Durand and M. Prigent
569
Characterization of Bimetallic Su$aces by 1801160 Isotopic Exchange. Application to the Study of the Sintering of PtRhlA1203 Catalysts S. Kacimi and D. Duprez
581
The Effect of Sulfur on Three-way Catalysts D.R. Monroe, M.H. Krueger, D.D. Beck and M.J. D'Aniello, Jr.
593
Effect of Severe Thermal Aging on Noble Metal Catalysts H. Shinjoh, H. Muraki and Y. Fujitani
617
Morphological Transformations in Reducing and Steam Atmospheres of Alumina-Supported Rhodium and Platinum Catalysts D. Duprez, F. Sadi, A. Miloudi and A. Percheron-Guegan
629
Miscellaneous The Influence of Three-way Catalyst Parameters on Secondaly Emission -B. Engler, E. Koberstein, D. Lindner and E. Lox
64 1
Improvements in Techniques for Reducing Emissions by Using Computer Simulation M. Hashimoto, R. Matsumura, F. Yukawa, M. Saitoh and M. Matsumoto
657
PdIA1203 Catalysts for the NO-CO-02 Reaction :"In Situ" Determination of the Palladium State under the Reactant Mixture 667 J.L. Duplan and H. Praliaud AUTHOR INDEX
679
STUDIES IN SURFACE SCIENCE AND CATALYSIS (other volumes in the series)
681
X
FOREWORD
In September 1986, we organized in Brussels the first International Symposium on “Catalysis and Automotive Pollution Control” known under the acronym CAPoC 1. At the end of this symposium, most participants expressed their hope that it would be repeated after a not too long interval of 3 to 4 years. A CAPoC 2 was thus organized in September 1990 and this volume constitutes the proceedings of the symposium. CAPoC 2 has been a great success from the point of view of its scientific interest, as evidenced by the content of this book, and also from the point of view of the participation, as 260 scientists, as compared with 177 in 1986, attended the symposium. This is the proof that exhaust catalysis remains a major topic of interest. As in 1986, most participants came from EEC countries, with large delegations from Belgium (48), France ( 5 9 , Germany (36), and the United Kingdom (31), but we again note the importance of the U.S. (23) and Swedish (17) delegations and the interest of people coming from all over the world, including Japan, P.R.China, Venezuela, South Africa and Algeria. About two-thirds of the participants came from the industrial world, mainly the car and oil industries and catalyst manufacturers. As in 1986, the Organizers choose to devote the first day of this fourday symposium to non technical, non catalytic problems, related either to pollution in general or specifically to automotive pollution. Nine general lectures on these subjects were presented and are published in this book. Concerning the strictly catalytic part of the Symposium, the Organizers received 66 papers for consideration, of which 42 were chosen for presentation and of which 34 are published here. This book also contains the text of three general lectures out of the four that were orally presented. In order to accomodate 42 papers and 4 general lectures within 3 days, a poster presentation of some of the accepted papers was necessary. The choice made by the Scientific Committee between oral or poster presentation was based not on the quality of the papers but on the topics. In this way, and with the aim of stimulating the discussion, it was possible to organize half-day sessions with oral presentations on the same topics.
XI
In these proceedings, the lectures and papers are grouped into the following headings: 1- The role of Ce02 in exhaust catalysts (1 1 papers) 2- Supports for automotive catalysts (6 papers) 3- Ageing of automotive catalysts (5 papers) 4- Depollution of Diesel engines (3 papers) 5- Fundamental studies (3 papers) 6- Base metals in exhaust catalysis (2 papers) 5- Fundamental studies (3 papers) 7- Unregulated emissions (2 papers) 8- Theoretical approach of the catalytic converter (2 papers) 9- Miscelleaneous It is worth well to mention that in 1986 not a single paper was presented concerning the role of Ce02 in exhaust catalysis. The Organizers thank all the people that have submitted papers for their interest in CAPoC 2, and all authors of presented papers for the high quality of their contributions. Thanks are also due to all delegates for their participation in the very fruitful discussions both inside and outside the conference room.
Dr. A.CRUCQ
XI1
ACKNOWLEDGMENTS
The CAPoC 2 conference has been held in the “Salle Dupreel” of the “Institut de Sociologie” of the University of Brussels. The Organizers thank the authorities of the University for their hospitality and for the welcome address given by the representative of the Rector of the University. The Organizers express their gratitude to E.Deworme, State Secretary to the Energy, for his interest for the symposium and for the welcome address given at the opening session. Special thanks are due to all lecturers of the introductory session: W.Hecq, K.Taschner, H.Henssler, C.Cucchi, T.Schmidt, K.E.Egeback, J.C.Guibet, M.C.F.Stee1 and W.Fierain - for the high quality of their contribution. The Organizers are pleased to thank Drs S.T.Gulati, E.Lox, D.Schweich and M.Shelef for the outstanding general lectures they have given as an introduction to the topics discussed in the papers presented during the following scientific sessions. Finally, the Organizers want to recognize the important but sometimes hidden contribution of all members of the ”Unit6 de Recherches sur la Catalyse” to the success of the symposium.
XI11
FINANCIAL SUPPORT
The Organizers of this Symposium gratefully acknowledge the financial support sponsorship from the following organizations: A.C.Rochester (Division of General Motors) Commission of the European Community Degussa AG Fabrimetal FEBIAC Ford Motor Cy Johnson Matthey plc Strategic Analysis Europe
XIV
ORGANIZING COMMITTEE : Frennet A. (Chairman) Crucq A. (Secretary) Campinne M. Derouane E. Gerryn C1. Koberstein E. Poncelet G. Schmitz J.C. Evans W.D.J. Germain A. Hecq W. Leduc B. Prigent M. -
UniversitC Libre de Bruxelles UniversitC Libre de Bruxelles Ecole Royale Militaire FacultC Notre Dame de la Paix Ford Europe Inc. Degussa AG UniversitC Catholique de Louvain General Motors Eur. Tech. Center Johnson Matthey plc UniversitC de Likge UniversitC Libre de Bruxelles UniversitC Libre de Bruxelles Institut Francais du PCtrole
SCIENTIFIC COMMITTEE : All members of the organizing committee, plus : Belot G. Peugeot S.A. Brunelle J.P. Rhone Poulenc Garin F. UniversitC de Strasbourg UniversitC de Lille Leclercq L. Universitk de Nancy Pentenero A. UniversitC de Leiden Ponec V. Johnson Matthey plc Webster D.E. -
1
OPENING ADDRESS by Elie Deworme
Belgian State Secretary for Energy Mr President, Mrs. Rector, Ladies and Gentlemen, The State Secretary of Energy, Elie DEWORME insists on thanking the organizers of this symposium for having invited him to take part in their activities and having offered him the occasion to give a speech. Unfortunately, the current international situation requires his full attention and this is the reason why he has asked me to represent him. However, he wishes to stress the great interest he takes in your activities. Indeed, the State Secretary is convinced that the subject of this meeting is important and he notes with satisfaction that an impressive number of experts from all over the world are taking part in this meeting, which will last for four days. Environmental protection constitutes a problem that knows no frontiers; this is why international collaboration is essential. When we evoke international collaboration, it means that we have to promote a harmonization of the policies and the regulations both on a European and a world-wide level. Furthermore, and this is undoubtedly the meaning of your approach, it seems equally important to combine our efforts on a scientific level, and to bring together regularly the knowledge in this field. However, we would like to draw your attention to another aspect of this problem, which brings us closer to the current events: saving energy. The recent events in the Gulf, and the international tension resulting from it, calls back the ghost of the oil crisis. It is obvious that we are not in the same situation as in 1973-1974 and 1979-1980, when the industrialized countries were subjected to an "embargo" of the oil producing countries. Today, the consuming countries supported by certain producers are organizing the embargo.
2
That does not prevent a large number of OECD-countries, including Belgium, from preparing themselves to deal with a supply deficit. Last week, we had a reunion with several regional ministers in order to discuss a number of measures to increase the consumer’s awareness of energy savings. In the course of the 198O’s, our country has conducted, both at its own initiative and by means of a concerted action within international organizations such as the EEC or the International Energy Agency, a series of actions on a national level, aiming at energy savings; those actions included information campaigns (distribution of brochures, spots on television, participation in fairs and exhibitions, etc.), financially encouraging measures (grants or fiscal advantages for private individuals or firms achieving energy saving investments) and even restrictive measures (for example to insure that the equipment put on the market possesses sufficient energy efficiency). Those measures have allowed us to obtain significant results, although it is not always easy to identify all the factors underlying the decrease of energy consumption. The evolution curves of energy intensity, that is the relationship between the energy consumption and the gross domestic profit, show a general decline that leaves no doubt, at least during the first half of the 1980’s. We are inclined to think that this period was favourable for the environment. One of the sectors in which energy saving is desirable and directly affects the public is obviously the transport sector, and more specifically the automobile sector. In this connection, we should salute the remarkable efforts of the researchers of the firms to reduce the consumption and the pollution due to automobiles; I often state that the only energy source that does not pollute is that which we do not use. It is true that for the last one or two years, the automobile constructors have realized enormous investments to reduce both the consumption of the cars and their pollution. On the other hand, the number of cars in circulation has considerably increased, as well as the yearly distances covered by each car. Consequently, we have cars that pollute less but in a larger number. This indicates the size of the efforts we still have to make together. You, the scientists, by finding new techniques to limit the pollution, and we, the politicians, by inciting the population to adopt a “nature-friendly’’ behaviour. The automobile constructors have introduced catalysts in a large number of models. These are solutions to promote and to encourage. In turn, the Belgian Government has taken measures to encourage use of catalysts by offering a premium to the owners of old cars who equip their car with a catalyst.
3
It is necessary that the public understands the value of its environment and the importance of the heritage we give to our future generations. This is the message that the State Secretary for Energy, Elie DEWORME, would like to give for the week ahead: to combine transport security with energy savings but also with struggle against pollution. I wish you fruitful activities during these coming days and I thank you.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZI 0 199 1 Elsevier Science Publishers B.V., Amsterdam
5
ENERGY AND AIR POLLUTANTS IN BELGIUM The Contribution of Automotive Traffic since 1980 Walter J. HECQ.
Political Economy Centre, Universite' Libre de Bruxelles During the seventies, three important events: the publication of the Rome Club's Meadow Report, the United Nations Conference in Stockholm, and the oil shocks revealed the dangers threatening the Western economies, 1.e. : - the long-term depletion of natural resources; - the atmospheric pollution produced by combustion gases, which is showing its international scale; - an over-dependence on imported oil fuels. This latter has been brought out by recent events. The measures taken in the meantime have made it possible to slow down the quasi continuous increase in energy consumption and the pollution associated with it. A number of indicators show this in various countries including Belgium. Firstly, primary energy consumption, which had been on the increase since the end of the Second World War rose from 19.1 Mtoe in 1950 to 45.3 Mtoe in 1973, slowed down and then reached 48.5 Mtoe in 1979. Secondly, since 1970, the daily average ambient concentrations of SO2 and smoke in the air started to decline in Brussels and elsewhere, as indicated by the monitoring station network. The decline in the use of coal and its progressive replacement by natural gas, and also reductions in the sulfur content of oil products has made it possible to make a reduction in the larger number of local pollution sources. However, these measures revealed themselves to be insufficient with a long distance pollution still adding to the latter and remaining. This long distance pollution has led to the establishment of a European level Policy. This took the shape of EEC Directives on the progressive restriction of exhaust emissions from vehicles and major combustion plants. For the latter, 1980 levels are used as reference levels which must not be exceeded. 1980 is also the first year in which energy consumption started to diminish in Belgium and some other Western countries. So, this year will be the starting point of the comparisons that I shall make later. If energy consumption - like emissions of certain pollutants - has become more or less stabilised since then, five forms of atmospheric energy-
Table 1. Five Exemples of Environmental Concerns. CONFORMITY WlTH
LEVEL
POLLUTION COMMON NAME
POLLUTANTS
CONTRIBUTION TO ADVERSE EFFECTS
ATMOSPHERIC TRENDS
ATMOSPHERIC
WHO STANDARDS IN
LIFE TIME
BELGIUM
~
PARTICULATE MAmAND
LOCAL URBAN
W O R
CARBON MONOXIDE EFFECT
co
CONSTANT OR LOW
TENS OF DAYS
DECREASE
SQ?
TOTAL
CONSTANT OR LOW GROWTH
YES EXCEPT EPISODES IN URBANIZEDAREAS
SEVERAL
ID.
MONTHS LOCAL AND REGIONAL
ACIDRAIN SOZ, N&, HCL
sqr
60%
LOW DECREASE
2 - 5 DAYS
NO
Nox
30%
GROWTH
1 - 2 DAYS
NO
03
N.D.
M E D m
--
YES, EXCEPT EPISODES IN SUBURBS AREAS
co"u0us INCREASE
a?
GLOBAL
W N E HOLE
CFCs, NO, CO; ....
50%
MEDIUM INCREASE
8 YEARS
N20
=lo%
ID.
170 YEARS
03
40%
ID.
--
NO
?
GROWTH
?
7
related pollution continue to cause concern. They have approximately three geographical ranges: local, local and regional (continental) and global. (i) Local (urban) atmosDheric Dollution:: Two kinds of pollution are observed in urban areas (table 1). The g e c t of particulate matter : particulates from mineral residue and unburned organic matter in fuels is associated with, inter alia, the use of heavy fuels (diesel and fuel oil) and coal. Their effects on building materials (appearance and accentuation of corrosion phenomena) are well known. In epidemiological and toxicological studies, these particulates are also held to be a factor leading to a growth in respiratory illness (chronic bronchitis) and mortality. The effect of particulates is reinforced by the presence of SO2 in the air (synergy effect). It is for this reason that the guide values counselled by the W.H.O. (short term, 24 hours : black smoke < 125 pg/m3, SO2 < 125 pg/m3; long term, 1 year : black smoke < 50 pg/m3, SO2 < 50 pg/m3) (W.H.O., 1987) and the limit values set in the council directives (Official Journal) link these pollutants. These established limit values are still exceeded during the winter periods. Domestic heating and diesel vehicles are the principle sources. The effect of carbon monoxide (CO) : carbon monoxide is an intermediary combustion product which is very common in the urban atmosphere. It is emitted principally by two types of source, involving incomplete combustion i.e. small coal fires and gasoline motors. With the disappearance of the former and the increasing number of motor vehicle, carbon monoxide is produced in great quantities by automobile traffic. Also, in unfavourable weather conditions (poor dispersion) and in the big cities, it is not uncommon to see mean concentrations rise above the limits counselled by the W.H.O. (11.5 mg/m3 over 8 hours, 30 mg/m3 per hour) (W.H.O., 1987). Given its well known affinity with haemoglobin in the blood, during such periodes, carbon monoxide is a danger (hypoxia) to at risk groups i.e. angina pectoris patients, bronchitics, young children and pregnant women.... (ii) Atmospheric pollution at local and regional level : Two types of pollution cause concern (table 1) on a larger geographic scale. Acid rain : this phenomenon, widely described in the press, takes the form of wet and dry deposits. These deposits attack building materials and health; they make the soil and surface waters acid at great distances from their origin. Two pollutants are the main contributors of acid deposits, namely sulphur dioxide ( S O 9 and nitrogen oxides (NO, : NO + NO2). Both are produced by burning fossil fuels. Sulphur oxides are formed from residual sulphur in petroleum fuels and coal. Nitrogen oxides are formed by
8
two mecha-nisms: either by the oxidation of nitrogen compounds contained in fuels, or by oxidation of atmospheric nitrogen. If gasoline contains reduced amounts of sulphur and nitrogen, taking account of combustion conditions, it is the nitrogen oxides from the air which form the family of polluting acids produced in significant quantities by gasoline vehicles; Photochemical pollution : this pollution is the result of secondary pollutants and among them, tropospheric ozone ( 0 3 ) is an aggressive oxidant. Briefly, ozone is formed by a process catalyzed by sunshine and NO,. The NO, are recycled thanks to volatile hydrocarbons (VOC)' . If the NO produced is greater than the NO --> NO2 photochemical conversion rate, the NO reacts with the ozone and the 0 3 concentration is reduced. This is a common phenomenon in urban areas. During daylight periods, when the NO produced is insufficient, NO2 - VOC photochemical conversion produces 0 3 . Concentrations of this gas in the atmosphere are therefore rising. The average annual concentration of 0 3 in the troposphere rose almost continuously from 30 to 50 ppb in 16 years. The W.H.O. guide values -150-200 pg/m3 for one hour and 100-120 pg/m3 for 8 hours - are exceeded from time to time and alterations to both animal and vegetable tissues result. (iii)AtmosDherk pollution at global level: While acid rain and photo-chemical pollution have spread at continental level, the "greenhouse effect" is found worldwide (table 1). This phenomenon, the effects of which are still poorly understood although without doubt of great significance, is closely associated with the use of fossil fuels. Four energy related gases, including the principal product of combustion C02, and also nitrous oxide (N20), methane (CH4) and ozone ( 0 3 ) are involved in the process to differing degrees (Table 1). These gases have the property of absorbing the IR re-radiated back by the earth, thus heating the troposphere. Given the quantities emitted and the time factor involved in the process, models predict that an increase of 1.5 to 5.5 C" in the temperature of the troposphere is likely by 2030-2050. Consequently, a general modification of the climate is expected, accompanied by a rise in the level of the sea (thermal expansion and the melting of polar ice). Changes in crop yields and the composition of flora, the disrippearance of coastal areas and the salination of estuaries are among the repercussions of the greenhouse effect most often mentioned. 1 - CO also play a role in this process. It controls the level of the OH radical which in turn, determines the lifetime of hydrocarbons
9
Seven air pollutants (particulates, CO,S02,NO,, VOC,C02,N20)are thus implicated in the five major current forms of pollution associated with the combustion of fossil fuels. Emissions of these pollutants are a function of five parameters : - the quantity of fuel consumed; - the systems of combustion used; - the conditions under which combustion takes place; - fuel composition; - any abatment technologies applied. The last four parameters are, to variable degrees, currently subject to regulations and can define the emission factors. As for the other parameter, the amount of fuel consumed, this also plays a significant role in pollutant emissions. It is also an important factor from the point of view of the energy dependence of our economies and the rational use of non-renewable resources - which include fossil fuels.
T
I
e
30
W
I 25 Y
Dlesel Gasollne
e 20 a
Transponatlon
1 I I 1
0 Resld.Agr.Com.PubL
15
Industry
10
1980
1981
1982
1983
1984
1985
1986
1987
1988
Figure I . Inland primary energy consumption in Belgium.
Developments in energy consumption in Belgium since 1980. As in other industrialised countries (France, the Netherlands, ...), Belgian energy consumption, which peaked in the seventies, started to decline in 1980 (figure 1). It reached its lowest level in 1983, by which time consumption had declined by 12% compared with 1980. Then, with the economic recovery and the reduction in fuel prices, energy consumption started to rise slowly once more. In 1988, the gross inland primary energy
10
consumption was very little higher than in 1980 (46Mtoe). During this period, gasoline consumption followed a similar curve, but with a time lapse of two years : a 15% decrease between 1980 and 1985 followed by a 17% increase between 1985 and 1988. Thus, gasoline consumption in 1988 returned to the same level as in 1980 (3.1 Mtoe), which represents 6.7-6.8% of gross inland primary energy consumption. In fact, during this same period the number of gasoline driven cars on the road had followed the same pattern. It decreased from 2.8 million units in 1980 to 2.6 million in 1985, increasing again to 2.7 million units in 1988. Studies (L. DE BORGER, 1987; T. MOROVIC et al., 1987; TestAchats, 1983-1989) also show that specific vehicle consumption decreased by more than 10% during this period (vehicle improvement, energy saving, etc ...). So it is overall annual vehicle kilometrage which grew between 1980 and 1988 (FEBIAC INFO, 1990). The diesel situation is different, consumption of this fuel increased steadily in 1980 (2.1 Mtoe) and 1988 (3.5 Mtoe) and numbers of diesel vehicles have also increased greatly: from 240,000 in 1980 to 810,000 in 1988. It should be noted that in Belgium, there is less tax on diesel oil than on gasoline, and that relatively speaking, "diesel" cars consume less than gasoline cars. Furthermore, they are reputed to be more robust. These observations explain the success of diesel cars in this country.
Combustion pollutants emissions between 1980 and 1988 and the role of traffic. As stated pertinently in the last W.E.C. report (W.E.C., 1989), "Breakdown of manmade pollutant emission by sector is an essentialstarting point for policy decisions concerning energy system structure and pollutant control". According to the figures 2, 3, 4, 5, there has been a general decrease' in combustion pollutants in Belgium since 1980. This decrease is explained h the first place by the significant increase during the 1980s in the numbers of nuclear power plants, which do not produce chemical pollution. The second explanation is the progressive implementation of regulations restricting emission of pollutants from fossil fuels (a reduction in the sulphur content of oil products and of CO, VOC, NO, emissions from gasoline vehicles, inter alia). A third explanation has to do with structural and economic'changes after the oil shocks which have influenced the development and distribution of the fuels used. This has resulted in overall reductions in: - particulates, SO2 and CO emissions whose effects are felt locally; 1
- Volatile hydrocarbon are an exception, their emissions have increased
11
1800000 1600000 1400000 ~
4
1000000
r
600000
;
Q Electricity 6 Heat
D Diesel
t 1200000
El Gasoline
800000
Transportation
0 Rwid.Agr.Corn.Pub1.
8 Industry
400000 200000 0 1980 1981 1982 1983 1984 1985 1986 1987 1988
Figure 2. CO emissionsfromfossil fuel combustion in Belgium. 450000
-I
400000
t
350000
0 Electricity & Heat
300000
0 Diesel
4 250000
N Gasoline
; 200000
E! Transportation
r
0 Resid.Agr.Corn.Pub1. W Industry
100000 50000 0
1980 1981 1982 1983 1984 1985 1986 1987 1988
Figure 3 N O x emissions from fossil; fuel; combu tion in Belgium (expressed in NO2 ) T
140000 120000 t
I
100000
Electricity h Heat
Dleset
Q Gasoline
80000 a
r
~
I
150000
160000
I
Transportation
60000
0 Resid.Agr.Com.Pub1.
40000
8 Industry
20000 0
Figure 4. VOC emissions from fossil fuel combustion in Belgium.
~
I I
PARTICULATESEMISSIONS
80000 70000
Irn
SO2 EMISSONS 700000
MWOO
OEMMh&h.(
50000 I
1
I 40000 I I
In
soowo I 400000
:
30000
r
0Iway.COnplaL
J00000
r
20000
200000
10000
lwOW
0
0
1980 1981 1982 1983 1984 1985 1986 1987 1988
1980191t 1982 19131911 19151916 19871918
Nx) EUSSIONS
C02 EMISSIONS 14mo
120000 100000
40000
JSOOO
rnEMH1yLH.M
I 30000
U
;
4sm
in {
80000
25000
:
I
'
60000
r
20000 11000
40000
1oooo
P0000
KO00
0nnuqI.corrrpubL
0
0 I980 1081 1982 1983 1984 1985 I986 I987 1988
1910 1911 1982 1985 1984 1986 1931 1087 1918
Figure 5 :Particulate, SO2 CO;!and N20 emissions from fossil fuel combustion in Belgium.
13
- emissions of acid pollutants (S02, NO,) and of one ozone precursor (NO,) which have effects at local and regional level1 ; - and, finally, emissions of greenhouse gases (C02, N20) which have worldwide implications. In this concept involving seven forms of pollutants, gasoline vehicle exhaust has a significant influence in three forms: - at local level: here we are talking about carbon monoxide (CO) emissions, 80% of which are caused by motor vehicles in Belgium (figure 2). If global CO emissions in Belgium have been reduced by 2.9%, the emissions from gasoline vehicles fell by 6.8% between 1980 and 1988 (from 1.394.106 t to 1.299.106 t ) thanks to the progressive adaptation of vehicles to EEC standards (R 15.00, 15.01, 15.03, 15.04). However, after decreasing by 16.6% between 1980 and 1985, CO emissions rose again by 11.6% between 1985 and 1988. - at local and regional level : this concerns one of the "acid' and ozone precursor pollutant, oxides of nitrogen (NO,). Total emissions of this pollutant dropped by 11.3% in Belgium between 1980 and 1988 (figure 3). The contribution of gasoline vehicles represents more than 30% of these total emissions in Belgium. On the other hand and despite the adaptation of vehicles to EEC R 15.02, 15.03, 15.04 standards, NO, emissions by gasoline vehicles increased slightly, by 0.7% between 1980 and 1988. From 142,600 t of NO2 in 1980 and 143,600 t of NO2 in 1988. In this case, increased vehicle kilometrage limited the effect of implementing the regulations. However, these emissions fell continuously until 1985 (- 14.3%), and then the trend was reversed from (+ 17.6%). - at local and regional level: this third form has to do with both the NO, emissions mentioned previously and emissions of volatile hydrocarbons (VOC), both precursors of photochemical pollution. In total, VOC emissions increased in Belgium by 1.6% between 1980 and 1988, largely due to the increased contribution of diesel engines (figure 4). The proportion due to gasoline engines in 1988 (97,650 t of VOC) is lower by 7.2% than that for 1980 (105,300 t of VOC). Such values represent about 64%-70% of total hydrocarbon emissions due to combustion in Belgium. As for CO and NO, emissions, VOC emissions fell between 1980 and 1985 (- 14%) then rose (+ 7.7%) from 1985 to 1988. As for other pollutant emissions (figure 5): particulates, S02, C02, N20, gasoline vehicles make relatively low contributions; < 10% of total emissions from other sources. The same is true for the proportion of diesel vehicle emissions in total emissions due to combustion, the role of these emissions are limited with the exception of particulates, and, to a lesser 1-
And even at global level since 0 3 is a greenhouse gas
14
extent, of NO, and CO. To summarize, the role of gasoline motors in pollutant emissions from combustion essentially concerns three pollutants -- CO, NO, and VOC -- the negative effects of which are mainly local (urban areas) and regional. If a significant reduction in CO and VOC emissions from gasoline engines is observable for the 1980-1988 period, this is mainly the result of a net decrease observed for the period between 1980 and 1985. This decrease was followed by an increase in these emissions from 1985 on, at which point, the price of petroleum products started to fall (a fall in the price of the $ and of crude oil). As for NO, emissions, these follow a similar "v" configuration, but were higher in 1988 than in 1980.
300,
NO7 (ug/m31
0KRO0I
03-24
DECEMBER 1986
CO ( p p m l
0KRO81
83-24
OECPlBER 1986
NUMB. CRRSJHR
0KR08l
83-24
OECM B m 1986
-1
Figure 6 . Relationship between trafic density (number - cars I hr.) and NO, N02. CO concentrations ( semi - hourly ) in Brussels.
15
When we know the relationship between traffic density and the concentration of pollutants in the ambient air as shown in the figure 6, it is not surprising that measures intended to reduce emissions of CO, VOC and NO, were adopted by both the Belgium and the Community authorities with the coming into force of improved standards provided for in the Luxemburg Agreement (Directive 88/76/EEC) and the signing, in June 1989 of the decrees applicable to vehicles with an engine capacity below 1400 cc. In order to conform these new standards, the solution chosen is the extensive use of the catalytic converter : one of the technologies which appears better suited at present to wiping out increases in emissions of these three pollutants (CO, NO,, VOC) which have been observed since 1985; this technology also has the indirect merit of causing a drop in the emission of lead particles into the atmosphere. In this context, we should emphasize the initiative of the Belgian authorities who, like several other European governments (Germany, Holland), have decided to grant premiums to private individuals buying vehicles with an engine capacity of two liters or less and equipped with a catalytic converter. This measure, which has been effective from July 1st this year, will remain in force for two years. Given the encouraging prospect of such measures and also the prospect which justifies our presence here i.e. the improvement of catalytic processes, we must keep in mind not only the diesel vehicles whose number is growing but also another goal in preparing for the future : namely that we must make ongoing efforts to economize oil, a non-renewable resource located in large amounts in politically instable states; a resource on which our country and many others are still very dependent. REFERENCES.
- DE BORGER, L. (1987): "Modelvonning van personenvervoer (vervoer en energie) op basis van een personenenquEte" - Ministerie van Economische Zaken, Programma Energie, 55 p., December 1987. - FEBIAC INFO (1990): "Dossier mobilis" - FEBIAC INFO, no 8.8 p., juin 1990. - Journal Officiel (1988): "Proposition de directive du Conseil modifiant la directive 80/779/CEE du Conseil concernant des valeurs limites et des valeurs guides de qualit6 atmosphkique pour l'anhydride sulfureux et les particules en suspension" - Journal Officiel des Communautt?s Europcknnes, no C 254/6,30/09/1988. - MOROVIC T.. GRUNDING F.-J., JAGER F.. JOCHEM E.. MANNSBART W., POPPKE H., SCHON M., TOTSCH I. (1987): "Energy Conservation.Indicators",SpringerVerlag CEE, FhG, 333 p.. 1987. - Test-Achats (1985): "Dossier voitures ? essence" i (1) et (2) - Test-Achats Magazine, no 273,... DD. 617. 1985 et no 274. DD. 4-18. 1986. - Test-Achats (1989): '*Dos&rvoitures no 1 et 2: L'expkriencedes automobilistes" - TestAchats Magazine, no 317, pp. 3645,1989 et no 318, pp. 29-37.1990. - W.E.C. (1989): "An Assessment of Worldwide Energy-Related Atmospheric Pollution" - World Energy Conference, Report 1989,224 p., London, August 1989. - W.H.O. (1987): "Air Quality Guidelines for Europe" - W.H.O., Regional Publication, European Series, no 23, Copenhagen, 1987.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 1991 Elsevier Science Publishers B.V., Amsterdam
17
ENVIRONMENTALISTS' VIEWS ON AUTOMOTIVE EXHAUST GASES
Karola Taschner European Environmental Bureau , Bruxelles, Belgium. INTRODUCTION
Automotive transport is very popular. It gives the possibility to move fast, far and in the first place self-determined and independent. The car means freedom for the driver, in the first years of the automobile for some " happy few", and nowadays, almost for everybody. The car was a success. It is only over the last decade that the disavantages emerged, due to the continuously growing car park. The all too numerous cars met physical barriers: lack of space, air pollution and noise were increasing. About 80 % of the EC population live in urban areas and so are concerned. But according to estimates of Commission, the situation will even aggravate, because car ownership rises continually The same is expected for vehicles for freight transport which will increase in number, weight and size due to the Single Market, which will result in a considerable rise in ton kilometers, between 30-50 % by 2000, according to the Task Force Report contracted by the Commission and up to 80 % according to the Institut fur Landes und Stadtentwicklung, Dortmund. Figure 1 indicates the evolution of goods and passenger traffic in the EC. Among noise, degradation of the urban and rural environment, consumption of land - which is a limited resource - and congestion, exhaust gases are the problem. Figure 2 shows the evolution and forecast of NOx and C02 emissions by sector. Clearly these emissions will not decrease in the next two decades. COMPOSITION AND EFFECTS OF NOXIOUS EXHAUST EMISSIONS
A number of damages are attributed to tailpipe emissions. the toxic pollutants are causing health impairments, plant damage and damages to material. The first pollutant to raise concern was carbon monoxide because it was emitted in so high quantities that ambient air concentrations were acutely toxic. Through concentrations have decrease meanwhile, carbon monoxide is still a problem because not much is known about chronic effects and because it
18
Car ownership
EUROPE 12
600 500 400 300
200 100 n "
I
1965
1970
1975
1980
1985
1990
1995
2000
2005
Evolution of goods traffic
180 160 140 120 100 80 60 1985
1990
1995
2000
2005
2010
Evolution of passenger traffic
1985
1990
1995
2000
2005
2010
Figure 1 :Transport indicators (ref 2 )
2010
2015
61 OLOZ OLOZ
oooz oooz
066 L 066 L
086 L 086 L
Figure 2: NOx and C02 emissions for Europe 12 by sector (ref 2)
20
contributes to the climate effect as its greenhouse potential is five times that of C 0 2 . In addition, it binds the "scrubbers" in the air thus reducing the selfpurification of the troposphere. Black fumes from cars disturb visibility and are also responsible for soot formation. But particulates which have a smaller diameter - white smoke- have risen concern to cause cancer because they contain on their very porous surface all kinds of carcinogenic compounds such as heterocyclic hydrocarbons. Particulates have such a small diameter that they pass right into the lung alveoles. An investigation into cancer statistics and occupation in London found that drivers of taxis and others diesel-powered vehicles had an increased incidence of bladder cancer. One pilot study among railway workers in the US has suggested that individuals exposed to diesel fumes stand a 42 % greater chance of developing cancer compared to individuals not exposed. In another study, of retired railway workers, an association has been found between lung cancer and exposure to diesel fumes. Automotive transport is responsible for more than 50 % of nitrogen oxides. This is demonstrated by the NOx emission map of the Federal Republic of Germany, which repeats accurately the grid of the most busy motorways. This is evident as long as only traffic emissions are reported, but still when total emissions are presented (Ref. 3). Nitrogen oxides are affecting in particular elderly people, asthmatics and people suffering from heart and lung disease. The limit value for air quality was fixed at 200 pg /m3, but that seems to be too high, especially for this group and in particular when NOx occur together with other air pollutants. The damages to vegetation due to NOx are manifold. In the acid rain discussion, nitrogen oxides were late comers, but they have not stopped ever since to raise concern. Acidification is nowadays caused to an increasing proportion by NOx, as sulphur dioxide emissions decreased over the last couple of years. Forest die-back is seen in connection with nitrogen oxides. The balance of sensitive ecosystems like moor, heather, is easily disturbed by excess input of acidifying substances, but also by the fact that nitrogen oxides act as fertilizer so bringing competitional disadvantages for plants which are sustained by extremely poor nutritional conditions. It is those plants which we find on the Red Lists of extinct or endangered species. Nitrogen oxides, together with hydrocarbons, play a role as precursors for tropospheric ozone and photochemical smog. - Ozone is a toxic gas for both man and animals. It reduces lung function. - Ozone affects plants, in particular in combinaison with other air pollutants, like NOx and SO2 - synergistic effects can be observed, entailing higher damages than one pollutant alone, e.g. reduction of harvest. -Tropospheric ozone affects the climate by contributing to the greenhouse effect.
21
Ozone data recorded between 1876 and 1910 near Paris were at a level of about 10 ppb all over the year. Nowadays, they show seasonal variations reaching a peak of 40 to 50 ppb in spring and summer. The differences in concentration and their variation are obviously due to human activities. Since 1970, the mean annual increase in ozone abundance in the Northern hemisphere has been between 0.5 and 1% per year, and about 2% in heavily polluted regions. High ozone values were reported from all countries of the European Community this summer, also from the North. Values exceeding 300 pg/m3 occurred in many Member States. WHO regards 120 pg/m3 as the threshold value where health effects occur. The Dutch Institute of public health has studied health effects at ozone concentrations higher than 240 pglrn3 and found them to provoke headache, cough, nausea, vertigo, irritation of eyes and nose, and a decrease of lung function. The importance of these effects varies with the exposition period and the quantity of ozone inhaled. The persons at risk are those working outside, people practicing sport and allergics. Frequent expositions might cause irreversible damages The Swiss, Austrian, and German ministers of the environment have agreed in August on a limit of 180-200 pg/m3 where the public should be warned to refrain from excessive outdoor physical activity. In Germany, a controversy has arisen if prohibition of automotive circulation during photochemical smog periods would reduce the risk. The Federal Minister of the Environment thought it to be useless, the Hamburg Senator was in favour of taking such a measure. In Belgium, in the region of Gent, ozone values higher than 300pg have occurred. The mayor of Gent demanded that industrial entreprises reduce their emission of air pollutants. So it can be concluded that the emissions from automotive transport have not improved: acid rain is still an unsolved problem.particulates. NO, and 0 3 continue to cause air pollution. REDUCTION STRATEGIES FOR EXHAUST GAS EMISSIONS
I In principle, it was evident since a long time that vehicles emit noxious exhaust gases and that it was necessary to reduce them by applying the best available technolow. I1 Environmentalists used to be critical with individual passenger transport. They have since long favoured public transDort for several reasons. Nowadays in the discussion, the concern of increasing air pollution and of the climate change prevails.
22
Another efficient measure to combat air pollution consists in speed limits. Reductions by 10 % could be reached by imposing speed limits which really should be applied, i.e. controlled effectively.
I11
I
-
Vehicle exhaust gas reduction by technical means.
There are two goals to be met: one consists in reducing C02 emissions by encouraging the development and the use of energy-efficient cars. The other aim is the reduction of pollutants from car exhausts to US 83 limit values by 1993 and US 93 standards by 1995 taking into consideration that only conformity of production is relevant for the environment and not type approval. This point of view is held by consumers as well and the European Parliament.(Table 1) Was there another convincing technology besides the 3-way catalyst? There were motoric measures, exhaust gas recirculation, etc. There were expectations in the lean burn engine. But nothing had ever been as efficient. The 3-way catalyst improved in performance and still does. Since the early eighties, European environmentalists have pleased for the introduction of US 83 limit values. I do not want to repeat the sad story of deceptions, when equivalence between US and EC was defined by the total amount of emissions over all the territory and not by what came out of the single tailpipe, and what was called the "Luxembourg Agreement", which fixed limit values for the nineties which were outdated already in the eigthies.
Table 1 : Limit values of exhaust emissions for cars (in g/km)
23
The proportion of cars applying state of the art pollution control has increased only slowly in Europe though faster in the contries of the Stockholm Group (EFTA countries) (fig. 3, 4 & 5). But the car market of the E R A countries is ten percent the EC's (fig. 6). Even measures taken now in adopting stricter limit values will be mitigated by the continous increase in the European vehicle fleet. Carbon monoxide and hydrocarbon emissions will rise by 2005 (fig 7 & 8) and nitrogen oxides by the turn of century already (fig. 9). 100% 90% 80% 70%
60% -
50%
-
40% -
30% 20% 10% -
0%
1985
1988
I
I
1991
1994
MODEL YEAR
S t d h l m Group
+
EmopeanCwmnnity
0
West Elrope
Figure 3 :Pollution controls on cars in Europe (state of the art controls) (Ref 1). Driving cycle The current driving cycle simulates urban driving conditions with a maximum speed of 50 km/h. Now an extra - urban part will be added, which goes up to 120 k d h . Both driving cycles will be combined and receive a common limit value (see table I), though two separate limit values would allow for a more exact determination of the pollutants, already two limit values for the urban and extra urban test cycle for CO would be an improvement. It would not come as a surprise if car types appeared on the market which could meet the commission's limit-values with an open loop three-way catalyst because they can compensate high NOx emissions in extra-urban driving conditions by low ones in urban conditions. This means compromissing at the expense of the environment. Here two separate limit values for NOx and HC would be preferable.
1: 1:
11
ia 9 8
7 8 i
5 4
3 2
f
1
0
1985
1
1
1
1988
1991
1994
MODEL YEAR Totnl New Car Sales
I
Slate of Art Cors
Figure 4 :Pollution controls on in Europe (European Community) (Ref 1) 1 .)
1 .:
1.; 1.1 1
'
0.7
0.0 0.5 0.4 0.3
0.2 0.1
0
'1985
1988
1991
1994
M W E L YEAR Tolnl New Car Sales
i
Stnte of Art Cars
Figure 5 :Pollution Controls on cars in Europe (Stockholm group ) (Ref 1)
25 14 13
12 11 10 9 -
!
7
.-
8
-
8
i
5 4
-.
3 2 1 -
0 1985
1988
1991
1994
MODEL YEAR
+
Total New Car Sales
Slate of Art Cars
Figure 6 :Pollution Controls on cars in Europe (Total Western Europe) (Ref 1) 3!
15
0 0
10
5
0
I
1985
1
I
1995
NOGROWTH
2005
+
1%
I
2015
0
2%
A
2025
3%
Figure 7 :EEC vehicle exhaust emissions trends :CO (alternative VMT growth rates) (Ref 1)
26 2.e 2.4 2 1 2
P
1 4
I? 1 2
Y
o.: 0.6
0 4 0.2
0
1985
1995
8
NOGROWTH
2005
I
1%
2015
0
2%
A
2025
31
Figure 8 :EEC vehicle exhaust emissions trends :HC (alternative VMT growth rates) (Ref 1) 4.5
4
3.5
3 2.5 2 1.5
1
0.5
0
I
1
I
1
I
1985
1995
2005
2015
2025
8
NOGROWTH
+
1%
0
2%
A
3%
Figure 9 :EEC vehicle exhaust emissions trends :NOx (alternative VMT growth rates) (Ref 1)
27
Durability testing. Durability testing is envisaged to guarantee the efficiency of pollution abatement systems. The US durability test which goes over 80.000 km will be valid also in the E.C., but provisions are made also for an alternative European test which requires testing only over 30.000 km though under particular stress, even deterioration factors would suffice in case an automobile producer does not want to carry out any durability testing. Testing after 80.000 km would be the most effective, the alternatives are compromises for economic reasons. However, these provisions only concern new cars. To guarantee the reduction of exhaust gases during the vehicle's whole lifetime, a testing requirement must be provided for in-use vehicles. Here the BEUC and the EEB would like to see the Commission coming up with a proposal for the testing of in use cars to make inspection and maintenance mandatory. US driving cvcle and limit values The recognition of US driving cycle and limit values is laid down in Annex 3A of the current Directive on car exhaust: This practice is to be phased out according to the proposal. Annex 3A was the guarantee that US limit values for car exhausts could be applied in the EC and make Member States and their authorities aware of the difference in standards. In suppressing Annex 3A we will lose the possibility to compare and an incentive to conform to stricter US limit values which should remain the goal. Otherwise we will widen the technical gap between the US and the EC market. Small commercial vehicles. As most of the small utility vehicles derive from passenger car concepts there is no reason why they should not meet the same limit values as passenger cars. Concerning energy efficiency, the EEB suggests to contemplate tax incentives for cars consuming less than 4 1 per 100 km.The use of natural gas would be useful as its combustion emits less C02 per energy unit. Natural gas had a good performance in city buses concerning the reduction of exhaust gas pollutants. Smaller, lighter and better designed cars will help to improve energy efficiency. It is not appropriate to have a vehicle weight of more than one ton to transport one person! Electric cars best fed by solar energy would be an ideal solution for some uses, but technically, they have not reached maturity in particular because storage facility and weight of batteries are still unsatisfactory.
28
Given the climate effect, diesel and lean bum rise in the appreciation of environmentalists and they regret the incompatibility of these engine concepts with the catalytic converter technology. Critical evaluation of the Catalytic technology to reduce air pollution from automotive vehicles. The 3-way catalyst has been a success. It has been the precondition for a considerable amelioration of car emissions. Platinum is a precious metal and so is not chemically reactive. Nevertheless, there have been questions about how it behaves in the environment, what kind of reactions it might catalyze. But a Swiss study gave evidence on its harmlessness. It has been tried to use oxidation catalysts to get rid of the particulates. But this was less successful as the use of an oxydation catalyst results in sulfate formation, which cloggs up the catalyst. Catalysts were poisened and had a short lifetime. In addition, there is the problem that the working temperature is too low: 140 "C instead of 300 "C. The discussion on trap oxidizers using copper and vanadium as catalysers have, not yet the expected performance, besides the fact that the use of copper might have led to copper pollution in the environment. Copper is toxic for plants and animals. Also considerations to use iron salts as catalytic additive for diesel fuel meets scepticism because under these circumstances, very small needle shaped iron salt cristals are formed. They would have the capacity to penetrate lung alveoles like asbestos fibers and it is not advisable to wait until the bad experiences repeat before iron salt additives are banned. To combat diesel particultes with the help of catalysts does not seem to be the adequate technical solution. Engine modifications for cars but also for trucks have proved to be more reliable and burner controlled particulate traps got good notes in the German field test on buses and lorries.
I1
-
Public transport.
Given the hopeless situation at the time of the Luxembourg agreement, it seemed impossible to obtain the strict limit values for car exhausts which were needed. Environmentalists looked for a resort to reduce automotive emissions and found it: public transport for passengers and rail transport instead of road haulage. This simple concept found much sympathy. Comparisons made between the subsidies received as state aid for road and rail revealed that the road user does not pay the full social and environmental cost he is generating. The amount he is paying as excise duty does not cover health effects and accidents (paid by the forest owners, farmers etc and the public at large).
29
Facts such as accidents and casualties on the road, hours spent in congestions, and the rising amount of the time which had to be dedicated to automotive transport are not to the advantage neither of the car, nor of the lorry. Transport is responsible for 58 percent of nitrogen oxide emissions and 22 percent of carbon dioxide (table 2).
Table 2 : Atmospheric emissions (9%)
-
1987
-
EEC
Source: Energy 2010 (1989). Comments to Table 2 : Transports contribute considerably to NOx emissions accounting for 60% of total emissions in 1987 in the Community. Its contribution to C02 emissions and thus to the green-house effect is also important and due to the burning of fossil fuels. The energy production sector accounts however for almost 40% of total C02 emissions. No technology exists at present capable of filtering C02. The only possibilities are less energy consumption, energy efficiency and cleaner fuels.
Of the different transport modes, road transport represents the most considerable environmental impact concerning air pollution and surface use (table 3). Air pollution from road transport is 13 fold higher concerning NOx and 20 to 60 fold more energy is needed. Only SO2 emissions are comparable (table 4). The surface covered by road infrastructure are 4,7 % in Belgium and 3 % in The Netherlands, 1,3 % in the average of the EC (table 5). The social and environmental costs of road transport -land use cost and long term cost along with the consequences of an intensification of the greenhouse effect not included - are running up to five percent of the GDP (table 6). Nonetheless, Member States of the EC continue to invest preferentially in road infrastructure instead of railways . In a market economy, state aids have to be balanced among competitors. Up to now the road has received a preferential treatement at the expense of the public transport. Environmentalists demand to put them on an equal footing in raising the excise duty on fuel by at least doubling the fuel price; the German Greens even demand an increase by the factor five.
30
Table 3 : Hierarchy of impacts per component of environment and of transport mode.
*: Small impact; **: Significant impact; ***: Severe impact; --:very small impact. (a) (b) (c)
Plus danger to transfer the problem to the electricity supply. Impact may become severe in case of mishap. Environmental effects of mobile sources at high altitude are still not well known
Comments on Table 3 : These rankings are based on conventional knowledge and experts' consensus. For a given transport mode, horizontal comparison between impacts is hampered by value judgements about the relative importance of different components of the environment (water, air, land, health). Road transport has the most significative overall impact on the environment. This is not only due to the fact that road modes are the most widely used but also to the more intense unitary impacts (per tonne-km or passenger-km) they provoke.
Table 4 : Emission factors and energy per mode of transportation
(a) Unit emissions only concern lomes : Charge factor for energy consumption equal to 50 % . With a charge factor close to 100 % , high capacity road transports should be almost as energy saving as rail.
Source : Strategic plan for rail freight. This study was ordered by the Netherlands Ministery of Transports and Waterways. (Rotterdam, July 1989 ) Comments to table 4 : By many standards, rail is more environmental-friendly than road transport. Data are not available for passenger traffic and C 0 2 emissions but it is currently suggested that the picture concerning emissions and energy consumption should be broadly similar to that of the table. This evidence provides the primary argument for a policy aimed at the encouragement of transport modes alternative to road transport (modal shift).
31
Table 5 : Road substructure (all roads), density of the road system and coverage of the ground (1985) (a). ~~
Length
Surface covered
1000 km
km2
I L NL P UK
127.9 70.1 491.2 34.5 151.8 804.7 92.3 303.0 5.2 96.3 51.9 348.3
3UR (b)
km of roads per 1000 km2
1445 792 5800 390 1600 8740 1043 3424 59 1110 586 3936
Surface covered in % of the total surf. 4.7 1.8 2.3 .3 .3 1.6 1.5 1.1 2.3 3 .O .6 1.6
4197 1627 1975 26 1 301 1466 1313 1006 2008 2582 564 1427
km of roads per inhabitant (EUR=100: 162 171 101 43 49 182 326 66 177 83 64 77
2577.2
28925
1.3
1143
100 (c)
USA
6600.0
87000
.9
704
345
JAPON
1127.5
9900
2.7
3028
117
B DK D ELL E F IRL
(a>
Colums 2 and 3 correspond to estimations made at the begining of the 8 0 s
(b)
Growth of the road system from 1975 to 1985 was of 3,7%, or 0,36%per year (0,47% from 75 to 80 and 0,25% from 80 to 85)
(c>
8 km per lo00 inhabitants
Origin: OCDE (1988) and CDP elaboration
32
Table 6 : The environmental costs of road transport (as % of GDP) Source Community of
I
Pollution (a) Congestion (b) Accidents (c) Total (d)
5
Environmental Costs in 1989 in ECU (e)
220
0.7 - 1.5 2.6 - 3.1 2.5 5.8 - 7.1
I
255 - 312
Socio-economic costs of pollution due to exhaust gases from road vehicles estimated for the Federal Republic of Germany. These costs are mainly health related. Time and energy use in the large european comdors : towns and links between towns (Bouladon report)). Socio-economiccosts of road accidents (France, Luxembourg, Belgium and the Federal Republic of Germany). The probability of accidents, in the EEC, is 125 times greater for private car users than for train passengers. Mostly non intemalised socio-economic costs. Costs related to land use are not available.
EEC's GDP in 1989 is estimated at ECU bn 4391 (EUROSTAT). Sources: OCDE (1988) (page 11) and Community of European Railways (1989) (page 24). Comments to table 6 : These data should be interpreted with caution for they come from various sources. The OCDE estimate of 5% of GDP for non-internalised costs of transports is perhaps the most solid. The other source helps to form an idea about the likely breakdown of the aggregate figure. Given that land-use costs and long-term costs associated to the consequences of an intensification of the green-house effect (C02 emissions) are not included, it is legitimate to interpret the 5% figure as a lower limit. According to all experts, these costs should be fully paid by the users of the different transport modes (via road pricing, vehicle mileage and fuel taxes, etc.). Today they are paid by tax-payers and victims and this suggests that the services of transport modes, and in particular road transport, are simply too cheap to the users.
The calculation finally on what and how much individual transport emits and how much could be avoided by using preferably public transport clearly stroke the balance into the direction of public transport In all countries, environmental organisations nowadays mobilize against the preferential treatment of automotive transport. An additional driving force is the concern raised by the greenhouse effect. Automotive transport is less energy efficient than public transport.
33
Quite new coalitions are formed and what is more important, people realize that there have been developments in transport policy of wich they are the victims and now they pronounce their disagreement. I only want to remind you of this summer, when alpine transit almost came to a standstill because one bridge collapsed. Inhabitants neighbouring the bypass routes were unwilling as an emergency to receive the extra burden of lorries. Heavy lorries are in the first place the target of displeasure. The car still maintains its popularity as can be read from the increase of cars sold over the past years. But also here changes are evident. Some urban municipalities take measures to reduce the number of vehicles in the city or ban them altogether. Nine renown Swiss holiday resorts advertised their having banned internal combustion vehicles. And these measures are by no means acompanied by a storm of indignation, on the contrary, people applaud. Automotive transport passes a period of criticism. Characteristic for the reflections was the title of a book: "Rethinking the automobile". All the different measures proposed are aimed to redress the balance which has certainly been too much into the direction of automotive transport. There are people who have never gone by train. This is an obvious sign that the possibilities are not well used. It is not to sweep the balance to public transport exclusively, but to create conditions without congestions, undue air pollution and noise, accidents and roads consuming the surface. Who wants to save automotive transport has to be for the proportionate development of public transport. That is common sense.
References 1
2 3
M.P.Walsh, "NOx emissions from road traffic in Europe" Workshop on projections of NOx emissions, Oslo, December 1989 "Major themes in energy", EC-Community Daten zur Umwelt 1988/89,pp 284-288, Umweltbundesamt Erich Schmidt Verlag
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZI 0 199 1 Elsevier Science Publishers B.V., Amsterdam
35
THE EVOLUTION OF THE EEC REGULATIONS ON VEHICLE EMISSIONS H. Henssler, Commission of the European Communities
1 - INTRODUCTION This paper deals mainly with the evolution of the regulations of the European Communities relating to the control of exhaust emissions from motor cars since the first CAPOC-seminar in 1986. As an introduction the main principles of the so called "Luxemburg Compromise" of June 1985, are briefly recalled as they represent the key for understanding the further evolution of the European emission standards. These principles were : (a) to set up European standards which in their effect on the environment were equivalent to the emission standards of the USA, however, taking into account the specific conditions of Europe concerning its car fleet and traffic patterns. (b) to allow for a choice of technologies to comply with the future European standards, these technologies being essentially the closed loop three-way catalyst and, as an alternative supposed to be cheaper, lean burn engine concepts, if necessary, fitted with simple oxidation catalysts. The realisation of these principles have led to a regulation containing 3 distinct sets of limit values according to 3 categories of engine capacities : 2 litres and more, 1.4 to 2 litres and less than 1.4 litres. In correspondance with the assumed availability of the concerned technologies also distinct implementation dates have been fixed : - For the category above 2 litres : 1 October 1988 for new car types, 1 October 1989 for the registration of new cars, - For the category between 2 litres and 1.4 litres : 1 October 1991 for new types and 1 October 1993 for all new cars and - For the category below 1.4 litres : 1 October 1990 for new types and 1 October 1991 for all new cars. The standards of the Luxemburg compromise are based on the existing European test procedure, the so-called urban driving cycle, but allowing, as a transitional measure, the alternative use of the US - FTP 75 driving cycle. These standards also require all spark ignition engines to be adapted to the use of lead-free petrol of 95 RON. They are equally applicable to diesel engines, however, engines with more than 2 litres being submitted to the standards of
36
the medium category. The Luxembourg compromise had to wait until 3 December 1987 to become an EEC Directive thanks to the European Single Act of February 1986, which introduced the possibility of decisions by a qualified majority, thus overcoming the Danish and Greek vetos. This Directive with the reference 88/76/EEC represents the 5th amendement to Directive 70/220/EEC which had established in 1970 the basic emission standards of the European Community. For the sake of completeness it may be added that the United Nations' Economic Commission for Europe translated in 1988 this latest EEC Directive into a new ECE Regulation no 83 intended to replace their traditional regulation no 15 covering motor car emissions.
-
2 FURTHER EVOLUTION The further evolution of the European emission standards since 1985 has been largely determined by a number of commitments which were part of the Luxemburg compromise or have been taken in connection with the adoption of Directive 88/76 : These commitments concerned : - the substitution of the interim standards for the car category below 1.4 litres by more severe European standards - the completion of the existing urban test procedure by sequences representing driving conditions on highways and motorways - the setting of standards for the emission of particulate matter by diesel powered cars - the introduction of requirements concerning the durability of emission control devices and the evaporation of fuel.
2.1 - The Extra-Urban Driving Cycle Immediately after the Luxemburg Council the Commission charged its well established expert group on vehicle emissions known under the abbreviation MVEG with the technical work resulting from these commitments. This group is composed of the experts of the interested national administrations, of the motor and petrol industry and of consumer and environmental organizations. Its work in relation to the future European extra-urban driving cycle started on the basis of studies of typical highway and motorway driving patterns in Germany and the United Kingdom and by the Committee of Common Market Constructors. The results of these studies were put together and resulted in a synthesized driving cycle which was presented in March 1988 as the European Extra Urban Driving Cycle "EUDC", caracterized by a relatively simple driving pattern and presenting a peak speed of 120 km/h and an average speed of 62.6 kmh. MVEG then examined the possible combination of this cycle with the
31
existing urban driving cycle with the help of tests carried out on a representative sample of 133 cars of different types and featuring different emission control systems. Eventually, the group agreed on a combination where the urban cycle precedes the extra-urban cycle, a combination which became known as "MVEGA" and which preserves the potential of the urban cycle to measure emissions at a cold start.
2.2
-
Diesel Particulate Standards
In parallel, MVEG worked on the future European standards for the emissions of particulate matter of cars powered by diesel engines on the basis of an investigation of the actual emissions from present European diesel car models. As sampling and analyzing method that of the present US regulation was retained and efforts were made to adapt it to the European test procedure. In June 1988 this work resulted in the adoption of Directive 88/436/EEC. Because of the lack of precision of the sampling method and the resulting relatively high limit values (1.1 dtest for type approval and 1.4 g/test for the control of production conformity), it was understood that this Directive could only represent a first stage of the particulate control in the EEC and the European Parliament fixed as the objective for a further more stringent stage with 0.8 g/test for type approval and 1.0 g/test for the control of production conformity. 2.3
- Emission Standards for Small Cars
Finally, MVEG examined the question of definitive European emission standards for the car category below 1.4 litres. The result of the technical work suggested that if the principles of "technical choice" and "reasonable costs" were to be retained, the limit values for the gaseous emissions of the small car category should be equal to those of the medium category i.e. 30gtest for CO and 8gtest for the combined mass of HC and NOx. These limit values were proposed by the Commission in 1987 and were approved in November 1988 as the common position of the Council and the Commission in accordance with the Single Act procedure. The European Parliament, however, with which Council and Commission have to co-operate under this new procedure, was of the opinion that the environmentai situation had becoms dramatic in the EEC, on the one hand, and that the closed loop three-way catalyst technology was now available to the whole European car industry, on the other hand. Consequently, the Parliament requested limit values which are at least as severe as those of the USA for the small car category which represents roughly 60% of the present European car fleet. After some discussions, the Commission followed the Parliament and
38
presented in May 1989 a revised proposal. The majority of the Member States agreed in June 1989 and adopted directive 89/458/EEC with the limit values of 19 gjtest for CO and 5 g/test for HC+NOx (type approval limits) to be applied on 1 July 1992 to new car types and on 31 December 1992 to all new cars. This decision represents another milestone of the motor vehicle emission regulations of the European Community : The principle of the technological choice has been abandonned and replaced by the principle of the best available technology on which the future European standards have to be based. Consistantly, the Council committed the Commission to present a proposal which aligns the limit values of all car categories to those of the small car category and, at the same time, to adapt these standards to the improved test procedure including the extra-urban driving sequences.
2.4
-
The Consolidated Directive
In compliance with these commitments, the previous ones from June and November 1985, and the recent ones from June 1989, the Commission decided to comprehend in one directive the different requirements relating to gaseous and particulate emissions of passenger cars and thus to consolidate the EEC emission standards for this vehicle category. Of course, the proposal is based on the results of the MVEG work described above : the driving cycle completed by the extra urban sequences, the improved particulate sampling method and the specifications for testing the durability of emission control devices and the evaporation of petrol from the fuel system of the cars. The limit values of this proposal for the gaseous emissions constitute a purely technical transposition of the small car standards into the new complete test procedure. The proposed particulate standards correspond to the objective fixed by the Parliament (0.8/1 .O gjtest). For this transposition the appropriate results of the 133 car sample established for the choice of the driving cycle combination has been used i.e. only those cars, both petrol and diesel, whose gaseous emissions were below the limit values of 19 g/test CO and 5 @test HC+NOx. The limit values which resulted from this operation are for the approval of new car types : 2.72 g/km C0.0.97 g/km HC+NOx and 0.19 g/km particles.The corresponding values for the control of production conformity are 3.16g/km CO, 1.13g/km HC+NOx and 0.24g/km particles. It is worth to note that these standards represent approximately 10% of the limit values of the first EEC directive of 1970 (70/220/EEC) relating to emission control. According to international practice, these standards are now expressed in "grams per kilometer" of the new European test procedure combining the urban and the extra-urban driving cycle representing a test distance of 1lkm. In accordance with the directive on small cars these limits are proposed to apply from 1 July 1992 to new car types and from 1 January 1993 to all new cars.
39
The Commission wanted of course to be sure that these limit values comply in fact with the postulate to be at least as severe as the existing US standards. Hence, a study was commissioned to 2 national testing laboratories in order to compare the actual emissions measured with the proposed European test cycle to those measured with the US test procedure FTP'75. The study was carried out with 44 representative European car models- 31 with petrol engines and equipped with closed loop three-way catalysts and 13 with diesel engines - complying with the CO standard of 19g/test and the HC+NOx standard of 5g/test of directive 89/458. This study showed despite an important scatter of the results, a clear trend of the European test procedure to yield generally higher emissions than the US procedure. This trend was most obvious for CO where the European test resulted in average in 45% higher values than the FTP'75 test. For HC+NOx and particulate emissions the average of the results were 15% higher than in FTP'75 test. In terms of limit values, this means, that the proposed European standards for gaseous emissions are indeed more severe - about 12% for CO and 4% for HC+NOx than the US standards. However, for particulate emissions the study showed that the objective fixed by the European Parliament and followed by the Commission has been too lenient : a value corresponding to the present US standard of 0.20gImile would be 0.14gkm In the European test and the corresponding standard for the control of production would be 0.18g/km. The Commission is prepared to correct its proposal in this respect.
3
-
ASSESSMENT OF THE PROPOSAL FOR A CONSOLIDATED EMISSION DIRECTIVE AND ITS POSSIBLE INDUSTRIAL EFFECTS
The main characteristics of the proposal are, on the one hand, the strict respect, by the Commission, of the previous agreements and commitments which are mentionned in the first part of this paper. The Commission believes in the continuity of the EC policy on vehicle emission standards because of its economic importance, in general, and of its bearing on the European automotive industry, in particular. On the other hand, this proposal is caracterized by the determination of the Commission to find a genuine European solution to the environmental problems caused by road traffic, a solution which is most adapted to the European conditions in respect of both its car fleet and its driving patterns. It needs to be said that the proposal which was presented to the Council of Ministers on 5 January 1990 and to the European Parliament on 22 February 1990 has not received a unanimous approval from these institutions. The majority of the European Ministers who discussed it at their meetings of 22 March and of 7 June 1990 declared it nevertheless to be a good basis for a rapid decision and confirmed their interest in a genuine European solution. The problems which have been identified so far appear to be more of a technical nature than fundamental. The European Parliament, at least its
40
Environmental Committee, appears, however, convinced that, in principle, the best for Europe would be to copy the American emission regulations. In the forthcoming dicussions, the Commission has decided to stand firm concerning its proposals and the following months will show what eventually, will be the solution which rallys the majorities of the European institutions. In any case, whatever the solution will be, from an industrial point of view, it appears clearly that from January 1993 on all cars produced for the market of the EEC in big series will be, independent of their engine capacity equipped with closed loop three-way catalysts if they are powered by petrol engines. As far as diesel powered cars are concerned, the proposed standards both for gaseous and particulate emissions, do not assume the use of catalysts or trap oxydizers. The reason for this is that expert consultations did not allow the Commission to assess neither the potential of these techniques, nor the time of their availability, with sufficient accuracy. The Commission will, of course, follow their development and, if possible, adapt the standards to their potential, in order to generalize their fitting to the concerned vehicles. 4
- OUTLOOK INTO THE NEXT FUTURE
The Commission is aware that catalysts deteriorate within a certain period of usage. The regulatory durability tests available now including the US - 50 000 miles- test which has been also included into the directive proposal do not offer a safe means to determine the moment when catalysts are no longer efficient and have to be replaced. The Commission believes that only a Community-wide system of periodic inspections of the cars in use is appropriate for this purpose and allows eventually to take the full environmental benefit of this technology. Consequently, the Commission is actively working on a directive proposal in this respect which it deems necessary to complete its present proposal relating to vehicle construction. Another complementary measure which the Commission will present in a few months, concerns the emission standards for all vehicles covered by the scope of this directive and not being passenger cars. Such vehicles, generally being light commercial vehicles of category N1 but also heavy passenger cars having more than six seats and/or a total mass exceeding 2.5 tomes, are presently governed by transitional provisions which submit them to more lenient standards than passenger cars. The reason for this is essentially that such vehicles have a power-to-weight ratio considerably inferior to that of typical passenger cars. Consequently, these vehicles have higher emissions than passenger cars under identical test conditions. At present, we are studying, a system of standards which is adapted to the technical conditions of these vehicles but represents the same level of severity as the proposed passenger car standards. Consequently, it can be foreseen that after the mid-nineties all motor vehicles driven by petrol engines will have to be equipped with closed loop three-way catalysts.
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V., Amsterdam
41
THETECHNOLOGICAL DEVELOPMENT OF THE EUROPEAN VEHICLES AND THE ENVIRONMENT
C. Cucchi and M. Hublin
CCMC 1. INTRODUCTION
On June '85, the Council of Environment reached an agreement on a plan for further tightening of the emission standards for passenger cars. This plan, known as the "Luxembourg Agreement", was motivated by the need to contain the impact of the road traffic on the environment and was intended to establish an equivalency between the total amount of pollutants emitted by the the European and the US car fleets, taking into account the difference in the composition and the patterns of use of the two vehicle parks. Table 1 summarizes the decision reached on the emission limit values and the corresponding dates of implementation. The basis for the following further actions were also layed down at that time: - Step I1 emission standards for small cars with an engine displacement below 1.4 litres - limit values for particulate matter emissions from passenger cars equipped with a diesel engine to be implemented in two steps. - introduction of an added driving cycle representative of extra-urban driving conditions. - limit values for vehicle evaporative emissions - durability requirements The Luxembourg Agreement was transposed into a Directive (Dir. 88/76/EEC dated 3 December '87), with a two-year-delay as a consequence of the late coming into force of the Single European Act Since '87, the Commission, with the support of the MVEG, has defined the issues that were left open. In particular: - a limit value of PM = l.lg/test was selected for diesel vehicles to be implemented on 1/10/90 for all vehicle types. - Step II emission standards for passenger cars below 1.4 It. was defined as: CO = 19g/test; HC + NOx = 5g/test and the corresponding dates of implementation put forward to 1/7/92 - 31/12/92 - the driving sequence shown in Fig. I was developed to reproduce the extra-urban driving conditions, and between the possible combinations
42
of the urban and extra-urban driving sequences Fig. 2, configuration A was retained as the new European driving cycle. - a test procedure to measure vehicle evaporative emissions within a European context was developed.
TABLE 1 Luxembourg Agreement Exhaust Emission Limit Values
I
CAR
NOx GfIEST
CATEGORY
Gasoline Engines >2000 cc
I
DATES IMPLETATION
I 3.6
NM 1/10/91 AT 1/10/93
1400-2000 cc
t
NM* 1/10/89 AT 1/10/89
I
el400 cc STEP 1
45
15
5
TO BE DEFINED
NM 1/10/92 AT 1/10/93
Diesel Engines** >1400 cc
c 1400 cc STEP 1
45
NM 1/10/90 AT 1/10/91
NM 1/10/90 AT 1/10/91
1
15
6
NM 1/10/90 AT 1/10/91
STEP 2
* **
NM-NEWMOD FOR DIRECTINJECTION DIESEL ENGINES. 2000 c c , THE DATES OF IMPLANTATION HAVE BEEN FIXED AT 1/10/96 FOR ALL TYPES.
While particulate emission limit values and Step I1 standards for small cars have already been finalized in Directives (Dir. 88/436/EEC of 16 June 1988 and Dir. 89/458EEC of 18 July 1989 respectively), the new driving cycle, the evaporative emission requirements and the durability requirements are planned to be introduced in a New Consolidated Directive which will impose a set of standards common for all car categories as from 31/12/92 (1/7/92 for new models). The Commission has recently published a proposal on this issue (COM (89) 662 dated 2 February 1990) as discussed in the paper “Evolution of the EEC Regulations on Vehicles Emissions” presented by Mr. Henssler
43
during this meeting. This fast evolution of the emission requirements has brought about an intensive engineering activity for the development of emission systems optimized within a European context. The auto industry has always felt that the technical solutions defined for the US market may not always be the most suited for a vehicle park which differs from the point of view of vehicle size, engine displacement/types and patterns of use. The economic aspects of new legislative requirements in terms of vehicle purchase prices and operating costs cannot be overlooked. 120
100
-
STATISTICS: LENGTHS :
-
VMAX.:
i5
i
6955 rn 400 S 120Kmh 62 Km/h
80-
s E3w
60-
n 40-
20
-
00
I
I
I
I
I
I
I
I
50
100
150
200
250
300
350
400
I
I
I
TIME (Sec.)
Fig. 1 EUDC Driving Cycle 2. EXHAUST EMISSION TECHNOLOGIES
The evolution of the car park in the US has been characterized by the following trends as a consequence of the emission requirements that are being enforced in this country: - exclusive use of electronically controlled fuel metering systems for gasoline engines - exclusive use of the three-way catalyst concept with or without an added oxidation catalyst - disappearance of the diesel engines despite the interest shown even by local manufacturers as a consequence of the corporate average fuel economy requirements and the large investments made. - very large percentage of vehicles equipped with fuel injection systems. Practically all US and European models are fuel injected.
44
120 -100 c 2 180 A;;W
40 v)
20 0
0
100 200
300 400
500
600 700 800
900 lo00 1100 1200
Time (s)
120
..-. 100 280
B Z W $40 a 20 0
0
100
200
300 400 500
600 700 m 900 lo00 1100 1200
Time (s)
1
120 -f
.
I I
80 60 40 20 0
0
m 40.
a
(I)
20.
100 200
n
300 400 500
600 700 800 900 loo0 1100 1200
I
nI7 I Time (s)
120 100 80
60 40 20 0
0
100 200
300 400
500 600 700 Time (s)
800 900 loo0 1100 1200
Fig 2 :Driving Cycle Configurations
45
The final decision of the European legislators concerning the emission standards in terms of driving cycle and limit values will determine the future situation in Europe.
2.1 Gasoline engines Various levels of severity, in terms of driving cycle and limit values, have been brought forward since the time of the Luxembourg Agreement. Each of them has a different impact on the technological choices available to the manufacturers and on the final price of the vehicles. The following paragraphs review the main proposals.
2.1 .I Urban Cycle - CO = 30gltest ;HC
+ NO* = 8gltest
These are the standards chosen for medium cars at the time of the Luxembourg Agreement and, at an early stage, also proposed by the Commission for small cars as Step II. They offer the choice between different technolgies such as: - lean-bum engine with an oxidation catalyst - three-way catalyst in an open-loop configuration - engine modifications + fuel injection system - engine modifications + EGR + oxidation catalyst As a consequence, the car manufacturer remains free to select the optimum solution on the basis of the following criteria: - emission reduction efficiency - engine performances including fuel consumption - driveability - system costs Of particular interest for this level of seventy is the lean-bum technology.On the basis of the response of gasoline engines to the aidfuel ratio Fig. 3, it is theoretically possible to reach optimum emission and fuel consumption levels in the lean region. However, under these conditions, engine operations become critical due to an increase in the combustion duration and in cycle to cycle scatter. To push forward the lean missfiring area, it is necessary to create high turbulance in the combustion chamber by resorting to high swirl intake ducts, air jets at the valve or butterflies valves in the intake ducts Fig. 4. The definition of a lean burn engine requires long development times. The optimum performances that can be achieved in terms of fuel consumption are counter-balanced by the following issues when compared with a conventional engine: - increase in HC emissions - higher NOx emissions during the cold phase
46
- higher dispersion of the emission levels as a function of the tolerances of the fuel metering system - driveability.
It
T
"1-
30
10
m ;lrH
N E L ~ Y s T I o N
mo
m$sa
M 1
14
1
1
18
1
1
18
1
1
20
1
8
1
p
iuno
1
24
Fig. 3 Engine Emissions and Specific Fuel Consumption vs AIF Ratio
Fig. 4 a :High swirl intake ducts
Fig 4b :Buttelflies valves at the intake ducts
Fig 4c :Air jet at the valve
47
As a consequence of the above, an oxidation catalyst and an improved carburetor or an injection system are needed to meet emission limit values as stringent as HC + NOx =8g/test. Further reduction in the emission levels are difficult if not impossible to achieve as NOx emissions can only be controlled through an optimization of the air/fuel ration setting. No EGR or three-way catalyst can be used in a true lean bum engine for obvious reasons. 2.1.2. Urban cycle - CO = 25gltest ;HC
+ NOX = 6.5 gitest
These are the standards that are in force in Europe since 10/89 for passenger cars with an engine displacement above 2.0 It.. This level of severity can only be met by the technolgy of the three-way catalyst in a closed-loop configuration. This technology is based on the property of Platinum and Rhodium to act as catalyst for both the oxidation of CO and HC and the reduction of NO,. Clearly, the highest efficiencies for the oxidation reactions are achieved when the exhaust gases flowing through the catalyst have an excess of O~(1ean mixture) and the highest efficiency for the reduction reaction when they are poor in 0 2 (rich mixture). Nevertheless, there is a window around the stoichiometry where the conversion efficiencies of the catalyst for both the oxidation and the reduction reactions are satisfactory (80-90%). An electronically controlled fuel metering system is required to ensure the correct composition of the exhaust gases flowing through the catalyst under all the engine running conditions. An oxygen sensor detects any deviation from stoichiometry , sends a signal to a control unit that corrects accordingly the amount of fuel delivered (closed-loop systems). In practical terms, the aidfuel ratio achieved with such systems is never set on a constant value but oscillates around a constant value (- stoichiometry). The frequency and amplitude of these oscillations are a function of system transport time (delay between the information given by the sensor and the corresponding modification of the A/F ratio at the inlet valve) and the control strategies. The overall efficiency of the three-way catalyst results from see Fig.5: - the base calibration of the fuel metering system - the frequency and amplitude of the aidfuel ratio oscillations around the above value. Therefore, the selection of the fuel metering system and the control strategies play a key role in the achievement of low vehicle emission levels. Electronically controlled carburetors, throttle body fuel injection systems and multipoint fuel injection systems are today available for use in combination with a three-way catalyst. In the category of passenger cars above 2.0 It, fuel injection systems are almost exclusively utilized.
48
1m
-
-
*
;E
t,
zp w U
aa-
)(01
m-
Hc
2
P
40-
v)
a
z8
10-
m
o
,
,
,
,
u
,
,
,
,
I
Fig. 5 (a) Catalyst Eficiency om
om
U11
1m
14.60
1.aY
1.m
IS09
14-
AIRIFUEL RAno AMPLITUDE 0.3 M FFIEWENCY 1.8 HZ
-
i RICH
om
I
0.1 I l
1
I
LEIN
1.w
1.02
I
A
RICH
A L E A N
too
I
I
NOx
HC
8,
NOx
10
WPUNDE
AIR'FUEL wno 1.0 M . FRCOUENCY 1 Hz
AYPUNDE
URlRlELRATIO 1.0 M - FREQUENCY 0.7 +k
+_
Fig. 5 (b) Three-Way Catalyst Efficiency
49
2.1.3 Urban cycle - CO = I9gltest ;HC
+ NOx = Sgltest
These are the emission standards chosen as Step II for small cars below 1.4 litres. They can be met only by the technology of the three-way catalyst in a closed-loop configuration. In practical terms, they correspond to a level of severity more stringent than that of the US '83 requirements as a consequence of the specific configuration of the European urban driving cycle. Their approval by the Council of the EC will have an evident impact on the vehicle prices (see Table 2) and their operating costs. It will also require large investments, especially for the auto industry suppliers, as a consequence of the widespread use of the electronics for the engine management and the phase-out of the production of carburetors. Electronically controlled carburetors are not always the most economical technology due to their intrinsic characteristics (poor control of the A/F ratio if compared with fuel injection systems). 2.1.4 New European cycle - CO = 2.72glkm ;HC
+ NO, 0.97glKm
The above limit values have been proposed by the Commission for all categories of passenger cars in connection with the introduction of the new European driving cycle. They are intended to meet the following prerequisites: - to be at least as stringent as the US '83 standards - to align the requirements for passenger cars with an engine displacement above 1.4 It. to the standards already approved for small cars (i.e. Step I1 - urban cycle - CO = 19dtest HC + NOx = 5g/test). The new European driving cycle is unique as it introduces driving modes at vehicle speeds higher than the ones covered by other test cycles including the US highway cycle. Under heavy load conditions, CO and NOx engine emissions rise sharply and the means to control them are limited without penalty on driveability and/or fuel consumption. As a consequence, the technologies today available to abate vehicle exhaust emissions must be reconsidered in function of the new cycle and the corresponding limit values. A comprehensive study that was recently carried out by CCMC shows that the three-way catalysts in a closed-loop configuration technology is already required if the limit values for the new European driving cycle are set at: CO = 5-
HC + NOx = 1.25 g/km see Fig. 6
50
TABLE 2 Emission Technologies and their Impact on Emission, Reduction, Price, Increase and Fuel Consumption Variation Consumer Fuel Price Consumption Increase Increase
Technologies Conventional Engine for "Lux. Agreement"(Step 1) Lean Bum Engine with Carburetor and Conventional Ignition Conventional Engine for Tax Incentives in
(%I
(%)
0.5
2
*
Overall Emission Increase
Limit Values
0 22 **
1.o
-2
22
1.5
3.5
34
Luxemburg Agreement step1 cQ=45g/t HC+NOx = 15 g/t
FRG-NL Pulsair and EGR LeanBum Engine with Carburetor and Recalibrated Conventional engine with EFI Lean-Bum Engine and EFI Lean-burn Engine
4.5
Davignon 1
I
Open loop-3Way-Cat Carburettor Lean-burn Engine Closed Loop - EFI Var. Intake System 3 Way-Cat
** ***
34
I
9*
34**
-15%
CO = 38 g/t HC+NOx = 12.8 g/t
2
34
-7
34**
-3
53***
Commission Proposal
+2
53***
CO = 30 g/t HC + NOx = 8 g/t
-7
approx. 60
+3
73
US '83 or equivalent
Fuel Consumption change is the average of 90/120/Urban consumptions Some of the technologies described above are only at an early stage of development. This varies from manufacturer to manufacturer
BASELINE = small vehicle, 1.4 L conventional carburetted engine meeting ECE 15/04
51
4
HC
+ NQx
3
2 1,25 1
0
CQ
0 0
2
4
5
6
8
10
Fig. 6. CCMC Study on the Impact of the New European Driving Cycle on Emissions- Lowest Levels Achieved by Vehicles not equipped with a three-Way Catalyst and Closed-Loop System. Other technologies such as a three-way catalyst in an open-loop configuration or an oxidation catalyst + E.G.R., which can still achieve very low emission levels in limited cases on the urban cycle, cannot control at the same time CO and HC + NO, emissions below the above figures when tested according to the new driving cycle . The Commission proposal, being even more stringent than the above values, will require added optimization of the present state of the art technology (i.e. three-way catalyst in a closed-loop configuration). Control strategies of the engine management system, catalyst composition and design, auxiliary devices such as E.G.R. must be reengineered on the basis of the constraints posed by these new European standards in order to achieve the desired result. Models meeting the US '83 requirements do not necessarily have emission levels below CO = 2.72g/km when tested according to the urban + extra-urban driving cycles as shown in Fig. 7. 2.2 Diesel engines
At present, the emission standards for pasenger cars equipped with diesel engines are set at CO = 30g/test, HC + NOx= 8g/test, PM = 1.1 g/test (but for engines below 1.4 It for which CO = 45g/test, HC + NOx= 15g/test). The only constraint is represented by the trade-off between particulate matter and HC + NO, emissions. The optimization of the engine calibration is sufficient to meet todays requirements. The Commission proposal concerning the introduction of the new European driving cycle provides also for a
52
further tightening of the particulate emission limit values. This means that diesel engines will have to meet: HC + NO, = 0.97g/km,
CO = 2.72g/km,
PM = 0.19 g/km.
Under these conditions, the trade-off PM/HC + NO, becomes critical within the technologies presently available to abate diesel exhaust emissions, i.e.: - combustion chamber design - fuel injection system design - EGR - improved diesel fuels. ---. .. -- .. . - M TEST I I t S U L l S IN t-i"/ K 5.0
-EEC PROC.
SLOPE
CO (With DF)
=
1.452
3.5 3.0 -
4.5
4.0
2.5
-
1.5 -
2.0
1.0
-
0.5
-
*/
US PROC.
0.0 0.0
I
I
I
I
I
I
I
I
I
1
0.5
10
1.5
2.0
2.5
3.0
3.5
40
4.5
5.0
0 GASOLINE VEHICLE
DIESEL VEHICLE
Fig. 7 Comparison between Emision Levels Achieved on the New European Test Procedure by Vehicles Designed for the US '83 Standards. It should be noted that NO, emissions cannot be dealt with an after-treatment system such as a three-way catalyst. At the same time, particulate filters for automotive applications are being investigated with still poor results as far as the filter life is concerned. Only oxidation catalysts have proved to be useful in reducing HC levels and, to a somewhat lesser extent, also particulate matter levels. The need of using oxidation catalysts as a possible essential mean to meet the future legislative requirements add to the urgency that the car manufacturers feel concerning the establishment of appropriate specifications for the European diesel fuels. Diesel fuel quality has an important impact on the exhaust emissions of diesel engines. In addition, the widespread use of catalysts requires the availability of sulphur-free diesel fuels.
53 3. EVAPORATIVE EMISSIONS
Passenger cars contribute to the emissions of Volatile Organic Compounds due to the evaporation of the gasoline during prolonged parking as a consequence of the fluctuations of the ambient temperature, when parked after use and, in a lesser amount, when used. In line with the mandate of the Luxembourg Agreement, the recent Commission proposal also introduces requirements in this area. A test procedure has been defined to measure the vehicle V.O.C. emissions during prolonged parking (diurnal losses) and when parked after use. A 2gltest limit value was proposed. The technology presently available to reduce this type of emissions and attain the above goal is based on the following principle: - the fuel system (fuel tank, carburetor bowl, etc ...) is vented exclusively through an activated carbon canister - the carbon canister is purged by fresh air drawn through the canister into the intake manifold during engine operations. This technology also allows the control of any evaporative loss when the vehicle is running. The constraints added by the evaporative emission requirements have an indirect bearing on the selection of the technology to abate exhaust emissions as: - the fuel metering system must be able to deal also with the canister purge flow which contains an unmeasurable amount of fuel - some fuel metering systems generate higher amount of fuel vapour than others (for example: carburetors versus fuel injection sytems). 4. CONCLUSIONS
The increased concern of the public opinion about the environment and the need to protect it has spurred the Commission towards the implementation of emission standards of increased severity. With the new Consolidated Directive recenlty proposed by the Commission, the legislative requirements concerning exhaust emissions will become more stringent than the corresponding requirements presently in force in the US. This will mean that all gasoline cars will be soon equipped with 3-waycatalysts in a closed-loop configuration while diesel engines will incorporate the latest development in terms of engine and fuel system design. This evolution is putting a strain on the industry that in a short period of time has to adapt the technology of the 3-way catalyst, known through the US experience, to the European requirements in terms of test procedures and specific patterns of use of the vehicles. There will be added costs due to the complexity of the new systems and the need for large investments in the component sectors, in order to meet the demand for new sophisticated products.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 199 1 Elsevier Science Publishers B.V., Amsterdam
55
INDUSTRIAL APPLICATION OF CATALYSTS FOR OFFGAS TREATMENT Dr.-Ing. Torsten Schmidt,
Kali-Chemie AG, Hannover, West Germany
Table of Content. 1 1.1 1.2 1.3 1.3.1 1.3.2 1.4 2 2.1 2.2 2.2.1 2.2.2 2.2.3 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2 3
Fundamentals Air pollution control: legislation Air Dollution control: process selection Principles of catalytic reactor desia Chemical reactions, heat and mass transfer Catalyst deactivation Plant design Examples of.apdication Selectivecatalytic nitric oxide reduction 0ffg.a~ treatment ef stationam internal combustion eneines or turbines Exhaust gas emission control for spark ignition engines according to the three way system Diesel engines Gas engines (two stroke), gas turbines Claus plant o f f a SULFREEN process Catalytic afterburning Catalytic $terburning of VOC for industrial production processes Survey Example: Laminating of paper Literature
SUMMARY Catalytic offgas treatment requires adaptation of catalysts to the existing production pmess and its emissions. Industrial catalytic offgas treatment offers opportunities in conversion of noxious emissions into environmentally compatible reaction products at low temperatures without formation of noxious by-products or residues. Due to the versatility of catalysts and offgas treatment processes their application yields various economies. Limitations include catalyst deactivation due to presence of poisons or the Occurence of concentrationpeaks. Those are tackled by chemical and engineering research and development of catalyst and plant manufacturers. Kali Chemie AG, Hannover @). offers optimized catalysts as a result of long term experience in cracking, isomerization, reforming, chemical, automobile exhaust and industrial offgas treatment catalysts.
56
-
1 FUNDAMENTALS 1.1 Air pollution control: legislation Technological measures for reduction of gaseous emissions to the atmosphere can be divided into three categories and priorities. 1. Avoidance of emissions by process optimization (primary measure) 2. Recycling of emissions (secondary measure) 3. Conversion of emissions (tertiary measure) [2] Catalytic processes for air pollution control are employed for the latter purpose. In 1987 the catalyst market consisted of three major fields of application: - refineries 950 Mio $ - chemical industries 1.496 Mio $ - air pollution control 568 Mio $ Market volume of air pollution control catalysts is divided into automobile exhaust purification (90%) and industrial offgas treatment (10 %)'. Performance of offgas treatment processes must comply with statutory regulations. Thus, design of plants and catalytic reactors must yield conversions of pollutants which ensure concentrations below the threshold levels specified for the particular pollutant and emission process. In West Germany threshold levels are mandatory, and are based on the Federal Emission Control Law (Bundesimmissionsschutzgesetz)and particular instructions (e.g. TALuft; GroPfeuerungsanlagenverordnung). 113 Some examples of emission threshold levels are given in table 1.
1.2 Air pollution control: process selection The primary goal of catalytic processes in air pollution control is the conversion of noxious emissions (e.g. hydrocarbons) into environmentally compatible reaction products (e.g. carbon dioxide and water) at minimum energy consumption and without formation of either noxious residual substances or noxious by products. Unlike many catalytic processes in refining or in petrochemicals, the basic principle in air pollution control is that the catalyst has to fit to the (existing) production process and its operational parameters.
1 - Source: Prof. J.T. Richardson: Introduction on Catalysis: Application of heterogeneous catalysts, Loughborough, 31.07.88-06.08.88
57
Table 1 - Emission Limit Values According to German Air Pollution Control Legislation POLLUTANT
THRE~HoLDLIMIT
THERMAL POWERPLANTS
NOx
0,2 G/M3
TWO STROKE GASENGINES
NOx
0,80 G/M3 0,65 G/M3
co
NM-HC ALDEHYDES SOLVENT APPLICATION
(E.G.P R I ” G , LAMINATING) CHEMICAL SYNTHESIS PLANT
0,13 G/M3 20 MG/M3
VOC CLASS III (E,G,ACETONE) CLASS II (E,G, TOLUENE) CLASS I (E,G, FORMALDEHYDE)
0,15G/M3 0,l G W 3
20 MGM3
The basic offgas parameters are: (i) atmospheric pressure (ii) low temperatures (iii) low concentration of pollutants (below 1 vol.%) (iv) wide variety of substances with several changes during catalyst service time (v) non stationary operation. [ 10,1] Catalytic processes are evaluated according to the following selection criteria: (i) efficiency (compliance with threshold levels) (ii) effort (energy consumption; demand for maintenance and supervision) (iii) reliability (iv) compatibility with production plant and energy supplies (v) avoid residues or by products (e.g. spent catalysts, filters, products of incomplete oxidation) Competing processes many include: condensation, adsorption, absorption, reverse osmosis, biological treatment, thermal oxidation. Recently catalyst application is state-of-the-art in the following fields of application: [ 123 - flue gas treatment (NOx control by selective catalytic reduction; SCR) - stationary internal combustion engine offgas treatment (oxidation of CO and HC; non selective NOx reduction) - treatment of Claus plant offgases (oxidation of H2S, CS2, COS) - complete oxidation of volatile organic compounds (so called catalytic af terburning)
Table 2. Chemical reactions in Catalytic Air Pollution Control
OXIDATION EXAMPLE
COMPLETEOXIDATION OF HYDROCARBONS CmHnOp+ (m+n/4-p/2) 0 2 -+ mC02 + n/2 H20
REDUCTION
DECOMPOSITION
SCR 4 NH3+ 4 NO + 0 2 + 6 H 2 0 + 4N2
DECOMPOSITIONOF OZONE 203 -+ 302
NSCR CO +NO + 1/2 N2 +C02 CmHn +2 (m +n/4) NO (m + n/4) N2 +n/2 H20 + mC02 TEMPERATURE RANGE
200 - 600 "C
RESIDENCE TIME
0.42 - 0.02 s
GHSV
5000-50000 h-
APPLICATION
CATALYTICAFI'ERBURNING; ENGINEOFFGAS
SCR NSCR 200 - 400 "C 2 - 0.3 s
0.42 - 0.02 s
1000-5000 h-1 5000-50000 h-1
DENOx
ENGINE OFFGAS
20 - 100 "C
0,7
- 0,3 S
5000-10000 h-l STERILISATION
59
1.3 Principles of catalytic reactor design
1.3.1 Chemical reactions, heat and mass transfer Chemical conversions arising in connection with chemical and biological exhaust gas treatment processes can be subdivided into - oxidation reactions - reduction reactions - decomposition reactions. Catalvtic exhaust gas purification refers to processes in which the polluting gaseous and vaporous substances (pollutants) are converted into harmless substances by means of chemical reactions on the surface of a solid, inorganic auxiliary material (catalyst). [ 5 ] For that reason, technical catalysts consist mainly of highly porous materials which provide a large surface area for the chemical reaction. Due to the fact that the reactants (pollutants and oxidation or reducing agents) are present in a different physical state than the catalyst, there are not only chemical reactions at the catalyst but also physical transport phenomena (table 2). [3] The essential mass transfer steps of such heterogeneously catalyzed reactions are (figure 1): -1. mass transfer between the offgas flow and the boundary layer, -2. diffusion of reactants through the boundary layer to the outer (geometric) catalyst surface and from there through pores to active sites on the internal surface, -3. adsorption at active sites, -4. chemical reaction, -5. desorption of reaction products from active sites, -6. diffusion of products through pores and boundary layer, -7. mass transfer into the offgas flow. The application of solid catalysts in gas reactions (heterogenous catalysis) increases the velocity with which the reactions in question are occurring through - compression of the reactant gases at the catalyst surface, which results in an increased number of collisions, - decrease of the activation energy. As a result of the increased velocity of reaction, the throughput of air polluting substances achieved by the chemical reaction is reached (i) at lower temperature (ii) at smaller reactor dimensions (shorter residence times) compared with thermal processes. High reaction rates over highly active catalysts lead to a decrease of the effectiveness factor. Thus, after light-off only active sites on the external surface are used for conversion, because the catalyst bed is heated up due to exothermal reactions. Effectiveness factor is high in low temperature
60
applications. Heat transfer between catalyst surface and the offgas flow preheat the catalyst surface to the ignition temperature of the gas purification reaction and removes the heat of reaction from the surface. During reactor design stationary and non stationary transfer phenomena have to be considered to ensure adequate contact time of pollutants with the catalyst and to prevent the catalyst overheating. [6,7, 81 Figure I :Mass Transfer in Heterogeneous Catalysis
1. Transport of Reactants from Bulk of Gas Flow to Boundary Layers 2. Diffusion of Reactants through Boundary Layer to External Catalyst Surface
3. Diffusion of Reactants through Pores to Internal Catalyst Surface 4. Adsorption of Reactants
at Active Sites
5. Chemical Reaction
6. Desorption of Reaction Products
11
7. Diffusion of Products through Pores
8. Diffusion of Products through Boundary Layer 9. Transport of Products into Bulk of Gas Flow Generally chemical engineering principles (mass and heat balance, plug-flow tubular reactor) can be applied to catalytic reactor design, if rate equations are known. This requires detailed investigation to determine activation energies and frequency factors for each substance. In practice, catalyst sizing is carried out empirically according to pilot plant information. These yield conversion versus either temperature or gas hourly space velocity (sv) curves. From these curves catalyst volume (v) and operating temperatures are chosen for the required conversion at a given flow rate (4)v). S v '4 )yV
61
1.3.2 Catalyst selection Catalysts for air pollution control are delivered as bulk material (beads, pellets) or honeycomb carriers. Carrier materials mainly consist of porous ceramic material (y-Al2O3 alumosilicate, silica). Dense materials such as cordierite or mullite or steel are covered with active layers (wash-coats) before impregnation with active materials. Active materials may be selected from the noble metals (e.g. platinum, palladium) or metal oxides (e.g.oxides of manganese, copper, cerium, titanium, vanadium, chromium). Metal oxide catalysts often are produced as mixed catalysts (prepared by precipitation methods). Noble metal catalysts are always supported catalysts (prepared by impregnation methods). [ 161 Catalyst for offgas treatment have to conform to the following technical and commercial selection criteria [9]: -high activity: high reaction rate of desired reactions at low temperatures and low catalyst volumes, -high selectivity: formation of noxious by-products must be avoided (e.g.: CO, N20, SO3) -high stability against chemical, thermal and mechanical impacts -low pressure drop to minimize electrical energy consumption for blowers 1 3.3 Catalyst deactivation
The economics of catalytic offgas purification require long service times (e.g. beyond two years). Catalyst lifetime can be reduced due to catalyst deactivation. The following factors can cause catalyst deactivation: [4]: - Chemical factors (so-called poisoning) Substances which are not desorbed or react irreversibly with catalyst carrier (e.g. acids) or active material (e.g. sulphur, arsenic, heavy metals) lead to changes in chemical composition, loss of active sites, reduction of accessible surface or covering of active sites (so-called fouling). - Thermal factors (so-called aging) Rates of change in structure of solid catalysts (e.g. crystallite growth, spinells, sintering) increase with increase of temperature. Structural changes can lead to decrease of surface area or reduction of dispersion. For each catalyst maximum operating temperatures must be used. - Mechanical factors: (i) attrition (e.g. by vibration, abrasive dusts), (ii) crushing (e.g. external stresses of reactor shell) Risks of catalyst deactivation must be considered during process and plant design. Minimization of risks start with process and catalyst selection. Pilot plants or trial tests can be camed out to check offgas composition for catalyst poisons or peak concentration, which could lead to catalyst
62
overheating. If catalyst deactivation occurs due to unforeseen reasons, analysis of phenomena can give hints, whether or not chances for reactivation exist. Reversible mechanisms like: - coking, - covering with aerosol or salt, - condensation of organics, - strongly adsorbed species (e.g. HCl, P2O5) may be removable by heating or washing. Otherwise (irreversible deactivation, e.g. sintering) the catalyst charge has to be replaced completely or partially or fresh catalyst has to be added.
1.4 Plant design considerations The equipment and instrumentation of the projected plant is determined by the following factors: [13] a) crude gas composition (average, peak) b) crude gas temperature c) crude gas pressure d) necessary conversion efficiencies (admissable residual emission) e) type of catalyst f ) possibilities of energy supply g) possibilities of heat recovery. With regard to the technical realization of catalytic exhaust gas purification processes it is necessary to adapt plant concepts to specific exhaust gas problems. A basic flow sheet for catalytic exhaust gas purification plants is represented in fig. 2. Figure 2 :Unit operations in catalytic off-gas treatment
Q I
I
1 - TAKE-OVER OF THE OFFGAS
2 - MECHANICAL CONVEYING 3 - SEPARATION OF DISTURBING COMPONENTS 4 - PREHEATING BY HEAT EXCHANGE 5 - EXTERNAL HEAT-SUPPLY 6 - MIXTURE W OXIDIZING OR REDUCING SPECIES 7 - CATALYTIC REACI'ION
8 - HEAT-RECOVERY 9 - DISCHARGE OF THE PURIFIED GAS
63
The following processing stages can be considered as unit operations of catalytic exhaust gas purification plants: 1.- Take-over of the offgas (e.g. by suction) 2.- Mechanical conveying of the offgas e.g. by ventilators, compressors) 3.- Separation of disturbing components for prevention of catalyst deactivation (e.g. filters, adsorbers, absorbers) 4.- Heating of the offgas by heat exchange (e.g. recuperative tubular heat exchangers or regenerative heat exchangers) [17 I] 5.- External energy supplies (e.g. burners, preferably using natural gas as additional fuel, electrical resistance heating) 6.- Mixture with oxidizing or reducing agents (e.g. oxygen or ammonia) 7.- Catalytic reactors. Commonly fixed bed reactors, honeycomb or bulk material type catalysts. Flow direction can be chosen arbitrarily if honeycombs are employed. Bulk material catalysts require flow direction top to bottom for prevention of bed movement (attrition) or channelling. Very rarely fluidized bed reactors are reported. [14] 8.- Heat recovery (e.g. steam or thermal oil heating) 9.- Discharge of the purified gases (e.g. stacks) Fig. 3 shows the simplified thermal energy balance for an afterburning plant. In the case where the energy demand of a process is designed to involve heat recovery by pollutant oxidation (table 3), the catalyst ignition temperature is reached by heat exchange, without the consumption of external fuel. Table 3 - Stoichiometry and heat of reaction for some complete oxidation reactions
Compound
I
Stoi chiometry
C7Hg + 902+ 7C02 + 4H20 CH30H + 3/2 0 2 + C02 + 2H20 C3H60 + 402+ 3C02 + 3H20 co co + 1/202+ c02 NH3 NH3 + 3/4 02- N + 3/2 H20 H2S H2S + 3/2 0 2 + SO2 + H20 Ethylacetate C4Hg02 + 5 0 2 + 4C02 + 4H20 Toluene Methanol Acetone
Heat of Reaction
IT pro 1g/m:
40.935 kJkg 21.119 H/kg 29.065 kJ/kg 10.423 H/kg 18.424 H/kg 15.355 kJ/kg 23.530 H/kg
30,6 K 15,8 K 21,9 K 7,5 K 13,9 K 11 K 18 K
64
Figure 3 :Thermal Energy Balance
T'WTa
QU
Catalytic Oxidation
P,CJ
External Energy Supply
W A
Thermal Energy Balance, Stationary
2. Convective Heat Transport
0 = OE,K+OE,F+Q&,S+OE,WR-~E,V
OE,K= @v,L. P O L . CL(T'WT,~ - T'wT,~:
3. External Energy Supply
@E,F = W,EG .A~U,EG
4.Combustion
QE,S =
5. Heat Losses (negligible due to thermal insulation)
@E,V
A ~ u , s. PS . @V,L
-)
0
Such a process is self supporting. These conditions (pollutant concentration and rates of heat transfer in the crude gadpurified gas heat exchanger) allow significant reduction in operating costs. Use of separation units (e.g. filter, adsorption, absorption, thermal afterburning) can reduce the risk of deactivation by poisons, dust or highpeak concentrations. Temperature measurement in front, behind and inside the catalyst bed and control (e.g. by additional fuel supply, heat exchange rate, fresh air dilution) reduces the risk of overheating. Economics of plants treating low concentration offgas can be improved by the incorporation of adsorptive concentration units. [19, 201
65
2 EXAMPLES OF APPLICATIONS
2.1
- Selective
catalytic nitric oxide reduction [21] (SCR, DeNOx, "denitrification")
Field of jxmlication Removal of nitric oxides from flue gases or process exhaust gases, e.g. boilers, nitric acid plants, glass melts, municipal waste incineration. PrinciDal reactions After admixture of ammonia, nitric oxides are converted selectively (also in the presence of oxygen) into molecular nitrogen. 4 NH3 + 4 NO + 0 2 4 6 H20 + 4 N2 Process description - SCR for Dower station flue gass Two process variants must be distinguished: (a) High-Dust process (Fig. 4) The flue gases are directly withdrawn from the steam generator (behind the economizer and before the air preheater) and are passed over the catalyst after the admixture of the reducing agents.
Bolier
sCR
D e Dusting
De-Sulphuritation
Fig. 4 :SCR process - crude gas operation
66
The largely nitric oxide free flue gas is fed to the air preheater for the purpose of preheating the combustion air (heat recovery). Depending on the fuels used, additional exhaust gas purification stages are installed downstream (dedusting, desulphurization). Crude gas preheating or fuel supply before the catalyst is not necessary, since the crude gases (= flue gases) are withdrawn at a place in the steam generator where the latter possess a sufficient temperature (350-450"C). Honeycomb mixed contacts are used as catalvsts. Material: Ti, V, Fe, W, Ni, Co, Cu, Cr, U, Mo, Sn as oxides Ag, Be, Mg, Zn, B, Al, etc. as metals [lo] Selectivity: SO2 + SO3 - conversion must be avoided Activity: as low NH3, consumption as possible Stability: high demands because of the abrasion resulting from the dust load of the flue gas oDeration (Fig. 5) dust and sulphur oxides are (b) In case of pure removed by electrostatic filters or flue gas scrubbers, before the crude gas is mixed with reducing agents and reaches the catalyst. The adjustment of the necessary reaction temperature (300-400"C) requires preheating of the crude gas and external energy supplies. In principle, similar, but more active honeycomb catalysts are used. Because of the preseparation of the disturbing components, lower demands are made on selectivity. Furthermore, a higher life time is expected.
Fue 1
Boiler
Dedusting
De-Sulphurization
Fig. 5 :SCR process - purified gas operation
67
2.2
Offgas treatment of stationary internal combustion engines or turbines
2.2.1 Exhaust gas emission control for spark ignition engines according to the firee-wav system 122.231 Exhaust gases from combustion engines contain the following classes of p011u tants - hydrocarbons (CnHm) - carbon monoxide (CO) - nitrogen oxides (NO4 By means of special catalysts and an appropriate carburetion, the three pollutant classes can be jointly removed for spark-ignition, four-stroke engines through one catalytic converter (three way system). catalytic converter
b
lambda-probe control-unit control
IL----------------Jl
temperature
I1
t-I
Throttle
I
t-I
IL-
OXIDATION
CmHn + (m+n/4)02 +mC02 +n/2H20 c0+1/202 + c02 H2 + 1/2 0 2 + H2O
Speed
REDUCI'ION
CmHn + (m+n/4) NO + (m+n/4) + C02 +n/2H20 CO + NO+ 1/2 N2 + C02 H2 + NO + 1/2 N2 + H20
Fig. 6 :Three-way catalyst system
68
As shown in Fig. 6 [ l 11, the catalyst is integrated into the exhaust system of the engine. Preheating or external energy supply is thus not necessary because of the given exhaust gas temperature. The necessary concentrations of oxidation agents (02, NO) and reducing agents (CnHm, CO, H2) are adjusted by the electronic oxygen sensor control (lambda probe) through a reactive effect on the carburation of the engine. The purified exhaust gases are discharged via the remaining exhaust system. In the case of gas engine driven block heating stations, the thermal energy contained in the pure gases is recovered. Precious metal impregnated honeycomb catalysts are used as catalysts. The precious metals used in this connection are mainly platinum and rhodium. The coatings and the supports are optimized with regard to a high temperature stability. 2.2.2 Diesel ennines Hydrocarbon and CO emission can be reduced by catalytic oxidation with noble metal supported catalysts (bulkmaterial or honeycomb). Catalysts must be optimized with respect to selectivity to avoid NO/N02 and Sods03 conversion.(Fig. 7)
catalytic afterburning
J i l l
Fig. 7 :Diesel engine ofSgas treatment
69
Particles can be removed by filtering. Some filters are catalytically impregnated to reduce temperature of coke/soot burning during regeneration. Due to high oxygen concentrations NOx emission control is not possible with TWC system (see 2.2.1). Feasible solutions are using offgas recycling or SCR system (see 2.1). 2.2.3 Gas emines (two stroke). gas turbines Offgases sometimes contain concentrations of hydrocarbons (alkanes, aldehydes) or carbon monoxide which exceeds the threshold limits. Noble metal or honeycomb catalysts are employed for oxidation of these pollutants. Catalysts contain high noble metal concentrations if temperatures are low (300 "C) and concentrations of ethane are high. Future tasks are to be seen in the removal of methane emission and other gases which until recently were not regarded as pollutants (non methanehydrocarbons, NMHC).
2.3 CLAUS plant offgases Typical Claus plant exhaust gases contain 1 % of H2S, COS, CS2 and 0.5 % S02. 2.3.1 SULFREEN process 1241 The flow sheet and the basic reactions of the SULFREEN process are presented in fig. 8. The reactions of the Claus process occur at lower temperatures (120140 "C). Modified pelleted alumina catalysts are used. The elementary sulphur arising as a product at the reaction of H2S with SO2 adsorbs on the inner surface of the aluminium oxides (pores). As a result, the catalyst is reversibly contaminated. Regeneration of the catalyst and recovery of sulphur is achieved by cyclical purging with hot inert gas. 2.3.2 Catalytic afterburning [15] Before released to the atmosphere residual sulphur compounds (H2S; COS, CS2) have to be converted to S02. Selectivity of the catalyst must avoid SO3 formation which could cause severe corrosion problems in afterbuming plants. Application of modified y-alumina catalysts allow lower energy consumption at high flow rates compared to thermal afterbuming. Process Darameters
temperature: space velocity:
350 "C 1500-2000 h-1
Table 4 Complete oxidation of volatile organic compounds catalytic afterburning : fields of application
INDUSTRIES
POLLUTANTS
EXAMPLES
FOODSTUFF
ODOURS
SMOKING; ROASTING ETHYLENEOXIDE; OZONE WIRE -ENAMELLING ARTIFICIAL LEATHER POLYESTER PHENOLIC-RESIN; PVC PROCESSING, LAMINATING PRINTING
STERILIZERS
MEDICAL
SOLVENTS
SURFACE COATING
SOLVENTS
PLASTIC PROCESSING
MONOMERS
GRAPHICAL PROCESSES CHEMICAL AND REFINERIES
SOLVENTS
SYNTHESIS
MONOMERS,BY-PRODUCTS, RESIDUAL REACTANTS
SOLVENTS
VAPOURS
DEPOTS SEWAGE WATER TREATMENT
CARBON-MONOXIDE
I OXYGENATED-HYDROCARBONS I
Parameters : Flowrate : 102-105m3/h ;
Temp.: 200 "C- 600 "C ;
FORMALDEHYDE; MALEIC ACID; PHTALIC ACID; VINYLCHLORIDE; FIBRE-PRODUCTION BITUMENE; FUEL, BENZENE; SOLVENTS OFFGASES OF W E T OXIDATION
Pollutant C o n t e n t :102-104ppm
71
CLAUS PROCESS : CONVERSION OF H2S INTO S
3 H2S + 3/202 + 2 H2S + SO2 + H 2 0 2 H2S + SO2 + 3/2 S2 + 2 H 2 0 3 H2S + 3/202
+ 3/2 S2 + 3 H 2 0 Afterburning
I'
c1
Fig. 8 :SULFREEN process 2.4
Catalytic afterburning of volatile organic compounds for industrial production processes
These processes are applied to reduce emission of volatile organic compounds by means of complete catalytic oxidation. They are tailor made to a variety of production processes. 2.4.1 Survev A survey on fields of application for complete oxidation of volatile organic compounds is shown in table 4.
72
2.4.2 Laminating of Daper [lo] During laminating of printed paper with polypropylene foils, solvents containing offgas are emitted. Due to flexibility of the production, concentrations and flow rates differ temporarily. Typical solvent mixtures contain ethanol, ethylacetate, methylethylketone, toluene. Metal oxide catalysts yield sufficient activity at ignition temperatures of about 250 "C. A catalyst volume of 600 1 is implemented to purify an offgas flow of 8.000 m3h and 8 g/m3 at maximum conditions. Table 5 shows purified gas composition and some process parameter for a service time of more than 8.000 h. Activity of the catalyst keeps concentrations of hydrocarbons within threshold limits (0,l g/m3). Due to selfsupporting plant operation, C02 emissions are equivalent to the carbon content in offgas. CO and NOx emissions are quite low. Gas chromatographic investigations (table 6) show that during catalytic complete oxidation the number of hydrocarbon compounds is reduced. Residual hydrocarbon contents in purified gas consist of unconverted parts of the feed components. Thus it can be concluded that no other hydrocarbon compounds are formed during contact with catalyst. Table 5 Purified gas composition after certain catalyst service periods Flow rate : 8000 m3 h-1 Concentration : 2- 8 g m-3 Solvents : Ethylacetate; MEK, Toluene; I-Propylacetate Catalyst : Mixed metal-oxides (KCO SG 21 16) Operating Hours
Operating Mode
Ignition Temp "C
Purified Gas Composition TOC CO NOx mg/m3 mgjm3 mg/m3
1275 3000 8000
SelfSupporting SelfSupporting Selfsupporting
238
2
250
10
240
18
30
1
C02 %
0.5
Implementation of plant ensures flexible adaptation of preheating (burner. aidair heat exchanger) and prevents catalyst overheating (450 "C). Catalyst temperature control is accomplished by thermocouples in the catalyst bed.
73
Dilution of offgas with fresh air allows operating, even if design concentrations are exceeded. Heat recovery (hot water, hot air) improves the economy of the offgas treatment plant (fig. 9). Table 6
-
Gas chromatographic investigation of reaction products of complete oxidation of organic solvents Solvents : Ethanol, Ethyl acetate, MEK, Toluene Catalyst : Mixed Metal Oxides KCO SG 2216 Ignition Temperature : 220 "C GHSV: 70o0h-1 Toc
OF DETECTED
'
~
OFFGAS
I
OTHERS
ETHANOL
ACETATE
8
4090 mg/m3
585 mg/m3
I
466 ppm C3
3
2,55 mg/m3
100%
I
99,9%
PEAKS
~
99,9%
mn-
Fig. 9 :Catalytic dterburning plant for laminating processes
74
3. LITERATURE 1 2 3 4 5
6 7 8 9 10 11 13 14 15 16 17 18 19 20 21 22 23 24
Janee, M,: Auswirkungen von gesetzlichen Vorschriften und technischen Regelwerken auf die Abgasreinigung. in: VDI-Bericht Nr. 730, VDI-Verlag, Dusseldorf, 1989. Zlokamik. Umweltschutz - eine standige Herausforderung. Chem.-Ing. Technik Nr. 5 (1989) S. 378-385. Schlosser. E. G;Heterogene Katalyse. Verlag Chemie Weinheim (1972). &L B;Reaction Kinetics on Reactor Design. Prentice Hall Inc. Englewood Cliffs, New Jersey (1980). firchner, K, Kaiczik, A.: Grundlagen der katalytischen Nachverbrennung. VDI-Berichte 525 S. 119-145, VDI-Verlag Dusseldorf (1985). Eigenberger. Q Stabilitat und Dynamik heterogenkatalytischer Reaktionssysteme. Chem.-Ing. Tech. 50 (1978) S. 924-933. Wicke. E.: Grundlagen der katalytischen Nachverbrennung. Chem.-Ing. Tech. 37 (1965) S. 892-904 Kanzler, W.; Schedler, J.; Thalhammer, H.: Theoretische und experimentelle Untersuchungen bei katalytischen Nachverbrennungsanlagen. Chem.-Ing. Tech. 59 (1987) S. 582-585 Szepe. S.;Levenspiel, k Proc.4th European Symp. on React. Eng., Brussel (1968) Pergamon Press, London (197 1) Schmidt. T.: Katalytische Verbrennung chlorierter Kohlenwasserstoffe. VDI-Berichte 730 S. 201-238, VDI-Verlag, Dusseldorf (1989) SuiveY, J. Complete Catalytic Oxidation of Volatile Organics. Ind. Eng. Chem. Res. 1987,26, S. 2165-218012 VDI-Richtlinie 3476 Sattler. K.; Umweltschutz, Entsorgungstechnik Vogel Verlag, Wiirzburg 1982 Hardison. L. C.: Dowd. E. J.: Emission Control via Fluidized Bed Oxidation CEP August, 1977, S. 31-35 Ketmer. R.: Lubke. T.: Betrieb von KNV-Anlagen nach Clausanlagen VDI-Bericht 730, S. 255-274, VDI-Verlag Diisseldorf (1989) Eneler. B.; Koberstein, E.:Katalysatoren in der Abgasreinigung. VDI-Bericht 730 S .97-120, VDI-Verlag Diisseldorf 1989 Taneer.Y; Energiesparendes KNV-Verfahren durch neuartige ProzePfuhrung mit regenerativen Warmeaustauschem. VDI-Bericht 730 S . 189-200, VDI-Verlag Diisseldorf 1989 . Martin. fL;Wirmeiibertrager. G. Thieme Verlag Stuttgart 1988 Brauer.fl& Adsorptive Reinigung schwachund mittelbelasteter Abluftstrome Vortrag auf dem 2. Fachsymposium Umweltschutz in Mainz, Mai 19 8 8 . Walter. B;Anwendungsbeispiele fur die katalytische Nachverbrennung. VDI-Berichte 525 (1985) S. 347-365 Richter.E;Die katalytische NOx-Reduktion - eine Ubersicht in: VDI-Bericht Nr. 730 S . 121-156, VDI-Verlag Diisseld. 1989 . Koberstein, Katalytische Motorabgasreinigung. in: VDI-Bericht 525 S. 217-246, VDI-Verlag Diisseldorf (1985). Publikation: Engelhard Kali-Chemie AutoCat GmbH, Hannover. Ruhl. E.: Katalytische Reinigung von Claus-Anlagen-Abgasen. VDI-Berichte 730 S. 255-274, VDI-Verlag (1989), Dusseldorf.
A. Crucq (Editor), Catalysis andAutomotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
75
CHARACTERIZATION OF EXHAUST EMISSIONS FROM TWO HEAVY DUTY VEHICLES FUELED WITH EIGHT DIFFERENT DIESEL FUELS Egeback K.E. (1) Mason G. (21, Rannug U. (3) and Westerholm R. (4) (1)-Environmental Section, The Swedish Motor Vehicle Inspection Company, Box 508, S-162 I5 Vallingby, SWEDEN (2)-Department of Medical Nutrition, Karolinska Institute, Huddinge University Hospital F60, Novum, S-141 86 Huddinge, SWEDEN (3)-Department of Genetics, Stockholm University, S-I06 91 Stockholm, SWEDEN (4)-Department of Analytical Chemistry, Arrhenius Laboratory, Stockholm University, S-106 91 Stockholm, SWEDEN
Abstract The impact on exhaust emissions of diesel fuel composition was studied by a comprehensive program compiled after discussions with the oil industry, the motor vehicle manufacturers and the Swedish Environmental Protection Agency. In order to reduce the emissions from diesel fueled vehicles different possibilities have to be taken into consideration such as improvement of the engine, use of exhaust purification devices and use of diesel fuels of improved quality. The aim of this investigation was to study the connection between certain fuel parameters and exhaust emissions.Eight different diesel fuels were used in the investigation and the emission tests were carried out using a city bus and a truck, driven on a chassis dynamometer. Both regulated andunregulated emissions were measured. In addition to the emission measurements samples were taken and prepared for biological testing.The investigation showed that there is a clear correlation between certain fuel parameters and the emissions and biological activity of the exhaust. Multivariate analysis was undertaken on groups of data derived from the eight blends of fuels in order to study the correlation between the fuel and emission parameters. A least square regression analysis of the correlation between 23 fuel parameters and 43 exhaust emission parameters resulted in correlation coefficients, which were higher than 0.95 for the two vehicles tested. It was confirmed that the most important fuel parameters to be controlled are the density, boiling point range, the content of aromatics and cetane index. In addition to these parameters the content of polycyclic aromatic hydrocarbons should be controlled and limited. Sulfur is also an important parameter to be controlled especially when the fuel is going to be used for vehicles equipped with catalysts or catalytically activated particulate filters.
76
INTRODUCTION
In many countries where actions have been taken to reduce the emissions from light duty vehicles there is now a growing concern regarding the emission from heavy duty vehicles. To reduce the emissions from such vehicles operating in city areas is the most urgent action to be taken. As work was going on in Sweden to prepare a definition of low emitting heavy duty vehicles operating in city areas there was a need to look into the area of diesel fuels and their impact on exhaust emissions. The Swedish Environmental Protection Agency which was authorized to prepare proposals for the Swedish Government concerning vehicle emissions and automotive fuels asked the Automotive Emissions Research Laboratory, organized within the Swedish Motor Vehicle Inspection Company, to define a program and to carry out an investigation in order to chemically and with bioassy tests characterize the emissions produced when changing some of the diesel fuel properties. The program was prepared in collaboration with the Swedish Urban Air Project and especially with the Emission Group Branch, who have also been carrying out analysis of unregulated emissions in addition to carrying out biological testing. The results in this article should be seen as an extended abstract of a report covering the whole investigation which will be published during the autumn 1990 (Westerholm and Egeback, 1990). EXPERIMENTALS
The objectives of the investigation were fourfold: - How do the aromatics in the fuel influence the emissions? - Does the sulfur content of the fuel influence the emissions? - Does the use of light diesel fuel influence the emissions? - How does the ignition improver used in diesel fuel influence the emissions ? In order to answer these questions the program was composed of a literature study, experimentals and analysis of the data generated by the experimentals.
Fuels Eight fuels were included in the investigation, see Table 1. The fuels were selected after discussions with the Swedish Environmental Protection Agency, different oil companies, diesel engine manufacturers and other experts. In order to answer the objectives of the investigation it was quite clear that at least for some of the fuels only the content of aromatics and sulfur should be changed so as to study their effect on the exhaust fuels should show the impact on the emission of sulfur and three of the fuels to be used should determine the impact of aromatics. Two light diesel fuels (kerosene) were to be
77
used as well as one with approximately the same composition as the light diesel fuels but somewhat heavier. One additional fuel would be regarded as the worst case. When the fuels were manufactured it appeared that the outcome was not as expected especially for the fuels which were meant to be used to determine the impact of aromatics. When changing the aromatic and sulfur content other properties in the fuel were also changed. However, the unintended changes were not too large and, therefore, there appeared to be a good possibility to study the effect on the exhaust emissions of fuel aromatics and sulfur, respectively. Of the eight fuels D1, D2 and D4 are from the same base fuel, table 1. The only main difference between these three fuels are that D2 has a higher content of aromatics compared to D1 and that D4 has a higher content of both sulfur and aromatics than D1. There is also a difference of fuel density. D1 has the lowest and D4 the highest density of the three fuels. The fuel D5 is a blend of different kerosenes. Its density is nearly as high as for a common commercial diesel fuel in Sweden. The contents of sulfur and aromatics are on the same level as for a low sulfur light diesel fuel in Sweden. The fuel D6 is a common commercial (summer) fuel and was regarded to be a reference for this type of commercial fuel in Sweden. The fuels D7 and D8 are the same fuels except that 2000 ppm ignition improver, ethyl hexyl nitrate (EHN) was added to D8. The fuel D9 is a blend of cracked gasoils with a low content of sulfur. Table1.-Fuel characteristics D6
D7
D8
D9
52.8 50.0 81 1.7 821.3
47.2 47.0 832.0 831.3
48.3 836.8
44.7 808.3
55.7 808.7
52.8 813.2
231 252 260 43.10
220 233 289 252 323 261 42.88 42.98
205 329 364 42.87
190 245 300 43.24
187 243 289 43.24
205 282 301 43.19
26.1 20.2 4.8 1.1 1.0 110 0.16
20.0 17.2 2.7
20.5 17.2 2.7
0.1
0.6
0.9 14.3 0.02 0.0
0.2 207 0.01 0.2
17.3 14.5 2.2 0.6 0.7 11.0 CO.01
D1 Cetan Nr Density (1) Distillation (2) 10%
90%
FBP Energy (3) Aromatics (4) -Total -Mono -Di -Tri
Olefms (4) Nitrogen (5) Sulfur (6) EHN(7)
238 251 26 1 43.20 1.8 1.8
D2
16.6 16.2 ~ 0 . 0 5 0.4
D4
23.0 18.1 4.9
D5
25.1 21.1 3.8
~ 0 . 0 5 ~ 0 . 0 5 ~ 0 . 0 5 0.2
1.4 2.0 2.2 0.2 12 0.34 3.9
1.6 29.2 0.02
(1) g/L, 15"C, (2) OC, (3) MJ/kg, (4) volume-%, ( 5 )
(6)weight-%,(7)ethyl hexyl nitrate, weight %.
78
Of the fuels tested only fuels D5, D6, D7 and D8 are commercially available on the Swedish market. The two fuels D7 and D8 (light diesel fuels or jet fuels) are commonly used for city buses and are regarded to be better from an environmental point of view than the heavier commercial fuels (such as D6). The fuel D5 has newly been introduced on the Swedish market to be used for city buses.Al1 fuels were analyzed for the same properties at the same facility, a well equipped laboratory with experienced staff. In addition to the presented fuel parameters, table 1, the PAH content in each fuel was quantified. The analysis and quantification of PAH are described in detail elsewhere (Westerholm and Egeback, 1990).
Vehicles Two vehicles: a bus and a truck were used during the generation of data. With kind support from Swedish vehicle manufacturers, the bus was loaned out by the Saab Scania Company and the truck by the Volvo Truck Company. Vehicle data are presented in table 2. Table 2.-Vehicle data. Vehicle Scania 113 Volvo FLlO Engine type DSC 1104i TD 101 F Displacement volume (L) 11 9.6 Maximum power (kW) 191 (1800 rpm) 229 (2050 rpm) Service weight (kg) 10480 8570 Maximum weight (kg) 15800 19000 In this report the Scania vehicle is denoted vehicle 1 and the Volvo vehicle is denoted vehicle 2, respectively. It must be pointed out that the vehicles should not be compared to each other. The aim of this investigation was to study fuel related emission and was not aimed as an evaluation of the vehicles. The vehicles were serviced by Saab Scania and Volvo, respectively. When the vehicles arrived at the laboratory the lubrication oil was changed to a special lubrication oil: Mobil Delvac 1. Before each test the vehicle was conditioned so as to warm up the engine. All tests were carried out with the engine hot. No adjustment of the engine or the engine fuel system was performed during the course of the testing.
Test programme The equipment used for sampling the exhaust gas is shown in figure 1. It consists of a chassis dynamometer (Schenk, FRG), a CVS-system (Constant Volume Sampler) and a dilution tunnel. The dilution tunnel is equipped with a critical flow venturi with a flow rate of 2.15 m3/s. The system was designed to
79
fulfil the specifications in the U.S. Federal Register (Federal Register, 1987). The Schenk chassis dynamometer for HDV is equipped with a set of inertias to simulate vehicle masses up to 20.000 kg. ABdMNSK BllPROVNlNG
,. ,,.,....-c ,",".,, m.
Sl**L
I
VEHICLE ON THE TEST STAND
Figure I .-Test equipment The instrumentation for analysis of
HC, CO and NO is a system
ABIZISMNSK BllPROVNlNC
Driving distance
c a 11.WOrn
Diiing time
1.740 s
Maximal speed
58.2 b / h
Average speed
22.5
Percent Idling:
22%
BUS CYCLE
Figure 2-Bus Cycle
.
manufactured by Beckman (Beckman Inc., USA). A heated flame ionization detector (FID) was used for hydrocarbon (HC) analysis, a nondispersive infrared analyzer (NDIR) was used for carbon monoxide and carbon dioxide analyses and a chemiluminescence detector for analysis of oxides of nitrogen. Samples for measurement of HC, CO and NO were taken in diluted exhaust. The procedures for the preparation of the filters and the sampling of the particles are as follows.:The particle filters, Pallflex T60A20, are conditioned in a climate chamber for at least 2 hours at 20°C and 50 % humidity before weighing.
80
After the filters have been used the same conditioning procedure was executed before reweighing. If the filters were to be usedfor analysis of polyaromatic compounds and for biological testing a special cleaning procedure was used (Westerholm and Egeback, 1990). Three different driving cycles were used: the bus cycle, the US transient cycle for heavy duty vehicles and the 13 mode test, ECE R49. Due to the fact that both regulated and unregulated pollutions were only measured during the bus cycle tests, these data alone will be presented in this summary. The bus cycle (Stochastischer Fahrzyklus fiir Stadtlinien Omnibusse) has been developed at the University of Braunschweig, FRG. It simulates the driving condition of a bus in city traffic. It has a duration of 29 minutes and the driving distance is about 11 km. The top speed is 58.2 km/h and the average speed is 22.9 km/h. a speed versus time scale of the bus cycle is shown in figure 2. In table 3 the total sampling during bus cycle driving has been summarized. Up to four repeated samples were analysed for each vehicle/fuel test case. Special analyses of the particles were carried out in order to study the different fractions such as the soluble organic fraction, carbon, sulfates, nitrate and phosphates, however not presented in the present article. Table 3-.Emission cornDonants measured
Component measured
Procedure for sampling or measurement
HC/CO/NO/C02 Dilution tunnel Particles j Filter Soluble organic fraction Filter/Gravimetric of particles Aldehydes Cartridge Oxygenates (*) Bag Light aromatics (*) Bag PAC FilterPUF Mutagenicity FilterPUF TCDD Receptor Affinity FilterPUF Fuel consumption Gravimetric (*)The results of these components will be presented and discussed in a final report (Westerholm and Egeback, 1990.)
81 RESULTS AND CONCLUSIONS
This section is a short presentation of the results obtained during the different measurements and it summarizes the conclusions drawn when analyzing the data. The complete data set was subjected to multivariate analysis which will be presented in the final report (Westerholm and Egeback, 1990).
Regulated emissions and aldehydes The results from the measurements of hydrocarbons (HC), oxides of nitrogen (NOx), carbon monoxide (CO), particulate emissions and aldehydes are shown in figure 3 for vehicle 1 and in figure 4 for vehicle 2. The main conclusions drawn from the analysis of the data are:
HC General The emission level is low for all fuels. Vehicle 1 : Among the three fuels of the same basic blend (Dl, D2 and D4) the fuels with lower density emit less HC. The ignition improver (D8) seems to cause a higher emission of HC Vehicle 2: No clear conclusion can be drawn. The engine in this vehicle seems to react to the differences in the fuel properties in a different way than the engine in vehicle 1.
NOx General: For both vehicles the variation of NOx emissions is very small when comparing the different fuel. The emission level is somewhat lower for fuels Dl and D8 than for other fuels. When comparing the three fuels of the same basic blend (Dl, D2 and D4) the lower density fuels seem to emit less NOx As NOx is a pollution of great concern, studies of the influence of certain fuel parameters on NOx emissions should be expanded
co General The emission level is low for all fuels. There is a certain variation of the CO emission when comparing the different fuels. No firm conclusions can be drawn by just examining the results in the figures. In this context CO is not considered as a pollutant of great concern.
82
...
1 "
L.1
1
.L
*..
,.a
.
1.1
s.0
0
L.0 0..
2.* 1.0
*.a
0.. 0..
s.0
0.i
e.*
0.0
0.0
Pmmculns cmiirion BF....,...,..
>......
?me.,.
Figure 3. Emissions of HC, NOx, CO, particles and aldehydes, vehicle 1 , mean values
Particles General: There is a clear relationship between the fuel properties and the particulate emissions. Heavier fuels emit more particles than lighter fuels in most cases. A vehicle dependency can also be seen. The relationship between the content of sulfur and aromatics respectively, and the emissions based on multivariate analysis are discussed below. Vehicle I: In line with the general conclusions there is a clear relationship between the fuel density and the emissions. Vehicle 2: Apart from the general conclusions, fuel D9 emitted more particles than D7 despite these fuels being of similar density. When comparing D9 to D7 no significant differences in their properties can be seen except for their cetane numbers, 52.8 for D9 and 44.7 for D7.Considering the influence of the cetane number on the combustion process one possible conclusion to be
83
drawn (and confirmed by the engine manufacturers) could be that the engine in vehicle 2 reacts to differences in cetane number in almost the opposite way with regard to particulate emissions than the engine in vehicle 1.
.".
S"CI.
0".
="ox.
0".
CYSI.
Figure 4 . Emissions of HC, NOx, CO, particles and aldehydes, vehicle 2 mean values
Aldehydes General: The emission of aldehydes is not only sensitive to the fuel composition but also to the type of engine. There is a large variation from fuel to fuel and a great difference in the emission pattern when comparing vehicle 1 to vehicle 2. Of all eight fuels only fuel D7 shows the same pattern for both vehicles. The high rate of aldehyde emission for D7 may be caused by the less complete combustion as a result of its low cetane number, 44.7. To fully understand the nature of the emission of aldehydes a more thorough investigation should be undertaken.
84
Vehicle I : There is more than a three fold difference between the lowest value of the sum of formaldehyde plus acetaldehyde (D6 and D7) and the highest value of the sum for D7. The ratio between the emission of formaldehyde and acetaldehyde is not the same for all fuels. The acetaldehyde part of the aldehydes is much smaller (16 %) for D7 than for all the other fuels. For these, D4, D6 and D7 it is still small (22 %) while the acetaldehyde part is 48% for D9. Vehicle 2: There is almost a three fold difference between the lowest value of the sum of formaldehyde plus acetaldehyde @2) and the highest value (D7). By examining the results presented in figure 5 it can be seen that the sum of aldehyde emissions is lower for fuels with a short distillation curve like Dl and D2. In addition to the general conclusion drawn about fuel D7, the tendency is that the lighter fuels emit less formaldehyde than the heavier. The acetaldehyde part of the sum of aldehydes ranges from 23 % to 64 %. The fuels which gave the highest proportion of acetaldehyde are D8 and D9, i.e. fuels with a high cetane number and a low initial boiling point. This may be an indication of the risk that the front end hydrocarbons contribute to the formation of aldehydes. PAH levels in fuels The polycyclic aromatic hydrocarbon (PAH) levels in the eight diesel fuels tested are presented in figure 5. The 28 PAH's analyzed ranged from three ringed (phenethrene) to six ringed (coronene). All the individual PAH analyzed in the fuels were added together and termed sum of PAH when presented in the figures. All figures represented by blocks are mean values (N=3), the standard deviation being represented by the overlying block on top of the mean value. The factor on the top of block is the multification factor by which the bar has been multiplied. From the figure it can be seen that fuels D1, D2 and D4 contain less than 5 mg PAHL, fuels D5, D7, D8 and D9 contain in the range of 150 to 350 mg PAHL and fuel D6 contains approximately 1000 mg PAH/L. Major contributors to fuel PAH contents are methylphenanthrenes and methyl anthracenes. PAH emissions
All PAH measured in the exhaust emission samples are the same as quantified in the fuel samples. All the individual PAH analyzed in the fuel have been added together and termed sum of PAH when presented in the figures. Particulate associated PAH was sampled on Pallflex T60A20 filters and the semivolatile phase PAH was sampled down stream of the filter on an absorbent material, poly-urethane foam (PUF). The particulate and the semivolatile phase associated PAH were chemically analyzed separately. The sampling and the
85
chemical analysis procedures are described in detail elsewhere (Westerholm and Egeback,l990). FUEL
m g / l
D1
D4
D2
m
M
V
D5
D6
D7
DB
D9
LLJSTD
Figure 5. Polycyclic aromatic hydrocarbon contents in fuels, mean value (N=3) and standard deviation, mg PAHIL. General: From the results of the emission measurements on both vehicles it can be concluded that the emission of both particulate , PAH and semivolatile phase associated to PAH are fuel dependent. In the most favourable case the lowest emission (considering both particulate and semivolatile associated) is approximately 35 pg/km and in the worst case the sum of PAH is approximately 450 pg/km i.e. by the selection of a better fuel the PAH emission is lowered by a factor of approximately 10. Vehicle I Particulate associated PAH emissions are presented in figure 6. The emissions from fuels D1, D2, D4 are in the range of 25 to 45 pg/km, and from fuels D5, D7, D8 and D9 in the range of 75 to 120 pg/km. The corresponding value from fuel D6 is approximately 220 mg/km. Regarding the semivolatile phase sum of PAH emission, displayed in figure 6, it can be seen that for the fuels D2, D4 and D7 emission levels are less than 35 pg/km. Fuels D l , D5, D8 and D9 were emitting in the range of 75 to 150 ug/krn and fuel D6 approximately 230 p g k m . Considering both the particulate and the semivolatile associated PAH emissions: Fuels D1, D2, D4, D7 and D8 were emitting less than 150 pg/km, fuels D5 and D9 in the range 150 to 300 pg
86
P A H b and finally D6 more than 300 pLg/km. The semivolatile phase contribution to the total measured PAH emissions variates between 8% and 70% (Dl). SUM o f P A H u g / k m
V e h i c l e
140
1
300 280
-
260
-
220
-
200
-
180
-
160
-
140
-
240
i
I I I
II
120 100
I
80 60 40
20 0
T D1 D1
D2DZ
D4 D4 I
m
M
V
___
D5D5 f
D6D6
D7D7
'liter
11---PUF
m
0
S
T
D8 08
D9 D9
Figure 6.-Polycyclic aromatic hydrocarbon emissions; mean value and standard deviation, pglkm. Vehicle 2 : The particulate associated PAH emissions are presented in figure 7. The PAH emissions from fuels D1,D2, D4, D7 are lowest (approx. 50-80 pg/km), fuels D5 and D8 intermediary (90-120 pg/km) and largest for D6 and D9 which are in the range of 160 to170 pLg/km. Also the semivolatile phase associated PAH emissions are presented in figure 6. From the figure it can be seen that the lowest emissions are obtained from fuels D1, D2, D5, D7 and D8 which are in the range of 10 to 40 pg/km. Intermediary are fuels D4 and D9 in the range of 70 to 120 pg/km. The highest emissions was detected in the exhaust originating from fuel D6, approximately 200 pgkm. When comparing the total emission of PAH i.e. both particulate and semivolatile phase associated emissions it is seen that the lowest emissions originate from fuels D1, D2, D4, D5, D7 and D8, less than 150 pg/km, intermediate D9 less than 300 pg/km and finally D6 more than 300 p g k m . These results imply that fuel D6 is the most unfavourable fuel with respect to PAH emissions and that fuels D1 and D7 best. The per cent contribution from the semivolatile phase varies between 12% (D7) to 63% (D4) which underlines the importance of sampling and analysing the semivolatile phase PAH emissions.
87 SUM o f P A H
u g i / k m
V e h i c l e
230
240 260
N o 2
1
I
I
fl I1
220 200 1.30 160
140
120 100
80
60 40
20 0
D1 D1
DZD2
D4D4
D5D5
D6D6
I----Filter m
M
V
a
S
11--
T
D7D7
DBDB
D9D9
P U F
D
Figure 7-Polycyclic aromatic hydrocarbon emissions; mean value and standard deviation, pglkm. Mutagenicity tests
General The results from the Ames tests show a fuel dependent mutagenicity. Both Salmonella tester strains TA98 and TAlOO gave in most cases a similar fuel dependent mutagenicity profile. The mutagenicity level also showed a vehicle dependence. However, in this summary of the investigation only the results from tester strain TA100tS59 are shown in the figures 8 and 9. Using multivariate analysis a good correlation between PAH emission and mutagenicity in Salmonella was found, especially in the presence of a metabolizing system. Particulate phase For both tester strains the particulate extracts showed a similar mutagenicity with and without S9 or, as in most cases, a higher mutagenicity in the absence of S9. Based on the mutagenicity of the particulate extracts the fuels could be classified in low (Dl, D4, D7, DS), medium (D5, D9) and high mutagenicy (D6) fuels. This classification was more clear out with vehicle 2. Semivolatile phase In general the semivolatile phase gave lower mutagenic effects compared to the particulate phase. The fuel dependent mutagenicity profiles of the semivolatile phase followed the profiles of the particulate phase. Also in this case fuel D1 showed the lowest mutagenicity.
88
T A 100 + S 9
r a v e r t o n t s / r n
V e h i ~ l c N O
1
220 200 180 160 140 120 1 00
80 60 40
20 0
D1 D1
DZDZ
D4 D4
D5D5
D7D7
D6D6
DBDB
D9D9
--- PUF
I --- Filter I1 S T D
Figure 8.-Mutagenic effects,vehicle 1 ; mean value and standard deviation, revertantslm.
,
T A l[DD+SS
r e v e r t a n t a / r n
6oo
N o 2
V e h i c l e
I
I
I
I
7
D1 D1
DZDZ
D4D4 I
B
M
V
D5D5
D6D6
--- Filter
--
I1 S T D
T
D7 D7
D8 D8
D9 D9
PUF
Figure 9.-Mutagenic effects, vehicle 2 ; mean value and standard deviation, revertantslm.
89
TCDD-receptor affinity tests For each fuel sample extracts of both the particulate and the semivolatile phase collected using the PUF technique associated emission products were tested for TCDD receptor binding activity. The binding affinities were expressed as EC5o values (Toftgard et a1.,1955) and are shown in figures.10-11 Vehicle 1 Particulate phase The ECso values for binding to the dioxin receptor of the particulate phase exhaust emissions are given in figure 10. Dioxin receptor binding activity was observed in all the emission samples tested. The EC50 values for binding to the dioxin receptor of the samples varied over an approximately 8-fold arange. Fuel D 1 displayed the lowest binding affinity. Fuel D6 produced the emission sample with the highest affinity for the dioxin receptor with the lowest EC5o value. This EC50 was significantly lower than those for fuels D1, D2, D5 and D8 but not significantly different from those of the remaining fuels.
TCDD
rn
V e h i c l e
0 6
No.1
I II
>0.5
0 5
0 4
03
0 2
01
0
. .
. . . . .
7
P2 DZ
D1 D1
D4 D4 I
r
n
M
V
. . . . .
7
D5 D5
--- F i I t c ~
r S
D7D7
D6D6 I1 T
,
,
.
DB Da
D9 49
--- P U F D
Figure 10 - EC50 values, vehicle one; particulate phase and semivolatile phase, mean values and standard deviation.
90
Semivolatile phase The ECso values for binding to the dioxin receptor of the semivolatile phase exhaust emissions are shown in figure 10. Dioxin receptor binding activity was observed in all the samples except that from fuel D7. Fuel D6 gave rise to the semivolatile phase emission with the highest affinity for the dioxin receptor with the lowest EC50 With the exception of the sample from fuel D1, the dioxin receptor binding activities of the semivolatile phase emissions were similar or lower than those of the corresponding particulate phase emissions. Vehicle 2:
Particulate phase Figure 11 shows the EC5o values for dioxin receptor binding of the particulate phase exhaust emissions. Dioxin receptor binding activity was again observed in all the samples. Due to technical difficulties an ECso value for fuel D4 could not be estimated and insuf'ficient sample remained to repeat the analysis. On examination of the fuel dependency of the ECso values obtained it was observed that fuel D6 once again produced the emission with the highest receptor binding affinity with an EC50. However, there were no significant differences between any of the fuels with regard to the ECso's of their emissions. Only a 2-fold variation in EC50 values was observed in the samples from vehicle 2. TCDD
rn
N o 2
V e h i c l e
0 6
I m
M
V
--- Filter ~
S
II
--- PUF
T
D
Figure 11.- EC50 values, vehicle two; particulate and semivolatile phase,mean values and standard deviation.
91
Semivolatile phase Figure 1 1 shows the EC50 values for dioxin receptor binding of the semivolatile phase exhaust emissions. No detectable dioxin receptor binding activity was seen in the samples arising from fuels D1,D4 and D8.Receptor binding activity was seen in the semivolatile phase emissions from fuels D2 and D9. CONCLUSIONS
Answers to the main objectives of this study can be summarized as: 1.-Aromaticity shows a clear influence on the emission, those fuels with high aromatic contents have high levels of PAH in their emissions measured. 2.-The sulfur content of the fuel does not seem to have a marked influence on the emission, but for particulate and SOXemissions. 3.-Light density diesel fuels are favourable in terms of emissions compared to high density fuels. 4.The ignition improver did not have a major influence on the exhaust emissions. Refemng to the outcome of this study, improvements of the diesel fuel quality will have a clear beneficial effect on the air quality in city areas.
Acknowledgments We would like to thank all participants involved in this investigation. The project was financed by the Swedish Environmental Agency.
References Impact of Fuels On Diesel Exhaust Emissions: A chemical and biological charactenzation. Westerholm R. and Egeback K-E Swedish Environmental Protection Agency Report Series. Stockholm, 1990 Codes of Federal Register 40 Parts 81 to 99 Protection of Environment. Revised as of July 1, 1987 Characterization of TCDD-receptor ligands present in extracts of urban particulate matter. Toftgard R., FranzCn B., Gustafsson J.A. and Lofroth G. (1985) Environment International, Vol. 1 1, pp 369-374.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
93
FUELS EVOLUTION IN EUROPE AND CONSEQUENCES ON EXHAUST CATALYSIS J.C. Guibet
Institut Franqais du Pbrole, Techniques &Application Energktiques 1-4, avenue du Bois-Prkau, F - 92506 Rueil Malmaison France SUMMARY
1 - The market for unleaded gasolines will certainly grow rapidly in Europe in the early 1990 s (use with and without catalysts) 2 - The Eurosuper fuel (RON 95) is the most widespread product in Europe with relatively low characteristics variation from one country to another. West Germany (regular-grade) and France (superpremium) represent today extreme situations in the quality of unleaded fuels. There is no evolution of characteristics of unleaded fuels likely to alter the efficiency of catalysts. Lead traces are very low; sulfur contents are relatively low and can be further improved Interest in the future will be focused on : - the share of superpremium in the overall unleaded gasoline pool - the relations between the chemical composition of unleaded fuels and exhaust emissions in relation with catalyst performance INTRODUCTION
The major event in the European motor fuel distribution context is the recent appearance and penetration, at different rates in the various countries, of unleaded fuels. This development logically comes before the generalized spread of vehicles equipped with catalytic converters. These are still relatively few in number. Their population in 1990 is estimated at about two million vehicles out of a total of 140 million units. This means that, today, the majority of european unleaded fuels are used in vehicles without catalyst. This situation is bound to change because, from 1992, the future emissions standards will require the use of catalytic converters on all new vehicles. We intend to examine here the main properties of unleaded fuels distributed in Europe in 1989. Some more recent data on the first half of 1990 are provided for certain countries, including France. Among the quality criteria considered, we will particularly describe those likely to affect the efficiency of the catalytic converter. Our aim is to see how, for Europe as a
94
whole, the mutual adaptation of unleaded fuels and the new vehicles in circulation is being prepared, in anticipation of the generalized use of exhaust catalysts. 1. THE SHARE OF UNLEADED MOTOR FUELS IN THE EUROPEAN DISTRIBUTION NETWORK IN 1989/1990.
The different types of unleaded motor fuel currently distributed in Europe are divided into three grades : - regular-grade, mainly used in West Germany : its RON is close to 92 to 93 and its MON between 82 and 83. - standard motor fuel corresponding to the European specification (minimum RON 95, minimum MON 85) : it is found in all the EEC countries, as well as Austria, Finland, Norway, Sweden and Switzerland, Table 1 :European and world market for unleaded motor fuels Situation between end 1989 and June 1990 Country Japan Canada U.S.A. Austria Belgium Denmark Finland France West Germany Greece Ireland Italy Luxembourg Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom
Share of Unleaded Fuels
Distri ution by PI duct Type tegular Eurosuper Superpremium 0 (%) A%L 14 100 86 23 4 65 92 90 53 30 50 50 10.5 67
x 60 30
28
6 23 28 50 50 2 33
0 0.2 4 21 40 45 0.2 0.5 42 58
0. 4 19 33 45 0.2 0.5 41.5 58
25
21
24 2
8.5 6
2 7
0.5 4
95
- high-octane premium grade (RON 98, MON 88) : France, the Netherlands and, to a lesser degree, West Germany and the United Kingdom currently consume this type of fuel. For each country, Table 1 gives an estimate of the share of unleaded motor fuel distribution in the total gasoline pool. The situation of nonEuropean countries, such as the United States, Canada and Japan, is given for reference, and as an example of geographic areas where the use of catalytic converters has become widespread. The figures given must be considered as approximate on account of the fairly rapid growth of the unleaded fuel market in many countries. Thus Table 2 shows that, in France, the situation has been changing from month to month since the beginning of 1990. Table 2 : Penetration of unleaded motor fuels in France in the first half of 1990 Relative Market Share Product Type Regular-grade Gasoline Conventional Leaded Premium Unleaded Premium* TOTAL
(%)
January 4.6 89.1 6.3 100
March
5.o
86.6 8.4 100
April 4.5 85.4 10.1 100
June 4.3 80.5 15.2 100
For Europe as a whole, the unleaded motor fuel market is estimated in June 1990 at about 31 Mt/year for a total of 118 Mdyear. Thus, the share of unleaded fuels would be approximately 26 % of the total gasoline european market. 2. 2. GENERAL PROPERTIES OF DIFFERENT GRADES OF MOTOR FUELS IN EUROPE
The data given below are taken from recent surveys published by OCTEL and analyses made by Institut Frangais du Petrole on samples taken at service stations. The information is merely indicative, and does not claim to describe accurately the present situation of the gasoline pool which, as we have already pointed out, is changing fast.
96
2.1. Leaded motor fuels These products, which are still largely in the majority, are only examined briefly here because they are incompatible with catalytic converters. However, it may be interesting to recall their general properties, which could have repercussions on refining and therefore on the quality of unleaded fuels. Today, the maximum lead content in conventional motor fuels is 0.15 g/l in nearly all European countries except France, where it is still 0.25 g/l, and Spain and Portugal, where it is 0.40 g/l. Table 3 : Average and extreme properties of leaded premium-grade gasolines in some European Countries.
I Country France R.F.A. Italy Benelux United Kingdom
Lead I Content g/l
0.25 0.15 0.15 0.15 0.15
~
MON
RON I
Min 96.8 97.9 94.9 97.9
Average 97.6 98.9 97.0 98.1
Max 98.3 99.9 98.1 98.3
Min 85.6 87.9 86.2 86.8
Average 87.0 88.8 88.1 88.4
Max 90,l 90.2 90,4 89,3
97.5
98.2
99.6
86.4
87.2
88,O
In most cases (Table 3), the octane numbers of leaded premium- grade gasoline average about 98 RON and 88 MON. Assuming that a lead content of 0.15 gll achieves a gain of about 3 points in the RON and MON, it is clear that the development of Eurosuper type unleaded gasoline (RON 95, MON SS), to the detriment of conventional leaded premium-grade gasolines, could take place without any important change in the refining process. On the other hand, the production of large amounts of superpremium unleaded gasoline (98 RON) instead of leaded gasoline would require a deep adaptation of refining process. Concerning other leaded gasolines it must be noted that regular grade was recently eliminated in West Germany. It is still distributed in very limited quantities in France (5 % of the market). 2.2. Unleaded regular-grade gasoline
This type of product is essentially found in West Germany and Austria, where it accounts for 42 and 57 % respectively of all unleaded fuels. Table 4 shows the average and extreme octane numbers of unleaded regular-grade gasolines in.West Germany. We point out that the RON varies
97
over a fairly wide range between 90.5 and 94.1, whereas the MON levels are all in the 82 to 83 zone. The average RON (92.8) and MON (83.0) values are only 2 points under the Eurosuper type unleaded premium-grade gasoline specifications.[SP] Table 4 : Octane numbers of unleaded regular-grade gasolines in West Germany RON MON
Min 90.5 82.3
Average 92.8 83.0
Max 94.1 83.5
2.3. Standard unleaded gasoline, Eurosuper type. This is the most widespread product in Europe today in the class of unleaded gasolines. Table 5 shows that its octane numbers only differ slightly from one country to another, and meet the requisite specifications. The MON is generally very close to 85, while the RON is well above 95, as much as 96 or even 97. This shows that, for the refiner, the strongest constraint in terms of fuel blending concerns the MON. Very often, to obtain the minimum value required (85 for MON), the RON has to be let far above 95 (give away). This situation occurs particularly if significant amounts of olefin-rich, low MON catcracker gasolines are blended in the fuel. Table 5 : Average and extreme properties of eurosuper type unleaded premium-grade gasolines Country West Germany Italy Benelux UnitedKingdom Scandinavia
Min 94.9 94.7 94.7 96.2 94.4
RON Average 96.1 95.6 95.1 97.0 95.4
Max 98.6 96.4 95.4 97.7 97.4
Min 85.0 85.1 85.2 85.5 84.7
MON Average 85.4 87.0 85.8 86.0 85.3
Max 85.8 87.0 86.4 87.6 85.8
2.4. High-octane unleaded gasoline For this type of product, the minimum values are 98 for RON and 88 for MON respectively. This is the most widespread unleaded fuel in France today. The octane numbers found in the distribution network (Table 6) show a very slight dispersion around 88 for MON. The RON very often exceeds 98, and could reach or even go over 100 (exceptionally up to 102). As in the
98
case of Eurosuper, this reflects a more severe constraint for MON than for RON. Table 6 : Average and extreme octane numbers of superpremium type unleaded gasolines distributed in France.(Spring 1990) RON MON
Min 98.4 88.1
Average 99.7 88.4
Max 101.1 88.8
3. EXAMINATION OF SOME PROPERTIES LIKELY TO AFFECT AUTOMOBILE CATALYSIS
Catalysts run satisfactorily with a wide variety of motor fuels, provided they do not contain lead or other impurities liable to poison the catalyst. Hence the influence of the fuel is not decisive. However, it may be important in terms of light-off temperatures, composition of unconverted hydrocarbons, emissions of by-products such as H2S, and the endurance behavior of the catalyst. We examine a number of points that appear interesting to consider.
3.1. Lead content The maximum permissible content in Europe is 13 mg/l. Although complete analytical results for all the countries are not available, it is very likely that this specification is easily obtained. Thus, in a survey conducted in France in summer of 1989 on 22 samples, the traces of lead observed were always less than 2.5 mg/l, except in one case, where they reached 3.2 mg/l.
3.2. Chemical composition We are going to examine the distribution of components by hydrocarbon family (paraffins, olefins, aromatics) and its possible influence on catalyst effectiveness. This subject is still not perfectly understood, and merits a further investigation. However, we can recall the main conclusions of a study carried out recently at Institut FranGais du PCtrole1 in an engine bench test, equipped with a three-way catalyst and tested with 12 motor fuels covering a very wide range of compositions. 1- M.F. PRIGENT, B.C. MARTIN and J.C. GUIBET. Engine Bench Evaluation of gasoline composition effect on pollutants conversion rate by a three-way catalyst. SAE 900153. Detroit, 1990.
99
It was demonstrated that the light-off temperature (corresponding to 50 % conversion) tends to decrease with an olefin-rich fuel, such as a catcracker
gasoline, and to increase with an aromatic fuel, such a reformate. The differences of 20 to 25°C shown in Table 7 are significant. The tendencies observed have been confirmed on commercial fuels which differ by their olefins and aromatics content (Table 8). Moreover, the presence of oxygenated components does not appear to have any clear effect on the activation temperature of the catalytic converter. Table 7 : Light-off temperatures for CO, HC and NO, on a three-way catalyst. Comparison of behaviour of a reformate with a catcracker gasoline.
Table 8 : Light-off temperatures for CO, HC and NO, on a three-way catalyst. Comparison of behavior of two unleaded gasolines of different chemical compositions.
Given the tendencies shown here, it may be interesting to examine the ranges of variation in the olefins and aromatics contents of European unleaded gasolines. Table 9, related to products of the Eurosuper type (RON 95, MON 85), shows relatively narrow dispersions. On the average, the olefins content ranges between 6 and 9 % in the different countries, and the aromatics content between 35 and 39 % (42 % in Scandinavia).
100
Table 9 : Olefins and aromatics contents of European unleaded motor fuels. (Eurosuper grade, RON 95, MON 85)
If we compare, for Europe, the chemical composition of unleaded fuels as a function of their classification (regular, Eurosuper, Superpremium), a very logical increase in the aromatics content is observed and a reduction in the olefins content, when going from regular grade to Superpremium (Table 10). The higher aromatics content is explained by the need to meet an increasingly higher RON requirement (93 + 95 + 98) and the drop in the olefins content results from growing constraints on the MON (83 + 85 + 88). The variable addition of oxygenated products may slightly disturb these tendencies, but without significantly affecting them. Table 10 : Comparaison of European unleaded motorfuels as a function of classification.
3.3. Sulfur content
The small amounts of sulfur present in the gasoline can be stored in the form of sulfates on the three-way catalyst, in normal operation, and then discharged in the form of hydrogen sulfide (H2S) when vehicle runs on a rich mixture, as in starting up period. The unpleasant odor of the exhaust gases is a significant drawback, and the car manufacturers wish the sulfur content of the fuels to be as low as possible. For Eurosuper, the maximum permissible content is 1000 ppm today. However, French manufacturers, for example, in their quality label,
101
recommend average and maximum values, not exceeding 200 and 300 ppm respectively. Table 11 : Sulfur content of unleaded motor fuels (superpremium) distributed in France in 1989.
A survey conducted in France in the summer of 1989 on Superpremium type fuels revealed very low sulfur contents, as shown in Table 11. Generally, much lower sulfur contents are found in Superpremium gasolines (RON 98) than in Eurosuper (RON 95). This is because the Superpremium products usually contain a high proportion of reforming gasoline, which is practically sulfur-free, whereas the Eurosuper products contain a certain proportion of cracking gasoline, which is richer in sulfur. Table 12, which compares the sulfur contents of several samples of Eurosuper and Superpremium type products, clearly illustrates this tendency.
Table 1 2 : Comparison of sulfur contents of two types of unleaded gasoline pool. (France Spring 1990)
-
3.4. Oxygenated compounds The main oxygenated component of gasoline in Europe is MTBE. Table 13 shows the average percentage of unleaded fuels which contained it in 1989, as well as the average and extreme contents found. Some samples of Superpremium contained up to 11 % in France, and 15 to 16 % in West Germany.
102
Table 13 : Estimation of MTBE contents in European unleaded gasolines. Country
Type of Gasoline
Benelux Eurosuper France Superpremium* West Germany Superpremium** West Germany Eurosuper** West Germany Regular United Kingdom Eurosuper Italy Eurosuper
* : Summer 1989
Share of Fuels containing MTBE (%)
100 64 100 42 42 25 50
Contents Observed (%weight) Min Average Max 0.8 2.1 4 .O 1.0 6.8 11.4 1.3 9.6 15.9 0.7 1.1 1.6 0.1 0.5 1.1 4.4 2.2 4.0 6.2
** : December 1988
It is important to note that MTBE may be found in all types of product conventional leaded premium-grade gasolines, unleaded regular-grade (in West Germany), Eurosuper (many countries) and Superpremium (France). Methanol and tert butanol were still found in 1989 in West German fuels, leaded and unleaded. The average contents were about 1.2 % for methanol and 1.4 % for tert-butanol. These compounds could be present in the fuels together with MTBE. In the operation of catalytic converters, the presence of oxygenated components does not appear to have any substantial effect. However, car manufacturers do not wish an excessively wide range of concentrations of oxygenated products. This is because, in developing the inlet system, they have to consider the possible extreme compositions for motor fuels, and obtain satisfactory combustion in all circumstances. For example, the discrepancy of equivalence-ratio due to the variation of MTBE contents could modify the emission of unburnt products (carbon monoxyde and hydrocarbons), in the starting up period where the catalyst is not completely effective. CONCLUSION.
The market for unleaded gasolines will certainly grow rapidly in Europe in the early 1990's. These products will continue their penetration, first in a large proportion of present vehicles in circulation not yet fitted with catalytic converters, and then on new vehicles equipped with catalysts to meet future European pollution standards. Around these two extreme situations, illustrated by unleaded regulargrade gasoline (West Germany) and Superpremium (RON 98) which is very
103
widespread in France, the conventional Eurosuper type fuel (RON 95) is the most widespread product in Europe, with relatively minimal variations in grades and characteristics from one country to another. An examination of the main properties of motor fuels likely to alter the operation of the catalyst fails to reveal any disturbing situation or development. The residual lead contents are very low. The sulfur contents are relatively low and can be further improved. Interest in the future will be focused on the growing share of Superpremium in the overall unleaded gasoline pool, and on relationship between the chemical composition of the products (content of aromatics, olefins, oxygenated components) and exhaust emissions, incorporating the operation of catalytic converters in this examination. Institut FranGais du PCtrole, which is engaged in this type of study, hopes to contribute in its way to further progress in motor fuels, vehicles and high-performance catalysts in order to reach a better protection of the environment.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 199 1 Elsevier Science Publishers B.V., Amsterdam
105
SUPPLY AND DEMAND OF PRECIOUS METALS FOR AUTOMOTIVE AND OTHER USES Dr M.C.F. Steel
Johnson Matthey PLC, London, United Kingdom
Introduction The three precious metals used in automotive pollution control are platinum, palladium and rhodium. These three metals form part of the platinum group, are closely related in terms of chemical and physical properties, and occur together in nature. Their geological abundance, detailed applications and supply/demand fundamentals are, however, significantly different and this paper will outline and discuss these differences.
Platinum G r o w Metals (PGM1 The platinum group of metals consists of six elements: ruthenium, rhodium, palladium, osmium, iridium and platinum. The metals are all quite rare and occur only in a few comparatively concentrated deposits.
G r o w Ib
Groun VIII Ru 0s
Rh 1
I
I
Au
I
These invariably contain all six metals but in varying quantities depending upon the mineral source. Separation of the platinum metals is a long and difficult process owing to the marked chemical similarities between both the metals and their compounds. As table 1 shows, the platinum metals occur in the same group of the periodic table as the ferrous metals. Although chemically the resemblances between them are slight they do tend to occur together in nature. The platinum metals are often referred to as precious metals along with gold and
106
silver. Although they share some properties, such as the nobility of gold and PGM, and applications, for example as jewellery and investment media, the PGM are much more important as industrial metals.
The sources of platinum group metals in the world that are suitable for economic exploitation are few. Figure 1 shows where these resources are located. The principal source is the Bushveld Igneous Complex in South Africa. Other significant mining for PGM occurs in the USSR, Canada and the USA. China is believed to produce same PGM but is still a net importer of the metals. Small amounts of PGM are also recovered elsewhere in the world either as placer deposits or as by-products of base-metal mining, but the contribution of these to world supplies is insignificant.
USSR
Figure I :World sources of platinum group metals
Although there has been much exploration in recent years for PGM in such countries as Australia and Canada no significant economically viable deposits have been found. In Zimbabwe, however, a well established but relatively low grade PGM deposit known as the Great Dyke is once again receiving attention as a possible source of platinum metals.
107
Table 2 : Reserves of platinum, palladium and rhodium (million 0 2 )
I I
Pt
Pd
Rh
333 437 160
141 365 175
17 83 12
CANADA, Sudbury
3
4
1
URSS, Noril'sk,
50
142
6
USA, Stillwater,
7 ------
------
------
850
120
Origin SOUTHAFRICA Merensky UG2 Platreef
23
3
As table 2 indicates, the proven reserves of platinum, palladium and rhodium in South Africa, the USSR, Canada and the USA are considerable and sufficient to last for between 250 and 400 years at current rates of consumption. The Bushveld Igneous Complex contains overwhelmingly the largest reserve of PGM (Refs 1,2). Although adequate reserves of PGM exist, the mining of these metals will always be expensive because of the enormous mechanical and human resources needed to mine the ore (it takes approximately 10 tonnes of ore to yield 1 oz of platinum in South Africa) and the complexity of chemically separating and refining the six platinum group metals to the high purity required in the market place. The relative mix of the six PGM - the 'mine mix' - is important, with the deposits in the northern hemisphere showing a preponderance of palladium whereas in South Africa platinum is the principal metal. Although the deposits in South Africa are mined principally for PGM, those in Canada and the USSR are essentially nickel and copper deposits with PGM being produced as by-products. This means that the output of PGM from Canada and the USSR is inelastic as far as PGM supply/demand fundamentals are concerned.
1 i (Ref 3) Automotive pollution control is a major application for PGM but the relative importance for the three autocatalyst metals differs considerably as Table 3 shows. In 1989 same 42% of the western world demand for platinum was for the manufacture of autocatalyst, whereas for palladium the
108
corresponding figure was only 8%. For rhodium, the market for which is only one-tenth the size of those for platinum and palladium, autocatalyst usage accounted for a massive 81% of total demand.
Table 3. Western World Demand for PGM in 1989.
In 1989 demand for platinum reached 3.425 million 0 2 . Figure 2 shows that autocatalyst accounted for 37% of demand (after allowing for recovery of spent autocatalyst), with jewellery taking 38%, investment 5% and a number of important industrial applications the remaining 20%. Industrial uses for platinum include petroleum reforming catalysts for gasoline production, ammonia oxidation catalysts for nitric acid production, bushings for glass fibre and high purity glass production, and thick film pastes and sputtering targets for the electronics industry. Supplies of platinum in 1989 are shown in Figure 3 and came from South Africa (77%), the USSR (16%), Canada (4%) and other miscellaneous producers (3%) . The relatively small shortfall of newly mined supply compared with demand was met from industry stocks held around the world.
Figure 2 :Platinum demand in 1989. Total demand = 3,425,000 oz
109
SOUTH AFRICA 77%
USSR SALES 16%
OTHERS 3% CANADA 4%
Figure 3 :Platinum supplies in 1989. Total supplies = 3375,OO oz
Demand for palladium of 3.31 million oz also exceeded newly-mined supplies by a similar amount. Figure 4 shows that for this metal the USSR was the principal supplier (51 %) with South Africa contributing 35%, Canada 6% and others 7%. Half the western worlds palladium was used for the electronics industry in 1989, with 30% going into dental alloys, 6% into autocatalyst (net), 5% jewellery and the remaining 9% a variety of industrial applications mainly in the chemical industry. Figure 5 shows the breakdown of palladium uses.
SOUTH AFRICA 35%
CANADA 6%
OTHERS 7%
Figure 4 :Palladium supplies in 1989. Total supplies = 3,205,000 oz
The market for rhodium is approximately one tenth that of platinum and palladium and in 1989 supply and demand were in balance at 0.33 million oz. As stated earlier, same fourth-fifths of demand was for autocatalyst, with other chemical catalyst uses accounting for a further 9%, electrical uses 4%, and miscellaneous applications the residual 8%. Supplies were from South
110
Africa (56%), the USSR (39%) and Canada (5%). Figures 6 and 7 show, respectively, demand and supply of rhodium in 1989. ELECTRICAL 50%
DENTAL 30%
AUTOCATALYST 6%
Figure 5 :Palladium demand in 1989. Total demand = 3,310,000 oz AUTOCATALYST(net) 79%
CHEMICAL 9%
GLASS 1% OTHER 7% ELECTRICAL 4%
Figure 6 :Rhodium demand in 1989. Total demand = 330,000 oz
CANADA 5%
Figure 7: Rhodium supplies in 1989. Total supplies = 330,000 oz
111
A significant difference between the markets for platinum and palladium and that for rhodium, in addition to their absolute size, is the relative liquidity of these markets. Platinum and palladium are widely traded around the world with futures markets existing in both New York and Tokyo and a well established spot market in London and Zurich. This and the wide spread of applications for each metal has led to the accumulation of stocks of metal around the world that can be used either to satisfy supply shortfalls or to absorb occasional surpluses. For rhodium, however, there are no formal spot or futures markets, the total market is much smaller and dominated by a single application, and industry stocks are therefore significantly lower than for platinum and palladium. In addition, supply of rhodium is rather inflexible due to the fact that rhodium is essentially a byproduct of platinum mining and that it takes six months or more from the mining of ore to the emergence of pure rhodium from the refinery.
PGM in Autocata lVsl When considering the outlook for the use of platinum, palladium and rhodium in automotive pollution control, it is important to consider the ratios of the metals in exploitable ore deposits and in autocatalyst formulations. The requirement for platinum, palladium and rhodium in autocatalyst is determined principally by the reactions that the metals are present to catalyse (Ref 4). These are shown in Figure 8.
Fig 8 Autocatalyst Reactions
HC
co
oxidation
H20 + CO2 + oxidation ---
---+
co*
In the early days of autocatalyst the principal requirement was to oxidise unburned hydrocarbons (HC) and carbon monoxide (CO). To do this catalysts of platinum, palladium, or a mixture of these two elements, were chosen. As legislation has become stricter and has focussed on additionally removing nitrogen oxides (NOx) the role of rhodium to catalyse the reduction of NO, to nitrogen and water has become critical. In the early nineteen-eighties dual bed systems were used containing a reduction catalyst (usually Pt/Rh) followed by an oxidation catalyst (usually PtPd). Modern engine technology employs sophisticated control of air-fuel ratios by using electronic fuel injection and sensors and has enabled the auto industry to converge on the use of a single catalyst to remove all three pollutants in one unit. Such 'three-
112
way' catalysts almost invariably contain platinum and rhodium, in a ratio of 5:1, as the active metals. N AMERICA 50%
JAPAN 24%
1- 1 W EUROPE 21%
Figure 9 :Platinum demand for autocatalyst by region in 1989 Total demand (gross) = 1,450,000 oz Figure 9 shows that the demand for platinum in autocatalyst in 1989 was 1.45 million oz and this is expected to grow substantially through the current decade. Initially the focus is on Western Europe, where by 1993 all new cars are expected to be fitted with catalysts. The use of catalytic converters in several Eastern European countries now also looks probable. In Brazil and Mexico legislation is in place that will result in catalysts being used in the next five years, and other countries with severe pollution problems may be expected to follow suit. In addition, the US Congress is currently progressing a new Clean Air Act which will contain tighter emission standards that will set a new world standard that other countries can be expected to adopt in years to come. This progressive tightening and geographical widening of automotive pollution control will lead to a growth in PGM consumption for autocatalyst. The question that follows is whether the necessary PGM will be available and from which sources.
Given the reserve base existing in the Bushveld Igneous Complex (Table 2) and the flexibility there of mining principally for platinum metals it is clear that South Africa must remain the prime source of PGM. Considerable expansion is already underway to satisfy the expected demand for autocatalyst production. A feature of the Bushveld Complex is that the PGM are contained in three different mineral formations. These are the Merensky Reef, the UG2 reef and the Platreef. As table 4 shows, the mine mix of PGM in these formations is significantly different, with the UG2
113
providing a mix that is closest to the current favoured autocatalyst ratio of Pt:Rh at 5:1. Table 4 : South African Ore Ratios
UG2
I Platreef I
14.0
I
15.3
I
1
I
Given the certainty of the growth of the market for autocatalyst during the remainder of this decade, it is not surprising to find that the main thrust of current expansion plans in South Africa is for greater exploitation of the UG2 reef (table 5). In 1988 some 92% of South Africa's platinum production was from the Merensky Reef and only 8% from UG2. Table 6 indicates that by 1996 it is expected that 70% will come from Merensky. 27% from UG2 and 3% from the Platreef.
Table 5 : PGM Production in South Africa Present Mining
Expansion
Ru stenburg Lebowa Potgietersrust
MNG2 M
UG2>M M P
Impala Messina
M / UG2
UG2 / M UG2>M
Western Eastern Karee
M / UG2 UG2 (new) M / UG2 (new)
UG2 UG2 M / UG2
Crocodile River Kennedy's Vale
UG2 (new)
UG2 UG2
Northam
Merensky UG2 Platreef
M
1988 92% 8% 0%
1992 80% 20% 0%
1996 70% 27 % 3%
114
Conclusion The market for platinum metals in automotive pollution control is expected to grow substantially over the next ten years, particulary for platinum and rhodium. To meet this growth mining companies are investing in new mines and in the expansion of existing mines to provide extra PGM. In making these investments attention is being focussed on the particular requirement for extra rhodium. This should ensure that the demands of the auto industry for PGM are satisfied as the market for automotive pollution control using PGM catalysts grows.
References 1
D L Buchanan, "Platinum Group Metal Production from the Bushveld Complex", (1979)
2
J R Loebenstein, "Mineral Facts and Problems", US Bureau of Mines, Bulletin 675, (1985)
3
J S Coombes, "Platinum 1990", Johnson Matthey, (1990)
4
M C F Steel, "Changing Patterns of Platinum Group Metals Use in Autocatalyst" SAE International Congress Paper 880127, (1988)
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 Q 199 1 Elsevier Science Publishers B.V., Amsterdam
115
THE VALUE OF SPENT EXHAUST CATALYSTS
W. Fierain MHO, BU HOBOKEN, BELGIUM
Abstract There is some logic in scheduling this paper at the end of the session since it deals with spent catalysts which, by definition, have reached the end of their useful life. Yet, the ultimate stage in life for a spent catalyst is a very valuable one since it becomes then part of the largest secondary source of platinum group metals worldwide. This paper will explain how and why and will indicate the obvious relationship between the development of this source and all the key elements governing the use of autocatalysts, meaning: - the environmental legislation planned or in place - the currently available technology - future new car sales - supply, demand and prices of the platinum group metals. INTRODUCTION.
Before going any further, a rapid sketch of MHO's activity might help understanding why we do consider spent exhaust catalysts as a "rising star" material. MHO, a Division of ACEC-UNION MINIERE, has two production sites in Belgium, one at Olen and one at Hoboken near Antwerp. Highly specialized in the treatment of complex precious metals bearing raw materials, the Hoboken Plant uses hydro and pyrometallurgical processes for the refining of - precious metals: Ag - Au - Pt - Pd - Rh - base metals: Pb - Cu - Sb - Sn - special metals: Se - Te - In - Bi On a yearly basis, the Hoboken Plant receives from all over the world more than 300,000 tonnes of various types of non-ferrous and precious metals bearing raw materials in the form of concentrates, metallurgical and chemical byproducts, spent petrochemical catalysts or industrial discards from the electronic industry. In addition to that and since the early 803, it has refined several thousand tons of spent catalysts from the automotive industry, mainly
116
of North American origin. With almost a decade of experience in the field, MHO is still actively pursuing its marketing and metallurgical research efforts with the aim of extensively implementing the optimal recycling scheme for a device which is or will be part of day-to-day's life for a vast majority of individuals in the industrialized world. THE VALUES
Different values can be associated with the spent exhaust catalysts market: strategic - ecological - industrial -economical. Let us examine each of these.
The strategic value ("putting something by for a rainy day): Biginning in the mid-70s, it has become a general concern to reduce the level of exhaust emissions from motor vehicles. Both Japan and the USA were the first to adopt tight standards for carbon monoxide, unburnt hydrocarbon and nitrous oxide emissions. The fitting of platinum group metals catalytic converters became the only efficient tool to achieve these standards. A similar package of standards applies or will soon apply in Europe and the installation of the so-called 3-way catalyst, which is mainly a platinum rhodium catalyst with a ratio 5 to 1 P m h , seems unavoidable. Because of public pressure and tax incentives, some European countries have been ahead of the EEC legislation: this is the case of Austria, Switzerland, Norway and Sweden. Likewise early this year, 80 % of the motor cars sold in Germany were already fitted with catalysts. This growing use of automotive catalytic converters has had a tremendous impact on the demand for PGM's. Starting from nil in the early ~ O ' SPt , demand from the automotive industry reached 40 % of total Western world platinum demand in 1989, as shown by Metals and Minerals Research Services. This is very much in line with the figures issued by Johnson Matthey in the 1989 Platinum review. For the same year 1989, platinum recovered from autocatalyst recycling was estimated at about 5.5 tonnes, equivalent to approximately 5 % of the total demand for platinum in the Western World. This is far from being an insignificant figure: indeed this secondary supply is, for instance, three times the estimated platinum output of the Stillwater Mine in Montana, the sole primary producer of platinum group metals in the United States Those 5.5 tonnes Pt are equal to the production of Canada, third largest supplier of platinum in the world, with Inco and Falconbridge producing
117
platinum as a byproduct of nickel smelting and refining. As a third comparison, this secondary supply is no less than the total US platinum demand from the chemical, electrical and glass industries all together. And last but not least, these 5.5 tonnes represent 50 % of the 1989 European platinum demand from the autocatalyst industry. Japan and United States together represent 75 % of the gross demand for platinum from the automotive industry i.e. about 34 tonnes; by comparison recycled platinum worldwide represents 16 % of their combined consumption in autocatalyst during 1989. The strategic value becomes obvious when remembering that more than 70 % of the western world supply originates from South Africa and that Soviet Union supplies more than 10 % of the Western market requirements. No doubt all producers have substantial quantities of PGM in the pipeline; no doubt PGM consumers own security stocks against possible disruption at mines or refineries due to political events or technical failures: pipeline and stocks altogether represent about one year of demand. In this respect, there is no doubt either that recycled PGM is definitely an hedge against any impediment in primary shipments or temporary tight supply. This has not escaped to Japan which is gearing up to import more platinum group metals derived from spent converters from the United States and has established its own collection network locally. In a similar approach, some German car producers are already today organized to have the spent catalysts collected by their dealers, decanned and the PGMs recycled and returned to the fresh catalyst producers. Palladium is showing a lesser strategic sensitivity in this context because regulations on emission standards are leading to the preferred use of platinum rhodium catalysts. By contrast, rhodium undoubtedly exceeds by far the strategic value of platinum because with a platinum/rhodium consumption ratio of 5 to 1 and a mine production ratio of 15 to 1, rhodium stocks are gradually vanishing in an already tight market.
The ecological value (Recycling the product): An article issued by John Griffiths in the Financial Times of May 22, 1990 announced that a major German car manufacturer was designing a pilot plant to search for and experiment the best way to achieve complete car recycling; the pilot facility will be used to disassemble and process a total of 1,500 cars over the next 12 months. One of the most important objectives of this project is to apply knowledge learned from the disassembly process to the design of future cars so that they can be specifically developed to be much more easily recycled than it is the case today.
118
After the pilot phase, a permanent recycling plant capable of processing
250,000 cars per year could be build within the next two years. The hope is that such plants would become a norm for the motor industry with the farreaching idea of recycling 100 % of every car built. Recycling of spent catalysts perfectly matches this valuable objective and a close cooperation should exist between collectors/refiners and catalysts producers/car manufacturers. Indeed, optimizing the complete recycling circuit still presents a lot of handicaps and challenging aspects. The next 2 topics will clarify what we are speaking about but before proceeding let me point out that the recycling industry is presently deeply concerned with two enhancing trends in environmental regulations: first, to classify all recyclable and secondary material as waste and second to prohibit the international transport of waste. We are very much worried by such a possibility which ignores the fact that the inherent value of precious metals bearing secondary raw materials already commands such careful management that it is inevitably protective of the environment.
The industrial value: The extensive process of recovering autocatalysts involves many different steps and players: - pipes and canisters removal: - car dismantling yards - car shredders - muffler shops - car dealers - garages - pipes
and canisters collection - sorting:
- scrap dealers - specialized collectors
- opening of the canisters (decanning): sorting of the catalysts lot preparation (possible crushing and pre-sampling)
- specialized collectors and consolidators
- sampling :
- refiners
- smelting and refining :
- refiners
The first part of the chain: the scrap collection network includes: - pipes and canisters collection - catalysts unloading
119
- catalysts sorting - catalysts defining - catalysts decanning - catalysts packinglstoring - catalysts empty shells The second part of the chain starts after receipt of the material at the refining plant; at MHO the sampling procedure for spent auto catalysts can be shortly described as follows, auto catalysts being delivered in the form of honeycombs (crushed, partly broken or fresh) or beads. Pretreatment Uncrushed or partly broken honeycombs are crushed with a hammer-mill to obtain a fine product that can be sampled in the sweeps automatic installation. Primary sampling The precrushed honeycombs or the beads are sampled with the automatic sweeps sampler. This unit has been especially designed to weigh, empty and sample fine materials packed in drums or big bags. The drums or big bags are unloaded in 2500 1. hoppers. Two of these hoppers are emptied in a 8000 1 biconical blender. After 15 min blending, the blender is unloaded over a rotary divider, which takes a 10 % sample. This 10 % sample goes in a 500 1. hopper, which is emptied in a 1000 1. biconical blender. After 15 min blending, the second blender is unloaded over another rotary divider, which takes a 1 or 2 % sample (i.e. 0.1 or 0.2 % of the original input). Final sample preparation The samples from the successive blending operations are transferred to the final sampling area where they are recomposed into a 3 kg sample with due proportion taken of their individual weight. This 3 kg sample is blended by means of a small biconical blender. After blending, it is crushed in a vibrating ring mill and screened on a 130 mesh screen. The fine material is then blended in a turbula blender and divided into analytical samples by means of a rotary splitter. The analytical sample-bags are sealed and sent to the analytical laboratories of the supplier and of MHO. Sealed bags are kept for possible umpire analysis. The final assay results will be determined by exchange of the figures determined on the official samples by the supplier and by MHO.
120
The so-determined final assay results will govern the settlement of the delivered material either by payment or by return of refined metals to the suppliers. Refining
The third part of the chain concerns the refining. Several flowsheets have been developed. These include: -hydrometallurgical technologies like: substrate dissolution with sulphuric acid or sodium hydroxide resulting in a PGM residue suitable for conventional refining -or selective dissolution of precious metals in a hydrochloric acid solution containing an organic oxidant plus proprietary chemicals -dry chlorination at 400/1000 "C followed by recuperation and refining of the volatile PGM chlorides: this process is not yet commercially proven. -pyrometallurgical technologies like: melting of spent autocatalysts in a plasma arc furnace or an electrical furnace using iron or copper as collector: the iron alloy contains 2 to 4 % PGM and can either be granulated and leached to produce PGM concentrate suitable for conventional refining or the iron or copper alloy at 2 to 4 % PGM can be directly sent to a refiner for further processing. or direct melting and refining of spent catalysts through the conventional nickel, copper or lead refining circuits. Each process has its own economical, ecological and metallurgical constraints. However, up to now, the pyro-metallurgical circuits seem to be the most efficient processes as they are the final outlet for more than 75 % of the spent automotive catalysts available on the market. Why is it so ? We all know that autocatalysts consist of a refractory oxide carrier such as alumina or cordierite (Al2O3, Si02, MgO) on which two or three PGM are spread in low concentration. The monolith type catalyst (honeycomb) is typically made of cordierite and the bead type (pellets) is made of gamma alumina. Average assays of the catalysts received by MHO from the USA during the past 10 years show the precious metals concentrations (drnetric ton) given in the following table and thus, as you can notice, very low PGM concentrations.:
121
In addition, the spent catalysts also pick up other metals from gazoline or motor additives like Pb, Mn, S, P or carbon and eventually stainless steel (nickel) or even dirt during the decanning. Furthermore, catalysts batches do not reach the refiner on a regular basis throughout the year. In order to cope with these unstable parameters we have chosen, at Hoboken, to load the spent catalysts after sampling at the sintering plant, together with a large variety of other raw materials such as concentrates, lead sulphate, dust or slimes for instance. The aim is to achieve a stable composition of the mix for sintering, therefore optimizing the recovery rates of the non ferrous metals including the PGMs. Another advantage of the Hoboken operations is to avoid the generation of by-products which would be economically or environmentally burdensome. Alumina, silica and magnesia are tapped from the blast furnace as a neutral slag while precious metals including the PGMs are collected in the lead bullion, the copper matte or the nickel speiss, which intermediate products are then further refined.
The economical value: With such strategic, ecological and industrial values associated to it as major driving forces autocatalyst recycling should have reached the "cash cow" stage. This is not yet the case, as shown hereunder. Justifications for low recycling rate in the USA are: - the anti-converter attitude on the part of many vehicle owners who remove and simply discard the converters in order to save money by using leaded rather than unleaded fuel and to possibly gain engine performance; the Environmental Protection Agency (EPA) did estimate that in 1984, 7 % of cars were running without converters. - the reconditioning of old converters as replacement equipment after quick cleaning and painting: this accounted for 10 to 20 % of the catalyst market in the.early 80s - the lack of an organized scrap collection network - the loss of PGMs due to mechanical attrition during use coupled with loss in the collecting and decanning process - finally, the difficulty for consolidators to maintain a regular feed flow because of price fluctuations arising from competition or PGM's price volatility. Nevertheless, there is now a definite trend for improvement in the reclamation rate thanks to: - the phasing out of leaded fuel and better campaigns promoting the merits of converters.
122
- the enforcement of EPA performance standards since November 1986 limiting the sale of reconditioned converters - the EPA legislation stipulating that dismantling yards must remove spent catalysts before scrapping vehicles because of the residual lead contained in some of them. Consequently, quantities of spent catalysts made available in the USA have developed as follows: 1985 1986 1987 1988 1989
3,600 4,500 5,000 6,000 7,500
tons tons tons tons tons
In Europe our forecast of theoretical availability of PGM in spent autocatalysts is as follows, based on: - trend in new petrol powered car sales - existing car exhaust emission legislation - an actual average of 1.8 g of PGM to be recovered per vehicle - disposal of spent catalyst 8 years after the first registration of the vehicle What does it mean in view of economies ? The total recovery of PGM from spent autocatalysts will reasonably reach 20 tonnes at the end of the decade as we have some room for improvement in the USA and as the collecting system in Europe is more developed and probably more cost efficient than it is in the United States. At the level of 20 tonnes, the primary producers will see a substantial part of the market requirements already filled up and this might possibly delay or cancel the development of the most expensive mines for which cost of production could reach US $ 350/oz. Such a scenario means that spent automotive catalyst recovery would be a determinant factor for the floor price of platinum. On the contrary, should the market be oversupplied, the platinum price would fall preventing recovery from being profitable; indeed at prices below US $ 450/oz, recovery becomes quickly unprofitable: in the US collectors buy for cash at prices between US $10 and 20 per unit depending on the type of catalyst, the PGM prices and the overall availability of scrap. However, in such a case, and contrary to what happened in the early 80s when the prices of platinum and palladium fell to levels which were so low that they discouraged any risky position, we are convinced that salvage companies will not throw catalysts away but will simply stockpile them for a time.
123
The result will be that at some future time the so-freezed PGMs will flow back onto the market anyway. This is commonly observed in the copper, lead, zinc, silver and gold market and it will apply to PGM as well. At MHO we have learned to live and operate with these realities and not to believe that the world is other than it is.
SOURCES Automobile catalyst and the PGM market by Jeffrey M. Christian, CMP Group Metal Bulletin 3rd Platinum Seminar 1987. Platinum 1990, Johnson Matthey PLC May 1990 Financial Times, Tuesday May 22, 1990 Charles E. Cunningham, A1 Specialized Services and Suppliers Inc One decade of commercial recovery of platinum group metals from autocatalyst: some change and implication New York Copper Club, March 1990. Francis Van Bellen - Sampling strategy for precious metals bearing materials at MHO IPMI, San Diego 1990 Platinum Lake Technology process for recovery of precious metals from automotive catalyst Recycling of Metalliferous Materials Platinum the way ahead - Metals and Minerals Research Services Ltd July 1990
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
125
CERIUM OXIDE STABILIZATION: PHYSICAL PROPERTY AND THREE-WAY ACTIVITY CONSIDERATIONS J. E. Kubsh, J. S. Rieck and N. D. Spencer W. R. Grace & Co.-Conn. Research Division Columbia, Maryland U.S.A.
Abstract Improved cerium oxide surface area stability in high temperature oxidizing environments has been demonstrated for stabilized cerias containing La, Nd or Y dopants. BET areas in excess of 30 m2/g can be achieved after a one hour, 980 O C calcination for a ceria stabilized with 10 mole % La. Three-way activity comparisons between catalysts promoted by stabilized and unstabilized cerias show significant activity penalties for La and Nd-stabilized ceria-containing catalysts after air oxidations for four hours at 980 "C. X P S characterizationof La- and Nd- stabilized cerias has revealed significant surface enrichment of ceria by La and Nd under high temperature oxidizing conditions. This ceria encapsulation is believed responsible for the observed loss in activity. INTRODUCTION
During the last five years, much of the focus of automotive exhaust catalysis research has been on catalyst thermal stability. This emphasis stems in part from the increasing number of vehicle applications that utilize closecoupled catalyst locations, in order to minimize vehicle emissions during cold starts. At high speeds and high loads, these close-coupled locations subject catalysts to temperature extremes that can often be in excess of 900 "C. Operation of three-way catalysts at these high temperatures can result in significant activity losses due to the large scale sintering of the alumina support, sintering of the noble metals, and sintering of other catalyst components including the cerium oxide promoters. Numerous recent studies have characterized this thermal deactivation process in terms of support surface area loss, or noble metal and cerium oxide crystallite growth [I-41. Improvements in three-way catalyst thermal stability have resulted from the incorporation of stabilized aluminas and high levels of cerium oxide promoters. Alumina stabilization has been the subject of several studies with La203 stabilization being generally preferred [5-81.
126
The increased levels of ceria present in three way catalysts have focused some attention on ceria physical properties and thermal stability in these systems. Both the patent literature [9-121 and the scientific literature [13,14] contain references to the synthesis of high surface area Ce02 or the stabilization of ceria for auto catalyst applications. In one recent study Miyoshi et al. [ 131 provided evidence for improved three-way catalyst thermal stability by incorporating lanthanum oxide in both the alumina support and in the ceria promoter The high ceria loadings currently used in three-way catalyst formulations are most readily achieved by physical mixing of bulk cerium oxide with alumina. In order to maximize the catalytic properties of the ceria additive, a high surface area is clearly desirable. On the other hand, the ceria should be resistant to sintering under high temperature conditions. In this paper we discuss the sintering characteristics and surface chemistry of stabilized and unstabilized ceria, and the performance of stabilized ceria in three-way catalysts.
Experimental Details Stabilized cerium oxides were prepared by co-precipitating aqueous solutions of cerium nitrate and the corresponding nitrate of the stabilization agent with a dilute NH40H solution. Precipitations were performed at 85 "C and at a constant pH=9. Stabilization agents investigated included La, Y, Nd and Al. After precipitation, the resulting mixed hydroxides were washed with deionizing water, dried at 120 "C and then calcined for one hour at 540 "C to yield a cerium oxide composite. Stabilized cerium oxides were also prepared by an incipient wetness impregnation of a commercially available high surface area ceria obtained from Novamet Specialty Products Corp. [Wyckoff, NJ U.S.A.] with nitrate solutions of either lanthanum or neodymium. After impregnation, the powder was dried at 120 "C and then calcined for one hour at 540 "C to yield the mixed oxide. XPS characterization of materials was carried out using a Perkin-Elmer 5400 Series ESCA instrument, operating with 300 W of Mg K a radiation. A hemispherical electron energy analyzer was used, giving an energy resolution of 1 eV [on Ag 3d5/2] at a pass energy of 18 eV. The spectrometer energy scale was calibrated with respect to the Cu 2 ~ 3 ~Au 2 , 4f712, and Ag 3d5/2 core levels at 932.7 eV, 84.0 eV, and 368.3 eV, respectively. Powder samples were examined by dusting them onto double-sided adhesive tape, mounted into a stainless steel sample stub. Charge correction was performed using the (CH), peak at 284.6 eV. Surface concentrations were calculated using PHI sensitivity factors.
127
Three-way activity measurements were made using a simulated exhaust gas reactor system described elsewhere [ 151. This simulated exhaust stream contained 20 vppm S2, 400 vppm hydrocarbons (3/1 molar ratio of propene/propane) , 1850 vvpm NO, 10 ;ol.% H20, 14,5 vol.% CO2, and appropriate levels of CO and 0 2 to cover the stoichiometric range of interest. The reactor system included provisions for oscillating the feed gas stoichiometry at 1 Hz to simulate closed loop operations typical of vehicle conditions. Conversion levels of hydrocarbons, CO and NO, were measured at 480 "C over a range of air/fuel ratios using a gas hourly space velocity of ca. 40,000 hr-1 (at STP). In these activity studies three-way catalyst washcoats were prepared by adding bulk cerium oxide or bulk stabilized cerium oxide to a La203-stabilized alumina that contained Pt and Rh (introduced by impregnation). This mixture was wet ball milled and then coated onto a corrugated 0.05 mm metal foil (Allegheny Ludlum, Alfa IV alloy). The coated foil was then spirally wound and inserted into a thinwalled stainless steel tubular case to form a sample 2.54 cm dia. x 2.54 cm long (total catalyst volume of 12.9 cm3). Noble metal concentrations on the washcoat used for these samples were chosen to yield 0.71 g/liter loadings with Pt/Rh=5/1 by weight. CERIASTABILISATION Stabilized cerias made by the standard co-precipitation process described in brief above were subjected to a high temperature calcination (one hour in air at 980 "C) to determine the surface area stability of these materials relative to a precipitated ceria containing no additives. Table 1 compares BET surface areas of these oxides as prepared (with 10 mole % stabilizer, calcined to 540 "C) and after the high temperature calcination. Included in this table are also surface area results for a commercial, high purity ceria. As indicated by the data in Table 1, the precipitation process used here was successful in producing ceria and mixed oxides with surface areas in excess of 70 m2/g after a 540 "C calcination. The second important result is that the incorporation of La, Nd and Y into the ceria matrix significantly improved the high temperature stability of the mixed oxides in comparison to the pure ceria made by precipitation or the commercial ceria. In all three cases La-, Nd- and Y-doped cerias exhibited surface areas in excess of 25 m2/g after this severe calcination. A1 incorporation into the ceria matrix resulted in some improved thermal stability relative to pure ceria but not to the same extent as La. Nd or Y stabilization.
128
TABLE 1 - BET Surface Area Data for Stabilized Cerias
* - including La and Nd Since the L a 2 0 3 / C e 0 2 composite provided the highest degree of surface area stability for the dopants examined here, a more detailed study was made of the effect of lanthana content on the thermal stability of this mixed rare earth oxide. Figure 1 compares BET surface areas for La203/Ce02 made by this constant pH precipitation as a function of the La content of the mixed rare earth oxide. The 10 mole % La level used in the initial investigations is close to the optimal stabilizer level for maximizing thermally aged surface area. This maximum, however, does not appear Mole 56 La sharp, with La levels between 5 and Figurel: BET areas for La203-Ce02 20 mole % yielding BET areas in co-precipitated mixed oxides excess of 25 m2/g after calcination at 980 "C. As-prepared surface areas also appear to track with thermally aged areas, with the highest surface area after 540 "C corresponding to about 10 mole % La. In the composition region between 0 and 20 mole % La, the two rare earth oxides form a solid solution, with Ce02 being the only crystalline phase present in the X-ray diffractograms of these solids, even after air calcinations to 980 "C. Both pure ceria and pure lanthana made by precipitation exhibited poor thermal stability relative to the stabilized La203/Ce02 composite.
129
THREE-WAY ACTIVITY Washcoats were prepared using the stabilized La203/Ce02 material described above in order to determine if the improved sintering characteristics of this material resulted in improved catalyst thermal stability as measured by three-way conversion levels after high temperature aging. As described previously, a standard washcoat formulation was used in these studies. In comparative evaluations, all washcoat ingredients were kept constant except the nature of the bulk ceria additive. The amount of ceriabased additive was also kept constant at 25 wt% on a washcoat-only basis. Before proceeding with - any high temperature treatments, the activities IConversion at Stoich. Conditions of all samples were evaluated at our 100 standard three-way conditions in the as-prepared state which included a Activity measured at 48OOC 80 mild calcination in air at 500 "C. All samples gave equivalent fresh conversion levels with respect to 60 hydrocarbons, CO and NOx. In the first thermal stability comparison, catalyst activities were 40 compared for a washcoat containing a relatively pure commercial ceria and 20 a coating containing a lanthanastabilized ceria (10 mole % La) made by co-precipitation. Catalyst samples 0 CO NOx Hydrocarbons were aged for four hours at 980 "C in air to simulate a severe high Unstabilized ceria La-stabilized ceria (,o La, co-precip.) temperature oxidation in a lean exhaust. Figure 2 compares threeFigure 2: Three-way activity way conversion levels for these two comparisonfor PtlRh washcoats after samples in a stoichiometric simulated 980 "C calcination ( 4h ). exhaust after this high temperature calcination. In all cases the catalyst prepared with the stabilized ceria showed lower conversion levels. Activity comparisons of these two catalysts were also made over a wide range of aidfuel stoichiometries typical of the range found under normal vehicle operations and again in all cases the stabilized ceria washcoat exhibited performance below that observed for the unstabilized ceria washcoat. Based on this result a more detailed study was undertaken to investigate stabilized ceria performance in high temperature environments after both lean and stoichiometric aging schedules. In these experiments both lanthana~
130
stabilized ceria and neodymia-stabilized ceria additives were compared against the commercial, high purity ceria-containing standard catalyst. In the case of stabilized ceria-containing catalysts, several levels of lanthana and neodymia stabilizers were evaluated. In each case the stabilizer was added to the high purity ceria by impregnation using an appropriate aqueous nitrate solution. Although not detailed here, this stabilizer impregnation scheme resulted in similar improvements in ceria surface area stability in high temperature calcinations as reported for the co-precipitation scheme. Figure 3 compares CO conversion levels in a stoichiometric simulated exhaust of the stabilized and unstabilized ceria-containing% CO Conversion at Stoich. Conditions catalysts after a four hour calcination ' O 8 1 at 980 "C. Both the lanthanastabilized and neodymia-stabilized ceria catalysts show performance Activity measured at 480% drop-offs with increasing levels of 60 stabilizer, confirming the poor Nd-Ceria Promoter performance observed for the coprecipitated L a 2 0 3/CeO2. CO 50 conversion levels for catalysts with about 10 mole % ceria stabilizers 40 were only about one-half those of the standard catalyst with unstabilized La-Ceria Promoter ceria. Conversion levels for hydrocarbons and NO, showed similar trends to the CO data of 20 8 IO 12 Figure 3. Mole % La or Nd A similar comparison was made using lanthana and neodymiaFigure 3 : co activity ComParisonfor stabilized cerias after heat treatments PtIRH washcoats after 980°C in a stoichiometric gas mixture calcination ( 4 h . ) . containing CO, 02, N2 and H20.
In these experiments, catalyst samples were treated for 95 hours at 960 "C in this flowing gas mixture and then evaluated for three-way activity. Figure 4 compares stoichiometric conversion levels for stabilized and unstabilized ceria-containing catalysts after this stoichiometric high temperature treatment. Unlike the air calcined samples, stoichiometrically treated, stabilized ceria-containing catalysts performed on a par with their unstabilized counterparts.
131
YO Conversion at Stoich. Conditions
80 I
Activity measured at 480°C
60
40
20
0
co
Hydrocarbons
=
Unstabilized c e r i a
: ,L
1 La-stabilized
ceria ( 6 mole % L a )
N Ox N d -s t a b i Ii ze d c e r Ia (5 mole Yo N d )
Fig.4: Three-way activity comparisonfor PtlRh washcoats afer 960 “C stoichiometric aging ( 9Sh ).
980 960 940 920 900 880 860 840 820 800 780 Binding Energy (eV)
Fig 5: Ce 3d and La 3d XPS signals for La-stabilized ceria (ca.10 mole% La)
132
XPS CHARACTERIS ATION In an effort to explain the poor three-way performance of the lanthanaand neodymia-stabilized ceria-containing catalysts after high temperature calcinations, these additives were characterized by XPS for relative surface compositions. The stabilized cerias containing approximately 10 mole % La or Nd (made by impregnation of the commercial ceria) and the high purity, commercial ceria were analyzed as-prepared (following calcination for 1 hr at 540 "C) and after a four hour calcination at 980 "C. Cer ia- 10% - Neod y mia h
v) c .-
C
3
2
v
x c .-
cn
K a, C
c
K
.-0 v) v)
._
5
0 0
Ce 3d
c
r
a
hv
1253.6 eV
= I
,
I
,
I
,
I
,
I
.
I
,
I
,
I
~
I
.
1030 1014 998 982 966 950 934 918 902 886 870 Binding Energy (eV)
Fig. 6: Ce 3d and Nd 3d XPS signals for Nd-stabilized ceria (ca.lOmole% Nd) Figures 5 and 6 compare signals in the 3d region for the lanthana- and neodymia-stabilized cerias, respectively. Both the stabilized cerias show significant surface enrichment of the stabilizer oxide relative to Ce02 after the 980 "C calcination. In both cases the Ce 3d signals have decreased and the La or Nd 3d signals have increased relative to the as-prepared oxides after the severe thermal treatment in air. An even more striking indication of this surface enrichment is shown in Figure 7a. In this figure the 0 1s signal for the lanthana-stabilized ceria is compared for the as-prepared oxide and after calcination at 980 "C. After the high temperature calcination the oxide exhibits a significant enlargement of the higher binding energy oxygen signal centered around 531.5 eV. A similar behavior was observed for the neodymia-stabilized ceria.
133
Ceria-10%-Lanthana
0 Is hv = 1253.6 eV
0 538 536 534 532 530 528 526 524 522 520 Binding Energy (eV)
Fig. 7a: 0 I s XPS signals for La-stabilized ceria (ca. 10 mole% La ).
Ceria
0 1s hv = 1253.6 eV
I
,
I
,
I
,
I
,
I
.
I
,
I
,
I
.
I
,
0 538 536 534 532 530 528 526 524 522 5: 0
Binding Energy (eV)
Fig. 7b: 0 I s XPS signals for a commercial, high purity ceria.
134
Figure 7b shows 0 1s signals for the commercial ceria after the same two calcination treatments. In contrast to the stabilized ceria data, these spectra do not show an increase in the higher binding energy oxygen signal. This higher binding energy oxygen signal is consistent with the more reactive nature of lanthana and neodymia than ceria towards C02. Both lanthana and neodymia readily form surface carbonates in the presence of C 0 2 181 and the higher binding energy oxygen signal is indicative of a surface carbonate species. As the ceria surface becomes enriched with either lanthana or neodymia, these carbonates become more abundant in the presence of C02, and the higher binding energy oxygen signal increases.
0
Binding Energy (eV)
Fig. 8: 0 I s X P S signals for La203 and h2(co3)3reagents. Confirmation of this carbonate assignment comes from XPS scans of reagent La2O3, reagent La2(C03)3, and La203 heated to 900 "C in low-CO2 air and transferred to the spectrometer under nitrogen. 0 1s signals for these materials are summarized in Figure 8. The oxygen signals of the two reagents are nearly identical with a peak at about 531 eV (similar location as the higher binding energy oxygen signal in Figure 7). This suggests that the lanthana is covered with a carbonate overlayer. The heated lanthana, on the other hand, shows a reduced carbonate band at 531.5 eV due to the decomposition of the carbonate overlayer and a lower binding energy oxygen signal that can be assigned to oxidic surface oxygen.
135
D IS c US SIO N The results of this investigation indicate that additives such as La, Nd, Y and to some extent A1 are capable of retarding ceria surface area loss after high temperature oxidations. The mechanism of this stabilization was not specifically addressed by this study but some inferences can be made about the ceria stabilization process. In the case of co-precipitated, lanthana-stabilized ceria, surface enrichment of the lanthana was observed by XPS even after drying and calcining the mixed hydroxides at mild temperatures (540 "C). La/Ce surface atomic ratios obtained from XPS for Ce02 containing 10 mole % La were about 70% higher than the levels associated with the bulk composition of this material. The La/Ce surface ratios were even larger for samples made by impregnation. TABLE 2 - Bulk and Surface Atomic Ratios for La-Stabilized Ceria
Co-precipitated La203/Ce02 Impregnated La203/Ce02
La/Ce, Surface
La/Ce, Surface (after 980 "C)
0.1 1
0.19
0.25
0.12
0.25
0.71
Table 2 summarizes La/Ce ratios for these materials. This lanthanarich surface phase may be associated with ceria grain boundaries or reactive ceria surface sites that initiate the sintering process. The lack of any discernable La-containing phases by X-ray diffraction before or after calcination to 980 "C is also consistent with a highly dispersed stabilizer phase. The presence of the lanthana-rich surface phase is then postulated to impede the growth of ceria particles at high temperatures. Similar surfacestabilizing phases such as LaA103 have been characterized as being responsible for the stabilizing effect of La on aluminum oxide [7,8]. Although effective at retarding ceria surface area loss, the rare earth stabilizers appear to drastically alter the promoter action of ceria in threeway catalysts following severe oxidations such as the 980 "C calcinations used in this study. The poor activity observed for the stabilized ceria-containing catalysts after high temperature calcinations appears directly related to the significant enrichment of non-cerium oxides on the surface of the bulk ceria additives that also occurs under these conditions. As shown by the XPS results, the resulting stabilized ceria surface becomes more like neodymia or lanthana in composition and reactivity. For comparison, a catalyst sample was prepared with a pure, low surface area lanthana in place of the ceria or stabilized ceria (all other catalyst details kept constant, including noble metal
136
content and total rare earth loadings). After a similar four hour calcination at 980 "C the lanthana-only promoted catalyst had almost no measurable conversion (<5%) for hydrocarbons, CO, or NO, under our standard activity conditions. Both this result for a pure lanthana promoter, and the results of Figure 3 that show a clear correlation between reduced activity and increasing levels of stabilizer, are consistent with the loss of ceria promoter action with increasing surface concentrations of the non-ceria stabilizer. A likely contribution to the poor promoter action of lanthana- or neodymia-enriched cerias is the fact that both these oxide surfaces are most likely covered with an unreactive carbonate overlayer, similar to those observed in Figure 8, given the large C02 partial pressure found in the exhaust stream. Results reported by Miyoshi et al. [ 131 and Ihara et al. [ 141 also discuss the impact of lanthanum and an unnamed rare earth additive, respectively, on three-way catalyst performance. Both these studies claim La or other rare earth additions to both the alumina support and ceria are beneficial in improving catalyst thermal stability. Besides retarding surface area loss of alumina and ceria, evidence is presented that supports enhanced oxygen mobility for La203/Ce02 promoters and improved Rh activity with La additions. Activity comparisons were made after aging catalysts in a stoichiometric engine exhaust in the case of Miyoshi et al. and after air oxidations in the case of Ihara et al. In both cases these studies showed benefits for samples containing La or the other unidentified rare earth additives. Catalyst preparation details are very sketchy in these studies and therefore a detailed comparison of the catalyst compositions with the formulations used here cannot be made. The present study focused only on rare earth additions to bulk ceria. The results of the two Japanese studies contain comparisons between catalysts that contain no rare earth stabilizers and catalysts with rare earth stabilizers added to both the alumina and the ceria by unspecified means. As such these results would appear to contain much broader effects than those discussed here. Both temperature and excess oxygen are important in facilitating the surface enrichment discussed above. This is evident from the fact that both fresh activity after a mild calcination (540 "C) and activity after high temperature treatments (960 "C) in a stoichiometric environment were equivalent for stabilized and unstabilized ceria-containing catalysts. The activity debit observed for stabilized cerias after a high temperature calcination was found to be reversible with reductions in hydrogen at moderate temperatures (500 "C). After a hydrogen reduction, stabilized and unstabilized ceria-containing catalysts showed higher and nearly identical three-way activities than the same samples calcined at 980 "C and tested under the standard conditions. Subsequent reoxidation of the reduced samples in air at 980 "C resulted in the large activity debit for the stabilized ceria-containing catalysts detailed previously.
137
100
% CO Conversion at Stoich. Conditions Activity measured at 480%
80
60
40
20
0 98OoC Oxidation ( 4 h in air) Unstabilized ceria
500°C Reduction (2 h in 5% H,/N,)
i zLa-stabilized ceria (6 mole 7' L a )
98OoC Reoxidation ( 4 h in air) Nd-stabilized ceria ( 5 mole % N d )
Fig. 9: CO activity comparisonfor PtlRh washcoats following oxidation and reduction treatments. Figure 9 details this reversibility with respect to CO performance for a catalyst containing ceria stabilized with 5-6 mole % La or Nd. Similar trends were observed for hydrocarbon and NO, performance in oxidation and reduction cycles. This reversibility implies that the stabilized ceria surface has been returned to a more ceria-like surface by the reduction. XPS experiments are in progress to further characterize stabilized ceria surfaces in reducing atmospheres to compliment the results presented here. CONCLUSIONS
1. Surface area loss of cerium in high temperature, oxygen-rich atmospheres can be retarded by the use of stabilizers such as La, Nd, or Y . These stabilizers can be incorporated by co-precipitation or impregnation. 2. The stabilization mechanism appears to involve formation of stabilizerrich surface phases that impede ceria crystallite growth in severe oxidizing environments. 3. The use of stabilized cerias containing lanthana or neodymia results in poor three-way activity relative to unstabilized ceria after high temperature oxidations.
138
This poor three-way performance is the result of significant enrichment of the stabilized ceria surface with lanthana or neodymia in high temperature, oxygen-rich atmospheres. The resulting promoter surface does not retain the same promoter action as Ce02. 5. The enrichment of the ceria surface is facilitated by high temperatures and excess oxygen. The resulting activity debit can be reversed by reduction in H2. The implications of these results for stabilized ceria applications in catalytic washcoats appear mixed. Although successful at reducing surface area loss in excess oxygen atmospheres, the resulting activity penalty observed here would present a serious disadvantage. This penalty might be expected to be short-lived in real service, however, due to the reversibility in performance reported here after reduction. The activity debit observed here may also only be associated with mixed rare earth promoters. Additional studies are required to determine catalyst performance over a wider range of conditions than detailed here, and the effect of non-rare earth stabilizers such as alumina. In this way the relative merits of stabilized ceria use in three-way catalysts may be better assessed.
4.
ACKNOWLEDGEMENTS
The authors would like to thank Nicole Ashley and Keith Halle for their assistance in sample preparation and testing, and W. R. Grace & Co.-Conn. for their permission to publish these results. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
R.H. Hammerle and C.H. Wu, SAE Paper No. 840549, 1984. L.L. Carol, N.E. Newman and G.S. Mann, SAE Paper No. 892040, 1989. L. Loewendahl and J.-E. Otterstedt, Appl. Catal., %,89, (1990) B. Stenbom, G. Smedler, P.H. Nilsson, S. Lundgren and G. Wirmark, SAE Paper No. 900273, 1990. H. Schaper, E.B.M. Doesburg and L.L. van Reijen, Appl. Catal., 211, (1983) M. Machida, K. Eguchi and H. Arai, J. Catal., 385, (1987). F. Oudet, P. Courtine and A. Vejux, J. Catal., 114,112, (1988). M. Bettman, R.E. Chase, K. Otto and W.H. Weber, J. Catal., 117,447, (1989) C.-Z. Wan and J.C. Dettling, U.S. Patent No. 4,717,694, 1987. J.-Y. Chane-Ching and J.-Y. Dumousseau, U.S. Patent No. 4,661,330, 1987. C. David, C. Magnier and B. Latourrette, U.S. Patent No. 4,859,432, 1989. T. Ohata, S. Terui and E. Shiraishi,U.S. Patent No. 4,708,946, 1987. N. Miyoshi, S. Matsumoto, M. Ozawa and M. Kimura, SAE Paper No. 891970, 1989. K. Ihara, K. Ohkubo and Y. Niura, Proceedings of the 4th Int. Pacific Conf. on Automotive Eng., PaDer No. 871 192, Melbourne, Australia, 1987. M.V. Ernest an; G. Kim, SAE Paper No. 800083, 1980.
m,
z,
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
139
THE ROLE OF CERIA IN THREE-WAY CATALYSTS A.F.Diwell, R.R.Rajaram, H.A.Shaw, T.J.Truex Johnson Matthey Technology Centre, BlountS Court Sonning Common, Reading RG4 9NH, UK
ABSTRACT The role of ceria in three-way catalysts (TWC) for the control of gaseous exhaust emissions has variously been described as: (a) oxygen storage under transient conditions; (b) catalytic promoter of precious metals for certain reactions such as water gas shift; (c) structural promoter for the stabilisation of precious metals and alumina against particle growth. However, it would appear that the interaction of ceria with the precious metal can significantly affect activity depending on pretreatment conditions and the precious metal involved. In this paper, we shall consider the role of the individual precious metals in TWCs and how their interaction with ceria affects performance. Catalysts containing Pt or Rh, with or without ceria, dispersed on alumina, have been prepared and pretreated in different atmospheres before characterisation and activity measurement in a simulated exhaust environment. Under oxidising conditions, a Pt-ceria complex is formed which maintains F’t stability against sintering. The complex fixes Pt in a high oxidation state and produces sites of lower activity than on alumina under lean conditions. However, mild reduction of a ceria-containing Pt catalyst produces sites which are highly active for the removal of CO and NO. Such sites are sufficiently stable to allow characterisationby XPS, TEM etc. but, following mild oxidation treatment, the high activity is destroyed. It is believed that the active sites are formed by surface Pt crystallites at the metal-support interface, inducing changes in the surface properties of ceria through the formation and oxidation of anionic vacancies. The presence of ceria also modifies the activity of Rh-containing catalysts by promoting the reduction of Rh.
1. INTRODUCTION Ceria has been traditionally applied as an additive in automotive catalysts because of its abilities to store oxygen [l-41 and to improve dispersion of noble metals [3,5,6].Its role as an oxygen storage component is manifested in the ability of ceria-containing catalysts to store oxygen under lean operating conditions - thus promoting conversion of oxides of nitrogen - and release it under rich conditions by reaction with carbon monoxide, hydrogen or hydrocarbons. Ceria has also been noted to promote a number of catalytic reactions on noble metal particles, including the water gas shift reaction [7,8].
140
It has gradually become recognised that cerium oxides and the noble metals are subject to mutual interaction, which is a function of the particular noble metal, ageing temperature and gaseous environment [9]. An interesting example of this interaction is the strong enhancement in catalytic activity of PtCe02 catalysts after reduction [2,10]. The exact nature of this interaction is not well understood, although several hypotheses such as: (a) formation of platinum-Ce alloys [11,12]; (b) promotion of redox reactions over ceria [2]; and (c) the involvement of anionic vacancies in ceria, either for stabilising Pt dispersion [ 131 or as partial catalytic sites [ 141, have been postulated. The purpose of this paper is to review the traditionally perceived role of ceria in autocatalysts and discuss, more specifically, how noble metal-ceria interactions can influence catalyst performance. 2. EXPERIMENTAL
2.1 Materials Catalysts were prepared by impregnation using standard platinum and rhodium precursors. The supports used were y-Al203, Ce02 or 75 wt % A1203 - 25 wt % Ce02. 2.2 CO Chemisomtion CO chemisorption was performed on Pt catalysts in a pulse injection system using a thermal conductivity detector. The samples were reduced at 300 "C for 30 minutes and cooled to room temperature before injection of pure CO in the H2 gas stream. From the amount of CO consumed, COP t values were calculated as a measure of the average Pt dispersion of the catalyst samples, assuming a C0:Pt stoichiometry of 1:l. CO chemisorption on the reduced ceria surface was found to be negligible. 2.3 Temuerature Programmed Reduction Temperature programmed reductions (TPR) were performed in a conventional apparatus (10% H2/N2. flow rate 40 cm3 min-1). The sample was heated to 900 "C at 10 deg min-1 in the reactive gas mixture. 2.4 XPS Analysis The XPS data were obtained on a Kratos XSAM 800 using A1 K alpha radiation at 65W and 40eV pass energy. The XPS binding energies were referenced to the adventitious C 1s level at 284.8 eV.
141
2.5 Activitv Measurement Catalysts were evaluated by studying their light-off behaviour in simulated exhaust gas mixtures, with a constant flow rate of the model gas, from 150 "C to 600 "C at a heating rate of 5 deg min-1 and a GHSV of 50,000. The conversion efficiencies of the catalyst were measured by continuously monitoring the outlet gases after the catalyst. The composition of the simulated exhaust gas to achieve either a lean mixture ( h = 1.02) or a rich mixture ( h = 0.98) is shown in Table 1.
TABLE 1. Components of simulated exhaust gas mixture in activity tests (Balance N2) Units % %
I h = 1.02
h = 0.98
15 9.2 1500 20 370 123 123 0.23 0.97 0.70
15 9.2 1500 20 670 223 223 0.43 0.72 1.30
PPm PPm PPm PPm PPm % % %
3. RESULTS AND DISCUSSION
3.1 Stabilisation of Pt Dispersion bv Ceria The presence of ceria in a TWC formulation is known to stabilise Pt dispersion [3,5,6]. The effect is illustrated in F i g . ] , where percent Pt dispersion is plotted versus ageing temperature (air for 2h) for 0.9 wt % pt on A1203 and 0.9 wt % Pt on A1203-Ce02. In the case of the A1203 support, substantial sintering of Pt occurs at temperatures greater than 600" C. Addition of Ce02 results in a reasonable stabilisation of the Pt dispersion up to temperatures of 700 "C - 800 "C, indicating significant Pt-Ce02 interaction, most probably through the formation of a surface complex involving Pt2+-02-Ce4+ [15,16]. The structural stabilisation of Pt by Ce02 is due to a stronger Pt-CeO2 interaction than that which exists between Pt and Al2O3. However, the
142
reduction of the platinum component of the Pt-Ce02 complex is more facile than that of platinum oxide dispersed on Al2O3, as indicated by the TPR profiles of 0.9 wt % Pt/A1203 and 0.9 wt % Pt/A1203-Ce02 samples calcined at 500 "C (Fig.2.) The reduction peak at 200 "C for the Pt/A1203-Ce02 sample is associated mostly with the surface reduction of Ce4+ with a contribution from the platinum (shoulder at c. 100 "C). The main Pt reduction peak for 0.9 wt % Pt/Al2O3 has a maximum at 220 "C and is comparatively broad.
Figure 1
Figure 2
Effect of Ce02 on Dispersion Stability Of WA1203 Temperature Programmed Reduction Traces for
0.9wt% Pt/A1203and 0.9wi% Pt/AI2O3-CeO2Catalysts
iI 50
b
100
zoo
500
400
600
1EMPERATURE't
3.2 Effect of Pt-Ce02 Interaction on Lean Activitv ( h = 1.02) The effect of these Pt-support interactions on CO and C3H8 oxidation activity under lean conditions is shown in F i g s . 3 - 4 , where turnover frequencies (ToF s-1) are plotted versus fraction of exposed metal for 0.9 wt% Pt/A1203 and 0.9 wt % PtIA1203-Ce02. The results show that the Pt on A1203 catalyst has much higher activity than the Pt on A1203-Ce02 catalyst. In addition, the ToF of Pt on A1203 increases significantly with decreasing fraction of metal exposed (increasing particle size), such that the atomic rate (rate per mass of Pt) is approximately constant as a function of dispersion. The atomic rate for the Pt on A1203-Ce02 decreases with decreasing dispersion. For catalysts containing the same mean Pt particle size, the specific activity is higher on A1203 than A1203-Ce02.
143
Figure 3
Figure 4
Turnover Frequency vs Fraction Exposed for CO Oxidation on Platinum
Turnover Frequency vs Fraction Exposed for C3Hs Oxidation on Platinum
8
I 0.01 0.01
.
p"*44-?
.\
I
0400.10 0 0 4 1 . 0 FAACTION EXPOSED
Kinetic studies of alkane [17-191 and CO [20] oxidation on Pt/A1203 have shown that, under conditions of excess 02, these reactions are demanding. The structure sensitivity of these reactions has been explained by changes in the Pts-0 bond strength [19], (small Pt particles have higher tenacity for 0 and are less active). The interpretation of the results is that, while Ce02 interaction with Pt aids dispersion, it does so by maintaining Pt in a higher oxidation state of lower activity. 3.3 Effect of CeO7 - on Rh Stabilisation, Reducibility and Activitv The thermal deactivation of Rh when dispersed on y-Al2O3 has been widely reported. The cause of this deactivation has been associated with theformation of a surface or subsurface spinel [21] or detrimental Rh-yA1203 interaction at grain boundaries [22]. The incorporation of Ce02 into a threeway catalyst decreases this detrimental Rh-Al203 interaction with substantial improvement in thermal stability and reducibility of Rh [23]. The beneficial effect of Ce02 on Rh activity is demonstrated in F i g 3 in the form of light-off curves for NOx conversion at h = 0.98 for 0.18 wt % Rh/y-A1203 and 0.18 wt % Rh/y-A1203-Ce02 catalysts which had been calcined in air at 800" C. A simiEar effect for CO oxidation under lean conditions is noted Fig.6. The detrimental Rh-y-Al203 interaction affects
144
especially the low temperature activity of the catalyst. The incorporation of Ce02 in the catalyst maintains the same low temperature activity even after thermal ageing. Figure 6 Figure 5 EHect of Ce02 on Rh for NOx Conversion at h- 0 98
Effect of CeOp on Rh for CO Conversion atA= 1.02
--- 0 1 M W l RhlALp-C.% 20
INLET TEYPEAATURE
%
0 l 1 r I S Rhl-
INLET TEMPERATURE
%
3.4 Role of Ceria as a Catalytic Promoter for Water Gas Shift and Steam Reforming Reactions The incorporation of ceria in TWCs has been noted to result in an improvement in CO performance at and rich of stoichiometry. This was originally ascribed to an oxygen storage and release mechanism as outlined below [7].
Lean
1/2 0 2 + Ce203
+ 2Ce02
Quantitative laboratory experiments subsequently showed that this simple mechanism could not account for the degree of enhancement in CO performance and that a major role of ceria was to enhance the water gas shift activity of Pt-Rh three-way catalysts [8]. The origin of this improved CO performance is demonstrated in Table 2. CO conversion was measured under rich spike conditions with and without water present. 1.08 wt% Pt-Rh [5,1] catalysts with 1.5, 4.0 and 8.0 wt % cerium levels were tested after a 900" C hydrothermal ageing treatment in 1% 02/10% H20/89% N2 for 4 hours. Cerium loading does not affect CO performance when H20 is absent from the feed gas stream.
145
When H20 is present, CO conversion is higher and increases with increasing cerium loading.
1.5 wt 96 Ce 4.0 wt % Ce 8.0 wt % Ce
Rich spike CO Conversion (%) with H2O 54 64 70
Rich spike CO Conversion (%) without H 2 0 49 49 49
The results are consistent with a PGM/Ce02 interaction resulting in enhanced water gas shift activity. Steam reforming of hydrocarbons, to produce CO and H2, provides an attractive route to curtail HC emissions under oxygen deficient conditions. The CO produced could be further removed by the water gas shift reaction. Promotion of the steam reforming reaction has been noted by the inclusion of ceria in TWC containing Pt-Rh, even in the presence of S02. The effect is demonstrated in Table 3, which gives propane conversion on Pt-Rh/A1203 catalysts, with and without ceria, and in the presence or absence of water. SO2 (20ppm) was present in the feed. When H20 is present, propane conversion is higher on the ceria-containing catalyst. The steam reforming reaction, however, is severely depressed by the presence of SO2 in the exhaust mixture at temperatures greater than 300" and under steady state rich conditions. The effect is illustrated in Fig.7,which gives the light-off curves for propane and hydrogen at h = 0.98 in the presence and absence of S02. Propane conversion is higher in the absence of S02, whilst the reverse applies for H2 conversion. The inhibition effect has been associated with the formation of reduced sulphur species which strongly adsorb on the active sites and has been observed [24] to be less severe under cyclic rich-lean conditions.
TABLE 3 The Influence of H 2 0 on Propane Conversion on Ceria and Non-Ceria Containing Catalysts (SO2 present) C3H8 conversion (%) at 350°C h = 0.98 with H 2 0 without H 2 0 Pt-Rh/y-A1203 Pt-Rh/y-A1203+Ce02
12 35
10 18
146
Figure 7 Effect of SO2 on Steam Reforming Reaction for P1-Rh/Al2O3-CeO2 Lbht-oll .I Lambdo 0.98
100,
1
90
10
* f
'O'
00 60
%
40
30
ao
K/
lo
'$60
200
260
300
380
400
4
INLET TEYPERITURE
%
3.5 Enhanced Pt-CeO7-Interaction Following Reduction Yu Yao [2,10] has reported the effects of in-situ reduction of Ce02A1203 supported Pt, Pd and Rh catalysts on CO and HC oxidation. All three catalyst systems exhibited enhanced activity after pretreatment in -1 % CO at temperatures greater than or equal to 300 O C. The effects are reported to be larger for CO than alkane oxidation and are transient in nature with a return back to the original activity state after exposure to oxidizing conditions. The short-term, transient nature of these effects with Pd and Rh catalysts precluded characterization or detailed kinetic studies. However, the high activity reduced Pt-Ce02-Al203 state was stable enough for kinetic measurements. A 1 to 2 order-of-magnitude increase in reaction rate for CO oxidation was observed after the reducing treatment. Based upon analysis of the kinetic parameters and comparisons with the kinetics of CO oxidation over Pt wire and Pt-Al203, which did not show a similar enhancement after a reducing treatment, Yu Yao concluded that the enhanced activity was due to a simple reduction of the Pt to a more metallic state on the CeOpA1203 support. This phenomenon has been investigated further in our laboratories. Specifically, three-way catalyst activities measured slightly rich of stoichiometry ( h = 0.98) and characterization data are presented for Pt-Ce02 catalysts after air calcination and H2 reduction treatments. Catalyst activity data measured at h = 0.98 for Pt-CeO2 catalysts after 500 "C calcination in air and subsequent reductions in 1% H2 at 900 "C and 600 "C are presented in Fig 8,9,10 . The reaction gas mixture was that shown in Table 1 which included SO2 in the gas stream. Fig. 9 and 10 show that prereduction of the Pt-Ce02 catalyst results in a dramatic improvement in CO, NO and propene conversion at low temperatures. The pre-reduction results in
147
an associated drop in propane and methane activities. Comparison of results in Fig.9 (900 "C reduction) and Fig. 10 (600 "C reduction) show that the effects are larger at the lower reduction temperature. It is interesting to note that the major activity changes are only observed at temperatures c c.400 "C. The high activity state decays slowly with time at reaction conditions and reverts back to the initial state (observed after the original 500 "C calcination in air) when the prereduced catalyst is subsequently aged in air at >500 "C as is shown in Fin.1 - I. These dramatic changes in low temperature activity are not entirely consistent with those previously observed by Yu Yao [2,10]. It may be the case that our use of H2 as a reducing agent results in an enhanced or different activation than that reported previously after CO reduction. The low temperature CO and NO activity combined with low alkane activity do not seem consistent with the simple reduction of Pt to a more metallic state. In particular, Pt is known to exhibit generally poor (relative to Rh) NO activity [25,26] which has been associated with CO inhibition [26,27] and SO2 poisoning [26]. The activity results presented here imply an induced Pt-Ce02 interaction which greatly affects the nature and activity of the catalytic sites. Table 4 summarizes the BET surface area and CO chemisorption (expressed as CO/Pt ratio) data for the Pt-Ce02 catalysts after various pretreatments. The initially high surface area Ce02 results in well dispersed Pt (0.71 COPt) after the initial 500 "C calcination in air. Upon reduction in H2 there is a decrease in BET surface area and CO uptake. Subsequent treatment of the reduced sample in air at 500 "C has minimal effect on the BET surface area but increases the CO uptake.
TABLE 4
- BET and CO Chemisorption
Pretreatment Air 500 "C Air 500 "C 4 H2 600 "C Air 500 "C 4 H2 900 "C Air 500 "C + H7 600 "C + Air 500 "C
I
of 0.9 wt 9% Pt/Ce02
I BET SA m2g-1 I
125 75 7 73
I
Copt 0.7 1 0.21
::"0
I
Recent TEM and preliminary EXAFS [28] analyses indicate that the reduced Pt-Ce02 catalyst still contains small Pt particles (cl0nm) and therefore the observed decrease in CO uptake cannot be accounted for by a Pt sintering process and associated growth in Pt particle size. An alternative explanation is that the changes in CO chemisorption result from an induced metal-support interaction. TPR results appear to confirm a significant change in Pt-Ce02 interaction associated with the H2 reduction treatments. Fig. 1 2 presents TPR profiles for the Pt-Ce02 catalysts after
Figure 8
Figure 9
Light-off curve atA= 0.98 for Various Gaseous Components on O.W%WCeO, Calcined in Air at sooOC 100.
00'
-Hz
t 00
........ ............
.-.co
-
Light-off curve atA= 0.98 for Various Gaseous Components on 0.9wt% WCe02 Calcined in Air at 5OOOC then Reduced in Hydrogen at 9oooC
....................
00
- - - NO.
10
1
,,,-' r.7
I'
40. 30'
I'
I'
'0
0160
INLET TEMPERATURE %
I'
........ .............. 260
200
------
.-
._._.-.-.
100
300- 9 6 0 460 460 INLET TEMPERATURE %
I
cI(,
..... w*
,,-I
80
660
Figure 11
Figure 10 Light-off curve at& 0.98for Various Gaseous Components on 0.9wt% PVCe02Calcined in Air at 5OOOC then Reduced in Hydrogen at 6OOOC 100
1 -n. I
.....
..
;
Light-off curve atA= 0.98 for Various Gaseous Components on O . M % WCeO, Calcined in Air at 5OOOC then Reduced in Hydrogen at 6oooc then re-calcined in air
:
:
'
..y.........;............ /-
10
_.',
1 'O
. . L _ .
.....,,' ,/;/
/@ 400 450 600 INLET TEMPERATURE %
co
--
I'
/'
20
I'
-HZ
---NO.
560
100 '!SO
200
26;
300
350
400
460
INLET TEMPERATURE %
600
660
16
149
initial air calcination, subsequent high temperature reduction with exposure to air at room temperature, and after a 500 "C air calcination of the reduced sample. The two samples which have Figure 12 received the air calcination prior to running the TPR show similar Temperature Programmed Reduction Traces features. A peak centred at -230 "C for O.W% Pt/Ce02 Catalysts is assigned to reduction of Pt and surface oxygen shared with cerium. A shoulder observed at -350 "C appears to be associated with surface CeO2 not in close proximity to the Pt, and the peak + centred at -800 "C is assigned to fa 6 ' reduction of bulk Ce02. The increase in the ratio of the bulk to surface peak ratio featured in the sample calcined after the high temperature reduction is consistent . . 0 (00 200 300 400 100 800 7 0 0 800 900 with the loss of BET surface area, TEMPERANRE % observed after the high temperature treatment.
.. ..
c.*M n r.blSM t M rr.*M in
n
The TPR of this sample first indicates that a reoxidation has taken place upon room temperature exposure to air since overall H2 consumption is similar to that for the calcined samples. The most interesting feature of the TPR profile of the prereduced sample is the existence of a low temperature reduction peak at -50°C that is not observed in the calcined samples. Although this reduction peak cannot definitively be assigned, it is speculated that it arises from a strong Pt-CeO2 interaction, since no similar low temperature peaks are observed either for Pt/A1203 or for ceria alone following the same pretreatment sequence. XPS results after initial air calcination and subsequent high temperature reduction are summarised in Table 5. After calcination in air, Pt is identified as being present as Pt2+; whereas, in the reduced sample, only Pt is observed. Cerium is present in the plus four oxidation state for both the air-calcined and reduced samples although, in the latter case, ambient oxidation in air has been shown to occur. The above results cannot consistently be explained by a simple increase in the metallic character of Pt, which was proposed to explain the enhanced activity of Pt-Ce02-Al203 catalysts after a milder reduction with CO [2,10]. Indeed, it would appear that the effects reported in this present paper may be
150
different in degree and/or nature from those previous studies due to the more severe reducing treatments used.
TABLE 5. Results of XPS Studies on Pt/CeOz
Air 500 "C Air 500 "C -+ H2 900 "C
Pt I1 0
Ce
Iv
IV*
* after ambient oxidation in air At present, we do not have a single model which will explain all of the results obtained, although the following appears to account for most of the observations. It is proposed that ionised vacancies are the active sites for NO reduction and the re-oxidised vacancies are active for CO and H2 oxidation. Lattice vacancies are formed during the reduction of surface lattice anions via the formation of hydroxyl groups. Subsequent ionization and stabilization occurs by electron transfer to the metal. A partial encapsulation of the metal by the support or partially reduced oxide could account for the low CO chemisorption [30,34]. This is consistent with the fact that only small metal crystallites are observed by TEM. Pt promotes the formation of the anion vacancies but may not be the active sites because increased electron density at the metal centre would promote stronger CO adsorption and inhibition of low temperature activity. It has been proposed [3 11 that similar anionic vacancies stabilized by electron transfer to metal crystallites may be the active sites in methanol synthesis. The reduction treatment may be inducing a Pt-CeO2 interaction causing an enhancement of the redox reaction on ceria oxidation of ionized vacancies by NO (or H 2 0 or 0 2 ) and reduction of oxidized vacancies by CO - with the metal particles acting as an electron donor and sink. The lower alkane activity would result from the encapsulation of the metal particles. At present, however, such a model does not appear to account for the increased low temperature alkene activity. It must be emphasized that, at present, the above represents a working hypothesis to help explain the experimental results, and that further work is in progress to assess its validity. Other models which can partially explain some of the observations include: (a) the formation of Pt particles with cationic character, arising from their interaction with oxygen ion lattice vacancies [13,32], or at the metal-metal oxide interface [ 3 3 ] ; and (b) active sites comprised of surface Pt atoms situated at the metal-support interface close to anionic vacancies in the support. The latter has previously been used to explain
151
CO-NO reaction over Rh/Ce02 [23] and enhanced CO-H2 reaction over Ni/CeOz[ 141. It is difficult to assess the extent to which the low temperature active state observed here is contributing to actual three-way catalyst performance on a vehicle. These laboratory experiments were designed to enhance the transitory condition observed so that it could be studied in detail and, as a result, the pretreatment conditions used are relatively unrealistic. However, it does seem feasible that, after high temperature rich operation on a vehicle, at least some of the effects observed in these laboratory experiments would be found. 4. CONCLUDING REMARKS
It has been demonstrated that ceria plays a wide variety of roles within three-way catalysts. It stabilises Pt dispersion by maintaining the metal in an oxidised state. This has an adverse effect on catalytic activity under oxidizing conditions. However, under reducing conditions, ceria promotes the reduction both of platinum and rhodium to the metals, and this leads to enhanced activity for water gas shift, steam reforming and NO reduction. The high activity state which has previously been reported to result from in situ reduction of Pt supported on Ce02-Al203 has been studied in more detail. Pt/Ce02 catalysts have been pre-reduced in hydrogen and tested and characterised. Dramatic improvements in the low temperature conversion of CO, NO and alkene (propene) have been observed, while alkane (methane and propane) activity is adversely affected. However, after subsequent ageing in air, this enhanced low temperature activity is lost. Although, at present, no single model explains all of our observations, the following appears to account for most. It is proposed that, during the reduction process, a strong Pt-Ce02 interaction occurs involving the encapsulation of Pt by the partially reduced oxide. Ionised vacancies in the Ce02 lattice are formed and these are the active sites for NO reduction. We must emphasize that the above is a working hypothesis and that further research is in progress to assess its validity. It is difficult to assess the likely influence of the low temperature active state on actual TWC performance on a vehicle. However, high temperature rich operation may induce its partial formation.
152
ACKNOWLEDGEMENTS The authors would like to acknowledge the contribution of Mr. G.P. Ansell, Mr. C.M. Brown, Dr. J.A. Busby and Mr. J.F. Pignon at Johnson Matthey Technology Centre. Thanks for their helpful comments are also due to Dr. S.J. Roth, Mr. R.A. Searles, Dr. D.E. Webster and Dr. M. Wyatt of Johnson Matthey Catalytic Systems Division.
REFERENCES 1.
2.
3. 4.
5. 6. 7.
8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
26. 27. 28. 29. 30.
31. 32. 33. 34.
Gandhi H.S., Piken A.G., Shelef M. and Delosh R.G., S A E Paper No.760201, Warrendale PA (1976) Yu Yao Y.F. and Kummer J.T.,J. Catal., 106,307, (1987) Yao H.C. and Yu Yao Y.F., J. Catal., 86,254, (1984) Shyu J.Z., Otto K., Watkins W.L.H., Graham G.W., Belitz R.K. and Gandhi H.S., J. Catal.,lJ& 23, (1988) su E.C. and Rothschild W.G., J. Catal., B,506, (1986) Su E.C., Montreuil C.N. and Rothschild W.G., Appl. Catal.,U, 75, (1985) Truex T.J., Presentation at 4th International Conference on the Chemistry of the Platinum Group Metals, July 1990 - to be published Kim G., Ind. Eng. Chem. Prod. Res. Dev., 2,267, (1982) Engler B., Koberstein E. and Schubert P.,Applied Catalysis&. 71, (1989) Yu Yao Y.F., J. Catal., 152, (1984) Summers J.C. and Ausen S.A., J. Catal., 58, 131, (1979) Meriaudeau P., Dutel J.F., Dufaux M. and Naccache C., Metal-Support and Metal Additive Effects in Catalysis, ed. B. Imelik, Amsterdam 1982 Sanchez M.G. and Gazguez J.L., J. Catal., 104,120, (1987) Henmann J.M., Rameroson E., Tempere J.F. and Guilleux M.F., 117,(1989) Applied Catalysis, Shyu J.Z. and Otto K., J. Catal., U, 16, (1989) Jawarska G.Z., React. Kinet. Cat. Letter, 19. 1-2,23, (1982) Hicks R.F., Qi H., Young M.L. and Lee R.G., J. Catal., 122, 280, (1990) Yu Yao Y.F., Ind. Eng. Chem. Prod. Res. Dev.& 223, (1980) Briot P., Auroux A., Jones D. and Primet M., Applied Catalysis,5p, 141, (1990) McCarthy E., Zahradnik J., Kuczynski G.C. and Carberry J.J., J. Catal, 3 , 2 9 , (1975) Yao H.C., Stepien H.K. and Gandhi H.S., J. Catal., 61, 547, (1980) Wang T. and Schmidt L.D., J. C a t a l . , X 187, (1981) 73, (1988) Harrison B., Diwell A.F. and Hallett C., Platinum Metals Review, Joy G., Lester G. and Molinaro F., SAE 790943, 1979 54, (1972) Bauerle G.L., Service G.R. and Nobe K., I. EC Prod. Res. Develop., Summers J.C. and Baron K., J. Catal., 51,380, (1979) v. d. Bosch-Drierbergen A.G., Kieboom M.N.H., v. Dreumel A., Wolf R.M., v. Delft F.C.M.J.M. and Nieuwenhuys B.E., Catalysis Letters, 2,235, (1989) Beagley B., Private Communication Shyu J.Z., Weber W.H. and Gandhi H.S., J. Phys. Chem., 92, 4964, (1988) Belton D.N., Sun Y.M. and White J.M., J. Phys. Chem., 88, 5172, (1988) Frost J.C., N a t u r e a m , 577, (1988) Yao H.C., Gandhi H.S. and Shelef M., Stud. Surf. Sci. Catal.,U, 159, (1982) Bell A.T., Catalyst Design Progress and Perspective, John Wiley 1987 Henmann J.M., J. Catal., & 404, (1984)
a
a
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZI 0 199 1 Elsevier Science Publishers B.V., Amsterdam
153
CHARACTERIZATION OF OXIDATION CATALYSTS BY CO-TPR AND SELECTIVE CARBON-CARBON BOND RUPTURE.
Ch.Serre1, F. Garid¶*, G. Belotl, G. Maire2, R. Rochel 1: Peugeot S.A. I E.R. I 8 rue des Fauvelles 92250 La Garenne Colombes. FRANCE. 2: Laboratoire a2 Catalyse et Chimie des Surfaces, URA 423 CNRS-Institut L.e Bel, ULP 4 rue Blaise Pascal 67070 Strasbourg Cedex. FRANCE. *: To whom all correspondance should be addressed.
Abstract The results obtained with TPR tests under CO/He are discussed. For ceria and Pt-Ce catalysts, they are well correlated with those of the literature. The reaction of methylcyclopentane hydrogenolysis is used as a chemical probe to study Pt, Pd and Ce catalysts. It is found to be a very sensitive test for the study of surface modifications.of exhaust catalysts These studies indicate that, under dynamic conditions, the catalyst surface is subjected to permanent evolutions.
INTRODUCTION.
Heterogeneous catalysis is extensively used for industrial application where the working conditions are mild and almost constant. The particularity of automotive exhaust catalysts is that, contrary to others, they are subjected to (i) high temperature, (ii) numerous poisons and (iii) gas composition fluctuations. In order to remain efficient in this hostile environment, the catalyst requires a quite complex formulation composed o f many precious metals (Pt, Pd, Rh [1,2]), additives (principaly ceria [1,2]) and a stabilized support (y-Al203 dopped with La [3,4]), where all ingredients are more or less in interactions. A lot of studies has already been devoted to understand the behaviour of these catalysts. A large range of test methods including, on one hand, physical characterisations and, on the other hand, chemical characterisations is available. In this study, we try to develop new in situ methods aimed to study the surface state of the catalyst. Two approaches are used. The first one consists of a temperature programmed reduction (TPR) performed under CO diluted in He. Compared with the classical TPR under H2, the main difference is that CO does not diffuse into the bulk of the particles. Thus, the reduction occurs via a chemical pump effect which causes the oxygen diffusion up to the surface. Moreover, working under CO atmosphere brings additional advantages. It enables to get a TPR spectum of Pd
154
catalysts, which is difficult with H2 because of hydrogen dissolution in Pd [ 5 ] and in terms of oxygen disponibility, it better simulates real working conditions under rich mixture, since the exhaust gas contains about 3 volumes of CO for 1 volume of H2. The second method is performed under H2 and uses the methylcyclopentane (MCP) hydrogenolysis reaction which is structure sensitive [6-81. Therefore, it enables us to follow the modifications of the surface state induced by different high temperature treatments.
EXPERIMENTAL. Catalysts preparation. The catalysts studied are composed of various combinations of Pt, Pd and Ce, deposited on y alumina pellets (table 1) by successive impregnations of each metal in the order: Ce, Pd, Pt for multimetallic catalysts, using the following procedure: the pellets are immersed in water containing the desired amount of precursor salt (Ce(N03)3, Pd(NH3)4C12, or H2PtC16 ). After l h contact, the solution is slowly evaporated ( = 80°C ) and the sample is calcined in air at 400°C for 2h. For each sample, the metal contents are determined at the “Service Central de Microanalyse du C.N.R.S.” (Vernaison - France) by atomic absorption.
Table 1 : Characteristics of the y-Al2O3 (Reference : GCO 64 purchased from Rhone-Poulenc)
Experimental procedures. Figure 1 shows a schematic diagram of the experimental apparatus used for all the tests performed in this work. The carrier gas (He or H2) flows through the reference side of the thermal conductivity detectors (TCD), whereas the test gas flows successively through the first TCD, the catalyst (C), the second TCD, and the I.R. carbon monoxide analyser (I.R.). The traps (TR) may be connected to the system, either for H2O trapping (in TR1) or for hydrocarbon injection (TR2), and the sampler (S) enables subsequent chromatographic analyses of the outgas. The flow rates are monitored by mass flow controllers (FC). The catalyst is heated by a regulated furnace and its temperature is measured by a thermocouple located in the pellets bed. All experiments are performed on 4.7 g of catalyst, at atmospheric pressure. All gases (>99.995%) are used without any further purification.
155
* Light-off - TPO - TPR. These tests are performed under helium at a flow rate of 100 cc/min (VVH=2350 h-1). First of all, the fresh catalyst is heated in He up to 700°C (10"C/min) for l h in order to dehydroxylate [9] the alumina support (H2O is trapped in TR1) and to avoid formate formation during the subsequent CO treatment [lo], then it is cooled at room temperature (R.T.), reduced in l.S%CO/He from R.T. up to 700°C (lO°C/min) plus l h isothermal at 700"C, purged in He and cooled back to R.T..
T W
FI
Figure 1: experimental apparatus. FC: mass flowcontroller TRi,TRz : traps TCD: thermalconductivity detector c:catalyst F furnace S : sampler I.R.: CO analyser F1: flow meter
After this pretreatment procedure, the catalyst surface is expected to be "clean and in a reproducible reduced state". The catalyst is then subjected to a sequential temperature programmed oxidation/reduction procedure, as shown in figure 2.
Two types of oxidation are performed, depending on the oxidizing atmosphere composition. The first one, referred as "light-off', is carried out
156
under a gas mixture containing 1.5%CO-1.5%02/He, i.e. in excess of oxygen according to the stoichiometry of CO oxidation. The second type of oxidation (noted "TPO") is simply conducted under 02/He (under various Po2). The thermal treatment is the same for both oxidations: after getting, at room temperature, a stationary outgas composition, the temperature is rised up to 700°C at a constant rate of 10"C/min, and kept l h isothermal at 700°C. The oxidative atmosphere is then sweeped out by pure He before catalyst's cooling. At the end of the oxidation step, the catalyst surface is more or less covered by different kinds of adsorbed oxygen atoms, which are characterized by a subsequent thermo-programmed reduction (see fig. 2). The TPR is performed under a flow of lS%CO/He, by the same procedure as for the previous oxidation: (i> stabilisation of the CO level at R.T., (ii) heating (10"C/min) until 7OO"C,(iii) l h isothermal at 700"C, (iv) purge with He, and (v) cooling back to R.T.. After this first oxidation /reduction cycle, the catalyst can be subjected to another cycle, similar to the leadrich atmospheres in the exhaust pipes, either to test the reproducibility, or to study the influence of the oxidative atmosphere ( Po2, presence of CO ) on the TPR results. Considering the widely used catalytic test of light off, let us stress the fact that, in our particular conditions, it is not possible to take into account this activity characterisation. Indeed, because of our very low space velocity, the temperature at which 50% of CO is oxidized into C 0 2 is in the range 100140°C for all catalysts, i.e. almost within the experimental error of the temperature determination. Furthermore, our apparatus is not equipped either with C02 nor with 0 2 detectors, and the advancement of the reaction is only followed by the desapearence of CO in the effluent gaz. Consequently, our light off curves consist of the superposition of two phenomena: (i) CO oxidation and (ii) CO adsorption on the catalyst surface. For these two reasons, the oxidation step of the procedure is not studied by itself, but our attention is focussed on its influence on the TPR results.
* Hydrogenolysis of Methylcyclopentane. This molecule is used as an "in situ" chemical test as this hydrocarbon reaction is structure sensitive. The aim of this study will be to point out the possible surface modifications that may occur during high temperature treatment (700°C) under various atmospheres: He, CO/He or 02/He. Therefore, the hydrogenolysis investigations are undertaken by reproducing the successive steps of our oxidation/reduction cycle i.e.: 1. fresh catalyst 2. neutral treatment: in He, RT+70OoC, 10"C/min + l h isothermal at 700°C " 3. reducing treatment: in 1.5%CO/He 4. oxidative treatment: in 5%02 /He '6
'6
157
After each step, the catalyst is reduced overnight with H2 at 300”C, and then tested using the methylcyclopentane (MCP) hydrogenolysis reaction. This catalytic test is also investigated with the apparatus described in figure 1,but with pure H2 as carrier gas and a flow rate of 71 ml/min. 10 pl MCP (puriss: >99.5%) are injected in TR2 which is maintained at -38°C in order to get a constant vapor pressure of MCP in the H2 flow. The reaction temperature is fixed at 200°C. After the passage of MCP on the catalyst (detected with the 2nd TCD), 5ml of the products are sampled in S and analysed on an independant FID chromatograph apparatus (DC 200 column).
Physical characterization. - The Pt catalysts’s dispersion is determined by H2 chemisorption at R.T. using a pulsed technique on prereduced catalysts (450°C under 10%H2 /Ar). - The mean Pt particle sizes are measured by the Pt line broadening of X-ray diffraction spectra. RESULTS AND DISCUSSION. TPR results. As mentioned previously, the temperature-programmed reductions are performed under a mixture of lS%CO/He, therefore they reflect the ability of the different type of surface oxygen to be reduced by CO. Generally speaking, the TPR spectra are reproducible and do not evolve with the number of the oxidatiordreduction cycle (for some catalysts, up to 10 successive cycles have been performed). This means that after our pretreatment procedure, the catalyst has reached a stable state. The effluent composition is analysed by two methods: (i) the %CO is measured by the IR detector and (ii) the second TCD gives a global composition of the gas. These data are entered in a computer and converted to plots versus temperature. Moreover, the TCD scale is recalculated knowing that after the stabilization of the effluent composition at R.T., both detectors are measuring 1.5%CO. As a result of this computer treatment we get two curves noted respectively “CO by IR” and “TCD” (for example on Fig.3). Assuming that the only reaction occuring during the TPR is the oxidation of CO in C02 , the difference between the two curves corresponds to the level of CO2 in the effluent. This calculated signal (noted “TCD-CO”) is represented on the graphs (see Fig.3 for example) but one must keep in mind the assumption that no parasite reaction will occur, and that this curve is drawn in arbitrary units.
158
Catalvsts without ceria. *0.2%PtIA1203. Platinum oxide is known to be unstable at 700°C even under an oxidative atmosphere [ 111, therefore under our experimental conditions this catalyst does not exhibit any reduction peak in the range 25400°C after the oxidation step either under 0 2 /He or under C 0 + 0 2 /He. The only CO consumption is very small and occurs after 600°C. Therefore we can exclude CO disproportionation as an important reaction below this temperature, probably because of the low CO% used. *1.06%Pd/A1203.In contrast with platinum, PdO is stable at 700°C in an oxidative environment, and the TPR spectra presents many peaks depending on the previous oxidative treatment as shown in Fig.3(a and b). At present, we do not have a proven assignment for the oxygen species associated with each one of the peaks, but it is worth noting that the relative peak intensities are affected by the presence (Fig.3.a) or the absence (Fig.3.b) of CO in the oxidative atmosphere. On the basis of this observation, adsorbed CO is thought to induce a surface restructuration of palladium particles [ 121 leading to a modification of the oxygen species distribution. Fig.3.b TPR of 1.062Pd/AlZOS (aft- dW0n 15202)
Jooad 2JQM)
“
8
rk
- w by 1 3 . --
‘b
.......... .iW
- CO bylR
21%
Gu
-- TCD -.‘“m-co
st&
so0
7bo
lc3 ‘V .(.=\cpCO
*0.2%Pt-l.7%PdIA1203. T h e data are presented in Fig.4 which have to be compared to the equivalent Pt-free spectra of Fig.3.a. One can notice that the peak positions in Fig.3.a remain almost unchanged by the addition of 0.2%Pt but that their intensity decreases markely. This effect is attributed to the existence of an interaction between the two precious metals and points out the real bimetallic character of the catalyst.
159
8
B
Catalysts containing ceria. *14S%CeIA1203. The shape of the spectrum (Fig.5) indicates a wide peak between 300 and 550°C accompained by an unresolved peak on the high temperature side. These results are very close to those obtained by using the classical TPR performed under hydrogen [4,12-131. *0.52%Pt-13.3%CelA1203. As previously mentioned, platinum reduction does not occur during our TPR experiments. The spectra in Fig.6 is thus only representative of ceria reduction. By comparing Fig.5 with Fig.6, the presence of platinum appears to promote the reduction of “capping oxygen” of ceria [12] (shift of the first peak from 500°C down to 350°C).In the case of TPR under H2, the same effect is observed [4,12-131 and could be attributed to H2 spillover on Pt particles, as proposed by Boudart [14] for the reduction by H2 of W 0 3 catalysts containing Pt. Taking into account this explanation, one might expect Pt to have no effect on ceria reduction by CO, which is not the case. Therefore, and in agreement with Yao [13], the results reported here are explained in terms of an interaction between Pt and ceria leading to the destabilisation of the Ce-0 bond. *I .51 %Pd-9.8%CelA1203. This W.7 TPR Of 1.51md-Q.BzCe/U203 catalyst exhibits many unresolved (it* an) TPR peaks as sketched in Fig.7. From 50000Fig.3.a (1%Pd/A1203) and Fig.5 2soOo. (14.5%Ce/A1203), one can deduce that the low temperature side of the spectra (T<300”C) is mainly representative of the precious metal reduction whereas the high temperature side is due to both - IXJbyIR -- TCD -.(?\CD_co reduction of the precious metal and ceria. For 1%Pd/A1203 (Fig.3.a), it
160
appears that the first reduction peak (visualised on the curve “TCD-CO” around 150°C) occurs without apparent consumption of CO (curve “CO by IR”). Therefore, this reduction is attributable to CO initialy adsorbed on the surface at R.T.. When ceria is added (Fig.7), the opposite effect is observed and the first TPR peak is accompagnied by an equivalent CO consumption. As a conclusion of this comparison and in accordance with Summers et al. [16], the addition of ceria is thought to inhibit CO chemisorption at room temperature on palladium particles which indicates that palladium and ceria are interacting. Others experiments have been performed on Pd-Ce/A1203 in order to study the effect of Po2 during the oxidative treatment. This study has revealed that an increase in the oxygen percentage (02%/He: 0.8, 1.5, 5 % ) has no influence on the subsequent TPR peak position, but it does globaly increase their intensity. This evolution is ascribed as an increase of the amount of oxidized Pd as the 0 2 percentage levels up, the metallic particles being not totally oxidized in the bulk. This result is consistent with the conclusion published by Shyu et a1 [17] on Pd-Ce02/A1203 where they observed a decrease in propane oxidation rate when increasing the Po2of the feed gas. Concerning the coadsorption effect of CO and 0 2 previously observed on the TPR profile of 2%Pd/A1203 (Fig.3.a and 3.b) it is difficult to say if it is still present when ceria is added because the peaks which were modified in Pd/A1203 are now in superposition with the reduction of ceria. +0.2%Pt-2%Pd-l3%CeiA1203.The TPR spectra of this catalyst does not exhibit additional points of interest. Similarly to the effect already observed for Pd/A1203, the addition of ceria to 0.2%Pt-2%Pd/A1203 appears to inhibit the CO chimisorption at RT.
Results of MCP hydrogenolysis. Previous studies performed in the laboratory [6-81 pointed out that the MCP ring opening is structure sensitive. In particular, for Pt catalysts, the probability for each C-C bond rupture in the C5 ring was found to depend on the metal particle size: for well dispersed catalysts, the C5 ring is statisticaly opened; for badly dispersed platinum catalysts n-hexane is not formed. This effect enables to correlate the Pt particle size and the “selectivity ratio”: M3P/nHex (3methylpentaneh-hexane) which corresponds to CII-CII / CII-CIII bond rupture. Within the scope of this article, we only discuss the results obtained for the fresh catalysts and after the whole series of high temperature treatments (the last step being the catalyst oxidation). Those results enable to investigate the global modifications that occur during our TPR pretreatment (i.e.
161
thermodesorption, reduction, oxidation) as mentioned in the experimental procedure.
Table 2 Physical characterization D: Dispersion; dia: diameter of Pt particles (1) for more details,see the experimental part
Catalvst
Treatment
(*I 1-fresh 4-Ome
Treatment
D
dia
(1) fresh
81%
<20A
02me
13%
908,
fresh
95%
<208,
02me
15%
160A
Conversion %
%
crack r.o.(I) 11.4 62.2 0
3.0
*13%CelA1203. This catalyst has no activity at this temperature of reaction (200°C). Therefore, when interpreting the following studies of PM-Ce/A1203 catalysts, one must keep in mind that the reaction always occurs on the precious metal (PM), which is more or less modified by the presence of ceria. * 2 % P t l A 1 2 0 3 . The data are presented in table 3. The fresh catalyst exhibits a very high activity and low particle sizes as shown by the M3P/nHex ratio which is around 0.5 [6] (i.e. statistical rupture of all C-C bonds of the cycle).The high temperature treatments lead to a tremendous decrease of activity and an increase of the Pt particle sizes (M3P/nHex =2.5).
2MPW 3MPW nHex(4) CycloP(5) 3MP/nHex +c1 37.0
27.6
31.7
3.7
0.87
70.9
20.9
0
0
2.5
162
* I .51 %Pt-12.5%CelAl203. The most stricking feature of table 4 is the relatively high selectivity for cyclopentane formation over Pt-Ce/A1203. On the oxidized catalyst, we have performed two MCP hydrogenolysis tests: (i) one before the night reduction under H2 at 300°C (noted "hexed.") and (ii) a second after a one-night reduction (noted “afxed.”). As shown in table 4, the cyclopentane percentage is well correlated to the oxidation state of the surface. This effect is attributed to a dipole-like interaction between platinum and ceria [18-191 resulting in a more or less positive charge on the precious metal and depending on the oxidation state of cerium. This electron deficient Pt is thought to promote the selective adsorption of the exocyclic carbon of MCP (slightly negatively charged [20]) which leads to the formation of cyclopentane and methane. Table 4: 1.51%Pt-12.51%Ce/AI203. Conversions in cracked products and isomers 1 and distribution in isomers for the MCP hvdrogenolvsis at 200°C.
Treatment
(*I 1-fresh 4-0fle be.red af.red
Conversion %
%
crack r,o.(l) 0.6 35.2 0.2 0.3
18.0 28.5
2MPW 3MP(3) nHex(4) CycloP(5) 3MP/nHex +c1 28.8
21.8
40.9
8.4
0.53
29.7 32.5
18.5 23.6
28.1 34.4
23.6 9.5
0.66 0.68
*For more details see the experimental part. (1) ring opening; (2): 2methylpentane; (3): 3methylpentane; (4):n-hexane; (5): cyclopentane.
Concerning the meaning and the value of the M3P/nHex ratio, 2methylpentane isomerisation tests where undertaken [21]. For this reaction, 3MP/nHex ratio have values very different from those obtained with the MCP hydrogenolysis. These discrepancies show that 3MP/nHex ratio from MCP reaction couldn’t be anymore related to the Pt particle size in the case of PtCe/A1203. This observation was confirmed by the physical analysis of the used catalyst (see table 2) which indicates that a Pt particle growing had occured although the M3P/nHex remains constant. A third point of interest emerges from table 4: compared with the results on Pt/A1203, the activity of the fresh Pt-Ce/A1203 catalyst is lower but it remains almost constant after the high temperature treatments in spite of the growth of the Pt particles.
163
Table 5: 2%PdlA1203. Conversions (in cracked products and isomers). and distribution in isomers for the MCP hydrogenolvsis at 200°C
*For more details see the experimental part. (1)ring opening; (2): 2methylpentane; (3): 3methylpentane; (4): n-hexane; (5): cyclohexane.
Table 6: 1.51 %Pd-9.80%Ce/A1203. Conversions (in cracked products and isomers), and distribution in isomers for the MCP hydroeenolysis at 200°C
*For more derails see the experimental part (1) ring opening ; (2): 2methylpentane; (3): 3methylpentane; (4):n-hexane; (5): cyclohexane.
*2.0Z%PdlA1203. The fresh catalyst exhibits a quite low activity (see Table 5 ) compared to that of platinum (Table 3). This result is consistent with previous studies obtained in the laboratory [22]. The hydrogenolysis test performed just after the oxidative treatment (“be.red.”) shows a large increase of activity which goes back to a small value after a one-night reduction under H2 at 300°C (“afxed.”). Following Hicks et al. [12], Pd particles are subjected to a surface roughing when treated under oxygen at high temperature. After our oxidation step, this “rough PdlzO surface” is thought to be reduced into a “rough Pdo surface” during the hydrogenolysis test (200°C under H2). This Pd restructuration may increase the number of accessible Pd atoms ( compared to
164
the fresh catalyst) and leads to a higher activity. The low activity detected after the night reducing treatment (“af.red.”) is tentatively attributed to a surface smoothing of the Pd particles induced by H2 at 300°C. *I 5 1 %Pd-9.8%Ce/A1203.By comparing Table 5 (Pd/A1203) with Table 6 (Pd-Ce/A1203), it emerges that ceria enhances the catalytic activity of Pd, but only for the fresh catalyst, In fact, Pd-Ce/A1203 is found to become totaly inactive after either the neutral (under He) or the reducing (under 1 .5%CO/He) high temperature treatments (those intermediate results are not reported in Table 6). This drastic evolution is ascribed to a partial coverage of Pd by rare earth oxide as already proposed by some authors [23-241. Following these authors, the coverage of Pd is suppressed by a subsequent exposure under 0 2 . Therefore, the 3.7% conversion exhibited by the reoxidated catalyst gives an additional evidence for this interpretation. Furthermore, the addition of ceria does not seem to modify the subsequent restructuration of Pd induced by the night reduction, as shown by the decrease of MCP conversion.
CONCLUSION. Our TPR results obtained under carbon monoxide can be correlated with those obtained by others authors under hydrogen. The hydrogenolysis of MCP is found to be a very sensitive test for the study of surface evolutions, starting from fresh catalyst. This chemical test may give informations about the state of the catalyst during the catalytic gas exhaust conversion. Nevertheless, complementary investigations are required in order to confirm our interpretations. These two approaches indicate that surface modifications of the catalyst occur when working under dynamic conditions. The following points have been underlined: - state changes in the surface of metallic particles during oxidation/reduction cycles - particle growing at high temperature - modification of mutual interactions for PM and ceria (decoration or electronic transfer) - changes in the oxidation state of the various catalyst components. All these electronic and geometric effects are involved in determining the activity and the selectivity of the catalyst. Finally, the fresh catalysts appear to age very quickly at high temperature, as shown by the hydrogenolysis results. Then, the catalysts reach a more stable state, as shown by the reproducibility of the successive oxidation/TPR cycles.
165
ACKNOWLEDGEMENTS. The authors are indebted to Dr. L. Hilaire for helpful discussions. We thank Dr. P Wehrer (chemisorption measurements) and Dr. G. Meunier (XRD) for their contributions.
REFERENCES: J.C. SUMMERS, W.B. WILLIAMSON and H.C. HENK,SAE Paper 880281 (1988) L.L. HEGEDUS, J.C. SUMMERS, J.C. SCHLElTER and K. BARON, J. Catal. 56 321-335 (1979) M. DAUZAT, M. PIJOLAT and M. SOUSTELLE, J. Chim.Phys. 85 (9) 865-869 (1988) B. HARRISON, A.F. DIWELL and C, HALLET, Platinum Metals Rev. 32 (2) 73-83 (1988) H.LlESKE, G.LIETZ, W.WANKE and J.VOLTER, Z.Anorg.allg.Chem. 527 135-149 (1985) F.G. GAULT, Advan. Catal. 30 1-95 (1981) F. GARIN, 0. ZARHAA, C. CROUZET, J.L. SCHMIlT, G. MAIRE and F.G. GAULT, Su$. Sci. 106 466-471 (1981) J.M. DARTIGUES, A. CHAMBELLAN, S. COROLLEUR, F.G. GAULT, A. RENOUPREZ, B. MORAWECK, P. BOSCH-GIRAL and G. DALMAI-IMELIK, Nouv. J. Chim. 3 591-601 (1979) B.C. LIPPENS and J.J. STEGGERDA, in: Physical and chemicals aspects of adsorbent and Catalysis. Ed. B. G. LINSEN (Academic Press, New York, 1970) 171-211 P.G. GOPAL, R.L. SCHNEIDER and K.L. WATTERS, J. Catal. 105 366-372 (1987) T. HUIZINGA, I. VAN GRONDELLE and R. PRINS, Appl. Catal. 10 199-213 (1984) R.F. HICKS, H. QI, A.B. KOOH and L.B. FISHEL, J. Catal. 124 488-502 (1990) H.C. YAO and Y.F. YU YAO, J. Catal. 86 254-265 (1984) H.C. YAO, Appl. Su$. Sci. 19 398-406 (1984) R.B. LEVY and M. BOUDART, J . Catal. 32 304-314 (1974) J.C. SUMMERS and S.A. AUSEN, J. Catal. 58 131-143 (1979) J.Z. SHYU, K. OTTO, W.L.H. WATKINS, G.W. GRAHAM, R.K. BELITZ and H.S.GANDHI, J. Catal. 114 23-33 (1988) Y.F. YU Y A 0 , J . Catal. 87 152-162 (1984) F. Le NORMAND, L. HILAIRE, K. KILI, G. KRILL and G. MAIRE, J . Phys. Chem. 92 2561-2568 (1988) R. H O F F M A “ J. Chem. Phys. 39 1399-1412 (1963) Unpublished results M. HAJEK, S. COROLLEUR, C. COROLLEUR, G. MAIRE, A. O’CINNEIDE, and F.G. GAULT, J. Chim. Phys. 71 1329-1336 (1974) J.S. RIECK and A.T. BELL, J. Catal. 99 278-292 (1986) W. JUSZCZYK, Z. KARPINSKI, J. PIELASZEK, I. RATAJCZYKOWA and Z. STANASIUK in “hoceedings, 9* International Congress on Catalysis, Calgary, 1988” (M.J. Phillips and M. Ternan, Eds.), vo1.3, pp 1238-1245. The Chemical Institute of Canada, Ottawa, 1988.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
167
DESIGN AND PERFORMANCE EVALUATION OF AUTOMOTIVE EMISSION CONTROL CATALYSTS
R. G. Silver, J. C. Summers and W. B. Williamson Allied-Signal Automotive Catalyst Company, Tulsa, Oklahoma 74158
ABSTRACT The performance of a series of single noble metal automobile emission catalysts was determined as a function of washcoat composition and aging conditions. The use of catalysts containing individual noble metals allowed the contributions of each type of noble metal to be identified, and revealed the role of the alumina supports containing cerium. Each noble metal was found to have a washcoat which optimized its performance under a given set of conditions. Addition of Ce to Pt and Pd-only catalysts improved performance after aging, but Ce did not improve Rh performance at 450 OC. The individual contributions of Pt, Pd, and Rh for aged three-way performance indicate significant advantages of using Pd-Ce over Pt-Ce, however when they are combined the predominant contribution to TWC activity comes from Rh. The influence of CO concentration in the feedstream on the light-off activity was determined for single noble metal catalysts on an alumina/ceria washcoat. Increasing the CO content of the feedstream led to an increase in light-off temperature for all three noble metal catalysts. The light-off characteristics varied with the noble metal, and are thought to be related to the stability of the noble metal oxides.
INTRODUCTION
Advanced automotive emission control technology coupled with sophisticated vehicle emission systems will be critical in helping to achieve improved global air quality. Improving the activity and durability of catalyst technologies are paramount to alternative catalyst control strategies such as increasing platinum (Pt) or rhodium (Rh) noble metal content or increasing catalyst volumes. Factors such as high Pt and Rh prices, noble metal scarcity, low recycling recovery rates for Rh, and increased global usage of noble metals make the latter a costly approach for improving the performance of automobile catalysts. Significant performance improvements in Pt/Rh three-way catalyst (TWC) formulations have been previously achieved by incorporating base metal promoters such as cerium (Ce) into TWC formulations. Cerium significantly enhances NOx and CO performance near stoichiornetric aidfuel
168
(A/F) ratios [1,2,3]. Increasing Ce levels in Pt/Rh TWCs substantially improves NOx and CO durability against thermal deterioration with only slight improvements in HC oxidation [4]. The increased use of palladium (Pd) containing catalysts for three-way emission control [5-91 could lower converter costs, and provide potential durability improvements. Recent Pd/Rh TWC technologies rival current high performance Pt/Rh technologies for a variety of vehicle applications [9]. In order to advance the design of automotive catalysts for maximum durability performance, the present study determines the relative individual contributions of Pt, Pd, and Rh in current high performance ceria-alumina catalyst formulations. This will allow future washcoats to be tailored to optimize each noble metal's contribution. Of course, these types of single component studies preclude any considerations of beneficial effects of noble metal synergism or deleterious effects of metal-metal interactions (e.g. Pd-Rh alloying [lo]). Noble metal durability performance over a variety of Cealumina supports are compared. The influence of exhaust CO concentration on CO and HC light-off activity over the individual noble metal catalysts has been determined, by comparing fresh, laboratory hydrothermal aged and engine aged catalysts. EXPERIMENTAL PROCEDURES
Catalyst Preparation Catalysts were prepared using 1.2 s/L Pt, 1.2 g/L Pd, and 0.24 g L Rh (33, 33, and 6.7 g/ft3, respectively) either alone or in combination with each other, adding the noble metals to washcoats containing varying amounts of calumina and cerium, and impregnating on monolithic cordierite substrates with 64 square cells/cm2 (400 cells/ i d ) . Cylindrical cores used for laboratory tests were 2.5 cm in diameter by 5 cm in length, the length being composed of four 1.25 cm segments taken from various locations down the monolith bed to minimize sampling errors. For engine dynamometer testing, the catalysts were sectioned into quarters. Eight quarters were cemented together to make two full-size catalyst pieces.
Laboratory Aging and Evaluation For the CO light-off study, fresh and hydrothermally aged samples were preconditioned at 590 OC in a simulated automotive exhaust gas [6] containing either 0.3, 1.43 or 2.0 mole% CO (corresponding to a lean, stoichiometric or rich automotive exhaust, respectively) for 15 minutes, before ramping in the unmodulated preconditioning gas (with a 90,000 h-1 gas space velocity) from 90 to 565 OC at 15 OC/ min. The conversion of CO and hydrocarbon during the temperature ramp was determined as a function
169
of temperature using an automated exhaust system analyzer. The samples were then cooled in nitrogen to 90 OC and then the ramp was repeated. This procedure was continued until duplicate results were obtained. Sample cores were hydrothermally aged by heating at 1000 OC for 4 h in a gas stream of 10% water in air in a Lindberg 3-zone tube furnace.
Engine Aging and Evaluation
The catalysts for the noble metal and washcoat durability studies were aged and evaluated using conventional automotive emission test methods which have previously been described [ 111. Two aging cycles of differing severity used gasoline containing 4 mg Pb/L and 0.5 mg P/L. The less severe aging cycle operated at the stoichiometric point with an inlet temperature of 760 OC during a cruise mode and included a lean fuel-cut mode with bed temperatures of 850 OC. A more severe aging cycle operated at the stoichiometric point with a higher inlet temperature of 850 OC and with lean fuel cuts having 910 OC bed temperatures. RESULTS AND DISCUSSION
Noble Metal Durability Study The relative durability performances of Pt/Rh and Pd/Rh catalysts, and the individual noble metals Pt, Pd and Rh on Ce containing alumina washcoats were determined after 100 hours of engine aging (850 OC inlet, fuel cut). Catalyst performances at 450 O C are shown in Figure 1 as integral conversions about the stoichiometric point (A/F = 14.56 k 0.15) [12]. The most noteworthy result is the poor durability performance of the Pt-only catalyst. The Pd-only catalyst has significantly higher conversions. The Rh catalyst at a lower loading generally has poorer HC conversion than the Pd catalyst, but both catalysts perform similarly for CO and NOx. Figure 1 indicates a major weakness in three-way Pt/Rh catalysts at high temperatures: Platinum alone is very susceptible to performance loss on state-of-the-art washcoats. This is due to sintering of the Pt-particles during aging to form large crystallites, leading to a lower dispersion and lower surface area of the noble metal [13]. This in turn results in a smaller number of active sites and, therefore, in a loss of performance. In contrast, Pd remains dispersed under normal operating conditions [ 141, and performs well after aging relative to Pt. The Pt/Rh and Pd/Rh catalysts have comparable durability performance, except for a slight advantage in rich side NOx for the Pt/Rh TWC. Since Pd alone is much more durable than the Pt, these results suggest that there is a counterbalancing loss of performance due to probable alloy formation of Rh with the Pd. The slight rich side advantage in gross NOx
170
conversion for Pt/Rh suggests that Pt does a slightly better job of activating Rh than Pd. Platinum is also known to be less selective for NO to N 2 formation and therefore results in more ammonia [6].
s
HC Evaluated at 450 C, 30000 S.V.
co Emission
NOx
Fig. I . Noble metal catalysts on an alumina-ceria washcoat evaluated at 450 O C afer aging 100 h at 850 O C .
These results indicate that Pd/Rh catalysts can have TWC performance comparable to Pt/Rh commercial catalysts. Significant advantages exist for Pd over Pt in Ce-alumina catalysts for CO and HC oxidation and NOx reduction. Even though the Rh is at a much lower loading than either of the other noble metals, it makes the predominant contribution to three-way performance. Based on the performance of the individual nobel metals, Pt/Rh might be expected to have lower conversions and Pd/Rh to have better conversions, than are shown in Figure 1. Since Pt/Rh is similar to Pd/Rh in performance after aging, it is possible that Pt and Rh interact to improve catalytic activity, while Pd and Rh interact in a manner which decreases activity.
Pt-only and Rh-only Durability Study Durability studies were carried out using Pt-only and Rh-only catalysts over various washcoat formulations containing increasing amounts of Ce. Two additional samples were made. One contained zirconia and the other did not. These latter catalysts explore the effect of Zr on Rh performance. The catalysts are listed in Table 1. The catalysts were aged for 100 hours in
171
quadrant reactors on a 760 OC inlet, fuel cut engine aging cycle, and then for an additional 50 hours on a 850 OC inlet, fuel cut engine aging cycle. For Pt-only catalysts, improved performance was obtained with increasing amounts of Ce in the washcoat. Figure 2 shows the promotional effects of Ce on Pt performance as a function of aidfuel (A/F) ratio, evaluated at 450 OC and a space velocity of 30,00O/h after 100 hours of the milder aging. Hydrocarbon and CO conversions improved in going from rich to lean conditions. The addition of Ce to the c-alumina washcoat results in improved HC, CO and NOx conversions, especially at stoichiometric conditions (A/F = 14.56). Table 1 Ce concentration Noble Metals Washcoat A1 0 Pt Al, Ce 1.3X Pt Pt Al, Ce 2x Pt Al, Ce 4x A1 Rh 0 Rh 1.3X Al, Ce Rh 2x Al, Ce Al, Ce Rh 4x Al, Ce with Zr Rh 2.4X Al, Ce without Zr Rh 2.4X
*
NoCe HC Conversion l
o
80
o
1.3XCe
-A-
2XCe 0
CO Conversion
-*
4XCe
NOx Conversion
! !
m
I I k’’
> c
0
?42
144
146
14d
Evaluated at 450 C, 30000 s v
144
146
148
1 2
144
146
148
I
AIF Ratio
Fig. 2. Effect of Ce content on Pt-only catalyst pe$orrnance as a function of AIF ratio, evaluated at 450 OC, after aging 100 h on the 760 OCfuel cut
cycle.
172
Figure 3 shows the effect of aging time and severity of aging conditions on conversions (450 OC inlet) over a Pt catalyst at the stoichiometric point versus the amount of Ce in the catalyst washcoat. Without Ce in the washcoat, the catalyst is essentially deactivated after 50 hours of mild aging (Figure 2). At this point, high stoichiometric conversions are seen with a 2X addition of Ce, and additional CO/NOx benefits are evident after doubling the Ce content (to 4X). After 100 hours of 760 O C fuel cut aging, the catalyst requires twice the amount of Ce (4X) to attain the same high conversions seen earlier with the 2X Ce loading. After an additional 50 hours of relatively severe aging (850 OC fuel cut), the catalyst with the 4X Ce loading shows some activity for CO conversion, but otherwise all Pt catalysts are severely deactived. Thus, increasing the Ce loadings resulted in improved catalyst performance after aging over the Ce range studied. 50 hours aging
150 hours aging -0-
100 hours aging
-8-
%
HC Conversion
Conversion
CO Conversion
0
1x
2x
3x
4x
0
:
w
3x
4x
Ce Content
Fig 3. Effect of aging time on Pt catalyst conversion at the stoichiometric point as afunction of Ce content, evaluated at 450 O C . Catalysts were aged 50 and 100 h using the 760 OC fuel cut cycle and then an additional 50 h on the 850 OC fuel cut cycle. As noted earlier, Pt loses significant activity during aging due to thermal deterioration. The above results suggest that Ce prevents the Pt from sintering, probably by keeping some of the Pt in a partially oxidized state [ 131. Increasing the Ce content of the washcoat may increase the amount of partially oxidized Pt, which in turn is harder to sinter than metallic Pt. In addition, ceria likely improves the performance of the Pt by acting as an oxygen source for CO conversion. Similar studies were performed in our lab using Pd/ Al, Ce catalysts. Although not included here, results obtained over
173
Pd-only and Pd/Rh catalysts show that Ce promotes Pd activity in a manner similar to its promotion of Pt [6]. HC Conversion
Evaluated at 450 C, 30000 S.V.
NoCe
1.3XCe
WCe
8
-A-
0
CO Conversion
-*
4XCe
NOx Conversion
NF Ratio
Fig. 4. Effect of Ce content on Rh-only catalyst performance as a function of AIF ratio, evaluated at 450OC, after aging 100 h on the 760 OCfuel cut cycle. For Rh-only catalysts, the conversion performance was not as sensitive to cerium content as for Pt-only catalysts. Figure 4 shows the effect of Ce content on Rh performance as a function of A/F ratio for catalysts aged 100 hours on the 760 OC fuel cut cycle, after evaluation at 450 O C (space velocity = 30,000 hours). Hydrocarbon and CO conversions were improved by making the A/F ratio leaner, and NOx conversions were improved by making it richer. The best hydrocarbon conversions occured with no Ce in the calumina washcoat, while the best CO and NOx conversions occured with the highest amount of Ce. Furthermore, these performance differences were much smaller than for the Pt-only catalysts. The 4X Ce loading catalyst also contains a separate, pure c-alumina phase in addition to the Ce-alumina, which probably accounts for the improved HC conversion [15] over the other Cecontaining catalysts. The results suggest that Rh is relatively less affected by the addition of Ce for COpNOx conversions at 450 OC, and that Ce may actually be detrimental to Rh hydrocarbon conversion under these conditions. However, a recent study suggests that Ce may enhance NO reduction over Rh at temperatures lower than those used here [ 161. Figure 5 shows the effect.of aging time and severity on Rh catalyst conversions at the stoichiometricpoint versus the amount of Ce in the catalyst
174
washcoat. After 50 hours of aging, HC conversions are not affected by Ce content, while CO and NOx conversions show some improvement with increasing Ce. At 100 hours of aging, the HC conversion initially deteriorates with increasing Ce content, but improves at the 4X Ce level as described above. A small improvement is seen in CO conversion with increasing Ce, but NOx conversion appears to be unaffected. After an additional 50 hours of the more severe aging, HC, CO and NOx conversions show a pattern similar to that seen with HC conversion after 100 hours of aging. While adding Ce to the washcoat may aid in CO and NOx conversion initially, Rh ultimately provides the best activity when it is not in contact with Ce. 50 hours aging --Ft
100 h z a g i n g
150 hours aging -€-
NOx Conversion
CO Conversion
0
1x
2x
1
3x
4x
0
IX
2x
3x
4x
Ce Content (g/cf)
Fig 5 . Effect of aging time on Rh catalyst conversion at the stoichiometric point as a function of Ce content, evaluated at 450 O C . Catalysts were aged 50 and 100 h using the 760 O C fuel cut cycle and then an additional 50 h on the 850 O C fuel cut cycle. The effect of zirconia on Rh durability performance is illustrated in Figure 6. After 50 hours of 760 OC fuel cut aging, the results obtained show performance advantages at 450 OC for the non-Zr containing catalyst, for HC and NOx. After 100 hours of aging, the two catalysts are nearly equivalent, due to minimal deterioration of the Zr-containing catalyst. After an additional 50 hours of the more severe aging, the situation is reversed with distinct advantages seen for the Zr-containing catalyst for HC, CO and NOx conversions under all conditions. Zr stabilizes Rh by preventing it from interacting with aluminum when it is exposed-to high temperatures [17,18,19].
175
co-
760c loo 8502 w 760c loo 850Cl5 Aging Time (h)and Inlet Temperature
Fig 6. Effect of Zr on Rh catalyst catalyst conversion at the stoichiometric point as a function of aging time and temperature, evaluated at 450 O C . Catalysts were aged 50 and I00 h using the 760 OCfuel cut cycle and then an additional 50 h on the 850 O C fuel cut cycle. Results at 350 OC indicated that the alumina-only catalyst had the best low temperature activity. This suggests that the Rh may light off more easily on an alumina washcoat surface, than on a more stabilized washcoat. Adding Ce to an alumina washcoat greatly improves the activity and durability of the Pt catalysts. In contrast, the Rh catalysts seem to perform the best over blank alumina, or alumina stabilized by zirconia. By testing the noble metals in this way, an environment can be designed for each of them which optimizes their durability and performance. Nevertheless, it still appears that high temperatures are detrimental to Pt performance.
CO Concentration Effects The effect of CO concentration on the light-off performance of Pt, Pd and Rh-only catalysts was determined for the single noble metal samples used in the Noble Metal Durability Study. All of these samples contained cerium. Catalysts were tested fresh, after lab aging in air with 10% water at 1000 O C for 4 hours, and after engine aging for 100 hours using the 760 OC fuel cut aging cycle. Figure 7 shows the performance of the fresh catalysts. Over Pt-only catalysts, increasing the amount of CO in the feedstream led to higher lightoff temperatures for both CO and hydrocarbons. The steady state conversion of CO and hydrocarbon also decreased. CO is converted via reaction with 0 2 to form C 0 2 . If the reaction is assumed to occur on the surface between
176
adsorbed species, and if CO competes for adsorption sites with 0 2 , then it seems plausible that increasing the amount of CO in the feedstream would lead to a decrease in CO conversion and an increase in CO light-off temperature. Since more of the 0 2 is used up in CO conversion, a decrease in hydrocarbon conversion occurs as the CO content in the feed is increased. CO light-off occurred before hydrocarbon light-off, due to the well-known CO poisoning effect on Pt. A similar pattern is seen over fresh Pd-only catalysts, except there is a much smaller difference between the hydrocarbon light-off temperatures at lean and rich conditions (Figure 7). Both CO and hydrocarbon light-off occurs at lower temperatures than over the Pt catalysts [6] for each test condition. Decreasing hydrocarbon conversion under rich conditions above 450 O C may be due to decreasing CO conversions via the water-gas shift (WGS) reaction [20]. The lower WGS conversions mean that CO must then compete directly with hydrocarbon for available oxygen. An inflection point is also seen in the hydrocarbon light-off curves at about 60% conversion. This corresponds to the point where the reaction is switching from conversion of unsaturated hydrocarbons to saturated hydrocarbons, which were present in the feedstream in a ratio of 67% propylene/ 33% propane. In another experiment (not shown here), the inflection point in the light-off curves was seen to change when the ratio of saturated to unsaturated hydrocarbons was changed. Under lean conditions, steady state hydrocarbon conversions for Pt are greater than those for Pd. At stoichiometry conversions are equal, and under rich conditions conversions are greater for Pd than for Pt. This might be explained by assuming that there are two competing rate limiting steps: hydrogen abstraction from the hydrocarbon and oxygen adsorption on the catalyst surface. Under lean conditions there is an abundance of oxygen which can adsorb on the surface to react with the hydrocarbon, and therefore the limiting reaction is hydrogen abstraction from the relatively rare hydrocarbons. Pt is proposed to be more efficient at abstracting hydrogen from hydrocarbons than Pd [21], therefore Pt should give better conversions under lean conditions. Under rich conditions there is a shortage of oxygen molecules, which must compete with the more numerous CO molecules for surface sites. Since more hydrocarbons are available for hydrogen abstraction, the rate limiting step in this case could be oxygen adsorption. Oxygen is more readily adsorbed over Pd than over Pt, so Pd would be expected to give the higher conversions under rich conditions. The light-off curves in Figure 7 over fresh Rh catalysts show several interesting features. In this case, increasing the amount of CO in the feedstream appears to have little effect on the CO light-off temperature. This may be due to Ce enhancement of the NO-CO reaction over Rh [16]. The hydrocarbon light-off is also different from that seen with Pt or Pd, as
Hydrocarbon
Hvdrocarbon
100
Carbon Monoxide
Carbon Monoxide
] ’ / - : ’
80 60 40 20
100
100
Lean
’
8 0 -
,
Rh 0
200
400
-0
2.40
4tm
Inlet Temperature (OC)
Fig 7. Hydrocarbon and CO lightoff curvesfor single NM catalysts as a function of the amount of CO in the feedstream. Lean = 0.3% CO, Stoich.= 1.45% CO, and Rich = 2.0% CO.
0
200
400
600
200
400
600
Inlet Temperature fC)
Fig 8. Hydrocarbon and CO lightoff curves for NM catalysts aged I00 h on the 760 OCfuel cut cycle as afunction of the amount of CO in the feedstream. Lean = 0.3% CO, Stoich.= I .45% CO, and Rich = 2.0% CO.
178
conversion starts at a lower temperature under stoichiometric and rich conditions, than under lean conditions. This may be because the oxidation of hydrocarbons over Rh is known to be inhibited by excess oxygen [22]. Clearly, the Rh-only catalyst on a cerium containing washcoat is very inefficient at converting hydrocarbons, with a maximum conversion of less than 70%. Like Pd, CO and hydrocarbon light-off over Rh occurs at lower temperatures than over the Pt catalysts for each test condition. The differences in light-off between the three noble metals may be related to the stability of their oxides. Assuming that the reaction occurs between adsorbed species on the catalyst surface, rapid adsorption of gas phase oxygen, as well as CO and hydrocarbon, is a critical step. Both Rh and Pd have more stable oxides than Pt and can therefore more readily adsorb oxygen to react with adsorbed CO and hydrocarbons. The relative stability of the oxides, and thus the relative surface coverage of oxygen anions, would also explain why Pt was most affected by higher CO levels and Rh was affected the least. Light-off evaluations were repeated after laboratory aging the catalyst samples at loo0 OC for 4 hours with 10% water in air. Lightoff temperatures were higher than for fresh catalysts, especially over Pt catalysts, where aged CO/HC lightoff could be as much as 100 OC higher than fresh light-off. Light-off temperatures over aged Pd and Rh were about the same under lean conditions. Light-off over Rh occurred before Pd under stoichiometric and rich conditions, and Pt was last in all cases. Aging the catalyst may result in a reduction of the number of sites where oxygen can adsorb on the surface, because of sintering of the noble metals. This would tend to affect Pt more than Pd or Rh, due to the relative stability of the noble metal oxides. Oxide stability, in turn, may be related to the sintering resistance of the noble metal, with the more stable oxides being more resistant to sintering. When CO feedstream content is increased in addition to aging, differences between Pd and Rh appear: Pd is more affected since its oxide is not as stable as that of Rh. Figure 8 shows the results of testing the catalysts which were engine aged using the milder aging cycle for 100 hours. A commercially available Pt/Rh catalyst was included in the evaluation as a reference point. Compared to the fresh catalysts, the engine aged samples had higher light-off temperatures and sometimes had lower steady state conversions, just as with the lab aged catalysts. Pt catalysts were again the most affected, having 135 O C higher rich-side HC lightoff temperatures than when fresh. Engine aged Pd had the lowest light-off temperature under lean conditions. Pt/Rh and Rh had the lowest light-off temperature under stoichiometric and rich conditions. Pt-only catalysts had the poorest light-off under all conditions. As expected, light-off occurred the fastest under lean conditions, followed by stoichiometric and then rich conditions, except for hydrocarbon lightoff over
179
Rh. These results suggest that lower light-off temperatures can be achieved using certain Pd-only catalyst technologies under lean conditions. The Pt/Rh catalyst appears to behave just like the Rh-only [23] catalyst for CO conversion, suggesting that most of the CO catalytic activity in aged catalysts is primarily due to Rh. Hydrocarbon conversion is initiated at temperatures similar to those of pure Rh, but Pt improves lean and stoichiometric steady-state conversions in this case. Compared to the laboratory aging done in this study, this engine aging cycle appears to be more severe for light-off of Rh under all conditions, more severe for Pd under stoichiometric and rich conditions, and more severe for Pt under rich conditions. Lab aging is more severe than engine aging for Pt light-off under lean conditions, and the two agings provide similar results for light-off over lean Pd and stoichiometric Pt. CONCLUSIONS
A careful study of single noble metal catalysts can reveal ways in which the design of automotive emission control catalysts can be modified to improve conversion performance. Increasing the Ce content in an alumina washcoat greatly improves the activity and thermal durability of Pt-only catalysts. The high temperature durability of Rh-only catalysts is improved via the addition of Zr to the washcoat, which has been attributed to decreasing Rh interaction with the alumina component in the washcoat. Pd-only catalysts are the first to light-off under lean conditions, while Rh-only catalysts provide the earliest light-off under stoichiometric and rich conditions. Significant durability and light-off advantages of Pd-only over Pt-only TWCs are indicated as well. However, the possible beneficial effects of Pt-Rh synergism or the detrimental effects of Pd-Rh alloying were not considered. ACKNOWLEDGEMENTS
D. G. Linden prepared the catalysts used in this study. J. R. Coopmans and R. J. Shaw made the stand dynamometer tests. J. H. White coordinated the engine testing. D. M. Thomason and D. H. Whisenhut performed the laboratory agings and evaluations. L. A. Butler typographed the manuscript. REFERENCES 1. 2. 3.
L. L. Hegedus, J. C. Summers, J. C. Schlatter and K. Baron, "Poison-Resistant Catalysts for the Simultaneous Control of Hydrocarbon, Carbon Monoxide, and Nitrogen Oxide Emissions", J. Catal. 1979, 56, 321. G. Kim, "Ceria Promoted Three-Way Catalysts for Auto Exhaust Emission Control", Ind. Ene. Chem. Prod. Res. Dev., 1982,21, 267. B.J. Cooper and T.J. Truex, "Operational Criteria Affecting the Design of Thermally Stable Single Bed Three-Way Catalysts", SAE Paper 850128, 1985.
180 4. 5.
6. 7. 8.
9. 10. 11. 12. 13. 14.
15. 16. 17. 18.
19. 20. 21. 22. 23.
W.B. Williamson, J.C. Summers, and J.F. Skowron, "Catalyst Technologies for Future Automotive Emission Systems," SAE Transactions Paper 880103, 1988, 97, 341. H. Muraki, H. Shinjoh, H. Sabukowa, K. Yokata and Y. Fujitani, "PalladiumLanthanum Catalysts for Automotive Emission control", Ind. Ene. Chem. Prod. Res. Dev. 1986, 25, 202. J.C. Summers, W. B. Williamson and M. G. Henk, "Uses of Palladium in Automotive Emission Control Catalysts", SAE Paper 880281, Transactions 1988,97, 158. J.C. Summers, J.J. White and W. B. Williamson "Durability of Palladium-Only Three-Way Automotive Emission Control Catalysts", SAE Paper 890794, 1989. R. Heck, K.S. Patel and J. Adomaitis, "Platinum Versus Palladium Three-Way Catalysts -Effect of Closed-Loop Feed-Back Parameters on Catalyst Efficiency", SAE Paper 892094, 1989. J.C. Summers, W. B. Williamson and J.A. Scaparo, "Role of Durability and Evaluation Conditions on the Performance of Pt/Rh and Pd/Rh Automotive Catalysts", SAE Paper 900495, 1990. G. W. Graham, T. Potter, R. J. Baird, H. S. Gandhi, and M. Shelef, "Surface Composition of Polycrystalline Pd- 15 Rh Following High Temperature Oxidation in Air" J. Vac. Sci. Technol. 1986, 44, 3. J.F. Skowron, W. B. Williamson and J.C. Summers, "Effect of Aging and Evaluation Conditions on Three-Way Catalyst performance", SAE Paper 892093,1989. W. B. Williamson, J.C. Summers and J.A. Scaparo, "Automotive Catalyst Strategies for Future Emission Systems", AIChE 1990 Spring National Meeting, Paper 52F, March, 1990, Orlando FL. Engler,B., Koberstein, E. and Schubert, P., "Automotive Exhaust Gas Catalysts: Surface Structure and Activity" , ADDIied Catal. 1989,48,71. Schlatter, J, and Taylor, K., "Platinum and Palladium Addition to Supported Rhodium Catalysts for Automotive Emission Control", J. Catal. 1977,49,42. H. S. Gandhi and W. L. H. Watkins, U. S. Patent 4,782,038, Nov. 1, 1988. Oh, S. H., "Effects of Cerium Addition on the CO-NO Reaction Kinetics over Alumina-Supported Rhodium Catalysts", J. Catal. 1990, 124,477. H.C. Yao, S. Japar, M. Shelef, "Surface Interactions in the System Rh/AI203", J1977,50, 407. H.C. Yao, H. K. Stepien and H.S. Gandhi, "Metal Support Interactions in Automotive Exhaust Catalysts: Rh-Washcoat Interaction", J. Catal, 1980,61, 547. H. K. Stepien, W. B. Williamson and H. S. Gandhi, "Development of ThermalResistant Rhodium catalysts", SAE Paper 800843, 1980. G. Lester, G. Joy, and F. Molinaro, "Water-Gas-Shift and Steam Reforming Ability of Group VIII Metals in Simulated Automotive Exhaust", paper presented at the 6th North American Catalysis Society Meeting, 1979. A. Schwartz, L. L. Holbrook and H. Wise, "Catalytic Oxidation Studies with Platinum and Palladium", J. Catal. 1971, 21, 199. J. T. Kummer, "Use of Noble Metals in Automobile Exhaust Catalysts", J. Phvs. Chem, 1986, 90, 4747. W.B. Williamson, H. K. Stepien and H. S. Gandhi, "Poisoning of PlatinumRhodium Automotive Three-Way Catalysts: Behavior of Single-Component Catalysts and Effects of Sulfur and Phosphorus", Environ. Sci. & Tech, 1980, 14, 319.
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 0 199 1 Elsevier Science Publishers B.V., Amsterdam
181
INFLUENCE OF Rh AND C e 0 2 ADDITION ON THE ACTIVITY AND + NO SELECTIVITY OF A Pt/A1203 CATALYST IN THE AND C o + NO + 0 2 REACTIONS
co
G. Leclercq, C. Dathy, G. Mabilon* and L. Leclercq Laboratoire de Catalyse He'te'rogene et HornogZne, U.R.A. C.N.R.S. N o0402, Universite' des Sciences et Techniques de Lille Flandres-Artois F - 59655 Villeneuve d'Ascq Cedex (France) *Institut Francais du Pe'trole, 4 et 6, rue de Bois Prkau, F - 92506 Rueil-Malmaison Cedex (France) ABSTRACT Four catalysts were prepared by impregnation of y-alumina :lwt% WAl2O3, lwt% Pt 0.2 wt% Rh/A1203, lwt% Pt -12wt% Ce02/A1203 and lwt% Pt - 0.2wt% Rh -12wt% CeOgA1203 and characterized by their activities and selectivities in CO + NO and CO + NO + 0 2 reactions in temperature programmed experiments from 100 to 550°C at a space velocity of 25000 h-l. In the reaction between CO and NO, at low temperature, NO gives mainly N20 the percentage of which in the products decreases when the temperature and the conversion increase. On the contrary for reactions CO + NO + 02, at low temperature and low conversion, the selectivity of NO transformation into N20 is low when oxygen is present in the gas phase because of a better adsorption of 0 2 than that of NO. Rhodium and ceria have a promoting effect on both CO + NO and CO + NO + 0 2 reactions, but the promoting effect of ceria manifests itself mainly in the presence of oxygen. During the reaction between CO + NO, besides its oxidation by NO, CO is oxidized in another process which is likely to involve the supports (alumina or ceria). INTRODUCTION
The CO-NO reaction is a major reaction pathway for removal of nitrogen oxides from automobile exhaust. This reaction has to be selective for N 2 formation which is favored at high temperatures. But at low temperatures, the formation of N 2 0 as a primary nitrogen-containing reaction product occurs as it has been reported in previous studies over supported Rh catalysts (1-4). Moreover, McCABE and WONG ( 5 ) have recently shown that N 2 0 formed during the CO-NO reaction can undergo further reaction with CO to produce N2. Owing to the noxious character of N 2 0 which is well-know as an "ozone killer" and can also contribute to the green house effect, it seems important to understand the conditions of N 2 0 formation from a fundamental
182
point of view related to the CO-NO reaction mechanism. In this paper we have first examined, in laboratory feedstreams, the conversion and the selectivity in the CO-NO reaction, with and without oxygen over an alumina-supported Pt compared to the bimetallic Pt-Rh on the same support. In a second part, we have studied the modification involved in the activity and the selectivity of the reactions by the addition of cerium oxide. EXPERIMENTAL
Catalvsts Four catalysts were prepared by impregnation of y-Al2O3 with solutions of hexachloroplatinic acid and rhodium trichloride. Their composition are as fOllOWS : 1 wt% Pt/A1203, 1 wt% Pt-0.2 wt% Rh/A1203, 1 wt% Pt-12 wt% CeOdA1203, 1 wt% Pt-0.2 wt% Rh-12 wt% Ce02/A1203. The y-Al2O3 had a surface area of about 100 m2/g and a total pore volume of 1.15 ml/g. It was used in the form of a powder with particle size between 80 and 124 pm. After impregnation with 12 wt% Ce02 the surface area of the support was 128 m2/g The dispersion of the metals was about 0.55 for Pt/A1203 and 1.00 for 1 wt% Pt-0.2 wt% Rh/A1203. It has not been possible to measure the dispersion of metals on CeOdA1203 either by hydrogen or oxygen chemisorption because of the large amounts stored on Ce02. Catalvtic testing The catalytic tests were performed in a plug flow reactor at a constant space velocity of 25.000 h-1 (gas flow rate per volume of catalyst) with 0.2g of catalyst diluted with 0.8g of a-AI203. The catalyst was heated following a temperature program at a constant heating rate of 1K per minute from 100150°C to 500-550°C. For the CO+NO test the reactant mixture contained 0.5 mole% CO and 0.56 mole% NO in helium. For the CO+N0+02 reaction the composition of the feedstream was 0.75 mole% CO, 0.28 mole% NO and 0.26 mole% 0 2 in helium. The analysis of the gas mixture at the inlet and at the outlet of the reaction was performed by gas chromatography using a column CTRl provided by ALLTECH and which is composed of two concentric columns : the inner one filled with Porapak Q and the outer one with molecular sieve 5A. By this way we were able to separate N2, 0 2 , NO, CO, N 2 0 and C02 in a single analysis and to calculate separately the conversions of CO and of NO. The detailed procedure of this analysis will be described elsewhere. Before being submitted to a catalytic run, the samples of catalysts were heated at 500°C in flowing helium (heating rate 3K per minute), then activated in the CO + NO reactant mixture at the same temperature and rapidly cooled down to room temperature.
183 RESULTS
1 - 1 wt% Pt/Al&& Reaction CO + NO The variations of the conversions of CO (TCCO) and of NO (TCNO) as a function of the reaction temperature are shown in figure 1a. One can see that while the stoichiometry for CO + NO reactions :
(1) (2)
2CO+2NO CO+2NO
+
+
2C02+N2 C02+N20
only allows TCCO/TCNO ratios comprised between 0.5 and 1, at low temperatures between 100 and 400°C the conversion of CO is consistently higher than that of NO. Moreover, starting from about 220°C, both CO and NO conversions level up, or even slightly decrease until about 300°C is attained, then the conversions increase again at higher temperatures. For this catalyst, the light off temperatures (at 50% conversion) are roughly the same for CO and NO - 410°C -. The selectivity of NO reactions has been characterized by the percentage of NO transformed into N 2 0 ( S N ~ O) the changes of which as a function of the reaction temperature are shown infig. I b . At low temperature (and conversion) S N ~ Ois 100% then it decreases slowly as the temperature is increased before decreasing drastically at the light-off temperature. Reaction CO + NO + 0 2 In the presence of oxygen the reaction rates together for CO oxidation and for NO reduction are much higher than without oxygen as seen infigure 2a , the light-off temperatures being respectively 290°C and 330°C for CO and NO conversions. The depletion of oxygen in the gas phase is indicated on the curve TCCO = f(T) by the presence of a small shoulder (marked with an arrow on fig. 2a ). As long as oxygen is present in the gas phase, the conversion of NO is very low and its start being noticeable only when oxygen is depleted in the gas phase. This is due to the well-known inhibiting effect of oxygen on NO chemisorption. Let us note that the levelling effect between 220 and 300°C observed for the CO + NO reaction is not observed here in the case of the CO + NO + 0 2 reaction. The selectivity S N ~ O(jig. 2b) is very different here than in the absence of oxygen since, at low temperature, when oxygen is present in large amounts in the gas phase, S N ~ Ois very low, then it increases when the 0 2 concentration decreases and when all 0 2 has been consumed then S N ~ Ois the same as for reaction CO + NO before, of course, decreasing when NO conversion is high.
CO (0,5%) - NO (0,5670)
CO (O,75%) - NO (O,28%) - 0 2 (0,257%)
TC CO - TC NO = f (T'C)
TC CO- TC NO = f (T'C) 100
100
e
60
60
40
40
20
20
- 80
8 E
-60
0
- 40 - 20
CO (0,5%) - NO (0,56%)
-ae
5 0 1 0 0 1 5 0 200 250 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 T ("C)
CO (0,75%) - NO (0,28%) - 0 2 (0,257%) TC NO - S N 2 0 = f (T'C) 100
80
80
60
60
40
40
20
20
100
r 100
1
w
4
e
0 0
0 5 0 1 0 0 1 5 0 200 250 300350 400 450 5 0 0 5 5 0 600
T ("C)
F i g . 1. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) of Pt/A1203 i n CO t NO r e a c t i o n s .
4 2
YO
0
TC NO - S N20 = f (T"C) 100
3
w
0 5 0 100 1 5 0 200 250 300350 400 450 5 0 0 5 5 0 6 0 0 T ("C)
0
100
-6
80
8
r
0
5 0 100 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 600
T ("C)
F i g . 2. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) of Pt/A1203 i n CO t NO t O2 r e a c t i o n s .
185
2- Influence of Rh : 1 wt% Pt - 0.2 wt% Rh / A1202 Reaction CO + NO The presence of Rh greatly enhances the activity for reaction between CO and NO since the light off temperatures are 130K lower than on Pt/A1203 (280°C for both conversion). The curves TCCO and TCNO versus the temperature start not very differently than for Pt/A1203 (fig. 3a) at low temperature but here the levelling of conversion was not observed. The conversion of CO is still slightly higher than that of NO. Concerning the selectivity, S N ~ Ois high at low temperature and low conversion ( f g . 3b) then it decreases and becomes very low after the light off temperature. At a given temperature, S N ~ O seems to be lightly lower on Pt-Rh than on Pt alone, however since the conversion of NO is not the same for the two catalysts one has to be careful in comparing selectivities of the two catalysts. Reaction CO + NO + 02 In the presence of oxygen in the gas phase, the curves TCCO and TCNO versus T Ifis. 4a) are shifted towards lower temperatures for Pt-Rh than for Pt, but the shapes of these curves are qualitatively the same. The differences in the light-off temperatures in the presence of oxygen compared to those of the reaction without oxygen is less for Pt-Rh than for Pt as evidenced in Table I where the light-off temperatures for Pt-Rh and Pt with and without oxygen are recalled. TABLE I
Light-off temperatures (at 50% conversion)
Moreover, the difference (AT) in these light-off temperatures for CO and for NO conversions is lower on Pt-Rh/Al2Og (AT = 30K) than on Pt/A1203 (AT = 40K). Hence the difference in reactivity of CO and of NO in the presence of oxygen is less on Pt-Rh than on Pt. This is also shown infigure 4 a where it can be seen that at low temperature the slopes of the two curves TCCO and TCNO = f(T) are less different than on Pt/A1203 (Jig. 2a). To illustrate better this assessment, when the conversion of NO is 5%, the
CO (0,5%) - NO (0,56%) TC CO - TC NO = f (T'C)
CO (0,75%) - NO (0,28%) - 0 2 (0,257s) TC CO - TC NO = f (T'C)
100
100
80
80
h
8
8
-z
u
v
'
60
60
4
40
40
e
20
20
0
0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 450 5 0 0 5 5 0 6 0 0 650 7 0 0 T ("C) CO (0,75?0 ')
- NO (0,28%) - 0 2 (0,257%)
TC NO - S N20 = f (T'C)
A
100
-
80
-
g
-
.
-
4 602
40
-
+ TCNO(%) Q
20
-
0-
:4o
S N20 (%)
- 20
u)
pa9
6o
60
2
40
40
20
20
0 . . . 01 0 0 I 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0 700
T P'C)
F i g . 4. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) o f Pt-Rh/A1203 i n C O t NO + O2 r e a c t i o n s .
8
*
187
conversion of CO is 19% on Pt-Rh, but it is 28% on Pt. On Pt-Rh the selectivity SN2O is higher at low temperature than on Pt (fig. 4b) and it does not start from zero at low conversion and low temperature. 3- Influence of CeOZ
3a.- I wt%Pt-12wt%Ce02lA1203 For the reaction CO + NO, at low temperature, the activity of PtCe02/A1203 is higher than that of Pt/A1203 and when the temperature is increased the activity goes through a maximum at about 250°C (fig. 5a), followed by a minimum at about the conversion of CO is always higher than that of NO, to varying extents at various temperatures. The light off temperatures are 380°C both for CO and NO reactions (with a gain of 30K compared to Pt/A1203). In the presence of oxygen such a minimum in activity does not occur (fig. 6a). The activity of Pt-Ce02/A1203 is higher than that of Pt/A1203 both for the oxidation of CO and for the reduction of NO. As with Pt/A1203 the conversion of NO remains low in the presence of 0 2 in the gas phase and becomes important only when 0 2 disappears from the gas phase; the depletion of 0 2 corresponding to a slower increase in CO conversion versus temperature is indicated by an arrow on figure 6a . Before the depletion of 0 2 , the curves TCCO and TCNO vs T have very similar shapes for the two catalysts with and without Ce02 (however with CeO2 the curves are shifted of lOOK towards lower temperatures). But after all 0 2 of the gas phase has been consumed, these two curves rise much more slowly for PtCe02/A1203, this result in a AT between the light off temperatures for CO oxidation and NO reduction of 70K, substantially higher than for Pt/A1203 or Pt-WA1203 (Table I). The selectivity S N ~ Oin the reaction CO + NO (‘jig. 5b) is not very different with and without CeO2. In the presence of oxygen, the curve S N ~ Ovs T (fig. 6b) has the same shape with Pt-Ce02 as with Pt alone, starting from zero at low conversion as a results between the competition of adsorption between 0 2 and NO. 3b. I wt%Pt-O.%wt%Rh12wt%CeOzlAlz03 Finally, the addition of 0.2 wt%Rh to Pt/CeOa results in a higher activity with curves TC CO and TC NO vs T qualitatively similar to those obtained with Pt/CeO2 but shifted towards lower temperatures vig. 7a and 8a and Table I). For the selectivity S N ~ O(‘fig. 7b and 8b), the curves are similar in shape to those obtained without Ce02, however it is difficult to compare the results without Rh or without ceria since both temperatures and conversions were different.
CO (0,75%)- NO (0,2870) - 0 2 (0,2577)
CO (0,5%)- NO (0,5670)
TC CO - TC NO = f (TOC)
TC CO - TC NO = f (TOC)
100
100
80
80
8
60
60
P
40
-ae
100
Y
Q
TCCO(7)
20
20 0
.
0
0
.
50 100 150 200 250 300350 400 450 500 550 600
0
50 100 150 200 250 300350 400 450 500 550 600 T ("C)
T ("C) CO (0,5%)- NO (0,567) TC NO - S N20 = f (T'C)
CO (0,7570) - N O (0,2870) - 0 2 (0,25770) TC NO - S N20 = f( T"C)
100
Iae
b
loo
-80
Y
l60
Q TC NO(%) + S N20 ("A)
20 0 0
50 100 150 200 250 300 350 400 450 500 550 600 T ("C)
F i g . 5. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) of Pt-Ce02/A1203 i n CO + NO r e a c t i o n s .
0 T ("C)
F i g . 6. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) of Pt-Ce02/A1203 i n CO t NO t O2 r e a c t i o n s .
- 40 - 20
f? z u)
CO (0,5%) - NO (0,56%) TC CO - TC NO = f (T”C) 100
-ae 8 2
CO (0,75%) - NO (O,28%) - 0 2 (0,257Y) TC CO - TC NO = f (T”C) 100
80
80
60
60
40
40
-c
P
Q
20
Q
+
0 0
TCCO (Yo) TCNO(%)
20
TCCO (Yo)
+ TC NO (“A)
0 5 0 1 0 0 1 5 0 200 250 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 50 600
0
5 0 1 0 0 1 5 0 200 250 3 0 0 3 5 0 4 0 0 450 500 550 600
T (“C) CO (0,75%) - NO (O,28%) - 0 2 (0,257Y)
T (“C) CO (0,5%) - NO (0,56%)
TC CO - S N 2 0 = f ( T T )
TC NO - S N20 = f (T”C) 100
ae
100
80
Y
8
e
60 40
TCNO (Yo) + S N 2 0 (%) Q
20
. 0
.
.
.
.
.
5 0 100 1 5 0 200 250 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 T (“C)
Fig. 7. A c t i v i t y ( a ) and s e l e c t i v i t y ( b ) of Pt-Rh-Ce02/A1203 i n CO + NO r e a c t i o n s .
- 40 .
u)
20
0 0
0 5 0 1 0 0 1 5 0 200 250 3 0 0 3 5 0 4 0 0 4 5 0 5 0 0 5 5 0 600 T (“C)
F i g . 8. A c t i v i t y ( a ) and s e l e c t i v i t ~ y( b ) of Pt-Rh-Ce02/A1203 i n CO + NO + O2 r e a c t i o n s .
e
00
W
190
DISCUSSION
For the CO + NO reactions on Pt/Al2O3 we have seen that an unusual curve TCCO vs T was obtained and that in a temperature range of 100 to 400°C the conversion of CO was always higher than that of NO which is not allowed by the stoichiometry of reactions (1) and (2) between CO and NO. Hence CO is also oxidized by "extra oxygen" other than oxygen coming from NO. Of course, the first explanation that comes into one's mind is that CO disproportionation occurs leading to CO2 and to carbon deposit on the catalyst: 2 co 4 c02 + c . But the temperatures at which this phenomenon was observed seems too low, since CO disproportionation usually does not occur at such low temperatures on this metal (6). Moroever, CO disproportionation would result in a carbon deposit on the catalyst and consequently in a deficit of carbon in the gas phase. However the carbon balance which was measured from the peak areas corresponding to CO and C02 in the chromatograms throughout the experiment was constant all over the temperature range as shown in Table 2 where some values of the carbon balance on Pt/Al2O3 at various temperatures are given. TABLE 2
Carbon Balance on Pt/A1203 a t various Temperatures T("C) I 1 0 5 1 1 5 6 1 1961 2491 2891 3421 4051 438 Cbalance (arb.units)l2487]245512442] 243712418123871 23911 2 4 0 9 The only remaining explanation is that this extra oxygen comes from the catalyst and probably from the alumina support at the vicinity of the metal particles. Such a participation of the oxygen of the support has already been proposed by Harrison et al. (7) for the reactions between CO with NO and H 2 0 on Rhkeria and by Otsuku and Kunitumi for the oxidation of CO over praseodynium oxides (8). Hence we have used the model proposed by Harrison et al. that we have slightly modified (fig. 9) and tried to explain with it most of the features observed here. It is assumed first that carbon monoxide adsorbs on the metal (step Mi), then in the absence of oxygen in the gas phase it migrates towards the limit between metal particles and the support where it can be oxidized by oxygen atoms of the support leading to oxygen vacancies on the support (step S i ) . We will not specify the nature of oxygen atoms, they could be oxygen atoms of the alumina itself or of hydroxyl groups or of water adsorbed at the surface or of water of cristallisation of transition alumina the formula of which being
191
Al2O3, xH20 (x < 1). Of course in the presence of 0 2 , CO adsorbed on the metal could react with 0 adsorbed on the metal according to step M7. NO first adsorbs on the metal (step M2) then dissociates on the metal (step M5) or migrates towards the support and is dissociated by reaction with an oxygen vacancy (step S2). Oxygen vacancies could also be refilled by reaction with NO in the gas phase as suggested by Harrison et al. (7) or by oxygen atoms migrating from the metal where they were previously adsorbed (step S3). The balance between steps S2 and M5 will depend on the ability of the metal to dissociate NO. It is already well-known that Pt has a poor activity in the adsorption and in the dissociation of NO, at least at low temperature (9, lo), while Rh is a better catalyst for this reaction (1 1, 12). Fig. 9 :HarrisonS Model (modified) METAL
(M1) (M2) 043) (M4)
(M5)
046) (M7)
co + * + co* NO + * + NO*
I
SUPPORT
(Sl)
co + "0"+ c02 + 0
(S2)
O+{
NO*
N+ + "0"
NO+* 2N* + N 2 + 2 * N* + NO* + N20 + 2" N20 + CO + C02 + N2 NO* + * + N* + 0" 0 + o*+ "0"+ * (S3) 0 2 + 2" + 2 0 * c o * + o*+ c02 + 2*
*
= Adsorption site on the metal "0"= Mobile oxygen on the support, at the vicinity of the metal particles
0 = Oxygen vacancy on the support Then N atoms adsorbed on the metal can either react with another Nads leading to dinitrogen via step M3 or react with NO adsorbed leading to dinitrogen oxide N20 (step M4). Hence the selectivity of NO reduction into N2 or into N20 will be controlled by the relative rate of steps M3 and M4, that is of course by their relative rate constants but also by the concentration of N* and of NO*. When the concentration of N* will be high, then the selectivity of NO transformation will be in favor of N2 formation (low SN,O ). Now coming back to our results on Pt/A1203 the shape of the curves TC CO and TCNO versus T in f i g u r e l a can be explained by the above mechanism, if, at low temperature the rate of step S1 is higher than that of
192
steps S2 and M5. This would result in a reduction of the support, hence to a decrease of the activity of the catalyst explaining the levelling of the CO conversion. The high selectivity in N 2 0 formation of N O conversion is well explained by the low NO dissociation leading to low concentration of N* compared to NO*. In such conditions step M4 which leads to N 2 0 formation is favored compared to step M3. In the presence of oxygen, the leveling effect was not observed probably because : -(I) steps M6 and M7 occur on Pt -(2) the filling of oxygen vacancies rapidly takes place via step S3. Here the zero selectivity in N 2 0 at low conversion and low temperature is well explained if 0 2 adsorption is much favored on Pt than that of NO. At low temperature, in the presence of large amounts of oxygen, 0 2 adsorption prevents NO adsorption, hence the concentration of NO* is very low and consequently the rate of step M4 is low. When 0 2 is consumed by CO oxidation, the adsorption of NO progressively increases explaining why S N ~ O increases. When all 0 2 has disappeared from the gas phase, NO adsorption occurs like in the CO + NO test and S N ~ Ois the same as in the absence of 0 2 like we have already mentionned. Pt-Rh/A1703 - - : influence of Rh With Rh the amount of CO oxidized by another oxidant than NO is much less important than on Pt alone. And the levelling effect mentionned for Pt/A1203 does not exist anymore. These two observations can be rationalized within the frame of Harrison's model : NO is adsorbed and dissociated more easily on Rh than on Pt as it has already been established by some authors (11). Since, both the concentrations of NO* and of O* are higher than on Pt alone, steps S2 and S3 of refilling of oxygen vacancies are faster on Pt-Rh than on Pt explaining why the "reduction" of the catalyst is less with Pt-Rh than with Pt. The concentration of N* resulting from steps M5 and S2 is also higher and the rate of step M3 is enhanced compared to that of step M4 explaining the slight lowering of S N ~ Oobserved for Pt-Rh in comparison with Pt. In the presence of oxygen, the slightly lower difference in the light-off temperatures for CO and NO conversions on Pt-Rh ( AT = 30K) than on Pt (AT = 40K) together with the lower difference of the slopes of the curves TCNO and TCNO vs temperature at low temperature can also be accounted for by the model, if the adsorption of NO is less unfavored compared to that of 0 2 on Rh than on Pt. In these conditions the difference between the rates of the 2 reactions CO + N O and CO + 0 2 is less than on Pt alone.
193
- - : influence of ceria Pt-CeO?/Ab03 - - - and Pt-Rh-CeO/Ah03
The higher activity for catalysts containing ceria is well explained by a better availability of oxygen atoms from ceria which has better oxidoreduction properties than alumina. The minima of activity at about 300°C are probably related to reaction rate lower for step S2 than for step S1 leading to a lack of oxygen atoms available from ceria hence to a slowing down of step S2. In conclusion, our results have shown that : (1) in CO + N O reactions
- C02 is produced in another process in addition with the oxidation by NO. We think that oxygen from the support in the vicinity of metal particles could be involved in this process leading to partial reduction of the support at temperatures between 250 and 300°C. - at low temperature the conversion of NO gives mainly dinitrogen oxide the addition of ceria increases the reaction rate at low temperature because of the better availability of "0"in Ce02, when the temperature is increased, the activity decreases, probably as a result of partial ceria reduction. At higher temperature, catalysts with or without ceria do not have very different activities.
(2) In the presence of oxygen :
- on Pt/A1203, with or without ceria, the conversion of NO is low until oxygen has disappeared from the gas phase, because of the much better adsorption of 0 2 compared to that of NO. This is also the reason why the selectivity of N 2 0 formation is significantly lowered, specially at low temperature when oxygen is present in the gas phase. - the adsorption of NO is less unfavored compared to 0 2 adsorption when rhodium is added. - finally, the promoting effect of ceria is particularly important (light off temperatures lowered by lOOK for CO oxidation and by 70K for N O reduction). Most of these observations can be explained by the model proposed by Harrison et al. that we have modified and generalized.
194 REFERENCES 1. 2. 3. 4. 5. 6a.
6b. 7.
8. 9. 10.
11. 12.
HECKER, W.C. and BELL, A.T., J. Catal. 84, 200 (1983). CHO, B.K., SHANKS, B.H. and BAILEY, J.E., J. Catal. 115,486 (1989). PRIGENT, M. and DE SOETE, G., SAE paper nx 890.492 (1989). OH, S.H., J. Catal. 124, 477 (1990). McCABE, R.W. and WONG, C., J. Catal. 121,422 (1990). POUTSMA, M.L., ELEK, L.F., IBARBIA, P.A., RISCH, A.P. and RABO, J.A., J.Cata1. 52, 157 (1978). RABO, J.A., RISCH, A.P. and POUTSMA, M.L., J. Catal. 53, 295 (1978). HARRISON, B., DIWEL, A.F. and HALLETT, C., Platinum Metal Review 32,73 (1988). OTSUKA, K. and KUNITONI, M., J. Catal. 105, 525 (1987). VAN SLOOTEN R.F. and NIEUWENHUYS B.E., J. Catal. 122,429 (1990). SOLYMOSI, F., SSRKANY, J. and SCHAUER, A., J. Catal. 46,297 (1977). ALTMAN, E.I. and GORTE, R.J., J. Catal. 113, 185 (1988). UNLAND, M.K., J. Catal. 31, 459 (1973). ACKNOWLEDGEMENTS
This work was carried out within the "Groupement Scientifique Pots Catalytiques" funded by the "Centre National de la Recherche Scientifique", the "Institut Fraqais du PCtrole" and the AFME (Agence Francaise pour la Maitrise de 1'Energie).
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
195
INFLUENCE OF WATER IN THE ACTIVITY OF CATALYTIC CONVERTERS
M. Weibell , F. Garinl,* , P. Bemhardtl, G. Maire1 , M. Prigent2 1:Laboratoire de Catalyse et Chimie des Surfaces, URA 423 CNRS-Institut Le Bel, ULP, 4 rue Blake Pascal 67070 Strasbourg Cedex. France. 2: Institut Francais du Pe'trole,B.P. 3 I I , 1 et 4 avenue de Bois-Prkuu, 92506 Rued Mulmaison Cedex. France. *: To whom all corresponhnce should be addressed.
Abstract Because the exhaust composition in closed-loop emission fluctuates about the stoichiometric point, an investigation of the behavior of catalysts in this environment seems to be necessary. In laboratory testing of RNA1203 and Rh-Ce/A1203 catalysts, it is demonstrated that cycling effects appear on the one hand when ceria is added to Rh/Al2O3 catalyst, and on the other hand when the catalysts work in presence of water. The participation of the water gas shift reaction in the transient response of these two catalysts is demonstrated in step change experiments. It is shown that following the lean-rich transition, an enhancement of the WGS reaction appears. This reaction occurs following the formation of Rh oxide. The Rh-Ce/A1203 catalyst shows higher WGS activity than the Rh/A1203 but the stability of the Rh oxide decreases more rapidely when ceria is added.
INTRODUCTION
Reviewing the literature concerning the automotive exhaust catalysts it seems that a lot of things has been undertaken (1,2,3) . This subject is very wide and very complex. For example as far as we know, in reforming catalysis supported bimetallic systems are widely used but their catalytic behavior is always under study (4) and about metal-support interaction, always more or less present in these catalysts, works are still in progress to understand their influences ( 5 ) . In automotive catalysts we are faced to all these problems together: (i) supported bimetallics or trimetallics are used, (ii) additives are added to the support to stabilize it and also to promote the reactivity. In addition these catalysts never operate under steady-state conditions (6): (i) catalyst temperature increases rapidly when the engine starts,(ii) flow rate and composition fluctuate rapidly under all modes of operations. Finally a large number of reactions occur on these three way catalysts: (i) oxydation reactions,(ii) reduction reactions,(iii) water gas shift reaction: to mention only the most important ones (7). In a such complicated situation we are going to focus on the catalytic behaviour of low-loaded Rh/A1203 and Rh-Ce/A1203 catalysts working under
196
transient conditions with and without water. CO and NO conversions are measured at various cycling frequencies and step change experiments are performed to follow the reducibility of Rh and Rh-Ce under various atmospheres. This study will give some informations about the state of the Rh under the working conditions. It is an approach to the study of the active sites. Our results are compared to those already obtained in such a way to understand the synergetic phenomena observed under dynamical conditions EXPERIMENTAL
Catalysts : The characteristics of the two catalysts used are listed in table I. They are prepared by wet impregnation of alumina powder,purchased from Rh6nePoulenc, with aqueous solutions of RhCl3 and Ce(N03)3. After drying the catalysts are calcined in air at 450 "C for 2 h and reduced in H2 at the same temperature and for the same time. Table I. Catalysts properties. 3
Metal loadings (wt %) support (A12031 Diameter pm 250-425 Catalyst Rh loading Ce loading Density g.cm-3 0.69 ~ ~ 1 2 0 3 0.003 Surface area m2.g-1 100 Rh-Ce/Al203 0.003 1.02 Total pore volumes ml-g-1 1.15 Apparatus : The design of the laboratory system we developped is similar to the one used by Schlatter et a1 (8). In this apparatus, three sections can be distinguished, the blending section, the reactor section and an analytical train. The blending section includes eight mass flow controllers and is duplicated in two blending systems independent of one another. The two feedstreams can be as simple as binary mixture or as simulated exhaust gases containing only CO, NO, HC, HC 0 2 and N2. We have for this work omitted to use H2 because we want only to study the CO reducing power on the alumina supported Rh and Rh-Ce catalysts. When working in presence of water vapor the feedstreams flow through two water evaporators in order to have 10 % water vapor in the feedstreams. This apparatus allows us to vary the following parameters: - The frequency: This gas mixing frequency is obtained by switching two electrovalves between the two gas streams. Their operating range in our apparatus is comprised between 0.075 and 1 Hz.
197
- The amplitude: This term is related to the gas composition. For example, if switching is operated between the two streams with respectively an A/F ratio equal to 14.25 and 14.75, the amplitude is equal to k 0.25 A/F around the mean A/F ratio equal to 14.5. - The mean A/F ratio from 14 to 15.25. The amplitude and shape of the composition arriving in front of the catalyst bed as a function of the oscillations frequencies determined in our laboratory set up is reported in figure 1.
0.075 Hz
0.2 Hz
1HZ
Figure 1. Catharometer output when electrovalves are cycling at different frequencies. Horizontal lines reflect the signal for separate streams. This gauging is done with a flow of N2 in the first gas section and a mixture of half N2 and half H2 in the second section. The shape of the composition encountered at the front of the catalyst is obtained with a catharometer sensitive to changes in the gas thermal conductivity. The total flow rate is about 250 ml./mn for each stream. The horizontal lines represent the catharometer responses to individual flows in steady state conditions. When the oscillations frequency is about 0.075 Hz the catalyst is exposed alternatively to each feedstream without blending. When the frequency increases the amplitude of the oscillations declines up to 1 Hz where the catalyst is exposed to a steady stream whose composition is the average of the two blends being cycled. The reactor section is composed by an oven and a quartz reactor (inlet diameter = 10 mm). Before flowing through the analytical train, the stream is homogenized in a mixing tank in order to analyse an average composition. The analytical train is composed by two infrared analysers (CO, C3Hg, C02), by one paramagnetism analyser (02) and one chemiluminescence analyser (NO). All the apparatus are connected to a computer in order to drive the different sections and to collect the data.
198 TEST CONDITIONS AND RESULTS
A. Cycling frequency experiments : These experiments consist in an investigation of the conversion efficiency of the three pollutants as a function of our oxidant/reducing ratio which may be named as our simulated A/F ratio. The oscillation frequency has a constant gas composition amplitude o f f 0.25 APF ratio. In this paper we report the catalyst performance for CO and NO conversions at two frequencies : 0.075 Hz and 1 Hz. The composition of the streams versus the A/F ratio is that already published by Schlatter et a1 (8). The NO and HC compositions are constant in the full A/F area and are equal respectively to 2050 and to 900 ppm. The experiment temperature is kept constant at 450 "C. The space velocity is about 100 000 h-1. In each case the catalyst is first stabilized under rich conditions for 30 minutes.at 450°C. 100 -
-
-sc 0 .Lo
-
50C 0
1
14.2
14.4
'
.
1
I
14.6
14.8
NF
. '
I
15.0
.
14.2
14.4
14,8
14.6
15.0
NF
(a) Figure 2 (b) CO and NO conversions measured at 0.075 H z and I H z cyclingfiequencies. Oscillation amplitude is f 0.25 AIF ratio around the mean value. RhIAl20.3 catalyst (a) Rh-CelA1203 catalyst (b). Experiments performed without water. The catalyst performances as a function of the simulated A/F ratio are plotted in figures 2 (a) and (b). As shown in these figures and in the work of Schlatter et a1 (8,9) it appears clearly that the catalyst needs to operate close to the stoichiometric point to convert simultaneously the pollutants. On the rich side the limiting factor is the CO conversion and on the lean side it is the NO conversion. It is noticeable that the catalyst efficiency for CO and NO conversion decreases at the stoichiometric point with cycling. On the other hand under cycling conditions the NO conversion is enhanced on the lean side. This last effect is not observed for CO on the rich side with the low loaded
199
catalysts but other experiments performed on high loaded catalysts (0.5% Rh) show this effect (10). In fact transient effect on catalyst efficiency can be observed in comparing experimental transient curve at 0.075 Hz with the expected theoretical curve if no memory effect will occur. This expected theoretical curve is calculated from the stationnary curve knowing that the catalyst spends half of the oscillation period under lean conditions and the other half of the period under rich conditions with an amplitude o f f 0.25 A/F. la) Rh catalyst : On the Rh catalyst the difference between the expected theoretical curve and the experimental curve is quite small for CO and NO conversions. A little transient synergetic effect is noticeable for NO conversion, on the contrary for CO a poisoning effect is observed which could be due to co-adsorption influences, these two molecules being electron acceptors (11). (b) Effect of ceria added to Rh :
When ceria is added to Rh catalyst no modification for NO conversion occurs, on the contrary a better one is observed for CO, comparison between figures 2(a) and 2(b). At this stage we can expect an influence of the oxygen storage capacity of ceria during the lean excursion and an enhancement of the CO conversion in cycling due to the reduction of ceria by CO on the rich side (12). But the conversion is always smaller than the theoretical expected one.
-
/.
L
0 14.2
14.4
14.6
14.8
AIF
15.0
14.2
.
.
,
'
14.4
.
,
.
.
14.6
I
A
.
14.8
L
.
,
.
15.0
AJF
(a) Figure 3. (b) CO and NO conversions,in presence of water, measured at 0.075 H z and 1 H z cycling frequencies on Rh-CelAl2Oj catalyst (a). CO conversionin presence of water, at 0.075 H z and theoretical curve on the Rh-Ce catalyst (b).
200
(c) Influence of water vauor on the activity of Rh-Ce/A12Q3 _ - : (Figure 3) The transient curve shows a decrease of NO conversion due to a poisoning effect of the water vapor (fig. 3(a) compared to fig. 2(b)). For CO conversion,the influence of the water gas shift reaction is evident,there is a better CO conversion: compare on the one hand to the results obtained on RhCe/A1203 catalyst without water and on the other hand to the expected theoretical curve (fig. 3 (b)).
Table 11. From these last four figures we Feedstream composition, vol 9%. have pointed out the effect of the ceria and the importance of the water (Balance N2) gas shift reaction. These two effects brought together may explain the presence of a synergetic effect due to transient operating. To try to confirm Rich 1 110 or 0 0 these points we have performed step change experiments.
I
I
B) Step change experiments : These experiments are undertaken in order to isolate features of catalyst response to cycling conditions. The feedstreams only contain C0,02,N2 and water vapor if necessary. The experiment consists in three steps respectively rich, lean and rich. The compositions of the two streams are given in Table IT. Step 1
i
Step 2
I
1%
-
co
i 4
1 Yo
0 2
10 % H 2 0 if present
Step 3
-
1%
long time
co
1
Figure 4 : Step change experiment This test is performed at the same temperature, 450 "C, than the cycling frequency experiments, with a space velocity of 100 000 h-1. The catalyst is first stabilized under rich conditions for 30 minutes, and the experiment starts with a rich step for 2 minutes, then the lean stream is directed through the
20 1
reactor for a time (t) which will vary from one experiment to another, after which the feed is changed back to the rich composition (see figure 4). We studied the following oxidizing times (t = 1 s-6s-20s-lmn-6mn-15mn-1 lh). The composition values of the different gases are collected each 3 s and we follow especially C 0 2 formation. la, Ce effect without water : Experiments are done with several oxidation times (lean stream). The catalyst, after the 30 minutes under rich conditions, is well reduced. No C 0 2 is formed. Then the experiment starts with the first rich step (the 2 minutes step under 1% CO). After the lean step ( 1 % 0 2 ) for a time "t", the gas composition is switched to the rich step (1% CO). Instantaneously a C 0 2 peak is formed.The area of the difference of the amounts of C02 formed following the lean step for Rh-Ce/A1203 and RWAl2O3 catalysts is related to the amount of 0 2 stored by ceria. Figure 5 reports the oxidation state of Ce as a function of lean step time, keeping in mind the fact that all the Ce IV is converted to Ce 111 under CO atmosphere. Two kinetic regims for ceria oxidation appear. A fast oxidizing rate occurs in the first 20 s where 51% of ceria is oxidized thanks to the interface Rh-Ce where dissociation of 0 2 and oxygen spillover occurs from Rh. For oxidizing times greater than 20 s, 78 % of ceria is oxidized after 15 min. under lean conditions, the rate of ceria oxidation becomes slower due to cerium oxide reaction with 0 2
80
-e
60
2
40
v
s
20
0 0
200
400
600
800
1000
oxidation time (s)
Figure 5 : Oxidation state of Ce following different oxidation times. Remembering the works of Herz et a1 (13) and Le Normand et a1 (14) who showed that the reduction of ceria in presence of Rh is quite fast, we can explain the enhancement of CO conversion in the frequency test when ceria is added to the Rh catalyst. In fact with a frequency of 0.075 Hz the half period is
202
*
about 6 s which means that 45% of ceria is oxidized under lean atmosphere, then after the other 6 seconds this surface is able to oxidize CO under rich conditions. These step experiments demonstrate the oxygen storage of ceria.
9h 30r
-
a
(a) Figure 6. WGS.activity following &era1 oxidation time at 450‘”Cfor RhIA1203 catalyst ( a )and Rh-CelAlZOj catalyst (b). fb) Influence of water . As already mentionned by Herz (16) and by Dictor (15) the water gas shift reaction is enhanced on Rh catalyst. First, following the same experimental procedure than in section (a) but with gases containing 1Oii water, we may note for the two catalysts a negligible activity under the 2 minutes of 1%C0/10%H20 following the 30 mn reduction under the same gas composition (part A of the curves in figures 6 (a) and 6 (b)). After the lean step, not mentioned in the figures, the gas composition is switched to the rich conditions (it corresponds to the time zero on the figures). It may be seen from figure 6 that oxidized Rh/A1203 and Rh-Ce/A1203 catalysts have significant WGS activity, but there is no proportionality between the oxidation time of the catalyst and the WGS activity. On the other hand the Rh-Ce/A1203 catalyst shows higher activity following the lean step compared to the Rh/A1203 catalyst, but the activity decreases more rapidely when ceria is present. The figures therefore suggest that oxidized Rh is active for WGS reaction, and that the catalyst deactivates as a result of reduction by the CO rich environment. As we have seen that the oxidation time has no direct proportionality with the WGS activity,we may ask the question: which is the best oxidation time to get the highest WGS activity and stabilityjn an other word it means which is the most favorable oxidation state of the surface?
203
I
oxidation time ( s )
(a) Figure 7. (b) Catalytic WGS activity following different oxidation times (a). Stability (tll2) of WGS activity following several oxidation times (b). RhIAl203 catalyst ( ) and Rh-CelAl203 catalyst ( ).
Figure 7 (a) represents the WGS activity following several oxidation pretreatments. As already mentionned, the activity is higher in presence of ceria. For each catalyst the curves suggest that there is some intermediate oxidation treatment which gives the greatest activity to the WGS reaction. An intermediate oxidation time of ca. 4 mn provides the greatest WGS activity for the Rh-Ce/A1203 catalyst and about 9 mn for the Rh/A1203 catalyst under the reaction conditions specified. In figure 7 (b) is reported the time it takes for the C 0 2 signal to decay to half of the greatest value observed immediately following the lean step (t112). It appears that whatever the oxidation time, ti12 is greater for the Rh/A1203 catalyst than for the Rh-Ce/A1203 catalyst. It means that the rate of reduction of Rh oxide is faster when Rh interacts with ceria (13). However according to Dictor's work (15), an intermediate treatment of ca. 15 mn provides the greatest stability for the WGS reaction when ceria is absent. On the other hand it appears that the decreasing rate of WGS activity is not influenced by oxidation times higher than 4 mn when ceria is present. DISCUSSION AND CONCLUSION
WGS reaction is enhanced when Ce is added to Rh/A1203 catalyst. A " one second oxygen pulse" is enough to enhance the activity (a "11 hours oxygen pulse" does not increase the activity). At this point we may think that (Rh H Ce) interactions regulate the formation of an "active Rh oxide" entity which is intantaneously formed, the 0 2 in excess is not adsorbed.
204
(Rh-Ce)
+
1 second 0, pulse
This surface oxygen formation and removal are easy and the adsorbed oxygen acts as a promoter. Following the proposed mechanism by Tinkle and Dumesic (17) and discussed by Chinchen and Spencer (18), our results lead us to this mechanism, we explain the decrease of C 0 2 formation versus time: 0
I
Oxygen vacancies are introduced into ceria lattice at sites near noble metal crystallites as suggested by Harrison (3). Reactions 1 and 2 are reversible, then an anion vacancy is formed as proposed by Kubsh et a1 (19) and the reaction 3 is irreversible. The water adsorption occurs (reaction 4) and we may think that there is a competition between reaction 5(1) (regenerative mechanism) and 5(2) where hydroxyl groups appear, in order to explain the fast decrease of WGS reaction when Ce is added to Rh catalyst. We know that the heat of adsorption of water on Rh is very high (20) thus the hydroxyl groups are adsorbed more strongly by Rh which means that reaction 6 is forbiden, Rh would more easily catalyse the formation of water which then leaves the metal surface. For Rh/A1203 catalyst, the interaction between Rh and the support promotes the formation of a stable oxide, an associative mechanism between CO and H 2 0 will occur which is slower than the above one, due to the fact that the surface oxygens are difficult to remove. Finally to try to understand the effects of oxidation-reduction treatments we may think that catalyst particles are submitted to microstructural changes as suggested by Schmidt et a1 (21). Adsorption of reactants and promoters will alter catalyst shapes and these redox treatments will lead to oxides formation
205
which spread over the support in the contrary to metals which do not "wet" the support. This dynamic of the catalyst particles has to be taken into account during these reactions; this reconstruction may also be induced by adsorbates as proposed by Somorjai (11,22) and G. Lindauer et a1 (23,24). In that complex situation, some EXAFS experiments are going to be performed to try to understand the environment of the metallic particles of the catalyst. ACKNOWLEDGEMENTS
This work was carried out within the "Groupement Scientifique Pots Catalytiques" funded by the "Centre National de la recherche Scientifique", the "Institut FranGais du PCtrole" and the "Agence FranGaise pour la Maitrise de 1'Energie". 1. 2. 3. 4. 5. 6. 7.
8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.
22. 23. 24.
REFERENCES H.S. Gandhi and M. Shelef, in: Catalysis and Automotive Pollution Control, 199,1987 A. Crucq and A. Frennet (Editors)-Elsevier Science Publishers B. V. Amsterdam "Catalysisand Automotive Pollution Control",1987,CAPOCl. A. Crucq and A. Frennet (Editors)-Elsevier Science Publishers B.V. Amsterdam B. Harrison, A.F. Diwell and C. Hallett, Platinum Metals Rev. 32,(2),73 (1988) F. Garin and G. Maire, Acc. Chem. Res. 22,(3),100 (1989) In "Metalsupport interactions in catalysis,sintering and redispersion" edited by S.A. Stevenson, J.A. Dumesic, R.T.K. Baker, E. Ruckenstein, Van Nostrand Reinhold Catalysis Series (New york, 1987) R.K. H e n in "Catalysis under transient conditions" (1982), p.59, A.T. Bell and L.L. Hegedus (Editors) A.C.S. Symposium Series 178 K.C. Taylor in Catalysis: Science and Technology",5 , 119 (1984) J.R. Anderson and M. Boudart (Editors) - Springer - Verlag J.C. Schlatter, R.M. Sinkevitch and P.J.Mitchel1, Ind. Eng. Chem. Prod. Res. Dev., 22,51 (1983) J.C. Schlatter and P.J. Mitchell, Ind. Eng. Chem. Prod. Res. Dev. ,19,288 (1980) M. Weibel, unpublished results (PhD work) G.A. Somorjai, J. Phys. Chem. ,94,1013 (1990) H.C. Yao and Y.F. Yu Yao, J. Catal. ,86,254 (1984) R.K. Herz and J. A. Sell, J. Catal. ,94,166 (1985) Le Normand, C. Prieto, J. El Fallah, J. Majerus and 0. Touret, submitted in XAFS6 (York, 8,1990) R. Dictor, J. Catal. ,106,458 (1987) R.K. Herz in "Catalysisand Automotive Pollution Control", p.427 (1987), A. Crucq and A. Frennet (Editors) - Elseviers Science Publishers B.V. Amsterdam M. Tinkle and J.A. Dumesic, J. Catal., 103,65(1987) and references therein G.C. Chinchen , M.S. Spencer, J. Catal. ,112,325 (1988) J.E. Kubsh and J.A. Dumesic, AIChE J., 28,793 (1982) G. Zakumbaeva, Geterog. Katal, 4,( 1),241 (1979) L.D. Schmidt and C.P. Lee in "Catalyst deactivation",p. 297 (1987), E.E. Petersen and A.T. Bell (Editors) - Marcel DekkerJnc. New york G.A. Somorjai and M.A. Van Hove, Prog. Surf. Sci. ,30,201 (1989) G. Lindauer, P. LCgar6 and G. Maire, Surf. Sci., 126,301 (1983) G. Maire,P. LCgarC and G. Lindauer, Surf. Sci., 80,238 (1979) "
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysisand Automotive Pollution Control 11 0 1991 Elsevier Science Publishers B.V., Amsterdam
207
THE ROLE OF THE OXYGEN VACANCIES AT THE SUPPORT IN THE CO OXIDATION ON Rh/Ce02 AND Rh/Ti02 AUTOCATALYSTS. G. MUNUERA, A. FERNANDEZ and A.R. GONZALEZ-ELIPE
Instituto de Ciencia de Materiales (CSIC-Univ. Sevilla) and Dpto Quirnica Inorgcinica., P.O.Box 1115., 41071-Sevilla (Spain).
Abstract The interaction of 0 2 , CO and C0/02 at 400K with Ce02 and Ti02 supported rhodium (ca. 2.5-3% Rh by weight) has been studied combining XPS and TPD-MS. CO oxidation occurs on the metallic rhodium as well as on the support and, in particular, at the metal-support interface where oxygen vacancies at the support are involved and rhodium is readily oxidized up to Rh3+. Hydrogen incorporated to these oxygen vacancies during the reduction of the catalysts suppress CO adsorption on all three type of sites due to SMSI effects and vacancies filling. Ar+ sputtering on a Rh/Ti02 catalyst, to generate oxygen vacancies at the support, confirms their role in the oxidation of the rhodium in contact with CO/O2. In these type of catalysts a mechanism is proposed for CO oxidation at T>773K.
INTRODUCTION
In a series of papers (1-4) we have previously examined the RhRi02 system from the point of view of the interaction of the noble metal with the reducible Ti02 support after reduction with hydrogen at ca. 773K what leads to a "Strong Metal Support Interaction" state (i.e. SMSI), a phenomenon first described by Vannice et al. ( 5 ) more than a decade ago. In principle, this SMSI-state greatly enhances the selectivity of Rh/Ti02 catalysts to give oxygenates in Fisher-Tropsch synthesis (i.e.COLH2 reaction) (6,7) as well as their catalytic activity for NOx reduction (i.e. NOx/CO reaction) (8). From this previous work on Rh/Ti02, we have concluded that hydrogen is incorporated into the Ti02 support during the reduction step in the preparation of the catalysts (2,4) producing, in a first stage, a partially reversible (upon outgassing) SMSI-state, which prevent H2 and CO adsorption on the metallic rhodium. The transfer of such hydrogen involves "spillover" from the metal to fill the oxygen vacancies mainly generated at the metal-support interface during reduction, thus leading to the formation of hydride-like species (i.e. TiH3+). In a second stage at higher temperatures, these species also promote a great mobility of the support favouring the decoration of the metal particles with TiOx(H) moieties which strongly
208
interact with the rhodium probably through the same hydride species (9). In principle, Ce02 is a much easily reducible oxide than Ti02 and it is well known (10) that when heated in hydrogen at T<600 K, incorporates considerable amounts of hydrogen leading to bronze-like compounds while, the generation of SMSI in Rh/Ce02 catalysts reduced at 773K has been reported in the literature (11). On the other hand, ceria is an important component in modem autocatalysts (three way catalysts), which include Pt-Rh on an alumina stabilized support. The promoting effects of ceria in these type of catalysts is now well established, being generally accepted that oxygen vacancies, probably at the metal-support interface, play an important role on it (12). In this work the interaction of CO and C0/02 mixtures with CeO2 and Ti02 supported rhodium has been studied combining XPS with TPD-MS to assess the role played by oxygen vacancies at the two oxides in the activity of these catalysts for CO oxidation.
EXPERIMENTAL The Rh/Ce02 catalyst precursor was prepared by incipient wetness impregnation of Ce02 (Rhone-Poulenc, 109 m2/g) with a Rh(N03)3 solution, to get a loading of 3 % in weight of rhodium. The impregnated precursor was dried in air at 383K for 24 h. and then calcined at 673K in an oxygen flow for 4h and finally stored in a desiccator until its use. The preparation of the RhlTi02 samples (2.5% of Rh by weight) used in this work has been reported in our previous papers (1-4), together with a full characterization by TEM, EPR, NMR, XPS, TPR/TPD and H2 adsorption. XPS spectra were recorded in a LHS-10 spectrometer (from Leybold) working with pass energy constant at 50 eV and stored in a HP-1000-E computer on line to the spectrometer. B.E. reference was taken at the spurious carbon C(1s) peak taken as 284.6 eV. The samples, as pressed pellets, were placed on a Mo holder that could be heated resistively while controlling the temperature by a thermocouple spot-welded at the rear of the holder. All the treatments were carried out in the pretreatment chamber of the spectrometer (base pressure
209
outgassed (hereafter sample Rh/Ce02-H) or directly outgassed at 773K up to reach the base pressure (sample RWCe02-HV). Adsorptions of H2 (10 torr), 0 2 (0.1 torr), CO (1 torr) or C 0 / 0 2 mixtures (1 and 0.1 torr respectively) were carried out by contacting the samples with these gases for 30 min. at the appropriate temperature (400K unless otherwise stated) in the preparation chamber, prior to XPS and/or TPD recording. TPD were carried out in the analysis chamber of the spectrometer (base pressure
RESULTS R h / C e 0 2 reduction with H2 Fig. 1 shows XPS spectra of Rh(3d) and Ce(3d) of the calcined precursor and the two Rh/Ce02-H and RWCe02-HV samples reduced in hydrogen as previously stated, together with the TPD profile up to 973K for the Rh/Ce02-H sample and for the same sample after this first TPD experiments, once cooled down to room temperature (300K) and a new adsorption of H2 at this temperature was carried out. As can be seen in this figure, the XPS spectra show that the treatment in hydrogen at 773K produces a total reduction of the rhodium to its metallic state (B.E. 306.9 f 0.2 eV) while the Ce02 support becomes partially reduced as deduced from the new peaks in the Ce (3d) spectra at 885.1 and 903.4 eV due to Ce3+ (13). The intensity of these peaks being very sensitive to traces of water in the gas phase. Outgassing at 773K up to 10-8 torr of sample Rh/Ce02-H, to remove hydrogen and generate sample Rh/Ce02-HV, does not produces any appreciable change either of the Rh(3d) or Ce(3d) peaks of XPS spectra (except for a small broadening in the former), though the TPD-profiles show a strong desorption of hydrogen from the H-sample with a sharp maximum at ca.570 K. Removal of this hydrogen during the TPD run allows the adsorption at 300K of a new form of hydrogen on the same sample, with a TPD maximum at ca. 410K that should be ascribed to desorption from the rhodium particles. In principle, these results are similar to those we have previously reported for Rh/TiO;! samples (7) and can be explained as a loss of the SMSI interaction after the removal of the hydrogen incorporated to the support. A similar treatment for the Ce02 support leads to slightly lower reduction as detected by the new XPS peaks due to Ce3+.
210
I
, .
315
311 307
I
I
1
a) 1
913 901 889
B. E. /eV
Fig.1.- Right: TPD of hydrogen: (a)porn RhlCe02-H ; (b) after H2 readsorption of hydrogen at 300K at the end of (a). -Left: Rh(3d) and Ce(3d) XPS-spectra: (a) calcined RhlCe02 (b)RhlCe02-H; (c) RhICe02-HV.
Interaction of Rh/Ce02 with 0 2 Interaction with oxygen of both Rh/Ce02-H and RWCe02-HV occurs in a rather similar way. Fig. 2 shows XPS spectra for Rh(3d) and Ce(3d) for the HV-sample exposed to 0 2 in different conditions. It should be noted that even under the milder conditions used in these experiments (i.e. 0.1 torr 0 2 , 300K, 30 min.) the Ce02 support was fully reoxidized and the rhodium was also partially oxidized to Rh+. This behaviour of the metal contrast with what we had previously observed for Rh/TiO2 (14) where reoxidation under similar condition did not produce any change in the Rh (3d) spectrum. When the temperature was raised up to 400K a net increase in the intensity of the XPS peak corresponding to Rh+ species occurs, while Rh3+ was now clearly detected as a shoulder in the spectrum This corresponds, again, to an oxidation of the rhodium much more effective than in R h R i 0 2 samples, treated under similar conditions (14). Only under more severe conditions (i.e. 1 torr, 473K) Rh3+ becomes the main species, though no further change
21 1
could be observed in the Ce(3d) spectra once all the Ce3+ had been oxidized at the early stages of the experiment under milder conditions. _-_ Wl/CeO,,tfJ (Ads. 0,)
Rh ( 3 d ) Rh3+ 1 torr
I
, ! '\'
\
.,
I
,
.
ce_L?2
-,v 7. *J4$\,
\
F+
v\,
0.1 t o q ,
d)
\ \%
I
4
b#'
t
J
r; ;\ J 0 \
1
F \,c)
. , /: :*I
I
S$.&/
;
-.
Fig.2.Rh(3d) and Ce(3d)XPS-spectra: -(a)RhlCe02 -HV; -(b-d)after contact with 0 2 for 30 min. at the indicated conditions.
I
I I.
I \
?I
original!I
*
; I\ ; ;\
: . 1
7
-
7
.
'.-
r
315 311 307 B . E . /PV
1
\, a) -
1
r
1
,
913 901 8R9
B.E. /eV
R h / C e 0 2 interaction with CO and C 0 + 0 2 A set of experiments were (carriedout combining TPD-MS with XPS to get information on the interaction of CO and 0 2 with our samples. Once reduced and outgassed the excess of hydrogen (either at 300K or at 773K to generate H or HV samples), the samples were contacted with ca. 1 torr of CO at 400K for 30 min. XPS spectra were then recorded before starting the TPD run, and once the sample was cooled down "in vacuo". The same type of experiments were repeated using, ca. 1 torr of a mixture of C 0 + 0 2 (C0/02 ratio ca. 20) under similar conditions. Fig. 3 and 4 show the corrected TPD profiles for CO (m/e=28) and C02 (m/e=44). TPD profiles recorded after the interaction with CO, of the two samples H and HV, in Fig.3(A), suggest that the presence of hydrogen incorporated into the Ce02 support prevents in some way the adsorption of CO on this RWCe02-H catalyst though a small desorption still occurs at very high temperatures (>800K) while H2 (m/e=2) and traces of CH4 (m/e=15) cou1.d be detected in the gas phase during all the TPD runs.
212
ADS :co sample _ _ _ -m/e= 28
0
Rh/CeO 2-H
ADS :co
A
B
Sample m/e=28 ~-
/-
-
/
-
+
m
+
Y
Y
i
i
h
m
/
m
m
/
B
B
/ /
.rl
*rl
/
vl
v)
/
vl
z
\
P
a
H
.
&\
400
.
.
600
. 800
.
. lo00
400
600
800
lo00
T/OK
T/OK
Fig.J.-(A) TPD from RhlCe02-H and HV samples after interaction f o r 30 min. at 400K with CO. -(B) The same f o r Ce02-HV. ms :co+02 Sample
400
600
T/OK
800
B
ADS :co+02
A m/e=28,44
Sample Ce02-HV
1000
400
m/e=28,44
0 0
600
800
lo00
T/OK
Fig.4.-(A) T P Dfrom RhlCe02-H and HV samples crfter interaction f o r 30 min. at 400K with C0+02 -(B) The same f o r Ce02HV.
213
However, the previous removal of hydrogen greatly enhances the amount of CO desorbed from sample HV, which gives now a well defined peak at 525K and a very strong increase in the CO pressure from T>800K. The new peak at 525K should be ascribed to CO adsorption on the metallic rhodium, which appears now enhanced by the removal of hydrogen from the support (loss of the SMSI state (7)) However, the other form of adsorbed CO in this Rh/Ce02-HV sample is very different from that recorded for a similarly reduced Ce02 support, shown in Fig.3(B), where only a sharp single peak at 800K occurs. When the above TPD-profiles for CO are compared with those obtained after interaction with the C0/02 mixture in Fig.4(A), several new features appear. First at all, TPD desorption profiles of CO are now very similar for both samples, indicating that the presence of 0 2 (0.1 ton-) in the gas mixture has removed the SMSI-state induced by the hydrogen incorporated in sample Rh/Ce02-H. Secondly, besides the CO, a considerable amount of C02 is now also desorbed from both samples. However while for sample Rh/Ce02-H this C02 appears as two well defined maxima at ca. 525K and >900K, for sample HV a third broad peak centered at ca. 730K is observed. The position of this peak coincides with that of the single one recorded in the TPD spectrum of the Ce02 support which had been under CO/O2 under similar conditions, as shown in Fig. 4(B). Finally, the comparison of the intensities of the TPD-profiles after adsorption of CO and C0/02 indicates that, as a whole (i.e. evolved as CO+CO2), the presence of a small pressure of oxygen has greatly enhanced the adsorption of CO, which is now being desorbed either as such or as C02 during the TPD experiments. XPS spectra for sample HV, corresponding to the two set of TPD experiments described above, are shown in Fig.5 and were very similar to those recorded for sample H. In this figure we can observe that adsorption of CO does not modify the Rh(3d) and Ce(3d) peaks. However, when the samples have been in contact with the C0/02 mixture a total oxidation of Ce3+ readily occurs while the rhodium is almost completely oxidized to Rh+/Rh3+, the amount of Rh3+ being somewhat higher for sample HV than for sample H. By comparing data in Fig. 5 and Fig. 2, we can conclude that, in the case of Rh/Ce02-HV, a deeper oxidation of the metal to Rh3+ occurs when CO is present together with 0 2 , in spite of the much less oxidizing conditions (i.e.C0/02 vs 02). This suggests that CO plays a certain role in the stabilization of this oxidation state of the rhodium in our Rh/Ce02 catalysts.
214
315 311 307 B.E. /eV
913 901 889 R . E . /eV
315 311 307 B.E./eV
913 901 B . E . /eV
FigS.-Rh(3d) and Ce(3d) XPS-spectra of RhICe02-HV: -Right: after interaction with C0+02.-43:with CO (see text).
Rh/Ti02 interaction with 0 2 and C 0 + 0 2 In order to try to go further in the comparison of these RhKe02 catalysts with the R W i 0 2 samples, previously studied by us (1-4), a similar set of experiments with 0 2 and C0/02 were carried out using the same Rh/TiO2 catalyst (either in its H and HV state). TPD experiments show in this case a single broad peak with maxima at ca. 625 and 675K for CO and C02 respectively for both samples. Meanwhile, as shown in Fig. 6, reduction of the Ti02 support after heating with hydrogen at 773K cannot be detected by XPS (though Ti3+ ions can be recorded by EPR (3)).This fact indicates the lack of a massive reduction at the surface layers of the Ti02 support in our conditions, in contrast with the observed reduction of the ceria support. On the other hand, the Rh(3d) XPS-spectra in the same figure show that, while for sample Rh/TiOz-H oxidation of the metal does not occur either with 0 2 or with C0/02, a small amount is oxidized by the C0/02 mixture in sample Rh/Ti02-HV, thus suggesting again that removal of hydrogen from the Ti02 support leads to a somewhat easier oxidation of the metallic phase. To check further the role played by oxygen vacancies in this process one experiment was made using sample Rh/TiOa-H and doing a very soft
215
sputtering with Ar+ ions (30 s.,0.5 Kv) to generate a small amount of oxygen vacancies at the Ti02 surface by differential sputtering of oxygen (9). As we can see in Fig. 6 , this very soft etching produces Ti3+ (i.e. TiVo3+ sites) at the uppermost layers of the Ti02 surface. When 0 2 was introduced at 300K in contact with this sample, a broad shoulder corresponding to Rh+/Rh3+, could be clearly observed in the Rh (3d) XPS spectrum, while the small shoulder due to Ti3+ disappeared from the Ti(2p) spectrum. Furthermore, when the same experiment was made using the C0/02 mixture a much deeper oxidation of the metal into Rh3+ occurs, as shown in the same figure thus clearly indicating the promoting effect of the CO in the oxidation of the rhodium up to Rh3+. _ Rh (3d) Rh (-1 Ti ( 2 p ) a) m:02
1 b) ADS:CO+02 /I
I!
c) REDUCED
(hr+-aichd)
312
304 B . E . /eV
316
312
308
304
B . E . /ev
B . E . /eV
Fig.4.-Rh(3d) and Ti(2p)XPS-spectra of RhlTi02: -(a)after interaction with 0 2 ; - ( b )after interaction with CO +02; -(c) effect of Ar+ etching on the Ti(2p) spectrum. DISCUSSION The comparative assessment of the results for Rh/Ce02 above with those obtained for RhfTi02 in our previous papers indicate differences in the behaviour of the two systems which can be ascribed to the degree of reducibility of both supports and their further interactions with 0 2 and CO. One of the most important differences observed is in fact the partial
216
reduction of the surface of the Ce02 support at 773K in H2 as compared with the lack of such deep reduction in the Ti02 (i.e. compare XPS in Figs. 1 and 6). We had previously found (9) that hydrogen reduction of Rh/TiOz is accompanied by an enhanced ionic mobility of the oxide lattice leading to Ti3+ migration toward the bulk while oxide ions diffuse backward to the surface, thus reaching a rather oxidized steady-state at the surface. Since, in principle, oxide mobility in Ce02 should be faster than in Ti02 its rate of reduction with H2 should be much higher to exhaust the surface ,thus leading to a deeper reduced steady-state within a thick layer at the surface of the oxide. As it is well known, TPR of ceria (12) gives two reduction peaks, one at ca. 773K ascribed to surface reduction while the other at 1073K is due to reduction of the bulk with formation of lower oxides of cerium. The addition of rhodium seems particularly effective in promoting the surface reduction of Ce02 down to ca. 625K (12). So, in our experimental conditions a deep surface reduction should be expected as found by XPS. Meantime, in our reduction conditions hydrogen must be incorporated to the reduced oxide as previously found by Fierro et al. (10). Part of this hydrogen is probably incorporated by spillover to oxygen vacancies (Vo) near Ce3+, as we have previously found in Rh/Ti02 (4), according to: (CeVo)3++ Rh-H
**
(CeH)3+ + Rh
(1)
many of them remaining close to the interface with the support. XPS spectra in Fig. I show that removal of this hydrogen does not produce any noticeable change in the Rh(3d) or Ce(3d) peaks except for a small broadening of the former what is also observed for the Rh/Ti02 catalyst in Fig. 6(C). Nevertheless, as we had previously found for R m i 0 2 (7) its loss leads to an increase in the capacity for H2 and CO adsorption on the metal due to a partial recovery of the non-SMSI state. This fact is actually observed in Figs.1 and 3 where TPD profiles indicate the existence now of peaks at ca. 410K and 500K for H2 and CO, respectively, due to the adsorption of these two species of the rhodium particles. The reduced state of the support in RWCe02 (H and HV) is completely removed in contact with a small amount of 0 2 at 300K as shown in Fig. 2 for the sample HV. It is worthy of note that even at 300K the Ce3+ becomes completely oxidized while simultaneously an important amount of the rhodium is oxidized to Rh+ what contrast with the stability of this metal under similar conditions in Rh/Ti02 (14). This fact suggests that oxygen should be incorporated to the reduced ceria according to:
217
ce4+-0;-m+
(3)
in fact, formation of 0;and 0; has been observed on reduced Ce02 by EPR (10) and m - I R (15) after 0 2 adsorption showing a great reactivity even at T<373K. On the other hand, the interaction of only CO with the sample Rh/Ce02-HV (see Fig.3A) indicates that, besides the adsorption on the rhodium, a new type of adsorption of CO occurs when the metal is present leading to a TPD-peak at T>800K. This new form of adsorption does not occur in the Ce02 support where only a single sharp peak at 800K is observed (see Fig.3B) which is less intense in sample H where oxygen vacancies should be filled by hydrogen. These two facts suggest that the presence of rhodium generates a new type of very stable adsorbed CO which, in principle, might be tentatively ascribed to CO interacting with both Rh and CeVo3+ sites at the metal support-interface, as previously suggested by Burch et a1 (16) for NiRiOz catalysts and recently by Lavalley et al. (17), to explain the broad i.r.-band centered at ca. 1725 cm-1 which appears when Rh/Ce02 catalysts are contacted with CO. XPS spectra in Fig.5 indicate that the interaction of CO with these two types of sites occurs without redox exchange since both Rh(3d) and Ce(3d) peaks remain unchanged in these spectra. The absence of the TPD-peak of CO characteristic of the CeO2 support at ca. 800K in the two Rh/Ce02 samples deserves some further comment. Since, in principle, a deeper reduction should be expected in the later (12) it is likely that oxygen vacancies far from the metallic particles would rearrange either forming "shear planes" or migration toward the metallic particles thus reducing their number in sample HV, while they will remains filled with hydrogen in sample H. So, they will not be ready to adsorb CO in our deeply reduced RhKe02 Catalysts. However, when the two Rh/Ce02 samples have been in contact with the C 0 / 0 2 mixture , TPD profiles in Fig.4B first indicate that oxygen is able to remove the SMSI-state of the metal in sample H, which now adsorbs CO on the rhodium (TPD-peak at 500K) as well as on the sites at the metal support interface as also does sample HV. In addition a certain amount of this CO is now oxidized to C02 in these two types of sites giving TPD-peaks at 525 and 900K. However, it is worthy of note that a third TPD-peak of C02 appears at ca. 730K only for the sample Rh/Ce02-HV in these conditions. This peak would correspond to the one observed for the oxidation of CO on the reduced Ce02 support. This fact suggest that in sample H CeH3+ sites far from the metallic particles cannot be activated by C 0 / 0 2 in our experimental
218
conditions while sites CeVo3+ in sample HV are activated by the 0 2 . Looking at the XPS spectra in Fig.5 we can see that the COIO2 mixture reoxidizes all the CeVo3+ at the support while part of the rhodium is also oxidized mainly to Rh3+. A comparison of the XPS spectra in this figure with those included in Fig2 indicates that, after interaction with 0 2 at 400K,oxidation of the metal to Rh3+ is much more important when CO is also present in spite of the much less oxidizing atmosphere in this case. This suggests a promoting effect of the CO in the stabilization of this oxidation state, though, as shown by the XPS spectra recorded after the TPD runs, the initial reduced state of the Rh/Ce02 surface is completely restored as could be expected in the cyclic catalytic process. This fact can be explained assuming that, in the presence of CO, the reaction (3) above progress according to:
the initial state of the catalysts being restored, at high temperatures, according to: A Ce4+ O= O= Rh3+(C0)2 (CeVo)3+ + Rh + 2C02 (5) TPD The importance of oxygen vacancies at the support (i.e.CeVo3+ sites) close to the rhodium particles in the redox processes involved in the oxidation of CO with 0 2 at high temperatures can be better observed on Rh/TiO2 due to the greater stability of Ti02 for deep hydrogen reduction. In fact, the results in Fig.6 confirm that 0 2 alone is much less efficient than the C0/02 mixture to partially oxidize the rhodium particles and that the presence of CO stabilizes mainly Rh3+. Moreover, the presence of Ti3+ in the Ti(2p) spectrum of the Ar+ sputtered sample, which indicates the formation of oxygen vacancies ( i.e. TiVo3+ sites), enhances considerably the oxidation of the rhodium to Rh3+ particularly in the presence of CO. In our view these results gives a clear evidence of the important role played by oxygen vacancies at the metal-support interface in the redox chemistry of the rhodium, and probably other noble metals, in these type of catalysts under CO oxidation conditions.
...
*
Acknowledgements Authors thank the CICYT project nr.MAT88-223 for financial support.
219
REFERENCES 1 -G.Munuera, J.C.Conesa, A.Muiioz, V.Rives, JSanz and J.Soria, Stud.Surf.Science and Catalysis.,l7 (1983) 149 2 -G.Munuera, J.C.Conesa, P.Malet, A.Muiioz, M.T.Sainz andJ.Soria, Proc. 8th Intern. Congress on Catalysis 5 (1984) 217,Elsevier,Amsterdam 3 -G.Munuera,J.C.Conesa, P.Malet, J.Sanz and J.Soria, J.Phys.Chem.,SS (1984) 2986 4 -J.Sanz, J.M.Rojo, P.Malet, G.Munuera, M.T.Blasco, J.C.Conesa and J.Soria, J.Phys.Chem.89 (1985) 5427 5 -M.A.Vannice and R.L.Garten, J.Catal.,SO (1977) 228 6 -A.F.Solymosi, A.Erdohelyi and T.Bansangi, J.Catal.,68 (1981) 371 7 -A.Muiioz, A.R.GonzBlez-Elipe, G.Munuera, J.P.Espinos andV.Rives-Amau, Spectrochimica Acta (Part A),43 (1987) 1599. 8 -G.Munuera and V.Rives-Amau, Appl.Surf.Science., 6 (1980) 122 9 -G.Munuera, A.R.Gonz6lez-Elipe, J.P.Espinos, J.C.Conesa, J.Soria and J.Sanz, J.Phys.Chem.,91 (1987) 6625. 10 -J.L.G.Fierro, J.Soria, J.Sanz and J.M.Rojo, J. Solid State Chem., 66 (1987) 154 11 -J.Cunningam, S.OBnen, J.Sanz, J.M.Rojo, J.A.Soria andJ.L.G.Fierro, J. Mol. Catalysis.,57 (1990) 379 12 -B.Harrison, A.F.Diwel1 and C.Hallett, Platinum Metals Rev.,32 (1988) 73 13 -F.Le Normand, L.Hilaire, K.Kili, G.Kril1 and G.Maire, J.Phys. Chem.,92 (1988) 2561 14 -A.Muiioz, G.Munuera, P.Malet, A.R.GonzBlez-Elipe andJ.P.Espinos, Surf. Int. Anal.,l2 (1988) 247. 15 -C.Li, K.Domen, K-1.Maruya and T.Onishi, J.Cata1.,123 (1990) 436. 16 -R. Burch and A. Flambard, J. Catalysis 78 (1982) 389 17 -J.L. Lavalley, J. Saussey, J. Lamotte, R. Breault, J.P.Hindermann and A. Kiennemann. J. Phys. Chem.94 (1990)5941.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V., Amsterdam
22 1
PHYSICO-CHEMICAL PROPERTIES OF Ce-CONTAINING THREE-WAYCATALYSTS AND THE EFFECT OF Ce ON CATALYST ACTIVITY
John G. Nunan, Heinz J. Robota, Michelle J. Cohn? and Steven A. Bradley?
Allied-Signal Research and Technology t UOP Research Center 50 East Algonquin Road, P. 0.Box 5016 Des Plaines, Ill., 60017-5016 ABSTRACT The activity of Pt,Rh,Ce/y-A1203 and Pt,Rh/CeO2 catalysts has been studied in a full complement synthetic exhaust gas mixture consisting of H2, CO, C3Hg. C3H8, NO, 0 2 , N2, C02, H20 and S02. Direct interaction between Pt and Ce02 was shown to lead to large improvements in catalyst performance after activation of the catalyst in the synthetic exhaust gas. Catalyst activation was shown to be due to reduction of the noble metals and surface Ce and this state of the catalyst was found to be particularly effective for CO oxidation. The key component responsible for activation was shown to be H2, even at the low level present in the exhaust gas (0.26 vol.%), reduction is shown to be very facile. The degree of Pt/Ce interaction and thus activity after catalyst activation could be controlled by controlling the Ce02 crystallite size. Decreasing the Ce02 crystallite size led to greater Pt/Ce interaction as shown by TPR and STEM analysis and resulted in greater activity for both fresh and laboratory aged catalysts. Direct Pt/Ce interaction was also shown to lead to a synergistic reduction of Pt and surface Ce andthis feature of the catalyst was shown to qualitatively correlates with catalyst performance after activation.
INTRODUCTION
The effectiveness of Ce02 in three-way-conversion catalysts is well established. However the detailed role of Ce02 in activity enhancement is still a subject of competing hypotheses. Ce02 has been reported to act as an oxygen storage component [ l , 21, to stabilize "/-A1203 [3], to promote water gas shift activity [3,4], and to stabilize noble metal (NM) dispersion [5].Ce02 has also been proposed to promote CO oxidation activity. Several mechanisms have been proposed to account for this selective promotion. These include retardation of CO inhibition during oxidation and a reduced activation energy f o r CO oxidation [6,7]. Ce02 has also been proposed as a direct source of oxygen during CO oxidation [ 7 ] , and reduced Ce has been linked to the formation of highly reactive 0 2 - radical anions[8,9] that are very effective in oxidizing CO.
222
Several authors have also described synergistic interactions between NM and CeO2. Some investigators have observed that reduction of Pt, Pd and Rh is more facile when in contact with Ce02 [3,10]. Also coupling of NM and CeO2 reduction was observed with reduction of both components at lower temperature than when they are not in contact. Yao et al [lo] proposed that such synergism enhances oxygen storage by the catalyst. Previous work has focussed primarily on model reactions involving limited reactants using carefully activated catalysts. The present work addresses the performance of Pt,Rh,Ce/y-A1203 catalysts in a full complement synthetic exhaust gas. Further the catalysts studied were not initially activated by reduction before testing. We attempted to correlate some of the physicochemical properties of our catalysts with conversion performance in the synthetic exhaust gas. We describe the impact of exhaust gas composition and the degree of NM/Ce interaction on the light-off activity of fresh, calcined catalysts and laboratory aged catalysts. Attempts were made to systematically vary the extent of NM/Ce interaction during catalyst preparation by using catalysts of varying CeO2 loading and Ce02 crystallite size. We show that direct interaction between CeO2 and Pt leads to dramatic improvements in catalyst light-off activity after activation in the exhaust gas.
EXPERIMENTAL A wide range of catalysts was prepared where the Ce loading, its method of introduction and the CeO2 crystallite size were varied. Table 1 describes the catalysts studied, including Ce loading, NM content and the Ce02 crystallite size as measured by x-ray diffraction (XRD). The NM and Ce contents were measured by inductively coupled plasma - atomic emission analysis. Different routes were followed for different Ce02 sizes. Fresh catalysts having Ce02 sizes in the range 60 - 110 A were prepared by impregnating a y-Al2O3 support with various Ce salts and by using colloidal Ce02. Samples with Ce02 crystallite sizes of 110 -350 A were prepared by air calcining Ce(N03)3 and Ce(OAc)3 before milling the resulting Ce02 together with y-Al203. A highly sintered reagent Ce02 powder was used to prepare samples 10 and 12. Samples 12-16 were laboratory aged before testing in 10%H20/90%air at 900°C for 4 hours and the CeO2 crystallite sizes are given after aging. Following Ce02 introduction, supports were sized to 425-850 pm with wire screens. Pt and Rh were introduced by impregnation with aqueous solutiuons of H2PtC16 and RhC13, the catalysts were oven dried and finally calcined in air at 600°C for 6 hours. Pure Ce02 supports were prepared by precipitation of Ce(N03)3 using
223
(NH3)2C03 followed by filtration, drying and calcination at 600°C for 6 hours. NMs were added as described. TABLE 1 Summary of catalyst compositions used in testing.
iample
Catalyst Description
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Pt ,Rh/y-A 1203 6% Ce/y-A1203 Pt,Rh,6% Ce/y-A1203 Pt,Rh/y-A1203 Pt,Rh,6% Ce/y-A1203 Pt,Rh,6% Ce/y-A1203 Pt,Rh,6% Ce/y-A1203 Pt,Rh,6%Ce/y-A1203 Pt,Rh,6% Cely-Al203 Pt,Rh,6% Ce/y-A1203 Pt,Rh,24% Ce/y-A1203 Pt,Rh,25% Ce/y-A1203 Pt,Rh,25% Ce/y-A1203 Pt,Rh,25% Ce/y-A1203 Pt,Rh,25% Ce/y-A1203 Pt,Rh/y-A1203
Metal content
Ce02 Crystallite Size(l11) A
Pt wt % Rh wt% Ce wt % 0.045 0.745 61 k 2 5.52 61 zk 2 5.52 0.044 0.7 1 0.80 0.033 72f 2 5.60 0.76 0.040 113 f 3 5.67 0.76 0.04 1 145 f 4 5.67 0.76 0.040 213 k 7 5.70 0.79 0.04 1 332 f 16 5.62 0.78 0.040 =1000 0.79 4.99 0.048 24.1 0.066 65 f 5 0.343 22.6 =1000 0.82 0.043 22.2 270 f 5 0.034 0.50 130 k 5 21.5 0.042 0.77 120 f 5 21.2 0.83 0.048 0.80 0.033
Finished catalysts were evaluated in a full complement synthetic exhaust gas consisting of H2, CO, C3H6, C3H8, NO, 0 2 , N2, CO2, H20 and S02. Catalysts were tested in a tubular reactor using l g of catalyst and a flow rate of 5 l/min. Catalysts could be exposed to rich (equivalence ratio (ER) 2.0), lean (ER=0.5) or stoichiometric mixtures. Different testing cycles were used with or without S02. Without S02, catalysts were heated at 5"C/min in the stoichiometric mixture to 450°C (Rise-1), held at 450°C for 0.5 hour, cooled to 5O"C, then reheated to 450°C in the stoichiometric mixture (Rise-2). Testing with SO2 was similar except that the maximum temperature for Rise-1 and Rise-2 was 600°C which was held for 1 hour. During the hold and drop between Rise-1 and Rise-2, catalysts could be exposed to various combinations of rich, lean, and stoichiometric exhaust gas mixtures.
224
The catalysts were characterized using XRD, x-ray photoelectron spectroscopy (XPS), scanning transmission electron microscopy (STEM) and temperature programmed reduction (TPR). Both TPR and XPS measurements were carried out on fresh samples and on samples that had been treated in the synthetic exhaust gas. Samples analyzed after pretreatments in the exhaust gas were removed from the reactor in N2 (02 content = 50 ppm) and the TPR was carried out without further pretreatment. RESULTS
Testing of the Pt,Rh,Ce/y-A1203 catalysts showed that they were subject to large activity changes following exposure to exhaust gas, especially when Pt, Rh, and Ce were all present. Some activity change was expected as the catalysts were not initially activated by reduction. However, the considerable variation in catalyst activity related to exhaust gas composition was unexpected. The nature of the activation and the experimental factors that control its onset were studied in detail over sample 11. This was done by exposing the catalyst to varying combinations of rich, lean and stoichiometric gas after Rise-1 and studying the effect on the Rise-2 light-off. The effect on CO oxidation during Rise-2 light-off is shown in Figure 1. It is evident that Rise-2 light-off activity is very sensitive to the exhaust gas composition seen by the catalyst prior to Rise-2. The lowest Rise-2 activity occurs after a lean hold/lean drop whereas the highest activity occurs after a rich hold/rich drop and, after a lean hold/rich drop. Clearly, catalyst activation is associated with exposure to rich exhaust gas. Activation of the catalyst after a lean hold by the rich drop further showed that the drop is critical in the activation. Further, a rich hold followed by a drop in N2 resulted in activation equivalent to a rich hold/rich drop. Thus, the rich drop can either activate the catalyst or maintain the catalyst in the activated state. The drop between Rise-1 and Rise-2 takes about 10 minutes indicating that activation is accomplished rapidly. The gas components contributing to activation were determined by removing various components one at a time from the rich drop, followed by another ramp in the full complement gas. Removal of CO, hydrocarbon (HC) and 0 2 had no effect on catalyst activation. However, subsequent removal of H2 was found to have a dramatic effect. Figure 2 compares Rise-2 light-off for CO oxidation with C02, NO, H 2 0 and H2 present during the drop to the case where only C02, NO, and H20 remain. Clearly, removal of H2 leads to complete loss of activation. Similar changes in catalyst performance were observed for HC and NO conversion. Apparently, catalyst activity can be dramatically altered by a low concentration of H2 (0.26 ~01%)in a relatively short time period.
225
0
Leon Hold / Leon Drop
0 Rich Hold / Lean Drop
lean @'dJ_Ric_h W E .
0
A Rich Hold / Stoich. Drop
RichHold(-~~~~~?~--.
100
90
80
c0
70
t
.3
v) 60
k
$
u8
50 C
1
I
I
'
40
30 20 D
0
100
I50
200
250
300
350
400
450
Inlet Temperature,"C
Figure 1 :Effect of varied in-situ conditioning in the SOz-free synthetic exhaust gas mixture on the Rise-2 light-off activity for CO. Catalyst: PtRh,24wt.%Cel yAl2O3; Pt = 0.343wt.%; Rh = 0.066wt.% Stoich.Hold / Rich Drop 0
GO5 N_Oo,&O,H~
is d
p
0 Con,
NO, H.0 in drop
~
110
100 90 80
fi
0
70
.4
L1
p
50
u
40
t g
so 20 10
0 -10
I 100
'
'
'
uo
200
250
300
350
400
450
Inlet Temperature,' C
Figure 2 :Removal of H2 during the drop results in degraded Rise-2 CO conversion activity. Catalyst: PtRh,24wt.%Cel yAl2O3; Pt = 0.343wt.%; Rh = 0.066wt.%
226
The effect of Ce02 crystallite size on catalyst performance was studied both with and without S02. Fresh catalysts 4-10 were tested without S02. Rise-1 and Rise-2 activities are compared in Figures 3 and 4 . The catalysts were exposed to a stoichiometric mixture after Rise-1. Relative activities are compared using the temperatures at 25% CO, HC and NO conversion plotted vs CeO2 crystallite size. Ce02 crystallite sizes shown were determined by XRD line broadening of the (111) reflection. As evident in Figure 3, Ce02 size has no effect on Rise-1 activity. However, this is clearly not the case for Rise-2. Figure 4 reveals dramatic improvement in light-off activity for all three exhaust gas components with diminishing Ce02 crystallite size, especially for sizes less than 250 A. It is also observed that the activity of the Pt,Rh/y-A1203 sample is less than that for the Ce containing samples. Further a comparison of Figures 3 and 4 shows that performance improvement is not the same for all components. The promotion of CO conversion (AT =: l5O0C) is appreciably greater than that for HC (AT = 100°C) or NO (AT = 100°C>. Since Ce02 size has essentially no effect on Rise-1 activity, it is also clear that Ce02 size is affecting the degree of catalyst activation. TABLE 2
Rise-2 light-off activity over a series of aged Pt,Rh, 25wt. % Ce/y-A1203 samples having varying CeO2 crystallite sizes. Activity expressed as the temperature for 25% conversion. Pt nominal loading = 0.77wt.%; Rh nominal loading = 0.04wt.%
Laboratory aged samples were tested with S02. Catalysts were exposed to stoichiometric exhaust during the hold and drop following Rise-1. Rise-2 light-off activities are summarized in Table 2. Also shown are the CeO2 crystallite sizes after aging and testing. Clearly, large improvements in Rise-2 activity are observed for all three componets with decreasing Ce02 size. An approximately 80°C reduction in CO, HC and NO 25% conversion temperature accompanies the reduction of CeO2 size from 270 A to 120 A.
221
250
I
o
100
zoo
300
IOO
500
600
700
CeO, Crystdlite Size
800
(A)
900
ma
ttoo
Figure 3 :Rise-1 activity as afinction of Ce02 crystallite size tested in a stoichioinetric S02-fiee exhaust gas mixture. Activiry is expressed as the temperature needed for 25% conversion. Catalyst: Pt,Rh,6wt. %Ce/ y-Al2O3; Pt = 0.77wt. %; Rh loading = 0.04wt. %
0
100
200
300
400
SO0
600
700
CeO, crystallite Size
800
$00
lo00
llDD
(A)
Figure 4 :Rise-2 activity as aJicnction of CeOz crystallite size tested in a stoichiometric SO2-free exhaust gas mixture. Activity is expressed as the temperature needed f o r 25% conversiorr. Catalyst: Pt,Rh,6wt. %Ce/ y-Al203; Pt = 0.77wt. %; Rh loading = 0.04wt. %
228
Several fresh samples with varying CeO2 size were analyzed by scanning transmission electron microscopy (STEM). It was observed that a larger fraction of the Pt was associated with Ce02 than expected based on the relative surface area exposed by Ce02 compared with y-Al2O3. Further, the fraction of Pt associated with CeO2 increased with decresing Ce02 size. Thus, for maximum Pt/Ce02 interaction, we need supports with the smallest Ce02 size. This observation suggests that the effect of Ce02 crystallite size is related to increased Pt/Ce interaction. In order to test this hypothesis, Pt was supported separately on Ce02 and y-Al2O3. The temperatures required to achieve 25% conversion are reported in Table 3. Unlike the previous tests, these catalysts were exposed to the rich exhaust during the hold and drop following Rise-1. It is evident that Pt alone supported on pure Ce02 undergoes large activations when exposed to the rich exhaust gas following Rise-1. It is further evident that the Rise-2 activities of the Pt/Ce02 samples are much greater than those for the Ptly-Al203 samples. Thus, the results shown in Table 3 and Figures 3 and 4 clearly demonstrate the beneficial effect of Ce02 in promoting catalyst activation in stoichiometric or rich exhaust gas. Further work showed that activations in the stoichimetric/rich gas mixtures were reversible upon exposure to lean exhaust gas conditions and could result in a net deactivation. TABLE 3
Effect of varying the support composition on Rise-1 & 2 light-off. Activity expressed as the temperature needed for 25% conversion. Testing was done in a stoichiometric exhaust gas mixture with a rich hold/rich drop following Rise-1. Pt = 0.6wt.%
Since activation of the Ce02 containing samples was clearly associated with exposure to H2 in the exhaust gas, several of the above catalysts were characterized using TPR. These included samples 1-3, 5, 8 and 10. The purpose of the analysis was to determine the effect of both Ce02 introduction and CeO2 crystallite size on catalyst reduction, Figure 5 shows the TPR spectra of the y-Al2O3 support, Pt,Rh/y-A1203, Cely-Al203, and Pt,Rh,Ce/yA1203 (samples 1-3). Quantitative H2 uptake values are tabulated in Table 4 .
229
0
100
200
300
400
500
600
700
800
Temperature, OC Figure 5 :Comparison of the TPR spectra for: (a) 3cAl203; ( b ) Pt,Rhl yAl2O3; (c) 6wt.%Cel y-Al20.3; (d) Pt,Rh,6wt.%Cel yAi2O3. Pt = 0.77wt%; Rh = 0.04wt.%. TABLE 4
Summary of quantitative H2 uptake as determined by TPR for: (a) y-Al2O3; Pt,Rh/y-A1203; (b) 6wt.%Ce/y-A1203; and (c) Pt,Rh,6wt.%Ce/y-A1203 Sample Description y - A1203 Support Pt, Rh/y - A1203 6 wt % Ce/ y - A1203 Pt, Rh, 6 wt % Ce/ y - A1203
Temperature Range 37 - 797 29 - 481 481 - 794 33 -439 439 - 560 560 - 736 736 - 795 35 -235 235 - 407 407 -790
Temperature Maximum 710 260 672 36 1 490 690 795 175 280 720
H2 Uptake m mole/g 0.064 0.085 0.069 0.06 1 0.034 0.070 0.090 0.099 0.030 0.184
230
The ?-A1203 substrate takes up some H2 at about 700°C. This feature is present in Pt,Rh/y-A1203 and Ce/y-A1203 over a similar temperature range. One broad, low temperature reduction, centered at 260"C, is assigned to reduction of Pt and Rh in the Pt,Rh/y-A1203 catalyst. Since the Rh content of these samples is very low, distinct features of Rh reduction were not distinguishable. The 0.085 mmoles of H2 consumed is consistent with the amount required for complete Pt + Rh reduction (0.083 mmoles). H2 uptake by the Ce/y-A1203 sample is shown in Figure Sc. Distinct peaks centered at 360°C and 490°C and a broad feature above 600°C are observed. The low temperature peaks have been previously assigned to surface Ce reduction[lO,ll] and the feature above 6000 is probably due to both H2 uptake by y-AI203 and some bulk Ce reduction. The TPR spectrum of Pt,Rh,Ce/y-A1203 is shown in Figure 5d. The spectrum is clearly not a superposition of thc previous two, but instead, has completely new features consisting af a sharp low temperature peak centered at 175"C, a shoulder at 280°C and a broad peak centered at 720°C. The shoulder at 280°C is likely due to reduction of the Pt,Rh on the y-Al203 component of the catalyst. We see no evidence for the surface Ce reduction peaks below 500°C and the Pt,Rh/y-A1203 peak is clearly attenuated, as shown by comparing H2 uptakes in Table 4. Further, the H2 uptake in the low temperature peak is greater than expected for exclusively Pt,Rh reduction. The absence of a distinct surface Ce reduction feature suggests that the extra reduction at low temperature is associated with surface Ce. Thus, this low temperature reduction feature appears to be a synergistic reduction of Pt,Rh + some surface Ce. Such a synergistic reduction in NM/Ce containing catalysts has been observed previously [3,10]. Again, we attribute the high temperature uptake to reduction of bulk Ce and uptake by y-Al2O3 The influence of Ce02 crystallite size on catalyst reducibility was also examined. Results for samples 5 , 8 and 10 are shown in Figure 6. Sample 1 is also shown for comparison. Changing the Ce02 crystallite size has a large impact on the TPR spectra, especially at lower temperatures. Decreasing the CeO2 size leads to a systematic increase of the low temperature peak at 175"C, which has been assigned to the synergistic reduction of Pt,Rh + surface Ce. The appearance of this peak and its growth with decreasing Ce02 size, follows the previously observed trend of improved light-off for smaller Ce02 size. The effect of catalyst preconditioning on the TPR spectra was studied in detail for a Pt,Rh/Ce02 sample. TPR was run on a fresh sample and after conditioning in rich and lean exhaust at 450°C for 0.5 hours. Results are summarized in Figure 7 and Table 5. Clearly, conditioning in rich or lean exhaust has a large impact on catalyst reduction. For the fresh sample we see a single sharp H2 uptake at 180°C which is about 4.5 times greater than that
23 1
100
0
200
300
400
500
600
700
800
Temperature, O C
Figure 6 :Effect of Ce02 crystallite size on the TPR spectra for a series of Pt, Rh, 6wt.%Cel y-Al20.3 samples. Pt = 0.77wt.%; Rh = 0.04wt.%. 7.5
-0.5
,
I
0
100
200
300
400
500
600
700
800
Temperature,'C
Figure 7 :Comparison TPR spectra for fresh, rich, and lean conditioned (450°C for 0.5 hours) Pt, Rhl Ce02. Pt = 0.6wt.%; Rh = O.lwt.%.
232
needed for Pt,Rh reduction. This suggests that most of the H2 uptake is associated with the CeO2 support. There is no evidence for distinct surface Ce reduction between 300-500°C as reported by other authors [3,10,11]. However, above 500"C, there is additional gradual H2 uptake. The H2 uptake in the temperature region between 500-650°C may be associated with some high temperature surface Ce reduction and uptake above 700°C with bulk Ce reduction. In the lean and rich treated samples, H2 uptake above 500°C is very similar to that of the fresh sample as seen by comparing the spectra in Figure 7 with the H2 uptake values in the temperature ranges 250-700°C and 700-800°C in Table 5 . Lean conditioning results in a moderate decrease in low temperature reduction with a distinct shift to 100°C. In contrast, rich conditioning causes a net decrease in uptake centered at 80°C. Thus, under rich gas conditions, extensive catalyst reduction occurs. As for Pt,Rh,Ce/yA 1 2 0 3 samples, we see no evidence of separate Pt and Rh reduction, suggesting that reduction of both noble metals occurs under the low temperature peak. TABLE 5 H 2 uptake determined by TPR for a fresh, rich, and lean
conditioned Pt,Rh/CeOz catalyst. Conditioning was done in the rich or lean synthetic exhaust gas at 450°C for 0.5 hours. Pt = 0.6wt.%; Rh = O.lwt.%
Pt loading : 3.08 10-2 mmole/g ; Rh loading : 1.00
mmole/g
Pt,Rh/Ce02 and Pt,Rh,Ce/y-A12Og samples were characterized using XPS after rich and lean conditioning in the exhaust gas at 450°C for 0.5 hours. The samples were removed under N2 and transferred to the XPS instrument. During the XPS measurements sample charging was compensated by using a low energy electron flood gun. The binding energies were then
233
internally referenced to the 01s peak located at 528.84 eV. Some of the XPS peak intensities were weak due to the low levels of NM present and only the peak positions for more intense peaks are reported The reported peak positions are accurate & 0.2 eV Peak positions and their assignments are summarized in Table 6. The lean conditioned Pt,Rh/Ce02 sample has peaks located at 75.79 eV (Pt4fsp) and 72.52 eV (Pt4f712) indicating the presence of oxidixed Pt [12] and at 74.53 eV (Pt4f5/2) and 71.26 eV (Pt4f712) following rich conditioning, indicating the presence of Pto [ 121. Also shoulders at 76.03 and 72.6 eV following rich conditioning indicate the presence of some oxidized Pt. The observed binding energies for the Pt4d peaks are also given in Table 6. The peak locations are consistent with oxidized Pt following lean conditioning and Pto following rich treatment. Pto peaks are expected at 314315.5eV whereas for oxidized Pt, the Pt4d5/2 peak occurs above 315.5 eV 1131. Rh was not detected by XPS in these samples. This may be due to coverage of highly dispersed Rh by Ce02. Both Pt/Ce02 [14] and Rh/CeO;! [ 151 have been reported to exhibit strong metal support interaction whereby the dispersed metal becomes covered by the support. TABLE 6
XPS analysis of Pt and Rh oxidation states for fresh, rich, and lean conditioned Pt,Rh,6wt.%Ce/y-A1203 and Pt,Rh/CeOz samples. For Pt,Rh,6%Ce/y-A1203; Pt = 0.6wt.%, Rh = 0.2wt.%.
x
SAMPLE PRETREATMENT
Pt,Rh 6 wt % Ce/ ')-A1203
Calcined
PEAK RATIOS
I
-
-
I -
Lean Rich
I 316.8 I 310.3
I
I
I
I
I 317.0 I 309.9
-
315.5
Pt,Rh/Ce02
Lean
72.5
75.8
315.7
Pt=0.60wt% Rh=O.lOwt%
Rich
71.3 72.6
74.5 76.0
314.2 316.5
308.0
I 0.074 0.053
+:I: 0.037 0.062
Results for the fresh Pt,Rh,Ce/ y-Al2O3 sample and after exposure to the rich and lean gas are also summarized in Table 6. For this sample the Pt loading was 0.6wt.% and the Rh loading was 0.2wt.%. The approximate Ce02
234
crystallite size was 60 A. Again evidence for oxidized Pt in the fresh and lean conditioned samples is provided by Pt4dgl2 peaks located at 316.84 and 3 17.01 eV respectively. Following rich conditioning, evidence for Pto is shown by a Pt4dgl2 peak at 315.5 eV. In this sample, Rh was also detected. Like Pt, it is oxidized in the fresh and lean treated samples. This is supported by peaks at 310.25 eV (Rh3d512) and 315.0 eV (Rh3dy2) for the fresh sample and at 309.9eV and 314.7eV for the lean treated sample [13]. Following rich treatment, Rh was found as reduced Rh with peaks at 308.0 eV and 312.8 eV [13]. The Pt/Ce and Rh/Ce peak ratios were also measured, and reveal a decrease in the Pt/Ce ratio but no change in the Rh/Ce ratio following gas treatments. Assuming no change in Ce02 dispersion which is supported by extensive STEM analysis, this indicates partial Pt sintering during the rich and lean treatments while Rh remains highly dispersed.
DISCUSSION We have found that catalysts formulated with CeO2 are very sensitive to their exhaust gas environmant and can undergo facile activation in a rich gas. The key components responsible for activation are Pt and Ce02. This is shown clearly by comparing Rise-1 and Rise-2 activities in Table 3 for Pt/Ce02 and Pt/y-A1203. We see that the Pt/Ce02 catalyst is less active during Rise-1 than Pt/y-A1203, whereas after activation the opposite is true. The light-off temperature for CO oxidation decreases 37°C over Pt/y-A1203 between Rise-1 and Rise-2, while over Pt/Ce02, the change is 250°C. Similar results were observed with S02, where, after activation, the Pt/Ce02 catalyst shows light-off advantages of 267°C for CO, 170°C for HC and 178°C for NO conversion to Pt/y-A1203. This demonstrates that Ce02 can be a far more effective support for Pt than ?-A1203 after exposure to rich or stoichiometric conditions. Activation of the catalyst is also very facile. This is shown clearly in Figure I where full activation of the catalysts was shown to occur during the rich drop after a lean hold at 450°C for 0.5 hours. During the rich drop the catalyst is exposed to 0.26vol.% H2 for a very short period of time. Figure I also shows that these catalysts can exhibit varying activity depending on their immediate exhaust gas history. Both the effect of Ce02 crystallite size on activity and the physicochemical properties of the catalysts suggest that Pt selectively migrates to CeO2 during catalyst preparation. Thus the Ce02 crystallite size effect reflects increasing Pt/Ce interaction with its subsequent improvement in catalyst performance. Direct evidence for this conclusion comes from STEM studies where Pt was found selectively associated with the CeO2, even though most of the catalyst surface area is y-Al2O3, Indirect evidence comes from
235 noting that performance features of the Pt/CeOz catalyst are most similar to the Pt,Rh,Ce/y-A1203 samples having the smallest Ce02 crystallite sizes. This is clearly seen by comparing performance results for Pt/Ce02 in Table 3 with the activity results in Figures 3 and 4. We see the same activation features in catalyst performance as observed for Pt/Ce02. Further evidence for extensive Pt/Ce interaction comes from the TPR studies outlined in Figures 5 and 6. Figure 5 indicates that direct Pt/Ce interaction leads to a synergistic reduction of Pt and surface Ce. This synergistic reduction was shown to be a function of CeO2 crystallite size in Figure 6 and to be very extensive for samples having small Ce02 sizes as shown in Figures 5 and 6. The results from the present work suggest that Ce02 is not an inert support but instead undergoes physico-chemical changes similar to the NM and that these changes relate to the catalyst achieving its optimum active state. These changes include reduction of the NM as observed by TPR and XPS and surface Ce reduction as suggested by TPR. NM and Ce reduction is coupled and appears related to the light-off performance of the catalyst after activation. This is supported by a comparison of Figures 4 and 6. The exhaust gas component responsible for bringing about catalyst activation is H2, which is present in low levels in the rich exhaust gas. These correlations suggest that NM in contact with reduced Ce plays a key role in catalyst performance and may represent the most active state of the catalyst. Direct evidence for the nature of the catalyst's active state comes from XPS and TPR. The XPS results in Table 6 indicate that the NM in fresh and lean conditioned catalysts are in oxidized states. Rich conditioning clearly leads to reduction of the NM. Since both fresh and lean conditioned catalysts exhibit low activity while rich conditioned catalysts have high activity, it appears that fully reduced NM result in the most active state of the catalyst. Evidence for reduced Ce was not found by XPS of the rich treated samples. This may have been due to partial reoxidation of the catalyst during sample handling or that insufficient reduced Ce was present for detection by XPS. The XPS peaks of Ce3+ tend to overlap with those of Ce4+ [ 13,161. Evidence for facile re-oxidation of Ce3+ in the N2 used for sample handling was obtained from in-situ reduction experiments in the XPS instrument. Clear evidence for Ce3+ could be obtained in the Pt,Rh/Ce02 sample following reduction at 300°C in 1 torr of H2. However, exposure of the reduced sample to the glove box N2 for 2 days at room temperature led to complete loss of Ce3+ features.(O2 concentration was found to be much greater than the 50 ppm values present during sample handling in the TPR). Since the XPS sample handling was done in N2 having higher 0 2 contents than for the TPR experiment, reoxidation of the reduced Ce could easily occur in preparation for the XPS measurement.
236
The TPR results provide further evidence for the nature of the catalyst's active state. TPR results in Figure 5 show that Ce not only promotes reduction of the NM but also undergoes a synergistic reduction with the NM. The fraction of Ce02 that couples with the NM reduction seems to be selectively the surface Ce component. This is supported by: (i) the dissapearance of the peaks assigned to surface Ce reduction when the NM are present; (ii) direct evidence for reduced Ce was observed by XPS following reduction at 300°C; and (iii) bulk Ce reduction is thermodynamically unfavorable at the low temperatures observed for synergistic reductions [ 171 A number of authors have reported H2 uptake by CeO2 [18] where H2 consumption was not necessarily associated with Ce reduction. However, the evidence here favors surface Ce reduction as the source of H2 uptake in TPR. It was further found that decreasing the Ce02 size led to greater coupling of NM/Ce reduction as shown in Figure 6, The TPR studies of the Pt,Rh/Ce02 before and after treatment in the exhaust gas give further insight into the nature of the NM/Ce interaction and its relation to the active state of the catalyst. For the fresh and lean conditioned catalyst we again clearly see evidence for synergistic reduction of the Pt and Ce. This is supported by the volume of hydrogen uptake in the low temperature reduction peak, as shown in Table 5, and the absence of surface Ce reduction peaks below 5OO0C, as reported by other authors [3,11]. After lean conditioning, less H2 is consumed by the fresh sample, along with a shift to lower temperature. A similar temperature shift is observed for the rich treated sample. The shift to lower temperature indicates more facile reduction after exhaust gas treatment. This may be explained by partial noble metal sintering in addition to removal of residual chloride. Sintering is supported by lower Pt/Ce XPS ratios following rich and lean conditing than observed in fresh sample in the case of the Pt,Rh,6%Ce/y-A1203 sample. It has been reported that reduction of partially sintered Pt is more facile than highly dispersed Pt [19]. The lower H2 uptake following lean gas treatment may be due to a combination of less Pt reduction and poorer coupling with CeO2 resulting from partial sintering. In contrast, the rich conditioned sample showed much lower H2 uptake indicating that considerable reduction of the catalyst had occurred. The small H2 uptake by the rich conditioned sample may have been due to reduction of surface 'PtO' species or incompletely reduced surface Ce. These observations suggest that the most active state of the catalyst after rich conditioning is reduced noble metal in contact with partially reduced Ce. This study shows that direct interaction between Ce02 and Pt leads to dramatic improvements in catalyst light-off activity. Since catalyst performance was measured under stoichiometric non-oscillating conditions, the impact of Ce02 cannot be explained by oxygen storage. Further
237
explanations such as promotion of the water gas shift reaction, promotion of NM dispersion, stabilization of the support cannot explain the large beneficial effect of Ce02 under the present testing conditions. An interesting feature of Ce02 promotion is the enhanced conversion of all the exhaust gas components. Also, fresh catalysts show exceptional CO light-off activity after activation. This is clearly seen by comparing Figures 3 and 4 and from the data presented in Table 3. Further, CO light-off is promoted to a greater extent than HC or NO light-off, especially in tests without S02, as seen by comparing light-off temperatures in Figures 3 and 4 and Table 3. Since CO can inhibit oxidation reactions over reduced Pt and Rh, the removal of CO via its effective oxidation at low temperatures may lead to more effective HC and NO conversion. In summary, our results indicate that the most active state of the catalyst is reduced surface Ce in contact with reduced NM which is particularly effective in CO oxidation. These conclusions are in general agreement with the proposals of Sass et a1 [8] and Oh et a1 [7] where the impact of CeO2 was on CO oxidation. For testing under more severe conditions, such as high temperature and in oscillating richDean exhaust gas mixtures, other factors, such as water-gas shift promotion or oxygen storage, may become important in catalyst performance. ACKNOWLEDGEMENT
The authors are indebted to their respective organizations for financial support, encouragement and provision of facilities. We acknowledge the following colleagues for their assistance at both Allied-Signal and UOP: Mark T. West for catalyst testing; David E. Mackowiak for catalyst preparation; Jeffry T. Donner and Sharon G. Varga for XPS analysis; Mary A. Vanek and Judy A.Triphahn for XRD analysis; and Tomatsu Imai for helpful discussions and encouragement. REFERENCES 12345-
678-
Gandhi, H. S., Piken, A. G., Shelef, M., and Delosh, R. G., SAE Paper No. 760201, 1976 Su, E. C., Montreuil, C. N., and Rothchild, W. G., Appl. Catal., 17, 75 (1985) Hamson, B., Diwell, A. F. and Hallett, C., Platinum Metals Rev., 32, 73, (1988) Kim, G., Ind. Eng. Chem. Prod. Res. and Dev., 21, 267 (1982) Sergeys, F., J., Masellei, J. M., and Ernest, M. V., W. R. Grace Co., U.S.Patent 3,903,020 (1974).; Hindin S. G., Engelhard Mineral and Chemical Co., U.S.Patent 3,870,455 (1973) Yao, Y. Y.-F., J. Catal., 87, 152 (1984) Oh, S. H., and Eickel, C. C., J. Catal., 112, 543 (1988) Sass, A. S., Shvets, V. A., Savel'era, G. A., Povova, N. M., and Kazanskii, V. B., Kinet. Katal., 27, 894 (1986)
238 91011121314-
1516171819-
Tarasov, A. L., Przheval'skaya, Shvets V. A., and Kazanskii, V. B., Kinet. Katal., 29, 1181 (1988) Yao, H. C. and Yao, Y. Y.-F., J. Catal., 86, 254 (1984) Johnson, M. F. L. and Mooi, J., J. Catal., 103, 502 (1987) Kim, K. S., Wingrad, N., and Davis, R. E., J. Amer. Chem. Soc.,93, 6296 (1971) Handbook of X-Ray Photoelectron Spectroscopy, Muilenberg, ed., Perkin-Elmer Corp., 1979 Meriaudeau, P., Dutel, J. F., Dufaux, M., and Naccache, C., in "Metal-Support and Metal Additive Effects in Catalysis", B. Imelik et al., Eds. Vol. 11, p.95, Elsevier, Amsterdam (1982) Cunningham, J., OBrien, S., Sanz, J., Rojo, J., M., Soriao, J., A., and Fierro, J. L. G., J. Mol. Catal., 57, 379 (1990) Shyu, J. Z., Weber, W. H., and Gandhi, H. S., J. Phys. Chem., 92, 4964 (1988) Bevan, D. J. M., J. Inorg. Nucl. Chem., 1, 49, (1955) Fierro, J. L. G., Soria, J., Sanz, J., and Rojo, J. M., J. Catal., 66, 154 (1987) McCabe, R. W., Wong, C., and Woo, H. S., J. Catal., 114, 354 (1988)
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 199 I Elsevier Science Publishers B.V., Amsterdam
239
STUDIES TO THE FUNCTIONING OF AUTOMOTIVE EXHAUST CATALYSTS USING IN-SITU POSITRON EMISSIONTOMOGRAPHY
K.A. Vonkeman, G. Jonkers and R.A. van Santen",
KoninklijkelShell-Laboratorium,Amsterdam, ABSTRACT Studying the reactions in an automotive exhaust catalyst is complex as the exhaust gases contain many components of which the composition constantly varies and the catalyst contains different active components. To elucidate reaction kinetics in exhaust catalysis, the application of a technique stemming from nuclear medicine, i.e. Positron Emission Tomography, has been developed. Short lived positron emitting nuclides 11C (t1E = 20 min), 13N (tin = 10 min) and l5O (tin = 2 min) have been used to synthesise actual reactant molecules such as 11C0, llC@., llCHq, 13N2, 1500,C'5O and C1500. The labelled reactants could be monitored in a catalyst bed right through the reactor wall using a Positron Emission Tomograph, with a resolution of 1 cm in place and 1 s in time. Hereby, a large amount of information on the residence time of the reactants is obtained directly from the catalyst surface under actual reaction conditions. The mechanism and the rate limiting step in the carbon monoxide oxidation on platinum under cold start conditions of a car was studied. A strong interaction was found between gas phase C 0 2 and CeO2 at the catalyst surface, whereby the oxygen atoms contained in lamce Ce02 exchange rapidly with the oxygen atoms of the adsorbed C02
INTRODUCTION Automotive exhaust gases contain carbon monoxide, nitrogen oxides, hydrogen, various hydrocarbons, sulphur oxides and a number of components in trace amounts, such as lead oxide and phosphorus oxides [ l ] . The composition of engine exhaust gases depends on many factors, such as type of engine in the car, lubrication of the engine, state of service of the engine, driving behaviour of the driver and, to a lesser extent, the type of fuel used. The main ingredients of most commercial automotive exhaust catalysts are platinum and/or palladium, rhodium and ceria [e.g. 2,3]. For these reasons the study of the functioning of automotive exhaust catalysis is complex. Here we report studies on the kinetics of CO oxidation catalysed by ceriumpromoted platinum.
240
Positron Emission computed Tomography (PET) is a relatively new 3D imaging technique emerging from nuclear medicine and is capable of mapping quantitatively the concentration of a positron-emitting tracer [4,5]. The application of positron-emitting nuclides in reactant molecules is a useful technique to study automotive exhaust catalyst kinetics both in-situ and noninvasively. The Positron Emission Tomography technique is in essence an extension of the transient techniques as it does yield direct information on the degree of occupation of the surface sites by reacting components. This information can be used to obtain the reaction rate parameters of the individual adsorption, desorption and reaction steps. The reactants or reaction products are labelled with positron emitting nuclides, in particular 11C, 13N and 1 5 0 . With the advent of Positron Emission computed Tomography as a research tool in nuclear medicine, the utilisation of short-lived positron emitting nuclides like 11C (ti12 = 20.39 min; 100% positron yield; Emax = 1.0 MeV), 13N (9.965 min; 100%; 1.2 MeV) and 1 5 0 (2.037 min; 100%; 1.7 MeV) has increased considerably. The positron is the anti-matter counterpart of the electron, which implies that upon their encounter the positron and the electron annihilate, i.e. they are converted into electromagnetic radiation. To conserve momentum two y-photons are emitted in opposite directions in this conversion process.(see Fig.1). The energy of the y-photons is dictated by the energy conservation law and is almost completely determined by the conversion of the mass (m) of the positron/electron pair into electromagnetic radiation according to E = mc2 = 51 1 keV.
INCIDENT POSITRON
Figure I :Schematic drawing of a positronlelectron annihilation event.
Owing to this annihilation process the distance over which a positron travels in the catalyst matter (p = 1 g/cm3) is very limited (ca. 1.6 mm for positrons with Emax = 1 MeV). However, the resulting y-photons may travel through a catalyst over elongated distances (half thickness value in catalyst
24 1
matter approximately 7.5 cm). The detection of positron emitters occurs indirectly by the registration of the 51 1 keV annihilation photons. This can be carried out using 3"*3" NaI(T1) scintillation detectors and with a positron emission tomograph equipped with Bi4(Ge04)3 (BGO) detectors. The use of positron emitters such as 11C in kinetic studies offers the following attractive aspects: - (1). Nuclides such as 11C, 13N and 150 can be incorporated into the actual reactant molecules and the resulting positron-labelled molecules are chemically not distinguishable from unlabelled ones. Isotope effects in the pre-exponential factors of adsorption and desorption rates are very small and will be neglected in this study. - (2). The state of the catalyst surface can be studied both in-situ and noninvasively, since the indirectly emitted radiation can be easily detected through a reactor wall (half thickness value in steel ca. 11 mm), thus yielding direct information on reaction kinetics. With another radioisotope of carbon, such as 14C, the product stream can be monitored, as the P--radiation originating from 14C cannot penetrate reactor walls. On the other hand, many surface science techniques can only yield direct information about the status of the catalyst surface at very low pressures. - (3). The utilisation of scintillation detectors as detection system for the annihilation photons offers the possibility of detecting coincidences and hence acquiring absolute positron emitter concentrations at a well-defined time and location.[4,6]. - (4). An advantage related to the use of short-lived radionuclides is their high specific activity, resulting in a high sensitivity of measurement with very small quantities of the labelled molecules (order of magnitude picomoles) in comparison with the amount of reacting molecules. Because of the short halflife of these radionuclides, there are relatively few problems with waste, product or environmental contamination. For more general information on the properties of positron-emitting nuclides we refer to Knoll [7]. EXPERIMENTAL
The Platinum-Ceria-Alumina Catalyst A standard highly dispersed platinum catalyst was synthesised on a y-alumina carrier containing highly dispersed ceria at the surface. This catalyst formulation is a simplification of a commercial automotive exhaust catalyst as it only contains one type of noble metal. The preparation and the characterisation of the catalyst is reported elsewhere.[8]. In Table 1 the characteristics of the platinum catalyst are summarised.
242
T
m
: Catalyst characterisation parameters. 3.9 g cat. in cat. bed.
Parameter support Particle size BET surface area Pore volume Ceria loading Ceria surface Noble metal loading Noble metal dispersion Maximum CO/Me ratio at surface CO ads. cap. of noble met. Noble met. surf. in cat. bed C02 ads. cap. of Ce02 of carrier C02 ads. cap. of CeO2 of cat.
A1203 / Ce02 30 - 80 mesh 111 m2/g 56 ml/g 3 0.6 %w highly dispersed on A1203 surface 12 %w 84% 1 2.4.10-5 mol/m2 1 m2 5.6.10-5 mol 3.1.10-5 mol
Positron-emitter-labelled Molecules The experiments have been carried out at the Institute for Nuclear Sciences (INW) of the State University of Ghent, Belgium.[9]. At the INW a cyclotron was present with which positron emitting nuclides could be produced. Expertise and equipment were available to synthesise gas molecules containing these positron-emitting nuclides, with a radiochemical purity greater than 99.9%. These purities were confirmed by means of (radio)gas chromatography. At the INW a NeuroECAT positron- emission tomograph (PET-camera) was available, which was used as detection system in the experiments described in this paper. At the INW the application of PET to the visualisation of industrial processes has successfully been explored for oil displacement in model reservoir rock and dehydration of water-in-oilemulsions using the NeuroECAT tornograph.[ 101. In this study the positron emitters 11C, 13N and 1 5 0 were used. With these nuclides the molecules 11C0, ClSO, 11C02, ClsOO, 1 5 0 0 and 13" synthesised, which were used in the pulse experiments. Details on the production of positron emitters and the synthesis of the positron-emitter labelled molecules has been reported elsewhere. The Experimental Set-up with the Reactor Oven Svstem Fig. 2 shows a schematic representation of the equipment in which the transient radioisotope experiments were carried out. The experimental set-up was identical to that described in our previous paper in this series.[8], apart
243
from the fact that an aluminium oven block was used and that the complete reactor oven system was now placed in the field of detection of a positronemission tomograph.(Fig. 2). INLET PULSE DETECTOR
I
;I
I
I '
I
I '
I
u
I
o
m
PULSE DETECTOR
II I/
r-/
xi
L RCC
i
]
I 11-Bank T O M O C W E
Fipure 2.: Schematic drawing of the experimental set-up.
The main features of the set-up and the experiments carried out are summarised below: - (1). Amounts of less than 0.1 nanomol of CO containing ca. 0.15 picomol (or 50 MBq) of 11CO or l l C O 2 in argon, were extracted from a leadshielded container. - (2). These amounts were taken up into a gas mixture, which was continuously flowing through the catalyst bed in the reactor, without disturbing the pressure or the chemical composition of this gas flow (ca. 0.3 pmol CO/s). - (3). The total gas flow of 40 mVmin STP, containing 1.0% CO, 0.5% 0 2 and 10%C02 was led over 3.9 g crushed catalyst, sieve fraction 30-80 mesh. This catalyst sample was placed in a 3/8-inch reactor tube (Hoke stainless steel, internal diameter 7 mm; catalyst bed length 14 cm). Under these conditions (100°C c T < 170°C; ambient pressure) the reaction rates were limited by kinetics and not by transport phenomena. 10% C02 was present in the gas mixture to suppress the effect of the strong adsorption of C02 to the ceria present at the catalyst surface, as described in the previous paper in this series [81.
244
- (4). The chemical composition of both reactant and product stream was determined using gas chromatography and mass spectroscopy. - (5). At the outlet, samples taken from the effluent stream were sent to a CO/C02-separating gas chromatograph equipped with a proportional counter for the detection of radiation. This counter was placed behind the Thermal Conductivity Detector (TCD). After separation of CO and C02 the amounts of 11CO and 11CO2 could be quantified so that the distribution of labelled l 1 C 0 and " C 0 2 in the effluent stream from the reactor outlet could quantitatively be determined in the course of an experiment. - (6). To ensure that complete steady state was reached at the moment the pulse experiments were carried out, the catalysts were submitted to the gas flow for at least 12 hours in advance. The Positron Emission TomograDh Positron emitters are detected indirectly by the recording of the 511 keV annihilation photons. For the photon detection Bi4(Ge04)3 (BGO) scintillation detectors are utilised, which are placed in detector banks in a positron emission tomograph. This tomograph was used for detection of radioactivity directly from the catalyst bed, while norma1.(3"*3") scintillation detectors were used for an accurate characterisation of the labelled reactor inlet pulse and reactor outlet signal.(Fig.2). For product identification of the labelled compounds a gas chromatograph equipped with a proportional radioactivity counter was employed. At the INW a NeuroECAT positron emission tomograph was available, which was used in this study. The utilisation of a NeuroECAT positron emission tomograph has the following advantages: - (1). A temporal resolution of about 1.2 s can be achieved in a detection mode, in which the scanning gantry is kept in a fixed position. - (2). A spatial resolution of about 8 mm Full Width at Half Maximum (FWHM) along the flow tube axis can be achieved, if the scanning gantry is kept in a fixed position. - (3). Twenty one measuring locations are equally spaced over the total length of the reactor tube. Furthermore, there is no detection sensitivity profile over the catalyst bed. Both characteristics of a positron tomograph are in contrast with those of the experimental method in which NaI (Tl) scintillation detectors were utilised, as reported in a former paper.[ 131. - (4).The pair of detectors, which records a coincidence event, determines the location of the centre of the volume from which the signal originates. This sensitive volume is determined by the spatial resolution and the radial cross-sectional area of the flow tube. - (5). A line source calibration of the tomograph determines a quantitative relation between the radioactivity in the sensitive volume (Bq/cm3) and the
245
response of the tomograph (counts/s). This allows us to obtain quantitative information on the concentration of labelled compounds in the catalyst bed. - (6). The presentation of the data as a 'reaction image' provides direct information on the distribution as a function of time and place of the labelled compound on the catalyst bed in the reactor. From this information, kinetic parameters can be derived with the help of simulations of the experiments by a mathematical model of the catalyst kinetics.
MODEL OF THE REACTION MECHANISM Reaction Mechanism of CO Oxidation bv 0 2 over a Pt/CeO~/Y-A17Q3 - - Catalvst
To simulate the reaction of CO oxidation by 0 2 over noble metal catalysts, a mathematical model of the reaction mechanism was constructed. Details on this mathematical model were reported in previous publications .[ 8,141. The model contains three reaction steps for CO oxidation over noble metal surfaces and one for C02 interaction with the ceria: CO adsorption at the noble metal surface (Me) CO desorption from the noble metal surface Irreversible dissociative 0 2 adsorption at noble metal surface Irreversible surface CO/O conversion and C02 desorption C 0 2 adsorption at the carrier (A) C 0 2 desorption from the carrier CO 02
+ +
MeCO+ A +
Me 2Me Me0 CO;!
f~
+
+
f~
MeCO 2Me0 2Me+C02 AC02
(1.1) and (1.2) (2) (3) (4.1) and (4.2)
The reaction rate constants for the adsorption steps (l.l), (2) and (4.1) are based on the sticking probabilities of the molecules to the surfaces. The desorption step (1.2) and (4.2) and the surface reaction step (3) are taken activated, using Arrhenius equations. As was reported by Yates et al. [15] and Oh et al. [ 161 the dissociative adsorption of oxygen can be modelled best to fit the data exploiting a first-order dependence of the fraction of free surface sites. The second-order dependence, which is expected from step (2) in the model was found not to agree in their experiments. Previous studies have shown that the oxidation of carbon monoxide by oxygen was promoted by the presence of ceria on the catalyst surface [8]. This is caused by an enhancement by ceria of the sticking coefficient for oxygen on the platinum surface.
246
RESULTS & DISCUSSION Reaction Images The data recorded by the tomograph can be displayed in the form of an "image" [13,17] in very much the same way as graphical representation is carried out in medical experiments with a tomograph, e.g. brain scans [11,12]. A "reaction image" is a 1D distribution of the labelled compounds along the axis of the tubular reactor. The place coordinate is plotted along the horizontal axis in the reaction image, with position 0 cm in the centre of the catalyst bed (Fig. 3, catalyst bed -7 to +7 cm).
Figure 3 . : A reaction image of the 1 5 0 0 pulse experiment over the PtlCe02IyAl203 catalyst under steady-state conditions. The intensity of the labelled compounds in the reaction image is indicated in the linearly scaled grey-bar; the lighter grey-shades represents the higher concentrations. The PRF curve represents the distribution of the labelled compounds over the catalyst bed along the axial position, during the 200th second after injection of the pulse into the system. The RTD curve shows the residence time distribution at position -3.2 cm in the catalyst bed.
247
Because of the high length (L)/diameter (D) ratio of the reactor tube ( L D = 20), gradients in the radial direction are neglected compared to those in the axial direction. The second dimension in the reaction image is formed by the time, which is plotted along the vertical axis. All data displayed in the reaction images are corrected for radioactive decay and are smoothed moderately [ 131. The lighter grey shades in the reaction image represent the locations with the higher concentrations of the radioactive label. The reaction image presented in Fig. 3 is the result of a 1 5 0 0 pulse experiment over the platinum catalyst in the standard CO, 02, C02 gas flow. The gas flow was from the left-hand side to the right-hand side in the reaction image and the pulse of 1 5 0 0 reached the entrance of the catalyst bed 14 seconds after the tomograph was started. Two cross-sections of the reaction image are given. A profile (PRF) curve is shown, which represents the distribution of the 1 5 0 label over the catalyst bed during the 200th second. An RTD curve is also given, which indicates the Residence Time Distribution at position 3.3 cm. This experiment will be discussed in more detail below. Dispersion in the catalvst bed
Fig. 4 shows the results of six pulse experiments. All six experiments were carried out under the same conditions of ambient pressure, 140°C and applying the Pt/Ce02/y-A32(33 catalyst, except for the experiment shown in Fig. 4a, in which the reactor tube was filled with Sic. During the pulse experiments shown, the standard CO, 02, CO2 gas mixture was continuously flowing through the catalyst bed. In the first experiment (Fig. 4a ) 13" was pulsed over a bed of inert Sic. As no interaction between 13" and the catalyst bed took place, Fig. 4a shows the behaviour of inert gas molecules. The gas phase flowed through the catalyst bed in 4 seconds. The intensity of the displayed signal was identical at all positions along the reactor axis in the catalyst bed due to the calibration of the detector pairs. This experiment indicates that no significant dispersion takes place in the catalyst bed so that the reactor can be described as a plugflow reactor. Simulations of the other experiments (which had a longer time scale than the experiment in Fig 4a) with the computer program confirmed that dispersion is not a significant effect in the reactor system. Irreversible Adsomtion of Oxvgen at Noble Metal Surfaces From Fig. 3 (Pt catalyst; 140°C; 76% conversion; pulse of 1 5 0 0 ) it can be derived that part of the labelled compounds reaches the reactor exit without delay and another part remains in the catalyst bed for a relatively long time (up to 700 s). The molecules that reach the reactor exit within 4 s flow with the gas-phase through the catalyst bed and do not adsorb. This is
248
confirmed by the results of an experiment with an inert catalyst, S i c Fig. 4a . Comparison of Fig. 3 with Fig. 4a reveals that the unconverted oxygen molecules flow undisturbed through the catalyst bed. The RTD curve given in Fig. 3 shows clearly the peak of 1 5 0 0 molecules flowing without delay through the catalyst bed, followed by the long tail representing the l5O species that have been adsorbed in the catalyst bed.
FiPure 4 : A set of six reaction images of pulse experiments over the PtlCe02/yA1203 catalyst. Under identical conditions (standard gas mixture and jlow rate; T = 140°C; 76% conversion), six diserent labelled compounds was pulsed over the were pulsed over the catalyst. In experiment a 13” catalyst, showing the behaviour of inert gas molecules. In b an 11CO pulse experiment is shown, in c ‘ICO2, in d 1 5 0 0 , in e Cl5O and in f C1500. Product identification of the labelled compounds at the reactor exit revealed that 24% of the 1 5 0 0 molecules flow without interaction with the catalyst bed through the reactor and leave the reactor unchanged, as 1 5 0 0 . 76% of the 0 2 is converted to C02 via reaction with CO. Fig. 3 indicates that this part of the 1 5 0 label remains for an extended period of time in the
249
catalyst bed. Product identification showed that the labelled compounds leaving the catalyst bed after the initial 1 5 0 0 peak (at times > 20 s after start of detection) are all in the chemical form of C15OO. This clearly indicates that indeed the oxygen adsorption at the catalyst surface is irreversible; after desorption the only possibility for an oxygen atom to leave the platinum surface again is via the CO-scavenging reaction, forming C02. When the oxygen adsorption would have been reversible, at least some 1 5 0 0 would have been found in the product gas at times >20 s. The proportional radioactivity counter used for product identification was very sensitive and allows detection of less than pro-mille quantities of the total amount of labelled compounds injected. The Exchange of C02 Oxvgen Atoms with Ceria Lattice Oxvgen The time scale of the reaction image of Fig. 3 indicates that the oxygen molecules, which adsorb at the platinum surface, remain for a relatively long time in the catalyst bed. After 500 seconds still significant amounts of labelled compounds are recorded at the catalyst surface. In Fig. 4 the reaction images of six pulse experiments are shown. In Fig. 4a the reaction image of inert gas molecules (13") is given, while the other five reaction images represent pulse experiments carried out over the platinum catalyst at 140 "C, at which 76% conversion was reached Fig. 4h - 4f. Identical conditions with respect to gas flow, composition of the gas mixture and catalyst were applied in all five experiments, but different labelled components were pulsed into the reacting system. In Fig. 4b the reaction image of an 11CO pulse experiment is given, while in Fig. 4e a pulse of C l s O was applied. Surprisingly, there is a considerable difference between both reaction images. The oxygen atoms of the CO molecules remain much longer in the catalyst bed than the carbon atoms of these molecules. This clearly points to a dissociation between carbon and oxygen atoms somewhere in the reaction mechanism of the catalytic process. Similar observations are made for the labelled C02 molecules shown in Fig. 4c (an 11C02 pulse) and Fig. 4f(a ClsOO pulse). Again it is found that the oxygen atoms remain much longer in the catalyst bed than the carbon atoms. In these C02 pulse experiments, the effect is even more pronounced than that observed for CO. The carbon atoms of the initial C 0 2 molecules all have left the reactor even before the first oxygen atoms of these C 0 2 molecules reach the reactor exit in this experiment. It was found that the difference between the residence time distribution of the oxygen and carbon atoms in the CO pulse experiments was dependent on the conversion, whereas it was independent of conversion in the labelled C 0 2 pulse experiments. In experiments where low CO conversions were reached in the catalyst bed, the overlap between the reaction images for 11CO
250
and ClSO was greater than at higher conversions. The importance of the ceria lattice oxygen atoms in CO oxidation reactions was reported by Jin et al. [18] and Daniel [19]. The interaction between C02 molecules and ceria present at the catalyst surface was addressed in previous publications [8,13]. This interaction follows an adsorption and desorption mechanism of the C 02 molecules at the ceria surface. From the experiments shown in Fig. 4 it was concluded that the adsorbed C 0 2 molecules dissociate at the ceria surface and associate to CO2 again before desorption. The dissociation of the C02 molecules can very well take place via a carbonate-formation mechanism. It was reported by Daniel [19] and Li et a1 [20] that carbonate groups are present at the ceria surface in catalysts. The following mechanism is proposed: - C02 molecules adsorb onto the ceria present at the catalyst surface; - the C02 molecules form carbonate groups at the Ce02 surface; - the ceria carbonate groups can dissociate to form C02 molecules and Ce02 again; the oxygen atoms incorporated in the C 0 2 molecule need not necessarily be the same as those initially adsorbed onto the ceria surface. A probability of 1 to 7 or 8 for an oxygen atom to form a C02 molecule with the same carbon atom that adsorbed initially was established from these experiments. As reported in a previous paper in this series [8], it was found that 34%mol of the ceria present in the catalyst formulation forms an adsorption site for CO2. It is therefore likely that all oxygen atoms of the ceria participate in the exchange mechanism with the carbonate groups at the surface. Barrault et a1 [2l]reported that ceria significantly enhances the rates of exchange reactions between lattice oxygen and the oxygen atoms of gasphase molecules. The 0 2 molecules which dissociatively adsorb to the platinum surface react with CO molecules and form C02, which desorbs from the platinum surface. When these C02 molecules are moving with the gas-phase through the catalyst bed, they get adsorbed to the ceria surface with formation of carbonate groups. These carbonate groups can dissociate again, to form C02 molecules, which desorb from the surface. The oxygen atoms exchange in this way with the ceria at the catalyst surface, and remain in the catalyst bed for an extended period of time. This mechanism is visible in the reaction image of the experiment in which 1 5 0 0 was pulsed over the catalyst, shown in Figs. 3 and 4d. The experiments were simulated with the mathematical model of the reaction kinetics. These simulations, together with those of experiments at different temperatures made it possible to establish the kinetic reaction parameters, such as sticking coefficients in the adsorption steps and activation energies and pre-exponential factors in the desorption and reaction steps. The results of these simulations are summarized and compared to literature values [22] in Table 2.
25 1
DULL
Values used for the kinetic reaction parameters in the model. Kinetic Reaction Parameters CO stick. prob. at metal Act. Ener. CO desorption CO desorp. pre-exp. fac. 0 2 stick. prob. at met. Act. Ener. CO/O reaction Pre-exp. factor reaction C 0 2 stick. prob. at Ce02 Des. const. C02 from Ce02
I
Values used in this stud
I
Values used by Lynch et al..[22]*) 2.10-5 9.103 K 1.8.10'O mol/m3/s 4.1O9 mol/m3/s 9.10-9 8.103 K 8.10'K Eam 5.8.1013 s-lm-' 5.8.1013 s-lm-l ko * 6.6.1 0-5 sco2 -* 7.9 s-1 Kdes
(*) No ceria present in this catalys
CONCLUSIONS
The use of positron emitters in catalytic studies makes it possible to establish kinetic parameters. A positron emission tomograph supplies a large amount of information about the catalyst surface. The concept of a 'reaction image' is a valuable tool for the display and analysis of the recorded data in this type of experiments. A clear relation between gas phase conditions and surface component concentrations can be established directly from the results obtained with the tomograph. Also the dynamics of the processes of adsorption, desorption and surface reaction under steady-state reaction conditions can be directly derived from the 'reaction images'. The reaction mechanism for CO oxidation by 0 2 on supported noble metal catalysts was studied and oxygen adsorption at the noble metal surface was found to be rate limiting. Also the irreversible dissociative character of the oxygen adsorption was demonstrated. In the CO oxidation by 0 2 under the reaction conditions prevailing during the cold start of a car, the surface of noble metals is predominantly covered by adsorbed CO molecules. This is an important observation for automotive exhaust catalysis as it inhibits the reaction to take place at lower temperatures. A strong adsorption of C02 to the ceria surface was found, which most probably leads to cerium carbonate formation. The oxygen atoms from the carbonates exchange with the ceria surface. The carbonates can dissociate again to form C02 molecules which desorb from the catalyst surface. The use of positron emitters and a PET camera in catalysis studies has been demonstrated. In principle there is no constraint in placing a complete automotive exhaust converter in a PET camera, passing real exhaust gases through it and pulsing positron-emitter labelled molecules into this gas
252
stream. In this way complicated aspects of the functioning of exhaust catalysts occurring under realistic driving conditions (such as the processes taking place at a cold start) can be studied in detail. ACKNOWLEDGEMENT We gratefully acknowledge the expertise and cooperation from Dr. K. Strijckmans and Dr. P. Goethals of the "Laboratory of Analytical Chemistry", Institute for Nuclear Sciences, Rijksuniversiteit Ghent, Belgium for the production of the radioisotopes, the synthesis of the labelled reactants and the help received in experimenting. We also gratefully acknowledge Dr. I. Lemahieu and Ing. K. de Kesel of the "Laboratory for Electronics and Metrology" of the same Institute for the help received in experimenting with the NeuroECAT PET camera.
REFERENCES 1. Williamson, W.B., Perry, J., Gandi, H.S. and Bomback, J.L., Appl. Catal. 15,277 (1985). 2. Taylor, K.C., Catalysis, Science and Technology, Anderson, J.R.and Boudart, M. Ed. jQ),119 (1984). 3. Walsh, M.P., Platinum Met. Rev. 106 (1986). 4. Phelps, M.E., Mazziotta, J.C. and Schelbert, H.R., Positron Emission computed Tomography and Autoradiography, Raven Press (1986) New York. 5. Webb, S., Medical Science Series, The Physics of Medical Imaging, Adam Hilger, Bristol(l988) 6. Hilger, A., The Physics of Medical Imaging, (S. Webb, Ed.), Bristol, U.K. (1988). 7. Knoll, G.T., Radiation Detection and Measurement, 2nd edition, John Wiley & Sons, New York (1989). 8. Vonkeman, K.A., Jonkers, G., Van der Wal, S.W.A., Oosterbeek, H., de Jong, J. and Van Santen, R.A., Part I, submitted to J. Catal. 9. Strijckmans, K., Annual Report 1987, Rijksuniversiteit Ghent, Lab. for Analytical Chemistry, Ed. Dams, R., p 99 (1988). 10. Van den Bergen, E.A., Jonkers, G., Strijckmans, K., Goethals, P., 407 (1989). Nucl. Geophys. 11. Williams, C.W., Crabtree, M.C., Burke, M.R., Keyser, R.M., Burgiss, S.G., Hoffman, E.J. and Phelps, M.E., IEEE Trans. on Nucl. Sci. NS-28, 1736 (1981). 12. Hoffman, E.J., Phelps, M.E. and Huang, S.C., J. Nucl. Med.24.245 (1983). 13. Vonkeman, K.A., Jonkers, G., Van der Wal, S.W.A., Oosterbeek, H., De Jong, J. and Van Santen, R.A., Part 11, submitted to J. Catal. 14. Vonkeman, K.A., Van der Wal, S.W.A., Jonkers, G. and Van Santen, R.A., submitted to Chem. Eng. Sci. 461 (1980). 15. Yates, J.T. Jr., Thiel, P.A. and Memll, R.P., J. Catal. 16. Oh, S.H., Fisher, G.B., Carpenter, J.E. and Goodman, D.W., J. Catal. 100,360 (1986). 17. Vonkeman, K.A., Van der Wal, S.W.A., Jonkers, G. and Van Santen, R.A., submitted to Chem. Eng. Sci. 18. Jin, T., Okuhara, T., Mains, G.J. and White, J.M., J. Phys. Chem.91, 3310 (1987). 19. Daniel, D.W., J. Phys. C h e m . 2 , 3891 (1988). 20. Li, C., Sakata, Y., Arai, T., Domen, K. Maruya, K. and Onishi, T. J. Chem. SOC.Faraday Trans. 1 8341,929 and m,1451 (1989). 21. Barrault, J. and Alouche, A., Appl. Catal. 58,255 (1990). 22. Lynch, D.T., Emig, G. and Wanke, S.E., J. Catal.97.456
m,
u,
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V.. Amsterdam
253
TPD AND XPS STUDIES OF co AND NO ON HIGHLY DISPERSED PT+RHAUTOMOTIVE EXHAUST CATALYSTS: EVIDENCE FOR NOBLE METAL-CERIA INTERACTION
P. Loof (la), B. Kasemo (la), L. Bjornkvist(Ib), S . Andersson(2) and A. Frestad(2)
( I ) Chalmers University of Technology, S-41296 Gothenburg, Sweden (a)Physics Department (b) Chemical Department (2) Svenska Emissionsteknik AB, S-44501 Bohus, Sweden Abstract Temperature programmed desorption(TPD) measurements of CO and NO, flow reactor measurements, and XPS analysis were performed on Pt,Rh and Pt+Rh supported on alumina, alumina+ceria and ceria. Ceria induces significant modifications in the CO, NO TPD spectra and in the activity of the catalysts, indicating noble metal-ceria interaction but no evidence for charge transfer between Pt and ceria was found in the XPS spectra. Successive TPD runs of NO showed that ceria-containing samples maintained their capacity to dissociate NO for longer time periods than catalysts without ceria and also that there was a gradual decrease in the low temperature N2 desorption peak associated with noble metal-ceria interaction.The observed influence of ceria on the CO TPD spectra and on the steady-state CO oxidation activity vanishes for samples exposed to SO2 demonstrating that SO2 poisons the noble metal-ceria interaction. This is probably associated with the formation of Ce3-sulphate, as demonstrated by XPS spectra, which blocks double bonded CO and/or dicarbonyl sites. Under oscillating conditions the SO2 poisoning is less severe. The noble metal ceria interaction may be associated with oxygen vacancies in ceria adjacent to the noble metal particles. 1. INTRODUCTION
Metal-support interaction and the effect of SO2 in catalytic reactions over supported metal catalysts has received considerable attention over the past years (1-12). In this work we have addressed these questions for the specific case of Pt,Rh and Pt+Rh supported on alumina, ceria and alumina+ceria. For earlier results on the role of ceria see e.g. (3,6). The main methods of characterization were temperature programmed desorption (TPD) of CO and NO and XPS (X-ray photon spectroscopy) of highly
254
dispersed platinum and rhodium. This is an extension of previous work (17). Kinetic measurements were made on the activity of the catalysts with and without SO2 in the feedgases. The results reveal significant metal-support interaction. 2. EXPERIMENTAL
2.1 Sample preparation Three different types of supports were used in this study: A1203 which contained the gamma and delta phases with a BET-area of 140 m2/g. High surface area Ce02 with a BET-area of 107 m2/g. (b) Ce02 on A1203 with a CeO2 content of 20 wt% and a total BET-area (c) of 120 m2/g. The A1203 was impregnated with an aqueous solution of Ce(N03)3, dried and calcined (5OO0C, 2 hr, air). Coated monoliths, ceria and alumina powder were impregnated with an aqueous solution of H2PtC16 and/or RhC13, dried and calcined at 500°C in air for two hours. The noble metal loading of the monolithic samples was 1.5 wt% Pt and/or.0.3 wt% Rh based on the weight of the washcoat support. The noble metal loading of the ceria and alumina powder was 1 or 3 wt% Pt or 1 wt% Rh. In the flow-reactor and TPD measurements monolithic samples (400 cells per square inch) were used. TPD measurements were also conducted with ceria powder samples. In the XPS measurements disc shaped pellets, obtained from powdered alumina and ceria, were used. (a)
2.2 TPD The TPD experiments were performed in a quartz flow reactor cell described previously (13). The reactor with the sample is connected on-line to a mass spectrometer which continously analyzes the composition of the gas sampled by a quartz capillary-type probe (14) positioned about 4 mm down stream the catalyst. The temperature of the reactor can be raised linearly up to 1000 K by an external heating coil. An inert carrier gas (Ar) flows continously over the sample at atmospheric pressure. The selected gas flow rates of the carrier gas were 60 ml/min and 30 mVmin for 1 gram and 0.5 gram samples, respectively. The catalyst temperature is measured by a O.lmm diam. chromel-alumel thermocouple located in one of the channels of the monolithic sample, or in the center of the powder in the case of ceria powder samples. Quantitative measurements were obtained by calibration of the mass spectrometer with known mixtures of CO,NO, etc in Ar. Control runs with
255
the reaction cell empty were performed at regular intervals to check that wall reactions did not influence the measurements. The gas mixtures used were ultra high purity 4% CO in Ar, 4% H2 in Ar, 4% NO in Ar and 2% 0 2 in Ar and 100% Ar, respectively (99.9997% purity). All measurements were performed at 1 atm. The samples (except for the samples exposed to S 0 2 ) were always pretreated in the following way, prior to experiments. First the sample was oxidized in 2% 0 2 in Ar for 5 minutes at 600°C to remove carbon contamination. It was then reduced in 4% H2 in Ar for the same time and temperature. The reaction cell was thereafter flushed with pure Ar for 1 minute and then cooled down to room temperature over a period of 7 minutes. CO was then adsorbed, by continous flow of the gas over the catalyst for 2 min. and NO by pulse injection into the carrier gas. After gas adsorption TPD runs were performed by heating the reaction cell by an external heating coil at a rate of 2"C/s. This rate was found an optimum compromise to avoid temperature gradients, to minimize readsorption effects, and to maximize the sensitivity. After CO adsorption the CO and C02-signals were continuously recorded during the TPD run, and after NO adsorption NO, N 2 0 and N2-signals were recorded. 2.3 XPS A Ce2(S04)3.xH20 standard was prepared by precipitation, mixing hot aqueous solutions of Ce(N03)3 and Na2S04. The precipitate was thoroughly washed with hot distilled water and subsequently pressed into disc-shaped pellets ( 13 mm diam.) in the wet condition. The pellets were dried at 120°C in air over night. The samples were characterized with X-ray powder diffraction and BET surface area measurements. For the XPS-analyses the powder was compressed in a die to disc shaped pellets. The Hewlett Packard ESCA instrument (5950A) used is equipped with an auxillary vacuum system. This includes a furnace (20+11OO0C) , in which the samples can be heat treated in a desired atmosphere. The oxidation of catalyst samples was made at 500°C and 200 mbar oxygen pressure and reduction was performed at 500°C and 40 mbar of hydrogen in nitrogen (250 mbar total pressure), for 15 min. Calibration The spectrometer was calibrated linearly against the Au 4f7/2 and Cu 2 ~ 3 1 2peaks at 84.0 and 932.7 eV respectively. The static charging of nonconducting samples was compensated for by the use of an electron flood gun, emitting low energy electrons to the sample surface (typically 4 eV and
256
0.2 mA). A small spot of gold was sputtered onto the reference samples to obtain an absolute binding energy reference (15). The accuracy of the calibration method used was checked by the reproducibility of repeated measurements and by correlating the measured binding energies (B.E.) of support signals, such as A12p and 0 1 s of y-Al2O3, with literature values. The accuracy has, in this way, been determined to better than f 0.2 eV.
2.4 Flow reactor measurements. This reactor consisted of a quartz tube placed in a furnace and a sample holder made of quartz (this is a different reactor than the one used for TPD). The gas mixture at the reactor inlet consisted of 0 2 , C02, H20 and N2. CO, C3H6, NO, SO2 and 0 2 were added at a mixing zone in the sample holder. All the gas flows were controlled by mass flow-meters. The water was pumped through a vaporizer. To simulate the lambda-oscillations part of CO and 0 2 were alternately added through two magnetic valves. They were controlled in such a way that the oscillation frequency and amplitude could be adjusted to predetermined values. The concentrations of the different gases at the reactor outlet were determined using conventional gas analysis equipment. In the beginning of a run the sample inlet temperature was kept at 150°C until the gas composition was stable. During a run, data were sampled with a sampling frequency of 1 Hz. Two types of tests were made:
(a)Steady-state conditions. The activity of the catalyst sample was measured as a function of reactor inlet temperature between 150°C and 450°C. The space velocity was 65.000 hr-1.Table 1 shows the gas composition. A first run was made to condition the sample which was kept at 450°C for 30 min. before cooling the sample to 150°C. In the second and third runs the activity was measured with 0 vppm and 20 vppm SO2 respectively. Table 1 Gas comDosition (steadv-state conditions). Nature of gas Concentration
CO 1.8%
02 0.24%
H20 10%
co2 10.9%
so2
(b)Oscillating conditions. The activity at a sample inlet temperature of 450OC was measured at five different gas compositions with a oscillation frequency of 0.5 Hz and a
257
space velocity of 65.000 hr-1.The mean gas compostion was characterized by the stoichiometric number S (16) according to:
s=
2 x ‘02’ “01 all in vol-%. [Col 9 [C3Htj1 Table 2 shows the gas composition. Note, that for a mean S-value of 1.03, the time-resolved S-values during the lean and rich pulses were 2.1 and 0.5 respectively. +
Table 2 Gas composition (oscillating conditions). 0.67
0.85
S-value 1.03
1.21
1.4
0.24H.2 1 0.63H.40 0.3 0.060 10.9 10 Balance
0.35f0.21 0.63k0.40 0.3 0.060 10.9 10 Balance
0.45*0.21 0.63f0.40 0.3 0.060 10.9 10 Balance
0.56M.21 0.63H.40 0.3 0.060 10.9 10 Balance
0.67f0.2 1 0.63k0.40 0.3 0.060 10.9 10 Balance
Gas 02*
co* NO
c3H6
co2 H20 N2
*Part of CO and 0 2 were alternatively added to the gas flow. 3. RESULTS
3.1 TPD of CO and NO The results in this section are essentially a summary of results presented elsewhere (17). The motivation for repeating them here in a condensed form is that they are vital complements to the new results in the following section and for the discussion. The reduced catalysts were saturated with CO in continous flow of CO/Ar at room temperature, before the TPD runs. Figs. 1 a-c show TPD spectra of CO after CO adsorption on Rh at room temperature supported on (a) Al2O3, (b) A1203+ceria, (c) ceria powder. Pt and Pt+Rh exhibit similar spectra (not shown). The dashed curve in Fig. l a shows the TPD spectrum from the blank support (A1203 and A1203+ceria exhibit similar CO TPD spectra indicating that ceria is inactive towards CO adsorption). The CO TPD spectra from Pt, Rh, and Pt+Rh on A1203 consist of 3 poorly resolved peaks centred around 100, 200 and 350OC-400OC. No significant change in the TPD spectra was observed after successive TPD runs, demonstrating that no appreciable sintering of the catalysts occured during the experiments. Fig.lb shows the spectrum obtained when 20 wt%
200
100
100
Temperature
PC)
Fig. 1. TPD spectra of CO afer CO adsorption at room temperature on Rh supported on : (a)Al20.3, ( b )AQO~+ceria,(c) ceria. Heating rate 2 U s .
40
300
200
300
Temperature rC1
Fig 2 Successive TPD runs of N2 and NO after NO room temperature adsorption on Pt supported on :(a-b)Al2O.3, ( c )A1203+ceria The first run was made on a reduced catalyst. No reduction was made between successive runs. Only weak or no N 2 0 desorption was observed.
259
ceria has been added to the A1203 support. The broad double peak in the region 200-4OOOC observed in the absence of ceria (Fig. la) is replaced by a peak around 470OC, indicating considerably stronger CO bonding which when combined with the results obtained with Pt and Rh supported on pure ceria powder (Fig. lc), can be associated with noble metal ceria interaction. The observed interaction between ceria and noble metals would be difficult to explain if the dispersion (i.e. the "size") of the noble metal particles were low since large particles would be "bulk like" and nearly unaffected by the support. Quantitative integration of the CO TPD curves in Figs. 1 a-c and assuming a chemisorption stoichiometry of 0.7 CO per Pt and Rh atom (which is close to the value observed for many CO-metal single crystal systems including the catalytically active noble metals (18)), yields dispersion values for the Pt, Rh and Pt-Rh samples supported on A1203 of 25, 60, and 35 %, respectively, and similar dispersion values for the samples supported on A1203+ceria. Thus the high dispersion indicates that a large fraction or all of the noble metal atoms are within a few atomic distances from the ceria support, making direct metal-support interaction possible. The value 0.7 CO per noble metal atom may be debated, and could be larger for small clusters e.g. due to the occurence of dicarbonyl species at high dispersion. However the conclusion of high dispersions remain unchanged although the absolute numbers may be smaller than above.
NO As described in a separate publication (19) A1203 adsorbs NO at a low rate. Therefore the NO adsorption had to be made by pulse injection into the Ar carrier gas. The pulse size was chosen so that most of the NO in the pulse was adsorbed on the noble metal particles, and only a negligable amount was adsorbed by A1203 or transmitted through the catalyst. The latter was controlled by the MS signal for NO accompanying the pulse injection. During the NO pulse injection a small N2 signal was detected simultanously with the NO signal indicating a small spontaneous NO dissociation followed by N2 desorption already at room temperature. This effect was more pronounced with Rh than with Pt and Pt+Rh catalysts. Addition of ceria to the catalysts enhanced this effect (a similar effect was also seen with pure reduced ceria samples) . Figs. 2 a-c show five TPD spectra of N2 and NO obtained by successive TPD runs on Pt/A1203 and Pt/A1203+ceria samples(Rh and Pt+Rh exhibit similar TPD spectra (not shown)). The first run was made on a reduced catalyst but no reduction was performed between successive runs. Only weak or no N 2 0 desorption was observed. The spectra 1 of Fig. 2 a,b representing the first run on the reduced catalyst show a minor N2 desorption plateau around 100°C and a major N2 desorption peak centered around 220-240°C. It
260
is clearly seen in Figs. 2 a,b how the NO signal increases simultaneously with the decrease in N2 signal after each run demonstrating that the oxygen atoms deposited after each run inhibit the capacity of the catalyst to dissociate NO. The addition of ceria to the alumina support has the following effects on the TPD desorption spectra (Fig. 2 c) (i) The main N2 peak appears at lower temperature (around lOO"C, see spectrum 1). N2 desorption was detected at nearly the same temperature (SOT) during TPD of NO on reduced blank A1203+ceria support and on reduced pure ceria but the amount was small compared to the samples containing noble metals. (ii) Although the main N2 peak moves to higher temperatures after successive TPD runs the amount of N2 desorbed is not significantly changed(see Fig. 2c) demonstrating that the poisoning effect of oxygen on NO dissociation is much weaker when ceria is added (i.e. larger oxygen doses are required to inhibit NO decomposition). This is attributed to oxygen spill over from the noble metal to the reduced ceria. Activity measurements for the C0+1/202+ C 0 2 and CO+NO+ C02+ 1/2N2 reactions for stoichiometric gas mixtures were performed on the various catalysts samples used in the TPD runs after the standard pretreatments. The most important observation, in line with the TPD results, is that ceria has a significant effect on both reactions for Pt, Rh, and Pt+Rh catalysts. The activity of the reduced samples is increased upon addition of ceria
3.2 XPS 3.2.1 The Ce3d XPS-spectrum The XPS studies focus mainly on the pure ceria supported catalysts. We therefore begin with a short description of the complex Ce3d spectrum. It has been demonstrated (20,21) that spatially extended f-orbitals, in e.g. Ce02, contribute to bonding. This property will give a valency mixing and should be responsible for the major differences between light lanthanide oxides and the transition metal dioxides. 0 2 p to Ce4f charge transfer processes have also been shown to add to the complexity of the Ce3d spectrum. The latter has received considerable attention in the literature (20-27). In this work standard spectra were recorded for reference and are presented in Fig. 3. The addition of Pt to a Ce02 support does not affect the Ce3d spectrum. Thus we have chosen to present the measurements on Pt/Ce02. The Ce3d spectrum of Pt/Ce02 samples shows six major peaks designated v for the Ce3d-j/2 and u for the Ce3dy2 transitions. Due to the uncertainity in peak assignment (22-27) it is at present preferable to rely on empirical information obtained by measurements on standard samples. Returning again to Fig. 3, we note that when Pt/Ce02 is partially reduced in
26 1
I
Ce 3d
I
920
I 910
I
I
900
I
880
890
binding energy (eV)
Fig. 3 Ce3d XPS spectra of (a) Ce2(S04)3,@) 3 wt% Pt/CeO2 oxidized in 0 2 at 5000C. (c) 3 wt % Pt/CeO2 partially reduced in H2 at 5000C.
Fig. 4 0 1 s and Pt4f XPS spectra of a I wt % Pt/CeO2 catalyst
in the oxidized and reduced condition,respectively.
262
H2, two new peaks, v' and vo appear. Simultaneously the peak v increases in intensity and v" decreases. Reductive treatment of Ce 0 2 in H2 also gives a chemical shift of the oxygen 1s peak of 0.6 eV towards higher B.E.,Fig. 4. Praline et al. (26) reported an 0 1 s shift of 0.7 eV when exposing a Ce-metal foil to controlled amounts of oxygen. The shift was attributed to the formation of CeO2 on top of the Ce2O3 surface oxide. The present study shows that the "reverse" process, surface reduction of Ce02 producing CeOx, where xe2, also yields an 0 1 s shift of equal magnitude. For these reasons it is obvious that no reliable internal standard is available for static charge referencing in ceria samples. However, cyclic reductiodoxidation of Ce 0 2 and Pt/Ce02, with gold spot reference, gives very good reproducibility for the respective 0 1 s binding energy so that it is possible to rely upon these values when the surface oxidation/reduction status is well characterized. (A more extensive treatment of XPS-analyses of Pt-ceria catalysts will be published in a separate paper (28)). The well resolved u"' peak can be used as evidence for the presence of C e W ) since this peak is absent for Ce(III) compounds (26,29). A reasonable explanation would be to assign this peak to the pure 4f0 finai state (22,24,25,30). However, regardless of peak assignment, the correlation between the appearance and intensity increase of the u"' peak and the formation of C e 0 2 has been nicely demonstrated through controlled oxidation of a Ce-metal foil (26). It is further demonstrated in Fig. 3 by comparing the trivalent cerium sulphate spectrum with the tetravalent oxide. 3.2.2 Surface chemical state of Pt/CeO?. - catalvsts The freshly prepared 1 wt% Pt/Ce02 catalyst was oxidized in situ in 0 2 at 500OC to minimize hydrocarbon contamination. The catalyst surface was subsequently analysed, and the Pt4f7/2 peak is found at 72.8 eV, Fig. 4 After sample reduction in H2 the peak shifts to lower B.E., 71.6 eV. The B.E. for the reduced catalyst is 0.4 eV higher than expected for metallic bulk Pt. Standard spectra of a polycrystalline Pt-foil recorded in this laboratory, gave a Pt4f7/2 B.E. of 71.2 eV in good agreement with previously published results (30,31). The small but significantly hgher B.E. for the Pt-metal peak could be interpreted as electron deficient Pr but also as caused by a reduced extra atomic relaxation (32) due to the extremely small particle size. Oxidizing the catalyst in 0 2 gives a chemical shift of 1.2 eV, which would mean an oxidation of Pt to PtO. Changing the reduction agent from H2 to CO also resulted in a metallic Pt state, with the same Pt4f B.E. as after H2 reduction.
263
It
Al2p
BINDING ENERGY lev1
Fig. 5. A12plPt4f XPS spectra of a 3 wt% PtlyA1203 catalyst, (a) oxidized in 0 2 at 500OC (b) reduced in H 2 at 500OC, ( c ) exposed to 2 vol% SO2 + excess 0 2 in N2 at 500OC for I h.
Analysis of the Pt4f peak obtained with y-Al2O3 supported samFles is difficult since the Pt4f peaks overlap with the strong A12p peak. After oxidation in oxygen it can be seen in Fig. 5a that the Pt4f peak is impossible to deconvolute properly. However, after reduction of the catalyst in H2, the Pt4f7/2 and A12p peaks are sufficiently resolved and the measured B.E. are 71.6 and 74.9 eV respectively, Fig. 5b. Thus no difference in Pt4f B.E. is found for the reduced Pt+alumina compared to the reduced Pt+ceria. Thus the metal support interaction caused by adding ceria, as shown in the TPD experiments, is not reflected in the XPS-spectra. Therefore, the Pt-ceria interaction does not seem to be a simple electronic perturbation of Pt. When exposing Pt-catalysts supported on Ce02, y-Al203 and Ce02/y-A1203, respectively, to a 2 % SO2 / 5%02 in N2 mixture the resulting Pt chemical state is the metallic state. This is illustrated in Fig. 5c for a 3 wt% Ptly-Al203 catalyst. We will return to the SO2 treatment later in a separate section.
3.3 Effects of SO2 exposure Since SO2 is present in exhaust gases from cars and is known to affect catalytic activity it was decided to further investigate its influence on the metal-support interaction caused by ceria.
7
100
200
300
400
Temperature
7
500
7
600
("C)
Fig. 6 TPD spectra of CO after CO adsorption at room temperature on Pt supported on A1203.i-ceria. (a) SO2 exposed, (b)fresh.
100
200
300
400
500
600
Temperature ("C)
Fig.7 TPD spectra of CO after CO adsorption at room temperature on Pt supported on ceria powder. ( a ) SO2 exposed and reduced at 7OO0C,(b)fresh
265
3.3.1 TPD of CO TPD of CO on Pt/ceria, Pt/ceria/A1203 and Rh/ceria/A1203 samples exposed to SO2 were performed in order to investigate if the influence of SO2 could be detected in the TPD spectra. The samples were exposed to SO2 in slightly different ways. The Pt/ceria sample was exposed to an atmosphere of 2 % S02/5%02 in N2 mixture at 500°C in a separate flow reactor system and then transferred to the TPD-reactor. The Pt/ceria/A1203 sample was the same as the one used in the activity measurements under oscillating conditions (S=1.03). The Rh/ceria/A1203 sample was exposed to 0.2% SO2 at 550°C for 15 min. directly in the TPD cell. The samples were reduced in 4%H2 in Ar at 600°C for 5min. prior to the TPD experiments (this step was omitted when the TPD spectrum b in fig. 8 was measured for the SO2 poisoned Rh/ceria/A1203 sample). Figs. 6,7 a,b show the resulting CO TPD spectra (upper panels) for P t / c e r i a / A 1 2 0 3 and Pt/ceria respectively. T h e spectra are dramatically different compared to samples unexposed to SO2 (lower panels). The high temperature peak (which indicates noble metal-ceria lbl interaction) was completely vanished for the Pt/ceria/Al203 sample and almost completely for the Pt/ceria sample. Figs. 8 a-c show the TPDcurves obtained from the Rh/ceria/A120 3 Id sample. The TPD curve in fig. 8a represents the fresh reduced sample and b,c are obtained from the SO2 exposed and the reduced sample (reduced after SO2 exposure) respectively. Temperature (“C)
Fig.8
TPD spectra of CO after CO adsorption at room temperature on Rh supported on AlzOj+ceria, (a)fresh, (b) SO2 exposed, (c) reduced.
266
The high temperature peak has vanished for the SO2 exposed and the reduced samples, as was the case for the Pt containing samples, demonstrating that the noble metal-ceria interaction is poisoned by SO2 and that the reduction treatment is not enough to restore it. More detailed results will be presented elsewhere (34). 3.3.2 Flow reactor measurements
Steady-state conditions Catalyst samples containing 0.7-0.8 wt% platinum, based on the weight of the support, were tested both for water gas shift (WGS)-and CO oxidation activity. The CO-conversion vs temperature is shown in Fig.9 for three different cases. By comparing runs number 1 and 2 it can be seen that the conditioning of the catalysts during the first run increases the activity at low temperatures in the second run, when ceria is present in the support. However this promotion effect of ceria is poisoned by SO2 as seen in run 3. Evaluation of the amount of 0 2 consumed at 450OC in the second runs showed that the maximum CO-conversion due to the CO + 0 2 reaction is 25% and 22% for the Pt/A1203 and Pt/ceria/A1203 respectively. Comparing these numbers with the observed CO-conversion at 450°C shows that ceria addition slightly increases the WGS-reaction. Comparing curves 2 and 3 shows that the CO-conversion decreases at high temperatures when SO2 is added to the gas mixture. The detrirhental effect of SO2 is mainly due to a decrease in the WGS activity. After run 3 the temperature was kept at 450OC and the flow of SO2 was shut off. For both catalysts tested, only a slight increase in activity was seen after 15 min.
Oscillating conditions. To illustrate how ceria and SO2 affect the activity of the Pt/A1203 catalyst, the conversions of CO, C3Hg and NO were measured as a function of the stoichiometric number. As can be seen in Fig.10, when no SO2 is present, the addition of ceria enhances the activity especially at the rich side of stoichiometry. When SO2 is introduced the conversion for all three gases decreases. The decrease is more pronounced at the rich side and when ceria is present in the support. 3.3.3 XPS-analvsis in connection with SO2 A freshly prepared, oxidized, 3 wt% Pt/Ce02 catalyst was exposed to an atmosphere containing 2 % SO2 /5% 0 2 in N2, at 500°C for 1 h. The
267
% 100
PWAl,O,
80
-
60
-
40
-
20
-
4/30 % CeOJAIAI,O, SO, free
SO, free
0
CO conversion with and without SO2 versus temperature
Fig. 9
for Pt/A1203 (lejl panel) and Pt/ceria/A1203 (right panel).
Flow reactor activity
dynamic conditions
effect of CeO, and SO2
Conversion % 100
80 NO SO free
-4.-
co so,
60
HC SO, 40
J
20 0
I
I
0.6
Fig. I0
PW30 % CeOJAI,O,
4/Al,O,
I 0,8
I
I 1
I
I 1,2
I
I 1.4
I S 0,6
I
I 0,8
I
I 1
I
I 1,2
I
I 1.4
CO-,HC- and NO-conversion with and without SO2 versus the stoichioinetric number Sfor Pt/A1203 (lefr panel) and Pt/ceria/A1203 (right panel).
268
resulting changes in surface composition were studied by XPS and the resulting spectra are shown in Figs. 5c and 11. In the Ce3d spectrum, Fig. 11, a doublet structure is found with the dominant peak at 887.4 eV. The S2p3/2 (Fig. 12) and 01s peaks have their B.E. at 169.7 and 532.7 eV respectively.
e v) 3
I
Z
W
C
b a
L
-
1
920
I
I
910
I
I
I
900
I
890
I
I
880
BINDING ENERGY (eV1
Fig. 11 (a) Ce3d XPS spectra of a 3 wt% PtlCe02 catalyst exposed to 2 vol% SO2 + excess 0 2 in N2 at 500OC for 1 h. (b) and (c)are recorded when the same treatment as in (a) has been followed by reduction in H2 at 500OC and 700OC, respectively, for 15 min. This is in good agreement with the Ce2(S04)3 standard spectum, Fig3.
By using this standard spectrum and normalizing the intensity quotients, ICe3dg/2/IS2p and ICe3dg/2/IO1S, respectively, it is possible to calculate the average atomic concentrations of Ce, S and 0 (31). This gives a Ce:S:O ratio of 1 : 1.06 : 4.35 or close to 2:2:9. Assuming that the ionic species present are limited to Ce3+, SO42- and 0 2 - the empirical formula for the surface composition becomes Ce2(S04)20. The sulphated samples were subsequently reduced in H2 at 500°C and 700°C respectively for 15 min and analysed. Reducing the SO2 exposed, Pt/Ce02 sample at 500°C produces only a minor loss of sulphate, Fig.12, (10% of the initial S2p intensity is lost). The Ce3d spectrum shows further that Ce still is in the trivalent state. Increasing the reduction temperature to 700°C causes a loss of all surface sulfur and the Ce3d and 0 1 s peaks correspond to Ce02-x (c.f. calibration section and Fig. 3).
269
r ul c
3
s z 2 ul
+w
z
I
I
I
175
170
165
BINDING ENERGY ( e V )
Fig. 12
Some preliminary XPS results from the S 0 2-exposed P t / A 1 2 O 3 and Pt/Ce02/A1203 catalysts have also been obtained. The surface sulphate groups formed on A1203 are reduced already at 500°C. For the Pt/Ce02/A1203 catalyst, the loss in S2p intensity upon reduction at 500°C corresponds mostly to the loss of aluminum-sulphate groups, the remainder being bonded to Ce(II1). However, it is interesting to note that the Ce3d spectrum, after 700°C reduction, shows that cerium remains as trivalent ions, in contrast to the behaviour of the pure ceria support.
S2p XPS spectra of a 3 wt% PtlCe02 catalyst exposed to (a) 2 vol% SO2 + excess 0 2 in N2 at 5OOOC fur 1 h. (b)and (c) treated as in (a) but subsequently reduced in H2 at 5OOOC and 7000C, respectively, for 15 min.
Evidence has also been found that sulphatation of ceria occurs not only as a surface process but in bulk. After reduction in H2 at 700°C of the sulphated Pt/A1203/Ce02 catalyst the apparent surface sulphate concentration is practically zero. But, after a consecutive oxidation in 0 2 at 500°C surface sulphate is again found indicating diffusion of bulk sulphate to the surface upon oxidation (i.e. oxygen-sulphur exchange takes place between the bulk and the surface). The Pt-crystallites are reduced to the metallic state on all three supports after the S 0 2 + 0 2 exposure. This is illustrated in Fig. 5 , which shows a comparison between an oxidized, a H2-reduced and a S 0 2 + 0 2 treated Pt/A1203 catalyst. After reduction of the catalyst in H2, Fig.3b and after SO2 exposure, Fig.5 c, the Pt4f7/2 and the A12p peaks are sufficiently resolved and the measured B.E. are 71.6 and 74.9 eV respectively. 4. DISCUSSION
The TPD-spectra of CO on Rh supported on A1203 consist of three overlapping peaks centered around 100,200 and 400°C (Pt and Pt+Rh exhibit similar spectra). These peaks may be attributed to the linear, bridge bonded, and dicarbonyl species respective€y,found in numerous infrared spectroscopy studies on supported Rh catalysts (35, 37-40). Dictor and Roberts (35) found that at 200°C the linear and the bridge bonded CO desorbed while the
270
dicarbonyl CO, associated with atomically dispersed noble metals, was stable at this temperature. The peak at around 400°C might therefore be attributed to a dicarbonyl species. The assignment of the peak is however uncertain. A second possibility which cannot be ruled out is that CO dissociates as suggested in (36) and then desorbs by C+O recombination during the TPD run. The addition of ceria changes dramatically the TPD spectrum of Fig. 1 and the result is a spectrum consisting of two weil resolved peaks at 80°C and 470°C (Pt and Pt+Rh exhibit similar spectra). This result is not in agreement with a study performed by Oh and Eickel (41) on Rh supported on A1203 and A1203+ceria. They observed no difference in the CO TPD spectra and IR spectra between the two types of samples. However Kiennemann et al. (37) found that when ceria is added to a rhodium catalyst a new absorption band appears at 1725 cm-1 in the infrared spectra. Similar results have been obtained with alkali additions to silica supported palladium (39) and for Rh promoted with metal oxides (40). The new IR band was attributed to a CO molecule bonded through both the carbon and oxygen atoms. The XPS measurements on reduced Pt samples show that the chemical state of the Pt particles is close to that of Pt metal and that no difference in Pt 4f B.E. is found for Pt supported on A1203 and A1203+ceria. Thus no electronic interaction was found by the XPS measurements.The ceria-induced shift in the TPD spectra of CO towards higher temperatures may therefore be attributed to an increase in the number of atomically isolated Pt and Rh atoms and/or the creation of a new type of sites where the CO molecules are bonded through the carbon atoms to the Pt or Rh metal particles and the oxygen atoms to the reduced ceria. The fact that the total amount of desorbed CO is almost unchanged upon addition of ceria supports the latter explanation. NO dissociates nearly completely on the reduced noble metals supported on A1203 or A1203+ceria. The results of sucessive TPD runs are a concerted decrease in the N2 signal and increase of the NO signal due to the dissociation inhibiting effect of the deposited oxygen. Ceria addition to the noble metal samples supported on A1203 alter the TPD spectra from a minor peak at 100°C and a major peak at 240°C to a peak centered around 100°C. Oh (42) performed TPD measurements of NO on Rh/A1203 amd Rh/A1203+ceria and obtained essentially the same results as ours. However, the extent of NO dissociation on the Rh/A1203 sample was less and addition of ceria enhanced only the low temperature peak (a peak centered around 300°C was also present in their TPD spectra). These differences are probably explained by differences in the reduction status of the samples. This interpretation is based on the observation that samples on which oxygen has been deposited, e.g. by repeated TPD runs, exhibit a lower degree of NO dissociation and also in the case of ceria containing samples, the appearance of two N2 peaks.
27 1
The inhibiting effect of oxygen on the NO dissociation is considerably smaller for the ceria containing samples, which we attribute to oxygen spill over from the noble metal to the reduced ceria. XPS-spectra reveal that the c e r i a in the reduced Pt/ceria and Pt/ceria/A120 3 samples are nonstoichiometric, thus containing oxygen vacancies. It is reasonable to assume that these vacancies are filled during the repeated NO-TPD runs, shown in fig. 2c. Note that no reductive treatment was made between the runs. Oh (42) proposed that the higher dissociation rate of NO which was observed on ceria containing Rh samples could be attributed to an adsorption site where the NO molecules are bonded through the N atoms to the Rh atoms and the oxygen atoms to the reduced ceria particles (similar type of bonding as proposed for the CO molecule). Although highly speculative the low N2 desorption temperatures obtained for the ceria containing samples could be explained by adsorption of NO on such sites. It could also be caused by oxygen spillover to the reduced ceria support,which prevents oxygen blocking of sites for NO dissociation. By comparing the TPD spectra of Figs. 2b and 2c one arrives at the conclusion that Pt is predominantly deposited on ceria on the Pt/ceria/A1203 sample. This conclusion is based on the observation that while the reduced Pt/A1203 sample produces one N2 peak centered around 240°C in the first run, no such peak is seen in the first run with the Pt/ceria/A1203 sample Knowing how the addition of ceria effects the TPD and XPS results it was of interest to try to alter the interaction between ceria and Pt and Rh. SO2 was chosen as an interesting probe molecule. As seen in Figs. 6-8 the effect of S02-exposure on the CO-TPD spectra is dramatic. The high temperature CO peak, associated with the atomically dispersed noble metals and/or the double bonded CO molecule, has disappeared and the total amount of CO desorbed has decreased. XPS spectra of Pt/ceria, Pt/A1203 and Pt on mixed supports show that Pt is reduced to the metallic state after S 0 2 + 0 2 treatment. Pt as a good SO2 oxidation catalyst converts Pt-0 to metallic Pt by SO2 oxidation. The oxygen excess present, is obviously not large enough to reoxidize the Pt-crystallites. XPS- analysises have shown the resulting surface composition of a Pt/ceria catalyst after SO2 treatment to be a trivalent sulphate or oxide-sulphate. This indicates that ceria is "locked" in the trivalent oxidation state and unable to vary its oxidation number and consequently its oxygen content. Furthermore, we have indications that this process is a bulk sulphating process in contrast to the SO2 exposed A1203 supported catalyst which shows only surface sulphate. When exposed to S02, the mixed ceria/A1203 support behaves much like the sum of the two, except that after reduction in H!: at 7Oo0C, trivalent sulphate is converted to trivalent oxide. Thus it seems to be difficult to recover the
272
C e 0 2 - x on A1203 after SO2 exposure. This could affect and reduce the promoting effect of ceria under real operating conditions. The disappearance of the high temperature peak in the CO TPD spectra after SO2 exposure is probably associated with the formation of Ce(II1)sulphate which blocks the double bonded CO and/or the dicarbonyl sites. The flow reactor measurements under steady-state conditions (fig. 9) demonstrate that the in situ conditioning of the samples at 450°C (after runl) gives an increase in activity at low temperatures for the ceria containing sample. It has been proposed (6) that ceria enhances the oxidation of CO under rich condition by donating lattice oxygen. However, when SO2 is introduced, this lattice oxygen is no longer available and the activity decreases. Note that this decrease in activity at low temperature is not seen when Pt is supported on Al2O3. TPD of CO showed that SO2 interacts strongly with the sites associated with the high temperature CO peak. It can therefore be argued that it is the dicarbonyl and/or the double bonded CO which reacts with lattice oxygen and forms C 0 2 especially at low temperatures. At temperatures above 350°C the WGS reaction contributes to CO-conversion. There seems to be a slight increase in WGS-activity when ceria is added to the support. This could be the reason for the larger decrease in activity for the ceria containing sample when SO2 is introduced. Under oscillating conditions, in a simulated exhaust, SO2 reduces the activity especially at the rich side of stoichiometry. This is true for both COYC3H6 and NO. Schlatter and Mitchell (10) found that, under oscillating conditions, SO2 decreased the WGS-activity. For Rh, they associated the WGS activity with a mildly oxidized state which was stabilized or more abundant when ceria was added to the support. Su and Rotschild (1 1 ) studied an engine dynamometer-aged TWC and found that, even in the presence of S 0 2 , part of the CO conversion under oscillating conditions, could be attributed to WGSactivity. Kim (43) noted that the promotion effect of ceria on WGS-activity was coupled with a decrease in the net NO conversion. The gross NO conversion, however, was not changed. This was taken as supporting evidence that hydrogen formed in the WGS-reaction is reactive towards NO. Grenoble et al. (44)concluded that the strong influence of the support on WGS-activity and steam reforming activity was due to the ability of the support to "activate water".
Summary Addition of ceria to Pt,Rh and Pt+Rh supported on A1203 changes dramatically the TPD spectra of NO and CO indicating significant noble metal-ceria interaction. The stability of adsorbed CO is increased and N2 desorption shifts to lower temperature. Successive TPD runs of NO showed that ceria-containing samples maintained their capacity to dissociate NO for
273
longer time periods then catalysts without ceria and also that there was a gradual decrease in the low temperature N2 desorption peak associated with noble metal-ceria interaction. Based on the XPS and TPD observations this was attributed to oxygen spillover from noble metal to oxygen vacancies in reduced ceria. XPS measurements on reduced Pt/A1203 and Pt/A1203/ceria samples do not reveal any large charge transfer between Pt and ceria since the Pt4f binding energy remains unchanged upon ceria addition. Thus the change of CO bonding upon addition of ceria is probably associated with either an increase in the number of dicarbonyl sites and/or with the creation of CO double bonded sites. NO bonded in the same way could be the cause of the low N2 desorption temperature obtained for the ceria-containing samples. The flow reactor measurements on supported Pt samples and the activity measurements performed in the TPD cell on noble metal samples both show that ceria addition increases the activity of the catalysts. The ceria induced shift in the CO TPD spectra completely vanishes for samples exposed to S02, demonstrating that the noble metal ceria interaction is poisoned by S 0 2 . XPS spectra reveal that Ce3 sulphate is formed and that Pt is then reduced to the metallic state. Flow reactor measurements under steady-state conditions show that the promoting effect of ceria on the activity is poisoned by S02. Under oscillating conditions, ceria increases the activity, especially on the rich side of the stoichiometric point. This activity increase is reduced by the presence of S02. References 1. 2. 3. 4. 5. 6. 7. 8.
9. 10.
11. 12. 13. 14. 15. 16.
Yao H.C. Japar S. and Shelef M., J. Catal. 50,407 (1977) Den Otter.G.J., and Dautzenberg F.M., J. Catal. 53, 116 (1978). Summers J.C., and Ausen S.A., J. Catal. 58,131 (1979). Pande N.K. and Bell, A.T., J. Cata1.,97,137 (1986) Jin T., Zhou Y., Mains G.J. and White, J.M., J. Phys. Chem., 91,5931 (1987). Oh S.H. and Eickel C.C., J. Catal. 112, 543 (1988). Shyu J.Z., Otto K.,Watkins W.L.H., Graham G.W.,Belitz R.K., and Gandhi H.S., J. Catal 114, 23 (1988). Engler B., Koberstein E. and Schubert, P., Appl. Cata1.48, 71 (1989). Shyu J.Z. and Otto K., J. Catal. 115, 16 (1989). Schlatter J. C. and Mitchell P. J., Ind. Eng. Chem. Prod. Res. Dev. 19, 288 (1980). Su E.C., and Rothschild W.G., J. Catal. 99, 506 (1986). Yao H.C., Stepien H.K. and Gandhi, H.S., J. Catal. 67, 231 (1981). Ltiijf P., Kasemo B. and Keck, K.-E., J. Catal. 118, 339.(1989). Kasemo B., Rev. Sci. Instrum. 50, 1602 (1979). Briggs D. and Seah M.P., Eds, Practical Sudace Analysis, John Wiley & Sons, Chichester, 1983 Muraki H., Shinjoh H., Sobukawa H., Yokota K., and Fujitadi, Y., Ind. Eng. Chem. Prod. Res. Dev. 25, 202 (1986). hrr
I
274 17 18. 19. 20. 21 22 23. 24. 25. 26. 27.
28. 29. 30. 31. 32. 33. 34. 35. 36 37. 38. 39. 40. 41. 42. 43 44.
Lijijf P. and Kasemo B., J. Catalysis,submitted. See Kasemo.B., and Tornqvist E., Phys. Rev. Letters vo1.44,23, 1555 (1980) and the references therein. Lijaf P. and Kasemo B., in preparation. Koelling D.D., Boring A.M. and Wood S.H., Solid State Commun., 47, 227 (1983) Ryzhkov M.V., Gubanov V.A., Teterin Yu. A. and Baev A.S., Z. Phys. B. Cond. Matter, 59, 1 (1985) Kotani A., Jo T. and Parlebas J.C., Adv. Phys., 37, 37 (1988) Jo T. and Kotani A., J. Phys. SOC. Jpn. 55, 2457 (1986) LeNormand F., Hilaire L., Kili K., Krill G. and Maire G., J. Phys. Chem., 92,2561 (1988) Burroughs P., Hamnett A., Orchard A.F. and Thornton G., J. Chem. SOC. Dalton Trans., 17, 1686 (1976) Praline G., Koel B.E., Hance R.L., Lee H.-I. and White J.M., J. Electron Spectrosc. Relat. Phenom., 21, 17 (1980) Sarma D.D., Vishnu Kamath P. and Rao C.N.R., J.Chem. Phys. 73,71 (1983) Bjornkvist L., work in preparation. Uwamino Y., Ishizuka T. and Yamatera H., J. Electron. Spectrosc. Relat. Phenom. 34, 67 (1984) Muilenberg G.E., Ed. Handbook of X-ray Photoelectron Spectroscopy, Perkin Elmer, Physical Electronics Division, Eden Prairie, MN, 1979 BarrT.L., J. Phys. Chem., 82 1801 (1978) Fung S.C., J. Catal., 76, 225 (1982) Ishi S-I, Ohno Y.and Wishwanathan B., J. Scient. Ind. Res., 46, 541 (1987) Bjornkvist L., Lijijf P., to be published Dictor R., and Roberts, S., J. Phys. Chem. 93, 5846 (1989). Castner D.G. and Somorjai G.A., Surf. Sci. 83, 60 (1979). Kienneman A., Breault R., Hindermann J.P. and Laurin, M., J.Chem. Soc., Faraday Trans. 1, 83, 2119.(1987). Yang A.C. and Garland C.W., J. Chem. Phys.61, 1504 (1957). Pitchon V., Primet M. and Praliaud H., Applied Cata. 62, 317 (1990). Ichikawa M. and Fukushima T., J. Phys. Chem. 89, 1564 (1985). Oh S.H. and Eickel C.C., J. Catal. 112, 543 (1988). Oh S.H., J. Catal. 124, 477 (1990). Kim G., Ind. Eng. Chem. Prod. Res. Dev. 21, 267 (1982). Grenoble D.C., Estadt M.M. and Ollis D.F., J. Catal. 67,90 (1981).
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
275
LASER RAMAN CHARACTERIZATION OF SURFACE PHASE PRECIOUS METAL OXIDES FORMED ON C e 0 2 L. L. Murrell, S. J. Tauster, and D. R. Anderson
Engelhard Corporation, Menlo Park, CN 40 Edison, NJ 08818 ABSTRACT Many surface phase oxide systems have now been characterized which are formed from reactions of Group IV, V and VIB metal oxides with the hydroxyl structure of Ti02, Si02, and Al2O3. In this paper, we present evidence that ceria has a significant capacity to form isolated, M - 0 groups with Rh, Ir, Pd, and Pt. This surface phase, precious metal oxide structure interacts strongly with the ceria surface, so that the latter is significantly stabilized under oxidizing conditions at 750-1000°C. Laser Raman of these systems is useful for tracking the M-0 phase, since a strong Raman band is observed for the M - 0 bond at about 700 cm-l The M - 0 surface phase may be recovered as a highly-dispersed metal phase upon reduction at 500°C in hydrogen. Addition of Pt beyond the capacity of Ce02 to form the PtO complex results in Pt sintering to form large metal particles. Treatment of all the Group VIII metals on Ce02 to cyclic redox aging conditions at 850°C causes severe sintering to occur with complete loss of the surface complex, M-0
INTRODUCTION
Laser Raman Spectroscopy (LRS) has proven to be a powerful tool in the characterization of Group IV, V, and VJB oxides and rhenium oxide bound to Ti02, Si02, and A1203 surfaces.[l,2] Recent Raman studies from the Ford laboratory has identified Raman bands for dispersed Pt on A1203 at 125, 335, and 590 cm-1.[3] We have found a quite strong Raman band at ca. 700 cm-1 for Rh, Ir, Pd, and Pt dispersed on Ce02. This Raman band is assigned to a surface oxide M-0 formed with specific sites on the Ce02 surface. Previous work [4-101 has proposed Pt+2 to be formed on the C e 0 2 surface, and that this precious metal oxide is stabilized as the oxide at high temperature conditions. In our investigations a strong Raman band was observed at 5 wt% Pt levels on CeO2 with no other phase being present except for this interactive M+2 oxide. By a combination of characterization techniques including LRS and conventional chemisorption it is possible to propose a detailed model of this precious metal oxide (PMO) surface complex phase. This model clearly indicates that the Group VIII precious metal oxides on Ce02 indeed can be classified under those systems which exhibit a Strong Oxide-Support Interaction (SOSI). The M - 0 surface complex on Ce02 stabilizes the Ce02
276
phase to sintering while being stabilized, in turn, in a highly-dispersed phase by the CeO2 support at high temperatures under oxidizing conditions. Cyclic redox aging conditions at 850°C for 8 hrs. suitable to simulate automotive exhaust conditions, however, result in very severe sintering of the PM. Sintering was so extreme that LRS failed to detect any residue species remaining on the CeO;! surface after cyclic redox aging.
EXPERIMENTAL The Group VIII metal samples in this work were prepared by the incipient wetness preparation method on high purity ceria of 130 m-2/g surface area. Samples were prepared using the following precious metal (PM) precursors; Rh(N03)2, H2IrC16, Pd(N03)2, and Pt amine complex. The Rh, Ir, Pd and Pd samples were all prepared at a metal content equivalent to 2.5 and 5 wt% Pt, i.e., 0.128 and 0.256 millimole/g respectively. The samples were first calcined at 350°C before investigation by LRS. However, severe fluorescence problems precluded obtaining spectra. Therefore, all the samples were calcined for 1 hr. at 530°C and none of the samples calcined at 530°C showed evidence of fluorescence after being calcined at the higher temperature. The Raman spectrometer employed in this work has been described in detail elsewhere.[l] All of the Raman intensities of the M - 0 band at ca. 700 cm-l were obtained using the Ce02 Raman band at 257 cm-1, or the Raman band at 557 cm-1 as an internal reference of the M - 0 band intensity. The CO chemisorption was carried out using a pulsed chemisorption unit of conventional design.
RESULTS AND DISCUSSION Considerable precedence exists in the literature for the interaction between precious metals (PM) and bulk Ce02, or between ceria particles or crystals dispersed on alumina.[4-10]. The literature suggests that this PM-ceria interaction stabilizes the PM component to sintering under automotive exhaust conditions. In this work the chemistry of this PM-ceria interaction has been probed by a combination of techniques including chemisorption, XPS, and Laser Raman spectroscopy (LRS). It is from the results of these combined techniques that a detailed model of the PMO-ceria interaction can be proposed. LRS has proven invaluable as a probe in elucidating the complex chemical behavior of the W03-A1203-A12( W 04)3 system after high temperature treatments [2]. In this previous work [2] it was found to be useful to evaluate changes in the Raman band intensities as a function of WO3 content, and also as a function of increasing severity thermal treatment. A very analogous strategy was adopted to probe the Pt-Ce02 system in this work. We chose calcination temperatures of 750, 900, and 1000°C for Pt on Ce02. In
277
addition, the Pt content was investigated over a wide range of compositions from 0.5 to 5 wt%. Raman spectroscopy was employed as a probe of the phases present after this series of high temperature treatments. One of the most pragmatic demonstrations of a strong oxide-support interaction (SOSI)[2,11,12] is the influence exerted by the surface phase oxide on the surface area of the support to which the oxide is strongly attached. Figure I shows the surface area of Pt on Ce02 samples calcined at increasingly severe conditions. It is clear from this figure that Pt on Ce02 when calcined at 750°C for 2 hrs. shows a strong dependency of surface area stability on Pt content. The 0.5 wt% Pt/Ce02 sample has a surface area of ca. 20 m2/g while the 5 wt% Pt/Ce02 sample has a surface area of over 100 m2/g. This surface area stabilization is strong evidence for a SOSI between Pt and Ce02. This surface area stabilization of Pt on Ce02 is very probably due to Pt+2 interacting with the Ce02 surface as proposed in previous work [4-lo]. It was due to the stabilization of Ce02 at over 100 m2/g at 750°C that we initiated a systematic investigation of these systems by LRS. Before proceeding to the results from the LRS investigation, it is useful to discuss the results in Figure I in more detail. Upon calcination of the 1-5 wt% Pt on Ce02 samples at 750°C for 24 hrs., a sharp dependency of Pt content on surface area is no longer observed as for the 2 hr. calcination period. Calcination at 900 and 1000°C shows no deDendencv of surface area on Pt content. However, at these severe conditions the surface areas are stabilized at 15 and 8 m2/g, respectively. It is apparent that the capacity of Pt+2 to stabilize Ce02 at severe conditions by a SOSI is limited, but far from insignificant as we will see in subsequent discussion. It is these changes in stabilities at different conditions that we hoped LRS would help to address. We established that calcination at 750°C for 2 hrs. followed by a 900°C calcination did not lead to enhanced stability compared to the sample calcined directly at 900°C. This suggests that there is a continuous decrease in the capacity of Pt+2 to stabilize Ce02 at severe conditions even though stabilization occurs at a lower temperature. This observation is similar to the limited capacity of MgO to stabilize ruthenium oxide as a highly-dispersed surface ruthenate at increasing severity conditions [ 131. The LRS spectra of Pt on Ce02 at 1 , 2.5, and 5 wt% loading levels calcined at 530°C are shown in Figures 2-4. It is clear from these figures that a Raman band at ca. 660 cm-1 is increasing in intensity as a function of increasing Pt content. No Raman active mode is present in region for CeO2 [14]. (Note that each sample is optimized for signal-to-noise so intercomparison between spectra is not possible in a quantitative way.) Figures 5-7 show the spectra for this same series of Pt on Ce02 samples when calcined at 750°C. These are the same samples shown in Figure I which have surface areas ranging from ca. 20 to 100 m2/g. Although the spectra quality is not as good as
278
o 750°C, 2 Hrs.
-
P
-
-
75OoC, 24 Hrs.
A 9W0C, 24 His. A 10W0C, 24 Hrs. 0 75O0C, 2 Hra, 900OC. 2 Hrs. loo m 750°C, 2 Hrs., 900°C, 24 Hrs.
120
-
U
E
80-
0
0
Wt. H Pt on Ce02
Figure I Surface area stability as a function of temperature for different Pt contents on bulk Ce02.
400
300
1 wt % Pt on Ce02, calcined 350 "C,2 Hrs, and 1 Hr 530 "C
200
2 79
r
n
it
24000
22000
20000 0
0
18000
F
X
3
16000
Y)
c
14000
12000
6000 1200 1100 1000
900
700
800
600
500
u
400
200
300
Wave Number (cm- 1)
Figure 3 2.5 wt % Pt on Ce02, calcined 350 "C, 2 Hrs, and I Hr at 530°C
35000
I
30000 0 F
C
a
8 20000 l5Ooo 10000
/
t /
1200 1100 1000 900
600 500 Wave Number (cm 1 )
800
700
-
400
300
200
GO
Figure 4 5 wt % Pt on Ce02, calcined 3.50 "C, 2 Hrs, and I H r at 530 "C
280
24000 r
20000 22000
0
18000-
8000 1200 1100 lo00
BOO
800 700 600 500 Wave Number (cm - 1)
I
I
I
400
300
200
Figure 5 I wt % Pt on Ce02, calcined 7-50 "C, 2 Hrs
Wave Number (cm - 1)
Figure 6 2.5 wt % Pt on Ce02, calcined 750 "c, 2 Hrs
28 1 8OOOo
-
78000
-
76000
-
74000
-
64000
-
1200 1100
lo00
900
800
700
600
500
I
I
I
400
300
200
Wave Number (cm - 1)
Figure 7 5 wt % Pt on Ce02, calcined 750 "C, 2 Hrs. Fluorescence problems probably due to P@ 500OC Calclned
Raman Intensity (Arbltuaq Unlts)
I
Raman lntenslty (Arbltuary Units)
I I I 700 600 500
Wave Number
a 75OoC Calclned
l
700
600
Wave Number
Figure 8 Relative Raman intensities of PtOx on Ce02 : I , 2.5 and 5 wt % Pt calcined at 500 and 750 "C
500
282
for the previous samples it is clear that a strong band remains in all samples at ca. 660 cm-1 A composite of the ca. 660 cm-1 band and the ca. 550 cm-1 band of Ce02 are shown in Figure 8. The band at 660 cm-1 was normalized with the band of CeO2 at ca. 260 cm-1 or the band of Ce02 at 550 cm-l, respectively, for the 530 and 750°C calcined samples. The normalized band intensity is shown in Figure 9 for these six samples as a function of Pt content. It is clear from this figure that there is an unmistakable relationship between this 660 cm-1 band and the Pt loading level. A similar relationship between Raman band intensity and W 0 3 content on A1203 was observed in previous work.[2] In fact, W 0 3 loading onto Ce02 shows this same relationship of W 0 3 content with the surface W 0 3 band when normalized with the Ce02band at 260 cm-l [15]. We feel that the very intense Raman band at 660 cm-l, which is frequency invariant with Pt content, is due to a Pt+2 oxide surface complex. The presence of oxidized Pt on Ce02 has been reported in many papers [4-lo], and it has also been suggested as the key to stabilization of a dispersed Pt phase at severe high temperature conditions [ 5 ] .Note that calcination of the 1 and 2.5 wt% Pt on Ce02 samples at 750°C appears to show little influence on the intensity of the Pt-0 band when referenced to a band of Ce02 as an internal reference, Figure 9 . In subsequent work it was demonstrated by CO chemisorption that the dispersion of reduced Pt on CeO2 was little changed from CO/M=1 for the 1 and 2.5 wt% samples, calcined either at 530 or 75OoC, see Table 1 .
co CHEMISORPTION
Table 1 RESULTS COUPLED WITH
ESCA
Results confirm that intermediate calcination at 750°C maintains high Pt dispersion
283
U C
0:
8-
A
Wt.Yo Pt
Figure 9 Normalized Raman band of PtOx versus wt% Pt for high loading levels of Pt on Ce02 calcined at 530 and 750 "C ( 0 530°C; A 750°C)
1200 1100 1000
SO0
000
700
600
Wave Number (cm - 1)
500
400
300
Figure 10 0.5 wt % Pt on Ce02, calcined lOOO"C, 4 Hrs .
200
284
The chemisorption results when coupled with the Raman results in Figure 9 make a strong case that the dispersed Pto phase on Ce02 following reduction is formed from the Pt-0 surface complex observed by the Raman band at 660 cm-1 serving as a "fingerprint." For the case of the 5 wt% Pt on Ce02 sample, the Raman spectrum was difficult to obtain due to fluorescence, probably due to large Pt metal particles that were observed for this sample by x-ray diffraction. Despite this difficulty, the Raman intensity corresponds to roughly that expected for ca. 1.5 wt% Pt on Ce02 sample calcined at 530°C. The CO/M ratio of 0.55 for the 5 % Pt on Ce02 calcined at 750°C suggests that about 50% of the Pt is in a dispersed phase analogous to the phase present for the 2.5 wt% Pt on CeO2 sample. There seems to be no question that 2.5 wt% Pt on Ce02 can be calcined at 750°C without any loss in a metal dispersion. Even a 5 wt% Pt on Ce02 sample retains surprisingly high metal dispersion, i.e. 50%, when calcined at 750°C. The Raman band at 660 cm-1 apparently is a fingerprint of the Pt-0 complex on CeO2 which, when reduced, leads to high metal dispersion for the 750°C calcined samples. The frequency invariance of the 660 cm-1 band argues that the Pt-0 complex is interacting with specific sites on the Ce02 surface where interaction between Pt-0 groups is not possible. This proposal model is not altogether surprising as isolated Fe+2 groups have been shown to be very strongly bonded to the Lewis acid centers of Ti02[16]. What is quite interesting is the apparent high density of sites on CeO2 which interacts strongly with the oxidized Pt-0 phase. The density of Pt-0 groups corresponds to ca. 1.6 micromole/m-2 of Ce02 surface area based on a CO/M ratio of 0.8 for the 500°C calcined sample. This density of interaction sites on Ce02 corresponds closely to the density of Lewis sites on high surface area alumina, 1.8 micromoles/m-2 [ 171. In summary, the Ce02 surface has quite a high capacity to form an interactive Pt-0 surface complex with platinum, and this leads to a SOSI which imparts high dispersion to the Pt phase even after high temperature calcination at 750°C. We will next turn our attention to the characterization of the Pt on Ce02 samples calcined at 900 and 1000°C. We initially focused our attention on characterizing the 0.5 wt% Pt on Ce02 sample calcined at 1000°C. The surface area of this sample was 8 m2/g after the high temperature treatment. Chemisorption using the static method gave a CO/M ratio of 0.7 for this sample when reduced at 500°C in H2. This result is indicative of retention of a fairly well-dispersed metal oxide phase. Interestingly, the H/M ratio was only 0.05, which suggests an intermediate Strong Metal Support Interaction (SMSI) for reduced Pt on Ce02. The Raman spectrum of the 1000°C calcined sample is shown in Figure 10. There is little question from the retention of the Raman band at ca. 660 cm-1 that the Pt-0 surface structure is retained in high
285
concentration on the CeO2 surface. This is especially significant in that the C e 0 2 surface area had undergone an order of magnitude change upon calcination at 1000°C. Based on the retention of the Pt dispersion of 70% for this sample coupled with retention of a strong M - 0 stretch in the Raman spectrum, Figure 10, it is proposed that the P t - 0 surface complex is immiscible in the Ce02 support and is retained in high concentration on the outer surface of the 8 m2/g CeO2 support. If one assumes that the CO chemisorption is a measure of the dispersed Pt-0 complex, then one can calculate the surface density to be 2.2 micromoles/m2. This value is not greatly different from the density of the Pt-0 complex for 2.5% Pt on Ce02 calcined at 750°C for 24 hrs. also based on CO chemisorption: 134 micromoles/g + 56 m2/g giving a surface density of 2.4 micromoles/m2. The number of Pt-0 complex units per 1 nm2 of surface area is about one. This gives an average Pt+2 - Pt+2 separation of 1 nm or 10 A. For the case of the higher Pt content samples calcined at 900" and 1000°C, XPS and chemisorption combined with x-ray powder diffraction provide clear evidence that as the CeO2 surface area decreases the Pt which can not be accommodated by the Ce02 surface as the complex M-0 sinters rapidly to form ca. 20 nm size Pto particles. Note that the 1 % Pt and 2.5% Pt on Ce02 samples calcined at 900°C which have a surface area of 15 m2/g both have almost the same CO uptake in absolute terms. These chemisorption values correspond to almost exactly that for the 0.5% Pt on CeO2 sample calcined at 1000°C. It can be proposed that there is a certain density of sites on Ce02 which strongly bond to the M - 0 complex for severe high temperature calcinations. This density of bound surface groups is largely independent of Pt content present initially, even if bound as M - 0 groups to the CeO;! surface. Upon high temperature treatment in air, phase segregation occurs leading to large Pto particles and only those strongly-bound M - 0 surface groups remain in a dispersed phase on the CeO2 with a surface area of 8-15 m2/g. This proposed model is shown in Figure 11.
PtO
PtO
PtO
PtO
A
1> PtO
PtO
PtO
Figure I 1 Model emerges from combination of techniques
286
All four of the Group VIII metals show strong Raman bands on Ce02 strong Raman bands on Ce02 at high metal loading levels, i.e. 0.13 and 0.26 millimoles/g. The Raman bands were at 654 and 630 cm-1 for Pt and Pd, respectively; and at 614 and 586 cm-1 for Ir and Rh, respectively. Since all four of the Group VIII metals form M - 0 species on the Ce02 surface at about equal concentration per unit surface area, then the sites on Ce02 capable of forming the M - 0 complex are not just unique to the Pt+2 complex as is commonly assumed in the literature [4-101. All four of the Group VIII metals investigated in this work, Pt, Pd, Rh and Ir, were investigated as regards their stability to sintering after cyclic redox aging conditions at 850°C for 8 hrs. This cyclic redox aging is very similar to that which would be present in automotive catalyst application. It was apparent that all of the Group VIII metals suffered severe sintering after this cyclic aging treatment based on the Raman spectra, see Figures 12-15. Calcination of these cyclic aged samples for 4 hrs. at 500°C failed to give any evidence for the presence of a M - 0 surface complex on the Ce02 surface. Apparently, sintering can occur in the reducing swing of the cyclic aging as retention of the M-0 complex is readily apparent under oxidizing conditions. This work suggests that the role of Ce02 in automotive catalysts is probably more one of a poison scavenger, probably for S02, and/or as an oxygen storage component than as stabilizer of precious metal dispersion.
CONCLUSIONS Ceria has a truly remarkable capacity to stabilize high surface concentrations of precious metal oxide groups, M - 0 , at intermediate and at high temperatures. The amount of M - 0 groups per unit surface area bound to the C 0 2 surface remains remarkably constant over a very wide range of surface areas after different high temperature calcination conditions. This density of M - 0 groups corresponds to about one group per 1 nm2 of surface area, or about one M - 0 group separated from another by l n m or 10 I$. This density of stable M - 0 groups is clearly about one order of magnitude lower than that expected from just steric packing of groups on the surface. Such steric packing appeared to control the density of surface RuO3 groups bonded to MgO at high temperature conditions, but Ce02 appears to have a much lower number of specific sites which can form the surface complex with the Group VIII precious metal oxides. Although Ce02 appears to have superb stabilization behavior of the precious metal oxides under continuous oxidizing conditions, cyclic redox aging conditions serves to cause essentially complete loss of the surface complex. This undoubtedly is related to severe metal
287
28000-
24000 22000 20000 18000 26000
5 u)
16000-
C
-
10000 8000
-
6000
-
1200 1100 1000
900
800
700
600
500
I
I
400
300
I 200
Figure 12 1.18 wt % Rh on Ce02 (Equivalent to 2.5 wt % Pt), Calcined 500°C
20000 22000
18000
-
16000 0
14000
-
X
2 12000C
"z
10000-
8000 6000 4000 1200 1100 1000
900
800
1
I
1
I
I
I
700
600
500
400
300
200
Figure 13 2.36 wt % Rh on CeO2, (Equivalent to 5 wt % Pt) Calcined 500°C.
288
12000 1 1000
10000 9000
8000 cn
c.
5
8
7000 6000
50001
4000
h
1200 1lOOlC
Figure 14 I .I8 wt % Rh on Ce02 , Cyclic Redox Aging 800 "C, 8 Hrs.
Figure 15 1.18 wt % Rh on Ce02 , Calcined at SOO"C, Reduced in Laser Raman beam.
289
sintering or agglomeration after exposure to cyclic aging conditions at high temperature, which, in turn, leads to substantial decreases in the surface area of CeO2 at high temperatures. REFERENCE§ 1. 2.
3. 4. 5. 6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17.
I. E. Wachs, F. D. Hardcastle, and S. S. Chan, Spectrosc. 1,30 (1986) S. Soled, L. L. Murrell, I. E. Wachs, G. B. McVicker, L. G. Sherman, S. Chan, N. C. Dispenziere, Jr., and R. T. K. Baker, in "Solid State Chemistry in Catalysis." Eds. R. K. Grasselli and J.F. Brazdil, ACS ymp. Series 279, American Chemical Society, Washington, DC, (1983, Chapter 10. K. Otto, W. H. Weber, G. W. Graham, J. Z. Shyu, Applied Surface Science, Vol. 37, 2,250 (1989). I. T. Kummer, Y. Yao, and D. McKee, SAE Paper No. 760143, (1976) J. C, Summers and S . Ausen, J. Catal. 3, 131(1979) H. C. Yao, H. S. Gandhi and M. Shelef in "M%d Support and Me-d-Additive Effects in Catalysis, ed. B. Imelik, Elsevier, Amsterdam, (1982). H. C. Yao and Y. F. Yu-Yao, J.Catal. 6 , 2 5 4 (1984). E. C. Su, C. N. Montreuil, W. G. Rothschild, Appl. Catal. 75,(1986) E. C. Su and W. G. Rothschild, J.Cata1. 99,502(1986). J. Z. Shyu, K. Otto, W.L.H. Watkins, A. W. Graham, R. K. Beilitz and H.S. Gandhi, J. Catal. 11423(1988). S. L. Soled, G.B. McVicker, L.L.Murrell,L.G.Sherman,N.C.Dispenziere, Jr., S.L. Hsa, and D Waldman, J. Catal 111,286 (1986) L. L. Murrell and N. C. Dispenziere, Jr., Catal. Letters, 4 235 (1990) S. J. Tauster, L. L. Murrell, and J. P. DeLuca, J. Catal. 48,258 (1977) L. L. Murrell, Unpublished Work. L. L. Murrell, Unpublished Work. L. L. Murrell and R. L. Garten, Application of Surface Science, B,218(1984). L. L. Murrell and N. C. Dispenziere, Jr., J. Catal. 117,275 (1959)
u,
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V.. Amsterdam
29 1
DIESEL EMISSION CONTROL E.S. Lox, B.H. Engler, E. Koberstein
Degussa AG ,Physical Chemistry Research Department P.O. Box 1345,0-6450 Hanau 1 ,W-Germany ABSTRACT
An overview is given on the kind and origin of tailpipe emissions from diesel engines and on their impact on the environment. The influence of engine type, engine operation condition, fuel composition and test cycle on the composition and properties of diesel engine exhaust are briefly discussed. The present and anticipated emission legislation is summarized. The various approaches to decrease the tailpipe emissions are described. These approaches are grouped in ways to decrease the raw emissions, such as engine design optimization or change of fuel type and properties, and in ways to aftertreat the engine out emissions. The latter involves various types of filtering devices, including catalyzed filters, combinations of filter devices with fuel additives as well as oxidation catalysts. Finally, the ways investigated at present to cope with future requirements, such as a decrease in the nitrogen oxide emission, are presented. INTRODUCTION
In major parts of the world legislation is presently in effect which aims at limiting the tailpipe emission of carbon monoxide, hydrocarbons and nitrogen oxides from internal combustion engines. The legislation was such that until recently mainly fossil fuel powered Otto engines needed a reduction of the amount of these components in the tailpipe exhaust gas. This was achieved by a variety of means, of which the incorporation of catalytic exhaust gas aftertreatment devices became a widespread accepted technology. Recently however the environmental concern also turned to the tailpipe emission of internal combustion engines operating according to the diesel principle. This intensified the research and development efforts of engine designers in this respect.
292
Just as otto engines, diesel engines are internal combustion engines with cyclic combustion, but with different aidfuel mixing and ignition principle, and with different operation pressure, temperature and aidfuel ratio. These differences are summarized in Table 1 . There exists a variety of diesel engine designs. As far as fuel admission is concerned, a division can be made between direct injection (DI) engines, where the fuel is injected directly in the cylinder, and indirect injection (IDI) engines, where the fuel is injected in a prechamber or a swirl chamber, which are small volume combustion rooms connected to the cylinders. Of course intermediate designs do exist also.The different fuel admission principles can be combined with various air admission systems. TABLE 1
Zomparison of combustion parameters for diesel engines versus Otto engines DIESEL ENGINE
$ $ $ $ $ $
Process Type Combustion Type Air/Fuel Mixing Ignition Type Operating Pressure Temperature at Compression $ h during operation $ Exhaust gas Oxygen content
Internal combustion Cyclic Heterogeneous Auto 30 - 55 bar
700°C - 900°C OIh<-J Lean
OTTO ENGINE
Internal combustion Cyclic Homogeneous External 15 - 25 bar
400°C - 600°C 0.8 5 h 5 1.2 Lean to Rich
A distinction can be made between natural aspiration (NA), where the air is sucked into the cylinder, and turbo charging (TC), where the air is compressed into the cylinder. The latter can be combined with cooling of the compressed air (TCI). Diesel engines are built in a broad range of engine displacement and engine rotational speed. The present diesel engine technology is such that there is a correlation between the application of the engine on one side and the design of the engine, i.e. displacement, operation speed, fuel- and air-admission principle on the other side. The present correlation for diesel engines used in on-road transportation vehicles is summarized in Table 2. It is obvious that the
293
chemical and thermodynamical properties of the engine exhaust emission are dependent on the design of the engine. TABLE 2
Diesel engine types for transportation applications HEAVYDUTY APPLICATION
$ Direct injection ;
TC; TCI $ Displacement per cylinder : 1 - 2 liters t; Multicylinder ( 6 , 8, 10, 12, 1 $ Lowrpm t; High power output (200 - 400 HP)
MEDIUM AND LIGHT DUTY APPLICATION
$ Direct and indirect injection ; NA; TC;TCI $ Displacement per cylinder : 0.5 - 1 liters $ 4 - 8 cylinders $ Low to medium rpm
PASSENGER
CAR $ Indirect injection:
NA; TC $ Displacement per cylinder : 0.5 liters $ 4 - 6 cylinders $ Highrpm $ Medium power
output (50 - 150 HP)
Diesel engines are used in a broader range of applications than Otto engines. A distinction can be made between stationary applications, such as engines used in cogeneration plants, and between instationary applications, which include on- and off-road transportation means. In the group of on-road transportation, diesel engines are used in heavy, medium and light duty trucks and busses as well as in passenger cars. For the 8 million multicylinder diesel engines produced world wide in 1989, the end use is represented in Figure I [ 13. A closer look at the diesel engines used for on-road transportation means, however reveals that major differences exist between the various countries. Indeed, whereas heavy duty trucks and busses are universally equipped with diesel engines, it is presently only in Western Europe where diesel engines are also intensively used in passenger cars and in light duty trucks. And even within Western Europe major differences exist in the relative importance of diesel engines applied to these vehicle categories, as is exemplified for model year 1989 in Figure 2. This picture was subject to major variations over the past twelve years; those variations being substantially induced by tax policies, also related to environmental matters [2].
294
3 1%
15%
trucks and buses
trucks and buses
Total production: 8 million units
Rot Borch l l 9 9 O l
Figure 1 World wide production of multicylinder diesel engines in 1989 [ 11
40
1 -
35.1 ~
30--
29. -
20--
-
lo--
5
0
2.9
0.6
_.
Figure 2 Relative importance of passenger cars with diesel engines in I989 for selected W-European countries [2]
295 EMISSIONS FROM DIESEL ENGINES
Kind and origin of emissions The engine out emissions from diesel engines contain components in the gaseous, the liquid and the solid state, as is schematized in Table 3. The presence of these components has various causes. TABLE 3
Kind of emissions from diesel engines GASEOUS
LIQUID
f N2 f co2 fco f H2
f
f NO/NO2
f swso3 f HC(C2-Cl5) f Oxygenates f Organic nitrogen
H20
f H2S04 f Hydrocarbons (C5
SOLID
f soot f Metals f Inorganic
Go)
f Oxygenates f Polyaromatics
f Sulfates
f Solid hydro carbons
and sulfur compounds Incomplete combustion gives rise to unburned fuel and lube oil hydrocarbons, as well as to their partial oxidation products such as oxygenates and carbon monoxide [3,4]. In the most oxygen deficient regions of the diffusion flame, the fuel hydrocarbons undergo a series of chemical transformations, during which their hydrogen to carbon ratio decreases to the extent that soot precursor molecules and subsequently soot nuclei are formed. As shown in Figure 3, these soot nuclei undergo first coalescent coagulation, by which soot spherules are formed, which finally form soot particles by chain forming coagulation [ 5 , 6 , 7, 8, 91. In the oxygen rich parts of the flame, part of the soot particles can be burned again. Fuel and lube oil hydrocarbons can also be transformed into a variety of polynuclear aromatic components (PAH), which in the cylinder and in the exhaust pipe can further react with other gaseous components, such as nitrogen oxides [ 10, 11, 121.
296
1-2 hconpkte combustion
1-10
nueloath 10-30
10-50
Figure 3 Physical and chemical processes during diesel soot formation in the engine The organic sulfur components in the fuel are oxidized to sulfur dioxide. A small portion however can be further oxidized to sulfates, whereby metal particles, formed by engine wear, might act as a catalyst [ 13, 141. The lube oil additives are oxidized also, causing inorganic oxides to be present in the exhaust. Nitrogen oxides are mainly formed from the reaction between the air nitrogen and the air oxygen, but a small part might come from the oxidation of organic nitrogen components in the fuel [15]. The constituents at the cylinder outlet can undergo further chemical and physical transformations in the tailpipe. Of particular importance for the present discussion are the transformations by which the tailpipe outlet particulates are formed. This is schematized in Figure 4 . From this description it is apparent that the particulate material has a complex chemical composition and physical structure. They are agglomerates of smaller particles, which have a soot nucleus, upon which hydrocarbons are condensed and inbetween which inorganic oxides and metal particles are occluded. Depending on temperature, also water and sulfuric acid can condense on these particles. Figure 5 shows Scanning and Transmission Electron Microscopy pictures of the particulates at the tailpipe outlet.
297
CyYndr
0 soot 0 unbwlnd hydrocarbons 0 hydrocarbon radkdr 0 SO2lSO~ 0 htab 0 WINO2
0 bmM*
oxydos
m carbon m Hydrocarbons 0Pobr fraction 0 Motels Watr
m Oxydosl matos
Figure 4 Physical and chemical processes during diesel particulate formation in the engine and the tailpipe
The impact of diesel exhaust on the environment has numerous aspects, which ranges from a particular smell, caused amongst others by the oxygenates, over a colored smoke, in which the condensed hydrocarbons play a role, to the suspected health hazard, which is thought to be related to the particulate matter [16, 17, 181. Although still a matter of debate, the particulates are suspected to have carcinogeneous properties, which recent research attributes mainly to the slightly polar polynuclear aromatic components, which can be adsorbed on the particulates. The suspicion is enhanced by the fact that diesel particulates have such dimensions that they can penetrate into the lungs and/or have a prolonged residence time in the lungs [19,20, 21,221. This suspicion leads a working group of the World Health Organization (WHO) to classify diesel exhaust as a potential carcinogenic substance (group 2A), as opposed to gasoline Otto engine exhaust, which is classified in group 2B, i.e. a substance with a suspect of potential carcinogenic action. This classification was also adopted in the 1988 MAK-list in W-Germany [23,24]. Relative imDortance of diesel engines in the total emission The contribution of diesel engines to the total emission of carbon monoxide, hydrocarbons, nitrogen oxides and particulate matter is exemplified in Table 4, which refers to the situation in W-Germany in 1987.
298
Also the comparison of the amount of these components, emitted by a passenger car equipped with a diesel engine, to that emitted by a comparable passenger car, equipped with an Otto engine, explains the directions of emission reduction developments.
Figure 5 The morphology of diesel particulates at the tailpipe outlet as measured by (a) Scanning Electron Microscopy and (b)Transmission Electron Micrsocopy
299
TABLE 4
Contribution of diesel engines to the emission of carbon monoxide, hydrocarbons, nitrogen oxides and particulate matter (W-Germany 1987) [26, 27, 421. Component
Contribution of diesel engine % of total
co Hc NOx Particulate
2.2 6.5 23.3 11.4
Relative contribution, in % of diesel engine part, of Heavy duty truck
Passenger car
Industria engine
64.2 65.0 71.9 60.0
17.9 15.9 7.1 22.6
17.9 19.1 21 .o 17.4
As shown in Figure 6, diesel engines emit much less carbon monoxide and hydrocarbons than gasoline Otto engines; they actually compete well with a gasoline Otto engine equipped with a closed-loop three-way catalyst. As far as nitrogen oxides are concerned, the diesel engine emits less than a gasoline Otto engine without exhaust gas aftertreatment, but considerably more than gasoline Otto engines equipped with closed-loop three-way catalysts. Finally, the emission of particulate matter from a diesel engine is more than ten times higher than from a gasoline Otto engine [25,26, 27, 281. Analytical procedures The analytical procedures used to determine the composition of diesel exhaust are complicated and lengthy, as they require both on-line and off-line methods [29, 30, 31, 321. As shown in Figure 7, most of the gaseous constituents can be measured by on-line analyzers. The particulate matter is currently defined as a material which can be sampled on a filter paper at a gas temperature of 5°C k 3°C. Therefore, part or all of the tailpipe exhaust has to be diluted with air in a dilution tunnel to cool it down to that temperature. Then the exhaust gas is sampled through a filter paper on which the particulates accumulate. For research and development purposes, the composition of the particulates can be further analyzed by a sequence of extractions combined with chromatographic procedures [ 3 3 , 341.
300
20
Emissions [g/mHel
Emissions [mg/mJ.l 400 1 335.5
1
15
10 150 5
Jm ,
0
100 -
4.24
n
2.11
co
HC Souc.:
vw
50 -
25.oB
10.54
NOx
/1*8*/
Figure 6 Typical emissions of passenger cars in the FTP 75-cycle. A comparison between diesel engines and Otto engines [25]
Engine exhaust Air
, I
Diluted exhaust b
Filter
impingers
-Particulate
mur
I
I
- Extraction CH&l
DNPH/HPLC Aldohyder C.rbony,.
On-line analysis
Organic fraction
-co2 co
inorganic fraction
I
I
HPLC
Extraction iPA/H20
02
-
-NOI
B
I
-- nc
so2
Soluble
w ion-Chromatography
Figure 7 Analytical procedures for measuring the composition of diesel engine exhaust Y
30 1
Figure 8 shows Scanning Electron Microscopy pictures of particulate filters in the various stages of analysis. Also methods were developed to measure the particulate size distribution [35]. The results of these analyses have to be carefully interpreted, however, as changes in the particulate properties during the sampling procedure may occur [36]. It should further be mentioned that the present particulate measurement method is an integral procedure; time-resolved methods are still under development [37]. LEGISLATION
The legislation in relation to the tailpipe emission from diesel engines is fundamentally dependent on the type of application. For diesel engines used in on-road transportation means a distinction is made between light and heavy duty applications. For light duty applications, the emission tests are made with vehicles on a vehicle dynamometer according to the FTP 75-, the ECE- or ECE+EUDC- and the 10-mode-procedures. These tests are inherently of a transient nature. Some countries do include a steady state smoke test. The components subjected to emission limits are summarized in Table 5, for USA, W-Europe and Japan. TABLE 5
Present diesel exhaust components limitations for light duty vehicles COUNTRY
USA
1 COMPONENTS
' I
W-EUROPE
$ CO,HC,NOx $ Particulates $ CO,HC, NOx
I $ OIC+NOx) $ Particulates $
JAPAN
Smoke
$ CO,HC,NOx
UNITS
TEST
g/mile g/mile
FrP-75 FTP-75
g/test g h g/test g h % opacity PPm
I
ECE,ECE+EUDC FrP-75 ECE,ECE+EUDC FrP-75
Steady state at average rpm
1 1 0 mode
For heavy duty applications, the emission tests are made with engines on an engine dynamometer. Both steady state and transient tests are applied, depending on the country considered.
Figure 8 a,b Scanning Electron Microscopy pictures of
( a ) a diesel particulate filter paper;
( b )a loaden filter paper;
303
Figure 8 c,d Scanning Electron Microscopy pictures of
(c) the loaden filter paper after extraction with CH3Cl
( d ) the loaden filter paper after the extraction with CH3C1followed by the
extraction with IPAIH20.
304
In the steady state tests, the engine is run at a well defined number of load-speed combinations, in a well defined order. An example of such a test is shown in Figure 9 [38, 391. The exhaust gas is collected in each of the loadspeed combinations. In a transient test, not only various combinations of engine speed-engine load adjustments, but also the transitions inbetween these states are part of the test. The exhaust gas is collected during the whole test and afterwards analyzed. An example of such a test is shown in Figure 10 [38, 39,401. Due to the complexity of the test equipment, steady state simulations of the transient tests are being used for development purposes also. In addition, smoke tests, which are of steady state and of transient nature, are generally applied. Table 6 summarizes the at present limited components and test procedures in the USA, W-Europe and Japan.
TABLE 6 Present diesel exhaust components limitations for heavy duty vehicles COUNTRY
COMPONENTS
UNITS
USA
$ CO,HC,NOx $ Particulates $ Smoke
g/hP*h @P.h % opacity
Transient test Transient test 3-modes with transient & steady state conditions
W-EUROPE
$ CO,HC,NOx
g/kwh
13-mode steady state 13-mode steady state 1 -mode steady ;tate at average rpn
JAPAN
$
Particulates
$
Smoke
$ CO,HC,NOx
$ Smoke
% opacity
PPm % opacity
TEST
6-mode steady state Steady state,full load,various rpm
9
0
0
C
0
0-
0
03
N c3
n w u w
w
z- w3
s
Figure 9 The ECE-R49-13-mode steady state test for heavy duty diesel engines represented in the load-speed diagram [MI
Figure 10 The US transient test cycle for heavy duty diesel engines represented in the load-speed diagram
“I 305
306 REDUCTION OF TAILPIPE EMISSION
General From the description above, it becomes clear that the emission from diesel engines is a complex phenomenon. The composition, the properties and the amount of these emissions depend not only on strictly technical parameters such as engine design and engine operation characteristics or fuel and lube oil composition, but also on how these emissions are measured and defined. This is schematized in Figure 11. Consequently, there are numerous ways to influence these emissions; the possible measures are highly interrelated and their application is dependent on the end use of the diesel engine. Test cycle I
Fuel cetane index
o Gaseous components
Fuel additives
I
Lube oil additives
t d
0 amount 0 composition
,
Engine design and operation principle
i,
Engine load, speed and displacement
o Particulates 0 amount
0 composition 0 size distribution
1 7
consumption
1I
Sampling and
Figure 11 Parameters affecting the emission of diesel engines
Engine design and fuel urouerties The design of an engine and the operation characteristics do considerably influence the composition, the properties and the amount of the emissions. This is exemplified by Figure 1 2 , which shows the amount and the composition of the particulate matter emitted as a function of the engine air to fuel ratio, which is the result of a well defined engine load-engine speed adjustment for a given engine [38]. Another example is shown in Figure 13, where the exhaust sulfur dioxide content is plotted for various engine air to fuel adjustments as a function of the fuel sulfur content.
307
Figure 12 Amount and composition of diesel particulates as a function of the engine air to fuel ratio (Direct Injection Engine; low sulfur fuel; AVL-&mode simulation of the US transient test cycle) [38]
I4O1 120
S-content
---
\
0 35 Gew -7. 0 25 Gew -7r
0 10 Cew -7.
Lambda
Figure 13 Sulfur dioxide content of the diesel engine exhaust gas as afunction of the fuel sulfur -content and the engine air tofuel ratio
308
It is to be expected that major improvements in the diesel engine emissions will result in the coming years from engine design modifications. However, it is not obvious that the emission targets can be achieved simultaneously by engine design modifications only, due to the different origins of the various emissions. This is exemplified in Figure 14: by improving the combustion of the fuel in the engine, lower particulate emissions can be achieved, but the higher temperature in the engine, resulting from the better combustion, might lead to an increased NOx-formation [41,
42,44,451. Also the formulation of the fuel does have a major impact on the emissions. For fossil fuels, the relation to emissions is complex. The only clear relation exists between the sulfur content of the fuel and the sulfur dioxide content in the exhaust gas. Recent research indicates that the amount of particulates emitted might be related to the aromatics content of the fuel, especially the diaromatics present [46,47,48,49, 50, 511. 0.7
1988 /Cart I01
0.6 0.5
. P
a r
9
0.4
u) I
-
3-
0.3
n b
0 .2 0.1
I 1994 e 9 1 Buses) L
Engineering h
l
1l 0 ~
y
l
I
~
1988 Llrnltl
II 0
2
3
4
5
6
7
NOx(ghW
8
9
1010.7 Source:
Rkardo 119891
Figure 14 Relation between the engine-out emission of particulates and the engine-out emission of nitrogen oxides for different heavy duty diesel engine designs 1441 Currently, a lot of interest exists in using non-fossil fuels, such as natural gas, alcohols or esters. Not all of these fuels are compatible with the current diesel combustion technology. Also this kind of fuels might have a positive effect on one kind of emissions and a negative effect on another kind of emissions, as is exemplified in Figure 15. Nevertheless, the kind of fuel used will remain an important parameter in the composition of the tailpipe exhaust gas, especially when exhaust gas aftertreatment devices are considered [52, 53, 541.
309 Aldehydes [mglml
50
PAH
m
20 0.26
10
0.2 -
co
NOx
Particulates
Aldehydes
0conventional diesel fuel 0 rapeseed methyl ester sourc.:
vw 119891
Figure 15 Passenger car diesel engine emissions in the FTP 275-cycle using conventional fuel and rapeseed methyl ester [54] Exhaust aftertreatment bv filtering devica A powerful way to drastically reduce the amount of particulates in the tailpipe outlet is the use of filtering devices. Basically, devices based on two filtering principles are used. A ceramic wall flow filter, of which the operation principle is shown in Figure 16, is a shallow bed filtration device. On the other side, deep bed filtration devices such as wire mesh filter with metallic or ceramic wires and ceramic and/or metallic foams are investigated also. Figure 17 shows an example of a metallic foam filter. The filtering devices operate to some extent discontinously: first particulate material is collected and then the loaden filter has to be regenerated. The regeneration consists in burning off the accumulated particulate material. The difficulty is that the exhaust gas temperature of a diesel engine is usually too low to regularly reach the minimum temperature needed for soot burn off. If an excess of soot is collected on the filter, the exhaust gas temperature raises due to the increased back pressure, leading to a sudden bum off, which might cause the filter temperature to raise above the melting point of the filter material [55, 56, 57, 58, 59, 601.
310
Figure 16 Schematic diagram of a wallfilter
[%I.
Figure 17 Example of a metallic foam filter
31 1
For a controlled regeneration two basic approaches exist: either the exhaust gas or the filter is heated in well defined intervals to the minimum temperature necessary to start the bum off, or the minimum temperature required for bum off is lowered by using a material that catalyzes the oxidation of the accumulated soot. These approaches are summarized in Figure 18. Of course, both approaches can be combined. The temperature of the exhaust gas can be raised by controlled throttling, or by applying external energy, for example electrical heating. To raise the temperature of the filter, several heating procedures were developed. These procedures include the use of electrical energy, in the form of microwave heating by using a conducting filter material or by equipping a ceramic filter with heating wires. Alternatively, the filter can be heated by a burner. To that aim also automatic dual filter systems were developed, in which one clean filter is loaded by the exhaust gas and simultaneously the second filter is regenerated off line by the burner [61, 62, 631.
o
Exhaust gas heating
o
o
Exhaust gas throttling
0 Electrical heating
o
Electrical heating
0 Burner systems
o
Filter heating
0 Lowering soot o Lowering soot ignition temperature ignition temperature before the filter in the filter 0 Fuel additives
o
Catalytic coating
Microwave heating
Figure 18 Schematic representation of the various diesel particulate filter regeneration principles For controlleable fleet operation, i.e. vehicles which return systematically after a controlled time of operation to the same place, an onboard regeneration system is not necessary. During the periods of nonoperation, the filter can be removed and regenerated off board. For the second approach to regenerate a filter material, i.e. by lowering the required soot bum off temperature, several procedures were
312
developed, as is also shown in Figure 19. One alternative is to add the material capable of lowering this soot ignition temperature to the fuel. To this aim, organic derivatives of iron or cerium for example are used. The organic derivatives are burned in the engine, allowing the inorganic oxide to be built into the particle, so that intimate contact with the soot is assured. A second alternative is to add continuously or discontinuously an additive into the exhaust stream. To this aim, organic derivatives of copper for example have been proposed [64,65,66,67, 681.
I
I Additive
/
soot particle
Addithm
Filter
V
I
soot particle
AdditiVO
I Air FWl
2
I Soot particle
PM+BM
or BM Coathg
Figure 19 Principles of diesel particulate filter regeneration by lowering the soot ignition temperature Finally, the third alternative is to apply the soot combustion catalyst as a coating on the filter. Several oxides, especially vanadiumoxides, were found to be very effective [69, 70,71,72, 73,741. The advantage of filter systems is that their application is widely independent of the composition of the particulate emissions and that their performance for reducing the particulate emission is very high. Some of the present day regeneration systems however require a lot of space and/or are costly due to their complexity. To reduce the emission of gaseous constituents, such as carbon monoxide and aldehydes a precious metal based coating can be applied upon the filter, or a precious metal based catalyst can be added in line with the filter system.
313
Besides the filter systems decribed above, the application of other types of gadsolid separation devices to diesel exhaust was discussed also. The patent literature mentions cyclones, electrostatic precipitators and gas washing equipment [75,76]. Exhaust gas aftertreatment - Non-filtering devices Recently the use of oxidation catalysts as a mean to also reduce the particulate emissions from diesel engines drew a lot of attention. By a precious metal based oxidation catalyst, carbon monoxide and gaseous hydrocarbons can easily be removed to a high extent from the exhaust gas, since there is always excess oxygen. The oxidation catalyst can also oxidize the liquid hydrocarbons, which are part of the particulate matter, and therefore reduce the amount of particulates to some extent. The performance in this respect is strongly dependent on the composition of the exhaust raw emission. The technical challenge was to formulate the oxidation catalyst such that it selectively oxidizes carbon-containing components at the low exhaust gas temperatures typical for diesel exhaust, and that it does not oxidize sulfur dioxide or nitrogen oxide in the range of exhaust temperature occuring during the engine use [77,78,79, 801. Although the oxidation of sulfur dioxide also occurs in the atmosphere to an appreciable rate, it is undesired since the sulfuric acid formed contributes to the amount of tailpipe particulates by definition, and since in street canyon situations enhanced sulfuric acid concentrations could occur [81, 821. Especially for the application to heavy duty engines, the oxidation catalyst has to be formulated to avoid deactivation by poisoning, for example by sulphur oxides such as to match the long service life of these engines [83, 84, 851. The operation principle of the oxidation catalyst is schematized in Figure 20, an example for its performance in reducing the particulate emission from a passenger car with a naturally aspired diesel engine is given in Table 7. Reduction of nitrogen oxide emissions As discussed above, the contribution of diesel engines to the emission of nitrogen oxides is becoming more and more important. For the reduction of nitrogen oxides in a lean gas mixture, several technologies were developed in the past. One of these technologies is the catalytic reduction of nitrogen oxides by using external reducing agents, such as ammonia. This technology is successfully applied to reduce the nitrogen content in the stack gas of fossil fuel power plants and has also been successfully adapted to reduce the nitrogen oxides in the exhaust gas of stationary diesel engines.
314
co Aldehydes HC PAH
so2 NO
0 0 0
$
I
H20
4
SO2 NO
Flow through monolith with catalytic coating
co + 112 0 2 +c02 HC + 0 2 PAH + 0 2 Aldehydes
d Cop+ H 2 0
+C o n + H 2 0 + 0 2 +COP+ H 2 0
Figure 20 Operation principle of a diesel oxidation catalyst
TABLE 7 Performance of a fresh diesel oxidation catalyst in the reduction of the particulate emission of a passenger car NA/IDI diesel engine, in the US-FTP-75 cycle Test phase cycle
Phase I without catalyst
Particulate total(mg)
'
'
HC-Lube oil derived(mg) HC-Fuel derived(mg) Sulfates and water(mg) Soot(mg) I
with without with catalyst catalyst .catalyst
without catalyst
with catalyst
0.38
0.16
0.56
0.25
0.72
0.27
0.28
0.04
0.35
0.14
0.28
0.06
0.18 4.54
0.11 3.45
0.14 3.47
0.14
0.21 4.05
0.07 2.57
2.60
315
The application to diesel engines in vehicles is more difficult, because of the varying catalytic reaction conditions (temperature, space velocity, exhaust gas composition) and because of the need to carry the external reducing agent in the vehicle [86, 87, 881. Therefore, the development of catalysts for the direct decomposition of nitrogen oxides and for the reduction of nitrogen oxides with the reduction agents present in the exhaust gas (carbon monoxide, hydrocarbons, soot) was intensified [89, 901. CONCLUSION
Diesel engines do emit a variety of components, which at the tailpipe outlet appear in the solid, the liquid and the gaseous state. The composition of the exhaust stream is dependent on the engine type, the engine operation conditions, the fuel and lube oil composition and consumption. The legislation relative to diesel engine tailpipe emissions depends on the end use of the engine: the kind of tests, the applicable emission limits and the required durability are different for heavy duty applications on one side and for light duty applications on the other side. The main environmental concern presently in relation to diesel engine emissions originates from the particulates in the exhaust. For the future also the amount of nitrogen oxides emitted will have to be reduced. By engine design modifications, major improvements in the emission of carbon monoxide, hydrocarbons, nitrogen oxides and particulate matter was achieved already and is expected to be further reached in the future. However it is unlikely that the emission limits for all these components can be met simultaneously by engine design modifications only. Also the formulation of the fuel plays an important role in the composition of the exhaust stream. For fossil fuels however the exact role of each of the fuel parameters is not well defined at present, except for the impact of the fuel sulfur content. The use of alternative fuels such as compressed natural gas or alcohols is still a matter of development, since not all of the alternative fuels are compatible with the present diesel engine technology. Also alternative fuels might be beneficial to reduce the amount of some of the exhaust components on one side, but might represent a disadvantage for the emission of other components. Therefore a lot of interest exists to apply exhaust aftertreatment devices to diesel engines. These aftertreatment devices can be divided in a group comprising filtering devices and in a group of non-filtering devices. For the group of filtering devices, the major developments are in the regeneration of the loaden filters. To this aim, external heat supplying devices, fuel additives and catalytic coatings have been developed. The group of non-filtering
316
devices includes special oxidation catalysts, besides applications of various gadsolid separation equipments. The reduction of the amount of nitrogen oxides in the exhaust gas, for which technologies applicable to stationary emission sources exist, still encounters difficulties in the instationary use. REFERENCES J .Coudre "Production of multicylinder diesel engine" Proceedings International Seminar "Diesel engines: Prospects 1990 - 2000" Societt des Ingenieurs de I'Automobile, Lyon (1990) G.Latapie "Situation on world market 1985 - 1989: Diesel engines for passenger cars" Proceedings International Seminar "Diesel engines: Prospects 1990 - 2000" SocietC des Ingenieurs de YAutomobile, Lyon (1990) N.D. Whitehouse,E.Clough,S.O.Uhunmwangho "The development of some gaseous products during diesel engine combustion" SAE Technical Paper Series 800028 (1980) T.W.Ryan, J.O.Storment, B.R.Right, R.Waytulonis "The effect of fuel properties and composition on diesel engine exhaust emissions - a review" SAE Technical Paper Series 810953 (1981) GGreeves, C.H.T.Wang "Origins of diesel particulate mass emission" SAE Technical Paper Series 810260 (1981) C.A.Amman, D.L.Stivender, S.L.Plee, J.C.Macdonald "Some rudiments of diesel particulate emissions" SAE Technical Paper Series 800251 (1980) K.Otto, M.H.Stieg, M.Zinbo, L.Banosiewicz "The oxidation of soot deposits from diesel engines" SAE Technical Paper Series 800336 (1980) G .W. Smith "Kinetic aspects of diesel scot coagulation" SAE Technical Paper Series 820466 (1982) L.B.Ebert "Is soot composed predominantly of carbon clusters" Science, 247, p. 1468 - 1471 (1990) W.Cartellieri, F.Ruhri, P.Tritthardt "Der B e i n g des Schmierols zur Partikelemission von Nutzfahrzeug-Dieselmotoren" 2. Aachener Kolloquium "Fahrzeug und Motoren-Technik" - RWTH Aachen (1989) W.J.Mayer, D.C.Lechman, D.L.Hilden "The contribution of engine oil to diesel exhaust particulate emissions" SAE Technical Paper Series 800256 (1980) W.H.Lipken, J.H.Johnson, C.T.Vuk "The physical and chemical character of diesel particulate emissions - measurement techniques and fundamental considerations" SAE Technical Paper Series 780108 (1978) N.J.Kham, J.H.Johnson, D.G.Leddy "The characterization of the hydrocarbon and sulfate ,fractions of diesel particulate matter" SAE Technical Paper Series 7801 11 (1978)
317
R.A.Baranescu "Influence of fuel sulphur on diesel particulate emissions" SAE Technical Paper Series 881 174 (1988) M.Chevrier "Toxicit6 compartk des gaz d'khappement diesel et essence" Proceedings International Seminar "Moteurs diesel pour vehicles automobiles et utilitaires" Societ6 des Ingenieurs de l'Automobile, Lyon (1990) K.D.Van den Hout, R.C.Rijkeboer "Diesel exhaust and air pollution" Report R86/038TNO (1986) M.P.Walsh "Controlling motor vehicle emissions" Platinum Metals Review, 33,4,p. 194 - 212 (1989) L.Ewetz, P.Camner "Health risks resulting from exposure to motor vehicle exhaust" Swedish Government official report SOV 1983:27,28; National Swedish Institute of -Environmental Medicine, Stockholm (1983) Y.A.Courtois, B.Festy, B.Vanrel1 "Biological effects of diesel exhaust: Experimental data" Proceedings International Seminar "Diesel engines: Prospects 1990 - 2000" Societt des Ingenieurs de l'Automobile, Lyon (1990) W.Stober "Estimates of health hazards from exposures to diesel exhaust" Annual Meeting Air and Waste Management Association, Paper 89-34A-2,82nd Anaheim (1989)" H.Klingenberg, H Winneke "Studies on health effects of automotive emissions: how dangerous are diesel emissions?" The Science of the total environment, 93, p. 95 - 105 (1990) J.Bond, J.L.Manderly, R.K.Wolff "Concentration and time-dependent formation of DNA adducts in lungs of rats exposed to diesel exhaust" p. 127 - 135 (1990) Toxicology, 60,(I-2), B.Festy "Biological effects of diesel exhaust - a review" Proceedings International Seminar "Diesel engi(es: Pr)spects 1990 - 2000" Societ6 des Ingenieurs de l'Automobile", Lyon (1990) Deutsche Forschungsgemeinschaft "Maximale Arbeitsplatzkonzentrationen und biologische Arbeitsstofftoleranzerte"MAK-Liste 1988" Bundesarbeitsblatt 12 (1988) Volkswagen A.G. "Unregulated Motor Vehicle Exhaust Gas Components" Volkswagen Research and Development, Wolfsburg (1990) N.Gorissen "Developments in air pollution caused by highway traffic in the Federal Republic of Germany: effects and preventive measures" The science of the total environment, 93, p. 21 - 42 (1990) G.A.Arhens, A.Friedrich, N.Gorissen "NOx-Emissions in the transport sector 1985 - 1988 and Measures to reduce them" GIT Fachzeitschrift Laboratorien, 7,p. 861 - 868 (1990) M.P.Walsh "Worldwide developments in motor vehicle diesel particulate control" SAE Technical Paper Series 890168 (1989)
A.Harmng, K.H.Lies "Schnellverfahren zur Bestimmung der PAK-Werte" Motortechnische Zeitschrift 51, 1, p. 12 - 17 (1990) J.M.Perez, F.Lipari, D.E.Seizinger "Cooperative development of analytical methods for diesel emissions and particulates - solvent extractables, aldehydes and sulfate methods" SAE Technical Paper Series 840413 (1984) K.H.Lies, J.Kraft "Schwefelverbindungenim Automobilgas: Meflverfahren und Ergebnisse" VDI-Berichte, 314, p. 17 - 19 (1978) Coordinating Research Council "Chemical methods for the measurement of unregulated diesel emissions" CRC Report No. 551, Atlanta (1987) K.H.Lies, A.Postulka, H.Gring "Particulate emissions from diesel engines - evaluation of measurement and results" SAE Technical Paper Series 830455 (1983) J.Winckler "Spezifsche Probleme bei Probenahme und Messung von Dieselabgasen" in "Kurzfassung der Tagung VDI": "Emissionsminderung Automobilabgase Dieselmotoren", Nurnberg (1985) D.F.Dolan, D.B.Kittelson, D.Y.H.Pui "Diesel exhaust particle size distribution measurement techniques" SAE Technical Paper Series 8001 87 (1980) M.K.Abbass, G.E.Andrews, P.T.Williams "Diesel particulate composition changes along an air -cooled exhaust pipe and dilution tunnel" SAE Technical Paper Series 890789 (1989) R.Whitby, R.Gibbs, R.Johnson, B.Hill,S.Shimpi, R.Jorgensen "Real time diesel particulate measmment using a tapered element oscillating microbalance" SAE Technical Paper Series 820463 (1982) P.Zelenka, W.Kriegler, P.L.Hermg, W.P.Cartellieri "Ways towards the clean heavy duty diesel" SAE Technical Paper Series 900602 (19%) F.X.Moser, E.Haas, H. Schlogl "Zur Partikelemissionvon Nutzfahrzeug-Dieselmotoren" Motortechnische Zeitschrift 51,5, p. 186-193 (1980) P.Zelenka "Minderung der Abgasemission von Dieselmotoren durch Oxidationskatalysatoren Fahxzeug- und Motortechnisches Seminar ;TH Darmstadt (1990) P.Zeranski "Uber die Zukunft des Dieselmotors) Automobilindustrie, 2, p. 201 - 212 (1989/ W.Haas "The future of the diesel engines: trends and development potential" Proceedings of the conference "Engine and environment"AVL Graz 1989) P.Walzer "Passenger car Diesel Engine; The main Development emphasis" Proceedings of the conference "Engine and environment"AVL Graz 1989) J.R.Needham, D.M.Doyle, S.A.Faulkner, H.D.Freeman "Technology for 1994" SAE Technical Paper Series 891949 (1989)
319 H.Bergmann "Schadstoffminderungspotentiale modemer Diesel-Nutzfahrzeuge" Proceedings Seminar "Umweltproblemedurch Dieselmotoren" Fortbildungszentrum Gesundheits- und Umweltschutz Berlin e.V. (1990) C.T.Flanigan, T.A.Litzinger, R.L.Graves "The effect of aromatics and cycloparaffiis on DI diesel emissions" SAE Technical Paper Series 892130 (1989) M.K.Abbass, G.E.Andrews, P.T.Williams, K.D.Bartle "The influence of diesel fuel composition on particulate PAH-Emissions" SAE Technical Paper Series 892079 (1989) T.J.Russe1, R.Fiat "Diesel fuel quality" Proceedings International Seminar "Diesel engines: Prospects 1990 - 2oOO" Societt des Ingenieurs de l'Automobile, Lyon (1990) F.Filippi "Which fuel for the truck of the year 2000" Proceedings International Seminar "Diesel engines: Prospects 1990 - 2000" SocietC des Ingenieurs de l'Automobile, Lyon (1990) K.Weidmann, H.Menrad, K.Reders, R.C.Hutcheson "Diesel fuel quality effects on exhaust emissions" SAE Technical Paper Series 88 1649 (1 988/ M.K.Abbass, G.E.Andrews, P.T.Williams, K.D.Bartle "The influence of diesel fuel composition on particulate PAH emissions" SAE Technical Paper Series 892079 (1989) T.L.Ullmaann, C.T.Hare, T.M.Barnes "Emissions from two methanol powered busses" SAE Technical Paper Series 860305 (1986) G.E.Hundley "Low emissions Approaches for Heavy-Duty Gas powered urban vehicles" SAE Technical Paper Series 892134 (1989) K.Weidmann, H.Menrad "Rapeseed methylester in the Diesel engine" Motortechnische Zeitschrift 50, 2, p.69 - 73 (1989) F.Brear "An introduction to diesel exhaust aftertreatment technology with special reference to the control of particulates" in "A short course on diesel particulates" The University of Leeds (1990) R.W.Horrocks "Particulate control systems for diesel engines" The Institution of Mechanical Engineers Paper C349/87, p. 319 - 334 (1987) P.Oser,U.Thomas "Particulate control systems for Diesel engines using catalytically coated and uncoated traps with consideration of regeneration techniques SAE, Warrendale (1 989) K.Esser, M.R.Montierth "The wall flow diesel filter system - a review" Paper E 2.6 presented at the second seminar ATA-MAT, Turin (1989) J.Kitagawa "Thermal shock failures of ceramic diesel particulate filters" Paper E 2.7 presented at the second seminar ATA-MAT, Turin (1989) T.Minah, A.Maurer, L.Gauckler, J.P.Gabathuler "Open-pore ceramic foam as diesel particulate fiter" SAE Technical Paper Series 890172 (1989)
320 W.R.Wade, V.Durga, N.Rao "Diesel engine exhaust particulates filter with.controlled incineration for regeneration" US Patent 463686 (Ford Motor Co.) (1983) G.Revenot "Regeneratingexhaust gas filter by periodically injecting cracked vapour of low boiling organic liquid* European Patent Application 84401 194 (Regie Nationale des Usines Renault) (1984) F.Pischinger, G.Lepperhoff, U.Pfeifer, K.Egger, G.Hiithwoh1 "Modular Trap and Regeneration System for Buses, Trucks and other applications" SAE Technical Paper Series 900325 (1990) AZ.F.Ahlstrom, C.V.1.Odenbrand "Catalytic combustion of soot deposits from diesel engines" Applied Catalysis, 60, p. 143 - 156 (1990) J.C.Frohne, H.J.Guttmann, H.Riesig, H.K.Schadlich "Modellierung der Diesehpverbrennung in der Thermowaage" Chem. Ing. Tech., 62, 1 (1990) R.K.Herz, R.M.Sinkevitch "Reactors for investigating soot combustion on filter surfaces" Carbon, 24 (4), p. 457 - 462 (1986) S.Meinrad, C.Giorgio "Laboratory results in particulate trap technology" SAE Technical Paper Series 890170 (1989) K.Pattas, ZSamaras, N.Patsatzis, C.Michalopoulou, G.Zagou, A.Stamatellos, M.Barkis "On-road experience with trapoxidiser systems installed on urban busses" SAE Technical Paper Series 900109 (1990) E.Koberstein, H.D.Pletka, H.Volker "Catalytically activated diesel exhaust filters - Engine Test Methods and Results" SAE Technical Paper Series 830081 (1983) B.Engler, E.Koberstein, H.Volker "Diesel particulate traps - new development and application" SAE Technical Paper Series 860007 (1986) B.Engler, E.Koberstein, H.Volker "Ceramic diesel filters with different catalytic activations and their test results" Paper E 2.9 presented at the second seminar ATA-MAT, Turin (1989) R.E.Marinangeli, E.H.Homeier, FSMolinaro "A laboratory method for determining the activity of diesel particulate combustion catalysts" in "Catalysis and Automotive Pollution Control", A. .Crucq, A. Frennet (Eds.), Elsevier (1987) R.Westerholm et al. "Chemical analysis and biological testing of emissions from a heavy duty diesel truck with and without two different particulate traps" SAE Technical Paper Series 860014 (1986) D.W. McKee "Metal oxides as catalysts for the oxidation of graphite" Carbon, 8, p. 623-635 (1970) R h o n h a r d , U.Projahn "Einrichtung zum Entfemen von Festkorperpartikeln,insbesondereRuptefchen, aus dem Abgas von Brennkraftmaschinen" DE 3841182 A 1 (Robert Bosch) (1988)
32 1
t 891
D.P.Thimsen, K.J.Baumgard, T.J.Katz, D.B.Kittelson "The performance of an electrostatic agglomerater as a Diesel soot emission control device" SAE Technical Paper Series 900330 (1990) B.Engler, E.Koberstein, E.Lox, K.Ostgathe, H.Volker "Diesel emission control with newly developed catalysts" Proceedings Autotech 89 - Birmingham, I M C E (1989) P.Zelenka, K.Ostgathe, E.Lox "Reduction of diesel engine emissions by using oxidation catalysts" SAE Technical Paper Series 9021 1 1 (1990) F.Anisits, G.Reibold "Moglichkeiten zur Partikelbegrenzungbei PKW-Dieselmotoren" Automobil-Industrie, 6,p. 759 - 766 (1989) G.Hunter, J.Schol1, F.Hibbler, S.Bagiey, D.Leddy, D.Abata, J.Johnson "The effect of an oxidation catalyst on the physical, chemical and biological character of diesel particulate emissions" SAE Technical Paper Series 810263 (1981) M.R.Hoffmann, S.D.Boyce "Catalytic autoxidation of aqueous sulfur dioxide in relationship to atmospheric systems" Adv. Env. Sci. Eng., 12,p. 148 - 189 (1983) R.J. Cadle "Formation and chemical reactions of atmospheric particles" Journal of Colloid and Interface Science, 39,p. 25 - 31 (1972) D.W.de Berry, K.J.Sladek "Rates of reaction of SO2 with metal oxides" The Canadian Journal of Chemical Engineering, 49 ,p. 781 - 785 (1971) J.R.Sohn, H.W.Kim "Catalytic and surface properties of zrO2 modified with sulfur compounds" Journal of Molecular Catalysis, 52,p. 361 - 374 (1989) S.W.Nam, G.R.Gavalas "Adsorption and oxidative adsorption of sulfur dioxide on y-Alumina" Applied Catalysis, 55,p. 193 - 213 (1989) W.Held, A.Konig, T.Richter, L.Puppe "Catalytic NOx-reduction in net oxidizing exhaust gas" SAE Technical Paper Series 900496 (1990) M.Iwamoto, H.Yahiro, K.Tanda "Catalytic decomposition of nitrogen monoxide over copper ion-exchanged zeolites. Influence of zeolite structure and aluminium content on the catalytic activity" in "Successful Design of Catalysts", T. Inui (Ed.), Elsevier (1988) Y.Fujitani, H.Muraki, S.Kondoh, M.Fukui "Catalytic removal of nitrogen oxides from waste gases such as from internal combustion engines or nitric acid manufacture" Japan Patent Application 86/247958(1986) K.Y oshida,S.Makino,S.Sumiya,G.Muramata,R.Helferich "Simultaneous reduction of NOx and particulate emissions from diesel engine exhaust" SAE Technical Paper Series 892046 (1989) B.Cooper, J.E.Thoss "Role of NO in diesel particulate emission control" SAE Technical Paper Series 890404 (1989)
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V., Amsterdam
323
KINETICS OF SOOT OXIDATION ON POTASSIUM-COPPERVANADIUM CATALYST P. Ciambelli *, P. Parrella, S . Vaccaro
Dipartimento di Chimica, Universita' di Napoli "Federico II" via Mezzocannone, 4 - 80134 Napoli. ABSTRACT Combustion of carbonaceous materials such as Diesel engine soot, carbon black and graphite on a potassium-copper-vanadium catalyst supported on a -alumina has been studied. Catalyst characterization by thermal analysis and X-ray diffraction indicates the formation of mixed K/Cu/Cl and K/V/O phases after calcination at 700OC. The presence of KCl has been also detected. Soot combustion experiments show a dramatic effect of the catalyst, burn off temperatures being about 30O0C lower than those obtained with thermal combustion. Kinetics of soot catalytic combustion has been investigated in a differential flow reactor. A linear dependence on both soot mass and oxygen partial pressure has been found. Comparable reaction rates with the same dependence on the reactant concentrations have been determined also in the case of carbon black. The apparent energy of activation for the catalytic process is 85-100 kJ/mol. The kinetic equation enables to describe the process of catalytic combustion up to conversions of 50%. INTRODUCTION
Pollutants in the exhaust gases of automotive engines represent a serious hazard for human health (1). In the case of Diesel engines such pollutants are CO, NOx, SO2, unburned hydrocarbons and soot (2). Among these the latter appears to be the most dangerous. Indeed, it is inhalable being mostly made of fine particles (90 9% by weight < 1 pm ) (3). Furthermore, soot particles are constituted by a carbonaceous matrix containing trace metallic inclusions and gases and liquids adsorbed on its surface. These represent a large number of chemical compounds (>10000) (2) some of which (Polycyclic Aromatic Hydrocarbons) are suspected to be cancerous (4). In the last years the growing concern of worldwide countries toward environmental problems triggered the development of new technologies for pollutant removal from engine exhausts. In particular, ceramic honeycomb wall-flow filters and wire mesh traps, employed for Diesel exhaust filtering,
*
Present address: Dipartimento di Ingegneria Chimica e Alimentare Universith di Salerno 84081 Baronissi (SA).
324
showed a high soot removal efficiency (5). In these systems periodic soot bum out is required to allow for filter regeneration. However, spontaneous thermal ignition of captured soot is difficult to obtain needing high temperature or high oxygen partial pressure in the gas exhausted by the engine. This leads to soot accumulation in the trap that causes high engine backpressure and filter overheating during regeneration, with consequent trap failure (6). The use of a catalyst to decrease ignition temperature allows for more frequent combustions of captured soot so keeping low both filter pressure drop and operating temperature. Research work in this field proceeded through two different lines. Car (7) and filter manufacturers (5) operated on full scale systems building catalytic traps and coupling them to the exhaust manifolds of Diesel engines. However, they used metal based catalysts often borrowed from spark ignition engine applications. The main role of the catalyst in this case is to catalyze the combustion of unburned hydrocarbons present in the gas, slightly increasing exhaust temperature and indirectly enhancing soot oxidation. On the contrary, academic researchers operated on a laboratory scale developing catalysts specific for soot combustion and testing their activity with samples of carbonaceous materials. The catalyst, generally, contains transition metal oxides or alkaline metal compounds and is derived from those employed for coal oxidation and gasification (8-1 1). This type of catalyst plays a direct role in the reaction of carbon combustion as it has been shown by studies carried out on graphite oxidation (12). A supported catalyst (1 37AA), containing vanadium, copper and potassium compounds and active already at 300-350°C in soot oxidation, has been developed (13) and patented (14) by the present authors. Its catalytic activity in the oxidation of Diesel soot has been investigated. The results of catalytic combustion experiments, carried out with a differential reactor, are discussed in order to derive a reaction kinetic equation. EXPERIMENTAL
Catalyst Catalyst 137AA was prepared by wet impregnation of a-A1203 powder with aqueous solutions of ammonium vanadate N H 4 V 0 3 , copper chloride CuC12*2H20 and potassium chloride KC1 (BAKER Chemicals). After drying at 120°C the solid sample was calcinated at 700°C overnight (14). Chemical analysis of the catalyst was performed by coulorimetry and atomic absorption spectrophotometry. Catalyst characterization was carried out by different techniques such as X-ray diffraction, thermal analysis and nitrogen adsorption for BET surface area measurement.
325
Carbonaceous materials Soot particulates were generated by a direct injection single cylinder Diesel engine (Ruggerini RP 170), naturally aspirated, aircooled, with a compression ratio of 18 and displacement of 746 cm 3 . Commercially available Diesel fuel, with 0.3 wt % of sulfur and cetane number = 50, was fed to the engine. Soot samples were collected during steady state operating conditions of the engine : 2000 r.p.m., ratio between air and fuel mass flow rates (a)= 17, 25 and 51, start of combustion -5" crank angle before top dead center. From now on soot samples obtained at different (a) will be recalled as DS17, DS25 and DS51, respectively. Their volatile content was determined by thermal analysis up to 800°C in nitrogen flow. Commercial carbon blacks (CBN110 and CBN330 DEGUSSA) and pure graphite were employed as reference materials. Reactivitv tests The performance of catalyst 137AA in the combustion of such different carbonaceous materials was preliminarly determined by simultaneous TG-DTG-DTA analysis carried out with a NETZSCH STA 409 analyzer. Samples, handly mixed with catalyst without pressing, were loaded in an alumina crucible and heated in flowing high purity cylinder air (200 cm3/min) at 20"C/min rate from room temperature. Soot-kaolin mixtures at the same weight ratio as for soot-catalyst mixtures were loaded to the analyzer when performing thermal combustion tests without catalyst. REACTOR
TC
OVEN
CO2 ANALYZER
I I
I
I k L J COMPUTER
b'ig. I :Scheme of the experimental apparatus for kinetic runs. (mfc, mass flow controllers; cw, cooling water).
Catalvtic kinetic runs Catalytic combustion tests were determined by the experimental apparatus shown in Fig. 1. Cylinder air and nitrogen (99.999 % purity) were mixed downstream the mass flow controllers (HI-TECH) for
326
obtaining different oxygen partial pressures. A 20mm I.D. glass tubular flow reactor, heated by an electrical furnace driven by a temperature programmercontroller, was loaded with 100 mg of catalyst 137AA mixed to different amounts of carbonaceous material with a mean size of 10 pm. Reactor temperature was measured by a Ni-Cr thermocouple which can be moved inside an axial sheat along the reactor. Gas outlet from the reactor was cooled down by water in a glass heat exchanger and fed to a continuous C 0 2 NDIR analyzer (URAS 3K, HARTMANN & BRAUN). The signal from the analyzer was sent to a personal computer for the calculation of carbon conversion and combustion rate. All kinetic runs were carried out under differential conditions and constant temperature. RESULTS AND DISCUSSION
The chemical content of the main components of catalyst 137AA is K 6.9, Cu 6.2 and V 3.4 wt %. Therefore copper content was partially lowered whereas metal amount and ratio of V and K resulted similar to those employed for catalyst preparation, indicating that calcination at 700°C caused only a small loss of active components. Results of the thermal analysis of the catalyst precursor are shown in Fig. 2a and 2b. From the first figure, where the TG curve (weight loss % of impregnated a-A1203) is reported together with its derivative (DTG curve, arbitrary scale), three main weight loss steps are evident. Up to about 150°C a 2% loss is associated to dehydration of copper chloride also corresponding in Fig. 2b to the endotherm peak of the DTA curve (differential thermal analysis, arbitrary scale) centered at about 130°C. The following TG step in Fig.2a (from 180°C to 350°C) corresponds to a complex set of DTA signals (Fig. 2b) which should be attributed to deammoniation with consequent formation of vanadates. In the range 350450°C Cu-K and Cu-V interactions occur leading to binary phases as effect of the stabilizing role of potassium (15). Indeed, in the absence of potassium complete decomposition of copper chloride has been observed (15). Partial decomposition of copper chloride should be responsible of the small weight loss up to 700°C (Fig. 2a) and of the corresponding exothermic signals (Fig. 2b). The surface area of the catalyst after calcination (2.7 m2/g) was comparable with A1203 surface area, showing that agglomeration of catalyst particles due to calcination was not significant. Prolonged calcination up to 20 hours did not result in any modification of the catalyst. X-ray diffraction data of calcined catalyst are reported in TABLE 1. Preliminary analysis of these results allows to associate the main signals to crystalline phases of the catalyst. Therefore, together with the diffraction peaks
321
of a-A1203, the signals of KC1 and K2CuC14 are present. The remaining peaks can be associated to the formation of mixed phases K V O,, whose X Y identification needs deeper investigation. Nevertheless, the presence of all these phases is compatible with the findings of thermal analysis discussed above. T ("C)
weight
I D S5 (%I
a
1
400
30
\
6001+
DTG
T
800
1000
0.0
*0° , 0 0 0 1
0.5
1.o
1.5
1"
u
t 0.0
1 ' O
20
10
T
a5
1.o
1.5 2 t (hr)
Fig. 2 :Thermal analysis of the catalyst precursor (50 mg samples; 20 "Clmin heating rate). a: TG and DTG curves; b: DTA curve. Performances of catalyst 137AA in the combustion of Diesel engine soot were at first investigated by simultaneous TG-DTG-DTA analysis. Soot samples, generated under the engine operating conditions specified in EXPERIMENTAL, were expected to have different content of volatile matter.
328
TABLE 1 :X-ray diffraction data of catalyst 137AA d(A) 1 5.48 3.96 3.48 3.22 3.19 3.14 2.97
w w m w w s w
(c) (c) (a) (d) (d) (b) (d)
d(A) 1
d(A) 1
2.80 w (d) 2.71 w (c) 2.55 s (a) 2.38 w (a) 2.22 w (b) 2.09 s (a) 1.96 w (a)
1.87 1.81 1.74 1.60 1.57 1.51 1.40
w w m s w w m
(c) (b)
(a) (a) (a,b) (a) (a,b)
(a) a-A1203, (b) KCl, (c) K2CuC14, (d) K,VyO, eight loss
I%( -.-'
if
i!
200
;L-.....
----_
+....-.- ........._....... ..........
400
600
800
T IT1
Fig.3 : Thermogravimetric curves of carbonaceous materials in flowing nitrogen. - _ _ _D S l 7 ; -DS2.5; .*--*-CBNllO; 50 mg samples; 20 "Clmin heating rate) This is confirmed by thermogravimetric analysis of soot in nitrogen flow, shown in Fig. 3 for DS17 and DS25. The former loses only 3% weight up to 800°C whereas the latter shows a higher volatile content reaching 15% weight loss. Carbon black CBNl 10 behaves like a dry soot (DS17) (Fig. 3). When the soot sample is heated in air flow the weight loss curve, corrected for the loss in inert gas, indicates that the combustion starts at about 500°C (curve T in Fig. 4). After 600°C T curve becomes steeper as ignition occurs and at 820°C
329
soot is burnt out. The effect of catalyst on soot combustion is shown in the same Figure (curve C). Temperatures of soot combustion are so much lowered than complete conversion is observed already at about 400°C. Moreover, the temperature at which the DTG curve displays a maximum decreases from 650°C down to 336°C. The comparison between TG analysis of CBNllO with and without catalyst, reported in Fig. 5 , indicates that catalyst activity does not change significantly, ignition and DTG maximum temperatures being comparable to those of soot. In contrast, the range of temperatures in which catalytic combustion of graphite occurs is larger resulting in a higher bum out temperature, especially for crystalline graphite (Fig. 6). Nevertheless, it must be remarked that also in these cases the catalyst allows the combustion temperatures to be lowered by about 300°C. After the preliminary investigations on the activity of catalyst 137AA in the soot combustion process, the study was aimed to derive a kinetic equation for the reaction. In fact, deriving such an equation is an essential step for designing a catalytic trap. Making use of the differential reactor described above, catalytic reaction was studied under different experimental conditions in order to find the dependence of reaction rate on temperature and on reactant concentrations. In Fig. 7 the times necessary to obtain 20% and 40% of soot bum off are reported as a function of the initial soot/catalyst weight ratio in the mixture. Being the weight of catalyst constant (100 mg) for all the runs , the abscissa in Fig. 7 is equivalent to the initial weight of soot loaded in the reactor. Changing that weight by about 5 times does not influence the time of burn off, suggesting that the rate of soot mass change by catalytic oxidation (-dm/dt) is linearly dependent on the mass m. Indeed, only in this case the time to reach a given conversion is independent from the initial amount of the reactant. Even in the case of carbon black the time of burn off is not affected by the soot/catalyst ratio as shown in Fig. 8 for CBNl 10. The influence of oxygen partial pressure on the reaction rate has been also investigated. In Fig. 9 the amount -dm/mdt, as evaluated at 20% soot conversion, is reported as a function of oxygen partial pressure in the range 0.03-0.21 atm. The rate of catalytic combustion is strongly dependent on gas oxygen concentration and the order of reaction is 1. The same order of reaction was found for carbon black catalytic combustion as evidenced by the linear plots of Fig. 10 and 11. After finding the dependence of reaction rate on sample mass and oxygen partial pressure we were able to evaluate the apparent kinetic constant, k,from the equation: dm/dt = - k m Po2
(1)
330
0
25 5075 -
100
C
I1 ___
1 .
Fig. 4 :Thermogravimetric curves of DSI 7 in flowing air. T, no catalyst; C, sootlcatalyst mixture, 15% weight of soot.
0
Fig. 5 :Thermogravimetric curves of CBNllO inflowing air. T, no catalyst; C, CBlcatalyst mixture, 15% weight of CB.
0
Fig. 6 :Thermogravimetric curves of graphite (a=amorphous, c=crystalline) in flowing air. T, no catalyst; C, graphitelcatalyst mixture, 15% weight of graphite,
33 1
3
1-
0
0
2-
-
0
0
r.-1
Fig. 7 :( o )20% and ( )40% burn off times in air of DSI 7 as afunction of the initial weight of soot (T=360"C,gasjlow rate= 500 cclmin).
-
0
0 m
0
"
0
Y
"
9
0
1
Fig. 8 :( o ) 20% and ( ) 40% burn offtimes in air of CBNIIO as a function of the initial weight of CB (T=340°C, gasjlow rate= 500 cclmin).
332
04
dmirndt h ~ n - ~ l
03-
02-
000
005
010
015
PO2 latml
020
025
Fig. 9 : Effect of oxygen partial pressure on catalytic combustion of D S l 7 (T= 340"C, 10% wt of soot, gasflow rate=500 cclmin).
0.0 000
0.05
0.10
0 15
PO2 latml
0.20
0.25
Fig. 10 :Effect of oxygen partial pressure on catalytic combustion of CBNlIO (T=330"C,5% wt of CB, gasflow rate=.500 cclmin).
Fig. I 1 :Effect of oxygen partial pressure on catalytic combustion of CBN330 (T=340"C,5% wt of CB, gasflow rate=500 cclmin).
333
and the apparent energy of activation from the Arrhenius type equation: k = ko exp [-Em (T + 273.1) ]
(2)
In Fig. 12 values of log (- dm/m dt) against the reverse absolute temperature of the catalytic combustion of DS17 and DS51 are reported. For both samples the apparent activation energy is 100 kJ/mol, markedly lower than the values reported for soot thermal combustion (16). The independence of the activation energy on the volatile content of soot indicates that catalyst 137AA produces a direct effect on the oxidation of the carbonaceous matrix of soot. This finding seems to be confirmed by the data obtained from the catalytic combustion of carbon blacks. The apparent energy of activation evaluated from data reported in Fig. 13 and 14 is 84 kJ/mol for CBNllO and 90 kJ/mol for CBN330. It must be pointed out that the values of -dm/dt in Figs 12-14 refer to 20% soot or carbon black conversion. The apparent energy of activation evaluated for higher conversions changes less than 15 kJ/mole up to about 50% bum off. At higher conversions a progressive reduction of the apparent kinetic constant k seems to occur, suggesting that at these values of conversion the process proceedes slower. This could be attributed to the progressively less efficient contact between the two solid phases with the advance of the reaction. Actually, due to the nature of compounds detected by X-Ray diffraction analysis, it should be possible the change to liquid phases. Nevertheless, thermoanalysis results do not give any evidence of thermal effects connected to those transformations. CONCLUSIONS
Catalyst 137AA is able to reduce ignition and combustion temperatures of all tested carbonaceous materials by about 300°C. In the presence of the catalyst the rate of combustion is enhanced by two or three order of magnitude. These are essential features for potential application in soot filtering traps for automotive engines. Given the complexity of the reacting system constituted by two solid phases and one gas phase, a rigorous approach to the reaction kinetics is not simple to be performed. Nevertheless, this work has shown that at least in the range of conversion up to 50% a simple kinetic equation can be used to describe the experimental results. In such an equation the catalytic combustion rate of soot is linearly dependent on both soot mass and oxygen partial pressure. The apparent activation energy is decreased by the catalyst to about 100 kJ/mol. The catalyst exhibits comparable performances in the combustion of carbon black.
334
Fig. 12 :Effect of temperature on the catalytic combustion of DSI 7 (0)and DS51 (0)in air at 20% conversion (5% wt of soot, gas flow rate = 500 cclmin). t
0.1 :
Fig. 13 :EfSect of temperature on the catalytic combustion of CBNllO in air at 20% conversion (5% wt of CB, gasflow rate = 500 cc/min).4 dm/rndt Irnin-'l
1,
t
0.OlL 1.5
1.6
lOOO/r IK-'I
17
1 1.8
Fig. 14 :Effect of temperature on the catalytic combustion of CBN330 in air at 20% conversion (5% wt of CB, gas flow rate = 500 cclmin).
335 LIST OF SYMBOLS
a E
k m PO2
R t T
ratio between air and fuel mass flow rates fed to the engine apparent activation energy, kJ/mol apparent kinetic constant, atm-1 min-1 mass of carbon oxygen concentration, atm gas law constant, kJ K/mol time temperature, "C ACKNOWLEDGEMENTS
The authors gratefully thank Dr. P. Corbo of Istituto Motori C.N.R. for providing Diesel engine soot samples. One of us (P.P.) acknowledges the C.N.R. of Italy for a post-graduate scholarship. REFERENCES M. Chinon Studies, Surf. Sci. Catal. 30 (1987) 1-10 M.P. Walsh Studies, Surf. Sci. Catal. 30 (1987) 51-67 C.P. Fang and D.B. Kittelson, SAE Paper No 840362, (1984) C.F. Funkebush, D.G. Leddy and J.H. Johnson, SAE Paper No 790418 (1979) B. Engler, E. Roberstein and H. Volker, S A E Paper No860007 (1986) S.T. Gulati, SAE Paper No 860008 (1986) 11-18 Y. Niura, K. Ohkubo and K. Yagi, SAE Paper No 860290 (1986) 163-172 T. Inui and T. Otowa, Appl. Catal. 14 (1985) 83-93 P.J. Goethel and T.Y. Yang, J. Catal. 119 (1989) 201-214 T. Inui, T. Otowa and Y. Takegami, J. Catal. 76 (1982) 84-92 C.A. Minus and J.R. Pabst, Fuel 62 (1983) 176-179 K. Hashimoto, K. Miura, JJ. Xu, A. Watanabe and H.Masukami, Fuel 65 (1986) 489-494 P. Ciambelli, P. Corbo, P. Parrella, M. Scialo' and S.Vaccaro , Thermochimica Acta 162 (1990) 83-89 P. Ciambelli, P. Corbo, M. Scialo' and S. Vaccaro, Italian Patent No A40421/88 (1988) P. Ciambelli, P. Parrella, and S. Vaccaro, Submitted for publication (1990) K.B. Lee, M.W. Thring and J.M. Beer, Comb. Flame 6 (1962) 137-145.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V.. Amsterdam
337
CATALYSTS FOR DIESEL POWERED VEHICLES Douglas J. Ball and Robert G. Stack
A.C.Rochester Division of G.M, Flint, MI-48556,USA
ABSTRACT A theory of how diesel oxidation flow-through type converters reduce gaseous and particulate emissions is explained. The effects of substrate, catalyst support and application of noble metals are discussed and investigated. Experiments were performed to determine, (1) the sulfur storage characteristic of silica and alumina catalyst supports, (2) the SO2 oxidation abilities of platinum and palladium and (3) the effects of ceramic and metal monolithic substrates on converter emission performance. Light duty FTP results are also presented.
INTRODUCTION
Diesel powered vehicles are of concern to many world governments. Some governments have chosen to legislate significant reductions of vehicle emissions, while others have them under consideration. Table 1 shows the current and proposed diesel emission laws for several major world markets. The most significant particulate requirements are: light duty US Federal 0.20 g/mile, light duty California 0.08 g/mile, and heavy duty US truck 0.10 ghhp-h. Several European countries have adopted the US Federal light duty requirements. Manufacturers of diesel engines are addressing these emission requirements through the modernization of engines. Changes such as: higher injection pressures, optimization of injection pressure, turbocharging and combustion chamber optimization to name 'a few, have shown potential for reducing particulate mass while maintaining vehicle performance (2). Additional methods such as; fuel additives (3) and exhaust aftertreatment have shown potential for reducing particulates (1,4,5,6,7). The use of diesel exhaust traps (or filters) that collect and oxidize particulates have demonstrated good performance at relatively low mileage (8,9), but have not proven to be effective with respect to high mileage durability. Other drawbacks are the initial costs (lo), vehicle performance and the difficulties the vehicle has in producing adequate trap regeneration conditions at appropriate driving intervals (8).
338
Table 1 - Current and Proposed Worldwide Diesel Emissions Laws Market
Test
Class
HC
CO
US 87
FTP g/mile
LD
0.4 1
3.4
1.0
-
0.20
Calif. 1989
FTP g/mile
LD
0.39*
7.0
0.4
-
0.08
us 91 Truck
HD Cycle gbhp-h
HD
1.3
15.5
5.0
-
0.25
us 94 Truck
HDCycle gbhp-h
HD
1.3
15.5
5.0
-
0.10
Japan 1990
10 Mode ~1,265kg 0.62 gkm >1,265kg 0.62
2.7 2.7
0.72 0.84
-
None None
Japan 1994
10 Mode ~1,265kg 0.62 g/km >1,265kg 0.62
2.7 2.7
0.5 0.6
-
0.2 0.2
EEC 1990
EEC g/test
19 30 25
3.5
5 8 6.5
1.10 1.10 1.10
2.72
-
0.97
0.19
EEC 1993
EEC +EUDC g/km
4.4L 1.4-2.OL >2.0L
-
ALL
-
NOx HC+NOx
Part
*- non-methane As observed by others, a diesel catalytic converter of a flow-through type design can be quite effective in lowering both particulate and gaseous tailpipe emissions (1,2,7,11). Additionally, such a design requires no regeneration and imposes virtually a negligible effect on exhaust system backpressure. To achieve maximum organic fraction and gaseous emissions oxidation, a diesel catalytic converter requires an understanding of the effects certain critical parameters such as; noble metals, substrate design, catalyst support materials and application environment. To obtain the lowest vehicle tailpipe emissions, a total systems approach is required where; the design of the engine, the engine control system and the catalytic converter is balanced.
339
This paper presents a theory on how diesel catalytic converters work. It also includes results of investigations designed to evaluate the effects of the before mentioned critical parameters on emissions. THEORY OF OPERATION
The combustion products of spark ignited gasoline and diesel engines vary considerably. Modern gasoline engines with closed-loop emission controls produce exhaust that is near chemical stoichiometry and at temperatures of 375 to 850°C. The exhaust contains significant amounts of hydrocarbons (HC), carbon monoxide (CO) and oxides of nitrogen (NOx) that must be reduced using a catalytic converter in order to meet the current US light duty requirements. Gasoline engine particulate emissions are within the US light duty and heavy duty requirements. Diesel engines, on the other hand, combust lean airfuel mixtures of >20/1 that produce exhaust temperatures of 100-450°C. This exhaust contains significant amounts of particulates that may require an aftertreatment device such as a catalytic converter in order to meet current and pending requirments. However, the concentrations of HC, CO, and NOx in diesel exhaust are typically below the current US light duty and heavy duty requirements. Gasoline converters using straight flow-through monoliths are very effective at reacting the gaseous constituents in the exhaust such as HC, CO and NOx to water vapor, C02, and nitrogen. These gases must diffuse and adsorb onto the catalyst surface before a reaction can take place. The driving force of diffusion, is the boundary layer concentration gradient of the gaseous species between the bulk gas phase and the catalyst surface. Once at the catalyst surface, the gases are chemically adsorbed. Chemical adsorption, or chemisorption, is a process that describes the adsorption of reactive gases onto an active catalyst surface. Chemisorptions generally occur at high temperatures and the rate of adsorption increases with temperature. This process is highly exothermic and the gases adsorb onto the catalyst surface in a monolayer. In order to properly describe how a diesel converter works it is necessary to first classify the constituents in the exhaust. See Table 2. The particulates are defined, as the species that collect on the filter during performance testing. The ash consists of inorganic compounds containing elements like zinc and phosphorous that originate from fuel and engine oil. The sulfate particulate and the associated waters of hydration originate from the combustion of sulfur containing fuels and the oxidation of SO2 by the catalyst. The organic particulate is a collection of relatively high boiling point compounds originating from the fuel, engine oil and their combustion products.
340
Table 2 Diesel Exhaust pecies
Other particulates, include catalyst and iron oxide from the exhaust system. As for the gaseous species, the Particulates Gases amounts of HC, CO, and NOx in the exhaust are not of primary concern, since Ash, Carbon HC Sulfate & H20 co they are typically below regulated requirements. SO2 however, can be NOx Organic Particulates Other so2 oxidized to sulfate particulate by the catalyst. SO2 is formed in the combustion chamber by the combustion of fuc s containing sulfur. Roughly, 99% of sulfur in the fuel is converted to SO2 in the engine. Since the exhaust temperatures of diesel exhaust are below the oxidation temperature of carbon, a diesel converter achieves particulate reduction by maximizing the oxidation of the organic particulate. In general, the organic particulate can exist in three states in the exhaust system (1); (i)- as a gas in the exhaust system, which will condense out in the dilution tunnel onto the particulate filter, (ii)- as a mist or aerosol, and (iii)- absorbed onto the carbon particulate. The ratio of aerosol versus gaseous organic particulate increases with decreasing temperature. In states 2 and 3 the organic particulates behave more like a particulate than a gas, having mass and momentum relative to the gas stream. A large portion of the organic particulates do not chemisorb onto the catalyst surface like HC, CO, and NOx do in gasoline converters, because of the relatively low exhaust temperatures and the high boiling points of the organic particulates. Some of the organic particulate that remains as a gas in the exhaust stream prior to the converter physically adsorbs. The physical adsorption of a species onto a surface occurs when it is near it's condensation temperature. The rate of physical adsorption increases with decreasing temperature and the species can be adsorbed in multilayer coverage. This statement implies that the effectiveness of a converter may decrease as exhaust temperatures rise. Studies performed by Horiuchi et al. (7) also suggests that the organic particulate is being physically adsorbed at lower temperatures and oxidized before they desorb. It is important to operate a diesel converter at temperatures where the physical adsorption of the organic particulate can occur and where the catalyst is kinetically active. In addition, a diesel converter can be designed to improve the collision of organic particulates onto the catalyst surface. A converter with a tortuous flow path (the exhaust through the converter changes direction) would aid in the collision of the organic particulates that have mass and momentum with the catalyst surface. Straight flow-through monoliths then should not perform as well as some metal monolith and pelleted converter designs that have tortuous flow paths.
34 1
Not only must a diesel converter be designed to maximize the oxidation of the organic particulate, it also must not create an sulfate particulate. Sulfate particulate can be derived from three sources. (i) - Sulfur trioxide (SO3) is formed in the combustion chamber and combines with water vapor in the exhaust to form sulfuric acid. The sulfuric acid and waters of hydration condense on the particulate filter during performance testing. (ii) - SO2 is oxidized by the catalytic converter to SO3 and combines with water vapor to form sulfuric acid. (iii) - Stored sulfur can be released from the catalyst support during high temperature transients to form sulfate particulate. Typically, SO3 can react with the alumina catalyst supports to form aluminum sulfate which will later decompose at elevated exhaust temperatures to alumina and SO3. Aluminum sulfate decomposes at 770°C,however experience with alumina supports has shown releases of sulfate particulate at lower temperatures (1). This suggests that the sulfur released from the alumina may be stored in a less fixed manner. Steps can be taken to minimize these sources of sulfate particulates. The use of lower sulfur fuels will directly lower engine out sulfate particulate values. A proper choice of noble metal catalyst can minimize the oxidation of SO2 to SO3. Catalyst support materials can be used that minimize the storage and release of sulfate particulate during exhaust temperature transients. In summary, for a diesel converter to maximize particulate efficiencies, it must operate within a temperature window that allows the physical adsorption and reaction the organic particulate without chemically adsorbing and reacting S02. In addition, the converter must be designed to maximize the contact of the organic particulate with the catalyst surface. However, it must be designed to pass the carbon, ash and other particulates through the converter. The storage of these particulates, can increase converter back pressure, cause a release of a black cloud during transient high flow conditions and if stored carbon were to ignite, it could cause irreparable damage to the catalyst. EXPERIMENTAL
Sulfir Storage Experiment It is important that a diesel converter is an emission stable device. More specifically, the converter must not release significant amounts of carbon or sulfate particulate during transient flow conditions. Release of black smoke or carbon can be minimized by manufacturing engines to produce less black smoke or by designing converters that do not store or release significant amounts of carbon. Transient sulfate emissions can be minimized by using lower sulfur containing fuels, or by operating the catalyst at temperatures that
342
do not oxidize SO2 to SO3 and/or by using a catalyst support that does not store or release sulfur during temperature transients. Experiments were performed to determine the sulfur storage characteristics of pelleted alumina and silica catalyst supports. Pelleted converters containing 2.6 liters of catalyst were used. The alumina supported catalyst had a BET surface area of 125 m2/g and a pellet diameter of 4-6mm. The silica supported catalyst had a BET of 275 m2/g and a pellet diameter of 4-6mm. Each converter contained 0.043 troy oz. of palladium. The converter were aged together on a 6.2L DDA V-8 engine, one converter on each exhaust manifold. The fuel contained 0.28 wt% sulfur. The aging schedule consisted of five, 50 hour aging intervals at temperatures of 200, 300, 400, 300, and 200°C, respectively. After each interval, a small portion of the catalyst was analyzed for sulfur using x-ray fluorescence. The results are shown in
Alumina
Silica
Fresh
200
300
400
300
200
Time at Temperature, 50 Hours @ each Temp. (Celsius)
Figure 1:Sulfur Storage on Silica and Alumina Catalyst Supports (I)
343
Figure 1 shows that the silica supported catalyst has an initial storage of sulfur to about 0.3% after the first aging interval. Then after the following four aging intervals, the amount of stored sulfur in the silica appears to be insensitive to changes in exhaust temperature. This implies that this catalyst support will not release any significant amounts of sulfur from the support during high temperature transients. The alumina catalyst support stored roughly 0.45% sulfur after the first aging interval at 200°C and continued to store sulfur to 0.83% after the second aging interval at 300°C. This implies that the alumina supported catalyst did not store sulfur to its equilibrium value after the first aging interval. The alumina support appeared not to store any additional sulfur after the third aging interval at 400°C. This would imply that an alumina support is at it's equilibrium storage value at 400°C. However, if all the alumina were to form aluminum sulfate, the weight percent of sulfur would be 28 wt%. The alumina support stored sulfur to 0.96% after the fourth aging interval and showed no further increase in sulfur storage after the fifth aging interval at 200°C. This experiment did not adequately show alumina's ability to release sulfur during high temperature transients. In conclusion, Figure 1 shows that the silica support stores less sulfur that the alumina support and that the silica does not store or release sulfur during temperature transients. SULFUR DIOXIDE OXIDATION
It is important to minimize the oxidation of SO2 to sulfate particulate, since oxidizing one gram of SO2 will produce 3.7 grams of sulfuric acid particulate, including the waters of hydration. SO2 oxidation can be minimized, by operating the converter at low temperatures where either oxidation of SO2 is minimal, adding base metals, and by chemically and thermally treating the catalyst. Noble metals also have different abilities to oxidize S02. Platinum is very effective at oxidizing SO2 in a lean environment. Adams et al. (12) showed that platinum oxidized three times the SO2 than palladium. Experiments were run in an effort to understand the ability of platinum and palladium to oxidize SO2 in diesel exhaust as a function of temperature. Three pelleted converters having 2.6 liters of catalyst volume were used. Silica catalyst support of 4-6mm pellet diameter were impregnated with platinum and palladium. Silica catalyst support was used based on previous experiments showing that it does not store or release any significant amounts of sulfur during temperature transients. The noble metal loadings of the three converters were 0.030 troy oz. platinum, 0.043 troy oz. palladium and 0.180 troy oz. palladium. Each converter was aged for 100 hours on a water brake dynamometer with a 6.2L DDA V-8 engine. The fuel contained 0.28 wt%
344
sulfur. See the appendix for a description of the aging schedule. After aging, the steady state performance of the converters were evaluated on a 6.2 liter DDA engine with an electric dynamometer at exhaust temperatures of, 500, 400, 300, 200, 150, 200, 300, 400, and 500"C,respectively. During each evaluation point the engine ran at 2000 rpm. The exhaust temperatures were achieved by varying the engine load with the dynamometer. Mini-dilution tunnels were used to collect the particulates before and after the converter.
FILTER APPARATUS
DRY TEST METER SAMPLE PUMP
~
Figure 2. Diesel Converter Performance Test Facility (1)
Figure 2 is a schematic of the converter performance test facility. Pallflex 47mm diameter filters were used. The particulate filters were analyzed for sulfate using a wet chemistry technique. It is assumed the SO2 oxidized by the converter is directly proportional to the sulfate particulate found on the filters. The performance evaluations at the same temperatures were averaged. Figure 3 shows that the platinum catalyst oxidized 3-5 times more SO2 than the palladium catalysts at temperatures above 200°C. The two palladium catalysts have similar SO2 oxidation profiles. However, the higher loaded palladium catalyst with 0.18 troy oz. of metal appears to oxidize less SO2 than the lower loaded catalyst at 0.042 troy oz. when evaluated at 400 and 500°C. This difference may be experimental error. If this is so, then it would appear that the oxidation of SO2 is not sensitive to palladium loading.
345 11 10 9
a Sulfate out Sulfate in
7
6 5 4
3 2 1
0
150
200
300 400 Temperature (Celsius)
500
Figure 3. Sulfate Particulate vs. Temperature (1) .28% wt Fuel Sulfur, Eval. on 6.2L DDA
Information from Figure 3 can be helpful in choosing a noble metal catalyst for an application. For light duty applications when exhaust temperatures do not exceed 250-300°C on the light duty US FTP cycle, palladium or platinum may be used. However, if off cycle temperatures exceed 30O-35O0C, platinum containing catalysts can produce significant amounts of sulfate particulate. For heavy duty applications where exhaust temperatures exceed 350-400°C on the heavy duty transient cycle, palladium should be used. It is estimated that if an engine manufacturer is to meet the US 1994 heavy duty particulate emission requirements, a catalytic converter can only a oxidize a.maximum 1% of SO2 in the exhaust to sulfate particulate. RESULTS
Evaluation of Fresh and Aged Alumina and Silica Supported Catalysts
Two metal monolith converters having 1.8 liters of catalyst volume were coated with alumina and silica washcoat technologies. The BET surface area of the alumina and silica washcoats were 125 and 200 m2/g, respectively. The metal monolith substrate was of a herringbone design at 35 cells/cm2. Each converter was impregnated with 0.05 troy oz. of palladium. Palladium was used to minimize SO2 oxidation at high exhaust temperatures so the effects of
346
sulfur storage and release could be seen. The converters were aged for 300 hours on the durability schedule described in the appendix. The particulate and sulfate efficiencies were determined at 400, 300, and 200°C before and after aging on a 6.2 liter DDA engine described in Figure 2. Sulfate efficiencies were measured at 400°C. Table 3 presents the results.
Table 3
-
Performance of Silica and Alumina Supported Catalysts
Converter support
Aging (hours
Alumina
0
400 300 200
-10 24 40
6
300
400 300 200 400 300 200
-120 -10 5 -33 56 81
16
400 300 200
-40 42 67
Silica
0
300
Particulate Evaluation Temperature ("C) (% Efficiency)
Sulfate Out Sulfate In
Both converters reduced particulate emissions during the fresh evaluation at 300 and 200°C. However, the silica supported catalyst had twice the particulate efficiencies at these temperatures. At the 400°C evaluation, both converters produced particulates. This is seen by the negative particulate efficiencies of -10 and -33 for the alumina and silica catalyst supports, respectively. The generated particulate is due to the formation of sulfate particulate from the catalyst oxidizing S02. After the 300 hours of aging the alumina supported catalyst performed poorer. The particulate efficiencies fell to -120, -10 and 5% for temperatures of 400, 300, and 200"C, respectively. This aged converter is generating significant amounts of sulfate particulate at 400°C and probably at 300°C also. The converter increased engine out sulfate emissions by 16 times at 400°C. The 400°C fresh evaluation of the alumina supported catalyst only increased sulfate emissions by six times. The difference in sulfate emissions of the fresh and aged alumina converter appears to be from the release of sulfur
347
(sulfate) from the catalyst support. The silica supported catalyst, on the other hand, did not exhibit a drastic change in emission performance after aging. After aging, the silica catalyst particulate efficiencies dropped about 15% at the three evaluation temperatures and the converter created about the same amount of sulfate particulate at 400°C. It is evident from Table 3 that the silica supported catalyst would be an emission stable device. The actual performance of alumina supported catalyst would probably be difficult to quantify as the converter is exposed to diesel exhaust for significant periods of time. The data shows that the performance of the alumina containing converter can change drastically with aging. If alumina is used, it should be stabilized in such a manner that it does not release significant amounts of sulfate particulate during high temperature operation. DIESEL CONVERTER PARAMETER STUDY
An experiment was performed to determine the effects of catalyst volume, substrate, and noble metal loading on fresh converter performance. The experiment compared the performance of converters having 1.8 and 1.4 liters of catalyst volume using either a 62 cell/cm2 ceramic monolith or a 35 cell/cm2 metal monolith substrate. The metal monolith substrate used a stacked herringbone design in which the flow of exhaust through the substrate follows a tortuous path. The substrates were coated with the same silica washcoat technology and impregnated with various combinations of noble metals. The converters were evaluated on a 1.9 liter turbocharged Lancia Dedra using the US light duty FI'P cycle. The average and peak temperatures during the FTP were 150 and 300°C, respectively. European specified fuel was used containing 0.20 wt.% sulfur. 30% of the particulate during the US FTP cycle was organic. Table 4 describes the converters that were built and tested. Prior to evaluation, each converter was aged for 24 hours on the dynamometer schedule described in the appendix. Each converter was evaluated three times on the vehicle. The averages of the FTP evaluations are shown in Table 5 in % FTP efficiency. The individual FTP results were regressed to determine the significance of the converter design parameters. Table 6 summarizes the analysis of the data. Keep in mind that the results of this study pertain to the silica washcoat technology investigated. The effects of different washcoat technologies, base metals additives, and other noble metals, such as rhodium, were not investigated.
348
Table 4
-
Diesel Converter Descriptions
I Substrate
No
Noble Metals (Troy oz per converter) I Pd Pt 0.00 0.15 0.00 0.09 0.15 0.09 0.15 0.03 0.15 0.03 0.15 0.09 0.15 0.00 0.09 0.00
Volume (liters) 1.4 1.4 1.8 1.8 1.4 1.4 1.8 1.8
Ceramic Ceramic Ceramic Ceramic Metallic Metallic Metallic Metallic
Table 5 - US Light Duty FTP Performance 1.9L Turbocharged Lancia Dedra ( % FTP Efficiency ) Converter No 1 2 3 4
co
HC 33 58 9 3 25 -5 21 31
5 6 7 8
Emissions Best Volume Best (liters) Substrate HC
NOx 4 2
2 87 1 3 12 2 7 64
co
1.4 1.4"
Ceramic
NOx Part.
1.4
Metallic*
5 0 9 1 8 -1
Noble Metal Pt Pd PtPd
+* -* +* -* -
Particulate 22 1 1 -8 14 8 23 19
Legend for Table 6.
* Parameter is significant to 90% level.
-* + Parameter is good for reducing emissions. +* - Parameter is poor for
+
reducing emissions.
All other listed parameters have F-Cal values of greater than 1.0. Unlisted parameters have F-Cal values less than 1.0. (F-Cai is defined as; the variance attributed to the parameter divided by the variance of the model)
349
From the results presented in Tables 5 and 6 the following statements are made. 1.- The smaller converters having 1.4 liters of volume appear to work best for HC, CO, and particulate emissions. This may be due to the higher concentration of noble metals in the smaller converters. 2.- The ceramic substrate may work best for CO emissions. However, the metal monolith worked best at reducing particulate emissions. This result appears to support statements make in the theory of operation, that straight flow through ceramic monoliths are effective at reducing gaseous emissions and that a substrate like the metal monolith with a tortuous flow path may be more effective at reducing particulates. In addition to the metal monolith having a tortuous flow path, it also has lower thermal inertia than the ceramic substrate. This unique thermal property may also improve particulate performance. Since the organic particulate physically adsorbs onto the catalyst surface at low temperatures, and the rate of physical adsorption increases with decreasing temperature, a substrate that gets cooler faster will physically adsorb more of the organic particulate. Also, if a portion of the organic particulate adsorbs at low temperature and is oxidized at higher exhaust temperatures, a substrate that heats up faster will oxidize a larger portion of the organic particulate before it desorbs. 3.- Platinum works best for reducing the HC and CO emissions, but is poor for NOx and particulate emissions. The particulate result is supported by Figure 3 which shows the ability of platinum to oxidize S02. It appears that platinum is producing considerable amounts of sulfate particulate. 4.-Palladium is poor for reducing HC and CO emissions. However, it appears to work well at reducing particulate and NOx emissions. The contrasting effects of palladium and platinum to reduce particulates is supported by Figure 3. Palladium is not producing any significant amounts of sulfate particulate during the evaluation. 5.- Do not mix platinum and palladium in a single converter. These converters performed poorly. SUMMARY
A diesel converter can be designed to specifically reduce engine out emissions. To maximize the reduction of particulates, the data and theory presented suggests that; 1 - A tortuous exhaust flow path would improve the contact of a portion of the organic particulate that has mass and momentum with the catalyst surface. 2 - A catalyst substrate with low thermal inertia may improve the adsorption and reaction of the organic particulate.
350
3 - Use a noble metal that does not oxidize S02. The choice of noble metal is application dependent. - For hot applications where exhaust temperatures exceed 300"C, palladium should be used. - For cooler applications, platinum or palladium may be used. However, when exhaust temperatures exceed 350°C significant sulfate emissions may be produced with platinum. 4 - Use a catalyst substrate material that does not store or release sulfur during temperature transients. To maximize the reduction of the gaseous species, HC and CO; 1 - Use platinum rather than palladium. 2 - Ceramic monoliths may work better than metal monoliths.
Acknowledgments The authors wish to thank the following groups and individuals; - Fiat Auto Spa DPSI for providing the vehicle for The Converter Parameter Study. - J. C. Schmitz and N. Meyrer who performed the FTP testing at The European Technical Center, in Luxembourg. - R. Beckmeyer, M. Tsang, W. Symons, S . Shaffer, S. Majkowski, J. Marklan, C. Metcalf, S. Barnhart, T. Kudza, and J. Finlayson of AC Rochester's Research and Test Lab Organization. APPENDIX
Diesel Dynamometer Aging Schedule Description Engine: 6.2L V-8 Cylinder w/split Exhaust Pipes Fuel: 0.28 wt.% Sulfur Capabilities: Ages two converters per engine. Step
Converter Inlet Temperature
1
2 3 4
5 6 7 8 9 10
1500 2500 1500 1000 1500 Repeat Steps 1-6 2500 1000 Repeat Steps 2-9
3.3 0.6 1.8 2.4 4.3
3.3 0.3
240 365
35 1 REFERENCES 1 - D. Ball, R. Stack, "Catalysts for Diesel Powered Vehicles", SAE Paper No. 9021 10, To be published October 1990. 2 - A. Gill, "Design Choices for 1990s Low Emission Diesel Engines", SAE Paper No. 880350, February 1988. 3 - T. Truex, W. Pierson, D. McKee, L. Casi and R. Baker, "Effects of Barium Fuel Additive and Fuel Sulfur Level on Diesel Particulate Emissions", Environmental Science and Technology, Volume 14, No. 9, September 1980. 4 - M. Arai, S. Miyashita and K. Sato, "Development and Selection of Diesel Particulate Trap Regeneration System", SAE Paper No. 870012, 1987. 5 - G. Lepperhoff and G. Kron, "Impact of Particulate Traps on the Hydrocarbon Fraction of Diesel Particulates", SAE Paper No. 859913, 1985. 6 - Y. Kiyota, K. Tsuji, S. Kume and 0. Nakayama, "Development of Diesel Particulate Trap Oxidizer System", SAE Paper No. 860294, 1986. 7 - M. Horiuchi, K. Saito and S. Ichihara, "The Effects of Flow-Through Type Oxidation Catalysts on the Particulate Reduction of 1990s Diesel Engines", SAE Paper No. 900600, February 1990. 8 - F. Indra, "Diesel Particulate Filters and Their Regeneration", SAE Paper No. 885151, 1988. 9 - M. Barris, "Durability Studies of the Trap Oxidizer Systems", SAE Paper No. 900108, February 1990. 10 - J. Wall, S. Shimpi and M. Yu, "Fuel Sulfur Reductions for Control of Diesel Particulate Emissions", SAE Paper No. 872139, November 1987. 11 - G. Andrews, I. Iheozor-Ejiofor and S . Pang, "Diesel Particulate SOF Emissions Reduction Using an Exhaust Catalyst", SAE Paper no. 870251, 1987. 12 - K. Adams and H. Gandhi, "Palladium-Tungsten Catalysts For Automotive Exhaust Treatment", Ind. Eng. Chem., Prod. Res. Dev. 198-3, 22, 207-212.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
353
CATALYTIC AUTOMOTIVEPOLLUTION C O N T R O L WITHOUT NOBLE METALS
Sander Stegenga, Nico Dekker, Jowi Bijsterbosch, Freek Kapteijn, Jacob Moulijn, Gerard Belot* and RenC Roche* Department of Chemical Engineering, University of Amsterdam Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands * P.S.A. Etudes et Recherches, 18 Rue des Fauvelles 922.50 La Garenne-Colornbes, France ABSTRACT The catalytic activity for the NO reduction and the CO oxidation, and the influence of several parameters on this activity, has been studied for Cu and Cu-Cr catalysts on pelleted and monolithic supports. It has been shown that 10 wt% Cu-Cr/A1203 catalysts are very effective for NO reduction, CO and HC oxidation. The highest activity is obtained at a Cu/Cr ratio of 2. Monolithic-type catalysts exhibit a three-way catalyst behaviour, comparable to noble metal catalysts under the same conditions. The conversion performance under oscillating compositions is even better than under stationary composition conditions. The presence of 0 2 leads to an inhibition of the NO reduction by CO. NO does not affect the CO oxidation. In the NO reduction with CO at low NO conversions predominantly N 2 0 is formed, whereas at higher conversion levels the N 2 0 formation passes through a maximum and at higher temperatures (T > 523 K) only N2 is observed. The catalyst system deactivates at higher temperatures (> 775 K) in stoichiometric CO/NO/02 mixtures, but the original activity returns in an overall oxidizing environment at 623 K. The CO oxidation activity level of a freshly prepared catalyst under reducing conditions is higher than under oxidizing conditions, after thermal treatment in a stoichiomemc mixture this is the reversed case. A tentative explanation is given. Cu catalysts prepared from EDTA complexes and a La-stabilized alumina-support appear to have a high thermal stability and a final activity comparable to that of the Cu-Cr catalysts.
INTRODUCTION
The activity of the catalysts used for automotive pollution control is based on that of the noble metals Pt, Rh and Pd [l].These noble metal catalysts are very active and fairly resistant to the sulphur present in the exhaust gas. There are, however, also some obvious disadvantages in the use of noble metal catalysts; they are relatively scarce and therefore expensive, they require a strict aidfuel ratio control, and because of the low amounts of metal present they are relatively sensitive to impurities. Mainly because of these disadvantages, our research was focused on the development of non-noblemetal alternatives. In the early years of exhaust gas catalysis research, base metals were recognized as promising active components for catalysts,
354
especially, combinations of Cu and Cr were found to be very active [2,3,4]. Forced by the pace of the evolution of the legislation, the attention was focussed on the noble metals, leaving the base metal catalysts underdeveloped. The last decade shows a renewed interest in the possibilities of base metal catalysts [5,6,7]. Previous research at our laboratory on carbon-supported catalysts showed that catalysts based on the combination Cu and Cr have an activity comparable to that of noble metal catalysts, both for the oxidation and the reduction reactions required [ 5 ] . The performance of alumina-supported catalysts has been tested in two, for the catalytic purification of exhaust gases, important reactions: NO reduction: CO oxidation:
2 N 0 + 2CO 2CO + 0 2
=3
2C02+N2 2c02
(1)
(2)
These reactions have been carried out both separately and simultaneously. A range of catalysts with varying Cu/Cr-ratio has also been tested in the reaction system containing NO, CO, and 0 2 . Special attention has been given to the thermal stability of differently pretreated catalysts. On the basis of the results with alumina-supported catalysts, also a monolith- supported catalyst has been prepared. These monoliths contained a Zr-stabilized alumina washcoat and here loaded with active phase resulting in a 10 wt% loading of the washcoat. These monoliths have also been tested on "bench-scale'' at the research laboratory of P.S.A. France for simultaneous NO reduction, CO and HC oxidation, both under stationary and oscillating feed conditions.
EXPERIMENTAL Catalysts As catalyst support a standard y-Al203 (Ketjen 000-1.5E, CK 300) was used (dp = 105 to 145 pm; Sa = 195 m2/g, BET-Nz ). The catalysts were prepared by either wet or pore-volume impregnation of the alumina support with a solution of the metal nitrates to result in a 10 wt% Cu-Cr/Alz03 catalyst. Catalysts with a fractional Cu loading of Xcu = 0, 0.33, 0.5, and 0.66 have been prepared. After overnight drying at about 353 K, the sample was heated in a 100 ml/min air flow. The heating rate was 2-5 K/min. First the temperature was increased to 353 K, and subsequently to 673 and 773 K, each of these temperatures was maintained for about 30 min. For one batch the maximum calcination temperature was increased to 1073 K. As monolitic support, batches drilled from a Zr stabilized Degussa monolith (OM 724 MLKV), containing a 10 wt% A1203 washcoat, were used (cut to dimensions of 1 cm diameter and 3 cm length). The preparation method was the same as for the particle catalysts. The monolith was impregnated to yield a 10 wt% Cu-
355
Cr loading of the A1203 washcoat. Also a La-stabilized alumina support, containing 2.5 wt% La, was used. This support was impregnated with CuEDTA and calcined at 725 K 181.
Apparatus For the activity measurements the experimental set-up [5] consists of a gas mix section enabling us to use a wide range of reactant mixtures, followed by a reactor section and an analysis section. Except for the U-shaped quartz reactor, the complete set-up is made of stainless steel tubing. The reactor is placed in a water-cooled, temperature controlled furnace. The combination of a continuous NO/NOx-analyser and a dual-column gas chromatograph enables us to quantitatively analyse the following compounds NO, NOx, N20, N2, C02, CO, and 02. Experimental procedures The catalytic activity was measured in temperature programmed experiments. Standard conditions for these measurements are given in table 1. For screening the monolith catalyst, a 1 g sample was used to make the absolute amount of Cu-Cr comparable with the experiments using particulate catalysts. For the separate and combined performance in NO reduction and CO oxidation, a set of four gas compositions was used (table 2). TABLE 1.- Standard conditions for the activity measurements Temperature range Pressure Catalyst: particles monolith Gas flow rate VHSV: particles monolith
293 - 1073 0.15 100 1.o
98 72 3.9
K MPa mg g pmolls 10 /h 1O h
The catalysts were screened by heating the sample at 2 K/min from 293 to 675-775 K and subsequently cooling at 2 K/min to 293 K. The thermal stability of the catalysts under these reaction conditions (gas comp. 1) was investigated by increasing the maximum screening temperature with 100 K in four consecutive experiments (Tmax = 773, 873, 973, and 1073 K), followed by an experiment up to 773 K to measure the final activity after these treatments.
356
1 2 3 4
1 % CO 1 %CO
1%co
0.2 % co
0.4 % 0 2 0.4%02 0.8%02
0.2% NO 0.2% NO
NO reduction and CO oxidation CO oxidation (reducing) CO oxidation (oxidizing) NO reduction
Apparatus and Experimental Procedures (at P.S.A France) The apparatus used to investigate the catalyst behaviour in closed-loop emission control systems is developed according to the guidelines given by Schlatter et a1 [9.] The relatively inert compounds, CO and N2 , are mixed and heated up to the reaction temperature. The reactants NO, CO, 0 2 , and HC are added to the heated gas. This reactant gas is led over the catalyst, either a packed bed or a monolith, at a flow rate corresponding to a VHSV of 75.000 h-1 . Both reactant and product gas can be analyzed, CO by I.R., HC by F.I.D., and NO by chemiluminescence. Rich and lean gas compositions are produced by either adding CO or 0 2 to a standard composition. For this purpose two other inlets are present, one for 0 2 and one for CO, constructed to open and close at a frequency of 1 Hz, to induce an oscillation in the gas mixture composition. In table 3 a characteristic rich and lean gas composition are given. h is used as a measure for the stoichiometry of the gas composition, and is defined as the ratio between the actual air/fuel ratio and the stoichiometric aidfuel ratio.
Lean ( h =1.02) 0.4*104 0.8* 1 04
02
Rich ( h =0.98) 1.4*104 0.4*104
HC (C3H6 &z C3H8) NO H20
500
500
1000 1.0*105
1000 1.0*105 1.4*105
Gas
co
co2
i4*in5
RESULTS In figure 1 the NO reduction activity (gas comp. 4) is shown for a standard catalyst, the concentration curves are plotted as function of temperature. It can be seen from this figure that all NO and CO is converted below 625 K. In the temperature range 420 - 520 K some N 2 0 is produced. Below 450 K the NO is predominantly converted to N20, above 520 K N2 is
357
the only reaction product. In figure 2 a comparison is made for the activities of NO reduction and CO oxidation, both separately and simultaneously. This figure shows that there is no influence of the addition of NO on the CO oxidation and also that there is only a small influence of the addition of 0 2 on the NO reduction. The NO concentration curve is clearly shifted to higher temperatures, while no N 2 0 formation was observed and the N2 formation is unchanged.
---I
f
P
P \
C
0 .c
I
+ I
C Q)
0 C
0
0
300
400
600
500
700
800
T/K
Figure 1.
NO reduction with CO. Course of the concentrations of reactants and products as function o j the temperature for 10 wt% Cu-Cr on Al2O3. Pco = PNO = 0.3 kPa.
Combined NO, CO, and 0 2 experiments have also been conducted with catalysts with varying Xcu,viz. 0.66(A), 0.5(B), 0.33(C) and O(D). Figure 3 shows the 0 2 and N2 concentrations as a function of the temperature, as a measure of the activity for CO oxidation and NO reduction, respectively. Clearly, N2 is only produced when all Xcu has been consumed. All three Cu containing catalysts show a similar reaction behaviour, a relatively steep 0 2 consumption curve and a rather smooth N2 production curve. A catalyst with Xcu = 0.66 seems to be the most active, full conversion of CO is achieved at 450 K and of N O below 600 K. The chromium catalyst (Xcu= 0) shows a very low CO oxidation activity and a very steep N2 production curve, full CO conversion is achieved at 610 K and full NO conversion at 650 K.
358
E
10
P P
8
0
0 F
6
\
C
.-40-
2
4
w
C Q1
0
2
c
0
0
0
300
400
500
600
700
800
T/K
Figure 2.
E
a
P 0
Comparison between NO+C0+02 , C 0 + 0 2 and NO+CO. Course of the concentrations as function of the temperature for 10 wt% Cu-Cr on A1203
-1
4 -
0,
0 7
\
c
0 .4-
F
c
c Q)
0 C
0
0
300
400
500
600
700
800
T/K
Figure 3.
NO reduction and CO oxidation f o r catalysts with varying X c u in Cu-Cr on A1203 , all 10 wt%. N2 (--- : NO reduction) and 0 2 (: CO oxidation) as function of the temperature . Xcu: 0.66(A), 0.5(B), 0.33(C) and O(D)
359
In figure 4 the results of thermal stability experiments are shown for a catalyst (Xc, = 0.5) freshly calcined at 1073 K. The 0 2 consumption curves became less steep with higher pretreatment temperature, while the N2 production curves are parallel shifted to higher temperatures. A pretreatment temperature of 873 K had a small influence on the CO oxidation and NO reduction activity of the catalyst. Pretreatment at 973 K resulted in a larger deactivation, while a consecutive pretreatment at 1073 K has only a small additional effect on the at 1073 K calcined catalyst.
300
400
500
600
700
800
T/K
Figure 4. Influence of pretreatment temperature on the NO reduction and CO oxidation. N2 production and 0 2 consumption as function of the temperature for 10 wt% Cu-Cr on A1203 , calcined at 1073 K. Similar experiments have been conducted with the Cu catalyst on Lastabilized A1203 (figure 5). This figure shows that, in contrast to the Cu-Cr catalyst, there is a slight delay after full 0 2 consumption before the N2 production starts. The activity of the Cu catalyst after thermal treatment at 773 K is lower than of any of the catalysts based on Cu-Cr. However, the thermal stability of this catalyst is much better, even after pretreatment at 973 K the activity of the catalyst has hardly changed. Only after pretreatment at 1073 K, the activity shows a remarkable decrease. In figure 6 the influence of the addition of NO on the CO oxidation is measured for a Cu-Cr/monolith catalyst (Xc, = 0.5; 1 wt% Cu-Cr). The concentrations of the gases has been plotted as a function of the reaction temperature. This figure clearly shows that there is no influence of NO on the CO oxidation, for both gas compositions the curves completely coincide. In agreement with the result described before NO reacts with the excess amount of CO as soon as all 0 2 has been consumed.
360
kP (1
0 r \
C
0 .w
z
w
C
a, 0 C
0
0
300
400
600
500
700
800
T/K
Figure 5.
Influence of pretreatment temperature on the NO reduction and CO oxidation. N2 production and 0 2 consumption as function of the temperature for 10 wt% Cu on La-stabilized A1203 .
ka
l2
I
0
0 r \
s
.-
.I-
z
.I-
C
a, 0 C
0
0 300
400
600
500
700
800
T/K
Figure 6.
The influence of NO on the CO oxidation : NO+C0+02 and C0+02. Course of the gas phase concentrations as function of the temperature over I wt% Cu-Cr on a monolith.
Figure 7 shows the influence of 0 2 addition on the NO reduction over the same monolith catalyst. The N2 production is plotted as measure for the NO reduction. The NO reduction is shifted to 100 K higher temperatures in the presence of 0 2 , and only starts when all 0 2 has reacted.
36 1
The thermal stability of the monolith catalyst has also been investigated. Figure 8 shows that pretreatment at 873 K resulted in a large shift of both the 0 2 consumption and N2 production curves to higher temperatures. Higher pretreatment temperatures, 973 and 1073 K, resulted in only a small additional shift. E
5
Q P
C
C
0
0
0-
300
400
600
500
700
800
T/K
The influence of 0 2 on the NO reduction : NO+CO+O2 and NO+CO. Course of the gas phase concentrations as functionof the temperature over I wt% Cu-Cr on a monolith.
Figure 7.
EP 0
0 l-
\
C
.-0 4-
z
c C
a, 0 C
0
0
300
400
500
600
700
aoo
T/K
Figure 8.
Influence of pretreatment temperature on the NO reduction and CO oxidation. N2 production and 0 2 consumption as function of the temperature for I wt% Cu-Cr on a monolith.
362
Directly after these thermal stability tests, CO oxidation experiments have been conducted with the thermally treated catalysts calcined at 1073 K and the supported monolith calcined at 773 K, in an overall oxidizing environment. The results are plotted as CO concentration as function of the temperature (figure 9), for the catalyst calcined at 1073 K and for the Cu/Crmonolith . For both catalysts the same pattern is observed. At the increasing temperature branch the activity is the same as shown by the most deactivated curve in figure 4 and 8. At 100% CO conversion the overall environment is still oxidizing, which apparently leads to a reactivation of the catalyst. The final activity is even slightly higher as compared to the initial activity before the thermal stability experiments. Cu-Cr based catalysts are very sensitive to their pretreatment or reaction conditions [ 101. Application of the catalyst in reducing environment increased the activity of both a catalyst freshly calcined at 773 K and a catalyst freshly calcined at 1073 K, the conversion level reached does not change during measurements in reducing environment. In oxidizing environment the activity is decreased to a lower, although stable, level compared to the level in reducing environment. Figure 10 and 11 systematically compare the CO oxidation activity in overall oxidizing or reducing environment before and after treatment in reactants at 1073 K. The activity of the catalyst freshly calcined at 1073 K is lower, both in oxidizing and reducing environment, than that of the catalyst calcined at 773 K. Treatment of these catalysts in a stoichiometric reaction mixture at 1073 K has a large influence on the catalyst performance.
300
400
500
600
700
800
T/K
Figure 9.
CO oxidation activity in overall oxidizing environment (gas comp. 3 ) after reaction in NO+C0+02 at 1073 K . CO2 production as firnction of temperature, for I0 wt% Cu-Cr on A1203 and on a monolith .
363
12
E
n n " 0 l-
\ N
0 0
3 00
500
700
T/K
Figure 10. CO oxidation activity in overall oxidizing or reducing environment before and after treatment in reactants at 1073 K. 10 wt% Cu-Cr on Al2O3, calcined at 773 K. Before treatment (---) ;reducing ( 0 ) oxidizing ; After treatment (-) ;reducing (0); oxidizing (D 12 10
8
6 4
2 0
3 00
500
700
T/K
Figure 11. CO oxidation activity in overall oxidizing or reducing environment before and aper treatment in reactants at 1073 K. 10 wt% Cu-Cr on A1203 , calcined at 1073 K. Before treatment (----); reducing (0); oxidizing 0) After treatment (--); reducing (0);oxidizing U,
364
Again stable activity levels of the catalysts are obtained in oxidizing and reducing environments, but now the use in a reducing environment results in the lower activity as compared to that in an oxidizing environment. After the thermal treatment the activity of the catalyst freshly calcined at 1073 K is higher than that of the catalyst calcined at 773 K. In figures 12 and 13 the results are shown of a standard three-way activity test of a Cu-Cr/monolith catalyst. The conversion of NO, CO and HC is plotted as function of lambda, for stationary conditions (figure 12) and for oscillating conditions (figure 13). These figures show that under stationary conditions in an oxidizing environment ( h > 1) the NO conversion level is low, while the HC and CO conversions are very high. In a reducing environment ( h < 1) the NO conversion level is relatively high, while the HC and CO conversions decrease with decreasing h. In the situation of oscillating gas composition the conversion curves are broadened. The HC, NO and CO conversion levels are higher at h < 1 and h > 1, only at around 1 the NO and CO conversions are at an intermediate level. Especially the overall conversion of NO and HC has increased. 1 .o
0.8 0.6
0.4 0.2
0 0.94
0.96
0.98
1.00
1.02
1.04
1.06
Lambda
Figure 12. Conversion of NO (A), CO ( 0 )and CHx ( 0 )as function of lambda under stationary Conditions jbr 1 wt% Cu-Cr on a monolith.
DISCUSSION Although slightly less active than carbon-supported Cu-Cr catalysts [11,12] , alumina- supported Cu-Cr catalysts are also shown to be very active for both CO oxidation and NO reduction. Full conversion of a stoichiometric mixture of 0 2 , CO and NO is reached below 625 K. In all situations, the CO oxidation occurs at lower temperatures than the NO reduction.
365
The production of N 2 0 is often observed in catalytic NO reduction [13-151. For alumina-supported Cu-Cr catalysts it is shown that, in a stoichiometric NO-CO mixture, a small amount of N 2 0 is formed between 420 and 520 K (figure 1). This is in agreement with previous results on carbonsupported catalysts [ l l ] , and also with results of Bauerle et a1.[15], who report the production of N 2 0 during the catalytic reduction of NO by CO in the presence of 0 2 , for both base- and noble- metal catalysts at temperatures below 650 K.
1 .o 0.8 0.6 0.4
0.2
0
0.94
0.96
0.98
1.00
1.02
1.04
1.06
Lambda
Figure 13. Conversion of NO (A), CO (0)and CHx ( 0 )asfunction of lambda under oscillating conditionsfor 1 wt% Cu-Cr on a monolith. The mechanistic aspects of the formation and reaction of N 2 0 over alumina-supported Cu-Cr catalysts is subject of a further study. The addition of NO to a mixture of CO and 0 2 has no influence on the CO oxidation activity of both the particle and monolith catalyst. The absence of an effect of NO is in contrast with many noble-metal catalysts, where the presence of low concentrations NO can severely decrease the CO oxidation activity, which is a result of strong chemisorption of NO on Pt above 473 K [16;19]. On the other hand, the addition of 0 2 to a NO, CO mixture results in a strong inhibition of the NO reduction, certainly for the monolith catalyst. For all NO reduction experiments in the presence of 0 2 , the conversion of NO is inhibited until all 0 2 is removed by reaction with CO. With the monolith catalyst this results in the NO reduction to occur at temperatures 100 K higher as compared to the situation without 0 2 .
366
For both the NO reduction and the CO oxidation over copper-based catalysts a redox mechanism is assumed [20,21]. In this redox cycle the partially reduced sites are vacant for oxidation by either NO or 0 2 . Apparently 0 2 reoxidizes these sites much faster than NO, which explains the inhibition of the NO reduction by 0 2 and the absence of any influence of NO on the CO oxidation with 0 2 . This conclusion is supported by results obtained from a kinetic modelling study of the CO oxidation, which shows that during CO oxidation, all the active sites are completely oxidized as long is oxygen is present in the gas phase [22]. Similar to the carbon-supported catalysts, a composition with Xcu = 0.66 results in the most active catalyst. This fraction does not correspond to any Cu-Cr compound as reported by Laine et a1.[6], but will more likely correlate to a compound in which the Cu-Cr compound might be present to stabilize the CuO. Previous results [ 111, which are in agreement with results reported by Shelef et a1.[4] and Gassan-Zade et a1.[23], show that supported Cr catalysts have a good NO reduction activity and a low CO oxidation activity, which would suggest a preference for the NO reduction during simultaneously performance of NO reduction and CO oxidation. As already found for carbonsupported catalysts, the preference for the NO reduction is not observed. The NO reduction is completely inhibited by the presence of 0 2 , and only when all 0 2 has reacted, NO reduction will occur, shown by a very steep N2 production curve. By comparing the results of the La-stabilized Cu catalyst [8] in figure 5 with figure 3 and 8 it can be seen that the initial activity of the Cu catalyst is lower than that of any of the Cu-Cr catalysts, either supported on A1203 or on the monolith, both for the CO oxidation and for the NO reduction. Compared to the most active Cu-Cr combination (figure 3) the CO oxidation is completed at a 50 K higher temperature, while the NO reduction sets on at a 90 K higher temperature. This lower initial activity seems to be compensated by the high thermal stability. After pretreatments above 973 K, in a stoichiometric NO, CO and 0 2 reaction mixture, the Cu-based catalyst shows a better CO oxidation performance and a comparable NO reduction performance as compared to the alumina-supported Cu-Cr catalyst. However, after treatment at 1073 K the Cu based catalyst exhibits a lower activity than Cu-Cr/A1203 . The Cu-Cr based catalysts show a decrease in activity as result of an increased pretreatment temperature (figure 4 and 8). Initially the activity of the catalyst calcined at 1073 K is lower than the activity of a catalyst calcined at 773 K. However, pretreatment of these catalysts in a stoichiometric reaction mixture up to 1073 K results in a comparable CO oxidation activity and a higher NO reduction activity of the catalyst calcined at 1073 K. These thermally treated catalysts can be reactivated by use in oxidizing environment at 623 K, the resulting activity is comparable to the initial activity of the
367
catalysts. The stable activities before and after thermal treatment at 1073 K in reactants show for the catalyst calcined at 773 K a decrease in the overall performance in oxidizing and reducing environment, whereas the catalyst calcined at 1073 K even shows a slight increase in overall performance. Both a catalyst freshly calcined at 773 K and a catalyst freshly calcined at 1073 K, exhibit a higher activity in reducing environment and a lower activity in oxidizing environment. After pretreatment of these catalysts at 1073 K in reactants this is reversed: a higher activity in oxidizing environment and a lower activity in reducing environment. For the understanding of these observations the following tentative explanations can be taken into account. In a fresh catalyst Cu(I1) is stabilized in or by the CuCr204 spinel structure [24]. In reducing environment at 100 9% 0 2 conversion Cu cations can migrate from the bulk to the surface, resulting in an increased surface Cu concentration [lo], and hence more active sites. Several authors report such a migration for spinel type structures, where Cu migrates to either bulk or surface in oxidizing or reducing conditions, respectively [10,25,26]. At least a part of this migration is considered being a fast, reversible process [24], for some Cu containing spinel type compounds even under surprisingly mild conditions [27]. The increased activity in reducing environment can also be a result of the reduction of Cu +2 to Cuo . The oxygen metal bond strength, which is the rate determining parameter in the CO oxidation is lower for Cu as compared to Cu +2 [ 5 ] . For thermal treatment in a stoichiometrical reactant mixture two phenomena are observed, the activity is decreased due to the high temperature, and a change in catalyst takes place, which causes activation to occur in oxidizing instead of reducing environment. For the catalyst calcined at 1073 K the activity does not decrease because it has already been subjected to a high temperature, but the change, causing the switch in activation-deactivation behaviour, is also observed. Calcination in air at 1073 K does not induce this change in behaviour, so this phenomenon has to be the result of the combined effect of high temperature and gas phase composition. The decrease in activity is probably due to the formation of Cu(I1) aluminate spinel of a part of the Cu present [28]. The mentioned change in the catalyst can be speculated to be a result of a thermally induced reduction of Cu(I1) to Cu(1) in the spinel structure, which would not occur during high temperature calcination in air, because of the stabilized Cu(I1) oxidation state. Similar activation behaviour in oxidizing environment has been observed for CuO/A1203 catalysts after treatment in N2 at 1073 K [8], but the explanation given by the author was not satisfactory. However, the present results can also not conclusively elucidate the mechanism for the reversed activation-deactivation behaviour in oxidizing and reducing environment. A extensive investigation in to this behaviour is presently executed.
368 The Cu-Cr/monolith catalyst shows a typical three-way behaviour. In overall oxidizing environment the conversion of CO and HC is high and of N O is very low. In overall reducing environment the conversion of NO is high and of CO and HC is fairly low. This behaviour fully corresponds to the three-way behaviour of noble-metal catalysts [9]. Only around stoichiometric conditions, all components show a high level of conversion. In order to provide the proper stoichiometrically balanced exhaust gas composition, a closed-loop electronic air-fuel ratio control is required. This continuous adjusting of h results in a perturbation of the exhaust gas composition. The introduction of this oscillation results in two changes in the conversion curves. First, at h< 1 the CO and HC conversions and at h > 1 the HC and N O conversions are improved and second, for NO and CO the maximum conversion at h= 1 is somewhat lowered. However, the conversion of HC is higher over the whole h range. As an overall result the NO and HC conversions are improved by an oscillating feed.
CONCLUSIONS It has been shown that 10 wt% Cu-Cr/A1203 catalysts are very effective for NO reduction, CO and HC oxidation. Monolithic type catalysts exhibit a three-way catalyst behaviour, comparable to noble metal catalysts under the same conditions. Bench scale tests under oscillating composition conditions show an even better conversion performance than under stationary conditions. A Cu/Cr ratio of 2 seems to result in the highest catalytic activity. The affinity of 0 2 for the catalyst is much higher than of NO, which leads to an inhibition by 0 2 of the NO reduction by CO. Consequently NO does not affect the CO oxidation. In the NO reduction with CO predominantly N 2 0 is formed at low N O conversions. At higher conversion levels the N 2 0 formation passes through a maximum and at higher temperatures only N2 is observed. The catalyst system seems to deactivate at higher temperatures (up to 1073 K) in stoichiometric CO/NO/O2 mixtures, but the original activity returns in an overall oxidizing environment. Before the thermal treatment the final CO oxidation activity level under reducing conditions is higher than under oxidizing conditions, after the thermal treatment this was the reversed case. Cu catalysts prepared from EDTA complexes and La-stabilized alumina supports appear to have a high thermal stability and final activities comparable to that of the Cu-Cr catalysts.
369
References 1. 2. 3. 4.
5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
J.T. Kummer, J. Phys. Chem., 90, 4747 (1986) J.T. Kummer, Adv. Chem. Series, 143, 178 (1975) T. Ohara, in "The Catalytic Chemistry of Nitrogen Oxides", R.L. Klimisch and J.G. Larson, eds., p. 191, Plenum Press, New York 1975 M. Shelef, K. Otto and H. Gandhi, J. Catal, 12, 361 (1968) S. Stegenga, A.J.C Mierop, F. Kapteijn and J.A. Moulijn, to be published J. Laine, A. Albornoz, J. Brito, 0. Carias, G. Castro, F. Severino and D. Valera, in "Studies in Surface Science and Catalysis", Vol. 30, A. Crucq and A. Frennet, eds., Elsevier, Amsterdam, 1987 p. 387 P. Gagneret, presentation at "Catalysts for the future", VDMA-Haus Frankfurt, 1989 1.I.M Tijburg, Ph-D Thesis, University of Utrecht, The Netherlands 1989 J.C. Schlatter, R.M. Sinkevitch and P.J. Mitchell, I. & E.C., Prod.Res.Dev., 22, 51 (1983) J. Laine, J. Brito, F. Severino, G. Castro, P. Tacconi, S. Yunes and J. Cruz, Catalysis Letters, 5 4 5 (1990) S. Stegenga, R. van Soest, F. Kapteijn and J.A. Moulijn, to be published S. Stegenga, R. van Soest, F. Kapteijn and J.A. Moulijn, Recl. Trav. Chim. Pays-Bas 109, 112 (1989) C.M. Fu, V.N. Korchak and W.K. Hall, J. Catal., 68, 166 (1981) J.O. Petunchi and W.K. Hall, J. Catal. 78, 327 (1982) G.L. Bauerle, G.R. Service and K. Nobe, I. & E.C., Prod.Res.Dev., 11(I), 54 (1972) J. Wei, Advances in Catalysis, Vol. 24, 110.57, Academic Press, New York (1975) S.E. Voltz, C.R. Morgan, D. Liedernian and S.M. Jacob, I. & E.C., Prod.Res. Dev., 12(4), 294 (1973) S.H. Oh and J.E. Carpenter, J. Catnl., 101, 114 (1986) E. Koberstein and G. Wannemacher, in "Studies in Surface Science and Catalysis", Vol. 30, A. Crucq and A. Frennet, eds., Elsevier, Amsterdam, 1987 p.155 J.W. London and A.T. Bell, J. Catal, 31, 96 (1973) R.T. Rewick and H. Wise, J. Catal, 40, 301 (1975) S. Stegenga, N.J.J. Dekker, F. Kapteijn and J.A. Moulijn, to be published G.Z. Gassan-Zade, M.Y. Woode and T.G. Alkhazov, React. Kin. Catal. Lett., 34( I), 225 (1987) A, Iimura, Y. Inoue and I. Yasumori, Bull. Chem. Soc. Jpn., 56, 2203 (1983) K.P. de Jong, J.W. Geus and J. Joziasse, J. Catal., 65, 437 (1980) J.O. Petunchi and W.K. Hall, J. Catal., 80, 403 (1983) R. Hierl, H. Knozinger and H-P. Urbach, J. Catal., 69, 475 (1981) M. Lojacono and M. Schiavello, in "Preparation of Catalyst", B. Delmon, P.A. Jacobs and G. Poncelet, eds., Elsevier, Amsterdam, 1976
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 1991 Elsevier Science Publishers B.V., Amsterdam
37 1
THREE-WAY ACTIVITY AND SULFUR TOLERANCE OF SINGLE PHASE PEROVSKITES D.Jovanovic(1) , V.Dondur(2) , A.Terlecki-Baricevic(1) and B.Grbic(1) (1)Institute
of Chemistry, Technology and Metallurgy, Njegoseva 12,11000 Belgrade, Yugoslavia
(2) Institute of Physical Chemistry, Faculty of Science, University of Belgrade, P.O.Box 550, I I000 Belgrade,Yugoslavia
ABSTRACT Single phase perovskites, as LaCo03, LaCr03, La Cog 5 Cro 503, Lao.7Sro gCrO3 and Lag 7Sro 3Cro 95Ruo 0503,were synthesized and three-way activity and sulfur tolerance were measured in a pulse-flame catalyst testing system in a wide range of redox-potential (R). Comparison of activities indicates that perovskite with ruthenium ion has the best redox behaviour in the redox potential interval 0.75
INTRODUCTION Perovskites comprise structures with densely packed cubic lattice of the general formula ABX3 where X being oxygen or halogen [ 1,2]. So far several hundred compounds with perovskite-oxide structure have been described; they may be simple or complex but they crystallize in the ABO3 form only if certain conditions relating to physico-chemical and crystallographical parameters of the ions A and B are satisfied. Interest for a perovskite-based catalysts, the properties of which are function of the spin and the valence state of the metal ion in the B site, has increased suddenly with the knowledge that certain ecological problems as, for example the purification of pollutants from the exhaust gases of internal combustion (IC) engines can be solved by their direct use [4-81. The interest in catalytic investigation of perovskite compounds should be considered on account of three main reasons: -in the B sites, which represent catalytically active centers of the structure, it is possible to incorporate a great number of catalytically active elements (Co, Cu, Ni, Cr, V, Pt, Pd, Rh, Ru, Fe, Mn etc.) which is of particular importance for their application. -a high thermal stability of this type of oxides, which are obtained by thermal treatment at temperatures above 700°C, and
372
-taking into account a high stability of perovskite oxide matrixes it may be assumed that the ions in the B sites will be also chemically stable in interactions with pollutants. Cobalt ion, well known as a low temperature highly active element in the redox reaction [9-111 , and ruthenium ion, which is not currently being used as stand-alone catalyst but is a potential additive in three-way catalysts [ 12-27], were selected for incorporation into perovskite matrix. The advantage of ruthenium as compared to other noble metals is its higher selectivity in NO, reduction to nitrogen in a wide range of values of the redox potential [28]. On the other hand, the main disadvantage of ruthenium, as a component of a supported noble metal catalyst, is its tendency to be oxidized to Ru+6 and Ru+* oxides, which are volatile and poisonous [29,30]- According to the published results [8,13,14,31-351 in most ruthenium perovskites investigated as threeway catalysts, besides ruthenium ion, there are also other catalytic active base metals (Co, Mn, Ni etc.) in the B site. The aim of this study was to follow a noble metal synergetic effects arising from the combination of the ruthenium properties with the sulfur resistance of the chromium perovskite matrix. The possibilities for such perovskite applications are conditionally dependent upon high catalystic threeway activity, thermal stability and resistance to sulphur oxides poisoning. EXPERIMENTAL METHODS
Preparation of the Catalysts The samples LaCo03, LaCr03, LaCoo.5Cro.503 and La0.7Sr0.3Cr03 were prepared by the coprecipitation of carbonates from the aqueous solutions of corresponding metal salts: La(N03)3 6H20, Co(N03)2 6H20, Cr(N03)3 9H20 and Sr(N03)2 (analytic grade). The precipitate was filtered, washed with distilled water, dried overnight at 140°C and calcined, with periodic regrinding and rehomogenizing, at appropriate temperatures: LaCo03, LaCrOg and La0.7Sr0.3Cr03 at 1100°C and LaCo0.5Cr0.503 at 1000°C. The catalyst La0.7Sr0.3Cr0.95Ru0.0~03 was prepared from the mixture of metal oxides La2O3, Cx-203, Ru02 and strontium carbonate (SrC03) (analytical grade). Oxides and carbonate, in corresponding quantities, were milled and homogenized in ethanol. The obtained precursor was calcined 72 hours in a few steps at a temperature of about 1000°C. Chemical and structural compositions were characterized by X-ray diffraction analysis and X-fluorescence analysis. X-Ray spectra were recorded on a Philips diffractometer PW 1710 using CuKa radiation filtered by Ni, while, X-fluorescence analysis were obtained on a X-Ray Si(Li) Detector System CAMBERA, Model 7333E. AUGER surface investigations were recorded on a RIBER OPS-105.
373
The Three-Way Activity Measurements The specially constructed and designed pulse-flame system, which has been described previously [27-291, was used for carbon monoxide (CO), and hydrocarbons (HC) oxidation and nitrogen oxides (NO,) reduction measurements and sulfur tolerance tests. The three-way activity of the catalysts were measured with a simulated exhaust mixture containing CO, H2, 02, NO, C1-C3 hydrocarbons, N2, C02 and H20, obtained by combustion of pure 2,2,4-trimethylpentane (isooctane). The inlet concentration of the pollutants after isooctane combustion, were kept constant during a single run with variations less than 10 percent. The H2, 0 2 , CO, total HC and NO, concentrations were varied to obtain the desired redox range (R) [40]. The redox potential R, a measure of the exhaust stoichiometry, of the reacting gas mixture is obtained by dividing the sum of the equivalent reducing components of the mixture (CO, H2, total HC) by the sum of the oxidizing components (02, NO,) .Thus:
where Cco,CH2,CHCtot.,Co2 and C N O are ~ concentration (vol.%) of c o , H2, total HC, 0 2 and NO,, respectively. Therefore, R > 1 represents an overall reducing gas mixture, R < 1 represents an overall oxidizing gas mixture and R = 1 a stoichiometric gas mixture. The redox potential (R) is directly related to the A/F ratio, which is a measure of the fuel mixture stoichiometry. The three-way activity measurements and catalysts aging were made at the space velocity of 32,500 h-1 and a temperature of 550°C. All three-way activity runs were performed at a redox potential range 3.0 > R> 0.4. The sulfur effect on the three-way catalysts activity was investigated in the redox potential range R < 0.8 by aging the catalysts with fuel containing 0.03 wt.% of sulfur, which is an average sulfur concentration corresponding to European gasolines. Sulfur wax added in the form of ethanethiol (C2H5SH, analytic grade) and the sulfur quantity in the isooctane was checked by a standard method for sulfur determining in fuels [41,42]. RESULTS AND DISCUSSIONS
The X-ray diffraction patterns of the synthesized catalysts showed that they were all of perovskite-type structure and no second components were detected. The amount of active transition metals in the B position of perovskite lattices, i.e. Co, Cr and Ru, in the catalysts, determined by X-fluorescence analysis, corresponds to the stoichiometric values of the proposed chemical compositions.
374
The pulse-flame catalysts efficiency testing runs were performed in a wide range of redox potential (R) by pure isooctane fuel combustion. All catalysts activity testing runs were done in the temperature range 100°C 700°C and 550°C was chosen as an optimal average three-way catalysts testing temperature . Effect of redox potential (R) on CO conversion over synthesized catalysts is presented in F i g . ] .
__-___-----
0
redox potential
(R)
Fig.] Percentage carbon monoxide conversion as a finction of redox potential (R) for synthesized perovskite-type catalysts.
All investigated catalysts for CO oxidation can be divided in two groups: very active catalysts (LaCoo.5Cr0.503 and La0.7Sr0.3Cr0.95Ru0.0~03) with an oxidizing efficiency over 80% for R < 1.5, and low activity catalysts (LaCrOg and La0.7Sro.3Cr03.) Unsaturated hydrocarbons (HC,,.) simultaneously testing runs (Fig.2) indicates that the LaCoo.5Cr0.503 sample presents similar catalytic properties as so called "low activity" catalysts. Meanwhile, samples like La0.7Sr0.3Cr0.95Ru0.0~03 and Lac003 have very high oxidation level over 80% for unsaturated HC at R c 1.5. In the same experimental conditions, influence of redox potential (R) on nitrogen oxides (NO,) reduction, is shown in Fig.3. Ruthenium perovskite catalyst and LaCoOg have satisfactory NO, reduction efficiency, as well as the LaCo0.5Cr0.503 sample. As it was expected, the ruthenium containing sample has a R shift towards lean mixture (R < 1).
375
From Figs.1-3, it is obvious that all perovskites with solitary chromium ion in B lattice position, have extremely low three-way activity for all pollutants. Only two catalysts, Lac003 and Lao.7Sro.3Cro.g5Ruo.0503,show high three-way conversion (over 80%) on the lean side of the redox potential scale, close to the stoichiometric value (R=l) and at R = 0.75, respectively.
redox potential
(R)
Fig.2 Percentage of unsaturated hydrocarbons conversion as a function of redox potential (R)for synthesized perovskite-type catalysts (Symbols as in Fig.1).
Fig.3. Percentage of nitrogen oxides conversion as a function of a redox potential (R) for synthesized perovskite-type catalysts (Symbols as in Fig.1).
376
A three-way activity "window" of a catalyst is defined arbitrarily as the width of the R scale over which a catalyst converts 80% or more of CO, NOx and HCun. Perovskite LaCo0.5Cr0.503 is less active than Lac003 The efficiency of LaCrOg and La0.7Sr0.3Cr03 in reducing CO, NO, and HCun is extremely low in the whole investigated range of R. Activity of the perovskite with a strontium ion incorporated into A site does not show any improvement in redox activity in relation to the base matrix LaCr03. Total HC conversion for all five perovskites is apparently low (less than 40%) as a results of high saturated hydrocarbons fraction (about 62 vol.%) in the total HC; these results are not represented in three-way activity curves (Figs. 1-3 ). The proposed "window" for all three pollutants are satisfactory only for LaCr03 and La0.7Sr0.3Cr0.95Ru0.0503 .
Sulfur poisoning resistant is one of the important catalyst characteristics, as well as a high three-way activity. In the combustion process, fuel sulfur is oxidized to sulfur dioxide (S02) [42]. The exhaust gases then proceed via the exhaust manifold to the oxidation catalyst. In the catalyst, excess CO and HC are oxidized to water vapor and C02, while some of the SO2 is oxidized to sulfur trioxide (SO3). Gaseous sulfur oxides can react with alumina carrier in catalysts, or with catalysts base metal components, to form sulfate or sulfite salts which lead to catalysts degradation [44-461. Because of that reason very carefully sulfur aging tests on the serial catalysts, with differerit transition metals in B site, were made. Catalysts were aged by combustion of isooctane containing ethanethiol. Change of the degree of CO, total HC and mixture obtained by C2HsSH combustion in isooctane is shown in Fig.4. All tests were performed in oxidizing atmosphere (R < 0.8) for six hours at 550°C and at velocity of 32,500h-1. Dynamics of SO, poisoning is different for all investigated samples. Lanthanum chromite (LaCrOg), La0.7Sr0.3Cr03 and Lao.7Sro.3Cro.95 RU0.0503 maintained almost initial activity during six hours of aging. Activity of Lac103 suddenly decreased during the first hour, whereas LaCo0.5Cr0.503 activity monotonously decreased within the first three hours and remained at that value during further aging. After three hours of aging, activity level of this sample was comparable to that of LaCrOg. Activity of the sample Lao.7 Sro.3 Cro.95 Ru0.05 0 3 , during the aging stage, remained practically unchanged. Small decrease of the total HC conversion was due to the slight decrease of ethane (C2H6) conversion. Taking into account very low initial activities of NOx reduction for all samples under oxidizing conditions (except for the ruthenium containing perovskite) it can be said that outlet NOx concentration practically did not change during SOx poisoning.
377
LaCo03 0
- co
0
- HC (tot.)
A
-
NOX
La0.7 sr0.3 Cro 3
aging time (h)
Fig.4 Catalysts eficiency for CO, total HC and NOx during aging with fuel containing sulfur ( R < 0.8).
378 CONCLUSION
In order to prevent possible degradation of perovskite structure due to metal sulfate formation, almost inactive (Figs 1-3), sulfur tolerant, matrix of LaCr03 perovskite was chosen for substitution of ruthenium ion. Substitution of Ru+4 for Cr3+ in sulfur resistant perovskite LaCr03 (Fig. 4 ) requires simultaneous partial substitution of La+3 for bivalent ion strontium (Sr). Preliminary investigation by AUGER method of SrO surface, aged under similar conditions as other perovskites, shows no trace of chemically bonded sulfur on the oxide surface [39]. According to the literature data [47] in this kind of substitution the charge would be compensated by transition of part of Cr+3 to Cr+4 and/or formation of anion vacancies. In order to examine only the effect of strontium substitution in LaCrOg lattice on the sulfur poisoning +3
+2
+3
+4
resistance the structure La0.7 Sr0.3 Cr0.7 Cr0.3 0 3 was synthesized. In this perovskite struture only a small fraction of chromium ion was substituted by iuthenium and the following investigated composition was obtained: +3
+2
+3
+4
+4
La0.7 Sr0.3 Cr0.7 Cr0.25 Ru0.05 0 3 In a redox potential range 1.40 - 0.75, at 550°C, and in the presence of sulfur oxides, the ruthenium perovskite reduced CO, NOx and unsaturated HC by over 80%. It can be concluded that the new synthesized ruthenium perovskite can be successfully used for the removal of carbon monoxide, unsaturated hydrocarbons and nitrogen oxides from exhaust gases of IC engines, within a wide range of temperature and oxido-reducing operating conditions. ACKNOWLEDGMENT
This work was supported by the US-Yugoslav Joint Board on Scientific and Technological Cooperation (Grants EPA 513 and 885).
REFERENCES 1. R.J.H.Voorhoeve, Perovskite-Related Oxides as Oxidation-Reduction Catalysts, Advanced Materials, Academic Press, Inc., New York,(1977) 2. F.Galasso, Structure and Properties of Inorganic Solids, Pergamon, New York, (1970). 3. O.Prakash, P.Ganguly, G.Rama Rao, C.N.R.Rao, D.S.Rajcria, V.G.Bhide, Mater.Res.Bull., 9,1173 (1974). 4. R.J.H.Voorhoeve, D.W.Johson Jr., J.P.Remeike, P.K.Gallaghar, Science, 195,827 (1977). 5 . R.J.H.Voorhoeve, J.P.Remeike, D.W.Johnson, Science, 180,62 (1973). 6. P.K.Gallagher, D.W.Johnson, F.Schrey, Mater.Res.Bul1.. 9, 134(1974).
379 7. R.J.H.Voorhoeve, J.P.Remeike, P.E.Freeland, B.T.Matthias, Science, 177,353 (1972). 8 R.J.H.Voorhoeve, J.P.Remeike, L.E.Trimble, Mater.Res. Bull., 9-1393( 1974). 9. S,Angelov, E.Zhecheva, LDirnitrova, A.Terlecki-Baricevic, D.Jovanovic, Proc. 2 nd Czeschoslovak Conference on Preparation and Properties of Heterogeneous Catalysts, Bechyne, CSSR, (1985). 10. S.Angelov, D.Mehandjiev, V.Zharkov, A.Terlecki-Baricevic, D.Jovanovic. Z.Jovanovic, Appl.Catal., 16,431 (1985). 11. A.Terlewcki-Baricevic, B.Grbic, D.Jovanovic, S.Angelov, D.Mehandjiev, C.Marinova, P.Kirilov-Stefanov, Appl.Catal.47, 145 (1989). 12. A.Lauder, US Patent 4,126,580 (1979). 13. A.Lauder, US Patent 3,897,367 (1975). 14. G.Mai, RSepmann, F.Kummer, US Patent 3,900,428 (1975). 15. A.Lauder, US Patent 4,182,694 (1980). 16. G.Mai, RSepmann, US Patent 3,905,918 (1975'. 17. T.E.Volin, US Patent 4,134,852 (1979). 18. E.L.McCann, US Patent 4,151,123 (1979). 19. A.Lauder, US Patent 4,110,254 (1978). 20. A.Lauder, US Patent 4,049,583 (1977). 21. S.Matsumoto, JP Patent 62,053,737 (1987). 22. M.Kawabata, S.Matsumoto, JP Patent 62,065,746 (1987). 23. K.Yamamura, K.Tachibana, S.Kondo, JP Patent 63,158,130 (1988). 24. K.Tabata, I.Matsumoto, JP Patent 61,149,245 (1986). 25. K.Tachibana, K.Yamamura, S.Sekido, JP Patent 61,082,843 (1986). 26. C.Koch, DE Patent 3,504,556 (1986). 27. T.Kanbara, K.Yamarnura, K.Tachibana, SSekido, JP Patent 61,097,032 (1986). 28. K.C.Taylor, The Catalytic Chemistry of Nitrogen Oxides, R.L.Klimisch and J.G.Larson Eds., Plenum, New York, (1975). 29. H.Remy, M.Kohn, Z.anorg.allgem.Chem.,m,365 (1924). 30. W.Bel1, M.Tagami, J.Phys.Chem., 67,2432 (1963) 31. R.A.Da1la Betta, H.S.Gandhi, J.T.Kummer, M.Shelef, US Patent 3,8 19,536 (1974). 32. H.S.Gandhi, J.T.Kummer, M.Shelef, US Patent 3,835,069 (1974). 33. T.P.Kobylinsky, B.W.Taylor, J-E.Young, SAE Paper 740250 (1975). 34. M.Shelef, H.S.Gandhi, Platinum Met. Rev., 18, 2 (1974). 35. H.S.Gandhi, H.K.Stepien, MShelef, SAE Paper 750177 (1975). 36. Yugoslav Patent Appl. P-1211/90, 2l.jun (1990) 37. D.Jovanovic, A.Terlecki-Baricevic, K.Petrovic, Gonva rnaziva, 23, 33 (1984). 38. A.Terlecki-Baricevic, D.Jovanovic, K.Petrovic, Motori motoma vozila, X-56/57, 310 (1984). 39. D.Jovanovic, PhD Thesis, PMF Belgrade, (1989). 40. HSGandhi, A.G.Piken, M.Shelef, R.G.Delosh, SAE Paper 760201 (1976). 41. ASTM - D 1266-80. 42. I S 0 - 2192. 43. J.M.KaweGki, EPA Report 600/9-78-028 (1978). 44. J.N.Braddock, EPA Report 600/2-77-237 (1977). 45. R.E.Gibbs, G.P.Wotzak, S.M.Byer, N.P.Kolak, EPA Report 600/9-79-047 (1979). 46. T.J.Truex, H.Windawi, P.C.Ellgen, SAE Paper 872162 (1987). 47. N.Iehisa, K.Fukaya, K.Matsuo, N.Horiuchi, N.Karube, J.Appl.Phys., 59, 317 (1986).
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 0 199 1 Elsevier Science Publishers B.V., Amsterdam
38 1
NO REDUCTION AND A1)SORPTION INTERMEDIATES O N Pt-Rh AI,I,OY CATALYSTS L. Heezen, V.N. Kilian, R.F. van Slooten, R.M. Wolf and B.E. Nieuwenhuys Leiden University, Gorlaeus hboratories, P. 0. Box 9502, 2300 RA Leiden, The Netherlands
Abstract The reduction of NO by H2 was studied over a pure Rh, a pure Pt and several Pt-Rh alloy catalysts supported on silica in a fixed-bed plug-flow reactor operating at atmospheric pressure. The results were compared with results obtained earlier for the reduction of NO with CO using the same catalysts. The results can be interpreted on the basis of earlier Pt-Rh single crystal studies. The catalytic behaviour of the Pt-Rh catalysts is determined by factors such as gas phase composition,bulk composition and reaction temperature. No synergism between Pt and Rh was observed and the differences in activity can be explained in terms of the surface composition and the behaviour of the constituent metals towards the reactants. INTRODUCTION
The automotive three-way catalyst is an excellent catalyst for purification of automotive exhaust gas. Its high activity and selectivity, combined with its outstanding thermal stability and resistance to poisoning make it a very effective catalyst. The active components include the precious metals platinum or palladium and rhodium. In 1989 the automotive catalysts already accounted for 40 % of the world total Pt consumption and for 80 % of the Rh consumption [l]. R is an effective catalyst for the oxidation of CO and hydrocarbons : 2c0+02 'HC' + x 0 2
+ -+
2c02 yH20+zC02
Nitrogen oxides are reduced over the three-way catalyst to nitrogen according to the following overall reactions : 2NO+2CO 2 NO + 2 H2
+
+
N2+2C02 N2+2H20
382
Unfortunately, N O and hydrogen can also react to the undesired N 2 0 and NH3. The overall reactions are:
The selectivity of Pt to promote the NO reduction to N2 rather than to NH3 is poor, especially at low temperatures. Therefore, rhodium is the essential ingredient in the three-way catalyst for the selective conversion of N O to N2. Recently, it has been found that the precious metals form alloy particles in the three-way catalyst [2]. Alloy particles often exhibit catalytic properties (e.g. selectivity, activity, stability and poison resistance) different from those of the constituent metals. Despite the widespread use of bimetallic Pt-Rh catalysts, only a few studies have been published concerning the performance of Pt-Rh alloy catalysts in the relevant reactions. This contrasts with the monometallic Pt and Rh catalysts, about which several investigations have been reported. In view of the enormous importance of bimetallic Pt-Rh catalysts we have investigated in our laboratory the properties of various Pt-Rh alloy single crystal surfaces and supported Pt-Rh alloy catalysts [3]. In this paper we describe comparative studies of NO reduction with hydrogen over several Pt-Rh alloy catalysts, a pure Pt catalyst and a pure Rh catalyst. These novel results will be compared with those previously obtained for the NO reduction with CO over the same catalysts [4].The adsorption of CO, NO and their interaction were also studied on the same catalysts using infrared spectroscopy [5].In addition, some preliminary results are shown concerning the effect of the metal particle size on the NO-H2 reaction in order to distinguish effects of alloy formation from particle size effects.
EXPERIMENTAL ADDaratUS : The experiments were performed in a single-pass fixed-bed flow reactor operating at atmospheric pressure. The gases used were research grade and contained a mixture of 5% of the reactive gases in helium. The gas flow in the system was 20 cm3/min. The gas stream passed the powdered catalyst which was supported by a porous glass filter in a Pyrex reactor. Temperatures were measured using a chromel-alumel thermocouple attached to the glass filter. Analysis of the gas stream containing the products was accomplished using a computer controlled quadrupole mass spectrometer, which was differentially pumped by a turbomolecular pump group. During the measurements the catalysts temperature was increased with a linear heating rate of 4.5 K.min-1. This heating rate was sufficiently slow to maintain steady-state reaction rate conditions at each temperature. Reaction
383
rates were calculated from measurements of the total flow rate and the composition of the gas stream leaving the reactor. For more detailed information see Ref. [4]. The IR study was performed using a Perkin-Elmer 580B doublebeam grating spectrometer (resolution 5 cm-1) equipped with a stainless-steel cell in which a self-supporting sample disk was placed. The cell had silicon windows and was part of a vacuum chamber (base pressure 1 x 10-4 mbar) with a gas introduction system. Because of the temperature dependence of the IR spectra, background spectra were collected at five selected temperatures : 300, 350, 425, 525 and 625 K. After dosing 2 mbar of the gases to the cell, spectra were recorded at those five temperatures. The proper background spectra were subtracted and the resulting spectra smoothed using a 25-point smoothing function. Materials : In order to avoid possible support effects, Si02 was used as the support for the noble metal particles. Pt and Rh catalysts were prepared by homogeneous precipitation of Aerosil 200 (200 m2/g) with aqueous solutions of either Rh(N03)2 or H2Pt(OH)6. The alloy catalysts were also prepared by homogeneous precipitation of Aerosil 200, but with both precursors in aqueous solution. The metal loading of all catalysts was 5 wt%. The resulting catalyst was calcined overnight at 383 K and, subsequently, reduced in flowing H2 at 673 K for 16 h. One extra reduction step was used for the IR experiments; after the reduction a self supporting disk was pressed and this was reduced in situ at 623 K for 16 h. All catalysts were examined using X-ray diffraction to determine the average particle size, and, in the case of the alloy catalysts, to ensure that alloy particles had been formed. Finally, metal surface areas were determined by CO adsorption assuming a 1:l stoichiometry between noble metal atoms on the surface and adsorbed CO molecules. These measurements showed that the difference in specific surface areas of the catalysts was less than 25 %. RESULTS
1
-
NO-CO reaction
In order to compare the present results for the NO+H2 reaction, the most relevant results which were obtained earlier for the adsorption of NO, CO, NO+CO and for the NO+CO reaction on the same catalysts, will be briefly presented. In table 1 the results of the adsorption of CO, NO and a COP40 mixture on a pto.5Rho5/Si02 alloy catalyst are summarized.
384
Table 1
-
Location of the main IR absorbance hands (in cm-1) for difTerent gases adsorbed on a Pt0.5Rh0.5 alloy catalyst at different temperatures
Adsorbate
co NO
co/No
I
I
625 300 350 2030 2070 2065 1870 1860 1855 1845 1830 1790 1795 1790 1645 1645 1665
2180 2180 2075 2070 1875 1860 1645 1645
-
Assignement Pt/Rh-CO (linear) (Pt-Rh)-CO (bridged) Rh-NO (linear) Rh -NO (bent)
Rh-NCO 2195 Pt-CO (linear) 2055 2045 2040 (Pt-Rh)-CO (bridged) 1850 1845 Rh-NO (bent) 1635 1645
The results can be summed up as follows. The absorption band at lo75 cm-1 found on the alloy catalyst after CO+NO adsorption is also observed when CO is solely adsorbed on the alloy catalyst. The band is stable up to at least 625 K just like the corresponding band on pure Pt. On Rh the linear band disappears already around 425 K in the presence of the CO-NO mixture. Furthermore, its shift with increasing temperature is similar to that found on Pt and different from that observed on Rh. Hence, it is likely that this band is largely due to CO bonded to a Pt atom. The band around 1875 cm-1 is associated with multiply coordinated CO. Its relative intensity, its location and its shift with increasing temperature and, hence, decreasing coverage, suggests that at least a part of this multiply bonded CO is bonded to both a Pt and a Rh atom. The spectra of NO on the Pt-Rh alloy catalyst are similar to those found on the pure Rh catalyst, suggesting that most of the NO is adsorbed on Rh sites. The most relevant conclusions are summarized as follows : -Both Pt and Rh sites are present at the surface of the Pt-Rh alloy catalyst, as is shown by the adsorption of CO and NO on both Pt and Rh. -The CO molecules have a tendency to adsorb on Pt atoms, while NO molecules are primarily adsorbed on Rh atoms. This is consistent with the relative heats of adsorption of CO and NO on pure Pt and Rh. -For multiply coordinated CO a mixed Pt-Rh adsorption site is proposed. The steady-state reaction rate was measured on five catalysts, viz. 100% Pt, 75% Pt-25% Rh, 50% Pt-50% Rh, 25%Pt-75% Rh and 100%Rh. The gas phase composition was varied from CO rich (NO/CO=1/4), stoichiometric (NO/CO=l/l) to CO lean (NO/CO=4/1). For illustration
385
figure 1 shows the temperature required to achieve a constant turnover frequency of 0.05 s-1 versus the catalyst composition. The turnover frequency is defined as the number of NO molecules reacting per second per metal (Pt and Rh) atom on the catalyst surface. The number of metal atoms on the surface was estimated on the basis of CO adsorption (see experimental) 680
1
660
s
640
2 620 2
c
m
600
a
$ 580
I-
560
0
25
75
50
100
-%Rhodium NO/CO=1/1 NO/CO=1/4 NO/CO=4/1
+
.--A--
0
Figure I . Temperature f o r a constant turnoverfrequency of 0.05 sec-1 for the NO + CO reaction over Pt - Rh alloys vs bulk compostion. The main reaction product found was N2 with only traces of N 2 0 . The activities of the alloy catalysts were between those of the pure components, synergetic effects were not observed. Rh and Rh-rich alloys are much better catalysts for the reduction of NO with CO than R and R-rich alloys, both under CO rich and CO lean conditions. The activities of Rh and the Rh-rich alloys vary only slightly with the flow composition, whereas for the Pt and Pt-rich alloys the influence of the flow composition is larger. On the pure Pt and Pt-rich alloys catalysts, a high conversion is obtained at lower temperatures under CO lean than under CO rich conditions. The activity of the m.25-Rh0.75 catalyst is almost equal to that of the pure Rh catalyst, independent of the flow composition. Under stoichiometric and CO lean conditions, the Pb.5-Rho5 catalyst has an activity almost equal to that of the pure Rh catalyst. Under reducing conditions its activity is between those of R and Rh. The Pt0.75-Rho25 catalyst shows an activity equal to that of pure Pt under net-reducing conditions. Under stoichiometric and net-oxidizing conditions its activity is intermediate to those of pure Pt and Rh.
386
2- NO-H2 reaction
To assess the effect of catalyst aging, three series of measurements on each catalyst were performed. First, the conversion and the product distribution were followed on the freshly reduced catalyst using the T-t program described in the experimental. The maximum temperature used was 673 K. These experiments were repeated twice. The results of the second series of experiments were similar to those of the first series within experimental accuracy.
300
', 0
I
25
-
I
,
I
50
75
100
% Rhodium
NO/HP=l/I NO/H2=1/3 +
Figure 2. Temperaturefor a constant turnover frequency of 0.005 sec -1 for the NO + H2 reaction over Pt - Rh alloys vs bulk composition.
In figure 2 results are shown for Pt, Rh and the three Pt-Rh alloy catalysts, using two NO/H2 ratios, viz. 1/1 and 1/3. The freshly reduced catalysts were slightly more active than the aged catalysts. This effect was completely reversible after reduction. XRD showed that the average particle size of the aged catalysts was equal to that of the fresh catalysts. The effect of aging is significantly larger for the alloy catalysts than for the pure Pt and Rh catalysts, especially for the NO/H2=1/1 ratio. The activity of the alloy catalysts was between those of pure Pt and pure Rh. However, after aging its activity resembles that of the pure Rh catalyst for the NOD32 ratio is 1/1. The results shown are those found for the aged catalysts. However, this is not relevant for the discussion, since plots for the fresh catalyst show qualitatively the same differences in the behaviour of the various catalysts. The figure illustrates the different behaviour of the five different catalysts for the two flow compositions.
387
275
325
375
425
475
525
575
625
--> Temperature (K)
Figure 3. Selectivity vs reaction temperature for the NOH2 = I / / reaction over Pt. 1
=! Y
Q
0
0 275
325
375
425
475
525
575
625
--> Temperature (K)
Figure 4. Selectivity vs reaction temperature for the NO/H2 I / ] reaction over Rh. The reaction products found are N2, N 2 0 and NH3. The temperature dependence of the product distribution is shown in figures 3 to 5 for pure Pt, Pto.sRho5 and pure Rh using a NO/H2 ratio equal to unity (stoichiometric composition for the formation of N2). For a NO/H2 ratio of 1/3 all the catalysts produce large amounts of NH3 at T below 525 K. The differences in selectivity are smaller than for the NOM2 ratio of 1/1 and are, therefore, not shown in this paper.
388
275
325
375
425
475
525
575
625
--> Temperature (K)
Figure 5. Selectivity vs reaction temperature for the NO/H2 reaction over Pto.5 Rho.5. The relevant observations can be summed up as follows : No synergetic effects, like enhancement of the reaction rate, for the Pt-Rh alloy catalysts are observed when compared with the pure metal catalysts. The activity of the alloy catalysts is always between those of the pure Pt and Rh catalysts. - Whereas the activities of Rh and the alloy catalyst for the NO+CO reaction vary only slightly with changing gas phase composition, this effect is much larger for the NO+H2 reaction. However, for the pure Pt catalyst the effect of the flow composition is smaller for the NO+H2 than for the NO+CO reaction. - When the relative activities are considered, Rh and Rh-rich alloys are the better catalysts for the NO+CO reaction. On the other hand Pt is the most active catalyst for the NO+H2 reaction, while all Rh containing catalysts are less active. - Rh has a better selectivity towards N2 formation, while Pt produces relatively large amounts of N 2 0 and NH3 at temperatures below 525 K. The selectivity of the alloy catalyst is different from both Pt and Rh. It resembles that of Rh but its N 2 0 production is larger in a broad temperature range. -
111 Effects of particle size
In order to distinguish effects resulting from alloy formation from particle-size effects, a series of WSiO2 catalysts was prepared with average particle-sizes ranging from 20 to 300 A. These were used as catalysts for the
389
NO+H2 reaction. No significant differences were found regarding the selectivities and activity, although catalysts with large particles tended to have a somewhat larger activity than those with small particles. These effects, however, were much smaller than the effects of composition of the catalysts (Pt,Rh and alloy catalysts). DISCUSSION
The reduction of NO to N2 with either CO or H2 is one of the important reactions taking place over the three-way catalyst. Both activity, viz. a high conversion at low temperatures, and selectivity, high to N2, are dependent on the composition of both catalyst and gas phase. For the reduction of NO with CO both platinum and rhodium have a good selectivity to N2, while rhodium has the better activity. This last observation is consistent with literature data [lo]. The mechanism of the NO + CO reduction proceeds via the following steps [3]:
NO NOa
co
N a + Na NOa+Na CO,+O, NOa+Oa
---* +
+
--+
+
+ --+
NO, Na+Oa COa N2 N2+Oa
(1) (2) (3) (4)
C@
(5) (6)
N20
(7)
It is often assumed that the dissociation of NO is the rate determining step. The lower activity of Pt compared with that of Rh can be understood on the basis of the activity of both metals for the dissociation of NO. The dissociation of NO is more difficult on platinum than on rhodium [ 3 ] . The IR experiments show that both CO and NO are adsorbed on the alloy surface as would be expected from the relative heats of adsorption. Indications are found of the existence of mixed Pt-Rhadsorption sites for CO. However, CO on mixed Pt-Rh sites has no apparent effect on the overall catalytic activity. Most likely the concentration of CO on such sites is too small to influence the catalytic properties of the alloy catalyst noticeably. When H2 is used to reduce NO, platinum has the better activity. When selectivity is considered it is rhodium that is the better catalyst, because NH3 is a main product on the platinum catalyst. Again this is in accordance with literature [7]. As is the case with the NO + CO reaction, all three alloys behave either like the pure metals or show an intermediate activity with no indications of a synergetic effect.
390
The most likely mechanisms for N2. N 2 0 and NH3 formation are dissociation of N O followed by reaction :
In these mechanisms products are formed via an atomic N species. This mechanism was proposed by Siera et a1.[8] who found that NH3 can be formed on Pt-Rh( 100) and Rh( 100) via nitrogen adatoms. The better activity of Rh for the NO+CO reaction, may be attributed to the better activity of Rh in the N O dissociation which is often assumed to be the rate determining step. However, this model fails to explain the better activity of Pt for the NO+H2 reaction, because N O dissociation proceeds more rapidly on Rh than on Pt. Therefore, a different rate determining step should be considered. One possibility can be found by looking at the temperature at which N O starts to desorb from Pt and Rh. Table 2 shows that this temperature is lower for N O desorption from Pt than from Rh. NO desorption is prerequisite to create free sites adjacent to NO molecules required for N O dissociation.
, Td
Pt (111) Rh( 111)
270
350
350
Ref.
El
~131
Pt(100) Rh( 100)
400
r141
Pt(l10)
270 350
~151
[61
[GI
Onset temperatures (in K) for the NO desorptiun from three Pt and Rh single crystal surfaces under the conditions of TDS ex eriments.
Since the light-off temperatures on the Pt catalyst for the NO+H2 reaction are not far above the temperatures at which N O dissociates under UHV conditions, it is not certain whether N O dissociation or N O desorption initiates the reaction on Pt. The observation that the reaction on Pt is not very sensitive to the partial hydrogen pressure may point to NO dissociation as the rate determining step on Pt. On Rh the reaction takes place at temperatures that are much higher than the N O dissociation temperature [9]. Apparently, the reaction rate is limited by the presence of NO on the surface.
39 1 The observation that the reaction rate on Rh is very sensitive to the partial pressure of hydrogen is consistent with this model. Introduction of CO instead of hydrogen changes the situation considerably. For Pt an increase of the reaction rate was observed when the NO/CO ratio was increased. On Rh the effect was much smaller. A study by Kobylinski et al. [lo] showed also that the reduction of NO over Pt is strongly inhibited by CO. Therefore, it is likely that CO inhibition is responsible for the high temperatures required for the NO+CO reaction, especially on Pt. The heats of adsorption of NO and CO are rather similar. Therefore, site competition may be expected. IR experiments [l 11 show that CO is able to displace preadsorbed NO on Pt at room temperature. CO inhibition is responsible for the high reaction temperature required for the NO reduction by CO over Pt catalysts. Rh seems to have some preference for NO. Hecker and Bell [12] found that a Rh catalyst is nearly saturated with adsorbed NO during the NO+CO reaction for NO conversions below 50%. Hence, during reaction NO is predominantly present on Rh and it inhibits the reaction. The selectivity for the reduction of NO with hydrogen towards N2 rather than NH3 or N 2 0 is the second factor determining the overall performance of the catalysts. Pure platinum catalysts produce much NH3, while pure rhodium catalysts produce mainly N2. This difference can be explained by looking at the relative concentrations of atomic hydrogen and atomic nitrogen on the surface of the catalyst [8] When the concentration of atomic hydrogen is relatively high, as it is on a platinum catalyst, the NH3 production will be high. On rhodium the NO dissociation proceeds more easily, the Rh-N bond strength is higher than the Pt-N bond strength, and thus the atomic nitrogen concentration will be higher, therefore producing more N2 [8]. The selectivity of the alloys is always between that of pure rhodium and pure platinum. A factor that could influence the behaviour of the alloy catalysts, is the effect of the particle size on both selectivity and activity. Our experiments with WSi02 catalysts showed only negligible particle size effects. However, unfortunately, no information is available about the influence of the particle size on the alloy catalysts. We cannot exclude that, for example, surface segregation of one of the components can occur preferentially on the edges. This could result in a different behaviour when the particle size changes, compared to a monometallic catalyst. There are, however, no indications that the contribution of this possible effect might have a larger contribution to the catalyst performance than that of alloying alone. Moreover, the difference in particle sizes was not more than 10%(around 170 A). Both the activity and selectivity of the Pt-Rh alloy catalysts are between those of the constituent metal catalysts. We have found previously
392 that the surface composition of the Pt-Rh alloys varies strongly with the experimental conditions such as the gas phase composition [3]. Clean Pt-Rh alloy surfaces are enriched with Pt. Adsorbates can easily induce segregation of Pt or Rh to the surface. Oxygen in the gas phase induces a Rh surface segregation because of the high Rh-0 bond strength relative to that of Pt-0. For the NO+CO and NO+H;! reactions the activity of the Pt0.5Rh0.5 alloy catalyst is like pure Rh under net-oxidizing conditions, whereas under netreducing conditions its activity is between those of Pt and Rh. This may suggest that the surface composition varies with the experimental conditions from almost pure Rh under net-oxidizing conditions to, perhaps, a bulk-like surface composition under net-reducing conditions. However, the selectivity of this catalyst for the NO+H2 reactions is different from both Pt and Rh. This indicates that this catalyst contains Pt atoms in the surface that influence the selectivity. The activity of the Pt075Rhfj25 alloy catalyst is almost equal to that of pure Pt under net-reducing conditions whereas its activity is between those of Pt and Rh under stoichiometric and net-oxidizing conditions. Again, this suggests that the surface composition changes with the feed composition. The properties of the Pt0.25Rh0.75 alloy catalysts are noticeable. Both for the NO+CO and NO+H2 reactions its activity is like that of pure Rh under net-oxidizing, stoichiometric and net-reducing conditions. Under netoxidizing conditions a behaviour like that of pure Rh may indeed be expected. However, under net-reducing conditions the presence of both pt an Rh atoms should be expected on the surface. The relatively high activity of this catalyst for the NO+CO reaction might be caused by a beneficial effect of the presence of both Rh and Pt : the low CO inhibition on Rh sites and a beneficial effect of Pt on, for example, the amount of N or NO on the surface. For the NO+H2, however, the presence of Pt atoms does not seem to diminish the inhibition effect of NO. In conclusion, it was shown that the catalytic properties of silica supported Pt-Rh alloy catalysts are strongly dependent on the gas phase composition, the bulk composition of the alloy particles and the reaction temperature. A complex of factors determine the activity of the different catalysts. Under oxidizing conditions at temperatures higher than 600 K, Rh segregation to the surface takes place [3]. This is reflected in the catalytic activity of the Pt-Rh alloys which can vary between that of pure Pt and pure Rh. No synergism, caused by alloying of Pt with Rh, was observed. The differences in activity, observed for the different catalysts, can be explained in terms of the specific properties of the pure metals towards the adsorbates.
393
Acknowledgment The authors acknowledge the Johnson Matthey Technology Centre (Reading, U.K.) for the loan of the Pt and Rh salts. This research was supported in part by NATO via grant = 86.352.
References 1. 2.
3. 4.
5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
Platinum 1989, Johnson Matthey, 24 (1989). B. R. Powell, Appl. Catal. 53, 53, (1989). S . Kim, M. J. d'Aniello, Appl. Catal., 56, 23, (1989). S . Kim, M.J. dAniello, Appl. Catal., 56, 45 , (1989). R.M. Wolf, J. Siera, F.C.M.J.M. van Delft and B.E. Nieuwenhuis Faraday Discuss. Chem SOC.,8 2 , 2 7 5 , (1989). A.G. v.d. Bosch-Driebergen, M.N.H. Kieboom, A.v. Dreumel, R.M. Wolf, F.C.M.J.M. v.Delft and B.E. Nieuwenhuis, Catal. Lett., 2, 235, (1989). R.F. v.Slooten and B.E. Nieuwehuis, J.Catal. ,229 ,429 , (1990). R.J. Gorte, L/D. Schmidt and J.L. Gland, Surf. Sci., 109 ,367 , (1981). T.B. Kobylinski and B.W. Taylor, J. Catal. , 33 , 150 , (1974). J. Siera, B.E. Nieuwenhuis, H. Hirano, T. Yamada and K.I. Tanaka, Cata. Lett. , 3 , 179 , (1 989), and to be published. R.M. Wolf, J.W. Baker and B.E. Nieuwenhuis, Surf. Sci., accepted for publication T.B. Kobylinski and B.W. Taylor, J. Catal. , 377 , (1974). G.M. Alilina, A.A. Davidov, I.S. Sazanova and V.V. Popovskii, React. Kinet. Catal. Lett. , 27 (2) ,279 , (1985). W.C. Hecker and A.T. Bell, J. Catal. , 289 , (1984). T.W. Root, L.D. Schmidt and G.B. Fisher.,Surf. Sci., 134,30, (1983). P. Ho and J.M. White, J. Chem. Phys. , fl , 103 , (1987)/ R.J. Baird, R.C. Ku and P. Wynblatt, Surf. Sci. 97,346 , (1980)
,a
,a
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis andAutomotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
395
A COMPARATIVE KINETIC STUDY OF THE C O - 0 2 REACTION OVER Pt-Rh (lll), (loo), (410) AND (210) SINGLE CRYSTAL SURFACES. J. Siera, R. van Silfhout, F. Rutten and B.E. Nieuwenhuys Gorlaeus labaratories, Leiden University, P.O. BOX 9502, The Netherlands ABSTRACT The CO-02 reaction has been studied as a function of temperature and partial pressure of CO and 0 2 over the (1 1l), (loo), (210) and (410) surfaces of a Pt025-Rho.75single crystal. The effects of alloying and surface structure are discussed. It is found that the reaction rates on the alloy surfaces show a positive order in oxygen at high temperature (%OK), whereas under these conditions the reported order in oxygen is negative for Rh(ll1). Furthermore, considerable differences are found in the steady-state rate of C02 formation for the four surfaces. On Pt-Rh(210)the CO oxidation starts around 500K, while over Pt-Rh(ll1) the rate has already reached its maximum value at this temperature. Comparison of the four surfaces shows that CO oxidation over terraces can proceed at lower temperature than over step sites. Hence, CO oxidation over stepped surfaces can proceed in two different stages. The first one occurs at relatively low temperature on terrace sites, and the second one on step sites at higher temperature. The possible origin of the controversy in the literature concerning surface structure sensitivity or insensitivity in the CO oxidation over metals is briefly discussed.
INTRODUCTION
Supported Pt-Rh catalysts are used as active components in the automotive three-way catalyst for controlling pollution from combustion products such as nitric oxide (NO), carbon monoxide (CO) and hydrocarbonsl. The reaction mechanism and kinetics of the CO-02 reaction over noble metals have been extensive y studied in the past [2-111. The formation of C 0 2 is usualTABLE 1 ly thought to proceed through a Langmuir-Hinshelwood mechanism between adsorbed CO and adsorbed oxygen atoms, as confirmed by molecular beam experiments [ 101.The reaction steps are summarised in table I . In this study we present results or the CO-02 reaction over Pt-Rh(ll1 Pt-Rh( loo), Pt-Rh(410) and Pt-Rh(210) under low pressure conditions. It will 9
396
be shown that there exists large differences in the behaviour of the four surfaces studied. At low temperature the inhibition of the steady state rate of CO2formation by CO is very high on the (210) surface. This effect is less pronounced for the two flat surfaces (1 11) and (loo), while the (410) surface shows an intermediate behaviour. The C 0 2 production found for the four surfaces differs also at high temperatures where CO inhibition does not take place. Under these conditions the (111) surface shows only low activity, probably caused by a low sticking probability of oxygen. EXPERIMENTAL
All experiments were carried out in a standard UHV system with a base pressure better than 2x10-10 mbar. It was equipped for quadrupole mass spectrometry (QMS), Auger electron spectroscopy (AES) and low energy electron diffraction (LEED). The single crystal was oriented by means of the Laue back reflection method to within 0.50 of the desired direction. The samples were cut by spark erosion and mechanically polished down to 0.25 pm grain size using standard techniques. The samples were cleaned by cycles of Ar-ion bombardment, oxidising treatments and flashing to high temperature (1400K) in vacuum. The crystals were heated by passing current through tantalum leads spotwelded to the edges of the crystals. The temperature of the crystals was measured by means of a Pt/Pto. 10-Rho.90 thermocouple spotwelded to the edge of the crystal. The system is pumped by an ion pump, a turbo-molecular pump and a titanium sublimation pump. During reactions the valve separating the ion pump from the main chamber was closed and the system was only pumped by the turbo-molecular pump. In this way, reaction rates were measured by using the main chamber of the UHV system as a continuous flow reactor. Gases were dosed to the main chamber through variable valves. It was assured that the background pressure contributed to less than 1% of the total reactant pressure. RESULTS
A
-
Surface composition
In the case of an alloy sample knowledge of the surface composition is required to understand the obtained kinetic results. The surface composition of Pt-Rh alloys has been studied extensively in the past few years [8,12-171. In general, a large surface Pt enrichment is found. In table 2 the platinum surface concentrations for the four surfaces are given as observed by AES after annealing the samples at 1300K in vacuum. The AES measurements on (210), (100) and (111) were done by directly measuring the signal intensities of the Rh222, Rh256, Rh302, Pt64 and Pt168 eV Auger transitions. In the evaluation of
397
the data the Gallon model was applied [18]. Calibration with pure Pt and Rh samples is needed to determine quantitatively the surface composition by means of this model. Furthermore, the assumption is made that the second layer has the same Pt concentration as the bulk. Tsong et a1 1141 found that the second layer is enriched in Rh atoms. However, the deviation from the bulk concentration is small and our assum tion will not lead to large errors. In the case of Pt-Rh(21O) it was Table 2. also assumed that the backscattering Composition of various factors and escape depths of the surfaces of a Pt0.25-Rh0.75 generated Auger electrons are the same as for the (100) surface. The AES alloy single crystal (at 9%) following annealing at 1300K. determination of the surface composition of Pt-Rh(410) has been done by Surface Pt concentration(%) oxygen titration, a method described by van Delft et a1 [ 121. Information conPt-Rh( 11 1) cerning the distribution of Pt and Rh 32f5 Pt-Rh( 100) atoms along the steps and terraces is, 40 f 5 P t - R h ( 2 10) unfortunately, not available.The open 55 k 5 Pt-Rh(4 10) (210) surface shows the largest Pt 4 0 k 10 enrichment, implying that on stepped I surfaces the Pt segregation may be irger at steps than on the terraces. Van Delft et a1 [17] that oxygen induces Rh surface segregation in the temperature range from 600 to 1000K. All experiments described in the rest of this paper were done after a standard annealing procedure in order to produce a well known surface concentration. However, it can not be excluded that the surface concentration changes during the course of the reaction, especially under oxidizing conditions at temperatures higher than 600K.
B - Temperature dependence of the reaction rate under steady state conditions. The formation of C02 was followed by monitoring the intensity of the amu=44 mass signal. It was assured that the reaction rate reached its steadystate value. In figures I to 4 the results are shown for the CO-02 reaction carried out over Pt-Rh(l1 l), Pt-Rh( loo), Pt-Rh(410) and Pt-Rh(210), respectively. Under stoichiometric reaction conditions the total reactant pressure was 3x 10-7 Torr. Partial pressures were also varied from oxidizing to reducing conditions in order to study the order of the reaction in CO or 0 2 at different temperatures. The first observation to notice is that there exists striking differences in the rate of COT formation as a function of increasing temperature for the four single crystal surfaces. For all the surfaces studied the reaction rate increases with increasing temperature until a maximum rate is
398
0
m
-0
0
.O b
0
m 0
0
x
m
Y
I-
-t -
t Y
I-
-
-0 0
0
.O PI
0 0 I n
0
0
Figures I to 4 ; Steady state rate of C02 formation versus temperature over : ffig.1) :Pt-Rh(ll1) ;(fig.2) :Pt-Rh(100) ;cfig.3) :Pt-Rh(4lO) ;cfig.4) :Pt-Rh(210)
399
observed at a temperature Tm. This value of Tm is one parameter which strongly depends on the surface structure. For Pt-Rh(ll1) the value of Tm is only 490K, the lowest Tm measured in this study. Pt-Rh(210) and Pt-Rh(410) have Tm 's of 615K and 650K, respectively. The Pt-Rh(100) shows a different behaviour, for this surface two local maxima are observed in the rate of C02 formation. The first maximum lies around 500K and the second one around 580K. A similar behaviour has also been observed for pure Rh(100) [7]. The positions of the local maxima in the case of pure Rh(100) are almost similar to those on Pt-Rh( 100). The relative contribution of the low temperature maximum is larger for Pt-Rh(100) than for Rh(100). In addition to the variation of the value of Tm for the four surfaces, also the increase in the rate of C 0 2 formation with increasing temperature is different in the low temperature range. These differences are reflected by the values of the activation energy of the process. At low temperature (<480K) the reaction rate increases with an apparent activation energy Ea of 65 k 5 kJ/mol for the Pt-Rh( 111) surface. The apparent activation energy was calculated from the rates measured in the temperature traject of 360-49OK. When a slightly smaller temperature traject is choosen, i.e. 360 to 465K a somewhat higher Ea is found, namely 85 f 5 kJ/mol. The latter value is close to the value of 84 k 8 kJ/mol reported for the CO-02 reaction over Rh(ll1) [7].
C
-
Dependence on partial pressures.
Pt-Rh(ll1) At distinct temperatures the partial pressure of one reactant is varied over a wide range while the other is kept constant. In this way, the effect of excess CO or oxygen is easily studied, i.e. information is obtained concerning the order of the reaction in the two reactants. In figures 5a and 5b the results for Pt-Rh(ll1) are shown. In these figures the logarithm of the C02 yield at a certain temperature subtracted by the logarithm of the C 0 2 yield at stoichiometric conditions at the same temperature is plotted versus the logarithm of the partial pressure of the reactant which is in excess present in the reactor. The reaction order in CO depends strongly on the reaction temperature. The reaction order in CO at 415K is -0.20, whereas the order switches to +0.17 at 528K. At 503K the order in CO pressure is only slightly positive, i.e. +0.04, The order in oxygen is positive for the three temperatures studied ranging from +0.85 to +1.00. It is interesting to compare these results with those reported for Rh(ll1) [7]. At low temperature (415K) the order in oxygen is positive and negative in CO, suggesting that CO saturates the Rh(ll1) surface below 500K. However, there is a striking difference between Pt-Rh(ll1) and Rh(ll1) at 528K. At this temperature the order in 0 2 is positive, as described above, however the order is negative in 0 2 in the case of Rh(ll1) [7]. Thus, it seems that at this temperature the oxygen coverage on Pt-
400
Rh( 111) must be considerably smaller than on Rh( 111). The negative order in 0 2 reported for Rh(ll1) was ascribed to oxidation of the Rh surface. Apparently, the oxidation of the alloy surface is lower under the influence of Pt atoms distributed in the surface.
Fig 5a Pt-Rh(ll1) 0.5
I
Fiaure 5
1t ,
-0.5-
41 5K I
-16
I
I
I
I
I
-1 5 In(P(C0)
Fig 5b Pt-Rh(l11)
I
I
I
I
,:J
-14 4
I
The logarithm of the C02 formation in excess of CO minus the logarithm of the C O 2 formation under stoichiometric conditons is plotted versus the logarithm of the CO (5a) pressure ( T o r r ) and 0 2 ( S b ) pressure (Torr) f o r PtRh(1 I I ) . The CO pressure is varied from 1x10-7 to I X I 0 - 6 Torr, the 0 2 pressure is varied from 2x10-7 to 1x10-6 Torr.
Pt-Rh(100) The experiments were carried out in the same way as described above for Pt-Rh(ll1). The results for Pt-Rh(100) are shown in f i g . 6. The CO inhibition on Pt-Rh(100) is similar to the CO inhibition on PtRh(ll1). At low temperature (415K)the order is -0.12 in CO, changing to
40 1
0.14 at 578K. At 503K the order in CO is +0.10. The order in oxygen is +0.33, +0.413 and +0.16 for 415K, 503K and 578K, respectively.Hence, both for the Pt-Rh( 111) and (100) surfaces no negative order in 0 2 is observed. The effect of oxygen is less pronounced as compared with Pt-Rh( 111). Even at the position of the second local maximum a slightly positive order in 0 2 is observed. Fig 6a Pt-R h(100) -
I
578K
.... ....
-2
.-... --w.
-16
-
-15 In(P(C0))
The logarithm of the C02 formation in excess of CO minus the logarithm of the C O 2 formation under stoichiometric conditons is plotted versus the logarithm of the CO (6a) pressure (Torr) and 0 2 (6b)pressure (Torr) for Pt-Rh(l00). The CO pressure is varied from 1 x 1 0 7 to 1x10-6 Torr, the 0 2 pressure is varied from 2x10-7 to 1x10-6 Torr.
-14
Fig 6b Pt-Rh(lO0)
528K
lrl(P(02))
+
-
The results for the Pt-Rh(210) surface are shown infig.7. Pt-Rh(210) CO inhibition at low temperature could not be measured because of the low steady-state rate of C02 formation. At 533K and 553K the order in CO was found to be positive, 0.11 and 0.15, respectively. The order in oxygen is positive, 0.43 and 0.45, respectively, just as was the case with Pt-Rh(100).
402
Fig.7a Pt-Rh(210) 0.5 1
I
FiPure 7
-16
-15 In(P(C0))
-14 -----C
Fig. 7 b Pt-Rh(210)
1.5
-
/
The logarithm of the C 0 2 formation in excess of CO minus the logarithm of the C 0 2 formation under stoichiometric conditons is plotted versus the logarithm of the CO (7a) pressure (Torr) and 0 2 (7b)pressure (Torr) for Pt-Rh(210). The CO pressure is varied from 1 x 1 0 - 7 to 1x10-6 Torr, the 0 2 pressure is varied from 2x10-7 to 1x10-6 Torr.
1 -
0.5
,-/
__.I-
__-.I.-
533K
_.I-
0
-15.5
Pt-Rh(410)
-15 -14.5 ln(P(02)) A
-14
The results for the (410) surface have been reported elsewhere17. In this study the surface concentrations of CO and 0 2 were also measured by AES. It was found that at low temperatures (T<580K) the surface is saturated with CO. Above this temperature the oxygen concentration increases and the CO concentration decreases. Around 600K the product of CO and 0 coverages is at maximum.
403 DISCUSSION
The CO-02 reaction has been studied under varying conditions over four Pt-Rh alloy single crystal surfaces, i.e. ( l l l ) , (loo), (210) and (410). The rate of C02 formation strongly depends on the surface studied. The surface composition of the surfaces was also determined to gain insight into the effect of alloying. All the Pt-Rh surfaces show a Pt-enrichment relative to the bulk concentration. It is known that the CO oxidation on metals like Pd and Pt proceeds at the boundary of CO islands and 0 islands [3]. One possible beneficial effect of alloy formation might be that one of the reactant molecules has a strong preference for adsorption on one of the alloy components while the other molecule is preferentially adsorbed on the other. For example, 0 is bound with a higher heat of adsorption on Rh than on Pt. Hence, it is likely that oxygen will be adsorbed on the Rh sites of a Pt-Rh alloy surface. This effect might result in an easier mixing of CO and 0 on the alloy surface and hence, in a faster reaction at lower temperature. No indication of such an effect was found in the present study, as will be discussed later on. Large differences were found in the steady-state rates of C02 formation for the four surfaces studied. These differences are found in the CO inhibition regime at low temperature, as well as in the regime at higher temperature where the orders in both CO and 02pressures are positive. The different behaviour of the four Pt-Rh alloy surfaces is well reflected in the values of Tm where the reaction rate has reached its maximum value: 490K for Pt-Rh( 1 1l), 615K for (210), 650K for (410) and 500K-580K for (100). At the temperature T m the concentrations of 0 and CO have reached their optimum concentrations for the reaction. A comparison between Pt-Rh(lOO), Pt-Rh(21O) and Pt-Rh(410) is very interesting because both (210) and (410) consist of (100) terraces, with similar structure of the step sites. The (210) surface has only 2 atoms wide terraces, whereas (410) consists of 4 atom wide terraces. In this way, by comparing the three surfaces not only information is obtained concerning the effect of the introduction of a surface defect, or step site, but also on the effect of the step concentration. At low temperature, around 500K, the reaction is faster on (100) than over (410) and it is slowest over (210). Hence, increasing the concentration of steps results in a reduced CO oxidation at low temperature. It has been reported that step sites can have an activation energy of desorption of CO enhanced by 29 kJ/mol relative to terrace sites [19]. McCabe and Schmidt [28] studied the CO desorption from Pt(l1 l), Pt( loo), Pt(1 lo), Pt(211) and Pt(210). It was found that activation energies for desorption of the most tightly bound states varies from 145kJ/mol on Pt(210) and (211) to 105kJ/mol on Pt(ll0) and Pt(ll1). In the CO inhibition regime it is therefore likely that CO
404
oxidation on terraces proceeds easier over terraces than on step sites. It is not very likely that the differences are caused by differences in noble metal atom concentration at the surface. The step atoms on (210) and (410) have the same coordination, it is therefore expected that the concentration of Pt and Rh along the steps is the same, and therefore the difference between the two surfaces must be due to the difference in concentration of steps. The Pt-Rh(l00) shows two local maxima in the C02 steady-state rate of formation. This was also observed for Rh(l00). The reason for the occurence of two maxima is not yet clear. It is planned to follow the concentration of the adsorbates CO and 0 during the reaction. Hopefully, these experiments will shed some light upon this matter. A comparison of the behaviour of Rh(100) with that of Pt-Rh(100) shows that the relative contributions of the first and second oxidation stage is different. The contribution of the high temperature maximum for Rh(100) is larger. The value of Tm in the case of Pt-Rh(ll1) is 490K, the lowest for the four surfaces studied. This can be understood in terms of a combination of a low adsorption energy of CO and a low sticking probability of oxygen. The low adsorption energy of CO will lead to desorption of CO at relative low temperature, thereby creating room for oxygen to become adsorbed. However, the oxygen concentration remains as a result of the low sticking probabilty of oxygen on the flat (1 11) surface. CO oxidation has mostly been studied on the densely packed (111) surfaces, and many similarities have been found for Pt( 111) and Rh( 111). A major difference between Pt( 111) and Rh(ll1) is that oxygen desorbs from Pt(ll1) in TDS below 900K [32,33], while it remains on Rh until 1300K [34]. Another major difference is the value of the sticking probability of oxygen, which is very low in the case of Pt(ll1) and Pt(l00) [32]. Based on these observations a positive or zero dependence in oxygen pressure is expected for Pt(l1 l ) , while for Rh(ll1) a negative one-half order kinetics in 0 2 is observed. The Pt-Rh( 111) surface shows a positive order in oxygen pressure at high temperatures (>500K). This observation may suggest that oxidation of CO under these conditions proceeds largely on Pt-sites. An alternative explanation may be that Rh sites are more difficult to oxidise in the presence of Pt atoms. Above the rate maximum the C02 production decreases giving rise a negative apparent activation energy. The value of this apparent activation energy is -9 kJ/mol for Pt-Rh(ll1). In the case of Rh(ll1) a value of -28 kJ/mol was reported. This difference illustrates the fact that the C02 production over the Pt-Rh alloy decreases less steeply with increasing temperature than on the Rh( 111) surface. This may be caused by an enhanced oxidation of Rh( 111) compared with Pt-Rh(ll1). The order in CO is negative at low temperature, switching to positive at higher temperature. At low temperature dense CO islands prevent oxygen adsorption, thereby inhibiting the reaction. At a temperature where CO
405
desorption starts sites are created on which oxygen can adsorb and subsequently react with adsorbed CO to form C02, which is released into the gas phase. The CO inhibition on the rough Pt-Rh(210) lasts up to relatively high temperature, i.e. 500K,at this temperature the CO oxidation rate over PtRh(ll1) has already reached its maximum. CO adsorbs in several binding states on Pt(210) [20], i.e. strongly bound and more weakly held states. Steadystate C02 production starts probably after considerable desorption of the tightly bound CO. This was also observed in the case of the CO oxidation over Ir surfaces [21]. The CO oxidation is usually considered as a typical example of a reaction that is insensitive to the surface structure [22]. For example, Oh et a1 reported that the specific reaction rate is the same for Rh(lll), Rh(100)and for high surface area supported Rh under high pressure conditions [4]. However, Fisher et a1 171 reported that at a total pressure of 2x10-7 TOITthe reaction proceeds much faster on Rh( 100) than on Rh( 111) at a temperature of 600K. FEM observations by Gorodetskii et a1 [30] also point to a large effect of the surface structure on the CO oxidation over Rh under low pressure conditions. Recently, Yates Jr. showed by ESDIAD that CO adsorbed on terraces of Pt(211) reacts at a lower temperature than CO adsorbed on step sites [19]. Rate oscillations in the C02 formation on Pt(100) are caused by a reconstruction of the surface from a (1x1) type of surface to a (5x20) type of surface. When the surface switches between the two different types of surfaces oscillations arise [23,31]. This model relies completely on the fact that the CO0 2 reaction on Pt is structure sensitive under the given experimental conditions. Structure sensitivity of the reaction is also reported for Ru [24,25] and Ir [21] surfaces at low pressures. However, Boudart [22] reported that the rate is independent of the particle size of Pd at high pressure. Moreover, the specific reaction rate was similar for Pd( 11l), (100) and (1 10) under low pressure conditions [3,26]. More recently, Kruse et a1 [27] found some evidence for a different behaviour of small Pd particles for the CO oxidation. On Ir(ll1) and (110) it was found that substantial desorption of the most tightly bound state of CO is required in order to achieve maximum activity in the steady state reaction of CO and oxygen to form C02 [21]. Furthermore, it was found that the maximum rate of C02 production occurs at lower temperature for Ir(ll1) than for Ir(ll0) [21]. Therefore, under various conditions structure sensitivity or insensitivity has been reported for several metals. At low temperature the CO-02 reaction is hampered by a large amount of adsorbed CO. Therefore, at low temperature the rate determining step is usually considered to be desorption of CO into the gas phase. After the desorption of a part of the CO, oxygen adsorption can take place. The last step, the reaction between CO and oxygen occurs relatively fast. Any structure sensitivity at low temperature must therefore be related to a difference in
406
adsorption energy of CO for different type of surface structures. For Pt it is reported that step sites show a desorption energy which is enhanced by 29 kJ/mol relative to the (111) flat surfaces [19]. However, the strong repulsion between adsorbed CO molecules could erase the structural heterogeneity, as has been suggested by Boudart in the case of Pd [22]. CO-CO repulsion on Pt(ll1) can only account for a 12-16 kJ/mol [28] decrease in adsorption energy of CO, if it assumed that the CO coverage does not exceed 0.5. Given the mild experimental conditions this assumption is justified. Thus an influence of step sites could still be expected, and has been observed in the present study. In conclusion, the CO oxidation by 0 2 is a surface sensitive reaction. However, under certain conditions the reaction could appear as being structure insensitive. Boudart suggested that lateral interactions between adsorbates erase the effect of different types of adsorption states, leading eventualy to structure insensitivity. A reactive mixture of CO and 0 2 could also modify the catalyst particle shape in such a way that one type of surface structure dominates. Another explanation could be that only the activity of one type of site is measured under experimental conditions. Recent results of Yates Jr [19]. et a1 are consistent with this model. They found that the CO + 0 2 reaction takes place at lower temperature on the terraces than on the steps of the Pt(211) surface using temperature programmed reaction conditions. However, at higher temperature the high diffusion rate of 0 adatoms over the surface obscures the differences of step and terrace sites.
References 1. 2. 3. 4.
5. 6. 7. 8. 9. 10. 11. 12. 13.
14. 15.
16. 17. 18. 19.
K.C. Taylor, Automotive Catalytic Converters (Springer, Berlin, 1984) I. Langmuir, Trans. Faraday SOC.17 (1922), 672 T. Engel and G. Ertl, Advances in Catalysis, 28 (1979), 1-78 S.H. Oh, G.B. Fisher, J.E. Carpenter and D.W. Goodman, J. of Cat. 100 (1986), 360-376 C.T. Campbell and J.M.J. White, J. Catal. 54 (1978), 289 W.M. Daniel and J.M. White, Int. J. of Chem. Kin., Vol 17 (1985), 413-417 S.B. Schwartz, L.D. Schmidt and G.B. Fisher, J. Phys. Chem. 90 (1986), 6194-6200 R.M. Wolf, J. Siera, F.C.M.J.M. van Delft and B.E. Nieuwenhuys, Faraday Discussions of the Chemical Society 87 (1989), 275 A.G. van de Bosch-Driebergen, M.N.H. Kieboom, A. van Dreumel, F.C.M.J.M. van Delft, R.M. Wolf and B.E. Nieuwenhuys, Cat. Let. 2 (1989), 73 T. Engel and G.Ertl, J. Chem. Phys., 69 (1978), 1267 B.E. Nieuwenhuys, Surf. Sci. 126 (1983), 307 F.C.M.J.M. van Delft and B.E. Nieuwenhuys, Surf. Sci. 162 (1985), 538 F.C.M.J.M. van Delft. A.D. van Langeveld and B.E. Nieuwenhuys Surf. Sci.189/190 (1987L 1129 T.T. Tsong, Surf. Sci. Rep. 8 (1988) 127 R.M. Wolf, M.J. Dees and B.E. Nieuwenhuys, J. Phys. (Pans) (1986), c7-419-/11 F.L. Williams and G.C. Nelson. ADD].Surf. Sci., 3 (1979). 409 F.C.M.J.M. van Delft, thesis Leidki, (1988) T.E. Gallon, Surf. Sci. 17 (1969), 486 J.T. Yates Jr., Private Communication v
407 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.
31. 32. 33. 34.
M. Ehsasi, J.H. Block, K. Christman and W. Hirschwald, J. Vac. Sci. Technol. A5(4), Jul/Aug (1987),821 W.H. Weinberg, W.F. Egelhoff Jr., V.P. Ivanov, V.L. Tataurov and G.K. Boreskov, R o c 7th Intern. Vac. Congr. and 3rd Intern. Conf. Solid Surfaces (Vienna 1977), 1151 M. Boudart and F. Rumpf, React. Kinet. Catal. Lett. Vol. 35,Nos 1-2, (1987)95-105 S.B. Schwartz and L.D. Schmidt, Surf. Sci. 206 (1988), 169-186 T.E. Madey, H.A. Engelhardt and D. Menzel, Surf. Sci. 48 (1975),304 P.D. Reed, C.M. Comrie and R.M. Lambert, Surf. Sci. 64 (1977),603 J. Koch, Thesis, Technische Universitat, Hannover (1972) V. Matolin, E. Gillet and N. Kruse, Surf. Sci. 186 (1987),L541 G. Ertl, M. Neumann and K.M. Streit, Surf. Sci. 64(1977),393 R.W. McCabe and Schmidt, Surf. Sci. 66 (1977),101-124 V.V. Gorodetskii, B.E.Nieuwenhuys, W.M.H. Sachtler and G.K. Boreskov, Appl. of Surf. Sci. 7 (1981),355-371 R.Imbihl, M.P. Cox and G. Ertl, J. Chem. Phys. 84 (1986),3519 J.L. Gland, B.A. Sexton and G.B. Fisher, Surf. Sci. 95 (1980),587 L. Gland, Surf. Sci. 93 (1980),487 T. Matsushima, J. Cat. 85 (1984),98-104
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZZ 0 199 1 Elsevier Science Publishers B.V., Amsterdam
409
MOLECULAR BEAM STUDIES OF CO OXIDATION ON Rh(ll0)
M. Bowker, Q. Guo, P.D.A. Pudney and R.W. Joyner Leverhulme Centre, Department of Chemistry and Surface Science Research Centre, University of Liverpool, P.O. Box 147, Liverpool L69 3BX England
Abstract The oxidation of carbon monoxide on Rh(ll0) at low pressures has been studied with a thermal molecular beam system. CO readily adsorbs on oxygen saturated Rh(ll0) at room temperature and oxidation follows a Langmliir- Hinshelwood reaction mechanism with an activation energy dependent on both oxygen and CO coverage. 0.75 monolayer(h4L) of CO preadsorbed on Rh(l10) completely blocks sites for oxygen adsorption. Above 530K,oxygen adsorption induces a series of surface reconstructions which strongly affect the rate of oxidation. It is proposed that the observed surface reconstructions may be associated with oxygen present in subsurface sites when the total coverage exceeds approximately 0.5 h4L and the rate of CO oxidation is limited by the presence of islands of reconstructed Rh with low reactivity.
1. INTRODUCTION The adsorption and oxidation of CO on group VIII transition metals has been studied extensively in the past [l--51 and continues to attract particular attention due to its technological importance in the pollution control of combustion products. The work reported here is part of our current programme on fundamental studies of reactions involving CO, NO, 0 2 and H2 on Rh(ll0) and modified Rh(l10) surfaces. Compared with the large quantity of work performed on Pt and Pd single crystal surfaces, much less has been done on single crystal rhodium which turns out to be the most important element in " three way catalytic converters " to perform the reduction of NO. Campbell et a1 have investigated CO oxidation on polycrystalline Rh wires[3]. Root et a1[6,7] have recently studied the reactions of NO and CO, NO and oxygen on Rh(ll1). The adsorption of CO on Rh(ll0) has been studied by Lambert[8] and Baird et a1[9]. Surprisingly no systematic study of CO oxidation has been done on Rh(ll0) which is a plane of high surface energy and therefore likely to better reflect the structure of highly dispersed particles in real catalysts. In a recent paper [lo] we reported molecular beam studies of CO and oxygen adsorption on Rh(ll0). A series of surface structures after oxygen adsorption were observed similar to those reported by Schwarz et al[ 113 who proposed simple overlayer structures of oxygen. Our study
410
proposes that significant surface reconstruction occurs due to oxygen adsorption. In this paper we report the study of CO oxidation on Rh(ll0) and the effect of surface reconstruction on the reaction rate.
2. EXPERIMENTAL The beam system and other equipment has been described eleswhere [12]. An effective pressure of 3x10-8 Torr and a beam diameter of 2.9mm at the sample is provided by the beam. The beam intensity is kept constant during experiments, but can be changed by adjusting the source pressure when necessary. The vacuum system contains a V.G. Micromass 200D quadrupole mass spectrometer for reactant and product analysis, LEED and AES for surface characterisation. The base pressure in the UHV chamber was less than 5 X 10-10 Torr. The rhodium crystal was supplied by Johnson Matthey PLC and has a bulk purity of 99.99% and was cut to within 0.5" of the (110) plane and polished afterwards; it is an ellipse with dimensions of 10 mm by 12 mm. After ion sputtering for a few hours the crystal was heated to lOOOK in 1 X 10-4 Torr of H2 for 20 hours. Sulphur was removed by cycles of ion sputtering and annealing. After the initial cleaning, the major impurity found during experiments was carbon which can be easily removed by heating the sample to 850K in 1 X 10-8 Torr of oxygen. A clean Rh(ll0) surface gives a sharp p(lX1) LEED pattern and has no sulphur, carbon or boron as checked with AES. The initial sticking probabilities of CO and oxygen were used as a reliable method to check if the surface is free from contamination.
3. RESULTS 3.1- CO and o x p e n adsorption Figure 1 shows the sticking probability(S) dependence on coverage for CO and 0 2 . The details of the method of measurement have been given elsewhere[l2]. The initial sticking probability for CO is 0.68 and is independent of sample temperature. S diminishes with increasing coverage and is zero at 1.OML coverage close to room temperature. TPD of CO shows that there are three desorption peaks which are possibly associated with three phases of CO on the surface. Although we observed only one ordered structure close to 1ML coverage at room temperature-(2xl)plgl-, Weimer et a1 reported at least three ordered structures at different coverages 1131 when CO is adsorbed at low temperatures. The oxygen sticking probability has a zero coverage limit of 0.62 and its dependence on coverage is rather similar to that of CO. However, the adsorption mechanism is probably different. CO adsorbs through a precursor intermediate, oxygen does not since sample temperature has little effect on the
41 1
shape of the S vs. 8 coverage curves. It was suggested that surface reconstruction takes place even at very low oxygen coverages [lo] and the reconstruction changes the property of the surface. The saturation coverage is 0.65ML at room temperature. At 540K and above, oxygen coverage reaches 0.72ML at which point S has decreased to 0.02 and though the coverage still increases slowly with further exposure. The full analysis of the LEED patterns and the proposed surface structures is presented eleswhere [lo].
S
0.0
0.2
0.4
0.6
0.8
1.0
1.2
COVERAGE(mono1ayers) Figure I The sticking probability ( S ) as a function of coverage for . ( o ) CO at room temperature ;( 0 ) 0 2 at 570K 3.2 - CO oxidation Figure 2 shows the rate of CO2 production as a function of time after CO was beamed onto the surface precovered with saturated oxygen at 330K. The initial sticking probability of CO on the oxygen saturated surface is still 0.48 which is only 30% less than So for CO adsorbing on the clean surface. In contrast to the adsorption on the clean surface where SO is independent of temperature, SO was found to decrease from 0.48 to 0.30 between 320K and 520K. This could due to a variety of factors such as structural ordering in the surface or a changed potential profile. CO2 formation occurs by reaction of adsorbed CO with atomic oxygen. Figure 2 also shows the calculated result of CO oxidation based on the following equations:
412
(3) where
= - RC02
deddt
ec0, 8,
are the coverage of CO and oxygen respectively. R,,,
is the
C 0 2 formation rate. The dependence of the sticking probability S on the coverage is represented by precursor kinetics with the observed zero coverage limit. The pre-exponential factor A1, the desorption energy El and its dependence on the coverage of CO are obtained from the analysis of the TPD of co [lo].
Mass Spec. Signal (a.u.1 I
0
1
2
3
4
&ML
1
5
‘0 6
TIME(MIN.)
Fiaure 2 The mass spectrometer signal of C02 (m) and the scattered CO (+) during titration of a saturated oxygen layer with a beam of CO at 320K. The related signals of C02 ( A ) and CO ( 0 )from model calculation are shown for comparison together with the calculated oxygen coverage ( X ) which is not experimentally measured
413
The beam intensity I and the initial coverage of oxygen are experimentally determined. The pre-exponential factor A2 is set to 3X10'2s-' which is of the right order for second order reactions when coverages are expressed in monolayers [2]. The activation energy E2 used is in the form of
plus an extra term of {-45(00-0.5)3'2) kJmol-' when oxygen coverage is higher than OSML and another term of {-12(~co-0.S))kJmol~1 when CO coverage is greater than 0.5ML. These changes with coverage are due to lateral interactions between species in the adlayer and are essential to give a good fit to the observed profile (fig.2). Although the exact choice of activation energy may not be correct, the important point is that the curves cannot be fitted with a constant value. It is to be expected that it will depend fairly strongly on the binding energy of oxygen and CO which in turn vary strongly with coverage. The E2 value is larger than that reported for Rh( 111) [14], but there an unreasonably low value of A2 was used for curve fitting (lo7 s-'). At this temperature oxidation is incomplete with about 0.05ML of oxygen left on the surface. This is possibly caused by the presence of very strongly bound oxygen at low coverages. When similar experiments were performed at 560K, the oxidation behaviour was very different as shown in figure 3. The surface was at first exposed to 2 min. of oxygen at 560K to give a coverage of 0.68ML and a np(2x3) LEED pattern. Then the CO beam was opened at time zero. The C02 rate remains constant for a considerable time and then it increases to a maximum before diminishing to zero when all the oxygen is consumed. The corresponding CO curve shows that the uptake of CO is directly proportional to the C02 formation rate. The life time of CO molecules on the surface is short under these conditions and the CO coverage is low. The availability of surface oxygen seems to determine the rate of oxidation at the start of the experiment. Observation of the evolution of the LEED pattern during this clean experiment showed the ~ ( 2 x 2 )structure present when the rate goes through the second stage (enhanced rate). Lowering the initial coverage of CO simply results in an initial rate similar to that shown in figure 3 for that oxygen coverage. Thus if 0.4 monolayers are predosed the rate begins near the maximum in figure 3. If the Rh(ll0) is first saturated with CO below 350K, subsequent exposure to oxygen does not lead to C02 formation because the sites for oxygen adsorption are all blocked by CO. Oxygen can adsorb when CO coverage is below 0.75ML.
414
Mass Spec. Signal (a.u.1
= J
0
1
2
3
4
5
6
TIME(MIN.) Figure 3 Mass spectrometer signals of C 0 2 (m) and scattered CO ( + ) during titration of a saturated oxygen layer 4. DISCUSSION
CO oxidation on Rh(ll0) follows a Langmuir-Hinshelwood reaction mechanism. However, quantitative descriptions are not always satisfactory because the repulsive interactions between adsorbates modify the apparent activation energy, and the exact dependence of activation energy on coverage is more complicated than the linear dependence which is assumed in many models [14]. The activation energy for CO oxidation on transition metals has been found to vary significantly with oxygen coverage. Gland et a1 [15] reported an activation energy of 166kJmol-1 for CO oxidation on Pt(ll1) at zero oxygen coverage and 68kJmol-1 at saturation coverage. Campbell et a1 [2] attribute this decrease in activation energy to repulsive interaction between the coadsorbed reactants. Previous analysis [lo] shows that there is a net repulsive interaction in the adsorbed CO layer. The initial high reactivity of the surface at 320K with saturated oxygen is caused by two things: first is the relatively low binding energy of oxygen to rhodium (evidenced in desorption spectra [10,11] at high coverages. As CO oxidation removes oxygen the
415
binding energy of the remaining oxygen on the surface increases. Secondly it is probably due to a weaker binding of CO on the oxygen covered surface, therefore, requiring less thermal activation for reaction. The decrease of S with temperature on oxygen saturated Rh(l10) does not seem to have a big effect on CO oxidation. Increasing the temperature from 320K to 420K causes a dramatic increase of the C02 formation rate in spite of the decrease of So from 0.48 to 0.3. This shows that within this temperature range the rate determining factor is probably the CO oxidation itself. Campbell et a1 [2] have found a reduction of C02 formation rate as temperature increases from 430K to 560K and attributes this to the observed decreasing of S on Pt(ll1). Similarly, the decreasing of reaction rate with temperature was also seen in our experiment on Rh(ll0). However, we find that the changing of S alone is not enough to account the changes in reaction rate even qualitatively. We propose that the oxygen induced surface structure is the most crucial factor that controls the reaction. The adsorption of oxygen shows that above 0.68ML coverage the surface still takes up oxygen but very slowly. It is thus assumed that at this stage oxygen begins to diffuse to the subsurface region which is a slow process due to the activated nature of diffusion. It is further assumed that while oxygen is absorbing, rhodium atoms move up towards the surface. The total effect is to bury oxygen in the second layer, so adsorbed CO cannot easily react with it. The proportion of subsurface oxygen is a function of temperature and exposure. By exposing to oxygen for a longer time at constant temperature more surface oxygen is converted to subsurface oxygen, and LEED indicates the top layer may become more close packed so reducing surface energy. When such a surface is exposed to CO, oxidation takes place at the top layer removing surface oxygen which is in low concentration perhaps due to equilibrium segregation. This leads to the back diffusion of subsurface oxygen as a result of the decreasing surface oxygen concentration. At constant temperature, the surface structure was found to depend only on oxygen coverage, and phase transitions occur at certain coverages. The increase of CO2 formation rate to a maximum shown in figure 3 is due to the change of structure from np(2X3) to (2X2)plgl as observed with LEED. There is little doubt that CO mobility in some form of the precursor state is important in determining the reaction kinetics, since, for a fixed initial high oxygen coverage, the initial rate decreases with increasing temperature above 450K. This is due to the decreased lifetime of the CO in this adsorbed precursor. Detailed considerations of the reaction kinetics and modelling will be given in the near future [ 151. The above discussions are complemented by our results for steady state CO oxidation with a mixed beam of CO and oxygen [MI. When the concentration of CO is equal to that of oxygen, oxidation at room temperature
416
is inhibited by too much CO on the surface. As the substrate temperature increases, CO2 formation rate increases and then decreases again above 500K where the reconstruction of the surface becomes significant and reaction more difficult. The issue of subsurface oxygen on Pd(ll0) has been discussed by He et a1 [16,17]. The reconstruction on Pd(ll0) is different from that on Rh(llO), but the basic mechanisms of formation of subsurface oxygen is similar. Recently Basset and Imbihl [ 181 have proposed subsurface oxygen can effect kinetic oscillations in the CO oxidation on Pd(ll0). No one has carried out a study on Rh(ll0) regarding kinetic oscillations, but based on our transient study of CO oxidation, oscillations on Rh( 110) may occur due to the different structural phases present on this surface.
Acknowledgements: The authors would like to express their gratitude to the SERC and to Johnson Matthey PLC for financial support for this work through a cooperative research award.
References 1.2. 3. 4. 5.
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
17. 18.
T. Engel and G. Ed,Adv. Catalysis 28 (1979) 1 and references therein C.T. Campbell, G. Ertl, H. Kuipers and J. Segner, J.Chem.Phys. 11 (1980) 5862 C.T. Campbell, S.K. Shi and J.M. White, Appl. Surf. Sci. 2 (1979) 382 J.L. Gland and E.B. Kollin, J. Chem. Phys. 78 (1983) 963 J.R. Engstrom and W.H. Weinberger, Sulf. Sci. 201 (1988) 145 T.W. Root, L.D. Schmidt and G.B. Fisher, Surf. Sci. 150 (1985) 173 T.W. Root, L.D. Schmidt and G.B. Fisher, Surf. Sci. 134 (1983)30 R.A. Marbrow and R.M. Lambert, Sulf. Sci. 67 (1977) 489 R.J. Baird, R.C. Ku and P. Wynblatt, Surf. Sci. W (1980) 346 M. Bowker, Q. Guo and R.W. Joyner, to be published E. Schwarz, J. Lenz, H. Wohlgemuth, and K. Christman, Vacuum 41 (1990) 167 M. Bowker, P.D.A. Pudney, and C.J. Barnes J. Vac. Sci. andTechno1. A8 (1990) 816 J.J. Weimer, J. Loboda-Cakovic and J. Block, J. Vac. Sci. and Technol. AS (1990) 2543 S.H. Oh, G.B. Fisher, J.E. Carpenter and D.W. Goodman, J. Card. 100 (1986) 360 M. Bowker R.W. Joyner and Q. Guo, in preparation J. He, U. Memmert, K. Griffiths, and P. Norton, J . Chem. Phys. 90 (1989) 5082 J. He, U. Memmert and P. Norton, J. Chem. Phys. 90 (1989) 5088 M.R. Bassett and R. Imbihl, J. Chem. Phys. 93 (1990) 811
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V., Amsterdam
417
ROLE OF SULPHATE DECOMPOSITION IN THE EMISSION AND CONTROL OF HYDROGEN SULPHIDE FROM AUTOCATALYSTS
A.F. Diwell, S.E. Golunski, J.R. Taylor and T.J. Truex.
Johnson Matthey Technology Centre Sonning Common, Reading RG4 9NH, UK Abstract Three-way autocatalysts comprised of Pt-RWCe02-Al203 are able to store sulphur and then to release it suddenly in the form of H2S. The storage occurs by the formation of sulphates under a fuel-lean exhaust gas; the release is induced by a change to rich conditions. Current production autocatalysts are suitably modified to minimise the effects of this storagehelease cycle. Thermodynamic calculations have been used to predict the stable phases of a catalyst, and to model the possible changes in pH2S after a lean-rich transition. For a series of metal oxides ,which may be components of a complete catalyst, substantial differences are expected in the composition of the gases evolved during the dissociation of the corresponding metal sulphates. For example, the decomposition of Ce2(S04)3 should yield CeO2, with S and 0 being released in the ratio 1:2.66, but NiS04 should ultimately form Ni3S2, with a ratio of 1:12 for S:O. Despite such differences, the calculations indicate that Ce, Al, Fe and Ni (but not Ca) should each act to reduce the steady-state H2S content of the fuel-rich gas stream, provided the gas stream is saturated with the decomposition products.The known ability of Ce02 to increase the emission of H2S well above the steady-state concentration implies that kinetic limitations must prevent final equilibrium being reached. This study reveals that even if a catalyst additive forms sulphates under lean conditions, it may still be able to attenuate H2S. In the ideal case, the sulphate will readily decompose on switching to a rich atmosphere, and will release a large excess of oxygen (relative to sulphur); the resultant oxide may also be able to scavenge any available H2S from the gas-phase.
INTRODUCTION
Under a fuel-rich exhaust gas, a continuous low-level emission of H2S typifies the steady-state reduction of SO2 in the presence of PGM (platinumgroup metals) [ 1-31. However, H2S emission at concentrations which exceed SO2 inlet levels can also occur if an autocatalyst is capable of retaining sulphur. Unmodified three-way catalysts, in which PGM is supported on Ce02-Al203, can effectively store sulphur during a lean phase and then release it in the form of a large "spike" of H2S when the exhaust gas becomes rich [2,3].
418
Adsorption studies [4,5] have shown that sulphur storage can occur by the direct interaction of SO2 with both Ce02 and Al2O3. Under typical operating conditions, the adsorbate is more likely to be SO3 [4,6], with the C e 0 2 providing the preferred adsorption sites [4] for the formation of sulphates [2-4,7,8]. Therefore, one approach to controlling the sudden emission of H2S is to prevent the formation of the sulphate species by modifying the CeO2 [5]. Another method, which has generally proved more successful, is to trap the H2S once it is formed [7-91. In the present study we have carried out a series of thermodynamic calculations to predict the stable phases in a standard three-way catalyst under steady-state conditions, and to model the effect of inducing the decomposition of metal sulphates on the composition of the gas-phase.The results are considered in light of the known H2S emission characteristics of the complete catalyst and its individual components.
STABLEPHASES Using a free-energy minimisation method (MTDATA [lo]), we have identified the stable phases of a three-way catalyst (Pt-RWCe02-Al203) in equilibrium with an exhaust-gas stream, at different equivalence ( h ) ratios.(Table 1) This computational method requires knowledge of the free energies of all gaseous species considered to be present and of all the solid compounds which could possibly be formed.In each calculation, the amount of every element and the reaction temperature are entered; the "output" provides the composition of the gas-phase and the amounts of any condensed phases formed. A sufficiently small amount of metal is assumed to come into contact with the gas stream, so that the formation of a compound does not alter the bulk gas composition. TABLE 1 :
Atomic composition of simulated exhaust-gas mixtures
h
0.92
H 0 C N S
20.90 39.25 15.90 147.4 0.002
I
0.98 I 0.99 I 1.oo Amount per 100 moles / mol 19.845 19.68 19.515 39.604 39.674 39.744 15.153 15.0305 14.908 148.83 149.03 149.23 0.002 0.002 0.002
I
1.02 19.205 39.854 14.703 149.63 0;002
The calculations show that, in an unmodified catalyst, sulphates should form on Ce02 and A1203 (Table 2) under lean or stoichiometric conditions, over a limited range of temperatures.These sulphates should then decompose
419
to the oxides when the exhaust-gas becomes reducing.The Pt and Rh are not expected to form sulphates, and should only retain sulphur under rich atmospheres.Similar conclusions have been drawn from a recent experimental study of sulfur-poisoning of supported PGM catalysts [ 1I]. TABLE 2
Stable phases of Pt-Rh/Ce02-A1203 under exhaust-gas conditions
h
700
T, 900
1100
1200
1300
1.02
Rh Ce02 1.oo
0.98
Rh Pt
Rh 0.92
Pt
-
-
Although the interaction of PGM with gas-phase sulfur has been extensively studied [ll-151, the mechanism by which an exhaust-gas yields H2S is still the subject of dispute. For example, a direct correlation has been reported [5] between the number of measured surface Pt atoms and the number of sulphur atoms contained within the sharpest portion of the H2S spike.following a lean-rich transition.The authors concluded that Pt can store sulphur during a lean phase, and that this sulphur is rapidly released on switching to a rich gas. Clearly,our results conflict with this interpretation, and are more consistent with the view [eg 111 that the PGM functions as a catalyst in the reduction of SO, (so42--S02or so3)to H2S. LEAN-RICHTRANSIENTS The same computational program has been used to model the effect of sulphate decomposition on the gas stream. The sulphates considered were those of metals which are either present in the catalyst support (Ce, Al) or can be used to attenuate the emission of H2S (Ca [l], Ni[7-9], Fe[8]).
420 It has been assumed that a relatively large amount of the metal sulphate is suddenly exposed to destabilising conditions ( h = 0.99; T = 923 K). The gas stream can be envisaged as a succession of small "packets", which reach local equilibrium with the decomposing compound. In the case of a metal sulphate dissociating to a metal oxide of the same valency, the reaction proceeds by adding S and 0 in the ratio 1:3 to the gas stream at h = 0.99 and T = 923 K.An equilibrium is reached between the gas-phase and a mixture of unreacted sulphate and oxide product. The gas is then considered to be saturated by the decomposition products. Further decomposition can only occur when the gas has been replaced by a new "packet". The effect of the decomposition reaction on the gas-stream (at a particular equivalence ratio and specific temperature) is particularly dependent on two factors : (i) the ratio of S:O released during decomposition; (ii) the degree of decomposition required before equilibrium is attained in each new "packet" of gas. (This is dependent on the thermodynamic stability of the sulphate, ie the more stable the sulphate, the lower will be the extent of decomposition before equilibrium is reached.)
s:o Process
(released)
A12(S04)3 -+ A1203 CaS04 + CaC03 Ce2(S04)3 -+ Ce02 Fe2(so4)3 -+ Fe203 k 2 0 3 +k 3 0 4
1:3 1:l 1:2.66 1:3.1
NiS04 + NiO NiO + Ni + Ni3S2 Ni -+ Ni3S2
1:12
Moles S released per 100 moles gas (at saturation) 2.84 0.028 0.397 12.70 -
pH2S I atm (at saturation)
0.366
1.60 10-24 4.03 x 10-6 9.89 x 10-6
1.28 2.40 2.70 5.59 x 1.25 x
10-25 10-4 10-23 10-26 10-12
The net ratio of S/O emitted during decomposition, and the expected solid products are shown in Table 3. (It should be noted that for Ni and Fe the decomposition occurs in multiple steps.). Calculations were carried out for each of these decompositions by considering the effect of progressively adding S and 0 (in the appropriate fixed ratio) to the elements present in the rich gas-feed (Table 1). The gas-stream composition was recalculated after each addition. The calculated changes in partial pressure of H2S, as a function of amount of sulphur added, are presented in Figures I ( i ) and I(ii). These plots
42 1
are both based on the same data, with the length of the x-axis in Figure I(ii) being chosen to allow the final equilibrium points for A1 and Fe to be shown. The curves reveal the relationship between added S and the resultant pH2S for different classes of reaction; the decomposition of a particular metal sulphate can then be represented by a single point on the appropriate curve. Similar trends are observed for four (of the five) S/O ratios, and hence for the decomposition of the corresponding sulphates (ie, Ce, Al, Ni, Fe). In these four cases, the addition of S (and 0) results in an initial increase in pH2S (above the level predicted for the steady-state conversion of S02), but eventually leads to a decrease. Thus, any tendency towards increased H2S formation through the release of sulphur is eventually offset by the removal of free hydrogen by the associated release of an excess of oxygen. Only when the S/O ratio is 1/1 (ie CaS04 + CaC03) is there insufficient release of oxygen to offset the tendency towards H2S formation.
ATTENUATIONOF H2S On the basis of the above results alone, it could be concluded that Ce, Al, Ni and Fe should all be capable of attenuating H2S, but that Ca should contribute to the emission. In practice, Ca additives have been used as H2Sgetters [ 11, whereas Ce is believed to be responsible for much of the emission from current catalysts [4,5,7].However, the predicted ability of Ce2(S04)3 to attenuate H2S is clearly dependent on a sufficient release of S and 0 to ensure that pH2S falls below the steady-state level (see Figure l(i),e); this occurs only when the gas-phase concentration of the released sulphur is >80% of the value at which equilibrium is achieved (Table 3). Clearly, if the rate of decomposition is low, then the release of sulphur may be insufficient to achieve 80% saturation. The belief that calcium functions as an H2S-getter [ l ] is difficult to justify in terms of our model. Even if it is assumed that kinetic limitations apply, such that the decomposition of CaS04 yields essentially CaO (ie the subsequent formation of the carbonate is very slow), our calculations indicate that Ce2(S04)3 should decompose before CaS04. The Ca should not, therefore, exert any control over the emission of H2S caused by nonequilibrium decomposition of Ce2(S04)3. Decomposition itself should raise pH2S. It must be stressed that the final H2S level (ie at saturation) depends on the thermodynamic characteristics of each sulphate. For example, the S/O ratio may be the same for two decomposition reactions, (eg A12(S04)3 > A1203 and La2(S04)3 > La2O3), and so both can be represented by the same pH2S curve, but the amount of decomposition necessary to achieve saturation can differ greatly.
422
+O steady-s tafe
(d) S
2.66 0
Ce
. 30
(a) S 0
+
+
12 0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 moles S added per 100 moles exhaust-gas
Figure I (i) :Sulphate decomposition trends 0
----
(el
-5
steady-state
-10
\
-15
-20 Fe
-25
-30 0
0
(a)
1
(d 2
3
4 5 6 7 8 9 1 0 1 1 1 moles S added per 100 moles exhaust-gas
Figure 1 (ii) :Extended decomposition trends
2
1
3
423
It has previously been assumed [8], that if a metal is capable of forming a stable sulphate under lean conditions, then it will not attenuate H2S during a subsequent change to a rich atmosphere. Our calculations indicate that this is not necessarily true. The decomposition of a sulphate has the potential to lower the pH2S below the steady-state value provided the proportion of S/O in the released gases is <1. Nevertheless, even if the stoichiometry is favourable, other factors (thermodynamic and kinetic) may prevent thesulphate from performing as an attenuator. EFFECT O F NICKEL
In the past, the role of additives, such as Ni, has been thought to be that of an H2S-getter [5,7-91. However, it has become apparent from the present work that Ni can also function by forming a sulphate, which decomposes to release a large excess of oxygen (compared to S). The effect of this decomposition is not limited to the attenuation of H2S formed by the steadystate conversion of S02, but can also influence the emission of stored sulphur. For example, if non-equilibrium dissociation of Ce2(S04)3 is assumed, the decomposition of NiS04 should ensure the complete or partial re-formation of the cerium sulphate (from any Ce02 formed). Only if all the NiS04 decomposes to NiO before the dissociation of Ce2(S04)3 begins, will the NiO then function solely as an H2S-getter. It is clear that nickel has a number of characteristics which enable it to perform as a very effective attenuator of H2S. It is essential that these are considered when searching for alternatives to Ni-additives. REFERENCES 1. J W Hightower, in "Preparation of Catalysts", eds B Delmon, PA Jacobs and G Poncelet (Elsevier, Amsterdam, 1976),p 615. 2. H Windawi and TJ Truex, 10th North American Catalysis Society Meeting, San Diego, USA (1987). 3. TJ Truex, H Windawi and PC Ellgen, SAE paper 872162 (1987). 4. AF Diwell, C Hallett and JR Taylor, S A E paper 872163 (1987). 5. ES Lox, BH Engler and E Koberstein, SAE paper 890795 (1989). 6. KC Taylor, Ind. Eng. Chem., Prod. Res. Dev., 15, 264 (1976). 7. JS Rieck, W Suarez and JE Kubsh, SAE paper 892095 (1989). 8. T Yamada, K Kayano and M Funabiki, SAE paper 900611 (1990). 9. MG Henk, JJ White and GW Denison, SAE paper 872134 (1987). 10. TI Barry, RH Davies, AT Dinsdale, JA Gisby, SM Hodson and NJ Pugh, "MTDATA" (National Physical Laboratory, Teddington, UK, 1989). 11. A Li-Dun and Dy Quan, Appl. Catal., 66,219 (1990) 12. R Maurel, G Leclercq and J Barbier, J. Catal, ,324 (1975) 13. LL Hegedus and RW Mc Cabe, in"Cata1yst Deactivation", eds. B Delmon and GF Froment 47 1(Elsevier, Amsterdam, 1980), 14. T Wang, A Vasquez, A Kato and LD Schmidt, J. Catal.,Zg, 306 (1982). 15. CR Apesteguia, TF Garetto, CE Brema and JM Parera, Appl Catal., u 2 9 1 (1984)
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZZ 0 199 1 Elsevier Science Publishers B.V.. Amsterdam
425
THE EFFECT OF AGING ON NITROUS OXIDE N 2 0 FORMATION BY AUTOMOTIVE THREE-WAY CATALYSTS M. Prigent, G . de Soete and R. Dozihre
Institut Francais du Pe'trole, B.P. 31I , F-92506 Rueil-Malmaison, France
Abstract Three series of Pt Rh three-way catalysts were aged in the exhaust of an engine running on a dynamometer stand for various lengths of time. Their activity was the11 measured as a function of temperature at stoichiometry and as a function of the air-fuel equivalence ratio at constant temperature (450°C). A peak of N 2 0 formation was observed close to the catalyst light-off temperature. Since the light-off temperature increases as a function of aging time, a shift in the N 2 0 peak formation was observed towards higher temperatures. The magnitude of the N 2 0 peak decreased simultaneously. At 450°C a minimum of N 2 0 formation was seen close to stoichiometry with two maxima below and above this point. N 2 0 formation was the highest on the rich side and increased progressively as a function of catalyst aging. Another series of eight catalysts was evaluated in parallel on a vehicle on a chassis dynamometer after engine bench aging. They were tested using the ECE driving cycles: ECE 15 urban with cold or hot starting and extraurban with hot starting. The amount of N 2 0 emitted did not vary very significantly with the catalyst type but was seen to increase 2 to 4.5 times after aging, depending on the driving cycle used. This has to be related to the shift in the light-off temperature to higher values during aging, which makes the catalyst work more at temperatures where N 2 0 is formed.
1 - INTRODUCTION Nitrous oxide ( N 2 0 ) is an absorber of IR radiation and thus contributes with other tropospheric molecules ( H 2 0 , 0 2 , CH4, CFC, etc.), around 8% to the so-called "green house effect". Since N 2 0 is a very stable molecule with a lifetime of 170 years, it can also be diffused into the stratosphere where it plays an important role in the ozone layer depletion (1 - 4). The actual tropospheric N 2 0 concentration is steadily increasing (0.7 ppbv per year) and the most recent calculations show that total yearly emissions are around 14 f 3 megatons (as N) of which 4 f 1 Mt is of anthropogenic origin (5). The main identified anthropogenic N 2 0 sources are fossil fuel combustion ( - 0.6 Mt/yr), biomass burning (- 0.5 Mt/yr) and fertilizer use ( 1 Mt/yr). The origin of the remaining 1.9 Mt/yr of
-
426
anthropogenic origin is still uncertain, but it was recently assumed that it could be due to the a posteriori formation of N20 from NO by gas-phase reactions in the atmosphere, or by biogenetic production from dry or wet ground deposition (acid rain) (5). In the exhaust from internal combustion engines there is relatively little N 2 0 . In a previous paper we showed that N20 represents less than 1% (between 0.4 and 0.75%) of the overall NOx emissions from a diesel or gasoline engine without catalytic aftertreatment. We also showed that threeway catalysts (TWC)used to decrease NOx emissions could increase the N20 concentration in the exhaust of spark ignition engines (6). New Pt+Rh TWCs were found to increase N20 mainly during catalysts light-off, but to have nearly no effect at medium temperatures around 400 to 500"C, and above to decrease N20 slightly . The present project was undertaken to evaluate the effect of catalyst aging o N20 formation and to determine whether the tendency of TWCs to increase N20 emissions was progressively lessened as a function of the mileage accumulation or enhanced, on the contrary.
-
2 EXPERIMENTAL
All the experiments were performed on engine test bench or with a vehicle on a chassis dynamometer. A four cylinder engine with 2.2 litres displacement, with electronic fuel injection and air-fuel control via an exhaust gas oxygen sensor, was used for aging the catalysts. This engine was run according to the following cycle:
Mode 1 simulates driving a car uphill (slope 4%) at 140-150 k m h r (inlet gas temperature = 78OOC). Modes 2 and 3 simulate a sudden deceleration: first, 20% of spark ignitions were suppressed to produce poor combustion and to cause a temperature spike in the catalyst of around 1000°C, and then air was injected to represent fuel cutting and to produce an oxidizing atmosphere. In addition engine oil consumption was increased by injecting of 40 cm3/hr of lubricating oil into the gasoline. For activity measurements a similar engine with means for the external control of the intake air-fuel ratio and of the exhaust gas temperature, as shown in Fig. I, was used. The gas space velocity
427
was maintained constant and equal to 50,000 h-1 during all the tests. Gases were analyzed for the determination of CO, HC, NO, NO2 and 0 2 by standard equipment. The equivalence ratio was determined by on-line computation from gas analyser outputs and also by the use of a specific lambda analyzer (Lamdascan Cussons).
Figure 1 - Engine bench set-up used for catalytic activity determinations.
The vehicle tests were performed with a vehicle having the same engine as the one used for the engine bench tests. The catalytic muffler was located at a distance of 1.4 m from the engine. The tests were performed first with a new catalyst after preconditioning at 80 km/hr for 1 hour the day before. They were repeated after aging of the catalytic mufflers for 24 hours on an engine bench under the aging conditions described above. The measurements were performed on a chassis dynamometer using the following driving cycles: European urban cycle ECE 15 (cold or hot start), European extra-urban driving cycle EUDC (hot start). N 2 0 was determined in a sample of raw (engine bench tests) or diluted (vehicle tests) exhaust gas stored in a stainless steel bottle for less than 30 minutes. The amount of N 2 0 contained in the dilution air was also determined and deduced from the results. The gas chromatograplly method using an electron capture detector, with a carrier gas composed of 5% methane in argon was used.
428
An initial separation of the gas sample was done in a Porapak-Q column at 80°C. The light fraction (including N20) issuing from this first column was then sent into a second molecular sieve column operated at 250°C for further separation and detection by an Ni 63 electron captor operated at 280°C. The heavy fraction of the sample was backflushed out of the Porapak column to prevent detector contamination. In this way perfect separation of N20 and C02 peaks was obtained together with an N 2 0 detection level of 0.02 ppm and high detector response stability. The catalysts used in these experiments were based on Cordierite monolithic substrates having a nominal cell density of 62/cm2 (400/in2) with 160 pm wall thickness. Round pieces with a diameter of 11.84 cm and a lengh of 7.62 cm were used in the engine bench tests. Race-track pieces with a minor x major axes of 8.08 x 16.97 cm and a lengh of 7.62 cm were used in the vehicle tests (two pieces per muffler with a total volume of catalyst of 1.8 litre). These monoliths were coated with around 100 g/L of an alumina-based washcoat and were subsequently impregnated with chloroplatinic acid and rhodium trichloride at a concentration of 0.706 and 1.413 g / L with a Pt/Rh mass ratio of 5 (Table 1).
TABLE1 Composition of the catalysts used for engine bench or vehicle tests
3
- RESULTS
AND DISCUSSION
Effect of temperature on N2O formation at h = 1-00. As observed previously the results obtained confirm that the amount of N 2 0 formed goes through a maximum at a temperature close to the catalyst
light-off temperature. This can be seen for example in Figures 2 a and b where the percentage of the initial NO converted into N20 (initial NO here was 2600 ppm), as well as the conversion percentage of CO, HC and NO converted, are plotted as a function of gas temperature for catalyst B aged for 10 minutes or
429
200 hours. After a rapid rise, the amount of N 2 0 formed decreases slowly up to temperatures around 500°C where it becomes nearly nil. 100
10
T
8e
80
0 2s 0
0 250
300
350
400
450
500
550
TEMPERATURE (DEG.C) Figure 2 (a ) :Catalyst B aged 10 minutes. I0
100 T
80
0 2s 0
0
250
300
350
400
450
500
550
TEMPERATURE (DEG.C) Figure 2 (b ) :Catalyst B aged 200 hours. Figure 2 Percent conversion of CO, HC and NOx (gross and NO converted into N20) measured on an engine bench at stoichiometry (A=I .OO)as a function of inlet gas temperature. GHSV = 50,000 h-1.
430
in
.-.
e
H
6 4
2
n
n TEMPEUATUUI:' (1)EG.C)
Fig3 (b) :Catalyst B (washcoat: yAl2O3 + CeO2)
F i g 3 (a) :Catalyst A (washcoat: yAl2O3) in
-i c
2 2 n 250
300
350
400
450
500
550
TEMPERATURE (DEG.C)
Fig3 (c) :Catalyst C (washcoat: y A1203 + CeO2 + promotor) Figure 3 Percent of NO converted into N 2 0 at stoichiometry ( A =I .OO)as a function of inlet gas temperaturefor catalysts aged on an engine bench for different lengths of time ( t 10 min, A 1 hr, 6 hr, 0 40 hr and + 200 hr). The evolution of the amount of N20 formed as a function of catalyst aging time is shown for the three different catalysts A, B and C and the five aging times (10 min, 1 hr, 6 hr, 40 hr and 200 hr) in Figures 3 a, b and c. A progressive increase in the N20 peak temperature is observed after 10 min, 1 hr, 6 hr, 40 hr or 200 hr of aging. Figure 4 shows the good correlation that exists between the N20 peak temperature and the catalyst light-off temperature, taken here as the temperature required to obtain 50% CO conversion.
43 1
I
I
300
I 350
I 400
T F O R SO?& CO CONVERSlOh’(deg.C)
Figure 4 Correlation between N20 formation peak temperature and catalyst light-ofl temperature (T.50 for CO), after engine bench aging for different lengths of time between I0 min and 200 hr: # catalyst A, catalyst B, + catalyst C .
Effect of air-fuel ratio on N20 formation at, 450°C. The CO, HC and gross NOy conversions curves obtained between an equivalence ratio of 0.96 (lean) and 1.04 (rich) are shown as an example in Figures 5 a and b for catalysts B by aged 10 minutes or 200 hours. The percentage of the initial NO converted into N 2 0 is also given in these two figures and is shown in Figures 6 a, b and c for the three catalysts A, B, C and the five aging times (10 min, 1 hr, 6 hr, 40 hr and 200 hr). It can be seen that the amount of N 2 0 formed goes through a minimum at stoichiometry with two maxima (or a plateau) below and above this point. The amount of N 2 0 formed generally increases as a function of the aging time. This is not the case however for the catalysts aged 200 hours, for which the amount of N 2 0 formed returns, for an equivalence ratio of 1.04, to the value (or below) obtained with the nearly new catalysts aged for only 10 minutes. But a s can be seen from Figure 5 a, the catalyst activity for NO reduction is very low under these conditions.
432
-
100 80
2
$ 9: 3
40
0
20
s
Fig.5 (a) : Catalyst B aged I 0 minutes
60
0 0.96
0.98
ID0
1.02
1.04
EQUIVALENCE RATIO
10
1W
Fig.5 (b) : Catalyst B aged 200 hours
0
0 0.96
0.98
ID0
1.02
1.04
EQUIVALENCE RATIO
Figure 5 Percent conversion of CO, HC and NOx (gross and NO converted into N 2 0 ) measured on an engine bench at 450 "C as a function of the air-fuel equivalence ratio. GHSV= 50,000 hN2O vehicle emissions. The gaseous emissions of a vehicle with electronic fuel injection and h control were determined on a chassis dynamometer with 8 different new or engine-bench-aged Pt Rh three-way catalysts (see Table 1 for catalyst composition). The tests were performed according to the European driving cycles, i.e. the urban driving cycle (with cold or hot start) or the extra-urban driving cycle (hot start only). The same measurements were also performed without any catalyst. A dummy monolith without a washcoat and noble metal was used instead.
433
111
,
I
I
I
P
s 2
$ 1 2
It
Fig.6(a) :Catalyst A (washcoat: yAl2O3)
Fig.6(b) :Catalyst B (washcoat: yAl2O3 + CeO2 )
1
0.96
I
0.98
I.IM
1.02
C
1
1.w
EQUI V ALEA'CI; h'A T I 0
Fig.6(c) :Catalyst C (washcoat: yAl2O3 + Ce02
+ promotor)
Figure 6 Percent NO converted into N 2 0 at 450°C as a function of the air fuel equivalence ratio for catalysts aged on-an engine bench for different lengths of time. ( * 10 min, A I hr, 6hr, 40 hr and + 200 hr).
I
434
Figure 7- Vehicle NOx and N 2 0 emissions without any catalyst and with 8 different new or engine bench aged catalysts.
435
Figure 7 shows the results obtained for NOx and for N20. The amount of N 2 0 emitted does not vary very significantly with the catalyst type (no great differences in formulations were noted here, however). In every case catalyst aging involves an increase in N 2 0 emissions of between 2 and 4.5 fold depending on the driving cycle used. NOx emissions by this vehicle with a new catalyst were 6.6 to 63 times lower than without a catalyst and 3.3 to 12 times lower with an aged catalyst (Tables 2 and 3). Concerning N20, emissions were 3.8 to 4.9 times higher with a new catalyst than without and 10 to nearly 17 times higher with an aged catalyst. The fact that N 2 0 is produced during catalyst light-off explains why rather large amounts of N 2 0 were formed during these low speed tests. It also explains why, after aging, more N 2 0 is observed due to the shift in the lightoff temperatures towards higher values.
NOx without Cat. NOx with Cat. N7 - 0 without Cat. N9O with Cat.
ECE Urban Cold Start 6.6
ECE Urban Hot Start 12.3
Extra Urban Hot Start 7.6
4.87
3.78
4.2
TABLE3 Three-way catalyst effect on NOx and N2O emissions. Average for 8 aged catalysts
I NOx without Cat. NOx with Cat. &O - without Cat. N20 with Cat.
I
ECE Urban Cold Start 3.36 10
I
ECE Urban Hot Start 12.3 16.8
I
Extra Urban Hot Start 7.6 16
1
436
4
- CONCLUSIONS
Pt Rh three-way catalysts convert a small portion of the NO present in the exhaust from spark ignition engines into N 2 0 . With a stoichiometric airfuel ratio the conversion varies between 0 and around 10% depending on the catalyst working temperature. NO formation is maximum at a temperature very close to the catalyst light-off temperature and then decreases progressively up to around 500°C where it becomes nearly nil. Aging of the catalyst, making the catalyst light-off increase, shifts the peak of N 2 0 formation towards higher temperatures. The study of N 2 0 formation as a function of the air-fuel ratio below or above stoichiometry has shown that the quantities are minimum at stoichiometry and tend to increase on each side, but more on the rich side that on the lean side. Aging of the catalysts increases N 2 0 formation whatever the air-fuel ratio may be. Since N 2 0 is mainly produced during catalyst light-off, vehicle N 2 0 emissions are increased less with a new catalyst because it remains for a shorter period of time in this temperature range. The shift in the light-off temperature to higher values, which results from aging, makes the catalyst work more at temperatures where N 2 0 is formed and thus increases vehicles N 2 0 emissions. It is necessary however to keep in mind that N 2 0 coming from ground transportation is the result of : (1) direct N 2 0 emissions by engines, (2) a posteriori N 2 0 formation from NO in the atmosphere and/or in the soil from acid rain; and that, whereas three-way catalysts have a negative effect on point 1 (direct emission), they have a clearly positive effect on point 2 (indirect N 2 0 formation) by lowering NO emissions.
References K.P. Bowman, "GlobalTrends in Total Ozone", Science, 239,48, 1988. L.B. Callis and M. Natarajan, "Ozoneand Nitrogen Dioxide Changes in the Stratosphere During 1979-1984",Nature, 323, 772, 1986. R.F. Weiss, J. Geophys. Res., 86, 7185, 1981. V. Ramanathan, R.J. Cicerone, H.H. Singh and J.T. Kiel, J. Geophys. Res. 90, 5547, 1985. LNETI/EPA/IFF' European Workshop on the Emission of Nitrous Oxide, Lisbon, 6-8 June 1990. M. F'rigent and G. de Soete, "NitrousOxide N 2 0 in Engine Exhaust Gases; a First Appraisal of Catalyst Impact", SAE paper No 890492.
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
437
FLOW, HEAT, AND MASS TRANSFER IN A MONOLITHIC CATALYTIC CONVERTER
by D. Schweich and J.P. Leclerc LSGC-CNRS ENSIC, I rue Grandville, BP 451, F-54001 Nancy Cedex, France
1 Introduction Monolithic converters are the most fascinating and complicated chemical reactors. They are most often operated in the transient state with respect to composition, flow rate, temperature, etc. There are several competing reactions, and these reactions are competing with the heat and mass transfer processes. Similarly to the catalytic reactors of the chemical industry, the user wants high conversions of the reactants. However, unlike industrial reactors, the user has not a lot of operating variables such as the space time, the feed and coolant temperatures, etc. The only available "operating parameter" is a clear scientific understanding of what is going on, in order to properly design the reactor. This paper aims at describing the present status of this scientific knowledge which is more limited than it can be believed from the literature published since the early 70s.
2 The flow pattern in a monolith Figure 1 schematically illustrates a typical monolithic converter The flow regime in the elements of constant cross-section (exhaust pipes and channels of the monolith) is determined by the Reynolds number, Re, defined bY p urn Dh Re = (1) rl where p is the density of the fluid, um the mean fluid velocity, rl the viscosity, and Dh the hydraulic diameter. When Re is smaller than about 2500 flow is laminar, and turbulent otherwise. In the channels of a typical monolith the Reynolds number is smaller than 1000 and flow is laminar. On the contrary, flow is turbulent in the inledoutlet pipes most of the time.
438
Divergent
Convergent
Figure 1: Schematic of a monolithic catalytic converter. The change of flow regime in the divergent and convergent cones results in a non uniform velocity distribution throughout the monolith cross-section. Consequently, the contact times between the pollutant and the catalyst are different from channel to channel, and one may wonder how this contact time distribution is reflected into the overall efficiency of the converter. The velocity distribution also affects the mass and heat transfer processes from the bulk gas phase to the catalytic wash-coat (See section 3 below) and catalyst activity (see Figure 2, below). Moreover, since the local temperature depends on axial convection, radial conduction, heat loss at the converter boundary, and heat generation by the reactions, it is concluded that the temperature field is intimately coupled with the velocity field. Eventually, it is concluded that predicting and understanding how the converter design is reflected in the pollutant removal efficiency is an overwehlming task because of the lack of an accurate knowledge of the flow pattern. Most of the flow studies have been devoted to cylindrical monoliths. Although non cylindrcal (oval) monoliths are now prefered, these studies suggest the main key factors which govern the flow pattern. The results can be summarized as follows (Lemme and Givens, 1974; Howitt and Sekella, 1974): -The higher the flow-rate, the more distorted the radial velocity profile. -The pipes, muffler, and convergent located downstream the monolith do not affect the flow pattern. -The diameter of the inlet pipe governs the radial velocity distribution. The smaller the inlet pipe diameter, the more non uniform the velocity distribution. The inlet pipe is responsible for a maximum velocity (jet flow) close to the center of the monolith. -A suitable flow deflector located in the divergent can make the velocity distribution more uniform, the steady-state conversion higher, and it reduces catalyst detorioration. Conversely, pressure drop generally increases, and light-off is delayed with respect to a monolith with no deflector. In the case of an oval monolith, these results are confirmed by Wendland and Matthes (1986) and our results (Leclerc et al., 1989a, b).
439
Figure 2 illustrates the sentivity of the conversion of CO and hydrocarbons to the flow pattern by comparing a standard converter with a converter equiped with a conical deflector in the divergent.
0.9 c 0 e,
0.6
6
I
I
I
I
100
200
300
400
time (hr)
Figure 2: Conversion of CO (circles) and h drocarbons (squares) versus time on stream. Full ots: conical deflector in the diver ent. Open dots: no deflector. Resultsfrom Howitt an2 Sekella (1974).
2
Figure 2 suggests that the high fluid velocity in the center of the monolith without the flow deflector is responsible for lower conversions and faster deactivation. These results clearly show that experiments are very sensistive to the flow pattern. Consequently, laboratory experiments must be performed in carefully designed reactors, especially when using very small amounts of crushed catalyst. It is advisable to dilute the catalyst with an inert carrier (crushed substrate) to get a more uniform bed and to avoid channeling. Upstream the catalyst bed, a few centimeters of inert carrier should be used for establishing a uniform flow.
3 Heat and mass transfer from the bulk gas phase to the wash-coat Because of the laminar flow regime in the channels, reactants must radially diffuse toward the wasch-coat. Similarly, heat transfer from the washcoat to the bulk gas phase mainly occurs by radial conduction through the gas. Chemical reactions in the wash-coat and radial mass and heat transfer are thus competing phenomena, and it is of prime importance to know which phenomenon is governing the converter efficiency.
440 3.1 The axial-convection-radial-conductionDroblem
For the sake of simplicity we will presently only deal with the steadystate radial conduction problem in a circular channel. The local heat balance equation is
JTe 5 = 0 , --0
a5
-
where z and 5 are the axial and radial coordinates respectively, e the thickness of the catalytic wash-coat (excluding the monolithic substrate), hm the thermal conductivity, AH the reaction enthalpy, and F the apparent chemical reaction rate per unit volume of wash-coat. Equations (2) defines a generalized Graetz problem, and unfortunately, it has no simple solution which can help to understand the true nature of the competition between the transport and reaction processes. Moreover, only the average temperature of the fluid at a given location z is required, whereas the radial temperature profile is overinformative. Thus, one must find a simpler and more efficient approach than the Graetz formulation. 3.2 The film model for heat transfer
The film model is a simplified form of the previous problem based on the following assumptions: -The fluid velocity, temperature and concentrations are uniform throughout the channel cross-section. -The mass and heat transfer resistances are confined to a thin film close to the channel wall surface. -The heat and mass transfer processes are accounted for by a linear rate law and the transfer coefficients are defined by matching the solution of the film and Graetz problems.
44 1
The heat balance is now
-
h (Te - Ts) = r (Cs, T,) e AH
where h is the heat transfer coefficient. The equivalence between the film and Graetz models is known for two situations: -Constant wall heat flux ( r e AH = constant). -Constant wall temperature (Ts = constant). The local heat transfer coefficient is obtained from the non dimensional Nusselt number, Nu, given by:
where f(z) is a decreasing positive function, and Num the limiting Nusselt number for a very long channel. f(z) is smaller than Nuoowhen
where Pr is the non dimensional Prandtl number defined by Pr =
G Am
Most of the time Pr is of the order of 0.7. In a standard monolith, made of 1x1 mm square channels and operated at Re = 500 (high value), the limiting Nusselt number is reached when z becomes greater than 18 mm. This suggests that a standard monolith about 15 cm long is mainly operated at the limiting Nusselt number, and one may assume that NU
Nuoo
(7)
The asymptotic Nusselt numbers are given in Table 1 according to the channel geometry, the wall condition, and the associated solution for the Graetz problem (Shah and London, 1978). Except for parallele plates, standard channel geometries are accounted for by a limiting Nusselt number of the order of magnitude of 3 to 4. Heck et al. (1974) experimentally confirmed these values with various monoliths.
442
Table I : Limiting Nusselt number Nu,.
Then, solving equations (3) with Nu = Nuoogives
for a constant wall temperature for a constant wall heat flux FQThe ratio z/l is often called the number of transfer units and it is abbreviated NTU. Using Nuoo= 3, one obtains 1 = 0.067 Dh Re
(9)
Using the constant wall temperature condition and equations (8a) and (9), the gas and the solid temperatures are identical within 1% when z/l = 5 , or z = 170 Dh with Re = 500. This is the standard length (about 15 cm) of a monolith made of square channels 1x1 mm. Although the chemical reactions in the wash-coat do not imply a constant wall temperature, the latter calculation suggests that heat transfer resistance in the gas phase of a channel is not negligible. The equivalence between the Graetz model and the film model with a suitable Nusselt number is only known for a restricted set of wall conditions (Shah and London, 1978). However the available results show that the Nusselt number is always of the order of 3. Consequently, it is generally assumed that the film model can be applied whatever the wall condition, and especially when chemical reactions takes place, although the wall temperature and the heat flux vary with the gas composition. The film model is also widely used for describing heat transfer in the transient state using the Nusselt numbers obtained from the steady state solution for the Graetz problem. This assumption seems acceptable and it greatly simplifies the models for transient states.
443
Although attractive, the film model as presented above fails to describe the establishing laminar regime at the entrance of the channels. Few theoretical considerations on that problem are available. The reader is referred to Shah and London (1974), Heck et al. (1974) and Wendland (1980). 3.3 The film model for mass transfer Similarly to the heat transfer problem, the mass transfer process can be described by the film model. The mass balance for a reactant in the gas phase is
where kD is the mass transfer coefficient. Equations (10) and (3) are formally identical. As a consequence, the results obtained for the heat transfer problem are applicable to the mass transfer problem using the conepondance given in Table 2. In particular, the limiting Sherwood numbers are given by Num of Table 1, and equations (8) are valid after substitution of Ce, Co, Sh, Sc, kD,
Table 2: Equivalence between the heat and mass transfer problems.
and FM for Te, TO,Nu, Pr, h, and FQ, where FM is the constant wall mass flux. Equation (9) remains unchanged. Dm is the average diffusion coefficient of the reactant in the mixture.
444
Assuming that a fast chemical reaction takes place in the wash-coat, the surface concentration, Cs, is close to zero, and it defines a constant wall concentration problem. About 95% of the reactant will be consumed when z/l = 3, or z = 100 Dh at Re = 500, or z = 10 cm for 1x1 mm square channels. This example shows that mass transfer limitations in the gas phase can be responsible for a limited efficiency even though the catalyst is highly active. This also suggests that mass transfer in the gas phase can be the rate determining step in the monolith as shown by Heck et al. (1976). Under such circumstances, improving the catalyst activity would be useless. The molar fluxes of each species through the film must obey the stoichiometric constraint. Consequently, only the species at the lowest concentration obeys the constraint Cs = 0 at the wall surface. From equation (lOa), it is seen that the reaction becomes first order with respect to the limiting reactant. Moreover, the rate constant being proportional to kD, the reaction is no more activated. A very low (or zero) activation energy and a first order reaction with respect to a limiting reactant must suggest a strong mass transfer limitation. 3.4 The film model for heat and mass transfer and chemical reactions The full problem is described by equations (3) and (10). When the reaction is strongly limited by mass transfer, it has been shwon in the previous section, that the corresponding Graetz problem is defined by a constant wall concentration condition. The proper wall condition for the heat transfer problem is probably different, and Hegedus (1975) supposed that the limiting Sherwood and Nusselt numbers could be different. Nevertheless, for the sake of simplicity, we will suppose they are identical: Sh = NU = Nuoo
(11)
Simple estimates of mass and heat transfer resistance can be made assuming there is no heat loss (adiabatic monolith). This assumption is acceptable for the innermost channels. Equations (3b) and (lob) yield
Making use of equation (1l), one obtains
445
where Je is the adiabatic reaction temperature at complete conversion calculated with the local concentration in the channel. Remark that Je decreases as the the conversion increases. p Cp/Ce is the specific heat of the fluid per mol of reactant. For a gas containing carbon monoxide at a molar fraction Xco about Te = 400"C, one finds AH = -2.8 105 j.mo1-1, Je
9500 XCO K,
C
c,
= 30 / Xco j.mol-1.K-1
Cte = 14 XCO
Consequently
With an active catalyst which makes the mass transfer limitation to be severe, Cs << Ce and
The temperature difference can be as high as 190C when the gas contains 2% CO. A more detailed calculation would require solving equations (3) and (10) for Ce and Te. Hegedus (1975) gave the solution for several competing reactions assuming that the fluid velocity and mass transfer coefficients are independent of the temperature, that the chemical reactions are not the limiting processes, and that the Nusselt and Sherwood numbers are different. For a single reaction Hegedus solution becomes
( %:)
C,(z) = Co exp - -
where Jo is the adiabatic temperature rise calculated with the inlet concentration of reactant. The term in square brackets in equations (17c, d)
446
suggests that the solid temperature can be different from the adiabatic reaction temperature rise Jo, except when assuming Sh = Nu (See equation (1 1)) and Pr = Sc. If the latter assumptions are realistic for carbon monoxide and hydrocarbons, they are not for hydrogen for which Sh Pr/(Nu Sc) = 3. A gas containing 2%CO, also contains about 0.7% H2 because of the water gas shift reaction. Using (17d), the temperature rise due to H2 can reach 17OoC,which must be added to the temperature rise due to CO and results in 360°C! Hegedus (1975) has experimentally illustrated this phenomenon. As a general conclusion, the main rate determining step in a monolith operated above the light-off temperature can be heat and mass transfer in the gas phase. However, below the light-off temperature (cold-start) the chemical reactions are the rate determining step, or they compete with the transfer processes.
4 Heat and mass transfer and chemical reaction in the wash-coat Similarly to the heat and mass transfer resistances in the gas phase, there are possible resistances in the catalytic wash-coat. However, the problem is more complicated than previously because the transfer resistances are intimately coupled with the chemical reactions. For an exothermic reaction, the competing processes are responsible for concentration and temprature profiles which are schematically illustrated in Figure 3.
t
r a
et
I
3 6 cc
Ce
I
Figure 3: Schematic of the concentration and temperature gradients in the wash-coat.
447
The apparent reaction rate r i s given by e - 1 r = - jr(C,T) dc e 0 Two main regimes can be defined: -The chemical regime, when the temperature and concentration are uniform and equal to Ts and Cs respectively. In this case
-The diffusion regime, when the temperature and/or the concentration are not uniform. In this case -
r
+ r(Cs7Ts)
(19b)
To relate the apparent and intrisic kinetic rate it is convenient to introduce the effectiveness factor
In the chemical regime, qs = 1, whereas qs differs from unity in the diffusion regime. Our main goal will now be to predict the effectiveness factor. 4.1 Mass transfer We will first assume that the wash-coat temperature is radially uniform (Ts = Tc = T(6)). Assuming that there is no convection inside the wash-coat, the mass balance equation for the reactant is
dC 6 = 0, -0 dc -
5 = e, C = Ce (21c) where r(C) is the intrinsic kinetic rate per unit volume of wash-coat, and De an effective diffusivity given by
448
p
being the porosity of the wash-coat, z the tortuosity factor (generally 1
For a n-th order reaction (n > 0) the regime is deduced from the Thiele modulus, Qs, (Froment and Bischoff, 1979; Villermaux, 1985)
-When Qs << 1, then qs = 1, and the chemical regime prevails. -When Q s >> 1, then qs = and the diffusion regime prevails. It is concluded that the higher the reaction rate, or the thicker the washcoat, the stronger the internal diffusional resistance. When the intrinsic kinetic rate, r, is not available, one must use the Weisz criterion, ($;, to detect the rate determining step
Comparing equations (23) and (24) and taking into account the rules given above, one concludes: -When Qs' << 1, then rls = 1, and the chemical regime prevails. -When Qs' >> 1, then QS = q s = l/Qs', and the diffusion regime prevails. Using equation (lob), the Weisz criterion becomes (PSI,
Assuming t = 2, b = 0.5, e = 10 mm, Dh = 1 mm, and Sh = 3, equation (25a) becomes
fs' = 0.12
Ce - Cs CS
As soon as the surface concentration, Cs, falls below 0.1 Ce, the internal diffusion regime should prevail (see below). This situation is easily met with
449
any catalysts. Typical simulation results obtained by Leclerc et al. (1990) are given in Table 3.
Table 3: Weisz criterion for CO oxidation in the leading part of the monolith, and assuming : r = 2 , p = 0.5,e = 10 p n , Dh = I mm, and Sh = 3. Kinetic rate law takenffom Chen and Oh (1988).
ce - cs
Oxygen rich mixture
Stoichiometnc mixture
178
3.2
21
0.4
CS
4 S'
Although these results indicate either a strong or a weak internal mass transfer limitation, they must be carefully considered for two main reasons: -The critical Weisz or Thiele modulus of unity which defines the threshold between the chemical and diffusional regimes is acceptable for a n-th order reaction with a positive order close to unity. Oxidation of carbon monoxide and some hydrocarbons are known to obey a LangmuirHinshelwwod kinetic law which leads to a negative apparent order at high reactant concentration. -It is not obvious that diffusion alone is responsible for the mass transfer process in a wash-coat layer as thin as a few tens of micrometers. We will see in the next two sections that these two problems make the internal mass transfer limitation much questionnable. 4.1 .I The generalized effectiveness factor
Froment and Bischoff (1979) have given a general method for estimating the effectiveness factor for an arbitrary rate law under isothermal conditions. When the intrinsic rate law is
and assuming the reaction to be irreversible, the Weisz criterion must be compared with the following threshold
450
For a n-th order reaction (g(C) = Cn) the above threshold is 2 @s,lim= n+l which generally is of the order of unity, although it diverges when n = -1. Assuming
and using equation (27a), one obtains
When KCs
> 1, one finds:
@s,lim+ 2 Ln(K C,)
(28c)
The latter threshold can be much larger than unity suggesting that the Weisz modulus given in the first column of Table 3 does not necessarily imply an internal diffusion limitation. Let us finally mention that when the reaction is strongly inhibited by the products, then @s,limcan be much lower than unity. Further details on this subject can be found in Froment and Bischoff (1979). When the reaction rate obeys equation (28a), internal diffusion limitations becomes beneficial at high reactant concentration because of the negative apparent order. Any diffusion limitation wili reduce the reactant concentration and thus favors the reaction.
4.1.2 Convective transport in the wash-coat Because of the finite permeability of the wash-coat, the bulk gas flow can induce convection inside the porous material. In standard catalyst pellets (a few mm or cm in diameter) this effect is negligible except for large pore catalysts (Rodrigues et al., 1982; Cresswell, 1985). With a very thin catalytic layer, internal convection cannot be excluded even if the pores are narrow. Moreover, the roughness of the wash-coat surface can enhance the internal convection.
45 1
Although there is no available literature concerning convection in thin catalytic layers, it is obvious that the convection process would reduce the mass transfer resistance. This brief analysis shows that the internal mass transfer problem remains unsolved. Only experiments with wash-coats of various thickness would help to get further insight in the problem. 4.2 Heat transfer In the previous section T, = T(<) = Ts was assumed. When this assumption is relaxed, estimating the effectiveness factor becomes a challenge. Fortunately, the thermal conductivity of the porous solid is high enough to make the temperature almost uniform. Let us illustrate this property with a simple example. The maximum internal temperature deviation is given by (Villermaux, 1985; Froment and Bischoff, 1979)
where he is the effective thermal conducivity of the alumina wash-coat. Typical values for 1% carbon monoxide at Ts = lOOOK are
Consequently (T, - Ts)max I 9K. When E/R = 15000K, neglecting the temperature gradient in the wash-coat is acceptable. As a general conclusion, one can assume that the temperature is radially uniform in the wash-coat, whereas it is not clearly known whether the internal mass transfer limitation is competing with the chemical reaction. 5 Estimating the rate determining step at laboratory scale
Catalyst studies at laboratory scale are often undertaken with a crushed catalyst. The competition between the reaction and the heat and mass transfer processes is thus different from the monolith case. Since the study of the internal mass transfer process is not conclusive, and since internal temperature gradient are generally negligible, we will only focus on the film mass and heat transfer phenomena. We will assume that the laboratory catalyst is made of spherical pellets of diameter dp. If the pellets are not spherical, dp is well approximated by 6 times the ratio Volume/external surface area.
452
In a fixed bed, the Sherwood and Nusselt numbers are given by the following empirical correlations
Comparing a catalyst sieved at d, = 100 pm and a monolith channel 1x1 mm, a channel Reynolds number Re of 500 becomes a particle Reynolds number Re' of 50. Consequently, one obtains
S h = Nu' = 15
when
Sc = Pr = 1
A mass balance over the external film around the catalyst particle gives
Rearranging equation (lob) and using equation (31) yields (CeLCs)
w
=(CeLCsk
ixed bed
onolith
Sh P d 2 w = 6 Sh'eDh
w is generally smaller than unity. With Sh = 3, Sh' = 15, dp = lOOpm, e = 10
pm, and Dh = 1 mm, w = 0.033. This suggests that mass transfer limitations in the monolith can be overlooked when laboratory experiments are made with a crushed catalyst. This could explain why catalysts which have different activities at laboratory scale, reveal to be similar in the monolith. Let us finally remark that reducing dp reduces the mass transfer limitation. Similar calculations can be made for heat transfer to obtain
453
r s i e T e ) f i x e d bed
-
r d, (-AH) 6 h Te
(33)
-
When r is experimentally measured, equations (30), (31) and (33) allows one to estimate the film resistances. Unfortunately it is not easy to measure the apparent reaction rate except in a differential reactor. A simple solution is to take advantage of the sensistivity of mass and heat transfer processes to d, and to the gas velocity. Mass and heat transfer resistances are not negligible when the conversion decreases when d, or increases at a constant ratio of the mass of catalyst to the gas flow rate (i.e., constant GHSV).
6 Competition between the reactions
-
The selectivity problem
The definitions of selectivities and yield are often confusing. Thus, we will first recall the basic definitions (Villermaux, 1985) on the following example 1 CO + 2 0 2 -+C02, ri (34d
CO + NO -+
1 C 0 2 + 3N2,
r2
assuming the following intrinsic rate laws
Yields are defined with respect to two species: a reactant A and a desired product P. If there were no undesired reactions consuming the reactant, the overall chemical transformation could be summarized by
A+
.... + v P + ....
(35)
v is the maximal stoichiometric number and is used to normalize the yields and selectivities. Inspection of the stoichiometric scheme (34) immediately gives the maximum stoichiometric numbers summarized in Table 4. Three yields can be defined
454
Overall yield Integral yield
F @P/A= V(FAOp-F ~ )
Differential yield
@'P/A= RP -VRA
where Fi and Ri are the molar flow rate and production rate (>0 or
r2 +
(37)
r2
Table 4 Some maximum stoichiometric numbers corresponding to reactions (34) Reactant
Desired product
co
co2 co2
NO N2
Stoichiometric number 1 1 1/2
co
Selectivities compare two products Pi and P2. Two selectivities can be defined Integral selectivity
Spip2 =
Differential selectivity
FPl v2
6
RPl v2
S'pi/p2 = ~ p 2
where Vi is the maximum stoichiometric number of Pi. Differential yields allows one to state two simple selectivity rules: -Increasing the concentration of a reactant favors the reaction which has the highest order with respect to this reactant. -Increasing the temperature favors the reaction with the highest activation energy.
455
These rules are immediately checked with
1
(39)
which is obtained from (34b) and (37). When n > n', increasing carbon monoxide concentration decreases $'N2/CO and thus favors the first reaction. The rules above are useful for ranking the reaction orders and activation energies at laboratory scale. Assuming there is no internal mass transfer limitation, equation (39) involve the gas/solid interfacial concentrations. When there is no film limitation, these concentrations equal the bulk concentrations. When there is a strong film mass transfer limitation, one of the surface concentration goes to zero. Consequently, if CO is the limiting reactant and n > n', a strong film limitation favors the second reaction. Let us remark that increasing the surface temperature, Ts, makes the film resistance more important. As a result the first reaction can be favored even if its activation energy is smaller than the activation energy for NO reduction. This phenomenon could explain why the conversion of some reactants decreases when the temperature increases, whereas high conversions are observed when the individual reactions are studied independently. As a conclusion, product distribution is affected by the nature of the intrinsic kinetic law and by possible heat and mass transfer limitations.
7 Multiple steady-states and transient behavior Monolithic converters are virtually always in the transient state (coldstart, speed-up, slow-down, change of gear,...). This implies that it is much difficult to give general rules concerning the rate detemining step which can change according to the flow rate, and the composition and temperature of the feedstream. The problem becomes even more complicated because of possible multiple steady-states due to various phenomena. We will essentially deal with the latter problem. 7.1 Multiple steady-states resulting from the rate law We will first consider the steady-state oxidation of CO at uniform temperature with no heat and mass transfer limitation. Voltz et al. (1973) suggested that the kinetic rate law is
456 Assuming that there is an excess of oxygen, the rate law becomes with k = k' Co2 = constant For the sake of simplicity subscript CO will now be omitted. Assuming that the reaction takes place in a continuous reactor of uniform composition (differential reactor), the mass balance for CO is r(C) =-
co- c to
where to is the space time defined by the ratio of the volume of catalyst to the volumetric flow rate. Figure 4 illustrates equation (40b) as a continuous curve, and (41) as dotted lines of the same slope -l&, and corresponding to different inlet concentrations Co, C'o, and C"0. The intersection between the rate curve and the mass balance lines defines the steady-state solutions. When the feed concentration is low enough (Co) there is a single solution illustrated by point A. For a high feed concentration (C"0) there is again a single solution C". Let us remark that the conversion at point C" is very low, owing to the strong inhibition (apparent order -1) at high reactant concentration. For an intermediate range of feed concentration, about C'o, there are three possible solutions. It can be shown that points A' and C are the only stable steady-state solutions. However, it remains to know which of the two solutions is actually observed. The solution to this problem can be obtained by intuitive arguments and knowing the past history of the system.
Fi ure 4 : Illustration of the rate law (continuous curve) and the mass bala.nce equation (straight lines) f o r different inlet concentrations.
457
We will assume that any process continuously changes as long as it can. Starting with a reactor free of reactant (Co = 0), let us slowly increase the feed concentration. The steady-state solution will slowly go from point 0 to A and will finally reaches A which is thus the real solution. Starting from a reactant rich mixture (C"0) and decreasing the feed concentration down to C'o, point C' will be found to be the solution. When the mass balance line becomes tangent to the rate curve making A"' and B"' to be identical, any further increase of Co is responsible for an instantaneous jump from A " to C"'. There is thus a discontinuous behavior of the system. The same reasonning applies when Co decreases, however the location of the tangent mass balance line is different. Figure 5 illustrates the resulting outlet concentration as a function of Co. The multiple steady-states described above are typical of negative order reaction rates. They can be observed within a limited range of feed concentration and for large enough space times. Eigenberger (1978a, b) gave a detailed study of this phenomena. Sheintuch and Schmitz (1977) gave a detailed theoretical study of CO oxidation, and of possible stationary oscillations. C
/
Figure 5:. Outlet concentration versus feed concentration when there are multiple steadystates
CO 7.2 Multiule steady-states resulting from the heat generation process We will now consider a first order reaction in an adiabatic CSTR The mass balance equation (41) still holds, although one must remember that r(C) depends on temperature owing to Arrhenius law. The heat balance equation is
Combining equations (41) and (42) gives
T = TO+ JOX,
X=-
co - c CO
(43)
458
Figure 6 illustrates the mass balance equation (41) as a continuous curve, and equation (43) as a straight line. As in the previous section, multiple steadystates are possible for a limited range of feed temperature To. When the reactor is not adiabatic, figure 6 is still qualitatively valid. However, the easier the heat transfer process, the steeper the straight line. In an isothermal reactor (vertical straight line) there is a single steady-state point. A catalyst pellet or a portion of the catalytic wash-coat is equivalent to the differential reactor except that they are fed by different processes, namely convection in the differential reactor, and diffusion in the catalyst. This means that the catalyst acts as a local differential reactor in the monolith. Consequently the feed temperature To of the differential reactor corresponds to the external temperature Te, and the outlet temperature T corresponds to the surface temperature Ts. The light-off curve X vs. Te of a catalyst is thus the curve X vs. To of a differential reactor. In case of steady-state multiplicity, Figure 6 suggests that the light-off curve is given by Figure 7 which shows that the light-off temperature is not uniquely defined.
x'
c,
, C
Figure 6: Possible steady-states f o r a first order reaction in an adiabatic differential reactor. For an intermediate range of temperature there are several possible steadystates. .At -T'o there is a sin le steadystate. oint B is unstable.
:#;
f
B
LX
--------__
Figure 7: Schematic of a light-off curve in case of stead state multiplicity. f n a narrow range of external temperature there are t w o possible steady -stat :'. a higher light-off temperature is obtained when increasing the gas temperature Te.
:?;
fi Te
459
7.3 Consequences in real converters and laboratory experiments There is no clear experimental evidence of multiple steady-states. However model results (Heck et al., 1976; Young and Finlayson, 1976; Oh and Cavendish, 1982) based on acceptable average data show that they can be observed, especially in 0 2 rich mixtures. Whatever the number of steady-states, the light-off phenomenon is responsible for strong temperature gradients. Figure 8 summarizes the process assuming the reactor is initially at the temperature of the feed gas. When the feed temperature is low, the chemical reaction is the controling step. The catalyst works at low temperature and the conversion is low. If the feed temperature is very low, then there can be no light-off. For an intermediate range of feed temperature, light-off occurs in the interior of the reactor. When there are two possible steady-states at the light-off point, the catalyst jumps from its cold state to the hot state (See Figure 6). When there is a single steady-state, the jump is more gradual. Downstream the light-off point, the chemical reaction can be so fast that film mass transfer controls, and the solid temperature is To plus the adiabatic temperature Jo (except when ShPr/(NuSc) # 1, see section 3.3). The bulk gas concentration and temperature vary according to the heat and mass transfer coefficients. When the feed temperature is high enough, light-off occurs at the inlet and all the reactor can be under film transfer control. 7.4 Start-uD of the converter The light-off of the converter can obey various senarios. First, Figure 8 suggests that light-off occurs downstream and then moves backward when the inlet temperature increases. However, at cold-start the intial temperature of the converter is much lower than the feed temperature. Consequently, the solid cools the gas and thus prevents any downstream light-off. When the inlet temperature is high enough, light-off occurs close to the inlet and then the hot zone progressively spreads downstream. For a limited range of feed temperature, light-off can occur in the reactor and then spreads upstream and downstream. A detailed description of the monolith behavior at cold-start is given by Chen et al. (1988) and Leclerc (1990).
7.5 Run-awav conditions Thermal run-away is responsible for monolith destruction. There are no clear explanations for thermal runaway other than qualitative. Misfiring of the engine can be responsible for high concentrations of reactants which result in a high adiabatic temperature rise which can be sufficient to reach the fusion temperature of the substrate (Morgan et al. 1973).
460
A H2 rich feed can make the temperature rise to be much larger than the adiabatic temperature rise, as it has been shown in section 3.3. Slow-down of the engine results in a decrease of the exhaust gas temperature and a higher hydrocarbons content. The high concentration of hydrocarbons makes the adiabatic temperature rise to be large. The heat previously stored in the exhaust tube and in the converter, can warm up the exhaust gas for a few seconds, and thus the feed temperature can be as high as before the slow-down. When Jo is of the order of 500K and the feed temperature about 1000K, the monolith melts at the light-off point. Light-off
Cold
4
Hot
-Chemical reaction is controling
Figure 8: Schematic of temperature and concentration pro iles in a g l u g L o w reactor. ownstream the li ht-offpoint the soliaB temperature can reach the adiabatic temperature and the reaction can be strongly limited by the film mass transfer resistance.
Film diffusion is controling in general
8 Conclusions Present knowledge allows one to ascertain that heat and mass transfer in the gas phase are rate controling processes above the light-off temperature, at least in an oxygen rich mixture and with a fresh catalyst. However, the intrinsic kinetics can become controling when the catalyst is deactivated. This suggests that improving the activity of a fresh catalyst above the light-off temperature is probably secondary, whereas improving its resistance to deactivation processes is of prime importance. Chemical kinetics is controling
46 1
at cold-start, and thus new catalysts with low light-off temperatures would be welcomed. Under stoichiometric conditions, the competition between the reactions and the transfer processes is much less understood. It would be much interesting to get further insight in the internal diffusion process. If internal diffusion is really a limiting process, a thick wash-coat would be advantageous for negative order reactions. Let us also mention that a thick wash-coat will prevent poison deposit on the innermost catalyst sites. Reliable kinetic data (especially for NO reduction) would be welcomed by modelers. Kinetic laws would allow one to numerically investigate the monolith behavior under various conditions at little cost, and to get a deeper understanding of what is going on. The structure of the kinetic laws is probably the same for many catalysts which are mainly made of the same ingredients. Only the rate and adsorption constants can vary from one catalyst to another. Thus, it would be advisable to obtain reliable kinetic laws for a standard catalyst which could serve as a reference for other catalysts. Little is known concerning transient states at the local level. Much work remains to be done on the effect of pulsating flow on the transfer processes in the gas phase, on 0 2 storage, and on heterogeneous catalysis in the transient state. Understanding and quantifying these processes will help to select better catalysts and operating conditions, especially those avoiding thermal run-away. No doubt that nowadays specialists of catalysis and chemical reactor engineering must work together. This is the price to pay for an efficient development of new converters. Finally, the authors think that scientists' imagination should work more. We will only suggest two possible research areas, one being presently under work, the other being more provocative. First, alternative and less expensive catalysts (non noble metals) would be much beneficial. Second, is it possible to regenerate spent catalysts, for instance when changing oil? What should be this regenerable catalyst? What should be the regenerating process? A lot of questions for the catalyst and reactor specialists.
List of symbols C CO
c,
c, cs
CP D De Dh Dm dP
Concentration (m01.m-3) Inlet concentration (m01.m-3) Concentration at the wash-coat-monolith boundary (m0l.m-3) Concentration in the gas phase (m0l.m-3) Concentration in the gas phase at the wash-coat surface (m0l.m-3) Specific heat (j.kg-1.K-1) Average diffusivity in the homogeneous re fluid (m2.s-1) Effective diffusivity in the wash-coat (mg-1) Hydraulic diameter, 4(cross~ectionalarea)/perimeter (m) Molecular diffusivity (m2.s-1) Equivalent diameter of a catalyst pellet (m)
462 e Fi FiO g(C) h Je JO
K
kD
k"
1 Nu, Nu, Nu'
n Pr Ri Re, Re' L
r
SrIA P/A
s
sc Sh, Sh'
urn
X
YPIA Xi Z
;;" Pi
AH
Thickness of the wash-coat (m) Molar flow rate (mo1.s-1) Inlet molar flow rate (mo1.s-1) Arbitrary function of the concentration Heat transfer coefficient (watt.m-2.s-1) Adiabatic temperature rise calculated with Ce (K) Adiabatic temperature rise calculated with Q (K) Adsorption constant (m3.mol-1 Mass transfer coefficient (m.s- ) Rate constant (various units) Defined by equation (8a) (m) Nusselt number in a channel, limiting Nusselt number Nusselt number in a fixed bed Reaction order Prandtl number Production rate of species i (mo1.s-1.m-3) Reynolds number in a channel in a fixed bed Intrinsic reaction rate (mo1.m-3 of wash-c0at.s-1 Apparent reaction rate (m0l.m-3 of wash-coats- ) Integral selectivity Differential selectivity Schmidt number Sherwood number in a channel, in a fixed bed Temperature (K) Inlet temperature (K) Bulk gas temperature (K) Temperature at the wash-coat substrate boundary(K) Temperature at the gas-solid interface (K) Space time (s) fluid velocity (m.s-1) Conversion Overall yield Mol fraction of species i axial coordinate (m) External thermicity criterion defined by equation (12) Internal porosity Defined by equation (29) Reaction enthalpy c.mo1-1) Thiele modulus threshold for the Weisz modulus Integral yield Differential yield Weisz modulus Dynamic viscosit (Pa.s) Effectiveness fac& Effective thermal conductivity of the wash-coat (j.rn-l.s-l.K-l) Thermal conductivity of the gas c.m-1.s-1.K-1) Maximum stoichiometric number Gas density (kg.m-3) Tortuosity factor Radial coordinate in the channel(m) Radial coordinate in the wash-coat (m)
1'
1
463
Literature cited: Chen D.K.S, S.E. Oh, E.J. Bissett , and D.L. van Ostrom: A three dimensional model for the analysis of transient thermal and conversion characteristics of monolithic catalytic converters. SAE paper 880282, Intern. Congress and Exposition, Detroit, Mich., 1988. Cresswell D.L.: Intraparticle convection: Its measurement and effect on catalyst activity and selectivity. Appl. Cat., 15, 103, 1985. Eigenberger G.: Kinetic instability in heterogeneously catalyzed reactions. I. Chem. Eng. Sci., 33, 1255, 1978a; 11. Chem. Eng. Sci., 33, 1263, 1978b. Froment G.F., and K.B. Bischoff Chemical reactor analysis and design. J. Wiley, New-York, 1979. Heck R.H., J. Wei, and J.R. Katzer: The transient response of a monolithic catalyst support. Adv. Chem. Ser., 133, 34, 1974. Heck R.H., J. Wei, and J.R. Katzer: Mathematical modeling of monolithic catalytic catalysts. AIChE J., 22, 477, 1976. Hegedus L.: Temperature excursions in catalytic monoliths. AIChE J., 21, 849, 1975. Howitt J.S., and T.C. Sekella: Flow effects in monolithic honeycomb automotive catalytic converters. SAE paper 740244, Automotive Engineering Congress, Detroit, Mich., 1974. Leclerc J.P., D. Schweich, and J. Villermaux: Transfert de chgaleur en rCgime transitoire dans un monolithe destinC i 1'Cpuration des gaz dCchappement automobile. S I T 89 Congress, "Thermique et GCnie des Proc6dCs", Nancy, 1989a. Leclerc J.P., D. Schweich, and J. Villermaux: Hydrodynamique et transfert de chaleur dans un monolithe destinC i 1'Cpuration des gaz dkchappement automobile. Symposium "GCnie des procCdCs", Toulouse, 1989. RCcents progrks en gCnie des procCdCs, Lavoisier Ed., 3(8a), 518-524, 1989b. Leclerc J.P., D. Schweich, and J. Villermaux: A new theoretical approach to catalytic converters. 2nd International Congress on Catalysis and Automotive Pollution Control, Brussels, 10-13 Sept., 1990. Lemme C.D., and W.R. Givens: Flow through catalytic converters - An analytical and experimental treatment. SAE paper 740243, Automotive Engineering Congress, Detroit, Mich., 1974. Morgan CR, D.W Carlson, and S.E. Voltz: Thermal response and emission breakthrough of platinum monolithic catalytic converters. SAE Paper 730569, 1973. Oh S.H., and J.C. Cavendish: Transients of monolithic catalytic converters; Response to step changes in feedsream temperature as related to controling automobile emissions. Ind. Eng. Chem. Prod. Res. Dev., 21, 29, 1982. Rodrigues A.E., B.J. Ahn, and A. Zoulalian: Intraparticle forced convection effect in catalyst diffusivity measurements and reactor design. AIChE J., 28,541, 1982. Shah R.K., and T.C London: Flow forced convection in ducts. Advances in heat transfer Laminar., Academic Press, New-York, 1978. Scheintuch M, and R.A Schmitz: Oscillations in catalytic reactions. Ctal. Rev. Sci. Eng., 15, 107, 1977. Villermaux J.: GCnie de la rCaction chimique. Conception et fonctionnement des rbacteurs. Tec & Doc, Lavoisier, Paris, 1985. Voltz S.E., C.R. Morgan, D. Liederman, and S.M. Jacob: Kinetic study of carbon monoxide oxidation on platinum catalysts. Ind. Eng. Chem. Prod. Res. Devel., 12, 295, 1973. Wendland D.W.: The segmented oxidizing monolith catalytic converter, theory and performance. J. Heat Transfer, 102, 194, 1980. Wendland D.W., and W.R. Matthes: Visualization of automotive catalytic converter internal flows. SAE paper 861554, International Fuel and Lubricants Meeting and Exposition, Philadephia, Pens., 1986. Young L.C., and B.A. Finlayson: Mathematical models of the monolith catalytic converter, Part 11, Application to automobile exhaust. AIChE J., 22, 343, 1976.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 199 1 Elsevier Science Publishers B.V., Amsterdam
465
A NEW THEORETICAL APPROACH TO CATALYTIC CONVERTERS
Leclerc J.P., Schweich D., Villermaux J .
Laboratoire des Sciences du Ge'nie Chimique-CNRS, ENSIC, INPL I , rue Grandville, BP 451,54001 Nancy Cedex, France
ARSTRACT A theoretical analysis to catalytic converter based on the heterogeneous cell model which accounts for flow distribution, gas-solid heat and mass transfer, chemical reactions and heat losses was developed. The converter efficiency is very sensitive to heat and mass transfer coefficients, especially to their variation with temperature. The effect of operating conditions, i.e;, feed temperature and composition or converter design controlling the velocity profile in the channels was also studied. Operating conditions determine the temperature profile and the catalyst utilization through out the monolith and consequently may change the conversion of pollutants and the thermal damage of the catalyst. The control of the velocity distribution may suppress the incompatibility between high conversion in permanent regime and light-off performance. Model results were used to suggest design improvements of catalytic converters, for example high upsteam activity of catalyst improves the light-off.
1 lntroduction
The competition between fluid dynamics, chemical reactions, and heat and mass transfer processes makes the theoretical and numerical study of the catalytic converter complex. The problem becomes even more challenging when dealing with the transient states associated with cold-start, change of gears, speed-up and slow-down. Several numerical models (Chen D.K.S. et al., 1988; Becker E.R. et al., 1986) attempt at describing the converter behavior. However, due to the complexity of these models, most of the physical and operating parameters are kept constant, and therefore few design improvements are deduced from the simulation results. Contrary to the available detailed models, we propose a tendency model which allows one, first to obtain good qualitative predictions rather than inaccurate quantitative results, second to test the sensitivity of the model to various parameters and physical assumptions, and finally to deduce possible design improvements.
466
2 Modelling
2.1 Fluid dvnamics, heat and mass transfer The monolith is modelled by several "macro-channels'' placed in parallel see figure 1. The macro-channel is a set of monolith channels where the fluid velocity is assumed to be identical. Fluid velocity can be different from one macro-channel to the next. It is known from the literature (Lemme and Givens, 1974; Howitt and Sekella, 1974; Leclerc et al., 1989a and 1989b) that the radial velocity profile strongly depends on geometrical properties and flow rate. As a consequence, a few macro-channels (i.e., one to three) are sufficient to account for the available information on the non uniformity of the radial velocity profile. A single macro-channel will be used to model a uniform velocity profile, whereas two or three parallel macro-channels will describe a more or less distorted profile. high now
1 to 1.5 d u r n
rate medium flow rate
El 0.5 to 1 d u r n
low now
0.1 to 0.5 d u
cross section: 61,M, ~ 3 ~4, solid mass: ml, m2, m3, m4
rate
no
0 0 to 0.1 d u r n
heat loss to the surrounding
now rate
Figure I : Sketch of the model. In thisfigure, the monolith is assumed to be composed offour zones fed by differentflow-rates. UlUm is the normalized velocity where um is the flow rate divided by the open section.
Each channel is composed of mixing-cells in series which are known to model a dispersed plug-flow (Villermaux J., 1985). The film model is chosen to describe heat and mass transfer processes. A brief description of the equivalence between Graetz problem and film model is given by Schweich. D. et al. (1990). A detailed discussion can be found in Young and Finlayson (1976).
467
2.2 Chemical reactions and kinetics In the literature, most of the kinetics rate are taken from Voltz et al. (1973). The inhibition factor (equation 3) can be a little different from one publication to another. Presently, the more detailed kinetics were proposed by Chen et al. (1988). There are no reliable kinetic data for NO reduction. Consequently our model includes only the oxidation reactions (equations 1a, lb, lc) and the NO mol fraction is constant. Since the thermal effect due to NO reduction is very small (Chen D.K.S., and al., 1988), the temperature predictions will be good. However, it will be impossible to estimate the chemical competition between reduction and oxidation. We consider the oxidations of CO, H2 and C3H6 as follows:
co + 1/2 0 2 + c 0 2 H2 + 1/2 0 2 + H20 rH = 2
k l xH,xo2
DEN
The value of DEN is given by equation (3): DEN = T 1 + Kal xco+ Ka2 xCpdZ( 1+Ka3xco
(
2
XCJI,
2
)(1+Ka4
~ ~ 2(3) 3
The reaction rate constants (mol. K. m-2.s-1) are given by equations (4a) and (4b): (44 13
k, = 1.392 10 exp
-14556
(4b)
The adsorption equilibrium constants are given by equations (5a) to (5d). Although theoretically questionable, the positive heat of adsorption of NO (see eq. (5d)) is consistent with experimental data (Voltz and al., 1973). (54
(5b)
(5c)
(54
468
2.3 Equations A single mixing-cell is represented in Figure 2. i
: Species k : Cells I : Reactions
Figure 2: Sketch of a mixing-cell and notations. The mass balance equations in the k-th cell is given by equation ( 6 ) for the bulk gas phase and by equation (7) for the gas phase in the washcoat. The flux CD is due to convection through the film. a (k) is a specific "activity" of the catalyst defined by the ratio of the surface of catalyst to the volume of the gas in the monolith.
m T,= TS7 +
Since the temperature is not constant in the film, it is not obvious whether we must use Tg or Ts in the mass transfer term. The results of the simulations are nearly identical whatever the temperature used. The thermal balance in the k-th cell is given by equation (9) for the gas phase in the washcoat, and by equation (10) for the bulk gas phase. In this paper, heat conduction in the solid is ignored.
m,
469
Molar fractions obey the following constraints:
Finally for J mixing-cells and n chemical species, there are J ( 2n+4) equations. 2.4 Standard parameters for the simulations The catalytic converter is operated in transient state with respect to composition, flow rate, temperature, etc... so it would be an enormous task to simulate all the possible conditions. We chose standard parameters for the simulations. Theses parameters are given in Table 1.
Feed properties 3 Flow-rate:1mol.sTi,l,=700K Flat velocity profile
Table I :Standard parameters for the simulations.
.. .
3 Sensitivitv to heat and mass t ransfer parameters As mentioned in the introduction, the simple model proposed can be used to test the sensitivity to various assumptions or underlying physical laws. Let us focus on the heat and mass transfer phenomena.
470
3.1 Limiting Nusselt and Sherwood numbers Limiting Nusselt and Sherwood numbers lump together the conduction processes and some geometrical properties of the channel. These limiting numbers are only known for ideal geometries and constant wall temperature or constant wall heat flux. In fact the channel walls are not perfectly regular and chemical reactions in the wash-coat do not imply a constant wall temperature or constant wall heat flux. So it is very important to test whether the results are sensitive to these transfer parameters. Figures 3 and 4 respectively show the time evolution of H!: mol fraction and C3H6 mol fraction in the first and in the last cell. Table 2 gives the residual fractions when steady state regime is reached. Light-off time and conversion at steady state are sensitive to the limiting Nusselt and Sherwood numbers. Accurate values are necessary in order to yield a good prediction. This also means that in addition to the macroscopic shape of the channel, the coating method, which is responsible for surface roughness, may affect the heat and mass transfer processes.
Table 2: Effect of the limiting Nusselt and Sherwood numbers on the C3H6 and H2 residual values when steady state regime is reached.
Figure 3: Time evolution of C3Hg mol fraction. Effect of limiting Nusselt and Sherwood numbers. Continuous line Num=2,4 Dotted line Num=3 Broken line Num=3,6
47 1
= 0.0024 -
Figure 4 Time evolution of H2 mol fraction; Effect of limiting Nusselt and Shenvood numbers. Continuous line NuM=2,4 Dotted line Nu-=3 Broken line NuM=3,6
1:
1:
1
0
6
12
18 24
30 36 42 48 Time (s)
3.2 Thermal conductivity and molecular diffusivity When the limiting Nusselt and Sherwood numbers are fixed, it is necessary to know the thermal conductivity and the molecular diffusivity to deduce the transfer coefficients. We tried both a constant diffusivity or a temperature dependent diffusivity according to:
CITd(6)
1.75
DO=
(12)
The temperature dependent thermal conductivity (J.m- .s-l . K - l ) was given by:. x,= f i ~ s ) =2.269 ~ o - ~ T , ~ . ~(13) ~ ~
Figures 5 and 6 show the effect of the temperature dependance of the thermal conductivity on the time evolution of CO mol fraction in the first and the last cell (constant molecular diffusivity) and the effect of the temperature dependence of molecular diffusivity on the time evolution of CO mol fraction in the first and the last cell (constant thermal conductivity). Table 3 gives the residual mol fraction of CO when steady state regime is reached. Thermal conductivity is found to affect the light-off time exclusively. The temperature dependence of diffusivity affects the light-off time, and the steady-state conversion.
472
h,= f(832K) parameters D = D (832K) cell number h, = f(832K) h, = f(T) D = D(832K) D = f(T) 1 8,767 10-3 8,766 10-3 8,767 10-3 8,922 10-3 1,44 10" 1,18 10" 9 1,44 104 1.44 10" Table 3: Residual molfractions of CO; effect of temperature dependent molecular diffusivity and thermal conductivity
c
Figure 5 Effect of the temperature dependent thermal conductivity on CO mol fraction (constant molecular diffusivity). Dotted curves: constant thermal conductivity. Continuous curves: temperature dependent thermal conductivity. Upper curves: first cell. Lower curves: last cell.
0.0120 -
Q0040
-
a0020 -
0 4 b 12 16 M2A 26 5 2 3 6 4 0 Time (s)
..---*. ........ ..............
'
j2
16
24
32 36 Time ( 3 )
Figure 6 Effect of the temperature dependent molecular diffusivity on CO mol fraction (constant thermal conductivity). Dotted curves: constant molecular difisivity. Continuous curves: temperature dependent molecular diffusivity .
473
3.3 Length-deuendent Nusselt number A standard monolith is mainly operated at the limiting Nusselt number except in the developing laminar flow regime zone where it is higher. A simulation using Nusselt number and Sherwood number equal to 1, 2 and 3 times the limiting Numbers in the first cell for a monolith of 16 cells, shows the effect of high mass and heat transfer in the leading centimeters Figure 7. A high value of the Nusselt number is responsible for rapid light-off and a higher conversion at steady-state. This confirms the results of Wendland (1980) concerning the usefulness of a segmented monolith in order to have several developing flow regime zones and consequently to improve the conversion.
Q0020
Figure 7 Effect of high Nusselt and Shenvood numbers in the first centimeters of the monolith on CO mol fraction. Continuous line Nu=Nu, Dotted line Nu=2Num Broken line N u = 3 N u ~ In these simulations, the molarflow rate is two mol. s-1 and the dimensions of the monolith are I I % smaller than standard values.
-
0 0
"
4
"
d
"
12
"
16
"
"
24
"
26
~
Time ( 3 )
4 Desian parameters 4.1 Feed temperature
An important design parameter is the location of the muffler with respect to the engine. Knowing that the gas temperature decreases by about 100 K per meter of exhaust tube, the location of the muffler can be simulated using different feed temperatures. Figure 8 illustrates the time evolution of CO mol fraction according to the inlet temperatures. Below 600 K the lightoff is strongly delayed.
474
ILU 3
4L -
650K 550K
0
o
.
o
o
2
0 ,..,~ , , , ,
’\.....
0 0
,
Figure 8 Dependence of CO molfiaction on inlet temperature. (from the left to the right:
750K,700K,650K,600K,550K)
20 40 60 60 100 120 140
Time (s)
Due to the external heat transfer resistance, the solid temperature in the first cell can be up to lOOK higher than the bulk gas temperature, whereas gas and solid temperature are nearly identical in the last cell. Therefore, a too high inlet temperature may result in thermal damage of the catalyst in the leading few centimeters. These results can be used to find a compromise between early light-off and resistance to thermal degradation. 4.2 Flow distribution The flow distribution can be changed by a deflector located in the divergent (Leclerc et al. 1990a, b). In order to test the efficiency of flow deflectors, we simulated the converter behavior assuming three differents distributions: 1-Uniform velocity with a single macro-channel and U=Um 2-Two identical macro-channel with u1=0.5 ummd u2=1.5 Um 3-Three identical macro-channels with u1=0.5 Um, u Z = U ~u3=1.5 , Um where Um is the average velocity.
II Distribution 1
1I CO mol fractiod, C?Hh .,- mol fraction
I
I , ” 1
2.03 10-41 4.12 10-4
2.3 10-5 4*3 3.6 10-5
1I
I
I
Table 4 : Pollutant molfraction at steady state for three different velocity distributions.
475
Early light-off occurs when there are some channels with high gas velocity (1 or 2 second earlier), pollutant mol fraction at steady state is then higher as shown in Table 4. Non uniform distribution is also responsible for thermal gradients. When the steady state regime is reached temperature differences up to 25 K can be observed from one macro-channel to the next.
4.3 Effect of the oxvgen concentration There are two main categories of catalytic converters. The first one is the oxydation catalytic converter, which is used only for the oxidation of CO and unburned hydrocarbon. The second one is the three-way catalytic converter. It is necessary for the elimination of CO, unburned hydrocarbon and NOx and it works under stoichiometric conditions whereas the first works with an excess of oxygen. At this time all the simulations in the literature are focused on the oxidation converter. Figures 9 and 10 show the time evolution of the gas CO mol fraction and solid CO rnol fraction in the first and the last cell for the stoichiometric mixture and with an excess of oxygen. With an excess of oxygen the reaction is totaly under mass and heat transfer control. With the stoichiometric mixture, there is competition between reaction and external mass and heat transfer.
0
6 16 24 32 40 48 56 64 72 60
T; me Is)
Figure 9 Time evolution of CO molfraction in a stoichiometric mixture Dotted curves: solid mol fraction Continuous curves: gas mol fraction Upper curves first cell Lower curves :last cell
o
6 16 24 32
40 4a 56 64 72 60 Time (s)
Figure 10 Time evolution of CO rnol fraction with an excess of oxygen Dotted curves: solid mol fraction Continuous curves: gas rnol fraction Upper curves first cell Lower curves :last cell
476
. .
5 Quantitv and distribution of c a t a m When the catalyst is chosen, there remains to know which amount should be used and how it should be distributed throughout the monolith. 5.1 Effect of the auantitv of catalvst Figure I 1 shows the effect of the quantity of catalyst on the time evolution of outlet CO mol fraction. High quantity of catalyst is responsible for rapid light-off, because the start of reaction is under kinetic control. But the residual fraction of CO is the same (0.000117) whatever a (k) because steady-state conversion is always under external diffusion control even for low quantity of catalyst. These simulations are also a representation of a uniform desactivation of the catalyst. In this case the activity of the catalyst decrease from 3 105 to 5 104 m2 of catalyst. m-3 reactor gas volume.
Figure I I EfSect of the quantity of catalyst on time dependent CO mol fraction. a.10-2 =500,1000,2000 and 3000 m2 of catalysis. m-3 reactor gas volume from the right to the left of the figure.
0
8
16
24
40 48 Time(s)
32
5.2 Effect of the distribution of catalyst
Since the conversion is very different from one mixing-cell to the next, the axial distribution of the catalyst is a potentially interesting design parameter. Three stepwise catalyst distributions corresponding to the same total quantity of catalyst were investigated at stoichiometry , and the results are illustrated in Figure 12. A high upstream activity improves the light-off, but the steady state conversion is nearly the same whatever the distribution.
477
Distribution mixing-cell 41) nO1 Uniformdistribution 500 n02 Hight dowstream activit:I 300 Hightupstreamactivity 700 n"3
42) 500 350 650
d 3 ) 44) 45) a(7) 4 8 ) 500 500 500 500 500 500 400 450 500 550 600 650 600 550 500 450 400 350
49)
u
500 700 300
Table 5: Catalyst distribution used in the simulations a(k).10-2 (m2 catalyst. m3 gas in the monolith)
Figure 12 Time evolution of CO mol fraction. Effect of the catalyst distribution. Continuous line: distribution n "I ; Dotted line: distribution n "2; Broken lines: distribution n "3.
- 0.0080a0040 -
Q0020
-
0
8
16
24
32
40
48
56
Time (s)
s Conclusions We developped a mathematical model of a catalytic converter in order to better understand its behavior and to improve its design. The results of the simulations show that conversion at steady state and light-off performance are sensitive to heat and mass transfer. Because high Nusselt number at the entrance of the monolith is responsible for rapid ignition of the reactions, light-off performance can be improved by used a segmented monolith. A non uniform radial velocity distribution, a channel shape with a high Nusselt Number, a high catalyst activity in the leading centimeters of the monolith, and a short distance between the engine and the catalytic converter are also responsible for good light-off performances. Conversion at steady state can be improved by a uniform radial velocity distribution. The performance of the catalyst is not very important in the case of an axiding monolith because the reactions are under external mass transfer control. If the gas mixture is really at stoichiometry in the film which is the case for a three-way catalytic converter, the conversion is both under mass transfer and kinectic control, so that it is not obvious whether a good catalyst activity or a good mass transfer efficiency are required. This problem should deserve futher study.
478
Acknowledae ments: This work was carried out within the "Groupement Scientifique Pot Catalytique" of the Institut Franqais du PCtrole, the PIRSEM (Programme Interdisciplinaire de Recherches sur les Sciences pour 1'Energie et les Matikres premikres) and the Chemistry Department of the CNRS. List of svmbols a Cp
Activity of the catalysis (m2 catalysis.m3 gas in the monolith) Heat capacity (J.kg-1.K-1) D Average diffusivity in the homogeneous pore fluid (m2.s-') DEN Quantity defied by equation 3 F MOIZ flow rate (mol. s-1) h Heat transfer coefficient (W.m-2.s-1) J Number of mixing-cells k rate constant (mol.m-2.s-1.K-1) Mass transfer coefficient (ms-1) Ka Adsorption constant Nu Nusselt number Nu- Limiting Nusselt number Pressure (Pa) P R Gas constant (J.mo1-l.K-l) S Washcoat area (m2) Time (s) t T Temperature (K) U Fluid velocity (m.s-1) V Gas volume (m3) X Mol fraction in the gas phase Mol fraction in the gas phase of the whast-coat XS AH Reaction enthalpy (J.mo1-1) CD Convection flux across the film Thermal conductivity of the gas (J.m-1.s-1.K-l) k Density (kg.m-3) P Stoichiometric coefficient V
Su bscrir>ts
g 1
j k m s
Gas Species Reactions Cells Averagevalue Solid
479
Literature cited: Chen D.K.S, S.E. Oh, E.J. Bisett, and D.L. Van Ostrom: A three dimensionam model for the analysis of transient thermal and conversion charateristics of monolithic catalytic converter. SAE paper 880282, International Congress and Exposition, Detroit, Mich., (1988). Howitt J.S., and T.C. Sekella: Flow effects in monolitic honeycomb automotive catalytic converters. SAE paper 740244, Automotive Engineering Congress, Detroit, Mich., (1974). Leclerc J.P., D. Schweich, and J. Villermaux: Transfert de chaleur en rCgime transitoire dans un monolithe destinC i l'kpuration de gaz dkchappement automobile. SFT 89 Congress, "Thermique et GCnie des PrwCdCs", Nancy, (1989a). Leclerc J.P., D. Schweich, and J. Villermaux: Hydrodynamique et transfert de chaleur dans un monolithe destine i 1'Cpuration de gaz dkchappement automobile. Symposium "GCnie des procCdCs", Toulouse, (1989). RCcents progrks en gCnie des procCdCs, Lavoisier Ed., 3(8a), 518-524, (1989b). Lemme C.D., and W.R. Givens: Flow through catalytic converters - An analytical and experimental treatement. SAE paper 740243, Automotive Engineering Congress, Detroit, Mich., (1974). Oh S.H., and J.C. Cavendish: Transients of monolithic catalytic converters; Response to step changes in feedstream temperature as related to controling automobile emissions. Ind. Eng. Chem. Prod. Res. Dev., 21, 29, (1982). Schweich D., and J.P. Leclerc: Flow, heat and mass transfer in a monolithic catalytic converter. 2nd International Congress on Catalysis and Automotive Pollution Control, Brussels, 10-13 Sept., (1990). Shah R.K., and T.C London: Flow forced convection in ducts. Advances in heat transfer Laminar., Academic Press, New-York, (1978). Voltz S.E., C.R. Morgan, D. Liederman, and S.M. Jacob: Kinetic study of carbon monoxide oxidation on platinium catalysts. Ind. Eng. Chem. Prod. Res. Devel., 12, 295, (1973). Villermaux. J.: Gtnie de la rkaction chimique. Conception et fonctionnement des rCacteurs. Tec & Doc, Lavoisier, Paris, (1985). Wendland D.W.: The segmented oxiding monolith catalytic converter, theory and performance. J. Heat Transfer, 102, 194, (1980). Wenland D.W., and W.R. Matthes: Visualization of automotive catalytic converter internal flows. SAE paper 861554, International Fuel and Lubricants Meeting and Exposition, Philadephia, Pens., (1986). Young L.C., and B.A. Finlayson: Mathematical models of the monolith catalytic converter, part 11, Application to automobile exhaust. AIChE J., 22, 343, (1976).
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis andAutomotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
48 1
NEW DEVELOPMENTS IN CATALYTIC CONVERTER DURABILITY Suresh T. Gulati
R , D & E Laboratory, Corning Incorporated, Corning, New York 14831 USA ABSTRACT
A number of new developments in the areas of material composition, cell geometry, wall porosity, substrate contour, washcoat formulation, thermal and acoustic insulation, and clamshell and heatshield designs have occurred in recent years to meet the more demanding performance and durability requirements of automotive catalytic converters. This paper reviews the major developments in ceramic and metal substrates, stable washcoat systems, nonintumescent mounting materials, edge-shielded intumescent ceramic mats, ceramic insulation rings, dual can and dual cone packaging designs, and perforated heatshields with convection cooling. These improvements offer higher conversion efficiency, lower pressure drop and extended durability. A brief comparison of ceramic and metal substrate durability, based on laboratory tests, is also included. INTRODUCTION
This paper reviews some of the new developments in catalytic converter durability. In particular, it focuses on the ceramic converter system due to its proven success over the last decade and a half in the automotive industry [l]. A brief comparison of ceramic and metal converter systems, based on laboratory testing, is also included. The need for a durable converter system continues to grow worldwide due to highter air quality legislation, namely faster light-off and higher conversion efficiency, lower pressure drop, and longer product life requirements without compromising vehicle performance. To meet these requirements simultaneously, it is imperative to take the systems approach [2-4] wherein each component of the converter assembly is carefully designed, tested and optimized. Starting with
482
the automakers' specifications with respect to converter size and location, available space, fuel management, engine emissions and acceptable power loss penalty, the converter designer is challenged with selecting appropriate materials, substrate contours, catalyst volume and washcoat formulation, and assembling them in a durable package. In view of the variety of materials, configurations and microstructures employed in various converter components, the task of designing the total system becomes even more complex and requires cooperative effort by all of the component suppliers to meet the automakers' space, performance and cost specifications. The durability requirements begin with the substrate design both in terms of material and geometry. The initial design of the substrate is generally dictated by performance requirements [5-71. To meet the durability requirements, however, certain trade-offs must be considered. On the one hand, the mechanical and thermal durabilities call for high strength, high deformation resistance, and excellent thermal shock resistance of washcoated monoliths which, in turn, require low porosity, good cell definition and low coefficient of thermal expansion. On the other hand, faster light-off, higher conversion efficiency, and lower pressure drop call for high cell density, low thermal inertia, and thinner cell walls which are contrary to durability requirements. However, certain composition and process modifications can help optimize ceramic monoliths to meet both the performance and durability requirements. Recent developments in the design and properties of ceramic substrates of dense cordierite and mullite aluminum titanate compositions are reviewed. In addition, the high temperature properties of ceramic and metal substrates are compared [8]. The role of high surface area washcoats is also important in this context because their interaction with the substrate has a major impact on physical properties which ultimately control its durability [ 9 ] . For example, the formulation (composition, particle size, and rheology), the loading (96 weight) and the processing (number of calcining cycles, peak temperature and total cycle time) of washcoat and catalyst affect the strength, thermal expansion coefficient, and structural modulus of coated monolith which, in turn, determine its long-term durability. Both the microstructure and substrate/washcoat interaction have strong influence on these properties. The next critical element in the converter system is the ceramic/wiremesh mat which provides mechanical and thermal insulation against vibrational and thermal loads [ 10, 113. The compression, shear, erosion and frictional properties of these materials, notably at high temperature, control the stress level in the monolith and hence its durability. These properties, together with the criteria for selecting the optimum mat, are reviewed. New developments in ceramic mats of special composition, designed for metal monoliths, are also discussed briefly. In certain dual-bed converters, a ceramic spacer ring is employed between the monoliths to shield the mat and
483
insulate the can against high temperature gases thereby preserving the can's rigidity and minimizing its deformation [12]. The critical data for ensuring the durability of these rings is highlighted. Finally, the pertinent design and physical properties of stainless steel can and heatshield are addressed [13] The key role of the can, besides housing the catalyst, is to ensure a positive mounting pressure on the monolith, under all driving conditions, without excessive deformation in the operating temperature range. The latter is best achieved by stiffening the can in critical areas via ribs, shoulders and end cones and by designing effective cooling measures to minimize its operating temperature. It is here that the heatshield design commands meticulous attention to meet the dual requirement of shielding the floor pan against the intense heat of the converter and keeping the can temperature as low as possible via effective convection cooling. However, the use of acoustical pads between the can and heatshield to minimize vibrational noise in certain applications negates some of these benefits and must be readdressed from durability point of view. Alternative can designs that do not require a heatshield and yet meet the insulation, cooling, and acoustical requirements are also reviewed [ 14,151. The paper concludes with a brief review of special packaging requirements for metal monoliths, with and without the outer mantle, to compensate for their low rigidity at high operating temperatures, e.g. in close-coupled applications. NEW DEVELOPMENTS IN SUBSTRATES
With stricter emission standards and low back pressure requirement, both ceramic and metal substrates have undergone significant developments over the past few years. The thrust in the ceramics area has centered on thinwall structures to minimize pressure drop and on high temperature refractoriness to permit higher use temperature in close-coupled, light-off, and heavy-duty applications [ 16 -191. The thin-wall structures are extruded from dense cordierite to provide strength and thermal shock resistance equivalent to those of standard cordierite substrate to meet the physical durability requirements; see Fig. 1. Known as Celcor XT, they are available in two different cell sizes: (i) 35"/5-5 for low back pressure and (ii) 47015 for higher conversion efficiency and lower back pressure than the standard 400/6.5substrate. The pertinent geometric and physical properties of four different cordierite substrates are summarized in Table 1.
484 TABLE 1
Nominal Properties of Standard and Thin-Wall Cordierite Substrates
I
COMPOSITION EX-20 (Stdl
EX-32 (St4
EX-22 EX-22 (Thin-Wall: :Thin-Wall)
Cell Structure Cell Shape Wall Thickness, mm Open Porosity Mean Pore Size, pm Wall Strength, MPa (Psi) Wall Modulus, GPa MIF Substrate Strength* TIF TSR OFA, % ( I ) GSA, cm2/cm3(2) Avg. CTE 8OO0C,(3) 10-7pC Dh, mm Ap(4)
40016.5 0 0.188 35% 3 .O 20.3 (2950) 26.1 0.025 73k 6.9 llOK 73 26.4
23611 1 A 0.279 42%
350/5.5 0 0.140 20%
7.0 14.3 (2080) 20.9 0.031 64k 7.4 105K 65 22.2
2.0 43.1 (6250) 42.9 0.012 75k 9.7 130K 80 26.4
470/5 0 0.127 20% 2.0 43.1 (6250) 42.9 0.013 81k 9.7 140K 80 30.4
6.0 1.106 172C
5.0 1.171 191C
4.5 1.218 129C
4.5 1.045 154C
Wall Density, dcm3
1.61
1.42
2.0
2.0
Substrate Density, g/cm3 Heat Capacity of Substrate, cal/cm3'C
0.43
0.49 1
0.39
0.41
0.11
0.12
0.10
0.10
(1) OFA: Open Field Aperture. (2) GSA: Gross Surface Area. (3) CTE: Coefficient of Thermal Expansion. (4) The constants k and C are normalization constants to help compare strength and pressure drop characteristics of cordierite substrates with different porosity and cell structures.
485
EX- 22 ,350/ 5.5
EX- 20 ,400/6.5
EX - 3 2 , 2 3 6 / 11
Figure I - Comparison of standard and thin-wall cordierite substrates
486
The first two are standard commercial products of EX-20 and EX-32 compositions while the last two are newly developed from the dense cordierite composition EX-22. These properties help compare light-off and steady state conversion activity through heat capacity and GSA values, engine performance through Ap, and physical durability through substrate strength, TIF and CTE [6]. The pressure drop values in Table 1 are based on constant flow rate i.e. given engine conditions) and constant total GSA (i.e. equivalent conversion activity). They are given by equations (1) and (2) for square and triangular cell respectively:
where
cs =
- 4.flow-rate.(substrate length) total surface area
4.4.1 4A(L-t)
2
L2 Ct = 443
+
2
- 4.flow-rate.(substrate length) total surface area (L-td?)
4.Q.1
2
L
in which Q denotes flow rate, A and 1 are cross-sectional area and length of substrate respectively, and L and t are side length and wall thickness of the unit cell. The constancy of total GSA requires that
which relates the cell dimensions and volume of the four substrates being compared in Table 1. Similarly, the strength of the substrate is proportional to the product of mechanical integrity factor (MIF) and wall strength with k as the proportionality constant [3] strength = kow (MIF)
(4)
Its values are shown in Table 1. The thermal shock resistance (TSR) of the substrate is related to its strength, CTE, thermal integrity factor (TIF) and wall modulus [5,6]:
487
TSR =
k cw(MIF) (TIF)
2E w a w k
TSR =
(TIF) 3Ewaw
for square cell
Gw(MIF)
for triangular cell
(6)
We may summarize the key points in Table 1 as follows: (i) thin-wall substrates enjoy similar or higher GSA than standard substrates thereby providing equivalent or improved catalyst activity; in particular, the 470/5 substrate offers 15% higher geometric surface area than the 400/6.5 standard substrate; (ii) thin-wall substrates result in lower pressure drop than standard substrates at constant flow rate and total surface area; in particular, the 350/5.5 substrate experiences 25% lower pressure drop than the 400/6.5 standard substrate; see Figs. 2 and 3; (iii) thin-wall substrates offer similar or higher physical durability as standard substrates; in particular, their thermal expansion is lower than that of 400/6.5 standard substrates; see Fig. 4, implying higher thermal shock resistance. Table 2 compares the crush strength data along A, B and C axes for standard and thin-wall substrates. In addition it lists the axial MOR and E-modulus data for these substrates. It is clear from Table 2 that thin-wall substrates are just as strong as the standard substrates and possibly stronger. The differential expansion strain values (DES) over the two different temperature ranges"(450" to 825°C and 600" to 9OO"C), are taken from Fig. 4 to compute the thermal shock parameter using eqn 7 : TSP = MOR E x DES
(7)
The higher the TSP value is, the more thermal shock resistant the substrate will be. It is clear from Table 2 that the thin-wall substrate has a higher thermal shock resistance than the standard substrate under both North American and European driving conditions.This is also borne out by the laboratory thermal cycling tests on full monoliths. *The 450" to 825°C range represents the nominal catalyst skin and center temperatures under North American driving conditions, and 600" to 900°C range represents the same temperatures under European driving conditions.
488
50
-
1
1
200
1
1
1
1
1
1
1
1
/
,
1
1
280 320 Flow Rote (SCFM)
240
1
,
1
,
400
360
Figure 2.- Pressure Drop vs Flow-rate through Uncoated Standard and Thinwall Ceramic Substrate. - 81mmx170mmxll7mm long 70 -
350/55
t h i n wall
-
20
200
'
/
"
'
240
/
'
I
"
280
'
'
'
I
320
'
I
'
/
360
'
400
Flow Rate (SCFM)
Figure 3.- Pressure Drop vs Flow-ratet hrough Coated Standard and Thin-wall Substrates.
-
1
5 - -Thin '600- ---
'
1
Wall Substrate
'
1
'
1
'
,
Figure 4.- Thermal Expansion Curvesfor Standard and Thin-Wall Ceramic Substrates.
489 TABLE 2
Comparison of Strength and Thermal Shock Parameter of Standard and Thin-Wall Substrates EX-20 400/6 5 (Stdl Axis Crush Strength MPa B Axis Crush Strength MPa C Axis Crush Strength MPa Axial MOR, MPa Axial E Modulus, GPa
EX-22 350/5 5 (Thin-Wall)
20
47
3.6
45
0.3 22 6.5
0.3 3 .O 8.4
DES
45OoC-825"C TSP 450°C-825"C
DES 600"C-900"C
TSP 600"C-900"C
525 x 10-6 0.64 485 x 10-6 0.70
430 x 10-6 0.83 433 x 10-6
0.82
A more refractory ceramic material, mullite aluminum titanate is under development for high temperature applications, e.g heavy-duty gasoline truck engine. Its composition has been optimized to eliminate high-temperature decomposition and minimize thermal cycling growth, both being undesirable from physical durability point of view [18,19]. Its density is 20% greater and the use temperature 200°C higher than the corresponding values for standard cordierite. The initial testing is centered on physical durability of 400/6.5 square cell structure with geometric properties (GSA, OFA, Dh) and performance parameters (TIF, Ap) identical to those of standard cordierite substrate.
490
Both the high temperature vibration test and exhaust gas simulation test were carried out for a heavy-duty truck converter comprising of three 6" diameter x 3.2" long MAT substrates wrapped in Interamtm IV ceramic mat, 8800 g/m2, with a mount density of 1.1 g/cm3 see Fig. 5. All three substrates survived the 20 hour vibration test at 1OOO"C under an acceleration of 33 g's and a frequency of 100 Hz there was no detectable movement of either the substrates or the mat. The EGS test was performed using the propane gas burner/Roots air blower at an inlet gas temperature of 950°C. The typical cycle consisted of 135 sec.heat-up, 600 sec. hold, and rapid Figure 5. MAT Substrate Assembly for cool-down. The typical cycle consisted of 135 sec heat-up, 600 sec High Temperature Vibration Test. hold, and rapid cooldown. A steady state radial gradient of 150°C was measured at midbed of the substrate and a temperature drop of 400°C was recorded across the insulating mat. All three of the MAT substrates survived five such cycles with no visible or audible damage, testifying to the excellent thermal-shock resistance of these substrates. Similar tests with coated MAT monoliths are in progress and will be reported in the future. The metal foil monoliths of Fe-Cr-A1 and Fe-Cr-Al-Ni compositions and 400/21 cell structure offer larger open frontal area, higher geometric surface area and bigger hydraulic diameter with potential light-off, pressure drop and conversion efficiency advantages relative to the standard cordierite substrates [20]. However, the field data though sparse, do not confirm these advantages in a consistent manner due to minimal differences in heat capacity and hydraulic diameter of ceramic and metal cell structures; there are also certain durability issues with the washcoat/metal adhesion. Furthermore, the metal monoliths have poor physical durability above 800"C[21]. They oxidize and become brittle, and/or they deform permanently under sustained operating stresses at high temperature [22]. While the foil thickness is 0.002", the average wall thickness of metal substrates as determinrd by image analysis can be twice as large due to overlaying of foil in certain image regionsduring manufacturing. 1
49 1
n"
-I
15
CelcorR Ceramic EX-20, 400/6 5
L
o
-
0 Lo
)
-
Emicat-S. 400/2
400
600
Temperature,
000
1000
1200
O C
Figure 6.- Measured Strength vs Test Temperature for Standard Celcor, Emicat-S and Metalit SQ Discs.
Time ( m i n )
Figure'/.- Percent Creep Strain in Emicat-S, 40012 Specimens at Various Temperatures under a Static Compressive Stress of 0.09 MPa. Figure 6 compares the high temperature biaxial strength of standard cordierite and commercial metal foil discs in the concentricring flexure test [22,23]. Not only are metal discs weaker, their strength decays rapidly above 400°C. The typical failure mode consists of either "peeling" or "telescoping" due to weakness of joints between crimped and flat layers which are unable to transmit the operational loads exerted by inlet gas pressure and the engine and chassis vibrations.
492
-s
I -10 -
I
I
I -
T= 600°C
P = 0 0 9 MPa-
c
z -70
-
-"O
! 20
1200°C
40
60
80
I00
120
Time (min)
Figure 8.- Creep Strain in Metalit SQ, 15012 Specimens at Various Temperatures under a Static Compressive Stress of 0.09 MPa. Similarly, Figs. 7 and 8 show the high temperature creep deformation experienced by 1 cm x 1 cm x 0.5 cm thick specimens of two different metal monoliths under a modest pressure of 0.09 MPa representative of canning load in service. As the creep sets in and the cells deform, the potential advantages of large frontal area geometric surface area and hydraulic diameter are steadily lost! It should also be pointed out that the differential expansion strain between the foil core and solid mantle also contributes to radial pressure responsible for high temperature creep. Identical tests on cordierite Ceramic specimens of 400/6.5 cell structure yielded no permanent deformation up to a test temperature of 1200°C. SUBSTRATE/WASHCOAT INTERACTION
When viewed individually, both the substrate and alumina washcoat offer optimum set of properties for their prescribed function. However, when processed together to form a catalytically active support, their composite properties may be altered due to an order of magnitude higher thermal expansion of alumina washcoat. Since the overall catalyst durability, i.e. mechanical, thermal and catalytic durability, depends on composite properties the fundamental expansion incompatibility between cordierite ceramic and y-alumina must be managed properly through judicious control of raw materials, substrate microstructure and washcoat processing. For prolonged durability, the coated substrate should have high strength, low E-modulus, and minimal increase in thermal expansion relative to uncoated substrate
P31.
493
I
Porous Alumina Washcoat
-t - - -
--
Figure 9.- Schematic of Washcoated Substrate Wall. It is only then that the catalyst possesses sufficient mechanical and thermal integrity to withstand vibrational and thermal shock stresses encountered in service. Figure 9 is a schematic of typical washcoated substrate wall which may be treated as a 3-layer symmetric composite with mechanical adhesion at alumina interface. The expansion coefficient of such a composite which controls the thermal shock resistance, is given by [25]:
where Es, a, and Ec, ac denote E-modulus and expansion coefficient of substrate and washcoat respectively, and ts and tc denote the thicknesses of substrate wall and washcoat layer respectively. Using the nominal properties of these two materials listed in Table 3, we compute a composite expansion coefficient of 29 x lO-7/"C - an increase of 400% over that of cordierite substrate! Obviously, such a high value of composite expansion coefficient is unacceptable from a thermal shock point of view. Moreover, the measured value of composite expansion coefficient, as shown in Figure 10, generally falls between 6 x lO-7/"C and 15 x 10 -7PC for different washcoats indicating that either the washcoat modulus Ec is not as high as shown in Table 3 or the interfacial adhesion between substrate and washcoat is less than perfect. Both of these scenarios are controlled by washcoat formulation and processing. It is here that the coaters can play a key role in ensuring the catalyst durability without compromising its activity and engine performance. On the one hand, higher and prolonged catalyst activity calls for either an increase in washcoat loading or the use of denser washcoat; on the other hand, these very elements can have an adverse impact on E-modulus and composite expansion coefficient both of which affect the thermal shock resistance.
494 TABLE 3
Nominal Properties of Standard Cordierite /Alumina Composite Cordierite/ Alumina Comuosite 30 1.60 27.8 0.25 0.225 29
35 1.61 26.1 0.25 0.165 6
Porosity (%) p, dcm3 E,GPa V
t, mm a,10-7/oc (Avg. Value: 25-80OOC)
TABLE 4
Effect of Washcoat Modulus on Composite Expansion Coefficient Ec GPa
32.3 30.0 25.O 20.0 15.0 10.0 7.5 5.0 2.5 0.0
a (10-7pc)
29.0 27.8 25.1 21.o 18.8 15.0 13.0 10.8 8.5 6.0
1400
,
,
,
,
,
Temperature,
I
'
.I
("C)
Figure 10. Thermal Expansion Curves for Three Standard Ceramic Catalysts with Different Washcoat Systems Through judicious formulation and process control, the coater can modify the washcoat micro-structure and its E-modulus. The latter can range from a high value of 32.3 GPa, as shown in Table 3, to as low as zero! The
495
impact of such a range of washcoat moduli on composite expansion coefficient, as given by eqn. 8, is shown in Table 4. It is clear from these values that a low modulus washcoat helps preserve the low expansion properties of cordierite substrate and is, therefore, highly desirable. Since the measured values of a range from 6 x lO-7/"C to 15 x 10-7 /"C, the practical range of Ec according to Table 4 is between 0 and 10 GPa, i.e. the bond between cordierite substrate and alumina washcoat is not a theoreticallyperfect-bond which is good news for physical durability. In view of its high expansion, the alumina washcoat can also induce significant tensile stresses in the substrate wall during heating. The magnitude and impact of such stresses on the mechanical integrity of catalyst depend on interfacial bond quality, microscopic distribution of alumina particles in the open pores of substrate wall, and the morphology of these pores. The maximum tension induced in the substrate wall due to a differential expansion strain of (AaAT) is given by [26]
while that due to constrained e: pansion of spherical alumina grains in filled pores is given by [27,28]
&+Y 2tc
2 E s m
I
where A a = ac - as and AT denotes the temperature difference between alumina and cordierite. Using the nominal properties of Table 3 and assuming A a = 80 x 10-7 /"C and AT = 300°C during washcoat processing, eqns. 9 and 10 yield osw = 21 MPa and o ~ =p 61 MPa stresses which exceed the wall strength shown in Table 1. Even with a more realistic value of washcoat modulus of 10 GPa, o ~ =p 32 MPa which exceeds the fatigue threshold of cordierite wall and can lead to strength degradation via slow growth of microcracks either during processing or in service. Figure 11 is a Weibull plot of tangential modulus of rupture of three catalysts with different washcoat formulations and thermal history; it demonstrates clearly that while most of the test specimens yielded a mean strength of 1.5 to 2 MPa, regardless of coating differences, washcoat C3 resulted in the largest variation with minimum strength approaching 0.25 MPa! Such a variability is detrimental to the long-term reliability of ceramic monoliths. It is also clear from eqn. 10 that OSP can be kept near zero by limiting the washcoat modulus, Ec, to very low values through process control. With Ec near zero,
496
the effective differential expansion strain, AaAT , also approaches zero with the result that the induced tension in substrate wall approaches zero, thereby preserving the mechanical integrity of the total catalyst. 0 999 0990 0900
0 800
0600 t 0400
a 0
e
a 2
0200
= 9
0100 0050
0 020 0010
0
Tang. Modulus o f Rupture (MPa)
Figure I 1 .- Weibull Distribution of Tangential MOR of Three Standard Ceramic Catalysts with Different Washcoat Systems. DEVELOPMENTS IN CERAMIC MATS AND INSULATORS
As the converter location moves closer to exhaust manifold to help meet light-off requirements, for both medium-and high performance engines, the need for thermal insulation of stainless steel can and end cones increases to minimize their high temperature deformation and potential loss of holding pressure [29]. Several approaches are being considered by the automotive companies to improve thermal insulation; e.g. dual cone and/or dual can designs with some form of insulation in the annulus; hybrid mat system comprised of wiremesh, ceramic fiber, and intumescent ceramic mat; and ceramic insulation rings between the catalysts. The introduction of metal monoliths for certain strategic applications has also led to the development of special ceramic mats to meet both thermal and acoustic insulation requirements simultaneously. And finally, to minimize thermal erosion of ceramic mat at high inlet gas temperature, the standard intumescent mat with edges shielded by wiremesh screen is now a commercial product. Thus, the major developments in insulation hardware are primarily those associated with the ceramic materials.
497
The standard intumescent ceramic mat is available in a wide range of free densities and initial thickness so that the desired mount density can be obtained by using one or more of these to meet mechanical, thermal and acoustical requirements; see Table 5. TABLE 5
Standard Ceramic Mat and Nominal Compressed Thickness Mat Weight g/m2
Initial Thickness (mm)
1050 2100 2600 3100 3662 4070 6200
1.7 3.3 4. I 4.9 5.4 6.0 9.8
pm=O.gg/cd 1.2 2.3 2.9 3.4 4.1 4.5 6.9
pm=I .Og/cm2 1.05 2.10 2.60 3.10 3.70 4.10 6.20
The room temperature holding pressure depends on mount density, Pm and can be estimated (in bars) from p = 2760 exp (-6.7/pm)
(11)
This pressure will relax somewhat following the welding of clamshells and end cones. An initial pressure of 1.6-3.4 bars, corresponding to a mount density of 0.9 to 1.0 g/cm3- is highly beneficial to monolith durability. At higher temperatures, as the mat expands, this initial pressure may increase by 50 to 100% depending on heat-up rate. The maximum pressure could approach 7 to 10 bars which is well below the isostatic strength of the monolith. An improvement over the standard mat, in terms of thermal erosion resistance, is the Series WS mat with stainless steel wire screen wrapped around the edges; see Fig. 12. This special mat is designed for high inlet/exit gas temperature in that the wire screen shields the edges from thermal erosion. In view of its higher mass density and elastic stiffness in the edge region. Series WS mat offers higher mount density and exerts larger holding pressure in that region. Consequently, it should be positioned carefully during the canning operation so as not to.line up with an inner stiffener rib and cause shear cracks in the monolith due to a localized pressure band. And yet, Series WS mats are easy to assemble; they provide acoustic damping against gas
498
pulsations; they help contain frayed vermiculite and fiber segments; and they are available in the same wide range of free densities as the standard mat (See Table 5).
Figure 12.- Series WS Ceramic Mat with Screen-Shielded Edges Special ceramic mats have also been developed to meet the packaging requirements of metal monoliths which differ from those of ceramic monoliths. These special mats, Series MM for example, are made from ultrathin “sol-gel’’ fiber of alumina-silica-boria composition stitch-bonded to polyethylene cloth; they are available with or without an intumescent mat layer for additional holding pressure. The special construction of Series MM mat provides excellent thermal and acoustic insulation, critical for metal monoliths, with significantly lower holding pressure so as to minimize hightemperature deformation of 0.05 mm thick cell walls. This mat is very effective for metal monoliths with or without the outer mantle. As the metal core expands against the mantle or outer can, it compresses the mat appreciably with modest increase in holding pressure. The special, high strength, high alumina, sol-gel fibers behave elastically even at high operating temperatures and provide the needed holding pressure in a reliable manner. The flexural stress in these fibers, induced by mat compression, is relatively small in view of their ultrathin size and does not produce fracture; thus the mat retains its resiliency and continues to operate reliably throughout its lifetime. The typical Series MM mat is 10.5 mm thick and weighs 1500 dm2.
499
Table 6 shows the room temperature holding pressure as function of compressed thickness and mount density. TABLE 6
Compressed Thickness (mm) 12.7
7.6 6.3 5.1 3.8 2.5
Mount Density (g/cm3> 0.12 0.20 0.24 0.30 0.39 0.59
Pressure at 25°C (bar) 0 0.28 0.62 1.52 3.45 9.93
The nominal mount density for assembling metal monoliths is 0.25 g/cm3 which translates into a holding pressure of 0.7 bars. Upon heating to 800°C followed by a 15 minute soak time, the mount density and holding pressure increase to 0.3 g/cm3 and 1.5 bars respectively. The experimental data show that the metal monolith assembled with this mat can survive 20 cycles of 900°C vibration test at 40g acceleration and lOOHz frequency; the same monolith when assembled with wiremesh or intumescent, or fiberfrax mats did not survive even the first cycle. The aluminum titanate spacer rings are employed in certain multibed converters to shield the insulating mat between two catalysts from direct impingement of hot gases [12]. These rings are exposed to 900°C gas on the inner surface and often develop both through-wall and peripheral temperature gradients due to nonuniform flow profile. In addition to thermal stresses arising from these gradients, the ring is also subjected to mechanical loads due to elastic deformation of clamshells during canning. To ensure the durability of insulation rings, the combined thermal and mechanical stresses must not exceed the material strength. The room temperature MOR data of approximately 20 slip-cast aluminum titanate rings were obtained in the diametral compression test. In addition, the fractured ring, sections were retested in 4-point bending to characterize the material strength more completely; both of these tests are shown schematically in Fig. 13. The MOR data are plotted on Weibull paper in Fig. 14 to demonstrate strength variability. Both tests yielded similar strength values which ranged from 9 MPa to 58 MPa! The failure origin in both types of specimens was a spherical void buried just below the surface;
500
such voids are generally associated with raw material impurities and the firing process. The thermal stresses in the ring due to through-wall and peripheral gradients are given by [30]:
where Er, Vr and Ctr denote E-modulus, Poisson's ratio, and expansion coefficient of ring material and AT denotes the temperature gradient. Figure 15 plots these stresses as function of temperature gradient using the following properties2 Er = 29.6 GPa ;
Vr = 0.25 ;
It shows that the thermal stress due to each of these gradients can approach a value of 55 MPa which can lead to fracture.Conversely, to limit the thermal stress to below the minimum ring strength the maximum AT value must be kept below 100°C by controlling the flow distribution. Alternatively, the strength of ring material must be improved significantly and/or its expansion coefficient must be reduced by a factor of 8 through compositional and process modifications. New ceramic materials with good refractoriness and low expansion coefficient are currently being developed for more durable rings.
a r 2 28 x lO-7/"C ;
1
t
Figure 13 Diametral Compression and 4-Point Bending of Aluminum titanate Insulation Rings
2 The elastic properties were measured by sonic resonance; the low value of Er is due to 13% open porosity measured by mercury porosimeter.
50 1
0 99
zI
080 060 -
0
2 a
020-
0 0
010-
.
a
9
I
005 -
002 0 01
I
I
i
0.
8.
0 Bar Data
t = 040n 0 n
-3-
I
I
I 0 Ring Data
O0
0.
. .
8"1'
O
-
:
-
-
-
I
I
10
20
I 30
I 40
I 50
I 60 70
Figure 14.- Weibull Distribution of MOR Data for Aluminum Titanate Insulation Rings.
-
50
0
P
5
Figure IS. Dependence of Thermal Sh-ess on Temperature Gradient.
f $!
40
301
in
g
20
L
al
t' lot
0'
'
400
I
I
I
I
300
200
I00
Temp. Gradient AT ( " C )
I
502
DEVELOPMENTS IN PACKAGING DESIGN
In view of the more demanding driving conditions in Europe, increased concern over warranty costs and future trends toward lOOK vehicle mile durability, the total canning system has undergone a number of design improvements, notably [4,12,14,15] (i) (ii) (iii) (iv) (v) (vi)
dual can system with multiple types of insulation; dual entry and exit cones with ceramic fiber insulation; more rigid stiffener ribs; ceramic insulation ring in transition zone; screen-shielded intumescent mat; heatshields with air scoop effect and special stiffeners to minimize resonant vibrations.
A successful package for catalytic converters is one that ensures positive holding pressure on ceramic catalysts, promotes symmetric entry of inlet gases, minimizes radial temperature gradient in the catalyst, and provides good vibrational dampinq under all operating conditions [31]. The above design improvements certainly help meet these requirements. Both a stiffer clamshell design and an effective thermal insulation system are key to good packaging design. Should the catalyst crack due to operating stresses, the successful package will help hold it together without jeopardizing its catalytic activity. The upper limit on holding pressure is dictated by the isostatic strength of washcoated substrates3 which, as mentioned before, depends on substrate/washcoat interaction. Table 7 summarizes the isostatic strength data for four different substrates coated with three different washcoats. It shows that the coated substrate is at least an order of magnitude stronger (>lo MPa or 100 bars) than the maximum holding pressure exerted by the mat at high temperature (10 bars). Moreover, a thick mat with high mount density which ensures positive pressure and excellent insulation will experience minimal erosion and continue to perform reliably; the can will also experience lower temperatures and undergo minimal permanent deformation. The use of internal stiffener ribs was originally intended for wiremesh mounted systems to ensure positive gripping at high temperature. Such internal ribs should be minimized or placed in the midbed region for matmounted systems to minimize frictional strain transfer from can to the monolith. The external ribs are equally effective as stiffeners although they require additional space which is at premium both in the engine compartment and under the chassis. Fortunately, with extra insulation, the can temperature is lower and requires shallower stiffeners thereby alleviating space 3
Or by crush strength along B and C axes.
503
constraints. Another option available to the canners is to use a thicker sheet of stainless steel for higher rigidity which will also enhance its durability.
x.AWL7
Isostatic Strength of Coated and Uncoated Cordierite Substrates (MPa)* --~ --
-
--
With With With Coating C: Coating C2 Coating C3 -~
EX-20 400/6.5 Sq. Cell 3.2" x 5" Oval
9.3
10.0
11.0
EX-20 400/6.5 Sq. Cell 3.18" x 6.68" Wide Oval
8.9
10.9
--
3x-20 300/11 Sq. Cell 4.16" Round
18.9
--
--
11.9
--
18.9
EX-32 236/11 Tri. Cell 3.4" x 5.0" Oval ____---
*
Uncoated
---
-
Above data are based on a sample of 15 substrates per test. NOTE:lMPa = 10 bars
In view of the local gap changes due to internal or external ribs, the mount density and the corresponding holding pressure on ceramic monolith will also experience step changes. Such changes result in localized strain discontinuities in the monolith skin and produce tensile stress in the axial direction. Figure 16 shows the results of finite element analysis of axial stress distribution in the vicinity of the stiffener [31] For maximum durability, this stress should be taken into account along with the thermal stress, and the combined stress should be limited to less than 50% of the monolith strength to prevent potential fatigue degradation [32]. One way to minimize the combined stress is to position the stiffener rib away from maximum thermal stress location, i.e. away from midbed region.
504
monolith
1 I mner con
fiber insulotm
A = 1.38MPo B = 1.05 MPo C=0.71 MPo D = 0.38MPa E = 0.04MPo
outer can
MIN = 0.29 MPo MAX = 1.72 MPc
Fig. 16. - Axial Stress a, in Ceramic Monolith Near the Stiffener Rib.
Fig. 17. - Close-Up View of Dual-Can and Dual-Cone Converter Assembly.
The thermal stresses can be minimized by reducing the radial temperature gradient in the monolith. Figure 17 shows a close-up view of dual-can and dual-cone converter assembly [12,14]. The former system uses a wiremesh mat next to the monolith and fiber insulation in the annular space between the two cans. Contrary to initial expectations, the monolith of this system experiences a lower temperature gradient due to (i) heat conduction from inlet gases to exposed wiremesh and from wiremesh to monolith skin and (ii) minimal heat loss through well-insulated dual can. The heat loss through the single mat layer and can in the latter system is higher and results in a larger temperature gradient; only in the dual-cone region is the heat loss minimized. Consequently, the thermal stresses in the monolith will be larger in the latter system despite the use of ceramic insulation mat In addition, the radial expansion of the inner cone at high inlet gas temperature could increase both the mat density and pressure loading causing deformatlon of inner and outer cones and exposing the mat to gas impingement. This should, of course, be avoided to minimize thermal erosion of the mat lest the support system disintegrate and expose the monolith to vibrational and impact damage. It
505
appears, therefore, that even the dual cone system may require stiffeners to resist deformation at high temperature. Alternatively, the thermal insulation used between the cones should be nonintumescent, e.g. Series MM mat, to limit the pressure build-up at high temperature. Returning to dual-can assembly, while the wiremesh mat plays a beneficial role of minimizing the radial temperature gradient, it does so at the risk of annealing and potential loss of mat pressure which is again undesirable. The use of high weavedensity austenitic steel wiremesh would reduce the risk of annealing. In certain converter assemblies, notably those employing a single can, one or two heatshields are used to protect the adjacent components, floor pan, and ground vegetation from thermalradiation [4]; see Fig. 18. However, their structural design and relative positioning are critical to converter durability. The annular space between the heatshield and can should be maximized to promote both the air scoop effect and convection cooling of the can; the latter can also be achieved by perforating the heatshields. Should the perforations reduce the flexural rigidity and produce unacceptable vibrational noise, the heatshield design and anchoring system may need to be stiffened. The use of acoustical pads to reduce the vibrational noise would defeat systems durability by raising the can temperature and causing excessive deformation; hence the acoustical pads between the heatshield and can should be avoided, if possible [ 141.
MONOLITH MOUNTING MATERIAL I . ,’
n
fi
HEATSHIELD
Figure 18. - Schematic of Converter Assembly Showing Heatshields
506 SUMMARY
A number of new developments in the areas of material composition, cell geometry, wall porosity, substrate contour, washcoat formulation, thermal and acoustical insulation, and clamshell and heatshield design have occurred in recent years to meet the simultaneous requirements of faster light-off, higher conversion efficiency, lower pressure drop, longer durability, lower warranty cost and affordable system cost. Noteworthy among these are (i) thin-wall ceramic substrates with equivalent or improved performance and durability, MAT ceramic substrates for 200°C higher use temperature, and metal foil substrates with equivalent performance and durability at lower use temperature than those of standard cordierite ceramic substrates; (ii) compatible washcoat systems with beneficial interaction with ceramic substrates; (iii) thick and monolithic intumescent ceramic mats for improved insulation; wire screen shielded ceramic mats for minimal thermal erosion; high alumina sol-gel fiber mat for low mounting pressure for metal substrates; and slip-cast ceramic spacer rings for shielding the mat and can in transition region; (iv) dual can and dual cone clamshell designs with multiple insulation for prolonged durability; (v) perforated and stiffened heatshields with air scoop effect for convection cooling of the can. Undoubtably, as we gain new field data and apply this technology to motor cycles, diesel buses and trucks, alternate fuel vehicles, and industrial emissions, additional developments will be needed to combat higher NOx and HC emissions, ash contamination, aldehyde formation and high amplitude/frequency vibrations. ACKNOWLEDGEMENTS
The author is grateful to many of his colleagues at CORNING, 3M Co., Carborundum, Johnson Matthey, AC Rochester, Mercedes Benz, Ford and Leistritz for helpful discussions and to Virginia Doud, Nancy Foster and Irene Tone of R,D and E Labs at CORNING for their assistance in the preparation of this paper.
507
REFERENCES 1. M.R. Montierth and J. P. Day; 7th CIMTEC Conference, Montecatini Terme, Italy; June, 1990. 2. S.T. Gulati; 5e SIMEA SIMPOSIO DE ENGEHARIA AUTOMOTIVA; Sao Paulo, Brazil; September, 1989. 3. S.T. Gulati; SAE Paper No. 850130; February, 1985. 4. P.D. Stroom, R.P. Meny and S.T. Gulati; SAE Paper No. 900500, Detroit, MI; Feburary, 1990. 5. S.T. Gulati; SAE Paper No. 750171; February, 1975. 6. S.T. Gulati; SAE Paper No. 881685; Portland, OR; October, 1988. 7. J. P. Day; SAE Int'l Fuels & Lubricants Meeting, Tulsa, OK,USA; October 22-25, 1990. 8. J.P. Day and D.F. Thompson; 7th ClMTEC World Ceramics Congress; Montecatini Terme, Italy; June, 1990. 9. S.T. Gulati, J.C. Summer et.al.; SAE Paper No. 890796; February, 1989. 10. "3M Interam Heat Expandable Ceramic Mat," 3M Co., St.Paul, MN. 11. S.T.Gulati and R.P. Merry; SAE Paper No. 840074; February, 1984. 12. D. Kattge; SAE Paper No. 880284; February, 1988. 13. ARMCO Praduct Bulletin, 1984. 14. J. Abthoff. H.D. Schuster, F. Nunnemann and W. Zahn; SAE Paper 900267; Detroit, MI; February, 1990. 15. H. Weltens, H. Bressler, and M. Doll; SAE Paper No. 880318; February, 1988. 16. L.M. Sheppard; Ceramic Bull; Vol. 69, No. 6, 1990. 17. H. Yamamoto, H. Hone, J. Kitagawa and M. Machida, S A E Paper No. 900614; Detroit, MI; February, 1990. 18. J.P. Day and R.L. Locker, 1st Int'l Symp. and Exposition Ceramics for Environmental Protection; Koln, W. Germany; December, 1988. 19. J.P. Day and R.L. Locker; U.S. Patent No. 4,855,265; August 8, 1989. 20. M. Nonnenmann; SAE Paper No. 900270; Detroit, MI; February, 1990. 21. M. Maattanen and R. Lylykangas; SAE paper No. 900505; February, 1990. 22. S.T.Gulati; SAE Int'l Fuels & Lubricants Meeting, Tulsa, OK, USA; October 22-25, 1990. 23. S.T.Gulati and R.D. Sweet; SAE Paper No. 900268; February, 1990. 24. S.T.Gulati et. al., SAE Paper No. 880101, Detroit, MI; February, 1990. 25. S.T.Gulati and W.A. Plummer; Thermal Expansion-1973; AIP Proc,No. 17; Am. Inst. Physica, New York, 1974. 26. S.T.Gulati and H.E. Hagy; J. Am. Ceram. Soc., Vol. 61, No. 5-6, 1978. 27. S.T. Gulati; Coming Incorporated Internal Report R-4545; 1971 . 28. R.J. Lee and R.A. Westmann; J. Comp. Mat; Vol. 4, 1970. 29. A. Verma, D.K.S. Chen and D.L. VanOstrom; Mechanical Engineering; Vol. 110, No. 11, 1988, pp 56-61. 30. S.P. Timoshenko and S.Woinowsky-Rrieger, "Theory of Plates and Shells," Sec. Ed., McGraw-Hill, NY, 1959. 3 1. D.K.S. Chen and A. Verma; ASME Winter Annual Meeting; Dallas, TX, November, 1990. 32. J.D. Helfinstine and S.T. Gulati; SAE Paper No. 852100, Tulsa, OK;October, 1985.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control I1 0 199 1 Elsevier Science Publishers B.V., Amsterdam
509
STUDY OF PLATINUM DEPOSIT ON FERRITIC STAINLESS STEEL CONVERSION COATINGS L. ARIES1, P. DE VEYRAC2, M. MANTEL3 and J.P. T R A V E R S E 1 1 - Laboratoire de Recherche sur I'Energie, Universite' Paul Sabatier - 31 062 TO ULOUSE CEDEX 2 - Service De'veloppement UGINE A.C.G., PARIS LA DEFENSE 3 - Centre de Recherche d'UGINE , 73403 UGINE CEDEX ABSTRACT An alternative solution to the alumina washcoat layer on a metal support is investigated. Chemical treatment of the refractory steel surface allows to form a conversion coating. The layers which are about 400 nm thick and present a fractal character, a large specific area and a high porosity, are perfectly apt for impregnation with platinum by electrolysis of hexachloroplatinic acid and for codeposition of platinum with lanthanum oxide. Homogeneous deposit of fine particules of platinum -a few hundred to a few thousand of Angstrom - is obtained. Codeposition with lanthanum oxide allows to reduce coalescence and diffusion of platinum during ageing. Problem of the formation of a superficial probably alumina film appears after heat treatment in the air at 1000 "C for a long period.
INTRODUCTION
We €ooked for an alternative solution to the alumina washcoat layer on a metal support [l]. We used conversion coatings on refractory ferritic steels. They were ferritic stainless steels with seventeen percent chromium stabilized with zirconium, titanium or niobium in order to preserve the ferritic structure at all temperatures. Table 1 Chemical Composition of F18 A1 Steel
Fe 78.30
I
I
Cr 17.43
I I
A1 1.77
I I
Weight per cent Si I c u I Ti 0.75 I 0.38 I 0.36
I I
Ni 0.18
I I
Mn 0.13
I
I
Mo 0.05
They also contained about 2 % aluminium to resist oxidation at high temperatures. They were about nought point one millimeters in thickness. Their aluminium content was lower than that of the steels usually used as a support for the alumina washcoat (4 %). We are aware of the problems that this can bring about for thin foils since the substrate plays the role of
510
aluminium reserve susceptible of being oxidized. Levels of about 2 % could possibly be insufficient to prevent aluminium exhaustion. However, these steels present economic advantages over the steels with 4 % aluminium. A conversion coating is obtained by the chemical tranformation of the surface of the steel. At the last congress at Louvain [2], we presented a study of the influence of the preparation parameters on the characteristics of the layers. It is a microporous medium with a catalytic activity of its own. We have developed a technique to impregnate the layer with platinum and the layer plays the role of support although it can also contribute to the catalysis by its composition and texture. PRESENTATION OF THE CONVERSION LAYER
An original method for catalyst preparation has been developped [3] [4]. The process involves either anodic oxidation or chemical treatment of ironchromium based alloys. Supported catalyst can be prepared in one main step from the substrate which furnishes constitutive elements of the coating.
/
Fig. 1 Typical anodic polarization curve of stainless steels in sulfuric acid solutions. Eoc :natural corrosion potential, Ep :passivity potential, Er :rupture potential. In the chemical treatment the surfaces were prepared by dipping of the steel into a bath. One of the main conditions of the treatment is the fitting of the electrode potential of the sample to the value of the natural corrosion potential of the steel (Eoc) in the active state Fig.]. It is then necessary to
511
control the surface potential during the treatment having previously determined the electrochemical characteristics of the interface metal solution, by means of polarization curves. For some alloys, this condition of potential is naturally fulfilled for the treatment baths used. Generally, the potential can be adjusted to the required value by cathodic activation of the surface in the treatment bath with the help of a current generator and counter electrode playing the part of anode. The coatings were prepared in an acid bath with suitable additives, particulary substances containing chalcogenides : Sulphur seems to give the best results, and it is preferable to put sodium sulphide or sodium thiosulphate in the bath. It can be profitable to add a corrosion inhibitor specific to the alloy and the treatment bath to further control the thickness of the coating. The presence of propargyl alcohol reduces the aggressiveness of the bath and leads to a decrease in the coat thickness. The compositions of baths used in this part of the study are given in table 11.
Sulfuric Acid (vol%)
Sulfured Species
(all)
Propargyl Alcohol (ml/l>
Temperature
("C)
Treatment Duration (min)
10
60
45
Na2S203.5H20 5
1
After the preparation of the conversion coating, the samples were washed with water. They were then dried in an oven at 90 "C or dried in ambient air for about 10 minutes. After rinsing, in some cases, the coatings were subjected to chemical oxidation treatment, in an aqueous bath or to heat oxidation treatment in air. This type of treatment has proved to be suitable for making coatings on substrates of very diverse types. It has been adapted, without any major difficulties, for the use on refractory steels. The conditions of pretreatment of the steel before immersion in the bath (degreasing and also possibly harsher cleaning) and the rinsing after pretreatment play a significant role. The optimal conditions determined are rinsing in the presence of agents to improve wettability before drying. The bath composition, temperature and duration of treatment used lead to coating thicknesses of about 400 nm with a high fractal dimension: 2.43 as opposed to 2.27 for the thinner coating studied elsewere [2][5]. The coatings are sufficiently porous for impregnation with platinum. Their composition is similar to that of the ferritic steel coatings as shown in Fig 2 [4]. Here, the main oxide component is a magnetite substituted with chromium three (Cr 111) and aluminium three (A1 111).
512
ME
STEEL ATOMIC PROPORTION
Fig. 2 : Representation of the composition of a typical ferritic steel coating : the scheme gives the atomic proportion against the depth. Atomic proportion is the ratio between the number of metal atoms in the different compounds to the total number of metal atoms in the diflerent compounds. (A) :the superficial film, ( B ) :the external zone (deep zone), (C) :the internal zone (deep zone), I :domain of Fe3+ and Cr3+ oxide and hydroxide, II :domain of Cr3+ substituted magnetite, III :domain of metallic iron and chromium (alloy), N :domain of metal sulphate (s), V :domain of metal sulphide (s). DEPOSITION OF PLATINUM
For impregnation with platinum alone and subsequent study of the deposit, the electrical conductivity of the coating was used to deposit the platinum by electrolysis. Aqueous solutions of hexachloroplatinic acid were used for cathode deposition. Typical conditions were : 10 grams per litre of hexachloroplatinic acid An electrolysis current of ten milliamps per square centimeter for an electrolysis duration of ten seconds. This gave about 1.5 grams of platinum deposited per square meter of steel sheet. The resulting material, examined by scanning electron microscopy, is shown in Fig. 3 , (top left) where the general appearance of the coating is seen to be hardly modified by the presence of platinum. EDAX analysis - on Fig. 3 (top right) - shows the contribution of platinum which is only very weak here owing to the depth of penetration of the probe which was about 1
513
Fig. 3 S.E.M. micrographs and microprobe analysis of conversion coating with non uniform platinum distribution.
514
pm. The figures in the centre and at the bottom illustrate a problem typical of this type of deposition. The platinum can have a non-uniform distribution. This was later overcome as shown in Fig. 4 where the morphology of the platinum grains is shown. Their size ranges from a few hundred to a few thousand angstroms. The large white particle is a grain of titanium nitride (covered with titanium carbite). The platinum is also deposited on its surface. The composition of the coating was analysed by SIMS. The profiles obtained are shown in Fig. 5. Oxidation is progressive from the substrate up to the surface. The ionic intensities of chromium, iron and aluminium show a corresponding decrease. There is, however, an enrichment of titanium. These analyses confirm the presence of platinum at the surface. The influence of qgeing was studied on samples which had been subjected to heat treatment in air for 24h at temperatures ranging from 400 to 1000 "C. Scanning electron microscope observations of samples heated in the air at 1000 "C for 24h are showp Irr Fig. 5 along-side the corresponding EDAX analyses. Coalescence of the grains of platinum occurs and they are roughly five times larger than before. Simultaneously, analysis shows an increase in the level of A1 related to oxidation of the substrate.Corresponding SIMS analysis of oxygen, aluminium aqd platinum indicates diffusion of platinum towards the substrate which, at 1000 "C, becomes quite extensive, Fig. 7. Diffusion of aluminium towards the surface from 800 "C was also confirmed. The oxidation state of platinum was shown by XPS to remain at zero, the metallic form persisting even after heating at 1000 "C for 24h. CODEPOSITION OF PLATINUM AND LANTHANUM OXIDE
In order to resolve the problems encountered - that is to say diffusion and dispersion of fie platinum and coalescence of the grains - we modified the treatment and electrolysed the platinum in the presence of lanthanum oxide dispersed in the medium ( 6 grams per litre) while the other conditions remained unchanged. At the same time the dispersion of the platinum is improved. The shape of the grains appears particularly clearly on the scanning electron micrograph, Fig. 8 , where the deposit can be seen on grain of titanium nitride. In fact, platinum distribution can be made more even in different ways : electrolysis conditions clearly play an important role. In any case the coatings obtained always show perfect adherence. SIMS analysis of the coating shows that a codeposit of platinum and lanthanum is obtained Fig. 9, top. On ageing testing Fig. 9, bottom SIMS analysis of a sample aged in air for 24 hours at 1000 "C shows a considerable reduction of the diffusion of codeposited platinum towards the substrate. Also the coexistence of platinum and lanthanum is maintained in a similar way to that observed before treatment. Aluminum is still seen to diffuse towards the surface but less than through layers coated with platinum alone. The shift observed along the
515
BIG
OUND T H
PARTIC
T E PART1
Fig.4 S.E.M. micrographs and microprobe analysis of uniform distribution of platinum grains on conversion coating.
516
INTENSITY
(a.u.1
IlEEEtl
EROSION TIME (sec)
' INTENSITY
E
E
734
¶W
E R O S I O N T I M E (sec)
Fig.5 S.I.M.S. intensity against sputtering time for constitutive elements of platinum impregnated coating.
517
Fig. 6 S.E.M. micrographs and microprobe analysis of platinum impregnated coatings after heat treatment in the air at 1000 "C for 24 h.
518
4tEdEE O X Y G E N
3IIW
_ _ _ _ _ _ 400°C 800°C
......1000°C
....._. ............. I
T L H ~
(secl --i
519
Fig. 8 S.E.M. micrographs and microprobe analysis of impregnated coating after codeposition of platinum and lanthanum oxide.
520
Cr x 1
Fe x 1
0
1200
2400
3600
Al x 0,l
4800
6000
7
TIME (sec)
Fig. 9 S.I.M.S. intensity against sputtering time for constitutive elements. of a coating after codeposition of platinum and lanthanum oxide. bottom :of the same coating aged in the air at 1000 "Cfor 24h.
m:
52 1
erosion time scale is attributed to the formation of very thin insulating layer but it was not possible to determine either its thickness, since it was extremelly thin, or composition. We can however suggest that it was mainly superficial alumina. We studied the behaviour of uncoated steel and of coated steel, with and without the layer of platinum, on oxidation at 1000 "C.The results are shown in Fig. 10 where the quantity of oxide formed is shown against time. From results given by thermogravimetry over a shorter time scale, oxidation is seen to be slowed down in the presence of platinum codeposited with lanthanum oxide
P
18Al
1W
50
24
0
18 Al
+
conversion
-k
18Al
+ conv +
TIME (h) Pt
Fig. 10 Weight of oxide formed in the air at 1000 "C against treatment time for alloys with and without coating. CONCLUSION Conversion coatings optimized for the purpose are perfectly apt for impregnation with platinum and for codeposition of platinum with lanthanum oxide. No degradation of the intrinsic properties of the coating was observed. It remains microporous and still presents a fractal character both after preparation of the layer and after heat treatment. Secondly, we have demonstrated the feasability of making a homogeneous deposit of fine particules of platinum. The material obtained
522
with codeposition conditions shows a relatively high chemical stability. The coalescence of the platinum particles was reduced as was the diffusion of platinum toward the substrate. Lastly a problem remains : that of the formation of what is probably a superficial film of alumina after heating in air at 1000 "C for long periods. We have not yet tested the catalytic performance of the coating after this type of treatment but it is possible that the film creates a barrier to the diffusion of gas molecules and thus to catalysis. Further studies will be performed on this material which seems to have retained its potential but heat treatment has to be restricted to shorter time at 1000 "C or to lower temperature.
REFERENCES 1
2
3 4 5
PRATT A.S. and CAIRNS J.A. Platinum Metals Rev., 2,74-83.( 1977) L. ARIES, A. KOMLA and J.P. TRAVERSE Roc. Vth Int. Symp. "Scientific bases for the preparation of heterogeneous catalysts". Louvain La Neuve. 3-6 sept. 1990. L. ARIES and J.P. TRAVERSE Brevet Fr no 86.18124 (1986) L. ARIES and J.P. TRAVERSE Brevet Fr no 88.08402 (1988) PCT/FR no 89.00295 L. ARIES, P. FORT, J.A. FLORES and J.P. TRAVERSE Sol. Energy Materials, l4,143-159.(1986)
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
523
MONOLITH SUBSTRATE EFFECTS ON CATALYST LIGHT-OFF T. S. Jasper (11, K. Robinson (2) , D. Anderton (2) and D. H. Cuttler (3) (1) Coventry Polytechnic (2) The Institute of Sound and Vibration Research, Soushampton University (3) Jaguar Cars Ltd ABSTRACT A test rig has been designed and built to examine the effects of catalyst construction in isolation. The test method utilises a fully stabilised engine with constant manifold downpipe temperature and mass flowrate. The catalysts examined were chosen to emphasize the thermal aspects of catalyst design. The light-off performance of four different catalysts was investigated to determine the effects of cell density (for three metallic catalysts) and monolith material (ceramic versus metallic catalysts). The cell density determines the thermal inertia and, for the metallics, it was found that the thermal inertia was directly related to the light-off performance - low cell density had low thermal inertia and produced faster light-off. Direct comparison between 400 cells/in2 ceramic and metallic monoliths showed the ceramic to be superior on light-off, despite a higher thermal inertia. The low thermal conductivity of the ceramic monolith kept the centre of the monolith at a higher temperature than the metallic and produced a better light-off performance.
INTRODUCTION In general, a large proportion (80% or more) of regulated emissions are emitted during the early period of the Federal US emissions test while the catalysts are not hot enough to work effectively. One of the objectives of a designer, therefore, is to minimise the time taken for the catalysts to attain their working temperature and thereby achieve significant reductions in toxic emissions. In the automotive industry, the transition from a non-working catalyst to a working catalyst is loosely termed "light-off '. During the design of new systems, choices have to be made in respect of substrate material, construction, cell density, volume, aspect ratio and precious metal coating. These choices will be influenced by factors such as mass flowrates (space velocities), gas temperatures, gas composition, total catalyst system configuration and catalyst durability (both chemical and physical), as well as cost, weight and size. Recent advances in substrate materials, such as thin-walled ceramics and more specifically metallic monoliths have further complicated the designer's choice. Claimed advantages of metallic over ceramic monoliths are lower
524
pressure drops for the same cell density, greater surface to volume ratio offering more effective conversion (or the same conversion for a smaller catalyst volume) and greater thermal and mechanical shock resistance. The high thermal conductivity of the metallic substrate is claimed to enhance lightoff and improve resistance to high temperature ageing [l]. Improvements in manufacturing techniques have opened up many more design options with an increased flexibility in construction. Balancing these advantages have been the drawbacks of higher cost, doubts about washcoat integrity at high temperatures under vibration and counter-claims about the light-off performance. Other work [2,3,4,5,6] has been conducted on parametric comparisons between ceramic and metallic substrates. These studies mainly address results obtained when using the 1975 Federal Test Procedure (FTP75) rather than focusing on the early stages and hence the mechanisms of light-off. Pelters et a1 [S] describe a development and application programme for a particular vehicle in which the balance of design considerations favoured the selection of a metallic catalyst. Faster catalyst light-off was listed as one of the advantages of this choice. Nashizawa et a1 [4] compared a conventional ceramic catalyst with a metallic catalyst of the same cell density but 24% less volume. Over the FTP cycle, little difference was observed in light-off performance. The work by Hawker et a1 [ 6 ] makes a similar comparison, and noted that there was a tendency for the metallic to achieve 70% conversion efficiency faster than the ceramic, but longer to achieve 90% efficiency. Jaguar Cars Ltd, in conjunction with Bosal Research Ltd and Johnson Matthey PLC sponsored a research project with the Automotive Group at the Institute of Sound and Vibration Research, Southampton University [7,8]. This research was designed to investigate the effects of cell density on the light-off performance of three metallic monolith catalysts and compare their behaviour with that of a conventional ceramic monolith catalyst. The catalyst volumes, cell densities, chemical and physical properties are shown in Table 1. In order to examine the effects of substrate material and construction on light-off, the complex engine warm-up and catalyst light-off processes were simplified. The experimental technique developed for this project used a fully warm engine producing a known gas temperature, flowrate, and composition. The test conditions were chosen so that the light-off performance of the catalyst was dominated by the response of the monolith to the changing inlet gas temperature. TEST EQUIPMENT AND PROCEDURE
The test apparatus consisted of a Jaguar 3.6 litre six cylinder in-line engine, coupled to a water-brake dynamometer. The exhaust system had two separate branches with manually operated ball valves to control the flow path. The test facility is shown schematically in Figure 1.
525
I IM
3pr
Ert
L L i =
0
m CI
Figure 1. A Schematic Representation of the Test Rig.
526
Prior to the start of each test, the exhaust gas was directed down the bypass section whilst the engine and coolant reached normal operating temperatures. The engine, now under closed-loop fuelling control ( h = l), was then set at the required speed and load for the test condition. A stabilisation period of fifteen minutes ensured that the exhaust gas temperature achieved a steady state condition. At the start of each test, the exhaust gas was directed from the bypass section to the test section while the time histories of the following were recorded on a data logger: (i) the %CO concentration of the gas exiting the catalyst; (ii) the gas temperature into the catalyst; (iii) the gas temperature exiting the catalyst; (iv) the monolith temperature, mid-length, centre axis; (v) the monolith temperature, mid-length, mid radial point; (vi) the exterior surface temperature of the converter case. The thermocouple positions are shown schematically in Figure 2. The length of each test was 200 seconds, by which time the CO conversion efficiencies were around 75% for all test samples. The use of steady state engine operation with a stabilised exhaust gas virtually eliminated variations in gas composition, flow rate and temperature. Two series of tests were conducted, each having a different gas temperature and mass flowrate (see Table 2). Cenaebed Thermocouple I
Skin Thermocouple
Gas flow
Half-Wd y-ill-bed -lkmKmuplC
catalyst
Monolith
Figure 2 . A Schematic Representation of the Thermocouples in the Catalysts.
527
TABLE 2 Engine Speed BMEP Downpipe Exhaust Mass Space Velocity Flow(g/s) (hr-1) (bar) Temp("C) (rpm) Series One 1500 1.8 465 18.0 175000 Series Two 1500 3 .O 510 23.4 2 loo00 Figure 3 shows the catalyst inlet gas temperatures for the four catalysts at the Series One and Series Two test conditions. The repeatability of the test procedure vindicated direct comparisons between the monoliths, as each was exposed to the same mass flowrate and inlet temperature - time history. Also shown on Figure 3 is the inlet gas temperature profile for the downpipe catalyst of a Jaguar vehicle over the first part of a Federal test cycle. RESULTS AND DISCUSSION
The temperature - time histories of the four catalysts at the Series One test conditions are shown in Figures 4-7. The Series Two test results, which are not shown, exhibit similar trends to the Series One results but the differences are much less marked. The post-catalyst %CO emissions for all the tests are shown in Figures 8 and 10. Figures 9 and ZZ show the cumulative %CO emissions. Cell Densitv Effects Close examination of the temperature - time histories for the metallic catalysts Figures 4 , 5 & 6 along with the CO conversion curves Figures 8 & 1 0 reveals that the 200 cell metallic catalyst has the fastest light-off performance. In terms of CO conversion, the 300 and 400 cell catalysts are very similar in the first series, but the 300 cell is much closer to the 200 cell in the second series. As most of the exothermic chemical heat that is released by the catalyst comes from the conversion of CO into C02, the quicker the reduction in CO, the quicker the rise in catalyst temperatures, both centre-bed and half-bed (and, consequently, gas-out). The 200 cell metallic shows the fastest rise in centre-bed and half-bed temperatures and the best light-off performance. The 400 cell has much slower and more rounded rises in bed temperatures and consequently a poorer lightoff performance. The 300 cell metallic has a faster rise in centre-bed temperature (though not as fast as the 200 cell) but a slower rise in half-bed temperature than the 400 cell. From the %CO emission results of the first series, it appears that the increased conversion through the centre of the 300 cell metallic monolith (owing to the higher temperatures present) balances the poorer conversion through the outer regions. The result is a light-off
Figure 3. Inlet Gas Terperature Profiles Gas I n To S t a r t r r C a t a l y s t s CXJ40, FTP75 Tost)------------
,
650 600 550 500
-
3
450 400
0
L D
350
L 0
n.
f 300
c
250 200 150
100
..0 A
I
I
40
I
1
I
80
I
120
Tino ( r o o s l
I
I
160
I
200
500
I
I
l
288 Cell Hetallic
l
I
500
I
I
I
388 Cell Hetallic
1
I
450
400
350
-z
300
L
258
L
LL
f
200
150
100
50
8
11111111111
8
48
120 T i r r (mrcm)
88
Figuro 4
160
208
A
0
Series One Tests
40
80 120 Tinr ( s r c e )
Figuro 5
168
208
530
Z CO Enissions For A 1 1 Catalysts
0.70
I
I
I
I
I
I
I
-
I
Cunulatiue X CO Enissions For A l l Catalusts
10
I
1
I
I
1
I
I
I
I
I
1
Hetallic 208
/’
9 lletallic 380 ----------
e ,613
netallic 480 --8
0.50
Ceraric 488
7 m
c 1.4 0
m
{
.li: 13.48
w
..4
0
I
u % 5
m ..4
C
W 0 0 %
6
C
m
>
.4
0.30
Y
O
3
4
0
u
3
0.28
2
e.ie 1
e .ee
e
40
88
128
168
2ee
e
I
e
Tire (secsl
I
413
I
I
88
I
I
128
1
1
168
1
2130
Tire (secs) ul
Figurc 8
Series One Tests
Figure 9
w
c
I
E
0.60
f\ \
;\:'
0.50
-\i'\\\
1
I
I
I
I
1 -
I
I
I
1
I
1
netaiiic 2ee
Metallic 288
------
fletallic 380
9 n e t a l l i c 388
__________
__-
Metallic 488
Metallic 488
---
8 -
Ceranio 488
I
I
I
I
1
-
Ceraric 488
7-
/A
m
I '
c 0
m
.r(
.Z
6
-
/
///
-
533
performance very similar to the 400 cell. In the second series, there was much less difference between all the half-bed and centre-bed temperatures with the result that the 300 cell metallic performed more like the 200 cell than the 400 cell. The results clearly indicate that the light-off performance of metallic monoliths under these test conditions was related to their cell density. As the cell density decreased, the light-off improved. Table 1 shows that the heat capacity (i.e. the thermal inertia) of the monoliths increased with cell density. The differences in light-off performance were attributed to this change in the thermal inertia. Active surface areas and precious metal loadings have been closely linked to light-off performance [9,10]. Although lower cell density monoliths have lower active surface areas upon which catalytic activity may take place, the precious metal loadings were the same for all the catalysts. On a lower cell density monolith, the precious metal was concentrated over a smaller area so the effect of active surface area upon light-off was minimal. Total active surface area only became important as the steady state was approached. This was particularly evident in the Series Two tests, when the CO conversion became limited by the mean residence time of the gas over the surface see Figures 10 and 1 I. Substrate Material Effects Comparison of the temperature profiles Figures 6 & 7 and also the CO conversion curves Figures 8 & 10 between the 400 cell ceramic and the 400 cell metallic clearly show that the 400 cell ceramic monolith has a superior light-off performance when compared with the equivalent 400 cell metallic monolith. Table 1 shows that the two catalysts have similar washcoat loadings, catalyst active surface areas and volumes, but rather different masses, thermal conductivities and heat capacities. The metallic had more mass but had a lower heat capacity owing to the much higher specific heat of the ceramic substrate. The ceramic centre-bed and half-bed temperatures rose significantly faster than those of the metallic, particularly from the point when they exceeded the inlet gas temperature. Over the latter portion of the test (160-200 secs), the 400 cell ceramic had a higher centre-bed but lower half-bed temperature than the 400 cell metallic. The ceramic thus had a larger radial temperature gradient than the metallic (c$ Figures 6 and 7 ) in spite of the matting wrapped around the ceramic monolith for packaging purposes. Vehicle packaging constraints usually force short entry cones with large included cone angles (70 or more), which are less than ideal. The gas flowrates encountered in automotive exhaust systems, coupled with these large cone angles, mean that the gas cannot flow smoothly around the sharp angles and flow separation results [11,12,13]. A plume of gas is directed at the centre of the monolith with recirculation zones in the outer parts of the cone. The result
534
is a significant flow maldistribution with high flow rates through a central 'core' of the monolith and lower flowrates in the outer regions. Most of the gas, and therefore most of the heating, is directed at the centre of the monolith. The higher thermal conductivity of the metallic substrate allows this heat to be conducted radially outwards, preventing any heat build-up. The ceramic substrate behaves in the opposite manner by effectively insulating the central core and allowing the temperature to rise rapidly. The higher core temperatures facilitate earlier exothermic chemical reactions, producing more heat and accelerating the light-off process. The results in Figures 6 and 7 are consistent with the ceramic catalyst rapidly generating a hot core which promoted early chemical heat release and resulted in a faster light-off. The metallic monolith, in contrast, did not produce a hot core but rather, it spread the heat through the whole monolith with a consequently slower light-off. Design Implications On a large vehicle such as the Jaguar XJ6, the catalysts used in these tests would make up approximately 20% of the total catalyst volume. The choice of a suitable starter catalyst will be determined by the position of the main catalyst. With the main catalyst positioned close to the starter (and both close to the manifold), the main catalyst will light-off in a reasonably short time (roughly 30 seconds). The starter catalyst should therefore be small with a fast light-off. The 200 cell metallic has these desirable light-off characteristics. Conversely, if the main catalyst is remote from the starter, the light-off of the main unit will take considerably longer (100 seconds or more). The starter catalyst must be able to achieve a rapid light-off and also be capable of dealing with the engine-out emissions (without breakthrough) until the main unit is working. For such a system, the 400 cell ceramic may be a better compromise. Catalyst performance deteriorates with ageing, and the various substrates will age differently. Any flow maldistribution will concentrate the ageing (both thermal and poisoning) to the central core of the monolith. The metallic, with its high thermal conductivity, will suffer less from the thermal ageing effects in the core than the ceramic. Thus, the characteristics of lightoff will change in different ways for the two types of monolith material during their working life, influencing a designer's final choice.
535
CONCLUSIONS A test method which can be used to rank the light-off performance of exhaust gas catalysts has been successfully developed. The physical design of the monolith substrate has a significant effect on the light-off performance. For a metallic substrate, it was found that the changes in cell density greatly affected the light-off performance. The cell density determines the mass of metallic substrate and therefore, all other things being equal, the thermal inertia. The light-off performance improved as the thermal inertia was reduced, i.e. as the cell density was lowered. Direct comparison between 400 cellsfin2 ceramic and metallic monoliths of similar dimensions showed the light-off performance of the ceramic to be significantly better. This difference was attributed to the poor thermal conductivity of the ceramic substrate effectively insulating the central core of the monolith. The resulting higher temperatures produced a faster light-off.
References 1 Nonneman, M., 'New High-Performance Gas Flow Equalizing Metal Supports For Automotive Exhaust Gas Catalysts', SAE 900274 2 Nonneman, M., 'Metal Supports for Exhaust Gas Catalysts', SAE 850131 3 Dulieu, C. et al, 'Metal Supported Catalysts for Automotive Applications', SAE 770299 4 Nishizawa, K. et al, 'Development of Improved Metal Supported Catalyst', SAE 890188 5 Pelters, S. et al, 'The Development and Application of a Metal Supported Catalyst for Porsche 91 1 Carrera 4, SAE 890488 6 Hawker, P. et al, 'Metal Supported Catalysts for Use In Europe', SAE 880317 7 Jasper, T. S., 'An Investigation Into The Parameters Affecting Catalyst Light-Off, M.Sc. Thesis, The ISVR, Southampton University, (1988) 8 Robinson, K., 'An Investigation Into Parameters Affecting Light-Off Performance of Exhaust Gas Catalysts' M.Sc. Thesis, The ISVR, Southampton University, (1989) Zygourakis, K. and Becker, E. Robert, 'Design Options for Rapid Light-Off of 9 Monolithic Catalytic Converters' submitted to Ind. Eng. Chem. Research Journal (1987) 10 Summers, J. et al, 'Improvements in Converter Durability and Activity via Catalyst Formulation' SAE 890796 1 1 Howitt, J. and Sekella T., 'Flow Effects In Monolithic Honeycomb Automotive Catalytic Converters', SAE 740244 12 Lemme, C. and Givens, W., 'Flow Through Catalytic Converters - An Analytical and Experimental Treatment', S A E 740243 13 Wendland, D. and Matthes, W., 'Visualisation of Automotive Catalytic Converter Internal Flows', SAE 861554
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
537
COMPARISON OF METAL FOIL AND CERAMIC MONOLITH AUTOMOTIVE CATALYTIC CONVERTERS G.L. Vaneman AC Rochester Division - General Motors Corporation ABSTRACT
Metal foil substrates are being promoted as the substrates of choice for automotive catalytic converters in the European market. Thinner walls are seen to offer both lower flow restrictions and improved catalytic performance, while metal's thermal characteristics are seen to benefit both performance and durability. AC Rochester has been investigating the advantages and disadvantages of metal foil monoliths for some time and recently developed a unique low cost design that is now in low volume production for @el. An extensive test was recently conducted to compare the aged performance of various metal foil and ceramic monolith converter designs. The herringbone-corrugated ACR metal units outperformed equivalent plate-fin metal and conventional ceramic units and approached the performance of 25% larger ceramic converters. However, metal's thermal characteristics may require the addition of insulation to compensate for rapid cooling during the ten minute shut-down and hot restart portion of the 23 cycle US FTP test. It was also found that tailpipe emission numbers and converter efficiencies may not accurately judge converter performance when the pollutant levels from the test engine are not constant, but vary with converter type.
BACKGROUND
Ceramic pellets, and more particularly ceramic monoliths, have been the substrates of choice for automotive catalytic converters over the past fifteen years. Recently however, many papers [ 1-41 have been published promoting certain advantages of metal foil substrates over conventional ceramic monolith substrates. Some of these advantages were seen to be especially important to the new European catalytic converter market. In particular, the advantage of thinner walls and the resulting lower solids fraction of the foil substrates should provide reduced flow restriction, thus promoting better engine breathing and increased power output. Alternatively, the cell density of foil units could be increased while holding the solids fraction and flow restriction constant, thus increasing the geometric surface area and thereby improving catalytic performance and reducing emissions. The significantly higher thermal conductivity of the metal foil should improve converter durability by quickly dissipating thermal "hot spots" produced by engine control problems, thereby decreasing both catalytic and
538
structural thermal damage. It has also been suggested that the higher thermal conductivity and lower heat capacity of the metal foil units should decrease emissions by reducing the time required to heat a cold converter to its lightoff temperature. However, ceramic monolith proponents have argued that the lower thermal conductivity of ceramic monoliths will help heat from the initial exhaust gas remain in the catalyzed washcoat layer, thus producing faster light-offs for ceramic substrates. Metal monoliths are also promoted as being more robust and less fragile than ceramic monoliths, and as being a better match for the thermal expansion of the converter shell. This eliminates the need for a mat support system, greatly simplifying the packaging of the substrate in the shell, thus offering both lower cost and improved converter durability. However, metal foil monoliths have also demonstrated several disadvantages that have limited their use in primary underfloor applications. Foil units inherently have substantially higher material costs and they also typically weigh more than their ceramic monolith counterparts. The nonporous nature of metal foil, coupled with the mismatch in thermal expansion between the foil and the ceramic washcoat, contributes to a washcoat adhesion problem during thermal cycling. In addition, the low strength of the foil at high operating temperatures has led to structural deformation problems in certain applications. DESIGN
AC Rochester, the automotive catalytic converter manufacturing division of General Motors, has been actively investigating metal foil substrates for more than ten years. The goal of this activity was to fully evaluate the manufacturability and costs of metal units while extensively investigating the performance and possible size reduction of foil substrates. After exploring many alternatives, the ACR design for metal monoliths took a different direction than that followed by most other foil substrate manufacturers. Following upon earlier work conducted at General Motors Research Laboratories, ACR adopted a whisker-producing oxidation treatment of the iron-chromium-aluminum foil that dramatically improved washcoat adhesion. In order to incorporate this treatment, and to avoid the high costs of brazing used by most manufacturers to rigidize their foil substrates, a design was adopted that involved the cutting, stacking, and interlayer welding of the foil sheets. This is then followed by the addition of retainer end rings that can subsequently be welded into pockets in the converter shell, as shown in Figure 1. Another very significant design change occurred when the conventional plate-fin (PF) foil corrugation pattern was replaced with the unique herringbone (HB) pattern, also shown in Figure 1. Unlike the PF design, in which alternate foil layers are uncorrugated and simply serve to separate the
539
corrugated "fin" layers, all foil in a HB unit is corrugated. "Nesting" between HB layers is prevented by reversing the HB pattern on alternate layers of the foil. By thus eliminating the need for separator sheets, the HB design uses approximately 30% less foil, thus offering a substantially lower solids fraction substrate that is therefore much less restrictive and significantly lower in weight. It also offers faster light-off and is significantly lower in cost.
AC ROCHESTER METAL MONOLITH CATALYTIC CONVERTER
OUTLET
HERRINGBONE SUBSTRATE
v-
ANGLED
CORRUGATIONS
Figure I , - Converter design using stack of H B foil layers captured in end rings welded into shell pockets.
Table I is a physical comparison of typical ceramic, PF, and HB monolith substrates and it reflects many of the differences described above. It is of interest to note that the bulk density of the HB design is quite low, indicating that such a converter would weigh approximately the same as a typical ceramic unit. And because of their differences in specific heat, the heat capacity of the HB design is approximately half that of the ceramic design. It should also be noted that the geometric surface area of all three designs is
540
approximately the same after washcoating. This is due to the significant losses suffered by the closed cell ceramic and PF designs when the washcoat slurry fills in the corners of the cells due to surface tension effects.
Table 1 PHYSICAL COMPARISON (Metal Foil vs. Ceramic Substrates)
Cell Density (#/cm2) Wall Thickness (mm) Solid Fraction Material Density (g/cm3) Bulk Density (g/cm3) Material Porosity (%)
STD. Ceramic 62.0 0.16 0.24 1.70 0.41 35
Plate Fin Metal Foil 62.0 0.05 0.097.10 0.63 0
Herringbone Metal-Foil 43.4 0.05 0.06+ 7.10 0.43
Geom. S.A. (cm2/cm3) - Washcoated
28 22
33 23
25 23
Specific Ht (KJKGOC) Heat Capacity (KJLOC)
1.05 0.43
0.5 0.32
0.5 0.22
Thermal Exp. (Rel.) Thermal Cond. (Re].)
1 1
17 20
17
0
The lack of discrete closed cells in the HB pattern, coupled with the stacked layer design, allows lateral internal flow of exhaust gases that contributes to the more complete utilization of the entire substrate, as well as to a more uniform distribution of temperature and flow. This contributes to both lower flow restriction and to reduced thermal stresses during thermal cycling. The unique HB pattern also offers better gas mixing and the disruption of gaseous boundary layers, thus contributing to both improved heat transfer and to better catalytic activity and reduced emissions. EXPERIMENTAL
Even with this lower cost herringbone-plus-retainer ring design, it became clear that only a smaller foil substrate with reduced noble metal content could successfully compete against ceramic monoliths for the primary underfloor catalytic converter market. Although some promising results had been seen in certain performance tests, it was difficult to consistently quantify a size/performance advantage for metal foil units, even for the new HB design. It was therefore decided to conduct a. comprehensive comparison test between various metal foil and ceramic monolith converters. This statistically designed experiment, described in a recent FISITA
54 1
paper [ 5 ] ,had three primary goals: first, to compare a standard ACR 1.4L PF metal foil unit to both an equivalent ceramic unit and to a larger 1.8L ceramic unit containing proportionately increased amounts of noble metals; second, to compare the HB design to the PF and the ceramic units; and third, to compare a new ACR washcoat technology to a conventional state-of-the-art washcoat applied by a commercial coater using PF substrates. A summary of the physical characteristics of the five converter types involved in this test is shown in Table 11. In all cases, noble metal loading was held constant at 3Og/ft3 of Pt:Rh at a ratio of 7:l. Note that flow restrictions, as reflected by pressure drops measured at approximately 480OC and a flow rate of approximately 23 s/sec, are directly related to the solids fraction values. To provide statistical validity, five randomly selected samples of each of the five different converter substrate types were carefully built, aged, and tested. Aging was done in random order on a group of 4.3L engine dynamometers running a severe accelerated schedule designed to simulate 80,000+ kms of customer driving in 100 hours. Performance was then measured on several different tests, often repeated to yield more accurate average values. Overall converter performance is probably best measured on an actual driving test such as the USA 23 cycle FTP schedule. That composite transient schedule includes a cold start, various urban-suburban driving cycles, and a shut-downhot restart. On that test, converter performance is determined by some complex combination of chemical kinetics and mass transfer effects. The mass transfer effects will in turn reflect the thermal characteristics of the engine, the exhaust system, and the converter itself. RESULTS
Table I11 summarizes the results from the FTP chassis dynamometer performance tests conducted on all 25 aged test converters. Each value in the Table is the average of 15 FTP tests (3 on each of 5 converters) run on the same 1.8L 'J' car by the same three drivers. Comparing the resulting tailpipe "bag" numbers for HC, CO, and NOx, the following conclusions were drawn: (1) the ACR washcoated PF unit was essentially equivalent to the PF unit coated by the commercial coater, thus proving the quality of the ACR washcoat technology; (2) both PF units had performances that were marginally worse than the equivalent 1.4L ceramic units, showing higher HC and NOx emissions and somewhat lower CO emissions; and (3) the HB units were judged better than the PF units and, due to their lower CO emissions, were also judged marginally better than the equivalent ceramic units. However, none of the smaller 1.4L metal converters approached the overall performance of the larger, more heavily loaded 1.8L ceramic units, although the metal units did approach the CO level of the larger ceramics.
TABLE I1 Substrate Summary From ACR Metal-Ceramic Monolith Comparison Test (1)(3)
Vol (1)
Frontal Cell Area Length Type Density Cell Coater (c/cm2) Shape (cm2) (mm) ~
(2)
Final Solid Fraction
Geometrical Surface
(2)
(2)
1.8 100 90(x2) Cer 92 150 Cer 1.4 1.4(1.57) 92(103) 150 Metal 1.4(1.57) 92(103) 150 Metal 1.4(1.57) 92(103) 150 Metal Notes:
wt. (8)
62.0 62.0 73.6 73.6 43.4
SQ SQ PF PF HB
-
A A
B ACR ACR
0.37 1050 0.37 810 1160(1390) 0.21(0.33) 1160(1390) 0.21(0.33) 810( 1000) 0.16(0.28)
(1) Noble metal loading 30 g/ft3 (1060 m a ) (2) MFM values in parentheses include rings and foil behind rings. (3) 7 : O : 1 = P t : P d : R h f o r a l l .
(2 1
23ds 480°C
3.90 3 .OO 3.67(4.17) 3.67(4.17) 3.31(3.76)
1.74 1.80 1.64 1.60 1.26
(m2)(Final)
543
Table I11 SUMMARYOF FTP RESULTS (Average of 15 runs ; 1.8L 'J' car) Engine Outs
Type 1.8LC
Coater A
HC 0.16
CO 2.17
Total g/mi Converted NOx Total Total (g/mi) (E.O. - T.P) 0.46 2.79 13.07 10.28
1.4LC 1.4LPF 1.4LPF 1ALHB
A B ACR ACR
0.19 0.26 0.24 0.20
2.52 2.13 2.32 2.30
0.57 0.60 0.61 0.58
Bag Emissions (g/mi)
3.28 2.99 3.17 3.08
12.97 13.15 13.18 13.58
9.69 10.16 10.01 10.50
However, in analyzing all of the FTP data, it was noted that, contrary to expectations, the total "engine out" emissions (HC -t CO + NOx) were not constant for all twenty-five converters, but in fact varied inversely with converter flow restriction. This variation was apparently caused either by the back-pressure-regulated exhaust gas recirculation system used on this 1985 test vehicle or by some sensitivity of its engine control scheme to exhaust backpressure. To correct for any adverse effect these differences in "engine outs" may have had on the tailpipe bag numbers, converter performance was instead measured on the basis of the total amount of pollutants eliminated by the converters, that is, by subtracting the tailpipe emissions from the "engine out" emissions. As shown in Table IV, the "amount converted" is a more valid reflection of converter performance than "tailpipe" or "efficiency" numbers when the engine out values are not held constant. Notice that when using that measure of performance, the PF units were superior to the equivalent ceramic units and the HB units outperformed all other converters, even the much larger and more heavily loaded ceramic units.
Tailpipe CO (g/mi) Efficiency EO - TP EO Engine Out CO (g/mi) CO Converted (EO - TP)
Conv. A Conv. B 3 .O 2.5 75% 75% 10.0 7.5
12.0 9.0
"A" Better "A" and "B" Equal
(11.0 +_ 10%) "B" Better
Further analysis of the FTP data revealed that the HB units had their poorest relative performance on the hot restart portion of the FTP schedule, that is, on cycles 19 and 20. Thermocoupled converters revealed that during
5 44
the ten-minute shutdown prior to the hot restart, the high conductivity, low heat capacity HB units were cooling below their light-off temperature of approximately 300OC and consequently they had poor efficiencies. An example is shown in Figure 2. However, the ceramic units were remaining well above 300OC and they therefore maintained their high efficiencies. Adding a wrap of insulation around such a metal converter was found to prevent this excessive cooling, and the converter performance was measurably improved.
FTP EMISSIONS AND TEMPERATURE ON COLD STARTlHOT RESTART
8gl o 95
285
475
665
855
1W5
1235
1425
1615
1805
TIME (SEC 1
Figure 2. - HC efficiency and bed temperature during 23 cycle FTP test for fast-cooling metal monolith converter.
Subsequent work on insulated HB converters revealed that although improved, such units continued to have relatively poor performance on the hot restart. Although the insulation kept the substrates above 300OC during the ten-minute shutdown, the HB units were quickly cooled below the lightoff temperature by the initially cool exhaust gases from early in cycle 19. The high heat transfer rates and high thermal conductivity of the HB units, advantages under warm gas/cool converter conditions, are thus seen as disadvantages under cool gas/warm converter situations. Work with an insulated downpipe prevented excessive cooling of exhaust gases early in cycle 19 and resulted in significant performance
545
improvements for the HB design. Since ceramic units never cooled below 300OC before or after the hot restart, they did not experience the same performance improvements as the HB units. However, ceramic units did show some improvement with the insulated downpipes, apparently because average converter temperatures were moderately increased throughout much of the FTP test. CONCLUSIONS
Based on earlier work, as well as on extensive metal/ceramic comparison tests described herein, it has been concluded that standard PF metal foil monoliths are unlikely to compete successfully with ceramic monoliths for the underfloor catalytic converter market because their costs are too high and their benefits too small. On the other hand, the HB design does compare well with ceramic monoliths. Performance advantages for the HB units should allow sufficient downsizing (approximately 20%) to make its costs very competitive. Other benefits of the HB design, such as low restriction, high thermal conductivity, and low thermal capacity, then become significant advantages. Accurate comparisons of converter performance are difficult to obtain. When such comparisons are needed, multiple samples and numerous replications should be incorporated into a statistically valid experiment. In addition, it should be noted that variations in converter and exhaust system restrictions can affect engine breathing and hence influence engine controls and "engine out" emissions. As a consequence, tailpipe emissions can also be affected and thereby confound any conclusions made about converter performance. In such circumstances, the concept of the "amount of pollutants converted" should be considered when comparing catalytic converter performance. REFERENCES 1. P. N. Hawker, C. Jaffray, A. J. J. Wilkins, and J. Alker, "Metal Supported Automotive Catalysts for Use in Europe," SAE paper 880317, International Congress, Detroit, Michigan, March 1988. 2 . P. Oser, "Novel Autocatalyst Concepts and Strategies for the Future with Emphasis on Metal Supports," SAE paper 880319, International Congress, Detroit, Michigan, March 1988. 3. K. Nishigawa, K. Matsuda, H. Horie, and J. Hirohashi, "Development of Improved Metal-Supported Catalyst," SAE paper 890188, International Congress, Detroit, Michigan, February-March 1989. 4. S. Pelters, F. Kaiser, and W. Maus, "The Development and Application of a Metal Supported Catalyst for Porsche's 91 1 Cdrrera 4," SAE paper 890488, International Congress, Detroit, Michigan, February-March 1989. 5 . G. L. Vaneman, "Performance Comparisons of Automotive Catalytic Converters: Metdl vs. Ceramic Substrates," FISITA paper 9051 15, XXIII International Congress, Torino, Italy, May 1990.
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 199 1 Elsevier Science Publishers B.V., Amsterdam
547
SOLS AS PRECURSORS TO TRANSITIONAL ALUMINAS AND THESE ALUMINAS AS HOST SUPPORTS FOR Ce02 AND ZrO2 MICRO DOMAINS L. L. Murrell, S. J. Tauster Engelhard Corporation, Menlo Park, CN 40 Edison, NJ 08818 ABSTRACT The possible relationship of alumina sols to the final three dimensional structure of calcined aluminas prepared by commercial procedures has been referred to only briefly in the literature. We have established that the pore sizes and pore volumes of transitional aluminas prepared by sol precipitation using ammonium hydroxide are strongly dependent on the size of the alumina sol precursor. Preparation of alumina from gelation of a 2 nm diameter alumina sol produces a relatively dense phase with an average pore diameter of 5.5 nm. Aluminas made by gelation of a 20 nm average diameter size alumina sol have three-fold the pore volume of the alumina produced by gelation of the 2 nm alumina sol. In the case of the 20 nm size sol the average pore size is 7 nm in diameter which is very close to that of aluminas produced by conventional aqueous procedures. The high temperature stability of the alumina formed from sols argues that very little amorphous alumina is formed as a separate phase. Ceria sols from 1 to 20 nm size were co-precipitated with the 20 nm alumina sol such that the Ce@ component is the "guest phase" within the three-dimensional alumina phase acting as the "host phase" of Ce02 micro domains. By varying the ceria content, highly dispersed Ce02 with rod-shaped morphology could be prepared within the relatively stable alumina host structure. It was established that these rod-shaped Ce02 structures would transform at increasing temperatures to cylinder morphologies which were isolated from each other within the alumina host structure. Studies with a Ce02 sol size of 12 nm size revealed that the micro domains formed within the alumina were very similar to the starting sol size. However, in the case of a Ce& sol of 1 nm size agglomeration to produce larger domains than the starting sol only occur at CeO2 contents of 5% or greater. The high temperature stability and characterization results for Ce-A1 and temperature stability and characterization results for Ce-A1 and Ce-Zr-A1 systems will be described. INTRODUCTION
Aluminas play a vital role as supports in a broad class of catalyst applications.[ 11. It is not generally appreciated that extruded aluminas owe their strength to dissolved and/or alumina sol species which crystallize to yield the "binder phase" of the non-soluble alumina phase. Since the "binder phase" of alumina can potentially produce a pore structure and surface chemistry which is distinct from the "non-binder" alumina phase it becomes critical to understand better the aluminas derived from sol precursors. In cases where a
548
slurry mixture is coated onto a ceramic or metal substrate, it becomes especially critical that the properties of the binder phase be well understood. Automotive catalysts rely on the alumina acting as a binder to be an integral support for the precious metals within the three-dimensional washcoat structure. This application is also the one which requires the alumina to exhibit excellent high temperature stability. In this paper we have studied the high temperature stability of sol-derived aluminas. The alumina made from the 20 nm sol precursor was also investigated as the "host phase" of Ce02 and Zr02 micro crystalline or micro domain structures. These micro domain structures were formed as the "guest phase" by co-gelation ceria and zirconia sols with the alumina sol. Characterization work established that the sol phases form micro domains which have thin rodshaped to cylinder-shaped morphologies depending on the starting sol size, concentration within the alumina host, and temperature of heat treatment. These "host" encapsulated CeO2 and Zr02 phase are much more stable than the bulk Ce02 and Zr02 phases. It appears that mixed sol preparation procedures may be a useful and general approach [2] to improve the stability of crystalline phases to severe conditions. EXPERIMENTAL
The sols used in this work were obtained from Nalco Chemical and Nyacol. The 2 nm size alumina sol (2A1) was a Nalco product and the average 20 nm size alumina sol (20A1) was a Nyacol product. The ceria and zirconia sols were obtained from Nyacol, and the sol sizes were provided by Nyacol, and were obtained from Coulter particle size analysis. The alumina sols were both precipitated by addition of ammonium hydroxide with rapid agitation until a constant pH of 9 was obtained. The 20 A1 sol was easily precipitated by addition of dilute nitric acid. The 2 A1 sol could not be gelled by addition of dilute nitric acid. In the case of deposition of the 20 A1 sol onto a flow-through monolith substrate by step-by-step build-up of ca 20 micron thick layers (measured from the comer of the channel to the top of the coating), it was found essential to calcine the sample after two layers were deposited by slow heating to 400°C in ca. 2 hrs. with an hour treatment at 400°C. The coating procedure was straight forward. The monolith was dipped into the sol slurry and then removed and the excess slurry in the channels was allowed to drain. The excess slurry was removed from the channels by gentle air knife pressure with the channels held in a horizontal posiposition. The air knife was used alternately in both directions to give a uniform coating on the monolith substrate.
549 RESULTS AND DISCUSSION
In a recent review article on producing aluminas with controlled pore structure, Trimm and Stanislaus state: "Improvements in pore size control can be expected, and studies of colloidal chemistry are the most fertile field from which improvements should originate." Despite this clear implication that alumina sols may be useful to control pore size, little information is available on the relationship between sol size and pore size in aluminas made by sol precipitation. Since sols play an important role as binders of washcoats in automotive application we investigated the properties of aluminas made from commercially available sol precursors. Two alumina sols were investigated, one having 2 nm size units and another having 20-30 nm size units, hereafter referred to as 2 A1 and 20 Al. Both sols were gelled with ammonium hydroxide until a pH of 9 was obtained. Both aluminas showed the characteristic broad lines in the x-ray diffraction of boehmite after calcining in air at 200°C. After calcination at 500°C both aluminas showed the characteristic x-ray pattern of delta-alumina, which also persisted for both samples when calcined in air at 900°C for two hours. The most important feature of these aluminas when calcined at 900°C was the of lines in the x-ray diffraction of alpha-alumina. Any amorphous alumina in these samples would have been converted to alphaalumina at this calcination temperature.[3]. It can be concluded that alumina sols are practical precursors to prepare high temperature stable aluminas with stabilities rivaling high purity aluminas prepared by conventional aqueous methodology.[ 1,4,5] When the physical properties of the 2 A1 and 20 A1 sols are compared it is quite clear that the starting sol has a remarkable degree of influence on the final alumina formed upon gelation. Firstly, the 2 A1 sol produces a much more dense alumina than the 20 A1 sol. The pore volume of the former is only 0.18 cc/g and the latter 0.44 cc/g. This physical characteristic difference alone argues that the 2 A1 sol is very reactive in a relative sense and must polymerize during the gelation step to produce a compact alumina structure. This hypothesis is supported by the fact that the average pore diameter of the 500°C calcined alumina from the 2 A1 sol has an average pore diameter of 3.6 nm. The size of the pores of this alumina argues that during crystallization, pores are formed which are larger than the size of the starting sol. Recent work [3] has also established that re-crystallized amorphous aluminas can have a wide range of pore sizes depending on the re-crystallization conditions of the amorphous phase. Therefore, it seems reasonable that the small 2 A1 sol can undergo polymerization and crystallization to form boehmite crystals where the final structure has relatively narrow pores compared to most aqueouslyprepared aluminas, and a dense three-dimensional structure. The surface area of this relatively dense alumina was 171 m2/g.
.
550
In the case of the 20 A1 sol-derived alumina, the average pore diameter was 7.5 nm with a surface area of 219 m2/g. This combination of pore volume, pore size, and surface area is almost identical to many aqueouslyprepared aluminas.[l,3]. These physical properties make a case that the larger A1 sol of 20 nm size is almost unreactive to re-crystallization during gelation. The structure of this transitional alumina is apparently made up of boehmite precursor particles which are of similar size to the alumina sol from which they are formed. This would make an overview, perhaps deceptively, straightforward: the link between alumina's pore structure and alumina sols in solution is simply that sols of size of ca. 20 nm are intrinsically stable and become" unreactive" because they can not "de-polymerize" to produce narrow pore size phases. Smaller size alumina sols, in contrast, are highly reactive, and undergo polymerization to produce aluminas with relatively narrow pores and in some cases low pore volume. If this is correct then it is no coincidence that the physical properties of the alumina made from the 20 A1 sol, and those made from conventional aqueous procedures are so similar. The surface area Stability aluminas at 900°C is informative because this temperature is sufficient to convert any amorphous alumina to the alphaphase. In earlier work [2,3] amorphous alumina would convert completely to alpha-alumina at 800°C. However, if this amorphous phase was contacted with water which allowed formation of crystalline boehmite particles, then an alumina was produced with good high temperature stability, i.e., 120 m2/g when calcined at 900°C. When calcined at 900°C the 2 A1 and 20 A1 sols gelled with ammonium hydroxide had surface areas of 114 and 142 m2/g, respectively. The difference in these aluminas' surface area is due to the difference in surface area of the starting aluminas, 171 and 219 m2/g, respectively. Both of these aluminas showed the characteristic shift to larger size pores as observed for calcined aqueous-prepared aluminas.[ 1,431. Both aluminas when calcined at 900°C showed no evidence of an alpha-alumina phase as noted previously. A sample of the 20 A1 sol gelled with nitric acid was also calcined for 2 hrs. at 900°C. In this case the surface area of the calcined sample was 100 m2/g with the pore distribution similar to the same sol precursor gelled under basic pH conditions. There was no alpha-alumina phase present for the 900°C calcined alumina precipitated under acidic conditions. This result shows that 20 A1 sol does not require basic conditions in order to form an alumina with good high temperature stability. This result is especially relevant to the preparation of coating slurries for automotive catalyst application where the desired pH of the slurry is usually between 3-4.5. Multi-coating of the 20 A1 sol onto a monolith honeycomb produced a washcoat of ca. 200 micron thickness. The surface area and pore volume of the alumina were 220 m2/g and 0.43 cc/g. This shows that a uniform coating of alumina can be obtained even after 8 sequential deposits of a sol precursor and that the alumina is essentially identical to the precipitated sol using ammonium hydroxide.
55 1
In summary, alumina sols can transform to boehmite phases even when formed in conditions not considered ideal for alumina re-crystallization [4]. Since it was established that the pore volume of the alumina made from the 20 A1 sol was 3-fold higher than that for the alumina made from the 2 A1 sol, then the 20 A1 sol was the clear choice as the "host" support for other oxide components. The concept of preparing mixed oxides from sol precursors has been disclosed previously in independent work.[2] We were interested in preparing ceria structures with bulk CeO2 surface redox properties but which were stable at temperatures of 900"C, or higher for automotive application. In our initial studies, a Ce02 sol of 12 nm size was mixed at three compositions with the 20 A1 sol and gelled with ammonium hydroxide. This Ce02 sol will be referred to as 12 Ce. The three compositions of interest were: 10 wt% Ce0290 wt% Al2O3, 20 wt% Ce02-80 wt% Al2O3, and 30 wt% Ce02-70 wt% Al2O3. The surface areas of these three compositions designated 10 Ce-90 Al, 20 Ce-80 Al, 30 Ce-70 A1 were 203, 195, and 150 m2/g when calcined at 500"C, respectively. The surface area decrease of the 30-70 composition compared to the other two is due to a real densification at the highest CeO2 level in this series, and not due to the density differences between Ce02 and Al2O3. We will see in subsequent work with similar Ce02 compositions containing low levels of a ZrO2 sol that only small differences in surface area are obtained over this same composition range. The pore volumes of these three compositions were ca. 0.44 cc/g, very similar to the sol-derived alumina "host." The surface area of these three compositions following 900°C calcination are 126, 116, and 103 m2/g, respectively. It is very important to note that the pore volumes and surface area stabilities of these three compositions were analogous to the alumina formed without the Ce02 sol present. The 20 nm size A1 sol appears to be acting as an ideal "host" for the C e 0 2 sol phase. The CeO2 domain size was obtained by x-ray diffraction (XRD) line broadening for these three compositions dried at 120"C, calcined at 500 and at 900°C. For all three compositions the estimated Ce02 domain size was estimated to be ca. 5.9 nm. This was somewhat of a surprise since the starting CeO2 sol size was almost double that estimated by x-ray diffraction. There are three pieces of evidence which point convincingly that XRD is significantly underestimating the size of the CeO2 domains. - Firstly, high resolution transmission electron microscopy (TEM) of the 30 Ce-70 A1 composition shows from the hundreds of Ce02 domains visible in the micrographs that CeO2 domains are composed of cylinders from 3-6 nm in width to 10-16 nm in length. The structure of these CeO2 domains are clearly consistent with the Ce02 sol being "locked" into the alumina acting as a host
552
where Ce02 domains are formed which are crystallizing as small cylinders which appear as "rice grain" structures within the host. - Secondly, when the three Ce-A1 samples were calcined at 900°C the XRD domain size for the three compositions: 10-90, 20-80, 30-70 were 12, 9.4, and 15 nm, respectively. Apparently, the 900°C calcination removes domains within the crystalline domains so that the line broadening is now more in accord with the starting sol size of 12 nm. - Thirdly, when 12 Ce was gelled with NH40H without A1 sol present, the bulk Ce02 obtained gave an XRD derived particle size of 5.9 nm. The physical properties of this CeO2 sample were very much in accord however, with significant re-crystallization of the CeO2 phase. The pore volume was only 0.2 cc/g with a surface area of 100 m2/g. The surface area appears quite high but all of the pores were of narrow distribution centered at 3 nm diameter. Formation of such a narrow pore diameter material with low pore volume argues that the Ce02 sol undergoes massive recrystallization to form a relatively dense bulk Ce02 phase. The XRD particle size of 5.9 nm for this bulk Ce02 phase again reflects domains within the bulk Ce02 particles. Calcination of this sol-derived Ce02 sample at 900°C showed a major collapse of the porous structure to give a material with a surface area of 3 m2/g. The XRD of this sample was consistent with formation of very large Ce02 primary domains of >60 nm diameter. The contrast in the domain size for the CeO2 within the host alumina structure to the bulk Ce02 phase serves to prove the remarkable enhancement in stability for Ce02 structures when encapsulated within an alumina host structure. We were fortunate that a Ce02 sol of only 1 nm size was available for preparation as a mixed Ce-A1 composite. This smaller size Ce02 sol could lead to smaller domains of Ce02 within the alumina host structure. Smaller Ce02 domains would lead to a much greater number of exposed Ce02 surface groups at a given CeO2 composition than for a sol size of ca. 10 nm. A Ce02 crystallite of ca. 10 nm size would only have about 1-in-10 CeO2 groups as a surface group whereas a 1 nm Ce02 crystallite would have every Ce02 group a surface group. The 1 nm Ce02 system would be referred to in catalysis terms as being highly-dispersed. For the 1 nm Ce02 sol-20 nm A1203 sol compositions a very wide range was chosen since it was expected that at some critical point the high dispersion of the Ce02 domains or crystals due to the starting sol would be lost. This loss of high dispersion would be associated with polymerization of the very "numerous" Ce02 sol units within the much larger A1203 host particles. These systems were prepared over a composition range from 2.5 Ce - 97.5 A1 to 45 Ce - 55 Al. The surface areas for 5 Ce - 95 Al, 15 Ce - 85 Al, 30 Ce - 70 Al, and 45 Ce - 55 A1 were: 196, 178, 180 and 181 m2/g, respectively. After 900°C calcination of these four compositions the surface areas were: 123, 127, 115, 90 m2/g respectively. The domain size
553
measured by XRD for these samples calcined at 500°C were 5.4, 4.6, 6.2 and 5.2 nm, respectively. This information makes it clear that little advantage has been realized in using the 1 nm CeO2 sol to form small CeO2 particles within the alumina host structure compared to the starting sol size of 12 nm described previously. The domain size for this series of samples calcined at 900°C even makes this point more clearly. The XRD domain sizes were 8.3, 9.5, 10.6 and 13.1 nm, respectively, for this series of samples. Note that these domain sizes are really very close to those described previously using the Ce02 sol of 12 nm size. Since the domain size only changes from 8.3 to 13.1 nm for samples calcined at 900°C over the composition range from 5 to 45% Ce02 content, then it is apparent that for high temperature application having a Ce02 sol of 1 nm in size is entirely irrelevant to obtain improved Ce02 dispersion. What is clear from this work is that ceria domain size is remarkably stable over a very wide composition range when co-gelled with an appropriate-sized alumina sol. In a final attempt to obtain evidence for small crystalline particles of Ce02 within an alumina host phase the 1 nm Ce02 sol was co-gelled with the 20 A1 sol. The compositions investigated were 2.5 Ce - 97.5 A1 and 5 Ce - 95 Al. TEM analysis clearly showed, for both of these compositions calcined at 500"C, that the CeO2 domains were thin rods of very high dispersion, or a high ratio of surface ceria groups to bulk ceria groups. The 2.5 Ce - 97.5 A1 composition showed that the rods were 2 nm in diameter and 45 nm in length. For the 5 Ce - 95 A1 composition the rods became thicker and much shorter, 4 nm in diameter and 20 nm in length. The TEM of the 5 Ce - 95 A1 composition calcined at 900°C gave very clear evidence for sintering of the Ce02 domains. Cylinder or "rice grain" structures were observed of similar size, 8 nm by 13 nm cylinders, to those observed for the 30 Ce - 70 A1 composition using the starting Ce02 sol of much larger size discussed previously. The XRD determined CeO2 domain size of 8.3 nm is in reasonable agreement to the 8 by 13 nm domains established by TEM for the 900°C calcined 5 Ce - 95 A1 sample. It is clear that high dispersion of CeO2 can be achieved in the form of thin rods within the alumina host structure which are stable to 500°C calcination temperatures. Calcination conditions of 900"C, however, result in transformation of these small micro domains to particles of size similar to those achieved by starting with a Ce02 sol precursor over an order of magnitude larger in size. In one series of samples a tri-gel of Ce-Zr-A1 was prepared using Ce, Zr, and A1 sols of 20, 10, and 20 nm respectively, gelled with ammonium hydroxide. In these systems three compositions were investigated: 10 Ce - 7 Zr - 83 Al, 20 Ce - 7 Zr - 73 Al, and 30 Ce - 7 Zr - 63 Al. All of these three compositions showed Ce02 and ZrO2 domain sizes obtained by TEM consistent with the starting Ce and Zr sol sizes. After 900°C calcination, XRD showed clear evidence that the CeO2 and Zr02 micro domains were of a size
554
similar to the starting sol size and that Ce02 and ZrO2 were similar to the starting sol and that CeO2 and Zr02 were intact as separate domains. The surface areas of these three compositions showed enhanced stability compared to the corresponding Ce-A1 systems when calcined at both 500 and 900"C, see Table 1. Apparently, the low level of the Zr02 phases serves to enhance significantly the stability of the composite which probably acts to retard the C e 0 2 particle sintering during the gelation step of the preparation. A stabilizing effect of the Zr02 particles on the alumina surface area can not be ruled out as is well established, at least for highly-dispersed ZrO2 phases on alumina. Table 1 Surface Areas for Ceria-Alumina and Ceria-Zirconia-Alumina Mixed Composites Calcined at Intermediate and High Temperature Sample (wt % as oxide)
Calcination Temperature "C
Surface Area m2/g
10 Ce - 90 A1 20 Ce - 80 A1 30 Ce - 70 A1 10 Ce - 90 A1 20 Ce - 80 A1 30 Ce - 70 A1
500 500 500 900 900 900
203 195 150 126 116 103
500 500 500 900 900 900
196 185 186 148 141 132
10 Ce -7 Zr 20 Ce -7 Zr 30 Ce -7 Zr 10 Ce -7 Zr 20 Ce -7 Zr 30 Ce -7 Zr
- 83 A1 - 73 A1 - 63 A1 - 83 A1 - 73 A1 - 63 A1
CONCLUSIONS
The size of alumina sols employed as precursors of aluminas plays a major role in the final pore structure of the calcined alumina phase. A 2 nm size alumina sol is apparently easily polymerized to give an unexpectedly dense alumina phase. In the case of a 20 nm alumina sol the final alumina is remarkably similar to that expected from little reactivity of the alumina sol during gelation.. The surface area, pore size, and pore volume of the alumina made from the 20 nm size alumina sol is very similar to that of many
555
aqueously-prepared aluminas. Perhaps the aqueous preparation methods simply produce a sol of size which is stable over a wide range of conditions of ca. 20 nm diameter which is then precipitated to give the aluminas which we are most familiar. In the case of co-gelation or tri-gelation of Ce-A1 and of Ce-Zr-A1 mixed composites it was established that the alumina host structure leads to micro domain structures of CeO2 or Zr02 within the alumina phase which were quite stable to high temperature conditions, This work would suggest that mixed sol procedures can be a new method to obtain a range of dispersions of bulk phases which are not presently accessible by non-sol routes. These encapsulated micro domain structures can be of unusual morphology as indicated by the thin rod-shaped structures observed in the case of the 1 nm Ce02 at low Ce02 content in the composite Ce-A1 phase. It has been shown that the size of the "guest phase" or micro domain formed within the alumina host is strongly dependent on the starting sol precursor of the host phase. Domains of Ce02 and ZrO2 were found to be isolated in the alumina host, so it appears that multi-component oxide systems can be obtained by gelation of mixed sol slurries. These micro domain systems will have the properties of their bulk counterparts, but will have the advantage of improved stability as demonstrated in the case of the stability of the alumina host in this work. REFERENCES
1. 2.
3. 4. 5.
D. L. Trimm and A. Stanislaus, Applied Catalysis, 21, 215-238, (1986) Baiher, A., Dollenmeier, P., Glinski, M., Reller, A. and Sharma, V. K., J. Catal. 111, 273 (1988). L. L. Murrell, N. C. Dispenziere, Jr., K. S. Kim, in "Novel Materials in Heterogeneous Catalysis," Chapter 9, p 97 Eds. R. T. K. Baker and L. L. Murrell, ACS Symp. Series 437, (1990), Washington, D.C., T. Ono, Y. Ohgachi, 0. Togari, in "Preparation of Catalysts 111", Eds. G. Poncelet, P. Grange, P. A. Jacobs, Elsevier, Amsterdam, p. 63 1,.( 1983) M. F. L. Johnson, J. Catal. 123, 245 (1990).
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
557
PREPARATION OF THERMOSTABLE HIGH-SURFACE-AREA ALUMINAS AND PROPERTIES OF THE ALUMINASUPPORTED Pt CATALYSTS
F. Mizukami*, K. Maeda*, M. Watanabe*, K. Masuda**, T. Sano* and K. Kuno**
*National Chemical Laboratory for Industry, -I Higashi, Tsukuba,Ibaraki 305, Japan **CentralIEngineering Laboratories, Nissan Motor Co.,Ltd., I Natsushima-cho, Yokosuka,Kanagawa 23 7,Japan ABSTRACT Thermostable high-surface-area aluminas were obtained by investigating the influence of preparation procedures, raw materials and additives on sintering of alumina. The complexing agent-assisted sol-gel method in which a complexing agent such as hexylene glycol was used as a solvent gave aluminas with large surface area and high durability at around 1OOO"C as compared with those prepared by conventional methods. Upon addition of BaO, La203 and SrO to alumina using the sol-gel method, the sintering of alumina was furthermore retarded and the effect was maximum when 10wt% BaO was incorporated to alumina. Alkoxide was the most suitable raw material for alumina, and alkoxide and acetylacetonate were suitable for BaO. Three lwt% Pt catalysts were prepared from a commercial alumina (A) derived from alkoxide, the sol-gel alumina (B) and the sol-gel BaOA1203 (C), and their catalytic properties were compared. The activity order of the three catalysts for the CO oxidation was (A) < (B) < (C). This sequence could be explained by the differences in the Pt crystal growth and a-Al203 appearance rates in the three catalysts.
INTRODUCTION
Alumina has such a superior thermal stability as compared with other high surface area materials that it is used as support to automobile exhaust gas catalysts and its demand is increasing from year to year. However, the thermal stability of the alumina is not always sufficient to satisfy the present requirements, and attempts at preparing thermostable aluminas with high surface area are still under way. There are two ways to obtain such aluminas. One is to retard the sintering of active aluminas with high surface area or prevent the transformation of the active aluminas into a-alumina with low surface area by adding certain additives. The other is to prepare an alumina showing high surface area and thermal stability without additives by investigating the preparation proeedure. For the former, although there are few systematic researches about the effect of preparation procedure on the structure and surface area of mixed metal oxide-aluminas, the addition of
558
yttrium, lanthanum, cerium, barium etc into alumina is known to be extremely effective, namely, the addition of yttrium or lanthanum by coprecipitation 11-21 and the addition of baryum by kneading or coprecipitation [3-51lead to the formation of thermostable p-alumina and magnetoplumbite (Ba0.6A1203) structures, respectively, and the addition of a rare earth by impregnation [6] inhibits the grain growth of transition aluminas and the formation of a-alumina. But for the latter, an efficient method has so far not yet been found . For the past several years, we have been studying the control of the particle size and structure of metal oxides with the sol-gel procedure (complexing agent-assisted sol-gel method) [7- 111 which is intended to regulate the hydrolysis and dehydration-condensation-polymerizationof metal ions or complexes with polyfunctional organic compounds (polydentate and bridging ligands). We applied the sol-gel procedure to the synthesis of aluminas or aluminas containing an additive, and tried to obtain thermostable aluminas with high surface area as well as highly active Pt catalysts supported on these aluminas. EXPERIMENTAL
Preparation of catalysts (i) Aluminas. Aluminium isopropoxide (120 g) as source of aluminum and hexylene glycol (2-methyl-2,4-pentanediol, 280 g) as solvent were heated at 120°C for 4h,a viscous homogeneous solution being obtained with ligand exchange and evolution of 2-propanol. To this solution 9 equiv. water was added to the alkoxide at 100°C. After the wet gel was allowed to stand for 1012h at 80°C, alumina (a) was obtained by drying in a rotary evaporator under reduced pressure. For comparison, aluminas were also prepared by two conventional methods which are known to give aluminas with high surface area; one method is the hydrolysis of aluminium isopropoxide in 2-propanol at 70°C (b), and the other is double decomposition in which 8.4wt% aluminium sulfate aqueous solution (830g) and 8wt% sodium aluminate aqueous solution (8298) were mixed at room temperature in the presence of gluconic acid (2.lg) to obtain alumina gel precipitate (c) [12]. These aluminas and a commercial alumina (Ketjen, CK-300 high purity) (d), as a reference, were calcined at various temperatures under the same conditions. (ii) Mixed aluminas. Aluminium isopropoxide (50 g) was dissolved in an ethanol solution (150 g) containing hexylene glycol (50 g) and the solution was heated in an oil bath at 120°C for 0.5h. A given amount of a raw material of a metal oxide as additive was added to the solution and heated for 3h. To the solution, 6 equivalent water to total metal ions were supplied. The remainder of the procedure used to obtain mixed metal oxide-alumina is identical to the method for the alumina described above. For comparison,
559
mixed metal oxide-aluminas were also prepared by the following coprecipitation method. A solution of aluminium sec-butoxide (87.6 g) dissolved in 2-propanol (50 g) was heated in an oil bath at 120°C for 0.5h. A given amount of a raw material of an additive component was added to the solution, heated for l h and the oil bath was cooled to 80°C. Then, water (320 g) at 80°C was added to the solution in the oil bath. After the precipitate occurred, it was filtered and treated in the same manner as above. (iii) lwt% Pt catalvsts Platinum catalysts were prepared by a conventional impregnation technique.with an aqueous solution of dinitro diammine platinum complex and the above aluminas previously calcined at 1000°C for 4h. They were calcined at 400°C for 3h, then at 1000°C for 20h, and activated in a hydrogen stream of 100 cm3/min at 400°C for lh.
CO oxidation Activities of the Pt catalysts for the CO oxidation were measured using a fixed-bed continuous flow microreactor with 0.1 g of catalyst. A gas mixture containing CO (0.47 mol%), 0 2 (0.25 mol%) and helium was flowed through the reactor at the rate of 100 cm3/min and analyzed by gas chromatography. The CO conversions and C 0 2 selectivities were calculated on the basis of the concentration of helium used as internal standard.
Characterization Specific surface areas of the aluminas were obtained from the nitrogen adsorption and desorption isotherms at 77 K, using a micro-BET apparatus (an AccuSorb 2100-01 of Micromentics). X-ray diffraction patterns of the aluminas were recorded on a Philips PW 1700 diffractometer equipped with a curved graphite monochrometer. With CuKa radiation, the aluminas were scanned over the range 28=5 - 60". Differential thermal analyses were carried out at 10"C/min in air with a Rigaku TG-DTA using a-Al203 as a standard. The Pt catalyst surfaces were examined with a Hitachi S-4000 scanning electron microscope. RESULTS AND DISCUSSION
Aluminas Figure 1 shows the variation of the specific surface area with calcination temperature of the sol-gel alumina (a) in comparison with the conventional ones, (b) and (c). The aluminas prepared from alkoxide, (a) and (b), show higher surface areas than (c). It should be emphasized that (a) keeps the highest surface area in the range of 600 to 1000°C and holds high values
560
(136 m2g-1) even after calcination at 1000°C for Soh, whereas the surface areas of (c) and of the commercial alumina (d), not originated from alkoxide, deteriorate remarkably on calcination at the same temperature (Fig. 2). Such a deterioration tendency was also observed in X-ray diffraction analyses, namely, (c) and (d) transformed into a-alumina at faster rates and lower temperatures than (a) and (b).
300 : m N
E
v
a p!
a
200
a,
8
't
3
v)
.o c 100 0 a,
a v)
0 500
600
800 900 1000 1100 Calcination Temperature ("C)
700
1200
Figure 1 Effect of the calcination temperature on the sugace area of the aluminas. Calcination time: 3h. (a) Sol-gel method (a); (0)Conventional alkoxide method (b); ( A ) Double decomposition method (c). Mixed aluminas First of all, metal oxides such as MgO, CaO, SrO, BaO, La2O3, Ce02 and Zr02 which have high melting points [9] were selected as the additives to alumina, and the effects on the specific surface areas of the metal oxide-mixed aluminas were examined. With the 10wt% mixed metal oxide-aluminas which were prepared from alkoxides by the sol-gel procedure and calcined at 1OOO"C for 3h, the aluminas containing SrO, BaO, La203 and ZrO2 showed specific surface areas of more than 100 m2g-1, while values lower than 40 m2g-1 were obtained for the others.
56 1
10
20 30 40 Calcination Time (h)
50
Figure 2 Variation of the specific sugace area of the aluminas on calcination at I000 "C. (m) The sol-gel method (a) ; (0)Conventional alkoxide method (b) (A)Double decomposition method (c); (0)Ketjen, CK-300 high purity (d) .
500
Q)
3 5 400
700
900 1100 1300
v)
Q)
500
700
900 1100 1300
0 500 700
900 1100 1300
3 5 400 v)
0 .'c
0 .'c
g cn
0
g 200
200
v)
0 500
700
900 1100 1300
Calcination Temperature ("C)
Calcination Temperature ("C)
Figure 3 Specific surface areas of IOwt% MxOylA1203 prepared by complexing-agent-assistedsol-gel procedures and coprecipitation. (m) Sol-gel method (0) Coprecipitaton method. MxOy :(a) La203, (b) SrO, (c) BaO, (d) Zr02
562
Figure 3 shows the effect of the preparation procedures on the specific surface areas of mixed La203-, SrO-,BaO- and ZrO2- aluminas prepared from alkoxides and calcined at the temperatures of 550 to 1200°C for 3h. It should be emphasized that the sol-gel aluminas show higher surface areas between 550 and 1000°C than the corresponding coprecipitated ones, although the differences between the two procedures are almost unexisting at calcination temperatures higher than 1000°C. The efficiency of the four additives for keeping high surface area is BaO > La203 > SrO > Zr02 at calcination temperatures of 1000°C or above. The effect of starting compounds was also examined with different mixed metal oxide-aluminas prepared by the sol-gel method. As a result, it was found that, in order to obtain mixed metal oxide-aluminas with large surface area and high thermal stability, simple alkoxides such as propoxide, butoxide etc. are suitable raw materials for alumina and zirconia, whereas alkoxides and acetylacetonates are preferred as sources of BaO, La203 and SrO. Figure 4 shows the X-ray diffraction patterns of the mixed aluminas prepared from aluminium isopropoxide and metal nitrates or metal alkoxides. The combination of alkoxide and nitrate yields more easily crystalline phases, which result in a lowering of the surface area, than the use of two alkoxides. In the four mixed aluminas prepared from alkoxides, the order in which crystalline phases are produced is Zr02 > SrO > L a 2 0 3 > BaO. This is consistent with the above order in which high surface areas are kept at high temperatures. The above results cannot be clearly explained at present, but they may be related to the mobility of additives. Generally, alkoxides and acetylacetonates can produce metal oxides interacting with aluminas through oxygen atoms by simple hydrolysis-dehydration reaction, but nitrates used here are hydrates; in this case, the dehydration is usually difficult unless the pH of the reaction mixture.is changed. Thus nitrates tend to produce metal oxides with high mobility. The mobility of a metal ion or oxide is also inversely proportional to its size, namely, the ionic size increases in the order: Zr+4<Sr+2=La+3
563
BaO-AI203 [ 161 : &A1203 (corundum), BaOeA1203 (spinel), Ba0*6A1203 (magnetoplumbite) and BaO (sodium chloride type). Besides, the specific surface areas of these phases decrease in the order of Ba0*6A1203 > BaOaA1203 > a-Al2O3 > BaO [4,5]. Furthermore, an amorphous metal oxide generally has a higher specific surface area than the corresponding crystalline metal oxide . Thus, the results of table 2 are consistent with the values of the specific surface areas given in table 1.
0
.
0
(c2)
(d)
-
20
40
-A
--k, ,
30
20
Calcined at 1000°C for 3h
10
20
40
30
20
10
Calcined at 1200 “cfor 3h
Figure 4 X-ray diffraction patterns of I Owt% MxOylA120.3 prepared by the sol-gel procedures. MXOY :(a) SrO, (b) BaO, (c) LazOj, (d) Zr02 Raw materials :A1203 Al(0Pr-i)~ ;( a ] ) ,(bl),( c l ) metal nitrates ;(a2) Sr(OPri)2 ;(b2) Ba(OBuS)2 ;(c2) La(OPrn)3 ;(d) Zr(OPrn)3 Assignment : (0)SrOeA1203, @) a-Al203, (0) BaO*A1203, (0)ZrO2
564
Table 1
Influence of BaO content on specific surface areas of BaO/A1203 prepared by the sol-gel procedure Calcination Temp. ("C)+ Calcination Time.(h)-+ BaO wt % 1 5 10 15 20 25
1450 800 lo00 1200 5 3 3 3 Specific Surface Area (m2/g) 1 162 111 17 4 219 171 96 7 227 176 32 8 202 141 13 11 111 75 10 6
550 3 444 375 488 340 364
Table 2
Calcination Temp. ("C)+ Calcination Time.(h)+ BaO wt % 5 5 10 15 20 25
lo00 3 a>>c a a a a>>s
1200 3 Structure c>>m>s a c<<m m m=s
1450 5 o m c=m c<<m m m
Figure 5 shows differential thermal analyses of dry BaO-Al203 gels corresponding to the samples in tables 1 and 2. The dry gels prepared by the complexing agent-assisted sol-gel method generally contain more or less organic moieties coming from complexing agents and solvents. Owing to the organic moieties, it is very difficult to obtain informations on the phase transformation of the mixed aluminas in the range of 300 to 500°C.
565
0
750
1500
Temperature ( " C )
Figure 5 Differential thermal analyses of dry BaOIAl203 gels BaO content (wt%) : (a)5, (b) 10, (c) 15, (d) 20, (e) 25 Raw material : BaO Ba(OBun)2;A1203 Al(OPri)3
But, exothermic peaks above 750°C give effective informations on the phase transformation. Thus, from the peaks, the phase transformation is estimated to occur twice with 5wt%, 15wt%, and 25wt% BaOA1203 gels, once with 20wt% BaO-Al203, and none with lOwt% BaO-Al203 in the range of 750 to 1300°C. This estimation is in good harmony with the results obtained by X-ray diffraction analyses. As can be seen from table 2, the amorphous structure is more thermostable in lOwt% BaO-Al203 than in the other BaO-Al203 samples. So, the durability and preparation procedure of 10wt% BaO-Al203 were examined into more detail. The specific surface area only fell to about 90% of the original one after calcination at 1000°C for 100h. It was also found that acetylacetonate is a more suitable source of BaO than alkoxide because of its high solubility and reactivity to aluminium alkoxide.
lwt% Pt catalysts Three lwt% Pt alumina catalysts were prepared using three aluminas, namely a commercial alumina derived from alkoxide, the sol-gel one prepared from alkoxide in hexylene glycol, and the sol-gel 10wt% BaOA1203 prepared from barium acetylacetonate and aluminium isopropoxide in hexylene glycol. The three supports were calcined at 1000°C for 4h before impregnation with an aqueous solution of the Pt complex. The specific surface areas of the commercial alumina, the sol-gel one and the sol-gel mixed one were 107, 170 and 186 m2g-1, respectively. The three Pt catalysts were furthermore calcined and activated under the conditions mentioned in the
566
experimental part, and used for the CO oxidation. The Pt contents of the three catalysts prior to the activation were checked by chemical analysis and were confirmed to be equal within experimental errors. Figure 6 shows the CO oxidation profiles for the three catalysts. The activity of the catalysts decreases in the order: sol-gel BaO-Al203 > the sol-gel alumina > commercial alumina.
C
.'2 60 a,
>
40
0
0 1
Ternperature ("C)
Figure 6 CO oxidation profiles with the three catalysts activated after calcination at IOOOOC for 20h. (a) Pt I the sol-gel IOwt% BaO-Al203, (A) Pt I the sol-gel alumina, (0) Pt I a commercial alumina derived from alkoxide. Figure 7 shows the observation of the surfaces of the catalysts calcined at 1000°C for 20h examined with SEM. White big masses correspond to platinum and their sizes increase in the order: sol-gel BaO-AI203 < the solgel alumina < commercial alumina. This order corresponds to the sequence of the sintering extent of the alumina surfaces. Actually, according to X-ray diffraction analyses, the transformation rate to a-alumina for the three catalysts was: 39% in the commercial alumina, less than 5% in the sol-gel alumina and could not be estimated in the sol-gel BaO-Al203. Thus it is considered that the sintering of aluminas proceeds with the transformation to a-alumina at least in these cases. As these results are in fair agreement with the reactivity order in fig. 6, we conclude that the aluminas whose crystallization or sintering rates are slower keep larger surface areas at high temperatures and give higher activities in CO oxidation.
567
Figure 7 SEM images of the three catalysts after calcination at I000 “c for 20h. (a) Pt I the sol-gel IOwt% BaO-Al203, (b) Pt I the sol-gel alumina, (c) Pt I a commercial alumina derivedji-om alkoxide. Magnification :30,000.
568
CONCLUSION
The complexing agent-assisted sol-gel method leads to aluminas with large surface area and higher thermal stability as compared with conventional methods. BaO, La203 and SrO are suitable additives to inhibit the sintering of aluminas and the mixed-oxide aluminas prepared by the sol-gel method show higher surface areas at the calcination temperatures of 550 to 1OOO”C than the corresponding ones prepared by the coprecipitation method. BaO keeps to alumina its amorphous character up to high temperatures as compared with the other additives and its adequate proportion to alumina is 10wt%. With the three Pt catalysts prepared from a commercial alumina, the sol-gel one and the sol-gel BaO-mixed one, obtained from alkoxides and barium acetylacetonate, the order of activities in CO oxidation is Pt/mixed BaO-alumina > Pdsol-gel alumina> Wcommercial alumina. The aluminas in the three catalysts sinter in the order: commercial alumina > the sol-gel alumina > the mixed BaO-alumina which is in good agreement with the order on the Pt crystal growth rates of the three catalysts. REFERENCES 1. S. Matsuda, Bull. Ceram. SOC.Jpn., 20, 189(1985). 2. S.Matsuda, A.Kato, M.Mizumoto and H.Yamashita, 8th Int.Congress on Catalysis, Proceedings, Berlin (1984),Vol4, p 879 3. J.K. Kesselring, Proceeding 3rd Workshop on Catalytic Combustion (1978). 4. M. Machida, K. Eguchi and H. Arai, Chem. Lett., 151(1986). 5. M. Machida, K. Eguchi and H. Arai, Bull. Chem. SOC. Jpn., 61, 3659(1988). 6. M.Ozawa, M.Kimura and A.Isaogai J.Mater.Sci;Lett, 9,709 (1990); J.Less-Common Metals ,162,297 (1990) 7. F. Mizukami, S.Y.Matsuzaki, K. Funrkori, S. Niwa, M. Toba and J. Imamura, J. Chem. Soc., Chem Commun., 678(1986). 8. S. Niwa, F. Mizukami, S. Isoyama, T. Tsuchiya, K. Shimizu, S. Imai and 1. Imamura, J. Chem. Tech. Biotechnol., 36, 236(1986). 9. F. Mizukami, S. Niwa, M. Toba, T. Tsuchiya, K. Shimizu, S. Imai and J. Imamura, In Studies in Surface Science and Catalysis Vol. 31, Preparation of Catalysts IV (B. Delmon, P. Grange, P.A. Jacobs, and G. Poncelet, Eds.), Elsevier (1987), p.45. 10. M.Toba, F.Mizukami, S.Niwa and K.Maeda, J.Chem.Soc.Commun.,l212 (1990) 11. K.Maeda, F.Mizukami, S. Miyashita, S.Niwa and M.Toba, J.Chem.Soc.Commun.,l268 (1990) 12. Jpn. Pat. 1978-19000 (1978) 13. I.E. Campbell, “High Temperature Technology”, John Wiley & Sons, (1956), p.31 14. F.A.Cotton and G.Wilkinson, “Advanced Inorganic Chemistry, A Comprehensive Text“, 2nd Ed., John Wiley & Sons (1962) 15. F.Baso1o and R.G.Pearson,”Mechanisms of Inorganic Reactions, A Study of Metal Complexes in Solution“, 2nd Ed., John Wiley & Sons (1968), Table 2.10. 16. E M Levin, C R Robbins and H F. McMurdie, “Phase Diagrams for Ceramists”, American Ceramic Society (1964) Fig. 206.
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
569
KINETICS OF THE PHYSIC0 CHEMICAL AND CATALYTIC ACTIVITY EVOLUTION OF A Pt-Rh CATALYST IN AUTOMOTIVE EXHAUST GAS G . Mabilon, D. Durand and M. Prigent Institut Francais du Pitrole, BP 31 I , F-92506 Rueil-Malmaison Cedex, France
ABSTRACT A Pt-RWAl2Og catalyst on cordierite monolith has been submitted to severe condition engine bench aging over extended periods. Sintering of alumina and precious metals was approximately proportional to the logarithm of the aging time. For aged catalysts, the ratio Pt/Rh in the metallic particles increased with increasing size. Phosphorus, zinc and lead coming from lubricating oil and gasoline accumulated on the catalyst at a steady rate. Sulfur concentration remained very low, probably due to high temperature inhibition of sulfate deposition. Testing of the catalyst under a constant composition stoichiomemc gas mixture, shows that the light-off temperature increased nearly proportionally to the logarithm of the aging time. Sintering of the precious metals explain the activity decrease for aging times up to 6 hours. For longer aging times both sintering and poisoning are responsible for the activity decrease.
INTRODUCTION Automotive post-combustion catalysis is characterized by severe conditions leading to a strong decrease in catalytic performances. Temperatures generally vary between ambient up to 1073-1173 K, and can even exceed with a poorly tuned ignition system. The oxide and metallic components of the catalyst sinter and interact under these conditions leading to a decrease and modifications of the surface of the support and of the active metallic surface [1,2]. Combustion of the gasoline and of part of the lubricating oil leads to the formation of water, carbon and nitrogen oxides, and also of compounds of phosphorus, sulfur, zinc, lead. These compounds can deposit on the active species of the catalyst, leading to either physical or chemical deactivation (enhancement of the diffusion limitation or poisoning by chemical interaction) [3]. These phenomena have been recognized for a long time, but have essentially been described by comparing the physico-chemical and catalytic properties of fresh and aged samples [4,5]. The purpose of this study is then to model the evolution of the physico-chemical properties and activity of a PtRh/A1203 catalyst as a function of aging time. Catalyst formulation without
5 70
ceria was chosen in order to avoid its adverse effects on precious metal characterisation. As sintering phenomena are frequently modeled by power laws versus time, aging time was varied exponentially between 0.17 and 200 h. A better knowledge of the important parameters of the aging kinetics under real exhaust gas could enable better vehicle aging prediction and better aging simulation under laboratory conditions.
EXPERIMENTAL The catalyst was made on a round cordierite monolith support (diameter 118.4 mm, height 76.2 mm) having 62 cells per cm2. It was coated with an alumina wash-coat and impregnated with chloroplatinic acid and rhodium trichloride at a concentration of 1.06 g/l of precious metals (about 1 5% wt with respect to the wash-coat) with a Pt/Rh mass ratio of 5. After impregnation the catalyst was calcined during 2 hours at 773 K. A four cylinder engine of 2.2 litre displacement with electronic fuel injection was used for catalyst aging. The accelerated aging cycle was composed of three modes [6]. Mode 1 simulates driving a car up-hill at 140150 km/h. It lasts 60 s and brings the exhaust gas temperature to about 1053 K. Mode 2 (3 s) simulates a sudden deceleration with suppression of 20 % of spark ignition. In mode 3 (3 s) lubricating oil (40 ml/h) and an excess of 30 % air is admitted in the engine. Temperature in the catalyst is of about 1043 K in mode 1, and rises up to nearly 1273 K in modes 2 and 3. The BET surface area was determined on a GIRA x SORB apparatus by nitrogen adsorption at 77 K. The surface area of the wash-coat was derived from that of the catalyst by dividing by the wash-coat mass fraction (cordierite has a negligible surface area of about 0.2 m2/g). Porous volume of the catalyst was measured on a Micromeritics 9200 mercury porosimeter. Cristallographic analysis of the wash-coat was made on a Siemens diffractometer. Metal dispersion was determined by hydrogen-oxygen titration. Catalyst samples were reduced in hydrogen at 723 K, cooled to room temperature in hydrogen; after having purged the reactor with helium, hydrogen adsorbed on the metals was titrated by oxygen. A second reduction in hydrogen at room temperature was done, followed by a second titration with oxygen. Only the second titration results were used. The first titrations were slightly modified by oxygen consumption by non precious metal compounds. The dispersion was calculated assuming that Pt and Rh had the same dispersion and that the titration equations were : PtH + 3/4 0 2
+
PtO + 1/2 H20
RhH + 02
+
1/2 Rh2O3 + 1/2 H20
57 1
Electron microscopy was performed on a JEOL 120 CX for the conventional mode and on a VG HB5 for the analytical scanning mode. For these studies catalysts were analyzed directly after engine bench aging without further treatment. Particle size distribution was determined from the measurement of about 200-300 particles. The mean diameter has been determined from the following equation :
Precious metal and poison distribution in the wash-coat was determined by microprobe analysis on a Camebax apparatus. The catalytic activity was determined for cylindrical samples (diameter 30 mm, height 76.2 mm), directly cut in the aged catalysts, on a laboratory test working at stoichiometry and under programmed temperature increase. Gas flowmeters were used to prepare a mixture containing : CO 1.5 %, NO 0.2 %, C3Hs 0.04 %, 0 2 0.85 %, C02 10 %, H20 10 %, the make-up being nitrogen. On-line analyzers allowed conversions of CO (IR detection), C3Hg (FID detector) and NO (chemiluminescence detection) to be followed as a function of temperature. Results have been expressed by the half-conversion temperatures for the three pollutants and by the apparent activation energy for CO oxidation. Half-conversion, or light-off, temperatures were measured at the inlet of the catalyst, as is usually made in engine bench or vehicle tests. For the determination of the apparent activation energy, catalyst temperature was chosen as the mean value between inlet and outlet, for conversions of no more than 10 %. As NO and C3Hs conversions generally took place after that of CO, their activation energies were not determined because of large thermal gradients in the reactor.
RESULTS Five identical catalysts were aged on the engine bench for increasing times of 0.17, 1, 6 , 40 and 200 h (plus respectively 4 and 2 min idle at start and before stopping the engine). At the end of the preparation step the colour of the catalysts is nearly white. The treatment in the exhaust gas at high temperature causes the reduction of the precious metals : the colour turns from white to grey. When comparing the catalysts aged after increasing times the colour lightens increasingly, probably due to the decrease of the exposed metallic surface. The following results will first concern the characterization of the aged catalysts, then their catalytic performances.
572 EVOLUTION OF THE PHYSICO-CHEMICALCHARACTERISTICS
Carrier Automotive post-combustion catalysts require a large porous volume with bimodal distribution in order to allow both the reactants and the products to diffuse and the precious metals to be well dispersed [7]. In the case of catalysts made on ceramic monoliths the porous volume is the sum of the contributions of the monolith and of the wash-coat. The porous volume contribution of the monolith is of the order of 0.18 cm3/g, and is only associated with the presence of macropores. As the manufacture of the monolith includes a calcination step at around 1673 K, its textural properties are assumed to remain unmodified during exhaust gas aging treatment. The porous volume of the catalysts is about 0.35 cm3g and appears to be unsensitive to aging time between 0.17 and 200 h (Fig. I). The macropore diameter remains nearly constant at about 3 microns. Nevertheless the micropore diameter increases from 20 to 40 nm. This increase is associated with a decrease of the surface area of the wash-coat with the aging time (Fig. I). This decrease is nearly proportionnal to the logarithm of the aging time. Various laws are used to model sintering. Those expressing surface area versus aging time are Figure l generally empirical and do not alEvolution of wash-coat sugace area and low the determination of a sintering catalyst porous volume of Pt-Rhl A1203 mechanism. For supported metals versus the logarithm of the time the rate of decrease of accessible area versus accessible area to the nth power (eq 2) is often used [8]. The value of the exponent has been correlated with the sintering mechanism. This expression has already been applied for alumina sintering [9].
These two kinds of law have been used. Considering that the initial wash-coat surface area is 160 m2/g it is possible to model the evolution of the surface area versus time according to the expression:
573
From the logarithmic representation of So - St versus t , it is possible to determine the following expression of S versus t :
If this expression is valid for aging times longer than 200 h, then it would take 1000 h to decrease the wash-coat surface area down to 50 m2/g. The integration of equation (2) leads to equation (5) :
For kt>>l, the slope of the straight line representing Ln (So / Sr ) versus Ln t is equals to l/(n-1) : n is found equal to 14. Values between 8 and 15 have been determined by JOHNSON [9] although no physical significance is given for these high n values. They are interpreted by a model of dehydroxylation at contact points of particles with subsequent formation of new Al-0-A1 bonds. The diffractogram of the original wash-coat shows only the peaks of gamma alumina with that of delta alumina. As the aging time increases the peaks due to theta alumina appear with increasing importance and eventually become the major diffraction lines.
Precious metals
Figure 2 Evolution of metal dispersion of Pt-Rhl A1203 catalysts versus the logarithm of aging time
The dispersion of the precious metals decreases monotonously with the logarithm of the aging time as shown in Fig. 2. Assuming that the dispersion of a catalyst undergoing only the heating and cooling phases of the engine bench aging (aging time equals zero) is 50 %, it is possible to calculate the sintering rate versus the dispersion using equation (2). The calculated exponent value (sintering order) is 4, which is not characteristic of a particular sintering process : it can either be related to sintering by particle (4
514
Nevertheless, several authors [ 11,121 have studied the sintering of Pt/A120 3 catalysts under different atmospheres and concluded that the sintering order is changing from large (8-20) to small values (2-5) after several hours. They associate this observation to a change in sintering mechanism : initially small particles migrate and coalesce, then sintering occurs essentially by atom migration. Because n has a constant value of 4 it seems that the only sintering process for Pt-RWA1203 is that observed for Pt/A1203 for long aging times: atom migration. Rhodium could favor this process because of its larger interaction with alumina than platinum. Electron microscopy (Fig.3) shows a bimodal particle size distribution for 0.17 h aging: a peak at about 1.0 nm and a broad contribution around 4 nm. Small particles of 1.0 nm always remain even after increased aging times, but the tail of the distribution extends to 10 nm and progressively to 70 nm.
200h
0.17h
1
2
3
4
5
Particle sire (nm)
Table. 1 Comparison between particle diameter derived from H2/02 titration and electron t (h)
d(W02) (nm)
d(EM) -
0.17 1 6
3.1 5.7 12 19 25
4.0
40
200
5.6 12 26 50
6
0
10
20
30
40
w7 50
60 7 0
Particle sire (nm)
Table 1 compares the mean diameters derived from titration - assuming d(nm) = 102D as for pure platinum -and from electron microscopy measurements They are generally very close with the exception of those at long aging times where the EM diameters are larger than those derived from titration. l%e large mean diameters obtained from electron microscopy measurements are probably overestimated due to the relatively important statistical weight of the largest particles. the statistical weight of large particles is probably too large.
575
The composition of the particles have been analysed for 0.17 and 200 h aged catalysts (Table 2). Table 2 Pt/Rh atomic ratio versus particle size for 0.17 and 200 h aged catalysts N is the number of analysed particles. d (nm) 3 .O 4 .O 5 .O 6.O
N 0.17 h 5 7 5 1
Pt/Rh (at)
d (nm)
2.1 1.2 2.1 0.6
70
N 200 h 2 2 2 6 1 1
Pt/Rh (at) 1.7 2.6 3.1 3.2 10.4 13.4
All the analysed particles are bimetallic. For the shortest aging time the atomic ratio PtBh seems to decrease when the particle size increases from 3.0 to 6.0 nm. But this variation is not continuous and the narrow range of particle sizes does not allow a clear correlation. The results are more evident for the 200 h aged catalyst. The particle size distribution is very broad and the Pt/Rh ratio increases continuously from 1.7 to 13 when the particle diameter increases from less than 10 nm to more than 70 nm. Poisons (P, Zn, Pb, S ) were not detected in the metallic particles. Thus during aging the metals sinter and form large particles that are enriched in platinum. Rhodium which is very reactive with alumina remains in the small particles where it can better interact with the support. Part of rhodium can be incorporated in alumina in an oxidised state as shown by STENBOM [ 131. Platinum, which is metallic at high temperatures, interacts only weakly with alumina and sinters readily to form large particles depleted in rhodium.
Poisons Automotive exhaust gas is not only composed of hydrocarbon combustion products from gasoline but also of other oxidised compounds coming either from the gasoline ( S , Pb) or from the lubricant (S, P, Zn). These elements are known to be strong poisons of metal catalysts [ 11. Fig. 4 shows the evolution of the mean concentration of these elements in samples taken along the entire length of the catalyst at different aging times on the engine bench. The sulfur concentration in the catalyst is independant of the aging time and is always less than 100 ppm. Lead, phosphorus and zinc
576
concentration increases nearly linearly with aging time. Poison concentration is not only time dependent. It strongly depends on the sample location, either in the length of the monolithic catalyst or in the depth of the wash-coat. After cutting the 76 mm long samples in three parts of about 25 mm each, it appears that the individual poison concentration ratio between the first and third parts is always between 2 and 14. The largest values are obtained for zinc. Microprobe analysis in the depth of the wash-coat shows that zinc and phosphorus concentrations are strongly decreasing from the external to the internal part of the wash-coat. Fig. 5 shows this decrease in a section of the wash-coat cut at 10 mm from the entrance of the 200 h aged catalyst. Lead appears to be more uniformly distributed throughout the washcoat depth.
-
+
E
CL 2000. 2
0
+ + 0
O
50
100 150 AGING TIME (h)
am
0 - P
200
250
Figure 4 Evolution of poison concentration in a Pi-Rhl A1203 catalyst versus aging time
Figure 5 Poison distribution in a section of washcoat cut at I0 nm from the entrance of a Pt-RhlA1203 catalyst aged for 200 hours
Zinc thiophosphates are commonly used antiwear agents in lubricating oils. Though their oxidation products have not been analysed in this study, it appears that zinc and phosphorus are deposited in the same outer regions of the wash-coat. One can assume that these oxidation products interact strongly with the wash-coat because their concentration decreases sharply both as a function of depth in the wash-coat and length from the entrance. The concentrations are almost proportional to the aging time indicating that the catalyst is a captation mass whose efficiency remains nearly constant. Lead is present in gasoline as tetraethyl lead at concentrations less than 2 mg/l (European limit is 13 mg/l). Even at this low concentration the accumulation of lead with respect to the aging time is clearly visible. The only
577
difference between lead and zinc and phosphorus is that lead appears to be more uniformly distributed in the depth of the wash-coat, perhaps due to better diffusion of lead compounds than zinc or phosphorus compounds. Sulfur concentration in gasoline is about 70 mg/l. Sulfur is also present in the oil as thiophosphates. Nevertheless, sulfur concentration in the catalyst is always less than 100 ppm. This is related to the aging conditions: over 1273 K sulfur oxides cannot adsorb on alumina [ 141. The poisons, which are probably in an oxidised state, have not been detected on the metallic particles, as already mentioned by KIM [15]. EVOLUTION OF THE CATALYTIC PERFORMANCES
The evolution of the light-off temperatures for CO, C3H8 and NO versus the aging time on engine bench were performed in laboratory test at stoichiometry. The increase in light-off temperature for CO is approximately linear with the logarithm of aging time between 1 and 200 h (Fig. 6 ). Assuming that the light-off temperature for a sample undergoing only the heating and cooling phases on the engine bench (aging time equals zero) is 498 K, it becomes possible to determine an empirical relationship between Tio and aging time for CO :
According to this relation the increase of light-off temperature associated with a 1000 h aging (very long in regard of the severity of the aging conditions) is 108 K. For NO and particularly for C3Hg, the light-off temperature increases more rapidly with aging time than that for CO, particularly between 40 and 200 h. The apparent activation energy for CO is 24 kcal/mole after 0.17 h aging. At longer aging times, this value increases to about 29 kcaVmole (see Fig. 6 ). CO oxidation is known to be structure insensitive [16]. This means that the turn-over number is independant of the metal dispersion. Fig. 7 shows that the turn-over number for CO oxidation at 550 K, calculated from the temperature required for 10 % conversion, the apparent activation energy and the dispersion of each catalyst, decreases for long aging times. The increase in apparent activation energy versus aging time indicates that although zinc and phosphorus are deposited on the surface of the washcoat they do not form an impervious layer preventing diffusion of the reactants. If this has been the case, the apparent activation energy should have decreased. The observed increase is probably related to modifications of the active sites, such as a variation of the Pt/Rh surface ratio, a variation of the
578
oxidation state of the metals, or chemical poisoning by stored poisons. As the first two phenomena should be associated with sintering, they might be linked with the increase of apparent activation energy at short aging times. On the other hand, poison storage is linearly proportional to aging time and might be associated with the decrease of the turn-over number at long aging times. This effect would probably be more important for catalytic tests performed under reducing gas mixture as was previously observed on a bench engine test of aged catalysts [ 6 ] . Poisons in a reduced state are known to interact more strongly with the precious metals [17].
-
2000
340
X
c
\:oO5l= a: w m
I
3 1000
3
Bz
z
-2
0
2
4
500 07
6
LNt)
Figure 6 CO, NO andC3Hg light-off temperatures and activation energy for CO oxidation for engine bench aged catalysts tested in a stoichiometric mixture
Turn-over number for CO oxidaion at 277 “c versus engine bench aging time with a Pt-Rhl A1203 catalyst
CONCLUSION Aging of a Pt-Rh/Al203 catalyst over increased time, in the exhaust of an engine bench, leads to the transformation of gamma and delta alumina to theta alumina. Small pores coalesce in larger ones but the total porous volume remains constant. This causes the surface area to decrease by a factor of two after 200 h. Phosphorus, zinc and lead compounds deposit at a steady rate on the external part of the wash-coat. Poison deposition does not seem to induce diffusion limitation for CO oxidation at low conversions up to 200 h aging time.
579
Precious metals particle size increases from about 4.0 nm to 70 nm as the aging time increases from 0.17 to 200 h. Rhodium remains in the small particles and additionally, though not proven, transfers to alumina. Sintering of the precious metals is probably the major cause of activity decrease during the shortest aging times. Above 40 h aging, poisoning by phosphorus, zinc and lead stored on the catalyst is probably responsible for the decrease of the turnover number for CO oxidation. REFERENCES 1 2 3
4 5
6 7 8 9 10 11 12 13 14 15 16 17
K. C. TAYLOR, Catal. Sci. and Tech. 5 (1984) 120 H. C. YAO, H. S. GANDHI and M. SHELEF, Metal-support and Metal-additive effects in catalysis, Stud. Surf. Sci. Catal. 11 (1982) 159 L. L. HEGEDUS, J. C. SUMMERS, J. C. SCHLATTER and K. BARON, J. Catal. 56 (1979) 321 R. K. HERZ, E. SHINOUSKIS, A. DATYE and J. SCHWANK, Ind. Eng. Chem.Prod. Res. Dev. 24 (1985) 6 R. A. DALLA BETTA, R. C. Mc CUNE and J. W. SPRYS, Ind. Eng. Chem.Prod. Res. Dev. 15 (1976) 169 G . MABILON, D. DURAND and M. PRIGENT, Sci. Tot. Env. 93 (1990) 223 P. NORTIER and M. SOUSTELLE, Catalysis and Automotive Pollution Control, Stud. Surf. Sci. Catal. 30 (1987) 275 E. RUCKENSTEIN and B. PULVERMACHER, J. Catal. 29 (1973) 224 M. F. L. JOHNSON, J. Catal. 123 (1990) 245 S. KIM and M. J. DANIELLO Jr., Appl. Catal. 56 (1989) 23 D. D. BECK and C. J. CARR, J. Catal. 110 (1988) 285 P. J. F. HARRIS, J. Catal. 97 (1986) 527 B. STENBOM, G. SMEDLER, P. H. NILSSON, S. LUNDGREN, G. WIRMARK, SAE Paper no 900273 S. W. NAM and G. GAVALAS, Appl. Catal. 55 (1989) 193 S . KIM and M. J. DANIELLO Jr., Appl. C a d . 56 (1989) 45 M. BOUDART and F. RUMPF, React. Kinet. Catal. Lett. 35 (1987) 95 J. BARBIER, Deactivation and Poisoning of Catalysts, J. Oudar and H. Wise eds, Dekker, New-York, 1985
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZI 0 1991 Elsevier Science Publishers B.V., Amsterdam
58 1
CHARACTERIZATION OF BIMETALLIC SURFACES BY 180/160 ISOTOPIC EXCHANGE. APPLICATION TO THE STUDY OF THE SINTERING OF PtRh/A1203 CATALYSTS. S. Kacimi (ab) and D. Duprez (a)
(a)Laboratoire de Catalyse en Chimie Organique, 40 Avenue du Recteur Pineau 86022 Poitiers France (b)Present adress :University of Sidi-Bel-Abbes, Algeria. ABSTRACT A series of PtRh/A1203 catalysts designated PtRhx, where x is the atomic percentage %Rh/Rh+Pt was prepared by successive impregnation of a delta-alumina, pretreated in H2 at 850"C, with aqueous solutions of H2PtC16 and RhC13. They were calcined at 450°C (fresh catalysts) and then sintered (2% vol.O2/Ar for 2h) at 700, 800 and 900°C. They were characterized by hydrogen chemisorption HC and oxygen titration and by oxygen isotopic equilibration (1602 + 1802 3 2 160180). The surface compositions were evaluated from the total metal surface area deduced from and from the rhodium surface area deduced from re (rate of equilibration at 300°C). re being very sensitive to the presence of chlorine, the catalysts were dechlorinated in a stream of H2 + H2O at 400°C. This treatment induces a decrease of the metal area of Pt while there is no change for Rh. No surface enrichment is found in these catalysts : a linear variation of re with x is obtained from pure Pt (0.16 x 1019 at.0 min-1m-2) to pure Rh (3.06 x 1019 at.0 min-1,-2). On the contrary a significant change in the surface composition is observed for the sintered catalysts which are strongly enriched in Rh at high Rh content and in Pt at low Rh content. The inversion of composition occurs for x = 25 Rh at.-% at 700°C and for x = 40 Rh at.-% at 900°C. Two models are proposed to explain these results; they take into account the structural and morphological changes of Rh and Pt in oxidizing atmosphere as well as the degree of interaction between the two composents of the bimetallics. INTRODUCTION
Surface composition of bimetallic catalysts in an essential parameter for their catalytic performances. This is particularly true for the PtRh system used in exhaust gas catalysis since each of the components possesses a specificity with respect to the oxidation reactions (platinum) or the reduction reactions (rhodium) (ref.1). The surface composition of bimetallics can be measured by physical methods (XPS, ISS ...) (ref.2). Complementary methods (X-Ray Diffraction, Temperature-Programmed Reduction) can also be used
582
for obtaining information on the degree of alloying in the bimetallics. Physical methods are well adapted to the study of bulk catalysts (powder, film...) or of highly loaded supported catalysts whereas physico-chemical methods are more effective for highly dispersed supported catalysts with a low metal loading. Nevertheless physico-chemical methods require the stoichiometries of chemisorption of the probe molecules on each metal to be quite different. Recently we showed that the precious metals used in threeway catalysts could promote the isotopic exchange between gaseous oxygen and the surface oxygen of the support and this reaction was used for the study of the oxygen mobility on the catalyst surface (ref.3). Moreover we showed that the same metals could catalyze the isotopic equilibration reaction : 1602 + 1802 2 160180 (1) As rhodium and platinum presented very different intrinsic rates of equilibration, it was decided to use this reaction to measure the surface composition of PtRh/AlzOs catalysts. The method was applied to a series of catalysts calcined at 450°C and then sintered at 700°C and 900°C.
*
PRINCIPE AND EXPERIMENTAL
The measurement of the surface composition is based on the two complementary experiments : (i) the determination of the total surface of the metals by appropriate chemisorption or titrations at room temperature, after reduction of the catalysts at 450°C.Hydrogen chemisorption (Hc) and hydrogen titration of chemisorbed hydrogen (OT) were used to determine the dispersion D, and the total surface area Am, of platinum and rhodium. We have : Am (m2 metal g-1) = a + b (2) where a and b represents the surface areas of platinum and of rhodium respectively (ii) the determination of the rate re of oxygen equilibration. For monometallics, re was shown to be proportional to the metal surface area. Assuming that there is no synergy effect between Pt and Rh, re will be equal to the sum of the contribution of each component in the bimetallics, i.e. : re = aa + pb (3) where a and p are the intrinsic rates of equilibration of platinum and of rhodium respectively. a and p being determined separately with monometallic catalysts, the measurement of Am and re give a and b by resolving the system (2) + (3). Eq.3 can be expressed in terms of intrinsic rate of equilibration (per m2 of metal area) : r*e = re/a+b = ~1 (a / a +b ) + p ( b / a +b ) or r*e = 01 (1 - XS)+ p XS (4) where xs represents the fraction of the surface occupied by rhodium atoms which is close to the atomic percentage of rhodium at the surface. Moreover,
583
if a linear plotting of r*e with x, atomic percentage of rhodium in the catalyst, is found, i.e. : r*e = m(l - x) + n x (5) then m = a,n = p and x = xs. This means that there is no surface enrichment in the bimetallics. For eq.4 to be valid, there must be no synergy effect between platinum and rhodium on the isotopic equilibration reaction, i.e. each surface atom behaves in the same manner as in the monometallics. A series of PtRWAlzO3 catalysts (E 1% metal) was prepared by coimpregnation of an alumina (pretreated at 900°C before use) with aqueous solutions of chloroplatinic acid and rhodium chloride. After drying at 120"C, the catalyst samples were calcined in air flow at 450°C. They are referred to as PtRhx where x is the atomic percentage of rhodium. Aliquot portions of these catalysts were sintered at 700°C and at 900°C in oxygen (oxygen pulses injected every 30s in an argon flow for 2h, corresponding to a mean value 0 2 of 2%).The sintered catalysts are designated as PtRhx700 or PtRhx900. Hydrogen chemisorptions and oxygen titrations were carried out in a pulse chromatographic apparatus using ultrapure argon (less than 1 vpm impurities) as a carrier gas (ref.4). H c values were corrected from the amounts of weakly adsorbed hydrogen (10 min. desorption). Temperatureprogrammed reduction (TPR) were carried out in the same apparatus. Isotopic equilibration measurements were made in a recycle reactor described elsewhere (ref.3). The catalyst samples (0.02 to 0.5g) were first treated at 450°C in natural oxygen in order to eliminate any carbonaceous impurities, then reduced in a flow of H2 at the same temperature (lh) and subsequently outgassed (450"C, lh, 10-4 mbar).They were cooled down to 300°C for isotopic equilibration measurements. (50 mbar, 1602 50% + l * 0 2 50%). The rates were deduced from the initial slopes dP34/dt of the formation of 160180.
RESULTS
Fresh catalysts (calcined at 450 "c) The characteristics of the chlorinated samules are given in Table 1 which shows that all catalysts are relatively well-dispersed and that their chlorine content is randomly distributed as a function of the atomic percentage of rhodium (x in PtRhx). The values of OT are found in accordance with those of HC as far as pure rhodium or rhodium-rich catalysts are concerned : a stoichiometry of OT : HC = 2 : 1 is then obtained in agreement with previous results on rhodium catalysts (ref.6 and 7). On the contrary, there is no agreement between the HC and OT values on pure platinum or platinum-rich catalysts, the stoichiometry H : Pt being close to 2. This is most likely due to the very high dispersion of platinum in these
584
catalysts (ref.8). That is why the dispersion characteristics (D and Am) given in table 1 were calculated from the values of OT. TABLE 1
Characteristics of the chlorinated catalyst samples (calcined and reduced at 450°C).
Pt Rh 12 Pt Rh 19 Pt Rh 28 Pt Rh45 Pt Rh 100
0.50 0.00
Rh wt % 0.00 0.036 0.061 0.1 1 0.22 0.5 1
wt % 0.49
pmole g-1 38 I 32
0.7 1 0.76 0.59
52 46
%
95
69
m2g-1 1.07 1.52 1.84 2.00
The TPR profiles of the chlorinated samples are shown in Fig.1. Rhodium reduces more easily than platinum in these catalysts : the difference between the two temperatures TM for the maxima of reduction reaches 100°C. However both metals seem to reduce together in the bimetallics : in every case a single peak of rduction is observed with a maximum increasing regularly from pure rhodium to pure platinum (Table 2). Moreover, the initial oxidation state of the metals is close to +4 for platinum and to +3 for rhodium and a regular change of the H/M values is observed from pure platinum to pure rhodium in the bimetallics. pmoi H min-lg”
Fig.1 TPR profiles of the chlorinated PtRhlA1203 catalysts. 1 : WA1203 ;2:PtRh45 ; 3 :Pt Rh28 ;4 : PtRh 19; 5 : Pt Rh 12; 6 : PdA1203
585
TABLE 2
Temperatures TM of the maxima of reduction and amounts of hydrogen taken by the chlorinated PtRh/A1203 during TPR from 25 to 450°C.
Dechlorinating the catalvsts (to less than 0.15 wt-% C1) in a stream of
H2
+ 10%H 2 0 results in a significant change in the metal accessible fraction
(Fig.2). Nevertheless the dechlorination treatment affects essentially platinumcontaining catalysts and has virtually no effect on rhodium. Temperature programmed reduction of dechlorinated catalysts shows that the metals are almost completely reduced after the dechlorination treatment : hydrogen uptakes at 25°C are observed on Pt and PtRh catalysts whereas Rh/A1203 shows a single peak for about 100°C (Fig.3). These hydrogen uptakes can be
Fig.2 :EJrfect of the dechlorination in H2 + 10% H 2 0 on the metal sulface area of the PtRhlA1203 catalysts. Cl = chlorinated samples ;DCl = dechlorinated samples.
586
ascribed to hydrogen titrations of chemisorbed oxygen on the metals : it is well-known that these titrations occur readily at room temperature on Pt catalysts whereas on Rh, they require higher temperatures for them to be attained. mol H min-lg-’
-o l
Fig.3 :TPR profiles of PtRh catalysts a f e r dechlorination a - RhlA1203 ;b - PtRh 19 ;c - PtIAl203. 1 6 0 / 1 8 0 isotopic equilibration experiments were carried out on the chlorinated and on the dechlorinated samples. The changes of r*e (rate of equilibration) with the atomic percentage of rhodium in the bimetallics are shown in Fig.4. The presence of chlorine in the catalyst provokes a very important decrease of r*e (by a factor of 10). Moreover no correlation can be found between r*e and the composition of the bimetallics (Fig.4a). As shown in Table 1, the chlorine content of the catalysts amounts to 0.62 f 0.15 wt-% but it seems that these small changes in C1 content can induce large variations of the rates of equilibration. On the contrary, a linear plotting of r*e with the atomic percentage of rhodium is obtained on dechlorinated samples (Fig.4b). This means that the surface composition of the dechlorinated, calcined bimetallics is very close to the bulk composition. This shows also that, there is no synergy effect, between Pt and Rh in this samples. A linear plotting r*e vs x would have been obtained only in the case of a hypothetical compensation between a surface enrichment in platinum and a synergy effect of the rhodium on the platinum. This would increase the activity of this latter metal. This unlikely hypothesis, can be rejected. We can thus consider that the equations 3 and 4 are valid for PtRWA1203 bimetallics with : 01 = 0.16 1019 at. 0 min-l(m2 Pt)-1 and p = 3.06 1019 at. 0 min-l(m2 Rh)-1 at 300°C.
587
r:
1 0 ' ~atoms
o
min-1 m2m-,m,
a
I
0
.
.
.
50
.
50
100
x
100
Rh at.-%
Fig.4 :Variations of the rates of 160t180 equilibration with the atomic percentage of rhodium in the PtRhlAl203 bimetallics. (a - chlorinated samples ;b - dechlorinated samples). Sintered catalysts The chlorinated catalysts were sintered in an oxidizing dry atmosphere at different temperatures. Their relative metal surface areas are given in Table 3. TABLE 3
Effect of the sintering (pulses of 0 2 in Ar for 2h at T"C corresponding to a 2% 0 2 average composition) on the relative metal surface area. The reference is the metal surface area of the catalyst calcined at 450°C denoted Am (450). Catalysts
Am(450) m2/g
Pt Rh 0 PtRh 12 PtRh 19 Pt Rh 28 Pt Rh 45 Pt Rh 100
1.07 1.08 1.29 1.52 1.84 2.00
Am(T)/Am(450), 96 T = 700°C T = 800°C T = 900°C 15.5 16.6 15.5 15.8 16.5 15.8
7.3 7.3 7.7 8.2 10.2
7.4
1.6 1.8
2.4 2.7 2.8
3.7
588
There is a continuous decrease of the metal surface area with the sintering temperature. At 700"C, the rate of sintering does not depend on the catalyst composition while at higher temperature, the effect of this composition is noticeable. At 8OO"C, it seems that the bimetallics resist better to sintering whereas at 900"C, there is an acceleration of the sintering rate for platinum and platinum-rich catalysts. These results will be discussed in the light of the different models of sintering for platinum and for rhodium. In the first column of Table 4, are given the values of the rates of equilibration on catalysts sintered at 700°C. These values show a non-coherent change of re with the rhodium content of the catalysts. Moreover the value of a (intrinsic rate of isotopic equilibration of Rh) calculated from the data obtained with PtRh 100 (2.66 x 1018 at. 0 min-1 g-1 for a rhodium surface area of 0.31 m2 g-1) is much smaller than the value obtained with the standard non-sintered catalyst (0.86 x 1019 instead of 3.06 x 1019 at. 0 min-1 m-2). It seems thus that the sintered catalysts still contain chlorine which can poison the equilibration reaction. Therefore this series of sintered catalysts was dechlorinated in a stream of H2 + H20. The results obtained after this treatment are reported in the second column of Table 4.
llUiLE4 Rates of 1 6 0 / 1 * 0 isotopic equilibration at 300°C on catalysts sintered at 700"C, before (a) and after (b) a further treatment of dechlorination (H2+H20 at 400°C). dechlorinated Pt Rh 0 Pt Rh 12 Pt Rh 19 Pt Rh 28 Pt Rh 45 Pt Rh 100
(4 sintered at 700°C 0.20
(b) sintered and dechlorinated 0.21
2.66
6.95
The total surface areas of metals (Pt+Rh) were not controlled after dechlorination. Nevertheless the values of a and p calculated from the metal areas measured with the non dechlorinated samples are 0.13 x 1019 and 2.3 x 1019 at. 0 min-1 per m2 of Pt and of Rh, respectively. These values, coherent with those measured on the fresh catalysts, were used for the determination of the surface composition of the samples sintered at 700°C. Similar measurements were carried out on the catalysts sintered at 900°C. The intrinsic rate of isotopic equilibration found for Rh (2.2 x 1019 at. 0 min-1 m-2) is coherent with the value obtained on the fresh catalyst.
589
Moreover no change was noted after a further treatment of dechlorination, which shows that the sintering at 900°C eliminates practically all the chlorine contained in the catalyst
X,
~h surface %
I
50
0
100
0
.
.
.
50
.
1 0
Fig.5 :Sulface composition of the PtRhlAl203 bimetallics sintered at 700"C (left) and 900 "c (right). The surface compositions deduced from the 160/1*0 isotopic equilibration rates are shown in Fig.5 for the catalysts sintered at 700 and 900°C. In every case, two kinds of phenomena are observed : (i) the metal surfaces are enriched in at least one of the metals (ii) an inversion of the surface composition can occur for a rhodium content x which depends on the sintering temperature. For x < 25 at.-% (catalysts sintered at 700°C) or x < 40 at.-% (catalysts sintered at 900"C), the bimetallics are strongly enriched in platinum while the opposite tendency (enrichment in rhodium) can be observed for x>25 or 40 at.-% . DISCUSSION
Chemical state of metals during isotopic equilibration During the experiments of isotopic e uilibration, the reduced catalysts are placed in contact with a dose of 1602 + 1 0 2 at 300°C. It has already been shown that, at this temperature, a surface oxidation occurs for platinum (ref.10-12) while rhodium is most probably oxidised in the bulk (ref.6,8,1315). The question arises whether or not the method itself could change the surface composition of the bimetallics. The results obtained on the fresh catalysts seem to show that, this change of surface composition, if it occurs, is
1
590
rather small inasmuch as no apparent surface enrichment is found in these samples. Moreover it can be argued that the characterization method by 160/*80 isotopic equilibration gives the surface composition of the bimetallics in oxidizing atmosphere, which is of the utmost interest in exhaust gas catalysis. After sintering at 700-9OO0C, there is a significant change of the surface composition with a rhodium enrichment for Rh/Rh+Pt > 25 at.-% (700°C) or 40 at.-% (900°C). Similar surface enrichments in Pt-Rh alloys or in supported Pt-Rh catalysts have been found by Williamson et al. (ref.16) and by Schmidt and coworkers (ref.17-18). These works report that apparently the rhodium surface enrichment occurs, even at low Rh content, on unsupported PtRh alloys. The inversion of surface composition found in the present study would be specific of aluminasupported catalysts.
Role of the support in the surj5ace segregation of PtRh bimetallics It is generally assumed that in an alloy treated in oxidizing atmosphere the element forming the most stable oxide segregates to the surface (ref.19). This is the case for Rh in PtRh alloys which forms Rh2O3 stable up to about 900°C while Pt02 decomposes beyond 550°C. The presence of platinum can accelerate the decomposition of Rh2O3 which starts then at 800°C (ref.20). However this does not change the general tendency for Rh to segregate to the surface of PtRh in 0 2 or air. The inversion of surface composition for low Rh contents can be related to the ionic mobility of Rh3+ ions in the alumina matrix : rhodium diffuse in alumina forming a "diffuse oxide phase" (DOP) no longer accessible to gases and, in particular, non-reducible in H2 (ref.8 and 14). Table 5 shows the change of the DOP content in Rh/A1203 with the sintering temperature. The DOP corresponds to the fraction of rhodium non reducible at 450°C, the reducible fraction being determined by oxygen uptakes at 500°C assuming a reoxidation of Rho into Rh2O3. TABLE 5
Degree of reduction of Rh (%R) in Rh/A1203 as a function of the sintering temperature Tox ("C). Tox ("C)
%R
450 (fresh) 100
700 18
800 9
900 4
It appears that the proportion of diffuse oxide phase depends largely on the rhodium content in the catalyst : the degree of reduction of a 2% Rh/A1203 catalyst was, after a treatment in oxidizing atmospher, 63% at 700"C, 26% at 800°C and 22% at 900°C (ref.8). Thus the proportion of
59 1
rhodium lost by diffusion in alumina increases significantly when the rhodium content is decreased. Two models, differing by the degree of interaction between Pt and Rh in the metal particles, can be proposed to explain the results of the present study : - In the model "with interaction", Pt and Rh are present together in the metal particles. Taking into account the TPR profiles shown in Fig.1, this is a realistic model, at least for the catalysts calcined at 450°C. Upon sintering in 0 2 at high temperature (700-9OO0C), rhodium oxidizes and segregates to the surface of the particles. Simultaneously part of the rhodium ions diffuse into alumina. For low Rh content, the fraction of this element remaining at the support surface is close to 0, the external particles becoming extremely rich in platinum because there is practically no longer Rh in the metal particles. For higher Rh contents, the fraction of rhodium remaining in the particles increases. As this element is segregated in the outer layer of the particles, the surface becomes enriched in rhodium. - In the model "without interaction", Pt and Rh are located in separate particles. The apparent enrichment in platinum at low Rh-content would be caused by the same phenomenon as in the model "with interaction", i.e. a significant diffusion of Rh3+ ions in the alumina matrix. The apparent enrichment in rhodium in Rh-rich bimetallics would be due to a good dispersion of the Rh2O3 phase remaining accessible at the alumina surface while Pt particles would suffer a severe sintering. For this model to be valid for catalysts calcined at 450°C, the particle sizes of Pt and Rh must be identical since the surface composition was found very close to the bulk composition. This is not the case in dechlorinated catalysts for which platinum was shown to sinter during the H2/H20 treatment. Thus it seems the most likely that in the fresh catalysts, Pt and Rh would form alloy particles. However in the course of sintering at high temperature, we could have a phase separation and with these catalysts, the two models (with or without interaction) could coexist. Further studies, including also the role of steam, are presently in progress to examine this eventuality. ACKNOWLEDGEMENTS
We thank the Groupement de Recherche sur les Catalyseurs de postcombustion automobile" (Centre National de la Recherche Scientifique, Institut FranGais du Ptrole et Agence FranGaise pour la Maitrise de 1'Energie) for its financial support. MM. Prigent, Mabilon and Gravelle are thanked for their constant interest and helpful discussions concerning this work.
592 REFERENCES 1. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
A. Crucq and A. Frennet (Eds.) "Proceeding of the 1st International Symposium on Catalysis and Automotive Pollution Control, Brussels, 1986" Stud. Surf. Sci. Catal. Vo1.30, Elsevier Publ., Amsterdam (1987). F. Delannay and B. Delmon in "Characterizationof heterogeneouscatalysts" Base1 (1984). (F. Delannay Ed.) p.1, M. Dekker Publ., New York and H. Abderrahim and D. Duprez, p.359 of ref.1. D. Duprez, J. Chim. Phys. 80 (1983) 487. H. Abderrahim and D. Duprez in "Proc. 9th Int. Congr. Catalysis, Calgary, 1988" (M.J. Philipps and M.Ternan, Eds) Vo1.3 p.1246. The Chem. Inst. Canada, Ottawa, 1988. T. Paryjczak, W.K. Jozwiak and T. Goralski, J. Chromatogr. 166 (1978) 65 and 75. D. Duprez and A. Miloudi, J. Catal. 86 (1984) 441. D. Duprez, G. Delahay, H. Abderrahim and J. Grimblot J. Chim. Phys.83 (1986) 465. M. Kobayashi, Y. Inoue, N. Takahashi, R.L. Burwell Jr, J.B. Butt and J.B. Cohen, J. Catal., 64 (1980) 74. H.C. Yao, M. Sieg and H.K. Plummer Jr, J. Catal., 59 (1979) 365. H. Lieske, G. Lietz, H. Spindler and J. Vlter, J. Catal., 81 (1983) 8. J. Barbier, D. Bahloul and R. Szymanski, Bull. Soc. Chim. F., 1(1988) 478. S.E. Wanke and N.A. Dougharty, J. Catal., 24 (1972) 367. H.C. Yao, S. Japar and M. Shelef. J. Catal. 50 (1977) 407. C. Wong and R.W. Mc Cabe, J. Catal., 119 (1989) 47. W.B. Williamson, H.S. Gandhi, P. Wynblatt, T.J. Truex and R.C. Ku Aiche, Symp. Ser. Nx201,76 (1980) 212. M. Chen, T. Wang and L.D. Schmidt, J. Catal., 60 (1979) 356. T. Wang and L.D. Schmidt, J. Catal., 71 (1981) 411. F.L. Williams and M. Boudart, J. Catal., 33 (1973) 438. A.J.S. Chowdhury, A.K. Cheetham and J.A. Cairns, J. Catal., 95 (1985) 353.
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 1991 Elsevier Science Publishers B.V., Amsterdam
593
THE EFFECT OF SULFUR ON THREE-WAY CATALYSTS David R. Monroe, Martin H. Krueger, Donald D. Beck and Michael J. D'Aniello, Jr.1
Physical Chemistry Department General Motors Research Laboratories Warren, M I 48090 U.S.A. 1Present address: Johnson Matthey Inc., Wayne, PA. ABSTRACT Both pelleted and monolithic catalysts have been tested in a laboratory reactor under conditions designed to simulate the operation of the catalyst in a vehicle. The warmed-up and light-off performances of both types of catalyst were deteriorated by the sulfur. The monolithic catalyst rapidly recovered all of the lost activity when sulfur was removed from the feed gas, but the pelleted catalyst only recovered a portion of the lost activity. Examination of the effect of sulfur on Pt, Pd, and Rh indicated that both Pt and Rh rapidly recovered all of its lost activity when sulfur was removed from the feed, but Pd did not. Additionally, a pelleted FVRh catalyst which did not contain any Ce also rapidly recovered all lost activity when sulfur was removed from the feed, but a similar catalyst which contained Ce did not. Hence, the slow recovery from sulfur poisoning is related both to the the high Ce surface area seen in pelleted catalysts and to the presence of Pd. The impact of the sulfur decreased when the cycling frequency was increased and the cycling amplitude decreased, indicating that sulfur will have less of an effect on cars with better A/F control. This result was confirmed in tests which used engine exhaust.
INTRODUCTION
Sulfur present in all commercially available gasoline, is widely acknowledged to deactivate three-way automobile catalysts in a laboratory environment. However, the extent and the implications of this deactivation on cars are not well understood. In the past, indications that sulfur was contributing to vehicle emissions were not sufficient to induce reduced sulfur levels in the fuel. Now, with the simultaneous drive to reduce the statutory limit on vehicle emissions as well as a movement to reformulate the composition of gasoline, it is important that we revisit sulfur poisoning to learn if emissions can be significantly reduced by reducing the sulfur level in fuel. Ultimately, the answer to the question of the impact of sulfur levels on vehicle emissions must come from car tests. However, to determine what car tests need to be run, how to run the tests and how to interpret the results,
594
some knowledge of sulfur poisoning under well controlled conditions will be helpful. To this end, this work was undertaken to examine some of the dynamics of sulfur poisoning, examining the effect of the sulfur as a function of time when the sulfur concentration is changed. Many of the previous studies on the effect of sulfur on automotive catalysts examined oxidation catalysts operating in an oxidizing environment. These catalysts usually contained only Pt and Pd supported on 141203. The three-way emission systems which are in use today are far more complex than the oxidation systems used 10 years ago. The three-way catalysts contain Pt, Rh, and sometimes Pd, but also contain up to = 30% Ce, as well as other additives such as La, Fe, Ni, Ba or Nd. The exhaust in which these catalysts operate is also more complicated than in the oxidation systems since the exhaust of three-way systems cycles rapidly between oxidizing and reducing conditions. All of these factors affect the nature of the interactions of the sulfur with the catalyst since the sulfur species which are present, be they S 0 2 , SO3, SO3=, H2SO4, SO4=, S=, or H2S, change with the exhaust stoichiometry, the temperature and the composition of the catalyst. Sulfur dioxide has frequently been shown to reduce the activity of three-way automotive catalysts. Summers and Baron (l),in a laboratory study using simulated exhaust, found that SO2 severly inhibits NO reduction by Pt and Pd catalysts, but has less of an effect on Rh catalysts. They also found that SO2 inhibits rich HC oxidation by all three of these noble metals and inhibits the CO oxidation by Pt. Joy et al. (2) found that SO2 has very little effect on catalyst activity under the cyclic conditions typically experienced in three-way automotive exhaust, but that it did reduce the oxygen storage capacity of the catalyst. This study was also a laboratory study. Furey and Monroe (3) found that when they increased the sulfur content of gasoline used for FTP tests of a car from 0.01 wt% to 0.03 wt%, the conversion efflciency of the catalyst fell 0.6% for HC's, 2.8% for CO, and 2.0% for N 0 2 . The conversions efficiencies fell an additional 2.8% for HC, 4.2% for CO, and 1.8% for NO2 when the fuel sulfur content was increased to 0.09 wt%. This study, however, used a 1980 car with a pelleted catalyst and it should be noted that both the catalysts and the engine control systems used in today's cars differ greatly from those used in that study. Williamson et al. (4),in a study that involved both laboratory and vehicle tests, found that SO2 reduced the performance of three-way catalysts in the laboratory and on cars with a rich calibration, but had very little effect on cars with a stoichiometric calibration. Sulfur poisoning of automotive catalysts is a more significant problem under rich and stoichiometric conditions than under lean conditions. There are two reasons for this. First, sulfur is a particularly severe poison for the water-gas shift ( 2 3 ) and steam reforming reactions (2). Under lean conditions there is an abundance of oxygen to remove both HC's and CO, hence, the water-gas shift and steam reforming reactions are not necessary to
595
remove HC and CO. Both of these reactions can be crucial to the rich and stoichiometric conversions where there is no longer a surplus of oxygen. The second reason that sulfur is a less severe poison under oxidizing conditions is that the sulfur can be removed from the surface of Pt by oxidation. Recently, there has been great interest in sulfur interactions with threeway catalysts, not because of its impact on the activity of the catalyst, but rather because of objectionable H2S emissions which occur under some conditions. The source of the odor has been attributed to sulfur species which are stored on the catalyst during stoichiometric or lean operation, then the subsequent reduction of these species during rich operation to form H2S (6). It has been found that sulfur storage by the catalyst increases as the Ce surface area of the catalyst increases. The H2S emissions can be reduced by several courses of action: the inclusion of a S= scavenger in the catalyst (such as Ni) (7), decreasing the Ce surface area (8), improving the A/F control (9,10), or reducing the sulfur content of the gasoline. This study examines the changes in activity exhibited by three-way catalysts with step changes in the SO2 feed concentration. The response of pelleted and monolithic commercial three-way catalysts is studied. The responses of model catalysts (Pt, Pd, Rh, Pt-Rh, and Pt-Rh/Ce) are also investigated to determine the role of each component.
EXPERIMENTAL Two commercial three-way catalysts were used for most of this study. The first was a pelleted three-way catalyst produced by W. R. Grace & Co. containing 0.11 wt% Pt, 0.045 wt% Pd, 0.014 wt% Rh, and 2.6 wt% Ce. The second, a commercial monolithic catalyst with 62 channeldcm2, contained 34 yg/cm3 Pt and 4.1 pg/cm3 Rh (30 g/ft3 Pt & 3.6 dft3 Rh) 4.7 wt% Ce and 2.5 wt% La, In addition to these two commercial catalysts, a series of noncommercial pelleted catalysts were prepared to determine the impact of composition on the sulfur tolerance of a catalyst. These catalysts contained 0.1 wt% Pt, 0.1 wt% Pd, 0.01 wt% Rh, or 0.1 wt% Pt + 0.01 Wt% Rh, all without Ce, or 0.1 wt% Pt + 0.01 wt% Rh + 5 wt%Ce. Both fresh and thermally aged samples were used for the experiment. Aging was accomplished by heating a fresh sample to 1000°C in a tube furnace (2.5 cm I.D.) in a feed which was cycled at 0.05 Hz between 5% 0 2 / N 2 and 5% H2/N2 for five hours. The gas rate was approximately 0.7 L/minute and it was passed through a bubbler prior to the catalyst to humidify the gas. The catalyst was cooled in the cycled feed. Activity measurements were obtained using a laboratory bench reactor. Three different tests were performed: (1) a dynamic test in which conversions of EC, CO, and NOx were monitored as a function of time using a timeaveraged stoichiometic mixture stream was net oxidizing and the other was
596
net reducing; (2) a net oxidizing light-off test in which conversions of HC, CO, and NO, were monitored as a function of temperature while cooling the catalyst at a rate of 5°C per minute from an initial warmed-up state; and (3) a stoichiometric light-off test similar to (2) only using the cycled feed of (1). Table 1 lists the net oxidizing and net reducing feed streams used throughout the experiment.
I
Table 1 : Feed Stream Compositions Net Oxidizing Feed
Net Reducing Feed
10.0 vol% c02 10.0 vol% c02 0.77 vol% CO 0.77 vol% CO 0.20 vol% H2 0.20 vol% H2 10.0 vol%H20 10.0 vol%HzO 300 ppm HC* 300 ppm HC* 500 ppm NO, 500 ppm NOx 1.0 vol% 0 2 0.2 vol% 0 2 0,5,20 or 90 ppm SO2 0 , 5 , 2 0 or 90 ppm SO2 Balance N2 Balance N2 h = 0.40 h = 1.80 I
* as propane (C3H8:
ir propylene(CgH6)
The reactor consisted of a 2.2 cm I.D. quartz tube inside of a tube furnace. The catalyst was mounted inside of the tube directly following the heating section of the furnace and was wrapped with Fiberfrax insulation. When pelleted catalysts were used 2.6 cm of the tube was filled with 10 cm3 of catalyst. When monolithic catalysts were used a portion of monolith 1.27 cm x 1.27 cm by 2.5 cm long (4cm3) was tested. The volume between the monolith and tube wall was packed with quartz wool to prevent the gas from bypassing the catalyst. The flow to the reactor was 8 L/min on a dry basis. This represents a space velocity of 48 000 h-1 for the pelleted catalysts and 120 000 h-1 for the monolithic catalysts. The impact of sulfur on the warmed-up cycled performance of a vehicle-aged (80 OOO km) commercial monolithic catalyst was also determined in the exhaust from a 1986 3.8 L engine. This engine operated under openloop computer control in order to give the desired cycle frequencies and amplitudes.
597
RESULTS AND DISCUSSION Pelleted Catalysts Oxidizing Light-off Results Figures I and 2 show the effect SO2 has on oxidizing the light-off for both the fresh and aged three-way catalysts using propylene as the hydrocarbon constituent. Light-off (defined as the temperature at which 50% conversion occurs) was determined initially without SO2 in the feed stream, then with 20 ppm S02, and then again without SO2 in the feed stream. Listed in Table 2, are the results from this experiment.
Table 2 : Effect of SO2 on Fresh and Aged Lean Light-Off of Pelleted Pt/Pd/Rh catalysts, "C (CO & Propylene)
I
Sample Fresh Aged
I
Before SO2 With 20ppm SO2 After SO2 Propylene CO Propylene CO Propylene CO 302 298 271 263 259 249 293 276 325 317 303 291
1
1
,
I
Evident from the data presented in Table 2 is the poisoning effect of S 0 2 . For both the fresh and aged samples, the temperature for 50% conversion of CO and propylene is greatly increased when 20 ppm of SO2 is introduced into the feed stream. Specifically, with the addition of 20 pprn S 0 2 , the CO light-off temperature increased 49°C in the case of the fresh catalyst and 41 "C for the aged sample, while propylene light-off temperature increased 43°C and 32°C for the fresh and aged catalysts, respectively. An indication of the magnitude of this effect can be gained by comparing the above mentioned results with the increase in light-off temperature resulting from the thermal aging procedure (i.e., 34°C for propylene and 27 "C for CO). After SO2 was removed from the feed stream, CO and propylene lightoff temperatures decreased; however, not to the original light-off temperatures. For example, the fresh sample exhibited a CO light-off temperature of 259°C before it was exposed to 20 ppm SO2 and 271°C after exposure to the S02. This suggests that the SO2 poisoning is having two effects on the catalyst. One of these effects is reversible, poisoning the catalyst only when SO2 is present in the feed. The other effect is much more persistent, causing a hysteretic effect on the catalyst. Figures 3 and 4 show the effect SO2 has on oxidizing light-off for both the fresh and aged three-way catalysts using propane as the hydrocarbon
598
100
HC z
0 -
$ w
50-
0 0
R
BEFORE SO2
ape!?
sw
&LEP.?.??.. 0 100
I
I
I
200
500
400
5
TEMPERATURE. C 100
co 50
-
8
BEFORE SO2
R
UPLm a 2
kmksi??.. a !
100
200
I
I
500
400
TEMPERATURE, C Figure 1 The effect of 20 ppm of SO2 as a hnction of temperature for a fresh pellered three-way cutalyst in lean exhaust. The tests with and afrer SO2 are the first tests following the addition and removal of S02. Propylene was used as the hydrocarbon .
.
599 100 -
t
HC
-6 Z
0
m
ac
g
50 50-
8 R
BEFORE SO2
............. AFTER SO2
0 100
R.. 1
200
I
300
400
TEMPERATURE, C
-
100
10
I
co z
0
m
6 > Z
50-
8 K
BEFORE SO2
;Lopemz 2
............. AFTER SO2
0 100
200
300
400
TEMPERATURE, C Figure 2 The effect of 20 ppm of SO2 as a function of temperature for a thermally aged pelleted three-way catalyst in lean exhaust. The tests with and afrer SO2 are theJirst tests following the addirion and removal of SO2. Propylene was used as the hydrocarbon .
600
100
HC
),c-.
c
N
BEFORE SO2
O f 100
200
300
400
TEMPERATURE, C 100
-4
J
co
-
t
I
Z
0
5
I> 5 Z
50-
8 K
BEFORE SO2
ape=292
............. AFTER SO2
0 100
I
I
I
200
300
400
5
TEMPERATURE, C Fiaure 3 The efsect of 20ppm of SO2 as afunction of temperature for a fresh pelleted three-way catalyst in lean exhaust. The tests with and after SO2 are thejirst tests following the addition and removal of S 0 2 . Propane was used as the hydrocarbon .
60 1 100 RE - S S
HC
z
Z p L m 532
............ A T R 502
0
5
er
W
> z
0
0 €R
#
50i I
4
,/ 0 200
300
d
*.-
..-.....
*.,'
I**....
400
so0
, I
P-
600
TEMPERATURE, C I
BEFORE SO2
Z P e F a22 AFTER SO2
............. 300
400
so0
6 0
TEMPERATURE, C
Figure 4 The effect of 20 ppm of SO2 as a function of temperature for a thermally aged pelleted three-way catalyst in lean exhaust. The tests with and aBer SO2 are the first tests following the addition and removal of S 0 2 . Propane was used as the hydrocarbon .
602
constituent. The same procedure was utilized as mentioned in the case of propylene, namely, the sample was first tested without S02, then with SO2 and again without SO2 in the feed stream. Listed in Table 3, is a summary of the CO and propane light-off temperatures resulting from this test.
Table 3 : Effect of SO2 on Fresh and Aged Lean Light-Off of Pelleted Pt/Pd/Rh catalysts, "C (CO & Propane)
As can be seen from Table 3, CO light-off temperature for both the fresh and aged samples has significantly increased with the addition of 20 ppm SO2 into the feed stream. This increase is even greater than in the case where propylene was used as the hydrocarbon constituent. Specifically, for the fresh sample, CO light-off temperature increased 95°C with the addition of 20 ppm SO2 into the feed stream. Likewise, CO light-off temperature increased 59°C in the case of the thermally aged sample. When the SO2 is removed from the feed stream, CO light-off temperature decreases for both the fresh and aged samples. The results are consistent with the two separate effects proposed for sulfur poisoning in that CO light-off performance is only partially recovered with the removal of SO2 from the feed. For example, the CO light-off temperature for the fresh and aged catalysts, after being exposed to 20 pprn S02, are 28 and 18°C higher, respectively, than before being exposed to the SO2. Thus, there appears to be a residual poisoning effect on the catalyst caused by exposure to SO2 even when the SO2 is subsequently removed from the feed stream. A curious result is the improvement of propane light-off performance with the addition of SO2 into the feed stream. Moreover, the lower propane light-off temperature in the presence of 20 ppm SO2 is substantial. For example, the fresh catalyst reached 50% propane conversion at 405°C before being exposed to SO2 while reaching the same point at 326°C with 20 pprn SO2 in the feed. This phenomenon has been reported previously (1 1).
Cycled Stoichiometric Light-off Results The conversions of the aged pelleted catalyst were also determined as a function of the temperature in a cycled, time-averaged stoichiometric feed as the feed cooled. This was done without S02, with SO2 and again without S02.
603
The results are summarized in Table 4 and the propane, CO, and NO, conversions (with the propane containing feed) are shown in Figure 5. The most was that the :?EYeri!?ss
prop!??
fell when bh& 202 was
added, then rose when the SO2 was removed. It did not rise to the level seen prior to the SO2 exposure. This behavior is similar to the propylene and CO conversions, but markedly different from the propane conversions in the lean feed where the conversion rose in the presence of S02. Comparing the 50% HC and CO conversion temperatures in the stoichiometric feed with those for the lean feed (Tables 2 and 3) we see that in all cases the temperatures were lower in the stoichiometric feed than in the lean feed. This was surprising in that oxidation reactions are generally enhanced b y higher oxygen concentrations. The inhibition effect, present in the oxidizing feed, was not investigated, though one explanation is that the reaction is actually inhibited by NO. The NO conversion is much lower in the oxidizing feed than in the stoichiometric feed, hence the NO concentration over the catalyst is higher in the oxidizing feed as the catalyst cools toward the point where the reactions are quenched. Nitrogen oxide is known to inhibit CO oxidation (12).
Table 4 : Effect of SO2 on Aged Light-Off of Pelleted Pt/Pd/Rh Catalysts, "C in Cycled Stoichiometric Feed Before SO2 With 20ppm SO2 After SO2 HC CO NO, HC CO NO, Hc CO NO, Propylene 236 220 214 300 282 268 261 244 222 Propane 1310 206 198 1368 279 268 1338 240 217
Hc
I
I
Warmed-Up Pelformance Figures 6 and 7 show the effect which varying levels of SO2 have on the dynamic performance of the fresh and aged catalyst, respectively. In these experiments, the catalyst temperature was maintained at 500°C while the feed stream was cycled at 0.5 Hz between the reducing and oxidizing mixtures. For the first 60 minutes no SO2 was present in the feed stream; then, SO2 was added (in the amount of 0, 5, 20, or 90 ppm) to the feed stream for the next 180 minutes; finally, the SO2 was removed from the feed stream while the experiment continued an additional 120 minutes. During the course of the test, HC (as propane), CO and NO, conversions were monitored as a function of time. The fresh catalyst (Figure 6) exhibited slight degradation of CO and propane oxidation performance during the blank experiment (i.e., 0 ppm S 0 2 ) with CO conversion dropping from 98% to 97% and propane conversion dropping from 92% to 91%. However, reduction of NO, during the blank experiment dropped significantly, falling from 93% conversion to
604
Z
9 v,
I5
I 8 be 100
200
300
400
500
TEMPERATURE, C Z
P v,
8
s
8 N
100
200
300
400
500
TEMPERATURE, C
100
200
300
400
500
TEMPERATURE, C
Figure 5 :The effect of 20 ppm SO2 as a function of temperature for a thermally aged pelleted three-way catalyst in cycled stoichiometric exhaust. The tests with and afer SO2 are the first tests following the addition and removal of SO2 . Propane was used as the hydrocarbon.
605
86%. The cause of this reduction in NO, was not rigorously investigated; however, the inlet conditions were checked at the end of the experiment. Both NOx and propane concentrations were low at the end of test, with NO, concentration down 9% to 455 ppm and propane concentration down 14% to 258 ppm. Presumably, the lost activity is a result of catalyst deactivation as one would expect conversions of NOx and propane to rise with falling inlet concentrations. Continuing with the fresh catalyst (Figure 6),propane oxidation is poisoned similarly when exposed to 5 or 20 ppm SO2. The poisoning effect on propane oxidation is much more dramatic when exposed to 90 ppm S02. Likewise, CO oxidation degradation is similar when exposed to 5 or 20 ppm SO2 and more profound when exposed to 90 ppm SO2. However, in the case of NOx reduction, 20 and 90 ppm SO2 levels mimic each other while the 5 pprn SO2 level is much less severe in deactivating the catalyst. Hence, there is a general trend to more severe deactivation with the increasing SO2 concentration. In most cases it also appeared that the magnitude of the SO2 impact was increasing even after three hours of exposure . The HC, CO, and NO, conversions all rose when the SO2 was removed from the feed, with the recovery of the HC and NO,, much more apparent than that of CO. In all cases the conversions two hours after the SO2 was shut off were below the pre-SO2 conversion, indicating that in this time scale the SO2 poisoning was not being fully reversed. The aged catalyst (Figure 7) exhibited similar results to the fresh in the blank experiment. During the course of the experiment with 0 ppm SO2 in the feedstream, CO conversion dropped from 96% to 93%, propane dropped from 78% to 71%, and NOx dropped from 77% to 67% conversion. Again, the lost activity was not rigorously investigated. For the aged sample, propane conversion was increasingly more poisoned with the addition of 5 ppm, 20 ppm, and 90 ppm of SO2 into the feed stream. In the case of CO oxidation and NO, reduction, the three sulfur levels deactivated the catalyst similarly. In these off/on/off sulfur poisoning experiments, the conversions two hours following the removal of the SO2 from the feed stream were below the conversions prior to exposure to S02. Hence, under the conditions investigated in these experiments, the sulfur poisoning was not fully reversible. Both the stoichiometric light-off results and the warmed-up cycled stoichiometric results showed the same general trend: Conversion activity falls on the introduction of SO2 and partially, but not fully, recovers when the SO2 is removed. The one situation where no recovery of activity was seen when the sulfur was taken from the feed was the CO conversion on the aged catalyst. It should be noted that there was very little difference in the CO conversion whether 5 or 90 ppm of SO2 was present, indicating that the
m
0
m
100
100
z 0
100
a h ¶
80-
s', -n
%@$RW4b -4
.#..a
8
*'"& -Una.=: .Sc'-m-
L c
K
co
HC 0
1
.
.
60
1
NOx 40
"
I
"
Fipres 6 & 7 :The efSect of SO2 on the conversions of a peileted three-way catalyst in cycied stoichiometric exhaust at 500 "C. SO2 was added to the feed after 60 minutes and remained in the feed for 180 minutes. Propane was used as the hydrocarbon . ) ; 5 ppm SO2 (- - - ) ; 20 pprn SO2 (- - - ) ; 90 ppm SO2 (. .. . ) OppmS02 (
-
.
607
activity poisoned by the SO2 (perhaps the water-gas shift reaction) can be fully deactivated by small amounts of sulfur. The slight fall in CO conversion seen when the SO2 was shut off can be attributed to the rise in the propane conversion that accompanied this shut off. Carbon monoxide must compete with the propane for available oxygen in this cycled stoichiometric feed. Reversal of Sulfur Poisoning The lack of full recovery from the sulfur poisoning during stoichiometric cycling at 500°C raises the question as to whether full recovery can be achieved under other conditions. Identifying the conditions under which complete recovery is obtained will provide insight into the effect of some of the diverse conditions to which a catalyst is exposed when used in automotive exhaust. The effect of temperature on the reversibility of the sulfur poisoning of the pelleted catalysts in a stoichiometric mixture was investigated. The catalysts were operated with a sulfur-free feed for one hour, then with 90 ppm of SO2 for 2 hours, and then no SO2 for 3 hours. These experiments were performed at 500, 600, and 700°C. Since the conversion efficiencies are affected by the operating temperature, direct comparisons of the conversions at different temperatures are not meaningful. Therefore, each catalyst was tested for its conversion efficiencies at 500°C. The results of these experiments are shown in Figure 8. Increasing the temperature for these treatments resulted in no improvement in the HC and NOx conversions at 500°C. The CO conversions of the catalysts operated at 600 and 700°C were slightly higher than the 500°C catalyst, but it is not clear that this is above the experimental error, and all conversions were well below the conversion efficiency before the catalyst was exposed to sulfur. Since increasing the operating temperature in the stoichiometric feed had so little effect in reversing the sulfur poisoning of the catalyst, attention was shifted to the impact of the exhaust stoichiometry. Four sets of catalysts were each aged in a feed containing 90 ppm of SO2 for 1 hour at 500"C, then exposed to feeds that varied in their relative richness. One catalyst was treated in a stoichiometric feed at 500"C, one in a rich feed at 5OO0C, one in a rich feed at 700"C, and one in a 5% H2/N2 feed at 700°C. The stoichiometric feed was produced by cycling between the rich and lean feeds every second. The rich feed was produced by operating on the rich side of the cycle for 2 s and the lean side for 1 s. The HC, CO, and NOx conversion efficiencies at 500°C in a stoichiometric feed of this catalyst before exposure to sulfur, following exposure to sulfur, and following exposure to each one of the four above treatments is shown in Figure 9. Consistent with the previous results, the HC, CO, and NO, conversions all fell on exposure to sulfur and the HC and NO, conversions rose when the sulfur was removed from the stoichiometric feed,
z
e 6 >
'0°1
v)
z 0 0 0
ao
I K 0
Figure 8 :Conversion eficiencies of a thermally aged catalyst at various temperatures : before exposure to S 0 2 ; after 1H with 90 ppm S 0 2 ; 1H following removal of SO2 tested at temperature; 1H following removal of S 0 2 , then tested at 500 "C 100
B ,\
Z
0
0 V V
I
N 0
ill B
C
D
I
I
I: C
-
F
L
I
Figure 9 :Reversal of surfur poisoning by exposure to various operating conditions. All conversions were measured at 500 "Cin a cycled stoichiometric feed. A: Before SO2 ; B : 90 ppm SO2 ; C: B + 0 ppm, 500°C, Stoichiometric; D : B + 0 ppm,500°C, Rich; E : B + 0 ppm, 700"C, Rich ; F : B + 0 ppm, 70O0C,H2/N2
1
F
609
but not to their previous levels. The HC conversion rose further when the catalyst was treated in the rich sulfur-free feed, but again there was no improvement in the CO or NO, conversions relative to the stoichiometric sulfur-free feed. The HC, CO, and NO, conversion efficiencies were all higher following the treatment in the rich exhaust at 700°C than following the treatment in the rich exhaust at 500°C. Finally, the HC and CO conversions rose to their pre-SO2 levels and the NO, conversion was close to its pre-SO2 level following the treatment of the catalyst in 5% H2/N2 at 700°C. Hence, the full recovery of activity is obtained under conditions which create an H2S odor - high temperature highly reducing operation.
Non-Production Catalysts All of the results presented to this point indicate that sulfur poisoning of pelleted catalysts can only be fully reversed by high temperature reducing treatments. This is contrary to the results of Summers and Baron (1) who found sulfur poisoning to be rapidly reversed upon the removal of the sulfur from the feed. A set of catalysts were prepared in order to investigate this inconsistency. These catalyst compositions were as follows: 0.1 wt% Pt, 0.1 wt% Pd, 0.01 wt% Rh, 0.1 wt% Pt + 0.01 wt% Rh, and a 0.1 wt% Pt + 0.01 wt% Rh + 5 wt% Ce. Their HC, CO, and NO, conversions were determined before exposure to sulfur, with 20 ppm of SO2 in the feed, and following the removal of the SO2 from the feed. The HC (propane), CO, and NO, conversions of the single component catalysts are shown in Figure 10. For the Pt/Al2O3 catalyst, the HC, COYand NOx conversions all fell upon the addition of the S02, then rapidly returned to their initial levels when the sulfur was removed from the feed. For the Rh catalyst, however, only the HC conversion fell markedly upon the introduction of the sulfur, with the NO, conversion falling slightly, and the CO conversion actually rising slightly. The increase in the CO conversion is probably a result of more oxygen being present due to the lower HC conversion. The oxygen concentration leaving the catalyst rose with the addition of the the sulfur, then fell when it was removed. All of the conversions of this catalyst rapidly returned to their initial levels when the SO2 was removed from the feed. The catalyst which contained Pd alone also showed a modest increase in the CO conversion on the introduction of S02, but had lower HC and NO, conversions when sulfur was present. The rate of loss of HC and NOx conversion efficiency was slower than for the Pt and Rh catalysts and was similar to the commercial pelleted catalyst (which also contained Pd). The HC and NO, conversions did not return to their pre-SO2 level when the SO2 was removed from the feed, a result which was also similar to the commercial pelleted catalyst.
100
80
1
I
I
60
0
180
360
m 0 c
l0O"1I
0
co
180
I
20
360
1 0
,
,
NOX,
1
,
180
360
TIME (MINUTES) TIME (MINUTES) TIME (MINUTES) Fiaure 10 :The effect of SO2 concentration on the conversions of fresh Pt, Pd and Rh catalysts in cycled stoichiometric exhaust at 500 "C. SO2 was added to the feed afer 60 minutes and remained in the feed for I80 minutes. Propane was used as the hydrocarbon. 0.1% Pt/A1203 ( - - - ) ; 0.1%Pd/A1203 ( - - - ) ; 0.01% Rh/A12O3 ( .) 100 1 o o y 1 100 7 f
pi p .. -
I
if6>5n !-=k7:-l 90
80
Z
80
Z
0 0 6\0
co
HC 80
0
180
O't-4
NOx
60 360
60 0
TIME (MINUTES)
180
TIME (MINUTES)
360
0
180
3 $0
TIME (MINUTES)
Figure I I :The effect of 20 ppm SO2 on the conversions of thermally aged PtlRh and PtlRhlCe catalysts in cycled stoichiometric exhaust at 500 "C.SO2 was added to the feed after 60 minutes and remained in the feed for I80 minutes. Propylene was used as the hydrocarbon. 0.1%Pt / 0.01% Rh ( ) ; 0.1%Pt / 0.01%Rh / 5% Ce ( - - - ) ~
61 1
The response of the Pt-Rh and the Pt-RWCe catalysts to the addition and removal of sulfur is shown in Figure 11. The Pt-Rh/A1203 catalyst showed a modest loss of HC, CO, and NO, conversion activity from the 20 ppm of SO2 and recovered virtually all of its activity once the SO2 was removed from the feed. The catalyst which contained Ce in addition to the Pt and Rh showed higher initial CO and NO, activity than did the equivalent non-Ce catalyst, but also showed higher sensitivity to the sulfur. Additionally, the Ce-containing catalyst did not recover all of the lost activity once the sulfur was removed from the feed. The differences between the CO and NOx conversions of the two catalysts were minimized upon the exposure to sulfur. The catalyst which contained Ce also demonstrated a more gradual loss of conversion activity than the catalyst without Ce and was similar to the production pelleted catalyst.
Monolithic Catalysts All of the work which has been presented so far has used a pelleted substrate. Monolithic catalysts differ from pelleted catalysts in several respects making it difficult to extrapolate the results from the pellets to monoliths. The first of these differences is that pelleted catalysts are composed entirely of a high surface area alumina (- 100 m2/g), while on the monolithic catalysts the high surface area washcoat used to support the noble metals is only 20-40% of the mass of the monolith. Secondly, the pellets are about 1600 pm in radius, while the washcoat on the monoliths is 10-100 p m thick, hence these two systems have very different dynamics for the storage and release of sulfur. The experiments which measured the catalyst conversion activity as a function of temperature and the experiments determining the impact of the sulfur concentration on the warmed-up activity of the catalyst were repeated using a commercial monolithic catalyst rather than the pelleted catalyst. All of the following data were obtained with thermally aged samples of the commercial monolith.
Stoichiometric Light-off of Monolithic Catalysts The HC, CO, and NO, conversions of a monolithic catalyst were determined as a function of temperature as the catalyst was allowed to cool from 500 to 100°C. This experiment was first performed in a feed that did not contain any S02, then with 5ppm S02, with 20 ppm S02, and finally again with no SO2 in the feed. The HC, CO, and NOx conversions are shown as a function of the inlet temperature in Figure 12. The conversion efficiencies of all three species fell as the sulfur concentration rose. However, contrary to the pelleted catalysts, there was essentially a full recovery of activity once the sulfur was removed from the feed.
612
Z
El v,
i> .3 z 0
50
0
K
.
.- , :
0 100
I
I
200
300
I
400
500
TEMPERATURE, C
100
,
1
R 0 1 100
I
I
I
200
300
400
500
TEMPERATURE, C Fiaure 12 :The effect of 5 and 20 ppm of SO2 as a function of temperature for a thermally aged monolithic catalyst in cycled stoichiometric exhaust. The test after SO2 is the first test following the removal of SO2 . Propane was used as the hydrocarbon.
613
Warmed-Up Cycled Pelformance The impact of sulfur on the warmed-up cycled performance of a monolithic catalyst was investigated by operating a sample for 1 h in a sulfur free feed, then for 3 h with either 0, 5, 20 or 90 pprn of SO2 in the feed, then for two hours with no sulfur in the feed. This is the same schedule as was used with the pelleted catalyst. These results are shown in Figure 13. The HC, CO, and NO, conversions all fell rapidly when the sulfur was added to the feed, then rose rapidly once the sulfur was removed from the feed. The conversions from the catalysts which had been exposed to 5 and 20 ppm of SO2 rose to their pre-SO2 level, while the conversions from the catalyst which had been exposed to 90 ppm of SO2 rose to close to its pre-SO2 level. The rate of loss of activity, the rate of recovery of activity and the the extent of the recovery of activity were all greater on the monolithic catalyst than on the pelleted catalyst. The slow response of the pelleted catalysts made it difficult to determine whether the steady state conversions were a function SO2 concentration or whether it only affected the rate at which they approached the steady state conversion. With the monolithic catalysts it is very clear that the stabilized conversions of HC, CO, and NO, all decrease as the SO2 concentration increases. All of the cycled results presented to this point were performed using a cycling frequency of 0.5 Hz. This is the fastest cycling rate at which the reactor could operate without attenuating the cycle amplitude because of mixing before the catalyst bed, It should be noted that cars operate over a range of cycling frequencies, with the cycle frequency frequently greater than 1 Hz. With this in mind, the monolithic catalyst was operated at both 0.5 and 1 Hz with 0, 5, and 20 ppm of SO2 in the feed to determine if there was a difference in the impact of the SO2 as the cycling frequency was increased. (It should be remembered that as the cycling frequency was increased in this reactor it was accompanied by a reduced cycling amplitude.) The results are shown in Figure 1 4 . All of the conversions were similar at the two frequencies in the absence of S02. The loss of conversion efficiency was much greater, though, as the sulfur was increased at the 0.5 Hz cycling than at the 1.0 Hz cycling, indicating that cars with poor A/F control would be more susceptible to sulfur poisoning. Impact of Pd A sample of the Pt/Rh monolith was impregnated with 0.4 mg Pd/cm3 (0.012 ozt/l). This sample was then thermally aged and tested using the 1 h of no sulfur, 3 h of 20 ppm S02, and 2 h of no SO2 schedule. The HC, CO, and NOx conversions after 1 h were 82.3, 76.9, and 77.1%, respectively, fell to
z
:'"
G
K2 w > 00
50-
Z
P
-
z G
tz;"'i**..i
5
Z
i
D
0
-
%
70->
CY
*
b'%::-.>/,.Z.
&?
HC I
v, Id
0
6\"
U
2
! ?--=
L
z
100
100-
100
80
"
0 V
.!
-<
L
b2: ---..*
-
40
1
f
NOx
R
co 1
I3
I
"
Figure 13 :The effect of SO2 concentration on the conversions of a thermally aged monolithic three-way catalyst in cycled stoichiometric exhaust at SO0 "C.SO2 was added to the feed after 60 minutes and remained in the feed for 180 minutes. Propane was used as the hydrocarbon . ) ; 5 ppm SO2 (- - - ) ; 20 ppm SO2 () ; 90 ppm SO2 (. .. ) 0 ppm so2 (
..
---
100
z P v)
z IA 0
E
t0
K
a
Y
w
z
50
z
5
a0
V 0
W
2 g a. K
100
100
0 VI
x 0
0 u
z
n
O
K 0
S U L W I CONCCNlRA11ON. ??Y
50
V
0 S U V U R CONCCNTRATION. W Y
FigureI 4I4:: Conversioneficiencies eflcienciesatat500 500"C "Cfor fordifferent differentLyycling cyclingfrequencies. frequencies. Figure Conversion Cyclingamplitude amplitudefell fellasascycling cyclingfrequency frequencyincreased. increased. Cycling
615
81.0, 63.1, and 56.0 %, respectively, following the three hour exposure to S 0 2 . These conversions rose to 83.2, 74.4, and 68.6%, respectively at the end of the test. The lack of full recovery of the CO and NO, conversions, as well as the slow rate of recovery of the CO conversion efficiency following the removal of the SO2 was similar to what was observed for the Pd containing pelleted catalysts. Performance in Engine Exhaust
The HC, CO, and NO, conversion efficiencies of a car-aged (80 000 km) 1.8 L catalytic converter were determined in the exhaust from a 3.8 L engine while the engine was operating with high (0.1 wt %) and with low (0.001 wt %) sulfur fuel. The low sulfur fuel was indolene clear gasoline, while the high sulfur fuel was indolene clear fuel doped with thiophene. The engine was operated under computer control so as to induce A/F oscillations o f f 0.5 A/F at 0.5,l.O and 2.0 Hz. These conversions are listed in Table 5. There was a clear and consistent reduction in the conversion efficiencies of this converter when the sulfur concentration in the feed was increased.
Table 5 : The Impact of Sulfur on the Cycled (k 0.5 A/F) Conversions in Engine Exhaust of a Vehicle Aged (80.000 km) Monolithic Converter
I Temp."C
Wt % S
HC
0.5 Hz CO NO
HC
1.0 Hz CO NO
HC
2.0 Hz CO NO
400 400
0.001 0.10
85.05 48.86 68.84 90.76 75.39 75.49 91.45 80.15 78.20 75.86 49.46 67.84 89.68 78.13 77.55 85.35 73.10 77.90
450 450
0.001 0.10
90.50 68.06 68.83 93.80 90.32 77.35 96.29 96.84 90.51 79.58 52.92 64.87 90.71 81.04 73.83 93.13 93.08 86.68
500 500
0.001 0.10
92.68 81.24 69.18 95.92 94.39 80.63 97.76 97.24 93.72 85.03 58.12 61.10 92.10 85.82 72.99 96.16 96.88 89.52
CONCLUSIONS Sulfur has been shown to inhibit the light-off and reduce the warmedup cycled performance of automotive catalysts. The impact of the sulfur was rapidly recovered for monolithic catalysts that did not contain Pd when the sulfur was removed from the feed, but did not fully recover for pelleted catalysts containing Pd. The difference between the two systems is thought to be due both to the difference in the support and to the presence of Pd.
616
References 1.
2.
3.
4.
9.
10.
11.
12.
Summers, I. C. and Barron, K., "The Effects of Sq;! on the Performance of Noble Metal Catalysts in Automobile Exhaust", J. Catal. 57, 380, 1979. Joy, G. J., Lester, G. R., and Molinaro, F. S., "The Influence of Sulfur Species on the Laboratory Performance of Automotive Three Component Control catalysts", S.A.E. Paper # 790943, 1979. Furey, R. L. and Monroe, D. R., "Fuel Sulfur Effects on the Performance of Automotive Three-Way Catalysts During Vehicle Emissions Tests", S.A.E. Paper # 811228, 1981. Williamson, W. B., Gandhi, H. S., Heyde, M. E., and Zawacki, G. A., "Deactivation of Three-Way Catalysts by Fuel ContaminantsLead, Phosphorus and Sulfur", S.A.E. Paper # 790942, 1979. Schlatter, J. C. and Mitchell, P. J., "Three-Way Catalyst Response to Transients", Ind. Eng. Chem. Prod. Res. Develop. 19, 288, 1980. Diwell, A. F., Hallett, C., and Taylor, J. R., "The Impact of Sulfur Storage on Emissions from Three-Way Catalysts", S.A.E. Paper # 872163, 1987. Henke, M. G., White, J. J., and Denison, G. W., "Sulfur Storage and Release from Automotive catalysts", S.A.E. Paper # 872134, 1987. Lox, E. S., Engler, B. H., and Koberstein, E., "Developmentof Scavenger Free Three-Way Automotive Emission Control Catalysts with Reduced Hydrogen Sulfide Formation", S.A.E. Paper # 890795, 1989. Petrow, R. S., Quinlan, G. T., and Trues, T. J., "Vehicle and Engine Dynamometer Studies of H2S Emissions Using a Semi-continuous Analytical Method", S.A.E. Paper # 890797, 1989. von Carlowitg F. J., Henke, M. G., and Gagneret, P. H., "Use of Mass Spectrometer to Continuously Monitor H2S and SO2 in Automotive Exhaust", S.A.E. Paper # 900272, 1990. Yao, H. C., Stepien, H. K., and Gandhi, H. S., "The Effects of SOZ. on the Oxidation of Hydrocarbons and CO over Pt/Al2O3 catalysts", J. Catal. 67, 231, 1981. Oh, S. H., &d Carpenter, J. E., "Role of NO in Inhibiting CO Oxidation over Alumina Supported Rh", J. Catal. 101, 114, 1986.
A. Crucq (Editor), Catalysis and Automotive Pollution Control IZ 0 199 1 Elsevier Science Publishers B.V.. Amsterdam
617
EFFECT OF SEVERE THERMAL AGING ON NOBLE METAL CATALYSTS
H. Shinjoh, H. Muraki, and Y. Fujitani Toyota Central R&D Labs. Inc. Nagakute-cho, Aichi-gun, Aichi-ken, 480-11,Japan ABSTRACT Both sintering and activity behaviours over noble metal catalysts aged in oxidative atmospheres with various 0 2 contents at 1100 O C (or 1000 "C) for 5 h were systematically characterized. With increasing 0 2 contents, catalytic activities over aged Pt, Rh, and Pt/Rh catalysts decreased, and, in contrast, those over aged Pd and Pd/Rh catalysts increased. While, an order of sintering for the noble metal particles on aged catalysts agreed closely to that of the each percentage conversions as well as to that of vapour pressures of respective catalyst species, such as noble metals or their oxides. Selectivity profiles of the aged catalysts for converting NO and 0 2 in a simulated exhaust gas were characteristic ones. It is confumed through the above results that the performance of aged catalysts are tightly governed by the properties of noble metals and the selectivity data are also important for exploring the catalytic activities, in particular, over multi-functional catalysts such as automotive exhaust ones.
INTRODUCTION
Characterizations of deactivation phenomena in thermally aged automotive exhaust catalysts are extremely important in terms, not only of improvement in the life of the catalysts, but also of suitable use of noble metal resources. For example, European driving conditions require catalysts to be more thermally stable than the practical catalysts used in the US and Japan. Generally, severe thermal experiences of catalysts lead to the sintering of noble metals and even the deterioration of support materials. Thus, in the first place, thermal stabilization of alumina have been conducted. Now, additions of alkaline-earth metals or rare-earth metals to alumina in a transition state have prevailed as one of the most useful techniques. Among such additives, lanthanum oxide is well known as the best [l-41. Secondly, for designing thermally resistant catalysts, the relationships between the sintering behaviour and the catalytic activity over aged catalysts must be clarified as a function of aging conditions. Sintering of noble metals, especially Pt, has been widely investigated [ 5 ] . The sintering rate of noble metals is largely dependent on the experimental conditions, e.g., substrate morphology, purities of catalyst species and support materials, composition of aging atmospheres,
618
temperature, and time. With respect to these issues, we know by experience that automotive noble metal catalysts can be deactivated more easily under oxidative atmospheres than under stoichiometric or reductive ones. However, the details of the sintering mechanism of automotive exhaust catalysts have not been clarified yet. Also, data concerning the catalytic activity during thermal aging are scattered. An attempt of this report is to systematically investigate the deactivation processes of noble metals on thermally stable La-doped alumina supports. In particular, we have concentrated on clarifying the effect of the 0 2 content in the aging gaseous atmosphere on the catalytic activity of aged catalysts. In order to do this, the 0 2 content was varied from 0 to 20 vol%, and these gases were flowed on the catalysts at 1100 "C (or 1000 "C) for 5 h. For aged catalysts, 1)percentage conversions for CO,propylene, and NO, 2) selectivities between NO and 0 2 , and 3) growth of noble metals particles were measured These results successfully explained the degradation of noble metal catalysts when considering the vapour pressure data of noble metals and their oxides. This work consists of two parts: the first deals with the simple catalysts, Pt, Pd, and Rh, and the second with the bimetallic catalysts, Pt/Rh and Pd/Rh.
EXPERIMENTAL Catalvst Spherical pellets of theta-alumina containing 1 mol% La (BET surface area: about 100 m2/g, apparent density: about 0.4 g/ml, diam.: 2-4 mm) were used as thermally stable support. These were prepared by impregnating gamma-alumina sphere with La(N03)3 aqueous solutions and then calcining at 1OOO"Cfor 20 h. The sample catalysts, such as Pt, Pd, Rh, Pt/Rh, and Pd/Rh were prepared by impregnation of the La-doped alumina with aqueous solutions of noble metal nitrates. The impregnated pellets were dried over night at llO"C, then calcined in flowing H2 at 450°C for 3 h. All loading amounts, as noble metal(s), were 0.5 g/l (about 0.14 wt%). In bimetallic catalysts, Pt or Pd was supported on the Rh catalyst and the loading ratio of both Pt/Rh and Pd/Rh catalysts was 9:l. The dispersion of noble metal on these catalysts was determined by CO up-take, which was standardized by the Committee of the Catalysis Society of Japan 161. Their values of Pt, Pd, Rh, Pt/Rh, and Pd/Rh catalysts were 0.21, 0.63, 0.17, 0.32, 0.38, respectively. Aging Catalysts were exposed to aging atmosphere gases with various oxygen contents at 1 100°C( or lOOO"C), for 5 h. The oxygen contents were 0 (N2), 3, 5 , and 2O(air) vol%, respectively. In the case of the 3- and 5 - vol% 0 2 , considering the effects of exhaust gas constituents, the respective 0 2 content
619
was added to a synthetic gas corresponding to the stoichiometric engine exhaust gas composition ( 0.23 vol% H2, 0.70 vol% CO, 0.16 vol% C as C3H6, 3 vol% H20, 10 vol% C02, and the balance N2). Aging was performed under gas flow conditions Catalytic activity measurement The laboratory reactor system used was described elsewhere [7]. The percentage conversions of CO, HC(propylene), and NO in the stoichiometric (synthetic) gas were measured under light-off tests in the range 100 to 500°C using a heating rate of 12"C/min. The space velocity was kept at 30,000 h-1. The exhaust gas, simulating stoichiometric conditions, was composed of 0.23 vol% H2, 0.70 vol% CO, 0.16 vol% C as C3H6, 0.12 vol% NO, 0.65 vol% 02, 3 vol% H20, 10 vol% C02, and the balance N2. The selectivity data were also obtained in the same manner. XRD measurement In order to measure the particle sizes of noble metals on aged catalysts, La-doped alumina was dissolved by a mixture of concentrated HF and H2S04. The residues collected on a filter were used for XRD sample. Particle size was determined based on the (1 11) diffraction peak of the noble metal crystallites using Scherer's equation. Lattice parameters were determined using the internal standard method. For Pt-Rh alloys, the lattice parameter can be easily related to the alloy composition [8]. Also, XRD data were partly used for identifying the chemical states of catalyst species after aging. RESULTS AND DISCUSSION
Simple noble metal catalysts The conversion profiles of NO, CO, HC over simple Pt, Pd, and Rh catalysts aged in 5 vol% 0 2 synthetic exhaust gas at 1100°C for 5 h are shown in Figure I , 2, and 3 , respectively. In addition, light lines in these Figures are conversion profiles over fresh catalysts. Clearly, the catalytic activities of aged catalysts are degraded more or less by aging, especially, the degradation of the Pt catalysts is significant (Figure I ) . Figure 2 shows that NO conversion profiles over Pd catalyst has singularly a maximum peak below 300°C. This has been explained by a contribution of the NO-H2 reaction [9]. The NO conversion profiles over the catalysts aged in gases with various 0 2 contents are shown in Figure 4 , 5 , and 6 . The order of catalytic activities over these catalysts is the following: Pt : N2 > 5 vol% 0 2 (lOOO°C) > 3 vol% 0 2 > 5 vol% 0 2 > Air Pd : Air > 5 vol% 0 2 > 3 vol% 0 2 > 5 vol% 0 2 (1000°C) > N2 Rh : N2 > 3 vol% 0 2 > 5 vol% 0 2 (lOOO°C)> 5 vol% 0 2 > Air
620
100
I-
Pt cat.
--
80
c
60
H
/-
’
.A
rn
$
40
I/
C
i
1,NO
U
20 n ”
100
200
400
300
500
600
(‘c)
Temperature
100
--
80
c
60
H
0 .d
rn
L QI
>
40
C
I
0
U
20
n L 100
7 //NO
W
Pd cat. .
.
200
300
.
400
Temperature
(‘c)
Temperature
(‘c)
. 500
.
E 0
Figure 1-3. Light-offpegormancein a simulated exhaust gas of afresh catalyst and of an aged one in 5 vol% 02 exhaust model gas at 1100°Cfor 5h.
62 1 100
--
80
!.
60
N
Pt cat.
ln
L
z
40
0
u
4
20 n
ioo
200
300
400
Temperature
500
600
(t)
Figure 4.: NO conversion profiles on aged Pt catalysts 100
-N
80
C
2
60
ln L
z
40
0
u
4
20 n 100
200
300
400
Temperature
500
600
('c)
Figure 5.: NO conversion profiles on aged Pd catalysts
" 100
200
300
400
Temperature
500
600
(t)
Figure 6.: NO conversion profiles on aged Rh catalysts
622
For Pt catalyst, with increasing 02 content in the aging atmosphere, the catalytic activity decreases. Table 1 shows the Pt-particle sizes obtained from XRD data on the aged catalysts. The particle size remarkably increases with increasing 0 2 content. It is known that Pt supported on alumina sinters easily under oxidative conditions [lo-111. The results of Figure 4 and Table 1 demonstrate that sintering of Pt particles is accelerated by increasing 0 2 content in the aging atmosphere, and that it leads to the decrease of the catalytic activity of Pt catalysts.
TABLE 1. Pt Particle size of the aged catalysts
1100 1100
5%02* Air
780 970
The result on Rh catalysts is similar to that on Pt catalysts. Namely, with increasing 0 2 content, the Rh-particle size increases (Table 2), and consequently, the catalytic activities decreases. While, according to XRD data, only metallic Rh is found on the N2-aged catalyst, but a mixture of Rh and its oxide is found on the 3 vol% 0 2 - , 5 vol% 02-, and air-aged Rh catalysts. In the latter cases, the more the 0 2 content increases, the more the fraction of Rh oxide increases. Carol et al.[ 121 studied high temperature deactivation of 3-way catalysts and suggested that oxidation of Rh was a major degradation mechanism at high temperatures and that sintering of noble metal had a minor contribution. While, for Pd catalysts, the effect of the 02 content on the catalytic activity is the reverse as compared with both Pt and Rh catalysts. That is, with decreasing 0 2 content, Pd-particles grow and the catalytic activity decreases (Table 2).
TABLE 2. Particle size of noble metals on the aged catalysts
623
Wanke [13], Fiedrow et a1 [14] studied the changes in dispersion of alumina supported noble metal catalysts during thermal treatments (250SOO0C) in 0 2 and H2 atmospheres. The sequence of thermal stability was found to be Rh > Pt > Ir in 0 2 atmosphere, while, Ir > Rh > Pt in H2 atmosphere. And, they predicted that the relative stability of noble metals originated from the heats of oxide formation and heats of sublimation. Vapour pressures of noble metals and their oxides at 800°C are shown in Table 3 [referred from 151.
TABLE 3. Vapour pressures of noble petals and their oxides at 800°C
The orders of vapour pressure of noble metals and their oxides are as follows: Metals : Pd >> Pt > Rh Oxides : Pt > Rh >> Pd These trends agree well with the order of catalytic activity on the aged catalyst. This fact suggests that the degree of sintering which is related to catalytic activity largely depends on the vapour pressure of the given species on the catalyst. Figure 7, 8, and 9, show the selectivities of the aged catalysts for converting NO and 0 2 by partitioning of reduced gases, such as H2, CO, HC, at elevated temperatures. These profiles are quite different from one another. It is interesting that, for Pt catalyst, the selectivities in NO and 0 2 are almost the same, regardless of the degree of sintering. For Pd catalyst, the selectivity of NO is higher than that of 0 2 in the low conversion regions. However, an opposite result is observed in the high conversion regions. In the low conversion regions, Pd catalyst had higher selectivity for the NO-H2 reaction than for the 02-H2 one [9]. For Rh catalyst, NO selectivity is the highest among the three noble metal catalysts, especially by less oxidative thermal treatments. This fact supports that metallic Rh is apt to convert NO rather than 0 2 .
624 100
/
-w 80
Pt cat.
N/
c
p, 60
Figure 7 :Pt catalyst
ffl L a,
2
40
0
u
:20 n 0
20
02
40
60
conversion
80
100
(%I
Figure 8 :Pd catalyst
0
20
40 60 02 c o n v e r s i o n
80
100
(%I
100
-w 80 c
.zffl
60
L
ar
40 0
u
8
20
n
Figure 9 :Rh catalyst
625
Bimetalic catalvsts The light-off performance of a fresh Pt/Rh catalyst and of a 5 vol% 0 2 aged one is shown in Figure 1 0 . The NO conversion profiles of Pt/Rh catalysts which aged in several conditions are shown in Figure 11. Catalytic activities of the aged Pt/Rh catalysts Temperature (‘c1 are dependent on the 0 2 Figure 10. content in the aging atmoLight-off performance in a simulated exhaust sphere. With increasing 0 2 gas of PtlRh catalysts (])fresh (2) aged at content, the catalytic acti1100°C in a simulated exhaust gas vities decreases. Table 4 shows the Pt-particle size and the amount of Rh diffused 100 into Pt-phase in the aged catalysts obtained by XRD 80 w data. The growth of PtC particles is clearly dependent z 60 on the 0 2 content. Their L UJ behaviour, the order of cata40 0 lytic activity and the trend of u particle growth, are the same 0 z 20 Pt-Rh as simple aged Pt and Rh catalysts. When the 0 2 0 200 300 400 500 600 content is low (for example, 100 less than 5 vol% 0 2 ), most Temperature of Rh diffuses into the PtFigure 11. phase and forms Pt-Rh NO conversion profiles on the aged PtlRh alloys. However, under aircatalysts aging condition, Pt-Rh alloy formation is not complete. It is known that Rh oxide diffuses into transition state alimina by oxidative treatment [16].Thus we examined the behaviour of low loading Rh the same Rh loading as in the present Pt/Rh catalyst, on the catalyst (0.05 La-doped alumina support.
-u)
a),
626
TABLE 4. Pt particle size and Rh amount in Pt-Rh alloy on the aged Pt/Rh catalyst Aging conditions Temp.("C) Atmosphere N2 1100 5%02* lo00 5%02* 1100 1100 Air
Pt particle Size (A) 85 276 344 496
Rh amount in Pt-Rh alloy (at%) 13 10 11 6
Figure 12 shows the NO conversion profiles on the aged low loading Rh Rh(low)cat. catalysts. The catalytic actiH2 red. I vity of N2-aged catalyst was high enough, but that of 3 I I vol% 0 2 -aged one is very I poor. It indicates that Rh on I the alumina surface is easily transported into the aluminan phase under oxidative condi100 200 300 400 500 600 tions. The dashed line in Temperature ('Cl Figure 12 is a NO converFigure 12. sion profile of the 3 vol% 0 2 NO conversion profiles on the aged low -aged catalyst after an loading Rh catalysts additional reduction by H2 at 1000°C for 3 h. The catalytic 100 activity of degraded catalyst was refreshed through this 80 H2 treatment. It suggests that H (1) Rh diffuses into the Lac 60 ." doped alumina phase and (2) 40 the diffusing Rh in the C alumina phase is revived and u 20 redispersed rather widely on Pd-Rh c a t the alumina surface. Thus, an 0 origine of noticeable degra1 200 300 400 500 E dation of the air-aged Pt/Rh Temperature "c) catalyst is not only the Figure 13. sintering of noble metals but Light-off pegormance in a simulated exhaust the loss of Rh from the gas of PdlRh catalysts (1)fresh (2) aged at alumina surface. 1IOO"C in a simulated exhaust gas 100
,
y1
/'/
627
Figure 13 and 14 show conversion profiles of the aged Pd/Rh catalysts. The catalytic behaviour of aged Pd/Rh catalyst is the same as the simple Pd one except for the N2-aged catalyst.
Pd-Rh 100
200
400
300
500
1 600
Temperature ('I3
Figure 14. NO conversion profiles on the aged PdlRh catalysts
01 0
1
I
I
I
20 40 60 80 02 c o n v e r s i o n (%)
J
100
Figure 1.5. Selectiviry of the aged PtlRh catalysts for converting NO and 0 2 in a simulated exhaust gas at elevated temperatures
Figure 15 and 16 show the selectivities of bimetallic catalysts for converting NO and 0 2 by reductive gases contained in the synthetic gas. Pt/Rh catalyst shows the same selectivity as the Rh one, while Pd/Rh catalyst is quite similar to the Pd one except for the N2-aged catalyst. Muraki et al. [7, 17-18] showed that homogenous alloys were produced in reductive atmospheres and, in contrast, surface enrichments in Rh on Pt/Rh alloy and in Pd on Pd/Rh alloy were observed under oxidative treatments. The above results on selectivity suggest that the surface of the Ptmh alloy is enriched in Rh, and that of the Pd/Rh one is enriched in Pd after the oxidative agings.
628 100
ap
-4
80 60
ffl L Q)
40 0 U
9
i
20
Ai r
0
0
40 60 02 c o n v e r s i o n
20
80
100
(%I
Figure 16. Selectiviry of the aged PdlRh catalysts for converting NO and 02 in a simulated exhaust gas at elevated temperatures
Unfortunately, the loading amounts of practical catalysts (about 2 g/l) are generally too low to precisely identify the surface states by microscopic analysis. Here, we recommend that the chemical characterization of the above mentioned surface states should become an useful technique, if the selectivity data are compiled in future. As a summary, throughout the above experimental results, it was found that the properties of noble metals largely contribute to the deactivation processes.
REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13
14 15 16 17 18
H. Schaper, and L. L. Van Reijen, in "Proceedings 5th International Roundtable Conference on Sintering, Portoroz, Yogoslavia, 7-10 September (1981), pp. 173-176, Elsevier, Amsterdam,(l982) H. Schaper, E.B.M. Doesburg and L.L. Van Reijen Appl. Catal., 7,211, (1983) H. Schaper, E.B.M. Doesburg, P.H.M. De Korte and L.L. Van Reijen, Solid State Ionics 16, 261, (1985). M. Bettman, R.E. Chase, K. Otto and W.H. Weber, J. Catal., 117,447, (1989). R. Burch in "Catalysis", ed. G.C. Bond and G.Webb, the Royal Society of Chemistry, London, vol7, p 175, (1985) T. Hattori, H. Matsumoto, Y. Murakami,in "Preparation of Catalyst 4", B. Delmon, P. Grange, P.A. Jacobs, G. Poncelet,(eds.), p815, Elsevier, Amsterdam,(l987) H. Muraki, H. Sobukawa, M. Kimura, and A. Isogai, SAE 900610, (1990). A S . Darling, Platinum Metals Rev. 5,58,(1961). H. Muraki, H. Shinjoh, H. Sobukawa, K. Yokota and Y. Fujitani, Ind. Eng. Chem. Prod. Res. Dev., vo1.25, No.2, (1986). P.J.F. Hanis, J. Catal., 97, 527, (1986). R.M.J. Fiedorow, S.E. Wanke, J. Catal., 43, 34, (1976). L.A. Carol, N.E. Newman and G.S. Mann, SAE 892040, (1989) S.E. Wanke, in "Sintering and Heterogegeous Catalysts",G.C. Kuczynski, A.E. Miller, G.A. Sargent,(eds.), p223, Plenum, New York & London,( 1984) R.M.J. Fiedorow, B.S. Chahar, S.E. Wanke, J. Catal., 51, 193, (1978). G.V. Samsonv,(ed.), C.C.N. Turton and T.I. Turton, (transls.), in "The Oxide Handbook", 217,397, Plenum, New York, (1973). H.C. Yao, J. Catal., 50, 407, (1977). M. Chan, J. Catal., 60,356, (1979). T. Wang, J. Catal., 71, 411, (1981).
A. Crucq (Editor), Catalysis andAutornotive Pollution Control ZI 0 199 1 Elsevier Science Publishers B.V., Amsterdam
629
MORPHOLOGICAL TRANSFORMATIONS IN REDUCING AND STEAM ATMOSPHERES OF ALUMINA-SUPPORTED RHODIUM AND PLATINUM CATALYSTS D. Dupreza, F. Sadib, A. Miloudib and A. Percheron-Gueganc
(a)Laboratoire de Catalyse en Chimie Organique 40, Avenue du Recteur Pineau 86022 Poitiers Cedex France. (b)L.aboratoire de Valorisation des Coupes Pe'trolidres, USTBA Dar el Beida Algiers, Algeria. (c)Laboratoire de Chimie Me'tallurgique des Terres Rares, CNRS I , Place Aristide Briand 92190 Meudon-Bellevue, France. ABSTRACT The formation of reduced centres in the vicinity of rhodium and platinum particles was studied by Isothermal Oxidation or Temperature Programmed Oxidation by Steam (IOS/TPOS) of Oxidation by Rh/Al2O3 and Pt/Al2O3 catalysts reduced at high temperatures (500-1000°C). steam of these reduced centres yielded H2 which can be easily detected and analyzed. The behavior in IOS/TpOS of RhAl and PtAl alloys was also examined to verify the reactivity of the reduced centres and to identify the solids formed by reaction with steam. It is shown that HT reduction of Rh and Pt/A1203catalysts give rise to reduced centres located round the metal particles, which explains why the number of these reduced centres depends little on the reduction temperature. The RhAl alloy reacts readily with steam to give Rho particles and alumina. For PtA1, the transformation is not complete and the result of the reaction is the formation of the neighbouring intermetallic phase Pt5A13 with a few amount of alumina. The behavior of Rh and Pt catalysts in exhaust gas conditions is discussed in the light of these results.
INTRODUCTION Exhaust gas catalysts are repeatedly subjected to severe temperature excursions (>9OO0C) in oxidizing or reducing atmosphere, which cause profound changes in the metal surface areas, in the surface compositions (bimetallics) and in the particle morphologies of the catalysts. These redox cycles can also create structural and chemical perturbations of the support particularly in the vicinity of the metal particles. It is now well established that ceria can be reduced or oxidized in mild conditions and that the redox properties of cerium can improve the catalyst behavior in transient regime owing to the "oxygen storage capacity" of the cerium oxide. By contrast, the
630 formation of reduced centres at the alumina surface is still a matter of controversy [Ref.l-31. In 1982, we proposed a new technique for the measurement of the number of reduced centres of Ti02 in Rh, Pt and Ni/Ti02 catalysts [Ref.4]. This technique is based on the measurement of the amount of H2 produced during the isothermal oxidation or the temperature programmed oxidation by steam (IOS/rPOS) of the reduced centres of Ti02:
02++ H20 + 2 e - d 02-+ H2 where 0 2+ is an anion vacancy (positively charged with reference to the network of Ti02) and 2e- represents electrons delocalized in the conduction band of Ti02. This technique presents three main advantages : (i) facility and effectiveness : reduction of the catalyst, titration of the reduced centres and eventually dispersion measurements are carried in situ in the same apparatus, (ii) high sensitivity, close to 0.01 pmole H2 (1016 reduced centres) and (iii) selective oxidation by water of the reduced centres of the support (and not the metals) provided that the reaction be carried out in cycling conditions (pulses of H20). It was therefore decided to apply this technique to alumina supported rhodium and platinum reduced in the range of temperature 500-900°C. Moreover, rhodium-aluminium and platinum-aluminium alloys were examined in IOS/TPOS to compare, their behavior with that of the new chemical compounds which have been formed at the metal-support interface during HT treatments. EXPERIMENTAL
Rh/A1203 and Pt/Al2O3 catalysts were prepared by impregnation of Rhne-Poulenc GFS C y -alumina (Na + Fe < 200 ppm, S < 200 ppm) with aqueous solutions of rhodium chloride and chloroplatinic acid. They were dried at 120°C and calcined at 450°C. Prior to impregnation, the alumina was pretreated in H2 at 900°C for 16h. Its surface area was then 120 m2g-I). RhAl and PtAl samples were prepared by arc melting rhodium (99.9%), platinum (99.97%) and aluminum (99.999%) under an argon atmosphere on water cooled copper hearth. To improve homogeneity the samples were remelted five times. X-ray diffraction patterns were obtaines with CuK radiation. The homogeneity and stoechiometry were checked by electron microprobe analysis. The surface compositions were analysed by XPS (with Mg source). 10s-TPOS experiments (catalysts and alloys) were carried out in a chromatographic microreactor described elsewhere [Ref.4]. The apparatus comprises a six-port valve for H2 and 0 2 injections, a heated injector for H20 injections, a quartz microreactor, a 5A molecular sieve column and finally a
63 1
TC detector. In the standard procedure, the catalyst (generally 0.2 g) is reduced in a flow of H2 (20 cm3 min-1) at Tr for 1 hour, then degassed in argon (less 1 ppm impurities) for 30 minutes and cooled down to the temperature of reaction with water (Tox). Pulses of water (generally 1 p1 or 54 pmole) were vaporized in argon and injected on the catalyst sample every ten minutes. Hydrogen produced in the reaction was analyzed by catharometry after trapping of the non reacted water (5A sieve column). A similar procedure was used for the alloy samples which were, however, directly heated at Tox in argon. Dispersion measurements (hydrogen chimisorption HC, oxygen and hydrogen titrations OT and HT) before and after 10s-TPOS were carried out in the same apparatus. RESULTS
Characterization of the starting materials. The characteristics of the catalysts are given in Table 1 which shows that the metal are well-dispersed in the fresh catalysts. The stoichiometries HC : OT : HT are close to 1 : 2 : 4 for rhodium in accordance with the values generally obtained with this metal [5,6]. For platinum, the stoichiometries (1.2 : 1.5 : 3) show an excess of hydrogen uptake during chemisorption. This is coherent with the previous results of Butt et al. concerning well-dispersed Pt/A1203 catalysts [7].
lLu3LEJ Characteristics of the fresh catalysts Catalysts Rh/Al2O3 pt/Al2O3
Metal Content wt % pmole/g 0.51 40.5 56.8 1.12
c1 wt 9%
0.26 0.61
I
OT I HT .mole/g 40.4 78.5 155 54.9 70.0 132
D(OT) %
.
80 83
The as-arc-cast samples were obtained single phase. All the lines of Xray diffraction pattrens were indexed in cubic structure for both compounds, in CsCl type and FeSi type respectively for RhAl and PtA1. The last compound crystallizes in the high temperature form. The values of parameters given in table 2 are in agreement with previous determinations [Ref.8-91. The results of analysis given in table 2 show a good homogeneity of compounds since the dispersion is less than 1%. X-Ray Photoelectron Spectroscopy reveals that the surface composition remains close to the bulk composition although a slight enrichment in aluminum is noted (Table 3). Binding energies observed for A1 and Rh show
632
that these elements are essentially at the zero-valent state. Nevertheless the alloy is largely covered with oxygen and carbon impurities under XPS conditions.
Compounds PtAl RhAl
Cubic Structure P213 Pm3m
Parameter 4.849 2.967
Microprobe Analysis ~~~980f.004~~1.019f.005 Rhl.01+.01A1.99+.01
TABLE 3
XPS data of the RhAl alloy (reference CIS at 285.0 eV) A1 2p Rh 3d 512
B.E.(eV) 73.5 307.5
% at
12.3 10.4
Rh 3d 312
c 1s
312.1 285.0 main peak 33.6 280.4 secondary peak 531.9 main peak 43.6
0 1s
527.4 secondary peak
Reaction with steam of the bare support
To verify the intrinsic reactivity of Al2O3, experiments were carried out on the bare support. Whatever the pretreatment of the support (pretreated in H2 at 900°C as in the catalysts, pretreated in air at 500°C or nonpretreated) and whatever the time and temperature of reduction, the time and temperature of degassing in argon, no H2 production was observed upon H20 injection. This proves that H2 is unable to create reduced centres at the alumina surface in the absence of noble metal. Reaction with steam of the RhIA1203 catalyst A hydrogen production was systematically observed in 10s experiments on the RhIA1203 catalyst reduced at Tr 2 500°C. The major part of this H2 was produced on the first pulse of H 2 0 (80-90% of the total amount); then the amounts of H2 decreased very rapidly with the pulse number and became negligible after the fourth or the fifth pulse. This behavior shows that reduced
633
centres are created during reduction of RWAl2O3 and that steam can titrate these centres with a correlative formation of hydrogen. However, the number of reduced centres (E 10 pmole g-1)) is one or two orders of magnitude less than the number of reduced centres created by H2 reduction in Rh/Ti02 catalysts [Ref.4]. The variation of Q* (total amount of H2 produced during 10s)with the reduction time shows that Q* varies little after 30 minutes of reduction (Table 4) : the reduced centres are created rapidly, which justifies the value of 1 hour used for the reduction time throughout this study. TABLE 4 Kinetics of formation of reduced centres on Rh/A1203 (Tr = 850"C, Tox = 500°C). Q*sat is arbitrarily taken after 16h of reduction.
Time (min.) Q*lQ*sat
2 0.21
5 0.39
10 0.75
30 0.94
60 0.96
Pulses of H2 were injected on the catalyst, just after the reaction with steam, to verify the reduction state of the metal : if Rh is oxidized by steam during IOS, it will be quantitatively reduced by H2 pulses. If steam reacts only with the reduced centres of the support a very low H2 uptake can be expected since the rate of formation of these reduced centres is not high enough for them to be formed again in significant proportion during H2 pulses. The results given in Table 5 show that the H2 uptakes are indeed very low and that the metal oxidation during 10s is insignificant. TABLE 5
H2 uptakes at 500°C after reduction and oxidation by steam of Rh/A1203.
Tr ("C) 500 705
850
Tox ("C) 500 500 550
H2 uptake (500°C) pmol H g-1 1.7 0.6 1.4
The effect of the temperature of reduction on Q* is shown in Fig.1. Although a maximum is observed for Tr = 700°C the values of Q* vary little with Tr, remaining in the range of 8-13 pmole H g-1). A similar behavior is observed when Tox is varied, while Tr is maintained constant (Fig.2) : the amounts of reduced centres titrable by steam increase slightly with Tox and then decrease for Tox > 700°C. Again, the values of Q* remain close to the
634
same mean value (10 pmol g-1)). It must be noted that the quasi-constancy of the Q* values when Tr and Tox are varied is not due to mass limitation effects : H20 conversions are in every case less than 20% and are generally close to 13% on the first pulse. The changes in the metal dispersion during the different treatments applied to the catalyst are reported in Table 6. 15
Tr : 650°C
1
°
.
5 - 1
Y
A
' .
.
Tr
Fig.1 :Amounts of H2 produced by reaction with steam at 500 "c of the RhIA1203 catalyst reduced for 1h at various temperatures (Tr)
Fig.2 :Amounts of H2 produced by reaction with steam at various temperatures (Tox).of the RhIA1203 catalyst reduced for I h at 850 "C.
Tr ("C)
Tox ("C)
HC (a)
650 750 850 900 850
500 500 500 500 900
29 23 20 12 20
I
HC (b) m o l e Wg 23 19 16 14 11
I
HC (c)
27 23 18 17 10
The dispersion decreases significantly during the reduction, particularly above 850°C. For the catalyst reduced at 900"C, the treatment in steam and the re-reduction at 500°C restore in part the chemisorptive capacity of the
635
catalyst. This behavior resembles the well-known SMSI effect already described for titania-supported catalysts [Ref.4 and 101. Nevertheless, there are significant differences between Rh/Al2O3 and m i 0 2 : the SMSI effect, probably generated by the reduced centres created in the vicinity of the metal particles (see the Discussion section) occurs at much higher temperatures on Rh/Al2O3 than on R W i 0 2 ; moreover, the effect is definitely less marked on Rh/Al2O3 than on RWi02. Reaction with steam of the PtIA1203 catalyst The same 1 0 s behavior is observed on Pt/A1203 (Fig.3 and 4) as on Rh/A1203 : a hydrogen production Q* is observed on the reduced Pt/A1203 catalyst and Q* vanes little with the temperature of reduction (a maximum is observed at 650°C). Two interesting differences can be noted however : (i) Q* is significantly higher on Pt/A1203 than on FWAl2O3 (about 20 instead of 10 pmole g-1) and (ii) Q* is more sensitive to the temperature of reaction with steam (Tox) on Pt/Al2O3 than on Rh/Al2O3.
TOX : SOOX
-
20
....
-qm El
I3
lo:,
.
I
.
I
.
1
.
I
.
500
700
900
Tox
Fig.3 :Amounts of H2 produced by reaction with steam at 500 "C of the PtIA1203 catalyst reduced for I h at various temperatures (Tr)
Fig.4 :Amounts of H2 produced by
reaction with steam at various temperatures (Tox) of the.PtlAl2Oj catalyst reduced for I h at 850 "C.
636
IOSITPOS on RhAl and PtAl alloys The behavior of the alloys in IOS/TPOS was examined to verify that their reaction with steam is not a limiting step of the titration of reduced species in alumina supported catalysts. Moreover the new phases formed during the reaction can easily be determined by XRD. The RhAl allov reacts easily with steam at Tox r 500°C. On a fresh sample, the first pulse of H20 yields an excessive amount of H2 which is not taken into account in the estimative of Q*. This high production of H2 is ascribed to the reaction with steam of carbon impurities. Fig.5 shows the variation with the pulse number of the amount of H2 during PTOS (by stage of 50°C). At each plateau of temperature, there is a slight deactivation of the alloy. Nevertheless, the initial activity is restored if the sample is maintained without H 2 0 injection for about 1 hour at the reaction temperature. Moreover, at a given temperature of reaction, a fresh alloy sample produces the same amount of H2 as a sample having already reacted at other temperatures. The initial values of Q* (Q*i) for each plateau of temperature are thus representative of the intrinsic activity of the alloy. The variations of Q*i with Tox are shown in Fig.6. An increase of the reaction rate with a very low activation energy is observed between 600 and 860°C. Above 860"C, there is a sharp increase of reactivity, Q*i being close to 10 mmol H/mol RhAl at 950°C and to 65 mmol H/mol RhAl at 1OOO"C. 2
d
mmol H / mol RhAl
1
0
663 ' C I
I
I
1
1
2
3
4
H P Pulsa numbor
Fig.5 :Amounts of H2 produced by reaction of RhAl with steam as a function of the pulse number (1 P l H 2 0 )
Fig.6 :Variation, with the temperature, of the initial amount of H2 produced by reaction with steam of RhAl alloy.
637
The sample oxided at 1000°C (20 pulses of water, total amount of H2 produced : 333 mmole H/mol RhAl, corresponding to 11.1% oxidation if only the A1 element is considered) was examined by XRD. The diffractogram recorded in Fig.7 shows that during oxidation by steam, the RhAl is transformed into metallic Rh and alumina : RhAl + 1.5 H20 Rho + 0.5 A1203 + 1.5 H2 (1)
IFig.7 :DifSractogram of the RhAl alloy treated in steam at 1000°C. Rho is quite visible on the diffractogram (new lines at 20 = 41.129, 47.850 and 69.885') but alumina is not, probably because a large part of the alumina produced by reaction (1) remains essentially amorphous, the a-Al203 microcrystals being too small. A baseline shift in the 20 region of 20-50" confirms the existence of an amorphous phase in the oxided alloy. Thus, alloy formation in reducing atmosphere appears quite reversible in the case of Rh/Al2O3 catalysts. Moreover, the continuous increase, with Tox, of the reactivity of the alloy definitely shows that the titration of reduced centres is not limited by the rate of reaction with steam of these centres. The initial amounts of hydrogen produced by reaction with steam of the PtAl allov are given in Fig.8 which shows that PtAl is about two orders of magnitude less active than RhA1. The H2 yields increase with Tox but remain, even at 900"C, extremely low (Q*RhAi/Q*ptAi = 180 at 900°C). The X-ray diffraction pattern of PtAl after heating at 1000°C under steam shows the transformation of the starting compound (fig.9) : the indexation of the lines leads us to conclude to the partial transformation of the starting compound in a richer platinum compound PtgA13 and in alumina. The X-ray pattern found for Pt5A13 is in agreement with crystallographic data given in Ref.11 (rhomboedric, structure type Ge3Rh5, a=0.541 nm, b=l.O70 nm, c=0.395nm, space group : Pbam).
638
0;pmol
H / mol P t A l
201
Fig.8 :Variation, with the temperature, of the initial amount of H2 produced by reaction with steam of PtAl alloy. roo
500
gooa
Tox
This result allows to conclude that the following reaction occurs during the heat steam treatment : 5PtAl+ 3H20 Pt5A13 + A1203 + 3H2 (2) The reaction yielding platinum and alumina : PtAl + 1.5 H20 Ptx + 0.5 A1203 + 1.5 H2 (3) does not occur with this intermetallic compound.
*
I
'
pTnL
--'-
1.m CIIKa1tz
0 PtAl
. A
iI 5
20
50.
35
Pt,AI, PtO
AI,O,
0
65
20 Fig.9 :Diffractogram of the PtAl alloy treated in steam at 1000°C.
639 DISCUSSION
10s experiments performed on Rh and Pt/A1203 catalysts show that treatments in H2 at Tr 2 500°C result in the appearance of reduced centres capable of dissociating H20 into hydrogen, in accordance with the reaction : reduced centre + H20 oxidized centre + H2 (4) The number of these centres apparently does not depends on the temperature of reduction. This result leads us to propose that the reduced centres are essentially located round the metal particles. The oxygen anions of the support bordering on the particles would be particularly reactive and could give, by reaction with H2, anionic vacancies and electrons capable of reducing the cations of the support, in accordance with the reverse reaction of the equation (I). The specific perimeter "10" of the metal/support interface is given by the equation (5) for hemispherical particles of rhodium rRef.121 : 10 = 8.81 x 105 x Xm x Do2 (m /g) (5) where Xm is the metal loading (wt.-%) and Do, the metal dispersion (%). The number of A1 ions present at the perimetric interface is : NAS= 10 x MS (ions / g) (6) where MS is the average number of A1 ions per meter (2.4 x 109). NAS can be compared to Q*/3, the experimental value of the number of reduced A1 ions (3e-/A1). For the Rh/Al203 catalyst (Do E 40% in most experiments), NAS = 3 pmol A1 ions per gram of catalyst whereas Q*/3 amounts to about 3-4 pmol A1 ions per g. The good accordance found between the two values supports the hypothesis concerning the location of the reduced centres. The formation of oxygen vacancies and the atomic rearrangement in the vicinity of the metal particles does not lead sfricfosenso to the formation of small particles of alloy since the formation of Rh-A1 or Pt-A1 bonds in the reduced centres is limited to pairs of atoms. These centres can react with steam much more easily than can the surface sites of the alloys. However the knowledge of the alloy reactivity give us precious information on the behavior of the reduced centres created in the catalysts. RhAl behaves exactly like the reduced centres of Rh/A1203 catalysts, i.e. both give rhodium atoms and alumina by reaction with steam. It can thus be stated that the structure of these reduced centres is equivalent to a rhodium-aluminium alloy. This apparently is not the case for the reduced centres of Pt/A1203 which can readily decompose H20 molecules, while platinum-aluminium alloys are relatively stable in steam (practically no oxidation). This proves that the structure of the reduced centres in Pt/A1203 catalysts is not equivalent to platinum-aluminium alloys. These centres are probably in an intermediary state of reduction and resemble PtAlOx suboxide species. The formation of reduced centres having a structure close to PtAl alloys requires certainly
640
higher reduction temperatures (T > 1000°C). Another relevant result of this study concerns the relative rates of the redox reactions in H2 and in steam : it seems that the formation of reduced centres is less rapid than their oxidation by steam (see Table 4). In cycling conditions, as in exhaust gas catalysis, the oxidized forms of the Rh-A1 centres will be relatively stable. A similar conclusion can be given for platinum. However if at very high temperature (T > 1000°C) Pt-A1 alloy could be formed, their high stability could cause drastic change in the catalytic behavior. ACKNOWLEDGEMENTS
The authors want to thank V. Paul-Boncour for her effective help in the XRD identification of platinum-aluminum alloy phases. F.S. thanks the Algerian Department of Higher Education for several leaves of absence from the University of Algiers. This work was performed in the framework of an International Cooperation Programme between the Universities of Poitiers and of Algiers (88 MES 87 Project). REFERENCES
1 2
3 4
5 6 7 8
9 10 11 12
G.J. Den Otter and F.M. Dautzenberg, J. Catal. 53 (1978) 116. K. Kunimori, Y. Ikeda, M. Soma and T. Uchijima, J. Catal. 79 (1983) 185. C.R. Apesteguia, T.F. Garetto and A. Borgna, J. Catal. 106 (1987) 73. D. Duprez and A. Miloudi in "Proc. Int, Conf. on metal-support and metal-additive effects in Catalysis, Lyon, 1982" (B. Imelik et al., Eds.), Stud. Surf. Sci. Catal. Vol.11 p 179, Elsevier Publ., Amsterdam (1982). D. Duprez, J. Chim. Phys., 80 (1983) 487. T. Paryjczak, W.K. Jozwiak and J. Goralski, J. Chromatog., 166 (1978) 65. M. Kobayashi, Y. Inoue, N. Takahashi, R.L. Burwell, J.B. Butt and J.B. Cohen, J. Catal. 64 (1980) 74. R. Ferro, G. Rambaldi and R. Capelli, Atti della accademia nazionale dei Lincei, 36 (1 964) 49 1. S. Bhan and H. Kudielka, Zeitschrift fur Metalkunde, 69 (1978) 333. S.J. Tauster, S.C. Fung and R.L. Garten, J. Am. Chem. SOC.,100 (1978) 170. R. Huch and W. Klemm, Z. Anorg. Chem., 329 (1964) 123. D. Duprez, P. Pereira, A. Miloudi and R. Maurel, J. Catal., 75 (1982) 151.
A. Crucq (Editor), Catalysis and Automotive Pollution Control II 0 199 1 Elsevier Science Publishers B.V., Amsterdam
64 1
THE INFLUENCE OF THREE-WAY CATALYST PARAMETERS ON SECONDARY EMISSION B. Engler, E. Koberstein, D. Lindner, E. Lox Degussa AG, Physical Chemistry Research Department ZN Wolfgang, P.O.Box 1345,6450 Hanau I (Federal Republic of Germany) ABSTRACT
Integral model gas tests and differential model gas measurements on model catalysts were performed under extreme operating conditions to elucidate the influence of different washcoat / precious metal combinations on the oxidation and the reduction of sulfur and nitrogen containing components at fixed reaction conditions. Major differences between the model catalysts were observed. The activity for hydrogen producing reactions such as the steam reforming and the water gas shift reaction was found to be parallel to the activity to convert SO2 to H2S and NO to NH3, for several of the model catalysts, but a unique correlation did not emerge. INTRODUCTION
Three-way catalysts based on Rhodium, Platinum and/or Palladium formulations are nowadays state of the art technologies to substantially eliminate carbon monoxide, hydrocarbons and nitrogen oxides from the exhaust gas of Otto-engines. The desired oxidation and reduction reactions proceed with high conversion rates in a closed loop catalyst arrangement when the exhaust gas composition equals a stoichiometric composition and when the catalyst achieves its normal operating temperature at appropriate space velocity. Conditions however can occur where the desired oxidation and reduction reactions cannot reach completion, so that intermediate oxidation and reduction products or products of side reactions may be formed. Although these so-called secondary emissions occur only with low concentrations, some of them could cause customers discomfort or are suspected to have an undesired influence on the atmospheric chemistry. So it is an ongoing challenge for the catalyst manufacturers to optimize the
642
catalysts in a way that they minimize the emission of secondary components under broad operating conditions, while still maintaining their major functions. The most important secondary components may be devided into two groups: 1- The formation of sulfur containing components like H2S and SO3 is due to the sulfur content in the fuel and causes either an unpleasant smell in the case of H2S or might result in a catalyst deactivation by the formation of surface oxides. Intensive research work was already carried out to elucidate the influence of the sulfur components on the performance of three-way catalysts ([ 13 - [4]). 2- The combustion process leads to the formation of nitrogen oxides, mainly NO, which contribute to the acid rain phenomenon [ S ] and to an enhanced immission of ozone [6].Under certain operating conditions of a three-way catalyst these oxides can be reduced to N20, which is suspected to contribute to the green house effect ([7], [8]), and to NH3. Under oxidizing exhaust gas compositions, NO2 can be formed. In the present study an approach is made to elucidate the role of the different washcoat/precious metal-combinations on the formation of secondary components H2S, SO3 and NH3, N20, N 0 2 . It should be emphasized that model catalysts were examined under extreme reaction conditions in a laboratory integral synthetic gas reactor. The activity of the model catalysts for the water gas shift reaction and the steam reforming reaction was measured in a Berty-type reactor. It was tried to establish a correlation between the activity for hydrogen production and the activity to reduce sulfur- and nitrogen-oxides. EXPERIMENTAL
Catalyst preparation and characterization The catalyst samples were prepared by coating a ceramic monolith (cell density: 400 cpsi, wall thickness: 6.5 mil) with an aqueous slurry of 'y-Al203 or with cerium nitrate. After drying the washcoated supports and calcination at 700°C for 2h in air, the samples were impregnated with aqueous solutions of chloroplatinic acid, rhodium (111) chloride or palladium (11) chloride, respectively, dried and calcined at 600°C for 2h in air. It should be pointed out that the precious metals are loaded in an equimolar ratio. The precious metal dispersion was determined by a COtitration technique; the BET surface area and the x-ray photoelectron spectrometric measurements (XPS) were reported previously [9]. The chemical composition of the catalysts and the results of the physico-chemical characterization are given in Table 1.
643 TABLE 1
Composition of the catalyst samples BET surface area of the washcoat and precious metal dispersion (Washcoat-loading = 200g/l support) Washcoat BET Surface Area (m2/g)
Precious Metal
Type 41203 A1203
A1203 CeO2 Ce02 Ce02
146 146 146 32 32 32
Pt Pd Rh Pt Pd
Rh
I
Precious Metal Dispersion (%)
Loading (g/ft3)
50 27 26 50 27 26
16 47 36 9 31 21
Integral model gas test The integral model gas reactor, schematically shown in Figure 1, consists of a gas mixing section, a reaction section and an analytical section. A detailed description is given in [lo]. On-line analysis is performed for the hydrocarbons by a flame ionization detector; for CO, C02 and N20 by IRdetectors, for NOx by chemiluminescence detector and for 0 2 by a paramagnetism detector. The concentrations of NH3, H2S, SO2 and so3 are determined by off-line wet chemical methods. For the measurement of H2S the product gas is absorbed during 15 minutes in an aqueous solution of zinc acetate buffered with sodium acetate. After the addition of an aqueous solution of N,N-dimethyl-l,4-phenylenediammonium dichloride and ammonium-iron-(11)-sulfate in sulfuric acid the S= ions form the methylene blue complex. The extinction measured by a photometer at 667 nm gives the sulfide concentration of the solution, from which the gas phase H2S concentration is calculated. The concentrations of SO2 and SO3 are measured by passing the gas mixture through a mistcatcher, which is kept at 90°C, so that SO3 reacts with water to sulfuric acid which condenses. The remaining SO2 containing gas is lead through an absorption bottle filled with an aqueous solution of H2 (3 weight-%), where SO2 is oxidized to SO3. The sulfuric acid contents in the mistcatcher and in the absorption bottle are determined by titration with a barium acetate solution.
644
NZ
COZ HCIN,
NOIN,
SO,” COIHJN,
Figure I - Schematic drawing of the integral test apparatus
To study the interactions of sulfur containing components with the model catalysts a three step test procedure was developed as shown in Figure 2. At a space velocity of 50,000 h-1 , temperature at the catalyst inlet is kept constant at 500°C.
V0l.-•
SO,
+
SS
+
3H,4
so,
f
1, 0,
2H20
-
phase 111 SOQ pure
y
SO,. SOs*
+
3H,+H,S 4H2-bH,S
+
ZH,O
4
340
1.02
CO: 4.21 : C,y:
0.88t
0.08
4.21
0.680
4 21
0.08
0.059
0.08
0 10
0,:
0.10 j
0.10
1.48
so,:
- i
0.002
0.002
4:
1.40 j
1.40
0.220
1.40
d
equilibrium conversion
Figure 2
80
2
15
time , min
- Test procedure for the evaluation of
the interaction of surfiir components with the catalysts
645
After conditioning the catalyst under a sulfur free net reducing gas mixture and awaiting the equilibrium conversion, 20 ppm of SO2 are added to the gas stream and the reduction of SO2 is investigated [phase 1 in Figure 21. After a fixed time the gas mixture is changed to net oxidation while the addition of 20 ppm SO2 is continued for one hour [Phase 2 in Figure 21. Here the direct conversion of SO2 to SO3 is measured and from the sulfur balance, the amount of sulfur oxides stored on the catalyst is calculated. 2 S 0 2 + 0 2 w 2SO3 (1)
After two minutes purging with pure nitrogen the original S02-free net reducing gas composition is adjusted again [Phase 3 in Figure 21. In this third phase the reduction of stored sulfur oxides to H2S is studied.
Y H ~ in S this case is related to the amount of sulfur stored on the catalyst. The detailed gas composition for the three step-test procedure is given in Table 2. Since some sulfur might be retained on the catalyst at the end of the experiments, the sulfur content of the used catalysts was measured with a LECO-apparatus. In this way a mass balance for sulfur could be calculated. The conversion of NO to NH3, N20 and NO2 was investigated in two different experiments. The formation of NH3 was examined at a catalyst inlet temperature of 400°C with a net reducing gas composition (while the space velocity was kept constant at 52,000 h-1 . The ammonia in the product gas mixture is absorbed in an aqueous solution of boric acid (1% by weight) for 15 minutes; the content of ammonia is determined by titration with HCI. The conversion of NO to ammonia is calculated from:
The gas composition of the NH3-test is given in Table 3.
646 TABLE 2
Gas composition for the three step-test procedure. Concentrations in vol 9% Balance N2
-
Coomponent
co
C3H8 02
Conditioning 4.21 0.08 0.10
so2 H20 NO
1.40 10 10 0.15
h
0.88
H2
co2
I
I
Phase 1 4.21 0.08 0.10 0.002 1.40 10 10 0.15
I
I
0.88
Phase 2 0.66 0.06 1.48 0.002 0.220 10 10 0.15 1.02
Phase 3 4.21 0.08 0.10
-
1.40 10 10 0.15 0.88
TABLE 3
Gas composition for the NH3-test ( h = 0.975 ) (Balance N2) Component CO Conc.(vol%) 1.5
C3H6 C3H8 0.033 0.017
02 H2 0.636 0.5
C02 10
H20 10
NO 0.1
The oxidation of NO to NO2 and the reduction of NO to N 2 0 were investigated in a light-off test, where the temperature was raised from 750°C to 450°C with a heating rate of lO"C/minute at a space velocity of 50,000h-1. The h value was adjusted to 1.02 and to 0.984 .The detailed gas composition is given in Table 4. The conversions of NO to NO2 is calculated from :
The conversion of NO to N20 is calculated from :
647
TABLE 4
Gas composition for the NzO/NO-test (concentrations in vol %; balance N2) Compound Concentration
h 0.984 1.02
co
C3H8
02
H2
C02
H20
NO
1.0 1.0
0.05 0.05
0.6 1.5
0.33 0.33
14.0 14.0
10.0 10.0
0.1 0.1
i
Differential model gas test The differential model gas tests were carried out in a Berty-type reactor. A detailed description of the experimental set up is given in [ 111.Two hydrogen producing reactions were investigated:
- the water gas shift reaction: CO + H20 3 C02 + H2
(8)
- the steam reforming reaction of propane C3Hg + 3 H20
3 CO + 7 H2
(9)
The reactor was heated from room temperature to 550°C with a heating rate of 3"C/minute. In both series of experiments the space velocity was 6,000 h-1, the gas compositions used are summarized in Table 5. TABLE 5
Reaction
Component
Shift Reaction
H20
Steam Reforming
C3H8 H20
co N2
N2
Concentration ' 2 10 88 4 61 35
648 RESULTS AND DISCUSSION
Catalyst characterization The BET surface area and the dispersion of the precious metals were already given in Table 1. Both on ceria and on alumina the dispersion follows the order: Pd > Rh > Pt. The higher dispersion of these metals on alumina might be related to the higher BET-surface area of alumina compared to that of ceria.
Integral model gas test The oxidation and reduction of sulfur dioxide The reduction of SO2 under reducing gas compositions (phase 1) leads to a complete SO2 conversion on the ceria based catalysts and to a nearly complete conversion on the PdAl203 catalyst as is shown Figure 3. For the alumina based catalysts, the conversion parallels the precious metal dispersion. In phase 2 of the test the interactions of the catalysts with So;! in a lean gas mixture are investigated. As shown in Figure 4 on the ceria based catalysts a lower SO2 conversion was obtained than on the alumina based catalysts. For the ceria based catalysts, the conversion of SO2 is dependent on the type of precious metal present. The conversion decreases in the order: Pt>Pd>Rh Conversion of SO,, % 100 'OOr-----l
10
40
PO
0
Ai,O,
CeO,
Figure 3 Conversion of SO2 to H2S under a reducing gas composition (test procedure, phase I )
Figure 4 Conversion of SO2 under oxidizing gas compositions (test procedure, phase 2)
649
This order parallels the oxidation state of the precious metals on ceria, as measured by XPS: Rhodium on ceria is deeply oxidized, while the Platinum on ceria shows only a small tendency to oxidize [9]. On alumina the precious metals hardly influence the conversion of SO2 : about 80% of the SO2 offered is either oxidized or stored on the catalyst surface. The different behaviour of the precious metals on alumina compared to that on ceria also parallels the different oxidation states of the precious metals on these supports; as measured by XPS the precious metals on alumina are generally less oxidized than on ceria [9]. Figure 5 shows the conversion of SO3 under the oxidizing conditions. The amount of SO3 leaving the reactor increases in the order: C e O m < CeOfld = Al2O3/Rh Pd > Pt and, as discussed before in Figure 4, the differences are less distinct for the alumina supported catalyst than for the ceria supported catalysts. Also here the parallels between the XPS-data [9] and the conversion are apparent. The results can be summarized as follows: on the ceria based catalysts nearly the total amount of sulfur offered is leaving the reactor: 38% as SO2 and 38% as SO3 -for CeOflt -for CeOfld 60% as SO2 and 14% as SO3 -for CeOflh 76% as SO2 and 2% as SO3 whilst on the alumina based catalysts only about 40% of the sulfur could be recovered during phase 2. In phase 3 of the test procedure the reduction of the stored sulfur oxides to H2S in a rich gas mixture was investigated. The results are given in Figure 6. Conversion of stored SO- to HS. X
A'P,
CeO,
Figure 5 Conversions of SO2 to SO3 under lean exhaust gas conditions (test procedure, phase 2 )
Conversion of SO, to
CQ,
Z
ma
I
A',%
CeO,
Figure 6 Conversion of stored suljkr to H2S under rich exhaust gas conditions (test procedure, phase 3 )
650
The conversion to H2S is related to the amount of sulfur not recovered in phase 2 of the test. Both, the complete reduction of SOX to H2S on the cena based catalysts and the uncomplete conversion on the alumina based catalysts parallel the precious metal dispersion and comply with the results obtained in phase 1 of the test (Figure 3).There seems to be no difference concerning the reduction of SOX to H2S whether the sulfur component is part of the gas phase or whether it is already stored on the catalyst surface. The sulfur balance during phase 2 and phase 3 of the Recovered Sulphur, % experiment is shown in Figure 7 100 and clarifies this effect: for the Ce02 80 based catalysts under the chosen reactions conditions, no sulfur is irreversibly stored, while on the 40 A1203 based catalysts 20 to 40% of sulfur is left on the catalyst. PO Because of the deficits in the 0 sulfur balance, a sulfur analysis was AI,O,IPi A 4 q l P 3 A/qIRh CeOJPi Ce0,IPd CeQlRh carried out with the alumina based DH,S W y s . 80, U y s so,. sq catalysts. The results are given in Table 6. The numbers are based on Figure 7 the SO2 offered during phase 2 and Sulfur balance for desorbed so2, neglect the amount of sulfur that so3 andH2S during Phases 2 and.? might have been stored during of the test phase 1.
.
IAu&
Sulfur balance for alumina based catalysts
I
I
I
Catalyst
1
Recovered Sulfur (%) Phase 2
I
Phase 3
Ion the catalyst
I Total
65 1
equilibrium conversions, U
.""
60
too 80
without SO, phaae I
addition of SO2 phaae I
without S q pha8e 111
A 100% recovery of sulfur could not be recorded for any of the alumina based catalysts what leads to the suspicion that some sulfur desorbs during the purging phase and/or that some of the stored sulfur leaves the catalyst as SO2 during phase 3.This is the subject of further investigations.
The effect of sulfur oxides on the Catalvtic activitv for CO. HC and NOx removal under reducing- p- a S conditions
60
The influence of sulfur on the three-way activity of the catalysts is 20 shown in Figure 8a-c. For a reducing gas composition the 0 conversions of CO, C3H8 and NO are compared with and without SO2 100 (phase 1) and at the end of the test again, in the absence of SO2 ao (phase 3). In all cases the oxygen was 00 completely converted. The oxidation of CO under 40 these conditions is strongly 20 suppressed by the addition of S02, whereby the alumina based catalysts almost completely loose their activity. Also the cerium based catalysts are affected, but CeOz/Rh still attain conversion rates of about 20% or more. During the test procedure the Figure 8 A120gPd-samples are not regeneInfluence of surfur on the activity of rated: in contrast to that, the the catalysts for CO (a),HC (b) and CeOflh-catalysts nearly regain their NOx (c) removal initial activity. 40
652
For the oxidation of C3Hg the losses in conversion are even more dramatic under the chosen reaction conditions. Initially, the highest conversion is found on Al203/Rh and decreases in the order:
None of the catalysts regain their activity for CgH8-remOVal. While the virgin catalysts convert NO nearly quantitatively, the reaction is strongly inhibited by SO2 on A1203/Pt,.A1203/Pd and Ce02Pd. Even in phase 3 of the test, these oxide/precious metal-combinations do not regain their original activity. On the contrary the influence of SO2 on Al203/Rh, CeOz/Pt and CeOflh for the the reduction of NO under the given test conditions is negligible. As can be seen from an oxygen mass balance, calculated from the given gas composition and the measured conversion rates in phase 1 of the experiment, i.e. in absence of S02, only in the case of Al203pt there is a sufficient amount of oxygen for the conversion of propane by the reaction with oxygen. This is due to the low conversion of propane on this catalyst. For all the other catalysts tested an oxygen deficit of 100% to about 200% was calculated. This might be explained by the steam reforming reaction, where water acts as the oxidizing agent. Similarly, the calculations show for the CO-oxidation that with the exception of Al203pd the measured CO-conversion can only be explained when the water gas shift reaction is taken into account. Both reactions are strongly affected by sulfur oxides, as was found by Gandhi et al. [12]. The reduction of nitrogen oxide to NH3 Under the reaction conditions chosen the NO almost completely is reduced to NH3; on all of the catalysts studied. The lowest ammonia conversions were measured with Al2O3pt and CeOflh. As was also found for the reduction of S02, the ability to reduce NO reflects the precious metal dispersion on the alumina based catalysts. The oxidation and reduction of NO The results for the conversion of NO to N 2 0 are shown in Figure 9a-c. Comparing Al2O3pt to Al2O3/Rh under oxidizing and under reducing conditions, the rhodium catalyst attains the higher total conversion of NO. Changing the gas mixture from a rich to a lean composition, the conversion of NO on the Rh-catalyst drops significantly: from 100 to 23%, whereas for the Pt-catalyst, only a small change in conversion, from 17 to 14% is observed.
100
Convorrlon of NO, %
Convorrlon to N 0, U
:F=
80
80
:onvorrlon to N 4 , %
,
653
--
40
PO
Lambda 0.084
Lambda 1.02
0.084
1.OP
0.914
1.02
Figure 9 Conversion of N O to N 2 0 and NO2 (T = 350 “c) Under the reaction conditions used, N 2 0 is formed in concentrations of less than 40 ppm; this corresponds to a maximum yield of 4%. In the reducing gas mixture no N20 is detected on Al203/Rh, while on Al2O3/Pt a small amount of 20 ppm N20 found. In an oxidizing atmosphere no N 2 0 was produced on Al203/Pt whereas on Al2Og/Rh, 40 ppm were emitted. The oxidation of NO to NO2 under reducing conditions is nearly totally absent; only on Al2O3/Pt a conversion to NO2 of less than 1% was measured. In the oxidizing gas mixture the conversion to NO2 is 7% for AI2O3Pt and 1.5% for Al203/Rh under the reaction conditions chosen. From these experiments the following conclusions can be drawn: Under reducing conditions the Al2Og/Rh-catalyst shows no N20 and NO2 formation, all of the NO offered is converted to N2. Under oxidizing exhaust gas compositions, the A1203/Pt-catalyst shows a high conversion of NO to NO2
Differential model gas test The results obtained in the Berty reactor for the water gas shift reaction and for the steam reforming reaction are shown in the Figures 10 and 11 respectively. Keeping the reaction conditions constant the conversions differ over an order of magnitude, especially for the water gas shift reaction. For both hydrogen producing reactions, independent of the precious metal, the ceria based catalysts are the more active. For the water gas shift reaction the Ce02Pt and Ce02/Rh combinations give the highest conversion; for the steam reforming reaction also the Al2Og/Pt-catalyst was very active.
654
Convoroion of C,H,
Convorolon of GO, lb
I 80
o
o
, lb
7
60
40
10
0
4 0 s
coo,
Figure 10 Results of differential reactor tests for the water gas sh$t reaction (T = 400°C)
w.
GOO,
Figure I1 Results of differential reactor tests for the steam reforming reaction of propane (T = 400°C)
A parallel emerges between the activity of hydrogen producing reactions and SO2 reduction that higher conversions occur on a ceria based catalysts as opposed to alumina based catalysts. CONCLUSIONS
From the results presented above the following conclusions can be drawn. For the reduction of SO2 or stored SOXto H2S under reducing gas conditions an enhanced activity was observed for Ce02 based catalysts as compared to the alumina based samples. This agrees with the results in Berty experiments, where the highest reaction rates for hydrogen producing reactions were found with ceria based catalysts, for a given precious metal. For precious metals supported on alumina the ability of the catalysts to reduce SO2 to H2S reflects the precious metal dispersion, whereas for the oxidation of so3 a parallel with the oxidation state of the precious metal emerges. The interactions of SO2 with the washcoat components A1203 and CeO2 differ distinctly: A1203 attains the higher irreversible sulfur storage which might lead to a severe catalyst deactivation, whereas on ceria most of the SO2 offered leaves the converter unreacted. A sulfur balance shows that on alumina only 60 to 80% of the sulfur can be recovered during the test procedure, on ceria the total amount of sulfur was found. The three-
655
way activity of all catalysts is decreased by the presence of SO2 especially the conversions of CO and C3H8 under reducing conditions are heavily affected. In both cases a mass balance demonstrated that the water gas shift reaction and the steam reforming reaction contribute significantly to the conversion of co and C3H8, respectively. For the conversion of C3H8 the sulfur leads to a permanent loss in activity for all catalysts; for the conversion of CO, the poisoning seems to be reversible on CeOflt and C e O m . For the formation of ammonia in a reducing atmosphere it was found that under the chosen reaction conditions all investigated catalysts convert 70% or even more of the offered NO into NH3. The lowest conversions were obtained with Al2O3/Pt and C e O m . As discussed for the reduction of SO2 here again the activity of Al203-based catalysts reflects the precious metal dispersion. Only small amounts of NO were reduced to N 2 0 on Pt/A1203 and Rh/A1203; under the test conditions chosen a maximum of 40 ppm N 2 0 could be detected. Acknowledgements The contributions of Dr. Albers (XPS-measurement), Dr. Leyrer and Mr. Gorsmann (Berty reactor experiments) are gratefully acknowledged. REFERENCES H.Rohlfing, M.Peters & A.Konig; Motortechnische Zeitschrift ,50 (1989) ,269/272 LGottberg, E.Hogberg & K.Weber, SAE Paper 890491 (1989). E.S.Lox, B.H.Engler & E.Koberstein; SAE Paper 890795 (1989). J.C.Summers & K.Baron; J. Catal. 57 (1979),380/389 J.C.Davies; Chem. Proc.Tech. Int. (1990), 101/109. G.A.Ahrens, A.Friedrich & N.Gorissen; GIT Fachz. Lab. (1990). R.T.Ellington & M.Meo; Chem. Eng. h g . (1990), 58/63. J.Jaccob & K.R.G.Hein; VGB Kraftwerkstechnik 68 (1988) 841/843 B.Engler, & P.Schubert; Appl. Catal. 48 (1989),71/92 E.Koberstein ;Chemie in unserer Zeit 18 (1984),177/199 E.Koberstein & GWannemacher; Studies in Surface Science and Catalysis, Vol. 30, Elsevier, Amsterdam (1987), 155/172 (Editors: A. Crucq and A. Frennet). H.S.Gandhi,A.G.Piken,H.K.Stepien& M.Shelef; SAE Paper 770196 (1977). SYMBOLS AND INDICES
X : conversion of, % 0 : initial
s : solid
Y : conversion to, 96 g : gas
This Page Intentionally Left Blank
A. Crucq (Editor), Catalysis and Automotive Pollution Control ZI 0 199 1 Elsevier Science Publishers B.V., Amsterdam
657
IMPROVEMENTS IN TECHNIQUES FOR REDUCING EMISSIONS BY USING COMPUTER SIMULATION M.Hashimoto1, R Matsumural, F. Yukawal, M. Saitoh2, M. Matsumoto2
( I ) Engine Design Department (2) Central Engineering Laboratories Nissan Motor Co Ltd. ABSTRACT In recent years environmental concerns are leading to demands for even lower automobile emissions and fuel consumption. It is becoming necessary to improve not only the emission control system but also the vehicle and engine characteristics, themselves. Investigating each factor that affects automobile emissions, such as emission control system, vehicle weight, tire-rolling resistance, and many others is important to carry out automobile emission reduction. Additionally, it is necessary to consider deterioration and irregularity of system characteristics. In order to reduce automobile emissions and optimize the emission control system, a time-consuming process involving repeating many different tests should be achieved. For these reasons, we have developed the total emission simulation model. And by using this simulation we have carried out the investigation of the effect of each parameter. OUTLINE OF CALCULATION MODEL
OVERALL-MODEL
The simulation method used for these parameter studies is described below. By this method, air and fuel flow and the reactivity status of the catalyst are continuously calculated for vehicles operated under specified speed and acceleration modes. At the same time, the test conditions are displayed on a computer screen and the total amount of each component of the emitted exhaust is calculated and displayed. The model is organized into operational blocks, relative to the vehicle operating mode. These are engine condition calculation blocks, air measurement and fuel transport calcu-lation blocks, air-fuel ratio control blocks, and emissions before/after catalyst calculation blocks. The calculation procedures are outlined in Fig.1 and 2.
658
1
Intake air volume Intake manifold pressure
Calculation of engine operating conditions
Enginespeed,Torque
I
~
Intake air volume
1
i
,
AIF
;'~ A p, Test sequence
1 Airflow metering
:E:ty
p a v i o r in intake manifold
time
1
converter inlet and outlet
+
[
ci
Response d e l e d
t=t+ot
t
test sequence
7
Fig.1 Procedure of simulation
Fig.2 Outline of calculation model ENGINE OPERATING CONDITIONS
of emissions and exhaust
Mass of intake air
Fig3 Calculation of engine operation
In the engine operating conditions, calculation blocks, based on the vehicle specifications for vehicle weight, drivetrahtires, etc., and tire-rolling resistance during each running mode, engine torque, engine speed, volume of air entering the cylinder, volumetric efficiency, intake manifold pressure, and open throttle position corresponding to the engine operating conditions are determined. The calculation procedures are shown in Fig.3.
AIR MEASUREMENTS AND FUEL TRANSPORT MODEL
The calculation procedures are shown in Fig 4. In the air measurements calculation block, an air flowmeter is used to measure the volume of air added to fill up the cylinder air intake volume when the collector pressure is low under transient operation conditions, and the detection delay is approximated as a first order delay. The system is illustrated in Fig 5. The volume of air flowing through the air flowmeter, QAFM,can be given as.
659
0
With respect to delayed fuel transport, the experimental equation is constructed and the attempt is made to express it using a few ---{Diameterof air flowmeter I Air flow rate parameters.[1],[2],[3] The model of t.. ~.~ r llowmeter outpfi h r flowmeter response delay I fuel transportation is described below. Part of the fuel discharged by the injector is directly taken into the cylinder and the rest is accumulated in .............. the manifold as wall flow. The wall Fuel mntml rylnal 1 flow will be vaporized and then taken into the cylinder.The system is illustrated in Fig 6. Mass flow of intake air
6 b
.+
pj/ I
Fuel quantity entering cylinder
Fig.4 Calculation of air measurement and fuel transport
Air flowmeter Onnl Fuel injected quantity M F Amount of fuel wall flow M ~ S air S Rar passing lhrough air HoWmeter M ~ SrSr Rov R m m g in10 cylinder CM Mass air I" intake manifold and mllBCtor
o<x<1 O
QAFM
mn
Fig.5 Calculation of air flow metering
X x QFIW
Fig.6 Fuel transportation in intake manifold
AIR-IWEL RATIO FEEDBACK-SYSTEM
The value of the feedback control signal was calculated from the characteristics of the 0 2 sensor. [4],[5] The calculations took into account the time lag for the combustion gas to reach the 0 2 sensor and the response delay inherent in the sensor. [6] The response of the 0 2 sensor was approximated using a first-order delay and the dead time of the device. The detection delay for the air-fuel ratio was calculated so as to include the time needed to process the 0 2 sensor signal in the engine control circuit. The calculation procedure is outlined in Fig 7.
660 CALCULATION OF CATALYST REACTIVITY CHANGES
The concentration of each exhaust gas component emitted from the engine was continuously calculated during engine operation. The calculations were based on the engine speed, torque, and air-fuel ratio in the cylinder found from the above mentioned calculations. The exhaust gas component concentrations were obtained by interpolation from experimental data recorded beforehand.
AIF ratio
i;
Feedback coefficient
Fig.7 Lambda control
The resulting concentrations were used to calculate the fluctuation in emissions flowing into the catalytic converter. The calculations took into account the dispersion and mixing of exhaust gas in the exhaust manifold and also the time lag for the gas to reach the catalyst. The modeling procedure outlined in Table 1 was used to prepare a simple model capable of expressing the dynamic-conversion performance of the catalyst under various operating conditions.[7] The main parameters of those conditions were experimentally determined in advance. A simple model expressing the dynamic characteristics of the catalyst was prepared according to the following formula.
where:
CiOUt :outlet concentration of component i (mol/m3) Ci'n :inlet concentration of component i (m0i/m3) Pij :reaction probability between gas component j and absorbed component i qi :probability that gas component i will be absorbed by catalyst ri :quantity of component i absorbed (mol) ArRi :decrease in absorbed component i as a result of reaction with gas component (mol) Arli :increase in quantity of gas component i absorbed by catalyst (mol)
66 1
TABLE 1 Catalyst effect
I
Exhaust
[
A/F atio Oxidized by 0 2 and NOx
CO
1
NOX
I
I I
02
~~
Oxidized by 0 2 and NO, Reduced to basic elements by HC absorbed by catalyst under rich A/F ratio and then absorbed by catalyst Oxidizes HC absorbed by catalyst under rich A/Fratio and then absorbed by
Rich Oxidized by 0 2 and NOx absorbed by catalyst under lean A F ratio and then absorbed by catalyst Oxidized by 0 2 and NOx absorbed by catalyst under lean A/F ratio Reduced to basic elements by HC and CO,(H2) Oxidized HC and CO,(H2)
LA4HOT EMISSION S I MULATlON
In “ 7
7
2 0
z 90In
2 ........
-0 1 m
....
.....Vehicle speed w 160.0
165.0
170.0
180.0 TIME(SEC)
185.0
Fig.8 Simulation result (AIF control)
190.0
195.0
662 STUDY OF CALCULATED RESULTS
EMISSIONS UNDER LA4-HOT MODE
In this simulation, the vehicle is operated in the LA4-HOT mode. The fluctuations in air-fuel ratio corresponding to this mode are shown in Fig 8, and the equivalent fluctuations in the concentration of each exhaust gas component are shown in Fig 9. The spike in the concentration of each exhaust gas component waveform are due to fluctuations in the air-fuel ratio during accelerations. LA4HOT EMISSION SlMULATlON I
I
I
I
I
I
I
x
"A
s?
o x
P
8
160.0
165.0
170.0
180.0
185.0
190.0
195.0
TIME(SEC)
Fig.9 Simulation result (emission component) AIR-FUEL RATIO FEEDBACK CONTROL
A simplified diagram of the air-fuel ratio feedback control waveforms is shown in Fig 10. The output of the 0 2 sensor increases or decreases depending on whether the air-fuel ratio is rich or lean. A threshold value was established relative to the output of the 0 2 sensor to serve as a reference for determining whether the present air-fuel ratio is rich or lean. As a result of repeating the feedback control compensation to increase and decrease the fuel supply, the actual air-fuel ratio repeats an alternating rich-lean pattern, as indicated by the curve at the top of Fig 10.
663
Lean
LA 4(HOt)
A= I Rich
* Calculated
2.0-
Oxygen sensor output
PL=4 25% vel
1OO(%)
I
I
0.4
0.2
00
HC (g/mile)
control signal
Fig.10 AIF feedback control
I
I
10
2.0
CO (g/rnile)
Fig.1 I Comparison between experimetal and calculated data for feedback gain PL
The simulation model was used to calculate HC,CO,NO, levels for various PL values. The results obtained under the LA4-HOT mode with experimental data are shown in Fig 11. Although the calculated levels of NO, were somewhat higher than the corresponding experimental data, the characteristics of the calculated results closely agreed with the experimental data. Thus, this simulation can successfully evaluate exhaust emissions. STUDY OF THE EFFECT OF EACH FACTOR ON EXHAUST EMISSION
The factors of present studies are shown in Table 2.
TABLE 2 Changed factors
664 STUDY OF EXHAUST EMISSION SENSITIVITY TO EACH FACTOR
Using this simulation in LA4HOT mode, vehicle specifications and emission control system characteristics were selectively changed, and the sensitivity of exhaust emissions to each change was measured. Sensitivity means the exhaust emission change rate relative to each factor change rate. The results are shown in Fig 12 and 13. (1) Although the vehicle weight affects the emissions before catalyst, the overall effect is low. (2) For the emissions after catalyst, however, the sensitivity is Fig.12 high for the 0 2 sensor response, volume of intake manifold, vehicle Exhaust emission sensitivity to each factor weight, and tire-rolling resistance. (calculation result after catalyst) The reasons for this are as follows: (i) Under transient operating conditions, changes in the air-fuel ratio control rate result from changes of the 0 2 sensor response. Also, increased differences between measured air flow and actual air entering the cylinder are due to increased volume of the intake manifold and vehicle weight. These changing conditions combined cause large fluctuations in the air-fuel ratio. As a result, the air-fuel ratio deviates from the most efficient range of the three-way catalyst. (ii) The volume of exhaust gas flowing into the catalytic converter Fig. 13 increases with increasing vehicle Exhaust emission sensitivity to each weight and tire-rolling resistance. As factor a result, catalytic conversion perfor(calculation result before catalyst ) mance falls.
665 STUDY OF THE EFFECT OF EACH FACTOR ON EXHAUST EMISSIONS
Using the simulation technique, the effect of each factor on exit emissions, in consideration of the deterioration and irregularity of each factor, added to the sensitivity in the foregoing discussion, were as follows. (1) The effect is high for the 0 2 sensor, catalyst, volume of intake manifold, vehicle weight, and tire-rolling resistance. The results are shown in Fig 14. (2) The studies on improving technique for reducing exhaust emissions from the view point of these effects are as follows. (i) The effects is exceptionally high for the 0 2 sensor. It is reasonable to improve the 0 2 sensor performance, itself, including deterioration and irregularity. But there are several important considerations for achieving optimum 0 2 sensor function. First is to lower the maximum temperature of the exhaust gas. Second is to choose a suitable temperature range that yields stable sensor performance. Third is to choose the best location for installing the 0 2 sensor. This location should be as close as possible to the combustion chamber to minimize the time lag for gas flow while ensuring equal exposure to exhaust gases from each cylinder. (ii) There are also several important considerations for achieving optimum catalyst conversion performance while minimizing deteriorating of the catalyst including, obviously, to improve the catalyst, itself. Similarly as for the 0 2 sensor, the maximum operating temperature of the vehicle should be lowered, and a suitable temperature range chosen that will yield stable catalytic conversion performance. (iii) The effect is also high for the air-intake system. It is necessary to investigate the design spec, along with consideration of power performance. (iv) Vehicle weight and tire-rolling Fig. 14 resistance have the effect of reducing all the exhaust emission components. Effect of eachfactor on t ~ h a u s t So it is important to reduce total emissions vehicle weight and tire-rolling (Calculation result after catalyst ) resistance.
666
Although the present study is carried out the hot mode, it can be seen by referring to the above F i g . 1 4 for deterioration and irregularity and the previous F i g . 1 2 for sensitivity that these values for each factor are quite different, In order to improve quality and development efficiency, it is important to investigate sensitivity by using the simulation technique, and then, by studying parts or system irregularity and deterioration, to proceed with development beginning with the most highly affected parts or system. CONCLUSIONS
(1) By using the total emission simulation technique, we have carried out an investigation of exhaust emission sensitivity to each factor and the effect of each factor on exhaust emission in consideration of the deterioration and irregularity of factor. These methods can be used to develop a system for reducing exhaust emission in consideration of the total balance of factors, which up to now has required a time-consuming process involving repeating different tests. (2) The present investigations were all carried out in the hot mode, and thus the number of factors that could be studied was not sufficient. We want to expand this simulation technique to include cold-mode and apply it to a wider range of improvements in the techniques for reducing exhaust emissions. REFERENCES 1
D. J. Boam, et al.,"A Model for Predicting Engine Torque Response during Rapid Throttle Transients in Port-injected Spark-ignitionEngines SAE Tech. Pap. Ser. 890565, (1989.) H. Wu, "A Computer Model for a Centrally-Located,Closed-Loop, Automotive Fuel Metering System ",ASME International Computer Technology Conference, Aug. (1980). Hatsuo Nagaishi et all, An Analysis of Wall Flow and Behavior of Fuel in Induction Systems of Gasoline Engines SAE Tech. Pap. Ser. 890837, (1989). H. U. Gruber and H. M. Wiedenmann, " Three Years Field Experience with the LambdaSensor in Auotomotive Control Systems ",SAE Tech. Pap. Ser. 800017, (1980). C. T. Young, "ExperimentalAnalysis of ZIo2 Oxygen Sensor Transient Switching Behaviour SAE Tech. Pap. Ser. 810380, (1981). Douglas R. Hamburg and Michael A. Shulman, A Closed-Loop A F Control Model for Internal Combustion Engines SAE Tech. Pap. Ser. 800826, (1980) Wei-Ming Wang, " Air-Fuel Control to Reduce Emissions, II. Engine-Catalyst Characterization Under Cyclic Conditions ",SAE Tech. Pap. Ser. 800052, (1980). "
2 3
"
'I,
4 5
'I,
6
"
'I,
7
A. Crucq (Editor), Catalysis and Automotive Pollution Control 11 0 199 1 Elsevier Science Publishers B.V., Amsterdam
667
Pd/A1203 CATALYSTS FOR THE NO-CO-02 REACTION : "IN SITU" DETERMINATION OF THE PALLADIUM STATE UNDER THE REACTANT MIXTURE
Jean Luc DUPLAN and Helkne PRALIAUDI
Institut de Recherches sur la Catalyse, Laboratoire Propre du C.N.R.S., Conventionne' a Wniversite' Claude Bernard Lyon I , 2 Avenue Albert Einstein, 69626 Villeurbanne Cedex, France. Telefax 78 89 47 69, Tel. 72 44 53 00. ABSTRACT The activities of various PdAl2O3 catalysts towards CO oxidation and NO reduction have been studied and related to the physico-chemical properties (metallic surface accessible to gases, redox properties, stability of the Pd-CO bonds) of the solids determined after contacting with the reactant mixture NO/C0/02/N2 without exposure to air. The activation under the reactants induces a particle sintering and formation of ionic palladium at the surface of the activated solids. The percentages of CO chemisorbed on ionic Pd and on metallic Pd have been determined by infrared spectroscopy and vary with pretreatments at high temperatures before contacting with the reactants, and with the composition of the feedstream.
INTRODUCTION
The catalytic control of the three major pollutants in automobile exhaust requires both reduction of nitrogen oxides and oxidation of hydrocarbons and carbon monoxide [ 1-51. Palladium based catalysts have been reported to carry out oxidation and reduction reactions in the same feedstream, for instance with NO/C0/02 mixtures. As thermal variations are expected to occur in driving conditions, especially with high speed rates, the properties of the solids (catalytic behaviour, metallic particle size, support modification) after thermal treatments have to be investigated [ 6 ] . Furthermore the state of these solids after contacting with the reactants has not been studied. This paper reports the activities per surface Pd atom of various Pd/A1203 solids for the CO oxidation and the NO reduction in presence of oxygen. The properties of the solids (metal surface accessible to gases, amounts of Pdn+, Pdo, stability of the Pd-CO bonds...) are determined after activation under the mixture NO/C0/02 without air contacting. The experimental techniques include transmission electron microscopy, 1
- Author to whom correspondence should be sent
668
chemisorption of H2 and infrared spectroscopy of chemisorbed CO. The activities are discussed in terms of the amounts of Pdn+ and PdO on the surface. EXPERIMENTAL
Several Pd/A1203 catalysts (around 1 wt% Pd) with various initial particle sizes were prepared. The y (6) alumina supports (around 100 m2/g) came from Degussa and from Rhone Poulenc. Palladium was deposited either by ion exchange with Pd chloride dissolved in hydrochloric acid, or by wet impregnation of the carrier with a known quantity of Pd acetylacetonate [7] dissolved in toluene, or by impregnation with Pd nitrate. Every catalyst was calcined in air at 723K (programmed rate 5K/mn, plateau 10 hours). The catalysts thus calcined are designates as "fresh" catalysts. Some batches were reduced overnight in a flow of hydrogen (41/h) by linearly raising the temperature up to 673K or 723K at a heating rate of 2K/min. For chemisorption measurements, the solids were evacuated at 623 or 673K for 2 hours. Some of the solids were treated at high temperatures, between 1073 and 1273K, in an oxidizing atmosphere (1.5% oxygen in nitrogen) in presence (10%) or in absence of water in the gaz stream. The following conditions were used: 0.5 g solid, flow rate 201/h, heating rate 5K/mn, plateau 10 hours. The solids thus treated are designated as "aged" catalysts. Adsorption experiments were conducted in a conventionnel volumetric apparatus. The irreversible chemisorption uptake of H2 was measured at 298K (Benson's method) [8] and at 348 K (Aben's method) [9] in order to take into account the formation of palladium hydrides. Both methods produced the same results. The metallic surfaces were calculated assuming that a Pd atom occupies 7.874 10-20 m2. The catalysts, subsequently aircontacted, were examined in a JEOL 100 CX electron microscope with a 0.3 nm resolution. Catalytic experiments were carried out in a flow system at atmospheric pressure. Nitrogen was used as a diluent. Reactant and product gases were analysed on-line by gas chromatography using a dual column (porapak and molecular sieves to separate 0 2 , N2, CO, C02, N20) and a thermal conductivity detector. NO was analyzed by infrared spectroscopy (Beckman Analyzer). The catalyst temperature was measured by a thermocouple placed just upstream of the bed. The experiments were conducted as follows: - the catalyst, previously reduced or not (100 mg), was charged to the quartz reactor with 400 mg of diluent (inactive a Al2O3), heated in a flow of nitrogen up to around 423K, then contacted with the reactant gases (201/h) and heated from 423 K to 773K at a programmed rate of 5K/mn.
669
- the activation at 773K under the reaction mixture was carried on during 3 hours. - the reaction temperature ranged from 773 to 423K (decreasing and increasing temperatures). At each temperature the conversion was measured at the steady value. The stoichiometry of the feedstream was characterized by the ratios: s = 2(02) + (NO)/(CO)= 0.93 or 1.07 or 2.2 (mixtures 0.75% CO + 0.1% NO + [ 0.30% or 0.35% or 0.75% 3 0 2 + N2) For the infrared spectroscopy the samples were pressed in order to obtain thin discs of known weight between 20 and 100 mg. The discs were placed in a sample holder made of quartz and introduced in an evacuable and heatable cell allowing to perform the various "in situ" treatments (reduction, evacuation, catalytic reaction...). Two kinds of experiments were performed: - the wafers were pretreated as described above in the reduction and chemisorption experiments and the IR spectrum of adsorbed CO at 298K on the reduced solids was recorded. - the infrared cell itself was used as the reactor in order to know the surface composition changes under the gaseous mixture. The wafer, initially reduced or not, was activated at 773K as described above for the catalytic reaction. The reactants were removed at 773 or 473K by a nitrogen flow and the solids were cooled down and evacuated at 298K. After recording the background spectrum, carbon monoxide was introduced at 298K under a pressure close to 15 Torr. The gaseous phase was evacuated at the same temperature. Infrared absorbance spectra were recorded at room temperature on a Fourier transform spectrometer (I.F.S. 110 from BRUKER). The plot function was set to absorbance. When overlapping the bands were graphically separated. The integrated absorbance for each individual CO band was related to the area under each band, determined by the gravimetric method and presented here in arbitrary unit..To compare the various solids the integrated absorbance was corrected taking account the weight of the pellet and the Pd content. RESULTS AND DISCUSSION
1 - DISPERSIONS MEASURED BEFORE ACTIVATION UNDER THE REACTION MIXTURE AND CATALYTIC ACTIVITIES.
The main characteristics of the Pd-supported catalysts are summarized in Table 1. When the particle size distribution is narrow enough (especially for the "fresh" catalysts) the mean particle sizes deduced from electron microscopy agree with those obtained from H2 chemisorption data. After high temperatures treatments in an oxidizing atmosphere ("aged" catalysts) the particle size distribution becomes broader and the dispersion deduced from
670 H 2 chemisorption clearly decreases. The B.E.T. area also decreases, especially
in presence of water when the temperature reaches 1273K. In this case the ( 6 ) A1203 support is partly transformed into a A1203 (X-Rays) as is well known [lo]. Figure I gives examples of the CO and NO conversions as a function of the reaction temperature for a "fresh" catalyst and for the corresponding "aged" catalyst. The maximum in NO conversion correspond to a maximum for the N20 production. The percentages of NO disappeared transformed into N 2 0 are respectively 37% and 51% for the "fresh " and "aged' catalysts.
A Conversion Figure I : CO and NO conversions as a function of the reaction temperature for the mixture 0.75% C0+0.35% 02+0.1% NO (S =1.07). Fresh catalyst PdJA1203 I (CO 473 K
573K
673K
,NOo)
Aged catalyst PdA1203 I' ( COANOA)
773K
Reactiar temperatun
TABLE 1
I
~~
Sample p r e c u r s o r
Pd(wt%) 1.o 0.85 0.8 1.0 0.85 0.85 0.85
Disp % (VH2)
D(TEM) (nm)
33 17 6
3.0-8.0 -6.0 >7.0 1.0-15.0 3.0-12.0 15.O- 100.0
. .
44
I
I
10 O\
S(BET) (m2/g)
1
I
120 112
I I
82 95 95 83
- Disp. (V H2): dispersion from H2 chemisorption after reduction. - D(TEM): range of the particle diameter (electron microscopy).
- S(BET): specific area.
- Pd/Al2O3 1': "aged" Pd/Al2O3 I solid (1173K, 10%H20, previous reduction at 723K) - Pd/Al2O3 11': "aged Pd/Al2O3 I1 solid (1073K, without previous reduction, without H20:
II'A, with H20: II'B) - PdJAl2O3 II": "aged PdJAl2O3 I1 solid (1273K, without previous reduction, without H20: II"A, with H20: 1I"B.
67 1
Figures 2 and 3 show the effect of the crystallite size on NO conversion and activity. It can be seen that the activities are only slightly affected by a reduction before activation but depend strongly upon the feed ratio CO/NO/02. The NO conversion declines as the feedstream becomes net oxidizing while the CO conversion is lowest at net reducing conditions, as already known [ll].
1
NO ~onvorrion%
100-
Figure 2 : NO conversion as afunction of the initial dispersion for the various catalysts, for three reaction temperatures, with the mixture s =I .07.
623 k 573 k
50523k
"fresh" catalysts"
A "agedt catalysts For the CO conversion the curves are similar
0.02
-
0.01
-
Figure 3 : Number of moIes of NO disappeared per hour and per m2 of Pd atoms present initially as a function of the initial dispersion for three reaction temperatures, with the mixture s =1.07. 023 k
573 k 523 k
10
I
20
30
50
40 initid
w
diaporsion
"fresh' catalysts /
A "aged" catalysts
%
The rates expressed as the numbers of moles of CO or NO disappeared per hour and per m2 of Pd atoms initially present increase strongly as the dispersion decreases to reach a maximum for around 510% of dispersion, as well for the CO oxidation than for the NO reduction (Figure 3). Such an apparent structure sensitivity has been reported for the propylene oxidation on Pt/Al2O3 [12] and more recently for methane oxidation over Pt and Pd on
672
alumina [13,14]. It should be noted that the initial state of the solids is not representative of the state after activation under the reactants. 2 - STATE^ OF THE PALLADIUM UNDER THE REACTION MIXTURE. 2-1 - Particle size. The activation under the reactant mixture induces a Pd particle sintering. The particle size distribution (electron microscopy) increases, the changes being more pronounced for the "initial" solid than for the "aged" ones. In general most of the particles lie in the range 10-15 nm except for the Pd/A1203 I solid. In that case two distributions are occurring : one around lnm and the other one around 6 nm. 2-2 - I.R. spectra of CO chemisorbed on the solids afteractivation under the reactives . The I.R. spectra of CO irreversibly adsorbed at 298K on the "fresh" and "aged" catalysts previously reduced are the spectra usually observed corresponding to CO/Pdo (v CO below 2100 cm-1). They present a high frequency band at 2080 (or 2070) cm-1 assigned to linearly bonded CO carbonyl complex and low frequency bands at 1970 and 1935 (shoulder) cm-1 assigned to multiply coordinated CO [15, 16, 171. Exposure to oxygen results in the disappearance of these bands corresponding to CO on metallic palladium. Thefigures 4, 5, 6 report some spectra of CO irreversibly adsorbed at 298K (10 minutes are long enough to evacuate the CO gaseous phase) on "fresh" and "aged" catalysts activated under the reactants and purged under N2. In addition to the bands corresponding to CO on PdO they show new bands at high wavenumbers (v CO above 2100 cm-1) related to CO adsorbed on Pdn+ ions [18, 191. This assignement was confirmed by the spectrum of CO adsorbed on Pd/A1203 wafers oxidized at 573K. It is clear that, after activation, ionic palladium is present at the surface of our catalysts. Such a result has already been mentioned during methane oxidation [13, 141 and neopentane hydrogenolysis [20]. From the present work we can conclude: - the redox state of the superficial Pd is not influenced by a cooling down under the reactants from 773 to 473K. However if the solids are contacted with the mixture s = 1.07 at 298K there is a further oxidation with an increase in intensity of the bands corresponding to CO on ionic Pd. the redox state of Pd is not widely influenced by a reduction before activation at 773K3,for a mixture s = 1.07 (solid Pd/A1203 11). The ionic sites are probably created under the mixture itself.
673
t
6
t
Figure 4 :I.R. spectra of CO irreversibly adsorbed at 298K on the PdIA1203 I1 catalyst (ex-HzPdC14) after activation at 773K under the reaction mixture and purge in a nitrogen flow at 773K (or 473K). B: mixture s = 2.2 A: mixture s = 1.07
- if the feedsteam is net-oxidizing, the intensity of the bands corresponding to CO on ionic Pd clearly increases (Figure 4 ). But even with a net-reducing feedstream and with a solid previously reduced, a small part of the surface palladium is oxidized (Figure 6 ). - after the aging treatments the intensity of the bands corresponding to CO on ionic Pd increases (Figure 5 ), whatever the solid considered. - for the fresh catalyst Pd/A1203 I (higher initial dispersion) the quantity of ionic palladium is higher than for the Pd/Al203 I1 catalyst. Two important remarks have also to be made: - for the Pd/Al2O3 I1 catalyst and the mixture s = 1.07 there is no appearance of gaseous C02 and of species chemisorbed on the support in presence of the probe molecule. But in others cases we have noted the formation of bands at 1650, 1430, 1225 cm-1. These bands are attributed to hydrogeno-carbonate groups resulting from the interaction of C 02 with the hydroxyl groups of alumina [21]. A part of the ionic palladium has been reduced by the probe molecule.
674
B
Figure 5 :I.R. spectra of CO irreversibly adsorbed at 298K on the PdIA1203 I and I' catalysts after activation at 773K under the reaction mixture 2.2 and purge under nitrogen. ~~
A: Pd/A1203 I: "fresh" catalyst ex-Pd(CSH702)2 B: P ~ A I I~': O ~ "aged" catalyst (1173K, 10%H20)
Figure 6: I.R. spectra of CO irreversibly adsorbed at 298K on the PdIA1203 I catalyst afer activation at 773K under the reaction mixture and purge under nitrogen. A: mixture S = 0.93 B: mixture S= 1.07
- as ionic palladium develops, there is a shift in the linear CO/Pdo (2070-2080 cm-l ) toward higher wavenumbers (2095 cm-1 ) as already observed during the adsorption of oxygen a fully CO-covered platinum surface [21]. It has been verified that this 2095 cm-1 peak is not due to ionic Pd. There will be a decrease in the bond strength in PdO-CO in presence of ionic Pd and with
675
increasing concentration of oxygen. Let us note that the Pd/A1203 I1 catalyst exhibits Pd2+ (2155-2150 cm-1 ) and Pd+ (2140-21 10 cm-1 ) [22,23] ions but that the Pd/A1203 I catalyst exhibit only Pd+ ions. - clearly the ratio of CO linearly-bonded to Pdo to CO multi-bonded to Pdo increases with the amount of Pdn. Here such a behaviour is not related to an increase in the metallic dispersion [24]. In fact the metallic Pd atoms become isolated one from another by the ionic palladium. 2-3 - Ouantitative evaluations of Pdm 2-3-1 Percentage of CO bonded to PdO and metallic surface after activation under the reactives and CO chemisorption From the integrated normalized absorbances it is possible to estimate the percentage of CO bonded to Pdo after activation under the reactives and CO chemisorption as well as the percentage of linear species on Pdo. These values are called respectively "%CO/Pdo" and "%lin/Pdo" (Table 2). A previous calibration [25] giving a linear relationship between the total integrated absorbance of the bands due to CO bonded to PdO and the metallic surface accessible to gases for completely reduced Pd/Al2O3 solids allows to calculate a metallic surface "S/Pdo" corresponding to "%CO/PdO". The previous values are excess values when there is appearance of HCO3- on the support i.e. when there is partial reduction of Pdn+ present after activation by the probe molecule itself. 2-3-2 Metallic surface formed by reduction of Pdn+ by the probe molecule. Considering the reaction CO gas + 0 adsorbed -+ CO2 and the formation of hydrogenocarbonate groups with the hydroxyl groups of alumina, and as there is no CO2 in the gas phase, it is possible to estimate, from the adsorption of C02 on the alumina support, via a calibration curve, the number of Pdo atoms formed during CO adsorption. The extra metallic surface due to the reduction of Pdn+ by the probe molecule is thus known. 2-3-3 The corrected metallic surface and the corrected percentage of CO bonded to Pdo obtained after activation under the reactants are thus calculated and designated as "S/Pdo(corr.)" and "%CO/Pdo (corr.) and reported in Table 2. The accuracy reaches 15%. The values thus obtained corroborate the conclusions arising from the qualitative considerations (section 2-2). Furthermore the conversions of CO and NO would reached maxima for a metallic surface of around 30-40 m2/g. The value thus obtained agrees with
676
the value deduced from hydrogen chemisorption after activation.
TABLE 2 -
S/PdO (corr.) m2/g 35 3 71 33
%
Solid
Mixture
lin/Pdo
I I' I1 I I' I1 II'A
1.07
47 30 11 71 100 34 87
I, I,
2.2 I,
It
% CO/Pdo % COPdO
79 70 77 78 68 51 53
(corr.) 46 56 77 33 44 28
.
S/PdO m2/g 54 4 71 44 4 90 26
2 .
77 13
CONCLUSIONS The initial state of Pd/A1203 catalysts is not representative of the state of the catalysts after activation under NO/C0/02 mixtures. An activation at 773 or 473K induces a palladium particle sintering and formation of ionic palladium on the surface of the catalysts. The quantities of CO adsorbed on Pdo and on Pdn+ are deduced from infrared spectroscopy and give a estimation of the percentage of PdO and Pdn+. They do not give the actual values since the extinction coefficients are unknown and probably not equal but they give proportional values. Since the redox state of Pd and the conversions are not influenced by a reduction or not before contacting with the reactants the catalytic sites (Pdn+ ions or Pdo atoms influenced by the presence of ionic palladium) are created under the reaction mixture. A relationship was found between concentration changes of the mixture, surface composition and catalytic activities. With increasing concentration of 0 2 the degree of oxidation of Pd increases and the bond strength between CO and Pdo decreases. The quantities of ionic palladium are also increasing after treatments at high temperatures in oxidizing atmospheres. The metallic atoms are thus diluted by the presence of the ionic Pd. It may be postulated that the active site is a Pdo atom modified by the vicinity of Pdn+. CO and NO conversions reach maxima for a given metallic surface and for a given Pdo/Pdn+ ratio. In the present state of our work the relationship between the number of ionic Pd atoms and the initial particle size is not clear.
677
ACKNOWLEDGEMENTS
This work was carried out within the "Groupement Scientifique Pots Catalytiques" funded by the "Centre National de la Recherche Scientifique", the "Institut Franqais du PCtrole" and the AFME (Agence Franqaise pour la Maitrise de 1'Energie).
REFERENCES 1. R. Prasad, L.A. Kennedy and E. Ruckenstein. Catal.Rev.Sci.Eng. 26, 1, (1984) 2. J. Wei, Adv. Catal., 57, (197 ) . 3 M. Shelef, Catal. Rev. Sci. Eng., 1_1 ,1, (1975) 4 K.C. Taylor in "Catalysis and Automotive Pollution Control", Studies in Surface Science and Catalysis (Ed. A. Crucq and A. Frennet, Elsevier) 3Q,97,.(1987) 5 L.L. Hegedus and J.J. Gumbleton, Chem. Techn., 630, (1980) 6 B.J. Cooper, W.D.J. Evans and B. Harrison, in "Catalysis and Automotive Pollution Control", Studies in Surface Science and Catalysis (Ed. A. Crucq and A. Frennet) Elsevier, 3,117,.(1987). 7 J.B. Boitiaux, J. Cosyns and S. Vasudevan, Appl. Catal., $41, (1983) 8 J.E Benson, H.S. Wang and M. Boudart, J. Catal. 2Q 146, (1973) 9 P.C. Aben, J Catal., 224, (1968) . 10 H. Schaper, E.B.M. Doesburg and L.L. Van Reijen, Appl.Catal., 7-21 1, (1983) 11 C.C Chang and L.L. Hegedus, J. Catal., 51,361, (1979) 12 L.M.Carballo and E.E.Wolf, J.Catai., 53,366, (1978). 13 R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal., 122,295, (1990) 14 R.F. Hicks, H. Qi, M.L. Young and R.G. Lee, J. Catal., 122,280, (1990) 15 A. Palazov, G. Kadinov, C. H. Bonev and D. Shopov, J. Catal., 44, (1982) 16 F.M Hoffman, Surf. Science Reports, 3,107, (1983). 17 M.A. Vannice, S.Y.Wang and S.H. Moon, J. Catal. 11,152, (1981) 18 A.L. Tarasov, V.A. Shvets, 396, (1989) and V.B. Kazanskii, Kinet.Katal., 19 Yu. A. Lokhov and A.A. Davydov 1515 and 1523, (1980) Kinet. Katal., 20 W. Jusczyk, Z. Karpinski, I. Ratajczykowa, Z. Stanasiuk, J. Zielinski, L.L. Sheu and W.M.H. Sachtler, J. Catal., 120.68, (1989). 21 M. Primet, J. Catal., 88, 273, (1984) 22 G.M. Alikina, A.A. Davydov, IS. Sazonova and V.V. Popovskii Kinet.Kata1. 418, (1987). 23 A. Palazov, G. Kadinov, Ch. Bonev, R. Dimitrova and D. Shopov, Surf. Sci., 225, 21, (1990). 24 L.L. Sheu, Z. Karpinksi and W.M.L. Sachtler J. Phys. Chem. 4890, (1989) 25 J.L. Duplan and H. Praliaud, Appl. Catal., in press.
a,
a,
This Page Intentionally Left Blank
679 AUTHOR INDEX
Anderson D.R.
275
Anderson S . Anderton D Aries L. Ball D.J. Beck D.D. Belot G. Bernhardt P. Bijsterbosch J. Bjornkvist L. Bowker M. Bradley S.O. Ciambelli P. Cohn M.J. Cucchi C. Cuttler D.H. D'Aniello M.J.,Jr Dathy C. de Soete G. de Veyrac P. Dekker N. Deworme E. Diwell A. F. Dondur V. Dozikre R. Duplan J.L. Duprez D. Durand D. Egeback K.E. Engler B.H. Fernandez A.
253 523 509 337 593 153,353 195 353 253 409 22 1 323 22 1 41 523 593 181 425 509 353 1 139,417 37 1 425 667 581,629 569 75 291,641 207
Fierain W. Frestad A. Fujitani Y. Garin F. Golunski S. E. Gonzalez-Elipe A. Grbic B. Guibet J.C. Gulati S.T. Guo Q. Hashimoto M. Hecq W.J. Heezen L. Henssler H. Hublin M. Jasper T.S. Jonkers G. Jovanovic D. Joyner R.W. Kacimi S. Kapteijn F. Kasemo B. Kilian V.N. Koberstein E. Krueger M.H. Kubsch J. Kuno K. Leclerc J.P. Leclercq G. Leclercq L. Lindner D.
115 253 617 153,195 417 207 371 93 48 1 409 657 5 38 1 35 41 523 239 37 1 409 581 353 253 381 291,641 593 125 557 465 181 181 64 1
680
Loof P. Lox E. Mabilon G. Maeda K. Maire G. Mantel M. Mason G. Masuda K. Matsumoto M. Matsumura R. Miloudi A. Mizukami F. Monroe D.R. Moulijn J. Munuera G. Muraki H. Murrell L. Nieuwenhuys B.E. Nunan J.G. Parella P. Percheron-Guegan Praliaud H. Prigent M. Pudney P.D.A. Rajaram R.R. Rannug U. Rieck J.S. Robinson K. Robota H.J. Roche R. Rutten F. Sadi F. Saitoh M.
253 291,641 181,569 557 153,195 509 75 557 657 657 629 557 593 353 207 617 275,547 381,395 22 1 323 629 667 195,425,569 409
139 75 125 523 22 1 153,353 395 629 657
557 55 437,465 Schweich D. 153 Serre C. 139 Shaw H.A. 617 Shinjoh H. 395 Siera J.. 167 Silver R.G. Spencer N.D. 125 337 Stack R.G. 105 Steel M.C.F. Stegenga S. 353 167 Summers J.C. Taschner K. 17 Tauster S.J. 275,547 Taylor J.R. 417 Terlecki-Baricevic 37 1 Traverse J.P. 509 Truex T.J. 139,417 Vaccaro S. 323 van Santen R.A. 239 van Silfhout R. 395 van Slooten R.F. 381 Vaneman G.L. 537 Villermaux J. 465 Vonkeman K.A. 239 Watanabe M. 557 Weibel M. 195 Westerholm R. 75 Williamson W.B. 167 Wolf R.M. 381 Yukawa F. 657
Sano T. Schmidt T.
68 1
STUDIES IN SURFACE SCIENCE AND CATALYSIS Advisory Editors: B. Delmon, Universith 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
Preparation of Catalysts I.Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the First International Symposium, Brussels, October 1417,1975 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 of the Chemical Processes in Relation to Practical Applications by V.V. Boldyrev, M. Bulens and B. Delmon Preparation of Catalysts II. 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 SocietB de Chimie Physique, Villeurbanne, September 24-28, 1979 edited by J. Bourdon Catalysis by Zeolites. Proceedings of an International Symposium, Ecully (Lyon), September 9-1 1, 1980 edited by B. Imelik, C. Naccache, Y. 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.F. Froment New Horizons in Catalysis. Proceedings of the 7th International Congress on Catalysis, Tokyo, June 30-July 4, 1980. Parts A and B edited by T. Seiyama and K. Tanabe Catalysis by Supported Complexes by Yu.1. Yermakov, B.N. Kuznetsov and V.A. Zakharov Physics of Solid Surfaces. Proceedings of a Symposium, Bechyiie, September 29-October 3, 1980 edited by M. UzniEka Adsorption at the Gas-Solid and Liquid-Solid Interface. Proceedings of an InternationalSymposium, Aix-en-Provence, September 2 1-23, 198 1 edited by J. Rouquerol and K.S.W. Sing Metal-Support and Metal-Additive Effects in Catalysis. Proceedings of an InternationalSymposium, Ecully (Lyon), September 14-1 6,1982 edited by 6.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. Jird and G. Schulz-Ekloff Adsorption on Metal Surfaces. An Integrated Approach edited by J. Benard Vibrations at Surfaces. Proceedings of the Third International Conference, Asilomar, CA, September 1-4, 1982 edited by C.R. Brundle and H. Morawitz
682 Volume 15 Volume 16
Volume 17
Volume 18
Volume 19
Volume 20
Volume 2 1
Volume 22 Volume 23 Volume 24
Volume 25
Volume 26
Volume 27 Volume 28
Volume 29 Volume 3 0
Volume 3 1
Volume 32 Volume 33 Volume 3 4
Heterogeneous Catalytic Reactions Involving Molecular Oxygen by G. I.Golodets Preparation of Catalysts 111. 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, LyonVilleurbanne, September 12-1 6, 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. JirS, 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, Portoroi-Portorose, September 3-8, 1984 edited by B. Driaj, 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 15-1 9, 1985 edited by D.A. King, N.V. Richardson and S. Holloway Catalytic Hydrogenation edited by L. Cervenq 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. Knozinger Catalysis and Automotive Pollution Control. Proceedings of the First InternationalSymposium, Brussels, September 8-1 1, 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 0. Delmon, P. Grange, P.A. Jacobs and G. Poncelet Thin Metal Films and Gas Chemisorption edited by P. Wissmann Synthesis of High-silica Aluminosilicate Zeolites 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
683 Volume 35 Volume 36
Volume 37
Volume 38
Volume 39
Volume 40
Volume 41
Volume 42 Volume 43 Volume 44
Volume 45 Volume 46
Volume 47 Volume 48
Volume 49 Volume 50
Volume 5 1 Volume 52
Volume 53
Keynotes in Energy-Related Catalysis edited by S. Kaliaguine 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 13-1 7, 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 7-1 1, 1987 edited by J. Koukal Heterogeneous Catalysis and Fine Chemicals. Proceedings of an International Symposium, Poitiers, March 15-1 7, 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. Pael Catalytic Processes under Unsteady-State Conditions by Yu. Sh. Matros Successful Design of Catalysts. Future Requirements and Development. Proceedings of the Worldwide Catalysis Seminars, July, 1988, on the Occasion of the 30th Anniversary of the Catalysis Society of Japan edited by T. lnui 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, Wurzburg, September 48,1988 edited by H.G. Karge and J. Weitkarnp Photochemistry on Solid Surfaces edited by M. Anpo and 1 .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 10-1 4, 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 AlChE 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 17-1 9, 1989 edited by J. Klinowski 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. Trimrn, S. Akashah, M. Absi-Halabi and A. Bishara
684 Future Opportunities in Catalytic and Separation Technology edited by M. Misono, Y. Moro-oka and S. Kimura New Developments in Selective Oxidation. Proceedings of an International Volume 5 5 Symposium, Rimini, Italy, September 18-22, 1989 edited by G. Centi and F. Trifiro Volume 5 6 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 5 7 A 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 5 8 Introduction t o Zeolite Science and Practice edited by H. van Bekkum, E.M. Flanigen and J.C. Jansen Volume 5 9 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. Perot, 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 6 1 Natural Gas Conversion. Proceedings of the Symposium on Natural Gas Conversion, Oslo, August 12-1 7, 1990 edited by A. Holmen, K.-J. Jens and S. Kolboe Volume 6 2 Characterization of Porous Solids II. Proceedings of the IUPAC Symposium (COPS 11). Alicante, May 6-9, 1990 edited by F. Rodriguez-Reinoso, J. Rouquerol, K.S.W. Sing and K.K. Unger Volume 63 Preparation of Catalysts V. Proceedings of the Fifth International Symposium on the Scientific Bases for the Preparation of Heterogeneous Catalysts, Louvain-laNeuve, September 3-6, 1990 edited by G. Poncelet, P.A. Jacobs, P. Grange and B. Delmon Volume 6 4 New Trends in CO Activation edited by L. Guczi Volume 6 5 Catalysis and Adsorption by Zeolites. Proceedings of ZEOCAT 90, Leipzig, August 20-23, 1990 edited by G. Ohlmann, H. Pfeifer and R. Fricke Volume 6 6 Dioxygen Activation and Homogeneous Catalytic Oxidation. Proceedings of the fourth International Symposium on Dioxygen Activation and Homogeneous Catalytic Oxidation, Balatonfured, September 10-14, 1990 edited by L.I. Simandi Volume 67 Structure-Activity and Selectivity Relationships in Heterogeneous Catalysis. Proceedings of the ACS Symposium on Structure-Activity Relationships in Heterogeneous Catalysis, Boston, MA, April 22-27, 1990 edited by R.K. Grasselli and A.W. Sleight Volume 6 8 Catalyst Deactivation 1991. Proceedings of the fifth International Symposium, Evanston, IL, June 24-26, 199 1 edited by C.H. Bartholomew and J.B. Butt Volume 69 Zeolite chemistry and Catalysis. Proceedings of an International Symposium, Prague, Czechoslovakia, September 8-1 3, 199 1 edited by P.A. Jacobs, N.I. Jaeger, L. Kubelkova and B. Wichterlova
Volume 5 4
685 Volume 7 0
Volume 7 1
Poisoning and Promotion in Catalysis based on Surface Science Concepts and Experiments by M. Kiskinova Catalysis and Automotive Pollution Control II. Proceedings of the 2nd International Symposium (CAPoC 2), Brussels, Belgium, September 10-1 3, 1990 edited by A. Crucq
This Page Intentionally Left Blank
