Studies in Surface Science and Catalysis 143
SCIENTIFIC BASES FOR THE PREPARATION OF HETEROGENEOUS CATALYSTS
This Page Intentionally Left Blank
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 Advisory Editors:
B. Delmonand J.T. Yates
Vol. 143
SCIENTIFIC BASES FOR THE PREPARATION OF
HETEROGENEOUS CATALYSTS
Proceedings of the 8t" International Symposium, Louvain-la-Neuve, Belgium,September 9 -12, 2002
Edited by E. G a i g n e a u x * , D.E. D e V o s * * , P. G r a n g e * , P.A. J a c o b s * * , J . A . M a r t e n s * * , P. R u i z * a n d G. P o n c e l e t *
* Universit# Catholique de Louvain, Louvain-la-Neuve, Belgium ** Katholieke Universiteit Leuven, Heverlee (Leuven), Belgium
2002 ELSEVIER Amsterdam - Boston - London - New Y o r k - Oxford - Paris - San Diego San Francisco - Singapore - Sydney - Tokyo
ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands
9 2002 Elsevier Science B.V. All rights reserved.
This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Science via their homepage (http://www.elsevier.com) by selecting 'Customer support' and then 'Permissions'. Alternatively you can send an e-mail to:
[email protected], or fax to: (+44) 1865 853333. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400, fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 207 631 5555; fax: (+44) 207 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Science Global Rights Department, at the fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.
First edition 2002 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for.
British Library Cataloguing in Publication Data A catalogue record from the British Library has been applied for.
ISBN: ISSN:
0444 511784 0167 2991
O The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). n__._
~ . _ ..1 - _
FOREWORD This volume of the Studies in Surface Science and Catalysis series contains the Proceedings of the Eighth International Symposium on Scientific Bases for the Preparation of Heterogeneous Catalysts, held on the campus of the "Universit6 catholique de Louvain" (UCL) in Louvain-la-Neuve, Belgium, on September 9-12, 2002. This symposium is jointly organized by the "Unit6 de catalyse et chimie des mat6riaux divis6s" of the "Universit6 catholique de Louvain", and the "Centrum voor oppervlaktechemie en katalyse" of the "Katholieke universiteit Leuven". The topic of this series of symposia, which was initiated in 1975 and organized at four-year intervals, has invariably been the discussion of the fundamentals behind the unit operations in the preparation of industrially relevant solid catalysts. Heterogeneous catalysis has always been a lively research field, although in the time period covered by these symposia, the emphasis of catalysis research has significantly shifted and spread from traditional applications in petroleum and bulk chemical production to, among others, the synthesis of fine chemicals, agro- and oleo-chemicals, pharmaceuticals and the environmental protection systems. This area, via the automotive applications, is presently dominating the catalyst market. The future for catalysis research seems bright, as it is now generally recognized as an essential element of sustainable development. In parallel with the broadening of the applications, the diversity of catalyst materials is booming, not only with respect to active elements and the sophistication of molecularly engineered active sites, but also with respect to the never ending discovery of new approaches for structuring porous matrices at the different relevant length scales, including the nanometer, the micrometer as well as the millimeter scale. Catalyst preparation techniques and physico-chemical catalyst characterisation tools likewise are subject to continuous developments. The organisers of these symposia, assisted by a Scientific Committee composed of eminent researchers holding an industrial appointment, try to keep trace of the relevant developments in the field and solicit contributions dealing with preparation aspects of relevant new generations of catalyst materials. In the organisation of this Eighth Symposium, the Scientific Committee was faced with the difficult task to select 139 papers (37 oral communications and 102 posters) out of the more than 200 submitted abstracts. When hesitating among excellent contributions, the decisive criterion was always the catalyst preparation aspect of the work. The organisers hope to be
able to offer a valuable platform for discussion of the science behind catalyst preparation. The attendance of over 250 suggests that the effort is appreciated by the scientific community. The organisers are indebted to public and private sponsors without whom the organisation of this symposium would have been financially very difficult. For obvious reasons, the sponsoring Companies and Agencies cannot be acknowledged properly by citing them in the Proceedings. The same holds true for all those who have contributed to the preparation of the meeting, secretaries, technicians, students and postdocs. Fortunately, the organisers are in position to express their appreciation towards the Rector of the UCL, Professor M. Crochet, and the "Service des Auditoires", for allowing the event to be patronized again by the university. Our grateful acknowledgements also go specifically to Ms Marianne Saenen, who has decisively contributed to the success of the symposium. Thankyou very much Marianne I
The Editors
Georges Poncelet has been a driving force of this Symposia series. Given his retirement next year, the moment has come to express our appreciation and gratefulness for his devotion and dedication to these meetings. Georges, it is with a warm heart that we dedicate this volume to you.
Eric
Dirk
Johan
Patricio
Paul
P~rre
vii ORGANIZING COMMITTEE Dr. D. DE VOS, Katholieke Universiteit Leuven Dr. E.M. GAIGNEAUX, Universit6 catholique de Louvain Prof. P. GRANGE, Universit6 catholique de Louvain Prof. P.A. JACOBS, Katholieke Universiteit Leuven Prof. J. MARTENS, Katholieke Universiteit Leuven Dr. G. PONCELET, Universit6 catholique de Louvain Dr. P. RUIZ, Universit6 catholique de Louvain SCIENTIFIC COMMITTEE Dr. A. ANUNDSKAS, Norsk Hydro, Norway Dr. M.P. ATKINS, BP Amoco, United Kingdom Dr. G. BELLUSSI, EniTecnologie, Italy Dr. J.-L. BOUSQUET, TotalFinaElf, France Dr. J.A. DELGADO, Repsol, Spain Dr. D. DE VOS, KUL, Belgium Dr. E.M. GAIGNEAUX, UCL, Belgium Prof. P. GRANGE, UCL, Belgium Dr. J. GROOTJANS, Atofina Research, Belgium Dr. K. HARTH, BASF, Germany Dr. G. HECQUET, Atofina, France Dr. S.D. JACKSON, Synetix, United Kingdom Prof. P.A. JACOBS, KUL, Belgium Dr. K. JOHANSEN, Haldor Topsoe, Denmark Dr. S. KASZTELAN, Institut Franqais du P6trole, France Dr. J.-P. LANGE, Shell International, The Netherlands Prof. J.A. MARTENS, KUL, Belgium Dr. L.R. MARTENS, Exxon Mobil, Belgium Dr. R. PARTON, DSM Research, The Netherlands Dr. M.A. PEREZ, CEPSA, Spain Dr. G. PONCELET, UCL, Belgium Dr. P. RUIZ, UCL, Belgium Dr. F. SCHMIDT, Sfid Chemie, Germany Dr. J.-P. SCHOEBRECHTS, Solvay, Belgium Dr. M. SCHOONOVER, UOP, USA Dr. C. STOCKER, Sumitomo Deutschland, Germany Dr. M. TWIGG, Johnson Matthey, United Kingdom
This Page Intentionally Left Blank
ix
Contents Foreword Aspects of scale-up of catalyst production Keld Johansen
Quantitative structure-activity relationships in zeolite-based catalysts: influence of framework structure J.L. Casci and M.D. Shannon
17
Cogelation: an effective sol-gel method to produce sinter-proof finely dispersed metal catalysts supported on highly porous oxides B. Heinrichs, S. Lambert, C. Alig, J.P. Pirard, G. Beketov, V. Nehasil and N. Kruse
25
Steam reforming of CH4 over Ni/Mg-A1 catalyst prepared by spc-method from hydrotalcite T. Shishido and K. Takehira
35
Toward a molecular understanding of noble metal catalyst impregnation J.R. Regalbuto, M. Schrier, X. Hao, W.A. Spieker, J.G. Kim, J.T. Miller and A.J. Kropf
45
Support modification of cobalt based slurry phase Fischer-Tropsch catalysts S. Barradas, E.A. Caricato, P.J. van Berge and J. van de Loosdrecht
55
The effects of nature and pretreatment of surface alumina support on the catalytic nickelsilicate membrane formation C. Constantin, V. Parvulescu, A. Bujor, G. Popescu and B.L. Su
67
Supports and catalysts preparation by using metal alkoxides grafting technique E. Santacesaria, A. Sorrentino, M. Di Serio and R. Tesser
77
Combinatorial approaches for speeding up heterogeneous catalyst discovery, optimisation and scaling-up C. Mirodatos
89
High surface area metal oxides from matrix assisted preparation in activated carbons M. Schwickardi, T. Johann, W. Schmidt, O. Busch and E Schiith
93
Effects of the impregnating and drying process factors on mechanical properties of a PCoMo/A1203 hydrotreating catalyst Dongfang Wu and Yongdan Li
101
Influence of CeO2 content on R h / Y i O 2 monolithic catalysts for N20 decomposition S. Suarez, M. Yates, F.J. Gil Llambfas, J.A. Martfn, P. Avila and J. Blanco
111
Pt combustion catalysts prepared from W/O microemulsions J. Rymes, G. Ehret, L. Hilaire and K. Jir6tov6
121
Preparation of stable catalysts for N20 decomposition under industrial conditions S. Alini, E Basile, A. Bologna, T. Montanari and A. Vaccari
131
The Anderson-type heteropolyanions in the synthesis of alumina- and zeolitesupported HDS oxidic precursors E. Payen, G. Plazenet, C. Martin, C. Lamonier, J. Lynch and V. Harl~
141
Sol-gel preparation of pure and silica-dispersed vanadium and niobium catalysts active in oxidative dehydrogenation of propane P. Moggi, G. Predieri, D. Cauzzi, M. Devillers, P. Ruiz, S. Morselli and O. Ligabue
149
Preparation of nickel-modified ceramic filters by the urea precipitation method for tar removal from biomass gasification gas D.J. Draelants, Y. Zhang, H. Zhao and G. V. Baron
159
Preparation of gold-titanosilicate catalysts for vapor-phase propylene epoxidation using H2 and 02 A.K. Sinha, S. Seelan, S. Tsubota and M. Haruta
167
Sol-gel synthesis of colloids and triflates containing hybrid type catalysts A.N. Parvulescu, B.C. Gagea, M. Alifanti, V. Parvulescu and V.L Parvulescu
177
Preparation of zeogrids through interposed stapling and fusion of MFI zeolite type nanoslabs S.P.B. Kremer, C.E.A. Kirschhock, M. Tielen, E Collignon, P.J. Grobet, P.A. Jacobs and J.A. Martens
185
Large scale synthesis of carbon nanofibers by catalytic decomposition of hydrocarbon L. Pesant, G. Wine, R. Vieira, P. Leroi, N. Keller, C. Pham-Huu and M.J. Ledoux
193
Synthesis and characterization of carbon nanofiber supported ruthenium catalysts M.L. Toebes, E E Prinsloo, J.H. Bitter, A.J. van Dillen and K.P. de Jong
201
Synthesis of high pore volume and specific surface area mesoporous alumina L. Sicard, B. Lebeau, J. Patarin and E Kolenda
209
xi Investigation on acidity of zeolites bound with silica and alumina X. Wu, A. Alkhawaldeh and R.G. Anthony
217
Preparation of BN catalyst supports from molecular precursors. Influence of the precursor on the properties of the BN ceramic J.A. Perdigon-Melon, A. Auroux, J.M. Guil and B. Bonnetot
227
Monitoring of the particle size of MoSx nanoparticles by a new microemulsionbased synthesis K. Marchand, M. Tarret, L. Normand, S. Kasztelan and T. Cseri
239
Transition metal phosphides. Novel hydrodenitrogenation catalysts V. Zuzaniuk, R. Prins, C. Stinner and T. Weber
247
The application of non-hydrothermally prepared stevensites as support for hydrodesulfurization catalysts M. Sychev, R. Prihod'ko, A. Koryabkina, E.J.M. Hensen, J.A.R. van Veen and R.A. van Santen
257
NiMo/HNazY(s)-A1203 catalysts for the hydrodesulfurization of hindered dibenzothiophenes: effect of the preparation method T. Klimova, D. Solis, J. Ramirez and A. L6pez-Agudo
267
Chiral dirhodium catalysts confined in porous hosts H.M. Hultman, M. de Lang, M. Nowotny, I. W. C.E. Arends, U. Hanefeld, R.A. Sheldon and T. Maschmeyer
277
Synthesis and characterization of zeolite encaged enzyme-mimetic copper histidine complexes J.G. Mesu, H.J. Tromp, D. Baute, E.E. van Faassen and B.M. Weckhuysen
287
Strategies for the heterogenization of rhodium complexes on activated carbon J.A. Diaz-Au~ton, L.C. Romdn-Mart[nez, C. Salinas-Mart[nez de Lecea and H. Alper
295
Heterogeneous metathesis initiators M. Mayr, B. Mayr and M.R. Buchmeiser
305
Preparation of physically heterogeneous and chemically homogeneous catalysts on the base of metal complexes immobilized in polymer gels A.A. Efendiev, T.N. Shakhtakhtinsky and N.A. Zeinalov
313
Hydrocracking catalyst to produce high quality diesel fraction R. Galiasso Tailleur
321
xii Thermostable yttria-doped inorganic oxide catalyst supports for high temperature reactions E. Elaloui, R. Begag, B. Pommier and G.M. Pajonk
331
Preparation and characterization of WOx-CeO2 catalysts M. Alifanti, C.M. Visinescu, V.1. Parvulescu, P. Grange and G. Poncelet
337
Preparation of iridium catalysts by deposition precipitation: room temperature oxidation of CO M. Okumura, E. Konishi, S. Ichikawa and T. Akita
345
New approach to preparation and investigation of active sites in sulfated zirconia catalysts for skeletal isomerization of alcanes N.A. Pakhomov, A.S. Ivanova, A.E Bedilo, E.M. Moroz and A.M. Volodin
353
Supported ruthenium carbido-cluster catalysts for the catalytic removal of nitrogen monoxide and sulfur dioxide: the preparation process monitored by sulfur K-edge X-ray absorption near-edge structure Y. Izumi, T. Minato, K.-L Aika, A. Ishiguro, T. Nakajima and Y. Wakatsuki
361
Catalytic transformation of dichloromethane over Y and X zeolites L. Pinard, J. Mijoin, R. Lapeyrolerie, P. Magnoux and M. Guisnet
369
Preparation of new solid super-acid catalyst, titanium sulfate supported on zirconia and its acid catalytic properties J.R. Sohn, E.H. Park and J.G. Kim
377
Superacid WOx/ZrO2 catalysts for isomerization of n-hexane and for nitration of benzene V.V. Brei, O.V. Melezhyk, S.V. Prudius, M.M. Levchuk and K.I. Patryliak
387
Preparation of copper-oxide catalyst systems for hydrogenation Y. Sakata, N. Kouda, Y. Sakata and H. Imamura
397
Application of experimental design for NOx reduction by Pd-Cu catalysts M. Rebollar, M. Yates and M.A. Valenzuela
407
Marked difference of catalytic behavior by preparation methods in CH4 reforming with CO2 over Mo2C and WC catalysts S. Naito, M. Tsuji, Y. Sakamoto and T. Miyao
415
Synthesis and properties of new catalytic systems based on zirconium dioxide and pentasils for process of NOx selective catalytic reduction by hydrocarbons V.L. Struzhko, S.N. Orlyk, T.V.Myroniuk and V.G. Ilyin
425
xiii Preparation of chitosan based catalysts for several reactions of liquid phase hydrogenation V. Isaeva, A. Ivanov, L. Kozlova and V. Sharf
435
Preparation of Mo/A1203 sulfide catalysts modified by Ir nanoparticles J. Cinibulk and Z. Vit
443
Peptization mechanisms of boehmite used as precursors for catalysts D. Fauchadour, E Kolenda, L. Rouleau, L. Barr~ and L. Normand
453
Influence of the treatment of Y zeolite by ammonium hexafluorosilicate on physicochemical and catalytic properties: application for chlororganics destruction R. Lopez-Fonseca, J.L Guti~rrez-Ortiz, B. de Rivas, S. Cibrian and J.R. Gonz61ez-Velasco
463
Preparation of SiO2 modified SnO2 and ZrO2 with novel thermal stability Y.-Z Zhu, J.-Y. Wei, L. Zeng, X.-D. Zhao, W. Lin and Y.-C. Xie
471
Control of the textural properties of cesium 12-molybdophosphate-based supports S. Paul V. Dubromez, L. Zair, M. Fournier and D. Vanhove
481
MnOx/CeO2-ZrO2 and MnOx/WO3-TiO2 catalysts for the total oxidation of methane and chlorinated hydrocarbons E. Kantzer, D. D6bber, D. Kiessling and G. Wendt
489
Catalytic behaviour of Rh-supported catalysts on lamellar and zeolitic structures by anchoring of organometallic compound C. Blanco, R. Ruiz, C. Pesquera and E Gonzalez
499
The use of sol-gel technique to prepare the TiO2-A1203 binary system over a wide range of Ti-A1 ratios A.Yu. Stakheev, G.N. Baeva, N.S. Telegina, L V. Mishin, T.R. Brueva, G.L Kapustin and L.M. Kustov
509
Catalytic performance in the complete acetone oxidation of manganese and cobalt oxides supported on alumina and silica A. Gil, S.A. Korili, M.A. Vicente and L.M. Gandia
517
Unsupported and supported manganese oxides used in the catalytic combustion of methyl-ethyl-ketone L.M. Gandia, S.A. Korili and A. Gil
527
Ni/Hfl zeolite catalysts prepared by the deposition-precipitation method R. Nares, J. Ramirez, A. Guttierrez-Alejandre, R. Cuevas, C. Louis and T. Klimova
537
xiv Sol-gel A1203 structure modification by Ti and Zr addition. A NMR study J. Escobar, J.A. de Los Reyes and T. Viveros
547
Promotion of Ru/ZrO2 catalysts by platinum A.M. Serrano-S6nchez, F. Blas-Sudrez, P. Steltenpohl, M.P. Gonzdlez-Marcos, J.A. Gonzdlez-Marcos, and J.R. Gonzdlez-Velasco
555
Catalysts based on RhMo6 heteropolymetalates. Bulk and supported preparation and characterisation C.I. Cabello, I.L. Botto, M. Mu~oz and H.J. Thomas
565
Metallosilicate mesoporous catalysts prepared by incorporation of transition metals in the MCM-41 molecular sieves and their catalytic activity in selective oxidation of aromatics (styrene and benzene) V. Parvulescu and B.L. Su
575
Controlled surface modification of alumina-supported Mo or Co-Mo sulfides by surface organometallic chemistry J.-S. Choi, C. Petit-Clair and D. Uzio
585
Novel one step synthesis of cobalt (II) phtalocyanine-hydrotalcite catalysts for mercaptan oxidation in light oil sweetening I. Chatti, A. Ghorbel and J.M. Colin
595
Structural and catalytic properties of Zr-Ce-Pr-O xerogels S. Rossignol, C. Descorme, C. Kappenstein and D. Duprez
601
Influence of the precursor (nature and amount) on the morphology of MoO3 crystallites supported on silica D. Navez, C. Weinberg, G. Mestl, P. Ruiz and E.M. Gaigneaux
609
Single step synthesis of metal catalysts supported on porous carbon with controlled texture N. Job, E Ferauche, R. Pirard and J.P. Pirard
619
A g - S i Q and Cu/SiO2 cogelled xerogel catalysts for benzene combustion and 2-butanol dehydrogenation S. Lambert, N. Tcherkassova, C. Cellier, E Ferauche, B. Heinrichs, P. Grange and J.P. Pirard
627
Preparation of zeolite catalysts for dehydrogenation and isomerization of n-butane M. Inaba, K. Murata, M. Saito, I. Takahara, N. Mimura, H. Hamada and Y. Kurata
637
XV
The application of well-dispersed nickel nanoparticles inside the mesopores of MCM-41 by use of a nickel citrate chelate as precursor D.J. Lensveld, J.G. Mesu, A.J. van Dillen and K.P. de Jong
647
Preparation of Ce-Zr-O composites by a polymerized complex method T.G. Kuznetsova, V.A. Sadykov, E.M. Moroz, S.N. Trukhan, E.A. Paukshtis, V.N. Kolomiichuk, E.B. Burgina, V.L Zaikovskii, M.A. Fedotov, V.V. Lunin and E. Kemnitz
659
Sol-gel routes for the preparation of heterogeneous catalyst based on Ru, Rh, Pd supported metals P. Moggi, S. Morselli and G. Predieri
669
Development of novel heterogeneous catalysts for oxidative reactions: preparation and performance of Co-Nx catalysts in partial oxidation of toluene and n-butane M.L. Kaliya, S.B. Kogan and M. Herskowitz
679
Synthesis and modification of basic mesoporous materials for the selective etherification of glycerol J.-M. Clacens, Y. Pouilloux and J. Barrault
687
Carbon nanotubes: a highly selective support for the C=C bond hydrogenation reaction J.-P. Tessonier, L. Pesant, C. Pham-Huu, G. Ehret and M.J. Ledoux
697
Raman studies of the templated synthesis of zeolites P.P.H.J.M. Knops-Gerrits and M. Cuypers
705
Templateless synthesis of catalysts with narrow mesoporous distribution N. Yao, G. Xiong, S. Sheng, M. He and K.L. Yeung
715
Control of pore structures of titanias and titania/aluminas using complexing agents M. Toba, S. Niwa, N. Kijima and Y. Yoshimura
723
Tungstophosphoric acid immobilized in polyvinyl alcohol hydrogel beads as heterogeneous catalyst L.R. Pizzio, C. C6ceres and M.N. Blanco
731
Functionalized SiMCM-41 as support for heteropolyacid based catalysts L.R. Pizzio, A. Kikot, E. Basaldella, P. Vazquez, C. Cdceres and M.N. Blanco
739
Influence of the preparation method on the surface properties and activity of alumina-supported gallium oxide catalysts A. Petre, B. Bonnetot, A. Gervasini and A. Auroux
747
xvi Preparation and properties of bimetallic Ru-Sn sol-gel catalysts: the influence of catalyst reduction J. Hajek, N. Kumar, H. Karhu, L. Cerveny, J. Vayrynen, T. Salmi and D. Yu. Murzin
757
A new insight into molybdate/boehmite interaction D. Minoux, F. Diehl, P Euzen, J.-P Jolivet and E. Payen
767
Controlled coating of high surface area silica with titania overlayers by atomic layer deposition J. Keriinen, E. Iiskola, C. Guimon, A. Auroux and L. Niinist6
777
Concept of the synthesis of novel platinum catalysts for selective hydrogenation of unsaturated carbonyl compounds J. Kijenski and P. Winiarek
787
Storage and supply of hydrogen mediated by iron oxide: modification of iron oxides S. Takenaka, C. Yamada, T. Kaburagi and K. Otsuka
795
Catalytic activity of bulk and supported sulfated zirconia I.J. Dijs, L.W. Jenneskens and J.W. Geus
803
New one-step synthesis of superacid sulfated zirconia L. Zanibelli, A. Carati, C. Flego and R. Millini
813
Elaboration and characterization of a realistic Phillips model catalyst for ethylene polymerisation P.G. Di Croce, E Aubriet, P. Bertrand, P. Rouxhet and P. Grange
823
Preparation of new basic mesoporous silica catalysts by ammonia grafting H. Yoshida, Y. Inaki, Y. Kajita, K. Ito and T. Hattori
837
Titania-silica catalysts prepared by sol-gel method for photoepoxidation of propene with molecular oxygen C. Murata, H. Yoshida and T. Hattori
845
Preparation of large surface area MnOx-ZrO2 for sorptive NOx removal M. Machida, M. Uto and T. Kijima
855
Preparation of CuOx-TiO2 nano-composite photocatalysts from intercalated layer structure M. Machida, S. Nagasaki and T. Kijima
863
Vanadia-doped titanium pillared clay: preparation, characterization and SCR activity of NO by ammonia L. Khalfallah Boudali, A. Ghorbel, P. Grange and S.M. Jung
873
xvii Advanced preparation by sol-gel method of the encapsulated Pd/A1203 catalysts for methane combustion S. Fessi, A. Ghorbel, A. Rives and R. Hubaut
881
Non-ionic surfactant templated synthesis of mesoporous silica in the presence of platinum salts M.A. Aramendfa, V. Borau, C. Jimdnez, J.M. Marinas, EJ. Romero and EJ. Urbano
891
Synthesis and acid-base properties of catalysts based on magnesium and sodium-magnesium mixed phosphates M.A. Aramendfa, V. Borau, C. Jimdnez, J.M. Marinas, R. Rolddn, EJ. Romero and EJ. Urbano
899
Preparation of Pd-Ce/ZrO2 catalysts for methane oxidation L.S. Escand6n, S. Ord6ftez, E V. Dfez and H. Sastre
907
The effect of cerium introduction on vanadium-USY catalysts C. Ramos Moreira, M. Schmal and M.M. Pereira
915
Rh-Co mordenite catalysts for the selective reduction of NO by methane C.E. Quincoces, M. Incolla, A. De Ambrosio and M.G. Gonzdlez
925
Surface characterization of WO3-TiO2/A1203 catalysts and reactivity on selective catalytic reaction of NO by NH3 S. Egues, N.S. de Resende and M. Schmal
933
Catalytic materials for the synthesis of hydrofluorocarbons P. Cuzzato, V. Giammetta, R. Trabace and E Trifiro
941
Preparation, characterization and reactivity in m-cresol methylation of new heterogeneous materials having basic properties E Cavani, C. Felloni, D. Scagliarini, A. Tubertini, C. Flego and C. Perego
953
The effect of glycols in the organic preparation of V/P mixed oxide catalyst for the oxidation of n-butane to maleic anhydride S. Albonetti, E Cavani, S. Ligi, E Pierelli, E Trifiro, E Ghelfi and G. Mazzoni
963
Synthesis and characterization of nanostructured M02C on carbon material by carbothermal hydrogen reduction C. Liang, Z Wei, Q. Xin and C. Li
975
Carbon composite-based catalysts: new perspectives for the low-temperature H2S removal J.-M. Nhut, R. ~eira, N. Keller, C. Pham-Huu, W. Boll and M.J. Ledoux
983
xviii Active carbon surface oxidation to optimize the support functionality and metallic dispersion of a Pd/C catalyst V. Dubois, Y. Dal and G. Jannes
993
n-Butane isomerization over Al-promoted sulfated zirconias. Influence of the sulfate content J.A. Moreno and G. Poncelet
1003
Influence of preparation procedure on physical and catalytic properties of carbon-supported Pd-Au catalysts P. Canton, E Menegazzo, M. Signoretto, E Pinna, P. Riello, A. Bedetti and N. Pernicone
1011
Preparation of mesoporous highly dispersed Pd-Pt catalysts for deep hydrodesulfurization X. Xu, P. Waller, E. Crezee, Z. Shan, E Kapteijn and J.A. Moulijn
1019
Preparation of highly ordered CMI-1 and wormhole-like DWM mesoporous silica catalyst supports using C16(EO)10 as surfactant A. L~onard, J.L. Blin and B.L. Su
1027
Synthesis and characteriation of nanostructured mesoporous zirconia catalyst supports using non-ionic surfactants as templating agents J.L. Blin, L. Gigot, A. L~onard and B.L. Su
1035
Effect of preparation parameters on the catalytic activities of sulfated ZrO2SiO2 catalysts obtained by sol-gel process R. Akkari and A. Ghorbel
1045
Non-aggressive way for preparation of zirconium sulfate pillared clay using zirconium acetate developing high sulfur thermal stability over 830~ S. Ben Chaabene, L. Bergaoui, A. Ghorbel and J.E Lambert
1053
Influence of the preparation conditions on the structure of the active phase and catalytic properties of Ni-Co-molybdate propane oxydehydrogenation catalysts M.M. Barsan, A. Maione and E C. Thyrion
1063
Oxidized diamond as a new catalyst support T. Suzuki, K. Nakagawa, N.-O. Ikenaga and T. Ando
1073
Preparation of catalytic membranes, micro-capsules and fabrics active in immobilized Fenton chemistry J. Fernandez, V. Nadtochenko, A. Bozzi, T. Yuranova and J. Kiwi
1081
Preparation of vanadium-based catalysts for selective catalytic reduction of nitrogen oxides using titania supports chemically modified with organosilanes H. Kominami, M. Itonaga, A. Shinonaga, K. Kagawa, S. Konishi and Y. Kera
1089
xix Design, preparation and testing of effective FeOx/SiO2 catalysts in methane to formaldehyde selective oxidation E Arena, F. Frusteri, L. Spadaro, A. Venuto and A. Parmaliana
1097
New Fe-Mo-Ti mixed oxides prepared via the sol-gel method: comparison of the textural properties with solids obtained by impregnation S.R.G. Carrazan, C. Martin, C. M. Pedrero and J. Saunders
1107
Index
1115
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
Aspects of scale-up of catalyst production Keld Johansen Research & Development Division, Haldor Topsoe A/S Nymollevej 55, DK-2800 Lyngby, Denmark, Tel: +45-45272491, E-mail:
[email protected] 1. S C O P E OF C A T A L Y S T P R O D U C T I O N Catalyst production has a significant influence on the economy, since 80-90% of the chemicals used in a modern society are exposed to a catalyst. The value of the US catalyst market was 2.2 billion $ in 2000 [1]. A number of specialised catalyst companies in the world, many global, produce and supply a large number of products to the industry. It is said that products corresponding to 10% of the GNP of the industrialised countries are dependent on the availability of catalyst. Catalyst plants produce quantities of 0.1100 t/day of each product dependant on the type. Thus, transfer of new products and implementation in catalyst plants are important technological disciplines for each catalyst company. 2. N A T U R E OF T H E SCALE-UP P R O B L E M Before the decision is taken on transferring a new recipe from research and development departments to an existing catalyst plant or a new investment, considerable work has already been carried out - many test samples have been prepared and activitytested. All candidates for a new product have one thing in common: in the starting phase they have been prepared and selected from the laboratory processes.
2.1. Description of laboratory prepared samples Laboratory or bench scale prepared catalyst samples for screening are typically made in gram scale (10-50 g). The catalysts can be prepared in many ways depending on the type, but steps for three commonly used preparation routes are shown below in Fig. 1 and, for each step, examples of the typical laboratory equipment are given.
Io Typical co-precipitated catalyst manufacture Preparation step Typical equipment 1. Dissolution of agents Beaker with stirrer 2. Precipitation Pump + beaker with stirrer 3. Ageing Electric heating, thermostatic bath 4. Filtration Buchner funnel 5. Washing Demineralised water on Buchner funnel 6. Drying Drying cabinet 7. Calcination Muffle furnace 8. Lubricant aid addition Powder mixer Single station excenter press 9. Tabletting Electrically heated muffle furnace 10. Calcination Small reactor with H/N2 once through 11. Activation
II. Typical impregnated catalyst carrier process Typical equipment Preparation step 1. Forming a support (from another route) Heated muffle furnace 2. Activation of support Beaker with stirrer (pure chemicals) 3. Dissolution of impregnation liquid(s) Net in a beaker 4. Impregnation Drying cabinet 5. Drying Muffle furnace 6. Decomposition 7. Re-impregnation back to 4. Reactor sulphidation with H2/N2 once 8. Activation through III. Typical process for mixed/compounded catalyst Typical equipment Preparation step From flasks 1. Powders Beaker with stirrer 2. Dissolution of active metals Beaker with stirrer 3. Dissolution of extrusion aids Laboratory kneader 4. Mixing Laboratory piston extruder 5. Extrusion Drying cabinet 6. Drying Muffle furnace 7. Calcination/decomposing Laboratory sieve 8. Sieving Small reactor with H2/N2 9. Activation through Fig. 1. Three examples of commonly used preparation methods
etc. once
The laboratory equipment used is normally characterised by: Small dimensions with short mass and heat transport distances - High energy intensity per volume for stirrer mixers and kneaders - Pure chemicals Small layers in muffle furnaces with low temperature gradients, but long hea cycle H2 activation with low pH20 Drying with low but undefined pH20 Precipitation within small volume dimension Filtration without respect of agglomerate size Washing without respect of time or leakage of particles or ions - Tabletting/forming of non-representative granules - Generated heat during processes easily dissipated to cooling surface -
-
-
-
-
-
-
A modern catalyst laboratory will analyse and describe in detail all intermediates final catalysts from the above three manufacturing routes by means of the metho&, follows: main chemical elements and trace elements, phases (if not amorphous), 1: distribution and BET or selective surface area. A more complete list of commonly u methods is given in Table 1 below: Table 1. Physical and chemical characterisation methods C h e m i c a l
Main chemical elements Trace elements Oxidation state Element distribution
ICP, AAS, XRF and electron micro probe analysis AAS, ICP Electron micro probe analysis, EDS in SEM, ED5 in TEM
P h y s i c a l
Surface area Pore volume- total Pore size distribution Phases, crystallite size Surface composition Surface properties Particle size distribution Structure and texture Thermogravimetry Specially for finished product form Abrasion resistance for granules Attrition resistance for fluid cat and powders Crushing strength
BET (N2, Ar, K), or specific area, chemisorption of H2,CO, N20 Water absorption-Hg intrusion He, or Hg intrusion, N2 adsorption XRD, TEM, Raman Spectroscopy XPS, SIMS IR, microcalorimetry, chemisorption/desorption (TPD,TPR) LLD, SAXS, sieves, TEM, SEM TEM, SEM, optical microscopy TGA, DTA, dilatometer
Attrition loss Texture analyser
However, all the above methods in Table 1 cannot give a scientifically exhaustive description of the intermediates nor of the final catalyst. Amorphous phases often obtained from precipitation cannot be characterised sufficiently (how many kinds of amorphous phases exist?) Furthermore, the final catalyst granule is formed from agglomerates of crystallites. Both crystallites (forming primary agglomerates) and the agglomerates have their own particle size distribution and binding properties. Particle size distribution of primary and secondary agglomerates controls the final pore size distribution. The pore size distribution and particle strength will have significant influence on the final performance of the product in the reactor. Even considering the methods listed in Table 1, there is no method or combined methods today that can give a full description of the crystalliteagglomerate multi-parameter system. To further illustrate the problem, it should be mentioned that even if the overall chemical composition is the same for two different manufacturing routes, the pore size distribution is most probably different. Thus, it is not possible to characterise an intermediate or final catalyst so you can be sure to have the same catalyst without preparing it in the same reproducible way.
2.2 Catalyst manufacturing- unit operations The catalyst plant is operating in ton scale (typically 1-100 t/day) with processes and equipment completely different from bench scale as sketched in Fig. 1, even if the preparation steps are the same. In the open literature, description of catalyst manufacturing processes and equipment is sparse. The reference list contains important monographs and papers [2-24]. The patent literature gives some information, but catalyst manufacturing technologies are often not patented but kept secret. The single step in manufacturing is called a unit operation and can be performed by several types of equipment. Table 2 shows most of the unit operations used and examples of equipment for each. Most of the typical equipment will have more different time constants, heat transfer, flow patterns, temperature profiles etc. than bench scale equipment and, therefore, the final catalyst will achieve other properties.
2.3 Optimal combinations For every commercial catalyst an optimal combination of unit operation sequence exists for the manufacture of that specific catalyst and there will for each unit operation exist preferential process equipment, i.e. fluid bed calciner for calcination. The sequence of unit operations with the special selection of process equipment and all process parameters forms the know-how for manufacturing a catalyst product of large commercial value. But know-how does not mean that you always know why the desired properties are obtained due to the insufficient scientific characterisation of the catalyst material as described above under 2.1. Even small adjustments of the process can change strength, pore size distribution, bulk density, crystallite size etc. of the product and, thus, harm the performance in the industrial reactor. It has normally been costly and time-consuming to reach the final recipe and, therefore, all catalyst companies want to keep it secret. I f a single unit operation is changed it will often influence the optimisation of most of the other unit operations, and much of the development will have to be redone.
Table 2. List of unit operations with typical equipment Unit operation Typical equipment Tanks with stirrer 1. Dissolution Pumps, specially designed reactors and stirrers 2. Precipitation Temperature-regulated tanks, autoclaves 3. Ageing and maturation, gel formation Belt filter 4. Filtration Drum filter Centrifuge Filter press Belt filter 5. Washing Drum filter Centrifuge Belt conveyor furnace 6. Drying Spray drying Fluid bed drying Rotary kiln Vacuum dryer Z-mixer 7. Wet mixing (kneading) Double screw mixer Nauta mixer 8. Dry mixing Double cone mixer Ribbon blender Jet mill 9. Grinding Roller mill Universal mill Pearl mill Screen 10. Sieving Tabletting 11. Forming Extrusion Granulation Spray drying Corrugation Belt conveyor furnace 12. Calcination Rotary kiln Shaft furnaces Chamber and muffle furnace Tunnel furnace Fluid bed Pore filling- incipient wetness 13. Impregnation Immersion in liquid Controlled chemisorption See under calcination 14. Decomposition Electrical hearth 15. Fusion Prereduction reactor 16. Activation Fluid bed, chamber and muffle furnace 17. Cooling and annealing Washcoater, dragee pan 18. Coating Tanks with stirrer 19. Leaching Tanks with stirrer, kneader 20. Reslurrying . .
2.3.1 Examples of combinations A process scheme example of a precipitated catalyst is given below in Fig. 2.
Metal salts
Bases
Dissolution of raw materials I
[
Precipitation
I
I
Ageing
I
I
Filtration
I
Washing
Recycle_= -=
Re-slurrying
]
l
Spray drying
I
Rotary kiln calcination
I
Powder mixing
I
Tabletting
I
I Lubricants i -i [
Salt Salt
}=
H20 ~.
I Conveyor belt calcination I
Recycle I I
Sieving
.]
Finished product
I
Fig. 2. The scheme for production of an extruded and impregnated catalyst can be as shown in Fig. 3. Raw materials Recycled
Additives
I~ Mixing of raw materials
-I
Extrusion
I r
Impregnation ~ liquid I
Recycle
[i E
Drying Calcination Impregnation Drying Calcination Sieving Finished product
Fig. 3.
H20
HaO HaO }= H20 + gases
3. P E R F O R M I N G SCALE-UP E X P E R I M E N T S When the unit operation processes and equipment have such a major influence on the final catalyst properties and performance, it would be logical and desirable to perform the catalyst test preparations in the catalyst plant process equipment or in a pilot production plant in order to avoid using development time on "wrongly prepared" samples. However, there must be at least some selection of recipes. The scale-up factor from laboratory to plant is approximately 1,000,000 corresponding to going from g/day to tons/day. A pilot manufacturing plant will typically produce 1-50 kg/day. Then the scale-up factors will be: - From laboratory to pilot: 1000 - From pilot to plant: 1000
3.1. Plant experiments Catalyst plants are large investments in a number of production lines. Some of the lines are dedicated to one product, others are multipurpose lines where various products can be made. The advantages of performing experiments in the plant with the realistic processes and equipment in tons/day scale must be compared to the drawbacks as follows: a. Expensive raw material, energy and labour. b. Expensive disposal/recycling of unusable product. c. Lack of flexibility of production lines. d. Production lines are booked and lost contribution will also be costly. The large economic consequences and lack of flexibility will dictate to perform most of the manufacture development work in a pilot scale, where the cost of experiments is 1001000 times lower, and only to use the plant lines for the last process adjustments.
3.2. Pilot production plants Catalyst pilot production plants have process and equipment facilities that are similar to equipment used in the large scale plant as listed in Table 2, but at the kg capacity scale typically 10-50 kg/day. Equipment in a catalyst pilot production is to a great extent industrial equipment downscaled by a factor 50-1000. It is necessary that equipment can easily be moved, so flexible production lines of new combinations of unit operations can be arranged. A modern pilot plant will contain most of the equipment listed in Table 2. 4. P H Y S I O C H E M I C A L D E S C R I P T I O N OF UNIT O P E R A T I O N S
4.1. Methods of study of unit operations The methods of characterisation of intermediates and final product coming from each unit operation are of course the same as given in Table 1. However, if the unit operation process itself is to be analysed and understood more profoundly in situ or in line, analytical equipment must be installed and mathematical models for heat and mass transport must be set up.
4.2. Examples of unit operation studies 4.2 91. Precipitation High surface activated alumina carrier is important for a large number of catalysts. The precipitation can take place from a number of aluminium salts: AI(NO3)3, A12(SO4)3, A1C13 and NaAIO2 with bases or acids as NaOH, NH3, KOH, HNO3, HC1 and H2SO4. Precipitation is carried out by a controlled mixing of the reactants in order to obtain a supersaturated solution from which nucleation takes place. Amorphous primary particles are formed that later crystallise into desired phases and in parallel agglomerate to larger secondary particles 9 Precipitation processes need in line pH meters and possibilities for automatic particle size distribution analysis coupled to the ageing vessel. The potential formed types of aluminium compounds present in the solution are numerous (25). The first phases formed are far from equilibrium and, as a consequence of Ostwald's rule of stages, transform within minutes and hours. This illustrates the importance of control of history 9The precipitation methods and parameters in combination with carefully controlled ageing determine agglomerate size distribution, agglomerate packing and agglomerate strength. The precipitation may be controlled by pH and mass flows, but the study of the particle size distribution of amorphous particles and agglomerates within milliseconds is difficult, since only few relevant methods as SAXS and SANS exist. As an example of SAXS measurements, crystallite size distribution of the long dimension of precipitated pseudoboehmite is given for two sets of precipitation conditions (Fig. 4). The samples are taken out during the precipitation.
9 "~
0.05
.4,,.a
'~ 9
0.04
N r~
0.03
C,J
0.02
03 0.01
i/ ,/
0.00 i
20
'
410
'
Particle diameter (nm)
Fig. 4.
!
60
4.2.2. Calcination Controlled calcination of alumina carrier extrudates or tablets is important for obtaining the desired pore volume, BET area and pore size distribution. All these parameters are influenced by the partial pressure of water vapour at a given time and temperature. As an example a dried alumina carrier is used, which should be calcined to the highest possible surface area, but the temperature must reach at least 500~ and the use of a rotary kiln with countercurrent flow of air is considered. The kiln is heated indirectly. From test calcinations of small samples in controlled atmosphere we know the influence of the water vapour pressure. It reduces the surface area and it is quite clear that the water vapour pressure in the kiln must be kept at the lowest possible values. This means drying the air fed to the kiln, if possible, and maximising the ratio between the air flow/carrier flow. The generation of water from calcining the alumina can be found by a simple thermo-gravimetric analysis: The profile of the water vapour pressure in a countercurrent rotary kiln follows the thermogravimetric profile - with adjusted scales and minimum values equal to the water vapour pressure of the feed gas - therefore, it can be expected to reach a high surface area, if the water vapour generated in the first part of the kiln is diluted to a sufficiently low partial pressure. If this is tried disappointing results will probably be obtained due to the diffusion resistance: a. The carrier in the kiln forms a bed of a certain thickness. The carrier is heated by a number of processes: heat transfer from the gas sweeping over the bed surface and from the kiln wall. Therefore, all the carrier at the bottom of the bed, as well, will generate water vapour at a given time. The water vapour can only escape the bed by diffusion or when the bed "folds over". To get an idea of the influence hereof, a one-dimensional equation for diffusion from a bed where all particles generate gas can be solved [26]:
(
o
_.r
j
I A f
90
.
A
/
.
/
~
/
=
"
V - ~ - - - V - - - - - - - V
85
cb
0 n
80 75
o~ ,-
v
I
9
i
9
i
9
i
9
i
9
i
9
i
'
6
O (1) >
c
O o
9
V ~ V ~
"~:~. ~ - ~
4 (D
"T" O
co
2
'
I
V ~ y ~
o--e,
60
9
I
'
I
'
I
'
I
'
I
'
I
'
-\ V~
c-'
.o
0")
'-{1) > cO ro
40 O ~ ~
20
. -R
A
9
A
--A
,,
0
'
16o
'
26o
'
ado
'
400
Time(min)
Fig. 4. Influence of the nature of support on propylene conversion, PO selectivity and hydrogen conversion. Au/Ti-MCM-41(II), Au/Ti/Ti-MCM-41(@), Au/Ti-Meso(O), Au/TiO-SiO(1) (A) and Au/TiO-SiO(2) (V)
174
The results of propylene epoxidation at 150~ over Au/Ti-MCM-41, Au/Ti/TiMCM-41, Au/Ti-Meso, Au/TiO-SiO(1) and Au/TiO-SiO(2) catalysts are compared in Fig. 4. The increasing order for propylene conversion after TOS > 150 rain. over Au catalysts supported on various mesoporous titanosilicates is: Au/TiO-SiO(1)< Au/TiMCM-41 1) show metallic behaviour [18,20]. The short M - M distance implies strong metal bonding, which results in metallic conductivity. The magnetic susceptibility of NiaP and other phosphides hardly depends on temperature, which is further proof of their metallic character and shows that they can be classified as Pauli paramagnets. The metallic character of the phosphides explains why the observed NMR signals are shifted to such high values (Table 2). In a first approximation, the observed Knight shift (Fermi contact interaction term) is proportional to the product of Pauli susceptibility and average s electron density probability at the Fermi level [15]. In heavy and transition metals, electrons in the p and d orbitals can make large contributions by the orbital (van Vleck) paramagnetism and by coupling of the d orbitals with the core s orbitals (core polarization) [15].
1 (a)
(b) I
700
I
600
I
500
I
400
I
I
300
[ppm]
200
I
100
I
0
,I
-100
Fig. 3: 31p NMR spectra of WP at a spinning rate of 10 kHz (the centerband is marked by a triangle): (a) observed and (b) fitted spectra
3.2. SiO2-supported transition-metal phosphides Our research on supported transition-metal phosphides started with the preparation of nickel phosphides supported on an amorphous SiO2 carrier. SiO2 was preferred to A1203 as it interacts to a lesser extent with phosphates, which are known to form aluminum phosphates in the presence of A1203 [21]. The latter was already used as a support for MoP [22], but the authors had to introduce high loadings of molydenum and phosphorus
253 onto the low surface area A1203 support (91 m2.g1) in order to be able to detect MoP phases by X-ray diffraction. At such high loadings, only part of the Mo and P will interact with the support and the danger of segregation is lower. There was a clear influence of the support on the reduction behaviour of the phosphide species: while pure bulk Ni2P could be obtained from the reduction of the corresponding oxidic precursor containing a stoichiometric Ni to P ratio of 2 to 1, an excess of phosphate had to be introduced onto the SiO2 support in order to synthesize SiO2-supported Ni2P. A Ni to P ratio equal to two lead to the formation of compounds such as NiI2P5 or Ni3P depending on the reduction conditions but Ni2P could not be obtained. Other parameters, such as the reduction temperature and the flow rate were also found to have an influence on the type of phase of the Ni phosphides synthesized. Three different Ni phosphides (Ni2P, NinPs, Ni3P) were then prepared on SiO2 and characterized by powder X-ray diffraction and 31p MAS NMR spectroscopy. The influence of the flow rate on the reduction behaviour is illustrated in Fig. 4, which shows the TPR profiles obtained from the reduction of the oxidic precursor Ni-P(2:I) (where the numbers in brackets refer to the stoichiometric ratio between Ni and P) under a flow of 4.8% H2 in Ar using two different flow rates (10 and 50 ml.min'l). Although the shape of the two profiles is similar, the temperature maxima are shifted towards higher values when a lower flow rate is used. While the reduction starts at around 520 K when 50 ml'min~ of 4.8% H2/Ar are used, it only starts at 600 K and finishes around 1100 K under a flow rate of 10 ml.min"l. This is explained by the formation of water during the reduction process, the removal of which being favoured by a higher flow rate [23]. This contrasts with the results obtained for the unsupported Ni2P, in which the flow rate was shown to have hardly any influence on the reduction behaviour.
5 t-
._o
./"
(b) / .............
/
\
,,,-----..._._...-
,,
'\
J
Q.
E
g i-
8
400
6~o
8~o
Temperature [K]
10'0o
12'00
Fig. 4. TPR profiles of the oxidic precursor of Ni-P (2:1) (a) 1023 K, 2 K.min1, 4.8%H2/Ar, 50 ml-min1 (b) 1173 K, 5 K.min"1, 4.8%HJAr, 10 ml.min"l Reduction experiments carried out in a separate micro-reactor under a flow of 5% H2/N2 confirmed the importance of the flow rate: different types of Ni phosphides were formed
254
when the flow rate was varied from 10 to 200 ml.min -1. With a flow rate as low as 10 ml'min -1, Ni3P was formed on SiO2, while NilzP5 could be observed when a flow rate of 50 ml'min -1 was used. NizP was not detected at any time under those conditions. An excess of phosphate (Ni:P=2:1.3) had to be used during the impregnation of the SiOz carrier, together with a high flow rate of 200 ml.min -1 in order to achieve the synthesis of NizP/SiO2. It is suggested that the reduction of the oxidic precursor supported on SiO2 to NizP/SiO2 occurs through different reaction steps, involving the formation of NilzP5 and Ni3P as intermediates. The different Ni phosphide samples supported on SiO2 were also studied by 31p MAS NMR and the results are summarized in Table 2. The high chemical shifts observed for the NMR signals are explained by the metallic character of the Ni phosphides. As for the unsupported NizP sample, NizP/SiO2 revealed two NMR signals with isotropic chemical shifts of 1487 and 4076 ppm. The NMR spectrum of the NilzPs/SiOz sample also displayed two signals centered at 1941 and 2259 ppm. On the contrary, one signal only was detected for Ni3P/SiOz with an isotropic shift of 1796 ppm. In addition to those signals, peaks were detected around and below 0 ppm (i.e. +2, -8 and-21 ppm) for all three supported samples. Such shifts are typical of phosphate species such as HnPO4(3-n)-, P2074- and (PO3-)n [24] These results differ from those obtained for the unsupported materials, in which no other signals than the ones corresponding to the phosphides were detected. This illustrates one of the possible effects of the support, which could segregate the Ni and phosphate and prevent the complete reduction of the latter. It has to be noticed, however, that no signals corresponding to silicon phosphate were detected below 30 ppm [25]. 4. CONCLUSIONS A variety of bulk transition-metal phosphides (CozP, NizP, MoP, WP, CoMoP, NiMoP) were successfully prepared by reduction of a metal oxide/phosphate precursor with H2 at different reduction temperatures. These materials were characterized by powder X-ray diffractometry and 31p MAS NMR spectroscopy. The latter technique enabled the determination of Knight shifts characteristic of metallic-type of phases. NizP, Nil2P5 and Ni3P supported on SiO2 were also prepared by reducing an oxidic precursor and by changing reaction parameters such as the P content, the reduction temperature or the reductant flow rate. The influence of the support on the reduction process was clearly demonstrated: pure NizP could be obtained when a Ni to P ratio of two was used during the preparation of the unsupported material, while an excess of phosphate had to be introduced onto the SiOz support to yield NizP. 31p MAS NMR proved to be a useful technique to characterize transition-metal phosphides and could be an alternative technique to the detection and characterization of small metal phosphide crystallites supported on a carrier. REFERENCES 1. R. Prins, V.H.J. de Beer and G.A. Somorjai, Catal. Rev.-Sci. Eng., 31 (1989) 1. 2. H. Topsoe, B.S. Clausen and F.E. Massoth, "Hydrotreating Catalysis", Springer, NewYork, 1996.
255 3. Th. Weber, R. Prins and R.A. van Santen, "Transition Metal Sulphides - Chemistry and Catalysis", Kluwer (Eds.), Dordrecht, 1998. 4. S. Eijsbouts, J.N.M. van Gestel, J.A.R. van Veen, V.H.J. de Beer and R. Prins, J. Catal., 131 (1991) 412. 5. M. Jian and R. Prins, J. Catal., 179 (1998) 18. 6. W.R.A.M. Robinson, J.N.M. van Gestel, T.I. Koranyi, S. Eijsbouts, A.M. van der Kraan, J.A.R. van Veen and V.H.J. de Beer, J. Catal., 161 (1996) 539. 7. W. Li, B. Dhandapani and S.T. Oyanm, Chem. Lett., 3 (1998) 207. 8. C. Stinner, R. Prins and Th. Weber, J. Catal., 191 (2000) 438. 9. C. Stinner, R. Prins and Th. Weber, J. Catal., 202 (2001) 438. 10. G. Nolze and W. Kraus, Powder Diffr., 13 (1998) 256. 11. J. Herzfeld and A.E. Berger, J. Chem. Phys., 73 (1980) 6021. 12. PDF 3-953 (Ni2P); PDF 32-306 (Co2P); PDF 24-771 (MOP); PDF 29-1364 (WP); PDF 31-873 (NiMoP); PDF 32-299 (CoMoP). 13. P. Villars and L.D. Calvert, Pearson's Handbook of Crystallographic Data for Intermetallic Phases, (2nd ed.) ASM International, Materials Park, 1991. 14. S. Ohta and H. Onmayashiki, Physica B, 253 (1998) 193. 15. W.D. Knight and S. Kobayashi, "Encyclopedia of Nuclear Magnetic Resonance", Eds. D.M. Grant and R.K. Harris, Wiley, Chiehester, 1996 (p. 2672). 16. R. Guerin and M. Sergent, Aeta Crystallogr. B, 34 (1978) 3312. 17. R.L. Ripley, J. Less Comm. Met., 4 (1962) 496. 18. E.D. Jones, Phys. Rev., 158 (1967) 295. 19. S. Wada, T. Matsuo, S. Takata, I. Shirotani and C. Sekine, J. Phys. Soc. Jpn., 69 (2000) 3182. 20. B.F. Stein and R.H. Walmsley, Phys. Rev., 148 (1966) 933. 21. P.J. Mangnus, J.A.R. van Veen, S. Eijsbouts, V.H.J. de Beer and J.A. Moulijn, Appl. Catal., 61 (1990) 99. 22. S.T. Oyama, P. Clark, V.L.S. Teixeira da Silva, E.J. Lede and F.G. Requejo, J. Phys. Chem. B, 105 (2001) 4961. 23. P. Burattin, M. Che and C. Louis, J. Phys. Chem. B 104 (2000) 10482. 24. K. Eichele and R.E. Wasylishen, J. Phys. Chem., 98 (1994) 3108. 25. T.R. Krawietz, P. Lin, K.E. Lotterhos, P.D. Torres, D.H. Barieh, A. Clearfield and J.F. Haw, J. Am. Chem. Soe., 120 (1998) 8502.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
257
The application of non-hydrothermally prepared stevensites as support for hydrodesulfurization catalysts M. Sychev a, R. Prihod'ko a, A. Koryabkina b, E.J.M. Hensen b, J.A.R. van Veen ~ and R.A. van Santen b aFaculty of Chemical Technology, National Technical University of Ukraine, 03056, Kiev, pr. Peremogy 37, Ukraine
bSchuit Institute of Catalysis, Eindhoven University of Technology, Den Dolech 2, PO Box 513 5600 MB Eindhoven, The Netherlands CShell International Chemicals B.V., Badhuisweg 3, 1031 CM Amsterdam, The Netherlands The stevensite-like materials containing Mg, Mg-Ni, and Mg-Co in the octahedral sheets were synthesized under non-hydrothermal conditions. The nature of divalent octahedral cations influences the surface area and pore volume of these materials. The thiophene HDS activities of stevensite-supported catalysts prepared by addition of Mo or W to the Nicontaining supports, especially to the NiZ+-exchanged stevensite, were superior in relation to those of their counterparts made by conventional co-impregnation of the Mg-stevensite with NiMo and NiW. The use of chelating agents, NTA and EN, affects beneficially the HDS activity of the catalyst which, however, is less obvious in the case of the CoMo catalysts. The positive role of the chelating agents is explained by weakening of the Ni(Co)-support interaction and by a change in the sulfidation sequence of Ni(Co) and Mo(W). The Ni z+- and Co Z+-exchanged Mg-stevensite or that isomorphously substituted with these cations are promising candidates for application as supports for hydro-processing catalysts. The thiophene HDS activity of the NiMo and NiW stevensite-supported catalysts is comparable with that of the corresponding ~,-AIzO3 commercial ones. 1. INTRODUCTION The necessity to develop hydrotreating catalysts with enhanced activity stimulates the search for alternative catalyst supports. It was shown that clay-supported transition metal sulfides can efficiently catalyze hydrodesulfurization (HDS) of thiophene [1-3]. However, the large scale application of the catalysts based on natural clays is still hampered, mainly due to the difficulties in controlling the chemical composition and textural properties. Synthetic clays do not suffer from these drawbacks. Recently, a novel non-hydrothermal approach was proposed for the synthesis of some trioctahedral smectites, namely saponite
258 and stevensite [2, 4-6]. Here, we report on the preparation and characterization of Mo(W)S~ promoted by the Co(Ni)-catalysts obtained using synthetic stevensites differing in the chemical composition as a support material. An attempt has been made to investigate the influence of chelating agents on the catalyst behavior in the thiophene HDS. 2. EXPERIMENTAL 2.1. Sample preparation Stevensite-like materials were synthesized on the basis of the theoretical composition of stevensite {N,jzZ+(M2+6_x.x)[Sis]O20(OH)4n H20}, where Nxz§ represents an interlayer cation and 9 a vacant site, respectively. The starting silica-aqueous suspension was obtained by heating (363 K, 1 h) the water-dispersed aerosil with the solid/liquid ratio of 1:50. Next, a solution containing 70 mmol of Mg(NO3)2 or a mixture of 63.7 mmol of Mg(NO3)2 with 6.3 mmol Ni(NO3)2 or Co(NO3)2 and 0.3 mol of urea was slowly added to this suspension and the reacting mixture was kept under stirring at 363 K for 36 h. The solid products were separated, washed, and dried at 383 K for 12 h. The samples with Mg2§ MgZ+-Niz§ and Mg2+-Co2+ (the cation ratio = 10:1) in the octahedral sheets are denoted as MgST, NiST, and COST, respectively. Ni2§ and Co2+-exchanged MgSTs were obtained conventionally and are referred to as NiexST and CoexST, respectively. The NiMo and CoMo catalysts were prepared by impregnation of the support with an aqueous solution containing ammonium heptamolybdate (AHM) or ammonium metatungstate (AMT) and Ni(Co)(NO3)2 with the Ni(Co) to Mo(W) molar ratio of 0.33 in the absence or in the presence of ammonia and a chelating agem, nitrilotriacetic acid (NTA) or ethylendiamine (EN). These chelating agents were chosen because of their different complexing properties. The NiW catalysts were obtained in the presence of the chelating agents only. For NTA, the ligand: Ni(Co) molar ratio was 1:1, while in the case of EN, this ratio was 4. 2.2.
Apparatus used and conditions of measurements
XRD: oriented powdered specimens, a DRON-3 diffractometer, CuIQ and CoK~ radiation. XRF: a high-vacuum device, Si(Li) detector (180 eV at 5.9 keV), sample weight of 200 mg. UV-visible diffuse reflectance spectroscopy (DRS): Shimadzu UV 2401, reference BaSO4. FTIR: Perkin-Elmer 2000, KBr technique. N2 adsorption: ASAP 2010 (Micromeritics), degassing at 403 K, 10.4 mbar, 5 h, pore volume calculated by the smethod. Temperature-programmed reduction (TPR): 66% H2 in Ar, flow-rate of 20 ml/min, temperature range of 298-1073 K, heating rate of 5 K/min, thermal conductivity detector. The thiophene HDS activity was measured at 673 K in a microflow reactor (1 bar pressure, 4 vol.-% thiophene in H2, 50 std cm3 minl). The samples were sulfided in situ using a gas mixture of 10 % H2S in H2 (60 std crn1, 6 K min1 from 293 to 673 K, 2 h at 673 K).
259
3. RESULTS AND DISCUSSION 3.1. Characterization of supports and catalyst precursors
The XRD patterns of the samples obtained (Fig. 1) are consistent with those of synthetic and natural stevensites [7,8], evidencing formation of the trioctahedral smectite structure, as indicated by the (060) reflection at about 1.52 A. However, intensities of the X-ray reflections vary for the particular solids, which is caused probably by differences in the crystallinity and/or in size and stacking of particles. All the samples show a typical smectite ability to swell in ethylene glycol and exhibit ion-exchange capacity comparable with that of the natural stevensite (of about 50 mequiv/g). The FTIR spectra of all the samples contain lattice vibration I ~" bands at about 1010, 660, 440, ]l o and 450 crn-1 and a shoulder at ,", 1 around 420 cm-1 assigned, respectively, to Si-O stretching, d S" OH libration, Si-O bending and ~ ' ~ ~" ~ 9 ~ g vibration of a Si-O-M 2+ bond ~ I ~ _ ~ ~ ~ _ _j~ characteristic of stevensite [8]. ~ -__.-._ _ 1" No silica hydrogel vSiO2 at 1100 t 2 cm-1 [7] was detected. The diffuse reflectance 3 spectra of NiST display bands at approximately 380, 670, and 0 ' 2'0 ' 4'0 ' 6'0 . . . . 80 740 nm, which originate from 2 Theta octahedrally coordinated Ni 2+ cations due to the transitions Fig. 1. X-ray diffraction patterns of stevensites air-dried from 3A2g(F) to 3Tlg(P) 3Tlg (F) at 383 K: (1) MgST, (2) NiST, (3) COST. and 3T2g(F) [9]. The DR spectra * CuK~ radiation. of CoST show a band at 528 nm and shoulders at 500 and 646 nm, assigned to the v3 (4m2g-')4Tlg) and v2 (4Tlg(P)'~4Tlg) transitions, respectively, typical of Co 2+ in the octahedral coordination [10]. Since Ni 2+ and Co 2+ admittedly accommodated in the non-framework positions were removed before the DRS measurements by the ion-exchange with NH4+, the observed DR bands can be attributed to the cations located in the octahedral sheets of the stevensite structure. The temperature programmed reduction (TPR) of NiST proceeds in two steps with maxima at around 700 and 815 K (Fig. 2). The observed TPR peaks are indicative for the presence of Ni 2+ with different reactivity for hydrogen toward the zero-valent state. Since the non-crystalline silica-alumina phase gave a much poorly resolved TPR profile (Fig. 2), the peaks should be related to the crystal structure of NiST. The nickel cations placed inside the octahedral sheets being coordinated with six framework oxygen atoms should be more stable than Ni 2+ exposed at the edges of the clay platelets. Consequently, the first reduction step can most likely be attributed to the reduction of the Ni 2+ cations located at those positions. Since the reduction of NiO takes place at approximately 570 K, these data show
260 absence of the Ni(OH)2 impurity. While in the case of CoSP virtually no Co(OH)2 phase was detected (623 K), the reduction takes place at considerably higher temperatures as compared to that for NiST. The TPR pattern of CoST contains also two reduction steps centered at 816 and 1074 K (Fig. 2). By analogy with NiST, they can be ascribed to the reduction of Co 2§ exposed at the edges of the clay platelets and inside the octahedral sheets, respectively. The sharper main reduction step, compared to that for NiST, shows that the Co 2§ cations are better distributed in the stevensite structure. 1074
682~06 ,
J 400
600
800
I000 400
4
600
800
3
I000
1200
Temperature [ K] Fig. 2. TPR patterns of (1) NiST and (3) COST. (2) Ni- and (4) Co-containing noncrystalline" silica-alumina phases are included as references. The N2 adsorption isotherms of all stevensites prepared are close to type II, being typical of mesoporous solids [11 ], with the H2 hysteresis representative of ink-bottle shaped pores (Fig. 3). The closure of the loop at P/Po = 0.45-0.5 indicates predominance of the mesopores with the size just below 4 nm. ,__,400
f
o
"~ 113 300
40
/
200
1
/,,
t
.,.o r
O
e.r 0
-,am-,-v o o~ ,=,7 ~.%....~" gX"" ":
#-_/....'+
/
..,.o- +" -"
.,o.,b,,/o'
1oo
o
0'.2 " 0.4
o16
oi~
1'.o
o~
0'.4
oi,5
oi~
~io
Relativepressure [PIP]
Fig. 3. N2 adsorption-desorption isotherms of (1) NiST, (3) COST, and (2,4) their sulfided counterparts.
261 The sulfidation of NiST and CoST at 673 K affects the shape of the N2 adsorption isotherms (Fig. 3), causing also some decrease in the surface area and pore volume (Table 1). The samples suffered from some loss of crystallinity, as indicated by broadening of the X-ray reflections. These results show that significant fractions of the Ni2§ and Co 2~ cations are sulfided and extracted from the octahedral sheets, which leads to partial amorphization of the stevensite structure. However, its main part persists with virtually no collapse. All the presented data confirm the formation of stevensite-like materials, the textural properties of which depend on the composition of octahedral sheets (Table 1). Table 1 Abundance of octahedral cations (XRF, in molar ratios), surface areas (SBET), and total pore volumes (Vto~0 of the studied air-dried stevensites and their sulfided counterparts Sample MgST NiST CoST NiSTsulf. Co STsulf.
Mg 5.62 5.11 5.12 5.11 5.12
Ni or Co -0.52 0.51 0.52 0.51
SBET(m2/g) 455 383 430 322 372
Vtotal(cc/g) 0.385 0.587 0.482 0.415 0.418
Treatment of the supports with impregnating solutions decreased their surface area and total pore volume by 10-15%, indicating incorporation of active phases in the stevensite interlayer space.
3.2. Thiophene HDS activity The sulfided Mg-stevensite itself does not exhibit any thiophene HDS activity. When Mo or W are introduced, the clays become active, exhibiting the conversion of 15 and 10% at 5 h run-time, respectively. The NiST and CoST samples (Ni2§ and Co 2§ are located together with Mg 2§ in the lattice) show comparable activities. The addition of Mo and W via impregnation substantially increases the HDS activity of these materials. A slight difference in activity is observed for the catalysts prepared from NiST or CoST and those obtained by the impregnation of MgST with the Ni-Mo or Co-Mo solutions (Fig. 4). This indicates that the Mo(W) cations can interact with Ni2§ or Co 2§ located in the clay lattice. The possibility that these cations can migrate from the framework upon sulfidation thus forming "NiMoS", "CoMoS", and "NiWS" phases cannot be excluded. This suggestion is consistent with the data obtained by means of the N2 adsorption. The thiophene HDS activity of the stevensite-supported catalysts was found to be sensitive to the mode of preparation. The catalysts prepared by addition of Mo or W to the Ni-containing supports (NiexST and NiST) were superior to their counterparts made by convemional co-impregnation of the Mg-stevensite with Ni and Mo(W) (Fig. 4). In the case of the CoMo catalysts, this effect was less pronounced. The beneficial effect of the presence of a complexing agent on the thiophene HDS activity was restricted to the catalysts containing both Ni(Co) and Mo(W). These findings are in accordance with those reported by Prins and co-workers [12]. The extent of the activity increase depends on the
262 ~
::::::::::::::::::::::::::::::::::::::::::::::::::::: :~:~'~'~:~-~'~ ":~-~:'
==================================================== ..........':q .......... :'::::"..... ~ , . M , ' , " . ' " ...........................
ata.ysts preparo wit
~
the
i!] i:' ": 3 0
............................ ~. .,,,,,,,,,,,,:~.-:-.~:--~:-.c::-.~:;--:~.-~ ,~.~...:~:~.::~:- "_l xL I, i l t." t,f.',' .~, . , ,
composition of the active phase and properties of the chelating agent. For the
i!~ ":
activity sequence was: NiMo NiW > "~:J!i!i ::'!i!..... ~'-ti[il::i......... ". :b1:~i~i Z0 CoMo, while the presence of EN resulted in !!~!i] i ~ ,l'!,!,, the following activity order: NiMo CoMo > NiW (Fig. 4). ~iii EN The obtained results can be explained ~?.:.-.---'-..-.......--.".'~~2 ~i:!l .....~ ! 1 ~~~" J NTA EI~O while taking into account several functions NiexST NiST MgST* of the chelating agents [12-14]. Firstly, they can inhibit formation of nickel silicates by Irarl 2o 7,e prohibiting hydrolytic interaction of the Ni2+ cations with silanol groups present on the support surface [12]. Most likely, this phenomenon can be propagated onto the 00 ..~-- 9 catalysts containing CoMo. NTA behaves differently than EN, forming much more ~!.".....,.. ......~.'_.:~:..'.:t.............~.!:..v..:.>.!....,,...~:.>..~ NTA CeexST CoST M~ST* stable complexes with Ni than EN does [13], and therefore better protecting nickel and hindering interaction with the support. This can explain higher activities of the NiMo and ~::|::: ::.';i~:." :::::::::::::::::::::::: ::::.'::::" NiW catalysts obtained from NiexST in the presence of NTA, as compared to that ~iWi 9 ~ililili!i ii~g :!!i!: ~'o prepared with use of EN. However, in the case of Co 2+, such a preventing functionality of the complexing ligands seems to be less pronounced (Fig. 4). N:c~:~T N~ST MgST* Secondly, adding NTA and EN to the Fig. 4. Initial thiophene HDS activity (5 impregnating solutions prevents the min run) of stevensite-supported catalysts sulfidation of Ni at low temperatures, thereby increasing the formation of the prepared in the presence or in the absence "NiMoS" phase, as was deduced from the of chelating agents in the impregnating increased activity in the thiophene HDS [12- solutions. 14]. Courier et al. [15] also showed that the * Samples obtained via co-impregnation. same effect, namely stabilization of cobalt against the sulfidation, explains the role of NTA in enabling the formation of "CoMoS" in the silica-supported CoMo catalysts. Some effect of the chelating agents on the dispersion of the MoS2 (WS2) particles also cannot be excluded [15]. This leads to the improved catalyst behavior. These conceptions, explaining the functions of the chelating agents, were mainly developed for the silica-supported HDS catalysts [12-15]. However, the data presented here indicate their vafidity for the stevensite-supported HDS catalysts. Nevertheless, taking into account differences in the crystal structure, surface chemistry of the smectites, and the corresponding characteristics of silica, additional study is necessary to elucidate the case.
d
!liiil
263 The catalysts under study deactivate relatively strongly during the initial 5 to 35 min of the test experiment (Fig. 5). This is probably due to the coke formation, although a contribution to the activity decrease arising from the establishment of a new equilibrium under reaction conditions cannot be excluded [16]. The extent of the catalyst deactivation depends on the active phase composition; the CoMo catalysts showed the strongest deactivation (Fig. 5). The use of the chelating agents, especially NTA, improves the catalyst stability. This effect can be attributed to a better dispersion of the sulfide active phase that hampers its sintering during the sulfidation and HDS test. As concerns selectivity, it can be stated that generally the thiophene HDS over the stevensite-supported catalysts resulted mainly in the formation of butane and butenes. Only during the initial stage of the catalytic test, when the conversion ofthiophene was very high, a small amount of C1-C3 hydrocarbons could be detected. Distribution of the C4 products depends on the composition of the metal-sulfide phase (Fig. 5B). The NiW catalysts produce more n-butane, likely because of their higher hydrogenation activity. Hence, catalytic tests in the hydrodesulfiafzation of thiophene showed that the behavior of the catalysts under study depends on the metal-sulfide phase composition and the mode of its formation. The presence of the chelating agents in the impregnating solutions influences both activity and stability of the catalysts. A.
35 =I
NiMoST
5o
u
NiWST
~30
,._.,
10
~--~
CoMoS T
I0
5
0
0
2
4 Run time [h]
6
g
-
-
-
n-butane l-butene 2-t-butene 2-c-butene
Figure 5. (A) Thiophene HDS over stevensite-supported catalysts: (1) NiexSTW(NTA), (2) NiexSTMo(NTA), (3) CoexSTMo(NTA), (4) NiexSTMo(H20), and (5) CoexSTMo(H20). (B) C4 products selectivity as a function of the metal-sulfide composition. 4. CONCLUSIONS The results obtained when using a variety of methods indicate that Mg-, Mg-Ni-, and Mg-Co-containing stevensite-like materials were synthesized under non-hydrothermal conditions. The composition of the octahedral layers affects the textural properties of these materials that are predominantly mesoporous.
264 The thiophene HDS activity of the stevensite-supported catalysts depends on the mode of their preparation. The materials prepared by addition of Mo or W to the Ni-containing supports, especially to the Ni2+-exchanged stevensite, were superior to their counterparts made by conventional co-impregnation of the Mg-stevensite with Ni and Mo(W). In the case of the CoMo active phase, introduction of the Co cations before those of Mo influences the catalyst activity less favorably. The activities of the NiMoST and NiWST catalysts prepared with NTA and EN exhibited an improvement, suggesting that Ni is responsible for the increase in the catalytic activity. This suggestion is based on the fact that these chelating ligands prefer coordination with Ni rather than with Mo and W. In the case of the CoMo catalysts, the beneficial effect of the NTA and EN is less obvious. The positive role of the chelating agents is explained by weakening of the Ni(Co)-support interaction and by a change in the sulfidation sequence of Ni(Co) and Mo(W). The Ni 2+- and C02+-exchanged Mg-stevensite or that isomorphously substituted with these cations are promising candidates for application as supports for hydro-processing catalysts. The NiMo and NiW stevensite-supported catalysts showed the thiophene HDS activity comparable with that of the corresponding %,-A1203commercial ones. ACKNOWLEDGMENTS
The authors thank Ms. M.C. Mittelmeijer-Hazeleger (University of Amsterdam, The Netherlands) for the adsorption measurements and Dr. K. Erdmann (Nicholas Copernicus University, Torun, Poland) for help. These investigations were supported in part by the Ukrainian Ministry of Education and Science and by a Spinoza grant (to R.A.v.S.) from the Dutch Science Foundation. REFERENCES
1. M. Sychev, V.H.J. de Beer, A. Kodentsov, E.M. van Oers and R.A. van Santen, J Catal. 168 (1997) 245. 2. R.G. Leliveld, W.C.A. Huyben, A.J. van Dillen, J.W. Geus and D.C. Koningsberger, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 106 (1997) p. 137. 3. E. Hayashi, E. Iwamatsu, M.E. Biswas, Y. Sanada, S. Ahmed, H. Hamid and T.Yoneda, Appl. Catal. A: Gen. 179 (1999) 203. 4. R..J.M.J. Vogels, M.J.H.V. Kerkhoffs and J.W. Geus, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 91 (1995) 1153. 5. M. Sychev and R. Prihod'ko, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 118 (1998) 967. 6. M. Sychev, R. Prihod'ko, I. Astrelin, P.J. Stobbelaar and R.A. van Santen, Book Abstract, EUROCLAY'99 Conf., Krakow, Poland (1999) 135. 7. N. Takahashi, M. Tanaka, T. Satoh, T. Endo and M. Shimada, Micropor. Mater., 9 (1997) 35. 8. G.T. Faust, J.C. Hathaway and G. Millot, Amer. Miner., 44 (1959) 342. 9. A.P. Hagan, M.G. Lofthouse, F.S. Stone and M.A. Trevethan, Stud. Surf. Sci. Catal.,
265 Elsevier, Amsterdam, 3 (1979) p.417. 10. J. Dedecek and B, Wichterlova, J. Phys. Chem. B., 103 (1999) 1462. 11. F. Rouquerol, J. Rouquerol and K. Sing, Adsorption by Powder and Porous Solids. Principles, Methodology and Applications, Acad. Press, San Diego, 1999, pp.439-441. 12. L. Medici and R. Prins, J. Catal., 163 (1996) 28. 13. R. Cattaneo, T. Shido and R. Prins, J. Catal., 185 (1999)199. 14. L. Coulier, V.H.J. de Beer, J.A.R. van Veen and J.W. Niemantsverdriet, J. Catal., 197 (2001) 26. 15. L. Courier, V.H.J. de Beer, J.A.R. van Veen and J.W. Niemantsverdriet, Topics Catal., 13 (2000) 99. 16. V.H.J. de Beer, C. Bevelander, T.H.M. van Sient Fiet, P.G.A.J. Werter and C.H. Amberg, J. Catal., 43 (1976) 68.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
NiMo/HNaY(x)-AI203
dibenzothiophenes:
267
catalysts for hydrodesulfurization of hindered the preparation method
effect of
T. Klimova a'*, D. Soils a, J. Ramirez a and A. L6pez-Agudo b aUNICAT, Departamento de Ingenieria Quimica, Facultad de Quimica, Universidad Nacional Aut6noma de M6xico, Cd. Universitaria, Coyoacfin, 04510 M6xico D.F., M6xico Fax: +52+55+56225366, e-mail:
[email protected] bInstituto de Catfilisis y Petroleoquimica, CSIC, Cantoblanco, 28049 Madrid, Spain Three NiMo-HNaY-alumina catalysts with similar composition were prepared by different methods and tested in the hydrodesulfurization of dibenzothiophene (DBT) and 4,6-dimethyl-DBT. It was found that the catalyst preparation method induces some changes of the characteristics of the deposited metallic species as well as of the acidic properties of the zeolite component. These changes affect the catalytic behavior in the hydrodesulfurization of DBT and 4,6-DMDBT. Acidic properties of the catalyst seem to be more important for the conversion of alkyl-substituted DBT. 1. I N T R O D U C T I O N In recent years, the interest in new efficient HDS catalysts is growing due to more severe environmental legislation with respect to sulfur level in fuels and the need to process increasing amounts of high sulfur containing crude oil. To achieve the necessary efficiency, the work should be directed towards the development of catalysts with high desulfurization activity with respect to strongly hindered sulfur containing molecules, such as dibenzothiophene (DBT) and its alkyl-substituted derivatives: 4-methyldibenzothiophene (4-MDBT) and 4,6-dimethyldibenzothiophene (4,6-DMDBT) [1]. The latter compounds particularly resilient with respect to HDS cannot be completely removed using current catalyst formulations, like NiMo or CoMo on ~,-AlzO3 [2, 3]. The extremely low reactivity of 4-MDBT and 4,6-DMDBT was ascribed to steric hindrance that occurs between the methyl groups of the dibenzothiophene molecule and the active sites of the catalyst. It was shown that hydrogenation of methyl-substituted DBT's prior to sulfur removal is able to alleviate this steric hindrance and, therefore, facilitate the HDS reaction [4]. Another possibility to increase desulfurization of 4,6-DMDBT is to use zeolite containing HDS catalysts [5, 6]. In this case desulfurization can take place more easily after cracking or inter and intramolecular migration of the methyl groups on the dibenzothiophene structure. However, there are no reports about the best method that can be used to prepare NiMo/~,alumina catalysts modified with zeolite and about the effect that the method of preparation
268
has on their catalytic behavior. It can be expected that, if the zeolite component is incorporated into the alumina matrix before the active phase deposition, some changes in Ni and Mo location and dispersion, as well as in zeolite acidity, can take place and they may affect the HDS performance of the catalyst. In order to obtain more information about zeolite containing systems, we characterized a series of catalysts with the same HNaY content (20 wt%) prepared by different methods. It is the object of the present work to study the effect that the preparation method has on the metallic and acid functionalities of the catalyst and its performance in the hydrodesulfurization of DBT and 4,6-DMDBT. 2. E X P E R I M E N T A L Three NiMo/HNaY-AI203 catalysts with the same zeolite loading (20 wt%) were prepared by different methods: i) mechanical mixing of the zeolite with a conventional NiMo/AI203 catalyst ((NiMo/AI203)+HNaY(MM)), ii) using y-alumina binder to join the zeolite and NiMo/AI203 ((NiMo/AI203)+HNaY(B)), and iii) by impregnation of Mo and Ni on the composed zeolite-alumina support prepared by the peptization method (NiMo/HNaY-AI203(P)) [7]. A commercial NaY zeolite (Si/A1 ratio = 2.42), 58% exchanged with ammonium acetate solution to obtain HNaY, was used as the zeolite source and pseudo-boehmite Catapal B as the 7-A1203 source. The impregnation of Ni and Mo (Mo first) on the supports was made by the pore volume method, using aqueous solutions of ammonium heptamolybdate and nickel nitrate to reach nominal composition of 12 wt% of MoO3 and 3.45 wt% of NiO. The catalysts were dried at 100~ (24 h) and calcined at 500~ (2h). The catalysts were characterized by N2 physisorption, XRD, thermodesorption of Py (FT-IR), NH3 TPD, TPR and NO chemisorption. The DBT and 4,6-DMDBT hydrodesulfurization activity tests were conducted in a 300 ml batch reactor at 300~ and 7.3 MPa total pressure. Before the activity tests, the catalysts were sulfided ex-situ in a tubular reactor at 400~ 4 h, in a stream of H2S (15 vol%)-H2. The course of the reaction was followed by taking liquid samples and analyzing them by GC or GC-MS. 3. RESULTS AND DISCUSSION
3.1. Catalyst characterization The results from N2 physisorption (Table 1) indicate that the use of binder during catalyst preparation produces some loss of zeolite microporosity, probably due to the micropore blockage. This effect was not observed in the (NiMo/AI203)+HNaY(MM) sample prepared by mechanical mixing.
269 Table 1 Textural properties of NiMo catalysts and HNaY support Sample
Surface area (mZ/g)
SBET NiMo/AlzO3 NiMo/HNaY-AIzO3(P) (NiMo/AIzO3)+HNaY(B) (NiMo/AIzO3)+HNaY(MM) NiMo/HNaY HNaY
Pore volume (cm3/g)
Smicropores
200 186 199 238 239 582
Vtotal
0 48 45 102 189 478
Vmicropores
0.33 0.30 0.36 0.34 0.14 0.31
Average pore diameter (A)
0 0.022 0.020 0.046 0.088 0.222
46 60 68 70 52 -
The XRD of NiMo catalysts (Fig. 1) reveal the presence of faujasite crystalline phase in all the cases. However, the intensity of the characteristic faujasite reflections is significantly higher in the diffractogram of the (NiMo/AIzO3)+HNaY(MM) sample than in those catalysts prepared with binder. A comparison of the diffractograms of the NiMo/HNaYAIzO3(P) catalyst with that of a mechanical mixture of the HNaY-AlzO3(P) support with 12
600
Q
o
=
400
=
200
,
i
I
20
,
I
40
,
a
I
60
,
80
De~rees (20} Fig. 1. X-ray powder diffraction patterns of NiMo catalysts: (a) (NiMo/AlzO3)+HNaY(MM), (b) (NiMo/AIzO3)+HNaY(B), (c) NiMo/HNaY-AIzO3(P), and (d) mechanical mixture of 12 wt% of MoO3 with HNaY-AIzO3(P) support (*faujasite, 9 molybdite). wt% MoO3 (curves c and d, Fig. 1), show that the impregnation of Mo and Ni species, and subsequent calcinations are responsible for the crystallinity loss of the faujasite phase. Previously, similar destruction of the NaY and HY zeolite structures was observed as a
270 result of Mo impregnation [8]. It was also found that the extent of crystallinity loss of NaY increased as the Mo content increased. Therefore, it can be supposed that in both catalysts prepared here with binder some destruction of the zeolite crystals takes place. It is possible that this deterioration of the zeolite structure is due to the penetration of the Mo species into the zeolite cages during the preparation of the catalysts, probably during the calcination step. The possibility of such migration of Mo species into zeolite was reported previously for MoO3/NaY [9]. The results from surface acidity measurements by NH3 TPD in the catalysts show that the total number of acid sites was proportional to the accessible zeolite area (Tables 1 and 2). The same trend was found also for the Br6nsted acid sites quantified by FT-IR of Py. Table 2 Acidity of NiMo catalysts Total acidity*
Br6nsted acidity** (gmol Py/g)
Sample
NiMo/AI203 NiMo/HNaY-AI203 (P) (NiMo/AIzO3)+HNaY(B) (NiMo/AI203) +HNaY (MM) NiMo/HNaY
pmol NH3/m 2
~tmol NH3/g
150~
250~
350~
25.71 38.45 34.92 46.31 64.74
5142 7268 6932 11031 15409
0 4.6 3.4 7.1 8.3
0 2.9 2.7 6.0 7.1
0 2.3 1.8 5.1 5.4
* Total number of acid sites determined by NH3 TPD ** The amount of pyridine adsorbed on Br6nsted acid sites per gram catalyst determined by FT-IR at different temperatures of Py desorption [10] In order to envision the effect that the catalyst preparation method has on the type and reducibility of the Mo and Ni species present in the catalyst precursor, TPR experiments were performed with all the catalyst samples in their oxide form. Fig. 2 shows the TPR patterns. For the pure alumina-supported catalyst, a TPR trace typical of a NiMo/AI203 catalyst is obtained. It presents two main reduction peaks, assigned to the reduction of octahedral Mo species (peak at about 400~ and tetrahedral Mo species in strong interaction with the alumina support (peak at 770~ [11]. It is also observed the presence of shoulders at 517~ and 600~ due to the reduction of octahedral Mo species with different degrees of polymerization. As HNaY is incorporated into the catalyst, by different methods, some changes in the proportion of tetrahedral and octahedral Mo species and in their agglomeration take place. Thus, in the thermogram of NiMo/HNaY-AIzO3(P) sample, a new reduction peak at 480~ is clearly observed. The position of this peak corresponds well with the reduction of octahedral Mo species supported on HNaY, where the surface agglomeration of anionic Mo species takes place during the impregnation of ammonium heptamolybdate [9]. This catalyst presents the highest proportion of octahedral Mo species. In contrast, for the (NiMo/AIzO3)+HNaY(B) sample, an increase in the intensity of hightemperature reduction peak, corresponding to tetrahedral Mo species, is observed (curve c,
271 Fig. 2). This increased proportion of tetrahedral Mo species is characteristic for low Mo loading MoO3/AIaO3 catalysts. It seems that in the (NiMo/AIaO3)+HNaY(B) catalyst the migration of Mo species towards the surface of alumina used as a binder takes place. In the case of the (NiMo/AIzO3)+HNaY(MM) catalyst the proportion of octahedral and tetrahedral Mo species is almost the same as in the initial NiMo/AlzO3 catalyst. Therefore, 'no significant migration of the Ni and Mo species is suggested.
// 300
200
d
J .~ = 1 o0
0 m
0
I
200
~
I
,
I
,
I
400 600 800 Temperature (~
t
//
,
1000
Fig. 2. TPR patterns of NiMo catalysts: (a) NiMo/Al203, (b) NiMo/HNaY-AlzO3(P), (c) (NiMo/AIzO3)+HNaY(B) and (d) (NiMo/AlzOa)+HNaY(MM). Isothermal period (1000~ after the axis break. Clearly, the changes in the characteristics of metal oxidic species in the catalysts, induced by the differences in the catalyst preparation procedure, may alter the morphology and dispersion of the final active MoSz phase and, therefore, the amount of catalytic sites. Therefore, chemisorption of NO was used to quantify the CUS (coordinatively unsaturated Mo sites) supposed to be the active sites of hydrotreatment catalysts [12]. The NO chemisorption results (Table 3) indicate that the amount of NO adsorbed on the three zeolite-alumina-containing catalysts is intermediate between NiMo/AlzOa and NiMo/HNaY samples. The results show that for the (NiMo/AlzO3)+HNaY (MM) and (NiMo/AlzOa)+HNaY(B) catalysts, the number of the accessible MoS2 active sites is lower when the binder is used in the preparation of the catalyst. The lowest number of active sites observed for the (NiMo/AIzO3)+HNaY(B) catalyst can be due to: i) the highest proportion of difficult to sulfide tetrahedral Mo species in strong interaction with the alumina support, observed by TPR, and ii) the partial coverage of surface Ni and Mo species by the alumina binder. The NiMo/HNaY-AIzOa(P) catalyst, prepared by the impregnation of Ni and Mo species on the HNaY-AIzO3(P) support, shows the highest amount of chemisorbed NO
272 among zeolite-alumina-containing samples, in line with the observed highest proportion of easy to sulfide octahedral Mo species. 3.2. Hydrodesulfurization activity The catalytic activity tests indicate that the activity in DBT hydrodesulfurization decreases in the following order: NiMo/AlzO3 > NiMo/HNaY-AIzO3(P) > (NiMo/AIzO3)+HNaY(MM) > (NiMo/AlzOa)+HNaY(B) > NiMo/HNaY (Table 3). This activity order correlates well with the amount of the MoSz active sites determined by NO chemisorption. For the transformation of DBT, no clear effect of the presence of zeolite on the total conversion is observed. In the 4,6-DMDBT HDS reaction, the highest conversion was found for (NiMo/AIzO3)+HNaY(B) catalyst, followed by the (NiMo/AIzO3)+HNaY(MM) formulation. In this case, there is no clear correlation between the observed catalytic activity and the amount of the MoSz active sites determined by NO chemisorption, or the number of the accessible acid sites of the zeolite. It is known that in this reaction, both types of active sites, metal sulfide and acidic, participate in the transformations of the methyl-substituted DBT molecules [5, 6]. Products formed on one type of sites can be subsequently transformed on the other. Therefore, it could be supposed that for the best catalytic activity in the hydrodesulfurization of alkyl-substituted DBT derivatives, an optimum ratio between acid and metal sulfide functions must be attained in the catalyst. Additional information about the catalytic performance of the catalysts can be obtained from the analysis of the product distribution, which is affected by the metallic and acid functionalities. Tables 4 and 5 compare the product distributions obtained in the DBT and 4,6-DMDBT reactions with the NiMo/AI203, NiMo/HNaY and NiMo catalysts with 20% of HNaY in their formulation. In the case of DBT, zeolite incorporation into the catalyst changes the contributions of the direct desulfurization (DDS) pathway, which yields biphenyl-type compounds, and of the desulfurization through hydrogenation (HYD) pathway, which gives cyclohexylbenzene-type compounds. Also, the proportion of CHB in the reaction products and the liquid yield decrease with the number of accessible zeolite acid sites in the catalyst. This effect is due to the cracking of CHB on the zeolite acid sites. On the other hand, the formation of DCH is enhanced on the catalysts where Mo precursor phase is more polymerized (NiMo/HNaY-AIzO3(P) and NiMo/HNaY formulations). In the case of the 4,6-DMDBT the DDS pathway was much more inhibited giving rise to the formation of hydrogenated products in larger proportion. The participation of the acid function of the zeolite is evidenced by the formation of significant amounts of cracking products like toluene, benzene, cyclohexane and light products and in the appearance of different isomers of MCHT and DMDCH.
273
Table 3 NO chemisorption results and catalytic activity of NiMo catalysts in the DBT and 4,6DMDBT hydrodesulfurization reactions ........ DBT Conversi6n (%) Catalyst
pmol NO/g
NiMo/AI203 NiMo/HNaY- A1203 (P) (NiMo/AIzO3)+HNaY(B) (NiMo/AI203) +HNaY (MM) NiMo/HNaY
130.5 51.1 45.8 49.6 25.5
DMDBT Conversi6n (%)
4h*
8h
4h
8h
57.5 57.0 45.6 56.6 25.0
90.9 90.7 82.0 88.5 42.0
27.7 26.4 43.7 38.7 18.3
61.1 58.5 65.3 61.3 36.3
* Reaction time Table 4 Product distribution (wt. %)* and liquid yield in the DBT hydrodesulfurization reaction Catalyst Compound**
NiMo/HNaY (NiMo/AI203) NiMo/HNaY +HNaY(MM) -A1203(P)
NiMo/AI203
(NiMo/AI203) +HNaY(B)
BP CHB DCH CH BZ LP
71.32 27.72 0.96 -
81.50 16.11 1.25 0.22 0.74 0.19
74.51 12.63 2.78 1.85 4.32 3.93
82.88 8.99 1.84 0.83 2.59 2.87
34.95 35.61 6.00 23.44
Liquid yield (%)
95.8
87.3
84.8
82.5
80.0
* At 60% DBT conversion for all the catalysts with the exception of NiMo/HNaY where DBT conversion was 42%. ** BP, biphenyl; CHB, cyclohexylbenzene; DCH, dicyclohexyl; CH, cyclohexane; BZ; benzene; LP, light products (C3-C6) 4. C O N C L U S I O N S The results obtained in the present work indicate that the method used for the catalyst preparation leads to some changes in the characteristics and performance of the obtained catalytic formulations. These changes are due to differences in the characteristics of both the deposited metallic species (their coordination state, location and dispersion) and the contribution of the acidic component of the catalyst (number of accessible acid sites and partial destruction of zeolite structure). Therefore, the importance of the method used for the preparation of zeolite-containing catalysts is clearly observed.
274 Table 5 Product distribution (wt. %)* and liquid yield in the 4,6-DMDBT HDS reaction Catalyst Compound** NilVIo/AI203 NiMo/AI203 NiMo/HNaY NiMo/AI203+H NiMo/HNaY +HNaY(B) -AI203(P) NaY(MM) THDMDBT HHDMDBT DMBP MCHT DMDCH BP CHB DCH TL CH BZ LP
4.64 1.05 27.70 54.37 7.94 1.45 2.64 0.21 -
6.04 1.41 24.28 16.38 5.99 3.04 0.95 5.55 1.23 35.13
6.30 1.28 39.42 26.53 4.15 0.56 0.59 1.16 13.42 0.64 0.99 4.95
36.69 22.72 3.35 3.02 2.53 3.14 1.79 26.75
6.45 0.35 51.43 0.17 8.22 6.52 4.49 22.4
Liquid yield (%)
92.5
87.5
90.8
82.5
80.0
* At 60% of 4,6-DMDBT conversion for all the catalysts with the exception of NiMo/HNaY where 4,6-DMDBT conversion was 36%. ** THDMDBT, tetrahydrodimethyldibenzothiophene; HHDMDBT, hexahydrodimethyldibenzothiophene; DMBP, dimethylbiphenyl; MCHT, methylcyclohexyltoluene; DMDCH, dimethyldicyclohexyl; BP, biphenyl; CHB, cyclohexylbenzene; DCH, dicyclohexyl; TL, toluene; CH, cyclohexane; BZ, benzene; LP, light products (C3-C6) ACKNOWLEDGEMENTS Financial support by DGAPA-UNAM (grant IN-103599), DGEP-UNAM, CONACyTCSIC program and IMP-FIES program are gratefully acknowledged. We would like to thank M. Cecilia Salcedo for obtaining XRD patterns.
REFERENCES 1. K.G. Knudsen, B.H. Cooper and H. Topsoe, Appl. Catal. A: General, 189 (1999) 205. 2. R. Shaft and G.J. Hutchings, Catal. Today, 59 (2000) 423. 3. F. Bataille, J.-L. Lemberton, P. Michaud, G. P~rot, M. Vrinat, M. Lemaire, E. Schulz, M. Breysse and S. Kasztelan, J. Catal., 191 (2000) 409. 4. T. Isoda, S. Nagao, X.L. Ma, Y. Korai and Y. Mochida, Energy Fuels, 10 (1996) 482. 5. M.V. Landau, D. Berger and M. Herskowitz, J. Catal., 159 (1996) 236. 6. P. Michaud, J.L. Lemberton and G. P6rot, Appl. Catal. A: General, 169 (1998) 343.
275 7. T. Klimova, D. Soils, J. Ramirez and A. L6pez Agudo, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 127 (1999) 373. 8. A. Lopez Agudo, R. Cid, F. Orellana, J.L.G. Fierro, Polyhedron, 5 (1986) 187. 9. Y. Okamoto, Catal. Today, 39 (1997) 45. 10. C.A. Emeis, J. Catal., 141 (1993) 347. 11. R. L6pez Cordero and A. L6pez Agudo, Appl. Catal. A: General, 202 (2000) 23. 12. L. Portela, P. Grange and B. Delmon, Catal. Rev.-Sci. Eng., 37 (1995) 699.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
277
Chiral dirhodium catalysts confined in porous hosts H.M. Hultman a'b, M. de Lang a, M. Nowotny a, I.W.C.E. Arends b, U. Hanefeld a, R.A. Sheldon b, T. Maschmeyer a* Delft University of Technology, Applied Organic Chemistry and Catalysis a and Biocatalysis and Organic Chemistry b, Julianalaan 136, 2628 BL Delft, The Netherlands. By immobilising homogeneous chiral catalysts on the inner surface of porous solids, considerations like separability, re-use and selectivity may be addressed simultaneously [1]. Dirhodium carboxamide complexes, Rhz(MEPY)4 and Rhz(BNOX)4, were attached to the surfaces of MCM-41 and silica through ligand exchange with carboxylic acid functionalised tethers. Their activity was probed in the cyclopropanation of styrene with ethyl and tert-butyl diazoacetate, as well as in the Si-H insertion of dimethylphenylsilane with methyl phenyldiazoacetate. Improvements found in regio- and enantioselectivity are discussed in terms of steric constraints imposed by the surface. 1. I N T R O D U C T I O N Metal complexes have been immobilised in many different ways, e.g.: covalent anchoring (by grafting or tethering) to inorganic supports, immobilisation by occlusion in zeolitic micro- or mesopores (ship-in-a-bottle concept), or as supported liquid-phase catalysts [2]. In recent reviews the potential of (chiral) metal complexes immobilised by these different methods has been evaluated [3,4,5]. Although silica has flexible Si-OH groups that are readily functionalised, its large pore size distribution prevents any confinement effects from occurring to a significant degree. An alternative carrier structure is MCM-41 [6,7], which is a mesoporous silica or aluminosilicate material. It has large channels ranging from 15 to 100 *, ordered in an hexagonal array [8], which can be prepared with an almost uniform pore size. Because of its mesopores, MCM-41 offers new opportunities for the encapsulation of large catalyst species, and for the catalytic conversion of substrates too large to fit into zeolites [9]. In this study chiral dirhodium catalysts (Rh2(MEPY)4) and Rh2(BNOX)4), developed by Doyle [10] (see Scheme 1) were immobilised. It can be anticipated that the spatial constraints induced by the carrier (MCM-41 or silica), and especially by the pores of MCM-41, are able to increase the influence of the chiral ligands. Earlier research [11,12] showed that enantioselective reduction catalysed by a palladium complex immobilised inside the pores of MCM-41 resulted in a threefold increase in enantioselectivity compared to the homogeneous palladium complex. In order to immobilise the homogeneous catalysts on the surface, an organic linker group was * The NRSCC (H.H.) and the KNAW (U.H.) are acknowledged for financial support.
278 introduced. The (inner) surface silanol groups were readily functionalised with tethers bearing a carboxylic acid group. Immobilisation proceeded via exchange of a carboxamide ligand.
x N
Y ....
Y....
oi,, Y y _l./O1[..-a ' ' J ' / + \. /R,h~Hh E X ~ . . ~ O O NI Y "~"~X
--O~ MCM-41 _ v/ or SiO2
I O I..N ~,. fRh~ .Rh toluene y---N:,,....IIO I RT, 5 days L'xr ON / N
/O si_R_c ,,
--O
.....
y \
OH
Y
C I R I Si /IX OOO I I I
R = (0H2)2 R = (0H2)3 R = P'C6H4
L = MEPY; X=CH 2, Y=COOMe
MCM-41 or SiO2
L = BNOX; X=O, Y=CH2Ph
Scheme 1. Immobilisation of dirhodium complexes on different carriers By introducing three different types of tethers (-(CH2)2-, -(CH2)3-, and -p-C6H4-, scheme 1) their influence on the catalyst could be evaluated. All silanol groups were silylated after the introduction of the tethers and before the immobilisation of the catalysts. Chiral dirhodium carboxamide complexes have attracted considerable interest as enantioselective catalysts [13]. As model reactions the Si-H insertion of methyl phenyldiazoacetate (1) with dimethylphenylsilane (2) (reaction (1)) as well as the cyclopropanation of styrene (4) with diazoacetates (5a/b) (reaction (2)) were investigated (scheme 2).
[•/COOMe +
Me,, , , P h Me/Sill
N2 1
@
C--C + 4
catalyst
CH2Ci2 ~. N2, reflux
2
f?
(1)
3
cata,,st
N2CH--C--OR 5a/b a: ethyl, b: tert-butyl
Scheme 2. Catalytic test reactions
COOMe H
Me2PhSr
~ RT/reflux N2
,~, cis 6a/b
+
* COOR trans 6a/b
(2)
279 2. E X P E R I M E N T A L SECTION
2.1. Materials and methods All reactions and manipulations were performed under an atmosphere of dry nitrogen using standard Schlenk-type techniques. Silica sources for the MCM-41 synthesis were CabO-Sil M5 (fumed silica, Fluka) and a solution of sodium silicate (14% NaOH, 27% SiO2, Aldrich). All other reagents were purchased from Aldrich, Acros or Baker and were used without further purification. Trans/cis ratios for cyclopropanes were determined by GC analysis using a CP-wax 52 CB column (50 m*0.53 mm, 2.0 om (film thickness)). Chiral GC analysis for cyclopropanes was performed using a B-DA or B-PH column (40m*0.25mm) at 110~ HPLC analysis for Si-H insertions was performed using a chiral OD-column (25 cm), using 98/2 hexane/isopropanol at 1 ml/min, with 254 nm UV detection. Rhodium contents were analysed by ICP-OES after dissolving the solid samples in l%v/v HF and 1.3%v/v H2804 in water. Loading of the COOH-tether was determined by CHN-elemental analysis of the corresponding CN-tether. Full conversion from CN to COOH was determined by IR analysis on a Perkin Elmer Spectrum One FT-IR spectrometer. BET surface analysis was performed by N2 adsorption at 77 K on a Quantachrome Autosorb-6B after drying the samples at 200~ in vacuum. XRD patterns were recorded using CuKz radiation on a Philips PW 1840 diffractometer equipped with a graphite monochromator. The samples were scanned in the range of 0.105 to 50.005~
2.2. Catalyst Preparation General synthesis of carrier with acid tether: 4.08 g MCM-41 (prepared according to Beck et al. [7] was activated at 200~ in vacuo for two hours. The outer surface silanol groups (10 % of the total amount of silanol groups) were protected by reaction with dimethoxydimethylsilane (0.197 g, 1.63 mmol) in refluxing toluene (40 ml). After 3 hours 1.04 mmol 4-(trichlorosilyl)butyronitrile or 3-(triethoxysilyl)propionitrile (equivalent to 10 % of the theoretical number of surface silanol groups) was added and the mixture was refluxed overnight. The remaining (inner surface) silanol groups were then protected by addition of dimethoxydimethylsilane (2.30 g, 0.0192 mol), and the mixture was refluxed for three hours. The solid was filtered off, washed with water and ethanol and dried at 100~ in vacuo. Hydrolysis of the nitrile was achieved by addition to 50% aqueous sulphuric acid and heating for two hours at 150~ After filtration, the solid was washed with water until the filtrate was neutral. It was then dried overnight at 80~ in vacuo. For silica samples (Aerosil 200) the same procedure was applied, only the second step (protection of the outer surface) was omitted. General immobilisation of Rh-complex: MCM-41-(CHz)3COOH (0,0397 g, 3,14"10 .5 mol carboxylic acid groups) and Rhz(5R-MEPY)4 (0,0235 g, 3,03"10 .5 mol) were stirred at room temperature for two days in toluene (6 ml). The resulting solid product was Soxhlet extracted with dichloromethane until the washings were colourless and then dried in vacuo.
2.3. Catalytic Procedures General Cyclopropanation Procedure: 0.2 g chlorobenzene, 3 ml dichloromethane and 4 (0.490 g, 4.70 mmol) were added to the catalyst (1 mol% maximum Rh-loading) and the mixture was stirred. Over a period of 3-5 hours a solution of 5a (0.0517 g, 0.453 mmol) or 5b (0.0599 g, 0.461 mmol) in 3 ml dichloromethane was added. After stirring overnight at room temperature (5a) or under reflux (5b), the solvent was evaporated in vacuo and the residue was chromatographed (silicagel, hexane/ethyl acetate 9/1). Before chiral GC analysis,
280 6a was converted into the corresponding methyl esters by treatment with a 0.1 molar solution of NaOH in MeOH. General Si-H insertion procedure: 0.2 g 1,2-dichlorobenzene and 1 ml dichloromethane were added to the catalyst (2 mol% maximum Rh-loading) and the mixture was stirred. Subsequently 1 (0.0973 g, 0.56 mmol) in 0,5 ml dichloromethane and 2 (0.0832 g, 0.62 mmol) in 1 ml dichloromethane were added. The resulting mixture was refluxed overnight. After evaporation of the solvent in vacuo, the reaction mixture was chromatographed (silicagel, petrolether/ethyl acetate 19/1). Leaching test: The cyclopropanation reaction was performed according to the general procedure. After stirring overnight at room temperature, the solid catalyst was allowed to settle. The supernatant solution was transferred to another vial through a syringe filter to remove traces of immobilised catalyst. 5a (0,2195 g, 1,92 mmol) was added to the filtrate and two minutes later a GC sample was taken. The mixture was left to stir overnight. Samples were taken after 15,5 h and 87,5 h. Recycling test: The cyclopropanation reaction was performed following the general procedure. After stirring overnight at room temperature, the solid catalyst was allowed to settle. The supernatant solution was transferred to another vial through a syringe filter to remove traces of immobilised catalyst. The rhodium content of this solution was determined by AAS. The trans/cis ratio of the products and the conversion were determined by GC. The remaining solid was washed with dichloromethane and dried. It was then used again following the same procedure. After three cycles, the rhodium content of the catalyst was determined by ICP OES. 3. RESULTS AND DISCUSSION 3.1. Catalyst preparation The preparation procedures of the carriers are outlined in Scheme 3. The alkyl tether groups were attached to the surface by the reaction of 3-(triethoxysilyl)propionitrile or 4(trichlorosilyl)butyronitrile with MCM-41 or silica. The nitrile function was hydrolysed with 50% aqueous sulphuric acid. The -p-C6H4COOH tether group was prepared by treating MCM-41 or silica with [4-(dimethoxymethyl)phenyl]trimethoxysilane. This acetal was then hydrolysed with 5% aqueous trifluoroacetic acid. This aldehyde, thus released, was finally oxidised with peracetic acid. Subsequently, the residual surface silanol groups were protected with dimethoxydimethylsilane. From C,H,N analysis of the CN-tether, the loading could be determined. IR analysis of the COOH-tether showed that no CN-tether was present after hydrolysis. CP MAS solid state NMR confirmed the presence of the tethers on the surface of the carriers. The analysis of the immobilised complexes by CP MAS NMR is hampered by the low loading of the complex on the carrier. Only very broad peaks with low intensity were observed. In order to validate the exchange of ligand as a way of immobilisation on the carriers, the model reaction of Rh2(5S-MEPY)4 with one equivalent acetic acid was followed by liquid 1H NMR [14]. Initially, the carboxylic acid exchanged with the axial ligands. This reaction was followed by exchange of chiral ligand with the acid. FAB-MS analysis indicated that more than one chiral ligand might be exchanged with an acid ligand. The theoretical maximum loading of dirhodium catalyst on the carrier (Table 1) was calculated from the loading of the CN-tether (determined by C,H,N analysis). Actual rhodium loadings are usually around 70% of the CN-loading, but drop to 15% in some cases.
281
-OH ~ ,O1~ MCM-41 --OH + (MeO)3Si---k~,~)/~---C,` -OH ~ OMe Silica
-OH surface --OH + --OH
___/~/CN
(EtO)3Si
,_ peracelJ&acid
,.~
"-CF3CO~
H20
H2SO ~ ~
"
"MCM-41
Silica
7si
_/~COOH
surface
Scheme 3. Synthesis of the different functionalised carriers Table 1. Immobilised catalysts used in this study with the maximum loading of dirhodium complex entry catalyst maximum loading (mmol catalyst/g) 1 MCM-41- (CH2) 2COO-Rh2 (5S-MEPY) 3 0,81 2 Si02-p-C6H4COO-Rhz(5S-MEPY)3 0,12 3 SiO2- (CH2) 2COO-Rh2 (5S-MEPY) 3 0,064 4 5
MCM-41 -p-C6H4COO-Rh2 (4S-BNOX) 3 SiOz- (CH2) 2COO-Rh2 (4S-BNOX) 3
0,80 0,085
6 7 8 9
MCM-41- (CHz)zCOO-Rh2 (4R-BNOX) 3 SiO2-p-C6H4COO-Rh2 (4R-BNOX) 3 SIC2- (CHz)2COO-Rha (4R-BNOX)3 SIC2- (CH2)3COO-Rh2 (4R-BNOX) 3
0,81 0,12 0,064 0,057
The nitrogen desorption measurements showed that the surface area, the total pore volume and the pore size decreased after attachment of the tether (Table 2). The XRD plots (Scheme 4) indicate, together with the N2 desorption experiments, that the structure of MCM41 remained intact when the CN-tether was introduced. The decrease in pore volume can be attributed to the presence of the tether inside the pore. However, after heating in aqueous sulphuric acid, the pore structure was partially damaged. However, 60% of the channel structure remained intact (determined by the decrease of the total pore volume, see Table 2, entry 2 and 3). Table 2. Results N2 desorption measurements entry sample ................... SBET(m2/g) 1 MCM-41 960 4- 13 2 MCM-41-(CH2)3CN 825 + 21 3 MCM-41-(CH2)3COOH 566 + 10
t0ial' pore vo]um'e""(cm3/g) 1.01 0.78 0.41
pore size (nm) 2.4 2.1 1.9
282
I
- - - - MCM-41 MCM-41-(CH2)3CN
8
--
10
20
30
40
MCM-41-(CH2)3COOH
50
2 theta
Scheme 4. Comparison XRD plots of MCM-41, (CHz)3COOH
MCM-41-(CHz)3CN and MCM-41-
3.2. Si-H insertion test reactions
In the Si-H insertion reaction (reaction 1) significant differences can be observed between the homogeneous and immobilised catalysts (Table 3). The homogeneous catalysts display no (entry 1) or low (entry 6) enantioselectivity. In the case of the catalysts immobilised on silica, however, the selectivities increased more than 10-fold (entries 2,3). This is clear evidence that despite the loss of one chiral ligand due to the method of immobilisation, the spatial confinement leads to a significant improvement of enantioselectivity. In contrast to the catalysts immobilised on silica, none of the catalysts immobilised inside MCM-41 showed significant activity. Even after refluxing overnight, large amounts of unmodified 1 remained. Possibly there is not enough space inside the pores of MCM-41 for the reaction to take place. Indeed the average pore diameter is 19 * (determined by nitrogen desorption analysis) and the catalyst size is similar (approximately 15 A). A transition state requiring a space demanding conformation might therefore be too constrained under these circumstances. Table 3. Results Si-H insertion reaction (1) entry catalyst Rhz(4R-BNOX)4 SiO2- (CH2) 2COO-Rh2 (4R-BNOX) 3 SiOz-p-C6H4COO-Rhz(4R-BNOX)3 MCM-41-(CHz)zCOO-Rhz(4R-BNOX)3 MCM-41-p-C6H4COO-Rhz(4S-BNOX)3 6 7 8
Rhz(5S-MEPY)4 SiO2-(CHz)zCOO-Rhz(5S-MEPY)3 Si02-p-C6H4COO-Rhz(5S-MEPY)3
yield of 3 (%) 73 88 67 only traces of product detected only traces of product detected
ee (%) 2 20 28 -
70 78 65
37 2 1
.................
Recycling experiments were performed for the Si-H insertion (reaction 1). Since all catalysts were immobilised in the same manner it can be assumed that these results are
283 representative for all the different immobilised catalysts described here. The yield of 3 decreased from 79% in the first cycle, to 31% in the second and 19% in the third cycle. After the third cycle, the liquid phase was removed from the solid catalyst and its activity was investigated. This solution did not catalyse the Si-H insertion. A rhodium analysis of the catalyst (SiO2-(CH2)2COO-Rh2(4S-BNOX)3) showed that no rhodium was left on the carrier. Further studies into the stability of this catalytic system are underway.
3.3. Cyclopropanation test reactions In the case of the cyclopropanation reaction (reaction 2) the trans/cis ratio is determined by the steric repulsion between the phenyl group of the styrene and the ester group of the intermediate carbene. In the homogeneous reaction, the small ligand does not significantly restrict the incoming styrene, and thus the difference in the formation of trans and cis product is not very large (Table 4, entries 1,4,7,10). In the heterogeneous reaction the steric hindrance by the bulky carrier surface forces the carbene slightly out of plane, away from the carrier surface. This, and the bulk of the ester group, directs the incoming styrene in such a way that more trans compound is formed than in the homogeneous reaction. If the more bulky tert-butyl diazoacetate (TBDA, 5b) is used instead of ethyl diazoacetate (EDA, 5a), both effects are even more pronounced. If the catalysts are immobilised on silica the effect is smaller than if the catalysts are immobilised inside MCM-41. This is due to the confinement in the pores that restricts the incoming styrene even more than the silica surface. A loss in enantioselectivity can be attributed to the loss of one chiral ligand during the immobilisation. The influence of the tether group was not significant. Only minor differences in selectivity were obtained. Table 4. Comparison EDA and TBDA inthe Cyc!0pr0panation reaction (2) entry catalyst diazo compound
yield 6a/b
trans/cis
(%) 1 2 3
Rhz(5S-MEPY)4 SiO2- (CH2)2COO-Rh2 (5S-MEPY) 3 MCM-41-(CHz)2COO-Rh2(5S-MEPY)3
5a 5a 5a
59 73 65
56/44 59/41 60/40
4 5 6
Rh2(5S-MEPY)4 SiO2- (CH2)2COO-Rh2(5S-MEPY) 3 MCM-41- (CH2)2COO-Rh2(5S-MEPY) 3
5b 5b 5b
50 62 50
60/40 71/29 74/26
7 8 9
Rh2(4R-BNOX)4 SiO2- (CH2) 2COO-Rh2(4R-BNOX) 3 MCM-41-(CH2)2COO-Rh2(4R-BNOX)3
5a 5a 5a
79 84 51
46/54 60/40 70/30
10 11 12
Rh2(4R-BNOX)4 SiO2- (CH2) 2COO-Rh2 (4R-BNOX) 3 MCM-41- (CH2) 2COO-Rh2 (4R-BNOX) 3
5b 5b 5b
64 53 51
59/41 66/34 72/28
3.4. Cyclopropanation leaching and recycling In order to evaluate whether any active rhodium complex leaches from SiO2(CH2)aCOO-Rh2(4R-BNOX)3 during the cyclopropanation reaction (reaction 2), the filtrate of a reaction mixture was tested for its catalytic activity. This filtrate displayed modest activity:
284 34% of the diazo compound was still present after 16 hours, decreasing to 11% after three more days. In comparison, in a typical cyclopropanation experiment, the conversion was already 100% after one minute. The activity of the filtrate is therefore approximately only 0.5% of the activity of the immobilised catalyst. Only a very small amount of catalytically active material leaches during reaction and the actual catalysis is performed by the heterogenised complex. In addition, we performed recycling experiments for the same reaction (SiO2C6H4COO-Rhz(5S-MEPY)3 as catalyst). The results (Scheme 5) show that the second and third cycle proceed more slowly. However, after stirring overnight, all diazo compound is consumed. From the results of the leaching and recycling experiments combined with rhodium analysis of the catalyst before and after recycling, it can be concluded that the catalyst remains active even after recycling for three times. However, 50% of the rhodium leached.
] 9
I.u,.. 40 60 . ~ "~
m -,,,_ cycle1 - -~ - cycle
o 200
0
i
200
i
400
i
600
l
800
i
1000 1200
reaction time (minutes)
Scheme 5: Recycling of SiO2-C6H4COO-Rh2(5S-MEPY)3 in the cyclopropanation of styrene with EDA 4. C O N C L U S I O N The immobilisation of the homogeneous dirhodium catalysts on silica surfaces and inside the pores of MCM-41 affords a significant improvement in regioselectivity (cyclopropanation reaction) and enantioselectivity (Si-H insertion) of these catalysts. The improvement is attributed to the confinement resulting from immobilisation. From leaching tests it was established that the reactions performed were indeed heterogeneous. The immobilised catalysts have been successfully re-used for three times in the cyclopropanation reaction. In the Si-H insertion the catalyst however, is less stable. The increased temperature utilised for this reaction is probably responsible for this difference. REFERENCES
1. J.M. Thomas, T. Maschmeyer, B.F.G. Johnson and D.S. Shephard, J. Mol. Catal. A, Chem., 141 (1999)139. 2. H.H. Wagner, H. Hausmann and W.F. H61derich, J. Catal., 203 (2001) 150 and references cited therein.
285 3. A. Baiker, Curr. Opin. Solid State Mater. Sci., 3 (1998) 86. 4. T. Bein, Curt. Opin. Solid State Mater. Sci., 4 (1999) 85. 5. K. Fodor, S.G.A. Kolmschot and R.A. Sheldon, Enantiomer, 4 (1999) 497. 6. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartulli and J.S. Beck, Nature, 359 (1992) 710. 7. J.S. Beck, J.C. Vartulli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T-W. Chu, D.H. Olsen, E.W. Sheppard, S.B. McCullen, J.B. Higgins and J.L. Schlenker, J. Am. Chem. Soc., 114 (1992) 10834. 8. A. Corma, Chem. Rev., 97 (1997) 2373. 9. C. Huber, K. Moiler and T. Bein, J. Chem. Soc., Chem. Commun., (1994) 2619. 10. M.P. Doyle, W.R. Winchester, J.A.A. Hoorn, V. Lynch, S.H. Simonsen and R. Ghosh, J. Am. Chem. Soc., 115 (1993) 9968. 11. B.F.G. Johnson, S.A. Raynor, D.S. Shephard, T. Maschmeyer, J.M. Thomas, G. Sankar, S. Bromley, R. Oldroyd, L. Gladden and M.D. Mantle, Chem. Commun., (1999) 1167. 12. S.Raynor, J. Thomas, R. Raja, B. Johnson, R. Bell and M. Mantle, Chem. Commun., (2000) 1925. 13. M.P. Doyle, in: Comprehensive Organometallic Chemistry II. A review of the literature 1982-1994, E.W. Abel, F.G.A. Stone, G. Wilkinson (Eds), Volume 12, Transition Metal Organometallics in Organic Synthesis, Elsevier Science Ltd., Pergamon Press, 1995. 14. Manuscript in preparation.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
287
Synthesis and characterization of zeolite encaged enzyme-mimetic copper histidine complexes J. Gerbrand Mesu*, Debbie Baute*, Henk J. Tromp*, Ernst E. van Faassen* and Bert M. Weckhuysen *'w Department of Inorganic Chemistry and Catalysis, Debye Institute, Utrecht University, P.O. Box 80083, 3508 TB, Utrecht, The Netherlands * Centrum voor Oppervlaktechemie en Katalyse, Departement Interfasechemie, K.U. Leuven, Kardinaal Mercierlaan 92, 3001 Leuven, Belgium * Interface Physics, Debye Institute, Utrecht University, P.O. Box 80000, 3508 TA, Utrecht, The Netherlands Histidine was complexed with copper cations and immobilized in zeolite Y by an ion exchange procedure. The influence of the initial copper concentration in the ion exchange solution and the ion exchange time on the structure of the final zeolite encaged copper complexes was studied. Two different copper complexes were found on the zeolite: a mono-histidine complex (A) and a bis-histidine complex (B). The complex ratio A/B decreases with increasing copper loading in the ion exchange solution. The dynamics of the exchange was also studied. The A/B ratio does not change during this process. The exchange process itself is very fast, as it takes place within an hour. After that the it is slowed down by orders of magnitude. 1. I N T R O D U C T I O N Histidine, a naturally occurring amino acid, was complexed with copper cations and incorporated inside the supercages of zeolite Y [1,2]. The obtained complexes mimic the active center of natural enzymatic counterparts, such as galactose oxidase. The pore geometry of the zeolite induces shape selectivity in reactions and allow for intra-particle transport of reactants and products. The zeolite host material also induces additional stability of the incorporated active center, thereby expanding the range(s) of operating temperatures and pressures as well as solvents. The zeolite encaged copper histidine (Curtis) complex has already been shown to exhibit promising catalytic activity for the epoxidation of alkenes with peroxides [3]. ESEEM experiments on the zeolite occluded Curtis systems showed the presence of (at least) two different Curtis complexes (A and B) on the zeolite [4,5]. Complex A exhibits 27A1 modulations, which indicates that the Cu(II) coordinates to zeolite framework wCorresponding author, e-maih
[email protected] Tel: +31-30-2534328, Fax: +31-30-2511027
288 oxygen. Complex B, however, shows no 27A1 modulation. Complex A is proposed to be a mono-histidine complex where both the amino and imino nitrogens of the histidine are coordinated to Cu 2+, whereas the other equatorial ligands are provided by a zeolite oxygen and a water molecule. The complex is stabilized by the presence of protons in the zeolite framework. Complex B is proposed to be a bis-histidine complex, situated in the center of the supercage. The two histidine molecules coordinate to Cu 2+ in a square planar geometry: the amino and imino nitrogens of one histidine molecule and the imino nitrogen and carboxylate oxygen of the second histidine molecule. A number of parameters might affect the A/B ratio. These parameters are: 1. The pH of the exchange solution; 2. The copper concentration in the ion exchange solution; 3. Duration of the ion exchange. The effect of the first parameter has been discussed in a previous paper [4]. In this paper, we investigate the effect of the other two parameters. Once it is possible to set the A/B complex ratio by tuning the synthesis conditions during the ion exchange procedure, the next step is to relate this ratio to catalytic yield and/or selectivity of the resulting heterogeneous Cu(II)-catalyst. 2. E X P E R I M E N T A L
2.1. Preparation Aqueous solutions of Curtis complexes, prepared in bi distilled water with a His:Cu(II) ratio of 5:1 at pH 7.3 were used for ion exchange with NaY (ZEOCAT, Si:A1 = 2.71). A series of zeolite samples, differing in their amount of Curtis complexes, were prepared using solutions with different copper concentrations (0.1, 0.25, 0.50, 1.0, 1.5 and 4.5 copper/unit cell (Cu/UC)), while keeping the His:Cu(II) ratio in the solution at 5:1 and the pH at 7.3. The pH was adjusted with 0.1 M NaOH and/or 0.1 M HC1 solutions. All samples were stirred for 24 hours at room temperature. The pH of the exchange solution was measured regularly and adjusted if needed. All samples were dried at 333 K after washing and filtration. The duration of ion exchange affects the copper concentration of the zeolite: longer exchange times lead to higher copper concentration. The exchange dynamics was studied by taking a series of zeolite samples from the ion exchange solution at different points of time. 2.2. Characterization CW-EPR X-band measurements were performed on a Bruker ESP 300E Spectrometer at a temperature of 120 K. The Curtis complexes are paramagnetic due to the S=1/2 spin of the Cu 2§ ion. Nitrogen physisorption was performed with a Micromeritics ASAP 2400 apparatus. Measurements were done at 77 K. Prior to the measurements the zeolite samples were degassed for 24 hours at 373 K in vacuum. Micropore volumes and pore size distributions were determined with standard BET and BJH theory. Diffuse Reflectance Spectroscopy of the Curtis complex encapsulated zeolite samples were taken on a Varian Cary 5 UV-Vis-NIR spectrophotometer at room temperature. The DRS spectra were recorded against a halon white reflectance standard in
289 the range 2500-200 nm. Atomic Absorption Spectrometry (AAS) measurements for quantitative analysis of Cu 2+ in the zeolite samples were performed using an Instrumentation Laboratory Inc. apparatus with a nitrous oxide-acetylene flame. Measurements were done at a wavelength of 324.7 nm using a hollow cathode lamp. The amount of Cu 2+ was determined after dissolution of known quantities of ion-exchanged zeolite materials in HF/HzSO4. 3. RESULTS AND DISCUSSION The X-band CW-EPR spectra, recorded at 120 K, are shown in Fig. 1. The EPR intensity is proportional to the number of copper ions taken up by the zeolite during the ion exchange. Each of the spectra consists of two distinct EPR subspectra, which can be attributed to two different complexes, viz. complex A and complex B.
.A 4.5 Cu."lJC 1.5 Cu/UC 1.0 Cu/UC 0.5 CLdt.JC 0.25 Cu/UC O. 1 Cu!UC
2.~
2.75
3.0
3.25
3.S
3.75
B (kG)
Fig. 1. CW-EPR spectra as a function of the external copper concentration in the ion exchange solution. A change in the relative amounts of these two subspectra becomes visible upon going to higher copper concentrations in the ion exchange solution. At low copper concentrations only the subspectrum of complex A is visible, but at higher copper concentrations also the subspectrum of complex B appears. The shape of the spectrum is particularly sensitive to the values of g//and A//. The larger g//and smaller A//of complex A indicate lower density of the unpaired electron at the site of the copper nucleus. It suggests that the electronic orbital is affected by fewer nitrogen atoms in the first coordination sphere of the Cu(II) ion compared to complex B. The EPR spectra can be simulated as a superposition of two different subspectra: A respectively B. The shape of each sub spectrum is chosen as a molecule with an axial Zeeman interaction plus an axial hyperfine interaction to the I = 3/2 copper nucleus.
290 Table 1.
Calculated EPRparame!ersof the zeolite-Y enc._a,~.s.u,!atedCuH~ c_0.mp!exes_ Complex A ComplexA ComplexB ComplexB gll
All
gll
All
2.32
154
2.27
173
100
o~
8O
o~ 6O
-'~'-Complex A ""D-Complex B
t~ t-
0 L
o.
40 20
O~ 0.0
2.0
4.0
Cu/UC in exchange solution
Fig. 2. Relative amounts of complex A and complex B on the zeolite as a function of the copper concentration in the ion exchange solution (estimated from EPR spectra). Hyperfine couplings to nitrogen are omitted because they are experimentally not resolved. The calculated values for g//and AH for both complexes are presented in Table 1. The experimental accuracy did not require simulation as a mixture of 63Cu/65Cuisotopes. The relative amounts of complex A and complex B on the zeolite can be estimated from the EPR spectra in Figure 1. These amounts are depicted in Fig. 2 as a function of the copper concentration in the ion exchange solution. At the lowest copper concentration only complex A is found on the zeolite. With increasing copper concentration in the ion exchange solution an increasing amount of complex B is found, with a maximum of approximately 40 % for the highest copper concentrations. The amount of copper in the zeolite was measured by quantitative analysis. The results are presented in Fig. 3 as a function of the copper concentration in the ion exchange solution. The EPR intensity also gives an indication of the amount of copper exchanged onto the zeolite. These results are also depicted in Fig. 3. At low copper concentrations (up to 1 Cu/UC) the amount of Cu(II) exchanged onto the zeolite increases linearly with the copper concentration. In this region, all the copper in the exchange solution is deposited on the zeolite. At higher copper concentrations in the exchange solution (above 1 Cu/UC), almost no extra copper can be deposited on the zeolite. An explanation for the observed behavior might be the congestion of the zeolite crystals by immobilized histidine or Curtis complexes in the outer pore system and supercages of the zeolite crystals.
291
07
2.0 j~) r
0
o6 x5
1.5 ~ m=
~4 c3
1.o
d~
~."
,-2 re1
0.5:3
LUO
o.o
.m=
0.0
O~
~"
1.0 2.0 3.0 4.0 Cu/UC in exchange solution
Fig. 3. EPR intensity and quantitative analysis (in Cu/UC) of the zeolite occluded copper complexes as a function of the copper concentration in the exchange solution.
0.35 0.34 0.33 "~ 0.32 o~, 0.31 .o 0.30 0 0.29 0.28 ~: =
0
1
2
3
4
Cu/UC in exchange solution Fig. 4. Micropore volume as a function of the copper concentration in the ion exchange solution. The process of ion exchange will affect the available pore volume inside the zeolite. The micropore volume has been measured by N2 physisorption. The N2 physisorption isotherms of the Curtis loaded zeolite samples are of Langmuir type I. The evolution of the micropore volume as a function of the initial copper concentration in the ion exchange solution is presented in Figure 4. The micropore volume decreases from 0.34 ml/g for a pure Y zeolite to 0.29 ml/g for the highest copper loading. The curve displays a bend in going from an initial copper concentration of 0.25 Cu/UC to an initial copper concentration of 0.50 Cu/UC. This phenomenon may be attributed to an increase in the relative amount of the more bulky complex B in the pore system of the zeolite.
292 E = E
760
"~
740
E ~,
720
o c
700
~0
680
" 99 >99 >99 >99
31 26 17 12 34 32 19 21
48 48 43 43 66 68 81 79
0 0 0 0 0 0 0 0
Select. (L/B) 2.4 2.4 2.5 2.5 2.0 2.1 1.2 1.9
Rh Leach
(%) 59
43
303
Fig. 2. Nonanal conversion over the heterogeneous catalysts
4. CONCLUSIONS Activated carbons can be effective supports for the heterogenization of Rh complexes to produce active heterogenized catalysts for hydroformylation. The carbon functionalization to create a ligand-support bond has shown to be a very promising method to give active and stable catalysts. With this kind of catalysts it is possible to obtain a nonanal conversion similar to that of Rh(COD) in homogeneous phase, even after 4 consecutive rims. ACKNOWLEDGMENTS
This study was made possible by the financial support from CICYT PB98-0983 and NSERC of Canada. REFERENCES
1. E. Lindner, F. Auer, A. Baumann, P. Wegner, H.A. Mayer, H. Bertagnoli, U. Rein6hl, T.S. Ertel and A. Weber, J. Mol. Catal. A: Chem. 157 (2000) 97. 2. J.M. Basset, J.P. Candy and C.C. Santini in: Transition Metals for Organic Synthesis, Vol. 2, p. 387, Eds. M. Belier, C. Bolm, Wiley-VCH. (1998), Weinheim (Germany). 3. V.A. Likholovov and B.L. Moroz, in: G. Ertl, H. Kn6zinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis, Vol. 5, p. 2231, Wiley-VCH Verlag Weinheim (Germany), 1997. 4. M. Lenarda, L. Storaro and R. Ganzerla, J. Mol. Catal. A: Chem., 111 (1996). 5. J.P. Arhancet, M.E. Davis, J.S. Merola and B. Hanson, Nature, 339 (1989) 454.
304 6. M. Iglesias-Hemhndez and F. S~nchez-Alonso in: Studies in Surface Science and Catalysis 1340, A. Corma, F.V. Melo, S. Mendioroz and J.L.G. Fierro (Eds.), Vol. 130D, p. 3395, Elsevier Science B.V., Amsterdam (The Netherlands), 2000. 7. J. Bahie and J.C. Bay6n, J. Mol. Catal. A: Chem., 137 (1999) 193. 8. K. Nozaki, F. Shibahara, Y. Itoi, E. Shirakawa, T. Ohta, H. Takaya and T. Hiymna, Bull. Chem. Soc. Jpn., 71 (1999) 1911. 9. A.M. Tzreciak and J.J. Zi61kowski, J. Mol. Catal., 88 (1994) 13. 10. A.J. Seen, A.T. Townsend, J.C. Bellis and K.J. Cavell, J. Mol. Catal. A: Chem., 149 (1999) 233. 11. R. M. Desphande, Pm-vvanto and H. Delmas, Ind. Chem. Res., 35 (1996) 3927. 12. V.L.K. Valli and H. Alper, Chem. Mat., 7 (1995) 359. 13. Z. Zhou, G. Facey, B.R. James and H. Alper, Organometallics ,15 (1996) 2496. 14. L.R. Radovik and F.R. Reinoso in: Peter A. Thrower (Ed.), Chemistry and Physics of Carbon, Vol. 25, p. 243, Marcel Dekker, New York (USA), 1997. 15. J.A. Diaz-Aufi6n, M.C. Rom~-Martinez and C. Salinas-Martinez de Lecea, J. Mol. Catal. A: Chem., 170 (2001) 81. 16. H. P. Boehm, High Temperatures-High Pressures, 22 (1990) 275 17. H.E. Van Dam and H. Van Bekkmn, J. Mol. Catal., 131 (1991) 335. 18. D. Briggs and M.P. Seah in: Practical Surface Analysis, Vol. 1, John Wiley and Sons, Chiechester (UK), 1993. 19. J. Hagen, in: Industrial Catalysis. A Practical Approach, p. 17, Wiley-VCH Verlag, Weinheim (Germany), 1999. 20. H. Alper and J-Q. Zhou, J. Chem. Soc., Chem. Commtm., (1993) 316. 21. M.P. Anderson and L.H. Pignolet, Inorg. Chem., 20 (1981) 4101. 22. S.C. Bourque, H. Alper, L.E. Manzer and P. Arya, J. Am. Chem. Sot., 122, (2000) 956. 23. C.M. Crudden, D. Allen, M.D. Mikoluk and J. Sun, Chem. Commun., (2001) 1154. 24. H.P. Boehm, Carbon, 32 (1994) 759. 25. M.T. Reetz and S.R. Waldvogel, Angew. Chem. Int. Ed. Engl., 36 (1997) 8.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
305
Heterogeneous metathesis initiators I M. Mayr, B. Mayr, M. R. Buchmeiser* Institute of Analytical Chemistry and Radiochemistry, University of Innsbruck, Innrain 52 a, A-6020 Innsbruck, AUSTRIA The synthesis of heterogeneous N-heterocyclic carbene- (NHC-) based metathesis initiators is described. Two entirely different approaches have been developed. The first consists of a "grafting from" approach, were polymerizable NHC-precursors have been grafted onto a norborn-2-ene (NBE) based monolithic support prepared via ring-opening metathesis polymerization (ROMP), taking advantage of the living character of ROMP. The second synthetic route is based on a "grafting to" approach and entails the synthesis of oligomeric NHC-precursors and their selective chain-end functionalization with tri(ethoxy)silane groups. These telechelic polymers were grafted on silica using standard silane chemistry. All heterogenized NHC precursors were successfully converted into the corresponding NHC-based second generation Grubbs catalysts and used for various metathesis reactions including ROMP, RCM and cross-metathesis. 1. I N T R O D U C T I O N Metathesis-based reactions represent valuable tools in synthetic organic chemistry, polymer chemistry and technology. So far, a broad range of well-defined homogenous systems including those for asymmetric synthesis is available.[1-3] In contrast to these well-defined homogeneous systems, only few reports exist on analogous, stable, permanently immobilized heterogeneous systems prepared by a molecular approach.[4-6] In this lecture, the synthesis of new, well-defined heterogeneous metathesis systems will be reported. We already reported on the synthesis of heterogeneous C-C coupling and ATRP systems[7-9]. In contrast to the grafting and precipitation polymerization techniques developed for the synthesis of these materials, two entirely different approaches were applied for the fabrication of new heterogeneous NHC-based metathesis catalysts. While other groups focus on the alkylidene moiety of metathesis initiators for immobilization purposes[10-13], we solely use NHCs for this goal since they are the most strongly bound ligands in these systems. In order to minimize polymer-analogue transformation to a minimum, we generally pursue a concept where entire ligands or at least their immediate precursors, which can be converted into the desired systems by a few simple synthetic steps, are attached to a carrier. This ensures a maximum analogy to the parent homogeneous systems and allows the direct comparison of catalytic data. The concepts for immobilization as well as selected results shall be outlined briefly in the following.
1 Grant number Y-158 provided by the FWF (Austrian Science Fund), Vienna, AUSTRIA.
306
2. RESULTS AND DISCUSSION 2.1 Heterogeneous metathesis catalysts based on monolithic supports Monolithic supports used for the present application consist of one piece and possess a permanent and interconnected porous structure. In principle they can be either inorganic (e.g. silica) or organic (e.g. PS-DVB). Nevertheless, in order to provide both maximum chemical stability and an easy access to functionalization, we already developed a completely new class of monolithic supports based on norborn-2-ene (NBE) and a NBEderived cross-linker[14-16]. These monolithic media are generally characterized by a high mass transfer within the interphase, which allows to run catalytic reactions in a continuous flow set-up at significantly elevated flow rates of up to 10 - 20 mm/s. They are synthesized within the confines of the reactor in a one-pot reaction procedure. Since the Ru-based initiator C12Ru(CHPh)(PCy3)2, which allows a "living" setup, is used during synthesis, the active catalytic sites can be used for derivatization purposes after synthesis of the support.
inner surface
1. n / ~
O
o CY3C
~' ~ .9
~~cy~
~2.~,,
~--x~
R-N~N- R I
o-~
BF4O_ /
i M,,c:z
H
FFN~N-R monolith
~v
1. base 2. CI2Ru(PCY3)2(CHPh)
~,
~
H
BF4-
O R..N~L--~N_R PCY3
Scheme 1. Surface-derivatization precursors.
of monolithic supports with polymerizable NHC-
In order to generate sufficient porosity, monoliths with a suitable microporosity (40 %) and microglobule diameter (1.5 + 0.5 lam) were synthesized. Consecutive ,,in-situ" derivatization was successfully accomplished using a mixture of norborn-2-ene and the corresponding NHC-precursor in methylene chloride (Scheme 1). The use of norborn-2ene significantly enhances grafting yields for the functional monomer. Using this setup,
307 tentacles of copolymer with a degree of oligomerization of 2 - 5 of the functional monomer may be generated. The free NHC necessary for recomplexation may simply be generated using 4-dimethylaminopyridine (DMAP). In a last step, excess base is removed by extensive washing and finally the catalyst is immobilized/formed by passing a solution of ClaRu(CHPh)(PCy3)2 over the rigid rod. Loadings of up to 1.4 % of Grubbs-catalyst on NHC base may be achieved. Monolith-immobilized metathesis catalysts prepared by this approach show high activity in various metathesis-based reactions such as ROMP and RCM. The cis/trans ratio of polymers (90 %) exactly corresponds to the one found with homogeneous systems. The use of chain-transfer agents (CTAs, e. g. cis-l,4-diacetoxybut2-ene, diethyldiallylmalonate, 2-hexene) allows the regulation of molecular mass, in particular in the case of cyclooctene. Typical values for the molecular weight and polydispersity (PDI) of poly(cyclooctene) were in the range of 1500 - 2500 and 1 . 2 - 1.9, respectively. The corresponding values for poly(norbornene) are 12000 and 1.2. The presence of CTAs additionally enhances the lifetime of the catalytic centers by reducing the average lifetime of the ruthenium methylidenes (Scheme 2).
monolith , ~
inner surface
0 ---/
CTA
regeneration
O R-N N R CI."~ CI~:Ru=-R PCy 3
[P-Ru=CH 2]
catalytic cycles
,,,,,
EtO2C CO2Et
+
EtOeC CO2Et
Scheme 2. Structure and reactivity of monolith-based, heterogeneous metathesis initiators. This is of enormous importance, since these methylidenes decompose in a unimolecular process and can only be suppressed by the use of a highly reactive CTA. In particular cis1,4-diacetoxybut-2-ene turned out to be well suited for these purposes. It allows the repetitive use of these systems, particularly important in RCM. Fig. 1 illustrates the enhanced long-term stability if CTAs are used in RCM. In terms of reaction kinetics, both
308 the tentacle-type structure and the designed micro structure of the support reduce diffusion to a minimum.
O O
O A A / % 6(]
5~ A
A A
0
. . . .
0
I
10
. . . .
I
20
. . . .
I
30 t/rain
. . . .
I
40
. . . .
I
50
. . . .
I
60
. . . .
I
70
Fig. 1. Difference in activity (A, expresses in % of the original value Ao) with (e) and without (A) the use of cis-l,4-diacetoxybut-2-ene. Thus, these systems behave as predicted by theory and must therefore be considered as successful alternatives to standard PS-DVB supports. The fast kinetics as well as an enhanced stability quantitatively translate into a high average turnover frequency (TOF) in RCM of up to 25 min-1, thus exceeding even the homogeneous analogue (TOF = 4 mini; 45 ~ Maximum tumover numbers are around 60 (homogeneous < 20). The catalytic systems presented here may be used as pressure stable catalytic reactors as well as one-way systems for use in combinatorial chemistry. The use of NHC-ligands successfully suppresses any bleeding leading even in RCM to virtually Ru-free products with a ruthenium-content of less than 0.07 %.
2.2 Heterogeneous metathesis catalysts based on silica 2.2.1 Surface-grafted silica Metathesis-based grafting techniques have already been successfully applied to the synthesis of other silica-based catalytic supports, e.g. those for heterogeneous ATRP[7, 18] as well as for heterogeneous Heck-type reactions[ 19]. This tempted us to investigate as to which extent these grafting techniques might be applied to the synthesis of silicaimmobilized NHC-precursors. The x-ray structure of such a polymerizable NHCprecursor is shown in Fig. 2.
309
F(I}
g(31
Fig. 2. X-ray structure of the polymerizable NHC-precursor 1,3-Di(1-mesityl)-4{[ (b icyc 1o[2.2.1 ] hept- 5-en- 2- ylcarbo nyl) o xy] methyl }-4,5- dihydro- 1H- imidazo 1-3- ium tetrafluoroborate. Scheme 3 summarizes the synthesis of a triethoxysilyl-telechelic oligomeric NHC precursor. This oligomer was grafted onto non-porous silica using standard silica chemistry[20]. Reaction of the grafted support with KO-tBu in THF at -30~ yielded the free carbene which was subsequently reacted with CI2Ru(CHPh)(PCy3)2 to yield the immobilized second generation Grubbs catalyst. After leaching of the support with aqua regia under microwave conditions, the ruthenium content of the solution was measured by inductively plasma-optical emission spectroscopy (ICP-OES). In terms of catalyst loading it is worth mentioning that only 13 % of the NHC ligand were converted into the corresponding catalyst, leading to a catalyst loading of 0.5 weight-%. This value is much lower than the one found in systems based on monolithic supports, were roughly 40 % of the NHC precursor could be used for immobilization, resulting in 1.4 weight-% catalyst loading[16]. Though a non-porous support should facilitate the accessibility of any surface-bound groups, this particular silica shows a reduced accessibility of the corresponding NHC-sites. We attribute this fact to the strong tendency of this support to agglomerate. Preliminary RCM experiments were carried out with diethyldiallylmalonate. The catalyst was added to a solution of this monomer in 1,2-dichlorobenzene and the mixture was heated to 50 ~ for 2 hours.
310
~,._~ BF4"
1. Mo(N-Ar')(CHCMe2Ph)(OR')2 2. (EtO)3Si-(CH2)3-N=C=O
(E(O)3Si~
[=
A
1~" - [ ' ~
.,~CM~Ph Jn
R Ar' = 2,6-/-Pr2-CeH3 R ' = CMe(CF3)2
n=7
O oA N.j~_~)
BF4-
Scheme 3. Synthesis of a triethoxysilyl-telechelic, oligomeric NHC precursor. Irrespective of the reaction conditions used (i. e. ultrasound, microwave, changing reaction times, temperature and solvents), the maximum turnover number (TON) that was achieved was 75. In principle, second generation Grubbs-type initiators immobilized on non-porous silica should behave similar to those immobilized on monolithic supports[16]. In fact, catalysts immobilized onto monolithic supports give similar maximum TONs (< 65) in the absence of any chain transfer agent (CTA). Ruthenium measurements by means of ICPOES revealed quantitative retention of the original amount of ruthenium at the support within experimental error ( 5 % ) , thus offering access to metal free products. 2.2.2 Surface-coated silica
A very simple approach to surface-functionalized supports lies in the use of copolymers that are used for simple coating techniques. While this method is certainly among the most straightforward ones, some general impediments need to mentioned. Generally, coating techniques result in a significant loss of specific surface and pore volume of the support. In due consequence, significant amounts of any catalytic site incorporated into such polymers are no longer accessible. In order to evaluate the general applicability of coating techniques for the synthesis of heterogeneous metathesis catalysts, copolymers of a NHC-precursor with norborn-2-ene were prepared and used for coating purposes as outlined in Scheme 4.
311
1. n
~ O
1 .COPOLYMER SYNTHESIS
O
R-N~..~-R [ BF4H r
CI2Ru(PCy3)2(CHPh)
3.
==~
~ m COPOLYMER
o--~
",,~.O Ph O ~ o BF4R-N~N-R H
2. COATING
COPOLYMER thermalcoating
r~
COATED SILICA
Scheme 4. Synthesis of coated silica supports. Typical amounts of NHC ligand that were immobilized by this approach were within a range of 90 - 130 mmol (ca. 5 - 7 %). Conversion of the polycationic precursor polymer into the polymeric NHC was accomplished using either KO-t-Bu or dimethylaminopyridine (DMAP). The former allows the synthesis of "protected" NHC precursors [17] that are thermally converted into the free NHC, while DMAP results in the instantaneous formation of the free NHC. In due consequence, ruthenium loadings are significantly higher in the case of the t-butoxide protected precursors (16 % v s 5 % for the DMAP route). Despite the high ruthenium loadings, the catalytic activity of such compounds is comparably low, in particular when compared with monolithic systems. Thus, typical values for the TON in the RCM of diethyldiallymalonate were < 10 (monolithic systems < 60!). Since the identical chemical approach in terms of monomers and conversion is used, these findings must obviously be attributed to diffusion-based processes within the pore structure of silica. This again underlines the high synthetic value of the monolith-based catalytic supports described in section 2.1.
312 REFERENCES
1. K.J. Ivin, J.C. Mol, Olefin Metathesis and Metathesis Polymerization, Academic Press, San Diego (1997). 2. M.R. Buchmeiser, Chem. Rev., 100 (2000) 1565. 3. A.H. Hoveyda and R.R. Schrock, Chem. Eur. J., 7 (2001) 945. 4. J.M. Basset and A. Choplin, J. Mol. Catal: A Chemical, 21 (1983) 95. 5. M. Chabanas, A. Baudouin, C. Cop6ret and J.-M. Basset, J. Am. Chem. Soc., 123 (2001) 2062. 6. R. Buffon, A. Choplin, M. Leconte, J.-M. Basset, R. Touroude and W.A. Herrmann, J. Mol. Catal: A Chemical, 72 (1992) L7. 7. R. Kr611, C. Eschbaumer, U.S. Schubert, M.R. Buchmeiser and K. Wurst, Macromol. Chem. Phys., 202 (2001) 645. 8. M.R. Buchmeiser and K. Wurst, J. Am. Chem. Soc., 121 (1999) 11101. 9. J. Silberg, T. Schareina, R. Kempe, K. Wurst and M.R. Buchmeiser, J. Organomet. Chem., 622 (2000) 6. 10. S.B. Garber, J.S. Kingsbury, B.L. Gray and A.H. Hoveyda, J. Am. Chem. Soc., 122 (2000) 8168. ll. J. Dowden and J. Savovic, Chem. Commun., (2001) 37. 12. A.G.M. Barrett, S.M. Cramp and R.S. Roberts, Org. Lett., 1 (1999) 1083. 13. Q. Yao, Angew. Chem., 112 (2000) 4060. 14. F. Sinner and M.R. Buchmeiser, Angew. Chem., 112 (2000) 1491. 15. F. Sinner and M.R. Buchmeiser, Macromolecules, 33 (2000) 5777. 16. M. Mayr, B. Mayr and M.R. Buchmeiser, Angew. Chem., 113 (2001) 3957. 17. S.C. Schfirer, S. Gessler, N. Buschmann and S. Blechert, Angew. Chem., 112 (2000) 4062. 18. U.S. Schubert, C.H. Weidl, C. Eschbaumer, R. Kr611 and M.R. Buchmeiser, Polym. Mater. Sci. Eng., 84 (2001) 514. 19. M.R. Buchmeiser, S. Lubbad and K. Wurst, Inorg. Chim. Acta, submitted (2002). 20. M.R. Buchmeiser, J. Chromatogr. A, 918/2 (2001) 233.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
313
In memory
of Professor
V!adimir Smetanyuk Preparation of physically heterogeneous and chemically homogeneous catalysts on the base of metal complexes immobilized in polymer gels A.A. Efendiev a, T.N. Shakhtakhtinskib and N.A. Zeinalovb alnstitute of Polymer Materials of the Azerbaijan National Academy of Sciences, 124 Samed Vurgun str., Sumgait 373204, Azerbaijan Republic bM.F.Nagiev Institute of Theoretical Problems of Chemical Technology of the Azerbaijan National Academy of Sciences, 29 H.Javid Avenue, 370143, Baku, Azerbaijan Republic A number of polymer gels have been prepared using tertiary ethylene-propyleneethylidenenorbornene copolymer as a rubber base with grafted poly-4-vinylpyridine, polymethacrylic acid and polymethacrylamide ligand chains. The grafted copolymers were crosslinked and complexes of nickel, zirconium and titanium were immobilized in the formed crosslinked copolymers. After treatment with organoaluminium compounds the obtained catalysts demonstrate high catalytic activity in the reactions of dimerization of lower olefins. 1. INTRODUCTION Supported complexes of transition metals combine the advantages of heterogeneous catalysts such as simplicity of separation from the reaction media and high stability with the advantages of homogeneous catalysts such as high activity and selectivity and the possibility of obtaining more accurate information about the structure of their active centers and thus, the mechanism of catalytic processes [1]. The use of polymer ligands as supports opens new possibilities to vary ligand surrounding and control the catalytic properties of complexes
[2,3]. We developed a new principle of preparation of metal complexes immobilized in polymer gels able to swell in hydrocarbon substrate thus, providing an easy access of the reagents to the active centers. These are two phase systems, wherein nonpolar rubber base is the dispersion medium containing fairly regularly distributed domains of graft chains of macromolecular ligands. The dispersion is crosslinked to a certain degree followed by treatment with transition metal compounds. As a result, metal complexes are formed in sites of the macromolecular ligands. Due to the rubber base, the catalysts can swell in hydrocarbon media up to several hundred vol. % forming a gel accessible for the reagents. The immobilized complex catalysts are actually physically heterogeneous but chemically homogeneous catalysts because the rate of diffusion in highly swollen polymers is comparable with that in liquids. On the other hand, the gel immobilized complexes can be easily separated from the reaction medium, as heterogeneous catalysts and used repeatedly.
314
The term "gel immobilized metal complex catalysts" was introduced by Kabanov and Smetanyuk [4,5], and then research continued by the authors of this paper in collaboration with laboratory of late Prof. Dr. V.I. Smetanyuk. This paper summarises the results of the preparation and investigation of nickel, zirconium and titanium complexes immobilized in rubber base with grafted macromolecular ligands [610]. 2. RESULTS AND DISCUSSION We have synthesized a number of polymer gels using tertiary ethylene-propylene-ethylidene norbornene copolymer (CEP) as a rubber base with grafted poly-4-vinylpyridine (PVP), polymethacrylic acid (PMA) and polymethacrylamide (PMAA) ligand chains [7,8]. The above mentioned monomers were added to the solution of tertiary copolymer in n-heptane together with 1-2% of azobis-isobutyronitrile and heated at 75-80 ~ for 6-10 hours. The grafted copolymers were crosslinked by adding 2-4 mass % of benzoperoxide to the solution. The crosslinked graft--copolymers were precipitated from the reaction medium in a form of swollen gels which were dried and granulated. The crosslinked graft-copolymers were contacted with hydrocarbon solvent (n-heptane, toluene) and in the swollen form they were treated with salts of nickel (nickel chloride, nickel acetylacetonate), titanium (dibutoxytitanium dichloride) or zirconium (dibutoxyzirconium dichloride). The resulting gel complexes were repeatedly washed with toluene-methanol mixture and nheptane to remove the excess of the metal salt until the washing gave a negative test for metal, and then treated with an organoaluminium compound (OAC). Diisobutylaluminium chloride (DIBAC), ethylaluminium dichloride (EADC), diethylaluminium chloride (DEAC), ethylaluminium sesquichloride (SCEA) and triisobutylaluminium (TIBA) have been chosen for such treatment. The swelling capacity of the obtained gel immobilized catalysts in nheptane was in the range of 600-800 vol.%. Comparison of IR-spectra of CEP and CEP-PVP shows that bands at 1620, 950 and 930 cm -1 characterizing non-saturation of CEP disappear after grafting of PVP and band at 1600 cm -~ associated with pyridine ring appears. After treatment of CEP-PVP with dibutoxyzirconium dichloride bands at 1640 cm -~, 1600 cm -1, and 1500 cm -1 with shoulder at 1529 cm -1 appear. Similar picture can be also observed in the case of treatment of CEP-PVP with dibutoxytitanium dichloride. One could assume that in these cases, coordination with zirconium and titanium takes place not only with nitrogen atom but also with ~-electron system of the pyridine ring, i.e. an arene complex is formed. Taking into account these data, as well as data of elemental analysis, the structures of zirconium and titanium complexes with CEP-PVP may be illustrated by the following schemes:
315
/ / / / _
/
CI
C4H9 Zr-Cl
N
.
""ke.~-ff" I C4H9
N
\ \ \ \
"-.
N
\"
In IR-spectrum of CEP-PMAA, there is a band in the 3320-3400 cm -1 region characteri the N-H bond of the amide group. After treatment with nickel chloride, this band is shifte the long-wave region. On the other hand, a band associated with the C=O bond of the aJ group does not change. One might assume that coordination of nickel takes place only N-H groups. Thus, the structure of nickel complexes of CEP-PMAA could be represente follows:
H\I~. C H
H
%o
NTiLz \ ' 'b
',
C
%O
H H
', \
H
o,
'N/
~C/
\N /
H
N/
~C/
H
In the IR-spectrum of the nickel complexes of CEP-PVP, we observe a band at 1640, which characterizes complexation of nitrogen atom and a decrease of intensity of the ba~ 1600cm -1 which characterizes free pyridine ring. From these data and elemental analysis can assume that the structure of the nickel complexes with CEP-PVP might be illustrate follows:
316
N
N
N
I
N
I
sNiL2,,
NiL2 I
% s
In the IR-spectrum of CEP-PMA there is a band at 1720 cm -a characterizing C=O group. After treatment of CEP-PMA with dibutoxyzirconium dichloride, this band is shifted to the long-wave region. From these data, it might be assumed that both CO and OH-groups take part in coordination with the metal. In the spectrum of zirconium complex with CEP-PMA bands at 1100 cm -1 and 570 cm -1 characterizing C=O and Zr-O bonds are observed, and bands characterizing Zr-CI bonds are absent. Based on these data, the structure of zirconium complexes with CEP-PMA might be illustrated as follows:
C4H9 I O
.O.. . - ' T Z r _'-'.
O .-~Zr "-
"y H
C4H9
%
~ C -
"'"o H
Similarly, the structure of complexes of CEP-PMA with titanium can be represented. After treatment of the complexes with OAC active gel-immobilized catalysts are formed. The catalytic activity of the obtained catalysts was studied in the reactions of dimerization of ethylene and propylene. It is known that in the dimerization of ethylene in the presence of homogeneous nickel complexes, 1-butene is initially formed, its major part being isomerized into cis-2-butene. The
317 latter, in turn, is isomerized into trans-2-butene which is a more stable compound in terms of thermodynamics. The quilibrium mixture resulting from the dimerization of ethylene in the presence of homogeneous nickel complexes has the following composition [11]: 1-butene 3 % mass cis-2-butene 27 % mass trans-2-butene 70 % mass Meanwhile, it is well known that 1-butene has more practical applications. Dimerization of ethylene in the presence of homogeneous titanium and zirconium complexes proceeds with mass selectivity up to 98-99 % with respect to 1-butene, but one always observes at least 0.5 - 1% mass of polymer formation which creates problems when scaling up. We carried out the dimerization of ethylene in the presence of the obtained gel immobilized complexes of nickel, titanium and zirconium. The reaction was carried out in 0,5-1itre thermostatted stainless steel reactor fitted with stirrer and manometer, n-Heptane was used as a solvent. Temperature range was 293-353 K; pressure range 0,2-4 MPa; molar ratio A1/Me varied in the range of 3 - 10. The catalytic activity of the catalysts was evaluated according to decrease of pressure in the reactor. Gas-liquid chromatography method was used for the analysis of the reaction products. Results of dimerization of ethylene at different temperatures in the presence of CEP-PVP-Ni (ac.ac.)z -DEAC are given in Fig.1.
g CzH 4 g Cat., h 120 100 80 60 40
I
I
I
I
293
313
333
353
p,,
T,K Fig.1. Dimerization of ethylene at various temperatures: catalysts CEP-PVPNi (ac.ac.)z, [] = EADC; 9 = DEAC; pressure- 0,2 MPa; molar ratio A1/Ni- 10. It is seen that optimum temperature range is 313-333 K. It is known that homogeneous nickel complexes are not stable at temperatures higher than 293 K [11] The process of dimerization is exothermic one and to prevent overheating at large scale complicated system of heat tapping is required. Gel immobilized nickel complexes remain active for a long time at higher temperatures, up to 353 K.
318
Fig.2 shows the dependence of dimerization of ethylene in the presence of CEP- PVP-Ni (ac.ac.)2- DEAC on pressure.
g C2H4 500 ~ g Cat., h 400 300 200 -
IOOF~i
l I 0,2
0,4
l
I
I
0,6 P, MPa
l
.-_
0,8
Fig.2. Dimerization of ethylene at different pressures: catalyst CEP- PVP- Ni (ac.ac)2 DEAC; temperature- 313 K; molar ratio A1/Ni = 10. As seen from Fig.2, the rate of dimerization increases linearly in the pressure range 0,1 1,0 MPa. Results of dimerization of ethylene using the same catalysts with various molar ratios AI/Ni are presented in Fig.3.
gC2H 4 g Cat., h 100 80 60 40 20 I
I
I
I
2
4
6
8
I
I
10 12 A1/Ni
I
I
I
14
16 18
i
,,..--
20
Fig.3. Dimerization of ethylene with different molar ratios of A1/Ni: catalyst CEP-PVP-Ni (ac.ac.)2-DEAC" pressure- 0,2 MPa; temperature- 313 K.
319 As it can be seen in Fig.3, the maximum catalytic activity is achieved with molar ratio 10. Further increase of the molar ratio does not lead to an increase of the catalytic activity. It is known that homogeneous nickel complexes are usually used when molar ratio A1/Ni is 50 100. Thus, it can be seen from the above mentioned data that gel-immobilized nickel complexes have significant advantages compared to homogeneous nickel complexes, as they can be used at elevated temperatures and with much lower A1/Ni molar ratio. The results of dimerization of ethylene in the presence of the obtained nickel, titanium and zirconium complexes with different macroligands and OAC are given in Table 1. The temperature in all the experiments was 313 K; pressure 0,2 MPa; OAC/Ni molar ratio of 10 and OAC/Ti or Zr of 4. Table 1 Dimerization of ethylene in the presence of gel immobilized complex catalysts Composition of the reaction Catalyst products, % mass 1-butenetrans-2-butene cis-2-butene 1 CEP-PMAA-NiClz-SCEA 86,0 11,0 3,0 99,9 traces traces CEP-PVP-Ti (OC4H9)zCIz-TIBA 67,0 33,0 traces CEP-PMA-Ti (OC4H9)zCIz-TIBA 99,9 traces traces CEP-PMAA-Ti (OC4H9)zClz-TIBA 99,9 traces traces CEP-PVP-Zr (OC4H9)zCIz-TIBA 99,9 traces traces CEP-PMA-Zr (OCzH9)zClz-TIBA 99,9 traces traces CEP-PMAA-Zr (OC4H9)zCIz-TIBA It can be seen from Table 1 that when using gel immobilized complexes of titanium and zirconium, very high selectivity with respect to 1-butene, up to 99.9 % mass practically, can be achieved without any formation of polymer. Besides, homogeneous complexes are not stable enough and loose their activity after a few hours, whereas gel immobilized complex catalysts remain active for hundred hours and more. It can be also seen from Table 1 that nickel complexes with PMAA macroligands demonstrate 86% selectivity with respect to 1-butene, whereas in case of homogeneous nickel complexes, as it was already mentioned, the selectivity does not exceed 3 %. We also studied the dimerization of propylene in the presence of gel-immobilized nickel complexes. It is known that in the dimerization of propylene with homogeneous nickel complexes a mixture of dimers containing 4-methyl-l-pentene, 4-methyl-2-pentene, 2-methy2-1pentene, 2,3-dimethyl-2-butene, hexene and other compounds is formed, and the content of 4-methyl-l-pentene does not exceed 8% mass [11]. It is also known that 4-methyl-l-pentene has more practical application as its polymer is widely used in electric power engineering, electronics, medicine, etc. The reaction was carried out using the same unit as with dimerization of ethylene. The conditions of the reaction were: pressure - 0,2 MPa; temperature - 313 K; molar ratio AI/Ni 10. CEP-PMAA-NClzo6HaO-SCEA was used as a catalyst. Analysis of the reaction products has shown that there was 46.0% mass of 4-methyl-l-pentene; 41,0%mass of 4-methyl-2pentene and 13,0% mass of other isomers in the mixture. Thus, using gel-immobilized nickel
320 complexes, one carl significantly increase the yield of 4-methyl-l-pentene in the dimerization of propylene. 3. CONCLUSION The results obtained show that immobilization of metal complexes in polymer gels allows to prepare physically heterogeneous and chemically homogeneous catalysts and leads to an important increase in their activity, selectivity and stability in the reactions of dimerizatiorl of lower olefirls. The immobilization of the complexes opens new possibilities of macromolecular design of the catalysts with desired structural organization and will contribute to the development of general principles of synthesis of highly efficient and environmentally friendly catalytic systems for liquid phase processes. ACKNOWLEDGEMENT The results discussed have been obtained in collaboration with V.A. Kabanov of Moscow State University. The authors would like to appreciate the contribution of late Professor Vladimir Smetanyuk. REFERENCES
1. F.R. Hartley, Supported Metal Complexes, D.Reidel Publ. Co., Dordrecht, 1985. 2. P. Hodge and D.C. Sheringtorl (eds.), Polymer-Supported Reactions in Organic Synthesis, Wiley & Sons, Chichester, 1983. 3. A.D. Pomogailo, Immobilized Polymeric Metal-Complex Catalysts, Nauka, Moscow 1996. 4. V.A. Kabarlov, V.I. Smetanyuk and V.G. Popov, Dokladi AN SSSR 225, (1975) 1377. 5. V.A. Kabarlov and V.I. Smetarlyuk, Macromol.Chem., 5 (1981) 121. 6. N.A. Zeinalov, A.V. Ivanyuk, A.I. Prudnikov, V.I. Smetarlyuk, M.V. Ulyanova and A.A. Eferldiev, Dokladi Akademii Nauk, 348 (1996) 207. 7. O.I. Adrov, G.N. Borldarenko, N.A. Zeinalov, A.V. Ivarlyuk, V.I. Smetanyuk, V.S. Stroganov, M.V. Ulyanova and A.A. Eferldiev, Vysokomolekulyarnye Soedineniya, Ser.B, 38 (1996) 1608. 8. N.A. Zeirlalov, A.V. Ivarlyuk, O.I. Adrov, G.N. Bondarerlko, M.A. Martynova, V.I. Smetanyuk, M.V. Ulyanova, A.A. Efendiev and A.I. Prudnikov, Vysokomolekulyarnye Soyedirlerliya, Ser.A., 39 (1997) 888. 9. N.A. Zeirlalov, O.I. Adrov, A.V. Ivanyuk, G.N. Borldarenko, V.A. Kabanov, M.V. Ulyarlova, A.I. Prudrlikov, V.I. Smetanyuk and A.A. Eferldiev, in: Book of Abstracts, International Symposium on Ionic Polymerization, Istarlbul, 1995, p.121. 10. A.A. Efendiev and N.A. Zeirlalov, in: Proceedings of XVI Mendeleev Congress of General and Applied Chemistry, Moscow, 1998, v.2, p.251. 11. V.Sh. Feldblyum, Dimerization and Disproportionation of Olefirls, Moscow, Chimiya, 1978.
Studies in SurfaceScienceand Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
321
Hydrocracking catalyst to produce high quality Diesel fraction Roberto Galiasso Tailleur PDVSA Intevep POBox 76343 Caracas 101 Venezuela,
[email protected] A hydrocracking catalyst oriented to the production of a high quality Diesel fraction was optimized by treating the AIzO3-SiOz support with steam-ammonia. The catalyst characterization shows that aluminum migrates from a tetrahedral coordination to pentahedral-octahedral coordination. This fact seems to increase the total conversion and selectivity of the VGO hydrocracking reaction. The results are associated to a higher metal dispersion and higher Lewis acid strength. 1. I N T R O D U C T I O N Catalytic hydrocracking is a modern refinery tool to produce clean fuels (Gasoline and Diesel). The versatility of this process is due to the catalyst formulation. When Diesel fraction quality is targeted, catalyst must produce a highly isomerized product to increase the cetane number, but still having a good cloud point. Hydrocracking of vacuum gas oil has been tested in a broad range of conversion levels, catalysts, and feedstocks [1,2]. A new generation of catalysts capable of transforming aromatics into iso-Paraffins has recently been introduced [3]. The 2000"s bring forward many challenges for the refining industry with more stringent environmental specifications for fuels, especially in diesel production and a general trend toward converting more difficult feedstocks. These changes are making hydrogen availability and catalyst stability a limiting factor in many refineries. As part of PDVSA-Intevep broad development, a new hydrocracking catalyst with improved diesel selectivity was targeted by modifying a "conventional" A1203-SiO2 support. 2. E X P E R I M E N T A L Three WNiPt/AlzO3-SiOz catalysts were prepared by steam ammonia treatment of the same AIzO3-SiOz support, characterized, and their hydrocracking activity tested using conventional pilot plant test. Previous studies [4-5] had shown the effect of the acidity and metal active center on activity and selectivity. Two hydrocracking severities have been used in this study to understand the catalyst impact in product quality. The catalysts were characterized using infrared spectroscopy (IR), nuclear magnetic resonance (NMR), X-Ray Photoelectron Spectroscopy (XPS), and programmed Thermal-Gravimetric Ammonia Desorption (TGAD).
322
2.1 Catalyst preparation Three hydrocracking catalysts (MHCK) were prepared by impregnation of an A1203-SIO2 support treated under steam ammonia atmosphere. The support was prepared by coprecipitation of silica and alumina gel. The product was dried at 120 ~ extruded in a lx5 mm cylinder shape and treated in air at 450 ~ for 4 hours to near constant weight. Then, it was dealuminated using steam-ammonia at 200 ~ 0.3 bar of ammonia partial pressure, 200 l/h of gas flow, during one, two and three hours, respectively, to generate support SI, SII and SIII. After that, the three supports were impregnated using the same amount of active species in two stages. First, they were impregnated by an ammoniumtungstate and nickel nitrate-water soluble solution, followed by two hours drying in air at 120 ~ Second, by impregnation with Pt-diammine water-soluble salt, drying at 120 ~ and calcined at 550~ in air during six hours. The three catalysts (CI, CII, and CIII) were then presulfided.
2.2 Catalyst characterization To understand the difference between the three supports and catalyst, SI-CI, SII-CII and SIII-CIII were submitted to a serie of analyses and tests. Chemical characterization: total metal analysis has been performed using atomic adsorption spectroscopy (Varian Techtron analyzer). Metals were reported in % by weight (bulk) of total metal oxides in the support (W, Ni, Pt / A1-Si support). See Table 1. Physical method: Surface, pore volume, and average pore diameter were measured using standard nitrogen adsorption and mercury porosimetry methods. See Table 1. XPS: Spectra were obtained in a Leybolh-Hereaus LHS-10 apparatus (Mg cathode) using Alfalfa with 50 eV of power. XPS method was used to assess the metal dispersion on supports using the peak area intensity (corrected) to measure atomic concentration. The Defoss6 et al [6] method was applied. Binding energies in sulfided catalyst were between Ni: 853.4 and 856.3 eV (2p3/2), W: 34,4 and 32,5 eV (4f5/2-7/2), Al: 74.2 eV (2p). In this way W, Ni, Pt and Al were measured using the peak deconvolution and integration to obtain the area and reported here as a ratio of metal/total metal in surface. Platinum at this low concentration was poorly detected as a large shoulder. See Table 2 for dispersion and Fig. 1 for typical XPS spectra of W and Ni species. Ratios of W+4/W+5 and Ni+2/Ni + were measured by deconvolution of the corresponding spectra. NMR: Solid 27A1 MAS NMR was used to determine the structure of the support (Si-A1) based on the method described by Nagy et al. [7]. The spectra provide information on the different type of aluminum structure in the support (tetrahedral-octahedral coordination). See Fig. 2 NHaTPD: Thermal desorption of ammonia was used to characterize the acid strength of the support and catalysts. A McBain microbalance was employed using 1 mg of sample. Ammonia was adsorbed at room temperature and the total mol of NH3*102/m 2 adsorbed were measured. Then the sample was heated using a ramp of 6 ~ and the remaining amount of NH3/m 2 calculated at three temperatures (200/300/400 ~ See Table 3. FT-IR: Infrared spectroscopy was used to determine the acidity of the MHCK catalysts. The apparatus was a Perkin Elmer 2865 with Fourier transform capabilities. The typical plots of absorbance as a function of increasing wavelength were obtained for each catalyst and reported in Fig. 3. To improve the plot, 4% and 8% of transmittance were added to the
323
Table 1- Physicochemical Properties Catalysts NiO wt% WO3 wt% PtO wt % A1203 wt% SiO2 wt% Surface m2/g Volume cm3/g particle diameter m aver. pore diameter A
CI CII CIII 4.0 4.2 4.3 13.4 13.2 13.5 0.1 0.1 0.1 15 14.7 14.5 Complement 222 234 228 0.52 0.54 0.51 0.001 0.001 0.001 120 105 110
Table_2- Surface Catalysts Ni/Total Me W/Total Me A1 /Total Me
Metal dispersion XPS CI CII CIII 2.2 2.7 2.1 6.4 5.8 6.0 4.4 3.8 3.5
Table 3- Ammonia TPD mmol*10-2/sm Temperature ~ SI SII SIII/CIII 200 5.3 4.6 4.3/2.8 300 1.2 1.3 1.1/0.4 400 0.4 0.6 0.7/0.2 _
signal for supports SII and SIII and catalyst CIII. Acidic absorption bands in the range 3500-3750 cm -1 were recorded because this region is associated to the stretching bands of the OH groups. Strong acidic bands appear in the region of 3600-3650 cm -1, and weak ones in the region of 3550 cm -1 [8]. See Fig. 3. Table 4. Reaction feed and products (wt%) Temperature ~ 380 380 380 400 400 LHSV h-1 0.75 0.75 0.75 0.75 0.75 Catalysts Feed CI-1 CII-1 CIII-1 CI-2 CIII-2 Conversion 100 60 64 69 70 72 Diesel 0.0 45.0 49.5 54.7 52.0 55.0 Nafta 0.0 10.8 10.2 9.9 13.0 12.0 Gas 0.0 4.5 4.8 5.0 5.6 5.8 Note: there is a dramatic change in percent conversion (10%) between CI-1 and CI-2 when going from 380 to 400~ while there is only a 3 % change f o r CIII-1 and CIII-2 in the same conditions and almost all went to the gasoline fraction (it seems selectivity shifts towards naphtha formation). Also note that there is no match between % conversion and the sum of individual fractions
Pilot plant The effect has been studied in a small-scale pilot plant (see detail of the plant in reference [1]). This unit has a 60 cm 3 down-flow fixed bed reactor that operates isothermally. The hydrogen and the hydrocarbon feed were preheated before entering the reactor. After reaction, the liquid product (C5+) was fractionated and analyzed using conventional ASTM method. In addition, a Mass Spectrometry coupled with gas chromatograph (GC MS) was used to measure aromatics, paraffins and naphthenics compounds distribution in the feed and in the products. In addition, a special NMR analysis was performed to determine the PNA. The VGO was desulfurized using commercial catalyst (not described here) and the product characteristics are shown in Table 4 as well as the feed. The MHCK catalysts were tested at 380 and 400 ~ LHSV=0.75 and 100 bar of total pressure, using 800 m3/m3 of HJHC ratio at the inlet of the reactor. The
324 HCK products quality are shown in Table 4. As example, CI-1 and CI-2 mean: catalyst I severity 1, and Catalyst I severity 2, respectively. To compare the catalysts, the same severity was used. Hydrogen purity was 100%; the catalytic system was diluted in the reactor with 50% inert material, and used a particle size of 0,1cm x 0,1 cm (cylinder). This special precaution was taken to ensure proper fluid dynamics according to De Bruijn results [9]. Catalysts were sulfided with light virgin gasoil at 300 ~ during six hours. Sulfur and carbon contents in all the fresh catalysts were nearly the same (6 wt% and 0.1 wt%) 3. RESULTS
3.1 Catalysts The three catalysts show nearly similar bulk composition in tungsten and Platinum, and a small difference in nickel, attributed to the impregnation method. Aluminum seemed to slightly decrease when the steam-ammonia treatment period increased. The small difference in surface, total pore volume, and average pore diameter (calculated by integrating the pore volume distribution curve in the range of 10 to 300 A) could not be correlated with the ammonia treatment and were in the range of the analysis errors.
3.1.1 Metals in surface (XPS) on sulfided catalyst The XPS spectra in the Nizp and W4f regions after sulfidation are shown in Fig. 1. The binding energy (BE, eV) of the support is in the range of 102.6 to 102.9 eV for the Sizp Table 5. XPS dispersion (IMe/ITotal) Sample CI W +6 0.121 WSz 0.334 IW/IT 0.455 NiO 0.09 NiSx 0.144 INi/IT 0.234 IAI/IT 0.452 Is/IT 0.314
CII 0.115 0.352 0.467 0.086 0.157 0.243 0.434 0.343
CIII 0.129 0.383 0.512 0.081 0.17 0.251 0.423 0.388
86618621858]8541850[848 W
C
...... .."..... ...-" .,..." :...., "....
I
~
....
and in the range of 74.5 eV for the Al2p. The sulfur 401 38 i 36 ! 341 321 30 was detected from 162.0-162.2 eV [10]. The shape Fig. 1. XPS spectra of the Ni2p envelope with a satellite peak at 860.6 eV shows almost the same presence of non sulfided Ni 2+ species in all catalysts (CI-CII-CIII) with no shift at all. The Nip3/2 signal (856.4 eV) is due to non sulfided Ni L+, probably in the Si-O-AI framework. It slightly increases from catalyst CI to CIII, with a maximum shift of 0.2 eV (856.4-856.2 eV). The second Ni2p3/2 peak corresponds to NiS [11]. The amount increases as a function of ammonia treatment with a shift of the signal by 0.4 eV (853.4853.7 eV). The NiS/NiO surface ratio is presented in Table 5. It increases from CI to CIII, indicating that the modification of the A1+2+O+3+-SIO2 framework changes to some
325
extent in the nickel structure at the active surface, and is probably also modified by Pt species during sulfating. The W4f7/2 doublets appear at 32.2. and 34.5 eV, and at 35.5 and 37.9 eV, which are ascribed to WS2 and non sulfided W § species. The position of the two doublets did not change for the three catalysts. The parameters for the "sulfided" species were obtained by curve fitting of sample CI and allowing the peak position and FWHM to relax into their local minimum. Table 5 shows that the proportion between WSz/W +6 species increases from 72 for CI to 78% for CIII, indicating larger sulfide species in catalyst with longer period of steam ammonia treatment and AIzO3-SiOz framework modification. Again the Pt species may have played some role in the active surface modification. Sulfur signal at 162.1 eV increases (with no relative shift) from CI to CIII in agreement with previous statement that surface sulfided species have increased. Previous sulfiding studies [12] speculated about the role of Pt on Ni and W migration from the framework, which could explain in part the present results. Table 5 shows that aluminum dispersion is reduced from CI to CIII, in agreement with the reduction of the bulk composition.
3.1.2 Acidity of support and oxide catalysts NH3 is one of the probing molecules for measuring the Lewis and Br6nsted acidity of the surface. The acidity of SI, SII, and Sill was measured as the amount of NH3 retained at each temperature. The results are presented in Table 3. The mass spectrometry analysis of the gases desorbed did not indicate any NH3 decomposition into amide, imido, hydrazine, and dimers species below 400 ~ Here the AIzO3-SiO2 sites are the main agent for the NH3 adsorption, but metals as W +6 and NiO contributed to the total acidity. Comparing the adsorption for the support Sill with catalyst CIII in Table 3, it can be concluded that impregnation reduced the total acidity by 40-60% and changed the acidity profile, as expected. Most of the metal oxides during the impregnation are deposited on top of the AIzO3-SiO2 framework, reducing the number of acid centers. The modification of the framework by the steam ammonia treatment changes these adsorption and metal dispersions, as shown above, and the exposed acid sites are reduced. Comparing support SI with SII and Sill, it was observed that the longer the steam-ammonia treatment period, the lower is the total acidity at 200 ~ but the higher is the acid strength (higher amount of NH3 retained at 400 ~ The infrared analysis of adsorbed pyridine (not shown here) confirmed that the strength of the Lewis MASNMR wppm fom [AI (H20)613+ acid sites was higher in tetrahedral pentacoordinates steam-ammonia sample CIII ~' octahedral (bands at 1350 cm -1). Figure 2 presents the Z7A1NMR sll analysis of CI CII and CIII samples without sulfiding. The CI Sl sample presents larger bands attributed to o Fig. 2. A1MASNMR spectra for SI, SII,SIII and CIII
326
aluminum in tetrahedral coordination and small bands attributed to aluminum occupying octahedral positions, with barely any penta-coordinated aluminum sites [14]. When the support is treated during a longer period with steam ammonia, the spectra changes. The AI in tetrahedral coordination decreases and turns to penta and octahedral coordinated sites, and no tetrahedral ones (see in the AI-MASNMR spectra above the peaks at 5, 30 and 60 ppm, respectively, for catalyst II and III in comparison with I). It seems that steam ammonia reorganizes the framework structure by dissolving aluminum and formation of A1203 over the -Si surface. Figure 3 presents the FTIR spectra in the region assigned to the acid centers. Bands centered at 3555 cm -1 are associated to Lewis acid centers and those of 3650 cm -1 to Br6nsted sites [10]. It can be seen that support SI has the lowest Lewis and the highest Br6nsted acidity while support Sill has the opposite surface composition with the higher Lewis to Br6nsted acid sites ratio. Support II has both type of sites in a similar amount/proportion. It is well known [11] that interactions between WO3-SiOz-AI203 are critical for the formation of Lewis acid centers. The new signal at 3570 cm -1 is attributed to non-framework OH stretching in silanol sites near a vacancy [12]. That signal decreases from catalyst CI to CIII. After metals impregnation, most of the bands are reduced (see dashed line in Fig. 3 for catalysts CIII).
~
CIII
"'" "LS-LS
'•
I
c mq 3680
_
"saIt - - - _- ~ _ ~ ' ~
I
3630
-
I
3580
14 12 10 8 !6 "4 2 3530
Fig. 3. FTIR of SI SII SIII and CIII 3.2 Catalytic test
The pilot plant test was done on desulfurized VGO. The sulfur, nitrogen and carbon Conradson properties are typical of those used in commercial hydrocracking for the conversion stage. Table 4 shows the results at two severities. Let us compare the catalyst at 380 ~ Activity is defined as the VGO conversion in weight, and selectivity as the diesel produced related to the total conversion. It can be seen that treatment produces an increase in VGO conversion and diesel yields at the expense of nafta production. Gas also slightly increases. This result is also confirmed at high severity (400 ~ The effect of temperature is almost the same in catalyst CI as in CII (same activation energy and similar selectivity). Catalyst CIII has almost the same pore structure, particle diameter and was tested under identical operating conditions. Neither fluid dynamics or diffusion differences can explain the higher activity. The active phase modification may be responsible of the activity and
327 selectivity improvement. The increase of acidity and the higher metal dispersion seem to be responsible for higher conversion and selectivity. The hydrogen consumption due to the hydrocracking reaction decreases from catalyst I to III due to the higher selectivity to diesel production (the higher the gasoline and C1-C4 production, the higher the hydrogen consumption). 3.3 Product quality The cracking and hydrogenation balance in the catalyst depends on acid center/metal ratio. In our case, the main target is the diesel quality. The results of diesel PNA analysis are shown in Table 6 for catalysts CI, CII and CIII operating at 380 ~ The analysis shows that the steam-ammonia treatment increases the paraffins content and decreases naphthenes and aromatics content. This indicates a higher hydrogenationhydrogenolisis activity in CIII than in CII and CI. In addition, the NMR analysis indicates Table 6 Selectivity to hydrogenation Catalyst n+isopar wt% Monocyclop. wt% Dicyclop. wt% Tricyclop. wt% Mono arom. wt% Di arom. wt% Triarom. wt% Cetane Number
CI-1 26.00 18.40 12.18 3.20 32.19 8.14 0.00 54.00
CII-1 26.90 19.15 12.60 3.00 30.80 7.60 0.00 55.00
CIII-1 28.20 19.14 13.40 2.20 30.00 7.20 0.00 57.00
a slightly higher iso/n-paraffins ratio in CII and CIII than in CI. As a consequence of the physicochemical modification of the catalysts, the cetane number increases from 51 to 54 (54 to 57 according to Table 6) and the cloud point is reduced f r o m - 2 2 ~ to -24 ~ in the hydrocracked products. 4. DISCUSSION Upon steam-ammonia treatment and then calcination at 500 ~ of the support samples, the -AI- coordination at the surface changed. Aluminum migrated from a well organized surrounding with tetrahedral coordination into a distorted penta and hexa coordination. Aurox et al. [14] discussed the heat of adsorption of ammonia on different types of alumina and concluded that the highest heat of adsorption of ammonia is associated to the abundance of pentacoordinated aluminum. They measured, for some superacid alumina with high content of pentacoordinated aluminum, a high LI+L2/NH3 ratio and the largest distribution of the acid centers (in agreement with their CO adsorption study [15]). Our FTIR results of Fig. 3 show a new band at 3570 cm -1, that may be attributed to the aluminum migration out of framework position. In addition, the change in
328
the AI-NMR signal (Fig. 2) at around 20ppm confirms the migration into a pentacoordinated-Al-framework position. When the support was impregnated using tungsten nickel and platinum watersoluble salts, the metals were preferentially adsorbed on the aluminum surface. The adsorbed metal salts were then decomposed in the air treatment, generating a more complex Me-AI-Si interaction (clusters) and most of the A1 NMR signal disappeared as well as the FTIR acid bands. Using ammonia adsorption, it was shown by XPS that metal dispersion changes from catalyst CI to CIII, which indicated the effect of steam-ammonia surface modification on the metals adsorption and migratiort The steam-ammonia treatment dissolved some aluminum, increasing the vacant sites by dealumination and migration to a different silica oriented environment. These vacant sites may be occupied by a tungsten and nickel species during impregnation. As a consequence, the electronic coordination of these metals is also modified in different ways after suffering an "equivalent" sulfiding procedure. The XPS spectra show the increase of the NiS/NiO and W+4/W+6 ratios, induced by support modification. Thus, not only are the acid sites modified, but the metals sites are also affected. At the present stage, it is not fully clear if the active phase is composed by nickel enclosed in a nickel-tungstate-sulfide layer where Pt could play an important role. The catalytic test proved that the steam ammonia treatment increased the initial conversion of VGO into diesel fraction, but did not proceed further to gasoline and gas formation by secondary cracking. This behavior is associated to a larger number of very strong (accessible) Lewis type acid centers in a non-shape selective support. Moreover, the change in aromatic hydrogenation shown in Table 4 confirms that the active metal phase was promoted, and the support modification (higher dispersion and higher Ni and W sulfur species). Both the acid and metal sites increased the isomerization (higher n/iso-paraffins ratio) in the complex bifunctional reaction path. 5. C O N C L U S I O N S It has been shown that steam ammonia treatment of an A1203-SIO2 hydrocracking support promotes the formation of pentacoordinated aluminum species. The mechanism is associated to a dealumination and aluminum migration from tetrahedral coordination into more distorted AI-Si environment. This generates a larger proportion of strong Lewis acid centers and a broad distribution of acid strength. The support modification also promotes a higher metal dispersion and a higher sulfur content at the surface. Both gave a higher activity and selectivity to diesel production. Diesel quality is also improved by a higher hydrogenation and a higher isomerization activity in steam- ammonia dealuminated catalyst. ACKNOWLEDGMENT I wish to acknowledge the efforts of the process engineers, and pilot plant and catalyst characterization people at PDVSA-Intevep, especially Jose Arroyo who did the bench scale catalytic tests. In particular, I thank Prof. Hercules who performed the XPS study and Z. Gabellica who carried out the AI-NMR studies. I would also like to thank Intevep for permission to publish this information.
329 REFERENCES
1. R. Galiasso, M. Di Marco and A. Salazar, 13th World Petroleum Congress, Proceedings (1991) 233. 2. J. Scherzer and A.J. Gruia, Hydrocracking Science and Technology, Marcel Decker, Inc. (1996) 96. 3. R. Prada, R. Galiasso, G. Romero and E. Reyes, US Pat. 4465792 (1989). 4. R. Galiasso and R Prada, Preprint 4 th International Conference on Refining Processing, Aiche Meeting, Houston A 22 (2001) 327. 5. R. Galiasso, Appl. Catal. (2002) (submitted). 6. C. Defoss6, P. Canesson, P.G. Rouxhet, and B.J. Delmon, J. Catal., 25 (1972) 407. 7 J. B.Nagy, Z. Gabelica, G. Debras, E.G. Derouanne, J.P. Gilson and P.A. Jacobs, Zeolites 2 (1969) 59. 8. J.B. Uytterhoeven, R. Schoonheydt, V.Liengme and W.K. Hall, J. Catal., 13 (1969) 425. 9. A. De Bruijn, 6th. Int. Congress on Catalysis, London, paper B34, 1976. 10. R.B. Shalvoy and R.J. Reucroft, J. Vac. Sci. Technol., 16 (1979) 567. 11. B. Pawelec, L. Daza, L.L.G. Fierro and J.A. Anderson, Appl. Catal. A: Gen., 145 (1906) 307. 12. R. Galiasso, WNiPt SIO2A1203 presulfiding, Appl. Catal. (in preparation). 13. D. Coster, A.L. Blumenfeld and J.J Fripiat, J. Phys Chem., 99 (1995) 321. 14. M. Aurox and M. Muscas, Catal. Lett., 28 (1994) 179. 15. V Gruver and J.J. Fripiat, J. Phys. Chem. 98 (1994) 8549.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
331
Thermostable yttria-doped inorganic oxide catalyst supports for high temperature reactions E. Elaloui
CH4
at
--i I
r--n
.~.
~
~
"/x-
-A-- - A - - ~
- --A- - A -
I Fig.3 Changes of the conversion in CH4-CO2 reaction over WC(I) catalyst at 1123K
- A - - -A
3O
o
c~
" ~-
-o-
-o-
- Q-
-o
- -o-
- o-
20
0 0
-(3
t
•
~
_1
L
1
2
3
4
5
Time
6
[h]
80F
I=cH
60
I~H,
70
COz
i_+_ c0
so
-
35
a0g~ ~ 25 ~ ~
=o 4 0
-
~ o
30
-15~
o o
20
-
10
-
0
'-n
20
10
5
~
Fig.4 Changes of the conversion in CH4-CO2 reaction over WC(II) at 1123K
o~
r
0 0
2
4
6 Time
8
10
12
[h]
Figs. 5 and 6 represent XRD patterns of these carbide catalysts before and after the reaction. Crystal structures of MozC(I) and MozC(II) as prepared were different from each other, the former being hcp and the latter fcc, as mentioned already. In the case of MozC(I), several new peaks assignable to MoOz e m e r g e d after the reaction as shown in (b). This was not observed in the case of MozC(II), which r e m a i n e d stable (shown in (d)). These results suggest that the oxidation of MozC to MoOz at the initial stage of the reaction might cause an abrupt deactivation in the case of MozC(I). The XRD pattern of MozC(II) after the reaction indicated that the crystal structure was t r a n s f o r m e d from fcc to hcp during the reaction, a c c o m p a n i e d by an increase in crystallite size. In the case of tungsten carbide catalysts (Fig. 6), no oxide patterns were observed after the first run, but small oxide peaks at 0=26 ~ and 53 ~ e m e r g e d after the second run. These results also suggest that the oxidation of WC to WOz at the second run is the main cause of the abrupt deactivation.
420
(b)
20
(a)
20
1 30
I 40
1 60
20[
I 60
~ ]
I 70
I 80
I
30
I
40
50
60
70
80
20[ ~ ]
9~ XRD patterns of WC(I) (a)" after 1st run, (b)'after 2 nd run
Fig. 5 XRD patterns of Mo2C(I) and (II) (a)(b)" Mo2C(I), (C)(d)" Mo2C(II) (a)(c)" before reaction, (b)(d)" after reaction,
(e)" MoO2 after reaction
Fig. 7 illustrates the Mo3d and C l s XPS spectra of Mo2C(I) and (II) before and after the CH4-CO2 reaction. As summarized in Table 2, both catalysts exhibited almost the same binding energies of Mo3d and C l s before reaction, which can be assigned Mo2C (Mo3d=227.4 and 230.6 and C1S=282.8 eV). After the CH4-CO2 reaction at 1123K, the binding energies for Mo3d transition shifted to the higher eV side in both catalysts, whose extent was larger in the case of MozC(I) catalysts (0.7-1.0 eV) than Mo2C(I) catalysts (0.3-0.4 eV). Moreover, characteristic Cls peak of 282.8-230.0 eV, which can be assigned to carbide carbon almost disappeared in the case of Mo2C(I) after the reaction, indicating that the carbide structure of the Mo2C(I) surface may be destroyed by CH4-CO2 reaction. These results are consistent with the XRD bulk information, which is the main cause for deactivation of the catalysts.
90
421
_••
Mo3d
I
2oo
. ~ 1
S
c)
295
290
285 B.E.
Fig. 7
B.E. leVI XPS spectra of (a),(b)" Mo2C(I) (a),(c)" after the reaction
280
275
[eV]
and (c),(d); MoZC(II) (b),(d)" before the reaction
Table 2. Binding energies of XPS data Catalysts Mo2C(I) Mo2C(II)
Treatments before reaction after reaction before reaction After reaction
Mo 3d3/2 230.8 231.5 230.6 230.9
Binding energy [ eV ] Mo 3d5/2 C 1s 227.5 284.8 282.8 228.5 284.8 227.4 285.0 283.0 227.8 284.9 228.9
According to Green's results for the carbides prepared by direct carburization (corresponding to MozC(I) in this study), elevated pressures were needed to maintain a constant high activity for 72 hours, while their activity dropped abruptly after 7 hours in the reaction at ambient pressure [11]. They proposed a redox type reaction mechanism for the formation of syngas. In this mechanism, after the dissociation of CO2 the formed O(a) reacts with carbon in the carbide surface to leave vacancies. These are then filled with either carbon from methane, reforming the carbide, or oxygen to form MOO2. For the former step to remain predominant, an elevated pressure is required. To elucidate the different catalytic behavior of MozC(I) and Mo2C(II) in this study, CH4-CD4 isotopic exchange reaction was carried out over both catalysts. The rate of CH3D and CHD3 formation was several times faster over Mo2C(II) compared to that over Mo2C(I) at 373K, suggesting that dissociation of methane was much easier over Mo2C(II). Accordingly, oxidation of vacancies with oxygen in the redox mechanism may be more
422 effectively prevented over Mo2C(II), resulting in the durability of the catalytic activity compared to MozC(I). These situations are schematically summarized in Fig. 8.
H2
CO
CO
j
j
/i .........
.......i MoO2
Inactive carbon
Fig. 8 Schematic view of the mechanism of CH4-CO2 reaction 4. CONCLUSION The dependence of the activity in reforming CH4 with CO2 was investigated in depth upon the preparation methods of molybdenum and tungsten carbides. In the case of
MozC, catalytic performance of the
catalyst, prepared through nitridation of the oxide before carburization, was considerably different from that prepared by direct carburization of the oxide. The deactivation was significantly suppressed in the former catalyst, although its TOF for syngas formation was smaller than the latter. The situation was rather different in the case of tungsten carbides, and both direct carburization and nitridation-carburization catalysts exhibited similar initial activity, but the durability was much better in the former catalyst. ACKNOWLEDGMENTS This study was supported by High Tech Research Project of Ministry of Education, Science, Sport and Culture of Japan.
423 REFERENCES
1. C.C. Yu, S. Ramanathan, F. Sherif and S.T. Oyama, J. Phys. Chem., 98 (1994) 13038. 2. F. Garin, V. Keller, R. Ducros, A. Muller and G. Maire, J. Catal., 160 (1997) 136. 3. H. Abe and A.T. Bell, J. Catal., 142 (1993) 430. 4. G. Djega-Mariadassou, M. Boudart, G. Bugli and C. Sagay,
Catal. Lett.,
31 (1995)411. 5. G.S. Ranhotra, A.T. Bell and J.A. Reimer, J. Catal., 108 (1987) 40. 6. J.-L. Dubois, K. Sayana and H. Arakawa, Chem. Lett., (1992) 5. 7. V. Keller, P. Weher, F. Garin, R. Ducros, and G. Maire, J. Catal., 153 (1995) 9. 8. V. Volpe and M. Boudart, J. Solid State Chem., 59 (1985) 332. 9. J.S. Lee, L. Volpe, F.H. Ribeiro and M. Boudart, J. Catal., 112 (1988) 44. 10. J.T. Wroloski and M. Boudart, Catal. Today, 15 (1992) 349. 11. J.B. Claridge, A.P.E. York, A.J. Brungs, C. Marquez-Alvarez, J. Sloan, S.C. Tsang and M.L.H. Green, J. Catal., 180 (1998) 85. 12. A.J. Brungs, A.P.E. York and M.L.H. Green, Catal. Lett, 57 (1999) 65. 13. A.J. Brugs, A.P.E. York, J.B. Claridge, C. Maroquez-Alvarez and M.L.H. Green, Catal. Lett., 70 (2000) 117. 14. M. Tsuji, T. Miyao and S.Naito, Catal. Lett., 69 (2000) 195. 15. J.S. Lee, L. Volpe, F.H. Ribeiro and M. Boudart, J. Catal., 112 (2000) 195. 16. T. Xiao, A.P.E. York, V.C. Williams, H. AI-Megren, A. Hanif, X. Zhou and M.L.H. Green, Chem. Mater., 12 (2000) 3896.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
425
Synthesis and properties of new catalytic systems based on zirconium dioxide and pentasiis for process of NOx selective catalytic reduction by hydrocarbons V. L. Struzhko, S. N. Orlyk, T. V. Myroniuk, V. G. Ilyin L. V. Pisarzhevsky Institute of Physical Chemistry of NAS of Ukraine Pr. Nauki, 31, 03039, Kiev, Ukraine This work is devoted to the synthesis of ZrO2 by various methods, the synthesis of zirconium-containing pentasils and ZrO2- W-zeolite based binary carriers. These materials were used as carriers of transition metal oxides (chromimn, cobalt) and their catalytic properties were characterized in the selective reduction of NO by methane and propanebutane mixture, the acidic properties of the samples were investigated by thermoprogrammed desorption and IR-spectroscopy methods. 1. INTRODUCTION Zirconium dioxide and zeolites of pentasil structure are widely used as catalysts and efficient carriers in many heterogeneous reactions, and particularly in the process of selective catalytic reduction of nitrogen oxides by hydrocarbons (SCR-process) [1,2]. Synthesis of new catalytic systems for NOx SCR-process by CnHm is therefore related with searching for their optimum composition and preparation methods to attain maximum activity in this reaction. It is well known that the composition, texture and crystalline structure of zirconium dioxide are rather sensitive to the conditions of its preparation, resulting in changes of their acidic and catalytic properties [3]. The transition metal oxides as dispersed onto the oxide or zeolite carriers surface are active catalysts in the process of nitrogen oxides selective reduction by hydrocarbons (NOx/HC/O2) [4]. For example, the systems Pt-ZrO2- A1203 and MnOy- ZrO2 have been studied in the reduction of NO with propane [5]. The reduction of nitrogen oxides (NOx) with propane on platinum deposited on La203, ZrO2, and their mixture has been studied [6]. Recently, much interest has been focused on zeolite systems, in which inorganic oxides are introduced into the matrix as catalysts, including the selective catalytic reduction process. Cobalt containing zeolites are known to have high activity when methane is used as a reductant for nitrogen monoxide [7,8]. We have previously reported the cation exchanging ZSM-5-type zeolites (Co- and Cacontaining samples), and MexOy/ZrO2 oxide systems were the active catalysts in the process of nitrogen monoxide SCR by methane and propane-butane mixture [9,10].
426 This work is devoted to the synthesis of ZrO2 by various methods, the synthesis of zirconium-containing pentasils and ZrO2- H+-Zeolite based binary carriers as well as the study of their acidic and catalytic properties in the SCR-process. The catalytic properties of samples doped with cobalt and chromium oxide, based on zirconium dioxide, zirconiumcontaining pentasils and ZrO2- H+-Zeolite based binary carriers have been characterized in the reduction of NO with methane and propane-butane mixture in an oxidizing atmosphere, and also the acidic properties of the various catalyst samples by thermoprogrammed desorption of ammonia (TPDA) and IR-spectroscopy. 2. EXPERIMENTAL 2.1. Preparation of the samples Pure zirconia was obtained by precipitation and the sol-gel method. Analysis of the literature data showed that precipitation is the most commonly accepted method. Thus, we prepared a sample of starting zirconia by precipitation of the hydroxide from 0.5 M aqueous solution of zirconium oxychloride by adding 2.5 M aqueous ammonia under vigorous stirring at 20 ~ and constant pH 9. After precipitation, Zr(OH)4 was maintained in contact with the mother liquor for five days and the precipitate was then washed with aqueous ammonia with pH 8 until there was a negative test for chloride ion. The sample was then dried at 170 ~ for 3-4 h. The subsequent zirconia sample was obtained by hydrolysis of zirconium isopropoxide in aqueous ethanol in the presence of NH4OH as the catalyst. Water and the organic solvent were removed from the zirconium hydrogel by drying in air at 150 ~ XRD (CuK) data show that upon thermal processing of zirconium hydroxide xerogels, at 460 ~ the tetragonal modification of zirconium dioxide synthesized by sol-gel method, and monoclinic modification of ZrO/synthesized by co-deposition were formed. Exothermic effect of amorphous Zr(OH)4 conversion to T-ZrO/ or M-ZrO2 occurs at 460 ~ [10]. The reaction mixture of initial zirconium silicagel was prepared by mixing aqueous solutions of NaOH and ZrOC12 8H20 in the presence of a complexing agent, [(C4H9)4N]J solid salt with aerosil. Zirconium-containing pentasils were synthesized via the crystallization of zirconium silicagels under hydrothermal conditions in an autoclave. After the crystallization terminated, the obtained deposits were separated, rinsed up to 7-8 pH, and dried in air at 120 ~ followed calcinating at 550 ~ for 6 h. When heating the initial o reaction mixture in the autoclave at 175 ~ 100~A-content crystalline phase zeolite was formed in 48 hours. X-ray photographic study of zirconium-containing zeolite revealed the identity of its structure with that of pentasil (the analog of zeolite ZSM-11). 100%-phase purity zirconium-containing zeolite has the following chemical composition: 0.057 Na20.1.00 SiO2.0.01 ZrO2.0.043 R20.18 H20. The presence of ZrO2 in formed pentasil was confirmed by X-ray fluorescence spectroscopy. NH4+ exchanged forms were prepared by multiple treatment of the original sodium formed by 0.5 N aqueous solution of NH4C1 at about 90 ~ accompanied by washing to remove the chloride ions. Samples were dried at 120 ~ and calcined at 540 ~ for 4 h in air. The sodium amount and degree of exchange in different H +- forms were determined by flame photometric analysis.
427
ZrOz-H+-Zeolite catalytic systems were synthesized in the following way: H+-Zeolite suspension was added to aqueous suspension of zirconium hydrogel washed off from salts, and the resulting mixture was then vigorously stirred for 45 rain to achieve more homogeneous sample. The deposit was then squeezed and dried at 100 ~ and calcinated at 500 ~ for 3 hours. To vary the amorphous and crystalline phases ratio, carriers with diverse compositions were obtained. The active phase of transition metals (Co, Cr) oxides was deposited by the precipitation method, ionic exchange from nitrate salts solutions and ionic exchange in solid phase as well [8,10]. The catalysts were obtained by impregnation of zirconia obtained by both methods with aqueous solutions of the corresponding salts (cobalt or chromium nitrate), subsequent drying at 100 ~ and roasting at 320 ~ for 6 h. The MexOy/ZrOz samples containing 5-10 wt% metal oxides on the support (relative to the metal) were prepared by this method.
2.2. Catalytic tests The catalytic activity of the catalyst samples obtained were characterized in the selective reduction of NO with methane and propane-butane mixture by conversion of NO to Nz (NzO) which was determined in a gradientless reactor with chromatograph analysis of the products. The NO concentration was determined with a gas analyzer with a chemiluminescence detector [8].
2.3. Investigation of acidic properties Studies of the acidic properties of the surfaces of samples by the TPDA method were carried out as follows. Samples (0.2 g) with 1-2 mm grain size were placed in a flow reactor (d=0.6 cm) and were conditioned in a stream of helium (V= 60 ml/min) for 1 h at 550 ~ After decreasing the temperature to 100 ~ the sample was saturated with ammonia. Completion of saturation was monitored by titration of the ammonia at the exit of the reactor. The saturated sample was treated with helium at 100 ~ to remove the physically adsorbed ammonia (30 min). The sample was then subjected to programmed heating in a stream of helium at a rate of 26~ The thermodesorption process was monitored with a catharometer and the amount of ammonia desorbed was determined by titration with HCI. The acidic properties were also studied by adsorption of carefully dried pyridine which was carried out at 150 ~ for 20 min, after which the sample was evacuated for 1 h at the same temperature to remove the physically adsorbed pyridine. Infrared spectra were recorded at room temperature on a Zeiss Specord 751R spectrophotometer. 3. RESULTS AND DISCUSSION The data on the catalytic activity of the samples MexOy/ZrOz in the process of NOx SCR by hydrocarbons are given in Table 1. It is seen that the activity of zirconium dioxidebased oxide catalysts dependent on a method to prepare ZrO2; and 10% CrzO3/T-ZrOz sample prepared by sol-gel method was found to be a more active catalyst in the reaction with propane-butane.
428 The Cr203/ZrO2 catalysts showed activity in the SCR of NO by a propane-butane mixture, which depended on the means of preparation of the zirconium dioxide. Thus, the conversion of NO to N~ was 13-17% at 350 ~ on 5-10 wt.% Cr203/ZrO~ catalysts obtained by precipitation, while the conversion of NO to N2 was 54% at 300 ~ on catalysts with analogous composition obtained through an alcogel step. This more active sample was also tested in the presence of SO2 (0.02%) in the reaction mixture. The conversion of NO in this case was also enhanced and reached 60% at 300-350 ~ This increase in activity by the action of sulfur dioxide may be attributed to the formation of sulfate since sulfated zirconium dioxide is a solid superacid and catalyzes the SCR of NO by hydrocarbons [ 11]. Table 1 Activity of synthesized MexOy/ZrO2 samples in the selective reduction of NO by hydrocarbons (HC)/0.05% NO + 0.09% CnHm+ 5% O2 + Ar; V = 6000 h-I/ No. Catalyst (preparation method) NO Conversion,% / T, ~ (HC) 1 10% Cr203/M-ZrO2 (deposition method) 13/300 (C3Hs-C4Hlo) 2 10% Cr203/T-ZrO2 (sol-gel method) 54/300 (C3Hs-C4HIo) 3 10% CoO/ZrOz (deposition method) 75/310 (CH4) 4 10% CoO/ZrO2 (sol-gel method) 72/300 (CH4) With 10% CoO/ZrO2 catalyst, conversion of NO (in reaction with methane) reached 75% at 310 ~ while the selectivity with respect to nitrogen decreased from 100% at 415 ~ to 63% at 310 ~ (the remainder was N20). There was no dependence of the catalytic activity of samples of CoO/ZrO2 on the method used to prepare zirconium dioxide. Both samples (No. 3 and No. 4, Table 1), in which zirconium dioxide was made by precipitation and the sol-gel method, respectively, had similar activity. This difference of behavior between samples of the CoO/ZrO2 and Cr203/ZrO2 catalysts may be explained by differences in the interactions of CoO and Cr203 with zirconium dioxide and, consequently, different influence of these catalysts on activation of methane and propanebutane. In order to elucidate the reasons for the dependence of the catalytic properties of these samples on their preparation method, we studied the acid surface properties of cobalt- and chromium-modified ZrO2 catalysts by ammonia thermoprogrammed desorption and IRspectroscopy. Our results again indicated that the activity of these catalysts in the SCR of NOx by hydrocarbons is a function both of the surface acidity and content of the active metal. The acid site concentration of the starting ZrO2 samples prepared by various methods is significant (0.13 and 0.23 mmol/g) but these samples are inactive, while 10% Cr203/ZrO/prepared by the sol-gel method displays considerable activity in the reaction studied with lower surface acidity. The acid site concentration of the sample with the same composition prepared by the precipitation method is reduced by a factor of 2.5 and, thus, this catalyst has much lower activity in the selective catalytic reduction. IR-spectra for zirconium dioxide samples obtained by the sol-gel method with and without 10% Cr203 shows that modification of zirconium dioxide by Cr203 leads to local IR vibrations, which cannot be attributed to characteristic modes of the ZrO2 and Cr203 frameworks. The finding of bands at 800, 1025, and 1170 cm1 in the spectrmn of sample
429 10%Cr203/ZrO2 may indicate the formation of new metal-oxygen bonds of the Zr-O-Cr type in the zirconium dioxide surface layer. Fig? 1 gives IR spectra for pyridine adsorbed on previously dehydrated samples. The spectrum of starting zirconium dioxide obtained through an alcogel step lacks the band characteristic for Br6nsted acid sites. The addition of Cr203 into zirconium dioxide leads to acidic B-sites characteristic for pyridinium ions with a band at 1540 cm-1. This may be related to formation of structure such as [3]:
H
/
H
L
I
O
O
mZrm
\ mCrm
] /\ It is known that for zeolite catalysts, BrOnsted acid sites are necessary for the selective reduction of NO by methane on Ga-H-ZSM-5, while the activity of Cu-, Ce-, and Cocontaining pentasils in the SCR of nitrogen oxide by hydrocarbons correlates with the strength of the Br6nsted acid sites of these catalysts. This correlation suggests that activation of the hydrocarbons reducing agent may occur specifically on the Bronsted acid sites [9]. Our results confirm the important role of BrOnsted acid sites in the reaction studied on ZrO2 systems. Thus, our results showed that zirconium dioxide modified by transition metal oxides (Co and Cr) displays significant activity in the selective reduction of NO by methane and propane-butane, which depends on the method of preparation of the ZrO2 sample.
I
r 9
|
J
9
9
~
|
I
Fig.1. IR spectra of zirconium samples obtained by the sol-gel method before and after pyridine adsorption: 1) ZrO2 aider vacuum heating at 550 ~ in vacuum, 2) ZrO2, and 3) 10% CrzO3/ZrO2.
,,
1600 1550 1500 1450 cm 1
The data on the catalytic activity of the synthesized samples of CoO/H+-pentasils in the NO+O2 NO2 reaction and acidic properties of these samples characterized by total surface acidity determined by TPDA method are given in Table 2. It is seen that 10% CoO
430 deposited by the precipitation on these carriers showed considerable oxidative activity in relation to NO, while only samples Nos. 1 and 3 showed the SCR-activity (conversion NO to N2 in reaction NO-CH4/O2 is 30-50%). It is also seen that these samples have crystalline structure and definite value of surface acidity which is necessary for activation of hydrocarbon-reductant. Table 2 Catalytic and acidic properties of CoO/I-F-pentasils /0. 05% NO+5% o2+mI'; V -- 6000 hl/ No. Catalyst SIO2/A1203 ~(~NH3, NO Conversion,%/ (ZrO2, A12Oaq-ZrO2) mmol/g T, ~ 1 i0%CoO/H +- pentasil 100 0.23 70/310
(SIO2,A1203) 2 3
10%CoO/H+-pentasil (SiO2,ZrO2) 10%CoO/i-F-pentasil
100
0.06
59/300
100
0.16
60/300
100
0.29
50/300
(SIO2,A1203, ZrO2) 4
10%CoO/H +-
(8iO2,A1203, ZrO2), amorphous 5 , !0%CoO/Si02(silicalite)
60/300
90-
o-%0. .070. r '" *-.
2
~o. 3020
0
J
Zr02
I
20
u
I
40
~
I
60
)
Content, %
I
80
~
I
1O0
H-TsVN --b-
Fig. 2. Dependence of the conversion of NO on the chemical composition of the binary carrier ZrO2 - H-TsVN (310 ~ 1) experimental curve; 2) calculated curve for direct additivity. The data on the catalytic activity of the samples on binary carriers 10% CoO/(H § pentasil - ZrO2) in the process ofNOx SCR by methane are given in Table 3.
431 During investigation of cobalt-zeolite catalysts the dependence of activity on the manner by which the active phase was introduced was established. Sample of 10% CoO/HTsVN (SiO2/A1203=37) (in which cobalt was introduced by soaking) had low activity in the selective catalytic reduction of NO with CH4. Conversion of 25% of NO was achieved at 320 ~ which is considerably lower than for cobalt containing cation-decationated form of zeolite with the pentasil structure, obtained by ion exchange in the solid phase (e.g., on Co-H-TsVN an 80% conversion of NO was obtained at 310 ~ [8]. Table 3 Activity of synthesized samples 10% CoO/(H-pentasils - ZrO2) in selective reduction of NO by methane/0.05% NO + 0.09% CH4 + 5% O2 + Ar; V - 6000 hl/ No. Catalyst NO Conversion, % / T, ~ 1 10% CoO/H-TsVN 25/320 2 10% COO/(35% H-TsVN-65% ZrO2) 67/300 3 10% COO/(50% H-TsVN-50% ZrO2) 72/300 4 10% COO/(65% H-TsVN-35% ZrO2 ) 81/310 5 10% COO/(35% H-TsVN-65% ZrO2 )* 69/300 */CoO was introduced by ion exchange in solid phase. Fig. 2 shows the dependence of the activity of cobalt containing catalysts CoO/ (zeolite-ZrO2) on the composition of the carrier. The observed activity of these systems is greater than that calculated on the assumption of addition of the catalytic properties of the components of the catalysts. Such differences indicate interaction between the components of the ZrO2-zeolite carrier, possibility forming new active centers. It is seen from Table 3 and Fig. 2 that increase in the zeolite content in the zeolite-ZrO2 system from 35 to 65% leads to increase in the catalytic activity which achieves 81% conversion at 310 ~ i.e., an activity was achieved equivalent to that of the cobalt containing cationdecationized zeolitic catalyst Co-H-TsVN [8]. To elucidate the reasons for differences in activity of cobalt-containing catalysts with the same amount of active phas, we studied the acidic properties of the surfaces of these samples by the previously described temperature programmed desorption of ammonia. Analysis of the results (Fig. 3, Table 4) indicated a complex correlation of the catalytic activity in the selective catalytic reduction of NO with CI-I4 with the chemical composition, the concentration and strength of the acid centers at the surface of the cobalt-zirconium, cobalt-zeolite, and binary systems based on them. The most important factor for the selective catalytic reduction activity is the localization of the metal (cobalt) active centers, which is determined by the method used to introduce cobalt into the catalyst. It may be claimed that, of two catalysts prepared in the same way, but with differing carrier composition (ratio of zeolite to ZrO2) the more active sample has the greater concentration of acid centers (samples No.2 and No.3 in Table 4). The low activity of the sample 10% CoO/H-TsVN (No.l) with considerable total acidity of the surface (0.66 mmol/g) and the presence of strong acid centers may be explained by different localization of the cobalt in comparison with the ion exchange sample Co-H-TsVN, obtained by ion exchange in the solid state [8].
432
,,m
r.f) r r
100
I
200
~
I
300
~
T,~
I
400
*
I
500
~
I
600
Fig. 3. Spectra of temperature programmed desorption of ammonia from the surface of cobalt catalysts on ZrO2 -zeolite binary carriers" 1) 10% CoO/H-TsVN, 2) 10% COO/(65% H-TsVN, 35% ZrO2), 3) 10%COO/(35% H-TsVN, 65% ZrO2). Table 4 Concentration of acid centers on CoO/(ZrO2 - Zt) catalysts determined by desorption of NH3 and their activity in SCR NO with CH4 No Catalyst Concentration of acid centers, )(No % / T, ~ mmol/g 150-260~ 400-500~ ~rfNH3 1 10% CoO/H-TsVN 0.38 0.28 0.66 25/320 2 10% COO/(65% H-TsVN0.19 0.15 0.34 81/310 35% ZrO2) 3 10% COO/(35% H-TsVN0.13 0.08 0.21 67/300 65% ZrO2) It has been established from these studies that the different catalytic properties of transition metal oxides (chromium, cobalt) on zirconium dioxide are attributed to their different acidic properties determined by TPDA and IR-spectroscopy. The most active catalyst is characterized by strong acidic Br6nsted centers. The cobalt oxide deposited by precipitation on the zirconium-containing pentasils has a considerable oxidative activity in the reaction NO+O2~NO2, and for SCR-activity the definite surface acidity is necessary for methane activation. Among the binary systems, 10% COO/(65% H-Zeolite - 35% ZrO2)
433 catalyst exhibits maximum activity, and the catalytic properties of such samples are not an additive function of the carrier composition. REFERENCES
1. M.P. Fokema and J.Y. Ying, Catal. Rev., 41 (2001) 1. 2. Y. Traa, B. Burger and J. Weitkamp, Microporous Mesoporous Mater., 30 (1999) 3. 3. K. Tanabe, Catalysts and Catalytic Properties [Russian translation], Mir, Moscow, 1993. 4. R. Burch and T. C. Watling, Appl. Catal. B, 14 (1997) 207. 5. K. Eguchi and T. Hayashi, Catal. Today, 45 (1998) 109. 6. V. Pitchon and A. Fritz, J. Catal., 186 (1999) 64. 7. Yu. Li and J. N. Armor, J. Catal., 145 (1994) 1. 8. S.N. Orlik, V. L. Struzhko and V. P. Stasevich, Teor. Eksp. Khim., 32 (1996) 47. 9. S.N.Orlik and V. L. Struzhko, Teor. Eksp. Khim., 35 (1999) 373. 10. T.V. Mironyuk, V.L. Struzhko and S.N. Orlik, Ibid., 36 (2000) 307. 11. H. Hamada, Y. Kintaichi and M. Tabata, Chem. Lett., 1 (1991) 2179.
This Page Intentionally Left Blank
Studiesin Surface Science andCatalysis143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
435
Preparation of the chitosan based catalysts for several hydrogenation reaction in the liquid phase V. Isaeva, A. Ivanov, L. Kozlova, V Sharf N.D. Zelinsky Institute for Organic Chemistry, Russian Academy of Sciences Leninsky pr. 47, Moscow 119991, Russia Novel chitosan based catalytic systems for hydrogenation of unsaturated compounds in the liquid phase were prepared. The catalytic performance of the obtained systems depended significantly on the chitosan forms (as the micro beads or chitosan deposited on the mineral supports), their preparation method and chemical modification of chitosan as well. The obtained chitosan based carriers and catalysts were examined by transmission and diffuse-reflectance FTIR spectroscopy. 1. I N T R O D U C T I O N It is only recently that chitosan (natural biopolymer, deacetylated chitin derivative) is being used as a carrier for metal catalyst preparation. Several papers and patents regarding chitosan application in reactions of selective and enantioselective reduction of organic compounds have been published in the last years [1-2]. Natural biopolymer chitosan (deacetylated chitin derivative) is attractive as carrier for heterogeneous catalysts due to its environmentally friendly nature, in particular, biodegradability. In addition, free amino groups in chitosan fragments facilitate metal deposition on chitosan from metal salts and metal complexes. Previously, we demonstrated the possibilities of chitosan as a macroligand for Rh and Ru complexes immobilization. The obtained metal complex systems showed activity and selectivity in transfer hydrogenation of several carbonyl compounds. Free amino groups in chitosan allow its immediate use as a macroligand without or with preliminary functionalization. The most attractive functionalization from our point of view is to modify the chitosan via reaction of chitosan aminogroups with carbonyl compounds leading to Shift's base formation. On the one hand, it allows to accomplish chitosan crosslinking with dicarbonyl compounds, commonly, with diglutar aldehyde [3,4]. On the other hand, it allows to introduce several functional groups in chitosan according to reaction with substituted carbonyl compounds. We chose 2pyridinealdehyde for such modification. The main obstacle for chitosan application in catalysis is to obtain stable regular micro beads separable after reaction for reusing. That is why this problem is of great interest [3]. The goal of this work was the investigation of alternative methods of Pd/chitosan based catalyst preparation for several reactions of hydrogenation in the liquid phase. The work focused on the following directions:
436 1. Preparation of the stable regular micro beads of desired size. 2. Development of cross-linking of chitosan and functionalization with 2pyridinealdehyde. 3. Deposition of cross-linked chitosan on the mineral support surface. The developed procedures were used for the synthesis of Pd catalysts for cyclopentadiene and 1,4-butynediol hydrogenation in the liquid phase. 2. E X P E R I M E N T A L
2.1.Catalyst preparation Several methods for the preparation of chitosan micro beads of regular size were tested. The main procedure was precipitating of hydrochloride chitosan solution by adding it dropwise in the bath containing the precipitating agent (alkali solution). Adding dropwise was performed with the equipment for micro drop forming through quartz die (d=0.1 - 0.2 mm) at 5 atm. Chitosan micro beads of diameter 0.5-1 mm were formed. The chitosan micro beads obtained by alcali precipitation method were cross-linked by reaction of chitosan amino group with diglutar aldehyde. Cross-linking extent was 7%. In addition to the chitosan native form, chitosan succinate form (70% of succine groups) was used. In several experiments, chitosan succinate form was used without crosslinking, as fibers or gels in aqueous and alcohol- aqueous systems. It should be noted, that cross-linking of chitosan succinate was performed via residual amino group. Chemical modification of the obtained chitosan micro beads was carried out by treatment with 2-pyridinealdehyde boiling solution in benzene for 24 h. The resulting chitosan forms were rinsed with benzene, THF and MeOH. Deposition of chitosan on SiOa (0.06-0.02 mm) and ZrOa was performed by multiple steeping of a portion of carrier in a solution of chitosan in 1% acetic acid and filtered. Wet or semi-dry chitosan-coated carrier was added directly to methanol or propanol-2 and stirred for 0.5-2h after addition of the calculated amount of glutaraldehyde. Calculated cross-linking extent was ~ 10-15%. Chemical modification of chitosan was performed by carrier treatment with pyridinealdehyde-2 boiling solution in benzene. Two procedures for metal introduction in chitosan base were used: impregnation and coprecipitation. According to the first procedure the metal deposition on chitosan micro beads was carried out from aqueous and alcohol solutions of NazPdCl4, H2PdC14, RhC13, Rh2(CH3COO)4, ZnSO4 and Pb(CH3COO)2. Pd and Pb/Zn in bimetallic catalysts was deposited by subsequent precipitation. Pd-Pb (Zn) atomic ratios were 1/1. Metal contents in the resulting samples were 0.5 - 4 % . According the second procedure, the metal complex with chitosan acidic (hydrochloride) form was synthesized and after that precipitated in the bath containing the precipitating agent. Metal contents in the resulting samples were 0.5-2%. The catalytic behavior of the catalysts was examined in reactions of hydrogenation of cyclopentadiene and 1,4-butynediol. Hydrogenation reactions were carried out at atmospheric pressure and 20~ (cyclopentadiene) and 45~ (1,4-butynediol). Hydrogenation rate was determined as the ratio of consumed H2 volume per unit time, ml/min.
437
2.2. Catalyst study by IR-spectroscopy The obtained chitosan carriers and catalytic systems on their base were studied by transmission and diffuse-reflectance FTIR spectroscopy. IR-spectra were obtained in "Nicolet Impact 410" equipment. To record the diffuse-reflectance spectra the samples were evacuated at 100~ for 2 h. The quantitatively spectrum analyses were performed using Kubelka-Munk equation according to the program OMNIC [5]. 3. RESULTS AND DISCUSSION 3.1. Catalyst characterization AB 3413 cm -1, characteristic for valence vibrations of NH pyridine group (Figs. 1, 2) and AB at 1591, 1567, 1475 and 1439 cm -~ corresponding to valence vibrations o f C=C a n d - C = N - bonds of pyridine ring are presented in the spectra of chitosan modified with pyridine fragments.
o~
,
o2a
024
,-'/""
Olt~
\
8'; oo8o
' S"
I
022
]
,,
,. '
\,
o t~
..i
\,
',7
\,
~
'
5
t V
\
9 ::.
~
"
~
\
\
3.~0
~O00
Fig. 1. Chitosan IR-spectrum
Fig.2. IR-spectrum of chitosan modified with 2-pyridinealdehyde
IR-transmission examination of chitosan/SiOz deposited systems gave little information, due to intensive absorption of chitosan and silica gel in the same region. In contrary, characteristic chitosan AB in the regions 1 6 6 0 - 1300 cm -1 and 1100 cm -1 are presented in the spectrum of chitosan/ZrOz system (Fig. 3). Diffuse-reflectance IR-spectra of CO adsorbed on Pd/chitosan/ZrO2 catalyst are presented in Fig 4. Two characteristic adsorption bands (AB) at 2070 cm -1 and 1900 cm -1, corresponded to the vibrations of CO, adsorbed on Pd ~ in the linear and in the bridge (three-fold coordinated) form, respectively, arc presented in the IR spcctrum of the system. The presence of bridged or three-fold coordinated CO points to the formation of Pd ~ metal clusters. Noteworthy is that frequencies of linear and bridged forms of CO are shifted toward lower wave numbers compared to conventional Pd/support systems (e. g. 2070 v s 2100 cm -1 for linear form) [6, 7]. This is indicative of negative charging of Pd ~ clusters. Negative charging of Pd ~ clusters is also confirmed by lower stability of the linear form of CO compared to the bridged form. After evacuation at 200~ the linear form disappears completely, while intensity of bridged bonded one remains almost the same.
438 0190 0 1BOO 170
i
0 160 0150 0 140 0 t30
A
0 120 0119 0 100 0.099
\
i
o oe9
oo5o
J i
\, '\
i
i
!
~ 1
!
oo4o i ,) 03o I
"
o 020
,'
/
t
\\..
I
"'-.,,< \
ooio /l ,I j ooo9 H .... ~:'!?'%/ -oo,o
/
1 i
~'wt$, i !
,,
U ;j
'7
~ 7 '
3000 Wavenumb~.r.~
20~ (cm-I I
1000
Fig. 3. IR-spectrum of chitosan/ZrO2 system
.
. . o 18o
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
--
-
o 17o o.16o o 15o
....
/V\o
.... 1
~o
i oo, o~ OO8O1 o,oo]
//
I
.... 1
!
,"
:=I .... / oo3oi 0020 oolo
It f f _ ~ .
~oo
l - ..
/
I
I
/
/
',;~
\
\
o t
i
i
A\
,'/
!
~/ !
'\ '\~,
/
/
,,"~~
I
i
"
~\
j , ,,
I
I/
I:
~\ \\
\\
\ \
\,\
\\ \\ \ \ ,~ . ),/
tr
2000 Wavenumbers
/
I
/!
term1)
Fig.4 Diffuse-reflectance IR-spectrum of CO adsorbed on Pd/chitosan/ZrO2 Presumably, small Pd ~ clusters are stabilized by NH2 -groups, which donate edensity to Pd ~ clusters. 3.2. Examination of catalytic performance of synthesized Pd-based systems. The hydrogenation activities of catalytic systems prepared by the coprecipitation method were low. It could be explained taking into account that the resulting samples with
439 low metal contents were formed because Pd was partially leached from chitosan during the precipitation in alcali solution. In general, activities of chitosan based catalysts prepared by impregnation method in hydrogenation of unsaturated organic compounds were comparable with those of traditional heterogeneous catalyst (as calculated per 1 mole of metal). It should be noted that the chitosan pretreatment influenced very much the catalyst activity. For instance, immediate Pd deposition from alcohol solution on dry chitosan fibers or micro beads led to almost completely inactive catalytic systems, regardless of the metal content. On the other hand, metal deposition on chitosan micro beads or fibers preliminary swollen in water dramatically improved the catalytic activity. 3.2.a. Hydrogenation of cyclopentadiene The obtained catalysts showed similar activity in cyclopentadiene hydrogenation. The selectivity data in cyclopentadiene hydrogenation are given in Table 1. The selectivity of consecutive reaction was determined as a rate ratio of cyclopentene/cyclopentane formation. On the contrary, the chitosan modification influenced essentially the selectivity of the catalyst on it basis. Table 1 Selectivity of Pd/chitosan based systetms in hydrogenation of cyclopentadiene Catalyst 0.1-0.5 g, Substrate 0.5 - 2 ml, 20~ 20 ml EtOH N ~ Catalyst composition Selectivity to Reaction rate ratio, cyclopentene, % Wolefine/Wdiene 1 Pd/chitosan 99.9 0.90 2 Pd/chitosan succinate 91.2 1.00 3 Pd/chitosan/Si02 94.9 0.02 4 Pd/chitosan-Pyr/SiO2 96.0 0.01 As can be seen from Table 1, the selectivity to cyclopentene obtained using the chitosan form containing carboxylic groups (succinate chitosan form, catalyst N ~ 2) was rather low. The highest selectivity to cyclopentene was achieved using catalyst N~ (Pd/chitosan). However, the cyclopentane formation rate (stage 2) decreased very slightly over this catalyst. Introducing pyridine fragments in chitosan led to a decrease of cyclopentene into cyclopentane hydrogenation rate up to ~ 3 orders. Deposition of chitosan on silica gel as well as modifying it with pyridine groups also resulted in a decrease of cyclopentene hydrogenation rate (catalysts N ~ 3,4). Taking into account the selectivity data and the hydrogenation ratio Wolefine/Wdiene, the best catalyst for this reaction was catalyst N ~ 4 (Pd/chitosan deposited on silica gel and modified with pyridine groups).
3.2.b. 1,4-butynediol hydrogenation The main product of 1,4-butynediol hydrogenation for this reaction is cisbutenediol. Simultaneously, the parallel reaction of 1,4-butenediol hydrogenation as well as cis-trans transformation takes place at the second stage. For this process, cis-l,4butenediol formation is most interesting from a practical point of view. Processes
440
selectivity was determined as ratio of product contents in the reaction mixture (cis/an+cis+trans, %) at 50% conversion (1 mole). As shown in Table 2, catalyst N~ exhibited a rather low selectivity. Introducing the second metal in chitosan (Pb or Zn) as well as pyridine fragments (catalysts N~ improved the selectivity regarding cis-l,4-butenediol. However, the second stage rate remained almost the same. Changing the support influenced remarkably the reaction performance, e.g. substitution SiOz (catalyst N~ for ZrOz (catalyst N~ enhanced the selectivity. In this case, Pb introduction (catalyst N~ almost completely suppressed the cis-trans isomerisation. Using succinate chitosan form very much improved the reaction selectivity: the maximum of selectivity to cis-l,4-butenediol was achieved over catalyst N~ But in this case, the reaction was not terminated at the stage of 1,4-butenediol formation. The rate of further hydrogenation of 1,4-butenediol (second stage) was rather high. Introducing Pb in chitosan completely suppressed the further 1,4-butenediol hydrogenation. The reaction was spontaneously finished over catalyst N~ after consumption of 1 mole of Ha. Simultaneously, the selectivity with respect to cis-l,4-butenediol was improved by that chitosan modification. Table 2 Selectivity of Pd/chitosan based systems in 1,4-butynediol hydrogenation. Catalyst 0.1-0.5 g, Substrate 0.25 - 2 ml, 20~ EtOH 20 ml Selectivity on 1,4Selectivity on Reaction N ~ Catalyst composition butenediol (cis/an + 1,4-butenediol rate ratio cis + trans) (cis) (Wz/W1)* 1 Pd/chitosan 0.88 0.84 0.6 2 Pd/chitosan succinate 0.97 0.91 1.3 3 Pd-Pb/chitosan 0.97 0.93 ~ succinate 5 Pd-Pb/chitosan 0.90 0.93 1 6 Pd-Zn/chitosan-Pyr 0.94 0.89 1 7 Pd-Pb/chitosan/SiO2 0.88 0.84 1,3 8 Pd/chitosan/ZrO2 0.88 0.87 1 9 Pd-Pb/chitosan/ZrO2 0.92 0.96 0.8 * W l - 1,4-butynediol hydrogenation rate (first stage), W a - 1,4-butenediol hydrogenation (second stage) 4.CONCLUSIONS Thus, our results demonstrated that the trends of selectivity for hydrogenation of 1,4-butenediol and cyclopentadiene are in tight connection with preliminary chitosan chemical modification as well as mineral support nature. Pd/chitosan modified with 2pyridinealdehyde deposited on SiOz demonstrated high selectivity in hydrogenation of cyclopentadiene into cyclopentene. 1,4-butynediol into cis-l,4-butenediol hydrogenation proceeded very selectively over Pd-Pb catalytic systems based on chitosan succinate form.
441 The catalytic systems based on chitosan/ZrOa exhibited improved selectivity in 1,4butynediol hydrogenation compared with those based on chitosan/SiO2.
REFERENCES
1. M.-Y. Yin et al., J. Mol. Catal. A: Chem., 147 (1-2) (1999) 93. 2. V Isaeva., V. Shaft, N. Nifant'ev, V. Chernetskii and Zh. Dykh., Stud. Surf. Sci Catal., 118 (Preparation of Catalysts VII), (1998) 237. 3. T. Ando and S. Kataoka., JP Patent No 61278354 A2 (1986). 4. W. Wang, F. Wood and G.A.F. Roberts, Advanc. Chit. Sci., II (1997) 920. 5. L.C.A van den Oetelaar et al, J. Phys. Chem. B, 102 (1998) 3445. 6. T. Rades, C. Pac, R. Ryoo, M. Polisset-Thfoin and J. Fraissard, Catal. Lett., 29 (1994), 91. 7. A.V. Ivanov, A. Yu. Stakheev, L.M. Kustov and Izv. An, Set. Khim. (rus.), No 7 (1999), 1265.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
443
Preparation of Mo/AlzO3 sulfide catalysts modified by lr nanoparticles J. Cinibulk and Z. Vit Institute of Chemical Process Fundamentals, Academy of Sciences of the Czech Republic, Rozvojovfi 135, 165 02 Prague 6 - Suchdol, Czech Republic. Preparation and catalytic properties of Ir/alumina and Ir-Mo/alumina sulfide catalysts were studied in hydrodenitrogenation (HDN) of pyridine and hydrodesulfurization (HDS) of thiophene. The Ir added to a Mo/alumina catalyst in amount 0.1-0.8 wt % increased activities in both reactions by a factor of about 2-3. The main factors, leading to improvement of activity of modified Ir-Mo catalysts, were Ir dispersion, and amount and state of the Mo phase before Ir deposition, i.e. whether oxidic or sulfided and the way of catalyst activation. The effects of starting Ir compounds and deposition order of Ir and Mo seem to be of a smaller importance. 1. I N T R O D U C T I O N Sulfides of some noble metals (Ru, Rh, Pd and Pt) are highly active in reactions such as HDS, HDN or hydrogenation (HY). Modification of conventional Mo catalysts by noble metals represents a possible way of improvement of their efficiency [1-3]. The majority of such modifications has been done up to now with Ru [1,2,4], whereas other noble metals have been studied less frequently. In contrast to a relatively large knowledge collected up to now on the addition of conventional promoters Co (Ni) to a Mo system, information about addition of noble metals or sulfides is rather limited. It was often reported that modification of a Mo catalyst by noble metals led to improvement of activity [1,2,4]. However, it some cases such modification did not bring any improvement or even led to decrease of activity. In contrast to conventional Co (Ni) promoters, it is desirable to keep the amount of noble metal in the mixed catalyst as low as possible. This is not only because of a high price, but also due to decrease of dispersion at higher loadings. Another problem is the way of deposition of noble metal and final catalyst pretreatment, affecting also dispersion and contact with the Mo phase. Preparation of conventional Mo/alumina system usually includes a calcination step in air after deposition of Mo and Co (Ni) salts, which are mostly ammonium heptamolybdate (AHM) and corresponding nitrates, before final sulfiding. However, for some noble metals such a step is not suitable because of possible decrease of the dispersion via oxidation or undesirable interaction with the support [1,5-7]. It was reported that the activity of mixed Ru-Mo catalysts was higher when these catalysts were directly sulfided and not calcined after Ru deposition [1,7]. It seems therefore that the preparation of systems based on a combination of noble metal and Mo needs a specific procedure taking into account the properties of noble metals.
444
One of the most active noble metal sulfides in hydrotreating reactions is Ir sulfide. Its exceptional activity was reported in HDN of different compounds [8] and in the parallel HDN of pyridine and HDS of thiophene [9,10]. Recently, we found a positive effect of the addition of a small amount of Ir to an Mo/alumina catalyst during HDS and HDN [11]. The aim of this contribution is to focus on some factors during preparation of mixed Ir-Mo sulfide catalysts and to evaluate their effect on catalyst activity. 2. E X P E R I M E N T A L The support was ~,-AlaO3 with BET surface area of 255 m2/g and pore volume of 0.76 ml/g (S/id-Chemie A.G., Germany), ground to particles 0.16-0.32 mm. The MoO3/alumina catalyst was prepared by the pore filling method using an aqueous solution of AHM. The product was calcined at 500~ for 2 h in air. The BET surface area was 213 m2/g. The Ir/alumina catalysts were prepared by impregnation of the support with cyclohexane or water solutions of Ir compounds (Ir4(CO)lz, acetylacetonate Ir(AcAc)3, HaIrCI6 and (NH4)zIrC16) by procedures described earlier [10]. Because of a low solubility of Ir carbonyl, which is 0.38 g per 1 of cyclohexane, a modified procedure facilitating the impregnation was developed. Ir carbonyl was placed in a cartridge of Soxhlet extractor and dissolved by hot circulating cyclohexane. The impregnation proceeded in the bottom flask in suspension of carrier and boiling solution of Ir4(CO)lz. A similar procedure was adopted for dissolution of Ir(AcAc)3, while a simple impregnation of the alumina by water solutions of HzIrCI6 and (NH4)aIrCI6 was used. The solvents were removed in a vacuum rotary evaporator. The surface area of all Ir catalysts was close to 230 mZ/g. Different mixed Ir-Mo/alumina catalysts were prepared. The details of procedures and catalyst characterization were published elsewhere [10,11]. In the first series, the Mo was deposited first. The samples were prepared from MoOJalumina or MoSz/alumina catalyst by adsorption of Ir4(CO)la from a cyclohexane solution. This procedure was the same as for the Ir catalysts. In another series, an inverse order of impregnation was used. The Ir was deposited first and then Mo was deposited from an aqueous solution of AHM. One catalyst was prepared by coimpregnation of alumina by an aqueous solution of AHM and (NH4)2IrCI6. The samples containing Ir were sulfided or reduced without calcination in air in order to avoid Ir sintering. The surface areas of mixed catalysts varied between 200208 m2/g. Reduction was performed by H2 using a temperature gradient 6~ up to 400~ and by keeping this temperature for 2 h. The dispersion of Ir in the reduced catalysts was determined by pulse H2 adsorption at 22~ and expressed by H/Ir ratios. The size of the Ir particles in reduced Ir catalysts was calculated under assumption of H/Ir=l stoichiometry [12] according to Anderson and Pratt [13] or estimated on some Ir-Mo samples by transmission electron microscopy (TEM) on a JEM-2000EX Jeol instrument. Sulfidation of catalysts was performed by H2S/Hz mixture (10 % H2S) using a temperature gradient 6~ up to 400~ and by keeping this temperature for 2 h. The TPR of sulfided catalysts was performed in a conventional apparatus by monitoring of H2 consumption. The catalysts were heated at a rate of 5~ in mixture 5 % of Ha in Ar (35 ml/min). Relative reducibilities of sulfided catalysts were defined as areas under the TPR curves in the range 100-600~ and related to the weight of catalyst. The content of metals and sulfur was
445 determined by the inductively coupled plasma (ICP). The contents of chlorine and carbon were determined by argentometric titration and by combustion, respectively. The BET surface area was measured by N2 adsorption on a Digisorb 2600 instrument. The activity of catalysts was tested in the parallel HDN/HDS of pyridine (PY) and thiophene (TH) in an flow reactor at 320~ and 20 bar. In case of reduced Ir samples, single HDN of pyridine was performed. The feed contained 220 ppm of PY and 240 ppm of TH (or only PY) in H2 at a flow rate of 0.4 mol/h. The HDS of thiophene was described by pseudo-first-order rate constant kTH. The HDN was described for simplicity by two rate constants for pyridine HY, kpy, and piperidine HDN, kc5. Further details concerning the evaluation of the catalyst activity can be found elsewhere [10,11]. 3. RESULTS AND DISCUSSION
3.1. Monometallic Ir/alumina catalysts In attempts to obtain information about the influence of the starting Ir compounds and Ir dispersion on the catalyst activity, the monometallic Ir/alumina catalysts were studied at first. The metal loading, content in carbon and chlorine and H/Ir values, evaluated from H2 adsorption, are summarized in Table 1. The values H/Ir of all catalysts exceeded slightly 1, which confirmed the presence of a well dispersed Ir phase. Differences between H/Ir values were rather small, despite different Ir precursors and preparation procedures. Table 1 Preparation, composition and dispersion of Ir/alumina catalysts Cat. Precursor Solvent Composition (wt%) Ir C C1 1 Ir4(CO) 12/alumina C6H12 0.85 0.23 2 Ir(AcAc)3/alumina C6H12 0.90 0.34 3 HzlrC16/alumina H20 1.10 0.12 0.86 4 (NH4)2IrC16/alumina H20 0.91 0.09 1.00 a Determined for reduced catalysts. Data taken from Ref. [10].
H/Ir a 1.06 1.19 1.26 1.18
The mean diameter of the Ir particles, calculated from H2 uptake, was near 0.91 nm. However, despite similar Ir amount and dispersion, the sample prepared from Ir carbonyl was more active than other samples (Fig. 1). We explain this difference by the lower amount of impurities (C and Cl) originating from the decomposition of the starting Ir compounds, rather than by the influence of the Ir dispersion. Differences in the H/ir values of the catalysts were rather small and did not allow to evaluate the effect of dispersion on activity. Thus, two additional samples with lower Ir dispersion were prepared by sintering of sample 3. This was achieved by calcination in air at 450~ As was shown by Foger and Jaeger [6], such treatment leads to the formation of IrO2 crystallites and to their segregation. After reduction, the H2 uptake on these catalysts was indeed much lower than on the original sample (Table 2). The mean diameter of the Ir particles, calculated from H2 uptake, increased roughly two times. Results of catalytic tests
446
of sintered catalysts and of the original sample 3 in HDN are shown in Table 2. Sintering of the catalysts suppressed significantly the rate constants of piperidine HDN, while it influenced only negligibly the rate constants of pyridine HY [10].
Fig. 1. Effect of precursor on activity of Ir/alumina sulfide catalysts in parallel HDN of pyridine and HDS of thiophene. Table 2 H2 adsorption, Ir dispersion and activity of reduced Ir/alumina catalysts in HDN Catalyst
Ha adsorption a (ml/gcat)
H/Ir a
dmean (nm)
Reduced It/alumina (No.3) 0.81 1.26 Sintered 450~ h 0.58 0.90 Sintered 450~ h 0.41 0.64 a Determined for reduced catalysts. Data taken from Ref.
0.9 1.2 1.7 [10].
kpy kc5 (mol/h.kgcat) 1.7 1.7 1.4
9.0 5.7 5.2
On the basis of these experiments, the carbonyl was chosen as the most suitable precursor for the preparation of the mixed Ir-Mo catalysts. At the same time, these experiments showed that it is desirable to achieve a good Ir dispersion, because it affects the rate of piperidine HDN and controls in this way the overall rate of transformation of pyridine.
3.2. Mixed Ir-Mo/alumina catalysts The prepared Ir-Mo catalysts contain almost the same amount of Mo and differ in the Ir loading and in the way of deposition of both components. The majority of these
447 catalysts was prepared by using the Ir carbonyl and the Ir amount was kept below 0.8 %. Precursors of the catalysts were prepared by three different routes, which are shown by the following scheme: Ir4(CO)12 MoO3/A1203
Ir4 (CO) 12-MOO3/A1203
MoSz/AI203
Ir4 (CO) 12-Mo 82/A1203 AHM
AHM-Ir4(CO)12/A1203 t:~
Ir4(CO) 12/A1203 t:~
Activation
IrSx+MoSz/AI203 (or
Ir+MoO3/A1203) (NH4)zlrCI6 +AHM A1203
~
AHM- (NH4) 2IrCI6/AI203
The precursors were transformed into active sulfide (or metallic Ir) state by activation, which was either sulfidation or reduction. The sulfided catalysts were examined by the TPR and by evaluation of activity in the parallel HDN/HDS reaction. Some reduced samples were studied by TEM in order to compare the size of the metallic Ir particles. The BET surface areas of all mixed catalysts was similar and close to 200-208 ma/g. The amount of sulfur found by ICP in the sulfided Mo/alumina catalyst corresponded to a S/Mo molar ratio of 1.9, similar to the values reported for the MoS2 phase. The ratios S/(Ir+Mo) of the mixed catalysts were between 1.6-2.3, which suggests the presence of elemental sulfur [11]. The list of precursors of the Ir-Mo catalysts is given in Table 3. Table 3 Preparation and composition of the mixed Ir-Mo/alumina catalysts Cat. 5 6 7 8 9 10 11 12 13 14 15
Precursor Ir4 (CO) 12-MoO3/alumina Ir4(CO) 12-MoO3/alumina Ir4 (CO) 12-MoSz/alumina Ir4 (CO) 12-MoO3/alumina (NH4)2IrCl6-AHM/alumina AHM-Ir4 (CO) 12/alumina Ir4(CO)12-MoO3/alumina AHM-Ir/alumina AHM-Ir4 (CO) 1z/alumina Ir4(CO) 12-MoO3/alumina Ir4(CO) 12-MoSz/alumina
Loading Ir 0.11 0.24 0.34 0.53 0.69 0.73 0.79 0.73 0.73 0.16 0.13
(wt%) Mo 8.4 8.4 8.4 8.4 8.7 9.1 8.4 9.1 9.1 8.8 8.8
Preparation Mo first Mo first Mo first Mo first Coimpregnation Ir first Mo first Ir first and reduced Ir first, simultaneous reduction Mo first, new series Mo first, new series
448
0 e,J) [] []
1
0 0.0
I
I
I
I
I
0.2
0.4
0.6
0.8
1.0
3
[]
0 0.0
0
b
I
I
I
I
I
0.2
0.4
0.6
0.8
1.0
I
I
I
I
I
0.2
0.4
0.6
0.8
1.0
m
r e~
m
~0
2
n
1
0 0.0
Ir loading, % Fig. 2. Activity of Ir-Mo sulfide catalysts in the parallel HDN/HDS reaction as a function of Ir loading, a- Thiophene HDS, b- Pyridine HY, c- HDN of piperidine. Mo (), Samples 5,6,8,11 (l-l), Sample 7 (11), Sample 9 (O), Sample 10 (~x).
Fig. 2 a-c shows the activities of the Ir-Mo/alumina sulfide catalysts in HDS of thiophene, HY of pyridine and HDN of piperidine during the parallel HDN/HDS, plotted against Ir amount in the catalysts. It is seen that addition of Ir to the Mo catalyst led to a substantial increase of activity. This increase was about 2 in HDS and about 3 in both steps of pyridine HDN. The data show that an optimum Ir amount in modified catalysts was found between 0.3-0.5 %. Above this Ir content, the activities in HDS and pyridine HY remained almost unaffected while activity in piperidine HDN clearly diminished. This decrease was explained by a diminution of the Ir dispersion, as evaluated from TEM measurements. The mean diameter of the majority of the Ir particles in reduced Ir-Mo sample with 0.53 % Ir was below 0.8 nm and some particles approached 1 nm. On the other hand, when the Ir amount increased to 0.79 %, the mean size of the majority of the particles approached 0.8-1.5 nm and the mean size of some smaller fractions (-10 %) increased up to 1.5-2.5 nm [11]. The effect of the way of deposition of Mo and Ir can be seen from the comparison of activities of samples 9-11. These samples had similar Ir loading but were prepared by different procedures, including deposition of Ir4(CO)12 from cyclohexane on an Mo/alumina catalyst, impregnation of an Ir4(CO)la/alumina precursor by an aqueous solution of AHM or coimpregnation of alumina by AHM and (NH4)2IrCI6. Fig. 2 a-c shows that the activities of these samples were in all reactions very similar. This suggests that the deposition order of both components, for Ir amount between 0.7 0.8 %, was of minor importance.
449 Correlation between the catalytic o activity in pyridine HY and relative ~176176 o~176176176176 reducibilities of the Ir-Mo sulfide cataZX ....... b ~0 lysts is demonstrated in Fig. 3. Similar m .~176176176176176176176176176176176176176176 [] ~176176176 dependence was also obtained for the ,~176176176176176176176176176176176176 ~1~ rate constants of thiophene HDS, as ~176176176176 recently shown elsewhere [11]. These ~176176176176176 .,.~.~176176 relations suggest that the activity of the mixed catalysts is closely related to the amount of hydrogen consumed during the TPR and, obviously, to the number I I 0 of reducible sulfur surface species. This 0.5 1.0 0.0 is in accordance with a generally accepted idea of the formation of anionic Relative reducibility, a.u. sulfur vacancies on the MoS2 phase, which are assumed to be the catalytic Fig. 3. Correlation between reducibility of sites in the HDS and hydrogenation the Ir-Mo sulfide catalysts and activity in the reactions. pyridine HY during parallel HDN/HDS. Denotations as in Fig. 1. Sample 12 (~), Sample 13 (v). A standard activation procedure for almost all mixed Ir-Mo catalysts was direct sulfidation of the precursors listed in Table 3. The only exceptions were samples prepared by deposition of AHM on reduced Ir/alumina catalyst (sample 12) or on an Ir4(CO)lz/alumina precursor and then reduced before sulfiding (sample 13). This aimed at keeping the Ir dispersion as high as it was in the starting It/alumina catalyst or, in the second case, at trying if such treatment in H2 could not lead to higher catalyst activity. The activity of samples 12 and 13 were compared with that of sample 10 with the same composition, but directly sulfided. Results of different activation procedures are shown in Table 4. The best activity was achieved in both HDS and HDN reactions after direct sulfidation of catalyst precursor as in sample 10. Table 4 Effect of activation on activity of Ir-Mo sulfide catalysts in parallel HDN of pyridine and HDS of thiophene kTH kpy kc5 Cat. Activation (mol/h.kgcat) 10 Sulfidation of AHM-Ir4(CO)12/alumina 1.9 2.5 3.0 12 Sulfidation of AHM-Ir/AI203 1.5 1.7 2.4 13 Reduction and sulfidation of AHM-Ira(CO)I2/AI203 0.9 1.5 1.1 Sample 12 was a little less active, which could probably be explained by a partial covering of an already developed Ir surface by Mo phase, decreasing in this way its accessibility. A third procedure, in which AHM and Ir4(CO)12 were reduced simultaneously before sulfiding, led to the lowest activity. In this case, it can be assumed that Ir4(CO)12 and AHM decompose easily to metallic Ir and MoO3 in H2 between 300-400~ [14,15]. On the
450
basis of our recent TPR study of the oxide Ir-Mo/alumina system, we speculate that the contact of well dispersed Ir particles with Mo oxide species could lead to the formation of some kind of non-reducible species [16]. Such an interaction can lead to a partial loss of Ir, keeping it in an Ir-O-Mo form. This assumption is consistent with earlier findings reported in the literature for the Ru-Mo system [1] and suggests to avoid thermal treatment of Ir-Mo catalyst precursor in the absence of HzS. The mixed Ir-Mo catalysts, in which Mo was deposited first, were prepared by impregnation of an oxidic or sulfided Mo/alumina catalyst. It was observed already that the sample prepared from presulfided Mo catalyst (sample 7), possessed a slightly higher activity than the samples prepared from oxidic Mo catalyst with comparable Ir loading. This is obvious from the comparison of the values k~i~ and kc5 plotted against Ir loading in Fig. 2. In order to confirm such an effect, an additional pair of mixed Ir-Mo samples was prepared by deposition of Ir4(CO)lz on a new charge of Mo/alumina catalyst (samples 14,15). The catalytic activity of both samples is compared in Fig. 4. The catalyst prepared from the presulfided Mo/alumina was clearly more active, approximately twice, than the catalyst prepared from oxidic Mo catalyst, in both reactions of thiophene and pyridine. A similar phenomenon was observed recently by Pinz6n et al. [17] after modification of sulfided and oxidic Mo/alumina catalyst by Ru, Rh and Pd, in HDS of dibenzothiophene and HY of naphthalene. The reason for the higher activity of the catalysts prepared by modification of a sulfided Mo/alumina matrix is not yet clear. We speculate that it could be connected with a closer contact between IrS• and MoSz phases in the catalyst, either through a higher Ir dispersion or due to a different distribution of Ir between the Mo phase and alumina.
Fig. 4. Effect of sulfidation of the Mo catalyst before Ir deposition on activity of Ir-Mo catalyst in parallel HDN/HDS. 4. CONCLUSIONS Deposition of Ir carbonyl, acetylacetonate, HzlrC16 and (NH4)zlrCI6 on alumina and subsequent reduction led to formation of Ir nanoparticles of about 1 nm. Direct sulfidation of deposited Ir precursors led to highly active Ir/alumina sulfide catalysts. A most active catalyst was obtained from Ir carbonyl. Addition of small amounts of Ir to Mo/alumina catalyst increased the reducibility of the MoSz phase, which possibly reflected in an enhanced activity
451 of the mixed catalysts in thiophene HDS and pyridine HY. However, activity in piperidine HDN decreased at Ir loadings above 0.5 %, in accordance with a decreased Ir dispersion. Other factors playing a major role were the state of the Mo phase before deposition of Ir, and the way of catalyst activation. Impregnation of the sulfided Mo catalyst by Ir4(CO)lz led to significantly higher HDS and HDN activities than impregnation of the oxidic Mo catalyst. The direct sulfidation of Ir-Mo catalyst precursors gave the most active catalysts. On the other hand, the starting Ir compound and the deposition order of Ir and Mo probably play a less important role. ACKNOWLEDGEMENT The authors thank the Grant Agency of the Academy of Sciences for financial support (grant A 4072103), Stid-Chemie A.G. (Germany) for providing the alumina carrier, and AIST (Tsukuba, Japan) for TEM measurements. REFERENCES
1. P.C.H. Mitchell, C.E. Scott, J.P. Bonnelle and J.G. Grimblot, J. Catal. 107 (1987) 482. 2. C.E. Scott, T. Romero, E. Lepore, M. Arruebarrena, P. Betancourt, C. Bolivar, M.J. P6rez-Zurita, P. Marcano and J. Goldwasser, Appl. Catal. 125 (1995) 71. 3. S.A. Giraldo de Le6n, P. Grange and B. Delmon, Catal. Lett. 47 (1997) 51. 4. A.S. Hirschon, R.B. Wilson Jr. and R.M. Laine, Appl. Catal. 34 (1987) 311. 5. A.G. Graham and S.E. Wanke, J. Catal. 68 (1981) 1. 6. K. Foger and H. Jaeger, J. Catal. 70 (1981) 53. 7. C.E. Scott, J. Guevara, A. Scaffidi, E. Escalona, C. Bolivar, M.J. P6rez-Zurita and J. Goldwasser, in: Stud. Surf. Sci. Catal., 130 (A. Corma, F.V. Melo, S. Mendioroz and J.L.G. Fierro, Eds.), Elsevier Science, 2000, p. 2813. 8. S. Eijsbouts, C. Sudhakar, V.H.J. de Beer and R. Prins, J. Catal. 127 (1991) 605. 9. Z. Vit and M. Zdrazil, J. Catal. 119 (1989) 1. 10. J. Cinibulk and Z. Vit, Appl. Catal. 180 (1999) 15. 11. J. Cinibulk and Z. Vit, Appl. Catal. 204 (2000) 107. 12. P. Marecot, J.R. Mahoungou and J. Barbier, Appl. Catal. 101 (1993) 143. 13. J.R. Anderson and K.C. Pratt, in "Introduction to Characterization and Testing of Catalysts", Academic Press (Harcourt Brace Jovanovish, Publishers), New York, p. 55, 1985. 14. K. Tanaka, K.L. Watters and R.F. Howe, J. Catal. 75 (1982) 23. 15. C. Thomazeau, V. Martin and P. Afanasiev, Appl. Catal. 199 (2000) 61. 16. Z. Vit and J. Cinibulk, React. Kinet. Catal. Lett. 72(2) (2001) 189. 17. M.H. Pinz6n, L.I. Merino, A. Centeno and S.A. Giraldo, in "Hydrotreatment and Hydrocracking of Oil Fractions", (B. Delmon, G.F. Froment and P. Grange, Eds.), Elsevier Science B.V., Amsterdam, 1999, p. 97.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
453
Peptization mechanisms of boehmite used as precursors for catalysts D. Fauchadour l, F. Kolenda 2, L. Rouleau 2, L. Barr~ 1, L. Normand 1 1 Institut Fran~ais du P6trole, 1-4 Avenue de Bois Pr6au, 92852 Rueil Malmaison Cedex, France 2 Institut Fran9ais du P6trole, CEDI "Rend Navarre", BP3, 69390 Vemaison, France In this work, using boehmite powders of different properties and under mild agitation, we determine key parameters, such as solid morphology, pH and ionic strength, and we identify the mechanisms which govern the peptization and account for observed behaviours (colloidal stability, sedimentation, gelification). We particularly reveal the role of dissolution on the dispersion and the aggregation of dense and open morphologies. This dissolution mechanism allows us to explain the behaviour of different boehmites during shaping processes and the textural properties of the elaborated supports. It is thus possible to explore better control of material forming and in particular of catalyst support elaboration. I. INTRODUCTION Boehmite is an important material in many fields such as petroleum, chemical or medicine industries. In the petroleum industry, it is mainly used as a precursor of gamma alumina (TAI203) which is a carder in the heterogeneous refining catalysts. Properties such as surface texture chemistry, porosity and thermal stability determine the performances of these supported catalysts [ 1]. In order to optimize the specific surface area, pore size and pore distribution of the alumina carrier, the best preparative schemes are sought. Making supported catalysts involves basically 4 steps : first is precipitation of boehmite (meaning synthesis of an aluminium oxyhydroxide) followed by washing and spray drying to produce a powder suitable for handling. The second step, called peptization, consists of the dispersion of the boehmite in an acidic solution in order to carry out a specific shaping technique such as oildrop or extrusion. The third step corresponds to a drying followed by a calcination at temperatures ranging from 500 to 700~ in order to transform the boehmite into the gamma phase structure of alumina. Metal deposition such as Pt, Mo or Co is finally performed by wet or dry incipient impregnation techniques [2]. Design of a tailored texture requires a good knowledge of the effect of each unit operation, and especially of the peptization, on the properties and characteristics of the support. For example, if one wants to obtain specific texture on the final alumina support, the particle size and the powder dispersion must be controlled during the peptization step. Moreover, use and performance of specific manufacturing process (such as oil-drop or extrusion) depends on the dispersion of boehmite and may even require the use of stable suspensions. If stability of boehmite has been studied as a function of pH, salinity and
454 concentration [3, 4, 5], very few studies take into account the nature of boehmite powders of different dispersability. This paper deals with 4 different kinds of boehmite used to identify key parameters controlling the behavior of boehmite during the peptization process. We begin to identify peptization mechanisms on a "model" boehmite that has been accurately characterized and which presents good peptization properties. Then we study mechanisms involved in the peptization process of 3 other kinds of boehmite with different properties in terms of peptization rate. Importance of dissolution is particularly emphasized. 2. MATERIALS AND PEPTISATION BEHAVIORS 2.1. Materials To succeed in this study, it is important to get a reference sample that can be dispersed to a rate of 100% in a specific test. This 100% peptizable boehmite powder (B 1) was provided by CONDEA Petrochemie Gesellshait. As indicated by the manufacturer, this boehmite is generated as a side product in the manufacture of alcohol straight-chain. We have verified this is a microcristalline boehmite of high purity and surface area (307 m2.g-1). The formula of the sample (A1OOHnHEO;n-0.45), as determined from thermogravimetric measurements, indicated a water excess from the well crystallized boehmite form (A1OOH), principally due to water physically adsorbed at the crystallite surface. X-ray diffraction and small angle X-ray scattering measurements showed the plate-like shape (length to width-~3) of the boehmite particles which have lengths of 100A and 30A along the a- and b-axes (planes (200), (020)) respectively. More detailed characterization on the powder initial state is given further in this paper. The three other boehmites were generated by aqueous precipitation. Two of them were provided by LA ROCHE CHEMICALS (B2 and B4) and one by PROCATALYSE (B3). They were chosen because they cover a large set of peptization properties in the reproducible peptization test described below.
Fig. 1. : Schematic representatlon ot Ume evolution ot -1 B1, B2, B3 and B4 suspensions in 0.1mol.l initial nitric acid solution. Note that B1 does not present any sedimentation
2.2. Boehmite peptization In order to be able to extrapolate further results to industrial conditions, boehmite sols were prepared by peptizing 6.67% weight of alumina in an aqueous solution containing various concentrations of nitric acid.
According to the weight concentration of boehmite and aggregation ionic strength limit determined elsewhere (0.1M to 0.15M for a monovalent salt [6]), initial concentration of acidic solution was 0.1 mol.1-1 (called classic peptization). Slow magnetic agitation was used
455 in order to obtain a homogeneous mixing of the powder with liquid without providing too much mechanical energy to the system. Following the peptization, one can identify 2 states: transitory and static. During the transitive state, 2 minutes after the end of the introduction of the boehmite powder, for 0.1mol.1l nitric acid initial concentration B1 is slightly opalescent and whitish. Under gentle agitation, B2, B3 and B4 are white and opaque homogeneous suspensions. After a few hours, an equilibrium pH is reached. If agitation is stopped, B 1 is still a stable suspension evolving slowly (over a few days) toward a low cohesive homogeneous gel. For the 3 other suspensions, stopping the agitation produces a fast sedimentation of a large part of the powder (sediment 1). If this sediment 1 is separated from this suspension, one can observe a slow sedimentation during a few weeks (sediment 2). These behaviors as well as the weight percentages of boehmite staying in the suspension are reported in Fig. 1 and Table 1, respectively. Table 1. Weight percentage of solid in suspension % weight of solid in susi~ensions (initial [HNOal = Suspension 1 Suspension 2 B1 100 100 B2 82 / B3 53 49 B4 33 29
In order to explain these differences we first characterized accurately properties of the powders of all the boehmite. Then we studied properties and behaviors of boehmite in acidic solution. We have then separated overfloatings from sediments in order to compare solids of these 2 phases to initial powder. Finally we discuss the results all together and conclude about the different mechanisms involved in the peptization process.
3. POWDER CHARACTERISTICS 3.1. Chemical and structural characteristics Characterizations by fluorescence X have shown various kinds of impurities, generally in small quantity, in the boehmites. Apart from giving us some hints on the synthesis process, these amounts of impurities can not be related directly to the peptization properties of the solids. XRD results have shown that there is no other crystalline phase in the initial powders and that the coherent domains are all around 3nm in thickness and 10nm in length. 3.2. Morphology characteristic SEM and TEM characterizations have shown that B1 boehmite is a "dense" agglomerate of platelet like crystallites developing a small dimension porosity (Fig. 2). B2,
456 B3 and B4 are made of two kinds of morphologies: a dense morphology similar to the one seen in B 1 and an open morphology with bigger porosity (Fig. 3). Although it is easy to identify tittle crystallites in the dense morphology (about 3nm thick and 10nm long like the coherent domains measured with XRD), it is impossible to identify clearly complete objects making up the open area. The large platelike curved crystals seen on TEM and SEM images are not independent from each others and this open morphology looks like a sponge. Thanks to XRD and HREM results, large plate-like curved crystals seem to be oriented agglomeration of small crystals. Because of the intimate mixing of the 2 phases, attempts to quantify these two morphologies from cross section images were not successful. N2 adsorption-desorption measurements show that BET specific areas are all about the same for the 4 powders (300m2g'l). However, total porosity is increasing going from B 1 to B4 and the isotherms show more and more macroporosity from B2 to B4. This suggests a higher quantity of open morphology.
Temperature(~ a ) .loo
o
0
'i I ....
0,05
~
0,1
"
lOO 200 ~ ','"
'I'
"'I
~1:
400 ~ '"'
I"
~
"I"
"
700 8oo 1....
I"~•
b'
~'''
TransitionoA1203
O H 2 ~ . . . Domaine 1 30a 140~ i
r~
,15
~ 0,2 r~ 0 0..~, 25 0,3 0,35
Fig. 4.
I ~,
B4
-
.......
B2
OHm--OH2 ,OH 2 OH 2 ~
'~ Domaine2
140/I230~
CRTA loss mass curve as a function of temperature for boehmites (a) and localisation model of water on crystallites for B 1 (b).
457 3.3. Boehmite-Air interface Because it has been shown that water at the surface could modify the dispersion properties [7], excess water of our boehmite has been studied by Controlled Transformation Rate Thermal Analysis (CRTA) [8]. This technique which respects the thermodynamic equilibrium allows to distinguish every type of adsorbed water and to estimate the apparent desorption activating energy [9]. 3 domains can be seen on the mass loss curve (Fig. 4) (apart from t h e 30/30~ domain corresponding to desorption of water due to relative humidity). The third domain corresponds to OH diffusion and OH desorption from the surface and from the structure due to phase transformation toward ~,-alumina. Domains 1 and 2 correspond to 2 different types of water. Considering the activating energies of these two domains (from 2 to 10 times lower than for phase transformation), it is clear that "excess" water (AIOOHnH20 ;n-0.45) compared to the alumina monohydrate A1OOH is not located between the octahedral double layers of the structure, but on the surface of plate-like boehmite crystals: water chimisorbed to A1 on [100] and [001] faces for the strongest links represented by domain 2 ; and water physisorbed to surface hydroxyls on [010] faces and to water molecules of the [100] and [001 ] faces. For the first time, this clarifies previous ATDATG experiments which showed an intermediate signal (between 100~ and 400~ attributed to possible strongly linked water on the surface [10,11 ]. This could validate as well the qualitative model proposed by Baker (Fig.4) [12]. Quantities of water on the surface of the 3 boehmites are not very different from each other and they correspond to 1 to 3 layers of water on surfaces of a model crystallite boehmite. However, the activation energy is a little bit higher for B 1, which could suggest a better crystallization or less amorphous microdomains in B 1 boehmite than in the other ones. We can finally conclude that B1 is of colloidal size and has a lot of water on the surface. B 1 could then behave as model particles weakly linked to each others. Although the other boehmite powders have about the same quantity of water on the surfaces, the open morphology identified on B2, B3, and B4 suggests particular contacts between elementary crystals. 4. BEHAVIOR IN LIQUID
Zero Charge Points (ZCP) measured by KNO3 addition technique showed that all boehmites have the same ZCP within the technique precision (8.9 0.1). Surface charge as a function of pH was measured on all the boehmites by comparing reference solutions (made from ultrafiltration of the corresponding boehmite suspensions) to boehmite suspemions. The results show that the variation of the surface charges for pH greater than 4 are all similar for the 4 boehmite powders. For pH smaller than 4, there is a slight divergence of the curves, probably due to a difference of dissolution below pH 4. Indeed, the study of pH change with time (which represents part of the dissolution) shows that there are important differences between powders, both in terms of pH variation rate and equilibrium pH (Fig. 5). These changes can be easily related to the variation of AI in solution as shown in Fig. 5 as well (measured by ICP on ultrafiltered solutions), pH is slowly increasing with time due to the dissolution kinetics of boehmite. Differences between the 4 boehmites are then due to differences of solubility (Fig. 6). In Fig. 6, one can clearly see that B1 has the lowest solubility since B1 solution always contains the smallest quantity of dissolved aluminum for a given equilibrium pH below 4.2.
458
Fig. 5.
: pH (a)and A1 in solution (b) variation as a function of time for an initial nitric acidic concentration of 0.1 mol.1-1 and 6.67% weight of boehmite.
However, it is surprising that at equilibrium and for a given initial quantity of acid, B 1 has the highest quantity of A1 dissolved (namely the lowest quantity of HNO3§ participating to the pH of the solution) and the lowest pH at equilibrium at the same time. In order to explain such a "paradox", calculations of surface charges were done by taking into account 1) the initial nitric acid introduced in the solution, 2) the A1 dissolution (by ICP), 3) the speciation of A1 (by NMR in order to know what kind of ion is present in the solution and what is the associated consumption of proton) and 4) the final pH at equilibrium. It was found that small differences of surface charge (estimated around 4 to 5 lxC/cm2) between B 1 and B4 can easily explain a greater dissolution of A1 and a lower pH at the same time.
4,0qg4.7-
~'~l
2'5 ~~ 2'05/ 1,5t-
1,01-
',
k'~
~
; \ ":~ /~ ~". "~ ~ ;~ _ 9 ,~. ~
~
,,., Fig 6 9AI in solution as a function ofpH at equilibrium for ---• .... B2 the 4 Boehmitepowders ---~--'B3 (solubilitycurves ) initial nitric ~ 1..acid eoneentralaons were 0, 0.035, 0.07, 0.1 and 0.15mol/, for
. . . . . .
ISI
"
"
"
3,2 3,4 3,6 3,8 4 4,2 4,4 4,6 4,8 5 pn We can finally stmmmrize that above pH 4.2 the 4 boehmite powders have the same behavior in terms of charge density. At pH 4.2 and below, there is dissolution of part of the alumina. Differences between powders can be interpreted in terms of difference of solubility probably due to the presence of a low crystallinity nano-phase on B2, B3 and B4 (that is not detectable in XRD). Dissolution of this phase is not zero at the beginning of the peptization but it is very low, especially for B 1. At the equilibritm~ charge density may be slightly different but keeps in the same order of magnitude. Charge density, dissolution and morphology seem then to be the main parameters to tmderstand the differences in peptization.
459 Characterizations by means of cryo TEM replica show that there are elementary particles dispersed in all the suspensions. This TEM work with light scattering studies of the overfloating suspensions 1 for the 4 powders show that they all contain aggregates with hydrodynamic mean Trace o~lementary p.a~_icles and aggregate~sofelementary particles radius in the order of 100nm (Fig. 7). For B 1, these aggregates are made of elementary particles as seen in the dense morphology of the dry powder. In B2, B3 and B4, it is interesting to see that some of these aggregates are made of elementary particles, whereas some others are clearly made of bigger objects identified in the open morphology. Thus these 2 morphologies have been separated by peptization and they can both constitute the suspension. Of course, the biggest aggregates will slowly segregate to form sediment 2, and suspension 2 has a smaller hydrodynamic mean radius (Table 2). Table 2 Hydrodynamic mean radius Rh (in nm) of suspensions and % weight of solid dispersed in suspensions 1 for different preparations. Classic Prep. No dissolution Increased dissolution (initial [HNO3] = 0,1mol.1l) Rh Rh % wt. Solid Rh % wt. solid Rh % wt. solid Suspens. Suspens. suspension Suspens. suspension Suspens. suspension B1 B2
1 110 110
2 / /
1 100 82
B3
130
90
53
B4
100
60
33
Fig. 8 SEM of sediment 1
1
1
1 100
170
1 100 38
150
250
39
185
85 94
150
21
showing only open morphology
By separating sediments from suspensions, we could see by SEM that sediments are all made of the open morphology with almost no trace of the dense morphology (Fig. 8). SEM on sediments 2 shows an agglomeration of
460 micro grains (of about 1~tm diameter) made of open morphology.
5. DISCUSSION 5.1. Peptization of BI B1 has a unique morphology made of a random arrangement of nanometric crystals. Furthermore, contacts between crystals, which are covered by 1 to 3 layers of water, are probably indirect and weak. This must involve an easy peptization under sott conditions as shown on alumina [7]. Actually, we could see that B1 is completely dispersed at the very beginning of the mixture, even when the dissolution is still very limited. This suggests that the increase of the charge density on surfaces of crystals (when [pH-ZPC [ increases) must be sufficient to "break" aggregates of the powder into colloidal particles. Peptization with no dissolution (with pH higher than 4.2), and with a comparable charge density adjusted by ionic strength shows a total peptization of B 1 with no sedimentation over a few weeks (table 2). This bears out that charging of surfaces is the main mechanism involved in the dispersion of B 1. Dissolution is not necessary for peptization of B 1 but it can modify the behavior of the suspension. Indeed, dissolution slowly increases the ionic strength of the solution. Taking into account the initial nitric acid quantity, pH, dissolved A1 in the solution, and speciation of aluminl'um ions (monomer [Al(H20)6]3§ by NMR), we can calculate the evolution
Fig. 9. Evolution of ionic strength calculated for B 1 classic peptisation from measured pH, AI concentration and speciation and initial nitric acid quantity introduced.
of the ionic strength I(t) with time (i.e. as a function of dissolution) with the formula: +
I(t) = 1/2 ([NO3"]initial + [H30 free](t) + 9[Alsolution](t)). Ionic strength reaches about 0.12moF1 after 3 days of peptization of 6.67% weight of alumina in an aqueous solution containing 0.1 mol.1-1 of acid nitric. Considering the aggregation ionic strength limit (around 0.1M to 0.15M for a monovalent salt), one can easily understand that this rise produces the formation of low cohesive gel due to aggregation kinetic of boehmite crystals over a few days.
5.2. Peptization of B2, B3 and B4 The dense morphology of the 3 other boehmite is completely dispersed during the classic peptization test used in this study. Similarities of properties with B1 suggest that this dispersion originates from charging of the elementary crystal surfaces. Sediments are made of open morphology only, indicating that contacts between objects of this morphology are more resistant than indirect solvated contacts between elementary crystals of the dense morphology. These contacts may be made of a lower crystallinity phase or an amorphous phase as it has been identified by CRTA and solubility measurements. However, part of the open
461 morphology can be dispersed in our test condition because of heterogeneity and dissolution. In a no-dissolution peptization, the quantity of solid in suspension is lower whereas it is largely increased when dissolution is increased (pH and ionic strength being constant, Table 2). This clearly indicates that open morphology is difficult to disperse with charging effect only. On the other hand, it is not an absolute obstacle to peptization since dissolution can break micro grains apart or separate micro grains from each others to make colloidal objects. Similarly to B1, if dissolution is too high, ionic strength may reach the aggregation limit and gel or flocculation may occur. 6. CONCLUSION This work allows us to propose a general representation of boehmite peptization, taking into account solid morphology, surface charge and dissolution. Powder made of a random agglomeration of nanometric crystals covered of water can only be dispersed by an increase of surface charge density (B 1). If the surface charge density is not high enough, the desagglomeration is not complete. In this case, aggregates up to a few hundreds of nm may subsist in suspension. They can create a bimodal pore size distribution like it has been seen in shaped alumina carrier. Other morphologies, like the ~ open morphology >>identified in this work, need more than only surface charging to be dispersed because they are made of bigger objects strongly linked together. Dissolution is then necessary to peptize such a morphology. Thus, one can understand why elaborated supports present mesopores around 9nm and macropores around 800nm when acidic concentration is not sufficient (2% for B4 with a few minutes of peptization, for example). When dissolution takes place (for 5 to 10% of acid for B4), macropores disappear because a progressive desegregation of open morphology. Finally, it is possible to play with the solid and the acidic concentrations and the ionic strength (that can be settled by different means) in order to control the rheology of the boehmite prior to the shaping and to obtain the targeted textural properties of catalyst supports. REFERENCES
1. R.K. Oberlander, in Applied Industrial Catalysis, Leach B.E. (Ed.), Vol. 3, 63, 1984. 2. J.F. Le Page, Preparation of Catalysts, Chapter 5, in Applied Heterogeneous Catalysts, Ed. Technip, p. 75-123, 1987. 3. J. Ramsay and S. Daish, J. Chem. Soc., Faraday Disc., (1978) 65. 4. C. Evanko, R. Delisio, D. Dzombak and J. Novak, Coll. Surf. A, 125 (1997) 95. 5. M. Van Bruggen, M. Donker, H. Lekkerkerker and T. Hughes, Coll. Surf. A, 150 (1999) 115. 6. F. Mange, Internal report IFP, 1998. 7. S. Desset, O. Spalla and B. Cabane, Langmuir 16 (2000) 10495. 8. D. Fauchadour, "Etude de la peptization de l'alumine boehmite", PhD Thesis, France, 2000. 9. J. Rouquerol, Thermochim. Acta, 144, (1989) 209. 10. P.A. Buining, C. Pathmamanohararg M. Bosboom, J.B.H. Jansen and H.N.W. Lekkerkerker, J. Am. Ceram. Soc., 73 (8) (1990) 2385. 11. P.A. Buining, C. Pathmamanoharan, M. Bosboom, J.B.H. Jansen and H.N.W. Lekkerkerker, J. Am. Ceram. Soc., 74 (6) (1991) 1303. 12. B.R. Baker and R.M. Pearson, J. Catal., 33 (1974) 265.
This Page Intentionally Left Blank
Studiesin Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
463
Influence of the treatment of Y zeolite by ammonium hexafluorosilicate on the physicochemical and catalytic properties: application for chlororganics destruction R. L6pez-Fonseca, J.I. Guti6rrez-Ortiz, B. de Rivas, S. Cibri~in, and J.R. Gonz~tlezVelasco* Departamento de Ingenieria Quimica, Facultad de Ciencias, Universidad del Pais Vasco/EHU, P.O. Box 644, E-48080 Bilbao, Spain. Phone: +34-94-6012681; Fax: +34-944648500; E-mail address:
[email protected] The objective of this work is to evaluate the dealumination via ammonium hexafluorosilicate treatment as an effective method for enhancing the catalytic performance of H-Y zeolite for oxidative destruction of chlorinated VOC. A series of Y zeolites with various Si/AI ratios was prepared from a commercial sample, then characterised and tested for the catalytic decomposition of chlorinated VOC (1,2dichloroethane and trichloroethylene). In general, these modified Y zeolites exhibited a higher activity with respect to that of the parent material, the zeolite subjected to 50% dealumination resulting in the most active catalyst. This increase in activity was associated with the development of strong Br6nsted acidity due to dealumination. 1. I N T R O D U C T I O N The increasing amounts of chlorinated volatile organic compounds (VOC), such as 1,2-dichloroethane (DCE) and trichloroethylene (TCE), released in the environment, together with their suspected toxicity and carcinogenic properties, have prompted researchers world-wide to find clean effective methods of destruction [1]. The abatement of chlorinated volatile organic compounds by catalytic combustion has been widely utilised in several technical processes. The lower temperatures required for catalytic combustion result in a lower fuel demand and can therefore be more cost effective than a thermal oxidation process [2]. In addition, the catalytic process also exerts more control over the reaction products and is less likely to produce toxic by-products, like dioxins, which may be generated by thermal combustion [3]. Most of the previous work related to catalysts for chlorinated VOC abatement is focused on the development of two type of catalysts, namely those based on noble metals and on transition metal oxides. By contrast, the utility of zeolites as effective catalysts for the decomposition of chlorinated organics has not been explored in detail, when it is reported that metal loaded catalysts employed in commercial applications are susceptible to deactivation by the HC1 and C12 produced during reaction [4]. In our previous works [5,6] it was found that H-zeolites showed a high activity for chlorinated VOC destruction under dry and humid conditions, and that their activity was controlled by the presence of
464 Br6nsted acidity. In the present study, an H-Y zeolite was dealuminated via the procedure described by Skeels and Breck [7,8] using ammonium hexafluorosilicate (AHFS) as the dealuminating agent under closely controlled conditions. The scope of this work is to analyse the catalytic behaviour of a series of H-Y zeolites with different Si/AI in the oxidative decomposition of chlorinated hydrocarbons (DCE and TCE) in air, at lean concentration conditions (around 1000 ppm) between 200 and 550~ 2. E X P E R I M E N T A L AND M E T H O D S
2.1.
Materials and zeolite preparation
The Y zeolite (CBV400) in its H-form (H-Y) was supplied from Zeolyst Corp. and used as received. The series of dealuminated samples H-Y(d) was prepared as follows: prior to dealumination the starting material was obtained by two successive ion exchanges with a 3 M ammonium nitrate solution of the commercial H-Y sample to reduce the sodium content. Then, the NH4H-Y zeolite was preheated in a 0.5 M ammonium acetate solution at 80~ An aqueous solution of ammonium hexafluorosilicate was added dropwise at a rate of 50 cm 3 h 1 under vigorous stirring. The (NH4)2SiF6-to-zeolite ratio was adjusted to remove 15, 30, 50 and 75% of the aluminium in the zeolite, respectively. Afterwards, the temperature was raised to 95~ and the slurry was kept at this temperature for 3 hours to ensure that silicon could be inserted into vacancies created by the extraction of aluminium. Finally, the zeolite was recovered by filtration and repeatedly washed with hot deionised water to remove the unreacted (NH4)2SiF6 completely. The zeolites were pelletised using methylcellulose as a temporary binder which was removed by calcination in air. Then the pellets were crushed and sieved to grains of 0.3-0.5 mm in diameter and used for catalytic runs without further activation.
2.2.
Catalyst characterisation
The BET surface areas of the zeolite samples were determined by N2 adsorptiondesorption at -196~ in a Micromeritics ASAP 2010 equipment. The adsorption data were treated with the full BET equation. The method was applied in order to obtain an estimation of the micropore volume. The determination of the compositions was carried out using a Philips PW 1480 wavelength dispersive X-ray fluorescence (XRF) spectrometer. The crystallinity and the unit cell size were established by a Philips PW 1710 X-ray diffractometer (XRD) with CuK~t radiation (~,=1.5406,&) and Ni filter. The number of aluminium atoms per unit cell, NAI, was calculated from a0 using the correlation given by Fichtner-Schmittler et al. [9]. The atomic framework Si/A1 ratio was derived from the calculated N AI. The number of extra-framework aluminium atoms per unit cell was calculated by the difference between the total aluminium, as determined by XRF analysis, and the framework aluminium N AI. Diffuse reflectance infrared (DRIFT) spectra of pyridine adsorbed on the zeolite samples were obtained with a Nicolet Proteg6 460 ESP spectrometer, equipped with a controlled-temperature and environment diffuse reflectance chamber (Spectra-Tech) with KBr windows and a liquid nitrogen-cooled HgCdTe detector. All spectra were collected in the range of 4000-1000 cm -1 averaging 400 scans at an instrumental resolution of 1 cm -1,
465
and analysed using OMNIC software. Temperature-programmed desorption (TPD) of ammonia was performed on a Micromeritics AutoChem 2910 instrument. Prior to adsorption experiments, the samples were first pre-treated in a quartz U-tube in a nitrogen stream at 550~ Subsequently, the desorption was carried out from 100 to 550~ at a heating rate of 10~ min -1 in an Ar stream (50 cm 3 minl). This temperature was maintained for 15 min until the adsorbate was completely desorbed.
2.3. Experimental device and product analysis Catalytic oxidation reactions were carried out in a conventional fixed bed reactor under atmospheric pressure [10]. The flow rate through the reactor was set at 500 cm 3 mini and the gas hourly space velocity (GHSV) was set at 15000 h -1. The residence time based on the packing volume of the catalyst was 0.24 s. Following the reactor, a portion of the effluent stream was delivered and analysed on-line using a Hewlett Packard 5890 Series II gas chromatograph (GC) equipped with an electron capture detector (ECD) and a thermal conductivity detector (TCD), and controlled with HP ChemStation software. The concentration of the chlorinated feeds was determined by the ECD after being separated in a HP-VOC column. 3. RESULTS AND DISCUSSION
3.1. Catalyst characterisation Expectedly, increasing amounts of AHFS added led to increased degrees of dealumination of the samples. For moderate dealumination levels (HY(d32%)>H-Y(d16%)>H-Y>H-Y(d64%). Hence, H-Y(ds0%) zeolite showed a light-off temperature or Ts0 (temperature at which 50% conversion was attained) of 265~ 100 H-Y lower than that of H-Y(d32%), H-Y(d16%) H-Y(d1 90 1 ~ H-Y H-Y(d3"/' and H-Y, 280, 300 and 325~ d5'2% respectively. H-Y(d64%), however, 80 t H.YId.o'/, f showed a less active behaviour with a Tso 70 i value of 350~ Unlike DCE, TCE combustion required significantly higher / temperatures [24,25]. Ts0 values were 475, 475, 500, 510 and 520~ over Ho 40 1 Y(dso%), H-Y(d32%), H-Y(d16%), H-Y and / H-Y(d64%), respectively. 30 The combined characterisation 211 ~ and catalytic evaluation of the Y zeolites 111 obtained by progressive dealumination via the (NH4)2SiF6 method revealed that 200 250 300 350 400 450 500 550 the strength of the acid sites had a Temperature, *C dominant effect on the catalytic behaviour [26,27]. The zeolite activity increased for Fig. 3. Light-off curves of DCE and TCE Si/A1 ratios from 2.6 to 6.2 since the combustion over Y zeolites. decline in the acid site density was more than compensated for by the concomitant increase in the population of acid sites with high strength. Upon further removal of aluminium (c.a. Si/AI=8.4) the catalytic activity destruction dramatically dropped due to
469 the decrease in the number of acid sites and a partial loss of crystallinity, as evidenced by the low conversion of H-Y(d64o/o) sample. Similarly, Greene et al. [28] and Prakash et al. [29] obtained a substantial improvement in C.C14 conversion when using a Y zeolite subjected to SIC14 dealumination. 4. CONCLUSIONS The scope of this work was to evaluate the catalytic performance of a series of (NH4)2SiF6-dealuminated Y zeolites for the oxidative decomposition of chlorinated VOC in dry air, at lean concentration conditions (around 1000 ppm) between 200 and 550~ The highly active performance of chemically AHFS-dealuminated zeolites for chlorinated VOC destruction could be accounted for by the generation of new strong acid sites, which were preferentially BrOnsted sites, due to dealumination treatment. It could be concluded that a zeolite with a modest concentration of BrOnsted sites, which were primarily of high acid strength, demonstrated to be effective for catalytic purposes. Likewise, it was established that chlorinated VOC oxidative decomposition was a type of reaction that required strong BrOnsted acidity. ACKNOWLEDGEMENTS
The authors wish to thank Universidad del Pais Vasco/EHU (9/UPV 0069.31013517/2001) and Ministerio de Ciencia y Tecnologia (PPQ2001-1364) for the financial support. R. L-F. acknowledges Ministerio de Educaci6n y Cultura for the FPI grant (QUI96-0471). REFERENCES
1. E.C. Moretti, Practical Solutions for Reducing Volatile Organic Compounds and Hazardous Air Pollutants, Center for Waste Reduction Technologies of the American Institute of Chemical Engineers, New York, 2001. 2. G.J. Hutchings and S.H. Taylor, Catal. Today, 49 (1999) 105. 3. J.C. Lou and Y.S. Chang, Combust. Flame, 109 (1997) 188. 4. J.J. Spivey and J.B. Butt, Catal. Today, 11 (1992)465. 5. J.R. Gonz~lez-Velasco, R. L6pez-Fonseca, A. Aranzabal, J.I. Guti6rrez-Ortiz and P. Steltenpohl, Appl. Catal. B, 24 (2000) 233. 6. R. L6pez-Fonseca, P. Steltenpohl, J.R. Gonz/tlez-Velasco, A. Aranzabal and J.I. Guti6rrez-Ortiz, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 130 (2000) 893. 7. D.W. Breck and G.W. Skeels, US Patent 4 503 023, 1985. 8. G.W. Skeels and D.W. Breck, in: D. Olson, A. Bisio, (Ed.), Proceedings of the 6th International Zeolite Conference, Butterworths, Guilford, 1984, p. 87. 9. H. Fichtner-Schmittler, U. Lohse, G. Engelhardt and V. Patzelova, Cryst. Res. Technol., 19 (1984) K 1. 10. J.R. Gonz/flez-Velasco, A. Aranzabal, J.I. Guti6rrez-Ortiz, R. L6pez-Fonseca and M.A. Guti6rrez-Ortiz, Appl. Catal. B, 19 (1998) 189. 11. Q.L. Wang, G. Giannetto and M. Guisnet, Zeolites, 10 (1990) 301.
470 12. A.P. Matharau, L.F. Gladden and S.W. Carr, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 94 (1995) 147. 13. A. Gola, B. Rebouis, E. Milazzo, J. Lynch, E. Benazzi, S. Lacombe, L. Delevoye and C. Fernandez, Microporous Mesoporous Mater., 40 (2000) 73. 14. H. Ajot, J.F. Joly, J. Lynch, F. Raatz and P. Caullet, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 62 (1991) 583. 15. A.V. Abramova, E.V. Slivinskii and E.A. Skryleva, Kinet. Katal., 39 (1998) 411. 16. J.A. Lercher, C. Cmindling and G. Eder-Mirth, Catal. Today, 27 (1996) 353. 17. T. Barzetti, E. Selli, D. Moscotti and L. Forni, J. Chem. Soc., Faraday Trans., 92 (1996) 1401. 18. G. Zi and T. Yi, Zeolites, 8 (1988) 232. 19. T. Masuda, Y. Fujiyata, H. Ikeda, S-I. Matsushita and K. Hashimoto, Appl. Catal. A, 162 (1997) 29. 20. A. Macedo, F. Raatz, A. Boulet, A. Janin and J.C. Lavalley, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 37 (1987) 375. 21. H.G. Karge and V. Dondur, J. Phys. Chem., 94 (1990) 765. 22. C.S. Triantafillidis, A.G. Vlessidis and N.P. Evmiridis, Ind. Eng. Chem. Res., 39 (2000) 307. 23. B. Chauvin, M. Boulet, P. Massiani, F. Fajula, F. Figueras and T. Des Couri6res, J. Catal., 126 (1990) 532. 24. R. L6pez-Fonseca, J.I. Guti6rrez-Ortiz, A. Aranzabal and J.R. Gonzalez-Velasco, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 135 (2001) 4995. 25. A. Aranzabal, J.A. Gonzhlez-Marcos, R. L6pez-Fonseca, M. A. Guti6rrez-Ortiz and J.R. Gonzhlez-Velasco, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 130 (2000) 1229. 26. R. L6pez-Fonseca, A. Aranzabal, J.I. Guti6rrez-Ortiz, J.I. Alvarez-Uriarte and J.R. Gonzhlez-Velasco, Appl. Catal. B, 30 (2001) 303. 27. A. Aranzabal, R. L6pez-Fonseca, J.R. Gonzhlez-Velasco, J.I. Guti6rrez-Ortiz, M.A. Guti6rrez-Ortiz and J.A. Gonz~.lez-Marcos, Abstr. Pap. - 221st Am. Chem. Soc. (2001) CATL-027. 28. H. Greene, D. Prakash, K. Athota, G. Atwood and C. Vogel, Catal. Today, 27 (1996) 289. 29. D.S. Prakash, K.V. Athota, H.L. Greene and C.A. Vogel, AIChE Symp. Ser., 91 (1995) 1.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
471
Preparation of SiO2 m o d i f i e d SnO2 a n d Z r O 2 w i t h novel t h e r m a l stability Y-X. Zhu, J-Y. Wei, L. Zeng, X-D. Zhao, W. Lin and Y.-C. Xie Institute of Physical Chemistry, Peking University, 100871 Beijing, China SnO2 and ZrO2 samples prepared by digesting precipitation in a glass flask or polytetrafluoroethylene beaker were investigated. It is found that silica could be dissolved from the glass flask during digestion in basic condition and existed in the samples obtained as a surface modifier, which significantly enhanced the surface area and thermal stability of the samples. Therefore, some silica-doped samples were prepared by adding silica sol into the precipitating system before digestion and similar results were observed. High surface area SiO2-ZrO2 and SiO2-SnOz samples with novel thermal stability were obtained. 1. I N T R O D U C T I O N Both SnOz and ZrOz are important catalysts and catalyst supports. SnOz is widely used in the selective catalytic reduction of NO because of its good hydrothermal stability as well as its fine oxidative selectivity [1,2] and SnOz-based composite oxides are very active catalysts for CH4 deep oxidation [3,4]. ZrOz as a catalyst or catalyst support is used in many catalytic processes [5]. Besides their wide applications in catalysis, SnO2 and ZrOz are useful materials as sensors, ceramics and solid electrolytes. Although the applications of SnO2 and ZrO2 are quite different, high surface area and good thermal stability are always indispensable for their properties. Much work has been done to enhance the surface area of SnOz and ZrO2. Suzuki et al. [6] reported a "solvent replacement" method to prepare high surface area SnO2 and got SnOz with a surface area of 108mZ/g after calcination at 500 ~ for 3 h. Xu et al. [7] investigated the promoting effect of additives on thermal stability of SnOz and obtained modified SnOz sample with a surface area of more than 40mZ.g -1 after calcination at 900 ~ for lh. For the preparation of high surface area ZrOz, the most interesting results were reported by Chuah and his coworkers [8-11]. They have obtained zirconia with surface area of > 90 mZ/g after calcination at 900~ for 12 h by adding 10wt% aqueous solution of zirconium(IV) chloride dropwise to 5M aqueous ammonia and then digesting the precipitate at 100~ for 96 h at a pH of about 9.4. They found that digestion of the hydrous zirconia in mother liquid with a pH value of 9-13 is the key to obtain high surface area zirconia without the necessity of adding other oxides or doping agents. We prepared SnO2 with a similar method [12] and
472
got SnO2 with a surface area of 53 m2/g after calcination at 1000~ for 2 h. This is really unusual for a pure SnO2 powder considering that the melting point of SnO2 is only about 1600 ~ We suggest that the digested sample might have some amount of silica on its surface to hinder its sintering because the hot basic solution can dissolve some silica from the glass bottle. Chuah [11] has reported that digestion at acidic condition (pH=3) resulted in a much lower surface area. This also gave a hint of the influence of silica, because silica could not come out of the glass flask at acidic condition. Satoshi Sato et al. [13] also reported that silica could be dissolved from the pieces of quartz glass tube immersed in the digesting solution and deposited on zirconia, and that ZrO(OH)2 precipitate could facilitate the dissolution of the glass chip. We prepared some SnO2 and ZrO2 samples with high surface areas using the digesting method and examined them with X-ray fluorescence spectroscopy to determine the content of silica. As we expected, some amount of silica existed in these samples and improved the thermal stability and surface areas of SnO2 and ZrO2. Therefore, some silica-doped samples prepared by adding silica sol into the precipitating system before digestion were also investigated. High surface area SiO2-ZrO2 and SiO2-SnO2 samples with novel thermal stability were obtained. 2. E X P E R I M E N T A L
2.1 Sample preparation The SnO2 and ZrO2 samples were usually prepared in a glass flask by adding NH3.H20 or NaOH solution and SnCI4 (or ZrOCI2) solution simultaneously to a certain amount of NH4HCO3 solution, keeping a pH 7-8 (this method is referred as co-current co-precipitation). The precipitate was digested at certain temperature for several hours in the mother liquid in a glass round-bottom-bottle, then filtered, washed with NH4HCO3 solution and distilled water until the filtrate was free of C1- ions as tested with 0.5M AgNO3. Then the product was washed twice with ethanol and dried at ll0~ followed by calcination at required temperature for 2 to 4 hours. The silica-flee sample P-ZrO2 was Table 1 prepared with the same method in a Physical properties of some SnO2 samples polytetrafluoroethylene beaker instead of a glass flask. Sample t (h) SiO2 (wt %) S (mZ/g) The silica-doped samples SnO2-D SnO2-1 0 0 34 and ZrO2-D were prepared in the SNO2-2 12 0.15 35 polytetrafluoroethylene beaker and a SNO2-3 24 0.26 38 certain amount of silica sol was added SNO2-4 36 3.6 74 Sn02-5 48 4.5 93 into the beaker after precipitation. The Calcined at 500~ for 4h. following digestion and after-treatment was always described.
the
same
as
above
473 2.2 Sample characterization The silica content of the samples was determined with a Rigaku 3271E X-ray spectrometer. BET surface areas were measured on a Micromeritics ASAP 2010 Analyzer. X-ray powder diffraction patterns were recorded on a Rigaku D/MAX-2000 with Cu Kc~ radiation (The common SnO2 samples were analysed with a BD-86 X-ray Diffractometer). DTA-TG measurements were carried out on a Thermal Analysis SDT 2960 with a heating rate of 10~ 3. RESULTS AND DISCUSSION 3.1 SnO2 samples prepared in glass flask Table 1 shows the compositions and surface areas of some SnO2 samples prepared by co-current co-precipitation method with NaOH solution as the precipitating agent and digested at 100~ for different times (t). Samples digested for less than 24 h contain no or very small amount of silica and possess lower surface areas. Sample number 4 digested for 36 h contains 3.6wt% silica and AI AI shows a surface area more than twice that of samples number 1 to 3. The sample digested for 48 h contains the largest amount of silica and consequently has the highest surface area. Fig.1 shows the XRD patterns of the samples calcined at 500~ ~ Sn02-1 (The A1 diffraction peak comes
' ~ ~ ~ a ~ Sn02.5 ~
Sn02-4
~
Sn02-2
'6'0'7'0
from the sample frame made of 2'0 3'0 4'0'5'0 aluminium). It can be seen that all samples show a diffraction pattern of SnOz without any peak Fig. 1 XRD patterns of some SnOz samples characteristic for silica, and that the longer the digestion time, the Table 2 weaker and broader the SnO2 peaks, Physical properties of SnOz-D samples indicating a smaller particle size and lower crystallinity. Sample SiO2 S (mZ/g) After calcination at higher (wt %) 500~ 800~ 1000~ temperature, namely 800 ~ and SnOz-D1 5.2 206 166 132 SnO2-D2 9.5 216 169 49 1000 ~ the surface area of sample SnOz-5 remained at 75mZ/g and 53mZ/g, respectively, while the
20/~
474 samples with no or less silica sintered severely and had very low surface areas. Evidently, silica has a significant effect on the surface area and the thermal stability of the sample.
_
SnO~-D2
,ooo~
20 i
~"
30 'i
',
40 i
'"
,
5'0
',
60 f"
20/~
,
~
1000~
800~
800oc
oooc
ooooc
7'0
2'0
,
3'0
,
40 i
,
""
50 !
,
60 i
'
'
'
70 i
20/o
Fig. 2 XRD patterns of silica-doped SnO2 samples calcined at different temperatures
3.2 Silica-doped SnO2 samples Two silica-doped samples were also investigated. Table 2 lists the compositions and surface areas of the samples. The XRD patterns of these two silica-doped samples calcined at different temperatures are shown in Fig. 2. It can be seen that the silica-doped samples show quite high thermal stability, especially SnOz-D1, with a surface area as high as 132 mZ/g after calcination at 1000~ Both samples with a silica content of 5.2wt% and 9.5wt% show no diffraction peaks of silica, only the peaks of SnOz. Samples calcined at 500~ and 800~ have much broader diffraction peaks than those calcined at 1000~ and the corresponding surface areas are also quite high, more than 160mZ/g. At low calcination temperature, sample SnOz-D2 with higher silica content has a higher surface area than sample SnOz-D 1 with lower silica content, but its thermal stability is not so good as sample SnOz-D1, its surface area decreases to 49mZ/g after calcination at 1000~ This is probably because of the aggregation of amorphous silica. According to the "close-packed" monolayer model [14], the utmost monolayer dispersion capacity of SiOz on the surface of the support is about 0.048g/100m 2 or 0.081g/g SnOz (169mZ/g), so 9.5wt% silica is higher than this value and therefore amorphous silica species besides monolayer-dispersed silica might be formed. Details still need further investigation with solid state NMR and other techniques.
475
3.3 ZrO2 samples prepared in glass flask and silica-free P-ZrO2 sample Some zirconia samples were Table 3 also prepared using NaOH or NH3 Physical properties of some ZrO2 samples solution as the precipitating agent digested for 48 h at 100~ (A) and 30~ (B) and digested at 100~ or 30~ Sample SiO2 (wt %) S(m2/g) respectively for 48 h followed by ZrO2-Na-A 3.2 176 calcination at 500~ for 4 h. The ZrO2-Na-B 1.8 125 silica contents and surface areas of ZrO2-NH3-A 2.3 163 these samples are listed in Table 3. ZrO2-NH3-B 1.9 122 Samples digested at 100~ have more silica than the samples Table 4 digested at 30~ This is because Composition and surface areas of some ZrOz samples more silica can be dissolved from SiO2 Surface area (m2/g) the glass flask at higher temperature. Sample (wt %) 600~ 800~ 1000~ Similar to the results of SnO2 P-ZrO2-13 a 0 35 24 19 samples, the higher the silica ZrO2-6 a 0.89 111 73 33 content, the higher the surface area ZrO2-9 a 2.0 157 85 34 of the sample. XRD analysis ZrO2-13 a 5.2 218 140 64 a pH after digestion 3,0
4.4.
,--- 2 . 5 =.
" .O'I
I--
c
4.2
2.0
1.5
.,_.,
4.0
"~
3.8
c
b 1.0
0.5
3.6
0
.
, 200
.
, 400
. T/~
, 600
.
, 800
.
., 1000
a
3.4 0
'
200
'
460
660
860 '1oo6
T/~
Fig. 3 DTA-TG results of some hydrous zirconium oxide samples The corresponding oxides are: a. ZrO2-6; b. ZrO2-9; c. ZrO2-13 (figures not shown) finds no silica species, only tetragonal ZrO2, indicating that silica is probably in a highly dispersed state on the surface of the sample as a surface modifier. Further experiments confirmed the above prediction. Several silica-containing samples were prepared by digesting hydrous zirconium oxide at 100~ for 24 h at different pH in a glass flask and a silica-free sample was prepared in similar conditions in a polytetrafluoroethylene beaker with a cover. The resulting hydrous oxides dried at l l0~
476
were analyzed with DTA-TG technique and the surface areas of the samples calcined at different temperatures were also measured. Table 4 lists the composition and surface areas of the zirconium oxide samples. Fig. 3 shows the DTA-TG results. The silica-free zirconium hydrous oxide gives no crystallization peak in DTA-TG measurement. XRD analysis (Fig. 4) shows that the pure zirconium hydrous oxide dried at l l0~ is in a well-crystallized monoclinic state with small amount of tetragonal phase. Obviously, the existence of silica
j, 2'0'3'0'4'0'5'0'6'0'7'0 20/~
Fig. 4 XRD pattern of pure zirconium hydrous oxide dried at 110~
P-Zr02-13 1O00~ 1ooo~
~ ~ 2~3
800~
8oooc
600~
600~
3'0
.
4'0'5'0 20/~
.
.
.
6'0
,
. . . .
7'0
7'0
2'0'3'0'4'0'5'0'6'0 20/~
ZrO2-9
Zr02-13
__j
~
..... _A.,,_ ....... 1O00~
1O00~
__.J 2'0
800~
800~ 600~
3'0
4'o
20/~
~'o
6'o
7'0
600~ 2'o
'
3'o
'
4'o
s'o
'
6'o
'
20/~
Fig. 5 XRD patterns of some ZrO2 samples calcined at different temperatures
7'O
477 significantly elevates the crystallization temperature (Fig. 3) of the hydrous oxide as well as the surface areas of the corresponding oxides. Higher digesting pH results in higher silica content and consequently higher crystallization temperature and larger surface area of the corresponding oxide. The oxide samples calcined at different temperatures Table 5 were also characterized by XRD (Fig. 5). Surface areas of ZrOz-D samples Similar to the above-mentioned results, Surface area (m2/g) there is no diffraction peak of silica in all Sample 600oc 800~ 1000~ the samples. The pure zirconium oxide ZrOz-D1 241 136 95 P-ZrO2-13 displays mainly the peaks of ZrO2-D2 233 196 139 monoclinic ZrO2. However, tetragonal ZrO2 is the dominant phase in the samples containing a certain amount of silica. Not surprisingly, the higher the silica content, the higher the percentage of the tetragonal phase, especially the sample ZrOa-13 with the largest amount of silica. It consists of little monoclinic phase even after calcination at 1000~ for 4 h. All the results reveal that silica in the sample exhibits a typical effect of surface modification.
3.4 Silica-doped ZrO2 samples Two silica-doped samples with silica content of 2.7wt% (ZrOz-D1) and 5.2wt% (ZrO2-D2) respectively were also prepared. Their surface areas are listed in Table 5. The XRD patterns are shown in Fig. 6.
I
ZrO2-D1
ZrO2.D2
1000~ 800~ 2'0'3'0'4'0'5'0 20/o
6'0'7'0
600~
2'0'3'0'4'0'5'0'6'0' 20/~
1000~ 800~ 600~ 7'0
Fig.6. XRD patterns of silica-doped ZrO2 samples
The surface areas of silica-doped samples are higher than those of the samples listed in Table 4 though the silica content is comparable. ZrOa-D2 with 5.2wt% silica has a
478
specific surface area of 139m2/g after calcination at 1000~ for 4 h. From Fig.5 and Fig.6, it can be seen that the phase compositions of the two kinds of samples are also different. Both ZrO2-D1 and ZrO2-D2 contain monoclinic phase after calcination at 600~ 800~ and 1000~ However, ZrO2-9 and ZrO2-13, when calcined at 600~ and 800~ show only tetragonal phase. When calcined at 1000~ ZrO2-9 and ZrO2-13 exhibit only very small amounts of monoclinic phase. This can be probably attributed to the difference in the precipitation process. ZrO2-D1 and ZrO2-D2 were prepared in a polytetrafluoroethylene beaker, and silica sol was added before the digestion, so there is no silica source during the precipitation. Since ZrO2-9 and ZrO2-13 were prepared in a glass flask, in the course of precipitation, traces of silica dissolved from the glass vessel could deposit on the freshly formed ZrO(OH)x precipitate and help to hinder the formation of monoclinic ZrO2. The inhibition of monoclinic phase by silica can also be observed in the silica-doped samples. As can be seen in Fig. 6, after calcination at 600~ 800~ and 1000~ ZrO2-D2 with higher silica content always contains less monoclinic phase than ZrO2-D1. 4. CONCLUSION SnO2 and ZrO2 prepared by co-current co-precipitation and digestion in basic conditions in a glass flask contain certain amount of silica, and the silica as a surface modifier can improve the thermal stability and surface areas of SnO2 and ZrO2 by hindering their sintering. This is the main reason for the high surface areas and good thermal stability of thus prepared SnO2 and ZrO2. Silica-doped SnO2 and ZrO2 were also prepared and investigated. These samples exhibit novel thermal stability. 5.2wt% SiO2/ZrO2 maintains a specific surface area of 139 m2/g after calcination at 1000~ for 4 h, while 5.2wt% SiO2/SnO2 exhibits a specific surface area of 132 m2/g after calcination at the same temperature for 2 h. Silica can also help hinder the formation of monoclinic ZrO2.The effects of digestion on silica-flee sample and precipitation process on ZrO2 structure need further investigation. ACKNOWLEDGEMENT We gratefully acknowledge the financial support from National Science Foundation of China (29803001) and The Major State Basic Research Development Program (Grant No. G2000077503) REFERENCES
1. M.C. Kung, E W. Park and D. W. Kim, J. CataL, 18 (1999) 1. 2. J. Ma, Y. X. Zhu and J. Y. Wei, Stud. Surf. Sci. Cata[, Elsevier, Amsterdam, 130 (2000) 617.
479 3. 4. 5. 6. 7. 8. 9. 10. 11.
X. Wang and Y. C. Xie, Chem. Lett. (2001) 216. X Wang and Y C Xie, Appl. Catal. B, 35 (2001) 85. T. Yamaguchi, Catalysis Today, 20 (1994) 199. K. Suzuki, A. Sutsuma and H. Yoshida, Chem. Lett., 1997, 279. C. Xu, J. Tamaki and N. Miara, J. Mater. Sci. Lett., 8 (1989) 1092. P. Fornasiero, R. Di Monte and J. Kaspar, J.Catak, 151 (1995) 168. G.K. Chuah and S. Janenicke, Appl. Catal. A: General, 163 (1997) 261. G. K. Chuah, S. Janenicke and B. K. Pong, J. Catal,. 175 (1998) 80. G. K. Chuah, S. H. Liu, S. Janenicke and J. Li, Microporous Mesoporous Materials, 39 (Z000) 381. 12. J. Y. Wei, Y. X. Zhu and Y. C. Xie, Acta. Phys. -Chim. Sift, 17 (2001) 577. 13. S. Sato, R. Takahashi and T. Sodesawa, J. Catal., 196 (2000) 190. 14. Y. Xie and Y. Tang, Adv. CataL, 37 (1990) 1.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
481
Control of the textural properties of cesium 12-molybdophosphatebased supports S. Paul w V. Dubromez w L. Zatr, " + M. Fourmer, " + D. Vanhove w Laboratoire de Catalyse de Lille (ESA8010 CNRS) Equipe G6nie Chimique, Ecole Centrale de Lille et E.N.S.C.L., BP 48 59651 Villeneuve d'Ascq Cedex France + Equipe Catalyse H6t6rog6ne, Universit6 des Sciences et Technologies de Lille, Batiment C4, 59655 Villeneuve d'Ascq Cedex France
* Corresponding author- dominique,
[email protected] An experimental design is applied in order to point out the influence of the more significant preparative parameters on the textural properties of Cs3PMol2040-based supports. With this in mind, a Hadamard matrix design including seven factors (i.e. stoichiometry, reactants concentrations, reaction temperature, addition rate, addition order, maturation time and thermal treatment procedure) has been used. Syntheses were carried out in an especially designed reactor in order to be able to control carefully mixing, temperature, concentrations and addition rate of reactants. Analysis of the results leads to the determination of optimal operating conditions to prepare supports with reproducible and controlled textural properties. 1. INTRODUCTION 12-Molybdophosphoric and 1-vanado, l l-molybdophosphoric acids are Keggintype heteropolyacids (HPA) well known as catalysts for mild oxidation reactions [1-15]. Nevertheless, these solids suffer from poor thermal stability and thus deactivate progressively under reaction conditions [16-19]. For this reason, coupled with their low surface area (1-5 m2/g), the influence of their dispersion on a support aiming at enhancing and stabilising the activity of the catalyst was tested. Various types of solids have been studied to play this role (silica, alumina, carbon, titania,...) but it is generally found that at low HPA contents strong interactions between the support and the active phase occur leading to the degradation of the latter [20, 21]. Promising results have been obtained by doping a silica support with alkaline ions [22]. In this case it is proposed that a heteropolysalt interface is formed between the silica network and the heteropolyacid crystal orientating and stabilising the structure of the active phase. Other interesting performances are also achieved when heteropolyacids are directly supported on a
482 heteropolyacid alkaline salt [23, 24]. Recently, we reported [25] that H4PMollVO40 deposited onto CsaHPMollVO40 is 15 times more active for the selective oxidation of isobutane to methacrylic acid and is also more stable than the pure acid alone. It was also observed in that work that apparently identical synthesis procedures lead to different solids as far as their textural properties are concerned. Moreover, Karmakar et al. [26-28] have shown a direct relation between the initial selectivity for acrylic acid and the proportion of porous volume related to large mesopores and macropores (pore radii Rp> 10 nm) in HPA supported-catalysts used for the selective oxidation of propane. This result shows that the textural properties of heteropolyanionic supports are the key parameters for the achievement of good selectivity. Lapham and Moffat [29] studied the influence of preparative parameters over several heteropolyoxometalates but the emphasis was put on 12-tungstophosphates. The role of thermal treatment was underlined and it was shown that the higher the temperature, the lower the surface area. In this context, cesium salts were much more stable than other heteropolysalts. Among the HPA alkaline salts, Cs3PMoI2040 (further noted Cs3) seems to be a judicious choice to play the role of support. Indeed, this solid is easy to prepare by a simple cationic exchange between H3PMo12040 (further noted H3) and a cesium salt. Moreover, Cs3 presents a good thermal stability and a high surface area is often reported [15,22,30,31]. Nevertheless, as far as this textural property is concerned, a great variation is found in the literature depending on the operating conditions used for the synthesis and the calcination [8,32,33]. In this study, we tried to clarify this important point by implementing a careful control of the Cs3 preparation procedure. A more particular attention was paid to the chemical engineering aspect of the reactor (agitation, temperature and concentrations homogeneity and addition rate of the reactants). Classically, the research of the optimal preparative conditions of a solid consists in a first step of listing the more significant parameters and then in varying intuitively a single one at a time while keeping the others constant. This work leads to a large number of tedious experiments and is uncertain to reach the objectives because of the potential interactions between the parameters. Experimental design avoids these drawbacks and enables the determination of the influence of the parameters with a minimum of experiments. Seven factors (i.e. stoichiometry, reactants concentration, reaction temperature, addition rate, addition order, maturation time and thermal treatment procedures) were studied using a Hadamard matrix design [34] with the objective to better understand their influences on the surface area, porous volume and porous volume distribution of the supports.
2. EXPERIMENTAL 2.1 [13 and Cs3 syntheses According to previous results [35], the synthesis of H3PMoI2040 was achieved in two steps: the formation of the acidic salt Na2HPMol2040,xH20 and its dissolution by acidification and purification by ether extraction. i) 218.8g of NaEMoO4,2 H20 (0.9 mole) were dissolved in 317 ml of deionised water. 5.15 ml of HaPO4 (85%) and then 250 ml of HC104 (60%) were added dropwise to
483 the solution. Pale yellow crystals of Na2HPMoI2040 precipitated and were collected by filtration and dried overnight at ambient temperature. ii) The disodic salt was dissolved in 4 ml/g of a 10% HC1 solution. A red orange solution was obtained. HaPMo12040 was extracted as heavy layer by diethyl ether, and then a quantity of water equivalent to half of the volume of the organic phase was added to it. After evaporation of the ether, the remaining aqueous solution was placed at 4~ to crystallise. The hydrated crystals (29 H20) were dried under air flow leading to the room temperature stable hydrated form (13 H20). Cs3 syntheses were performed by a simple cationic exchange between Ha (issued from the same batch for all preparations) and Cs2CO3 in a thermostated vessel; cesium carbonate being chosen to avoid the presence of residual counter-anions in the final solid. The reactor was especially designed to permit a constant and controlled addition rate of the reactants and an efficient and reproducible mixing. To this purpose, the vessel was equipped with baffles avoiding a vortex formation in the liquid and achieving therefore a quick mixing which was checked using a colorimetric tracer. Ha and Cs2CO3 solutions were both thermostated at reaction temperature (one in the reactor and the other (the so-called added reactant) in a separate vessel) before starting the addition, the rate of which was controlled by a peristaltic pump. In the reactive media, the pH was constantly monitored during the reaction. During the maturation, the reactor was kept under constant stirring and at constant temperature (the same as during the reaction). The mixture was then evaporated under vacuum at 70~ and ground in a mortar. In order to try to stabilize the properties of the support, a calcination was carded out. Two different procedures were followed. In the first one, the cold solid was placed in a furnace at 100~ and immediately heated up to 200~ (50~ The temperature was kept constant at this level for 2 hours and was then risen at 350~ (100~ and stabilised for 3 hours. The furnace was then switched off and allowed to cool down to ambient temperature overnight. The second procedure consisted in putting the cold solid in the furnace directly at 200~ and then in following the same thermal treatment as above. All the supports were then analysed by N2 adsorption-desorption over an ASAP 2010 Micromeritics apparatus after outgassing for 4h at 200~ BET [36] and BJH [37] methods were used to determine surface areas, porous volumes and porous volume distributions. The reproducibility of the synthesis and calcination procedure as far as textural properties are concerned was checked and validated.
2.2 Experimental design Hadamard matrices of experiments are generally used to point out the more influent qualitative and/or quantitative factors within a given experimental domain. In this method, two levels are attributed to the factors (noted -1 and +1) as presented in Table 1. To study the seven factors mentioned above, eight experiments are needed. The matrix of experiments is presented in Table 2 where each line corresponds to a synthesis while the columns correspond to the factors. Estimations of the effect of each factor were calculated by adding the responses modified by the sign of the level for the considered factor and by dividing this sum by eight.
484
The responses studied were the surface area, the porous volume and the porous volume distribution split in 3 classes of pore radii (Rp10nm), a synthesis has been done with the operating conditions presented in Table 5. It can be noticed that the values have been changed compared to the Hadamard matrix levels in a view to amplify the effects. The textural properties of the solid obtained are in
487 good accordance with the expected ones. The proportion of porosity attributed to large pores is actually 77% which is higher than the best result obtained in the Hadamard experimental design. The reproducibility of this particular synthesis and calcination as far as textural properties are concerned have been checked and validated as shown in Fig.1.
E
0,0020
f\
i, !
C
E
0,0015
i,/ I J"
E
/ 'L /
0
> 0,0010
ill'
0
>
,\ \
i
0,0005 r~f
\
\
'k,
0,0000 1
10
100
Pore Radius (nm)
Fig. 1 9BJH desorption derivative pore volume 5. C O N C L U S I O N This study has shown the strong influence of preparative operating conditions on the textural properties of the cesium 12-molybdophosphate-based supports. The use of stoichiometric quantities of reactants leads to a high surface area and microporous solid probably close to pure Cs3PMo12040 whereas high reactants concentrations favour the porous volume. Pores distributions can also be adjusted by modifying the preparative operating conditions. The formation of large pores is actually favoured by the use of an excess of Cs2CO3 and a long maturation time. In a way, the selection of the adapted operating conditions allows to design a support with "tailor made" textural properties. Experiments are now in progress in order to disperse an active phase on these supports and compare the reactivity of the catalysts thus obtained. REFERENCES 1. T. Okuhara, N. Mizuno and M. Misono, Adv. Catal., 41 (1996) 113. 2. N. Mizuno and M. Misono, Curr. Op. Sol. Sta. & Mat. Sci., 2 (1) (1997) 84. 3. I.V. Kozhevnikov, Chem. Rev., 98 (1) (1998) 171. 4. N. Mizuno and M. Misono, Chem. Rev., 98(1) (1998) 199. 5. M. Ai, J. Catal., 71 (1981) 88. 6. H. Mori, N. Mizuno and M. Misono, J. Catal., 131 (1991) 133. 7. N. Mizuno, T. Watanabe and M. Misono, Bull. Chem. Soc. Jpn., 64 (1991) 243. 8. K. Eguchi, I. Aso, N. Yamazoe and T. Seiyama, Chem. Lett. (1979) 1345.
488 9. L. M. Deusser, J. C. Petzoldt, J. W. Gaube and H. Hibst, Ind. Eng. Chem. Res., 37 (1998) 3230. 10. Y. Konishi, K. Sakata, M. Misono and Y. Yoneda, J. Catal., 77 (1982) 169. 11. J. Hu and R. C. Burns, J. Catal., 195 (2000) 360. 12. T. Ilkenhans, B. Herzog, T. Braun and R. Schl6gl, J. Catal.,153 (1995) 275. 13. M. Akimoto, H. Ikeda, A. Okabe and E. Echogoya, J. Catal., 89 (1984) 196. 14. T. Haeberle and G. Emig, Chem. Eng. Technol., 11 (1988) 392. 15. G. B. Mc Garvey and J. B. Moffat, J. Catal., 132 (1991) 100. 16. C. Rocchiccioli-Deltcheff, A. Aouissi, M.M. Bettahar, S. Launay and M. Fournier, J. Catal. 164 (1996) 16. 17. O. Watzenberger, T. Haeberle, D.T. Lynch and G. Emig, New Devel. in Select. Oxid., Stud. Surf. Sci. Catal., Vol. 55, Elsevier Science Publishers, Amsterdam, 1990, 843 18. G. Lischke, R. Eckelt and G. Ohlmann, React. Kinet. Catal. Lett., 31(2) (1986) 267. 19. G. Mestl, T. Ilkenhans, D. Spielbauer, M. Dieterle, O. Timpe, J. Kr6hnert, F. Jentoft, H. Kn6zinger and R. Schl6gl, Appl. Catal. A: Gen., 210 (2001) 13. 20. P. G. Vazquez, M. N. Blanco and C. V. Caceres, Catal. Lett., 60 (1999) 205. 21. M. Prevost, Y. Barbaux, L. Gengembre and B. Grzybowska, J. Chem. Soc., Faraday Trans., 92(24) (1996) 5103. 22. C. Desquilles, M. J. Bartoli, E. Bordes, G. Hecquet and P. Courtine, Erdol Erdgas Kohle, 109(3) (1993) 130. 23. K. Briickman, J. Haber, E. Lalik and E. M. Serwicka, Catal. Lett., 1 (1988) 35. 24. K. Brfickman, J. M. Tatiboui~t, M. Che, E. Serwicka and J. Haber, J. Catal., 139 (1993) 455. 25. M. Sultan, PhD thesis report, Universit6 de Technologie de Compi6gne, n~ D1215, 19/07/1999. 26. S. Karmakar, A. F. Volpe Jr., P. E. Ellis Jr. and J. E. Lyons, US Patent 6043184, 03/28/2000, assigned to Sunoco Inc. and Rohm and Haas. 27. J. E. Lyons, A. F. Volpe Jr., P. E. Ellis Jr. and S. Karmakar, US Patent 5990348, 11/23/1999, assigned to Sunoco Inc. and Rohm and Haas. 28. A. F. Volpe Jr., J. E. Lyons, P. E. Ellis and S. Karmakar, Prep. Am. Chem. Soc., Div. Pet. Chem., 44(2) (1999) 156. 29. D. Lapham and J. B. Moffat, Langmuir, 7 (1991) 2273. 30. B. Che|ighem, S. Launay, N. Essayem, G. Coudurier and M. Fournier, J. Chim. Phys., 94 (1997) 1831. 31. M. Akimoto, Y. Tsuchida, K. Sato and E. Echigoya, J. Catal., 72 (1981) 83. 32. C. Marchal-Roch, N. Laronze, R. Villaneau, N. Guillou, A. T6z6 and G. Herv6, J. Catal., 190 (2000) 173. 33. N. Mizuno, M. Tateishi and M. Iwamoto, J. Catal., 163 (1996) 87. 34. R. Perrin and J.P. Scharff, ~ Chimie industrielle ~, Vol.1, Paris, Masson, 1993. 35. C. Rocchiccio|i-Deltcheff, M. Fournier, R. Franck and R. Thouvenot, Inorg. Chem., 22 (1983) 207. 36. S. Brunauer, P. H. Emmet and E. Teller, J. Am. Chem. Soc., 60 (1938) 47, 309. 37. E. P. Barret, L. G. Joyner and P. O. Halenda, J. Am. Chem. Soc., 73 (1951) 373, 104, 114, 116, 127.
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
489
MnOx/CeO2-ZrO2 and MnOx/WO3-TiO2 catalysts for the total oxidation of methane and chlorinated hydrocarbons E. Kantzer, D. D6bber, D. KieBling and G. Wendt Institut Rir Technische Chemic, Universit~it Leipzig, Linn6straBe 3, 04103 Leipzig, Germany Zirconia and titania-supported manganese oxide catalysts for the combustion of methane and chloromethane were prepared by precipitation and impregnation and characterized by various techniques. The characterization studies showed that, in contrast to titania-supported catalysts the manganese oxide phases on zirconia-supported manganese oxide catalyst is highly dispersed. The different states of the manganese oxide phase are reflected in the reducibility. Addition of ceria to zirconia and tungsten oxide to titania enhances the reducibility of the manganese oxide species. Catalytic activity in the methane oxidation is related to the dispersity of the catalytically active manganese oxide phase. In contrast to the methane oxidation the zirconia and titania supports are catalytically active themselves for chloromethane oxidation. Zirconia-supported manganese oxide catalysts showed a lower catalytic activity than ceria- zirconia catalysts with low ceria content. The best results for the titania-supported catalysts were obtained on catalysts with low manganese oxide and tungsten oxide loadings. 1. INTRODUCTION Catalytic total oxidation of volatile organic compounds (VOC) is widely used to reduce emissions of air pollutants. Besides supported noble metals supported transition metal oxides (V, W, Cr, Mn, Cu, Fe) and oxidic compounds (perovskites) have been reported as suitable catalysts [1,2]. However, chlorinated hydrocarbons (CHC) in industrial exhaust gases lead to poisoning and deactivation of the catalysts [3]. Otherwise, catalysts for the catalytic combustion of VOCs and methane in natural gas burning turbines to avoid NOx emissions should be stable at higher reaction temperatures and resists to thermal shocks [3]. Therefore, the development of chemically and thermally stable, low cost materials is of potential interest for the application as total oxidation catalysts. Manganese oxides have long been known to be catalysts for a variety of gas clean-up reactions. Manganese/copper mixed oxide (Hopcalite) is the catalytically active component in gas mask filters for CO; CO is converted to CO2 at room temperature [4]. Further applications of manganese oxide catalysts are the NH3 oxidation to N2 [5], the combustion of VOC [6,7] and methane [8], the oxidation of methanol [7], the 03 decomposition [9] and the NOx reduction[14]. Perovskite-type oxide catalysts (e.g. LaMnO3) have been proven to be effective catalysts for the total oxidation of chlorinated hydrocarbons [10]. Several studies have shown that besides preparation method and calcination temperature the kind
490 of support materials determines the structural and catalytic properties of the resulting manganese oxide catalysts [9,11-13]. The aim of the present work was to examine supported MnOx catalysts and to clarify the influence of the support (TiO2, WO3-TiO2,ZrO2, CeO2-ZrO2) on the textural, structural and catalytic properties for the total oxidation of methane and chloromethane, considering catalyst deactivation and formation of by-products. 2. EXPERIMENTAL
ZrO2 and CeO2-ZrO2 support materials were prepared by precipitation of the hydroxides from the nitrates with NHa-solution followed by drying and calcination at 600 ~ TiOE-anatase and TiOE-rutile were supplied by the Sachtleben Chemic GmbH (Germany). WOa-TiO2 samples were prepared by impregnation of TiO2 with aqueous (NH4)10W12041-solution and, after drying, calcined in air at 400 ~ for 3 h. The MnOxAVOa-TiO2 catalysts were prepared by impregnation of the calcined supports with an aqueous solution of Mn(NO3)2. MnOx/CeOE-ZrO2 catalysts were obtained by coprecipitation of the hydroxides from the metal nitrates with NHa-solution in presence of H~O2. The dried precursors were calcined in air at 600 and 800 ~ resp. for 6 h. InFfigures and Tables, the following abbreviations were used: xMn/yCe(W)Zr(Ti) for x mol% MnEO3/y mol% CeO2 (WO3) - (100 - y) mol% ZrO2 (TiO2 anatase). The catalysts were characterized by thermoanalytical (DTG/MS), powder X-ray diffraction (XRD), nitrogen adsorption (BET) and temperature programmed reduction (TPR) measurements. A continuous fixed bed reactor coupled on-line with a GC (FID/TCD) was used for the catalytic experiments. The amount of catalyst (particle size diameter: 0.1 - 0.3 mm) loaded was 0.6 g. The catalytic behaviour of the catalysts was investigated in the total oxidation of methane and chloromethane (1 vol.% in air, feed stream 5 l/h) considering by-product formation. 3. RESULTS AND DISCUSSION 3.1. MnOdCeO2-ZrO2 catalysts XRD analysis of the calcined ZrO2 supports revealed a mixture of monoclinic and tetragonal ZrO2. With increasing MR203 content the cubic ZrO2 modification was favoured and stabilized at higher MnOx content (Table 1). ~-Mn203 was detected at Mn203 contents of 40 mol%. Depending on the Mn203 content, a maximum of the specific surface areas was observed between 20 and 40 mol% for the precipitated catalysts; the specific surface areas for the impregnated catalysts are lower [15]. Using TPR, it was shown that reduction of the MnOx/ZrO2 samples is a multistep process and obviously depends on the MnO• loading. With increasing temperature peaks for amorphous MnOx species, ~-Mn203, MR304 and MR3+ ions incorporated in the ZrO2 lattice were determined (Fig. 1). In comparison to pure -Mn203 supported MnOx species are reduced at lower temperatures. Incorporation of CeO2 in the ZrO2 framework leads to a higher mobility of surface and bulk oxygen and can thus influence the catalytic properties [ 16,17]. At low CeO2 contents, beside monoclinic and tetragonal ZrO2, the Zr0.84Ce0.1602phase was identified. In addition,
491 crystalline CeO2 was found at higher CeO2 contents. The specific surface areas of the CeOz-ZrO2 samples did not differentiate significantly; a decrease was observed at higher CeO2 contents (Table 1). The reduction of CeO2 and CeOa-ZrO2 samples with high CeO2 contents occurred in two steps. The low temperature peak is assigned to the reduction of amorphous CeO2 and that at higher temperature, to crystalline CeO2 (Fig. 1). Table 1 Specific surface areas (aBET) and phase analysis by XRD of selected catalysts calcined at different temperatures (Tcalc.) Catalyst Tcalc.: 600 ~ Zcalc.: 800 ~ SBET(mZ/g) phases SBET(m2/g) phases 10Mn/Zr 84 c(t)-ZrO2 13 m-ZrO2, t-ZrO2 20Mn/Zr 147 c-ZrO2 39 c(t)-ZrO2, m-ZrO2 40Mn/Zr 144 Mn304, c-ZrO2 25 Gt-Mn203,t-ZrO2 20Ce/Zr 78 t-ZrO2(CeZr), m-ZrO2 45 t-ZrOz(CeZr), m-ZrO2 80Ce/Zr 59 CeO2 32 CeO2 20 Mn/20CeZr 108 t-ZrO2 9 t-ZrO2, Mn304 20Mn/80CeZr 72 CeO2 13 CeO2 3Mn/Ti(A) 42 10Mn/Ti(A) 29 3Mn/3WTi(A) 95 10Mn/3WTi(A) 72 3Mn/10WTi(A) 87 10Mn/10WTi(A) 68 m, t, c: monoclinic, tetragonal, R: rutile
Gt-Mn203,A 4 Gt-Mn203,A 4 A 7 Gt-Mn203,A 6 Gt-Mn203,A 10 0t-Mn203,A 6 cubic; CeZr: Zr0.84 Ce0.1602; MnTi:
~t-Mn203,A, R a-Mn203, A, R WO3, A, R a-Mn203, R WO3, A, R Gt-Mn203,R, MnTi MnTiO3; A: anatase;
The loading of MnOx on the CeOz-ZrO2 supports increases the specific surface areas (Table 1). It is important to point out that there are CeO2-MnOx interactions on unsupported as well as on supported catalysts. Using cerimetric redox titration [16] it was found that in comparison to the unsupported MnOx an increase of the Mn oxidation number is observed for CeO2-MnOx samples calcined at low temperatures, whereas a decrease was observed at higher calcination temperatures [15]. The obtained results are in accordance with those of Imamura et al. [16]. TPR measurements on MnOx/CeO2-ZrO2 samples revealed that CeO2 incorporation in the ZrO2 structure leads to a small shift of the low temperature reduction peaks of MnOx phases to lower temperature; thus, the presence of Ce ions in the ZrO2 structure enhances the reducibility of the MnOx species. The catalytic activity was investigated for the total oxidation of methane and chloromethane as testing reactions. Selected results are presented in Fig. 2 and Fig. 3 Compared with the impregnated MnOx/ZrO2 catalysts the catalytic activity of the precipitated catalysts for the methane conversion is higher [15]. The best results were obtained for MnOx loadings between 20 and 40 mol% (Fig. 2). With respect to the structural investigations it is suggested that the amount of X-ray amorphous (dispersed) MnO• species on the ZrO2 surface is responsible for the catalytic activity. Moreover, with
492 increasing calcination temperature of the samples the catalytic activity decreases. This is explained mainly by the decrease of the specific surface areas and by the migration of Mn 3+ ions in the ZrO2 framework under formation of solid solutions which lead to a depletion of the catalytically active sites. The catalytic activity of the ZrO2 and CeO2-ZrO2 supports is low. For the CeO2-ZrO2 samples, an activity maximum was found on the catalyst with a CeO2 content of 80 mol% [15]. 1~0 3
~ho~ - - ~ o ~ - - ~ o
,", / ', ~
10MIV3WTi 10Mn/10WH 10~Zr
Tert~alure (~ Fig. 1. TPR profiles of selected catalysts (3 l/h 8 vol.% H2 in Ar; heating rate: 10 l/h) Fig. 2 shows that at a constant MR203 content of 20mo1%, MnOx/CeO2-ZrO2 catalysts exhibit nearly the same catalytic activity as the MnOx/ZrO2 catalysts. However, comparing the catalyst behaviour of both catalyst systems, the activity of MnOx/CeOa-ZrO2 catalysts are lower than that of the MnOx/ZrO2 catalysts. On the basis of textural investigations, this effect is explained by the lower specific areas of the CeO2 containing catalysts. Furthermore, the decrease of the Mn oxidation number determined by cerimetric titration [15] is due to MnOx-CeO2 interactions at calcination temperatures > 400 ~ which affect the catalytic behaviour of the three-component system MnOx/CeO2-ZrO2 for the total oxidation of methane.
493
'~176,o~, 80
lON,kl/3WIi lOMn/lOWTi / /
-
p~'->~
/?
2o~
r
/
/f
~Zr
=o
:z
=>
'0t o. c ~ - _ - ~ _ ~ ~ ~ - o - - , 300 400
, 500
ge~on~
|
6~
700
(~
Fig. 2. Conversion of methane on selected catalysts vs. reaction temperature
y.
100
- ~ ;
-
,
0-
~6o. lOMCfi 3Mu3WIi IOMa/IOWIi --9 20MC/r v 40MqZr --
=> 40.
20-
~ y ~Y
2 0 ~
-v
300
'
~0
'
~o
'
&
Fig. 3. Conversion of chloromethane on selected catalysts vs. reaction temperature
494 The catalytic activity for chloromethane conversion over mixed oxide catalysts is characterized by a reversible deactivation of the catalysts and depends on the kind of CHC and reaction conditions [ 10]. After an initial period of up to 60 min, a nearly constant CHC conversion is observed. The conversion of chloromethane as a function of reaction temperature for selected catalysts is shown in Fig. 3. The results were obtained at steady state regime. With increasing MnOx content up to a loading of about 40 mol%, the catalytic activity increases. As in the case of methane conversion, the catalytic activity in chloromethane conversion is attributed to the dispersed MnOx species at the ZrO2 surface. Moreover, in comrast to methane oxidation, the ZrO2-support itself is catalytically active [ 15]. The investigations of CeO2-ZrO2 samples over a wide composition range showed that the best results were obtained with the sample 20 mol% CeO2-80 mol% ZrO2, which is active at low temperatures. It is suggested that the activity of these catalysts is determined by the incorporation of Ce 3§ ions in the ZrO2 matrix, which leads to a higher mobility of bulk and surface oxygen species. Loading of the CeO2-ZrO2 samples with MnOx causes a decrease of the catalytic activity (Fig. 3) which is explained by the CeO2-MnOx interactions (see above). Besides the main reaction products, HC1, CO2 and H20, several by-products were determined in the exit gases (Table 2). Only higher chlorinated chloromethanes were formed up to a reaction temperature of 500 ~ Remarkably low amounts of by-products were found over the catalyst 20 mol% CeO2-80 mol% ZrO2. With the MnOx comaining catalysts, more by-products were obtained. Furthermore, chlorine was formed by Deacon reaction. Table 2 Concentrations of organic by-products in exit gas (vpm) in the oxidation of chloromethane on selected catalysts Reaction temperature (~ Catalyst By-products 300 350 400 450 500 550 20Mn/Zr CH2C12 30 160 450 270 CHC13 30 CCh . . . . 10 40Mn/Zr CH2C12 50 340 730 110 CHC13 10 20 40 CCh
20Ce/Zr 20Mn/20CeZr
10Mn/Ti 3Mn/3WTi 10Mn/10WTi
CH2C12 CHC13 CH2C12 CHC13 CCh CH2C12 CHC13 CH2C12 CH2C12 CHCI3 fEb
-
60 . . .
-
-
10 220 .
. 50) is also in agreement with the data reported by Klimova, Linacero, and Toba [4, 6, 11]. 3.2. X-Ray Powder Diffraction
The X-ray powder diffraction patterns are shown in Fig. 2. The X-ray diffraction pattern of alumina precipitated from gel shows two broad maxima at 20=46 and 67 ~ characteristic of ~'-A1203. Introduction of small amounts of titania markedly reduces the intensity of these peaks. Accordingly, the sample with 15% appears to be virtually amorphous to X-rays. This result is in agreement with the data of Ramirez [15]. Amorphisation of A1203by TiO2 also explains the increase in SBET(see above).
I
9- anatase o- rutile
o
~1 ~ ~..,__ ]l il
0,
,
Mech. mixture: 50 mol % TiO 2
o
"
A
~-"~" ~___
._^
+ AI203
,~
85 mol % T i O 2
~~'f
~'~~..,,,.,.~.,....,_~~
50 mol O/oTiO2
~,~
' " ~ ~ ~ _ , ~
15 mol % TiO 2
I
20
30
40
A 50
60
70
80
20, ~
Fig. 2. X-ray diffraction pattems of the samples with different TiO2 contents. The data for mechanical mixture of bulk TiO2 and sol-gel A1203 are given for comparison
512 Appearance of crystalline TiO2 phase was observed only when TiO2 content reaches 50 mol %. Diffraction pattern of this sample exhibits broad maxima indicating the presence of minor amount anatase phase. The crystallite size of titania in the 50%TiO2-A1203 sample estimated from the Scherrer equation shows values around 100,~. No rutile phase was detected in the sample. This picture is essentially different from the pattern yielded by an equimolar mixture of TiO2-AI203 prepared by mechanically mixing parent titania and alumina. Diffraction pattern of an equimolar mixture TiOa-A1203 precipitated from gel The mechanical mixture reveals sharp diffraction peaks due to anatase and rutile. The diffractogram of the sample with 85% of TiO2 shows sharper peaks then the 50% TiO2 sample. An increase of the titania content from 50 to 85% considerably increases the crystallite size of titania. However, anatase is nearly the only phase in this sample. The estimation of rutile fraction from the intensities of the [101] and [110] reflection planes for anatase (Ig) and rutile (IR) respectively by applying the equation XR=1/[1+1.26 (IA/ IR )] gives a proportion of 55 and 5% respectively, since the fraction of rutile does not exceed 3-5% in the high-titania samples. The results obtained by XRD are in accordance with the results reported by Linacero [6] and Ramirez and Gutierrez-Alejandre [15] for A1203-TiO2 mixed oxides with different TiO2 content. These authors also reported a formation of TiO2 crystalline phases when TiO2 content exceeds ~ 70 mol %. The samples with lower TiO2 contents were found to be x-ray amorphous. 3.3. Study the surface composition by XPS Surface concentrations o f Ti a n d Al. In order to evaluate deviations in the surface concentrations of Al and Ti from the overall content, the coefficients of surface segregation (/5) were calculated by the method similar to that proposed by Seach [16]:
[JTi =
XPS
bulk
where ~Ti and ~Al are the coefficients of the surface segregation of Ti and AI, respectively (Ti/A1)xps and (Al/Ti)xps
are the element atomic ratios on the surface calculated from XPS data;
(Ti/Al)xps and (Al/Ti),,ps
are the overall (bulk) element atomic ratios determined by chemical analysis
513 Variations of 15~iand 15A1with the TiO2 content are displayed in Fig. 3. Evidently, titania does not show a pronounced tendency to segregation. In the samples with a low Ti content, the surface composition remains essentially the same as the bulk composition. Therefore, the coefficients of surface segregation [~Ti and I~A~are close to 0. We can conclude that at low TiO2 contents, Ti ions are homogeneously distributed in TiO2-AI203 like in a solid 4 solution. _
3 ................. !~ ......' ~........."..................."1"
"t2 a7 ~
'~2
.....................................
0"1 I . . . . '1~'I .... ~/ ............... . . . . . . . . ~J.............. [ ". .D. ." ~ ~, ............... . ; ; . . . ~ ":i~. .............. ,. t -
~
0
l
!
i
l
!
20
40
60
80
100
-1
tool % TiO2 Fig. 3. Dependence of the coefficients of surface segregation of Ti and A1 on the TiO2 content.
Unlike Ti, A1 demonstrates very pronounced tendency to segregate on the surface of mixed oxide. This tendency becomes particularly distinct with decreasing A1 content. Thus, for the sample containing 15 and 3 mol % A1203 (the TiO2 content is 85 and 97 mol %, respectively), the A1/Ti surface atomic ratio exceeds the bulk value by a factor of 5-6 ([~Al "" 4-5).
We can conclude that TiO2 tends to interact with A1203 at low TiO2 contents, which results in homogeneous distribution of Ti. A 1 2 0 3 shows a tendency to segregate on the surface of T i O 2 species. This tendency is particularly pronounced at low A1203 contents.
Chemical states of O, Ti, and A1. XPS spectra in the O ls, Ti 2p, and A1 2p regions for the samples with different molar contents of TiO2 are shown in Fig. 4. Analysis of variations and XPS line positions and shapes allow us to reveal several evident relationships. 0 ls region. For the samples containing 2.5-15 mol % TiO2, the binding energy of the 0 Is peak is - 531.5 eV, the peak is symmetrical, which is typical for plain A 1 2 0 3 . The position and lineshape of the O ls peak remains essentially unaltered even for the sample containing 50 mol % TiO2. However, the furher increase in the TiO2 content up to 85 mol % leads to broadening of the O ls peak and the peak shifts toward higher binding energies by -~ 1.5 eV. Curve fitting analysis reveals two peaks: one at-~ 531.7 eV (characteristic of A1203) and another one at -~ 530.3 eV with pronounced asymmetry, which is typical of TiO2. Note that the appearance of the second peak is in agreement with XRD data showing the formation of the TiO2 phase in this sample. Ti 2p region. Analysis of the Ti 2p peak allows us to reveal two components at-~ 458.5 and 460.5 eV in the samples containing 2.5-5 mol % TiO2. The component at 458.5 eV can be assigned to the Ti ions in the octahedral coordination typical for TiO2. The component at 460.5 eV implies the presence of lower coordinated Ti ions, presumably in
514 the tetrahedral coordination [17]. With increasing Ti content, the intensity of the signal of octahedral Ti increases. This peak predominates in the spectra and the signal of tetrahedral Ti becomes invisible. Presumably, the signal of tetrahedral Ti points to certain incorporation of Ti ions into the A1203 lattice and occupation of some tetrahedral positions. Probably, this process is not pronounced due to the difference in the A1 and Ti radii, and at higher TiO2 contents, the octahedral coordination predominates, which points to the preferable formation ofTiO~ species.
AI 2p region. The position and shape of the A12p line remain the same in the whole range of TiO2 contents. We can tentatively conclude that even for the samples with a low A1 content, A1 tends to form separate A1203 species without marked interaction with the TiO~ phase. Pronounced segregation of A1 on the surface (see above, Fig. 3) indicates that A1203 species are located on the surface of TiO2 particles. 4. CONCLUSIONS The data obtained allow us to propose the following mechanistic scheme describing oxide-oxide interaction in the TiO2-A1203 system. At low TiO2 contents (0-15 mol % TiO2), there is a pronounced interaction between two oxides resulting in the formation of AI-Ti mixed oxide. In this oxide, a part of Ti ions appears to occupy tetrahedral positions of the A1203 lattice. Incorporation of Ti ions lowers the crystallinity of A1203, which, in turn, results in an increase in the specific surface area of the mixed oxide. As a result of Ti-A1 interaction, formation of the TiO2 phase is negligible and Ti is uniformly distributed in the oxide particles. At higher TiO2 contents up to 50 mol %, the TiO2 phase begins to form. However, A1203 appears to segregate on the surface of TiO2 and encapsulate TiO2 clusters thus preventing their agglomeration. Therefore, the size of TiO2 clusters remains relatively small (- 100 A) and SBET does not decrease significantly. Upon the following increase in the TiO2 content up to 85 mol %, formation of a bulk TiO2 phase takes place, which is accompanied by an abrupt decline in SBET.
99%) and 0.1 g of catalyst. The products were identified and quantified with an HP 6899 series chromatograph equipped with an HP-1 capillary column. 3. RESULTS AND DISCUSSION The deposition-precipitation of Ni(II) shows the typical pH versus time behaviour found in high surface area Ni/SiO2 systems [8-11]. The pH-curve displays a maximum; after this value the nucleation and growth of the solid phase start up rapidly and the rate of generation of hydroxyl ion is lower than its consumption leading to a temporary decrease in the pH value (Fig. 1). 5.5 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0
L
......................
,
0
,
,
,
,
,
,
,
,
,
,
,
40 80 120 160 200 240 280 320 360 400 440 480 Time, (min)
Fig. 1. pH-curve of the Ni/HI3 samples by hydrolysis of urea at 90 ~ as a function of the DP time. Table 1 displays the textural properties and the Ni loading for the catalysts prepared by DP and competitive cationic exchange. With one h DP time a Ni concentration close to that obtained by cationic exchange is obtained. Also, the microporous area of the Ni/H[51 and Ni/HJ3-CE catalysts is similar. The nitrogen adsorption isotherms of the HI3 zeolite and the Ni/H[3-CE samples (Fig. 2) are, as expected, very similar. In the sample prepared by DP, the presence of a hysteresis caused by the formation of a secondary porous system is evidenced in the Ni/H[~2 and Ni/HI34. The shape of the newly formed hysteresis suggests the formation of a laminar type porous structure [21, 22], possibly nickel phyllosilicates.
540 Table 1. Ni loading and textural properties of the HI5 zeolite, Ni/H~5 and Ni/H[5-CE samples. HI5 Ni/HI3 Ni/HI31 Ni/HI32 Ni/HI34 zeolite CE Wt % Ni 3.8 5.4 14 18 Surface area, mZ.g-1 579.4 540.4 584.9 520.2 452 Microporous area, ma.g-1 359.2 280.2 288.1 200 120 Microporous volume, cmg.g-1 0.166 0.125 0.133 0.094 0.053
~
= 100
d r
Z
b
>
o.o'o11 'oi2'ols'o~'ols'd6'dT'ois'019'1.o Relative Pressure, P/Po
Fig. 2. Nitrogen adsorption-desorption isotherm: a) HI5 zeolite; b) Ni/H[3-CE" c) Ni/HI32; d) Ni/HI34. X-Ray diffraction. The XRD patterns for the HI5 zeolite, the Ni/H[5-CE sample and some Ni/H[3 samples prepared at different DP times are shown in Fig. 3. The incorporation of Ni to the zeolite by either of the two methods, cationic exchange or DP, produces a small loss in the crystallinity of the zeolite. However, only the samples prepared by DP show clearly two new asymmetric reflections at 20 = 33.2 and 59.2 ~ with d spacing of 2.66 and 1.544 ,~ respectively. These two new reflections according to published results [8, 9, 12, 13] may be attributed to the formation of Ni hydrosilicates. The intensity of these two reflections increases with DP time indicating an increase in the crystallinity of the responsible phase. TPR of dried samples. The TPR profiles of the HI5 zeolite and the dried Ni/HI3-CE, Ni/H[32 and Ni/H[54 samples, shown in Fig. 4, display an asymmetric reduction peak at 400 ~ whose intensity decreases in the Ni/H[M sample. According to previous work on bulk and SiO2 supported Ni prepared by DP [14, 15], the peak at 400 ~ can be assigned to the reduction of nickel hydroxide or highly disordered Ni phyllosilicate. The high temperature peaks can be assigned to the reduction of 1:1 nickel phyllosilicates with different degrees of crystallisation [8, 9, 16, 17].
541
~= 50
c
o
u+.+
t 0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
2 Theta Degree
Fig. 3. XRD patterns of the a) HI3 zeolite; b)Ni/HI3-CE" c) Ni/H[32; d) Ni/HI34.
~= 100
j
d O cl.
__J
E O
o
:ff a 0
1o 0
2o 0
300 '
4~o 500 . . 600 . . 700. Temperature, ~
800
900
1000
Fig. 4. TPR profiles of dried samples a) Ni/HI3-CE; b) Ni/HI32" c) Ni/HI34. FTIR characterization. The mid-infrared region of 2000-400 cm -1 contains the fundamental framework vibrations of the Si(A1)O4 groupings [18-20]. In the same way, nickel phyllosilicates exhibit in the 1200-400 cm -1 range bands characteristic of stretching and bending of SiO vibrations, and bending vibrations specific of structural OH groups [7]. In the IR spectra of our samples several features are evident. The band that in the HI5 zeolite appears at 618 cm -1 grows and develops with DP time into one broad band with two contributions, one at 641 cm -1, assigned to the presence of Ni(OH)2, and another at 665 -1 cm , assigned to the presence of 1"1 nickel phyllosilicate. It appears that in the Ni/H[3 samples prepared by DP there is a mixture of 1"1 nickel phyllosilicate and nickel hydroxide. For the sample prepared by cationic competitive exchange there is no clear evidence of these bands. The band that in the pure zeolite appears at about 797 cm 1 and that corresponds to symmetrical stretch of SiO4 tetrahedra, diminishes with DP time indicating a probably dissolution process of the siliceous framework during the DP process. This also occurs with the band at about 1086 cm -1. This behaviour supports the formation of Ni hydrosilicates at the expense of the siliceous framework in the DP prepared samples. The
542 sample prepared by cationic competitive exchange shows similar features indicating also the partial dissolution of the siliceous framework of the zeolite, but the bands at 640 and 665 cm -1 are not clearly evident in this case. Fig. 5 shows the IR spectra in the fundamental region of the HI3 zeolite, Ni/HI3-CE, Ni/HI32 and Ni/H[34.
3000
2000 Wavenumbers(cm4)
1000
Fig. 5. IR spectra a) HI3 zeolite, b)Ni/HI3-CE; c) Ni/HI32; d) Ni/HI54.
I=0.5
1090 A
___.__~Ni/HI~4Dp=4h ~ _ _ . _ . . ~ 2193
~
Subtraction
8
/
5
463
\ 792/'---
1049
~,~oo5
_~ /"-
~6o
~
~
I
465
HI5zeolite1629~___~ 2500
2250 20()0 1750 1500 1250 1000 750 Wavenumbers (cm -1)
500
Fig. 6. IR spectra of the HI3 zeolite treated with urea at 90% during 4h, Ni/H[34 and subtraction spectrum In the DP samples, after two hours deposition-precipitation, there is formation of a clear shoulder with DP time at about 1005 cm -1, which according to the literature [7], points to the presence of 1:1 nickel phyllosilicate. This is more clearly seen in the subtraction spectra. It appears that in the case of the cationic competitive exchangeprepared sample the formation of hydrosilicate species is only incipient. In fact, these
543 poorly crystallized species were only detected by TPR. Fig. 6 displays the subtraction of the Ni/HI34 and Hf~ zeolite treated with urea during 4h. Electron microscopy. TEM micrographs of the reduced Ni/H[32 and Ni/H[3-CE samples are shown in Fig. 7. The micrograph of the DP reduced sample (Fig. 7a, b) shows a homogeneous distribution of Ni metal particles. In contrast, the sample prepared by cationic competitive exchange shows a highly inhomogeneous distribution of Ni particles, which in general are larger than the ones found in the DP samples.
Fig. 7. TEM micrographs of a) Ni/H[31 and b) Ni/H[3-CE (bar= 50 nm) Catalytic Activity. The results of catalytic activity in the hydrogenation of naphthalene are well in line with the above findings. By comparing the concentration versus time curves for the Ni/H[51 and the Ni/H[5-CE samples, it is easily observed that the DP method leads to higher catalytic activities. In fact, with the DP catalyst, the total conversion of naphthalene is reached at 120 rain while the catalyst prepared by cationic competitive exchange takes 300 rain to convert all the naphthalene. The amounts of products resulting from the hydrogenation of the second aromatic ring, cis and trans decalins, also make clearly evident the supremacy of the DP prepared catalyst. Fig. 8 (a-b) shows the catalytic activity of Ni/H[51 and Ni/HI3-CE in the hydrogenation of naphthalene.
0.050 T ,J --~ o 0.040
.e
- 0.030 ._
I~
(1)
(3)
ai i
0.050
- 0.030 ._
0.020 8
b
;
0.020
8
c
)
.J o 0.040
8 0.010
0.010
0.000 ~
0.000
0
60
120
180
240
0
Time, (min)
50
1O0
150
200
250
Times, (min)
(1) Naphthalene (2) Tetraline (3) Trans-Decaline (4) Cis Decaline
Fig. 8. Hydrogenation reaction of naphthalene at 220 ~
a) Ni/H[31 and b) Ni/H[3-CE
300
544 4. CONCLUSIONS From the above results one can conclude that the DP preparation method leads to Ni/H[3 catalysts with better Ni dispersion than those prepared by cationic competitive exchange. This result seems to be due to the formation of a stronger support metal interaction (formation of Ni hydrosilicates) in the case of the DP method. The better deposition of Ni achieved by the DP method is clearly reflected in a superior activity in the naphthalene hydrogenation reaction. ACKNOWLEDGMENTS We acknowledge the financial support from the IMP-FIES Program. We are grateful to Mr Ivan Puente for the microscopy work. REFERENCES
1. J.W. Geus, Dutch Patent Applications, 1967, 6705,259, and 1968, 6813,236. 2. J.A. van Dillen, J.W. Geus, L.A. Hermans and J. van der Meijden, In Proceeding of 6th International Congress on Catalysis, London, 1976; G.C. Bond, P.B. Wells, F.C. Tompkims, Eds., Elsevier, Amsterdan, 1977; p 667. 3. L.A.M. Hermans and J.W. Geus. in Preparation of Catalysts II, B. Delmon, P. Grange, P.A. Jacobs, G. Poncelet, Eds., Elsevier; Amsterdam, (1979) 113. 4. J.W. Geus, Preparation of Catalysts III, G. Poncelet, P. Grange, P.A. Jacobs, Eds. Elsevier, Amsterdam (1983) 1. 5. O. Clausen, M. Kermarec, L. Benneviot, F. Villain and M. Che, J. Am. Chem. Soc., 114 (1992) 4709. 6. M. Kermarec, J.Y. Carriat. P. Burattin, M.Che and A. Decarreau, J. Phys. Chem. B 98, (1994) 12008. 7. J.Y. Carriat, M. Che, M. Kermarec, M. Verdaguer and A. Michalowicz, J. Am. Chem. Soc., 120 (1998) 2059. 8. P. Burattin, M. Che and C. Louis, J. Phys. Chem. B 102 (1998) 2722. 9. P. Burattin, M. Che and C. Louis, J. Phys. Chem. B 103 (1999) 6171. 10. J.W.E. Coenen, Appl. Catal. 54 (1989) 65. 11. G. Ertl, H. Kn6zinger and J. Weitkamp, Preparation of Solid Catalysts; Wiley-VCH, Weinheim, (1999) 460. 12. H. Mod6sir and A. Decarreau, Bull. Min6ral. 110 (1987) 409. 13. O. Clause, L. Benneviot, M. Che and H. Dexpert, J. Catal. 130 (1991) 21. 14. B. Mile, D. Stirling, M.A. Zammitt, A. Lovell and M. Webb, J. Catal. 114 (1988) 217. 15. J.P. Espin6s, A.R. Gonzfilez-Elipe, A. Caballero, J. Garcia and G. Munuera, J. Catal. 136, (1992) 415. 16. P. G~nin, A. Delahaye-Vidal, F. Proteger and F. M. Tekaia Figlarz, Eur. J. Solid State Inorg. Chem. 28 (1991) 506. 17. A. Decarreau, H. Mod6sir and G.C.R Besson, Acad. Sci. Paris (S6r. II) 308 (1989) 301. 18. R. Szostak, Molecular Sieves Principles of Synthesis and Identification; Van Nostrand Reihold: New York, (1989) p. 282.
545 19. Zeolite Chemistry and Catalysis; J.A. Rabo, Ed.; ACS Monograph 171 American Chemical Society: Washington, D. C. (1976) p. 80. 20. D.W. Breck, Zeolite Molecular Sieve: Structure, Chemistry, and Use, John Wiley and Son: New York, (1974) p. 415. 21. S.J. Gregg and K.S. Sing, W. Adsorption, Surface Area and Porosity; Academic Press, London and New York, (1967) p. 173. 22. G. Leofanti, M. Padovan, G. Tozzola and V. Venturelli, Catal. Today, 41 (1998) 207.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
547
Sol-Gel A1203 structure modification by Ti and Zr addition. A NMR study J. Escobar a*, J. A. De Los Reyes b and T. Viverosb aInstituto Mexicano del Petr61eo, Tratamiento de Crudo Maya, Eje Central Lfizaro Cfirdenas 152, San Bartolo Atepehuacan, G. A. Madero, M6xico, D. F., M6xico 07730 bArea de Ing. Q., UAM-Iztapalapa, San Rafael Atlixco 186, Col. Vicentina, Iztapalapa, M6xico, D. F., M6xico 09360 In order to assess the effect of various synthesis parameters and the presence of a second oxide (TiO2 or ZrOz) at different concentrations on the AIaO3 structure, the corresponding samples were prepared by low-temperature sol-gel method. The oxides were characterized by N2 physisorption, XRD and 27A1 MAS-NMR. In all samples calcined at Tc_ CID > CAD = AD >CAC. Bimetallic catalysts showed selectivities much higher than the theoretically expected for a NSM. AD series showed a good selectivity towards isoparaffinic products, and ID serie presented the lowest value from the theoretical. In CAC and CID catalysts, the fraction of 2-MP remained constant, but that of other C6 was higher in CAC. with the creation of dual sites Pt-Ru. However, it can be concluded that the impact of platinum on ruthenium dispersion is small. Reduction temperature seems to have positive influence on metal dispersion for both mono- and bimetallic catalysts. There was no reduction temperature that improved the activity in all catalysts. Cracking or deep hydrogenolysis was directly related to higher ruthenium content and to higher mean cluster size. Catalysts reduced at 773 K are more suitable to produce isoparaffins. ACKNOWLEDGMENT The authors acknowledge the Universidad del Pals Vasco/EHU for the financial support (UPV 069.310-G40/98) to this research and for a FPI Grant to one of the authors (A.M.S.S.). REFERENCES
1. H. Miura, T. Suzuki, Y. Ushikubo, K. Sugiyama, T. Matsuda and R.D. Gonzfilez, J. Catal., 124 (1984) 194. 2. J.R. Gonzfilez-Velasco, M.A. Guti6rrez-Ortiz, J.A. Gonz~ilez-Marcos, P. Pranda and P. Steltenpohl, J. Catal., 187 (1999) 24. 3. L. Maya, J. Inorg. Nucl. Chem., 41 (1979) 67. 4. G.H. Van Den Berg and H.Th. Rilnten, Preparation of Catalysts II. Scientific Bases for the Preparation of Heterogeneous Catalysts, B. Delmon. P. Grange, P. Jacobs and G. Poncelet (eds.), p. 265, Elsevier, Amsterdam, 1979. 5. J.L. Falconer and J.A. Schwab, Catal. Rev.-Sci. Eng., 25 (1983) 141. 6. N.W. Hurst, S.J. Gentry, A. Jones and B.D. McNicol, Catal. Rev., 24 (1982) 233. 7. A. Baiker, AIChE, 25 (1985) 30. 8. A.S. Sass, N.A. Antonova and N.M. Popova, Kin. Catal., 37 (1996) 105. 9. M.U. Kilsyuk and V.V. Rozanov, Kin. Catal., 36 (1995) 13.
563 10. F.G. Gault, Adv. Catal., 30 (1981) 1. 11. G. Diaz, F. Garin and G. Maire, J. Catal., 82 (1983) 13. 12. G. Diaz, F. Garin, G. Maire, S. Alerassol and R.D. Gonzfilez, Appl. Catal. A: General, 124 (1995) 33. 13. Y. Zhuang and A. Frennet, Appl. Catal. A: General, 134 (1996) 37. 14. Y. Zhuang and A. Frennet, Appl. Catal. A: General, 177 (1999) 205. 15. B. Coq, A. Bittar and F. Figueras, Appl. Cata|. A: General, 59 (1990) 103. 16. M.J. Dees, M.H.B. Bol and V. Ponec, Appl. Catal. A: Gen., 64 (1990) 279. 17. J.M. Brunelle, Preparation of Catalysts II. Scientific Bases for the Preparation of Heterogeneous Catalysts, B. Delmon. P. Grange, P. Jacobs and G. Poncelet (eds.), Elsevier, Amsterdam, 1979. 18. J.R. Gonzfilez-Velasco, J.A. Gonzfilez-Marcos, M.A. Guti~rrez-Ortiz, J.I. Guti6rrezOrtiz and S. Amaiz, Spanish patent No. 9.500.364 (1995). 19. S. Amaiz, Ph.D. Thesis. University of Basque Country/EHU (1997). 20. D.L. Hoang, H. Berndt and H. Lieske, Catal. Lett., 31 (1995) 165. 21. P. Betancourt, A. Rives, R. Hubaut, C.E. Scott and J. Goldwasser, Appl. Catal. A: Gen., 170 (1998) 307. 22. J.G. Goodwin, J. Catal., 68 (1981) 227. 23. G. Del/imgel, V. Bertin, P. Bosch, R. G6mez and R.D. Gonzfilez, New J. Chem., 15 (1991) 643. 24. P. Villamil, J. Reyes, N. Rosas and R. G6mez, J. Mol. Catal., 54 (1989) 205. 25. Bond, G.C. and Hooper, A.D., Appl. Catal. A: General, 191 (2000) 69. 26. Bond, G.C. and Pafil, Z., Appl. Catal. A: General, 86 (1992) 1.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
565
Catalysts based on RhMo6 heteropolymetallates. Bulk and supported preparation and characterization. C. I. Cabello .1, I. L. Botto 2, M. Mufioz 1 and H. J. Thomas ~. ~Centro de Investigaci6n y Desarrollo en Procesos Cataliticos, CINDECA-CONICETUniversidad Nacional de la Plata, (1900) La Plata, Argentina. 2Centro de quimica Inorgfinica CEQUINOR- CONICET-Universidad Nacional de La Plata, (1900) La Plata, Argentina.
The ammonium salt of Rh(III) Anderson type heteropolymolybdate [RhMo6024H6] 3- has been prepared and characterized by powder X-ray diffraction, spectroscopic [FTIR-Raman, DRS (UV-visible)] and SEM-EDAX electron microscopy techniques. The water soluble salts were used in the design and preparation of 7-A1203 supported catalysts. The varied Mo:Rh ratio of both olution and solid samples was measured by AAS technique. The supported oxidic system was characterized by DRS spectroscopy and SEM-EDAX microscopy. The HDS and HYD activity for different bimetallic catalysts was measured in a high-pressure reactor. In addition, some conventional catalysts and some CoM06 and combined supported systems [(RhMo6 + AIM06)] have been tested for comparative purposes. The discussion about the performance of the new catalysts is made on the basis of the structural and physicochemical heteropolyanion properties as well as the preparation conditions. 1. I N T R O D U C T I O N Heteropolyoxometallates are polymeric species formed by the condensation of more than two different oxoanions. There is a great variety of these structures [1,2]. The compounds with this type of anions containing molybdenum are very interesting for chemistry and particularly for hydrotreating catalytic processes [3,4]. Although practical applications of some of these phases, especially those belonging to the Keggin structure, have been tried for a long time, it is only recently that the relationship between the crystal structure, the physicochemical properties and the catalytic functions has been carefully investigated [2]. Our recent research on a large series of hexametallates named Anderson [XM6024H6] 3-, [1] [with M = Mo; W or Mo(6_x)Wx and X = Co(III); Cr(III); Rh(III); Fe(III); Ni(II); Cu(II); Te(VI) etc.] [5,6,7] enabled the design and preparation of a variety of mixed bi- or tri-metallic phases which show interesting structural and redox properties, and can be used especially in heterogeneous catalysis. These planar heteropolyanions are precursors that provide welldispersed bi- or tri-metallic ensembles on supports and have been proved successfully on some catalytic processes such as hydrotreating, ammoxidation, etc. [3,8,9]. * Corresoondinu author. Member of the research staff of CICPRA. Fax" +54-221-42q4277 F.-mnil"
566 Particularly, Mo(VI)-Co(III) and Mo(VI)-Ni(II) containing species (here in CoM06 and NiM06 monolayer supported phases) have shown to be important precursors for HDS and HYD reactions [3,4]. In this sense, it seems important to prepare and characterize the RhM06 species as both bulk and 7-A1203 supported phase (by equilibrium impregnation with aqueous solution of RhM06 ammonium salts) in order to analyze the catalytic behaviour. In the last years, rhodium-based catalysts have shown a relatively high activity and selectivity for NO reduction to N2. The highest selectivity distinguishes the Rh catalyst from other Pt- and Pd-based ones in the ability to promote N-pairing in adsorbed NO molecules before the N-O bond is broken. This property is related to the highly dispersed supported Rh catalysts used in practice [10]. Hence, clean combustion (CO elimination and NOx reduction) can be another promising use of the catalysts based on RhM06, in addition to the hydrotreating processes. The present work shows the preparation and characterization of RhMo6/7-AI203 based catalysts and the preliminary results for hydrotreating (HDS of thiophene and HYD of cyclohexene) tests, discussed on the basis of the relationships between catalytic activity and spectroscopy and structural behaviour of RhM06 heteropolyanion. In addition, similar tests have been performed with [RhMo6+A1Mo6]/7-AI203 combined system; RhMo6/7-A1203 pretreated catalyst in reducing atmosphere; CoMo6/7-AI203 and Co-Mo or Rh/7-A1203 commercial catalyst for comparative purposes. 2. EXPERIMENTAL
2.1. Synthesis and characterization of the pure phases (NH4)3[RhMo6Oz4H6]. 7H20 was prepared by solution reaction from Rh(III) chloride hydrate and ammonium heptamolybdate (AHM), as previously described [11, 12]. The pure samples were characterized by X-ray powder diffraction analysis (DRX), using a Philips PW 1714 diffractometer (Cu Ka radiation, Ni filtered), Fourier transform infrared spectroscopy, using a Bruker IFSS 66 FT-IR equipment (KBr pellet technique); Raman spectroscopy using a Spex-Ramalog 1403 double monochromator spectrometer, equipped with a SCAMP data processor (excitation line: 514.5 nm of an Ar-ion laser), Scanning electron microscopy (SEM) using a Philips 505 with an EDAX 9100 for the electron probe microanalysis and Diffuse Reflectance spectroscopy (DRS) by a UV-VIS Varian Super Scan3 Spectrophotometer with a diffuse reflectance chamber with integrating sphere. The range covered was 200-800 rim, using BaSO4 or ~-AlzO3 as reference for pure and supported phases respectively. 2.2. Catalysts preparation The supported catalysts were prepared by equilibrium impregnation of ~,-AlzO3 with aqueous solutions of RhMo6 ammonium salts in the Mo range of 20-100 ~tmolMo/ml. The ~,AlzO3 powder used had a specific surface area of 226mZ/g, a pore volume of 0.36 cm3/g and a grain size of 200 ~m. All impregnations were performed at room temperature employing an excess of solution with continuous stirring for 24 h. The solid was then separated by centrifugation and dried at 350 K. Chemical analyses of the solutions before and after impregnations were made using an IL-457 atomic absorption spectrometer (AAS). This enabled calculation of the values of C~ (initial concentration of the impregnating solution) and Cf (final concentration of the impregnating solution) expressed as ~tmolMo/ml solution. From these values and the use of a simple material balance equation, it was possible to calculate Ca (the concentration of adsorbed metal, i. e. Rh, Co, and Mo) expressed as g metal/g support. In
567 addition, the [RhMo6+A1Mo6]/7-Al203 combined system was prepared by using a mixed solution of both rhodium and aluminium heteropolymolybdates 0.5:0.5 ratio) in attempts to analyse the synergic effect of rhodium, (since the Al(III) species is inactive for hydrotreatment processes). Some RhMo6 samples were pretreated by TPR up to 473 K before being introduced in the catalytic reactor.
2.3. Adsorption isotherm The adsorption isotherms for the RhMo6 Anderson phase were measured at room temperature using the Ci, Cf and Ca values determined for M=Mo. When the concentration of Mo adsorbed (Ca) expressed as monolayer (%) was plotted against the Mo concentration in solution at equilibrium (Cf), the shape of the resulting curve was found to follow the Langmuir model [13]. Hence, by plotting the linearised form of the Langmuir equation, i. e. Cf/Ca = [1/(Kad S)]+ Cf/S
(1)
and extrapolating the subsequent straight line obtained, it was possible to calculate the total number of active sites present on the surface (S) expressed in gMo/g 7-A1203. The equilibrium adsorption constant (Kad), expressed in ml/g Mo could be obtained from the slope of the line.
2.4. Catalytic activity The thiophene hydrodesulfurization and cyclohexene hydrogenation activity of the catalysts were measured in a high pressure reactor. The experimental conditions for the activity tests were a feedstock of thiophene (15000 ppm), cyclohexane (90%) and cyclohexene (10%), flow rate 0.353 ml/min, total pressure = 26 Kg/cm2 and LHSV=52 1/h. Additional experiences with toluene (90%) and cyclohexene (10%) were also carried out. The operative conditions for the hydrotreating test were selected according to the recent experience about the CoMo6 [3]. The products were analysed by gas chromatography by means of a Varian Start 3400 gas chromatograph, with FID detector. 3. RESULTS AND DISCUSSION
3.1. Morphological and structural features Structurally, the [XMo6024H6] group is made up of a compact package of six MoO6 octahedra surrounding one XO6 polyhedron in a planar hexagonal configuration, with approximate overall D3d symmetry [14]. In general, the ionic radius of the metallic central ion or heteroatom is 0.5 to 0.7A, so that the preparation of a great variety of phases is possible, with X(III): AIMo6; CoMo6; CrMo6; RhMo6; FeMo6, with X(II): CuMo6, NiMo6 and with X(VI): TeMo6. It is also possible to prepare solid solutions, by partly replacing the heteroatom by any other of the above mentioned ions by means of a stoichiometric mixture of the corresponding nitrates or sulfates with ammonium heptamolybdate, e.g.: Cr0.25A10.75Mo6 or Cr0.75A10.25Mo6 [6]. Moreover, the method enables the synthesis of phases having W, which can partially or totally substitute the Mo, producing phases of [XMonW(6-n)O24H6] type. From the crystal-chemical point of view, the phases containing X (III) are isomorphic, therefore the X-ray powder diagram of the Rh phase is similar to that corresponding to Co, Cr and Al. Also, scanning electron microscopy (SEM-EDAX) is a good method to characterize these phases, as the obtained microphotographs show the same crystalline characteristics (square plates) for all the series of phases [5]. On the other hand, the semi-quantitative chemical
568 analysis by microsound (EDAX), showed Rh and Mo contents of 1 5 . 9 1 % and 84.09 %, comparable to the theoretical values of 15.16 % and 84.83 %, respectively, in pure samples. 3.2.Characterization by FTIR and Raman spectroscopy. Fig. 1 shows the FTIR spectra for the RhMo6 phase in the region 4000-400 cm -1 55
55
50
(a)
50
45
45
40
40
o~ 35
/
o~' 35
~" 30 25 ~ v ~
~ ~
20 15 4000
(b)
~" 30 25
"
~ '
~ 3500
20 ' v
3000 [cm -1]
'
2500
'
2000
15 2000
'
'
1500 v
1000
'
' 500
[cm -1]
Fig. 1: FTIR spectra of (NH4)3[RhMo6Oz4H6].7H20 in the range (a): 4000 to 2000 cm -a and (b): 2000 to 400 cm -1. In general, the spectra for an ammonium salt can be divided into the following typical regions: 3600-2800 cm -1 (O-H and N-H stretchings), 1650-1400 cm -1 (OH and N-H bendings), 1000-800 cm -1 (Mo-Ot vibrations), 750-550 cm -1 (fundamentally Mo-Ob modes) and below 450 cm -1 (Mo-Oc and some other modes). This low spectral region is rather difficult to assign. Between 550-500 cm -1, it is possible to observe some bands attributable to water librations. Below 450 cm -1, the bendings of the Mo-Ot and Mo-Ob bonds can be mixed with the Mo-Oc stretching vibrations. In relation to these last modes, only the Mo-Ot vibrations can be considered as pure stretching. [15]. On the other hand, it is possible to observe from the crystallographic information [16], that the heteropolyanion size does not differ significantly from the X-ionic radius. The molybdic framework of this type of phases usually presents a great capability for a rearrangement when some atomic replacement occurs. This can lead to slight changes in the Mo-O-Mo angles, but not noticeable variations in the heteropolyanion-size. This effect was also observed in some Keggin phases [17,18]. In some previous vibrational studies for Anderson phases, the observed bands seem to be independent from the heteroatom types [12]. However, by careful measurements between 950-200 cm 4 of the Raman spectra for all members of this series, it is possible to notice that the bands associated to the Mo-Ot bonds are the most affected by the X change. The vibrational modes associated to the Mo-Ob and Mo-Oc bonds are less affected by the change of heteroatom, undoubtedly because they cannot be considered as pure stretchings and show some bend- character [15]. The Raman spectra, showing very sharp bands, make the shift clearer. Generally, Raman spectra of compounds with an octahedral coordination exhibit main lines in the high frequency range (800-1000 cm -l) [19]. Table 1 shows Raman frequencies for the RhMo6 and some other Anderson phases for comparative purposes.
569
Table 1 Raman spectra of ammonium salts of RhMo6, CrMo6, CoMo6 and A1Mo6, (registered between
1000- 180 cml). RhM06
CrM06
CoMo6
AIM 06
947 s 947 vs 945 vs 906 m 900 s 900 s 572 w 569 w 570 vw low 480 vw 481vw resolution 363 m 363 m Ref.: vs: very strong, s: strong, m: medium, w: weak, vw: very weak
944 vs 902 m 578 w 482 vw 364 m
Assign ation VsymMo-Ot VasymMo-Ot
vX-O vMo-Oc 6Mo-O, 6X-O
The most intense Raman line clearly shows that the Mo-Ot symmetric stretching is affected by X substitution. The general trend is a weakness of the Mo-O terminal bonds when the X(III) -size decreases (rAl(III) = 0.530A, r Co(re)lowspin -- 0.545/~, r Cr(III) = 0.615A, r Rh(In) = 0.665 A). This can be attributed to the X(III) polarizing power which produces an inductive effect, affecting the mentioned type of Mo-O bonds. On the other hand, the existence of the hydrogen bonds, proceeding of the NH4 + ions and the H20 out of the XMo6 ring, seems to contribute to this effect.
3.3. Characterization by diffuse reflectance spectroscopy Fig. 2 shows the diffuse reflectance spectrum of the RhMo6 phase, which is compared to CoMo6 and CrMo6 isomorphous phases. It is known that there are more than two types of chromophores in the polyoxometallates [20]. The charge- transfer transitions of the Mo-O bonds usually appear in the highest energy region of the DRS spectra (~250 nm for the Mo-O terminal bonds and between 250-350 nm for the Mo-O-Mo bridge bonds related to polymeric species) [21]. In the lower energy region, some other bands can be observed (usually due to the d-d transitions). The charge transfer transitions seem to be independent from the heteroatom. This behaviour is similar to that observed in other related compounds such as the Keggin and Dawson phases [21]. For the rhodium(III) compound, there are two bands at 511 and 409 which are assigned to the 1Tlg ~-lAlg and 1T2g CrMo6>CoMo6 >TeMo6-AIMoa>NiMo6 [7]. The adsorption parameters obtained for the system containing Rh get closer to the values corresponding to Cr and to Co. Although the number of active sites is smaller than those observed in other phases and AHM, the fact that Kaa values for all Anderson phases are higher by several size values than AHM is even more significant. The presence of trivalent heteroatoms of ionic radius between 0.50 y 0.66 A leads to a good deposition on ]t-A1203 support due to a possible X-AI interchange. Likewise, Raman results have shown that a lower heteroatom size is associated to higher Mo-Ot stretching frequencies. However, the Kaa calculated values present an inverse ratio which can be attributed to a different M o - O -
571 support interaction, while the active sites decrease when the heteroatoms increase due to possible geometrical considerations. All these facts surely play a very important role in the adsorption process of the planar species which contribute to the global impregnation process, characterized by a series of homogeneous and heterogeneous equilibria (anion deposition, A1 dissolution of the support, cationic interchanges, diffusion, etc). Such reactions have been recently characterized by means of spectroscopic and thermal methods [7, 23]. The characterization of Rh phases supported on alumina has been carried out by DRS. Although the spectra present less intense bands than those of pure phases, it is possible to observe the bands of load transference of octahedral Mo ( ~300 nm) as well as a band of wide transition and lower resolution attributed to Rh(III), around 400 nm. The surface of catalytic systems based on CoMo6/3,-AlzO3 and NiMo6/3,-AIzO3, sulfided and untreated, has also been recently studied by XPS and EXAFS [4, 24], and the corresponding RhMo6/7-A1203 study is in course. 3.5. Catalytic activity The thiophene hydrodesulfurization and cyclohexene hydrogenation activity of the catalysts have been measured in a high pressure reactor, the operative conditions having been selected according to the recent experience about the CoMo6 and NiMo6 based systems because of the common structural and physical-chemical properties [3, 4]. Table 3 shows chemical data and conversion obtained for selected catalysts based on RhMo6. In addition, the data for CoMo6 Anderson, CoMo commercial and Rh commercial 3,-A1203 supported catalysts are included for comparative purposes. Table 3 Chemical data and conversion of thiophene and cyclohexene for different RhMo6 catalytic and reference systems a, b 7A1203
supported Catalysts
X(%)
Mo(%)
HYD(%)
SHYD
CoMo6 0.80 8.00 30.00 3.75 CoMo a 1.80 9.30 15.55 0.86 RhMo6 1.00 6.00 84.68 8.47 RhMo6 b 1.12 6.80 66.80 5.96 [RhMo6 + AIMo6] 0.50 6.00 54.98 10.00 Rh a 5.00 64.96 1.30 Note: X(%) and Mo(%) are total metal concentration in gM/100g support. a Commercial catalysts. b Catalyst after TPR conditions treatment at 473 K. SHVDand S riDS = HYD/X and HDS/Mo
HDS(%)
SHDS
70.81 71.00 72.40 13.70
8.85 7.63 12.06 2.01
50.14 22.38
8.35 -
From the analysis of results shown in Table 3, it can be deduced that: 9 The comparison between the catalytic behaviour of Co and Rh Anderson phases reveals that, although the Mo content is lower for the latter, the HDS activities are similar. Hence, the synergic effect of Rh on the Mo is higher than that observed for the Co system. This is in agreement with literature for pure sulfides. This behaviour is more clearly observed from the comparison among the specific activities (Sm,Dand S HDS).
572
9 In HYD, the heteropolymetallate systems also show a better performance, especially the catalyst based on RhlMo6 in which the activity is enhanced more than five times. 9 The reducing thermal pre-treatment of RhMo6/7-AI203 produces a substantial decrease in HDS activity and not so noticeable in HYD. The presence of Rh induces the Mo(VI)-Mo(IV) reduction at lower temperature. Thus the Mo(IV) species, poorly sulfided, affect the catalytic performance, as it is well known. 9 The dilution of the impregnating solution RhM06 with A1M06 (which is poorly active for hydrotreatment processes) shows that the catalyst behaviour is function of Rh content. CONCLUSIONS Taking into account the specific activities a better performance of heteropolymetallate systems than commercial catalysts is clearly observed. The bifunctionality of the RhMo6 system is an important feature. This advantage seems related to the adsorptive interaction process of the planar anion that offers a good site distribution and involves a synergic effect. REFERENCES
1. M.T. Pope, "Heteropoly and Isopolyoxometalates", Springer-Verlag, Berlin, New York, (1983). 2. M.T. Pope and A. M/.iller, "Polyoxometalate Chemistry from topology via self-assembly to applications", Kluwer Academic Publishers, London (2001). 3. C.I. Cabello, I.L. Botto and H.J. Thomas, Appl. Catal. A: General, 197 (2000) 79. 4. I. Pettiti, I. L. Botto, C. I. Cabello, S. Colonna, M. Faticanti, G. Minelli, P. Porta and H.J. Thomas, Appl. Catal. A: General, 220 (2001) 113. 5. C.I. Cabello, I. L. Botto and H. J. Thomas, Thermochim. Acta, 232 (1994) 183. 6. I. L. Botto, C. I. Cabello, H. J. Thomas, D. Cordischi and P. Porta, Mater. Chem. Phys., 62 (Z000) 254. 7. C. I. Cabello, I. L. Botto, F. Cabrerizo, M. G. Gonzfilez and H. J. Thomas, Adsorption Sci. Tech., 18(7) (2000) 591. 8. I. L. Botto, C. I. Cabello and H. J. Thomas, Mater. Chem. Phys., 47 (1997) 37. 9. T. Liu, K. Asakura, U.Lee and Y. Iwasawa, J. Catal., 135 (1992) 367. 10. M. Shelef and G. W. Graham, Catal. Rev. Sci. Eng., 36(6) (1994) 433. 11. R.D. Hall, J. Am. Chem. Soc., 29 (1907) 692. 12. K. Nomiya, T. Takahashi, T. Shirai and M. Miwa, Polyhedron, 6 (1987) 213. 13. a) C.H. Giles, D. Smith and A.J. Huitson, J. Coll. Interf. Sci. 47 (1974) 755. b) C.H. Giles, H. P. D'Silva and I.A. Easton, J. Coll. Interf. Sci. 47 (1974) 766. 14. Jr. H. T. Evans, J. Am. Chem. Soc., 70 (1948) 1291. 15. C. Rocchiccioli-Deltcheff, M. Fournier, R. Frank and R. Thouvenot, Inorg. Chem., 22, (1983) 207. 16. A. Perloff, Inorg. Chem., 9 (1970) 2228. 17. H. Kondo, A. Kobayashi and Y. Sasaki, Acta Crystallogr., C45 (1980) 661. 18. R. Thouvenot, M. Fournier, R. Frank and C. Rocchiccioli-Deltcheff, Inorg. Chem., 23 (1984) 598. 19. T. L. Barr, Ch.G., F. Cariati, J.C.J. Bart and N. Giordano, J. Chem. Soc. Dalton Trans., (1983) 1825. 20. K. Nomiya, Y. Sugie, K. Amimoto and M. Miwa, Polyhedron, 6 (1987) 519. 21. H. So and M. T. Pope, Inorg. Chem., 11 (1972) 1441.
573 22. A.B.P. Lever, "Inorganic Electronic Spectroscopy"(Sec. Edition), Elsevier (1984) 23. X. Carrier, J.B. D'Espinose de la Caillerie, J.F. Lambert and M. Che, J. Am. Chem. Soc., 121 (1999) 3377. 24. P. Porta, G. Minelli, G. Moretti, I. Petitti, I. L. Botto, C. I. Cabello and H. J. Thomas, J. Mater. Chem., 4 (1994) 1641.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
575
Metallosilicate mesoporous catalysts prepared by incorporation of transition metals in the MCM-41 molecular sieves and their catalytic activity in selective oxidation of aromatics (styrene and benzene) V. PgtrvulescuI and B. L. Su* Laboratoire de Chimie des Mat6riaux Inorganiques, ISIS, The University of Namur (FUNDP), 61 rue de Bruxelles, B-5000 Namur, Belgium MCM-41 analogue metallosilicates containing the redox active metals V, Co, Cr and Ni or bimetals VCo and CrNi have been synthesized from aqueous gel by hydrothermal treatment. The mono- and bimetallic molecular sieves were characterized by various techniques, such as XRD, N2 adsorption-desorption, SEM, TEM and FTIR spectroscopy. The catalytic activity in the selective oxidation of styrene and benzene with H202 has been evaluated. The catalytic properties and the characteristics of the structure were correlated with the synthesis conditions, the nature and concentration of the metal, the association of two metals and their molar ratio in the bimetallic incorporated silicates and the oxidation state of the metal cations in the precursors. 1. INTRODUCTION The newly discovered hexagonal mesoporous molecular sieves such as MCM-41 offer new opportunities for transition metal incorporation into silica framework in order to generate potential catalytic activity [1-5]. Transition metals containing MCM-41 mesoporous redox molecular sieves have shown a high activity and selectivity in the oxidative transformation of organic molecules [6-12]. The properties of these high selective catalysts named as "mineral enzyme" can be tailored over a broad range, allowing the easy preparation. Transition metal ions can be introduced into molecular sieves in three different ways: ion exchange, impregnation [13, 14] and direct introduction by hydrothermal synthesis [9, 10, 15]. In the first two cases, extra-framework metal ions dispersed on the internal surface of channels were in general observed and only very small amount of metal ions were found to be incorporated in the framework. While in the last case, lattice metal ions in principal, but extra- lattice metal ions in small quantity may be found, depending on the metal content. In this work we present our results on the synthesis parameters and characterization of a series of MCM-41 molecular sieves containing active metals (V, Cr, Co and Ni) and bimetals (V-Co and Cr-Ni). The synthesis conditions and the effects of nature and content of the metals on the ordered hexagonal structure and morphology of the mono- and
1On leave from: Institute of Physical Chemistry I.G. Murgulescu, Spl. Independentei 202, Bucharest, Romania, SSTC and DGRE-DRI-RWresearch fellow * Corresponding author (
[email protected])
576 bimetallic mesoporous samples were discussed. The catalytic activity in selective oxidation of styrene and benzene is evaluated. 2. EXPERIMENTAL 2.1. Synthesis The synthesis of mesoporous molecular sieves by hydrothermal treatment was carried out using two synthesis procedures. First, V (1.6 and 3.2%), Cr (1.6%), Co (1.8-9.2%), Ni (1.8-3.6%), V-Co (Co/V molar ratio was fixed at 1 for VCol-MCM-41, 0.3 for VCo2MCM-41, and 3 for VCo3-MCM-41) and Cr-Ni (Cr/Ni molar ratio = 1.0 and 3.4 % in total metal contem) containing-MCM-41 catalysts were synthesized from sodium silicate, as the silica precursor, cethyltrimethylammonium bromide as the surfactant and VOSO4-5H20, Cr(NO3)3-9H20, CrO3, Co(NO3)2-6H20, Ni(CH3COO)2-4H20, as the metal precursors. The pH was adjusted to 11 with H2SO4 solution. These mixtures were heated at 373K for 3 day. In the second procedure, the V (1.6 %) and Ni (1.8 %) mesoporous molecular sieves were synthesized with tetraethylorthosilicate (TEOS), VOSO4.5H20, or Ni(CH3COO)24H20, ethanol, 2-prophanol and an acidified mixture of CTMAB in water. The gels with pH 13 were loaded into teflon-lined steel autoclave and heated at 373K for 3 days. The pH of the gels was ft~her modified to 11 with H2SO4. The gels were then subjected to another 3 days hydrothermal treatment at 373K. The solid products dried in air at 373K and calcined at 773K. 2.2. Characterization The obtained materials were characterized by XRD (Phih'ps PW 170 diffractometer), N2 adsorption-desorption (Tristar, Micromeritics), SEM (Philips XL-20 microscope), TEM (Philips Tecnai microscope) and FTIR (Spectrum 2000, Perkin Elmer) techniques. The oxidation of aromatic hydrocarbons (styrene and benzene) was carried out in the thermostated glass reactor with magnetic stirring in the presence or absence of the solvent (acetonitrile). The reaction temperature and time were 343K and 24h, respectively. The molar ratio of hydrocarbon/solvent/hydrogen peroxide was 1/-/3 for benzene and 1/1.8/3. After reaction, the catalyst was separated by centrifugation and the oxidation products were chromatographically analyzed. 3. RESULTS AND DISCUSSION
Using the procedures described in the experimental section, the mono- and bimetallic ions incorporated mesoporous structures, V-MCM-41, Co-MCM-41, Cr-MCM-41, NiMCM-41 VCo-MCM-41 and CrNi-MCM-41 could be generated with different M/Si and Co/V molar ratios (0.02-0.1 and 0.3-3 respectively). The synthesis conditions, the nature and concentration of the metal in monometallic silicates, the association of the metals and their molar ratio in the bimetallic silicates and the oxidation state of the metal cations in the precursors can affect the regularity of hexagonal arrangement of the MCM-41 structure, the incorporation of the metal into these structures and the physico-chemical and catalytic properties of the obtained solids.
577 3.1. Effect of nature and precursors of metal ions, M/Si molar ratio, MI/M2 molar ratio and silica sources on the synthesis of mesoporous molecular sieves
3.1.1. Nature and oxidation state of the metal cations The XRD patterns (Fig. 1) of all the mono- and bi-metallic (the ratio of V/Co and Cr/Ni is equal to 1) samples show 4 diffraction lines at the small angle range, characteristic of well ordred MCM-41 type mesoporous materials with a regular hexagonal array of their cylindrical channels. N2 adsorption-desorption isotherms are type IV (see Fig. 6), typical for mesoporous materials. The pore size distribution (see Fig. 6) obtained from BJH method give a monodal peak centred at around 2.7 0.2 nm. The regular hexagonal arrangement of cylindrical channels of our materials can be clearly viewed by TEM technique. Fig. 2 depicts the TEM images of one mono-metallic incorporated and one bimetallic ions incorporated samples. When electron beam of TEM is in the parallel direction to the channel axis of our materials, a honeycomb structure can be observed while when the electron beam of TEM is perpendicular to the channel axis of our materials, layer-type materials with equal distance between layers can be visualized. All the above technical analysis results such as XRD, N2 adsorption-desorption isotherms, pore diameter distributions
Fig. 1. X-ray diffractograms of the mono- and bimetallic mesoporous molecular sieves Fig.2. TEM images of the molecular sieves
578
Fig.3. SEM images of the monometallic samples: (a)" CrlN4, (b): Cr2N4 and (c): NiN4 and TEM (Fig. 2) confirm strongly the highly structured mesoporous materials with well regular hexagonal array of cylindrical channels of our mono- and bi-metallic materials. It has to be noted that the higher ordered and stable hexagonal phases are obtained when the oxidation state of the cation, introduced during synthesis, is favourable for a tetrahedral coordination, i.e. RM(n+)/Ro(2.)ratio in the range of 0.3 and 0.5. The XRD diffractogram of sample synthesized with CrO3 (Cr2N4) is better resolved compared with that of sample synthesized with Cr(NO3)3 (CrlN4) (Fig. 1), showing higher order in the mesoporous structure of Cr2N4 than that of Cr~N4. The initial oxidation state of metal Fig.4. SEM images of NiCrprecursor is thus important for the regularity of final MCM-41 structure of materials. The morphology of aggregates formed by a large number of small spherical particles observed by SEM for most of monoand bi-metallic ions incorporated samples such as Cr2-MCM-41, Ni-MCM-41 (Fig. 3) and CrNi-MCM-41 (Fig. 4) samples, except Cr~-MCM-41 (Fig. 3a), is typical for metals modified MCM-41 as reported previously [9, 10]. A "sandy-rose" morphology is visualized for Crl-MCM-41 (Fig. 3a). It is evident that the metal precursors can modify the ordered structure and the morphology. The characteristics of the mesoporous structure of our mono- and bimetallic molecular sieves are listed in Table 1. From this table, it can be concluded that the above described mesoporous silicates have heteroelements in the framework. The unit cell parameter ao confirm incorporation of the metal cation in the framework of our materials. It is observed that for a defined metal and metal source, the unit cell parameter ao increases with increasing the metal content and then decreases when the metal content is too high. As the bond length of M-O is higher than that of Si-O, the increase in unit cell parameter with increasing the metal content suggests the incorporation of metal ions in the framework of mesoporous structures. The loss in unit cell parameter observed when the metal content is too high is due to the loss, at least a part, of the mesoporous structure. The
579 decrease in surface area and the loss in the intensity of XRD peaks support our explanation. IR-spectra of the transition metal containing mesoporous silicas are similar to that of amorphous silica. Only an supplementary absorption band at around 940 nm is present in all the spectra of the mono- and bimetallic incorporated samples but its intensity is variable. This band can be attributed to the M-O-Si bond and often used as a prove of the incorporation of metal ions in the framework of silicas. Table 1. Characteristics of the mono- and bi-metallic ions incorporated mesoporous MCM-41 molecular sieves
Catalyst
M %
SBET O ( m 2 / g ) (nm)
ao Catalyst (nm)
M %
SBET O ( m 2 / g ) (nm)
ao (nm)
VN4 1.6 1055 2.5 4.65 NimN4 1.8 945 2.8 3.97 VN1 1.7 1032 2.8 4.55 NizN4 3.6 828 2.6 4.70 CrgN4 1.8 958 2.6 4.48 Ni3N4 5.5 777 2.6 4.60 CraN4 1.8 862 2.7 4.56 N4N4 7.4 654 2.7 4.30 Co1N4 1.8 990 2.8 3.99 NsN5 9.2 568 2.7 4.20 Co2N4 3.6 905 2.8 4.70 VCo1N4 3.4 948 2.8 4.70 Co3N4 5.5 644 2.7 4.70 VCo2N4 3.4 1013 2.7 4.60 Co4N4 7.4 568 2.6 4.80 VCo3N4 3.4 973 2.9 4.70 Co5N4 9.2 681 2.5 3.92 CrNiN4 3.4 914 2.7 4.32 NimN1 1.9 1021 2.8 4.51 N=MCM-41, N4: synthezised from sodium silicate, N l: synthesized from TEOS, CrlN4 synthesized from Cr(NO3)3 and Cr2N4 from CrO3; Col-CO5: different metal content and obtained from Co(NO3)3 and Nim-Nis: different metal content and synthesized from Ni(CH3COO)a-4H20
3.1.2. Effects of the M/Si molar ratio The variation of the M/Si molar ratio, between 0.02 and 0.1, for Co-MCM-41 and NiMCM-41 can induce a modification in the ordered hexagonal structure and surface area. Xray diffractograms (Fig. 5) show a less ordered hexagonal structure when M/Si >0.06. Surface area and unit cell parameter decrease with increasing M/Si molar ratio (Table 1 and Fig. 5). For a high quantity of the metal, the incorporation of the metal in the framework of mesoporous structure is difficult. The surface area decreases remarkably when the metal content is higher than 3.5 %. Above this value, the regular hexagonal arrangement was progressively lost. Nitrogen adsorption-desorption isotherms of the samples with different Co/Si molar ratio (Fig. 6) are characterized by a very steep increase at the relative pressure P/P0 of around 0.38, indicating the homogeneity of our bimetallic incorporated mesoporous molecular sieves.
580
800
'~
"
500 . 2
.
Col
. 4
.
6
8
II lllIl~ll~r
-
10
2
i
I
I
I
2
4
6
8
> 10
2 Fig.5. X-ray diffractograms of the Co-MCM41 and Ni-MCM-41 samples with a variable metal content
0 .... 0.0
,
,
,
,
0.2
0.4
0.6
0.8
1.0
RelativePresane (P/P0) Fig. 6. N2 adsorption-desorption isotherms and pore size distribution of a series of Co-MCM41 samples with a variable metal content
This value decreases only slightly from 0.38 to 0.34 with increasing metal content. The adsorption capacity decreases also with metal content. The BJH pore size distribution
581 shows a very narrow monomodal peak for all the M/Si molar ratio values and the average pore diameter decreases insignificantly (Fig. 6 and Table 1). The decrease in surface area and adsorption capacity with increasing metal content suggest that not all the metal ions introduced in the synthesis gel can be incorporated into the mesoporous framework. A part of the metal ions introduced in the synthesis gel will be present as extra-framework species and dispersed on the internal surface of mesopores. These extra-framework species can be easily removed by leaching since they are less strongly linked with internal surface of mesopores. SEM (Fig. 7) images show the typical morphology of aggregates of small spherical particles, characteristic of mesoporous metallosilicates. It has to noted that a notable quantity of the amorphous phase was observed in the samples with a high metal content (Co4N4 and Co5N4) while the particle size decreases with increasing Co/Si molar ratio. Highly organized mesoporous structures were observed with low M/Si ratio by TEM (Fig. 8) and lower ordered structure was viewed in the samples with high M/Si ratio. 3.1.3. Effects of the C o N molar ratio
The M1/M2 molar ratio can modify the structure, the morphology and the catalytic activity of the mesoporous redox molecular sieves. A highly ordered hexagonal arrangement (Fig. 9) is evidenced for Co/V molar ratio of 0.3 (VCo2-MCM-41). This sample has the high surface area (Table 1). It is very interesting to note that the calcination ~00_ increases the intensity of the first Bragg peak and calcined metallosilicates have a highly ~000ordered pore system with a high porosity. It is possible that the calcination can allow a better .~ 4OO0 penetration of metal ions into the framework and "a a reorganization of structure into a well ordered " ~00 framework. TEM (not shown here) images confirm the 0 higher ordered pore systems for V-C02 and VC03 samples. SEM (Fig. 10) micrographs show the typical morphology of mesoporous Fig.9. X-ray diffractograms of the metallosicates. VCo-MCM-41 samples 3.1.4. Effects of the synthesis method
Transition metal containing mesoporous silicas were obtained by hydrothermal treatment from two different silica sources: sodium silicate (VN4 and NiN4) and tetraethylorthosilicate (VN1 and NiN1). The sol of TEOS is treated in autoclave in basic medium to obtain highly hydrolysed and reticulated silica species during the condensation. The condensation of silica species in the presence of the metal cations favors their incorporation into framework of silica. These silica species with metal cations are ordered around the well self-organized micelles of CTMABr at pH 11. All the samples obtained from TEOS have a very high ordered structure (Fig. 11), a very high surface area (Table 1) and are a very stable structure. TEM and SEM images show a highly ordered hexagonal structure for all the samples synthesized both with sodium silicate and TEOS. However, it
582 is observed that the particle size evidenced by SEM is smaller for the samples prepared with TEOS.
Fig.10. SEM images ofthe VCo-MCM-41 with V/Co molar ratios: (a): 0.3, (b)" 1.0, (c): 3.0 3.2. Evaluation of catalytic activity in selective oxidation of styrene and benzene Metalosilicates synthesized here with MCM-41 structure are very active in oxidation of styrene and benzene in liquid phase (Table 2). The selectivity tD to benzaldehyde and phenol are very high for all the catalysts. For Ni-series of catalysts, in oxidation of styrene, the activity increases, however, the selectivity decreases with increasing the metal 2 content. In oxidation of benzene, the Fig. 11. X-ray diffractograms of the V and Niconversion is not very attractive (4MCM-41 samples synthesized with TEOS (VN1 11%) although the selectivity is high. and NiN1) and with sodium silicate (VN4 and For Co-series of catalysts, in oxidation NiN4) of styrene, the catalyst with lowest metal content gives the highest conversion and selectivity. The conversion is then practically insensitive to the metal content while the selectivity decreases significantly with increasing metal content. In oxidation of benzene, the variation in conversion is opposite. The conversion increases sharply with metal content. Comparing the two Cr-modified catalysts synthesized with different metal precursors, although the selectivity for benzaldehyde and phenol from styrene and benzene, respectively, remains quite high, the variation in conversion in these two reactions is opposite too. The sample with less well organized structure CrlN4 synthesized with Cr(NO3)3 gives a higher conversion for styrene oxidation while CrEN4 synthesized with CrO3 shows a higher activity for benzene oxidation. The oxidation state and the location of metal ions and the regularity of structure can play an important role in benzene and styrene oxidation reaction. Comparing the samples
583 synthesized with different silica sources, VN1 and VN4, NilN1 and NilN4, the results given in Table 2 demonstrate that the catalytic activity observed for N1 samples synthesized with TEOS do not differ strongly from those observed for N4 samples synthesized with sodium silicate under same reaction conditions. However, the efficiency of the hydrogen peroxide and the stability under reaction conditions of the redox molecular sieves are higher when the samples are synthesized by TEOS (N1). For mono-metallic samples, for a given metal content (1.7-1.9%), the conversion varies in an order of VN 1 o o T rv CO
76 74 72 70 68
84 o~ 82
DS6 DS5
c-
0
lDs4 DS3~ ,
,
50
1O0
Surface area (rr~/aD
Fig.4. Mercaptan conversion as a function of surface area.
DS6
80
~ 78 9 > 76 co 74 0 -r 72 co ,v 70 68
5 DS4 DS3 i
i
i.
0,1
0,2
0,3
.
.
.
i
0,4
C__.d3dtcontent (mmd/g)
Fig.5. Mercaptan conversion as a function of cobalt content.
599 9 8
E 7ca. 6
@
-
,o,
I
03 5 CK ~4 .x2_ 3
~ j
n,, 2 1
o o
@
9
I
r
i
f
i
1
2
3
4
5
--
6
7
8
9 10 11 12 13 14 15 16 17
Time (day)
Fig. 6. Residual mercaptans as a function of time for DS6 catalyst in a pilot test. The selected catalyst, DS6, was tested over a long period (17 days) with a pilot modeling the fixed-bed sweetening process. Knowing that the actual norm of mercaptan amount in carbureactor product is about 15 ppm maximum, we observed after one day, a residual mercaptan of about 0.8 ppm corresponding to 99.4 % removal (Fig. 6). The amount of RSH increases slowly until 8.4 ppm atter 17 days testing. This low deactivation is probably due to the numerous impurities contained in the industrial hydrocarbon feed. Addition of active charcoal by mechanical crossing in the fixed-bed reactor or incorporated it into the final catalyst could probably remove these impurities and improve the lifetime of the catalyst. 4. CONCLUSION In conclusion, we have shown that without alkali medium a very good mercaptan removal using a bifimctional catalyst was obtained confirming that the adequate tuning of basic and oxidant properties lead to a promising catalyst. This catalyst consists of cobalt phtalocyanine parallel-intercalated MgA1 hydrotalcite obtained by a direct synthesis method. Moreover it exhibited a good mechanical resistance and did not need further post treatment to improve physical properties. ACKNOWLEDGMENT We thank Dr. Anne Davidson and Mr. Bernard Morin from Laboratoire de Rractivit6 de Surface for their help in ESR characterization.
600 REFERENCES
1. B. Basu, Satapathy and A.K. Bhatnagar, Catal. Rev.-Sci.Eng., 35(4) (1993) 571. 2. A. Corma and R.M Martin-Aranda, J.Catal., 130 (1991) 130. 3. J.Muzart, Synth. Commun., 15 (1985) 285. 4. J.Muzart, Synth. Commun., 1 (1982) 60. 5. A. Corma, A. Fornes, R.M Martin-Aranda, H. Garcia and J. Primo, Appl. Catal., 59 (1990) 237. 6. M.J. Climent, A. Corma, S. Iborra and J. Primo, J. Catal., 60 (1995) 151. 7. D. Tichit, M. H. Lhouty, A. Guida, B.H. Chiche, F. Figueras, A. Auroux and D. Bartalini, E.J. Garrone. J. Catal., 50 (1995) 151. 8. E. Narita, T. Yamagishi, K. Tamaza, O. Ichizo and Y. Umetsu, Clay Sci., 9 (1995) 187. 9. F. Cavani, F. Trifiro and A. Vaccari, Catal.Today, 11(1991) 173. 10. S. Miyata, Clays Clay Miner., 31 (1983) 305. 11. W.T. Reichle, Chemtech., 16 (1986). 12. E. Suzuki, S. Idemura and Y. Ono, Clays Clay Miner., 37 (1989) 173. 13. T. Kwon and T.J. Pinnavaia, J. Mol. Catal.,. 23 (1992) 74. 14. M. Del Arco, M.V.G. Galiano, V. Rives, R. Turjillano and P. Malet, Inorg. Chem., 35 (1996) 6362. 15. K.J. Balkus Jr and A.G.Gabrielov, J. Incl. Phenom. Mol. Recog. Chem., 21 (1995) 159. 16. F. Bedoui, Coord Chem. Rev., 39 (1995) 144. 17. N. Herron, Chemteeh., 19 (1989) 542. 18. S. Kannan, S.V. Awate and M.S. Agashe, Recent Advances in Basic and Applied Surface Science and Catalysis., 113 (1998) 927. 19. K.A. Carrado, J.E. Forman, R.E, Botto and R.E. Winans, Chem. Mater.,5 (1993) 472. 20. M.A. Dredzon, Inorg. Chem., 27 (1988) 4628. 21. M. Perez-Bernal, R. Ruano-Casero and T.J. Pinnavaia., Catal. Lett., 11 (1991) 55. 22. V.I. Ileiv, A.I. Ileva and L.D. Dimitrov, App. Catal. A., 126 (1995), 333. 23. Mercaptan sulphur in gasoline, kerosene, aviation turbine and distillate fuels (potentiometrie method). ASTM D 3227-83. 24. W.T.Reichle, J .Catal., 94 (1985) 547. 25. P.K. Dutta and M. Puri, J. Phys. Chem. 93 (1989) 376. 26. T.J. Wallace, A. Schriescheim, H. Hurwitz and M.B. Glaser, Ind. Chem. Prod. Res. Dev., 3 (1964) 237. 27. Constantino V.R.L. and Pinnavaia T., Catal. Lett., 23 (1994) 361. 28. Constantino V.R.L. and Pinnavaia Y., Inorg. Chem., 34 (1995) 883.
Studies in SurfaceScienceand Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
601
Structural and catalytic properties of Zr-Ce-Pr-O xerogels S. Rossignol, C. Descorme, C. Kappenstein and D. Duprez Laboratoire de Catalyse en Chimie Organique, UMR 6503, CNRS et Universit6 de Poitiers, 40 Avenue du Recteur Pineau, 86022 Poitiers, France. Zr0.10(Cel-xPrx)0.90Oz mixed oxides (x between 0 and 0.75) were prepared by coprecipitation (nitrates) or by the sol-gel route. Zirconium n-propoxide and cerium and/or praseodymium nitrates were used as precursors. "Sol-gel" oxides calcined at 900~ were shown to be cubic with a fluorite-type structure. Coprecipated oxides could not be obtained as solid solutions. The BET surface area of these samples rapidly decreases when x > 0.50. A Raman study confirmed that all oxides were cubic and put in evidence the presence of oxygen vacancies. The optimum oxygen storage capacity was obtained for Zr0.10(Ce0.50Pr0.50)0.9002. It appears that the substitution of cerium by praseodymium in Zr0.10Ce0.90Oz mixed oxides leads to a material with improved redox properties. The presence of vacancies, associated with pr3+/pr4+ ions, is thought to be responsible for these enhanced OSCs. 1. I N T R O D U C T I O N Ceria CeOz is a crucial component of modern automotive pollution control catalysts [1]. This oxide is a promoter in the water gas shift reaction and prevents the sintering of the noble metals used as active centers. More important is the ability of this oxide to store oxygen. This phenomenon is associated with a fast Ce4+/Ce 3+ redox process in the solid, also involving anionic vacancies [1-4]. It was shown that up to 50% zirconium could be substituted to cerium into the ceria framework to form solid solutions with facecentered cubic structures. In a previous work, we have evidenced that among the Zr-Ce-O mixed oxides, the compound Zr0.1Ce0.902 prepared by sol-gel procedure displays the highest oxygen storage capacity [5, 6]. Praseodymium is one of the possible "new" additives, which today attracts more and more attention. In fact, it was demonstrated that the oxygen exchange occurs at lower temperature on cerium-praseodymium mixed oxides than on ceria [7]. Furthermore, high temperature pretreatment does not affect the oxygen exchange capacity (OSC) of the mixed oxides. Moreover, high surface area PrOy-ZrO2 materials with a fluorite-type structure were also prepared by the sol-gel way [7, 8]. The main objective of our study was to modify the composition of CeO2 and Zr0.10Ce0.90Oz samples, in order to follow the influence of cerium substitution by praseodymium and to enhance both thermal stability and oxygen storage capacity. The results will be discussed on the basis of DTA, XRD, Raman spectroscopy, BET surface area and OSC data.
602
2. E X P E R I M E N T A L 2.1. Preparation Two series of samples were prepared. Series 1 consists of Zr0.10(Cel-xPrx)0.90Oz samples while, in the second series, oxide samples do not contain any zirconium (Cel_xPrxO2). Preparation methods were described earlier [5]. Solution concentrations were adjusted in order to synthesize 3 g of solid in one batch. All reactants were Aldrich products, 99% purity. They were used without any further purification. The synthesized materials were dried in air at 60~ for 1 h and at 120~ overnight before further calcination in air at 900~ for four hours. 2.2. Characterization Differential Thermal Analysis (DTA) experiments were carried out in dry air (Air Liquide, < 5 ppm impurities) between 25 and 500~ using a Thermal Analyst 2100 TA apparatus. Samples previously dried at 120~ were heated at 5~ -1. Preliminary experiments showed that no DTA peak was observed above 500~ Specific surface areas were determined by N2 (Air Liquide, 30% N2 in He) adsorption at -196~ (one point BET method) with a Micromeritics Flow Sorb II. XRD analysis was performed on a Siemens DS00 powder diffractometer using CuKct radiation (~(CuKctl) = 0.15406 nm). Standard diffractograms recording parameters were: dwell time = 1 s, step = 0.05 ~ 20 range = 10 to 90 ~ divergence slit = 1~ for precise cell parameter determination: dwell time = 10 s, step = 0.02 ~ Raman spectra were recorded using a Perkin Elmer spectrometer: Nd-YAG laser (1064.4 nm, 100 mW), dwell time = 120 s, number of scans = 20. Oxygen storage capacity (OSC) was measured at 400~ under atmospheric pressure. A 20 mg sample was continuously purged with helium (30 cm3.min-1). Successive or alternate pulses (0.265 cm 3) of Oz (Air Liquide, < 5 ppm total impurities) and CO (Air Liquide, N20) were injected every minutes in order to simulate lean and rich operating conditions as those encountered in an Otto engine coupled with a three-way catalytic converter. The Oxygen Storage Capacity (OSC) was calculated from the CO consumption after stabilization of the sample in alternate pulses condition. 3. RESULTS 3.1. Structural characterizations (DTA, XRD,
Raman)
DTA and XRD. After the drying step at 120~ all the samples correspond to complex mixtures containing major M iu (precursor oxidation state) and minor M TM oxides and hydroxides (M = Ce or Pr) and no well-defined phase could be evidenced from the XRD results. Fig. 1 shows the thermograms obtained under air of all series 1 samples and of one series 2 sample (x = 1). All the thermograms consist in endothermic peaks due mainly to the elimination of a small quantity of the residual water molecules and to the transformation of hydroxide into oxide ions. Therefore, the possible exothermic reactions such as the formation of a crystallized phase or the oxidation of Ce Iu or Pr m atoms are hidden by the endothermal events.
603 For x=0 in series 1, the peak at 250~ is characteristic of the appearance of a fluoritetype solid solution Zr0.1Ce0.9Oz [6]. For the other samples, this peak is shifted to higher temperature and additional endothermic peaks appear at lower temperatures. Diffractograms of the samples after calcination at 900~ are displayed in Fig. 2 in the 260-36 ~ 20 range, including the calculated crystallite size. The samples of series 1 (with x between 0 and 0.75) present only symmetrical diffraction peaks in agreement with the presence of one phase identical to Zr0.]0Ce0.9002 compound (x = 0) described in a previous paper [5]. This mixed oxide has been shown to be a solid solution with a fluorite-type structure and lattice parameter a = 5.3962 A. The lattice parameters of some samples are given in Table 1 and we observe a slight contraction of the cell as praseodymium atoms are substituted to cerium atoms, in agreement with a smaller ionic radii for Pr 4+ [9]. For the sample Zr0.10Pr0.9002 (Fig. 2, x = 1, series 1) the distortion of the diffraction peaks indicates the presence of two phases: a major Pr6011 phase and a fluorite-type structure solid solution, in agreement with the work of Sinev et al. [10]. X=I series 2 I
Pr60 ,
x=0.75 series 1
II
'
Particle size l
~
x=1series2
x=O series 1
I
~T = 0.01
100
200
300
1"/~
400
500
. ~
~ ~ ~ j ~ x
=0series 2
600
Fig. 1. DTA profiles of Zr0.1o(Cel_ xPrx)0.9002 oxides prepared by sol-gel (series 1) and Cel_xPrxO2 oxides prepared by coprecipitation (series 2).
20/ degree
Fig. 2. XRD patterns of Zr0.10(Cel_ xPrx)o.90Oz oxides prepared by sol-gel (series 1) and Cel_xPrxO2 oxides prepared by coprecipitation (series 2).
604
In series 2, the oxides CeOz (x=0) and Pr6Oll (x=l) appear to be single phase stable after air calcination at 900~ The non-stoichiometric praseodymium oxide can be better formulated: [prIV4 prnI2][O-IIll (Vo~176 which displays the importance of the vacancies. For both series, the samples containing no cerium (x= 1) display different DTA profiles by comparison with the single phased samples (Fig. 1) and this behavior agrees well with the final presence of the praseodymium oxide phase Pr6Oll and the successive formation of intermediate phases [ 11]. Table 1. Lattice parameters and experimental and calculated OSC. Series
Praseodymium atom. fraction
Lattice
parameter
O S C cal
(OSCexp/OSCcal)
(gmolCO.g-)
O S C exp 1
([amolCO.g -1)
Ratio
(A) Series 1
Series 2
x=0
5.3962
197
180
1.09
x = 0.25
5.3913
272
175
1.56
x = 0.50
5.3937
284
108
2.63
x= 0.75
5.3953
234
67
3.51
x = 0.1
42
Not calculated
x=0
Phase mixture 5.4138
60
34
1.77
x= 1
5.4641
35
6
6.28
Raman spectroscopy. Four Raman spectra only are presented in Fig. 3. In fact, for samples with x higher than 0.25, fluorescence effects due to praseodymium were so important that no spectra could be recorded. The band at ca. 460 cm -1 is attributed to the T2g vibration mode of the metal-oxygen bond in the fluorite-type structure [12-14]. In the case of series 1 samples, the presence of this band confirms that these oxides are solid solutions. In order to explain the shift of this band as a function of x, samples were analyzed again after calcination at different temperatures (Fig. 4). This band is shifted towards higher frequencies as the calcination temperature increases (200, 400, 900~ This phenomenon could be related to structural ordering. On the contrary, a shift towards lower frequencies can be observed as zirconium or praseodymium atoms are substituted to cerium atoms. This shift indicates that the incorporation of Zr or Pr atoms induces ~i perturbation of the M-O bond involving some kind of disordering or structural distortion. Mc Bride et al. [15] gave two reasons for the observed frequency shift: (i) an increase in the oxygen vacancy density and (ii) an expansion or contraction of the cell. In fact the presence of oxygen vacancies in this material could be explained by the presence of reducible praseodymium atoms (Pr4+/pr3+). Moreover, in the case of Ce0.75Pr0.2502 (Fig. 3), a small and broad band appears at about 570 cm -1. The same authors have tentatively attributed this band to the presence of oxygen vacancies in the material. In
605
the other samples, this small band could not be observed. This fact can be linked either to the very weak intensity of the recorded spectra or to the absence of praseodymium atom in the oxide samples.
4
>,
3
X= 0.25 series 1
4-1 X= 0.25 series 2
~ 46~ cm-1
A
~4~4crn'
~
2
l/
5701 cm 1
o
120 ::;
100
X= 0 series 1 465 cm 1
80
.~- 60
. I
1500-1 1200 ] 900
II
II
800
600
1
~176176
3~ 400
X= 0 series 2 466 cm 1
660
200 800
R a m a n shift / cm 1
460
2()0
R a m a n shift / cm 1
Fig. 3. Raman spectra of Zr0.10(Cel-xPrx)0.9002 and Cel.xPrxO2 samples with x = 0 and x = 0.25 after calcination at 900~
470
u U
C
E
465 460 t"" ~ x= 0 series 2 r x = 0 series 1 I x = 0.25 series 2] x = 0.25 series 1}
455 450
~
...
~
~
~
~
o
445
i
0
200
i
i
400 600 Temperature / ~
i
800
1000
Fig. 4. Raman shift as a function of the calcination temperature for the two series of samples with x = 0 and x= 0.25.
606
3.2. Surface characterization and oxygen storage capacity All BET surface areas are reported in Fig. 5. Whatever the series, introduction of Pr results in a decrease of the BET surface area. For series 1 samples, this decrease is even more pronounced for x > 0.5. It seems that the presence of P r 6 O l l could be responsible for this decrease of surface area. Furthermore, the presence of zirconium remains crucial to stabilize cerium-praseodymium oxides: BET surface areas of series 1 samples are much higher than those of series 2, except for x = 1. Praseodymium can stabilize the texture of ceria but its influence in obtaining high surface area oxides appears to be much weaker than in the case of zirconium addition. [2-4, 16]
40
--0- series 2 l # series 1
35
,:,
~ 3o 9 25 u
20
~
15
0
u 111 ~
5
---....... ""
"-0 .
.
.
.
.
.
"-0
J
,
,
i
0,2
0,4
0,6
0,8
1
Fig. 5. Specific surface areas of Zr0.10(Cel-•215 oxides prepared by sol-gel (series 1) and Cea_xPr,,O2 oxides prepared by coprecipitation (series 2) after calcination at 900~ The evolution of the oxygen storage capacity (OSC; in t.tmol-CO.g -1) is given in Fig. 6 as a function of the amount of praseodymium in the solids. For both series, the introduction of praseodymium induces an increase of the OSC up to x = 0.50, whereas above x = 0.50, a decrease is observed. Therefore, an excess of praseodymium has a negative effect on the OSC [17-19]. In order to understand how praseodymium can influence the oxygen storage capacity, "calculated" OSC values were obtained as follows:
OSCcal =. b S Na 2 with:
(1)
- S = BET surface area (m2.g -1) - a = lattice parameter (m) - N - Avogadro number - b = fraction of reducible elements (Ce + Pr) in the unit cell (0.9 for series 1, 1 for series 2)
This calculation is based on the following assumptions: - all the C e Iv o r Pr Iv atoms of the surface monolayer are reduced to oxidation state III"
607
- the surface faces are { 100} faces and contain 2 metal atoms and 4 oxygen atoms per unit surface area a 2 (a = cubic parameter)" therefore one oxygen atom out of four participates to the OSC process as indicated by equation (3); oxygen atoms bonded to reducible elements are the only one to participate to the OSC.
300
";"r,.200250 )d~ ~
' ' ~ serie21 s
0
E 150 o 100 0
, .-"" o
50
.
0
i
i
0,2
0,4
Fig. 6. OSC values measured at 400~ calcined at 900~
x
i
l
0,6
0,8
1
as a function of x for the two series of samples
Calculated and experimental values, expressed in lamol-CO.g -1, are collected in Table 1. The ratio of these two values may represent the number of oxygen layer involved in the redox process. In all cases, the calculated values are lower than the experimental ones. This observation shows that 2 to 3 layers are involved in the redox process. The increase of the OSC along with the amount of Pr shows that the presence of this element induces the creation of anionic vacancies. In fact, in substitution to cerium cations, Pr ions may undergo a redox process (pr3+/pr 4+) whereas it is not possible for zirconium cations. 4. C O N C L U S I O N S We prepared by sol-gel some thermally stable Zr0.a0(Cel-• mixed oxides (x between 0 and 0.75) with a fluorite-type structure. This structure was confirmed by the presence, in the Raman spectrum, of a single band ca. 460 cm -1, characteristic of the M-O vibration in the fluorite-type structure. Moreover, the band position and the shoulder at 570 cm -1 indicate the presence of oxygen vacancies probably associated with praseodymium cations. Consequently, high OSCs appears to be the result of the presence of both cerium and praseodymium atoms. Thus, addition of praseodymium atoms into zirconia-ceria oxides appears to be very promising for the design of new automotive catalysts. R E F E R E N C E S
1. C. K. Narula, J. E. Allison, D. R. Bauer and H. S. Gandhi, Chem. Mater., 8 (1996) 984. 2. T. Bunluesin, R.J. Gorte and G.W. Graham, Appl. Catal. B, 14 (1997) 105. 3. Y. Sun and P. A. Sermon, J. Mater. Chem, 6 (1996) 1025.
608 4. A.Trovarelli, F. Zamar, J. Llorca, C. Leitenburg, G. Dolcetti and J. T. Kiss, J. Catal., 169 (1997) 49. 5. S. Rossignol, F. Gerard and D. Duprez, J Mater. Chem., 7 (1999) 1615. 6. S. Rossignol, Y. Madier and D. Duprez, Catal. Today, 50 (1999) 261. 7. C. K. Narula, L. F. Allard and G. W. Graham, J. Mater. Chem, 9 (1999) 1155. 8. M. D. Krasil'nikov, I.V. Vinokurov and S.D. Nikitina, Fiz Khim. Electrokhim., 3 (1979) 123. 9. N. N. Grenwood and A. Earnshaw, Chemistry of the Elements, 2nd edition, 1997, Butterworth Heinemann, Oxford, UK, p. 1295 10. M. Yu Sinev, G. W. Graham, L. P. Haack and M. Shelef, J. Mater. Res., 11(8) (1996) 1960. 11. See ref. 9, p. 643. 12. S. Mochizuki, Phys. Stat. Sol. (b), 114 (1982) 189. 13. W. H. Weber, K. C. Hass and J. R. Mc Bride, Phys. Rev. B, 48 (1) (1993) 178. 14. G. Groppi, C. Cristiani, L. Lietti, C. Ramella, M. Valentini and P. Forzatti, Catal. Today, 50 (1999) 399. 15. J. R. Mc Bride, K. C. Hass, B. D. Poindexter and W. H. Weber, J. Appl. Phys., 76 (4) (1994), 2435. 16. S. Imamura, J. I Tadani, Y. Saito, Y. Okamoto, H. Jindai and C. Okamoto, Appl. Catal. A: Gen., 201 (2000) 121. 17. G. Balducci, J. Kaspar, P. Fornasiero, M. Graziani and M. Saiful Islam, J. Phys Chem. B., 102 (1998) 557. 18. L. Mubmann, D. Lindner, E.S. Lox, R. Van Yperen, T.P. Kreuser, I Mitsushima, S. Taniguchi and G. Gaff, SAE Technical Paper 970465 (1997). 19. C. Leitenburg, A. Trovarelli, F. Zamar, S. Maschio, G. Dolcetti and J. Llorca, J. Chem. Soc., Chem. Comm., (1995) 2181.
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
609
Influence of the precursor (nature and amount) on the morphology of MoO3 crystallites supported on silica D. Navez ~, G. Weinbergb, G. Mestl~, P. Ruiza and E.M. Gaigneauxa• a
Unitt~de catalyse et chimie des mat&iaux divis6s, Universit6 catholique de Louvain, Croix du Sud 2/17, B-1348 Louvain-la-Neuve, Belgium.
b Department of inorganic chemistry, Fritz-Haber-Institut der Max Planck Gesellshafl, Faradayweg 4-6, D- 14195 Berlin, Germany. NanoScape AG, Frankfurter Ring 193A, D-80809 Munich, Germany. The influence of the nature of the precursor and the amount used is investigated in the preparation of MoO3 catalysts supported on a mesoporous silica by a wet impregnation method. Four precursors are compared: ammonium heptamolybdate, peroxopolymolybdate, a citrate complex of Mo and an oxalate complex of Mo. The MoO3 loading is adjusted between 0 and 2 theoretical monolayers of MOO3. Characterization of the samples was performed by N2 physisorption, X-ray diffraction and scanning electron microscopy with as main objective the evaluation of the dispersion of MoO3 on the silica and of the morphology of MoO3 crystallites eventually obtained. At low loadings, MoO3 appears as amorphous material remaining at the outer surface of the silica. At high loadings, MoO3 appears as molybdite crystals. The loading above which MoO3 crystals are obtained depends on the precursor used. The morphology of MoO3 crystals, when obtained, also depends on the precursor used. Isotropic crystals are obtained with Mo oxalate and peroxopolymolybdate while Mo citrate leads to crystals with largely developed basal faces. Ammonium heptamolybdate leads to crystals with an intermediate morphology. Our results thus show that changing the nature and the amount of precursor used to impregnate MoO3 on silica offers a promising possibility to adjust the morphology and, to some extent, the dispersion of MoO3 crystals at the surface of the support. 1. INTRODUCTION AND OBJECTIVES In addition to its dispersion, the morphology of the crystallites of the active phase is another important parameter that influences the performance of supported catalysts. The morphology of the crystallites may become particularly crucial when the supported material exhibits "structural specificities", i.e. when the different faces of the active phase crystals have different behaviors in a given process, either in terms of activity and selectivity or of surface dynamics under the catalytic conditions. MoO3 possesses "structural specificities" in oxidation reactions and related processes carried out in the presence of oxygen (oxidative dehydrogenation, oxygen-assisted dehydration *Corresponding author: E.M.G. is Chercheur qualifi6 (Research associate) for the National Foundation for Scientific Research (FNRS) of Belgium
610
of alcohols and of amides, etc). Typically, on the one hand, the (010) basal faces are weakly active and preferentially perform the total oxidation of olefins, but the (100) lateral faces selectively transform olefins to partially oxygenated compounds [1-3]. On the other hand, the synergetic effects between MoO3 and a-Sb204 or BiPO4 (or any other spillover oxygen donor phase, as defined in the Remote Control Mechanism theory [4]) are much more intense when the MoO3 crystals develop larger (010) basal faces [5,6]. The phenomenon is due to the pronounced ability of (010) faces to reconstruct under the influence of spillover oxygen to new more active structures, namely nanometric pits and steps with walls oriented as (100) facets, while (100) faces do not undergo such modification during the reaction. These differences of behavior most likely originate, at least partially, from a difference of reducibility of the different faces of MoO3 crystals, the later being itself related to a difference in the co-ordinations of the superficial oxygen atoms exposed on the different faces. Namely, the accessible oxygen atoms on the (010) face are doubly linked to Mo atoms (Mo-O), while both doubly linked oxygen and oxygen species bridging two Mo atoms (Mo-O-Mo) are exposed on the (100) face [7,8]. The morphology for MoO3 crystals in an oxidation catalyst should thus be carefully adjusted to an optimum depending on the particular catalytic application to which it is dedicated. For the preparation of selective oxidation supported catalysts using MoO3 as the active phase, it is thus crucial to dispose of synthesis methods that allow efficient tuning of the morphology of MoO3 crystals at the surface of the support. The methods proposed in the literature to do so, e.g. spin-coating [9], thermal evaporation [10], chemical vapor deposition [11], flash evaporation [12], laser deposition [13] and r.f. reactive sputtering [14], are rather scarce and complex. Moreover, they are often more dedicated to the deposition of active phase on flat and/or monolithic supports (to produce model catalysts for surface science purposes) than on powder supports. These methods thus usually only allow the production of samples at a small scale, so that they are often inadequate for the production of pulverulent real catalysts in large amounts. In this contribution we demonstrate that, using an easy wet impregnation method that is suitable for deposition of active phase on powder supports, the morphology of the active phase crystallites of silica-supported MoO3 catalysts can be efficiently adjusted by simply varying the nature and the amount of the precursor used in the synthesis. 2. E X P E R I M E N T A L 2.1. Preparation of the samples The support used for the preparation of the samples was a commercial mesoporous silica (Grace 532). Prior to the impregnation, the silica was activated overnight in air at 773 K. After activation, the surface area was 314 m 2 g 1 and the pore volume was 1.75 cm 3 g-1 with an average pore diameter of 112.9 A. Four different precursors in aqueous solutions were used to impregnate MoO3 at the surface of the silica: 1~ commercial ammonium heptamolybdate (Vel, p.a.), hereinafter noted HEPTA, dissolved as such in distilled water, 2 ~ peroxopolymolybdate, hereinafter noted PEROXO, prepared by reacting a metallic Mo powder (Aldrich, 99.95%) in H202 (Aldrich, 30% wt, diluted to 10% wt in distilled water) following the procedure described in [9], 3 ~ citrate complex of Mo 6§ hereinafter noted CIT, prepared from HEPTA in aqueous solution and an equivalent amount of citric acid (Merck, 99.5%),
611 4 ~ oxalate complex of Mo 6+, hereinafter noted OXA, prepared from HEPTA in solution and an equivalent amount of oxalic acid (Aldrich, 98%). For all the samples, the impregnation procedure was carried out by dispersing the silica in the precursor solution. The solvent was then removed under vacuum from the suspensions maintained at 303 K in a rotavapor. The obtained solids were dried at low temperature. A precalcination at 413 K was applied in the case of the samples prepared from the PEROXO precursor. A precalcination at 573 K was applied for the samples prepared using the CIT and OXA precursors. The solids were finally calcined in air under conditions identical for all the samples. The concentrations and the volumes of the precursor solutions, and the masses of support involved in the impregnation, were adjusted to synthesize samples with loadings of MoO3 on the silica, between 0 and 2 theoretical monolayers (hereinafter abbreviated ML). Calculations of the quantities used for the syntheses were based on the assumption that a [MOO6]6" octahedron occupies 0.17 nm 2 on the support [ 15]. Table 1 summarizes the conditions for the syntheses of the samples and the loadings of MoO3 prepared starting from the four different precursors. Table 1 Experimental conditions for the syntheses of the silica-supported MoO3 catalysts and loadings of MoO3 prepared expressed in number of theoretical monolayers (ML) Precursors HEPTA PEROXO CIT OXA Drying air/22 h/393 K air/22 h/393 K vacuo/16 h/353K air/20 h/393 K air/5 h/413K air/16 h/573 K air/20 h/573K Precalcination Calcination air/90 min/723 K air/90 min/723 K air/90 min/723 K air/90 min/723 K v
Loading (ML)
0, 0.1, 0.25, 0.375, 0.5, 0.75, 1, 1.5, 2
0, 0.25, 0.375, 0.5, 1, 2
2.2. Characterization N2 physisorption was used to evaluate the specific areas of the samples, the volumes of their pores and their mean pore diameters. Specific area measurements were performed with a Micromeritics Flowsorb II 2300 equipment. During the whole analyses, the samples, analyzed as batches of 0.1 g, were flowed with a gas mixture (hereafter termed reference gas) containing 30% vol of N2 in He. Samples were first degassed at 433 K during 90 minutes. Adsorption of N2 was then performed by flowing the reference gas on the samples maintained at 77 K. When the adsorption equilibrium was reached, desorption was finally performed by rising the temperature of the samples back to room temperature. The quantities of N2 having physisorbed at the surface of the samples were measured with a catharometer integrating the defaults of N2 and the excesses of N2 in the gases recovered at the cell outlet during the successive steps of the analyses with respect to the reference gas. Calculation of the specific area was then made considering that the occupation of a uniform monolayer formed by 1 ml of gaseous N2 is 2.84 m 2. Pore volumes and mean pore diameters of the samples were measured with a Micromeritics ASAP 2010 equipment. Prior to the analyses, the samples were degassed at 433 K down to a pressure of 0.1 Pa. Analyses were then based on the determination of the hysteresis between the adsorption isotherm of N2 at 77 K and the corresponding desorption isotherm for partial pressures of N2 between 0.1 Pa and 101325 Pa.
612 Powder X-ray diffraction (XRD) was performed with a Kristalloflex Siemens D5000 diffractometer using the K radiation of Cu ( =1.5418 /~) and equipped with a secondary curved graphite monochromator. The analyses were done in the continuous coupled /2 reflection mode. Two- angles were scanned between 5 ~ and 70 ~ at a rate of 0.45 ~ min1. Scanning electron microscopy (SEM) was done with a $4000 microscope from Hitachi using a field emission gun and interfaced with an energy dispersive X-ray spectrometer (EDX) device. Samples were analyzed without any coating. Observations were made with an accelerating voltage of 5 kV. 3. CHARACTERIZATION RESULTS AND DISCUSSION 3.1. N2 physisorption Existence of two steps in the deposition of MoO3 on the silica Fig. 1 shows the evolution of the specific area of the samples as a function of the amount of MoO3 impregnated at the surface of the silica. Very similar tendencies are observed regardless of the precursor used for the impregnation. As expected, the specific area of the samples decreases with the amount of MoO3 deposited on the support increasing. However, a break in the decrease is observed corresponding to an amount of MoO3 impregnated of about 0.75 ML. For the low loading domain (less than 0.75 ML of MoO3 applied), a fast decrease of the specific area is observed. But for the high loading domain (more than 0.75 ML of MoO3 applied), the specific area decreases more slowly with the amount of MoO3. This observation suggests that the deposition of MoO3 at the surface of the silica proceeds in two distinct steps depending on the amount of material impregnated.
+ PEROXO 9 HEPTA I--10XA • CIT
~t~ 300
o250
O ,,
o,=~ o tD
200
150
;t ,
0
,
,
|
I
0,5
,
,
!
.
.
.
.
i
,
,
,
,
1
1,5 2 MoO 3 loading (ML) Fig. 1. Evolution of the specific areas of the silica-supported MoO3 catalysts as a function of the MoO3 loading and the precursor used in the synthesis.
Further on the mechanism of deposition of MoO3 on the silica A plausible hypothesis to account for this interpretation could be that, at low loadings, MoO3 preferentially deposits inside the pores of the support, while at high loadings, it also
613 deposits in significant amounts at the outer surface of the silica. However, this hypothesis must be discarded considering pore volume and mean pore size data in the case of the samples with low MoO3 loadings prepared from PEROXO as precursor (Table 2). Table 2 Pore volumes and mean pore diameters for the samples prepared using PEROXO as precursor MoO3 loadings 0 MC 0.1 MC 0.25 MC 0.375 MC 0.5 MC Pore volume (cm3 g-l) 1.75 i.68 1.60 1.53 1.46 Mean pore diameter (A) 112.9 112.0 113.4 112.0 113.7 The volume developed by the pores of the silica decreases linearly with the amount of MoO3 deposited on the support. But the mean pore diameter remains unchanged independently of the number of MoO3 monolayers applied on the silica. If the MoO3 material had been deposited inside the pores, one should have observed, regardless of the mechanism by which the pores would get filled, a variation of the mean pore diameter with the quantity of MoO3 impregnated. Precisely, a decrease of the mean pore diameter would suggest that, for increasing amounts of MoO3 impregnated, the pores would get progressively filled, all together and independently of their sizes, through the coating of their inner walls by progressively thicker layers of MoO3 material. On the contrary, an increase of the mean pore diameter would indicate that, for increasing amounts of MoO3 impregnated, the pores would get entirely filled sequentially by increasing order of size. Namely, the narrowest pores would first be entirely filled, then the just slightly wider pores would be filled, and so on until no empty pores remain. None of these hypothetical mechanisms match our observation that the mean pore diameter does not vary with the amount of MoO3 impregnated. Therefore, the most plausible conclusion is that, using PEROXO precursor (and presumably the others also) MoO3 does not penetrate inside the pores of the support, but only deposits at the outer surface of the silica particles, thus covering the mouth of the pores and blocking them, independently of their size.
3.2. X-ray diffraction Existence of two steps in the deposition of MoO3 on the silica Fig. 2 shows the X-ray diffraction patterns of the silica-supported MoO3 samples impregnated with CIT as precursor. Attention is directed to the influence of the amount of MoO3 deposited on the silica. When low loadings of MoO3 are applied, no diffraction peaks are detected and only the signal typical of the fresh silica is observed. When 0.375 ML of MoO3 is deposited on the support, tiny diffraction peaks appear at 20 angles typical of molybdite [16]. For higher loadings, intensities of molybdite peaks then increase with increase of the amount of MoO3 impregnated on the silica. These results indicate that for loadings lower than 0.375 ML, MoO3 remains at the surface of the silica as an amorphous material. Crystallites of molybdite only appear when at least 0.375 ML of MoO3 is applied. The number and/or the size of the crystallites increase when the amount of MoO3 impregnated increases. This interpretation of the X-ray diffraction data clearly confirms the existence of two distinct steps in the deposition of MoO3 on the silica depending on the amount of material impregnated.
614
.q,,~ r.~
=o
d 9
b
9
~
15
25
35
a
45 55 65 2 Theta Angles (o)
Fig. 2. XRD patterns for MoO3 supported on silica prepared with the Mo citrate precursor: a) 0 ML, b) 0.25 ML, c) 0.375 ML, d) 0.5 ML, e) 1 ML and f) 2 ML.
5
15
25
35
45 55 65 2 Theta Angles (o)
Fig. 3. XRD patterns for MoO3 (1 ML) supported on silica prepared with as precursor: a) the peroxopolymolybdate, b) the Mo oxalate, c) the ammonium heptamolybdate and d) the Mo citrate.
This interpretation is fully consistent with the understanding of the system from N2 physisorption data. More precisely, the deposition of amorphous material most likely leads to an efficient blocking of the pores of the support. This would match closely the rapid decrease of specific area observed for increasing amounts of MoO3 impregnated in the low loading domain. Respectively, the slower decrease of specific area observed for increasing amounts of MoO3 impregnated in the high loading domain most likely matches the formation of MoO3 crystals which is likely to not be so efficient in pore blocking.
Dependence of MoO3 deposition on the nature of the precursor The tendencies in the X-ray diffraction patterns observed when impregnating MoO3 starting from PEROXO, OXA or HEPTA precursors are almost identical to those, previously commented, observed when using CIT precursor. However, two differences of behavior must be quoted when comparing samples prepared from different precursors: 1~ the minimum loading of MoO3 for which the diffraction peaks of molybdite are detected slightly depends on the precursor used. When using PEROXO, CIT and OXA precursors, molybdite crystals are formed from 0.375 ML. But, when using HEPTA precursor, the crystals only appear when at least 0.5 ML of MoO3 is deposited.
615 2 ~ the relative intensities of the molybdite diffraction peaks, and thus the morphology of the MoO3 crystals when present, depend on the Mo precursor used for the impregnation. Fig. 3 illustrates the influence of the nature of the Mo precursor used for the impregnation on the morphology of the MoO3 crystals obtained. When CIT precursor is used (except for the sample with 2 ML of MOO3), peaks of the (0k0) series of molybdite are the most intense. This suggests that the corresponding MoO3 crystals possess a morphology with basal faces larger than the lateral faces. On the contrary, when performing the impregnation of the support with PEROXO and OXA precursors, the (110) and the (021) peaks are the most intense, suggesting that the corresponding MoO3 crystals possess a more isotropic morphology with lateral and basal faces exposed in a similar extent. Following the same criterion, the MoO3 crystals obtained when impregnating the silica with HEPTA solutions are suggested to possess an intermediate morphology.
3.3. Scanning electron microscopy Fig. 4 shows scanning electron micrographs of the fresh silica support and of some silica-supported MoO3 catalysts prepared using PEROXO and HEPTA precursors. As can be seen on the micrograph of the sample containing 0.1 ML of MoO3 prepared using PEROXO precursor (Fig. 4 - top right), when low loadings of MoO3 are deposited on the silica, no visual distinction can be made between the fresh silica support (Fig. 4 - top left) and the impregnated samples. For the latter, however, EDX reveals that silica particles are homogeneously covered by Mo-containing material. As a counterpart, as illustrated by the micrographs of the samples containing 1.5 ML of MoO3 prepared using PEROXO and HEPTA precursors (Fig. 4 - bottom), when high loadings of MoO3 are deposited, crystals are observed at the surface of silica particles. The crystals are confirmed by EDX to be made of Mocontaining material. Two differences appear when comparing the formation of crystals as a function of the precursor used for the synthesis of the silica-supported MoO3 samples: 1~ for the syntheses using PEROXO, OXA and CIT precursors, the crystals are observed for the samples containing at least 0.375 ML of MOO3, while when HEPTA precursor is used, crystals only appear when 0.5 ML of MoO3 or more is impregnated on the silica. 2 ~ the morphology of the crystals depends on the precursor used for the impregnation. Crystals observed for the samples prepared using CIT and HEPTA precursors are very similar. Their average thickness is about 100 nm and their basal faces have approximate dimensions ranging from 1.5 gm up to 3 gm (and sometimes more) by about 1 gm (Fig. 4 - bottom left). As a counterpart, crystals typical of the samples prepared using OXA and PEROXO precursors are much more isotropic as their thickness is about 200 nm and the dimensions of the basal faces are about 1 [am by 750 nm (Fig. 4 - bottom right). These observations are consistent with the conclusions drawn from the interpretation of Xray diffraction patterns.
616
Fig. 4. SEM micrographs of the fresh silica support (top-left) and of the silica-supported MoO3 catalysts prepared using PEROXO precursor (top-right: 0.1 ML, bottom-left: 1.5 ML) and HEPTA precursor (bottom-right: 1.5 ML). For the top micrographs, bar lengths correspond to 2 lam; for the bottom micrographs, bar lengths correspond to 5 lam. 4. CONCLUSION All characterization data are fully consistent. They lead to an unambiguous understanding of the deposition of MoO3 impregnated on the silica. Namely, 1o two modes of deposition for MoO3 at the surface of the silica exist: at low loadings, MoO3 deposits as amorphous material and homogeneously covers the support; at high loadings, crystals are formed; 2 ~ the amount of impregnated MoO3 required to induce the formation of crystals depends on the precursor used: for HEPTA precursor, crystals appear for a quantity of deposited MoO3 larger than for the other precursors; 3~ the nature of the precursor used for the impregnation influences the morphology of the crystals: OXA and PEROXO precursors lead to more isotropic crystals than those obtained for the samples prepared from CIT and HEPTA solutions.
617 Changing the nature and the amount of precursor used to impregnate MoO3 on the silica thus offers a promising possibility to adjust the morphology of the crystals obtained at the surface of the support. In addition, the dispersion of MoO3, precisely obtaining a crystalline or an amorphous material for a given amount of active phase impregnated, can also be mastered to some extent by adjusting the same two parameters. ACKNOWLEDGMENT
ESEM analyses were carried out at the Fritz-Haber-Institut der Max Planck Gesellschatt (Berlin, Germany). The FNRS (Belgium) and the "Communaut6 Fran~aise de Belgique" are gratefully acknowledged for financial support for the acquisition of XPS and XRD equipments. REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
R.A. Hemandez and U.S. Ozkan, Ind. Eng. Chem. Res., 29 (1990) 1454. K. Briickman, R. Grabowski, J. Haber, A. Mazurkiewicz, J. Sloczynski and T. Wiltowski, J. Catal., 104 (1987) 71. J.C. Volta and J.M. Tatibouet, J. Catal., 93 (1985) 467. L.T. Weng and B. Delmon, Appl. Catal. A: General, 81 (1992) 141. E.M. Gaigneaux, H.M. Abdel Dayem, E. Godard and P. Ruiz, Appl. Catal. A: General, 202 (2000) 265. E.M. Gaigneaux, P. Ruiz and B. Delmon, Stud. Surf. Sci. Catal., 112 (1997) 179. E.M. Gaigneaux, P. Ruiz and B. Delmon, Catal. Today, 32 (1996) 37. E.M. Gaigneaux, P. Ruiz, E.E. Wolf and B. Delmon, Appl. Surf. Sci., 121/122 (1997) 552. E.M. Gaigneaux, K. Fukui and Y. Iwasawa, Thin Solid Films, 374 (2000) 49. N. Miyata, T. Suzuki and R. Ohyama, Thin Solid Films, 281-282 (1996) 218. A. Abdellaoui, G. Leveque, A. Donnadieu, A. Bath and B. Bouchiki, Thin Solid Films 304 (1997) 39. C. Julien, A. Khelfa, O.M. Hussain and G.A. Nazri, J. Cryst. Gr., 156 (1995) 235. Zs. Geretovszky and T. Sz6renyi, Appl. Surf. Sci., 109-110 (1997) 467. A. Gorenstein, J. Scarminio and A. Louren~o, Solid State Ionics, 86-88 (1996) 977. J. Sonnemans and P. Mars, J. Catal., 31 (1978) 209. Powder Diffraction File, Joint Committee on Powder Diffraction Standards, International Center for Diffi'action Data (JCPDS-ICDD), 1996, Card 05-0508.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
619
Single step synthesis of metal catalysts supported on porous carbon with controlled texture N. Job, F. Ferauche, R. Pirard and J.-P. Pirard Universit~ de Liege, Laboratoire de G6nie Chimique, Institut de Chimie (B~t. B6a), B-4000 Liege, Belgium This paper presents an original preparation method of metal catalysts supported on carbon porous material [1]. Carbon porous materials can be synthesised by evaporative drying and pyrolysis of aqueous resorcinol-formaldehyde gels whose operating variables are correctly chosen. The porous texture of these materials is mainly controlled by the precursor solution pH. Texture analysis shows that it is possible to tailor the morphology of these materials : indeed, micro-macroporous, micro-mesoporous, microporous or non porous materials can be obtained by varying the pH value in a narrow pH interval. This texture control is slightly influenced by the introduction of a metal in the gel : the limits of the pH interval can slightly differ when a metal salt is added to the resorcinol-formaldehyde aqueous solution. Carbon supported nickel and palladium catalysts prepared by this method have proven to be active for ethylene hydrogenation. 1. I N T R O D U C T I O N Porous carbon materials prepared by sol-gel process from hydroxylated benzene and aldehyde in a solvent have been extensively studied for the past ten years [2-10] and used in electrochemical applications [11-12]. This type of material is generally obtained from resorcinol-formaldehyde gels synthesised in water in presence of a catalyst (sodium carbonate). After solvent exchange (water to acetone), these gels are dried by CO2 hypercritical drying process in order to avoid material shrinkage due to capillary forces observed when the gels are dried by simple evaporation. The dried gel is finally pyrolysed in order to obtain pure carbon (containing small amounts of sodium). Under appropriate synthesis conditions mainly defined by the Resorcinol/Catalyst molar ratio (R/C), these materials (called 'carbon aerogels') exhibit high specific surface areas and high pore volumes. The procedure can be simplified if acetone is used as synthesis solvent [13] (in this case, the reaction between resorcinol and formaldehyde is catalysed by perchloric acid). This method eliminates the time-consuming solvent exchange step. Nevertheless, this process remains too expensive and difficult to apply at an industrial level. Researches tend to develop other drying processes. The evaporative drying process, which is cheaper and easier to use than hypercritical drying, generally leads to densified non porous materials. Large capillary forces at the liquid-vapour interface cause the gel to shrink and crack. Attempts to overcome this
620 limitation led to the use of solvent exchange technique before evaporative drying : water can be exchanged with acetone, and acetone reexchanged with cyclohexane, which can be removed by simple evaporation [9, 10, 14]. In this case, capillary forces are less severe, so the dried gel remains porous, but this technique is long, complicated and uses large toxic solvent volumes. Freeze drying was also envisaged [15-16], but freezing induces a high volume expansion of the solvent when water is used and the samples never remain monolithic. A porous material can nevertheless be obtained. The present study shows that it is possible to synthesise still largely porous materials by evaporative drying (vacuum drying) of aqueous resorcinol-formaldehyde gels when the gel synthesis conditions are appropriate. This paper presents a synthesis method using an evaporative drying process that leads to hyperporous carbon materials. Synthesis pH is the key operating variable which controls the texture of the carbon material obtained after drying and pyrolysis of resorcinol-formaldehyde aqueous gels. Both pure carbon materials and metal-containing carbon materials can be prepared. The material texture is still controllable when a metal is incorporated to the gel by dissolution of a salt in the precursor solution in order to synthesise carbon supported metal catalysts. To our knowledge, previous attempts to incorporate metals in carbon aerogels (prepared by dissolution of a metallic salt in the resorcinol-formaldehyde solution and hypercritical drying) have failed to control the material texture [17-18]. Carbon supported catalysts could be used in various chemical reactions, and more specifically as fuel cell electrodes, the carbon support being a good electrical conductor. 2. E X P E R I M E N T A L
2.1. Synthesis Organic gels were prepared by polycondensation of resorcinol with formaldehyde in water as a solvent. Three series of carbon samples were prepared : one without addition of metal ; the two others containing nickel and palladium (about 1% weight) respectively. The incorporation of metal was achieved by dissolution of a metal salt: nickel acetate (tetrahydrate) and palladium acetate were used. As resorcinol plays the role of nickel complexant, nickel is easily soluble in resorcinol-formaldehyde aqueous solutions. On the contrary, significant amounts of palladium cannot be dissolved without using an additional complexant. Diethylenetrinitrilopentaacetic acid (DTPA) was then added to the solution. No sodium carbonate as polymerisation catalyst was used. In this work, the pH was adjusted to a chosen value by the use of sodium hydroxide (aqueous solution) and measured by a pH-meter. The synthesis variables are : (i) the Resorcinol/Formaldehyde molar ratio : R/F ; (ii) the Dilution ratio : D = Total Water/Reactants molar ratio. Note that 'Total Water' includes both added deionized water and water contained in the formaldehyde solution. Reactants refers to resorcinol, formaldehyde, metal salt and complexant.
621 (iii)
the theoretical Carbon/Metal molar ratio. This was calculated assuming that no carbon loss occurs. The real metal weight percentage was re-evaluated after pyrolysis. (iv) the Complexant/Metal molar ratio : DTPA/M; (v) the pH of the resorcinol-formaldehyde solution. In the case of metal incorporation, the Carbon/Metal molar ratio was chosen in order to theoretically obtain 1% weight metal samples (without carbon loss during drying and pyrolysis). The complexant/metal molar ratio (DTPA/M) was kept at 1 for palladiumDTPA samples. The obtained aqueous gels were dried by vacuum evaporation. The unsealed flasks were kept at 60~ and the pressure was progressively reduced from 105 Pa to 103 Pa. This step was performed over 5 days. The samples were then heated to 150~ (103 Pa) for three days. After drying, the gels are generally monolithic despite a significant shrinkage (40% vol. for low pH to 60% vol. for high pH). After drying, the gels were pyrolyzed at 800~ under inert gas (N2). The pyrolyzed materials are black, matt (for low pH) or bright (high pH). The samples generally stay monolithic, but pyrolysis also induces shrinkage. The total shrinkage (50% vol. for high pH to 80% vol. for low pH), which must be taken into account for the manufacture of monolithic parts, can be evaluated prior to synthesis. These materials possess a very good mechanical strength.
2.2. Sample characterisation and catalytic tests The porous texture of the dried gels and the pyrolyzed gels was characterised by the analysis of nitrogen adsorption-desorption isotherms, performed at 77 K. The analysis of the isotherms was performed according to the methodology proposed by Lecloux [19]. Samples bulk density was obtained by mercury pycnometry. Infrared and X-ray spectra analysis allowed to obtain data about the elementary composition of the samples and the aggregation state of the metals. Nickel and palladium catalytic activities were tested on ethylene hydrogenation into ethane in an isothermal differential reactor. 2.3. Samples nomenclature The three series of samples are denominated as follows : i) C1-C6 : pure carbon samples; ii) Nil-Ni5 : nickel loaded samples; iii) Pdl-Pd5 : palladium loaded samples. As the index increases (i.e., from C1 to C6), the pH of the resorcinol-formaldehyde solution increases. A 'p' is added for pyrolyzed samples (i.e. Clp). 3. RESULTS AND DISCUSSION
3.1. Texture analysis Fig. 1 shows the nitrogen adsorption-desorption isotherms of pure carbon samples after drying (series C1-C6). As pH increases, the samples evolve from micro-macroporous material to exclusively microporous material. Lower pH than the inferior limit leads to
622 losing the mechanical strength of the material. Intermediate samples contain both microand mesopores. The hysteresis becomes smaller and moves to lower P/p0 values as pH increases. This shows that a pH increase of the precursors solution leads to the decrease of the broader pores size and porous volume. The sharp increase of the adsorbed gas volume at low relative pressure is quite the same for every sample (except C1) : the micropore volume must then be constant. These isotherms show that the dried gel texture can easily be defined by pH control.
800 --~- C l
700
-a9 C2 --~- C3
600
---e- C4
"~ 500
--~- C5
E
-x-C6
,9, ~ 400 I/}
300 200 100 T
0
i
i
i
i
i
0.2
0.4
0.6
0.8
1
p/po
Fig. 1. Nitrogen adsorption-desorption isotherms after drying 9series C1-C6. After pyrolysis, the material exclusively contains carbon (and metal in the case of Niand Pd-gels) and traces of sodium. This was shown by infrared spectrum measurements : no C-H, O-H or C-O bonds could be detected. X-ray spectra showed that nickel and palladium are in metallic state after pyrolysis. Fig. 2. shows isotherms relative to the same series of samples after pyrolysis (Clp-C6p). If the pH is not too high, the effect of pyrolysis is to increase the microporous volume (lengthening of the sharp increase of the adsorbed gas volume at low relative pressure) and to slightly decrease the total adsorbed gas volume and the size of the pores (hysteresis smaller and shifted to lower values of p/p0). On the contrary, for higher pH (C5p and C6p), the adsorbed gas volume roughly drops, and the material becomes non porous. Similar results are found for metal loaded samples. In each case (with or without metal), it is possible to find a pH interval that leads to micro-mesoporous materials after drying and pyrolysis. This pH interval is always very narrow. If the pH is lower than the inferior
623 limit, the obtained material is micro-macroporous and tends to lose its mechanical strength: the sample becomes friable and brittle. The pore size is probably still controllable below this pH limit, but as our goal was to obtain monoliths, this pH range was not investigated. Nevertheless, such materials can be interesting if the mechanical strength of the material is not a key factor. If the pH is greater than the superior limit, the sample is exclusively microporous after drying and becomes quasi non-porous after pyrolysis. Between these two extremes, the sample texture can be adjusted as desired. The position and width of this pH interval are nevertheless slightly influenced by the introduction of metal and complexant.
1200 1000
---~Clp .-a-- C2p --..~ C3p ---~- C4p --x9 C5p
800
E
600 400 200
1
0
0.2
0.4
0.6
0.8
1
p/po
Fig. 2. Nitrogen adsorption-desorption isotherms after pyrolysis 9series Clp-CSp. Tables la and lb show the quantitative results of the nitrogen adsorption-desorption isotherm analysis, i.e., BET specific surface area (SBET), pore volume (Vp) calculated from the adsorbed volume at saturation, and total micropore volume (VDuB) calculated by the Dubinin-Radushkevich equation [19]. Bulk density of each sample is also displayed. Before pyrolysis, specific surface areas (SBET) vary from 280 to 510 ma/g, and pass through a maximum when the pH increases. The microporous volume slightly increases with the pH and seems to reach a step (0.20 to 0.25 cm3/g). The total porous volume (Vp) decreases when pH increases. This pore volume varies from 0.25 to more than 1.30 cm3/g. This last value is very high for materials dried by simple evaporation. Note that the total porous volume of samples containing macropores (i.e., C1, Nil, Pdl and Pd2) is underestimated by N2 adsorption-desorption isotherm analysis because of the technique
624 limitations (pore width < 50nm). The total porous volume, macropores included, probably continues to increase when pH decreases. Concurrently, bulk density increases when pH increases, and varies from 0.49 to 1.01 g/cm 3. Table 1a Texture after drying
Table lb Texture after pyrolysis Sample SBET VDUB v o Sample SBET VDUB (cmVo3/g)dbulk dbulk (mE/g) (cm3/g) (g/cm3) (mZ/g) (cm3/g) (cm3/g) (g/cm 3) C1 330 0.15 1.06" 0.49 Clp 625 0.27 1.44 0.53 C2 435 0.19 0.95 0.63 C2p 635 0.27 0.92 0.68 C3 475 0.22 0.79 0.75 C3p 610 0.27 0.66 0.98 C4 505 0.23 0.59 0.85 C4p 565 0.24 0.41 1.10 C5 510 0.24 0.40 1.01 C5p 99% purity purchased by "Maxima LTD") were fed from cylinders. Brooks mass-flow meters controlled the gas flow rate. Liquid was condensed in a trap at 3~ and gases flowed to GC for analysis. The analysis of volatile reaction products in butane oxidation was performed on line with GC HP-5890 that contained four columns: 45/60 molecular sieve 13X, 10 ft x 1/8"; A1203,50 m x 0.53 mm; 80/100 Hysep Q, 4 ft x 1/8" and 1 ft x 1/8" with internal switching valves and two detectors TCD and FID controlled by ChemStation analytical software. Analysis of water collected in the condenser displayed only traces of oxygenates. Blank experiment without a catalyst indicated conversion less than 1% at 550~ [5,8]. HPLC analysis was used for the analysis of liquid products of toluene oxidation. Toluene, benzaldehyde and bezoic acid were analyzed on HP-1050, column C-18, Zorbax, feed of eluent containing 60 and 40% of acetonitrile and water was 1 ml/min, ~, = 254 nm. Two supports, u (Norton 6175) and SiO2 (PQ 1030) were selected for catalyst preparation. Water-soluble sodium salt of (4,5-carboxy) cobalt phthalocyanine (NaCoPc) was kindly prepared by Prof. E. Lukyanetz (Organic Intermediates and Dyes Institute, Moscow). Catalysts were prepared by impregnation of the supports in a rotavapor with an aqueous solution containing NaCoPc, 4 g/L at pH 8.3 (NaOH, NH4OH). The catalysts were dried for 3 hr at 100~ and calcined at 700~ for 5 h in the stream of He. Standard cobalt oxide catalyst was prepared by impregnation of y-A1203 with an aqueous solution of cobalt nitrate hexahydrate (Aldrich) followed by drying at 100~ 3 h and calcination at 550~ 6 h. V-Mg-O catalyst was prepared according to the procedure reported elsewhere [ 17]. Analysis of the catalysts is given in Table 1. SEM, XRD, XPS and BET methods were used for catalyst characterization [8].
681 Table 1 L Catalyst characterization Catalyst [ Support Co no.
Content, wt % N Na V
1 dried 1 calcined 1 tested 2 dried
A1203
04s
08
A1203
2tested
A1203
(.4 (.4 (.8 (.9 (.7
06 06 12 11
(.7 2.3
05
4
SiO2 SiO2 Mg M~O *) 3-6- calcined
5.6 0.6 0.5 10. 0.8
Surface area m2/g
Pore volume cc/g
Average D pore nm
199
0.63
13.0
191 190 343 330 45
0.69 0.65 0.46 0.45 -
14.5 13.5 11.5 10.7 -
3. RESULTS AND DISCUSSION
Catalysts 1 and 2, that are 0.4 and 0.8 % CoNx/y-ml203,indicate similarly high performances in oxidative dehydrogenation of n-butane. Even at 400~ the conversion exceeded 40 % (Fig. 1) and yield of olefins reached 25 %. The special feature of this catalyst consists of a high conversion of n-butane yielding mostly light olefins, ethylene and propylene. 90 wt% of olefins formed at 400~ and molar ratio O2/n-butane of 1.5, were C2-C3 olefins;53 wt.% was ethylene and the rest propylene. Therefore, the selectivity to ethylene on the novel catalysts was much higher than over traditional Co-O/A1203 (Fig. 2). Experimental data and comparison of the performance and products selectivity of all the catalysts studied are given in Table 2. Although the highest total olefins selectivity and highest selectivity toward butylenes and butadiene was obtained over the traditional vanadia-magnesia catalysts no. 6, conversion of n-butane and yield of light olefins were significantly higher over samples 1 and 2. It is interesting that total olefin selectivity with promoted samples 1 and 2 displayed little change with an increase of temperature, and was close to 58% at 400-600~ (Fig. 1) while the selectivity on traditional systems is usually high at low temperature and low at high temperatures. Increasing of temperature in the range 400-600~ (Figs. 1,2) significantly affected n-butane conversion which reached 82% at 600~ In spite of the comparatively low value of total selectivity at 58 wt.%, yield of olefins over sample 2 peaks at 47.6 wt.%, with yield of ethylene of 31 wt.%. The specific performance of nitrogen containing samples 1 and 2 was compared with results described in literature for oxidative dehydrogenation of n-butane. A summary of the catalytic data is given in Fig. 3. Most catalysts, based on vanadia, yielded optimal performance toward olefins in the range 520-560~ and yielded preferably butylenes. A general feature of most catalytic systems is the low selectivity to light olefins resulting in yield of ethylene and propylene OI~176 oi80_aJ~. I
c=
"i=} f~
E
I
i
I
A
(-.
U
I
CNTs AC
6040-
2o-
r9~
U
0 0
I
I
I
I
5
10
15
20
I
i
25
"--"
I
30
Time on stream (h) Fig. 3. Cinnamaldehyde hydrogenation activity on the Pd/CNTs and Pd/AC at 80 ~ and under atmospheric pressure of hydrogen. The higher degree of crystallinity, i.e. ordered stacking of graphene planes, of the carbon nanofibers compared to the activated charcoal, i.e. turbostratic carbon with no long range order, could also explain the relatively high activity observed. Similar results have been reported by Rodriguez et al. [15] for the Fe-Cu bimetallic catalyst supported on carbon nanofibres and activated carbon during the hydrogenation of unsaturated hydrocarbon. The influence of the catalyst particle location with respect to the support on the catalytic performance should be studied in more detail in order to obtain a better understanding of the mechanism.
702 Fig. 4 shows the reaction pathway involved in the cinnamaldehyde hydrogenation, leading to compounds containing hydrogenated C=C and/or C=O bonds. Cinnamaldehyde
Hydrocinnamaldehyde (3-Phenylpropionaldelyde)
_Cr o
~
OH
Cinnamyl alcohol
3-Phenyl propanol (3-Phenylpropyl alcohol)
Fig. 4. Reaction pathway of the cinnamaldehyde hydrogenation.
The CNTs based catalyst also displays an oustanding performance when compared to the commercial activated charcoal catalyst in terms of product selectivity. A totally different selectivity was observed between the two catalysts: on the CNT-based catalyst, the ratio of the C=C bond hydrogenation product versus the complete hydrogenation product was 80:20 whereas on the commercial catalyst, the selectivity was significantly modified toward the total bond hydrogenation which leads to a ratio of only 45:55 (Fig.
5B). Such a difference in terms of product selectivity was attributed to the complete absence of any acidic sites on the carbon nanotubes surface and also to the absence of micropores which could induce re-adsorption and consecutive reaction [16]. The presence of micropores could artificially increase the contact time and as a consequence, modify the hydrogenation pathway. The influence of the support nature on the electronic properties of the metallic phase could also be put forward to explain these results. Depending on the metal-support interaction, the metal particles could exhibit different exposed faces and as a consequence, significantly modify the chemisorption of the reactant on their surface. According to the interaction between the C=C bond and the faces exposed by the palladium particles, the residence time and the desorption of the intermediate could be different and thus, lead to a different selectivity. The presence of palladium aggregates on the activated charcoal as compared to the individual palladium dispersion on the CNTs could be the illustration of this difference in exposed crystalline faces.
703
100 8O" -o i1) >..
60
~23
4o-
a.
20-
"o o
I
I
A
I
I
= 1
100
I
CNTs
C=Cbond / hydrogen/ation/
~-- 80 "~ i1) >.
t~
"Z3 o
e
0~ 0
I1
5
10
15
20
Time on stream (h)
25
30
I
60
40
I
I
I
I
I
AC
B
C=Cbond hydrogenation
201 ~ ~ Z ~ O ~n d 0~,~---I~, , , hydrogenation 0 5 10 15 20 25 30 Time on stream (h)
Fig. 5. Distribution of the reaction product during the hydrogenation of cinnamaldehyde on the Pd/CNTs and Pd/AC at 80 ~ and under atmospheric pressure of hydrogen.
4. CONCLUSION Carbon nanotubes are efficient catalyst supports for liquid phase reactions, i.e. hydrogenation of cinnamaldehyde. The palladium particles are well dispersed inside the CNTs tubules when using a classical incipient wetness impregnation followed by a mild thermal treatment and reduction. The peculiar morphology of the support and the location of the active phase allows an active and highly selective catalyst for the C=C bond hydrogenation in a,/~unsaturated compounds to be obtained, when compared to the commercial high surface area activated charcoal catalyst. Such results could open a new field of catalytic investigations for fine chemical applications.
REFERENCES 1. S. Iijima, Nature, 354 (1991) 56. 2. J-M. Bonard, L. Forro, D. Ugarte, W.A. de Herr and A. Chfitelain, Eur. Chem. Chronicle, 1 (1998) 9. 3. G. Che, B. B. Lakshmi, C.R. Martin and E. R. Fischer, Langmuir, 15 (1999) 750. 4. F. Salman, C. Park and R.T.K. Baker, Catal. Today, 53 (1999) 385. 5. R . T . K . Baker, K. Laubernds, A. Wootsch and Z. Paal, J. Catal., 193 (2000) 165. 6. C. Park and R. T. K. Baker, J. Phys. Chem. B 102, 5168 (1998). 7. A. Chambers, T. Nemes, N.M. Rodriguez and R.T.K. Baker, J. Phys. Chem. B, 102 (1998) 2251. 8. E. Dujardin, T. W. Ebbesen, H. Hiura and K. Tanigaki, Science, 265 (1994) 1850. 9. S.C. Tsang, Y.K. Chen, P.J.F. Harris and M.L.H. Green, Nature, 372 (1994) 159; P.M. Ayajan, O. Stephan, P. Redlich and C. Colliex, Nature, 375 (1995) 564.
704 10. J. Sloan, J. Hammer, M. Zwiefka-Sibley and M.L.H. Green, Chem. Commun., 347 (1998). 11. J. Sloan, D. M. Wright, H. G. Woo, S. Bailey, G. Brown, A.P.E. York, K.S. Coleman, J.L. Hutchison and M.L.H. Green, Chem. Commun., 699 (1999). 12. C. Pham-Huu, N. Keller, G. Ehret and M.J. Ledoux, J. Catal., 200 (2001) 400. 13. Y. Gogotsi, J.A. Libera, A.G. Yazicioglu and C.M. Megaridis, Mater. Res. Soc. Symp. Proc., Vol. 633 ~ Nanotubes and Related Materials >>A.M. Rao, Ed., p. A.7.4 (2001). 14. R. Schl6gl, in: Handbook of Heterogeneous Catalysis, Eds. G. Ertl, H. Kn6zinger, J. Weitkamp, Wiley-VCH, Weinheim, Vol. 1, p. 138 (1997). 15. N.M. Rodriguez, M.S. Kim and R.T.K. Baker, J. Phys. Chem., 98 (1994) 13108. 16. L. Zhang, J.M. Winterbottom, A.P. Boyes and S. Raymahasang, J. Chem. Technol. Biotechnol., 72 (1998) 264.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
705
Raman studies of the templated synthesis of zeolites P.P.H.J.M. Knops-Gerrits* and M.G.L.J. Cuypers D6partemem de Chimie, Universit6 Catholique de Louvain (UCL), Batiment Lavoisier, Place L. Pasteur n~ B-1348 Louvain-la-Neuve, Belgium T61 : (32)10 - 47 29 39 Fax : (32)10 - 47 23 30 E-mail :
[email protected] * To whom correspondence should be addressed. The application of confocal Raman and FT-Raman in the understanding of zeolite synthesis is a goal that can be reached by studying different levels of complexity. First, the Raman properties of TEOS and its polymerization products under acid conditions is investigated. Structural characterization of zeolite framework vibrations is related to their building blocks such as prisms and sodalite cages. The Raman study of zeolite synthesis with structure directing agents permits the study of the geometry of these organic molecules, such as in ZSM-5 synthesis, the tetra-propyl-ammonium ion (TPA+), in EMT synthesis the Na-18-Crown-6 that can direct hypo- (and hyper-)cage formation, Can Li and co-workers performed zeolite X synthesis studies and FAU (X,Y), MFI, MOR and BETA characterization with UV-Raman. Key-words vibrations
9 Raman spectroscopy, TEOS, synthesis, zeolite, templates, framework
1. INTRODUCTION In comparison with this vast amount of infrared spectroscopy work, the field of zeolite Raman spectroscopy seems rather evolving [1-2]. The examination of highly characterized surfaces such as single crystals to which are sorbed sub-monolayers of reactive gases or vapors must obviously be far more likely to yield results of significance on catalytic processes [3-4]. If we wish to carry out a simple transfiectance measurement the sample thickness would be -2 angstroms i.e. -10 -5 of the conventional infrared pathlength. If one planned to try Raman the sample size is absurdly small. So the outlook looks bleak before one starts studying microporous oxide materials but in the presence of zeolites the good sorption capacities allow to study adsorbed or ship-in-the-bottle complexes quite well. Many catalysts are highly porous. Compounds such as silicaaluminas or zeolites have huge areas to which molecules can sorb. In microporous materials the holes may be of molecular size and limit access to sites nearer the surface, but the surface area of these materials can be huge indeed - values of 100 to 1000m2g 1 are not unusual. Sorbing a monolayer to these materials does provide a decent amount of
706 material and so vibrational spectroscopy started with this type of catalyst system. Pioneers such as Norman Sheppard adsorbed methane to silicas and using infrared were able to show how the molecules interacted [3-4]. In the development of zeolite science, IR has been one of the major tools for structure and reactivity characterization. The reason for this choice is that there is considerable difficulty in obtaining Raman spectra with acceptable signal-to-noise ratios from highly dispersed materials such as zeolites [56]. The Raman effect is intrinsically a weak phenomenon. In order to pass about the low sensitivity an increase of the signal can be obtained ( v 4 law) by an increase of the excitation frequency [7-9]. Raman spectra of zeolites are often obscured by a broad fluorescence. Several causes for fluorescence have been identified. (1) small amounts of aromatic, strongly luminescent molecules might be present in the samples. (2) the presence of (reduced) transition metal ions or Fe impurities in the lattice is known to cause luminescence. The latter problem can be overcome via high-purity synthesis, starting e.g. from metallic Al. (3) proton super-polarizability and basic surface OH groups are two other reasons. Dehydroxylation can only be used if the sample resists such a treatment. The advent of Fourier transform Raman (FT-Raman) spectroscopy with excitation in the near infrared (NIR) domain offers new perspectives in reducing background fluorescence. 2. EXPERIMENTAL 2.1.Materials TEOS 98% was obtained from ACROS Organics. Commercial samples of ZSM-5 (PQ Company Si/Al ~ 81, ALSI-PENTA Si/Al ~ 11), MOR (TOSOH with Si/A1 ~ 10), FER (PQ Company Si/Al ~ 10), NaY (Zeocat; Si/A1 ratio of 2.47) were obtained. Intergrowths of FAU/EMT were synthesized using crown-ethers as structure directing agents. The quantification of the FAU and EMT phases is done by independent calibration with pure 1,4,7,10,13,16-hexaoxacyclo octadecane 18-Crown-6 containing EMT and 1,4,7,10,13-pentaoxacyclopentadecane 15-Crown-5 containing FAU. In EMT only Na-18-Crown-6 can direct hypo- (and hyper-)cage formation, whereas in a FAU phase both Na complexes of 18-Crown-6 or 15-Crown-5 can direct supercage formation. [26-27].
2.2.Spectroscopy Raman spectra were recorded on a Renishaw Raman Microscope. The zeolite samples, in powder form, were placed on a glass microscope slide. The power on the sample was about 2mW/mm 2. The collection time varied from sample to sample and was between 15 and 20 minutes. The spectra were background corrected and a Fourier deconvolution procedure, described elsewhere [3], was applied to resolve the overlapping bands in the OH stretching region. FT-Raman spectra were recorded on a Bruker IFS100. The zeolite samples, in powder form, were pressed in metal sample-holders. FT-Raman spectroscopy has been used to characterise FAU/EMT intergrowths. The vibrational properties of the crown-ethers are discussed in the context of their symmetry and complexation. In the CH stretching region planar or puckered coordination around the
707
Na + cations can occur. The conformation of the cation 15-Crown-5 and 18-Crown-6 complexes in the different faujasite structures is discussed. FT-IR spectra were recorded on a Nicolet F-730 spectrometer. 3. RESULTS AND DISCUSSION 3.1. Raman studies of TEOS and the polymerization of silicon oxides. The few bands observed in the Raman spectrum of tetra-ethyl-ortho-silicate (TEOS) led to the assumption of a high symmetry for this compound. A detailed vibrational characterization of TEOS has been reported by Mondragon et al. [12]. They carried out a normal mode analysis to confirm the experimental assignments and to obtain the force field parameters. TEOS has 3n - 6 = 93 normal modes and with double or triple degenerated bands this spectrum thus becomes simpler. The vibrational spectra can be analyzed according to structural models of Dzd and $4 symmetry. The Dad structural models has 43 normal active modes in IR and 67 in Raman, with 20 polarized active Raman modes. The $4 structural model has 47 normal active modes in IR and 70 in Raman, with 23 polarized active Raman modes. If it is considered that the -OC2H5 groups rotate around the Si-O axis they describe revolutional cones, the TEOS structure could be considered as pseudo-tetrahedral Ta, in this case structural models has 17 normal active modes in IR and 34 in Raman, with 14 polarized active Raman modes. For the structural fragment -OC2H5 of C s symmetry, the symmetric and anti-symmetric stretching normal modes of the -CH3 and-CH2- groups, the group frequencies are well known in Raman and are observed at 2964, 2925 and 2881 cm -1. The 6 (O-Si-O) and 6 asym (O-Si-O) are observed at 301 and 391 cm -1. The v (Si-O) can be found at 995 and 1024 cm -1. The attribution of most (intense) Raman bands is given in Table 1. By varying the water alkoxide ratios R (0.5 < R < 2.0) the progressive hydrolysis and polycondensation of tetraethoxysilane (Fig. 2) in acidic medium was investigated (HCl 0.03M). The chemical reactions of hydrolysis and polycondensation are the following : Si-O-CHe-CH3 + He0 "-) Si-OH + CH3CHzOH Si-OH + HO-Si ") Si-O-Si + HzO Si-O-CHz-CH3 + HO-Si ") Si-O-Si + CH3-CHz-OH In order to appreciate the effects of the alcohol on the reactions involved, part of the solutions can be prepared with different amounts of ethanol as has been presented by Dhamelincourt and co-workers [13]. The Raman band characteristic for the monomer TEOS appearing at 650 cm -1 has been assigned to the symmetric SiO4 stretch. In the scheme of the formation mechanism of polymeric silicate species at 40, 60 and 80~ with R - 1 different intermediates can be observed, which are shown in Fig. 3 and Table 2. It can be noted that at 80~ the bands of ethanol are largely absent as ethanol is evaporated at this temperature. At 40~ mainly monomers and dimers are observed, at 60~ mainly monomers, dimmers, trimers, tetramers and polymers are observed, at 80~ mainly dimers and polymers are observed.
708 Table 1. Raman data of tetra-ethyl-ortho-silicate (TEOS). wavenumber (cm- 1)
intensity
attribution
301
w
(O-Si-O)
391
w
asym (O-Si-O)
614
m
not used in the calculations
647
s
sym (Si-O4)
789
m
(Si-O + C-O)
928
m
995
vs
1024
m
1084
s
(C-C) (Si-O) (Si-O) asym (SiO-CO)
1146
w
(CH3)
1176
m
(HCH)(CH2)
1191
m
(HCH)(CH2)
1291
m
(H-C-H) and CH2 twist
1385
w
(O-C-H / C-C-H)
1449
m
asym (H-C-H)
1479
m
(O-C-H)
1597
m
(O-C-H)
2881
m
sym(CHa)
2925
rn
asym(CH2)
2969
w
, asym(CH3),
= elongation ; = in plane angular deformation; = rocking; = out of plane angular deformation; Table 2. Structural Assignmems of mono-, oligo- and polymeric species in the Raman spectra of water/TEOS mixtures with [HC1]=0.03M. Raman (cm") Structural assignments polymer 500-530 530-550 Si(O-Et)3-O-Si(0-Et)2- O-Si(O-Et)2-O-Si(O-Et)3 550- 570 Si(O-Et)3-O-Si(O-Et)2-O-Si(O-Et)3 600 Si(O-Et)2 (OH)-Si(O-Et)3 and Si(O-Et)3-Si(O-Et)3 Si(O-Et)4 et Si(O-Et)3 OH 650
709
,,,. 2500u'l
~2000-
.Q i,,,
m
r 1500-
~ '1000,=..-
r
,~ 500,.=.
O---7
. . . . . . . . . . . . . .
3OOO
i . . . . . . . . . . .
2500
q
+
2OO0
+
~
'
cm -1
-+
i
"
1500
T
1000
500
Fig. l. Raman spectrum ofTEOS (tetraethoxysilane).
.....~00*ILLJI~LIIL
JI ..J~--~ .... I J-~--J i ..i-i-
,.a
.~3000-
~2000I/%
~ 1000-
ira,
. . . . . . .
I
3500
..............
i
3000
. . . . . . . .
i
|
I
2000
t500
........
2500
cm -1
. . . .
I
..................
1000
I
-'
500
Fig. 2. Scheme of the formation mechanism of polymeric silicate species at 40~ (top), at 60~ (middle) and at 80~ (bottom) with R = 1.
710 3.2. Raman studies of zeolite synthesis. Just like IR spectroscopy, Raman can detect small, X-ray amorphous zeolite particles. Therefore Raman has been used to examine both the liquid and the solid phase of zeolite synthesis mixtures [28]. Ex situ methods (with separation of solid and liquid) and in situ methods have been applied. In studying the liquid phase [10-11], one should remember that (i) minimum concentrations for detection of spontaneous Raman from liquids are typically 0.05 - 0.1 M, [15-27, 29] (iO that the cross-section of the AI(OH)4 species is much stronger than e.g. for silicate or aluminosilicate anions [30]. Thus species which are present in low concentration or with variable structures may easily be overlooked in Raman spectra of the synthesis liquors. When zeolites are synthesized starting from a silica sol, the Raman spectrum of the initially formed solid phase resembles that of vitreous silica, with a broad band around 450-460 cm 1 [29, 31,32]. For zeolites A, X and Y, aluminate is available in solution, as shown by a Raman band at 620 cm 1 [30-34]. A1 can be rapidly incorporated into the solid phase. In particular for zeolites X and Y, this process is accompanied by the appearance in solution of monomeric silicate species, such as SiO2(OH)22 (780 cm "1) or dimeric silicates [30,32,34]. There is no Raman evidence for the presence of aluminosilicate ions in solution. Analysis of the 400-550 cm 1 spectrum during heating or eventual aging of the synthesis gel provides information on the building blocks present in the still relatively disordered solid phase. Again this analysis is based on the correlation between ring-size and ~r frequency. Thus for zeolites A and X, the band close to 500 cm -1 indicates the presence of 4MR at the beginning of the gel heating [31 ]. In zeolite Y synthesis, the gel aging causes a shift towards lower frequencies (440, 361 cml), showing the formation of 6MR. During the subsequent heating, sodalite units with 4MR are formed and the 500 cm 1 band gains intensity again [32]. Syntheses from Si sources other than colloidal silica have also been studied [34]. UV Raman spectra of the liquid phase of the synthesis of the framework of zeolite X indicate that AI(OH)4" species are incorporated into silicate species, and the polymeric silicate species are depolymerized into monomeric silicate species during the early stage of zeolite formation. An intermediate species possessing Raman bands at 307, 503,858 and 1020 cm -1 is detected during the crystallization in the solid phase transformation. The intermediate species is attributed to the 13 cage, the secondary building unit of zeolite X. A model for the formation of zeolite X is proposed, which involves four-membered tings connecting to each other via six-membered ring to form 13 cages, then the 13 cages interconnect via double six-membered tings to form the framework of zeolite X [6,8]. Mordenite and ZSM-5 are synthesized at lower pH values, and it is not surprising that in these conditions silicate species are not usually observed in solution [27,29]. In both cases, the solid initially resembles vitreous silica. In the case of ZSM-5, new vibrations in the solid phase are only observed when ZSM-5 crystals appear [27]. More information concerning the intermediate synthesis steps can be retrieved from mordenite synthesis spectra [29]. In mordenite synthesis, a 495 cm 1 band is observed at an early stage in the spectrum of the solid fraction of the gel. In analogy to what is observed for A and X zeolites, this band is ascribed to 4MR aluminosilicate units. Only at a later stage, broader bands appear at 402 and 465 cm 1, indicative of the formation of still rather
711 disordered 5MR mordenite-like units. These bands eventually sharpen into the characteristic framework vibrations of mordenite.
Bim IF
L._J 6'il
am 4
Ai,,m~N,~nc:a~
$O1' a n d 111~ ~ ' * llwia e ~ g e
2 9 E . M D a n d d~14 e m "~ ZB~Ih~ X
Fig. 3. Scheme of the formation mechanism of zeolite X (after Ref. 6 ).
3.3. Template observation via Raman. Observation of the organic template against the weak zeolite background is highly facilitated in Raman in comparison with IR. In some particular cases, the incorporation and conformation of the organic template can be followed. In ZSM-5 synthesis, the tetrapropylammonium ion (TPA § is initially incorporated into the solid amorphous aluminosilicate in the all trans configuration, which it also possesses in solution. The eventual ZSM-5 crystals however contain a high-energy TPA + conformer, in which one of the N-C bonds is rotated [27,35]. The strong non-bonded interactions between-CH2groups of different propyl groups are evidenced by the changed rocking and wagging modes. This conformation has also been proved in XRD studies. In syntheses of LTA zeolites, a tetramethylammonium ion (TMA+) can be incorporated in the sodalite cages. The tightness of this fit induces a frequency increase in the symmetric C-N stretching vibration of the template [33-36]. Raman studies indicate that at a fixed pH, organic cations (e.g. TMA +) and alkali ions stabilize different silicate species [37]. This clarifies the role ofTMA + in synthesizing LTA zeolites with high Si/A1 ratio, e.g., ZK-4. Crown-ether templates are used in the synthesis of cubic and hexagonal faujasites. The Na + forms of the cyclic ethers 18-crown-6 (18C6) and 15-crown-5 (15C5) can be used to direct the synthesis of faujasites towards the cubic FAU or the hexagonal EMT topology [26-27]. FT-Raman spectra of as-synthesized FAU, EMT and a structural intergrowth of both topologies (MIX) are shown in Table 1. The intergrowth was synthesized with a 3 91 18C6 915C5 mixture. The FT-Raman technique yields superior spectra compared to visible excitation, based on other Raman reports on this system [3940]. In the spectra, features of framework and crown ether template are superimposed. The band at 503 crn1 is the framework symmetric bending vibration; most other bands are crown-ether vibrations. There are a few differences between the spectra ofNa+-15C5
712 in FAU and Na+-I 8C6 in EMT. Both ethers have a characteristic intense band between 800 and 900 cml, which has C-O stretching and CH2 rocking character. A sharp band is observed at 1001 crn1 in FAU and to a lesser extent in the intergrowth. This band is lacking in as-synthesized pure EMT, for 18-crown-6 and its complexes, the relation between Raman spectra and the symmetry of a compound (D3d, Ci or CI) has been studied in depth. K+-18C6 always assumes D3d symmetry, and in crystalline (Na+18C6)(SCN) it has an envelope-like C~ structure. For dissolved Na+-I 8C6, the geometry may fluctuate between different conformers (D3d or CI) depending on the solvent. While Na+-18C6 undergoes some distortion in the EMT hypocage, the EMT hypercage provides sufficient space for a more relaxed crown ether conformation, similar to solution structures [39-42]. Table 3. Characteristic differences in the FT-Raman spectra of the FAU/EMT containing the crown-ethers 15-Crown-5 and 18-Crown-6. MIX FAU EMT (15-Crown(18-Crown5) 6) 348 347 (Na-O)lattice 354 355 (Na-O)lattice 503 503 503 (S i-O-S i)sym lattice I 860 860sh 15-Crown-5 (C-O-C)sym 869 869 18-Crown-6 (C-O-C)sym II 1001 1002 15-Crown-5 CH2 rocking complex III 1040 1040 15-Crown-5 (C-O-C)asym 1092 1082 18-Crown-6 (C-O-C)asym 1137 18-Crown-6 CH2 wagging complex IV 1137 15-Crown-5 CH2 wagging complex 1152 1152 1270s 15-Crown-5 CH2 twisting complex V 1264 1294 18-C-6/15-C-5 CH2 twisting 1294sh 1296 18-Crown-6 CH stretching complex VI 2859 ,i
i
ACKNOWLEDGMENTS
PPKG thanks the UCL and ESA PRODEX for a research grant. MC is a DEA fellow of UCL. REFERENCES
1. I.R.Lewis and H.G.M.Edwards, Handbook of Raman Spectroscopy, 2001, Marcel Dekker. 2. N.J. Ortinis, T.A. Kruger and P.J. Dutta, Anal. Applics of Raman Spect. Ed., M.J. Pelletier Blackwell Science Oxford (1999) and refs contained therein. 3. R. Ferwerda, J.H. van der Maas and P.J. Hendra, J. Phys. Chem, 97 (1993) 7331.
713 4. P.J. Hendra. Intemet J. Vib. Spec.[www.ijvs.com] 5, 2 (2001) 4. 5. W. Pilz, Z. Phys. Chem. (Leipzig), 271 (1990) 219. 6. G. Xiong, Yi Yu, Z-C. Feng, Q. Xin, F-S. Xiao and C. Li, Micropor. Mesopor. Mater., 42 (2001) 317. 7. P.-P. Knops-Gerrits, D.E. De Vos, E.J.P. Feijen and P.A. Jacobs. Micropor. Mater., 8 (1997) 3. 8. C. Li and P.C. Stair. Catal. Today, 33 (1997) 353. 9. Y. Yu, G. Xiong, C. Li and F-S. Xiao, Micropor. Mesopor. Mater., 46 (2001) 23. 10. P.K. Dutta and B. Del Barco, J. Phys. Chem., 92 (1988) 354. 11. P.K. Dutta and J. Twu, J. Phys. Chem., 95 (1991) 2498. 12. M.A. Mondragon, V.M. Castano, J. Garcia M., C.A. Telles S., Vibrational Spectr., 9 (1995) 293. 13. J. Gnado, P. Dhamelincourt, C. P616gfis, M. Traisnel and A. Le Maguer Mayot, J.Non-Cryst. Solids, 208 (1996) 247. 14. P.K. Dutta and B. Del Barco, J. Phys. Chem., 89 (1985) 1861. 15. P.K. Dutta and B. Del Barco, J. Chem. Soc. Chem. Commun., (1985) 1297. 16. A. Miecznikowski and J.Hanuza, Zeolites, 7 (1987) 249. 17. P.K. Dutta and M. Puri, J. Phys. Chem., 91 (1987) 4329. 18. P.K. Dutta, D.C. Shieh and M. Puri, Zeolites, 8 (1988) 306. 19. P.K. Dutta, K.M. Rao and J.Y. Park, J. Phys. Chem., 95 (1991) 6654. 20. A.J.M. de Man, W.P.J.H. Jacobs, J.P. Gilson and R.A. van Santen, Zeolites, 12 (1992) 826. 21. P.P. Knops-Gerrits, P.A. Jacobs, A. Fukuoka, M. Ichikawa, F. Faglioni and W.A. Goddard III, J.Mol.Cat., A, 166 (2001) 3, P.P. Knops-Gerrits, M. Witko, W. Goddard and R. Millini Eds. 22. P.-P. Knops-Gerrits, H. Toufar, X.-Y. Li, P. Grobet, R. A. Schoonheydt, P.A. Jacobs and W. A. Goddard III, J. Phys. Chem.A.,104, 11 (2000) 2410. 23. P.P. Knops-Gerrits and W.A. Goddard III, J.Mol.Cat., A, 166 (2000) 127. 24. G. Mestl, P. Ruiz, B. Delmon and H. Kn6zinger, J. Phys. Chem., 98 (1994) 11269, 11276 and 11283. 25. G. Mestl, J. Mol. Cat., A, 158 (2000) 45. 26. A. Kozlov, A. Kozlova, K. Asakura and Y. Iwasawa, J.Mol.Cat., A, 137, (1999), 223. 27. E. Feijen, K. De Vadder, M.H. Bosschaerts, J.L. Lievens, J.A.. Martens, P.J. Grobet and P.A. Jacobs, J. Am. Chem. Soc., 116 (1994) 2950. 28. P.A. Jacobs, E.G. Derouane and J. Weitkamp, J.Chem.Soc. Chem.Commun., (1981) 591. 29. J. Twu, P.K. Dutta and C.T. Kresge, J. Phys. Chem., 95 (1991) 5267. 30. F. Roozeboom, H.E. Robson and S.S. Chan, Zeolites, 3 (1983) 321. 31. P.K. Dutta and D.C. Shieh, J. Phys. Chem., 90 (1986) 2331. 32. P.K. Dutta, D.C. Shieh and M. Puri, J. Phys. Chem., 91 (1987) 2332. 33. B.D. McNicol, G.T. Pott and K.R. Loos, J. Phys. Chem., 76 (1972) 3388. 34. J. Twu, P.K. Dutta and C.T. Kresge, Zeolites, 11 (1991) 672. 35. C. Peuker, W. Pilz, B. Fahlke, E. Loettler, J. Richter-Mendau and W. Schirmer, Z. Phys. Chem. (Leipzig), 266 (1985) 74. 36. P.K. Dutta, B. Del Barco and D.C. Shieh, Chem. Phys. Lett., 127 (1986) 200.
714 37. P.K. Dutta and D.C.Shieh, J. Raman Spectrosc., 16 (1985) 312. 38. K. Nakamoto, IR & Raman Spectra of Inorganic & Coordination Compounds, Wiley (1986) 124. 39. F. Delprato, L. Delmotte, J.L. Guth and L. Huve, Zeolites, 10 (1990) 546. 40. S.L. Burkett and M.E. Davis, Microporous Mater., 1 (1993) 265. 41. J.J.P.M. de Kanter, I.E. Maxwell, P.J. Trotter, J. Chem. Soc. Chem. Commun., (1972) 733. 42. C. Br6mard and M. Le Make, J. Phys. Chem., 97 (1993) 9695.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
715
Templateless synthesis of catalysts with narrow mesoporous distribution 9 a* N. Yaoa, G. Xlong , S. Sheng a, M. He b and K.L. Yeung e
a State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, P. O. Box 110, Dalian 116023, P. R. China bResearch Institute of Petroleum Processing SINOPEC, Beijing 100083, P. R. China c Department of Chemical Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong, SAR, P. R. China In this study, amorphous silica-alumina nanomaterials with narrow mesoporous distribution can be obtained by two novel sol-gel processes, without the use of any templates. The results of our experiments show that the preparation method has a great influence on the precursor sol structure as well as the specific surface area and mesopore volume of the final product, but has no effect on the pore size distribution. 1.
INTRODUCTION The tendency of the refining industry to process heavier fossil oils has raised the
interest in catalysts which will have pore sizes optimized to cope with large molecules. Microporous zeolites, which are widely used in catalytic processing, are limited in their applications in this area of the treatment of large molecules due to their small pore geometry. Thus, the demand for mesoporous materials has triggered major synthetic efforts [1]. Since the discovery of mesoporous molecular sieve (M41S) in 1992 [2], it has now been well accepted that the formation of such mesoporous materials could occur through several templating pathways by using various types of organic templates [3]. Until now, none has described a method for the synthesis of mesoporous material without the aid of a surfactant.
Corresponding author Email:
[email protected] 716 The objective of the present work is to obtain narrow pore-sized, mesoporous materials without the use of surfactants. Two distinct templateless synthesis routes defined as vacuum-sol method and ultrasonic-sol method were developed, respectively. In addition, attempts have been made to identify and evaluate the effect of the synthesis method on the texture of the materials. 2. EXPERIMENTAL
Method A (Ultrasonic-sol method): The pH value of a 50ml water glass solution was adjusted to 10 through the addition of 0.94 M nitric acid. The prepared alumina sol was then added to the water glass solution until the desired Si/A1 molar ratio was attained. The resulting precipitate was peptized using nitric acid and then ultrasonically treated for 1 minute to obtain a stable sol. When the sol finally formed a gel at room temperature, 70ml of NH4NO3 solution (1.2M) was added to the gel to remove the sodium ions. At the end of 24 hours, the solution was drained offby centrifugation, and this procedure was repeated 3 times. The gels were calcined in air at 550 C for 10 hours to obtain the solid samples. In this work, the sol and solid samples prepared by the method A were named as USG samples [4]. Method B (Vacuum-sol method): A 0.96M aluminum nitrate solution was added to 50ml water glass solution until the Si/A1 molar ratio was attained. The precipitate was collected by centrifugation and washed 7 times to remove the sodium ions. After washing, the precipitate was put into 200ml of water and then added a certain quantity of 0.94M nitric acid to peptize the precipitate to obtain the sol sample. The prepared sol was dried at room temperature in a vacuum box until it began to form the gel sample. The gels were also calcined at 550 C in air for 10 hours. The sol and solid samples produced by method B were defined as VSG samples [5]. A N4 plus laser scattering particle meter (Coulter) was used to measure the sol particle diameter distribution at a 90 angle to the light beam. The material structure and chemistry were characterized by X-ray diffraction (XRD, Rigaku D/MAX-RB), atomic force microscopy (AFM, Nanoscope III), transmission electron microscopy (TEM, JEM2010) and N2 physi-adsorption (Omnisorp- 100CX).
717 3. RESULTS AND DISCUSSION 3.1. Effect of preparation method on the precursor soi's properties
Fig.1 and Table 1 summarize the effect of the Si/AI molar ratio and preparation method on the sol samples' pH values and the particle diameter distributions. It is seen that the VSG sol samples have large particle diameter in the range 200nm to 2000nm. Increasing the Si/A1 molar ratio results in a decrease in pH value as well as broadens the particle diameter distribution as shown in Fig. 1A. In contrast, every USG sol sample has small particle diameter and narrow particle diameter distribution (e.g. Fig. 1B). Fig. 2 shows USG1 and VSG1 sol samples' AFM images, respectively. The large secondary particles existed in VSG1 sample are formed by random aggregated of spherical primary particles with size of 10-20ran, while the uniform sized primary particles within 13-25nm disperse well in the USG1 sol sample. It is thus that USG1 and VSG1 sol samples have similar sized primary particles, although they have different particle structure. These results mean that the preparation method take great contribution to precursor sol's structure and particle diameter. The method A favors the formation of monodispersed and stable sol particles in the system. 100 USG1
75
ii
50 25
~50~
20
.40
40
lq~O
60 USG2
9
.
_ _
,
|
,
I
,
,
'
10 0541 411
30
VS~
,,/ A
size(m)
20
40
60
80
ltbO
USG3
20 10
~
B
o
40 60 Size (nm)
80
Fig. 1. Particle diameter distribution of USG and VSG sol samples, respectively.
100
718 Table 1 pH value of every precursor sol sample Sample No. VSG1 VSG2 VSG3 USG1 USG2 USG3
Si/A1 molar ratio 10 7 3 10 7 3
pH value 2112 2.41 2.50 2.55 2.48 2.01
3.2. Effect of the preparation method on the solid's texture
The XRD analysis shows that all of synthesized materials are amorphous. Table 2 reports the solid samples' specific surface area and main pore texture characteristics. The solids have large specific surface area, and this value diminish as decrease of Si/A1 molar ratio in both USG and VSG series. Fig. 3 displays the pore size distribution of USG1 and VSG1 samples, respectively. It is clear from the figure that the both samples have a narrower pore size distribution (i.e., 2- 10nm). Fig. 4 shows a representative micrograph of the solid sample analyzed by TEM. There is no apparent order in the pore arrangement unlike the ordered hexagonal array observed in M41S mesoporous molecular sieves. Also, the sample is made of spherical particles with narrow size distribution from 12nm to 25nm. This result indicates that the present methods enable to produce nanomaterial and the pores existed in the solids may be created through the random packing of spherical particles. Based on the TEM and N2 sorption results, it is concluded that the developed methods are able to synthesize silica-alumina nanomaterials
719 with large specific surface area and narrow mesoporous distribution absent of organic templates.
Fig. 3. Pore size distribution of USG1 and VSG 1 samples, respectively.
Fig. 4. TEM morphology synthesized solid sample.
of
Table 2 Physic chemical characteristics of the synthesized materials Sample Preparation S i / A 1 MEPV" MIPV + Specific surface Pore size No. method molar ratio (ml/g) (ml/g) area (m2/g) distribution (nm) VSG1 B 10 0.244 ' 0.03 542.99 2.12-10.61 VSG2 B 7 0.294 0.00 493.89 2.12-12.08 VSG3 B 3 0.364 0.00 432.96 3.21-12.03 USG1 A 10 0.291 0.01 692.54 2.04-10.20 USG2 A 7 0.355 0.01 648.59 3.24-11.23 USG3 A 3 0.305 0.01 587.24 3.20-11.70 Mesopore volume calculated by BJH method from adsorption branch. + Micropore volume calculated by t-plot method from adsorption branch. i
i
As can be seen in Table 2, it is found that the USG samples have larger specific surface area and pore volume than those of VSG samples, except for VSG3's pore volume. This phenomenon can be ascribed to two reasons. First, as described in experimental section, the drying processes of two methods are totally different. The sample derived from
720 vacuum-sol method should dry under the vacuum condition, otherwise, it is impossible to produce mesoporous materials with narrow size distribution. However, the ultrasonic-sol method does not take any drying action prior to calcination. The gel was formed under the general air pressure and the room temperature. This difference makes a great contribution to the gel's structure. As can be seen in Fig. 5A, many sol particles densely aggregated to form large island-like particles in VSG1 gel sample due to the effect of capillary force appeared in the vacuum drying process. Nevertheless, the sol particles only form an incompact gel network in USG1 gel sample. Obviously, the denser particle aggregation causes more voids collapse in the VSG gel network, which leads to decrease the pore volume and specific surface area.
Fig. 5. (A) AFM image of VSG1 gel sample. (B) AFM image of USG1 gel sample. In order to confirm the above explanation, typical TG analysis was performed and its pattern is displayed in Fig.6. In the TG curves, the weight loss between the 40 C and 150 C can be ascribed to the desorption of physically adsorbed water [6]. In this region, it should be noted that the weight loss of USG1 gel sample is much greater than that of VSG1 sample. That is to say, the shrinkage of USG1 gel sample caused by capillary force is restrained mostly so that its network has much more pore structure to contain the water before the heat treatment. This result is in good agreement with the above
discussion. 110 ,
100
'~
90
80 .~ 70 -" .~ 60 ~ 50 40 30 20 lo o o
:
~
VSG 1
.......
USG1
,! ~
-
~.! ~._.
:
"~'"'~'i,
I
100
,
I
,
Rtgion III I
200 300
|
I
,
i
,
i
400 500 600
Temperature
,
i
700
,
i
800
~C)
Fig. 6. TG curves of USG1 and VSG1 gel samples, respectively.
721 The second possible reason is the role of base-exchange procedure in the ultrasonic-sol method. In fact, such a step is not only to remove the sodium ions in the system, but also to age the gel. The latter takes a significant effect on the gel structure because the amount of shrinkage that occurs during drying is dependent on the stiffness of the gel's network. If the gel is aged under a suitable condition, the network may be strengthened and stiffened so that it can resist compression by capillary forces during the subsequent calcination [7]. Accordingly, it has a profound effect on the decrease of the pore collapse. It is hereby reasonable to find that USG samples have larger pore volume and specific surface area than VSG samples. Another noticeable feature in Table 2 is the pore size distribution. It is seen that materials have similar and controlled mesoporous distribution, although they have different Si/AI molar ratios and are prepared by different methods. A possible explanation can be obtained by analysis of the particle diameter of the precursor sols. It is found in Fig.2 that the VSG1 and USG1 sol samples have primary particles with similar size. After gelation, the primary particles maintain their size and morphology but form regularly packed aggregate clusters. The fluid-filled inter-particle voids are the precursors for the mesoporous channel network in the final material. The size of these void spaces is dictated mainly by the size of the primary sol particles and their packing order. Thus, primary sols of similar size and distribution (cfr. Figs. 2 A and B) should lead to final mesoporous materials of similar pore size distribution (cfr. Table 2). A more detailed explanation about this point has already described in our previous work for every case, respectively [4,5]. Thus, the N2 sorption analysis results mean that the preparation method only affects the specific surface area and pore volume of the materials, but has no relationship to the pore size distribution. The pore size distribution of the final products is only related to the diameter and size distribution of the primary sol pai'ticles. 4. CONCLUSION In the present study, amorphous silica-alumina nanomaterials with controlled mesoporous distribution have been synthesized by two templateless approaches: (1) vacuum-sol process, and (2) ultrasonic-sol process. It is found that the preparation method affects the precursor sol properties and the specific surface area and pore volume of the final materials. Ultrasonic-sol method favors the formation of monodispersed sol particles with narrow size distribution. Because of several base-exchange cycles and absence of drying process prior to heat treatment, the gel derived from ultrasonic-sol method may have enough stiffness to protect the network from pore collapse by capillary force, thus, leading to produce the materials with
722 larger specific surface area and pore volume than vacuum-sol method. Moreover, the analysis results also indicate that the synthesis method has no effect on the pore size distribution of the materials due to the fact that their precursor sols have similar-sized primary particles.
ACKNOWLEDGMENT
The authors are indebted to Ms. Zhang Yan (The Materials Characterization & Preparation Faculty, HKUST) for acquiring JEM 2010 Microscope. The financial support from the Chinese Academy of Sciences, the Research Institute of Petroleum Processing SINOPEC and the National Sciences Foundation of China are also acknowledged. REFERENCES
1. S. Biz and M. L. Occelli, Catal. Rev.-Sci. Eng., 40(3) (1998) 330. 2. C. T. Kresge, M. E. Leonowicz, W. J. Roth, J. C. Vartuli and J. S. Beck, Nature, 359 (1992) 710. 3. Q. Huo, D. I. Margolese, U. Ciesla, P. Feng, T. E. Gier, P. Sieger, R. Leon, P. M. Petroff, F. Schtith and G. Stucky, Nature, 368 (1994) 317. 4. N. Yao, G. X. Xiong, S. S. Sheng, M. Y. He, W. S. Yang and X. H. Bao, Catal. Lett., in press. 5. N. Yao G. X. Xiong, M. Y. He, S. S. Sheng, W. S. Yang and X. H. Bao, Chem. Mater., 14(1) (2002) 122. 6. C. J. Brinker and Ct W. Scherer, Sol-gel Science. The Physics and Chemistry of Sol-gel Processing, Academic Press Inc., Harcourt Brace Jovanovich, 1990. 7. C. J. Brinker and G. W. Scherer, Sol-gel Science. The Physics and Chemistry of Sol-gel Processing, Academic Press Inc., Harcourt Brace Jovanovich, 1990.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
Control of pore structures of titanias andtitania/aluminas complexing agents
723
using
M. Toba, S. Niwa, N. Kijima and Y. Yoshimura Research Institute of Green Technology, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8565, Japan In order to control the surface area, pore volume and pore distribution, titanias were prepared by using complexing agents and the relationship between the structure of complexing agents and pore structures weas examined. The structure around the hydroxy group of the alcohol used as a complexing agent has large influence on the pore structure of titania. The modification of titania sol by using complexing agents was also effective for controlling the pore volume. The specific surface area and pore volume of precipitated titania/aluminas were proportional to those of alumina supports prepared with various complexing agents. 1. I N T R O D U C T I O N Titania and titania containing mixed oxides are extensively used as catalysts and supports. Recently, many preparation techniques have been developed to control their pore structure. However, some of them are not suitable for a systematic control of the pore distributions. In addition, pores usually disappear after calcination at high temperature [1]. To avoid this problem, catalysts were calcined at relatively low temperature [2-4] or an extraction method was used to remove organic residues [2]. Under such mild conditions, it is impossible to remove completely the organic residues which are bonded to a titania precursor. In this work, we prepared titanias using two procedures: sol-gel [5, 6] and modified sol methods using complexing agents which were able to be easily removed. Titania/aluminas were prepared by titania precipitation on alumina supports which were modified by complexing agents. The effect of the structure of complexing agents on their surface areas, pore volumes, pore distributions and crystallinities were examined. 2. E X P E R I M E N T A L The sol-gel titanias were prepared using mono alcohols as complexing agents. Titanium iso-propoxide (TIP) was mixed with a complexing agent and stirred at 393 K for 3 h. After removing 2-propanol which was formed by ligand exchange reaction under reduced pressure at 353 K, water was added to the solution to hydrolyze the various titania complexes formed. The modified titania sols were obtained by neutralization of basic titania sol modified by complexing agent under reflux condition. Titania/aluminas were
724 prepared as follows. Alumina supports prepared by a complexing agent-assisted sol-gel method were calcined at 823 K for 2 h. Titanium iso-propoxide was added to alumina and 2-propanol suspension and then water was added to the suspension to hydrolyze the alkoxide. Ligand exchanged titanium alkoxide was also used as a titania source. The obtained gels were dried at ca. 413 K under reduced pressure. Finally, the dry gels were calcined at 823 K for 2 h. Specific surface areas, mesopore volumes and pore distributions were determined from nitrogen adsorption data at 77 K using a BELSORP 28SA (Nippon BEL. Co.). All the samples were outgassed at 413 K for 3 h before measuring the nitrogen adsorption. The X-ray powder diffraction patterns were obtained on a MAC Science MXP-18 instrument using Cu-Ka radiation with a Ni filter. 3. RESULTS AND DISCUSSION
3.1 Sol-gel titanias Fig. 1 shows the typical nitrogen isotherms of titanias measured at 77 K. The isotherms of the sol-gel titanias are type IV of IUPAC isotherm classification [5]. Effect of the amount of the complexing agent on the rate of ligand exchange, surface area and pore volume is shown in Table 1. The rate of ligand exchange was calculated by the amount of 2-propanol evaporated by ligand exchange reaction. The rate of ligand exchange increased with increasing the amount of alcohol (2-ethyl-l-hexanol; 2-EHA) used as a complexing agent. However, the pore volume increased with increasing 2-EHA/TIP ratio from 1 to 3, and remained constant at 2-EHA/TIP ratio of 3 to 5. The specific surface area slightly decreased with increasing the amount of alcohol. The effect of the amount of complexing agent on the pore distributions is shown in Fig. 2. V/cm3g -1 300
'''
dV/dR/mm3nm-lg -1
I'''
I'''
I'''
I''
'-{
j,/~
250
i00{2,,,,....'....'....'....'....I....,""_{ 80 I
i
.....2-EHA/TIP=I~ 2-EHA/TIP=2 "~ 2-EHA/TIP=3 ] -f"~"j .....2-EHA/TIP=4 ~'.. " Z
200 adsorption desorption
150
( ,
60I ~
.~,~t. f;.,.'.,j~"
Z
100 50
"
~
-
f~
,,, l,,, l,, , l , , , l , ,
,
0.2 0.4 0.6 0.8 1 Fig. 1 Typical nitrogen isotherm of titania Complexing agent, 2-ethyl-l-hexanol; calcination, 823 K. 2h.
0
0
5
10 15 20 25 30 35 40 Pore diameter / n m Fig. 2 Effect of the amount of the complexing agent on the pore distributions of titanias. Complexing agent, 2-ethyl-l-hexanol; calcination, 823 K, 2h.
725 Table 2 shows the effect of calcination temperature on the specific surface area and the pore volume of titania. The surface area and pore volume were almost constant between 723 K and 773 K and decreased between 773 K and 823 K. Table 1. Effect of the amount of the complexing agent on the rate of ligand exchange, surface area and pore volume 2-EHA/TIP Rate of ligand Specific surface Mesopore (mol/mol) exchange (%) area (m2/g) volume (cm3/g) 1 2 3 4 5
13.2 34.5 42.3 62.6 71.9
125 123 113 117 113
0.32 0.38 0.41 0.41 0.41
Calcination, 823 K, 2h. Table 2. Effect of calcination temperature on the specific surface area and pore volume of titania Calcination temperature Specific surface Mesopore (K) area (m2/g) volume (cm3/g) 723 773 823 Complexing agent, 2-ethyl-l-hexanol
139 136 117
0.46 0.47 0.41
The rates of ligand exchange, the surface areas and the pore volumes of titanias prepared with various complexing agents are shown in Table 3. The rate of ligand exchange decreased in the order primary alcohol > secondary alcohol > tertiary alcohol. Secondary and tertiary alcohol gave titanias with higher surface area and larger pore volume. For example, the specific surface area and the pore volume of sol-gel titania prepared with 1-phenylethanol are much larger than those of the titania prepared with 2-phenylethanol, while the molecular sizes of 1- and 2-phenylethanol are almost equal. These results mean that the structure around the hydroxyl group of the complexing agent molecule has a large influence on the pore structure of titania. Among the non-branched primary alcohols, the specific surface area and mesopore volume decrease with increasing chain length, while the ligand exchange rates are almost equal to each other. Branched primary alcohol such as 2-ethyl-l-hexanol also shows higher surface area and larger pore volume. Fig. 3 shows the effect of the complexing agent on the pore distributions of titanias. Most of the primary alcohols (Group A) gave titanias with small pores (3~6 nm). Branched primary alcohol (2-EHA), secondary alcohols and tertiary alcohols (Group B) gave titanias with large pores (10 ~13 nm). Comparing three alcohols of Group A, a more bulky alcohol gave a titania with larger pore size. Asimilar tendency was observed in Group B. The ligand exchange rates of primary alcohols are higher than those of secondary and
726 tertiary alcohols. These results indicate that titanium atoms in the alkoxide species formed by ligand exchange reaction with primary alcohol were surrounded by more bulky alkoxy groups. These alkoxy groups seem to prevent the hydrolysis reaction. As expected, precipitates were slowly formed when 1-octanol, 1-decanol, benzyl alcohol and 2-phenylethanol were used as complexing agents. These results also mean that a considerable amount of alkoxy groups bonded to titanium still remains in the gels prepared by using aromatic and straight chain aliphatic primary alcohol after hydrolysis and drying. During calcination, these remaining organics are eliminated at relatively high temperature and the structural change caused by the elimination results in decreasing the surface area and the collapse of mesopore. In contrast, the titanium alkoxide species formed by ligand exchange reaction with secondary and tertiary alcohol easily hydrolyzed and excess amount of organics could be removed during drying. When the ligand exchange was depressed under mild condition, primary alcohol also gave a titania with relatively high surface area and mesopore. For example, when the ligand exchange rate is 51%, the surface area and mesopore volume of titania prepared by using 1-octanol are 96 mZ/g and 0.22 cm3/g, respectively. Therefore, in order to control the pore structure, both control of the ligand exchange rate and selection of optimum complexing agent are important.
.
Table 3. Specific surface areas and pore volumes of titanias prepared by using various complexing agents. Rate of ligand Specific surface Mesopore Complexing agent exchange (%) area ( m a / g ) volume (cm3/g) Primary alcohol 1-Hexanol 74.5 95 0.21 1-Heptanol 74.4 87 0.17 1-Octanol 82.2 7 0.04 1-Decanol 77.3 11 0.03 2-Ethyl- 1-hexanol 62.6 117 0.41 Benzyl alcohol 66.3 33 0.07 2-Phenylethanol 61.6 46 0.07 Secondary alcohol 2-Heptanol 72.7 91 0.43 2-Octanol 43.7 108 0.37 1-Phenylethanol 44.6 122 0.45 Cyclohexanol 72.6 92 0.29 1-Butoxy-2-propanol 53.6 114 0.30 Tertiary alcohol 2-Methyl-2-butanol 49.8 89 0.18 2-Phenyl-2-propanol 39.1 122 0.44 Calcination, 823 K, 2h; alcohol/TIP (mol/mol) = 4. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Fig. 4 shows X-ray powder diffraction patterns of sol-gel titanias. Most of samples showed patterns characteristic of anatase form and weak diffraction patterns of brookite
727 titania. The extent of crystallinity decreased in the order (A) > (B) > (C) = (D) = (E). The diffraction peak derived from brookite titania (2[]=31 ~ was getting smaller with decreasing specific surface area and pore volume. These results indicate that crystallinity of titania is influenced by complexing agents. dV/dR/mm 3nm-lg -1 100 F"
80
. . . . 2-Heptanol ...... 1-Phenylethanoi"/'~!'2-P---henyl-2-prop ano~
6O
If-'\
:
~
20 0
~1 5 10 15 20 25 30 35 40 0 Pore d i a m e t e r / n m
0
5
10 15 20 25 30 35 40 Pore d i a m e t e r / n m
Fig. 3. Pore distributions of titanias prepared by using primary alcohol (A), secondary and tertiary alcohol (B). Calcination, 823 K, 2h; alcohol/TIP (mol/mol) = 4. cps 5000
'"'
'
I
' ' ' '
I ' ' ' '
I ' ' ' '
I ' ' ' '
I ' ' '
'
I
' ' ' '
I ' ' '
"
-
4000 3000
ooo
Z
1ooo _0 "
(E) 10
20
30
40
_~.~/~
60
70
so
Fig. 4. X-ray powder diffraction patterns of titanias prepared by using various alcohols. Calcination, 823 K, 2h; alcohol/TIP (mol/mol) = 4. (A), benzyl alcohol" (B), 2-phenylethanol; (C), 2-ethyl-l-hexanol; (D), 1-phenyl ethanol; (E), 2-phenyl-2-propanol
728
3.2 Modified
sol titanias
The specific surface areas and the pore volumes of titanias prepared by a modified sol method are shown in Table 4. The mesopore volume increased with increasing the molecular size of the complexing agent, while the specific surface areas of three samples are almost equal. The pore size of modified sol titania also increased with increasing the molecular size of the complexing agent (Fig. 5). In this method, a complexing agent reacts with the surface hydroxy group of titania sol at the first step. This combined complexing agent has an influence on the structural changes during drying and calcination. These changes seem to result in different pore structures. The modified sol method was also effective for controlling the pore structure. Table 4. The specific surface areas and the pore volumes of titanias prepared by a modified sol method. Specific surface Mesopore Complexing agent area (m2/g) volume (cm3/g) Ethylene glycol Diethylene glycol Triethylene glycol
77 82 78
0.09 0.21 0.24
Calcination, 823 K, 2h; alcohol/Ti (mol/mol) = 4. dV/dP~,~l:0m3nm-lg-1 o u , I'", ...... " I ....
dV/dR/mm3nm -lgq 100
80
400
60
.
300 40 20
200~-
o
lO 0
r
ia
2mO e t
/
20
' .... ""i 89 A1203
-
....
' TiO2/A1203 (2-ME) 2::::: A1203 (2-EHA) TiO2/Al203 (2-EHA~ - - - A1203 ( 1 -PEA) ~ TiO2/A1203 (1-PEA)S_ i=''=" TiO2(1-PEA) /A1203 (1-PEA) __ ! ,-,
'
.
.
.
~t
,~ : ~
k
-,.
o ~ ~ i ~ . . '.':. 9.-r..i~ Fig. 5. Pore distributions of titanias prepared by 0 5 10 15 20 25 30 35 40 Pore d i a m e t e r / n m modified sol method. Calcination, 823 K, 2h; alcohol/Ti (tool/tool) = 4. Fig. 6. Pore distributions of precipitated titania/aluminas, 2-ME, 2-methoxyethanol, 2-EHA, 2-et hyl- 1-hexanol, 1-PEA, 1-phenylethanol. Calcination, 823 K, 2h; alcohol/Al (mol/mol) = 4, Ti/AI=I.
729
3.3 Precipitated titania/aluminas Table 5 and Fig. 6 show the specific surface areas, pore volumes and pore distributions of precipitated titania/aluminas (Ti/AI=I) and alumina supports prepared by using complexing agents. Generally, the specific surface areas, pore volumes and pore sizes of precipitated titania/aluminas were proportional to those of the corresponding alumina supports. However, the mesopore of precipitated titania/alumina prepared by using an alumina (1-PEA) with extremely large pore disappeared. In this case, the titania precursor was filled in the pore on the alumina support during mixing and most of the pore might be covered by titania after calcination. However, pore volume and pore size remarkably increased when ligand exchanged titanium alkoxide was used as a titania precursor. Titania/aluminas with large pore volume and pore size could not be obtained with this method. However, their surface areas are larger than those of pure titanias, depending on the structure and coordination ability of the complexing agents. Table 5. Specific surface areas and Por e volumes of precipitated titania. Mesopore Material Complexing agent Specific surface volume (cm3/g) area (mZ/g) AlzO3 2-Metho xyet ha no I 282 0.58 TiOz/Alz03 2-Methoxyethanol 153 0.29 AIzO3 2-Ethyl- 1-hexano 1 315 0.89 TiOz/AlzO3 2-Ethyl-l-hexanol 183 0.40 AlzO3 1-Phenylethanol 338 1.43 TiOz/Alz03 1-Phenylethanol 142 0.08 TiOz/Alz03 a' 1-Phenylethanol 187 0.45 Calcination, 823 K, 2h (support)+2h(precipitated sample); alcohol/Al (mol/mol) = 4, Ti/AI=I. a) Titanium alkoxide which was prepared from titanium iso-propoxide and 1-phenylethanol by ligand exchange reaction was used as a titania source. ,
,
4. C O N C L U S I O N The preparation of titanias and precipitated titania/aluminas using complexing agents enables a systematic control of their pore structures. This method is effective not only for modification of titanium alkoxide which is used as a titania precursor but also for modification of titania sol.
REFERENCES 1. 2. 3. 4. 5. 6. 7.
K.C. Song et al., J. Colloid Int. Sci., 231 (2000) 289. Y. Miyake and T. Kondo, J. Chem. Eng. Jpn., 34 (2001) 319. S. Cabrera et al., Solid State Sci., 2 (2000) 513. S. Takenaka et al., J. Sol-Gel Sci. Tech., 19 (2000) 711. M. Toba et al., J. Mater. Chem., 4 (1994) 585. M. Toba et al., J. Sol-Gel Sci. Tech., 19 (2000) 695. K . S . W . Sing et al., Pure Appl. Chem., 57 (1985) 603.
This Page Intentionally Left Blank
Studies in Surface Science andCatalysis143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
731
Tungstophosphoric acid immobilized in polyvinyl alcohol hydrogel beads as heterogeneous catalyst Luis R. Pizzio, Carmen V. Cficeres and Mirta N. Blanco Centro de Investigaci6n y Desarrollo en Procesos Cataliticos (CINDECA), UNLP-CONICET, 47 N ~ 257, (1900) La Plata, ARGENTINA e-mail: hds @dalton.quimica.unlp.edu.ar The preparation of tungstophosphoric acid supported on beads of polyvinyl alcohol hydrogel and polyethylenglycol, by means of equilibrium impregnation, was studied. The catalyst was characterized by FT-IR, 31p MAS-NMR, XRD, SEM-EDAX and TG-DTA. The results showed that the acid retains its Keggin structure upon impregnation. Moreover, the anion is firmly attached to the support, as seen by leaching studies. The acidity measurements by means of potentiometric titration with n-butylamine showed that the catalyst presents similar acid strength than the bulk acid, although the number of acid sites is lower. A high selectivity to ester formation was observed in the reaction of esterification of acetic acid with isoamyl alcohol. 1. I N T R O D U C T I O N The strong acidity of heteropolyacids (HPA) make them suitable as catalysts for many acid-catalyzed reactions. Since HPA are less corrosive and produce lower amount of waste than conventional acid catalysts, as sulfuric acid, they can be used as replacement in environmentally benign processes. The use of a support allowing the HPA to be dispersed over a large surface may result in an increase of its catalytic activity. The performance of supported HPA catalysts depends on the carrier, the HPA loading, conditions of pretreatment, among other variables. Acidic or neutral solids such as active carbon, SiO2 and ZrO2 are suitable as supports [1]. But HPA often leaks out of catalyst supports even in vapor-phase reactions. It is important, for practical purposes, to develop supported catalysts which can be applied to several reactions with no leakage of HPA. Interesting supports are the polymeric materials, notwithstanding their thermal instability at high temperatures. In the electrocatalysis field, the use of polypyrrole, polythiophene and polyaniline as heteropolyanion supports was reported [2]. The catalytically active species were introduced, in this case, via electrochemical polymerization. Hasik et al. [3] studied the behavior of polyaniline supported tungstophosphoric acid in the isopropanol decomposition reaction. The authors established that a HPA molecular dispersion can be attained via a protonation reaction. The different behavior of the supported catalysts with respect to bulk acid, namely, predominantly redox activity versus acid-base activity, was attributed to that effect.
732
Molybdophosphoric acid was also used to prepare HPA-polymer composite film catalysts, using polyphenylene oxide, polyethersulfone and polysulfone as polymers [4]. The membrane-like materials were tested as catalysts in the liquid-phase synthesis of tertbutanol from isobutene and water, showing higher catalytic activity than the bulk acid. On the other hand, polyvinyl alcohol is used in biotechnology for the encapsulation of enzymes or cells, due to its ability to form hydrogels by crosslinking with boric acid or formaldehyde [5]. Its use for the encapsulation of metal catalysts was also reported [6]. Taking into account the above-mentioned reports, we prepared a tungstophosphoric acid (TPA) based catalyst by means of equilibrium impregnation and using beads of polyvinyl alcohol hydrogel (PVA) and polyethylenglycol (PEG) as support. The aim is to obtain materials without TPA leakage. The characterizations were carried out by different physical-chemical techniques. The activity of the catalyst is measured in the esterification of acetic acid with isoamyl alcohol. 2. E X P E R I M E N T A L
2.1. Catalyst preparation The PVA-PEG gel beads (mean diameter: 2 ram) were prepared using the freezingthawing method [7]. The PVA solution was prepared by solving PVA (Mallinckrodt, 2.2 g) in a hot mixture of PEG (Mallinckrodt, 2.0 g) and water (18 g). The solution was slowly dropped into liquid nitrogen. After freezing, the beads were slowly thawed after the nitrogen was evaporated, using a Dewar vessel. The beads (1 g) were impregnated in equilibrium at 20 ~ with an ethanol-water (50 % v/v) solution (0.004 dm 3) of TPA (Fluka) of 130 g W/dm 3 concentration for 72 h and dried at room temperature. Tungsten concentration in the catalyst was calculated on the basis of the decrease of tungsten amount in the solution, by means of a mass balance. The tungsten concentration in the solutions, both before and after contacting the beads, was determined by atomic absorption spectrometry. The calibration curve method was used, with standards prepared in the laboratory. The equipment used was an IL Model 457 spectrophotometer, with single channel and double beam, and monochromator of 330 mm focal distance. The light source was a hollow monocathode lamp. The analyses were carried out at a wavelength of 254.9 nm, bandwidth 0.3 rim, lamp current 15 mA, phototube amplification 800 V, burner height 4 mm and acetylene-nitrous oxide flame (11:14). Catalyst thus obtained was washed with toluene, acetonitrile or chloroform, at 70 ~ for 6 h, in a system with continuous stirring. Finally, catalysts were thermally treated in the same conditions as before washing. 2.2. Catalyst Characterization 2.2.1. Fourier transform infrared spectroscopy Spectra of PVA-PEG beads and TPA-PVA-PEG samples dried at room temperature were recorded. For these analysis, a Bruker IFS 66 FT-IR equipment, pellets in BrK and a measuring range of 400-1500 cm- 1 were used. 2.2.2. Nuclear magnetic resonance spectroscopy The same solid samples studied by FT-IR were analyzed by 31p MAS-NMR. For
733
this purpose, a Bruker MSL-300 equipment with a sample holder of 5 mm diameter and 10 mm in height was employed, using 5 [as pulses, a repetition time of 10 s and a frequency of 121.496 MHz, being the resolution of 3.052 Hz per point and the spin rate 2.1 kHz. The repetition time was 3 s, and several hundred pulse responses were collected. Phosphoric acid 85% was employed as external reference. 2.2.3. X-Ray diffraction XRD patterns of the solid samples were recorded. The equipment used to this end was a Philips PW-1732, with built-in recorder. The operative conditions were: Cu I ~ radiation, nickel filter, 30 mA and 40 kV in the high voltage source and scanning angle between 5 and 55 ~ of 20 at a scanning rate of 1~ per minute. 2.2.4. Scanning electron microscopy The distribution of TPA molecules over the radius of the beads of PVA-PEG was measured using a Philips Model 505 scanning electron microscope with energy dispersive X-ray analysis (EDAX) system. The secondary electron micrographs of selected solid samples were obtained. 2.2.5. Thermogravimetric and differential thermal analysis The TG-DTA measurements of representative samples dried at 70 ~ were carried out using a Shimadzu DT 50 thermal analyzer. The thermogravimetry and differential thermal analysis experiments were performed under argon or nitrogen respectively, using 25-50 mg samples and a heating rate of 10 ~ Quartz cells were used as sample holders with c~A1203 as reference. The studied temperature range was 25-700 ~ 2.2.6. Acidity measurements Acidity of solid samples was measured by means of potentiometric titration. A small quantity of 0.05 N n-butylamine in acetonitrile was added to a known mass of solid suspended in acetonitrile, and agitated for 3 h. Later, the suspension was titrated with the same amine solution at 0.05 ml/min. The electrode potential variation was measured with a digital pHmeter. 2.2.7. Catalytic activity The esterification was carried out in a glass batch reactor at atmospheric pressure. Isoamyl alcohol (0.058 tool) and acetic acid (0.058 tool) were dissolved in toluene (0.114 tool). Then, the TPA-PVA-PEG catalyst (0.100 g) was added and the resulting mixture was heated to reflux. The reaction was followed by gas chromatography, using a thermal conductivity detector. Also, hydrochloric acid was used as catalyst for comparative purpose. 3. RESULTS AND DISCUSSION
3.1. Fourier transform infrared spectroscopy FT-IR spectrum of bulk TPA dried at 70 ~ (Fig. la) shows bands at 1081, 982, 888, 793, 595 and 524 cm -~, which coincide with those referred to in the literature for the acid
734
H3PW12040 [8]. The first five bands are assigned to the stretching vibrations P-Oa, W-Od, W-Ob-W, W-Oc-W, and to the bending vibration Oa-P-Oa, respectively. The subscripts indicate oxygen bridging the W and the heteroatom (a), corner-sharing (b) and edge-sharing (c) oxygens, belonging to WO6 octahedra, and terminal oxygen (d). Fig. lb shows the FF-IR spectrum of TPA-PVA-PEG catalyst (290 mg W/g support). It presents bands at 1081, 982, 897 and 812 cm -1 assigned to [PW~2040] 3- anion, although some of them are overlapped with those of the support (Figure lc). The W-Od band (982 cm -1) exhibits a splitting that may be a result of a direct interaction between the [PWlzO40] 3- anions and C-OH2 + groups of the PVA-PEG support. Similar observations have been reported for the case of W-Od in Cs2.sH0.sPW]2040 [9] and Cul.sPWl2040 [10] salts.
a
23
(1) o
C
E 03
t"-
I--
c
I
1400
,
I
1200
,
I
1000
,
I
800
,
I
600
,
400
Wav enumber (cm -1)
Fig. 1. FT-IR spectra of bulk TPA (a), TPA-VPA-PEG catalyst (b), PVA-PEG support (c). 3.2. Nuclear magnetic resonance spectroscopy 3ap MAS-NMR spectrum of TPA-PVA-PEG beads shows an intense line a t 14.6 ppm which can be attributed to the [PW12040] 3- anion [11]. Neither the [PWllO39] 7lacunar species nor [P2W21071] 6- dimeric anion were detected by this technique. Then, the study of the catalyst by FT-IR and 3]p MAS-NMR allowed us to verify that the species present in the solid coincided with those in the impregnating solution of the support, that is [PW12040] 3- anion [1]. 3.3. X-ray diffraction and scanning electron microscopy The XRD pattern of the TPA-PVA-PEG catalyst did not present lines corresponding to crystalline structures and, as a consequence, it is similar to that of the support. This may
735 be due to a high dispersion of TPA on the support surface and/or to the presence of amorphous TPA. On the other hand, according to the results obtained by energy dispersive X-ray analysis (EDAX), a uniform distribution of TPA along bead radius was obtained. The secondary electron micrographs of the bead inner shows an sponge-like gel structure.
3.4. Thermogravimetric and differential thermal analysis The DTA of the PVA-PEG beads (Fig. 2a) shows three endothermic peaks at 51, 221 and 317 ~ The first one is associated with the loss of physisorbed water. The other two peaks are assigned to evolution of water and other volatile compounds (aldehydes, ketones and ethers), as a result of the thermal degradation of the PVA and PEG [12]. From the TG diagram, it was calculated that the amount of physisorbed water is less than 11% of PVA-PEG bead weight, and the degradation of the support is almost complete at 500 ~ (95 % weight lost of the total amount). The DTA of TPA-PVA-PEG catalysts (Fig. 2b) presents an endothermic peak at 145 ~ associated with the dehydration of H3PW12Oa0.6H20 phase and another exothermic peak at 580 ~ assigned to the Keggin's anion decomposition. According to Mioc et al. [13], Keggin's anions are transformed at about 600 ~ in a new monophosphate bronze type compound PW8026. The TG diagram shows that the decomposition takes place without appreciable weight loss.
3.5. TPA leaching The amount of TPA removed from TPA-PVA-PEG catalyst washed with toluene, acetonitrile or chloroform, was less than 0.5 %. This indicates that [PW12040] 3- anion is firmly attached to the support. This can be explained assuming that the interaction between the tungstophosphate anion and PVA-PEG support groups is of electrostatic nature due to transfer of TPA protons to the OH groups of the support. However, positive charge may be distributed through conjugate bonds giving, as a result, relatively soft cations. Then, the adsorption might not even be purely electrostatic involving, in addition, interactions of covalent nature in variable degree, as in the interaction between soft cations and anions.
3.6. Acidity measurements by potentiometric titration The curves obtained by titration with n-butylamine for bulk TPA, PVA-PEG beads and TPA-PVA-PEG sample are shown in Fig. 3. A criterion for interpreting the results is that the initial electrode potential (Ei) indicates the maximum acid strength of the surface sites and the range where the plateau is reached (meq/g solid) indicates the total number of acid sites [14]. Bulk TPA presents very strong acidic sites (Ei = 631 mV) (Fig. 3a). TPA-PVA-PEG sample shows acidic surface sites with essentially the same strength (Ei = 640 mV), but a lower total number of acid sites (Fig. 3b), as a result of the lower acidity of the protons engaged to OH groups. On the other hand, the PVA-PEG beads present very weak acidic sites (El =-63 mV).
736
a
b
5
5 ,< (5
(5
s
145 ,
o
'
2~o
'
!
4~o
600
Temperature (~
J
o
t
,
200
I
|
400
I
600
Temperature (~
Fig. 2. TG-DTA of PVA-PEG beads (a) and TPA-PVA-PEG catalyst (b). 3.7. Catalytic activity
The catalytic activity of TPA-PVA-PEG catalyst for the esterification of acetic acid with isoamyl alcohol is presented in Table 1. A high selectivity to the ester was obtained, although traces of 3-methyl-l-butene, 2-methyl-l-butene, and isovaleric aldehyde were detected by means of GCMS. On the other hand, the catalytic activity value obtained with HC1 as catalyst is also shown in Table 1. Besides, a previously prepared catalyst [15], TPA supported on functionalized silica (TPA/SF-T), was tested and the value given in the same table for comparative purpose. a 600
b
_~mm \
mm
|
t
400
200
-200 0,0
i 0,5
,
n,m
~mmmmmmmmm
2o0
Tm ,
\
40o
0
I 1,0
-200 0,0
,
I 0,5
,
, 1,0
....
meq/g TPA
Fig. 3. Potentiometric titration of bulk TPA (a) and TPA-PVA-PEG catalyst (b).
737
Table 1 Catalytic activity of TPA-PVA-PEG, TPA/SF-T and HC1 catalysts Catalyst TPA-PVA-PEG TPA/SF-T
HC1
TON 8.02 10 z 6.96 10 z 4.09 10 z Turn over number (TON): moles of ester formed at 5 hours/moles of H + in the catalyst It is observed that the activity of the TPA-PVA-PEG catalyst is slightly higher than that of TPA/SF-T catalyst. The different acidity of these catalysts may explain the behavior mentioned. Both catalysts have very strong sites, nevertheless the acid strength of the TPA supported on the functionalized silica (Ei = 185 mV) is lower than that of TPA supported on the polymeric material (Ei = 640 mV). Also, the catalysts based on TPA show TON values appreciably higher than that corresponding to HCI. So, the TON value of TPA-PVA-PEG is twice that of the mineral acid. These results are in accordance with the higher acidity of TPA with respect to HCI. On the other hand, the TPA-PVA-PEG catalyst were reused several times without appreciable loss of catalytic activity. It was determined by atomic absorption spectrometry and UV spectroscopy that, during the reaction, leaching of TPA did not occur. Therefore, this type of catalysts can be used in liquid phase reactions without an appreciable loss of TPA, so being attractive for many processes in replacement of conventional homogeneous catalysts at temperatures under the decomposition of support (200 ~ 4. CONCLUSIONS The use of PVA-PEG beads as a support of TPA enables to retain the primary Keggin structure of the heteropolyacid, as seen through the physical-chemical characterizations. A new TPA supported catalyst was obtained, which proved to give a high yield and selectivity in the preparation of isoamyl acetate by means of the liquid-phase catalytic esterification of isoamyl alcohol and acetic acid. A comparison with the yield obtained with a mineral acid, HCI, showed high performance of the TPA-PVA-PEG catalyst. Moreover, it has the advantages of an easy catalyst separation from the reaction medium and lesser problems of corrosion. As a consequence, it leads to an ecofriendly technology for the preparation of isoamyl acetate, which is important for flavor and fragrance industries and another fine chemical products. REFERENCES
1. L.R. Pizzio, C.V. Cficeres and M.N. Blanco, Appl. Catal., 167 (1998) 283. 2. B. Wang and S. Dong, Electrochim. Acta, 38 (1993) 1029. 3. M. Hasik, W. Turek, E. Stochmal, M. Lapkowski and A. Pron, J. Catal., 147 (1994) 544. 4. S.S. Lim, Y.H. Kim, G.I. Park, W.Y. Lee, I.K. Song and H.K. Youn, Catal. Lett., 60 (1999) 199. 5. M. Watase, K. Nishinari and M. Nambu, Cryo Letters, 4 (1983) 197. 6. U. Pr/isse, S. H6rold and K.-D. Vorlop, Chem. Ing. Tech., 69 (1997) 100.
738 7. U. Prfisse, B. Fox, M.F. Bruske, J. Breford and K.-D. Vorlop, Chem. Ing. Tech., 21 (1998) 29. 8. C. Rocchiccioli-Deltcheff, R. Thouvenot, and R. Franck, Spectrochim. Acta, 32A (1976) 587. 9. S. Choi, Y. Wang, Z. Nie, J. Liu and C.H.F. Peden, Catal. Today, 55 (2000) 117. 10. T. Okuhara, T. Hashimoto, T. Hibi and M. Misono, J. Catal., 93 (1985) 224. 11. R. Massart, R. Contant, J.M. Fruchart, J.P. Ciabrini and M. Fournier, Inorg. Chem., 16 (1977) 2916. 12. K.J. Voorhees, S.F. Baugh and D.N. Stevenson, Thermochim. Acta, 274 (1996) 187. 13. J.B. Mioc, R.Z. Dimitrijevic, M. Davidovic, Z.P. Nedic, M.M. Mitrovic and PH. Colomban, J. Mater. Sci., 29 (1994) 3705. 14. R. Cid and G. Pecchi, Appl. Catal., 14 (1985) 15. 15. L. Pizzio, P. Vfizquez, C. Cficeres and M. Blanco, Proceedings ENPROMER 2001, Vol. II (2001) 979.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
Functionalized
SiMCM-41
as
739
support for heteropolyacid based catalysts
L.R. Pizzio, A. Kikot, E. Basaldella, P. Vfizquez, C.V. Cficeres and M.N. Blanco Centro de Investigaci6n y Desarrollo en Procesos Cataliticos (CINDECA), Facultad de Ciencias Exactas, UNLP-CONICET, 47 N ~ 257, 1900-La Plata, ARGENTINA. e-mail: vazquez @dalto n. quimica.unlp.edu, ar Different characterization techniques were used to evaluate the immobilization and acidic properties of Keggin structured tungstophosphoric and molybdophosphoric acids impregnated on an amine-functionalized SiMCM-41 support. The characteristics of prepared catalysts were correlated to the activity in the esterification of acetic acid with isoamyl alcohol to obtain isoamyl acetate. 1. I N T R O D U C T I O N Nowadays, significant changes are taking place worldwide in the area of classical fine chemicals. As a consequence, the development of profitable production of already established fine chemicals can only be achieved with innovative methods that have ecological and economical benefits. In comparison with heterogeneous catalytic systems, homogeneous catalysts often show attractive selectivities under mild conditions and are generally better understood on a molecular level [1]. Nevertheless, their disadvantages, due to the difficult product recovery and environmental pollution as a consequence of harmful wastes, among others, lead to the use of heterogeneous catalysts as a possible and desirable solution. The development of catalysts obtained by means of heteropolyacids (HPA) and related compounds is a very important growing field. Since HPA are less corrosive and produce lower amount of wastes than conventional acid catalysts, they can be used as replacement in ecofriendly processes. Within the HPA, there is a special interest in those which present a Keggin type structure. A disadvantage of HPA as catalysts lies in their relatively low thermal stability. It has been tried to stabilize them by supporting the HPA on several carriers as silica, alumina, titania [2, 3] and functionalized silica [4]. Nevertheless, the leaching of supported heteropolycompounds cannot beforehand be excluded when catalysts are used in heterogeneous liquid reactions. In order to avoid such phenomenon, the behavior of SiMCM-41 supported tungstophosphoric (TPA) and molybdophosphoric (MPA) acids, with Keggin structure, is studied. This material was selected as the host structure because it has been shown to be an excellent support for preparing bifunctional catalysts due to its high surface area coupled with its regular hexagonal array of uniform pore sizes within the mesoporous region. The SiMCM-41 surface was also modified by covalent attachment of primary amino groups through the grafting of 3-aminopropyltriethoxysilane to the siloxy groups.
740
For a better understanding of these supported HPA catalysts, the characterization of the adsorbed species nature by alp NMR, FT-IR and XRD techniques and their acidity by potentiometric titration were performed before and after the leaching with ethanol/water. Besides, the catalytic activity in the esterification of acetic acid with isoamyl alcohol to obtain isoamyl acetate is discussed.
2. EXPERIMENTAL 2.1. Catalyst Preparation Mesostructure synthesis. Pure siliceous, SiMCM-41 (MS) support (SBET: 915 mE/g) was hydrotermally synthesized at pH = 9-10 in our laboratory according to the methodology described in [5]. A solution of commercial waterglass (SiO2, 26.8 % w/w; Na20, 9.2 % w/w; H20, 64 % w/w) was used as the silica source, employing cetyltrimethylammonium bromide (CTABr, 98 %, Aldrich), as the framework structure director. The molar ratio of the starting mixture was 1.89 SiO2:l CTABr:0.738 Na20:0.267 H2SO4:160 H20. The resulting gel was stirred for about 1 h, then it was transferred in a teflon container, and placed in an oven at 100 ~ for 4 days. To provide a control for the pH, the synthesis was carried through to completion with addition of appropriate 1M H2SO4 solution each 24 h. The solid product was recovered and washed by filtration on a Buchner funnel, and dried in air at room temperature (r.t.). The surfactant was subsequently removed from the mesostructure by calcination at 650 ~ for 6 h. Functionalization. MS functionalization was performed by addition of 3aminopropyltriethoxy-silane to a suspension of organic free MS in refluxing toluene and stirred for 5 h. The solid (MS-F) was filtered, washed in a Soxhlet apparatus with diethylether and dichloromethane and dried at 120 ~ [6]. Catalyst Preparation. MS and MS-F were impregnated using the equilibrium adsorption technique at 20 ~ The support (1 g) was contacted for 72 h with 4 ml of a solution obtained by dissolving the corresponding HPA in an ethanol/water (e/w) solvent. The solution concentration was 110 g W(Mo)/1, using Fluka H3PW(Mo)IEO40.nH20 as precursors. The solid was separated from the solution by centrifugation and dried at r.t., thus obtaining samples HPA-MS and HPA-MS-F. In order to evaluate the HPA retention in the mesoporous structure, these samples were leached in e/w, with continuous stirring, for two periods of 24 h (hereinafter named HPA-MS-L and HPA-MS-F-L).
2.2. Characterization Textural properties. The specific surface area (SBET), the pore volume and the mean pore diameter of supports and catalysts were determined by nitrogen adsorption/desorption technique using a Micromeritics Accusorb 2100E equipment. A Bruker IFS 66 equipment, pellets in BrK and a measuring range of 400-4000 cm -1 were used to obtain spectra of solids. A Bruker MSL-300 equipment linked to a "SOLIDCYC.DC" pulse program was utilized to obtain 31p MAS-NMR spectra of solids. The mesostructure supports were characterized by small-angle X-ray scattering (XRD), using a Phillips PW-1714 diffractometer. A small quantity of 0.1 N n-butylamine in acetonitrile was added to a known mass of solid, and shaken for 3 h. Later, the suspension was potentiometrically titrated with the same base at a flow of 0.05 ml/min. The electrode potential variation was
741 measured with an Instrtmaentalia S.R.L. digital pHmeter. Catalytic activity. The esterification was carried out in a three-neck flask (100 ml) equipped with a water-cooled condenser, a thermometer and a glass tube to extract the solution. The reactants were acetic acid (0.1162 mol) and isoamyl alcohol (0.1162 mol); toluene (0.2317 mol) was used as solvent. After the catalyst (0.200 g) was added to the solution, the resulting mixture was heated to reflux and the reaction followed by gas chromatography, using a thermal conductivity detector. The specific conversion was calculated as the molar ratio of the formed ester and the amount of HPA present in the solid.
3. RESULTS AND DISCUSSION 3.1. Support characterization XRD. XRD patterns corresponding to SiMCM-41 before and after functionalization (MS and MS-F, respectively) are shown in Figure 1. It can be seen that MCM-41 was the only phase formed. The pattern shows the strong reflection for the (100) plane of MCM-41 and also well --resolved secondary peaks. The secondary peaks indicate long-range .~" ordering of the MCM-41 structure. Additionally, it can be seen in Fig. 1 -= that organosilane grafting to the mesostructure (MS-F) causes a significant decrease in peak MS intensities. These results could be MS-F attributed to the occurrence of contrast matching between the silica 2 4 6 8 10 framework and the grafted organic 0 2 tetha groups [8]. Fig. 1. XRD patterns of SiMCM-41 before FT-IR. Fig. 2 shows the FT-IR (MS) and after functionalization (MS-F). spectra of MS, MS-F and 3aminopropyltriethoxysilane (ANH). The main difference between the MS-F s~ectrum and that of the MS is due to the presence of small shoulders in the 3000-2750 c m region of the MS-F spectrmn, which can be assigned to the C-H2 carbon-hydrogen stretching of ANH [9]. The ANH used in the graining process presents another strong band in the region 1110-1050 cm1, assigned to the Si-O-C aliphatic groups [9], which is overlapped with a band of MS. It is present as a small shoulder in the functionalized support. r
'
I
'
l
'
I
'
I
'
I
742 On the other hand, the surface area decreases from 915.5 to 307.3 m2/g by the grafting of the MS surface. These results suggest that a noticeable attachment of the basic amino groups and the silanol groups takes place at the working conditions.
Fig. 2. FT-IR spectra of MS, MS-F and ANH. ai
3.2. Catalyst characterization FT-IR. The FT-IR spectra of bulk and supported HPA are shown in Fig. 3. The main characteristic features of bulk TPA are observed at 1081 (P-O), 982 (W=O), 888 and 793 (W-O-W) cm-l (Fig. 3a) and at 1064 (P-O), 962 (Mo=O), 869 and 787 4000 3500 3000 2500 2000 1500 1000 500 (Mo-O-Mo) cm-1 for bulk MPA (Fig. 3b). wavenumber (cm-1) In the HPA/MS and HPA/MS-F spectra, a support band masks the HPA band placed at the 1100-1050 cm-1 zone. Anyway, information can still be obtained from the less masked regions. For MPA/MS, an intensity increase of the band placed at 962 cm1 and a small nonoverlapped band at 869 crnl is observed (Fig. 3b), thus confirming the presence of the undegraded heteropolyanion. TPA/MS catalyst shows the bands at 982 and 793 cm 1 as an increase in transmittance of support bands, while the band at 888 crn-1 is observed without overlapping (Fig. 3a). For TPA/MS-F and MPA/MS-F samples, small bands and shoulders are observed in the 800 - 950 cm-1 zone, which could be possibly assigned to the [PW(Mo)IIO39] -7 lacunar species. The FT-IR spectra of the catalysts, after leaching with ethanol/water (TPA(MPA)/MS-L; TPA(MPA)/MS-F-L) do not present important changes, when they are compared with the spectra before leaching (Figure 3c and 3d). NMR. Fig. 4 illustrates the 31p MAS NMR spectra for MPA and TPA supported on MS. Previously reported bulk acids chemical shills are between -14.8/-15.3 ppm for TPA and 2.9/-4.8 ppm for MPA [2, 3]. In our samples, obtained spectra exhibit one wide line with maximum at -3.6 ppm for MPA/MS and at -15.1 ppm for TPA/MS. These measurements confirm the presence of the acids, in accordance with the FT-IR results.
743
, /.-
TPA. y
" ~ ........
/I\/
-',.
\
MPA
,,.y, '",\ //-, :
//
. ................
-
,
:
i~.
\/
~/
\
1-
k
"7:;" -~'......... -'x
t\ /~J~
.....
I
/ \/
5
a
(1) O
1200 ' 12'00 ' 10~00 ' 8()0 ' 6;0
E (#}
TPNMS-F-L
E fill
'
/ '',,
L
14'oo
400
'
1200
1000
.
\
\.........
':.j'
:./'
\-
\\
.:
'
~2'oo
400
/ \/
I
\ :t
L
/'
C /4'00
600
MPNMS-F-L
i-
TPA/MS-L
800
'
lo'oo
'
8~o
'
6~o
'
!
,
1400
400
-
\
1200
/
1000
800
~venumber
waven umber (cm -1)
600
400
(cm -1)
Fig. 3. FT-IR spectra of MS and MS-F supported TPA (MPA), and the bulk acids (a, b) and the catalysts after leaching with ethanol/water (c, d).
-15.1
-3.6 ppm
MPNMS
,
,
.
.
-100'-80-60-40-20
.
.
0
.
.
20
.
chemical shift (ppm)
Fig.
4.
3Zp
MAS
NMR
ppm
TPNMS
40
.
.
.
6~) &
spectra
.
.
.
.
.
.
.
.
100-100-~0-60-;0-20
0
2=0 gO gO 80
chemical shift (ppm)
for
TPA
and
MPA
supported
100
on
MS.
744 Leaching in ethanol/water. MPA and TPA contents of the catalysts, before (CT) and after (CL) leaching with e/w at 20 ~ are shown in Table 1. Table 1 TPA(MPA) contents in the catalysts before and after the leaching Catalyst MPA/MS MPA/MS-F TPA/MS CT 0.25 0.21 0.16 CL 0.23 0.19 0.13 CT, eL: concentrations in mmol TPA(MPA)/g cat
TPA/MS-F 0.15 0.14
The results show that there is a decrease of the HPA amount for catalysts on MS-F support with respect to MS. For MPA-MS, CT is nearly 18 % higher than in MPA-MS-F; this difference is smaller, around 7 %, for TPA catalysts. This behavior could be due to MS surface area undergoes a high decrease during grafting process, thus diminishing the contact of MPA, or TPA, with MS surface. In addition, the MPA amount in both supports is higher than the corresponding to TPA. In a previous paper, we have reported that the MPA interaction with silanol type groups is stronger than that of TPA [4]. On the other hand, TPA and MPA solubility (So) during the leaching is shown in Table 2. The TPA amount removed from both supports is higher than that of MPA, though for MS-F, the difference between the two So values is lower. These low values are very attractive because they imply that these solids are promissory heterogeneous catalysts to be used in liquid phase. Table 2. TPA(MPA) solubility during the leaching Catalyst MPA/MS MPA/MS-F %So 10.73 7.16 O-foSo--(CT-CL/CT)1O0
TPA/MS 16.74
TPA/MS-F 8.72
Potentiometric titration. The catalyst acidity measurements by means of potentiometric titration with n-butylamine enable the evaluation of the total number of acid sites and their acidic strength [3]. The titration curves obtained for the catalysts, before and after leaching, are shown in Fig. 5. MS and MS-F were titrated in order to compare their acidities, although both carriers exhibit weak sites, MS shows higher acidity than MS-F (Fig. 5a). It is evident from this result that the grafting process attached the amine groups of 3aminopropyltriethoxysilane to the acidic sites of MS. Additionally, MPA and TPA present similar acid strength values when they are supported on MS, 1062 mV for TPA/MS and 1017 mV for MPA/MS (Figs. 5b and 5c). These results can be attributed to the presence of Keggin structures that remain unaltered onto the MS surface. The acidity considerably decreased for TPA when it is supported on MS-F (Fig; 5b). For MPA/MS-F, though the acid strength slightly decreased with respect to the system MPA/MS, the acidity is lower (Fig. 5c). This behavior could be due to different interaction of HPA on the functionalized support with respect to MS as a consequence of the different superficial groups. Another factor would be the Keggin structure degradation to less acidic lacunar species in the catalyst based on MS-F. In addition, this difference in acidity could be assigned
745 to a change of the HPA proton positions. The protons would be localized on the most highly charged oxygen atoms, they could migrate from bridged to terminal oxygens. The acidity remains practically unchanged for the MS supported catalysts after the leaching with e/w (Fig. 5d). A similar behavior was observed for the samples MPA/MS-F and TPA/MS-F. 1250.
a
50
1000
o
100-
~ o
-
%
750 -. 0
150-
o 0
500 200 250-
250 -
~
- 350
0 - Cl
MS-F
300 .
'
.
.
I
.
.
'
.
I
'
I
"~:~-~'"~
-250
'
'
T PA/M S-F '
'1
'
I
1250
LU 1250
ooo
1000.,
750
t~
150 -
MPA/MS-L
500-
o4/
\
-[~
250
'1
] /
I
0,0
'
I 0,5
cb~,.,~_
MPA/MS
"~ MPA/MS-F '
I 1,0
meq/g
'
I 1,5
. '. . .
~_50-
oi :# ~pA/lVlS_F_LT PA/MS-F-L - ~ ~ ~ r P / ~ IVlS"EI' 25
2,0
' 0,0
I 0,5
'
I 1,0
rneq/g
'
I 1,5
"
' 2,0
Fig. 5. Potentiometric titration of MS and MS-F (a), supported MPA (TPA) (b, c) and the catalysts after leaching with e/w (d).
3.3. Catalytic activity The catalyst behavior was tested in liquid medium reaction for the esterification of acetic acid with isoamyl alcohol to obtain isoamyl acetate. This compound is an important product for flavor and fragance industries. The isoamyl acetate can be used as for instance flavor in mineral waters and syrups, perfume in diverse products, as shoe polish, and in the manufacture of materials like artificial silk and other textiles, among other uses.
746 Table 3 Specific conversion of the catalysts Catalyst MPA/MS-L MPA/MS-F-L SC 1.99 1.35 SC (Specific conversion)" tool ester/mmol MPA(TPA)
TPA/MS-L 2.47
TPA/MS-F-L 2.08
The specific conversion (SC) values for MPA(TPA)/MS-L and MPA(TPA)/MS-F-L are shown in Table 3. The TPA catalysts present a SC higher than the corresponding to the MPA system. Nevertheless, the acidity of these systems (Fig. 5d) is similar. This behavior can be explained taking into account that the [PW12040] 3- heteropolyanion presents a higher softness than [PMo12040]3-. In addition to the acidity, the softness of the heteropolyanion is an important characteristic in catalysis and greatly influences the catalytic behavior in organic solution [4]. On the other hand, the catalysts obtained from MS-F, after leaching, showed a slightly lower conversion than those on MS, also washed with e/w. This fact is correlated with the acidity measured by potentiometric titration (Fig. 5a). However, it is important to point out that these catalysts show low TPA(MPA) solubility during the leaching, as above-mentioned. 4. CONCLUSIONS The MPA and TPA based catalysts supported on a new amine-functionalized SiMCM-41 carrier showed important activity in the esterification reaction of acetic acid with isoamyl alcohol. On the other hand, low TPA (MPA) solubility during the leaching was observed. Then, these catalysts are very attractive for their use in liquid medium reactions. REFERENCES 1. M. Belier, Stud. Surf. Sci. Catal., 108 (1997) 1. 2. L. Pizzio, C. C~ceres and M. Blanco, Appl. Catal. A: General, 167, (1998) 283. 3. P. V~zquez, M. Blanco and C. C~ceres, Catal. Lett., 60, (1999) 205. 4. L. Pizzio, P. V~zquez, C. C~ceres and M. Blanco, Proceedings ENPROMER 2001, II (2001) 979. 5. K. Edler and J. White, Chem. Mater., (1997) 1226. 6. M. Lasp6ras, T. Lloret, L. Chaves, I. Rodriguez, A. Cauvel and D. Brunel, Stud. Surf. Sci. Catal., 108 (1997) 75. 7. W. Zhang, M. Froba, J. Wang, P. Tanev, J. Wong and T. Pinnavaia, J. Chem. Am. Soc., 118 (1996) 9164. 8. J. W.Cooper (ed.), Spectroscopic Techniques for Organic Chemists, John Wiley & Sons, New York, (1980).
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
747
Influence of the preparation method on the surface properties and activity of alumina - supported gallium oxide catalysts Alice Luminita Petre, a Bernard Bonnetot, b Antonella Gervasini e and Aline Auroux a a Institut de Recherches sur la Catalyse, CNRS, 2 avenue Albert Einstein, 69626 Villeurbanne Cedex, France; e-mail:
[email protected] b Laboratoire des Multimat6riaux et Interfaces, UMR CNRS 5615, UCB Lyon I, 69622 Villeurbanne Cedex, France c Dipartimento di Chimica Fisica ed Elettrochimica, Universit~ di Milano, via Golgi 19, 20133 Milano, Italy As alumina seems to be the best support for gallium species in terms of activity in the reaction of NOx reduction by hydrocarbons, alumina-supported gallium oxide catalysts presenting similar surface areas were prepared by two different methods, an innovative pseudo sol-gel method and for comparison a conventional impregnation method. The bulk and surface compositions of the samples as well as their structural and textural features confirmed the better dispersion of the pseudo sol-gel sample. The acid-base properties of the samples were determined by adsorption microcalorimetry of ammonia and sulfur dioxide. The sample prepared in non-aqueous mixture showed a more homogeneous distribution of the acid sites, accompanied by an increase of the acid strength and a decrease of the number of basic sites. Moreover the catalytic activity for NOx reduction by ethene in oxygen excess was superior for the sol-gel sample. It is likely that this better activity could be associated with the higher dispersive effect of this synthesis method. 1. INTRODUCTION Gallium-containing materials have received a sustained attention over the past two decades as catalysts for aromatization and/or dehydrogenation of light alkanes. It is only recently that the high activity of supported gallium oxide catalysts in the NO reduction by hydrocarbons has been discovered, making them one of the most effective catalytic technologies for the abatement of NOx in diesel and lean burn engine exhausts [1,2]. The influence of the Ga203 content added to various supports on the catalytic activity for NO reduction by ethene or propene in the presence of high amounts of oxygen has been recently for a series of conventional commercial supports (A1203, SiO2, TiO2) [3]. It has been revealed that Ga203/A1203 exhibits high activity and selectivity with a good tolerance against water in the NO reduction [3,4].
748
The preparation method is decisive for the creation of highly dispersed and coordinatively unsaturated gallium species covering completely the alumina support; these species should play the role of active sites. Moreover, the support can alter the electronic and geometric configuration of Ga sites according to the acidity of the oxide surface binding the metal phase in different ways. It is expected that the physico-chemical properties and the activity of gallium oxide surfaces should be dependent on the type of preparation method. Ga203/AIzO3 prepared by a sol-gel method exhibits a much higher activity for NO reduction by propene than samples prepared by an impregnation method [5]. However, the important differences in surface area of the samples as well as the peculiarities of the preparation methods (co-precipitation or addition of gallium oxide precursor into the aluminium boehmite sol solution) make it difficult to compare the samples and understand the role of dispersion and/or of the presence of a composite oxide [5-71. As alumina was found to be the best support for gallium species in terms of activity towards NO reduction, we prepared alumina-supported gallium oxide catalysts presenting similar surface areas by two different methods, an innovative pseudo sol-gel method and for comparison a conventional impregnation method. In the pseudo sol-gel synthesis, the deposition of gallium a|koxide was obtained from gallium (III) acetylacetonate {Ga(CH3COCHCOCH3)3 or Ga(acac)3}, which was decomposed under strong basic conditions in a non aqueous multidentate Lewis base using published methods [8,9]. In this work the surface properties (dispersion, acidity) of thesupported gallium oxide catalysts and A1203 are presented comparatively, in relation with the preparation method and catalytic activity in deNOx. 2. E X P E R I M E N T A L A series of gallium oxide supported catalysts was prepared by incipient wetness impregnation of a "~-A1203 (surface area 108 m2.g-1, non porous, reference Oxid C from Degussa) with appropriate amounts of an aqueous solution of Ga(NO3)a,9H20. A second type of samples was prepared by a pseudo sol-gel method using the same alumina support. The gallium oxide was obtained from gallium acetylacetonate (Ga(acac)3) decomposed in a non aqueous medium. 5 g of dried ~,-alumina were added to a mixture of 150 mL of dried tetraglyme (methylether of tetraethylene glycol) and 50 mL of 2-propanol in a three necks glass flask under argon atmosphere. The amount of Ga(acac)3 required to obtain the expected GaaO3/A1203 ratio was dissolved in 50 mL of tetraglyme and added to the alumina suspension. A mixture of 6 mL of hydrogen peroxide (50 % solution in water) and 3 mL of concentrated ammonia hydroxide (28 % NH3 in water) was added to the reaction before heating. The mixture was refluxed under argon (the temperature was raised from 378 K to 443 K in order to eliminate the volatile species) for 48 hours under vigorous stirring and then cooled down to room temperature. During the reaction, the acetylacetonate was displaced by ammonia and an exchange of ligand took place leading to a gallium alcoholate with 2-propanol. The crude catalyst was separated from the mixture of solvents by filtration, and the remaining organic parts were removed by washing twice the solid with 50 mL of toluene. No gallium acetylacetonate was remaining in the organic solution. This reaction can be considered as a pseudo sol-gel method, because no gel of
749 gallium isopropanolate could be characterized and because the high temperature reached by the reflux at the end of the synthesis favoured the formation of oxide from alcoholate. The solid was then dried under vacuum up to a temperature of 373 K and until the pressure reached 10"l hPa (around 48 h). Both series of samples were subjected to a further calcination at 773 K under nitrogen flow (6 h) and then oxygen flow (6 h). The bulk and surface compositions of the samples as well as their structural and textural features have been determined by chemical analysis, BET, XRD, TEM and XPS measurements. The acid-base properties of the samples were investigated using adsorption of appropriate probe molecules, namely ammonia and sulfur dioxide, monitored by microcalorimetry. The microcalorimetric studies were performed at 353 K for sulfur dioxide adsorption and at 423 K for ammonia adsorption in a heat flow calorimeter of Tian-Calvet type (Setaram C80), linked to a conventional volumetric apparatus. Before each experiment the samples were outgassed overnight at 673 K. The Ga203-supported catalysts were tested in NO reduction by ethene under lean conditions (NO-C2H4-O2). The catalytic tests were carried out with about 0.1 g of sample placed in a quartz tubular microreactor (5 mm ID). The reactant stream was provided from a set of mass flow controllers (Bronkhorst, Hi-Tee) supplying 3000 ppm of NO and of C2H4 and 40,000 ppm of 02 in helium at a total flow rate of 85 cm 3 minl, with the reactor at close to atmospheric pressure. The contact time was maintained constant at 70 g s~ L -~, corresponding to a space velocity of about 50,000 h "~. The interval of reaction temperature from 473 K up to 823 K was investigated. The exit gas stream from the reactor flowed through an FT-IR gas cell (path length 2.4 m, multiple reflection gas cell) in the beam of a spectrometer (FT-IR from Bio-Rad with DTGS detector). The spectrometer provided analyses for NO, N20, and NO2 for the N-containing species, and C2H4,CO and CO2 for the C-containing species. The measurements were carried out at a resolution of 0.50 crn"~ with an accuracy of + 10 ppm for NO, and + 4 ppm for N20 and NO2, using bands centered at 1876, 2225, and 1619 cml , respectively. The sample was placed in the reactor between plugs of quartz wool and initially pretreated in a 20 % O2/He flow while raising the temperature in stages up to 623 K and maintaining it for 4 h. NO conversion was measured with respect to its initial effective concentration in the gas stream as determined from FT-IR analysis. The amount of NO flowing into the reaction line differed from that introduced in the mixture, due to the presence of high amounts of 02 that transformed part of the NO into NO2. The effective feed gas composition flowing into the reaction line was 2380 pm of NO, 600 ppm of NO2 and some amount of N20. Conversion of NOx (NO plus NO2) corresponded to N2 production. 3. RESULTS AND DISCUSSION
The main physicochemical characteristics of the alumina and of the differently prepared supported gallium oxide catalysts are reported in Table 1. The impregnated sample presented a small decrease in surface area compared to the alumina support. For the theoretical geometry monolayer coverage (approx. 20 wt % Ga203), the surface area was about 99 m2.g~. On the contrary, the surface area of a similar
750 sample prepared by our sol-gel method was higher than that of the support, about 124 m2 g-1. The XRD patterns were the same for the supported samples as for the alumina support. This showed that gallium oxide was deposited in an amorphous phase but did not allow to differentiate the quality of the dispersion. The better dispersion of the sol-gel sample in comparison with the impregnated sample was confirmed by TEM and XPS measurements. Using different probe molecules such as ammonia and sulfur dioxide, the acid-base properties of the samples were determined by adsorption microcalorimetry.
Table 1. Physicochemical characteristics of the alumina and of the differently prepared supported gallium oxide catalysts . . . . . . Sample Ga203 SBET Theoretical Chemisorption uptake weight % /m2.g1 coverage Acidity Basicity /% layer ~tmolNH3.g"l ~tml S02.~."1 7-A1203 108 105 209 Ga-A1 (i)
22.3
99
1.10
123
184
Ga-A1 (sg)
16.7
124
0.83
138
164
(i) impregnation; (sg) pseudo sol-gel
Fig. l a. Differential heats of ammonia adsorption versus coverage for the alumina and the supported gallium oxides. Fig. l a shows the differential heats of adsorption of NH3 at 423 K for the alumina support and for the two differently prepared supported gallium oxides. The alumina and Ga-A1 (i) sample show similar and very high initial heats of adsorption around
751 200 kJ mo1-1. As the coverage of these two samples increases, the initial heats decrease rapidly to reach values of =150 kJ moli and then more slowly but continuously, a result indicative of heterogeneous acidity. The plots of Qd~rvalues v s . coverage for sample Ga-A1 (sg) show the appearance of a plateau, and this profile can be related mainly to a significant homogeneity of the gallium acid sites. The number of acid sites corresponding to a given strength interval is represented in Fig. lb. The energetic distribution of acid sites in Ga-A1 (sg) leaves almost no room for doubt that more strongly acidic sites appear in comparison to Ga-A1 (i).
Fig. 1b. Strength distribution of the acid sites for the alumina and the supported gallium oxides. The increase in acidity upon deposition of gallium oxide on alumina is also confirmed in Table 1 by the increase in NH3 chemisorption uptakes, while the basicity, as measured by the acidic probe SO2, concomitantly decreased. The differences in sulfur dioxide adsorption behaviour between the two samples can be evidenced in Figs. 2a and 2b which represent, respectively, the differential heats of SO2 adsorption v s . coverage and the strength distribution of basic sites. For both samples, the total number of basic sites decreased approximately in the same manner as the acidity increased, the greatest differences in comparison with alumina being observed for the sol-gel sample. The sol-gel preparation allowed intimate contact of the gallium oxide with the support, resulting in a maximized strength (and hence energy) of the interaction. Ga203 deposition contributed to decrease the number of basic sites and increase the number of acid sites of the amphoteric alumina support. The isomorphism between alumina and gallia renders a comparative study difficult, but the observed differences in heats of adsorption for the two samples show that the nature of the interaction between active phase and support, and thus the dispersion, are different, a situation for which the preparation method should be responsible. Indeed, in aqueous solution the gallium oxide deposition occurs preferentially on the strongest OH
752 groups of the alumina support, while in non-aqueous mixture the deposition is more homogeneous on the alumina sites. The sol-gel samples also displayed a greater activity in the catalytic reduction of NO by C2H4 in oxygen excess. The reaction of selective reduction of NOx (SCR) was carried out with C2H4 as reducing species in high oxygen atmosphere (NO-C2H4-O2) working at very high space velocity. Conversions of both NO and NO2 (formed by the homogeneous oxidation of NO in the feed mixture) to N2 were followed together with conversion of C2H4 to CO and CO2.
Fig. 2b. Distribution of the strength of the basic sites for the alumina and the supported gallium oxides.
753 As expected, the two catalysts (Ga-A1 (i) and Ga-A1 (sg)) are active towards NOx reduction at high temperatures [3]. Therefore, we did not observe the typical volcanoshaped curve for NO conversion to Nz that is observed for de-NOx catalysts working at lower temperatures, such as copper-based catalysts [10]. The NOx conversion to N2 as well as the C2H4 conversion were continuously increasing with temperature. Quantitative conversions of NOx as well as of C2I-I4 were not observed even at the highest reaction temperature. Under the severe conditions employed in terms of concentrations and contact time, both catalysts were highly active and selective towards N2 formation. Starting from about 3000 ppm of NOx in the feed, N2 yields of 53 and 7 1 % were obtained at 823 K for Ga-Al (i) and Ga-A1 (sg), respectively. This corresponded to very high specific activities for N2 formation: 4.4 10.4 and 7.7 10.4 mOtN2, s'lmol-]Ga for Ga-A1 (i) and Ga-A1 (sg), respectively. Formation of N20 was detected only at temperatures higher than 723 K, the maximum amount formed being around 140 ppm for both catalysts. It is interesting to calculate the so-called SCR selectivity (SscR, %), defined as the ratio between the amount of C2H4 consumed to reduce NO to N2 and the total amount of C2H4 consumed. The SscR was around 20 % at 723 K and decreased to 14 % at 823 K for both catalysts. These values are among the highest reported in the literature and indicate that gallium oxide species remain unable to oxidize hydrocarbons even at very high temperatures, a very interesting behaviour for deNOx catalysts. Fig. 3 compares the NOx and C2H4 conversions over the two Ga-based catalysts. The results have been plotted as the conversion of NOx to N2 as a function of the extent of the C2H4 conversion, considered to be an index of the extent of the reaction. C2I-I4 is indeed the common species simultaneously able to reduce NOx to N2 while it can also be oxidized by 02 in the parallel side reaction. In this representation, the curve of Ga-A1 (sg) lies above that of Ga-A1 (i). Fig. 4 shows the NO and NO2 concentrations versus reaction temperature for the two catalysts. Starting from 673 K, the NO concentration decreased in a marked way as well as that of NO2, leading to N2 formation. The curve of NO concentration is steeper for Ga-A1 (sg) than for Ga-A1 (i), indicating the superior activity of Ga-A1 (sg). N2 production over the two catalysts is also reported in Fig. 4. 100 --.
80
o :
60
~-
4o
0 z
20
- e - Ga-AI (i) - ~ - Ga-AI (sg)
0
20
40
60
80
100
C2H4 conversion 1%
Fig. 3. Conversion of NOx to N2 as a function of the extent of C2H4 conversion
754 4. CONCLUSION The difference in deNOx activity between the two differently prepared samples further highlights their differences in structure and surface chemistry. The catalytic activity of supported gallium oxide is likely to be governed by the surface concentration ratio of acidic/basic functional groups. The differences in the heats of adsorption of NH3 and SO2 allow some insight into the nature of the interaction and hence the type of surface functional groups with which Ga/O3 interacts; namely in the impregnation method the strong acid sites of the alumina and in the pseudo sol-gel method a much larger population of acid sites.
2500
E
! 80
2000
60
o.
-A-Ga-AI (sg) ; NO
c 1500
-O- Ga-AI (i) NO2 9
50
~. e-
0=
-~-Ga-AI (sg)" NO2 -X-Ga-AI (i)" N2 (%)
40
o
o
L_
c 1000
o
:3
o 30 a.
-X-Ga-AI (sg)" N2 (%)
O C
o o
~
20 z
5O0-
10 0 300
,
,
400
500
0 600
700
800
900
Temperature I K
Fig. 4 - NO, NO2 concentrations and N2 production as ftmctions of reaction temperature for gallium-containing catalysts.
REFERENCES
1. 2. 3. 4. 5. 6. 7.
T. Maunula, Y. Kintaichi, M. Inaba, M. Haneda, K. Sato and H. Hamada, Appl. Catal. B, 15 (1998) 291. M. Haneda, Y. Kintaichi, T. Mizushima, N. Kakuta and H. Hamada, Appl. Catal. B, 31 (2001) 81. A.L. Petre, A. Auroux, A. Gervasini, M. Caldararu and N.I. Ionescu, J. Thermal Anal., 64 (2001) 253. K. Shimizu, A. Satsuma and T. Hattori, Appl. Catal. B, 16 (1998) 319. M. Haneda, Y. Kintaichi, H. Shimada and H. Hamada, J. Catal., 192 (2000) 137. K. Shimizu, M. Takamatsu, K. Nishi, H. Yoshida, A. Satsuma, T. Tanaka, S. Yoshida and T. Hattori, J. Phys. Chem. B, 103 (1999) 1542. Yu.N. Pushkar, A. Sinitsky, O.O. Parenago, A.N. Kharlanov and E.V. Lunina, Appl. Surf. Sci., 167 (2000) 69.
755 R.C. Mehrotra, R. Bohra and D.P. Gaur, "Metal -Diketonates and Allied Derivatives" Academic Press, London 1978. S. Kawagushi, Inorg. Chem. Concept, Vol 11, Springer Verlag (1988), p 89-90. 10. P. Camiti, A. Gervasini, V.H. Modiea and N. Ravasio, Appl. Catal. B, 28 (2000) 175. .
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
Preparation and properties influence of catalyst reduction
of bimetallic
757
Ru-Sn sol-gel catalysts:
J. Hajek 1, N. Kumar 1, H. Karhu 2, L. Cerveny3, J. Vayrynen2, T. Salmi I and D. Yu. Murzin 1. 1Laboratory of Industrial Chemistry, Process Chemistry Group, Abo Akademi, Turku/Abo, Finland ZDepartment of Applied Physics, Laboratory of Electron Spectroscopy and Surface Physics, University of Turku, Finland 3Department of Organic Technology, Institute of Chemical Technology in Prague, Czech Republic Ru-Sn/SiOz catalysts were prepared by the sol-gel method. The influence of the reduction procedure and modification with sodium was investigated. The properties of the chemically reduced catalysts were compared to non-reduced and sodium-modified catalysts. The influence of Ru/Sn metals ratio was also studied. Physical characterization and liquid-phase hydrogenation of cinnamaldehyde demonstrated the high importance of the chemical reduction in the preparation of tested sol-gel catalysts. The highest selectivity to cinnamylalcohol was achieved on catalysts of 5%Ru-2.5%Sn/SiOz type (70%). Sodium modification of catalysts decreased the formation of acid catalysed side products and increased the yield of saturated aldehyde. The hydrogenation properties were dependent on Ru/Sn ratio. 1. INTRODUCTION Selective preparation of unsaturated alcohols from corresponding unsaturated carbonyl compounds is a difficult task to achieve with heterogeneous catalysis. Thermodynamically preferred reduction of double C=C bond can be restricted only with difficulties. Industrial relevance of unsaturated alcohols [1,2] in conjunction with economically expensive methods based on chemical reduction of unsaturated carbonyl compounds calls for the development of new catalysts for highly selective preparation of unsaturated alcohols. Specific hydrogenation features of ruthenium are widely exploited [3,4]. Good intrinsic selectivity for unsaturated alcohols makes ruthenium very attractive for the abovementioned task. Modification of conventional ruthenium catalysts can increase selectivity to the desired alcohols. In the series of reactions, the best result in the formation of unsaturated alcohols was proved over tin doped ruthenium catalysts [5,6]. Corresponding author. E-mail:
[email protected] 758 Sol-gel chemistry is a versatile tool for the preparation of more active and selective catalysts [7,8]. Sol-gel technique was applied for the preparation of a variety of catalysts, including also metals supported on silica and alumina [9,10]. Several papers were dedicated to catalytic hydrogenation over sol-gel catalysts [11,12]. Utilisation of sol-gel technique with respect to the preparation of supported bimetallic catalysts allows to produce catalysts with homogeneous distribution of finely dispersed metals. Another advantage includes improved thermal stability of the metals, higher surface areas, well-defined pore size distribution and ability to control the microstructure of the carrier. Concerning the preparation, the sol-gel catalysts properties can be easily effected by a number of contributing factors comprehended in a synthesis route. One of the steps in the catalyst synthesis is chemical reduction of prepared catalysts. This work is focused on the investigation and improvement of sol-gel Ru-Sn/SiO2 catalysts. Main attention is dedicated to study the chemical reduction benefits involving metals reduction and chemical and physical stabilization. Influence of Ru/Sn ratio and sodium modification on catalysts activity and selectivity will be also discussed. For clarification of the reduction effect, the catalysts were characterized with a wide range of physical methods (BET, PSD, XRF, EA, XPS). Catalysts hydrogenation properties, namely selectivity and activity were compared to Ru/SiO2 catalyst and evaluated during liquid-phase hydrogenation of 3-phenyl-2-propenal (cinnamaldehyde). 2. E X P E R I M E N T A L 2.1. Preparation of sol-gel 5%Ru-5%Sn/SiO2 catalysts A solution of a pertinent amount of ruthenium chloride (RuCl3.x HzO (x < 1), Aldrich) and tin precursor (SnCl2, Aldrich) was stirred for 30 minutes at 333 K in 1,2-ethanediol (p.a). Added molar amount of 1,2-ethanediol was 2/1 compared to carrier precursor (Si(OC2H5)4) molar amount. Tetraethoxysilane (98%, Aldrich) was inserted to the cooled solution of metal precursors under stirring at a room temperature. Acquired mixture was heated to 343 K and stirred at this temperature for 3 hours. Then, a stoichiometric excess of distilled water (90 ml) was added to the solution and the solution was further stirred at 343 K until a gel formed. The produced gels were left to mature for 12 hours. To remove water, solvent and organic residues from preparation, aged gels were dried at decreasing pressure. At the first stage, drying was realized in a water-rotary evaporator. Water bath temperature was at pressure of 1.9 kPa slightly increasing (20 K/hour) to 363 K. The temperature of 363 K was kept for 12 hours. Final stage of gels drying was performed at a pressure of 0.5 kPa (oil-vacuum pump) for 2 hours. Drying temperature at this stage was 473 K. Prospective reduction of catalysts was carried out with 10%-solution of NaBH4 in distilled water. The amount of NaBH4 (97%, Fluka) was selected to apply the following relation: nNaBH4/ (nRu + nsn) = 10. Reduced catalysts were washed several times with small amounts of distilled water and finally with small amounts of ethanol. Washed catalysts were dried for 2 hours under inert atmosphere (N2 4.0, Linde Technoplyn, CR) at the temperature of 473 K.
759 Sodium modification of non-reduced catalysts was carried out with ethanolic-NaOH. Crushed and sieved catalysts were inserted into intensively stirred sodium hydroxide dissolved in ca 50 ml of ethanol. The amount of NaOH was tuned to match the value of nNaOH / (nRu + nsn) = 50. The stirring followed for a couple of minutes, until the coagulation of the catalyst powder was observed. Subsequently the catalyst was carefully washed several times with small amounts of ethanol. The obtained catalyst was dried in nitrogen atmosphere for 2 hours at 473 K. Prior to hydrogenation, prepared catalyst were activated in hydrogen flow at the temperature of 473 K for a period of 2 hours. 2.2. Characterization of
catalysts
Surface areas were determined from nitrogen adsorption-desorption isotherms (Sorptomatic 1900, Carlo Erba Instruments) at 77 K. For the calculation of the specific surface areas the BET method was used. Pore size distribution was obtained by the DollimHeal method. Electrochemical coulometric method was used [13] for the determination of active metal surface. In this method, metals in catalysts were transformed to their oxides by oxidation in air at 473 K before the surface determination. The method is able to determine only a relative surface of metals since it is not possible to verify the values of measured surfaces by another conventional method. XRF analysis was used for the determination of ruthenium, tin, sodium and chlorine content. Solid catalysts samples were examined by automatic sequence RTG spectrometer (ARL 9400 XP). Elements loadings were evaluated by UNIQUANT analyzer. The program used a universal calibration method. Elementary analysis measurements were carried out on an automatic CHN-analyser (Perkin-Elmer 2400, USA). The chlorine content was determined by AgNO3 titration. The exact amount of carbon and chlorine residues in the catalysts was determined from three independent measurements. The activated catalyst sample was examined by ESCA spectrometer (Perkin Elmer PHI 5400, USA). During the analysis an energy pass of 35 eV was used, with pressure lower than 2.7.10 -9 kPa. Binding energy calibration was based on carbon impurity peak at 284.6 eV. Shirley background removal method was applied to remove the background of inelastically scattered photoelectrons. Liquid-phase hydrogenation (H2 4.0, Linde-Technoplyn) experiments were carried out in a 300 ml batch reactor in kinetic regime under 7 MPa of total pressure at 433 K. Typically 3.0 g of cinnamaldehyde (98%, Aldrich) and 0.5 g of catalysts were stirred in 2propanol (p.a.). Total liquid phase volume in autoclave was 200 ml. The products were identified with GC-MS and analyzed by gas chromatography. Hydrogenation samples were analyzed with a gas chromatograph HP-5890 Series II Plus (Hewlett-Packard, USA). Chromatograph was equipped with FID detector and capillary column HP 20M. Content of individual components in the reaction mixture was determined by the Internal Standardization Method (n-decane, Aldrich, USA).
760
3. RESULTS AND DISCUSSION
3.1. Catalysts characterization The catalysts surfaces were strongly dependent on the catalyst type and on metal content. As shown in Table 1, the highest surface areas were by the exhibited non-reduced catalysts. The lowest surfaces areas were observed for the sodium modified catalysts. Generally, the surface area decreased with increased tin content; 5%Ru-5%Sn/SiO2 and 5%Ru-10%Sn/SiO2 showed almost identical results. Pore size distribution of non-reduced and chemically reduced catalysts was well defined. Non-reduced catalysts showed narrow pore distribution of mesoporous character with maximum of pores in the range of 1.5-2.5 nm. Chemically reduced catalysts exhibited a microporous structure with pores of maximum 1 nm. Surprisingly, the pore distribution did not depend on tin content. In the case of sodium modified catalysts, their low surface areas made the evaluation of pore size distribution meaningless. Table 1. Surface area of catalysts. Catalyst
Non-reduced
Reduced
Sodium Modified
5%Ru/SiO2
411
436
121
5%Ru-2.5%Sn/SiO2
333
202
19
5%Ru-5%Sn/SiOz
254
156
17 10.4 mol/1), the same gaman lines (952 and 570 cm ~) are still present but sharper and well defined (Fig. 2-c). No Mo-containing species other than the Anderson-type structure have been detected by LRS. 3.2. Adsorption mechanism Adsorption can result either from electrostatic interactions between molybdates and boehmite surface or from chemical interactions, i.e. from the formation of a iono-covalent bond through a chemisorption mechanism. The interaction mode is governed by the boehmite hydroxyl surface groups as well as by the solution molybdate species. The determination of the nature and concentration of the molybdenum species involved in the experiments (before and after the adsorption equilibrium) has been carried out by computer simulations (cf w 2.2). Concerning hydroxyl surface groups, we referred to MUSIC modeling [11,12] as well as to the work of gaybaud et al. [13], who performed DFT studies on boehmite and so determined boehmite morphological and structural surface properties.
770
3.2.1. Low molybdenum loading 9[AHM]i < 9 1 0 -4 mol/! In solution, at low molybdenum loading and at pH=4.8, the initial AHM entities give rise mainly to the monomer, non or monoprotonated ( M O O 4 2" o r H M o O 4 ) species because of the dilution effect (Fig. 4).
r~
0.
4
0.5
r.t9
04
o
3
c~
Fig. 3. Change in the reduction degree of iron oxides during Cr-Fe-oxide(imp.)= reduction with hydrogen (Step 1) and reoxidation with water vapor Fe-oxide(urea). It is likely that the addition of Cr ions (Step 2). 9 9Cr-Fe-oxide(urea), C)" Cr-Fe-oxide(imp.)and into iron oxides by /x :Fe-oxide(urea). coprecipitation with urea decreases the crystallinity of host oxide. We examined the redox performances of the iron oxide samples added with various metal ions by coprecipitation with urea. Fig. 4 shows average rates of hydrogen consumption in Step 1 at 603 K (left) and those of hydrogen production in Step 2 at 653 K (right) during the repeated cycles. The amounts of added metal cations (Mg, A1, Ca, Sc, Ti, V, Cr, Mn, Co, Ni, Cu, Zn, Ga, Y, Zr, Nb, Mo and Ce) were adjusted to 3 mol% of the total metal cations. The rate of hydrogen consumption and of hydrogen production were estimated from the change in reduction degree from 35 to 50% and from 70 to 55%,
800 respectively. As for the samples with AI, Sc, Ti, V, Cr, Ga, Y, Zr, Nb, Mo and Ce cations, the rates of hydrogen consumption in Step 1 became faster with repetition number, although the rates in the first reduction were slower by the addition of these ions compared to that of Fe-oxide(urea) (unmodified sample). For the reduced samples added with AI, Sc, V, Cr, Ga and Mo, the rates of hydrogen production (Step 2) were generally higher in all the cycles than those of Fe-oxide(urea) and the rates were kept high during 3 cycles. The rates of hydrogen production at the first cycle for the samples added with Ti and Zr were slower than that for Fe-oxide(urea). However, the rates for the samples with Ti and Zr became higher after the second cycle. On the other hand, the addition of Co, Ni and Cu ions into iron oxides increased the rate of the first reduction significantly. However, the reduction rates were reduced considerably with repeated cycles. The rates of hydrogen production (Step 2) for the samples added with Co, Ni and Cu ions also decreased with repeated cycles. On the basis of the results described above, we concluded that the iron oxides added with AI, Sc, Ti, V, Cr, Ga, Zr and Mo ions are effective mediators for storage and recovery of hydrogen.
oxidation at 653 K
reduction at 603 K
lll 1 st cycle 3 rd cycle 2 nd cycle
l
none +Mg
I ,, "IilIII/I//A
+A1
I
I
+Ca +Sc
I/II/)A
//////////~A
+Ti Nil
+v +Cr
"111111///1 , , I/I/////////J
+Mn +Co +Ni
/////A'
I
.1
I
I l/ll//l/l/i
+Ga
I
I
/II/III
"///////A
+Cu +zn
,
I
"7///'//////////J
+v +Zr +Nb +Mo +Ce
"111/111/i//I ~ ~ , ~ 1 , , , I , , ,
-~,,I,,,,I,,,,I
I
,,, ,I, ,,,
0 0.02 0.04 0.06 0.08 0.1 0 0.05 0.1 0.15 0.2 0.25 rate of 142consumption rate of I42 production /mol (mol-Fe)-I min-1 /mol (mol-Fe)-I min-1 Fig. 4. Average rates of hydrogen consumption in Step 1 and hydrogen formation in Step 2 for the iron oxide samples with different metal ions.
3. 3. Effect of addition of different metal ions on the structure of iron oxides Fig. 5 shows the SEM images of Fe-oxide(urea), AI-Fe-oxide(urea) and Cr-Fe-oxide(urea) before the first reduction and after the third oxidation. Al and Cr cations are effective additives for redox performance of iron oxides, as described earlier. The SEM images of the samples before the first reduction showed that all the samples consisted of small particles with uniform size and the particles sizes of AI-Fe-oxide(urea) (Fig. 5 (b)) and Cr-Fe-oxide(urea) (Fig. 5 (c)) were smaller than that of Fe-oxide(urea) (Fig. 5 (a)), suggesting that the addition of Al and Cr cations into iron oxides divides the oxides into small particles. In the SEM image of the Fe-oxide(urea) after the third oxidation (Fig. 5 (a)), larger particles with different size were observed. Therefore, the iron species in Fe-oxide(urea) were sintered during redox cycles. On the other hand, the SEM images of
801
Fig. 5. SEM images of the iron oxide samples. (a), (b) and (c): Fe-oxide(urea), AI-Fe-oxide(urea) and Cr-Fe-oxide(urea) before the first reduction. (a'), (b') and (c'): Fe-oxide(urea), AI-Fe-oxide(urea) and Cr-Fe-oxide(urea) after the third oxidation. I
I
Fig. 5 (b~ and (c') indicated that the particle sizes Fe-oxide ] of iron species in AI-Fe-oxide(urea) and (urea) I! Cr-Fe-oxide(urea) were kept small even after the third oxidation. The changes of particle sizes AI-Fe-oxide ~ J during redox cycles could be also suggested from (urea) ~ " the changes in specific surface areas of the _ samples. Fig. 6 shows specific surface areas of the iron oxide samples before the first reduction Cr-Fe-oxide ' ] and after the third oxidation. The specific surface (urea) ~ ~ I I I, I areas before the first reduction for 0 10 20 30 40 50 A1-Fe-oxide(urea) and Cr-Fe-oxide(urea) were specific surface area / rnz g-1 larger than that for Fe-oxide(urea), which was consistent with the results of SEM images shown in Fig. 5. The redox cycles decreased the specific Fig. 6. Specific surface areas of the iron surface areas of all the samples. However, the oxide samples. I-'-1. before the first surface areas after the third cycle for reduction and m after the 3rd oxidation. AI-Fe-oxide(urea) and Cr-Fe-oxide(urea) were significantly higher than that for Fe-oxide(urea). Thus, we concluded that Cr and A1 cations prevented sintering of iron species during the redox cycles, which is one of the reasons why the redox activities of the iron oxide samples with A1 and Cr cations were kept high during repeated cycles. However, we can not explain the favorable effects of these ions only from suppression of sintering, because the rates of reduction and/or oxidation of the samples with A1, Sc, Ti, V, Cr, Ga, Zr and Mo cations were enhanced with repeated cycles while the specific surface areas became smaller. The addition of these metal ions into iron oxides may produce amorphous compound oxides which activate water and/or
802
hydrogen molecules to accelerate the redox reactions. If the compound oxides between the added ions and iron oxides were formed gradually during the repeated redox cycles, the rates of the redox should be improved with repeated cycles. At the moment, the formation of the compound oxides were not confirmed by XRD analyses of the iron oxides added with AI, Sc, Ti, V, Cr, Ga, Zr and Mo cations during repeated cycles. Further studies about the structure and the state of added metal ions are needed to clarify the effects of these ions on the redox performance of iron oxides. REFERENCES 1. 2. 3.
K. Otsuka, A. Mito, S. Takenaka and I. Yamanaka, Inter. J. Hydrogen Energy, 26 (2001) 191. P.B. Terman and R. Bijetina, Coal Process Technol., 5 (1979) 114. M.E. Dry, J. A. K. du Plessis and G~M. Leuteritz, J. Catal., 6 (1966) 194.
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
803
Catalytic activity of bulk and supported sulfated zirconia Ivo J. Dijs, Leonardus W. Jenneskens, and John W. Geus Debye Institute, Department of Physical Organic Chemistry, Utrecht University, Padualaan 8, 3584 CH Utrecht, The Netherlands To assess whether Lewis acid sites are present on zirconium sulfate, we prepared waterfree bulk zirconium sulfate. Furthermore, a water-free silica-supported zirconium sulfate catalyst was prepared by deposition-precipitation of zirconia on silica and subsequent gasphase reaction with SO3. The activity of these catalysts was compared with that of two conventionally prepared sulfated zirconia catalysts. The different catalysts were extensively characterized. XPS indicated that the conventionally prepared sulfated zirconia catalysts contained sulfuric acid. The activity of the catalysts was determined with the gas-phase trans-alkylation of diethylbenzene with benzene and the solvent-free liquid-phase addition of acetic acid to camphene. Both water-free zirconium sulfate catalysts did not exhibit a significant activity; Lewis acid sites are therefore not active in these sulfated zirconia catalysts. Upon exposure to water vapor the initially water-free catalysts were active. The stability of the conventional sulfated zirconia catalysts appeared to be determined in the gas-phase by the volatilization of sulfuric acid. As a result, a highly porous catalyst was more effective than a catalyst based on zirconia of a relatively low porosity. With liquid-phase reactions extraction of sulfuric acid proceeds leading to an acid liquid, which is catalytically active also after separation of the solid catalyst from the reaction mixture. 1. INTRODUCTION The presence of catalytically active Lewis acid sites in sulfated zirconia catalysts is much debated [1-5]. The conventional preparation of sulfated zirconia catalysts inx~olves reaction of freshly precipitated zirconium hydroxide with diluted sulfuric acid or impregnation of zirconium hydroxide with sulfuric acid or ammonium sulfate [6,7]. The final solid acid catalyst results by calcination at a temperature of 723 to 873 K. Provided thermodynamic equilibrium has been reached, all water and free sulfuric acid should have evaporated upon calcination at 673 to 873 K and only chemically bonded sulfate groups remain [8]. Above 890 K, bulk anhydrous Zr(SO4)2 decomposes [1 ]. When uptake of water by the calcined catalyst is prevented or after loading of the catalyst in the reactor physisorbed water is removed by thermal treatment, only Lewis acid sites are present. Since it is difficult either to prevent the uptake of water vapor or to remove adsorbed water completely, it is difficult to attribute the acid activity of sulfated zirconia catalysts unambiguously to Lewis acid sites.
804 In view of the fact that complete removal of water vapor cannot be readily achieved, we prepared water-free bulk and silica-supported zirconium sulfate. The bulk anhydrous Zr(SO4)2 was obtained by reaction of zirconium tetrachloride with oleum [1]. The silicasupported zirconium sulfate resulted from deposition-precipitation of zirconium hydroxide on silica, calcination at 723 K and subsequent reaction with gaseous sulfur trioxide. The catalytic activity of the sulfated zirconia's was measured in the gas-phase trans-alkylation of benzene (1) with diethylbenzene (2) to ethylbenzene (3, reaction 1) [8,9] and the liquidphase hydro-acyloxy-addition reaction of acetic acid (4) and camphene (5) to isobornyl acetate (6, reaction 2) [8,10]. With the trans-alkylation we used an amorphous silicaalumina catalyst as a reference. For comparison purposes we prepared and investigated also two different sulfated zirconia catalysts prepared conventionally [6,7]. One catalyst was prepared by reaction of 0.5 M H2804 with freshly precipitated zirconium hydroxide and calcination at 773 K [H2SO4/ZrO2(prec.)], the other in a similar way but with calcined zirconia [H2SO4/ZrOz(Gimex)] [8]. It is interesting that it has been concluded from their infrared spectra that addition of water vapor leads to formation of sulfuric acid [5]. As the volatility of sulfuric acid and its constituents is higher than that of metal sulfates, the transport of water and sulfur oxides out of the porous structure of the zirconia is important.
2
*
1
2
(1) 3
H3C~ jCH3 HO..qlfCH3
CH3
+
O 4
?vc. ~
~
~../~O...ff..CH3
~H2CH3 5
(2t
O 6
The difficult transport of strongly adsorbing molecules out of a porous system may give an explanation for the result mentioned in the literature that the reaction of zirconium hydroxide with sulfuric acid leads to a fairly active catalyst, whereas reaction of calcined zirconia with sulfuric acid providing the same sulfur content did not result in an active catalyst [11 ]. Accordingly, the removal of the constituents of sulfinSc acid out of calcined zirconia will proceed much more smoothly than that out of zirconium hydroxide. The activity of catalysts prepared from zirconium hydroxide thus may be due to water and sulfilfic acid remaining in the catalyst providing Bronsted acid sites. The sulfated zirconia catalyst based on non-porous calcinied zirconia [H2SO4/ZrO2(Gimex)], on the other hand, will loose readily most of its sulfuric acid during thermal treatment. Furthermore, we investigated the effect of water on the activity of the above-mentioned sulfated zirconia catalysts and the observed activities were compared. We have extensively characterized the different catalysts by XPS, physical adsorption, analytical electron microscopy, and thermogravimetry.
805 2. EXPERIMENTAL
2.1. Catalyst preparation The water-free bulk zirconium sulfate Zr(SO4)2 was prepared by treatment of ZrCI4 with oleum as published earlier [1 ]. The preparation of the silica-supported, water-free sulfated zirconia catalyst started with the deposition-precipitation of zirconia on silica. In a reaction vessel (2 L) equipped with a pH-meter, thermometer, baffles and a stirrer (1000 rpm), 13.5 g (0.225 mol) SiO2 [Aerosil OX50 (Degussa-Htils), 50 m2/g] was suspended in 750 mL water. Under stirring, both 4.0 M HC1 and 4.0 M NH3 were separately injected via narrow tubes (i.d. 1.0 mm) ending below the level of the liquid using two Gilson Minipuls III peristaltic pumps. Whereas the 4.0 M HC1 was injected at 0.25 mL/min, the injection of the 4.0 M NH3 solution was automatically regulated to maintain a pH of 4.5. When a constant pH of 4.5 was reached, the 4.0 M HC1 was replaced by a 4.0 M HCI (250 mL) solution containing 3.92 g (12.2 mmol) ZrOC12.SH20. After addition of these solutions at a pH of 4.5, the pH was raised to 6.5 with the 4.0 M NH3 solution. The wet residue was re-suspended three times in water (300 mL) for one day followed by filtration in order to remove remaining NH4C1 impurities. The final residue was dried at 393 K for one day and subsequently sieved; the 500-850 m fraction was isolated. Calcination was performed under carefully controlled conditions using a quartz fixed bed reactor (i.d. 10 mm) equipped with a K-type thermocouple. The 500-850 m sieve fraction was calcined at 723 K for 10 h (heating/cooling rate: 5 K/min) in a dry air flow (50 mL/min). On top of the sieve fraction, glass beads (d 1.0 mm) were placed to achieve effective preheating of the air. The next step involved the sulfation of the small zirconia particles. For the oxidation of SO2 to SO3 a layer of a finely powdered 2 wt% Pt/SiO2 catalyst was installed at half-height of the glass bead layer. Helium was used to maintain a gas flow of 50 mL/min. The temperature was raised to 675 K at 5 K/min. Subsequently 5 vol.% of SO2 and 5 vol.% of 02 were added to the helium flow, which was maintained for 5.5 h. Via a heat-traced tube (423 K) the gas-flow leaving the reactor was passed through a stirred suspension of Ca(OH)2 in water. Next, the temperature was lowered to 573 K and the SO2 and 02 were switched off, after which the temperature was decreased to room temperature at 5 K/min. A literature procedure was used for the preparation of calcined H2SO4-impregnated ZrO2 catalysts [6,7]. Two different ZrO2 sources were used. (1) Precipitated ZrO2:20.23 g (62.8 mmol) ZrOCI2.SH20 was dissolved in water (160 mL) and 25 wt% NH3 was added dropwise under vigorous stirring until pH = 8. Next, the NH4C1 was removed re-suspending the wet residue by four times in water (300 mL) for one day followed by filtration. The ZrO2 residue was dried at 433 K for 16 h. (2) Commercial ZrO2 (Gimex B.V., The Netherlands, surface area 60 m2/g, monoclinic). Sulfation of both zirconia's was performed by stirring 4.0 g (32.46 mmol) of each zirconia sample with 20 mL 0.5 M H2SO4 for 3 h and drying at 433 K for 16 h (no filtration). Sieve fractions (500-850 m) were calcined in a flow of dry air (50 mL/min) at 773 K for 3 h (heating/cooling rate: 10 K/min). The catalysts were isolated from the reactor as well as stored under dry air prior to analysis and application.
806
2.2. Catalyst characterization XPS analysis was performed on a Vacuum Generators (Fisons Instruments) MT-500 with a non-monochromatic A1 X-ray source (Ka 1486.6 keV) and a CLAM-2 hemispherical analyzer for electron detection. The samples were supported on carbon adhesive tape. Spectra were corrected for charging using the Si(2p) peak and scaled on the Si(2s) peak. For the determination of the binding energies a background correction was applied. Thermostabilities were determined by analyzing the relative loss of weight in a dry N2 flow (50 mL/min) as a function of temperature and time with a PC-controlled PerkinElmer TGS-2 TGA apparatus, autobalance AR-2. Temperature program: 1 h at 323 K, heating rate 10 K/min to 1123 K followed by 15 rain at 1123 K. Samples of ca. 3.5 mg were used. Transmission electron microscopy was performed with a Philips EM420 and a Philips CM200 equipped with a field-emission gun and an EDAX detector for elemental analysis. Ground and ultrasonically dispersed (in dry n-hexane) samples were brought on copper grids covered by a thin polymer film on which carbon was deposited. SEM analysis was performed with a Philips XL 30 FEG equipped with an EDAX detector for elemental analysis. The samples were supported on carbon adhesive tape and covered with a carbon layer by vapor-deposition. 2.3. Gas-phase trans-alkylation of benzene (1)and diethylbenzene(2) The gas-phase trans-alkylation reaction was performed in an automated micro-flow apparatus containing a quartz fixed-bed reactor (i.d. 10 mm) at 105 Pa [ 16 vol% benzene (1, p.a., dried on molsieve), 3.2 vol% diethylbenzene (2, consisting of 25% ortho, 73% meta, 2% para isomers, dried on molsieve), N2 balance (50 mL/min), WHSV = 1.5 h-1] with 2.0 mL of the tube reactor filled with catalyst particles (500-850 ~tm sieve fraction, typically 1.4 g). Two separate saturators were connected to the inlet of the reactor for the supply of 1 and 2. The partial vapor pressure of 1 and 2 was controlled by adjusting the temperature of the saturator-condensers and the N2 flow rate. After equilibration for 30 min at the applied reaction temperatures (473 K and 673 K, heating rate 10 K/min) within a dry N2 flow (50 mL/min), benzene (1) and diethylbenzene (2) were passed through the reactor. To prevent condensation of both reactants and products prior to GC analysis [Hewlet Packard 5710 A, column: CP-sil 5CB capillary liquid-phase siloxane polymer (100% methyl) 25 m x 0.25 mm, 323 K, cartier gas: N2, FID, sample-loop volume: 1.01 ~tL], tubes were heat-traced (398 K). FID sensitivity factors and retention times were determined using ethene (99.5 %, dried over molsieve) and standard solutions of 1, 2, and ethylbenzene (3, 99%) in methanol (p.a.). The conversion of 2 was measured as a function of time [8]. 2.4. Liquid-phase hydro-acyloxy-addition of acetic acid (4) to camphene(5) A mixture of glacial acetic acid (4 p.a., 0.70 tool), camphene (5 95%, 0.70 tool) and acetic anhydride (p.a., 9.05 mmol)was mechanically stirred (1500 rpm)ovemight at 328 K under a N2 atmosphere. Subsequently, 2.5 g of catalyst was quickly suspended in the reaction mixture. The composition of the soluble fraction of the reaction mixture was analyzed by capillary GC as a function of reaction time; samples were prepared as follows: 1.00 mL of the reaction mixture was added to water (25.00 mL) followed by an extraction with n-heptane (25.00 mL). 1.00 mL of the n-heptane fraction was diluted with n-heptane
807 to 25.00 mL in a volumetric flask. 1.0 txL of the diluted solution was injected into the GC [Varian 3400, column: DB-5 capillary liquid-phase siloxane polymer (5% phenyl, 95% methyl), 30 m • 0.323 mm, temperature program: 5 min at 333 K, 10 K/min to 553 K, 10 min, carrier gas: N2, FID]. In the case of hydro-acyloxy-addition reactions performed in the presence of water, 320 ~tL (17.78 mmol) 1-120 was added instead of acetic anhydride. To establish whether leaching occurs, the insoluble catalyst particles were removed from the reaction mixture by filtration with a double-ended glass filter under a dry N2 atmosphere (before equilibrium had established), whereas the composition of the reaction mixture was further measured as a function of time. Since no solid residue remained after evaporation of the reaction mixture to dryness in vacuo, removal of the solid particles was complete. 3. RESULTS AND DISCUSSION 3.1. Catalyst characterization The chemical composition of the different catalysts investigated are collected in Table 1. Also the sulfur contents calculated for complete conversion to Zr(SO4)2 are indicated. The experimental sulfur contents are lower than the calculated values. The reaction of the silica-supported zirconia with gaseous sulfur trioxide is therefore not complete and the reaction of zirconium hydroxide and zirconia with sulfuric acid involves only a limited fraction of the zirconia. As to be expected, the specific surface area of the catalyst prepared from zirconium hydroxide is much larger than that of the other catalysts. The catalyst based on calcined zirconia exhibited the X-ray diffraction pattern of zirconia and the catalyst based on zirconium hydroxide showed broadened reflection of zirconia. The bulk water-free zirconium sulfate did not display an X-ray diffraction pattern; after exposure to ambient air (relative humidity 50 to 60%) for two weeks the sharp X-ray diffraction pattern of Zr(SO4)2.4H20 appeared [ 1].
Table 1. Quantitative analysis of elements by ICP-AES and Flash-combustion GC for the 10 wt% ZrO2/SiO2 subjected to gaseous SO3 and for the conventional H2SOa/ZrO2 catalysts. Residual atom (%)" oxygen. Sample
S03/ZI02/Si02 H2SO4/ZrO2(prec.) H2SO4/ZrO2(Gimex)
Element
Atom (%)
S Zr Si S Zr S Zr
Calc. ICP-AES / (100% Flash-combustion sulfated) GC 3.0 1.7 1.5 1.3 27.6 23.7 7.3 4.0 23.9 23 9 7.3 18 23.9 24.2
BET area (m2/g)
50
217 50
808 Deposition-precipitation of zirconium hydroxide on silica at a constant pH level of 4.5 leads to very finely divided zirconia. Fig. 1 shows an electron micrograph of the resulting catalyst precursor. Tiny zirconia particles have been deposited onto the non-porous silica spheres. Fig. 1. Transmission electron micrograph of 10 wt.% ZrO2/SiO2 prepared by pH-static deposition-precipitation onto silica [8]. Large light-gray spheres: silica support (Aerosil OX50, Degussa-Htils) Dark dots: zirconia.
Thermogravimetry indicated a continuous weight loss of the sulfated silica supported zirconia of only about 3 %. The thermogravimetric data on the catalysts prepared by reaction of zirconia with sulfia'ic acid were more informative (see also ref. [12]). When the temperature was raised with 10 K/min, the catalyst prepared by reaction with calcined zirconia showed a much more smooth weight loss, which set on already at about 350 K. Apparently, it is much more difficult to remove the constituents of sulfia'ic acid out of the much more porous structure of the zirconium hydroxide. The bulk anhydrous Zr(SO4h catalyst exhibited a ratio of the areas of the S(2p) and the Zr(3ds/2,3;2) peak of 0.50. Employing XPS atomic sensitivity factors of S(2p) = 0.54 and Zr(3dsa,3a) = 2.1, the S/Zr atomic ratio is 2.0, which agrees with the bulk chemical composition. The energy of the S(2p) peak of the silica-supported zirconia after treatment with sulfur trioxide was at 170.0 eV, which agrees nicely with that exhibited by anhydrous bulk zirconium sulfate, which was at 170.3 eV [1 ]. Bulk anhydrous zirconium sulfate has the Zr(3dsa,3t2) peak at 185.6 eV and zirconia at 183.3 eV. The partial conversion of the supported zirconia into zirconium sulfate is not only evident from the chemical analysis, but also from the energy of the Zr(3dst2,3/2) peak, which was at 184.1 eV. Together with the broadening of the peak, which was 0.3 eV, the sulfur-to-zirconium peak ratio being 0.25 instead of 0.5 as measured with the bulk zirconium sulfate, indicates the incomplete reaction of the tiny zirconia particles. It is significant that the zirconia catalysts prepared by reaction of sulfuric acid with zirconium hydroxide exhibit a S(2p) binding energy of 169.3 eV, which is nearly identical to that of liquid H2SO4 (169.4 eV [13]). The Zr(3dst2,3t2) peak of the catalysts is broadened and is positioned at an energy lower than that measured for bulk anhydrous zirconium sulfate. The XPS results therefore point to the presence of
809 sulfuric acid adsorbed on a zirconia surface that has reacted at most to a limited extent to the sulfate.
3.2. Gas-phase trans-alkylation The anhydrous bulk zirconium sulfate preparation did not display any activity in the
trans-alkylation of benzene (1) and diethylbenzene (2) to ethylbenzene (3). At 473 K the silica-supported, gas-phase sulfated zirconia showed a very small activity, which rapidly dropped to a negligible level (Fig. 2). The conclusion is that Lewis acid sites are not active with sulfated zirconia catalysts. The low activity of the silica-supported catalyst is due to adsorption of some water leading to Bronsted acid sites. Desorption of water at 473 K leads to the decrease in activity with time. Pre-hydration of the supported catalyst brings about a slightly higher activity as apparent from Fig. 2; the activity drops again due to the loss of water. Conversion (%)
Fig. 2. Conversion of 2 in the transalkylation of benzene (1) and diethylbenzene (2) at 473 K on SO3/ZrO2/SiO2 catalyst; !"7 before hydration and A after hydration (2 h, 2 % H 2 0 , 50 mL/min).
5
A A 0
A 0
0
A
Zk ~ A
100
/k A
/~
200
300
400
500
600
Time (min) Conversion (%)
Conversion (%)
15 o
15! O
o O O o O
9
[]
O0
0
O O O 0
10
5 []
o []
0
~
0
~
~
|~
100
~
~
~|
~
200
~
~, ~
~
300 Time (min)
~ |~
400
~
~ |~
500
~
~
!
600
~
0
0
100
~
~
~
200
300
400
500
600
Time (min)
Fig. 3. Conversion of 2 in the trans-alkylation of benzene (1) and diethylbenzene (2) on HzSO4/ZrOz(prec.) (left-hand side) and on HzSO4/ZrOz(Gimex) (right-hand side) 1"7 473 K and 0 673 K. Fig. 3 represents the catalytic activity of the two catalysts prepared by reaction with liquid sulfuric acid. The activity has been measured at 473 and at 673 K. In agreement with the result mentioned in ref. [10] that calcined zirconia does not exhibit activity upon reaction with sulfuric acid and calcination, the activity of the HzSO4/ZrOz(Gimex) catalyst
810 at 473 K is low. The activity rapidly decreases with time on stream. At 673 K the catalyst did not show any activity. The catalyst prepared by reaction of precipitated zirconium hydroxide with sulfiafic acid and calcination [H2SO4/ZrO2(prec.)], on the other hand, exhibited a significantly high conversiQn, which did not drop with time on stream. At 673 K, however, the latter catalyst also showed a very low activity. Since an activity decreasing with temperature is unusual, we compared the behavior of the sulfated zirconia catalysts with that of an amorphous silica-alumina catalyst. Fig. 4 shows the conversion of the silica-alumina catalyst at 473 and 673 K. Fig. 4 also indicates that the amorphous silica-alumina catalyst displays the expected dependence of the temperature; at higher temperature the conversion is higher. The higher activity is partly due to reaction to ethene; at 473 K the selectivity to ethylbenzene (3) is 100 %, but 50 % at 673 K. The anomalous behavior of the sulfated zirconia catalysts is due to the loss of sulfuric acid at elevated temperatures. The catalyst prepared from calcined zirconia looses its sulfuric acid at 673 K and, consequently, is not active at this temperature. As a result, decreasing the temperature of the catalyst to 473 K does not restore the activity. Also the more highly porous catalyst prepared from zirconium hydroxide releases sulfuric acid, but in narrow pores some sulfia-ic acid is lett. The loss of sulfitdc acid at 673 K is obviously irreversible. When the catalyst prepared from zirconium hydroxide is, however, kept at 473 K, the transport of water out of the porous structure is thus low that a stable activity is exhibited. Pre-hydration of the bulk anhydrous zirconium sulfate does not provide an active catalyst. That no catalytic activity is induced in this case is due to the fact that bulk anhydrous zirconium sulfate readily reacts to a stable tetrahydrate, viz., Zr(SO4)2.4H20 [1 ]. As a result the hydrolysis of the sulfate by water vapor is suppressed. Conversion (%)
Fig. 4. Conversion of 2 in the transalkylation of benzene (1) and diethylbenzene (2) on amorphous silicaalumina. 1"7 473 K andO673 K.
15
o o 0 0 0 O 0 r 1 6 2
[]
0
I~
o
o |
100
o
o
~ |
200
o
o
o
,
o
0 r
o
o
300
o
,
n
400
o
n
o
irl
500
r
O
[]
n
,
600
Time (min)
3.3. Liquid-phase hydro-acyioxy addition Neither the anhydrous bulk zirconium sulfate nor the silica-supported, sulfated zirconia were active in the addition of acetic acid (4) to camphene (5). The lack of activity is due to the fact that addition of acetic anhydride removes water completely from the reactants. Also the liquid-phase reaction thus demonstrates that Lewis acid sites are not active in our catalysts. Addition of water leads to a well measurable activity with both catalysts. Fig. 5 represents the activity of the silica-supported sulfated catalyst after pre-hydration. The
811 activity is considerable, but a homogeneous catalyst, such as, sulfuric acid or BF3 in acetic acid, is more active raising the conversion to about 70 % in 1500 min. It is interesting that filtration of the catalyst did not stop the reaction. Apparently some sulfuric acid has been formed by reaction with the water, which has been released by the catalyst into the solution. Yield 6 (%) 30
[]
\ 9
.
10 0
0
i
2000
4000 6000 Time (rain)
i
i
8000
10000
Fig. 5. Course of the reaction to isobomyl acetate (6) by the hydro-acyloxy-addition of 0.70 mol acetic acid (4) to 0.70 mol camphene (5) at 338 K (stirred tank reactor, N2 atmosphere). O 2.5 g SO3/ZrO2/SiO2 without H20 and !-i 2.5 g SO3/ZrO2/SiO2 and 320 ml (17.78 mmol) H20. Fig. 6 shows the activities of the sulfated zirconium hydroxide and the sulfated calcined zirconia catalyst. In contrast to the activities displayed in the gas-phase reaction, the calcined zirconia catalyst now shows the higher activity. Since the mass transport in the liquid is much slower, the rate of the reaction is now more strongly transport-limited with the catalyst prepared from the more porous zirconium hydroxide. Upon removal of the solid catalyst by filtration the reaction continues also with these two catalysts. Yielc 6 (%) 3O
0
20 / ~ 10 l
[]
Removalof solid
pirticlesbyfiltration
0
,
0
Fig. 6. Course of the reaction to isobornyl acetate (6) by the hydro-acyloxy-addition of 0.70 mol glacial acetic acid (4) to 0.70 mol camphene (5) at 338 K (stirred tank reactor, N2 atmosphere). El 2.5 g H2SO4/ZrO2(Gimex) and 0 2.5 g HESO4/ZrOE(prec.).
500
,
1000
Time (rain)
|
1500
,
|
2000
812 4. CONCLUSIONS The sulfated zirconia catalysts prepared and investigated in this research do not exhibit activity due to Lewis acid sites both in a gas-phase and in a liquid-phase reaction. The positive effect of water as well as the XPS evidence together with infrared results from the literature suggests that sulfated zirconia catalysts are actually zirconia-supported sulfuric acid catalysts. The fact that sulfi,ic acid is the active component leads to a drawback of sulfated zirconia catalysts. In gas-phase reactions at temperatures where the vapor pressure of the constituents of sulfuric acid is considerable, de-activation of the catalyst has to be taken into account. A highly porous structure can significantly slow down the loss of the active constituent of the catalysts. In the liquid-phase dissolution of sulfuric acid can lead to corrosive properties and to contamination of the reaction products. Furthermore deactivation of the catalyst will eventually result. References
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13.
I.J. Dijs, IL de Koning, J.W. Geus and L.W. Jenneskens, Phys. Chem. Chem. Phys,. 3 (2001) 4423. K. Tanabe, M. Misono, Y. Ono and H. Hattori, New Solid Acids and Bases, Their Catalytic Properties, Elsevier, Amsterdam, 1989. G.D. Yadav and J.J. Nair, Microporous Mesoporous Mater. 33 (1999) 1. X. Song and A. Sayari, Catal. Rev.-Sci. Eng., 38 (1996) 329. F. Babou, B. Bigot, G. Coudurier, P. Sautet and J.C. V6drine, Stud. Surf. Sci. Catal., 90 (1994) 519. K. Arata and M. Hino, Mater. Chem. Phys., 26 (1990) 213. A. Corma, A. Martinez and C. Martinez, Appl. Catal. A, 144 (1996) 249. I.J. Dijs, J.W. Geus and L.W. Jenneskens, J. Phys. Chem. B, submitted for publication. T.-C. Tsai, S.-B. Liu and I. Wang, Appl. Catal. A, 181 (1999) 355. J.O. Bledsoe, Terpenoids, in: J.I. Kroschwitz, M. Howe-Grant (Eds.), Kirk-Othmer Encyclopedia of Chemical Technology, 4th Edition, Wiley, New York, 1997, Vol. 23, pp. 833-882. F.R. Chen, G. Coudurier, J.-F. Joly and J.C. Vedrine, J. Catal., 143 (1993) 616. S. Ardizlone, C.L. Bianchi and E. Grassi, Coll. Surf. A, 135 (1998) 41. D.H. Fairbrother, H. Johnston and G. Samorjai, J. Phys. Chem., 100 (1996) 13696.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
813
New one-step synthesis of superacid sulfated zirconia L. Zanibelli, A. Carati, C. Flego, R. Millini EniTecnologie SpA- Via Maritano, 26 - 1-20097 S. Donato Mil. (MI) - Italy email:
[email protected] A new one-step synthesis of Sulfated Zirconia (SZ) is proposed. After calcination in the range 500-600~ the resulting SZ shows a pure tetragonal phase with S content close to the theoretical value for the monolayer coverage of the surface hydroxyls. The superacidity of SZ has been confirmed by FT-IR analysis in presence of pyridine. After Pt impregnation on SZ, a new bifimctional catalyst is found for isomerisation of n-heptane, active also at low temperature. I. INTRODUCTION Sulfated zirconia (SZ) materials are usually prepared by multi-step synthesis [1,2], involving the preparation of an amorphous hydrous zirconia powder and its contact with a source of sulfated groups, followed by calcination at temperatures varying from 500 to 650~ More recently several new preparation routes of SZ involving one-step synthesis have been described, including a single step in acidic medium [2]. According to Arata and Hino [3], SZ exhibits an acidity 104 times stronger than 100% sulfinqc acid, as determined by Hammett acid function, therefore it is recognised as a superacid. Even if the presence of superacidity can be debated [4], several studies have been devoted to the synthesis of SZ with the aim to study new catalytic applications. The crystal structure of zirconia and the catalytic properties of SZ generally depend on the synthesis method and thermal treatment adopted. In particular zirconia crystallises in three different polymorphs characterised by monoclinic, tetragonal and cubic symmetry. Among them only the tetragonal SZ phase displays significant catalytic properties [5-7]. Unfortunately, the synthesis of the pure tetragonal polymorph is difficult and, in the absence of promoted oxides [8], it could be stabilised only through an accurate control of the synthesis parameters, with particular attention to the thermal treatments. In this work a sol-gel synthesis of sulfated zirconia (SZ) in basic medium is presented. The new synthesis route, claimed in patents [9, 10], permits to obtain in a single step SZ materials that, after calcination in the range 500-600~ show a pure tetragonal phase stabilised by the nano dimension of crystallites. From the earlier sixties the application of SZ to the isomerisation of hydrocarbons (C5-C6 cut) has been extensively investigated [ 1,11-13]. In refinery, with isomerisation processes, normal C5-C6 paraffins are isomerised into their higher-octane branched isomers, then blended into the
814 gasoline pool. The new gasoline requirements for reducing aromatics and benzene, obtained primarily from reforming of C7-C9 fractions, are pushing towards the isomerisation also of C7 cut added to C5-C6 paraffins. With conventional catalysts, the main problem is due to the high tendency towards cracking of paraffins while increasing the molecular weight. The catalyst described in this work shows good performances in n-heptane isomerisation. 2. EXPERIMENTAL
2.1. SZ synthesis SZ1 is considered as a reference of this series. All the other materials have been prepared modifying one synthesis parameter each with respect to SZ1. SZ1 synthesis. A transparent sol is formed by adding 66.0 g of Zr(OC3H7)4 (70 wt % in nPrOH) and 10.0 g of tetrapropylammonium hydroxide (TPAOH, 40 wt % in water) to 364.0 g of n-PrOH. After 2 hours of ageing under stirring, the sol is added with 50 g of aqueous solution of 0.44 M H2SO4. The dense slurry obtained is stirred for 4 hours at room temperature, followed by 4 hours at 60 ~ The sample is dried for 8 hours at 100 ~ under vacuum and calcined at 550 ~ for 5 hours. SZ2 synthesis. The effect of the presence of organic base is studied performing the synthesis of SZ1 without TPAOH. An opalescent sol is formed from the solution of 66.0 g Zr(Of3H7)4, 364.0 g of n-PrOH and 6 g of water. The synthesis continues as for SZ1. SZ3 synthesis. The effect of the presence of complexing agent (acetylacetone, AcAc) during hydrolysis is studied. 0.14 g of AcAc are added to the solution of 364.0 g of n-PrOH and 66.0 g of Zr(OC3H7)4 (70 wt % in n-PrOH). The synthesis continues as for SZ1. SZ4 synthesis: The effect of increasing S content is studied. In the synthesis of SZ1, an aqueous solution 1.32 M HzSO4 is added. Pt-SZ catalysts preparation. 10 g of SZ samples are impregnated using wet imbibition technique with 16 ml of an aqueous solution of HzPtCI6 containing 0.031 g of Pt per ml. The resulting products, named Pt-SZ, are dried at 100~ and calcined at 550 ~
2.2. Physico-chemical characterisation Textural analysis was performed by Nz physisorption at-196~ with a Carlo Erba 1990. The surface area of calcined samples was calculated with B.E.T. method and the pore distribution with Dollymore-Heal model. XRD (X-Ray powder Diffraction) analysis was performed with a Philips X'PERT diffractometer equipped with a secondary monochromator; data were collected in the 159
0 0 80 35
a Mixture (2.5 ml toluene as solvent, 0.7 mmol benzaldehyde 1 and 0.5 mmol reactant 2) was stirred with catalyst at 323 K. See also eqn. 1. b See Table 2. c From ref. [19,20]. Table 2. Results of the Knoevenagel condensation a Product Amount of S iNHz b Pre-activation NH3-treatment Yield / % / p,mol g-1 TF / h -1 Entry Sample temperature / K temperature / K 1 FSMN-1 673 473 4.0 0.528 568 c 2 FSMN-2 873 473 12 7.60 158 3 FSMN-3 1073 473 16 11.2 143 4 FSMN-4 923 923 30 107 28.0 5 FSMN-5 1073 923 35 140 25.0 6d FSMN-5 1073 923 19 140 27.1 7e FSM-16 1073 .... 0.0 0.0 0.0 a Mixture (2.5 ml toluene as solvent, 0.7 mmol benzaldehyde, 0.5 mmol ethyl cyanoacetate and 0.1 g catalyst) stirred at 323 K. Reaction time 30 min. b Values estimated using an absorption coefficient of SiNHz band at 1553 cm 1. Coefficient calculated as molar amount of desorbed NH3 in NH3-TPD above 923 K per decrease amount of the integrated intensity of SiNHz band for FSMN-5 above 923 K. Listed values are refined from values previously published in ref. [17] c Very high TF would be due to underestimation of SiNHz band. d Half amount of FSMN-5 catalyst (0.05g) used. e Reaction time 18 h. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
840 When it is considered that the base-catalysed reaction starts with the formation of carboanion on the methylenic group by the abstraction of proton, the acid dissociation constant (pKa) would be one of the indexes for the difficulty of the reaction. Although there are no good relationship between the product yield and the pKa value, it is clear that this catalyst could not promote the reaction with reactant 2 of high pKa value. The tendency of catalytic property for these four condensations (entry 1-4) is similar to that over 3-aminopropyl-functionalised silica gel catalyst prepared through silylation [18]. 3.2. Effect of p r e p a r a t i o n condition of F S M N on the catalytic activity Table 2 shows the results of the Knoevenagel condensation of benzaldehyde and ethyl cyanoacetate on the FSM-16 and the FSMN samples prepared in various conditions. The reaction occurred on all FSMN samples (entry 1-5), while reaction did not occur over unmodified FSM-16 (entry 7). 1H-NMR and GC did not detect any by-products, confirming the progress of the selective reaction. Moreover, it was confirmed that basic species would not elute during the reaction, since the filtrate after the reaction over active catalyst did not exhibit the activity. The catalytic activity of FSMN increased with increasing the pre-activation temperature for FSM-16 prior to NH3-treatment at 473 K (entry 1-3) or 923 K (entry 4, 5). These results suggest that the sites on which NH3 molecules are stabilised would be generated by evacuation at higher temperature. NH3-treatment temperature much influenced the catalytic activity. Increase of the treatment temperature from 473 K (entry 3) to 923 K (entry 5) enhanced the activity significantly. In the present study, the most active catalyst was FSMN-5 prepared through the pre-activation at 1073 K and the NH3-treatment at 923 K.
~
3600 3400 Wavenumber / cm-1
f I A'T ~ [
16'00 15'00 Wavenumber / cm-1
\\\
'
891[
950 900 850 Wavenumber/ cm-1
Fig. 1. [A] FTIR spectra of FSM-16 after evacuation at 1073 K (a), followed by NH3-treatment at 473 K (b) and 923 K (c). [B] Difference spectra (b) and (c) obtained by subtraction the spectrum Aa from Ab and Ac, respectively. [C] The same curves as [B]. The intensity was normalised by the sample weight.
3.3. The generation of base sites The generation of the base sites on the evacuated FSM-16 was confirmed by FT-IR spectroscopy as shown in Fig. 1. Evacuation at high temperature such as 1073 K gave rise to the absorption band at 891 cm -1 with a small band at 910 cm 1 (Fig. 1Aa). These bands
841 were assigned to the strained siloxane bridge formed by dehydroxylation of the isolated hydroxyl groups (eqn. 2) [14]. Intensities of these bands on FSM-16 increased with increasing the evacuation temperature above 673 K (not shown). Evacuation at high temperature is required for dehydoxylation of isolated hydroxyl groups on the silica surface. This surface strained siloxane was known as reactive sites towards some molecules including NH3 [14]. OH
OH
I
Si
.o
_ H20
J
(2)
'-
Si
Si"
Si
After NH3-treatment at 473 K, the strained siloxane bands at 891 and 910 cm -1 completely disappeared (Fig. lAb), and new absorption bands were generated at 1553 cm 1 (Fig. 1Bb), 3444 cm -1 and very weak band at 3525 cm -1 (Fig. 1Cb). These new bands could be assigned to bending, symmetric stretching and asymmetric stretching vibration of Si-NH2, which were formed by dissociative chemisorption of NH3 on the strained siloxane bridges (eqn. 3) [14]. In the spectrum, no other N-H bands than the SiNH2 bands were observed. At the same time, the intensity of Si-OH band at 3740 cm -1 increased (Fig. 1Bb) by NH3-treatment, indicating that the reaction of eqn. 3 occurred during NH3-treatment at 473 K. ,O
NH 3
NH2
OH
(3)
/
Si"
Si
473 K
Si
Si
When the NH3-treatment was done at room temperature, some bands due to physisorbed NH3 were observed in addition to the bands due to the Si-NH2. In the present study, NH3-treatment was done at 473 K to exclude the physisorption of NH3 that might elute into the solvent during the reaction. Higher temperature NH3-treatment at 923 K resulted in generation of three intense bands at 1553, 3452 and 3540 cm -1 (Fig. 1Bc and 1Cc). These could be also assigned to Si-NH2 bands [14]. Even Si-N stretching of Si-NH2 at 932 cm -1was also observed, which was probably due to the huge amount of Si-NH2 (Fig. 1Ac). It was reported that another reaction would occur to form Si-NH2 (eqn. 4) at high temperature such as 923 K [13, 14]. In the present case, much stronger SiNH2 bands were formed than the case at 473 K. However, Si-OH band intensity did not increase so much (Fig. 1Cc), and the strained siloxane band at 891 cm -1 remained partially (more than a hal0 (Fig. 1Ac). These results suggest that the reaction of eqn. 4 should mainly occur at 923 K than that of eqn. 3. OH
NH 3
]
Si
NH 2 ~
923 K
+
H20
(4)
Si
In these ways, by the NH3-treatment at the both low and high temperature, the Si-NH2 species were formed. However, the wavenumbers of the N-H stretching vibration mode for Si-NH2 obtained at 473 K (3525 and 3444 cm -1 in Fig. 1Cb) were shorter than those
842
obtained at 973 K (3540, 3452 cm -1 in Fig. 1Cc). This would be originated from some interaction between the Si-NH2 and Si-OH moieties in the pair sites formed at 473 K as described in eqn. 3. These FT-IR studies indicate that the reaction of eqn. 3 occurred by NH3-treatment at 473 K, while the reaction of eqn. 4 should mainly occur at 923 K. This means that we can produce two kinds of Si-NHz sites; one is the pair site of Si-NHz and Si-OH, and another is the single Si-NH2 site. Fig. 2 shows the X-ray diffraction patterns of FSM-16 sample and FSMN-5 sample prepared by the pre-activation at 1073 K and NH3-treatment at 923 K. The FSMN sample showed almost the same clear XRD pattern as that of FSM-16, meaning that the structure was maintained even after the treatment at high temperature.
1
20000
Q.. o
_ (b)
(3)
r,o
..= (a) 2
4 6 20 / degree
8
Fig. 2 XRD patterns of FSM-]6 (a) and FSMN-5 (b).
0
50 100 Amount of SiNHz / l~nol
150
Fig. 3 Product yield for Knoevenagel condensation against the amount of Si-NHz sites. Numbers in the graph correspond to the entry number in Table 2.
3.4. Two kinds of base sites
The amounts of Si-NH2 sites were estimated by using NH3-TPD and the intensity of the band at 1553 cm -1 (see footnote text in Table 2) and listed in Table 2. The yield in Knoevenagel condensation increased with increasing the amount of Si-NH2 sites (Table 2), suggesting that Si-NH2 would be the basic active sites catalysing this reaction. Fig. 3 shows the plot of the product yield in Knoevenagel condensation against the amount of Si-NH2. It is noteworthy that the plot gives two straight lines (Fig. 3). NH3-treatment temperature for each line was different: The one with steep slope was obtained by NH3-treatment at 473 K (entry 1-3), while another line was obtained at 923 K (entry 4-6). The activity per amount of Si-NH2 (TF) was apparently larger for the former than for the latter as listed in Table 2. As mentioned above, the pair site of Si-NHz and Si-OH was formed by NH3-treatment at 473 K (eqn. 3), while single Si-NH2 site was formed at 923 K (eqn. 4). The higher TF for FSMN treated at 473 K (entry 2-3) would be attributed to the presence of neighbouring Si-OH with the basic site Si-NH2 (eqn. 3).
843 3.5. High activity of the pair site The pair sites on the FSMN catalyst prepared at low temperature showed higher specific catalytic activity. The role of the Si-OH moiety in the pair site is discussed here. There are alternative possibilities, the direct or indirect participation of the Si-OH moiety to the catalysis. The indirect one is that the Si-OH enhances the base property of the Si-NH2. However, at present, it is difficult to consider a suitable model that could enhance the base property. The direct one, which is claimed in the present study, is the concerted mechanism where the neighboring Si-OH takes part in the catalytic reaction as an acid site as shown in Scheme 2. Angeletti et al. [18] also proposed a similar mechanism to the latter one, i.e., the Knoevenagel condensation over propylamine catalyst bounded on the surface of amorphous silica was promoted by participation of residual surface silanols. In the scheme 2, the surface hydroxy group as a weak acid site activates the carboxyl group in the benzaldehyde. The surface amino group as a base site abstracts the proton from methylenic group of ethyl cyanoacetate to form carboanion. These two activated molecules react more easily to produce the product. In this way, the pair sites would function cooperatively as acid and base sites in the reaction. This might be one of the merits of using the heterogeneous catalyst in comparison with homogeneous system. _
H
\COzEt
?H NHz
/
O
/
\
OzEt
H
CO2Et
~ CO~E~
I I I I I Si I I I I I I I I I Si IIIIII
(~-~/
Pair sites
IIIIIIIIIIIIIIIIIIIIIIII
Si
iH 3
Si
Scheme 2 Proposed reaction mechanisms of Knoevenagel condensation on the pair site consisting of the Si-NH2 and the neighboring Si-OH. The neighboring Si-OH functions as an acid site to activate the carbonyl group.
Both types of FSMN catalysts, prepared by pre-activation at 1073 K followed by NH3-treatment at 473 K and 923 K, had hydroxy groups on the surface with very low density; 0.72 nm -2 and 0.60 nm -z, respectively, which were estimated from the data in the previous paper [15]. On the other hand, the densities of the surface amino groups coexisting on the surface were further low; 0.015 nm -z and 0.098 nm -2, respectively. This means that both catalysts had larger amount of hydroxy groups than the amino groups. However, on the FSMN prepared at high temperature, the Si-NH2 sites and the Si-OH sites are isolated each other since the Si-NHz was generated by substitution reaction (eqn. 4) with the Si-OH isolated originally, in other words, the distance between the Si-NHz and Si-OH is long. On the other hand, the Si-NH2 on the FSMN prepared at low temperature is paired with Si-OH, as expected from both the formation scheme (eqn. 3) and the shift in FT-IR spectrum (Fig. 1Cb). Thus, the higher TF on the FSMN prepared at low temperature would be originated from the shorter distance between the Si-NH2 and Si-OH moieties of the pair site in comparison with those on the catalyst prepared at high temperature. This means that the acid site separately located from the base site could not effectively enhance this catalysis reaction, while the neighbouring acid site would be effective. This would be one of the merits of the pair site.
844 4. CONCLUSION The NH3-grafted mesoporous silica, FSMN, exhibited the base catalytic activity for some reactions. The catalytic active site was clarified to be the surface Si-NH2 group. The highest activity in Knoevenagel condensation was obtained on the FSMN sample that pre-activated at 1073 K and NH3-treated at 923 K since the catalyst possessed the largest amount of Si-NHz. The pair sites of Si-NHz and Si-OH were formed on the FSMN catalyst prepared by the NH3-treatment at low temperature (473 K) while the Si-NHz single site was formed on the catalyst prepared by the NH3-treatment at high temperature (923 K). The former pair sites exhibited higher TFs in Knoevenagel condensation. This FSMN of high TF was prepared by the simple grafting method using surface reaction between molecules and activated silica surface. It is expected that this simple grafting method could be widely applied to the preparation of other functionalised mesoporous or amorphous silica catalysts. REFERENCES
1. 2.
3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
S. Inagaki, Y. Fukushima and K. Kuroda, J. Chem. Soc., Chem. Commun., (1993) 680. J.S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmitt, C. T.-W. Chu, D. H. Olson, E. W. Sheppard, S. B. McCullen, J. B. Higgins and J. L. Schlenker, J. Am. Chem. Soc., 114 (1992) 10834. H. Hattori, Chem. Rev., 95 (1995) 537. K. Tanabe and W. E H61derich, Appl. Catal., A, 181 (1999) 399. J. Weitkamp, M. Hunger and U. Rymsa, Microporous and Mesoporous Mater., 48 (2000) 255. D.J. Macquarrie, Chem. Commun., (1996) 1961. D.J. Macquarrie and D. B. Jackson, Chem. Commun., (1997) 1781. D. Brunel, Microporous and Mesoporous Mater., 27 (1999) 329. B.M. Choudary, M L. Kantam, P. Sreekanth, T. Bandopadhyay, F. Figueras and A. Tuel, J. Mol. Catal. A, 142 (1999) 361. M. J. Climent, A. Corma, V. Forn~s, A. Frau, R. Guil-L6pez, S. Iborra and J. Primo, J. Catal., 163 (1996) 392. S. Delsarte, A. Auroux and P. Grange, Phys. Chem. Chem. Phys., 2 (2000) 2821. S. Ernst, M. Hartmann, S. Sauerbeck and T. Bongers, Appl. Catal. A, 200 (2000) 117. E.F. Vansant, P. Van Der Voort and K. C. Vrancken, Stud. Surf. Sci. Catal., 93 (1995) 383. B.A. Morrow, I. A. Cody and L. S. M. Lee, J. Phys Chem., 80 (1976) 2761. Y. Inaki, H. Yoshida, K. Kimura, S. Inagaki, Y. Fukushima and T. Hattori, Phys. Chem. Chem. Phys., 2 (2000) 5293. Y. Inaki, H. Yoshida and T. Hattori, J. Phys. Chem. B, 104 (2000) 10304. Y. Inaki, Y. Kajita, H. Yoshida, K. Ito and T. Hattori, Chem. Commun., (2001) 2358. E. Angeletti, C. Canepa, G. Martinetti and E Ventullo, J. Chem. Soc. Perkin Trans. I, (1989) 105. R. G. Pearson and R. L. Dillon, J. Am. Chem. Soc., 75 (1953) 2439. S. Singer and E Zuman, J. Org. Chem., 39 (1974) 836.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
845
Titania-silica catalysts prepared by sol-gel method for photoepoxidation of propene with molecular oxygen Chizu Murata, Hisao Yoshida, and Tadashi Hattori Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan Photooxidation of propene to propene oxide (PO) by molecular oxygen was performed over TiO2-SiO2 binary oxides prepared by sol-gel method (TS(s)) and TiO2/SiO2 supported oxides prepared by impregnation method (TS(i)). On both the TS(s) and TS(i) systems, the selectivity to PO increased with decreasing Ti content. Diffuse reflectance UV spectra showed that the samples of low Ti content consisted of the isolated tetrahedral Ti species predominantly, while the [TiOz]n clusters were also formed and became larger with increasing Ti content. Thus, it was concluded that the isolated tetrahedral Ti species were the active sites for PO production. The TS(s) sample of low Ti content exhibited much higher selectivity to PO (60 %) than the corresponding TS(i) sample (40 %) although the dominant Ti species seemed to be the same in both samples. It was confirmed that the consecutive reaction of PO was much suppressed on the sample prepared by the sol-gel method. 1. INTRODUCTION The direct gas phase epoxidation of propene by molecular oxygen has been desired, and has been attempted by many researchers [1-4]. As a new approach, 'photoepoxidation' of propene using only Oz has been investigated over several systems such as TiO2, Ba-Y type zeolite, Nb205/SiO2, MgO/SiO2 and SiO2 [5-10]. But these activities were still low and it was not clear whether the reaction proceeded catalytically or not. In our previous screening study of silica-supported metal oxides for propene photoepoxidation, TiO2/SiO2 showed the highest PO yield [11]. In the present study, we prepared titania-silica catalysts by two kinds of preparation method, the two-stage sol-gel method [12] and the conventional impregnation method, and compared their propene photooxidation activity. 2. EXPERIMENTAL TiOa-SiOa mixed oxide samples were prepared by the sol-gel method consisting of two-stage hydrolysis procedure, by which Ti species are expected to be highly dispersed in silica since the aggregation of Ti is suppressed during the hydrolysis process [12]. A mixture of Si(OCaH5)4, CaH5OH, HaO and HNO3 (0.5, 1.9, 0.5 and 0.043 mol, respectively) was stirred at 353 K for 3 h to hydrolyze Si(OC2H5)4 partially, and the obtained sol was cooled down to room temperature. A 2-propanol (20 ml) solution of
846 titanium isopropoxide (0.0005-0.05 mol) was added to the sol and stirred for 2 h. Then, an aqueous HNO3 solution (H20 and HNO3 were 0.5 and 0.043 mol, respectively) was added to the sol and stirred until the gelation was completed (about 3 - 14 days). The gel was heated up to 338 K at 0.2 K m-in-1 and dried for 5 h. After additional drying for 5 h at 373 K, it was calcined at 773 K in a flowing air for 8 h. Ti content was determined by inductively coupled plasma (ICP) measurement. Titania-silica samples thus prepared by the sol-gel method were denoted as xTS(s), where x (mol% of Ti) = NTi/(NTi + Nsi)xl00. TiOz/SiO2 supported oxide samples were prepared by the conventional impregnation method; amorphous silica (1.0 g) was impregnated with an aqueous solution (50 ml) of (NH4)2[TiO(C204)2]" 2H20, then dried at 383 K for 12 h and calcined at 773 K in flowing air for 5 h. Amorphous silica was prepared from Si(OC2H5)4 by another sol-gel method followed by calcination in a flow of air at 773 K for 5 h [11]. Titania-silica samples prepared by the impregnation method in this way were denoted as xTS(i) similarly to above. The TiO2 sample employed was a Japan Reference Catalyst (JRC-TIO-4; equivalent P-25). The BET surface area of the samples was determined by N2 adsorption. Prior to each reaction test and spectroscopic measurement, the sample was treated with 100 Tort oxygen (1 Torr = 133.3 N m -2) at 673 K for 1 h, followed by evacuation at 673 K for 1 h. The photooxidation of propene was performed with a conventional closed system (123 cm 3). The sample (200 mg) was spread on the flat bottom (12.6 cm 2) of the quartz vessel. Propene (100 l.tmol, 15 Torr) and oxygen (200/Ltmol, 30 Torr) were introduced into the vessel, and the sample was irradiated by a 200 W Xe lamp. After collecting the products in gas phase, the catalyst bed was heated at 573 K in vacuo to collect the products adsorbed on the catalyst by a liquid nitrogen trap. These products were separately analyzed by GC. The results presented here are the sum of each product yield. The reactivity of PO over titania-silica catalysts was tested in two ways. (a)Thermal isomerization of PO; PO (20 [.tmol) was introduced to the reactor in the dark, then the sample was heated at 573 K to collect the products. (b)Photooxidation of PO; PO (20 ptmol) and O2 (200 ~mol) were introduced to the reactor, then the sample was irradiated for 1 h. After collecting products in gas phase, the sample was heated at 573 K to collect the adsorbed products. Diffuse reflectance UV spectra of the samples in vacuo were recorded on a JASCO V-570 spectrophotometer at room temperature. The crossing-point between the base line and the tangential line of the inflection point was employed as the wavelength of absorption edge. Temperature-programmed desorption of NH3 (NH3-TPD) was examined to estimate the amount of surface acid sites. The pretreated sample (200 mg) was exposed to NH3 at 373 K for 0.5 h, and evacuated at 373 K to remove weakly adsorbed NH3. The temperature of the sample increased linearly by 5 K min -1 in a He carrier stream. The amount of desorbed NH3 was monitored with the mass spectrometer at m/e = 16. 3. RESULTS
3.1. Photooxidation of Propene Table 1 shows BET surface area of the SiO2, TiO2, titania-silica samples and the results of photooxidation of propene over them. The specific surface area of each
847 titania-silica and silica samples were similarly high (400-550 m2g'l). The SiO2 sample showed a low conversion as previously reported [10], while the TiO2 sample showed high activity for the complete oxidation of propene to CO2. Partial oxidation of propene took place over all the titania-silica samples, both the TS(i) and TS(s) samples, under photoirradiation. The major products were propene oxide(PO), ethanal, propanal, CO, CO2, acetone and acrolein; and small amount of2-propanol, ethene and butene were also formed. It was confirmed that propene was not converted on the titania-silica in the dark, as typically shown over the 0.34TS(s) sample. Over both the TS(i) and TS(s) systems, the conversion of propene increased with an increase of Ti content, while the yield of PO increased with increasing Ti content up to about 1 mol % and then decreased. Fig. 1 depicts the propene conversion and the selectivity in the photooxidation of propene over the TS(i) and TS(s) systems from Table 1. In the TS(i) system (Fig. l a), the samples containing small amount of Ti (0.01-0.1 mol %) showed high selectivity to PO such as 40 %. Increasing Ti content, the selectivity to PO decreased and the selectivity to COx and propanal increased. Also in the TS(s) system (Fig. lb), the distribution of products changed according to Ti content similarly to that over the TS(i) system. However, comparing at the same content of Ti, the TS(s) system always showed higher PO selectivity than the TS(i) system. Especially, the TS(s) sample containing small amount of Ti (0.08 mol %) showed the best selectivity to PO such as 60 % among all the samples tested in the present study and the previously reported photoepoxidation systems [5, 8-11 ]. Each TS(s) sample showed lower selectivity to ethanal and acrolein than TS(i) sample (Table 1). Table 1. Results of the photooxidation of propene over the SiO2, TiO2 and titania-silica samples. SA a Conv. b PO ~ Selectivity / C% Sample /mEg q / C % /C% PO d propanal acetone acrolein ethanal HC e CO t. SiO2 558 0.7 0.2 22.3 3.5 25.8 15.2 18.1 10.0 5.1 TiO2 34 59.5 0.0 0.0 0.4 0.2 0.0 0.1 0.4 98.8 0.0ITS(i) 544 1.9 0.7 37.0 3.0 15.1 7.6 26.8 2.7 4.9 0.1TS(i) 478 9.1 3.7 40.8 3.3 11.3 4.2 24.5 2.3 10.1 0.5TS(i) 495 17.8 5.4 30.1 11.0 12.4 4.2 20.0 1.8 16.4 1.5TS(i) 473 24.4 4.7 19.2 17.6 9.3 8.7 21.5 2.4 14.1 5.0TS(i) 453 23.8 1.1 4.8 10.2 12.2 2.6 26.4 3.7 37.0 0.08TS(s) 387 4.4 2.6 60.2 2.9 10.4 1.1 17.7 3.0 3.9 0.34TS(s) 423 9.2 5.3 57.5 2.7 5.8 1.2 21.1 5.1 6.6 1.0TS(s) 535 12.5 6.3 50.5 6.2 8.3 1.7 22.1 3.3 7.4 4.1TS(s) 483 21.0 4.5 21.5 19.1 11.4 2.4 24.1 3.4 14.4 8.3TS(s) 416 32.1 1.8 5.7 24.5 13.6 3.1 26.4 2.9 19.1 0.34TS(s) g 423 0.5 trace 5.5 1.6 2.9 0.0 70.0 20.0 0.0 Catalyst 0.2 g, propene 100 tool, 02200 moi, reaction time 2 h. aBET surface area. b Conversion based on propene, c PO yield, d Propene oxide, e Ethene and butenes, f CO and CO2. g In the dark for 2 h at 323 K.
848 40
40
60~TS(s)
601 (a) TS(i) 30~
30 ~40
~ 9
~-
"
2 4 6 Ti content / mol %
o
20"~
R2 ~___.x'-"_~_ . . . . _Y. . . . . . . . . . . . .
0
20 ~~D
N 20 0
8
- >
0 " r -~
0
,
2
,
,
,
4 6 8 Ti content / mol %
Fig. 1 Results of photooxidation of propene over the TS(i) (a) and TS(s) (b) samples. Conversion (C)), and selectivity to PO (O), propanal (A), ethanal (Y) and COx (•
60
40~
40
30-~,, ~ O ~.4020 "~ -~ ,
> .,..q
ID
60. (b) 4.1TS(s) / ~ ~ ~
!40~...
> Io)r
~" i"
30 "~, j
O
- 20
t~
20
10 "~
~20
o 10 "~
ID
0
2 4 6 8 Irradiation time / h
0
10
0 ~ 40 ~
~D
0
:
0
1 2 3 4 Irradiation time / h
0
0 ~
(c) 0.1TS(i)
60
o
- 30
~40
-*..4
>
".~
20
/"
0
O r 20 ~ =o 10 "~ "r
,
~
"Y,,
......---X" . . . . . . . _~
08 10
ethanal (v) and COx (• are plotted.
4 6 8 Irradiation time / h Fig. 2 shows time courses of propene photooxidation over the 0.34TS(s), 4.1TS(s) and 0.1TS(i) samples. Over all these samples, the conversion increased with an increase of irradiation time. With increasing conversion, the PO selectivity decreased slightly, and the selectivity to ethanal and propanal much decreased, while the selectivity to COx increased. This indicated that ethanal and propanal were consecutively oxidized to COx more easily than PO was. Over the 0.34TS(s) sample the PO yield reached 15.7 % at 10 h, while over the 0.1TS(i) sample it reached 9.2 % at 8 h (Fig. 2). Even if all the Ti atoms were assumed to 0
2
-
Fig. 2. Time course of photooxidation of propeneover the 0.34TS(s) (a), 4.1TS(s) (b) and 0.1TS(i)(c) samples. Conversion (o), PO yield ( , ) , and selectivity to PO ("), propanal (zx),
849 be the active sites, the turnover number, TON = (the amount of produced PO) / (the amount of active sites), was 1.4 and 2.8, respectively. This means that the photoepoxidation of propene over titania-silica proceeds catalytically [13]. In all the runs mentioned above, most of products except for CO2 were collected by heating at 573 K in Vacuo. Thus, we examined the desorption temperature of each product over the 0.34TS(s) and 4.1TS(s) samples by heating the catalysts stepwise at 323 K, 373 K, 473 K and 573 K. Over both of the samples, most of PO and ethanal were collected by heating up to 373 K. These results suggest that PO and ethanal weakly adsorbed on the catalysts. On the contrary, propanal was mainly collected by 573 K heating. Propanal adsorbed on the catalyst more strongly than PO and ethanal. 3.2. The reactivity of propene oxide (PO) To know the possibility of the conversion of produced PO, the thermal isomerization of PO and the photooxidation of PO were examined over the 0.34TS(s), 4.1TS(s) and 0.1TS(i) samples (Table 2). By the thermal isomerization of PO (Table 3 (a)), PO was converted mainly to propanal without irradiation. A very small amount of PO was converted over the 0.34TS(s) and 0.1TS(i) samples, while 18.8 % of PO was converted over the 4.1TS(s) sample. By the photooxidation of PO with molecular oxygen (Table 3 (b)), ethanal, acrolein and COx were obtained in addition to propanal over all the samples. These results suggested that propanal was produced by the thermal isomerization of PO in the dark predominantly, and others such as ethanal and CO,, were mainly produced by the photooxidation of PO. The amount of PO converted by the photooxidation, which would correspond to the difference between (b) and (a), was 2.1%, 20.3 % and 11.7 % over the 0.34TS(s), 4.1TS(s) and 0.1TS(i) samples, respectively. Table 2. Reaction of propene oxide (PO) over the representative titania-silica catalysts. Yield/C% b Method a Sample
propanal acetone acrolein ethanal alcohols HCCCOx d
Total yield / C%
(a) 0.34TS(s) 1.7 0.1 0.1 0.1 0.0 0.2 0.1 2.3 (a) 4.1TS(s) 14.3 0.7 0.0 0.0 0.0 1.1 2.7 18.8 (a) 0.1TS(i) 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.2 (b) 0.34T S (s) 1.3 0.2 0.0 1.4 0.0 0.2 1.3 4.4 (b) 4.1TS(s) 18.4 2.6 2.4 6.1 3.5 1.3 4.8 39.1 (b) 0.1TS(i) 1.5 0.6 1.5 2.8 0.7 0.5 4.3 11.9 Catalyst 0.2 g, propene oxide 20 lamol, a (a) The thermal isomerization of PO and (b) the photooxidation of PO. See text. bBased on PO. c Propene, ethene and butenes, d CO and COz.
850
3.3 Characterization
of the titania-silica
samples
Fig. 3 shows diffuse reflectance UV spectra of the samples. The TiO2 sample showed a large absorption band below 380 nm. The SiO2 -,,, \ sample scarcely showed absorption. The titania-silica samples of low Ti content less than 0.4 mol %, in both \ ,, \,, ,.,,,, \ the TS(i) and TS(s) systems, \ ',\', ,\, exhibited a narrow absorption band ~ 2. ~,, below 250 nm, which was assigned to the LMCT (ligand-metal charge transfer) from O to Ti of isolated tetrahedral Ti species [14, 15]. The samples containing more than 0.4 ' 315 ' I ' ~1 mol % of Ti in both the TS(i) and 2q)0 250 waveleng~... / nm 350 400 TS(s) systems showed the additional Fig. 3 Diffuse reflectance UV spectra of SiO2 (---), absorption in the region above 250 TS(i)(----), T S ( s ) ( ~ ) and TiO2(. . . . ). The samples nm. The absorption edge was shifted were evacuated at 673 K. SiO2 (a), 0.0ITS(i) (b), to longer wavelength with increasing 0.08TS(s) (c), 0.1TS(i) (d), 0.34TS(s) (e), 1.0TS(s) ~, Ti content and became close to that 0.5TS(i) (g), 4.1TS(s) (h), 1.5WS(i)(i), 8.3TS(s) ~,'), of the TiO2 sample. The absorption in 5.0TS(i) (k) and TiO2 (/). the 250 - 330 nm region are assigned to the [TiOz]n clusters, and it is generally known that the absorption edge is shifted to longer wavelength as the size of the [TiOa]n clusters become larger [14, 15]. From these results, it was indicated that both the TS(s) and TS(i) samples containing less than 0.4 mol o% of Ti consisted of the isolated tetrahedral Ti species predominantly. In the titania-silica samples containing more than 0.4 mol % of Ti, the [TiOz]n clusters were also formed, and the size of the [TiOz]n clusters become larger with increasing Ti content. Fig. 4 shows the relationship between the Ti content and the absorption edge in UV spectra. In the region of more than 0.4 mol % of Ti content, the TS(s) system showed the absorption edge of shorter wavelength than the TS(i) system. This indicates that the TS(s) system have more dispersed Ti species than the TS(i) system when they were compared at the same Ti content. Fig. 5 shows the NH3-TPD profiles of the 0.34TS(s), 4.1TS(s) and 0.1TS(i) samples. The 4.1TS(s) sample showed 54.7 ~tmol g-1 of desorbed NH3, which corresponds to about 8 % of Ti atoms in this sample. On the other hand, the 0.34TS(s) and 0.1TS(i) samples showed very small amount of desorbed NH3, which means that both of them have almost no acid site.
,,,
851
400
J
E q) ca3
0.8 ea3 o
350
0.6
= 300 .2
=~" 0 . 4 -
.o
(b)
.,.a
o 250
o 0.2
,.o
200
0
2 4 6 8 Ti content / mol % Fig. 4. Relationship between the Ti content and the absorption edge in UV spectra of TS(s) (a) and TS(i) (b).
Z
400
500 600 700 Temperature / K Fig. 5. NH3-TPD profiles of 0.34TS(s) (a), 4.1TS(s) (b) and 0.1TS(i) (c).
4. DISCUSSION 4.1. The structure of Ti species and the acidity over titania-silica samples When Ti content was very low (< 0.4 mol %), the titania-silica samples consisted of the isolated tetrahedral Ti species predominantly, regardless of the preparation method (Fig. 3). In the titania-silica samples containing more than 0.4 tool % of Ti, the [TiO2]n clusters were also formed, and the clusters became larger with increasing Ti content (Fig. 3). The TS(s) samples contained more dispersed Ti species than the TS(i) samples when they were compared at the same Ti content (Fig. 4). By the sol-gel method employed in the present study, the aggregation of Ti atom would be reduced in comparison with the conventional impregnation method, as expected. The 0.34TS(s) and 0.1TS(i) samples predominantly having the isolated tetrahedral Ti species possessed almost no acid site, while the 4.1TS(s) sample containing the [TiOz]n clusters 60~oCk(a had acid sites (Fig. 5). It is known that Ti-O-Si ~" ) bridges where the Ti atoms reside in ~ 40 pentahedral or octahedral sites cause the charge i~ A"~ I imbalance and generate Bronsted acid sites [14] 9 ~5 Thus, the acid property of the 4.1TS(s) sample ~ 20 would be due to the presence of [TiO2]n clusters, which contained the octahedral or pentahedral 0 I I - ~ Ti moieties 9 200 250 300 ~ 3 5 0 400 absorption edge / n m 4.2. The effect of Ti content on the Fig. 6. Relationship between the PO photocatalytic activity selectivity and the absorption edge in UV On both the TS(s)and TS (i) systems, the spectra of the TS(s) (a) and TS(i) (b) PO selectivity increased and the selectivity to samples. propanal and COx decreased with decreasing Ti
852 content (Fig. 1). As mentioned above, the dispersion of Ti species varied with Ti content. Fig. 6 shows the relationship between the PO selectivity and the absorption edge of UV spectrum of each sample. The selectivity to PO increased when the absorption edge was shifted to shorter wavelength. This means that more dispersed Ti species show higher PO selectivity. The 4.1TS(s) sample showed higher selectivity to propanal, ethanal and COx than the 0.34TS(s) sample at similar conversion (Fig. 2). This result would be explained by the result of PO conversion shown in Table 2. PO was not so much converted over the 0.34TS(s) sample, while a large amount of PO was converted into propanal over the 4.1TS(s) sample in the dark (Table 2(a)). Generally, it is known that PO is isomerized into propanal on acid sites, and into acetone on basic sites [16]. Thus, it was suggested that a part of PO produced on the 4.1TS(s) sample in the photooxidation of propene cannot be desorbed as PO, and was converted into propanal due to the acid sites arising from the [TiOz]n clusters. In addition, much larger amount of PO converted to ethanal and COx over the 4.1TS(s) sample than over the 0.34TS(s) sample under photoirradiation in the presence of molecular oxygen (Table 2(b)). The [TiOz]n clusters promoted not only the thermal isomerization of PO to propanal, but also the consecutive photooxidation of PO during the propene photooxidation. As a conclusion of this section, it is found that the isolated tetrahedral Ti species on SiO2 are effective for propene photoepoxidation, while the [TiOz]n clusters for other products such as propanal, ethanal and COx production. 4.3. The effect of the preparation method on the photocatalytic activity The TS(s) system showed higher selectivity to PO than the TS(i) system at any Ti content (Fig. 1). In Fig. 6, the plots of the TS(s) and TS(i) samples seem to be on the common curve in the region from 270 to 380 nm of the absorption edge. This means that PO selectivity would be determined by the dispersion of Ti species regardless of the preparation methods in this region. On the other hand, in the region less than 250 nm of the absorption edge, although the samples in the both systems consisted of the isolated tetrahedral Ti species predominantly, the plots of the TS(s) system showed higher PO selectivity than the TS(i) system (Fig. 6). The 0.1TS(i) sample showed higher selectivity to ethanal, acrolein and COx than the 0.34TS(s) sample (Table 1). This result agrees with the result of photooxidation of PO; more PO converted to ethanal, acrolein and COx over the 0.1TS(i) sample than over the 0.34TS(s) sample (Table 2(b)). The yield of propanal converted from PO was similar over both samples (Table 2), and the 0.1TS(i) sample showed almost no acid sites similarly to the 0.34TS(s) sample (Fig. 5). Therefore, the conversion of PO over the 0.1TS(i) sample should not be attributed to the acidity arising from [TiOz]n clusters. Although no evidences were obtained to clarify the structural differences between the TS(s) and TS(i) samples containing a small amount of Ti, some possibility can be considered. One is that the local structure of the isolated tetrahedral Ti species may be different. For example, Lamberti et al. [17] suggested the existence of two kinds of the isolated tetrahedral Ti species in TS-1; [Ti(OH)(OSi)3] and [Ti(OSi)4]. As a conclusion of this section, it should be noted that the sol-gel method gave more selective catalysts for propene photoepoxidation than the impregnation method.
853 5. CONCLUSION The titania-silica catalyst of low Ti content prepared by the sol-gel method exhibited the highest selectivity to PO in the photooxidation of propene by molecular oxygen. The selectivity to PO decreased with increasing Ti content. It was concluded that the isolated tetrahedral Ti species were active sites for propene photoepoxidation, while the [TiOz]n clusters for the formation of propanal and CO2. Comparing samples TS(s) and Ts(i) which exhibited similar UV absorption band attributed to the isolated tetrahedral Ti species, the TS(s) sample showed much higher selectivity to PO, up to 60 %, than the TS(i) sample. It was found that the sol-gel method could provide effective titania-silica catalysts for the photoepoxidation of propene by molecular oxygen. ACKNOWLEDGEMENT This work was supported by a grant-in-aid from the Japanese Ministry of Education, Science, Art, Sports and Culture, and by Nippon Sheet Glass Foundation for Materials Science and Engineering. REFERENCES 1. Y. Wang and K. Otuka, J. Catal., 157 (1995) 450. 2. T. Hayashi, K. Tanaka and M. Haruta, J.Catal., 178 (1998) 566. 3. G. Lu and X. Zuo, Catal. Lett., 58 (1999) 67. 4. K. Murata and Y. Kiyozumi, Chem. Commun., (2001) 1356. 5. E Pichat, J. Herrmann, J. Disdier and M. Mozzanega, J. Phys. Chem., 83 (1979) 3122. 6. E Blatter, H. Sun, S. Vasenkov and H. Frei, Catal. Today, 41 (1998) 297. 7. Y. Xiang, S. C. Larsen and V. H. Grassian, J. Am. Chem. Soc., 121 (1999) 5063. 8. T. Tanaka, H. Nojima, H. Yoshida, H. Nakagawa, T. Funabiki, and S. Yoshida, Catal. Today, 16 (1993) 297. 9. H. Yoshida, T. Tanaka, M. Yamamoto, T. Funabiki and S. Yoshida, Chem. Commun., (1996) 2125. 10. H. Yoshida, T. Tanaka, M. Yamamoto, T. Yoshida, T. Funabiki and S. Yoshida, J. Catal., 171 (1997) 351. 11. H. Yoshida, C. Murata and T. Hattori, J. Catal., 194 (2000) 364. 12. R. Lange, J. Hekkink, K. Keizer and A. Burggraaf, J. Noncryst. Solids, 191 (1995) 1. 13. H. Yoshida, C. Murata and T. Hattori, Chem. Commun., (1999) 1551. 14. X. Gao and I. E. Wachs, Catal. Today, 51 (1999) 233, and references therein. 15. S. Bordiga, S. Colucia, C. Lamberti, L. Marchese, A. Zecchina, E Boscherimi, E Buffa, E Genomi, G. Leofanti, G. Petrini, and G. Vlaic, J. Phys. Chem., 98 (1994) 4125. 16. Y. Okamoto, T. Imanaka and S. Teranishi, Bull. Chem. Soc. Jpn., 46 (1973) 4. 17.C. Lamberti, S. Bordiga, D. Arduino, A. Zecchina, E Geobaldo, G. Span6, E Genoni, G. Petrini, A. Carati, E Villain and G. Vlaic, J. Phys. Chem. B, 102 (1998) 6382.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
855
Preparation of large surface area MnOx-ZrO2 for sorptive NOx removal M. Machida, M. Uto and T. Kijima Department of Applied Chemistry, Faculty of Engineering, Miyazaki University Gakuenkibanadai-nishi, Miyazaki 889-2192 Japan
Surfactant-controlled coprecipitation was applied to the preparation of large surface area MnOx-ZrO2. Coprecipitation in the presence of cationic surfactants, CTAB or CTAOH, and subsequent heating in air yielded homogeneous noncrystalline mixtures of Mn203 and ZrO2. The resulting porous materials with large surface areas in the range 200-300 m2/g were found to be useful for oxidative adsorption of NO. For this application, precursors and surfactants containing halide ions (C1- or Br-) should be avoided, because these species are strongly bound to surface sites for the oxidative NO adsorption. 1.
INTRODUCTION
A number of metal oxides show various interactions with NO, which result in adsorption, absorption and solid-gas reactions [1]. Recently, binary oxides containing Mn and tetravalent ions (Zr or Ce) have been reported to show the high ability to adsorb gaseous NO in the presence of Oa at lower temperatures [2-7]. The binary oxides are effective in promoting oxidative NO adsorption to form nitrate and/or nitrite species on the surface. In the previous reports, we have pointed out that the NOx adsorbing materials are applicable to not only adsorptive NOx removal, but also catalytic deNOx by loading noble metals. For the Pd-loaded MnOx-CeOz, hydrogen activated on Pd is readily spilt over onto the MnOx-CeO2 surface and reduce NOx (NO2-or NO3-species) adsorbed thereon to Na [8,91. Because the amount of NOx adsorption on these materials is less than surface monolayer, preparation of larger surface area materials is requested to increase the NO uptake as well as the catalyst efficiency. MnO~-ZrQ is more promising in this regard. The larger surface area, ca.150-180 m/g, was obtained for noncrystalline MnOx-ZrO2 materials, compared to ca. 60-70 m2/g for MnOx-CeO2 with a fluorite-type crystalline phase. Due to the noncrystalline nature, the surface area of MnOx-ZrO2 is strongly sensitive to the preparation route. One possible method to achieve large surface areas is surfactant-template precipitation. In the present work, preparation of MnOx-ZrO2 fine particles was studied by the use of coprecipitation in the presence of cationic surfactants, which would cause electrostatic interactions with Mn/Zr hydroxide ions. The thermal stability and NO adsorbability of as prepared oxides were also evaluated. 2. E X P E R I M E N T A L 2.1
Sample preparation MnOx-ZrOz was prepared from corresponding metal chlorides or nitrates as shown
856 surfactants, CTAB (cethyltrimethylammonium bromide) or CTAOH (cethyltrimethylammonium hydroxide), were dissolved in distilled deionized water. To the solution was added an aqueous solution of NaOH or NH4OH under vigorous stirring. The pH of the resultant suspension was controlled from 5 to 13. After subsequent aging at 75-90 ~ a solid product was centrifuged and washed with water, and finally calcined at 450 ~ in air. The sample prepared using different sources and surfactants is designated as X-Y, where X is NO3 (nitrates) or C1 (chlorides), and Y is B (CTAB) or OH (CTAOH) as listed in Table 1. MmOx-ZrO2 was also prepared by conventional coprecipitation from an aqueous nitrate solution (NO3-1). An aqueous solution of dinitrodiamine platinum nitrate was impregnated onto as prepared binary oxides and calcined at 450 ~ for 5 h (1.0 wt% loading as Pt). As prepared powder samples were pressed and crushed into 20 mesh granules before use for adsorptive NOx removal.
I
Chlodde or nirate Na0H or NH3OH CTAB or CTAOH stirring
I
I,Mn-Zr, hydr~
:1
rinsing
I,
dr~ng under reduced pressure ~calcination in air
! Mn~176
!
Fig. 1. Diagram of surfactantcontrolled coprecipitation of 2.2 Characterization MnOx-ZrO2. Crystal structure of calcined samples was determined by powder X-ray diffraction (XRD, Table 1 Samples prepared in this study Shimadzu XD-D 1) using monochromated CuKa sample source surfaetant base radiation (30 kV, 30 mA). The BET surface area was obtained by measuring N2 adsorption NO3-1 nitrate none NH 3 isotherms a t - 1 9 6 ~ The XPS measurement CI-Br chloride C T A B NaOH was performed on a Shimadzu-Kratos AXIS-HS NO3-Br nitrate CTAB NaOH spectrometer with a magnesium anode ( M g K ) NO3-OH nitrate C T A O H NaOH operated at 15 kV and 10 mA. DRIFT spectra of NO3-OH-2 nitrate CTAOH NH3 NOx species adsorbed onto MnOx-ZrO2 were ' recorded on a Jasco FT-IR610 spectrometer. A temperature-controllable diffuse reflectance reaction cell (Jasco DR600A) was connected to a gas flow system and a vacuum line. The sample was heated in a stream of 20vol%O2/He at 400 ~ for 1 h and then exposed to the reaction gases containing 0.08vol%NO, 2vo1%O2, and He balance at 25 ~ for 30 min. After the treatment spectrum was recorded in a flowing He at ambient temperature.
2.3 Adsorptive NO removal The sorptive NO removal was carried out in a conventional flow system at atmospheric pressure. Gas mixtures of 0.08 vol% NO and 10% 02, balanced with He were fed to the granular sample (0.2 g) at W/F=0.24 s-g" cm ~. The effluent gas was analyzed by an on-line gas chromatography (TCD) with molecular sieve-5A and Porapak-Q colunms, and a chemiluminescence NOx analyzer.
857 3. RESULTS AND DISCUSSION 3.1
Phases of MnOr-ZrOz
Coprecipitation is the most widely used procedure for preparing a precursor of mixed oxides. Its application to the present system caused noticeable difference in the phase of as calcined oxides with different compositions. Fig. 2 shows powder X-ray diffraction patterns of (n)MnOx-(1-n)ZrO2 prepared by coprecipitation from nitrate sources and subsequent heating at 450 *C. The diffraction patterns at n=0 and 1.0 consisted of tetragonal/monoclinic ZrO2 and Mn203, respectively. However, the increase of n converted the crystalline phases into noncrystalline products at 0.210 by aging at higher composition temperatures (90 ~ The largest surface area in the present study, 300 mZ/g, was obtained by using nitrate sources in the place of chlorides (NO3-Br). We believe that this is one of largest values for Mn-oxide materials reported so far. It was found that CTAOH is also applicable to obtain the large surface area products (NO3-OH). This is t-
859 important because the halide-flee synthesis is essential for the preparation of NOx adsorbing material as described below. 3.3
Thermal stability of MnOx-ZrO2 Since the large surface area of MnOx-ZrO2 is due to their noncrystalline fine particles of Mn203 and Zr02, thermal stability should be a key point to be evaluated before practical uses. The effect of calcination temperature on the C1-Br-2 product was examined. Fig. 6 represents the XRD diffraction patterns after calcination at elevated temperatures. Clearly, significant crystallization of tetragonal ZrO2 took place at 450 ~ caused the considerable sintering as a result of crystallization of ZrO2. The sintering of this material caused complete deterioration ofNOx adsorbability [3]. Since a similar result was obtained for the oxides synthesized by conventional coprecipitation (NO3-1), the thermal stability of MnOx-ZrO2 cannot be improved by modifying the preparation route. The use of this material must be limited below 450 ~ 3.4
NO adsorption property The sorptive NO uptake of as prepared MnOx-ZrO2 is summarized in Fig. 8. On contrary to the improved surface area, both CI-Br and NO3-Br samples exhibited small NO uptakes. Since the solid-gas reaction in this case should be limited on the oxide surface, the XPS measurement was conducted to detect surface contaminations for these samples. The spectra suggested the presence of chloride or bromide anions, which were totally absent on the surface of NO3-1. These residual halide species originated from metal sources and/or surfactants would be strongly bonded to the NO adsorption site. This is consistent with that NO3-OH samples, k_...... n-0.8 which were prepared from halide-free precursors ............-- " ~ ............ ~................ and CTAOR showed much larger NO uptake. The use of halide anions must be avoided for the n-0.6 preparation of NO adsorbing materials. The NO uptake was further improved by loading 1 wt% Pt n=0.4 via wet incipient impregnation of a Cl-free ~-~--___.-_---_---- - ~ ' ~ ' - - - ............. - : . . . . . platinum complex. A Zr02 tetragonal
Structure of adsorbed NO In situ DRIFTS measurement was --:~_.... _ _o o _ _ ~ conducted to evaluate the structure of NOx adsorbed on MnOx-ZrO2. Fig. 9 compares the , , , , , , , , spectra taken from MnOx-ZrO: (NO3-OH) (a) and o lo 20 30 40 so eo 70 Pt/MnOx-ZrO2 (b) after NO adsorption at 25 ~ 2 /deg The admission of NO immediately yielded several a)~.at,a ,.,. b)Detcnnined by EDX measul'emcstt different bands in the range 1200-1600 cm1, Fig. 4. Powder XRD patterns of depending on the 02 concentrations in the gas feed. (n)MnOx-(1-n)ZrO2 prepared by The band at ca. 1270 cm] observed for unloaded coprecipitation in the presence of NO3-OH in the absence of 02 is due to nitrite CTAB and subsequent calcination at 450 ~ pH7.0-7.4 .__...
. . . . . . . . .
;..
9 ._...,~
__
o Zr02 monoc/inic
~,m.m4mvu
~..*Jr
it,,,..~
_1,.~_
-_-.,,,~
9
-:~:--,---.,.
n=0.2 3.5 . . . .
860 (NO2) type species. The intensity of this band decreased with increasing 02 concentration as a result of the conversion of nitrite to nitrate (NO3, 1450 cm'l). These bands are quite similar to those observed for Mn203 [7]. The manganese oxide is an active catalyst for NO oxidation to NO2, but the single phase Mn203 could not adsorb large amount of NO because of the low surface area. In this respect, the NO3-OH is quite adequate as an NOx-adsorbing material. The Pt-loaded NO3-OH showed more intense bands at ca.1560, 1450, and 1360 cm~. The bands at 1560 cm~ may be assigned to nitrate adsorbed on ZrO2, which was formed by NO oxidation over Pt and subsequent adsorption. As pointed out in our previous report [12], the oxidative adsorption onto ZrO2 requires a catalyst, such as Pt and Pd, to promote the NO oxidation. In accordance with this effect, the Pt loading intensified the nitrate bands in the oresence of 02.
Fig. 5.
SEM photographs of NO3-1 CI-Br-2 after calcination at 450 ~
and
~,~
400"C
9Zr02 tetragonal
0
I
10
I
20
A
I
550oc
I
30 40 2#/deg
I
50
I
60
Fig.6. Powder XRD patterns ot CI-Br-2 prepared by coprecipitation in the presence ot CTAB and subsequent calcination.
I
70
861
Fig.8. Temperature dependence of NO uptake of MnOx-ZrO2 prepared by using different metal sources and surfactants 0.08% NO, 10% 02, He balance.
[~ ~~ Fig.9.
in situ DRIFT spectra of and b) 1wt~ after NO adsorption at 25 ~ 0.04 % NO, 0-10 % O2 He balance, 30min.
a)MnOx-ZrO2
,!7/
.~, ,~ 2000
150(
1000
Wavenumber/cm'
2000
500
I000
V~venumber/cm -~
862 REFERENCES
1. M.Machida, in Catalysis, Vol.15, The Royal Society of Chemistry, Cambridge, 2000, p.73. 2. M.Machida, Catal. Survey (2002) in press. 3. K.Eguchi, M.Watabe, S.Ogata and H.Arai, Bull. Chem. Soc. Jpn, 68 (1995) 1739. 4. K.Eguchi, M.Watabe, M.Machida and H.Arai, Catal. Today, 27 (1996) 297. 5. K.Eguchi, M.Watabe, S.Ogata and H.Arai, J. Catal., 158 (1996) 420. 6. M.Machida, M.Uto, D.Kurogi and T.Kijima, Chem. Mater., 12 (2000) 3158. 7. M.Machida, D.Kurogi and T.Kijima, J. Chem. Mater., 11 (2001) 900. 8. M.Machida, D.Kurogi and T.Kijima, Chem. Mater., 12 (2000) 3165. 9. M.Machida, D.Kurogi and T.Kijima, Stud. Surf. Sci. Catal., 138 (2001) 267. 10. Phase Equilibria Diagrams, ver.2.1, Am. Ceram. Soc. (1997). 11. Z.R.Tien, W.Tong, J.Y.Wang, N.CtDuan, V.V.Krishnan and S.L.Suib, Science, 276 (1997) 926. 12. M.Machida, A.Yoshii and T.Kijima, Int. J. Inorg. Mater., 2 (2000) 413.
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
Preparation of C u O x - T i O 2 intercalated layered structure
nano-composite
863
photocatalysts
from
M.Machida*, S.Nagasaki and T.Kijima Department of Applied Chemistry, Faculty of Engineering, Miyazaki University, Gakuenkibanadai-nishi, Miyazaki 889-2192, Japan
Nano-composite photocatalysts, CuOx-TiO2, were synthesized from Cu(OAc)2intercalated fibrous layered titanates by thermal decomposition in different atmospheres (Na, air, and Ha). The structural characterization using XRD, UV-vis, XPS, and SEM implied that the composite of partially reduced CuOx and anatase-type TiO2 in a waffle-like texture would be a reason for the excellent photocatalytic activity for Ha production from CH3OH/H20 mixtures. 1. INTRODUCTION The pillared layered materials are well-known nano-composites, which can be prepared via intercalation of large polymeric inorganic cations or metal alkoxides into ion-exchangeable layered hosts [1-9]. The structure of composites obtained after heat treatment is characterized by periodic alternative stacking of a host layer and a guest oxide. One interesting feature of such composites is a wide variety of possible chemical combinations between hosts and guests. In particular, many researchers are interested in the layered titanate, niobates, and tantalates, because of their excellent photocatalytic property [10-14]. We have previously reported that the photocatalytic activity of SiOz-pillared layered titanates is influenced by partial replacement of transition metals for Ti [13]. However, no attempt was made to incorporate various transition metal oxides into the interlayer. In the present study, a series of composite photocatalysts were synthesized by stepwise ion exchange and intercalation of transition metal acetates into fibrous layered tetratitanate microcrystals (K2Ti409) and subsequent calcination in different atmospheres. The highest photocatalytic activity obtained for the CuOx-composites was studied in relation to the microstructure. 2. EXPERIMENTAL
2.1. Sample preparation Fibrous potassium tetratitanate (K2Ti409) powders were supplied from Otsuka Chemical Co., Ltd. They were obtained from K2MoO4 flux containing mixtures of KzCO3 and TiO2 as was first reported by Fujiki et al. [15]. The pillared samples were prepared by a stepwise exchange process. The synthetic route followed was a modification of the method adopted by Yan et al.[9]. First, the sample was exchanged in lmol/1 HCI at room temperature for 1 week to give protonated derivatives. Intercalation of hexylamine was carried out by mixing the protonated sample in an aqueous solution of 2 mol/l n-C6HlaNH2
864 at room temperature for 1 week. This was followed by the treatment with 3 mol/l tetramethylammonium chloride for 3 days. The solid product was next treated with 0.1 mol/l aqueous solutions of transition metal acetates at 60 *C or at room temperature for 3 days. The resulting solids were centrifuged, washed with distilled water and air-dried at 60 ~ Finally, the products were heated at elevated temperatures (300-500 ~ for 2 h in a stream of Nz, Ha, or dry air. We denote the synthesized composites as M-X-Y, where M=Cr, Mn, Fe, Co, Ni, or Zn, X=heating temperature (~ and Y=heating atmosphere (N2, Ha, or air). 2.2. Characterization The crystal structure of as prepared samples was identified by using a powder X-ray diffractometer equipped with CuK~ radiation (30kV, 20mA) and a monochromator. An infrared spectrometer was used for the chemical structure analysis. Chemical composition of samples was determined by EDX analysis. To determine the content of organic species in the composites, thermal gravimetric (TG) analysis was carried out at a heating rate of 10 ~ in air. The BET surface area was determined by measuring N2 adsorption isotherms at 77 K. The microstructure of samples was observed by FE-SEM. Diffuse reflectance spectra were recorded with a UV-vis spectrometer. 2.3
Photocatalytic reaction The photocatalytic Ha evolution was conducted in an inner irradiation Pyrex cell, which was connected to a closed gas-circulating system consisting of a circulation pump, a pressure sensor, gas sampling valves, and stainless steel tubing. As calcined powder nano-composite (0.2 g) was suspended in 4mol/1 CH3OH (200 cm ~) in the cell by use of a magnetic stirrer. Prior to the reaction, the mixture was deaerated by evacuation and then
Fig. 1.
Crystal structure and SEM photograph of KzThO9.
flushed with Ar gas (20 kPa) repeatedly to remove Oz and COz dissolving in water. The reaction was carried out by irradiating the mixture with light from a 400W high-pressure Hg lamp. Gas evolution was observed only under photo-irradiation, being analyzed by an on-line gas chromatograph.
865 3. RESULTS AND DISCUSSION 3.1
Nano-composites derived from intercalation
Fig. 1 shows a SEM (200) photograph and the crystal 0.83nm structure of KzTi409 microcrystals H2Ti409 employed in the present study. The 2.23nm powders consist of well-developed fibrous microcrystals with smooth surface, the geometrical surface area of which is close to the measured BET surface area (11 mZ/g). From electron diffraction measurement in TEM observation, 1.12nm it was confirmed that the fiber axis 06H13NH3-Ti409 is parallel to the b axis of the 2.18nm monoclinic structure, which is composed of TiO6 octahedral sheets stacking along the a* direction. As shown in Fig. 1, the crystal structure is built up from a unit of four T i O 6 octahedra (CH3)4N-Ti409 arranged in a line by edge sharing [16,17]. These units are joined to 1.64nm t 11.17nm similar blocks above and below to ! I 079nm . . . . . . . . .___ .... ~ form zigzag strings. The strings are combined by sharing corners of I . I I J i . I I 10 20 30 40 50 60 octahedral to form staggered 0 28 / deg sheets. The structural change of the Fig. 2. Change of XRD patterns of HzThO9 layered titanates during stepwise by stepwise reactions with C6H13NH2, exchange to pillared derivatives was (CH3)4NCI, and Cu(OAc)2. studied by XRD (Fig. 2). Here, the (200) reflection corresponds to the interlayer distance between two adjacent TiO6 sheets. The interlayer distance of the protonated phase, H2Ti409, was increased from 0.83 nm to 2.23 nm by intercalation of C6H13NH2. The corresponding interlayer spacing of ca.1.5 nm is much lager than the chain length of n-hexylammonium ions (ca.1.0 nm), suggesting the bilayer configuration of alkyl chains as was pointed out by Hou et al. [7]. Mixing the hexylamine-intercalated phase with an aqueous solution of copper (II) acetate led to the decreased interlayer distance (1.64 rim). According to FT-IR measurement, this structural change was accompanied by the appearance of bands due to vco at 1450 and 1550 cm-1. The reaction was also conducted using aqueous solutions of other transition-metal acetates in the same manner. As shown in Table 1, the M/Ti ratio for the as prepared composites was in the range of 0.10-0.47, suggesting that the bulk-type reaction took place between metal acetates and the layered titanate. No deposition of metal acetate was confirmed on the surface of the fibrous microcrystals by the SEM for all as prepared samples before calcination. These results demonstrated that metal acetates were accommodated in the interlayer.
(CH3COO)2C_u-Ti,0?
866 Table 1 Chemical composition and BET surface area of M-400-N2 M M/Ti Surface area/m2g '~ " Cr 0.35 100.5 Mn 0.41 24.1 Fe 0.10 32.1 Co 0.35 26.4 Ni 0.34 44.5 Cu 0.33 36.6 Zn 0.47 23.3 ............
M=Cr -,
.
.
.
.
,
-
-
-
.
.
.
.
.
.
.
.
.
.
3.2. Structure and photocatalytic ,A M n O ~ 9 Mn property of nano-composites after calcination Fig. 3 shows the XRD patterns u Fe~03 after calcination of as prepared [] u Fe solids at 400 ~ in N2 (M-400-N2). The disappearance of basal reflections at 20400 ~ in N2. When the precursor was heated in air (Cu-450-air), however, these phases were scarcely observed. This means that residual organic molecules in the interlayer would play
868 as a reducing agent for Cu acetate to produce the metallic state. The reduction process caused the collapse of pillared layered structure to produce the Cu/TiOz composite, but the fibrous morphology was retained. The photocatalytic activity was strongly dependent on the heating temperature as well as heating atmosphere as shown in Fig. 6. Calcination at 400
_ ~-
__~_:-
_,,,__ ~ _
. . . . . . . . . .
.
.
.m
o
and Cu-450-N2. The former exhibited a position of the highest filled bands ca. 0.9 eV higher than the latter. However, such an alteration of the valence band was not effective in improving the photocatalytic activity.
200
/ .
0
250
-~
300
~
in Air ,
m
1
i
350
400
A
450
i
500
55O
Calcination temperature / ~
Fig. 6. Photocatalytic H2 evolution from 4mol/l CH3OH solution over CuOx-composites calcined at different temperatures.
869 Table 3 Effects of pretreatment on photocatalytic activity of CuOx-composites Atmosphere Rate of Hz evolved a) / ~tmol.h-1 air(5h) 37 Nz(5h) 1780 air(5h) + Na(5h) air(5h) + Hz(5h)
478 603
Table 4 Photocatalytic activity of unloaded layered titanates Catalyst Rate of a)/~tmol h-1
Cu-loaded gas
evolution
Hz 15.6 44.6 0.4 802.8
KzTi409 26wt%Cu/KzThO9 b) HzThO9 26wt%Cu/Hz Ti409 b)
and
CO2 0 0 0 11.6
All treatment was carried out at 450 ~ a) 4mol/l CH3OH 200 cm 3, catalyst 0.2 g a) 4mol/l CH3OH 200 cm 3, b) Prepared by wet impregnation of Cu(OAc)2 catalyst 0.2 g and subsequent heating at 450 ~ in N2.
ii
300 ~ in N2 300~
in N2
350 ~ in N2 45
II 450 500 ~ in N2
450 ~ in air 980 250
350
450
550
650
750
Wavelength / nm
Fig. 7. UV-vis reflectance spectra of CuOx-composites derived from Cu(OAc)2- intercalated tetratitanates after calcination at elevated temperatures.
850
I
I
I
I
I
970
960
950
940
930
920
Binding energy / eV
Fig. 8. Cu2p CuO• and air.
XPS spectra of after calcination in N2
870 Table 3 exhibits the effect of heating atmosphere on the structure and the photocatalytic activity of Cu-450. Contrary to Cu-450-N2, Cu-450-air exhibited much less activity. However, it was found that the activity of Cu-450-air was much improved by additional heating in N2 as well as H2. Since the treatment in N2 did not give rise to metallic Cu, the reduction of Cu to a metallic state is not essential to the photocatalytic activity. Fig. 8 shows the XPS spectra of Cu2p region. As is evident from a main peak of Cu2p3/2, all samples contained Cu in the divalent state, but a shoulder on the right side of the peak was observed for the exception of Cu-450-air. This means that the metallic Cu as well as partially reduced Cu species (Cu+) was formed after heating in N2. The formation of composites between partially reduced Cu and anatase-type TiO2 should be a key to yield the high photocatalytic activity in the present system. Table 4 shows the photocatalytic activity of CuOx supported on the pristine samples of layered titanates. The loading (26 wt% Cu) corresponds to the Cu/Ti ratio (0.33) for the composite prepared by the intercalation process. The two pristine layered titanates showed very low activity without loading CuOx. In contrast, the activity of the CuOx-loaded samples was quite different; the protonated phase produced ca. 18-times higher rate of H2 evolution. Copper oxides in these materials were deposited only on the surface of the fibrous crystals. However, the protonated phase was decomposed to produce anatase-type TiO2 as in the case of the intercalated composite (Fig. 5). This result also supports that the formation of anatase is essential for the photocatalytic activity. ACKNOWLEDGMENT One of the authors (M.M.) gratefully acknowledges the financial support from Grant-in Aid for Scientific Research from the Ministry of Education, Science, and Culture and Shiseido Foundation for Science and Technology. REFERENCES 1. I.V. Mitchell (ed), Pillared Layered Structures: Current Trends and Applications, Elsevier Applied Science, London, 1990. 2. S. Cheng and T.C. Wang, Inorg. Chem., 28 (1989) 1283. 3. M.W. Anderson and J. Klinowski, J. Am. Chem. Soc., 29 (1990) 3260. 4. M.E. Landis and B.A. Aufdembrink, P. Chu, I.D. Johnson, G.W. Kirker and M.K. Rubin, J. Am. Chem. Soc., 113 (1991) 3189. 5. W. Hou, Q. Yan and X. Fu, J. Chem. Soc. Chem. Commun., (1994) 1371. 6. W. Hou, B. Peng, Q. Yan, X. Fu and G. Shi, J. Chem. Soc. Chem. Commun., (1994) 253. 7. W. Hou, Q. Yan, B. Peng and X. Fu, J. Mater. Chem., 5 (1995) 109. 8. C. Guo, W. Hou, M. Guo, Q. Yan and Y. Chen, J. Chem. Soc. Chem. Commun., (1997) 801. 9. Y. Chen, W. Hou, C. Guo, Q. Yan and Y. Chen, J. Chem. Soc. Dalton Trans. (1997) 359. 10. J. Kondo, S. Shibata, Y. Ebina, K. Domen and A. Tanaka, J. Phys. Chem., 99 (1995) 16043. 11. S. Uchida, Y. Yamamoto, Y. Fujishiro, A. Watanabe, O. Ito and T. Sato, J. Chem. Soc., Faraday Trans., 93 (1997) 3229.
871 12. M. Machida, J. Yabunaka, H. Taniguchi and T. Kijima, Stud. Surf. Sci. Catal., 118 (1998) 951. 13. M. Machida, X.W. Ma, H. Taniguchi, J. Yabunaka and T.Kijima, J. Mol. Catal. A, 155 (2000) 131. 14 M. Machida, J. Yabunaka and T. Kijima, Mol. Cryst. Liq. Cryst., 341 (2000) 249. 15. Y. Fujiki and N. Ohta, Yogyo Kyokaisi, 88 (1980) 11. 16. H. Izawa, S. Kikkaw and M. Koizumi, J. Phys. Chem., 86 (1986) 5023. 17. T. Sasaki, M. Watanabe, Y. Komatsu and F. Fujiki, Inorg. Chem., 24 (1985) 2265.
This Page Intentionally Left Blank
Studies in Surface ScienceandCatalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
873
Vanadia-doped titanium pillared clay: preparation, characterization and SCR activity of NO by ammonia L. Khalfallah Boudali a, A. Ghorbel a, P. Grange b and S. M. Jung b Laboratoire de Chimie des Mat6riaux et Catalyse, D6patement de Chimie, Facult6 des Sciences de Tunis, Tunis, Tunisia.
a
b Unit6 de catalyse et chimie des mat6riaux divis6s, Universit6 catholique de Louvain, Croix du sud, 2/17, 1348 Louvain-la-Neuve, Belgium. Vanadia doped sulfated Ti-pillared clays were prepared and characterized by BET, XRD, XPS, TPD-NH3 and compared with sulfated Ti-pillared clays and vanadia-doped unsulfated Ti-pillared clays. When sulfated Ti-pillared clay was doped with vanadia, the BET surface areas decrease whereas the diffraction line (001) are not significantly affected. The acidic properties of sulfated catalysts are higher than vanadia doped sulfated samples. The comparison of the activity of the catalysts in the selective catalytic reduction (SCR) of NO by ammonia in presence of oxygen show that vanadia doped sulfated Tipillared clay were highly active for the SCR NO. Therefore sulfated Ti-pillared clay appears as a good support for vanadia catalysts for the SCR reaction. 1.
INTRODUCTION
The selective catalytic reduction (SCR) of NO with ammonia is an industrial process for the abatement of NO from gaseous emissions [1,2]. Among the large number of catalysts active in such reaction, vanadia-based solids have been described as the best ones. Various methods for their preparation have been studied [3, 4]. Many supports have been reported to be active for this reaction, such as vanadia supported on AlzO3, SIO2, TiOz/SiOz, etc. [4, 5], but VzOs/TiOz catalysts present the highest activity and resistance to poisoning by H20 and SO2 [5]. It is accepted that TiO2 in the anatase form is the appropriate support [6]. Commercial SCR catalysts are mostly based on vanadia/titania and WO 3 and/or MoO 3 are usually added to VzOs/TiO2, partly as promoters [7]. Both Lewis and Br6nsted acid sites exist on pillared clays. The acidity depends on the exchanged cations, the preparation method and the starting clay [8]. It is known that surface acidity is important for SCR reaction of NO by NH3 [4, 5, 9]. Different pillared clays were synthesized and tested for their activities in the SCR NO [10, 11]. Research on titanium pillared clays was initiated by Sterte [12], who first reported the synthesis of titanium pillared montmorillonite using TIC14 solution in hydrochloric acid. Bernier et al.
874 [13] later studied the influence of temperature on titanium pillared clay. Khalfallah Boudali et al. [14] prepared titanium pillared montmorillonites by intercalating polymeric cationic species via hydrolysis of TiCI4 with HC1 and H2SO4. Smaller BET surface areas but stronger Br6nsted acid sites were observed on sulfated Ti-modified pillared montmorillonites than on the unsulfated ones. It appeared that the nature of the polymeric Ti species depends on the mineral acid used. Del Castillo et al. [15] carried out a detailed study on the experimental conditions needed to synthesize sulfated Ti-pillared montmorillonite. The sulfation of the pillared clays was performed by intercalation of the clay with the titanium species in the presence of (NH4)2SO4 or H2SO4 and by impregnation of Ti-PILC. The catalytic behavior of the Ti-modified pillared montmorillonites in the selective catalytic reduction of NO by NH3 with or without SO2 has been reported by these authors [16]. It has been shown that vanadia-doped pillared clays represent interesting potential catalysts for SCR NO with ammonia and that the catalytic performance of these systems depends on the method of preparation and on the V content [17]. It was also shown that the addition of SO42- species on TiO2 is attractive since sulfated titania exhibits a good activity for high temperature SCR reactions [18, 19]. It has been proposed that the sulfate produce a strong acidic site at high temperature [20-23], since it is a way to tune the acid-redox properties required. It appears interesting to investigate vanadia doped sulfated Ti-pillared clay for the selective catalytic reduction of NO by ammonia and to compare their performance with sulfated Ti-pillared clay and vanadia doped Ti-pillared clay. In this work, all the catalysts were synthesized under identical conditions and were characterized by different techniques. These techniques included surface area measurement, pore size distribution, Xray diffraction, X-ray photoelectron spectroscopy, TPD-NH3 and chemical analysis of Ti retained by the clay. The catalysts were then tested in the selective catalytic reduction of NO by ammonia in the presence of oxygen at different temperatures. 2. E X P E R I M E N T A L 2.1. Catalyst preparation The initial material used for intercalation was a volclay montmorillonite from CECA. The pillaring solution was obtained by slowly adding TiCl 4 into H2SO4 (6 M) under vigorous stirring. Final concentrations of 0.82 M in titanium and 0.2 M in H2SO4 were reached by adding water; H+/Ti value of 0.24 was thus obtained [14]. The fresh pillaring solution was then added dropwise to 500 cc of a suspension containing 2 g of clay to obtain sulfated titanium pillared clay, in such quantities that final Ti/clay ratios of 10, 20 and 40 mmol/g were obtained. The same procedure was employed using HC1 (6 M) for hydrolysis of TiCh to obtain unsulfated titanium pillared clays. After 24 h of stirring the solid fraction was separated by centrifugation and filtration, then washed several times with distilled water and dried at room temperature. Vanadia doped sulfated and unsulfated Ti-PILC, containing 3 % of vanadium, were prepared by impregnation of the supports with a solution of ammonia vanadate (NH4VO3) 0.1 M dissolved in water acidified by oxalic acid. All the samples were then dried at 80~ for 20 h and calcined at 400~ for 3 h. The catalysts are referenced S-xTi, V-xTi and V-S-xTi in which x is the Ti/clay ratio equal to 10, 20 or 40 mmol/g clay.
875
2.2 Characterization The basal spacings d001 of the samples were evaluated by XRD on CGR theta 60 instrument using monochromatized CuK~ radiation at ~ - 1.54 A. The specific surface areas were measured by the BET method from nitrogen adsorption isotherms using a Micromeritics ASAP 2000 equipment. The thermal stabilities were investigated for all samples at the heating rate of l~ The quantities of titanium retained by the clay were determined by colorimetry using a Philips PYE Unicam PU 8650 spectrophotometer after dissolution of the sample in 6 N sulfuric acid at 80 ~ and addition of 5 ml H202 per mg of Ti [24]. The total acidity measurement was evaluated by temperature-programmed desorption of ammonia (TPD- NH3). For these experiments, a sample weight about 0.150 g calcined at 400 ~ was placed in the cell, evacuated at 400 ~ for 2 h and then cooled to 100 ~ Pure ammonia gas (50 cc.min -1) was adsorbed at 100 ~ for 20 min. Thereafter, while increasing the sample temperature to 400 ~ at a constant rate 10 ~ and maintaining the carrier gas flow rate at 50 ccHe/min, the desorbed ammonia was passed through a 20 wt% HBO3 solution in order to check the amount of NH3 evolved. The NH3 amount was obtained by Kjeldhal method. XPS analysis were performed at room temperature with a Surface Science Instruments SSX-100 model 206 spectrometer with a monochromatized AIKa source operating at 10 KeV and 12 mA. Samples calcined at 400 ~ were compressed in a small cup under a 5 Kg/cm 2 pressure for 30 s and supported on a holding carousel. The binding energies (B.E.) of Ti2p, V2p and S2p lines were referenced to the C ls band at 284,8 eV. Activity measurements were performed in a continuous flow fixed bed reactor operating at atmospheric pressure between 100 ~ and 400 ~ The total flow rate was 100 cc/min and feed composition was: NO (0,1 vol. %), NH3 (0,11 vol. %) and 02 (2,5 vol. %), in helium. The catalyst amount of 0,150 g calcined at 400 ~ and the space velocity were kept constant for all experiments. The inlet and outlet gas compositions were measured using a quadrupole mass spectrometer QMC 311 Balzers coupled to the reactor. 3. RESULTS AND DISCUSSION
3.1. Catalyst characterization The BET surface areas, pore volumes and pore size distributions for all catalysts investigated are summarized in Table 1. S-10Ti sample showed high specific surface area. But, according to the increase of Ti/clay ratio, the surface area was decreased. This result indicates that the higher Ti/clay ratio leads to a progressive plugging of the internal structure of the clay. After addition of vanadium (3 %) to sulfated samples, the surface area and the micropore volume decreased sharply, whereas the pore diameter increased. This effect can be explained by the blocking of small pores due to the building up of the vanadia layer. The BET surface areas of vanadia-doped unsulfated Ti-PILC were near 100 mZ/g. The average value is larger than that of vanadia-doped sulfated Ti-PILC. This may be due to the blockage of small pores by both the vanadium oxides and sulfate ion in the case of vanadia-doped sulfated Ti-PILC. The d001 spacings of the pillared clays are also presented in Table 1 and range from 15.1 to 18.6 ,~ with a maximum of 48 % TiO2 retained by the
876
clay. The basal spacings are not significantly affected by vanadia addition. A study by Einaga [25] on Ti4§ and polymerization suggested the existence of a polymeric cationic species, (TiO)8(OH)84§ However, the structure of the complex is presently unknown. Table 1 Physico-chemical properties of the catalysts prepared in the present study (preparation with H+/Ti = 0,24 and Ti/clay = 10, 20, 40 mmol/g). Catalysts S.A. VpTotal Viap d001 TiO 2 NH3 (mZ/g) (cma/g) (cm3/g) (,~) (wt%) (~mol/m 2) S-10Ti S-20Ti S-40Ti
255 137 100
0.146 0.082 0.058
0.060 0.041 0.033
18.6 15.4 16.2
38.9 32.9 40.0
2.5 5.9 9.0
V-10Ti V-20Ti V-40Ti
103 92 98
0.132 0.068 0.125
0.010 0.020 0.001
17.2 15.2 18.5
39.2 48.5 48.4
6.97
V-S-10Ti V-S-20Ti V-S-40Ti
60 91 87
0.083 0.092 0.078
0.003 0.015 0.012
17.7 15.1 15.9
38.9 32.9 40.0
4.41 3.85 2.75
S.A. :Specific surface areas, VpTotal: Total pore volume, VIap" Micropore volume, d001: basal spacing The effect of thermal treatment was investigated on the samples of the 10Ti series. The specific surface areas were measured according to the different temperatures. As shown in Fig. 1, the BET surface areas of sulfated Ti-PILC decreased with the increase of the thermal treatment temperature, while the surface areas for vanadia Ti-PILC increased. At 400 ~ which is the calcination temperature for SCR reaction, all samples showed almost the same value of specific surface area (near 160 mZ/g). In order to evaluate the total acidity, TPD- NH3 was carried out. The desorbed amount of NH3 was also compared in Table 1. In the case of S-xTi series, it is evidenced that the amount of desorbed ammonia increases with the added amount of sulfate ion during the preparation. In other words, the different Ti/clay ratio also induced the different amount of sulfuric acid added. This result shows that the different amount of sulfate intercalated contributes to the increase of the acidity. When vanadia was added to sulfated Ti-PILC, the amount of NH3 desorbed decreased significantly. R. T. Yang [26] reported that the vanadia create more Br6nsted acid sites with the increase of vanadia from 2 to 6%. Thus, the decrease of the acidity observed in our samples may be explained by the assumption that some of the ammonia desorbed from the surface was oxidized by lattice oxygen of the catalysts. It is known that lattice oxygen of VzO5 can oxidize ammonia to Nz and nitrogen oxides at high temperatures (27).
877 350 ~- S-10Ti
300 "
V-10Ti
--x-
250 -
Ti
t~
200 150 r~
100 50-
50
!
!
I
150
250
350
450
Temperature (~ Fig. 1. BET surface areas of catalysts after calcination at increasing temperatures. Table 2 XPS Data of sulfated Ti-PILC and vanadia doped sulfated Ti-PILC. Binding energy (eV) Catalyst
S/(Ti+V) atomic ratio Ti 2p 3/2
V 2p 3/2
S 2p
S-10Ti
459.3
-
169.2
0.163
S-20Ti
459.2
-
169.1
0.306
S-40Ti
459.2
-
169.1
0.432
V-S- 10Ti
459.1
517.3
169.0
0.336
V-S-20Ti
459.3
517.3
169.2
0.339
V- S-40Ti
459.2
517.4
169.3
0.345
The XPS results of Ti 2p3/2, V 2p3/2 and S 2p on the vanadia doped sulfated Ti-pillared clay catalysts are summarized in Table 2. A broad XPS band near 459 eV was observed on all samples investigated. This value is almost the same as the binding energy of 2p3/2 of Ti on TiO2 [28]. The intensity of this band increases with the Ti concentration, which reflects an increase of the surface concentration of Ti species. A band centered near 517 eV was detected on the vanadium containing samples. This indicates that vanadium on the fresh
878
catalysts was present mainly as the +5 valence form, as VzO5 [28]. During calcination at 400 ~ in air, vanadia was oxidized to V205 by oxygen. Contrary to the previous investigations of the VzOs/TiOz catalysts [29], no asymmetry was detected in the V 2p spectra, testifying the absence of V 4§ contribution. These results indicate that compared to VzO5/TiO2 catalysts [29] the state of V 5+ in the Ti-pillared clay is much more stable. The S 2p spectra of both sulfated Ti-PILC and vanadia doped sulfated Ti-PILC exhibit a band with a binding energy near 169 eV which is attributed to sulfate SO42, typical of S 6§ in SO bonds [19, 28]. Therefore, the XPS results lead to the conclusion that the vanadium was present mainly as the +5 valent form and the sulfur-containing species on all sulfated catalyst surfaces is the S 6§ form as sulfate. The amounts of sulfur in the vanadia-doped sulfated Ti-PILC measured by XPS atomic surface composition evaluated as S/(Ti+V) atomic ratio are also shown in Table 2. The increase of the S/(Ti+V) atomic ratio of the sulfated Ti-PILC is proportional to the sulfate concentration in the Ti-pillaring solution without vanadia. For vanadia doped Ti-PILC, the S/(Ti+V) atomic ratios became almost similar which evidences the homogeneous distribution of sulfate after vanadia doping. 3.2. Catalytic activity The present results of SCR NO in presence of oxygen indicate that vanadia-doped sulfated Ti-pillared clay catalysts are highly active for ammonia SCR (Fig. 2). The activity of the sulfated Ti-PILC increases with temperature of reaction up to 400 ~ but NO conversions are small at 200-350 ~ The activity of these samples is essentially related to the presence of sulfates which enhances the acidity of the catalysts and there is a direct correlation between the SCR activity and acidity as has been already reported [4, 5, 9]. It was shown that the introduction of sulfate ions on the Ti-PILC not only increased Lewis acidity but also the strength of acidic Br6nsted sites [14, 15]. Jung [18] reported that the strong acidic sites, especially the strong Lewis sites generated by modifying TiO2 with SO42- are responsible for the higher reactivity of TiO2-SO42- in the SCR at high temperature. When a small amount of vanadia (3%) was added to the unsulfated Ti-PILC, the NO conversion increased significantly with temperature reaction from 200 to 400 ~ The SCR activity of vanadia-doped unsulfated Ti-PILC was higher than that of sulfated Tipillared clay, suggesting that vanadium played an important role in this reaction. This result is confirmed by the catalytic performance of vanadia sulfated Ti-pillared clay. On these latter catalysts, NO conversion was increased at low temperature (< 200~ but declined slightly at high temperatures (> 350 ~ due to the oxidation of ammonia by oxygen. The NO conversion was higher on vanadia-doped sulfated Ti-PILC than on the other catalysts investigated at any temperature between 100 and 400~ The high activity of vanadia-doped sulfated Ti-PILC in the SCR reaction could originate from the requirement of a surface redox site with a surface acid site for this reaction. For all the catalysts investigated, the increase of the Ti concentration (Ti/clay = 10, 20 or 40 mmol/g) does not have a main role in the SCR reaction.
879 100
-" S-10Ti S-20Ti S-40Ti f V-40Ti r + V-S-10Ti / ~ ---o--V-S-20Ti ///
80 "-" ~9 60 g
--~v
40
,~"
/" /f /
I r /
//?
//i/ ~// [1~'
9
20 0 0
100
200
300
400
500
Temperature (~ Fig. 2. NO conversion vs. temperature
4. CONCLUSION Vanadia doped sulfated Ti-pillared clay are highly active for the reduction of NO by ammonia in the presence of oxygen. The results of SCR reaction show that vanadium present mainly astthe +5 valence played an important role in this reaction. SCR activity was found to be correlated to acidity at high temperature. At low temperatures, the redox properties play a key role for the activity. REFERENCES
1. 2. 3. 4. 5. 6.
V.I. Parvulescu, P. Grange and B. Delmon, Catal. Today, 46 (1998) 233. F. Anssen and R. Maijer, Catal.Today, 16 (1993) 157. G.C. Bond and S.F. Tahir, Appl. Catal., 71, (1991) 1. H. Bosch and F. Janssen, Catal. Today, 2 (1988) 369. G. Busca, L. Lietti, G. Ramis and F. Berti, App. Catal. B Environ., 18 (1998) 1. I.M. Pearson, H. Ryu, W.C. Wong and K. Nobe, Ind. Eng. Chem. Prod. Res. Dev., 22 (1983) 381. 7. L.L. Sloss, "Nitrogen Oxide Control Technology Fact Book." Noyes Data Corporation, Park Ridge, 1992. 8. F. Figueras, Catal. Rev. Sci. Eng., 30 (1988) 457. 9. N.Y. Topsoe, J. A. Dumesic and H. Topsoe, J. Catal., 151 (1995) 241. 10. R. T. Yang, J. P. Chen, E. S. Kikkinides, L. S. Cheng and J. E. Cichanowicz, Ind, Eng. Chem. Res., 31 (1992) 1440. 11. R. T. Yang and J. E. Cichanowicz, Pillared clays as catalysts for selective catalytic reduction of NO, US Pat. 5, 415 (1995) 850. 12. J. Sterte, Clays Clay Miner., 34 (1986) 658.
880 13. A. Bernier, L. F. Admaiai and P. Grange, Appl. Catal., 77 (1991) 269. 14. L. Khalfallah Boudali, A. Ghorbel, D. Tichit, B. Chiche, R. Dutartre and F. Figueras, Microporous Mater., 2 (1994) 525. 15. H. L. Del Castillo, A. Gil and P. Grange, Catal. Lett., 43 (1997) 133. 16. H. L. Del Castillo, A. Gil and P. Grange, Catal. Lett., 36 (1996) 237. 17. K. Bahranowski, J. Janas, T. Machej, E. M. Serwicka and L. A. Vartikian, Clay Miner., 32 (1997) 665. 18. a) S.M. Jung and P. Grange, Catal. Today, 59 (2000) 305; b) S.M. Jung and P. Grange, Appl. Catal. B, 27 (2000) L11. 19. J. P. Chen and R. T. Yang, J. Catal., 139 (1993) 277. 20. K. Tanabe, M. Misono, Y. Ono and H. Hattori, Stud. Surf. Sci. Catal., 51 (1989) 199. 21. H. Hino and K. Arata, J. Chem. Soc. Chem. Comm., (1979) 1148. 22. Y. Tsutomu, Appl. Catal., 61 (1990) 1. 23. J. R. Sohn and H. J. Jang, J. Catal., 136 (1992) 267. 24. G. Charlot "Chimie analytique quantitative" Vol 2, Masson Ed, Paris (1974). 25. H. Einaga, J. Chem. Soc. Dalton Trans., (1974) 1917. 26. R. Q. Long and R. T. Yang, Appl. Catal. B Environ., 24 (2000) 13. 27. Y. Kosaki, A. Miyamoto and Y. Murakami, Bull. Chem. Soc. Jpn., 52 (1979) 617. 28. C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and G. E. Mullenberg (Eds) "Handbook of X-ray Photoelectrons Spectroscopy", Perkin-Elmer, Eden Prairie, 1979. 29. J. Ph. Nogier and M. Delamar, Catal. Today, 20 (1994) 109.
Studies in Surface ScienceandCatalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
881
Advanced preparation by sol-gel method of the encapsulated P d / A I 2 0 3 catalysts for methane combustion S. Fessi a, A. Ghorbel a, A. Rives b and R. Hubaut b aLaboratoire de Chimie des Mat6riaux et Catalyse, D6partement de Chimie, Facult6 des Sciences de Tunis, Campus Universtaire 1060 Tunis, Tunisie. bLaboratoire de catalyse H6t6rog6ne et Homog6ne, URA CNRS 402, Universit6 des Sciences et Technologies de Lille, 59655 Villneuve d'Ascq, France.
In this work, the influence of the preparation method and the thermal treatment on the textural, structural and catalytic activity of Pd/AlzO3 catalyst is investigated. The palladium occluded into the alumina support appears on the surface when the encapsulated catalyst is calcined at 700~ Compared to the catalysts prepared by impregnation method, higher resistance to sintering during calcination up to 800~ and during ageing under reaction mixture at 500~ for 72h is observed on the encapsulated catalyst. The nature of the metalsupport interaction seems to be the reason. Furthermore, the catalytic activity in methane combustion is observed to be dependent both on palladium particle size and metallic surface area. Moreover, during the catalyst ageing under reaction mixture at 500~ less reactive catalytic sites are obtained. The large palladium particles are more inhibited than the small ones. 1. I N T R O D U C T I O N The supported palladium are the most effective catalyst in the combustion process [1]. However sintering is a serious problem causing catalyst deactivation [2]. Many studies were devoted to the thermal stability improvement of the support, either by the help of additives (Ba, La and Si oxides),[3-5] or by the preparation of a thermally stable support, using an adequate synthesis procedure, such as the sol-gel method [5-8]. Unfortunately, this was insufficient to maintain a good metal dispersion at high temperatures. Many studies suggest that a strong metal-support interaction should be among the required properties needed to improve the supported palladium catalysts [5,7]. For these reasons, in this work, two different sol-gel method are used to improve the thermal stability of alumina supported palladium catalysts. Moreover, since the catalysts are applied to function at around 700~ during the ignition step of the methane combustion process [2], the thermal treatment effect on the texture, the structure and the activity of the catalysts are investigated in this paper.
882
2. EXPERIMENTAL 2.1. Catalyst preparation Sol-gel alumina, (AI203-SG): a mixture of aluminium-sec-butoxide (Across, 97 %), (AsB) and sec-butanol (Across, 99 %), (sB) is aged for 20 mn, ([AsB] =IM). Then, acetic acid (Across, 99.8 %), (AcA) is added as AsB modifier and water producer in-situ with the concentration molar ratio [AcA]/[AsB] = 4. The obtained gel is finally oven dried at 70~ for 24 h and calcined in flowing oxygen (1.8 l/h) at 800~ for 2 h.
Sol-gel alumina supported palladium catalysts: Method A, (SG catalyst): a mixture of AsB and sB is aged for 20 mn, ([AsB] =IM), followed by addition of palladium acetylacetonate, Pd(acac)2, (Fluka, 34 wt.-% Pd), with a designed metal loading of 2 wt.-%. The mixture is then kept for an appropriate time under constant stirring. After, AcA is added, ([AcA]/[AsB] = 4). The obtained gel is finally oven dried at 70~ for 24 h, calcined in flowing oxygen (1.8 l/h) at 500, 700 or 800~ for 2h and reduced in flowing hydrogen (1,2 l/h) at 500~ for 1 h. Method B, (SGI catalyst): the sol-gel alumina calcined at 800~ is stirred for 24 h with an acetone solution containing dissolved Pd(acac)2. The designed metal loading is 2 wt.-%. The solvent is then evaporated and the residual solid is dried in oven at 110~ for 24h, calcined in flowing oxygen (1.8 l/h) at 500, 700 or 800~ for 2 h and reduced in flowing hydrogen (1,2 l/h) at 500~ for 1 h. Reference alumina supported palladium catalyst, (I catalyst): A commercial alumina (oxide C, from Degussa) is impregnated with palladium(II) chloride solution acidified with hydrochloric acid to facilitate the dissolution of PdC12 (Fluka, 60 wt.-% Pd) as [PdCl4] 2-. The designed metal loading is 2 wt.-%. Excess water is then removed by evaporation, followed by drying in oven at 110~ for 24h. The catalyst is finally calcined in flowing oxygen (1.8 l/h) at 500, 700 or 800~ for 2h and reduced in flowing hydrogen (1,2 l/h) at 500~ for lh.
2.1. Characterization methods Specific surface area was determined by the BET method from the nitrogen adsorption at 77 K, using an automatic Micrometrics ASAP 2000. Palladium metal dispersion was determined by dynamic pulsed hydrogen chemisorption. The metallic average particle size of palladium was examined by transmission electron microscope (JEOL 100 CX) with a resolution of 0.3 nm. Chemical analysis allowing the determination of palladium and chloride contents were performed respectively by inductively coupled plasma spectrophotometry and by potentiometry. XPS analyses were performed on a Leybold-Heraeus LHS10 spectrometer using a non-monochromatized Al Ka source (hv = 1486.6) operated at 300 W (15kV, 20 mA). The energy resolution of the instrument was 0.80 eV. The standard deviation of the analysed result is 0.02 eV. The error from the charge effect was calibrated using the binding energy of the Al 2p photopeak as a internal standard assuming a value of 74.70 eV. Catalytic activity for methane combustion was determined over the reduced catalyst (fresh catalyst, 100 mg), in a dynamic microreactor. The flow of 1 vol.-% methane, 4 vol.-% oxygen and balance helium were mixed. The total flow rate was regulated at 6 l/h and admitted at 500~ The effluent gas was led directly to the six port gas sampling valve of a gas chromatograph (Intersmat IGC 120 ML) for analysis using a
883 thermal conductivity detector and a porapak Q column heated at 100~ under stream reaction was fixed at 72 h.
The ageing time
3. R E S U L T S AND DISCUSSION 3.1. BET Surface area measurements The nitrogen physical adsorption results show that SG-500 catalyst has a higher BET surface area SGI-500 and 1-500 samples (Table 1). When the catalysts are calcined at 700 or 800~ when they are aged under the reaction mixture at 500~ for 72 h, no significant BET surface area loss is observed for samples SGI and I. However, a considerable decrease of the surface area is observed for SG-500. In addition, it is important to note that the SG samples still show the highest BET surface areas for the different calcination temperatures and after the ageing step.
3.2. X-ray diffraction measurements The XRD measurements show an amorphous phase for the SG-500 catalyst and a mixture of 3' and 6-alumina for SGI-500 and 1-500. When the calcination temperature increases to 700 and 800~ the same alumina phases are conserved in SGI and I. However, transformations to ~,-alumina and to a mixture of ~,and 6-alumina are observed, respectively, for SG-700 and SG-800. After ageing under the reaction mixture, only the amorphous alumina of sample SG-500 changes to the ~/phase. According to these results, a good correlation between the textural and the structural characterisation is observed. In fact, the surface area drops are related to the formation of a thermal stable alumina structure. Moreover, the sol-gel method appears to favour the thermal stability of the alumina support, since the surface area of the SGI catalysts are higher than that of the I samples. In addition, the introduction of Pd(acac)2 during the first step of the SG catalyst preparation seems to have also a positive effect on the alumina resistance to sintering, since the surface area of the SG solids is higher than that of the SGI samples. 3.2. Metal dispersion measurements The hydrogen chemisorption measurements performed on the catalysts calcined at 500~ show a relatively high palladium dispersion for the SGI-500 catalyst (D -36%), and similar values for the SG-500 and 1-500 samples (around 20 %, Table 1). However, in spite of the similar palladium contents (around 2 %) of SG-500 and 1-500, the TEM results show larger average particle diameter (5.3 nm) on the 1-500 catalyst than for the SG-500 sample (2.7 nm). In addition, compared to the SGI-500 and 1-500 catalysts, only the average palladium particle diameter of the SG-500 catalyst deduced from TEM micrographs is not in accordance with the value determined from hydrogen chemisorption measurements. This can be explained by the fact that the fraction of palladium unable to chemisorb hydrogen is not only in the bulk of the SG palladium particles, but also in unreachable regions of the support. This metallic fraction could be inserted in inaccessible pores or surrounded by alumina. Similar results were reported by Khelifi et al. [9] and Balakrishnan et al. [8] when supported palladium or platinum alumina catalysts were prepared by the metal incorporation during the first step of alumina synthesis by a sol-gel method.
884 When the calcination temperature increases, the metal dispersion on catalysts I and SGI decreases. However, it increases from SG-500 to SG-700 and then decreases in SG800. The palladium particle size distributions deduced from TEM characterisation show principally small palladium particles on SG-500 sample (1 to 10 nm) with a narrow metallic particle size distribution and an average particle diameter of 2.7 nm (Fig.l), and larger palladium particles on SG-800 sample (1 to 50 nm), with a wide metallic particle size distribution and an average particle diameter of 7.6 nm. Nevertheless, in spite of the mixture of small and large particles on SG-700 catalyst (1 to 40 nm), the corresponding metallic particle size distribution is narrow and the average particle diameter (1.9 nm) is lower than that obtained for SG-500. This apparent contradiction can be explained by the increase of the small number of palladium particle caused by the re-dispersion of the occluded palladium particles during the calcination from 500 to 700~ This is illustrated by the intensification of the peak of 1 nm particle size distribution of the SG-700 (Fig.l) and by the increase of the metal dispersion on this catalyst. Moreover, the accordance between the average particle diameter determined from hydrogen chemisorption and from TEM micrographs of sample SG-800 suggests that practically the total amount of palladium (2 %) has appeared on the alumina surface. Table 1 BET Surface area, metal dispersion, average particle diameter and chloride content of fresh and aged catalysts. Sample
S BET (mZ/g)
D (%) Ha-chemisorp.
Average particle diameter (nm)
500 700 800
500 700 800
SG fresh state aged state
337 215
21 18
45 41
15 14
5.4 2.5 6.3 2.8
7.5 8.0
SGI fresh state aged state
148 131 146 127
132 137
36 25
16 14
5 5
3.1 7.0 4.5 8.0
22.5 22.5
96 94
20 12
10 9
4 4
5.6 9.4
192 167 194 152
500 700 800
C1 Pd (wt-%) (wt-%) 500 800
500 1.95
-
-
2.05 -
I
fresh state aged state
106 108
97 96
11.3 28.2 0.74 0.67 1.90 12.5 28.2
0.02
o
0.01 0.00 10
1000
100 log (D (A))
Fig.1. Pore size distribution of cerium-modified catalysts
m
----B ~A
228
=i 246
h~
50
250
450 Tempemture (~
650
Fig. 2. TPR profiles of zirconia catalysts
850
83
150
--,,
i
l
i
i
100
150
200
250
Temperature
(*(2)
300
Fig. 3. Detail of TPR profiles of zirconia catalysts
911 A strong reduction peak appears for catalyst B, caused by the reduction of cerium oxides. However, this peak does not appear for catalyst A. This result suggests that cerium oxide is present as a separate phase in catalysts B, whereas there is a strong interaction between Ce and Zr oxides in catalysts A. A marked influence of the procedure for the addition of cerium was observed in XRD profiles. These profiles suggest that the proportion of zirconium oxide with tetragonal structure, considered as the most stable phase, is higher for procedure A. Important changes with respect to unmodified zirconia were not observed in the case of procedure B. No cerium-containing phases were detected, probably because of the low cerium content of the catalysts. 3.2. Reaction experiments The performance of the three 1% PdO catalysts for the combustion of methane was tested both obtaining the light-off curves and with ageing experiments. The catalytic activity of the supports with no Pd loaded was checked at the reaction conditions of these experiments, methane conversions attained being negligible. Light-off curves were obtained after 15 h on stream at 450~ increasing then the temperature at 2~ from 150~ to 550~ Light-off curves were obtained both for 5000 ppmV methane and methane and SO2 (5000 ppmV CH4 and 40 ppmV SO2). The light-off curves were characterised by the parameter Ts0, defined as the temperature needed for achieving 50% conversion. The values of this parameter are shown in Table 1. Table 1. Parameters obtained from light-off curves Absence of SO2 Catalyst Ts0 (~ k 450~c (s -1) Ea (kJ/mol) A 354 2.91 78.0 B 387 1.15 62.4 U 406 0.79 40.1
. . . .
T50 (~ 470 --484
Presence of SO2 k 4s0-oc (s-i) Ea 0.20 0.05 0.24
(kJ/mol) 112.0 68.0 88.0
On the other hand, the kinetic-controlled region of the light-off curves (conversion below 80 %) was fitted using a pseudo-first order kinetic model. Although it is accepted that the best kinetic mechanisms to model the oxidation of methane over palladium catalysts are Mars-Van Krevelen ones, in a given range of conditions pseudo-first order models can be accurate enough and suitable for comparison purposes. So, considering zero order with respect to oxygen (due to its high concentration), order 1 with respect to methane, and Arrhenius dependence for the kinetic constant with respect to temperature, the kinetic equations are the following: (1)
(2)
912 (-r) being the reaction rate for methane, P the partial pressure of methane, k the kinetic constant, R the gas constant, T the absolute temperature and k0 and Ea the kinetic parameters that will be fitted to the experimental data. The values obtained for these parameters are given in Table 1. Ageing experiments were carried out at 450~ (Fig. 4). The operation temperature was chosen because in previous experiments it was observed that self-deactivation of palladium catalysts is important at this temperature. Space time was 4.5 g-h/mol CI-I4 in all the experiments. The gas feed consisted of 5000 ppmV methane in synthetic air. In order to study the poisonous effect of sulphur compounds, 40 ppmV SO2 were added to the feed in some experiments.
100
o
.~.,q
=o
j
., "~4q,94,ve,e4q,ee,e,e4q4,~4j,e e 4 ~ milL--
0
0
2
_ _
-
-
In,
L
99 --
i
i
i
!
i
i
4
6
8
10
12
14
16
tine ~) Fig. 4. Ageing experiments in absence of m B, 9 U).
8 0 2 (0
A, [] B, o U) and in presence of
8 0 2 (#
A,
As it can be observed in Fig. 4, when only methane is added to the feed, catalyst A shows the best behaviour, followed by catalysts B and U. According to this, the addition of cerium to the zirconium hydroxide increases the activity of the zirconia-supported palladium catalysts. Comparing the performance of the cerium-containing catalysts, it is remarkable that catalyst B presents a poorer performance, than catalyst A (slightly lower initial conversion and faster deactivation). This result suggests that the interaction between Pd and Ce, revealed in the TPR experiments, does not enhance the activity of the active phase (Pd). In contrast, the interaction Ce-Zr in catalysts A increases the thermal stability, considered as the main factor for preventing catalyst deactivation in these reactions [9]. On the other hand, in the experiments carried out in presence of SO2, catalyst A shows a higher resistance to deactivation and higher conversions than catalyst B. However, both perform worse than the unmodified zirconia catalyst (Fig. 4). So, it can be concluded that the thioresistance of zirconia-supported palladium catalysts is not increased by the addition of cerium.
913 In previous works of our group [8], it was observed that an important factor to the thioresistance of palladium catalysts is the ability of the support to adsorb SO2, higher adsorption capacities leading to higher thioresistance. According to this, it seems that the addition of cerium decreases the capacity of the catalyst to adsorb SO2. At this point, it is important to remark that the characterisation of deactivated catalysts (XRD, BET) does not reveal any important change in the catalyst morphology and crystalline structure. ACKNOWLEDGEMENTS
This research was financed by the Environmental Research Programme of the European Union (ENV4-CT97-0599). REFERENCES
1. S. Vigneron, J. Hermia and J. Chaouki (eds.), Characterization and Control of Odours and V OC in the Process Industries, Amsterdam, 1994 2. E.C. Moretti and N. Mukhopadhyay, Chem. Eng. Prog., 7 (1993) 20 3. J.H. Lee and D.L. Trim, Fuel Process. Tech., 42 (1995) 339 4. K. Narui, K. Furuta, H. Yata, A. Nishida, Y. Kohtoku and T. Matsuzaki, Catal. Today, 45 (1998) 173. 5. S. Yang, A. Maroto-Valiente, M. Benito-GonzAlez, I. Rodriguez-Ramos and A. Guerrero-Ruiz, App. Cat. B, 28 (2000) 223. 6. P. Hurtado, S. Ord6fiez, H. Sastre and F.V. Diez, Proceedings of the 7th North American Catalysis Society Meeting, Toronto, Canada (2001) 7. O.D. Simone, T. Kennelly, N.L. Brungard and ILL Farrauto, Appl. Catal., 70 (1991) 87 8. L.S. Escand6n, S. Ord6fiez, H. Sastre and F.V. Diez, Proceedings of the 5th European Congress on Catalysis, Limerick, Ireland (2001) 9. J.J. Spivey and J.B. Butt, Catal. Today, 11 (1992) 465
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
T h e effect of c e r i u m introduction on v a n a d i u m - U S Y
915
catalysts
C.R. Moreira 1, M. Schmal z* and M.M. Pereira 1 1 Institute of Chemistry, Federal University of Rio de Janeiro (UFRJ), Ilha do Fund~o, CT - Bloco A, sala 637, CEP: 21941-590, Rio de Janeiro, RJ, Brazil,
[email protected]. z Chemical Engineering P r o g r a m - COPPE/NUCAT/UFRJ,
[email protected] USY modified by cerium and vanadium tolerance of theses catalysts were studied. Three methodologies were used for cerium introduction: aqueous precipitation at 25 and 90~ wetness impregnation and ion exchange. The results suggest that the ratio of total cerium and cerium exchanged is very different for each catalyst, decreasing from impregnated to exchanged cerium introduction methodologies. These differences become much more evident in the presence of vanadium. All characterization, DRS in situ, DRX, TPR, BET after steam and cracking activity are in agreement that cerium and vanadium interacted. For cerium located on zeolite it is not possible to detect vanadium in the oxidation state V § by DRS in situ. On the other hand, increasing cerium into the framework the vanadium in oxidation state V § increased. In this way, the impregnation methodology leads to the most vanadium tolerance. 1. I N T R O D U C T I O N The fluid catalytic cracking unit (FCCU) is used for vacuum distillates and residues into olefinic gases. The great demand in processing heavy feedstocks and the high amounts of metals in Brazilian oils, forced to develop novel catalysts that are more resistant to metal contamination. Since the FCCU is a cyclic process, the catalyst passes through reduction and oxidation conditions. Indeed, the reductive atmosphere and coke observed in the riser favor the deactivation and reduce the life time of the catalysts. The burn off in the regenerator releases steam, CO, NOx, SOx, and other compounds. The catalyst is recycled in the reaction-regeneration (reduction-oxidation) between 10000-50000 times and, therefore, the change in the metal environment is very complex, affecting the metal oxidation state, which is an important parameter. Although the reaction conditions are well known in the literature, there are few reports concerning the environment of reduced vanadium species. Vanadium is the most important deactivation compound in FCC catalysts. In steam atmosphere, the zeolite framework is completely destroyed, and therefore the rate of make-up catalyst in the unit is very high [1,2]. The low melting point of V205 (690~ and the possibility of formation of acid species, as reported in the literature, are responsible for this deleterious effect [3,4]. The control of vanadium
916 migration and oxidation state are important to preserve the FCC catalyst. Rare earth elements have been used as metal traps, but there are still fundamental questions concerning the rare earth zeolite interactions and vanadium oxidation states. The objective is to study the effect of cerium introduced in USY zeolite and the resistance of these modified catalysts to vanadium in steam. 2. E X P E R I M E N T A L
2.1. Catalyst preparation The ultra-stable (USY) zeolite (SAR=13) was exchanged twice with a NHnNO3 aqueous solution at 343 K for 1 h, reducing the sodium content to less than 0.5% [5]. After a calcination at 873 K for 2 h in a muffle, the zeolite (HUSY) was modified by addition of cerium using three different methods: precipitation of an aqueous solution of cerium (III) chloride at 298K(PP25) and 363K (PP90); by wetness impregnation (IMP), and ion exchange (EX) as reference. A 1.5 M solution of cerium (III) chloride was poured together with a 1 M ammonium hydroxide solution in the HUSY suspension at pH 8, forming a precipitate of a cerium (III) hydroxide which, after calcination at 873K for 2 h, was transformed in cerium(IV) oxide. In the second procedure, the required amount of cerium (III) chloride was dissolved in ethanol and added to the zeolite at 268K. After drying overnight, it was calcined in a muffle at 873K for 2 h. In the ion exchange method, an aqueous solution of cerium (III) chloride was added to a HUSY suspension at 353K. After 1 h at this temperature, the system was washed with deionized-water and then calcined at 873K for 2 h. In the impregnation method, vanadium was introduced using vanadyl octanoate in toluene, followed by drying and calcination at 873K [6]. All the catalysts were submitted to a hydrothermal treatment for 3 h at 1073K using water at partial pressure of 0.3 atm. The metal contents were determined by Inductive Coupled Plasma Atomic Emission Spectroscopy (ICP-AES, Perkin Elmer 1000), after dissolving 100 mg of the sample in 5 drops and 5 ml of hot HNO3 and HF, respectively.
2.2. Textural properties The BET surface areas of the catalysts were determined in a Gemini 2360 Micromeritrics equipment. Samples were first calcined at 873K for 1 h and then transferred at high temperature to a vacuum unit at 50 mTorr for 1 h and cooled to 473K. Finally, the sample was transferred to the equipment for N2 adsorption. Isotherms were taken in a range of relative pressure of 0.06 to 0.21atm.
2.3. Temperature Programmed Reduction (TPR) The TPR analysis was carried out in continuous flow system, as described elsewhere [7]. Before reduction the catalyst was heated with an argon flux at 873K for 2 h to eliminate water. After cooling to 373K, the catalyst was reduced in a 1.53% H2/Ar flow (30 ml/min) at 10K/min up to 1273K. The hydrogen consumption was measured using a
917 thermal conductivity detector.
2.4. X-ray diffraction (DRX) The X-ray diffraction patterns were obtained using a Rigaku Miniflex equipment with Cu K a radiation of 154.18 pm. The angular interval scans were carried out over the range 14-35~ in 0.5 ~ steps and counts of 1.5 s per step.
2.5. Cracking activity The catalysts were evaluated by measuring the activity and the conversion with time on stream by using cracking of cyclohexane. Before reaction, the catalysts were reduced under flowing 10% Hz/Nz at 60 ml/min, rising the temperature at 10K/min up to 773K and held at this temperature for 5 min, according to the literature [8]. The reaction was performed at 703K, using a saturator with cyclohexane at a constant temperature of 284K, and pure hydrogen as carrier gas (20 ml/min). The reaction products were analysed by on-line chromatography (Shimadzu GC-17 A) with a packed Chrompack column (60 m length and 0.32 nm diameter) at 453K.
2.6. UV-VIS Spectroscopy in diffuse reflection mode (DRS) ~in sitm> The analysis was performed in a Varian Cary 5 spectrophotometer equipped with a Harrick diffuse reflectance chamber. The first spectra were taken after drying at 473K for 1 h, then after a reduction at 773K for 1 h passing a 20% H2/N2 flow at 10K/min. The spectra were taken at room temperature in the range 200-2000 nm, using zeolite HUSY as reference [9]. 3. RESULTS
3.1. Cerium-zeolite catalysts The preparation method and the main properties are presented in Table 1. The loss of surface area is presented as the ratio of BET area after and before the hydrothermal treatment. The surface area of the samples prepared by ion exchange and impregnation, respectively, EX and IMP, are very similar to the surface area of zeolite HUSY. However, the precipitated catalysts (PP25 and PP90) presented a lower surface area, which can probably be attributed to some blocking of pores. After hydrothermal treatment with steam, the loss of surface area of the reference zeolite was 28%, while the loss of surface area of the catalysts containing cerium was not more than 21%. The PP25 sample presented a small loss of surface area, and was therefore more resistant. The EX and IMP catalysts presented similar losses of surface area, around 20-21%. The reduction profiles catalysts are presented in Fig. 1. The PP25 and EX catalysts exhibited similar reduction profiles with a broad peak with a maximum at 836K and at 786K. On the other hand, the PP90 sample showed a profile with two reduction peaks at 897K and at 1059K. Note that the temperature of precipitation is an important parameter in this method. The impregnated (IMP) sample showed a different behavior with no significant hydrogen uptake.
918 Table 1. Properties of cerium-zeolite catalysts and the lost of area after hydrothermal deactivation by steaming. Catalyst Ce Method of cerium introduction BET area Loss of area (%) (m2/g) after steam (%) HUSY 648 28 EX 2.2 Ion exchange 631 20 PP25 3.0 Precipitation T = 25~ 612 12 PP90 3.5 Precipitation T = 90~ 609 17 ................!M P 2,0 ~ Wetness Impregnation 639 .......................................21
(o)
__ 624~
563~
~~--~--._.__..~_..
(b)
513oc
(a)
26o
~o
~
Temperature(~
~o
,doo
Fig. 1. Temperature Programmed Reduction (TPR) profiles [8]" a) EX; b) PP25; c) PP90, d) IMP.
3.2. Vanadium-cerium-zeolite catalysts Table 2 presents the main properties of these samples, the loss of surface area after hydrothermal treatment and the surface area after a TPR analysis. The H2 consumption was calculated based on the metal content (~mol H2/~tmol Ce + V). In addition, a ratio of extrapolated and observed H2 uptake from the TPR profile (H2 ext/obs) was calculated, taking in account the H2 consumption of isolated cerium and vanadium oxides in the bimetallic system, in order to see if interaction between both occurred or not, or if they are isolated particles. As observed, there is a loss of the surface area after steam treatment, which markedly depends on cerium addition method. The PP25V catalyst after steam treatment led to a higher damage of the zeolite structure. Although the TPR results of the PP25V and EXV samples indicate some similarity, the BET surface areas suggest noteworthy differences in their morphology, due to the cerium location. Indeed, the impregnated sample (IMP) was the only one that protected the zeolite structure and presented the lowest loss of surface area (25%).
919 Table 2. Properties of cerium-vanadium-zeolite catalysts including: BET area after TPR measurement, H2 consumption during TPR analysis and cracking activity of cyclohexane. Catalysts V Loss of area Area after TPR ~mol H2/ C6H12 Ha (ppm) after steam (mZ/g) ~mol Ce + V (ext/obs) (ext/obs) (%) activity HUSYV EXV PP25V PP90V IMPV
3000 2000 3900 2600 2000
33 31 45 28 25
455 33 239
1.15 0.78 0.48 0.57 0.28
4.23 3.35 2.82 2.75
0.5 1.0 1.2 3.2
However, the BET surface area of the catalysts after TPR experiment presented a drastic crystalline damage of the EXV catalyst and some crystallinity retention for the IMPV catalyst. These results agreed with the X-ray diffraction pattern after TPR measurements, as shown in Fig. 2. The crystalline difference for both catalysts was 20 %. Vanadium reduction was strongly affected on the modified zeolites. The amount of Hz consumption in the TPR analysis related to the metal content (~tmol Hz/~tmol Ce + V) decreased largely. Moreover, the ratio of the extrapolated and observed Ha consumption (Hz ext/obs) was higher on the IMPV catalyst compared to the others. This behavior was supported by the UV-VIS spectroscopy data (Figs. 2 and 3), showing similarity of both profiles, the dry catalyst and after the reduction treatment. Both did not present absorption bands attributed to d-d transition in the 800-1800 nm region. These results suggest that almost all the vanadium is in the oxidation state V +5 in the IMPV catalysts. On the other hand, the ratio H2 (ext/obs) < 1 of the EXV catalyst indicates an increase of the amount of hydrogen consumption. The UV-VIS spectra on this catalyst (Fig. 3) support this result, showing a large band in the of d-d transitions region. According to the literature [3,10], the peak at 800 nm corresponds to the octahedral vanadium species in oxidation state V +4. The catalytic activity, C6Hlz (ext/obs), decreases depending on the preparation method (EXV > IMPV) of the catalysts, in opposition to the increase of H2 (ext/obs) ratio. The highest activity observed on the EXV catalyst indicates that, in this case, the acid sites of the zeolite were better protected by the presence of rare earth than on the IMPV catalyst. The activity of the cyclohexane cracking are presented in Table 2. The C6H12 (ext/obs) activity is presented as the ratio between the extrapolated and the observed activity. The extrapolated activity was calculated assuming isolated metals without interaction, while the observed value was obtained experimentally, meaning that the extrapolated activity is the sum of the isolated activity, pondered to the metal content in the catalyst.
920 0,10.
~3.
..-,~ 0,05
(~ o,1 am
(a) c~o 800
1000
1200
1400
V~de-cJh(n~
1630
18~0
2[]30
i 800
,
i 1000
,
! 1330
,
! 1400
,
i 1600
,
i 1800
,
2000
Figs. 2. UV-VIS Spectroscopy in situ after Fig. 3. UV-VIS Spectroscopy in situ after dry (a) and reduction (b) treatment for dry (a) and reduction (b) treatment for IMPVcatalyst. EXV catalysts. 4. DISCUSSION 4.1. State of cerium in the zeolite The results have shown that the introduction of a rare earth element in the zeolite really depends on the preparation method that directs the location of this metal in the zeolite. As expected, when cerium oxide is introduced by impregnation, the surface area decreases due to formation of particles blocking the pores. On the other hand, the ion exchange method would favor the introduction of cerium in the zeolite framework and, therefore, it does not affect the surface area but favors the dispersion of cerium. Noteworthy is the influence of the temperature in the precipitation method during the introduction of cerium in the zeolite. As seen, the BET surface area of the modified zeolites with cerium at 298K (PP25) and 363K (PP90), decreases slightly compared to HUSY, or to ion exchanged catalysts(EX). Therefore, the best method is when cerium is introduced in the zeolite framework and, therefore, a better dispersion of cerium is expected. The TPR results have shown very similar profiles for EX and PP25 catalysts. It supports the previous results that cerium on PP25 catalyst was exchanged in the framework. However, in this case, after steam treatment the destruction in the zeolite framework was very large. It suggests that the cerium environment in the PP25 catalyst is different compared to the EX catalyst. On the other hand, the reduction of the PP90 catalyst showed a profile which is very similar to the reduction of cerium (IV) oxide, and this suggests the presence of superficial cerium species and bulk species [11,12,13]. Therefore, there are larger cerium aggregates on the external surface of the zeolite and
921 isolated cerium species. On the contrary, the impregnation of cerium evidences the reduction of external particles and the existence of different easily reducible cerium bulk species. The presence of different cerium species would probably influence the vanadium environment and the catalytic behavior of these catalysts containing both elements. Indeed, the Hz(ext/obs) ratio which is a measure of the reduction degree and therefore indicates if there is an interaction with the zeolite or between cerium and vanadium, exhibited different values, depending on the way of introduction and species formation. The catalyst treated with steam, EXV, presented a low H2 (ext/obs) ratio, which indicates a better reduction. On the other hand, the impregnated catalyst (IMPV) presented a high H2 (ext/obs) ratio, and thus low reduction. This could explain the indication that an interaction occurred during the treatment, with the formation of bimetallic or alloys or even the formation of aluminum silicate-metal interaction. DRS measurements support the TPR results. The impregnated catalysts and steam treated (IMPV) did not show the presence of V +4 after the reduction. Probably, the hydrogen consumption in the TPR profile is due (a) to the reduction of cerium. The band (b) in the d-d transition can be attributed to the formation of alloys like cerium vanadate, according to the literature [14]. Baugis et al. [15] reported that the presence of vanadate with rare earth decreases the diffusion of vanadium in the zeolite structure [14]. The existence of these compounds Fig 4. X-ray diffraction patterns of a) EXV may affect the oxidation state, the and b) IMPV catalysts after TPR dispersion, morphology and location measurements. of cerium species in the catalyst. The DRS spectrum of the EXV catalyst after reduction showed the presence of vanadium in V § oxidation state. Based on thermodynamics redox results, it is expected that when vanadium and cerium present some interaction, the last one should present an easier reduction. The reduction of cerium is favored because of its higher potential (1.64 eV), that should maintain vanadium in the V § oxidation state [16]. Therefore, the formation of rare earth vanadate is favored. On the IMPV catalyst, where probably cerium is dispersed over the zeolite, the reduction process would be preserved. On the other hand, for the EXV catalyst, the reduction of cerium exchanged in the presence of vanadium leads to an easier reduction of both components. But it is not possible here to distinguish and to quantify the formation of V +4 and cerium in a Ce +3 oxidation state. The DRX diffractograms after TPR suggest a possible model of crystalline destruction of the catalysts using their reduction potentials. The highest crystalline damage is observed on the IMPV catalyst compared to the catalyst containing exchanged cerium (EXV) that is completely amorphous (Fig. 2). This sustains the proposed model that the introduction of cerium by wetness impregnation leads to more cerium species outside the
922 zeolite structure. If cerium species are exchanged in the framework, the catalyst should stay amorphous like the other one, due to the formation of vanadates inside the zeolite framework. The cracking activity suggests that the amount of active sites of the zeolite poisoned by vanadium depends on the cerium location in the zeolite. The higher the activity value of C6H12 (ext/obs), the lower is the poisoning effect of vanadium. (Table 2). Therefore, the most active catalysts are those that protect the zeolite better, and this was observed on those catalysts that contain exchanged cerium, resulting in a higher proximity between cerium and vanadium. This is because the cerium species exchanged in the acid sites would not allow that vanadium interact with the acid sites, keeping vanadium close to them. On the other hand, on the IMPV catalyst, part of the actives sites of the zeolite would be affected by vanadium. The literature has reported methodologies to quantify the oxidation state of vanadium species using TPR [ 17] and EPR/DRS [10]. In summary, this work shows for the first time a marked influence of cerium species on the vanadium reducibility. Since rare earth elements in FCC catalysts depend on different preparation morphologies, it is necessary to develop a model to quantify the oxidation state of vanadium. 4.2. Resistance of zeolite under steam
The BET results after steaming show that the catalysts containing cerium exhibited higher hydrothermal stability. The literature reported that rare earth exchanged in zeolites enhance the thermal and hydrothermal stability [5]. After the introduction of vanadium, it is possible to verify that the catalyst containing cerium introduced by impregnation (IMPV) protected the zeolite structure. Therefore, the protection depends on the dispersion of cerium species on the zeolite surface, decreasing the vanadium mobility. However, the mechanism which explains the damage of zeolite by vanadium is unclear [1,3]. The higher cerium contact increases the framework damage. Probably the steam effect on PP25V catalyst could explain the non-homogeneous distribution of cerium exchange in the framework as it is in the EXV catalyst. The heterogeneous cerium distribution leads to a high local damage and a higher effect of steam. In this way, it is possible to observe that the preservation of the zeolite structure depends very much on the cerium location in the catalyst. Probably vanadium introduction first localizes vanadium outside the zeolite framework, as expected in real cracking catalyst, which increases the probability of formation of cerium-vanadium compounds on the IMP catalyst. 5. CONCLUSIONS The different ways of cerium introduction lead to different morphology and location on the zeolite. The ratio of total cerium and cerium exchanged into zeolite decreases from cerium impregnated to cerium exchanged. The precipitated catalysts lead to intermediary systems between IMP and EX catalysts, showing exchanged and superficial cerium species. After vanadium introduction, different behaviors in the vanadium reduction/oxidation capacity were observed depending on the cerium location and the
923 morphology, but cerium exchanged and dispersed on zeolite presented high interaction with vanadium. The last modified zeolite did not show, after the reduction treatment, vanadium in low oxidation state, by in situ UV-Vis spectroscopy. On the other hand, V +4 was observed when cerium was exchanged in the zeolite. Finally, cerium provides better thermal resistance to the zeolite and, after hydrothermal treatment with and without vanadium, the crystallinity of the zeolite depends on the cerium species in the catalyst. Zeolite modified with cerium are good models compounds to study the vanadium oxidation state. ACKNOWLEDGMENT
CNPQ and CTPETRO are gratefully acknowledged for financial support. REFERENCES o
2. 3. 4.
5. 6. 7. 8.
9. 10. 11. 12. 13. 14. 15. 16. 17.
M. Torrealba, and M.R. Goldwasser, Appl. Catal. A: General, 90 (1992) 35. J. Biswas and I.E. Maxwell, Appl. Catal., 63, (1990) 197. C.A. Trujillo, C.A. et al., J. Catal., 168 (1997) 1. G. Martino, Stud. Surf. Sci. Catal., 130 (2000) 83. M.L. Occelli, Catal. Rev.-Sci. Eng., 33 (3-4) (1991) 241. B.R. Mitchell, Ind. Eng. Chem. Res. Dev., 19, (1980) 209. L.T. Santos et al, Stud. Surf. Sci. Catal., 139 (2001) 343. J. Abbot, J. Catal., 123 (1990) 383. M.A. Bafiares, M.A. et al., S. Surf. Sci. Catal., 130 (2000) 3125. G. Catana et aL, Phys. Chem. B, 102 (1998)8005. A. Piras, A. Trovarelli and G. Dolcetti, Appl. Catal. B: Environ., 28, (2000) L77 B. Ernst, L. Hilaire and A. Kiennemann, Catal. Today, 50, (1999) 413. F. Giordano, J. Catal., 193 (2000) 273. R. Zhuo, F. Wang and W. Wu, in: 215th National Meeting American Chemical Society, Dallas, 1998 A. G.L. Baugis et al., in: 11~ Congresso Brasileiro de Catfilise e 1~ Congresso de Catfilise do Mercosul, 2 (2001) 916. J.G. Nery et al., Zeolites, 18 (1997) 44. E.F. Souza-Aguiar et al, Zeolites, 15 (1995) 620.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
925
Rh-Co mordenite catalysts for the selective reduction of NO by methane C.E.Quincoces, M. Incollfi, A. De Ambrosio, M.G.Gonzfilez 1 CINDECA (CONICET, UNLP), 47 Nro 257, 1900 La Plata, Argentina
The performance of RhCoMOR catalysts for the NO selective reduction by methane in the presence of excess oxygen has been determined. Incorporation of 1% w/w of Rh and 5% w/w of Co to mordenite provides a solid that is active and selective for SCRCH4 reaction showing the highest activity between 400 and 500~ with a CH4/NO ratio=l and 2% of oxygen in the feed, and a GHSV= 20,000 h 1. The Co-catalysts were prepared by ionic interchange of Na Mordenite (MOR) with aqueous solution 0,025 M of Co acetate and, after calcination, Rh was added to the CoMOR sample by incipient wetness impregnation at room temperature. The Co mono- and bimetallic catalysts showed a good performance for the reaction of NO reduction by methane. The addition of 1% of Rh improved the activity of SCR-CH4 on CoMOR catalyst, decreasing the temperature of maximum NO conversion. The TPR characterization of RhCoMOR catalyst showed a signal that could be attributed to intermetallic compounds, and another signal at high temperature, attributed to Co § which would be responsible for the catalytic activity. 1. I N T R O D U C T I O N The selective reduction of nitrogen oxide with methane is one of the most promising technologies to control the NOx, both in stationary emission sources, where it represents an alternative for the use of ammonia as reducer, and in mobile sources that use natural gas. Methane offers the benefits of low cost and wide availability compared to other hydrocarbons and is much less corrosive than ammonia. The use of methane as NO reducer offers the opportunity of reduced capital and operating costs compared to other reducing agents. This technology may allow to find a solution to exhaust from natural gas fired power plants and lean-burn natural gas engines. In both cases, methane may be available at low level in the exhaust or can be easily injected into exhaust downstream of the burner from on-hand supply of natural gas. It has been shown by Li and Armor [1,2] that the selective reduction of NOx by methane is produced with high selectivity on zeolitic materials interchanged with Co. Even though the presence of zeolites plays an important role by enabling a high dispersion of Co, its disadvantage is that it is very sensitive to the presence of water vapor. The catalyst systems for the SCR of NOx with methane that will have the greatest impact and 1 Corresponding author: Fax:+54-221-425-4277. e-mail:
[email protected] 926
applicability must be both highly active and stable. Bicomponent systems, generally a transition metal and noble metal [3-8] have been suggested to solve inconveniences resulting from the presence of water vapor and higher temperatures. This paper studied Co catalysts interchanged on mordenite to which Rh is added over the activity. An analysis of Rh effect on the activity and selectivity for the selective reduction of NO by methane in oxidizing medium is presented. 2. E X P E R I M E N T A L The CoMOR catalysts were prepared by means of ionic exchange of NaMordenite (NORTON 2900, Si/AI = 5.9) with 0,025 M aqueous solution of Co acetate by stirring 24 h at different temperatures (25 and 70~ in order to modify the metallic content. The interchanged material was washed and dried for 24 h and then calcined at 400~ for 14 h following the heating pattern of 2~ min -1, with 2 isothermal intervals of 90 min at 110 and 210~ according to [8]. Part of the interchanged material was treated at 25~ in a second stage, previous to calcinations, with a fresh solution for another 24 h at room temperature. Bimetallic samples, having between 0.1 and 1% of Rh, were obtained from the monometallic catalysts by incipient wetness impregnation with aqueous solution of CI3Rh.2H20 at room temperature. After adding Rh, the solid was calcined at 350~ for 2 h with a heating rate of 2.5~ min 1. The samples were denominated Rh(m)Co(n)MOR, where m and n represent the percentage of the elements contained in the catalyst. The monometallic and bimetallic solids were characterized by adsorption of Nz, energy-dispersive X-ray analysis (EDX), temperature programmed reduction (TPR) and Xray diffraction (XRD).. The BET surface area and pore volume distribution were measurement by nitrogen adsorption in an Accusorb 2100E Micromeritics analyzer. The chemical analysis using EDX were carried out on a DXPRIME 10 system connected to the scanning electron microscope. The TPR experiments were carried out in a conventional equipment. The samples (0.06g) were heated from R.T. to 1000~ at a rate of 10~ min -1 in a 10 % H2-N2 stream at a rate of 20 ml min -1. Sample diffractograms were obtained with a Philips PW 1732/10 equipment and CuKct radiation at a rate of 20/min The measurements of catalytic activity were carried out in a fixed bed reactor having 0.18 g of catalyst, in the temperature range 200-650~ and at a GHSV of 20000 h -1. The feed mixture to the reactor consisted of 1000 ppm of methane, 1000 ppm of NO, 2% of Oz and He as balance. The feed and effluent were analyzed by an "on-line" gas chromatograph using a column CTR1 at 40~ and a thermal conductivity detector. The conversions of NO and hydrocarbon were calculated from the amount of N2 and CO2 produced. 3. RESULTS AND DISCUSSION During the preparation stage, the effects of temperature and the number of interchanges on the Co amount added to zeolite were determined. Table 1 shows the Co
927 content in the different samples. The Co content was determined by EDAX, and the ion exchange level was calculated, assuming that one divalent cobalt ion is exchanged for two monovalent Na cations. It is observed that, as long as the successive interchanges scarcely affect the content of Co interchanged, the temperature has a distinct effect, allowing the doubling of the percentage of active component. In agreement with different authors, Co(5)MOR was selected to prepare the bimetallic RhCoMOR solids. Table 1. Catalysts composition Catalysts Ion exchange Temperature(~ Co(4)MOR 25
Ion exchange steps .... 1
Co(w/w%) 4.3
Ion exchange level 90
Co(5)MOR
25
2
5.1
106
Co(10)MOR
70
1
9.8
203
The monometallic and bimetallic catalysts showed to be active and selective for the reaction of NO reduction by methane. CoMOR was active at higher temperatures than RhCoMOR catalysts. All the samples show good stability during a test of 24 h at 500~ The NO and methane conversions for the different samples are shown in Figs. 1 and 2.
35 30 25 A
0 =" x
20 15 10
200
400 T ern
600 p eratu
ra
800
(~
Fig. 1. Nitric Oxide conversion as a function of temperature on: (A) Rh(1)Co(5) MOR, (11) Co(4)MOR,(O) Co(5) MOR,(O) Rh(1)Co(5)MOR reduced
928 The methane was oxided only to CO2 and H20 over the catalysts; CO was not detected as reaction product. At the maximum NO conversion, the methane oxidation was near 100%. Bi- and monometallic samples achieved the maximum methane conversion at 400 and 600~ respectively. The results obtained in this study over CoMOR samples are in general agreement with the data reported by several authors [2,7]. Gutierrez et al [7] reported that Co(2)MOR gives 30% of NO conversion at 450~ but with only 6500 h-1 GHSV.
100 80
o'-?,
60
I
o X
40
20 0 4-,,._,l~-~____~r~ i v , , , , '='` ~
200
~
300
.r ..
I
I
400
500
600
700
T e m p e r a t u r e (~ Fig. 2. Methane combustion as a function of temperature on: (A) Rh(1)Co(5) MOR, (11) Co (4) MOR, (O) Co(5) MOR,(O) Rh(1)Co(5)MOR reduced The addition of Rh improves the activity of Co/MOR catalysts, increasing the NO conversion to Nz and decreasing the temperature of maximum conversion. At 430~ the NO reduction increases from 19 to 32% by the addition of 1% of Rh to the monometallic catalysts. The reduction of bimetallic catalyst decreases considerably the NO conversion and enables the methane combustion. At the opposite to observations of Petunchi et al. [9] for PtCoMor catalysts, the reduction does not increase the catalyst activity of RhCoMOR sample. The bimetallic catalysts submitted to Hz-He stream at 400~ during 2 hours show a great drop in the NO conversion without modification of the temperature window for maximum activity. At low temperatures, the reduced and unreduced samples present the same behaviour for the NO conversion. At temperatures higher than 400~ the NO conversion decreased for the reduced catalyst. This behavior can be associated with the increment of methane combustion at high temperatures over the reduced sample, which would affect the selectivity of the reaction. In order to analyze the effect of the Rh content, bimetallic samples containing 5% of Co and between 0.1 and 1% of Rh were prepared and tested for the SCR-CH4 reaction.
929 It is observed in Fig. 3 that the performance of the bimetallic samples is affected by the amount of Rh incorporated to CoMOR. Whereas the addition of 1% Rh increased the conversion of NO to Nz without modifying the selectivity, the incorporation of smaller amount of Rh to CoMOR decreased the catalytic activity (Table 2) and increased the temperature of maximum NO conversion. These results suggest that Rh has a promoting effect on the activity of CoMOR for a content of 1%.
100 c
80
.o 03
60
>
40
o
20
I,.,.
tO
"
- - ~ , - - r - ' - -r-
0
--=l
~
200
i~
j
J
400
600
800
Temperature(~ Fig. 3. Effect of Rh content on the NO (black symbols) and CH4 ( white symbols) conversion on (0) Rh(1)Co(5)MOR, (0) Rh (0.5)Co(5)MOR, (A) Rh(0.1) Co(5)MOR. Table 2. Co and RhCo Mordenite Catalytic Behavior a Catalyst
XNO (%)
Tmaxb
XCH4 ( % ) c
Selectivityd
Co(4)MOR
29.1
500
53.8
0.54
Co(5)MOR
30.0
490
54.6
0.55
Rh( 1) Co (5) MOR
33.0
438
61.5
0.536
Rh(0.5)Co(5)MOR
15.0
453
82.1
0.18
Rh(0.1)Co(5)MOR
10.5
454
49.2
0.20
Rh (1)Co(5)MOR red.
12.0
435
84.3
0.14
a: reaction conditions: GHSV=20000 h-1, NO=1000 ppm, CH4=1000 ppm, 02=2%. b:Maxima NO conversion temperature. C" CHnto COz conversion, d 9NO/CH4 selectivity
930 At about 440~ the NO conversion to Nz is near 35% for the Rh(1)Co(5)MOR, suggesting a promoter effect of Rh. MonometaUic samples containing the same % of Co as bimetallic ones show a drop of NO conversion near to 20%, when evaluated under the same conditions. In order to understand the effect of the active sites on the SCR-CH4 reaction, the different samples were analyzed by TPR technique. Fig. 4 shows the reduction thermograms of the different samples. For comparison, the TPR profiles of Na mordenite and mordenite impregnated with Rh are also included in the figure. The TPR profiles of mono and bimetallic solids show a peak whose maxima is observed about 1000~ This signal can be assigned to the presence of interchanged CO +2 which would be responsible for the formation of NO2. The NOz reacts with the adsorbed methane in the acid places, beginning the SCR-CH4 reaction [8,9]. These Co +2 ions at exchange sites may be considered as the active sites. The peak at 400~ in the Rh(1)Co(5)MOR catalyst, compared with Rh/MOR sample, is attributed to the reduction of Rh. The shoulder at the higher temperature side present in the bimetallic sample, can be attributed to intermetallic compounds. This intimate contact of cations would contribute to the promoting effect of Rh on CoMOR activity. During the calcination step, the formation of CO304 is quite probable, but the peak at lower temperature (230 and 390 ~ assigned by Wang et al. [11] to the reduction of Co oxo-ion and Co304, is not observed. This result suggests that there is no formation of Co +3 during the preparation step of the catalysts.
E 0.4
~
m c
c d
~ o.a CJ
e
~ I 0.2
0
I
500 Temperatura (oC)
I
1000
Fig. 4. TPR patterns of a) Rh(1)Co(5)MOR, b) Rh/NaMOR, c) Co(5)MOR, d) Co(4)MOR, e)NaMordenite The diffractograms obtained for the mono- and bimetallic samples were similar to those of the original Na-mordenite. No signal corresponding to Co and Rh species were observed in the diffractograms of calcined and reduced samples, which indicates that the metallic particles have a size below 40 A, detection limit of the instrument. Surface area measurements and pore volume confirm that the mordenite structure is not affected during the calcination treatment.
931 4. CONCLUSIONS The results obtained indicate that the addition of a small amount (1%) of Rh to the CoMOR catalysts improves its performance for the reduction of NO to Nz in excess of Oz, increasing significantly the NO conversion to N2 and decreasing the temperature of maximum conversion. Intermetallic compounds and Co +z at exchange positions would be responsible for the improved performance of the catalysts. These solids could be an alternative in the development of effective catalysts for the SCR-CH4 reaction in excess of O2 ACKNOWLEDGMENTS. The authors thank CONICET and UNLP for financial support for this research project.
REFERENCES 1.Y. Li and J.A. Armor, Appl. Catal. B, 1 (1992) L31. 2. Y. Li and J.A. Armor, Appl. Catal. B, 2 (1993) 239. 3. H. Hamada, Y. Kintaichi, M. Sasaki, T. Ito and M. Tabata, Appl. Catal., 75 (1991) L1. 4. E. Kikuchi et al., Catal. Today, 27 (1996) 35. 5. H. Ohtsuka and T. Tabata, Appl. Catal. B, 21 (1999) 133. 6. M. Ogura, Y. Sugiura, M. Hayashi and E. Kikuchi, Catal. Lett., 42 (1996) 185. 7. L. Gutierrez, A. Ribotta, A. Boix and J. Petunchi, in 11th International Congress on Catalysis, J.W. Hightower, W.N. Delgas, E. Iglesia and A.T. Bell (eds.) Elsevier, Amsterdam, 1996, 631. 8. A.V. Boix, M.A. Ulla and J.O. Petunchi, J. Catal., 162 (1996) 239. 9. J-Y. Yan, H.H. Kung, W.M.H. Sachtler and M.C. Kung, J. Catal., 175 (1998) 294. 10. L. Gutierrez, A. Boix and J. O. Petunchi, J. Catal., 179 (1998) 179. 11. X. Wang, H-Y. Chen and W.M.H. Sachtler, Appl. Catal.B, 26, (2000) L227.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
933
Surface characterization of WO3-TiOz/Alz03 catalysts and reactivity o n selective catalytic reaction of NO by NH3 Silvia Egues a, Neuman S. de Resende a and Martin Schmal a* aNUCAT/PEQ/COPPE - UFRJ, Cidade Universitfiria, Rio de Janeiro, Brasil. Caixa Postal 6 8 5 0 2 - CEP 21945-970. E-mail:
[email protected] In this work, the structural and physico-chemical properties of WO3/TiO2, WO3/A1203 and WO3-TiOz/A1203 catalysts and their activity in the SCR of NO and NH3 were investigated. The catalysts were prepared with titanium content between 7 and 13 % (w/w), keeping tungsten at 7 % (w/w). The catalysts were characterized by BET, XRD, UV-Vis DRS, Raman spectroscopy, and the catalytic activity was observed by TPSR. The results suggested the presence of well dispersed tungsta and titania phases on the support forming amorphous phases or microcrystallites. DRS spectra showed that the WA sample has only tetrahedral tungsten species, but the incorporation of TiO2 to the support leads to a higher content of octahedral tungsten species. Raman spectra of mixed supports indicated that TiO2 forms a surface layer on alumina, which has a polymeric structure. The TPSR profiles showed that the catalytic activity increases in the following order WO3/TiO2 > WO3TiOz/AI203 > WO3/m1203 and that the presence of oxygen increases the NO reduction rate. 1. I N T R O D U C T I O N Commercial oxide catalysts for the selective catalytic reduction (SCR) of NOx by NH3 usually consist of TiO2 (anatase) as support, V205 as active component, and WO3 (or MOO3) as promoter. Vanadia is responsible for the activity and selectivity of the catalyst in NOx reduction but also favors the transformation of anatase to rutile, leading to sintering, loss of surface area and to undesired oxidation of SO2 to SO3 in the case of sulfur fuels [1]. Accordingly, the V205 content is generally low, 0.3-1.5% w/w. WO3 is employed in large amounts (--10% w/w) increasing the working reaction temperature window, the thermal and mechanical stability. An additional advantage is that WO3 stabilizes the titania structure and hinders both surface area loss of anatase and its transformation to rutile [2]. Therefore, the use of tungsta as active phase seems a good alternative for SCR catalysts. Attention has been paid to titania-alumina supported catalysts, which use high surface area and thermal stability of inert alumina to sustain the structure of titania, enhancing the textural properties of the catalysts [3,4]. The literature presents a number of studies dealing with the characterization of WO3-AIzO3 and WO3-TiOz catalysts; however, little has been reported on the properties of the WO3-TiO2/AI203 system. It was found that the support composition and the physico-chemical properties determine the tungsten species at the surface, presenting tetrahedral species on alumina, and octahedral species on titania [5].
934 It seems interesting to study the changes that occur in the tungsten surface species as the TiO2/A1203 support changes with variable amounts of titania until monolayer, and the relation between the type of tungsten surface species with the catalytic activity in SCR reaction. Therefore, this work reports the preparation of TiO2/A1203 and WO3-TiO2/A1203 with various TiO2 contents, and their characterization by different techniques (BET, X R , DRS, Raman spectroscopy). Also, an attempt has been made to relate the results from these characterizations to the changes of the catalytic activity in NO reduction with NH3 observed by TPSR. 2. EXPERIMENTAL
2.1. Supports and catalysts preparation The TiO2/Al203 supports were prepared by dry impregnation method with variable amounts of TiP2 (7.2%, 9.7% and 13.5% w/w) over alumina calcined at 600~ [4]. Titanium isopropoxide (Ti(OC3H7)4) from Aldrich was used as precursor. Titanium dioxide (anatase) was obtained by slow hydrolysis of the isopropoxide. The supports were named 7TA, 10TA and 13TA according to the TiP2 content. The series of W catalysts containing 7% w/w WO3 were prepared using ammonium paratungstate ((YH4)6H2W12040) as precursor in wet impregnation. Later on, all the solids were dried at 120~ for 17 h and calcined at 500~ for 5 h. The catalysts were named W7TA, Wl0TA and Wl3TA. 2.2. Supports and catalysts characterization The specific surface area and pore volume distribution were obtained with nitrogen physisorption using BET method in a Micromeritics ASAP 2000. X-Ray diffraction (XRD). A Philips PW 1410 powder diffractometer with CuK~ radiation (40 KV, 30 mA) plus a Ni filter was used with an angular range varying between 19 to 80 ~ in a 0.02 ~ per step and 0.8 seconds counting per step. Diffuse reflection spectroscopy (UV-DRS). UV-Vis diffuse reflectance spectra were recorded in a Varian Cary 5 spectrophotometer (Harrick Scientific) with a diffuse reflectance accessory of Praying Mantis geometry. The samples were set in holds with 2 mm thickness. The spectra were recorded in the 200 nm and 800 nm range with a scanning speed of 1800 mm.min1 at room temperature against supports as reference. Laser Raman spectroscopy (LRS). The spectra were recorded on a Nicolet 950 FTRaman spectrometer instrument, equipped with a nitrogen cooled Ge detector. A Nd:YAG laser (1064 nm) was used as excitation source. The measurements were performed with a power at the sample of 100-200 mW in order to avoid decomposition and thermal effects. The samples were rotated to provide a noncontinuous irradiation of any given spot on the samples. The spectral slit width was typically 4 cm -1. Temperature-programmed surface reaction (TPSR). Measurements were performed in a flow apparatus with a BALZERS quadrupolar mass spectrometer to monitor effluent products versus temperature. Samples of 300 mg were used in each measurement. The gases employed were He (99.9%) and mixtures 4% NH3/He, 1% NO/He and 5% O2/He. TPSR measurements were obtained after NH3 adsorption at 70~ under continuous flow until base line stabilization. The reactor was heated to 500~ at 20~ and kept at this
935 temperature for lh. Concentration of desorbed compounds were determined from peak intensities, as follow: NH3 (m/e = 15), NO (m/e = 30), H20 (m/e = 18), N2 (m/e = 28), 02 (m/e=32) e N20 (m/e = 44). 3. R E S U L T S AND DISCUSSION
3.1. Textural properties The structural and morphological properties of the catalysts were investigated. Table 1 presents the specific surface area (SBET), pore volume (Vp), TiO2 content of the mixed supports and the WO3 content at the surface of the catalysts. Assuming a monolayer capacity of 0.083 %w/w TiO2/m 2 [6], the titanium coverage (0wi) was calculated considering TiO2 content and specific area of the samples. As shown in Table 1, the theoretical TiO2 monolayer (10.4 [amol Ti/m 2) was not exceeded in any mixed support samples. The results of the surface area measurements of the supports present a small decrease relative to alumina, as follows: A1203 (SBET--197 m2/g) and 13TA (SBET--180 m2/g). The addition of WO3 caused only a slight drop of the surface area of the catalysts relative to mixed supports, as seen for W13TA (SBET--174 m2/g). However, higher surface areas were obtained for Ti-A1 mixed oxides than for pure TiO2 (101 m2/g). All the catalysts showed a monomodal pore volume distribution curve, which gives an indication of the absence of segregation of the different oxide supports in the catalysts formulation. 3.2. X-ray diffraction (XRD) The diffractograms of the catalysts are shown in Fig. 1. The XRD patterns of the WO3/ZiO2 and WO3/A1203 catalysts present peaks characteristic of anatase and alumina, respectively. The diffraction lines of the catalysts WTA showed peaks related to T-alumina phase. Only the W13TA sample showed an incipient peak of anatase phase (20 = 25,2 ~ corresponding to a slight segregation of crystallized titania. No diffraction lines of any tungsten phase were detected in the different samples. The above observations indicate that, at this measuring scale, both titania and tungsten phases were well dispersed over the different catalysts, forming either an amorphous phase or microcrystallites not detectable by XRD. Table 1 Textural properties of the catalysts samples Sample SBET Vsp TiO2 0.997, 63000 m2/g) was prepared by carbothermal hydrogen reduction method. The MoO3 precursor and MozC have been characterized by X-ray diffraction, nitrogen adsorption, high-resolution transmission electron microscope (HRTEM) and temperature-programmed reduction-mass spectroscopy (TPR-MS). The data show that nanostructured 13-MozC can be formed on the ultrahigh surface area carbon materials by carbothermal hydrogen reduction at 700 ~ The particle sizes of [5-MozC increase with the increase of the temperature of carbothermal hydrogen reduction. The carbothermal hydrogen reduction includes two successive steps: reduction of MoO3 precursor by hydrogen and reaction between partially reduced molybdenum oxides and surface carbon atoms of carbon materials. 1. INTRODUCTION The transition metal carbides have received considerable attention as advanced materials, especially as catalytic materials, because of their unique physical and chemical properties [1, 2]. Many methods including gas-phase reactions of volatile metal compounds, reaction of gaseous reagents with solid state metal compounds, pyrolysis of metal complexes and solution reactions, have been developed for the preparation of high surface area carbides [2, 3]. These carbide materials show exceptionally high activity in hydrogen-involved reactions [4-10]. It is also found that the catalytic properties of carbide materials strongly depend on their surface structure and composition, which are closely associated with the preparation methods. Among those preparation methods, the typical one is the temperature-programmed reaction between oxide precursors and flowing mixture of hydrogen and the carbon-containing gases, such as CH4 [11-13], C2H6 [14, 15], C4H10 [16], and CO [17]. However, there exist also some problems in this method of preparation: the carburization and passivation processes must be carefully controlled; and the resultant carbide surface is usually contaminated by polymeric carbon from the pyrolysis of the containing-carbon gases. The carbon is blocked in the pores, covers the
976 active sites, and is difficult to remove. Recently, Ledoux and coworkers developed a novel synthesis route to preparing high surface area carbides which can avoid the formation of carbon residues on the surface [18-20]. It involves the reaction of solid carbon with vaporized metal oxides at very high temperatures, even above 1000 ~ Activated carbon was used as the carbon source and the final carbides appear to remain the characteristic of porous structure of the activated carbon. Mordenti et al. [21] modified this method and prepared activated carbon-supported MozC samples under moderate temperatures. But this method is still in an early stage of development. A broad range of supports and metal compounds need to be tested and studied in detail because metal compounds can be highly dispersed on carbon materials, and the carburized materials are potential catalysts in hydrogenation reaction. In the present work, we have tried to prepare nanostructured MozC using ultrahigh surface area carbon material (>3000 mZ/g) [22], a kind of novel carbon material with uniform pore sizes, as carbon source and template by carbothermal hydrogen reduction. 2. EXPERIMENTAL Ultrahigh surface area carbon material is made by a direct chemical activation route in which petroleum coke is reacted with excess KOH at 900 ~ to produce carbon materials containing potassium salts. These salts are removed by successive water washings. The surface area of the carbon materials measured by BET method is about 3234 mZ/g and the pore volume is about 1.78 m3/g (Table 1). Table 1. Surface area and porosity of HSAC, Mo/HSAC and the samples with carbothermal hydrogen reduction Sample
BET surface area (mZ/g)
Pore volume (cm3/g)
HSAC
3234
1.78
Mo/HSAC-RT
2446
1.31
Mo/HSAC-600
2505
1.38
Mo/HSAC-700
2341
1.35
Mo/HSAC-800
2180
1.30
The ultrahigh surface area carbon material was impregnated in a rotary evaporator at room temperature with aqueous solution of ammonium heptamolybdate. The molybdenum content is 10 wt. %. After evaporating and drying in air at 120 ~ overnight, the materials were transferred to a quartz reactor inside a tubular resistance furnace controlled by temperature programmer. The amount of the sample was about 4 g/batch. Pure hydrogen was passed through the sample at a flow rate of 200 cm3/min. The temperature was
977 increased at a linear rate of 1 ~ to the final temperature, which was held for 1 h. The samples were quenched to room temperature at flowing argon, then passivated by 1% O2/Na mixture. X-ray diffraction analysis of the samples was carried out using a Rigaku D/Max-RB diffractometer with Cu K ct monochromatized radiation source, operated at 40 KV and 100 mA. Temperature-programmed reduction (TPR) of the sample was carried out in a stream of 95% argon and 5% hydrogen with a flowing rate of 30 cm3/min. The catalyst bed was heated linearly at 20 ~ from room temperature to 950 ~ Mass spectroscopy was used as detector. Transmission electron microscopy (TEM) studies were carried out on a JEOL 2000 electron microscope. High-resolution transmission electron microscopy (HRTEM) and electron diffraction were performed on a JEM-4000EX electron microscope with an acceleration voltage of 200KV. Nitrogen adsorption and desorption isotherms at 77K were measured using Micromeritics 2010. Surface areas were calculated from the linear part of the Brunaure-Emmett-Teller (BET) plot. 3. RESULTS AND DISCUSSION Surface area and porosity of the ultrahigh Surface area carbon and the samples after carbothermal hydrogen reduction are complied in Table 1. After impregnation of the carbon material, its surface area decreases from 3234 mZ/g to 2446 mZ/g and its porosity from 1.78 m3/g to 1.30 m3/g. It can be assumed that molybdenum precursor fills and blocks a fraction of the pores. The surface area and porosity of the samples shows a decrease, to
m
0
44 m , o
......
20
30
40
HSAC
~ .....
50 2-theta
60
70
80
(~
Fig. 1. XRD patterns of HSAC, Mo/HSAC precursor and the samples with carbothermal hydrogen reduction
978 some extent, with the increase of the temperatures of carbothermal hydrogen reduction. The changes of the surface area and porosity may be attributed to the reaction of carbon materials with hydrogen and molybdenum precursors, and the changes of molybdenum species. The phases and dispersion of molybdenum compound were also measured by XRD after the supported samples were reduced in hydrogen. Fig. 1 shows XRD patterns of the samples with carbothermal hydrogen reduction at different temperatures, together with ultrahigh surface area carbon and the supported Mo precursors. XRD pattern of the ultrahigh surface area carbon shows clearly that the carbon material is non-graphitizable. XRD pattern of the supported molybdenum precursor sample does not show any diffraction peaks of molybdenum precursors, indicating that molybdenum species is well dispersed on the carbon material and the particle size is smaller than 4 nm. The result is consistent with that obtained using high-resolution TEM by Mordenti et al. [21].
Fig. 2 TEMs of samples with carbothermal hydrogen reduction: a obtained at 700 ~ b obtained at 800 ~ XRD pattern of the sample with carbothermal hydrogen reduction at 600 ~ does not show any diffraction peak due to the MoO2 phase, indicating highly dispersed MOO2. But this result is different from that of molybdenum precursor supported on activated carbon [21], where MoO2 was detected by XRD. The good dispersion of MoO3 precursor and MoO2 may be due to the ultrahigh surface area and pore volume of the carbon material used in..this study. XRD pattern of the sample with carbothermal hydrogen reduction at 700 ~ shows a diffraction peak at 39.4 o, which is due to [3-Mo2C with hexagonal close-packed structure. With further increase of the carbothermal hydrogen reduction temperature up to 800 ~ the typical diffraction peaks due to fS-Mo2C clearly show up at
979 39.4, 37.8, 34.3, 61.7, 52.0, 69.6, 74.5, and 75.7 o 20. The above XRD results also imply that the particle sizes of 13-Mo2C increase with the increase of the temperature of carbothermal hydrogen reduction. In order to estimate the particle sizes of 13-Mo2C, TEM and high-resolution TEM analyses were carried out. The TEMs of the samples with carbothermal hydrogen reduction at 700 and 800 ~ are shown in Fig. 2. It can be seen that the particle sizes of 13-Mo2C at 700 ~ are remarkably uniform, about 10 nm in diameter, dispersed on the outer surface of the carbon materials (Fig. 2a). The particle sizes of 13-Mo2C at 800 ~ are non-uniform, with about 25 nm diameter (Fig. 2b), thus larger than at 700 ~
Fig. 3 HRTEM of sample with carbothermal hydrogen reduction at 700 ~ HRTEM picture of Fig. 3 shows the molybdenum carbide containing a high density of planar defects and dislocations after synthesis and passivation at room temperature. The EDS analysis performed on this phase indicated the presence of Mo, C, and O. The existence of O is due to formation of passivated layer, which avoids a violent oxidation of the sample when it is exposed to air. To understand the carbothermal hydrogen reduction process, the reactant and principal products were monitored with mass spectroscopy in real time. The sample was first dried at 150 ~ in a stream of nitrogen for 2 h, then cooled to room temperature in order to avoid the desorption peak of water at about 100 ~ Fig. 4 shows the synthesis traces of masses (M) 2, 16, 18, 28 and 44 with the increase of temperature. The signals at M=2 and 16
980
represent hydrogen and methane, while the signals at M=18, 28 and 44 represent water, carbon monoxide, and carbon dioxide, respectively. The TPR-MS traces show that the carbothermal hydrogen reduction reaction proceeds in three stages in the range of temperatures studied. First, there was a considerable amount of hydrogen consumption and water formation at about 400 ~ From it, one can conclude that MoO3 was reduced into Mo oxide with low chemical valence. But the actual Ha consumption is higher than needed for the following reaction. MoO3 + Ha 50%. They were characterized by various techniques and their performance in deep HDS of 4-ethyl, 6-methyl-dibenzothiophene measured at 633 K and 6.0 MPa in the presence of carbazole and HaS. The Pd-Pt catalysts showed good activity as compared to typical conventional HDS catalysts. Their activity decreased with a decreasing A1 content in the support (thus support acidity): ASA > MPAS > MPS. Enhanced support acidity seems to favor the stabilization of the dispersed metal particles against sintering. 1. I N T R O D U C T I O N Organic sulphur- and nitrogen-compounds in motor fuels are a source for acid rain and harmful to the environment. Moreover, they are poisonous to the auto exhaust catalysts. To meet new developments in EU regulations on the S-concentration, a commonly applied one-step hydrodesulfurization (HDS), using conventional catalysts, e.g. Co-Mo/7-A1203, is insufficient. A second HDS step, viz. a deep HDS step, can be more economical to reduce the S-content to the currently allowed European level of 350 ppm. This level will be reduced further to 50 ppm in 2005 [1]. In the first HDS step, often the heavy organic sulfur-containing polyaromatics survived, such as dibenzothiophene (DBT) and (4-, and/or 6-) alkylated DBTs [2,3]. They are the most refractory. In crude oils, there are also aromatic N-compounds, which suppress the performance of the HDS catalysts. Hence, a model feed for representative HDS-activity measurements should contain characteristic S- and N- compounds for practical relevance. Previous studies by Reinhoudt et al. [4-7] using a model feed, showed that PdPt/ASA catalysts are suitable for deep HDS. It was reported that at the same total loading of Pd and Pt, the Pd/Pt atomic ratio also influences the performance [8]. Pd-Pt
1020 catalysts for deep HDS on three carriers, viz. ASA, MPS and MPSA, were studied due to their interesting textural properties and different Al-contents, hence acidic properties. 2. E X P E R I M E N T A L
Three supports, viz. a novel mesoporous silica (MPS), a mesoporous silica-alumina (MPSA, Si:Al ratio of 6) [9-11] and a commercial amorphous silica-alumina (ASA, Grace Davison HA) were used to prepare Pd-Pt catalysts with a Pd/Pt atomic ratio of 4 [8]. The MPS and MPSA carriers were prepared according to [9-11]. The Pd-Pt catalysts were prepared by ion-exchange [12,13] of 3.5 g carrier using one liter of aqueous solution containing 0.001 M Pd(NH3)4C12 and 0.00025 M of Pt(NH3)4CI2 (both from Alfa Aesar) under vigorous stirring. After one hour the pH was adjusted to 9.5 by 12.5 N NH4OH. Then, it was filtered and dried at 353 K overnight. Similar procedures were used to prepare Pd or Pt on MPS catalysts. This procedure is a modification of the method described in [12,13], which makes the preparation more simple and costefficient. The catalysts and their precursors were characterized using various techniques, viz. TGA (TGA/SDTA 851 e from Mettler TOLEDO), CO chemisorpiton, nitrogen physisorption (Quantachrome Autosorb-6B), Temperature Programmed Reduction (TPR, at 10 K/min in a flow of 7.6% H2 in Ar), elemental analysis (ICP-OES) and High Resolution Transmission ElectroMicroscopy (HRTEM), to optimise the procedures of preparation, calcination and pretreatment. For chemisorption, the sample was reduced in 100 % Hz (50 ml/min) at 573 K for 3h, followed by evacuation at 573 K for 2 h. Then, the temperature was lowered to 308 K and CO chemisorption started. The amount of CO chemisorption was calculated from the difference in CO uptake between two successive measurements (with an interim evacuation). Transmission Electron Microscopy (TEM) was performed using a Philips CM30T electron microscope with a LaB6 filament as the source of electrons operated at 300 kV. Samples were mounted on microgrid carbon polymer supported on a copper grid by placing a few droplets of a suspension of ground sample in ethanol on the grid, followed by drying at ambient conditions. The catalytic performance was determined in a micro batch reactor at 633 K and 6.0 MPa, using 200 mg catalyst (particle size 100 - 200 ~tm) and a model diesel fuel containing 150 mg 4-ethyl, 6-methyl-dibenzothiophene and 27.5 mg carbazole in 100 g n-hexadecane with n-octadecane as internal standard [5]. The catalyst was reduced in H2 flow (50 cm3/min) at 0.1 MPa at a heating rate of 2 K/min up to 573 K and kept at the temperature for 2 h. After reduction and cooling to about 300 K, the catalyst was introduced into the batch autoclave reactor without any air contact. The reaction mixture and the catalyst were heated under hydrogen to 633 K and pressurized to the desired level. The reaction was started by stirring at 2500 rpm using a capillary swing stirrer [14]. The samples were analysed by a gas chromatograph (Chrompack CP 9001) with a flame ionisation detector and a fused silica capillary column (CPSIL-8, 60 m, 0.22 mm internal diameter and a 0.25 ~m dimethylsiloxane film thickness). Samples for GC analysis were taken at intervals of ca. 30 min.
1021 3. R E S U L T S AND DISCUSSION CO chemisorption at 308 K was used to measure the metal dispersion and a stoichiometry ratio of M/CO (M = Pd or Pt) = 1:1 was assumed in the calculation for both Pt and/or Pd. Due to the possible formation of Pd hydride and the uncertainty on the metal to hydrogen ratio, hydrogen chemisorption was not used. Our previous results on Pd/MPS catalysts showed that the hydrogen concentration during reduction has little influence on the dispersion obtained. The metal dispersions of some Pd and/or Pt catalysts are presented in Table 1. Table 1 Pd and/or Pt dispersion of some reduced catalysts Pd wt% Pt wt% % Dc Support Preparation 2.75 0 0.7 MPS impregnated 2.75 0 6.4 MPS impregnated 2.75 0 43 MPS ion-exchanged 2.75 0 54 d MPS ion-exchanged 4.6 0 53 d MPS ion-exchanged 0 2.5 53 d MPS ion-exchanged 0 2.5 33 MPS ion-exchanged 2.56 1.16 56 MPS ion-exchanged 2.56 1.16 51 MPS ion-exchanged 2.5 1.1 51 ASA ion-exchanged 1.76 0.76 54 MPSA ion-exchanged a. At 623 K in air for 2 h. b. At 573 K in air for 2 h. c. Measured method, d. particle size of 1-2 nm observed in HREM.
Calcination Tred/K uncalcined 453 calcined a 453 calcined a 573 calcined a 453 calcined a 453 uncalcined a 548 uncalcined 673 uncalcined 573 calcined b 573 calcined b 573 calcined b 573 by CO chemisorption
The dispersion of properly reduced Pd/MPS catalysts, prepared by the ion-exchange method, is substantially higher than the corresponding ones, prepared by impregnation. It follows that impregnation is not a good method for preparing Pd on MPS catalysts. Hence, Pt and Pd-Pt catalysts were prepared by ion-exchange method only. Table 1 showed that calcination prior to reduction is favorable for Pd/MPS catalysts, but not for the Pd-Pt catalysts. The results of Pd-Pt/MPS showed that air calcination at 573 K for 2 h prior to the reduction reduces the dispersion from 56% to 51%, so a prior calcination is not necessary for the Pd-Pt catalysts as for the Pt catalysts. High dispersions (51-53%) are obtained for the properly reduced Pd-Pt catalysts on all the three carriers, prepared by the ion-exchange method, similar to the corresponding Pd/MPS catalysts, which are higher than the Pt or Pd catalysts prepared by ionexchange, described in [12,13]. These dispersions are also higher than those reported by Navarro et al. [7], though their Pd-Pt/ASA catalyst has a lower total metal content (Pd: 0.27wt% and Pt: 0.94 wt%). Note that their catalyst was prepared by co-impregnation. This confirms that ion exchange is a better method to prepare the Pd-Pt catalyst than impregnation. Moreover, it appears that a too high reduction temperature is unnecessary. For example, 573 K is proper for the Pd-Pt catalysts and 453 K is enough
1022 to reduce the Pd/MPS catalysts. The lower dispersion of the 673 K-reduced Pt/MPS catalyst by ion-exchange as compared to a 548 K-reduced one (33% versus 53%), is thought to be caused by a too high reduction temperature, which is above the H/ittig temperature of Pt (THtittigPt ~ 608 K), leading to possible sintering of the reduced Pt particles. TEM/HRTEM measurements of some samples were performed and Fig. 1 shows some HTREM images of Pd-Pt catalysts
Fig. 1. High resolution transmission microscopy image of Pd-Pt/MPS catalysts a: after chemisorption and recalcination (left) and b: spent (right). The metal particle sizes observed for three MPS supported Pd or Pt samples marked with a in Table 1 are between 1-2 nm. As CO chemisorption data showed that the Pd-Pt catalysts have similar dispersion, it may be expected that these have similar metal particle sizes. A calculation of Pd and Pt particle sizes at 100% dispersion using a spherical model yield 1.12 and 1.13 nm, respectively. It follows that the dispersion measured by CO chemisorption may represent a lower limit, the actual dispersion could be higher than those presented in Table 1. For Pd on silica or alumina, Moss [15] used a CO:Pd stoichiometric ratio of 0.6 and a value of 0.82 was used by Navarro for Pd-Pt on ASA [7]. Were these ratios used in calculating the chemisorption data, it would indeed lead to a higher dispersion. HRTEM results show that no metal-containing particles have been found for uncalcined samples, the only material visible in the HRTEM images in uncalcined PdPt/MPSA is the MPSA. Pd-Pt samples on MPS (after reduction, CO chemisorption and recalcination), MPSA (calcined) and ASA (calcined) were measured. The former contain mainly larger metal oxide particles of about 5-15 nm in diameter, Pd-Pt/MPSA showed only small particles of 1-5 nm, whereas Pd-Pt/MPS showed many small particles (1-5 nm) next to a few large clusters of large particles. HRTEM photos of a spent Pd-Pt/MPS sample showed Pd-Pt particles ranging from a few nm up to ca. 200 nm, indicating that at least part of the noble metal particles agglomerated in the HDS reaction [7] and the fast sintering of this active phase might be the cause for the quick deactivation. Obviously, the lower activity of Pd-Pt/MPSA as compared to that on ASA is not caused by a lower dispersion.
1023 Table 2 Textural properties* Sample dpmax(nm) SBET (mZ/g) Vp (cm3/g) Vmicro (cm3/g) Al wt% *: reduced catalyst.
and AI content of Pd-Pt catalysts Pd-Pt/MPS Pd-Pt/ASA 13.3 5.8 270 440 0.97 0.7 0.024 0 0 22.7
Pd-Pt/MPSA 2.4 520 0.31 0 5.63
The textural data in Table 2 shows that the reduced catalysts based on all the three carriers have mainly mesopores. The BET surface area of the reduced Pd-Pt catalysts decline in the following order of the carriers: MPSA > ASA > MPS, whereas the maximum pore diameter in pore distribution and pore volume follow the opposite sequence. Note that the dispersion of the reduced Pd-Pt catalysts has the following order of the carrier: MPSA > MPS - ASA. Fig. 2 shows the pore distribution of the three catalysts.
~, 9
30
a
25 0
E
20
8 7"6
15
4
).0
0.2
0.4
p/pO
0.6
0.8
1.0
Fig. 2. Adsorption isotherms of the reduced Pd-Pt catalysts on: a. MPSA, b. ASA and c. MPS. Here, for sample a, the peak at p/p0 value of 0.5, corresponding to 3.5 nm is caused by tensile stress effect, thus an artifact. It is clear that they are all dominated by mesopores. Table 3 gives the catalytic HDS performance of the catalysts, expressed as a first order reaction rate constant in h-lgcat-1. Those of some typical relevant HDS catalysts are also presented for comparison. For the Pd-Pt catalysts (at similar Pd/Pt ratio), it decreases in the followin.g order of the support" ASA >> MPSA > MPS. This is also the decreasing order of A1 content in the catalyst. Note that the total content (wt%) of Pd and Pt for Pd-Pt/MPSA is less than those on MPS or ASA, although the same preparation procedure was used. This may be caused by a lower amount of surface hydroxyl groups on the surface of MPSA, available for ion-exchange.
1024 Table 3 First order reaction rate constants (hqg-lcat) in deep HDS and the metal dispersions Catalyst Base Case With carbazole % Dispersion Co-Mo/AlzO3 (commercial) 0.34 0.14 0.38 0.24 Ni-Mo/AlzO3 0.72 0.24 Ni-W/AIzO3 Pd-Pt/ASA [4] 2.82 1.68 46 a Pd/Pt=2, 2 wt% Pd+Pt 3.44 51 c Pd-Pt/ASA (3.5wt% Pd+Pt) b 0.77 (3.9 d) 51 c Pd-Pt/MPS (3.5wt% Pd+Pt) b 0.92 53.6 c Pd-Pt/MPSA (Si:AI=6, 2.5 wt% Pd+Pt) b a. by Hz chemisorption, b: Atomic ratio of Pd/Pt = 4:1. c: by CO chemisorption. d: initial activity. The first order reaction rate plot of Pd-Pt catalysts on ASA and MPS are given in Fig. 3.
~3 o
,- 2
n
e--
m
0
0
~
200 Reaction time / min
400
0
0
~
200 Reaction time / min
400
Fig. 3. First order reaction rate plot of various catalysts, a. Pd-Pt/ASA and b. Pd-Pt/MPS (Table 3). For Pd-Pt catalysts on ASA and MPSA (not shown here), linear regression gives a straight line passing through the origin. For Pd-Pt catalyst on MPS, at the beginning, the rate was very high, later on it slowed down and followed a line, which did not pass through the origin. This indicates a fast deactivation at the beginning of the reaction, in accordance with a fast growth of the metal particles in the spent catalyst, as observed by HRTEM. Table 3 shows that the Pd-Pt catalysts are more active than the commercial catalysts, e.g. Co-Mo/7-AlzO3, using the model feed. Although the Pd and Pt loading in our PdPt/ASA catalyst is 75% higher than that used in [5] (3.5 versus 2.0 wt%), its rate constant is about twice as high. This may be partially attributed to its higher dispersion and/or to its different Pd:Pt ratio (4 instead of 2) [8].
1025 Although all three Pd-Pt catalysts are mesoporous and have similar initial dispersions, their performance differs enormously. The activity decreases in the decreasing order of their A1 content, viz. ASA > MPSA > MPS. The replacement of Si by AI in the silica skeleton and its subsequent charge unbalance are often used to explain the acidity in silica-alumina. It is expected that acidity will decrease in the same order, i.e. ASA > MPSA > MPS. This shows the impact of acidity on noble metal catalyzed deep HDS [5,6,8]. The texture is of secondary importance to the performance. For the Pd-Pt/MPS catalyst, a high initial activity was observed, which declined fast (Table 3). Apparently, the active sites on MPS are less stable as those on MPSA or ASA, they sinter fast according to our HRTEM observation. Although this catalyst has a higher pore volume, larger pore diameters and a higher Pd or Pt content than those of Pd-Pt/MPSA, its activity is lower. Calculations show that the adsorption of the substrate on the active sites cannot explain this initial high activity. This catalyst does not contain any AI. This implies that strong acidic sites may not necessarily be essential for the active sites for deep HDS. It follows that the presence of AI (thus strong surface acidity of the carrier) contributes to the stability of the noble metal active sites in deep HDS. Besides the stabilization of small Pd-Pt particles, it may also affect the metal-sulfur bond strength and the nature of the active sites, via creating electron-deficiency, leading to changes in the reaction kinetics [5], but in view of the high initial activity of the MPS sample this seems less probable. ACKNOWLEDGEMENTS We wish to thank Dr. P.J. Kooyman of the National Center for HREM, Delft University of Technology for HRTEM measurements and helpful discussions, Ing. J. Groen of DCT/TNW for nitrogen physisorption and CO chemisorption measurements, and ABB Lummus for financial support. REFERENCES 1. European Standard 590. 2. P. Waller, Reaktivit/it organischer Schwefelverbindungen beim Hydrotreating von Gas61en unterschiedlicher Herkunft, Ph.D. Thesis, 1997, University Karlsruhe (TH), Germany. 3. H. Schulz, W. Bohringer, P. Waller and F. Ousmanov, Catal. Today, 49 (1999) 87. 4. H.R. Reinhoudt, R. Troost, A.D. van Langeveld, S.T. Sie, J.A.R. van Veen and J.A. Moulijn, Fuel Proc. Tech., 61 (1999) 133. 5. H.R. Reinhoudt, The development of novel catalysts for deep hydrodesulfurisation of diesel fuel, Ph.D. Thesis, 1999, Delft University of Technology, The Netherlands. 6. H.R. Reinhoudt, R. Troost, S. van Schalkwijk, A.D. van Langeveld, S.T. Sie, J.A.R. van Veen and J.A. Moulijn, Fuel Proc. Tech., 61 (1999) 117. 7. R.M. Navarro, B. Pawelec, J.M. Trejo, R. Mariscal and J.L.G. Fierro, J. Catal., 189 (2000) 184. 8. H. Yasuda and Y. Yoshimura, Catal. Lett., 46 (1997) 43. 9. Z. Shan, Th. Maschmeyer and J.C. Jansen, WO 00/15551 (2000).
1026 10. Z. Shan, Th. Maschmeyer and J.C. Jansen, US Patent Appl. no. 09/390276 (2000). 11. J.C. Jansen, Z. Shan, W. Zhou, L. Marchese, N. van de Puil and Th. Maschmeyer, Chem. Commun., (2001) 713. 12. P.C. Aben, J. Catal., 10 (1968) 224. 13. H.A. Benesi, R.M. Curtis and H.P. Studer, J. Catal., 10 (1968) 328. 14. S. Tajik, P.J. van der Berg and J.A. Moulijn, Meas. Sci. Technol., 1 (1990) 815. 15. R.L. Moss, Catalysis (Specialist Periodic Report), The Chemical Society, London, 4 (1981) 31.
Studies in Surface Science and Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
1027
Preparation of highly ordered CMI-1 and wormhole-like DWM mesoporous silica catalyst supports using C16(EO)10 as surfactant Alexandre L6onard #, Jean-Luc Blin and Bao-Lian Su* Laboratoire de Chimie des Mat6riaux Inorganiques, ISIS, The University of Namur, 61, rue de Bruxelles, B-5000 Namur, Belgium phone" +32-81-72-45-31, Fax" +32-81-72-54-14, e-mail"
[email protected] Concentrations of polyoxyethylene alkyl ether surfactants located in the hexagonal H1 micellar domain typically afford disordered mesoporous structures of wormhole-like channels after polymerization of the inorganic precursor in solution. The effect of hydrothermal treatment duration and temperature on the characteristics of the obtained disordered DWM (Disordered Wormhole-like Mesostructure) compounds has been investigated. Thermal stability of these samples has also been studied and appears to be superior to that of MCM-41 type materials. If the weight percentage of surfactant in solution is however decreased, well-ordered hexagonal CMI-1 can be directly obtained. In this case, a co-operative mechanism leading to spherical particles of organized channel structures has been proposed. Influence of the surfactant concentration on the arrangement has been discussed in detail. 1. INTRODUCTION Since their appearance in 1992, ordered mesoporous silicas have been widely used, due to their high surface area and uniform pores sizes, as catalyst supports, to graft a series of catalysts or to disperse metallic or semi-conductor clusters [1-5]. However, the pore size control as well as the reduction of the synthesis costs remain still great challenges. The latter can be reduced significantly if easy-to-use, non-toxic and biodegradable polyoxyethylene alkyl ethers are employed as templating agents. Besides, the micelles have bigger sizes than the "conventional" ionic templates and the recovery as well as the re-use of the surfactant are made possible because of the H-bonding type interactions that exist between surfactant and inorganic species [6-8]. The materials obtained however possess a disordered structure of wormhole-like channels (MSU or DWM-type compounds) for concentrated micellar solutions unless working in strong acidic media like reported by Stucky et al. [9] or by removing the methanol released from tetramethoxysilane hydrolysis from the synthesis medium [10]. The present paper reports a systematic study of the influence of hydrothermal treatment on the final features of such disordered materials obtained with concentrated micellar solutions. The thermal stability of these compounds has also been evaluated. Moreover, Mobil researchers showed that highly ordered mesoporous could be obtained with CTMABr concentrations well below CMC-2 (Critical Micellar Concentration), composition at which micelles adopt a hexagonal ordering in solution. In that way, we studied the effect of decreasing the non-ionic C16(EO)10 surfactant weight percentage. This will lead directly to well-ordered CMI# :FRIA fellow
1028 lmaterials with a hexagonal empilement of their channels. These structures can be advantageous if they are to be used as encapsulating hosts for nanowires or as nanoreactors for the production of low-branched polyethylene fibres, for instance [4-5]. On the other hand, the 3-dimensional access to the reaction centres of the disordered MSU or DWM frameworks seems to be more suitable for catalytic purposes, so the study on the influence of the treatment on the final texture of such compounds is of considerable interest if the pore diameters are to be well-controlled for size seleetivities for example. Different characterization techniques (SEM, TEM, XRD and nitrogen adsorption-desorption analysis) have been used to shed light on the morphological, structural, and textural features of the prepared DWM and CMI-1 compounds. 2. EXPERIMENTAL 2.1. Synthesis The first series of micellar solutions leading to disordered DWM compounds were prepared with 50 wt.% C16(EO)10 (Brij 56~) by dissolving the surfactant at 70~ in an aqueous solution. The pH of the micellar solution was then adjusted to 2.0 with H2SO4. After homogenization at 70~ during 3 hours, the silica source, tetramethoxysilane (TMOS), was added dropwise, the surfactant / silica molar ratio being equal to 1.5. After further stirring for 1 hour, the synthesis gel was poured into teflon cartridges sealed in stainless steel autoclaves. Hydrothermal treatment was then performed during 0 to 15 days at 60, 80 and 100~ Surfactant was removed from the inorganic framework by ethanol extraction with a soxhlet apparatus and calcination at 550~ under nitrogen and oxygen. The same procedure was applied for the preparation of ordered CMI-1 materials. In this case, less C16(EO)~0 was dissolved in the same aqueous volume, still for 3 hours. The surfactant / silica molar ratio was still equal to 1.5 and hydrothermal treatment was performed for 1 day at 80~ 2.2. Characterization To investigate the structure of the samples, XRD at low angles was performed on a Siemens D-5000 diffractometer and transmission electron microscopy on a Philips Techna~" 100 kV microscope. The powdery samples were embedded in an epoxy resin before being sectioned with an ultramicrotome and deposited on carbon coated copper grids. The textural properties of our compounds were assessed by nitrogen adsorption-desorption measurements. Analysis took place over a wide range of relative pressures on a Micromeritics ASAP 2010 or Tristar 3000. The pore diameters and the pore size distributions were determined by the BJH method. Morphological features have been investigated with the use of a Philips XL-20 scanning electron microscope. For conductivity purposes and in order to enhance the yield in secondary electrons, powders were first covered by a thin layer of gold.
3. RESULTS AND DISCUSSION 3.1. Influence of hydrothermal treatment on the features of disordered DWM materials Fig. 1 reports the XRD patterns of materials obtained at 60~ for different treatment durations. At this temperature, only one peak located at 13.0, 11.6 and 16.4 nm for heating times of 2 (Fig.lb), 3 (Fig.l c) and 11 days (Fig.ld) respectively, is detected. The presence of only one reflection line indicates that the compounds belong to the MSU or DWM family. The structure of these molecular sieves consists of regularly sized wormhole-like channels. The single broad peak arises from the average pore-to-pore separation in the disordered
1029 wormhole framework, which presents a lack of long-range crystallographic order. However, according to the phase diagram, a hexagonal H1 phase of micelles is present in aqueous solution at such a surfactant concentration (50 wt.%) before adding the silica source [11] and thus, mesoporous possessing this structure could be expected. However, the results obtained by ~-~ = ~~-- 11.6mn XRD and TEM are contrary to our expectations. We suggest that the interaction between the hydrophilic head of decaoxyethylene cetyl ether and TMOS as well as the methanol released during hydrolysis of the latter could disturb the hexagonal micellar array preformed in solution, thus leading to a disordered framework. The TEM picture of one of our typical s m i l e s reported in Fig. 2 confirms the wormhole-like structure of our materials and its adherence to MSU (or DWM) family. If the heating temperature is increased to 80 or 100~ (diffractograms are not reported), no peak is detected any more after 2 days of hydrothermal treatment. This results from the fact that, as will be discussed further, the pore size distributions are very likely to broaden and to become more or less inhomogeneous for the 2 4 8 8 io 12 higher temperatures. The regular size of the channels could then 2 (o) be partly lost. It should be noted that for the sample which has Fig. 1 : XRD patterns of not been subjected to a hydrothermal treatment, one reflection compounds synthesized at located at 7.5 nm (Fig. l a) is present on the XRD pattern. So, 600C for a :0, b :2, c : 3, when tetramethoxysilane is added to the prepared micellar d: 11 days. solution and pH is adjusted, the condensation and polymerization of the silica directly take place around the micelles of surfactant. This is suggestive of the formation of mesoporous materials even before the hydrothermal treatment. The latter will then affect the textural properties and in particular the pore diameter of the final compounds. The textural characteristics of the samples as a function of heating temperature and duration are listed in Table 1. From these data, it appears that the increase of heating temperature favours the formation of large pore mesoporous molecular sieves. Too long heating at higher temperature, 100~ for Fig. 2 9TEM micrograph example, will however lead to the destruction of the of a typical DWM sample. compounds. Their X-ray diffraction pattern will not show any peak any more and the pore size distributions broaden a lot. Moreover, specific surface area decreases indicating a partial collapse (surface remains at about 400 m2/g) of the mesoporous framework. For the shorter durations of treatment, homogeneous mesoporous materials, showing pore sizes from 4.0 nm up to 13.4 nm with very high specific surface areas of around 950 m2/g have been synthesized.
1030 It is observed that the pore diameter strongly depends on the heating time and temperature (Table 1). This is in agreement with Heating Pore size (nm) aider heating at" results reported previously with time (days) C13(EO)6 [12] and Cls(EO)10 [13, 60~ 80~ 100~ 14] surfactants and observed by Pinnavaia et al. [15]. Two main 0 1.7 tendencies can be drawn from Table 0.5 1.7 5.2 7.5 1. First, with increasing 1 1.7 4.7 8.0 temperature, the pore diameter 2 9.4 4.0 13.4 grows as mentioned above. 3 6.7 Secondly, for a given temperature, 4 5.8 12.9 15.8 in particular for 60 and 80 ~ the 5 5.2 pore diameter first sharply increases 6 4.5 9.9 17.1 and then decreases when 8 4.8 12.1 14.6 hydrothermal treatment is 11 7.9 12.1 prolonged. Concerning the first - : no data tendency, it appears that decaoxyethylene cetyl ether can be solubilized in water in different ways. Under acidic conditions, large amounts of hydrogenbonded water molecules are present around the hydrophilic heads of the surfactant and the equilibrium between the contracted and extended conformation of C16(EO)10 is strongly shitted towards the contracted form. A low treatment temperature would not alter this conformation and interaction of the ethoxy oxygens with the silanol groups of the silica would not be favoured. The effective cross-sectional area of the hydrophilic headgroups of the surfactant is important and, according to Kunieda et at [16], micelles with a high curvature and a small diameter are therefore formed. If the heating temperature is raised, due to the thermal motion, the water molecules tied around ethoxy oxygens through hydrogen bonds progressively disappear and the oxyethylene head can extend. The curvature of the micelles will then decrease and this conformation allows for more interactions with silica. This results in an increase in the pore diameter. Finally, if the heating temperature is further raised, the surfactant - silica interface becomes less important and the size of the mesopores increases with the stretching of the surfactant molecules. This growth in length of surfactant chain or size of micelles can induce a breakdown of the wall separating adjacent pores. Larger-sized channels are therefore formed, but porosity and relatively high specific surface area can be maintained. This phenomenon of wall breakdown was previously reported by Sayari et al. [17] to explain the process of pore size enlargement of pure siliceous mesoporous materials during a post-synthesis treatment, using an amine as swelling agent. Pairs or triplets of adjacent pores can transform into single pores in a similar way. However, when thermal treatment is prolonged over a wider period of time, for a given temperature, the pore diameter decreases (Table 1). Indeed, it is observed that between 2 and 6 days of hydrothermal treatment at a heating temperature of 60~ the pore diameter decreases from 9.4 to 4.5 nm instead of increasing as expected. The possibility of reorganization of the micelles should be considered in this case. Some molecules leave the formed micelles and, in order to maximize the hydrogen bonding interactions between the molecules of surfactant and water molecules or silica source, a rearrangement of the micellar Table 1. Variation of the pore diameter determined by the BJH method with synthesis heating temperature and time
1031 system in the gel occurs, which leads to the formation of more micelles, but with a smaller size. For durations inferior to 6 days, the materials result from a compromise between the reorganization of the formed micellar phase to get more stable micelles and the stretching of the surfactant molecules with increasing hydrothermal duration. 3.2. Thermal stability of the disordered mesoporous materials obtained
This study was performed using three samples (a, b and c) synthesized under different hydrothermal treatment conditions (1 day at 100~ 1 day at 80~ and 6 days at 60~ respectively). The pore diameters of these compounds are respectively 8.0, 4.7 and 4.5 nm. The materials were calcined at 550, 600, 700, 800, 900 and 1000~ and the variation of specific surface area was monitored as a function of this calcination temperature. A large pore MCM-41 (sample d) with 7.5 nm pore size is also reported in Fig. 3 for comparison. From Fig. 3, it appears that the specific 1000 c surface area of the disordered DWM materials t:l b ... ~.. remains very high till 800~ Beyond this ~ 800d "~"~-,,.{ temperature, for sample B for instance, its value decreases from 800 to 20 m'/g if calcination temperature is raised from 800 to 1000~ 400 Whatever the hydrothermal treatment conditions of 200 the synthesis are, the thermal resistance of the .=. materials is similar. Even at 1000~ the recovered *~~' 0500 660 760 860 "960 1000"1100 materials exhibit a type IV isotherm, characteristic Calcination temperature (~ of mesoporous compounds. A part of mesoporosity Fig. 3. Specific surface area vs. is thus maintained with quite a narrow pore size calcination temperature for disordered distribution, but the maximum adsorbed volume is DWM structures (a, b and c) and for a MCM-41 type compound (d). sharply reduced. This is in accordance with the very broken appearance of the particles observed by SEM. Thermal stability of these disordered materials is, however, very superior to MCM-41, whose structure does not resist beyond 600~ This behaviour can be related to the different preparation method that affords compounds with a different structure and also thicker walls. 3.3. Obtention of ordered hexagonal CMI-1 materials
Fig. 4 shows the XRD patterns of compounds prepared at surfactant weight percentages of 40 (A) and 10 (B), respectively. For samples synthesized at a low concentration of surfactant, in addition to the sharp 100 reflection line, two weaker peaks, suggestive of a hexagonal organization of the channels can clearly be pointed out. \__ A However, these secondary reflections vanish with increasing 2 4 6 8 10 12 surfactant concentration and finally completely disappear at 20 (o) weight percentages above 30. Beyond this composition, only Fig. 4. XRD patterns of A: one peak, characteristic of the regular repetition of the pore disordered compound and diameter, can be observed, suggesting that the compounds B: highly ordered CMI-1 evolve towards a disordered array of very uniformly sized sample. channels like MSU materials. TEM pictures show a disordered wormhole-like framework at surfactant weight percentages above 30 (Fig. 5A) whereas a very highly ordered hexagonal structure can be evidenced for lower concentrations (Fig. 5B and C). This is also confirmed by the inserted FFT of the micrograph which clearly
1032
Fig. 5 : TEM pictures of: A : a disordered wormhole-like structure, B : a regular hexagonal framework and C : a ~t f'mgerprint >>-likearrangement of channels. exhibits a sixfold symmetry. Interestingly, in addition to this honeycomb-like arrangement, a fingerprint-like channel array (Fig. 5 C), which could be related to the morphology, (see belo w) can be observed. As explained above, the hexagonal micellar mould is only present for the micellar solutions that contain between 30 and 65 wt.% of surfactant. The addition of TMOS probably disturbs this array, leading to a disordered framework. However, in the case of less concentrated micellar solutions, where only individual micellar rods exist, it appears that the addition of the silica source plays the key role in the obtention of very regular hexagonally arranged channels. Such a co-operative mechanism could be analogous to the LCT pathway initially proposed by Mobil researchers [2]. From nitrogen adsorption-desorption measurements (Fig. 6), we can observe that all the samples exhibit a type IV isotherm characteristic of mesoporous compounds. However, there is an evolution of hysteresis loop from type H2 (IUPAC), usually encountered for disordered structures, towards H1 commonly observed for MCM-41 when the weight percentage of template decreases below 30 wt.%. For the lower concentrations, the adsorption branch of the isotherm becomes much steeper and the pore size distributions (PSD) become much narrower (see inserts) and centred on about 4.4 nm. Specific surface area also increases 1600~_
,
j
~2oo~=J\
//
....
~ool
/ /
400
2o
C
0|.
0.0 ;~
0.2
0.4
0.6
0.8
Relative pressure P/Po
1.0
0.0 0.2 O.4 0.6 0.8 Relative pressure P/Po
1o0
0.0
,
,
,
0.2
0.4
0.6
"
,
0.8
'
1.0
Relative pressure P/Po
Fig. 6 : Nitrogen adsorption-desorption isotherms of: A: disordered wormhole-like sample, B: highly organized CMI-1 compound and C: sample prepared at 30 wt.% showing secondary mesoporosity. to values higher than 1000 m2/g. A more peculiar behaviour is noted at 30 wt.% as the sample shows secondary mesoporosity (Fig. 6C). This could arise from the competition between the ordered CMI-1 and wormhole-like DWM structures that appear at this composition.
1033
Fig. 7 9SEM pictures representing" A" sample prepared at a concentration above 30 wt.%, B 9 the spongy particles obtained at 30 wt.% and C: spheres corresponding to the lower concentrations of template. As can be seen from Fig. 7A, the compounds prepared at high surfactant loadings (>30%) exhibit particles of different sizes and shapes. At 30 wt.%, large spongy particles, which are very likely to emerge from a competition between the spheres and the irregular edge-shaped particles can be found. However, as the concentration of template decreases beyond this limit, the morphology remains unchanged and can be described by an assembly of very small spheres (1-2~m) and the section of such a sphere could explain the ~ fingerprints >> observed by TEM (Fig.7C). At this low surfactant concentration, it is also possible to obtain some "exotic" gyroidal, toroidal and rope-like morphologies like already encountered by Ozin et al. [18-20]. Thus, not only the channel structure but also the morphology of MCM-41 can be successfully reproduced with a non-ionic surfactant. All these results led us to postulate two different synthesis mechanisms [21 ]. First, for the higher weight percentages where, according to the phase diagram, a hexagonal liquid crystal phase is present in the micellar solution, it appears that the polymerization of the silica source as well as the methanol released during its hydrolysis disturb the hexagonal array. The interface interactions between the hydrophilic heads of the template and the hydroxyl groups of the silica and the polymerization of the inorganics will provoke a slight gliding of the channels in order to minimize steric and electrostatic energies. In this case, the prepared compounds possess a disordered wormhole-like structure. Irregularly-shaped particles with a high specific surface area are then obtained. On the contrary, when the surfactant concentration is lowered, only isolated micelles exist in solution. The silica can easily polymerize around the single micelle-rods present in solution and a complete condensation would imply that these rod-like supramolecular assemblies join in order to form the regular hexagonal framework. We propose that in this case, a co-operative mechanism analogous to Mobil's LCT pathway affords the regular empilement. Thus, the addition of the silica source favours the formation of highly ordered CMI-1 mesoporous molecular sieves leading to spheres with very high surface areas and very narrow PSD's. This co-operative mechanism is also compatible with the obtention of ropes, toroidal, gyroidal and spherical shapes as the Interaction at the Single rod-micelle PEO-silica interface .
Eo ~-,/~, Eo 9
~~o~::~
~"
"~
o,,
:-~0,,
. . . . Surfactant.TMOS,o , interaction ~.~ #%"0~~
Silica-covered rod-micelles agglomerate owing to a cooperative pathway ,:oE.o~o
~"
TMOS . ~, ~o~ .'f~ Hydrothermal polymerization ~"~[~,~o~' treatment,extraction, Eo~ " to~ . t ~_,~u~o o calcination
Fig. 8 : Cooperative pathway leading to the ordered CMI-1 materials.
1034 change in morphology occurs when the amount of added silica is varied at a constant low weight percentage of template. 4. CONCLUSION This work has shown that hydrothermal treatment strongly affects the texture of disordered wormhole-like mesoporous catalyst supports, and in particular the ability of the hydrophilic oxyethylene head of the non-ionic surfactant to adopt different conformations with temperature. These compounds show a remarkable thermal resistance up to 800~ suitable for a series of catalytic reactions. On the other hand, ordered hexagonal CMI-1 can be directly obtained by lowering the surfactant concentration below the domain of existence of hexagonally ordered micelles in solution. ACKNOWLEDGEMENTS
Alexandre L6onard thanks FNRS (Fonds National de la Recherche Scientifique, Belgium) for a FRIA scholarship. We thank Prof. P.A. Jacobs and Prof. P. Grange for giving access to their XRD diffractometer. REFERENCES
1. C.T. Kresge, M.E. Leonowicz, W.J. Roth, J.C. Vartuli and J.S. Beck, Nature, 359 (1992) 710. 2. J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, C.T.W. Chu, D.H. Olson, E.W. Sheppard, S.B. McCullen, J.B. Higgins, J.L. and Schlender, J. Am. Chem. Soc., 114 (1992) 10834. 3. A. Sayari, Chem. Mater., 8 (1996) 1840. 4. K. Kageyama, J.I. Tamazawa and T. Aida, Science, 285 (1999) 2113. 5. K. Moiler and T. Bein, Chem. Mater., 10 (1998) 2950. 6. G.S. Attard, J.C. Glyde and C.G. G61tner, Nature, 378 (1995) 366. 7. S.A. Bagshaw, E. Prouzet and T.J. Pinnavaia, Science, 269 (1995) 1242. 8. S.A. Bagshaw and T.J. Pinnavaia, Angew. Chem. Int. Ed. Engl, 10 (1996) 1102. 9. D. Zhao, Q. Huo, J. Feng, B.F. Chmelka and G.D. Stucky, J. Am. Chem. Soc., 120 (1998) 6024. 10. N.R.B. Coleman and G.S. Attard, Microporous Mesoporous Mater., 44-45 (2001) 73. 11. D.J. Mitchell, G.J.T. Tiddy, L. Waring, T. Bostock and M.P. McDonald, J. Chem. Soc. Faraday Trans., I, 79 (1983) 975. 12. J.L. Blin, A. Becue, B. Pauwels, G. Van Tendeloo and B.L. Su, Microporous Mesoporous Mater., 44-45 (2001), 41. 13. J.L. Blin, G. Herrier, C. Otjacques and B.L. Su, Stud. Surf. Sci. Catal., 129 (2000) 57. 14. G. Herrier, J.L. Blin and B.L. Su, Langrnuir, 17, 14 (2001) 4422. 15. E. Prouzet and T.J. Pinnavaia, Angew. Chem. Int. Ed. Engl., 36 (1997) 516. 16. H. Kunieda, K. Ozawa and K.L. Huang, J. Phys. Chem., 102 (1998) 831. 17. A. Sayari, M. Kruk, M. Jaroniec and I.L. Moudrakovski, Adv. Mater., 10 (1998) 1376. 18. H. Yang, N. Coombs and G.A. Ozin, Nature, 386 (1997) 692. 19. H. Yang, G. Vovk, N. Coombs, I. Sokolov and G.A. Ozin, J. Mater. Chem., 8 (1998) 743. 20. G.A. Ozin, H. Yang, I. Sokolov and N. Coombs, Adv. Mater., 9 (1997) 662. 21. J.L. Blin, A. L6onard and B.L. Su, Chem. Mater., 13 (2001) 3542.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
1035
Synthesis and characterization of nanostructured mesoporous zirconia catalyst supports using non-ionic surfactants as templating agents J.L. Blin, L. Gigot, A. L6onard # and B.L. Su* Laboratoire de Chimie des Mat6riaux Inorganiques, ISIS, The University of Namur, 61, rue de Bruxelles, B-5000 Namur, Belgium The formation of nanostructured porous zirconia catalyst supports using non-ionic surfactants as templating agents has been studied in order to optimize the synthesis conditions without addition of structure stabilizing agents such as sulfate or phosphate anions. The effect of the quantity of added zirconium source with respect to the surfactant concentration in solution has been studied. We also examined the role played by the surfactant during the synthesis. Further investigations have shown that the pore diameters could be increased towards the mesoporous domain if heating time and temperature are raised. The present work reveals that the control of the balance between the precipitation rate of zirconia and the interaction of zirconium source will be fatal in the formation of nanostructured porous zirconia. 1. I N T R O D U C T I O N Among the non-silica mesoporous oxides, zirconia is of particular interest due to its large field of applications ranging from catalysis to ceramics [1]. However, the surface area of crystalline zirconia is usually rather low, in the order of 20-50 mVg. Thus, to develop particular catalytic properties such as high conversion and selectivity, the synthesis of ordered mesoporous ZrOz with high specific surface area and narrow pore size distribution is of capital importance both from the scientific and industrial application point of view. In previous work [2,3], we have shown that it was possible to obtain mesoporous zirconia using the CTMABr-ZrOClz.8H20 system. However, the packing of the channels was not well-ordered as expected even by varying the weight percentage of template or the surfactant/zirconium molar ratio. On the basis of TEM, SEM XRD and N2 adsorptiondesorption results, a synthesis mechanism has been proposed. It is observed that at low temperature or for short durations at higher temperatures, the obtained materials are first supermicroporous, then a breakdown of the walls separating adjacent pores allows the transformation to mesopores with increasing hydrothermal treatment time. The obtained materials have a uniform pore size and their surface can reach 300 mZ/g. But the channel array is, at least part of samples, wormlike. However, if the hydrothermal treatment is performed at too high temperature or for too long durations, mesoporous compounds are no longer obtained. Indeed, the final phases observed correspond to tetragonal and monoclinic crystalline zirconium oxides with very low specific surface area. However, the major drawback that has been encountered in the synthesis of these materials is the removal of the surfactant. Without addition of phosphate or sulfate anions [4, 5], which # : FRIA Fellow * : Corresponding author
1036 delay the crystallization of the amorphous ZrO2 into the tetragonal or monoclinic form, it is very difficult to remove the template in order to free the pores. This step of the synthesis is facilitated when the template belongs to the family of polyoxyethylene alkyl ether [Cm(EO)n]. In this case, the removal of the surfactant can easily be achieved by solvent extraction, using ethanol for instance, because of the weaker interactions between the entities (H-bonds). The synthesis of mesostructured oxides, that is difficult or impossible by electrostatic assembly can be performed through this N~ ~ process. Moreover, these templating agents are cheaper, more biodegradable and less toxic than their ionic analogues. Thus, based on the experience acquired in the synthesis of mesoporous silicas using non-ionic [Cm(EO)n] templates [6-9], we have established a synthesis protocol of porous zirconia through a neutral C13(EO)6-Zr(OC3H7)4 assembly pathway. The effect of different physico-chemical parameters such as heating time and temperature on the synthesis of pure nanostructured porous zirconia has been investigated, in order to shed some light on the synthesis mechanism and to determine the important physicochemical variables on the preparation. 2. E X P E R I M E N T A L
2.1. Synthesis A 50 wt.% micellar solution of C13(EO)6 was prepared by dissolving the surfactant at room temperature in an aqueous solution during 3 hours. The obtained medium was further stirred for three hours at room temperature before adding drop by drop the inorganic source : zirconium propoxide [Zr(OC3H7)4]. The surfactant / zirconia molar ratio was varied from 0.5 to 10. The obtained gel was sealed in Teflon autoclaves. Hydrothermal treatment was performed during 2 days at 60~ in a first time. The surfactant was removed by ethanol extraction.
2.2. Characterization The XRD patterns were obtained with a Philips PW 170 diffractometer, using CuKct (1.54178 ,A,) radiation, equipped with a thermostatisation unit (TTK-ANTONPAAR, HUBER HS-60). The transmission electron micrographs were taken using a 100 kV Philips Techna'i microscope. For TEM observations, sample powders were embedded in an epoxy resin and then sectioned with an ultramicrotome. The thin films were supported on copper grids previously coated by carbon to improve stability and reduce the accumulation of charges. The morphology of the final phases was studied using a Philips XL-20 Scanning Electron Microscope (SEM) with conventional sample preparation and imaging techniques. Nitrogen adsorption - desorption isotherms were obtained at -196~ over a wide relative pressure range from 0.01 to 0.995 with a volumetric adsorption analyzer ASAP 2010 or TRISTAR 3000 both from Micromeritics. The samples were further degassed under vacuum for several hours before nitrogen adsorption measurements. The pore diameters and their distribution were assessed by the BJH (Barret, Joyner, Halenda) method [10].
1037 3. R E S U L T S AND D I S C U S S I O N 3.1. Role of the surfactant: Effect of the variation of the C 1 3 ( E O ) 6 [ Zr ( O C 3 H 7 ) 4 molar ratio The XRD patterns of some samples C C prepared without template (A) and with different C 1 3 ( E O ) 6 / Z r ( O C 3 H 7 ) 4 molar ratio (B) are depicted respectively in Fig. 1A and B. The concentration of Zr(OC3H7)4 in the absence of template corresponds to the same fictive surfactant / "7. b zirconium molar ratio. Except a broad band, located in the range of 2502 nm for Fe-particles. 3.4 Modified block-copolymers thin films loaded as innovative catalysts. . . . . More recently [7], Fe 3 + , Fe203 and T102 have been lmmoblhzed on low cost polyethylene@ modified copolymers films containing maleic anhydride anchoring groups. The observed rates of degradation of Orange II and halocarbons were only slightly below the rates observed during homogeneous Fenton photo-assisted degradation or with TiO2 suspensions. Polyethylene is known to be the second most inert Dupont polymer after Teflon, a C-F polymer of widespread use. IR spectroscopy suggests that Fe 3+/Fe203 interacts with the negatively charged conjugated carboxylic groups through -COO-Fe 3+. In the case of the copolymer-Fe3+, the bands in the IR spectrum at 1523 and-1557 cm "1 correspond to two types of iron-carboxyl species. The strong attraction between the negative anhydride groups and the Fea+-ion leads to formation of metal to ligand transfer charge bond (MLTC) between the negatively anhydride group and the Fe-ion on the copolymer thin films. This accounts for the lack of leaching of metal ions into the solution. It is also the reason for the attachment of a one layer of complexed species on the thin film surface that will not come out of the polymer surface due to mechanical factors. During the last decade it has been repeatedly invoked that homogenous Fenton processes involve three reactive intermediate species: a) the formation of oxidative radicals like OH ~ HO2~ and other oxidative radicals due to the H202 added in solution b) the possible formation but controversial solution species (Fe(IV) and finally c) Fe-chelates intermediates like oxalates, maleates, pyruvates that decompose to CO2 with accelerate kinetics in Fenton photo-assisted reactions and with slow kinetics in dark reactions. Fig. 3 shows the mechanism of the reaction intermediates in the dark and under for many organic compounds The importance of the [R-COO] as a preferred reaction pathway in the decomposition of pollutants on the surface [polyethylene-COO--Fe3§ is rationalized by the observation that these copolymer thin films should be attacked by OH ~ HO2~ and other oxidative radicals available in solution. But this was not the case for [polyethylene-COO-Fe 3§ even when used over long times (300 hours) and repeated recycling below the polyethylene flowing temperature (80~ Moreover, the observation reported recently [8] that degradation of organic compounds is able to take place during Fenton photo-assisted treatment in the presence of 3000 ppm of Cl-ion
1087
Fig. 3 3.5 T r e a t m e n t of azo-dyes and textile effluents by immobilized Fenton photo-assisted reactions on inorganic silica membranes. Fe-clusters on silica fabrics have been prepared in our laboratory | [9] and used in the degradation of textile waters and waters of the SMDK treating station for toxic industrial waters in K611iken, Switzerland. The pH of the waste waters as received was in the range 7-9. Treatment with HzO2 without pH adjustment in the dark was ineffective in the reactor. Under UV light irradiation the TOC was seen to decrease by less than 5% with the latter treatment. But in the presence of silica/Fe fabric under 254 nm (75 W) light irradiation, the reflecting immersion concentric photo-reactor decreased the initial COD values of the textile waste waters with pH 9 by about 75% in less than 25 minutes. More surprisingly, compared with homogeneous photo-Fenton processes the degradation was the same in terms of COD reduction. But the BOD increase when using the silica/Fe fabric was more significant than in the case of the homogeneous photo-Fenton treatment. Adsorption of toxic organic compounds on from the waste waters was occurring on the Fe-silica fabrics and turned out to be beneficial in the latter case.
1088 ACKNOWLEDGMENT The financial support of KTI/CTI TOP NANO 21 (Bern, Switzerland) under Grant N ~ 5320.1 TNS is appreciated.
REFERENCES
1. J. Fernandez, J. Bandara, A. Lopez, Ph. Buffat and J. Kiwi, Langmuir, 15 (1999) 185. 2. T. Gierke, G. Munn and F. Wilson. J. Polym. Sci. Polym Phys., 19 (1981) 1687. 3. A. Lopez and J. Kiwi, J. Eng. Ind. Chem. Res., 40 (2001) 1852. 4. M. Dhananjeyan, J. Kiwi. P. Albers and O. Enea, Helv. Chim. Acta, 84 (2001) 3433. 5. A. Harmer, E. Farneth and Q. Sun, J. Amer. Chem. Soc., 118 (1996) 7708. 6. J. Fernandez, M. Dhananjeyan, J. Kiwi, Y. Senuma and J. Hilbom, J. Phys. Chem. B, 104 (2000) 5298. 7. M. Dhananjeyan, E. Mielczarski, K. Thampi, Ph. Buffat, M. Bensimon, A. Kulik, J. Mielczarski and J. Kiwi, J. Phys. Chem. B, 105 (2001) 12046. 8. J. Kiwi, A. Lopez and V. Nadtochenko, Environ. Sci. Technol., 34 (2000) 2162. 9. A. Bozzi, T. Yuranova, P. Lais and J. Kiwi, article in preparation 2002.
Studiesin SurfaceScienceand Catalysis 143 E. Gaigneauxet al. (Editors) 9 2002 ElsevierScienceB.V. All rightsreserved.
1089
Preparation of vanadium-based catalysts for selective catalytic reduction of nitrogen oxides using titania supports chemically modified with organosilanes H. Kominami, a M. Itonaga, a A. Shinonaga, a K. Kagawa, b S. Konishi, a and Y. Kera a aDepartment of Applied Chemistry, Faculty of Science and Engineering, Kinki University, Kowakae, Higashiosaka, Osaka 577-8502, Japan bTechnical Research Center, The Kansai Electric Power Co. Inc., Nakoji, Amagasaki, Hyogo 661-0974, Japan A titania (TiO2) sample with a large surface area (300 mZg-1) was chemically modified with 3-aminopropyltrimethoxysilane (APS) under a reflux of toluene. The thermal stability of the modified TiOz (APS-TiO2) and the adsorptivity of metavanadate anion (VO3-) on APS-TiO2 from an aqueous solution were investigated. Modification with APS suppressed crystal growth and transformation of anatase crystallite to rutile upon calcination, and the anatase phase was preserved even after calcination at 1000~ while transformation to rutile in the unmodified TiOz samples was observed at around 800~ Since there was little crystal growth in APS-TiO2, it possessed a large surface area of 205 mZg-1 after calcination at 700~ The amount of VO3- adsorbed on APS-TiO2 was ca. 1.5-times larger than that on unmodified TiO2 due to a strong affinity caused by acid-base interaction between VO3-and the amino group of APS-TiO2. A supported V205 catalyst that was prepared by decomposition of a VO3--adsorbing APS-TiOz sample had a large surface area despite the large V205 loading and exhibited higher activity than that of a catalyst prepared from an unmodified TiO2 sample. 1. I N T R O D U C T I O N Vanadia/titania (VzOs/TiO2) catalysts are widely used as commercial environmental catalyst for the selective catalytic reduction (SCR) of nitrogen oxides (NOx) with ammonia [1-3]. Due to current strict regulations, a high level of NOx removal is now required, and one of the most effective approaches for the removal of a large amount of NOx is to use a TiO2 support that has a large surface area. Generally, a large surface area is required for a catalyst support to disperse a catalyst material effectively and increase the number of active sites in the catalyst. Since a catalyst is usually used at high temperature, a high degree of thermal stability is also important. We have shown that chemical modification of metal oxide supports, such as silica (SiO2) and TiO2, with silane-coupling agents (organosilanes) having amino groups was effective for dispersion of a catalytic material such as phosphododecatungstate and
1090 phosphotetradecavanadate [4-8] and that the thus-obtained catalysts exhibited higher activities than those of catalysts without chemical modification in some reactions such as partial oxidation of methanol to formaldehyde and oxidative dehydrogenation of isobutyric acid to methacrylic acid [5, 7]. We applied this technique to the preparation of VzOs/TiO2 catalysts and here report the effect of organosilane on the thermal stability of a TiO2 support with a large surface area, the adsorption properties of metavanadate anion (VO3-), and the SCR activities of supported V205 catalysts prepared from VO3--adsorbing TiO2 supports. 2. EXPERIMENTAL
2.1. Modification of TiO2 with organosilane TiO2 powder with a large surface area (ST01) was kindly donated from Ishihara Sangyo, Osaka, Japan. Organosilane, 3-aminopropyltrimethoxysilane (H2N-(CH2)3-Si(OCH3)3; APS), was supplied from Shin-Etsu Chemical Co., Ltd., Tokyo, Japan and was used without further purification. Chemical modification of the ST01 TiO2 with APS was carried out according to the procedure previously reported [4-8]. The TiO2 powder was dried in advance at room temperature under reduced pressure for 24 h. Two grams of the TiO2 powder (2 g) was then suspended in a toluene solution (20 cm 3) of APS (22.1 mmol) and heated at ll0~ for 2 h. After completion of the reaction, the powder was filtered, washed with toluene, diethylether and finally methanol, and then allowed to stand in a desiccator for 30 min. Finally, the sample was dried at l l0~ for 1 h. Hereafter, the TiO2 powder modified with APS is designated APS-TiO2. 2.2. Adsorption of VO3" on modified and unmodified TiO2 Ammonium metavanadate (NH4VO3) (Kanto Chemicals, Tokyo, Japan) was dissolved in H20 to give solutions (0.04-2.00 • 10.2 mol din-3). A hundred milligrams of unmodified TiO2 or APS-TiO2 was added to 30 cm 3 of NH4VO3 solution and stirred to make VO3- adsorb at room temperature for 24 h. The samples were then collected by filtration under suction and dried at l l0~ for 1 h. The amount of VO3- adsorbed was determined using a UV-vis spectrometer (UV-2400, Shimadzu, Kyoto, Japan) from the difference in the concentrations of the solution before and after addition of TiO2. 2.3. SCR of NOx with NH3 over a V205[TiO2 catalyst Unmodified TiO2 and APS-TiO2 samples adsorbing VO3- were decomposed at 550~ for 1 h in air to form supported V205 catalysts. Two V205 catalysts were used for deNOx reaction. The catalytic reaction (catalyst weight: 80 mg) was carried out in a fixed bed flow reactor at temperatures ranging from 300 to 400~ (flow rate: 1.2 dm3 min -1) using a model gas containing 200 ppm NO, 240 ppm NH3, 500 ppm SO2, 33 ppm CO, 3% 02 and 12% H20. The concentration of NOx in the outlet gas was determined by using a NOx meter. 2.4. Characterization Powder X-ray diffraction (XRD) (MultiFlex, Rigaku, Tokyo, Japan) was measured using CuKct radiation with a carbon monochromater. Crystallite size (d101) was calculated
1091 from the half-height width of the 101 diffraction peak of anatase using Scherrer's equation. The value of the shape factor, K, was taken to be 0.9. The specific surface area was calculated using the BET single-point method on the basis of nitrogen (Na) uptake measured at 78 K at the relative pressure of 0.3. Before the N2 adsorption, each sample was dried at 403 K for 30 min in a 30% Na-helium flow. Differential thermal analysis (DTA) (TG-8120, Rigaku) was carried out at a rate of 10~ min -1 in air flow. The morphology of the sample was observed under a JEOL JEM-3010 transmission electron microscope (TEM) operated at 300 kV. Chemical analysis of the APS-TiO2 was performed at Galbraith Laboratory Inc., TN USA.
(a)
IU
(b)
1000~C
b
10001:,C 9001~C
9001:,C
800~C
..~.
-..
8001=C 7001~C .....
7001~C
-
~
.
_. - , , -
5501~C
5501:,C 'before calcination ,
20
I
30
I
40 20 /
,
I
50
,
degree
I
60
I
,
70
20
I
30
I
I
t
I
I
I
40 50 60 20 / degree
t
70
Fig. 1. XRD patterns of samples obtained by calcination of (a) unmodified TiO2 and (b) APS-Ti02 at various temperatures. 3. RESULTS AND DISCUSSION 3.1. Effect of modification with APS on thermal stability of TiOz In the DTA curve of APS-TiO2, a large exothermic peak was observed at around 300~ due to combustion of the aminopropyl group of APS-TiO2. Chemical analysis
1092 revealed that the degree of modification in the APS-TiO2 sample, D(APS), was 2.21 mmol-APS.g-TiO2 -1 and that the Si/Ti ratio was 0.177. Since the initial amount of APS in the toluene solution was 22.1 mmol for 2 g of TiO2, the degree of deposition of APS onto TiO2 was calculated to be 20%. The specific surface area of unmodified TiO2 (bare ST01), S(TiO2), was determined by the BET method to be 300 m2g-1, which is in good agreement with the value reported by the supplier. From D(APS), S(TiO2) and Avogadro's number, the number of APS per nm 2 of TiO2 was calculated to be 4.4, which is almost equal to the numbers (4.5-4.9) of surface hydroxyl groups per nm 2 of anatase TiO2 [9]. The surface coverage of APS on TiO2, 0 (APS), can be estimated by the following formula: 0 (APS)
= D(APS)
9 or (APS) / S(TiO2),
where cr(APS) denotes the cross-sectional area (m 2 mo1-1) of APS. By applying the values for D and S given above and cr = 7.8 • 104 m 2 mo1-1 by estimation from the value given by the supplier (minimum covered area, 436 m 2 g-l), 0(APS) of the present APS-TiO2 sample was estimated to be 0.58.
60
300 (a)
E
~,~ 2OO
4O
"~ .
(b) 9
9
I-U.l rn
m
20
0
I
~
I
,
i
~
!
,
I
200 400 600 800 1000 Tcal /13C
100
0
~
w
200 400 600 800 1000 TcaI/13C
Fig. 2. Changes in (a) crystallite size and (b) BET surface area of TiO2 samples upon calcination at various temperatures. Circles and squares show data of unmodified TiO2 and APS-TiO2, respectively.
The XRD patterns of samples obtained by calcination of unmodified TiO2 and APS-TiO2 at various temperatures are shown in Fig. 1. Transformation of the anatase phase in the unmodified TiO2 sample to the rutile form occurred at 800~ and was almost completed at 900~ (Fig. 1(a)), whereas the anatase crystallite in the APS-TiO2 sample was stable even after calcination at 1000~ (Fig. l(b)). Changes in dl01 and BET surface area of the unmodified TiO2 and APS-TiO2 samples upon calcination at various temperatures are shown in Fig. 2. Unmodified TiO2 exhibited a small dl01 of 8 nm before calcination and, as
1093 mentioned above, possessed a large surface area of 300 mZg-1. However, calcination of the TiOz sample at 550~ drastically increased d101 to 17 nm and decreased the surface area to 91 mZg-1, indicating low thermal stability of the original TiO2 powder. Modification of the TiOz powder with APS inhibited the growth of the anatase crystallite as well as its transformation to the rutile crystallite, as shown in Fig. 2. For example, the sample obtained by calcination at 700~ showed the almost same dl01 as that before calcination and retained a large surface area (>200 mZg-1) due to the inhibition of crystal growth. As can be seen in Fig. 3, TEM observation clearly showed a difference between the behaviors of crystal growth of unmodified and modified TiO2 upon calcination. It had been reported that silica-modified TiO2 samples synthesized from a mixture of titanium alkoxide and silicate ester by glycothermal reaction and thermal decomposition in toluene showed a high degree of thermal stability [10, 11]. Most of the silicon (silica) was incorporated into the anatase structure of both products and suppressed crystal growth and transformation of the TiOz phase. On the other hand, in the present APS-TiOz, silicon originating from APS was distributed only on the surfaces of TiOz crystals (particles). It should be noted that surface modification of TiO2 with APS results in significant improvement of the thermal stability of the original TiOz if the original TiOz has a large surface area.
Fig. 3. TEM photographs of samples after calcination of (a) unmodified TiO2 and (b) APS-TiO2 at 550~ 3.2. Effect of modification with APS on adsorption properties of TiOz Fig. 4(a) shows the amounts of VO3- adsorbed on unmodified TiO2 and APS-TiO2 (Ca~) as a function of equilibrium concentration of VO3-(Ceq). Both TiO2 samples gave Langmuir-type isotherms. From linear plots (Ceq vs. Ceq/Ca~s) (Fig. 4(b)), the limiting
1094 amounts of Cads (maXCads) were estimated to be 0.91 and 1.4 mmol g-1 for unmodified and modified TiOz, respectively, indicating that adsorptivity of VO3- onto TiO2 was improved by modification of TiOz with APS. Since the methoxy group of APS is reactive to HzO or moisture, pKa of APS can not be determined directly in the presence of HzO. Here, pKa of propylamine (10.7) is regarded as that of the amino group of APS-TiOz. The large pKa indicates that the amino group of APS-TiOz is protonated in the NH4VO3 solution (pH 6.8). Therefore, the strong interaction between VO3- and the protonated amino group in APS-TiO2 is attributable to the larger amount of VO3- adsorbed. From the results of chemical analysis and maXCads, the ratio of VO3- adsorbed to APS was calculated to be 1.0, indicating that all of the amino groups of APS-TiOz interacted with VO3-. Therefore, despite the higher loading, greater dispersion of VO3- on the APS-TiOz support is expected because one VO3- anion interacts with the protonated amino group of APS. 1.5
(a)
=
"o,
20 l
(b)
~ 15 0
0~
E E co
0
"o
0.5 5
0 1
0
=
0
5 10 15 20 0 5 10 15 20 Fig. 4. (a) Isotherms of adsorption of VO3- (Ceq-Cads plot) and (b) Ceq-Ceq/Cads plot. Circles and squares show data of unmodified TiO2 and APS-TiOz, respectively.
3.3. Properties and deNOx activities of supported V205 catalysts The unmodified and modified TiOz samples adsorbing VO3- in the most-concentrated NH4VO3 solution (2.00 • 10.2 mol dm -3) were decomposed at 550~ for 1 h in air to form two supported V205 catalysts. The VzO5 catalyst, prepared from the VO3--adsorbing unmodified TiO2, had a small surface area (81 mZg-1), as expected from the low thermal stability of the unmodified TiOz. It is well known that an excess loading of VzO5 on a TiO2 support generally induces sintering and pore-mouth plugging of the support, resulting in a decrease in surface area of the catalyst. Therefore, the amount of VO3- loading (1.0 mmol g-i) on the unmodified TiOz from the NH4VO3 solution exceeded the limited amount of the support for effective dispersion of V205 on it. On the other hand, the other V205 catalyst, prepared from the VO3--adsorbing APS-TiOz sample, retained a sufficient surface area (144 mZg-1) as expected from the improved thermal stability of the APS-TiO2 support. This indicates that VzO5 species were effectively dispersed on the modified TiOz support despite
1095 the large amount of V205 loading. Thus the modification of a TiO2 support with a large surface area with APS enables preparation of V205 catalyst with both a large surface area and a large amount of V205 loading.
60
~
40
> 0 E •
0
20
Z
I
I
I
300 350 400 Temperature/bO Fig. 5. Removal of NOx over V205 catalysts. Circles and squares show results of the catalysts prepared from VO3-adsorbing unmodified and APS-modified TiO2 samples, respectively. Reaction conditions are described in the Experimental section. These two V205 catalysts were used for SCR of NOx with ammonia in a temperature range from 300 to 400~ and the results are shown in Fig. 5. With elevation in the reaction temperature, the amount of NOx removed increased. Due to the large amount of highly dispersed VzO5 species, i.e., large number of active sites, V=O, the V205 catalyst prepared from APS-TiO2 exhibited higher activity than that of a catalyst prepared from unmodified TiO2. 4. CONCLUSIONS Chemical modification of a TiO2 support with a large surface area with organosilane having an amino group suppressed the growth and sintering of anatase crystallite, resulting in improved thermal stability of the support. The adsorption properties of VO3- were also improved by the strong interaction with VO3- and the protonated amino group in the modified TiO2. A supported V205 catalyst prepared from VOa--adsorbing modified TiO2 satisfies both the requirements of large surface area and large V205 loading and, in SCR of NOx with ammonia, exhibits higher activity than that of a catalyst prepared from unmodified TiO2.
1096 REFERENCES
1. S. Matsuda and A. Kato, Appl. Catal., 8 (1983) 149. 2. M. Inomata, A. Miyamoto and Y. Murakami, J. Chem. Soc., Chem. Commun., (1980) 233. 3. E Luck, Bull. Soc. Chim. Belg., 100 (1991) 781. 4. M. Kamada, H. Nishijima and Y. Kera, Bull. Chem. Soc. Jpn., 66 (1993) 3565. 5. Y. Hanada, M. Kamada, K. Umemoto, H. Kominami and Y. Kera, Catal. Lett., 37 (1996) 229. 6. M. Kamada, H. Kominami and Y. Kera, J. Colloid Interface Sci., 182 (1996) 297. 7. M. Kamada, H. Kominami and Y. Kera, Nippon Kagaku Kaishi, (1996) 300 [in Japanese]. 8. Y. Kera, M. Kamada, Y. Hanada and H. Kominami, Composite Interfaces, 8 (2001) 109. 9. H.P. Boehm, Disc. Faraday Soc., 52 (1971) 264. 10. S. Iwamoto, W. Tanakulrungsank, M. Inoue, K. Kagawa and E Praserthdam, J. Mater. Sci. Lett., 19 (2000) 1439. 11 H. Kominami, M. Kohno, Y. Matsunaga and Y. Kera, J. Am. Ceram. Soc., 84 (2001) 1178.
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
1097
Design, preparation and testing of effective FeOx/SiOz catalysts in methane to formaldehyde selective oxidation F. Arena a,,, F. Frusteri b, L. Spadaro b, A. Venuto a, A. Parmaliana a a. Dipartimento di Chimica Industriale e Ingegneria dei Materiali, Universit~ degli Studi di Messina, Salita Sperone 31, 1-98166 S. Agata (Messina), ITALY b Istituto CNR-TAE, Via Salita S. Lucia 39, 1-98126 S. Lucia (Messina), ITALY A preparation method of highly effective methane to formaldehyde selective oxidation (MPO) FeOx/SiO2 catalysts, based on "adsorption-precipitation" of Fe II ions under controlled conditions (ADS/PRC), is reported. The performance of ADS/PRC catalysts in the MPO reaction at 650~ has been compared with that of conventional systems prepared by the "incipient wetness" of silica carriers with aqueous solutions of Fe Iu (INC/WET). The ADS/PRC method, enabling a higher dispersion of the active phase, provides very effective MPO catalysts featuring CH4 turnover frequency (TOF) and HCHO productivity (STYHcHO) values larger than those of the counterpart INC/WET systems. 1. INTRODUCTION A large number of research papers devoted during the last decade to the catalytic partial oxidation of methane to formaldehyde (MPO) allowed the peculiar functionality of the silica surface in driving HCHO formation to be highlighted [1-3]. The preparation method affects the catalytic performance of silica samples, according to the following activity scale: "precipitation > sol-gel > pyrolysis" [1,4], as it determines the density of "strained siloxane bridges" early claimed to be the active sites in MPO [1,4]. In fact, recent studies dealing with the origin of the catalytic functionality of commercial silicas [5-7], really pointed to constitutional Fe 3+ ions incorporated into the matrix during preparation as the active centres enabling the occurrence of redox cycles under MPO conditions [5-11]. Direct relationships between the concentration of "isolated" Fe 3+ species in a rhombic-like coordination and either reaction rate or HCHO productivity (STYHcHo) have been outlined, whereas Fe203 clusters, bearing Fe 3+ ions in an octahedral symmetry, mostly drive COx formation owing to a high availability and mobility of lattice oxygen ions [5-7]. Such findings undoubtedly pointed to dispersion as the key-factor affecting the performance of the FeOx/SiO2 system in MPO [5,6], stimulating further research efforts aimed to disclose effective preparation methods enabling a high dispersion of the active phase [7,11]. Actually, due to a relatively high level and dispersion of iron impurities [5,6], we found that a "precipitated" SiOz sample (Si 4-5P grade, Akzo product) features the best "activityselectivity" pattern and highest HCHO productivity (STYHcHo) in the range 650-750~ * e-mail address:
[email protected] 1098
[3,12]. Such promising catalytic performance prompted us to investigate in some detail also the kinetics and mechanism of the MPO reaction on the "precipitated" silica catalyst [8-10] putting the scientific basis for further catalyst improvements [8]. Therefore, this contribution is aimed at providing basic evidences into the suitability of the preparation method based on the "adsorption-precipitation" of Fe H precursors to attain highly dispersed FeOx/SiO2 systems featuring a superior catalytic performance in MPO.
2. EXPERIMENTAL 2.1. Catalysts A series of silica supported iron catalysts (FS-x) was prepared by the "adsorptionprecipitation" (ADS/PRC) method [7,11] according to the following procedure using either Si 4-5P (Akzoproduct) or F5 (Akzoproduct) "precipitated" silicas as carrier [11]. 10 g of a powdered silica carrier were put into 0.3 g of distilled water (pH=3) and the suspension was vigorously stirred under a nitrogen flow for lh at r.t. to remove any dissolved oxygen avoiding any air admission. Then, an amount of FeSO4, corresponding to the designed Fe loading, was added to the stirred suspension, raising progressively the pH to a value of 7.07.5 by adding a 10% NH4OH solution. The suspension was kept at the final pH value for lh to attain a complete adsorption of the precursor, with an efficiency higher than 98%. Thereafter, the catalyst was filtered, washed, dried at 100~ and then calcined at 600~ in air for 16 h. For the sake of comparison, a series of silica supported iron catalysts (x-FS) was prepared by the "incipient wetness" (INC/WET) method, using powdered silica samples and aqueous solutions (pH =3) of Fe(NO3)3. After impregnation, the catalysts were dried overnight at 100~ and then calcined at 600~ (16 h). The list of the studied catalysts is reported in Table 1. Table 1 List of SiO2 and FeOx/SiO2 catalyst samples Code SiO2 Preparation method - Fe precursor
S.A.BET (m2.g"1) 385 607
S.L. (Feat.nm"2) 0.011 0.003
Si 4-5P F5
Si 4 5P F5
precipitation precipitation
Fe loading (wt%) 0.020 0.015
1-FS 2-FS 3-FS
F5 Si 4 5P Si 4 5P
INC/WET - FeIII INC/WET - FeIII INCAVET- FeIII
0.095 0.10 0.43
593 402 398
0.018 0.027 0.118
FS-1 FS-2 FS-3
Si 4 5P F5 F5
ADS/PRC- FeII ADS/PRC - FeI1 ADS/PRC - Feu
0.35 0.09 0.37
399 601 597
0.096 0.081 0.018
support
2.2. Catalyst Characterization The redox properties of FeOx/SiO2 catalysts were probed by Temperature Programmed Reduction (TPR) measurements in the range 200-800~ using a 6% H2/Ar reducing mixture flowing at the rate of 30 stp cma'min-1 and a heating rate of 20 K-min -1.
1099 Quantitative calibration of peaks area was made by monitoring the reduction of known amounts of CuO.
2.3. Catalyst Testing Catalyst testing in MPO was performed by a specifically designed recirculation batch reactor operating at 650~ and 1.7 atm total pressure with a flow rate of 1,000 stp cm 3rain-1 (He:N2:CH4:O2=6:1:2:1) and a catalyst sample of 50 mg unless otherwise specified [5-11]. 3. RESULTS and DISCUSSION 3.1 Theoretical basis to FeII "adsorption-precipitation" preparation route The marked tendency of Fe 3+ ions to form the insoluble hydroxide (Kps Fe(OH)3, 1.2-1038) in aqueous solution prompted since 1969 the use of Fe 2+ precursors to attain a high dispersion of the active phase, basically, across the ZSM-5 zeolite support [13,14]. That is, conventional impregnation methods based on aqueous impregnation (e.g., "incipient wetness", "ion exchange" or "electrostatic adsorption") of Fe III precursors, preventing an effective interaction of the positive Fe-hydroxo-ions with hydroxyl groups of acidic oxide carriers at pH>6, cannot m principle enable a high dispersion of Fe species in supported systems [7,11,13]. On the other hand, special preparation methods like gas-phase grafting of Fe 3+ precursors (i.e., FeC13), although enabling a high dispersion, implies the deposition of foreign ions (i.e., C1-) on the support requiring further treatment steps for their removal [13,14]. Then, the "adsorption-precipitation" of Fe 2+ ions (ADS/PRC) in aqueous solutions, under N2 atmosphere and controlled pH conditions [13], has been exploited to attain a higher dispersion of MPO FeOx/SiO2 catalysts [7,11]. Such a preparation route, preventing the formation of insoluble hydroxides (Kps Fe(OH)2, 1.8-10 -12) also at high pH values (pH>6), allows for a progressive interaction of the negatively polarised silica support surface with Fe 2+ ions leading thus to a selective tailoring of "isolated" Fe 3+ moieties onto the matrix [7,11]. .
.
.
.
.
.
III
9
3.2 Catalyst performance A first evidence for a different effectiveness of ADS/PRC and INC/WET preparation routes in promoting the dispersion of Fe uI on the silica carriers stems from a different colour of the two series of catalysts, mostly at Fe loadings higher than 0.1 wt%. Namely, while a typical brownish dye of INC/WET catalysts, the intensity rising with loading, likely signals the presence thereon of Fe203 clusters at any loading (0.09-0.43 wt% Fe), ADS/PRC samples resulted colourless up to the highest investigated Fe loading (0.37 wt%). Then, the effectiveness of the ADS/PRC method in promoting the dispersion of the supported phase and consequently the reactivity of FeOx/SiO2 system in MPO, has been evaluated by comparing the performance of catalysts prepared using either Si 4-5P or F5 silica carriers by INC/WET and ADS/PRC methods, respectively. Activity data at 650~ of the various catalysts in the MPO reaction, with reference to Si 4-5P and F5 silica carriers, are presented in Table 2 in terms of hourly CH4 conversion (XcH4,h, O~), product selectivity (Sx, %) and reaction rate (rate, ~molcu4"S-l'g-1), while the relative space time
1100
yield (STYHcHO, g'kgcat-l'h -1) and CH4 turnover frequency (TOF, atmcH4-1.s-1) values, calculated on the basis of the actual Fe loading (Table 1), are compared in Fig. 1. According to our previous findings [5,6], Fe-addition always results in a marked promoting effect on the activity of whatever silica support, though it is evident that the ADS/PRC route is considerably more effective than conventional INC/WET one in enhancing the catalytic performance of the FeOx/SiO2 system in MPO at any loading [7,11]. In particular, the bare Si 4-5P silica support features a hourly CH4 conversion of 3.5%, corresponding to a methane TOF of 3.0 atmcH4-1-s-1, along with a SHCHOvalue of 80% (Sco, 14%; Sco2,6%) which account for a STYHcHO of 310 g'kgcat-l"h-1 (Fig. 1). In spite of its larger surface area (Table 1), the F5 silica carrier displays both lower activity (XcH4, h, 2.9 %) and SHCHO(64%) with respect to Si 4-5P sample, resulting in a STYHcHO of 195 g'kgcatl'h-l. Yet, while a TOF value of 3.0 atmcHn=l-s-~ (Fig. 1), equal to that of the Si 4-5P silica catalyst, proves that the decrease in activity is the consequence of the lower Fe content (see Table 1), the drop in SHCHO, counterbalanced by a specular rise in Sco, reflects a higher tendency of such system to drive the consecutive oxidation of the primary oxidation product [1-3,5-7], likely because of the larger development in surface area [11]. Table 2 Activity d a t a of SiO2 a n d FeOx/SiO2 c a t a l y s t s in M P O at 650~ Wcat XCH4,-h SHCHO Sco Sco2 rate Sample (mg) (%) (%) (%) (%) (~tmolcH4"Sl'g 1) Si 4-5P 50 3.5 80 14 6 3.6 F5 50 2.7 65 27 8 2.8 1-FS 2-FS 3-FS
50 50 50
8.6 13.4 13.2
55 56 39
28 22 30
17 22 31
8.8 13.8 13.6
FS-1 FS-2 FS-3
50 50 50
37.2 14.9 34.2
33 64 35
29 22 28
31 12 27
38.3 15.0 35.4
FS-1
5.0
4.6
55
23
22
47.3
Addition of ca. 0.10 wt% of Fe to the Si 4-5P silica sample by INC/WET (2-FS) implies a ca. four-fold rise in XCH4,h (13.4%) and rate (13.8 txmOlcH4"Sl"g 1) along with a lowering in SHCHOfrom 80 to 56% (Table 2). With reference to the Si 4-5P silica carrier, the above figures account for a decrease in TOF from 3.0 to 2.1 atmcH4-1"s-1 but in a marked rise of STYHcHO from 310 to 830 g'kgcat-l"h-1, respectively (Fig. 1). A comparable amount of Fe (0.095 wt%) added to F5 silica carrier still by INC/WET (1-FS) attains a lower promoting effect on the activity (Xcn4, h, 8.6%), while the product distribution (SHcHO, 55%) looks similar to that of the previous sample. These catalytic data reflect in lower reaction rate (8.6 ~tmolcH4"Sl"g1) and STYHcHO (525 g'kgcat-l'h -1) values which, on the basis of a TOF equal to 1.5 atmcHn-l's q (Fig. 1), is ascribable to a poorer dispersion of
1101 the active phase [11]. On the other hand, the FS-2 system, bearing 0.095 wt% of Fe loaded by the ADS/PRC method, displays the highest activity (XcH4,h, 14.9%,) and an even larger SHCHO (64%) accounting for TOF and STYHcHO values equal to 2.7 atmcH4-1"s-1 and 1,040 g'kgcat-l"h-1 respectively; these being considerably larger than those of both 1-FS and 2-FS catalysts (Fig. 1). Compared to the 2-FS system, addition of 0.43 wt% of Fe III to Si 4-5P carrier by INC/WET (3-FS) has, yet, a negative impact on all the catalytic evaluation parameters (e.g., SncHo, TOF, STYHcHO) [11]. Indeed, while a XCH4,.h of 13.2%, practically equal to that of the 2-FS sample, and a lower SHCHO (39%) accounts for a STYHcHO of only 570 g'kgcat-l"h-1, the rise of both Sco (30%) and Sco2 (31%) and the concomitant marked decrease in TOF (0.5 atmcH4-1"s-1) altogether are diagnostic of a weak catalytic functionality consequent to a drop in FeO,, dispersion [7,11]. Despite the comparable Fe loading (0.35 wt%), the FS-1 sample, obtained by ADS/PRC of Fe 2+ ions on Si 4-5P silica, features an impressive activity (Xcu4,.h, 37.2%) larger by ca. one order of magnitude than that of the relative silica carrier and ca. three times higher than that of the counterpart 3-FS catalyst, though SHCHO (33%) keeps at a comparable level of the former system. Considerably higher TOF and STYHcHO values, equal to 1.9 atmcH4-1-s-1 and 1,360 g'kgcat1.h-I respectively, indicate that the FS-1 catalyst is ca. four times more active than the counterpart 3-FS sample, though it maintains a satisfactory functionality towards HCHO formation.
Fig. 1. MPO on silica and FeOx/SiO2 catalysts at 650~ Methane turnover frequency (TOF, atmcH4-1-s-1) and space time yield (STYHcHO,g'kgcat-l"h-1) values.
1102 Herewith, such findings undoubtedly prove that the ADS/PRC method confers to FeOx/SiOz catalysts a considerably superior activity in MPO with respect to the conventional INC/WET route, also irrespective of the silica carrier. In fact, considering the FS-3 sample (0.37 wt% Fe), prepared by ADS/PRC on F5 silica carrier, an activityselectivity pattern quite similar to that of the homologous FS-1 system (XcH4, h, 34.2%) is noticed, while also the TOF and STYHcHO values, resulting equal to 1.7 atmcH4-1"s-1 and 1,340 g'kgcat-l"h-1 respectively, well compare with the values of the latter sample (Fig. 1). Since the high value of CH4 conversion (ca. 13.5%) of the above two system which could result in reaction kinetics controlled by the 02 conversion level [8-10], by tuning a timely decrease in contact time (Wcat, 0.005 g), the activity-selectivity pattern of the FS-1 catalyst at a CH4 conversion level comparable with that of the Si 4-5P silica carrier has been evaluated. Such data, also included in Table 2, outline a XCH4, h of 4.6% corresponding to a TOF equal to 2.3 atmcn4-1"s-1, larger by ca. 20% than the value (1.9 atmcn4q.s -1) at higher z. A SHCHOlevel of 55% coupled to such high specific activity allows for a steep rise in STYHcHO to a value of 2,810 g'kgcat-l'h -1. Notably, such a figure results one order of magnitude greater than that of the related bare silica sample and, really, represents a breakthrough in view of a potential industrial exploitation of the MPO reaction [7,10-12].
3.3. Structure and redox properties of FeOx/SiO2 catalysts In order to ascertain if the catalytic pattern of the studied systems is fairly related to the dispersion of the active phase, the redox properties of representative FeOx/SiOz catalysts have been probed by TPR measurements with the aim to shed lights into the reduction pattern of the various Fe 3+ species and getting insights into their surface structures. Then, to highlight the influences of the Fe loading and preparation method, the TPR profiles of F5 silica and differently loaded (0.015-0.43 wt% Fe) INC/WET FeOx/SiOz catalysts are compared in Figure 2A, while those of Si 4-5P silica, FS-3 and 1-FS catalysts are shown in Figure 2B. Moreover, for the sake of comparison the TPR profile of bulk Fe203 (d) is reported, the values of the peak maxima (TMi) and the extent of H2 consumption being summarised in Table 3. Since an analogous very low Fe loading level (Table 1), the bare F5 (Fig. 2A, a) and Si 4-5P (Fig. 2B, a) silica samples feature an analogous TPR profile, spanned in the range 400-800~ being characterised by a main reduction peak with a broad TM2 maximum at 540-550~ and an unresolved one at ca. 600~ (TM3), evidently skewed on the high T side. This spectrum likely denotes the reduction of highly dispersed Fe 3+ moieties, in the form of "isolated" ions in a strong interaction with the silica matrix [5,6]. Indeed, according to literature data, pointing to a hard reduction of supported Fe 3+ ions strongly interacting with carriers [15], a Hz/Fe ratio close to 0.5 accounts for the reduction of Fe 3+ ions to Fe z+.
1103
Fig. 2. TPR profiles of silica and FeOx/SiO2 catalysts. Legend: (A) F5 silica (a), FS-2 (b) and FS-3 (c), Fe203 (d); (B) Si 4-5P silica (a), 3-FS (b); FS-1 (c); Fe203 (d).
Table 3 T P R of SiO2 a n d FeOx/SiO2 c a t a l y s t s
(*C)
(*C)
TM3
Hz consumption
Sample
TMI
TMZ
(*C)
(Hzmoi/Feat)
Fe203
419
541
680
1.48
Si 4-5P F5
420
530 558
583 604
0.45 0.48
3-FS
399
FS-1 FS-2 FS-3
423 409 421
0.57 510 479
570 651 -
0.51 0.49 0.56
1104
Addition of 0.095 wt% Fe by ADS/PRC to the F5 silica carrier (FS-2) implies a corresponding rise in the intensity of peaks area, while no significant differences in qualitative terms are noticed with respect to the spectrum of the related support. In particular, besides to a systematic shift of the reduction maxima to lower T (Table 3), a relatively larger intensity of the TM3 peak is recorded, while the TM1 component still looks as an unresolved shoulder of the main TMZ peak (Fig. 2A, b). At higher Fe laoding (Fig. 2A, c), the TPR profile of the FS-3 catalyst still outlines a resolved TM2 maximum, further shifted to lower T (479~ revealing also a concomitant increase in the intensity of the TM1 component and an almost total disappearance of the TM3 shoulder. Taking into account the TPR profiles of the highly loaded 3-FS and FS-1 catalysts (Fig. 2B) some peculiar differences connected with the different preparation method, are observable. Actually, the 3-FS catalyst (Fig. 2B, b) features one sharp and symmetric reduction peak centered at 399~ being virtually absent any other component at higher T. By contrast, the FS-1 catalyst (Fig. 2B, c) outlines a broader and asymmetric peak centered at 423~ clearly skewed on the high T side because of the contribution of other unresolved components. Quantitative data indicate for all the studied FeOx/SiO2 catalysts (Table 3) a Hz/Fe ratio ranging between 0.49 (FS-2) and 0.57 (3-FS). The TPR profile of the bulk FezO3 (Fig. 2A and B, d) practically spans the same T range of supported species (Fig. 1A and B), indicating minor differences in the reducibility of the various Fe 3§ moieties. However, quantitative data signal a quite different behaviour, since at 800~ a stoichiometric reduction of Fe In to Fe ~ (e.g., Ha/Fe=l.48) is attained in bulk Fe203 [15], clearly in contrast to that found for all silica-supported systems (Table 3). However, the related qualitative TPR features provide basic clues towards the identification of the various Fe 3§ moieties in supported systems taking into account also the general effects of loading and preparation method on dispersion of the active phase in supported systems. In particular, according to the TPR features of bare silica samples, the reduction of "isolated" Fe 3§ ions gives rise to the broad TMZ peak because of their distribution all over the various defect sites of the silica surface [5,6]. Accordingly, the low loaded (0.095% Fe) FS-2 sample presents the largest concentration of isolated species and Fe 3+ ions strongly interacting with the silica matrix (TM3) [5-7,11,15], while the specificity of the ADS/PRC method prevents an extensive formation of more easily reducible "small Fe203 clusters" [5-7,11], monitored by the TM1 peak. Notably, the same TM1 value recorded for both bulk Fe203 and supported catalysts (Table 3) points to the same origin of such a component. On the FS-3 catalyst (0.37% Fe) a further rise in the extent of isolated species is paralleled by an increased concentration of small FezO3 clusters which enhance the reduction kinetics at lower T [5,6], well evidenced by the larger intensity of the TM1 peak. Accordingly, the spectrum of the 3-FS sample (Fig. 2B, b) confirms that the INC/WET route, at such a "high" loading, leads more favourably to the formation of small Fe203 aggregates, whereas the broader TPR profile of the FS-1 system (Fig. 2B, c), signals a more "homogeneous" distribution of Fe Iu ions across the silica matrix allowing foe the formation of both "isolated moieties" and "small FezO3 clusters" [5,6,15].
1105
4. CONCLUSIONS A versatile preparation method of effective FeOx/SiO2 catalysts via the "adsorptionprecipitation" (ADS/PRC) of FeII precursors from aqueous solution is outlined. The ADS/PRC method ensures a high dispersion of the active phase, conferring a considerably higher performance to FeOx/SiO2 catalysts in MPO in terms of specific activity, selectivity and HCHO productivity. Basic evidences into the surface structures, redox properties and catalytic pattern of FeOx/SiO2 catalysts are provided. Formaldehyde productivity values as high as 9-10 kgncno'kgcat-l'h-l('atmcn4-1), ensured by "ADS/PRC" FeOx/SiO2 catalysts, represent a breakthrough in view of a potential exploitation of the MPO reaction on an industrial scale.
ACKNOWLEDGEMENTS This work has been realized in the framework of a research contract between SOD CHEMIE Ag (Miinchen, GERMANY) and University of Messina (Messina, ITALY). REFERENCES
1. 2. 3. 4.
Q. Sun, R.G. Herman and K. Klier, Catal. Lett., 16 (1992) 251. M.M. Koranne, J.G. Goodwin and G. Marcelin, J. Catal., 148 (1994) 378. A. Parmaliana and F. Arena, J. Catal., 167 (1997) 57. K. Vikulov, G. Martra, S. Coluccia, D. Miceli, F. Arena, A. Parmaliana and E. Paukshtis, Catal. Lett., 37 (1996) 235. 5. F. Arena, F. Frusteri, J.L.G. Fierro and A. Parmaliana, Stud. Surf. Sci. Catal., 136 (2001) 531. 6. A. Parmaliana, F. Arena, F. Frusteri, A. Martinez-Arias, M. Lopez-Granados and J.L.G. Fierro, Appl. Catal. A: General, 000 (2002) 000-000. 7. A. Parmaliana, F. Arena, F. Frusteri and A. Mezzapica, Get. Patent n. GEM 174 filing number 100 54 457.6 in the name of SOD CHEMIE AG (November, 2000). 8. F. Arena, F. Frusteri and A. Parmaliana, J. Catal., 207 (2002) 00-00. 9. F. Arena, F. Frusteri and A. Parmaliana, Appl. Catal. A: General, 197 (2000) 239. 10. F. Arena, F. Frusteri and A. Parmaliana, AIChE J., 46 (2000) 2285. 11. F. Arena, T. Torre, A. Venuto, F. Frusteri, A. Mezzapica and A. Parmaliana, Catal. Lett., (2002) accepted for publication. 12. A. Parmaliana, F. Arena, F. Frusteri and A. Mezzapica, Stud. Surf. Sci. Catal., 110 (1998) 665. 13. X. Feng, and W.K. Hall, Catal. Lett., 41 (1996) 45. 14. X. Feng, and W.K. Hall, J. Catal., 166 (1997) 36. 15. S. Yuen, J.E. Kubsch, J.A. Dumesic, N. Topsoe and H. Topsoe, J. Phys. Chem., 86 (1982) 3022.
This Page Intentionally Left Blank
Studies in Surface Science and Catalysis 143 E. Gaigneaux et al. (Editors) 9 2002 Elsevier Science B.V. All rights reserved.
1107
New Fe-Mo-Ti mixed oxides prepared via the sol-gel method: comparison of the textural properties with solids obtained by impregnation S.R.G. Carraz~n*, C. Martin, C.M. Pedrero and J. Saunders Departamento de Quimica Inorg~inica. Universidad de Salamanca., 37008-Salamanca, Spain
Iron-molybdenum-titanium oxides were prepared via the sol-gel method. Either (NH4)6Mo7Ozn'4HzO, and ferric nitrate (FEN), Fe(NO3)3"9HzO, or FeC13 and MoOCh were used as precursors together with Ti-isopropoxide. These solids were characterised by chemical analysis and N2 adsorption. The sol-gel samples developed higher surface areas (c.a. 100 mZg-1), except for the sol-gel solid prepared with Fe and Mo chlorides as precursors, than those prepared by impregnation. 1. I N T R O D U C T I O N The main industrial processes based on iron molybdate-containing catalysts are the production of formaldehyde from methanol [1,2] and the oxidation of toluene to benzaldehyde [3,4]. Since the first work of Adkins and Peterson [5], the growing industrial importance of Fe-Mo oxide catalysts has led to a large number of investigations, but controversial opinions has been found with respect to the relation between the method of preparation, the solid properties of the catalyst and its catalatytic behaviour. Several researchers have found that the catalytic behaviour of iron molybdate mostly depends on the Mo/Fe atomic ratio [6-10]. Stoichiometric iron molybdate (FezMo3Olz) is likely to be the active component of these catalysts. Boreskov [6] and Okamoto [11] showed that an excess of MoO3 in Fe-Mo-O catalysts promoted the formation of stoichiometric ferric molybdate, while other researchers [8,10] believe that the active sites of Fe-Mo-O catalysts are associated with Mo atoms in octahedral coordination, such coordination being only present in FeMo mixed oxides with an excess of Mo. Introduction of a support (i.e. TiO2) to these systems can induce positive effects on the active phase (FeMo), as to avoid an excessive sintering of the particles during the thermal treatment and/or modification of the reduction capacity and the acid properties. The synergy of iron molybdate and the support is therefore another way for improving the catalytic performance of these solids. Such benefitial effects have been detected in bismuth-molybdenum-titania mixed oxides prepared via sol-gel, in addition these solids resulted to be amorphous materials with a unique morphology and extraordinary dispersion of the active phase [12]. These results encouraged us to extend this field to iron molybdenum oxide catalysts.
1108 The aim of this work is to develop new Fe-Mo containing mixed oxides highly dispersed in a titania matrix, prepared by the sol-gel method, and to compare these materials to those of iron molybdate prepared by conventional methods (i.e. impregnation). Here, we report the preparation of sol-gel derived iron molybdenum titanium mixed oxides. The bulk composition and the textural properties of these materials are investigated by elemental chemical analysis and N2 adsorption, respectively. 2. EXPERIMENTAL
2.1. Sample Preparation Distilled water and analytical grade reagents (Fluka, Switzerland) were used in all preparations. Bulk Fe2Mo3012 (FM) was prepared by coprecipitation of ammonium heptamolybdate (AHM), (NH4)6Mo7024"4H20, and ferric nitrate (FEN), Fe(NO3)a'9HzO [13]. This oxide was calcined in the same conditions as the sol gel materials (see below). FeMoO catalysts were included in a TiO2 matrix using three methods, simultaneous impregnation (FMT-I samples), consecutive impregnation (FMT-IS samples) and the solgel (FMT-SG samples). 2.1.1. Samples prepared by impregnation: samples FMT-I and FMT-IS have been prepared by simultaneous or consecutive implregnation, respectively, of the titania support (Degussa P25, ca. 80% anatase, ca. 50 m g-, previously calcined at 500~ in air for 3 h) with aqueous solutions of ammonium heptamolybdate (AHM), and ferric nitrate (FEN. The precipitates were dried at 100~ and calcined at 500~ for 3 h after each impregnation. The relative amounts of AHM, FeN and titania were chosen to yield solids with a geometric monolayer of MoO3 after calcination, and molar Mo/Fe ratios of 3/2 for samples FMT-I and FMT-IS1. 2.1.2. Samples prepared by the sol-gel method: In Diagram 1, the steps followed for the preparation of the FMT solids via the sol-gel method [12] are indicated. First, a solution of titanium (IV) isopropoxide (11.5 ml) was prepared in methanol (40 ml). Then, the appropriate amounts of the Fe and Mo compounds were introduced by syringe along with 0.1 ml of concentrated HNO3 acting as an hydrolyzing agent. The H20:alkoxide:acid molar ratio was 5:1:0.01 and the molar Mo/Fe ratio was 3/2 (sample FMT-SG1). A second solid (FMT-SG2) containing an excess of MoO3 has also been prepared, with a Mo/Fe molar ratio of 6/2. Gelation always occurred within seconds at ambient temperature. The gels were stirred at 500 rpm for 20 h before being dried in vacuo at room temperature. A third solid (FMT-SG3) has also been prepared by the sol-gel method, but using Mo and Fe chlorides as precursors. In this case, an aqueous acid solution was preliminarily added to the Ti-alkoxide. After gelation has occurred, the solution containing FeCl3 and MoOC14 in isopropanol was injected via a syringe. The molar Mo/Fe ratios was also 3/2 for this sample.The gel was redispersed after the addition of the solution by vigorous stirring at 500 rpm for 20 h and the isopropanol was removed in vacuo at room temperature.
1109
A
B
Ti(i-OC3H7)4 in isopropanol
Fe(NOa)3.9H20
Ti(i-O~H7)4HNO3 + H20
(NHa)6Mo7024.9H20 (Mo/Fe =3/2) HNO3 + H20 (HzO:alcoxide:acid = 5:1:0.01)
]
Gelation, gmin
I
I Drying in v a c u o at r.t.
I
MoOC14 + FeC13 in isopropanol
(Mo/Fe =3/2)
]
Aging, 24 h Stirred at 500 r.p.m
Aging, 24 h Stirred at 500 r.p.m
Calcination in 02 atm.
] Gelation ]
I !
I Drying in vacuo at r.t.
I FMT Xerouel
Calcination in 02 atm.
[
I
Diagram 1. Experimental steps followed to obtain the FMT solids via the sol-gel method using, A) FeN and AHM and B) Fe and Mo chlorides, as precursors All the solids were calcined in flowing 02 at a heating rate of 5 C min -1 up to 250~ this temperature was maintained for 30 min in order to remove the organic residues. Then it was raised at 5~ min -1 up to 500~ and held for another 30 min. The solids were then cooled to ambient temperature whilst still in the oxygen flow. 2.2. Sample characterisation
Elemental chemical analysis (ECA) of iron, molybdenum and titanium were carried out by atomic absorption in a Mark-2 ELL 240 apparatus at Servicio General de Anfilisis Quimico Aplicado (University of Salamanca, Spain). The nitrogen adsorption-desorption isotherms for specific surface area and porosity assessment were recorded at -196-~ in a Gemini instrument from Micromeritics. The specific surface areas were determined by the Brunauer-Emmett-Teller (BET) method. The pore size distributions were obtained from the desorption branch, and the micropore volume was determined by the t-plot method, using literature software [14].
1110
3. RESULTS AND DISCUSSION
3.1. Elemental chemical analysis (ECA) Table 1 lists the results of ECA for some of the samples studied. The solids were not filtered during the synthesis procedure, and consequently the percentage of Fe, Mo and Ti determined by chemical analysis practically coincide with the expected values. Considering the sol-gel materials, the atomic ratio of 3/2 has been preserved throughout the preparation procedure only when iron nitrate and ammonium heptamolybdate were used as precursors (sample FMT-SG1). When using Fe and Mo chlorides precursors (sample FMTSG3), a loss of molybdenum was observed and the Mo/Fe atomic ratio was 1.28 instead of 1.5. Tablel. Bulk composition of the samples determined by ECA Sample [Bulk Mo/Fe(at/at)]exp FM 1.55 FMT-I 1.52 FMT-SG1 1.52 FMT-SG3 1.28
3.2. Textural properties The specific surface areas, calculated following the BET method are given in Table 2. The adsorption capacities of the samples prepared via sol-gel method are noticeably larger than for samples obtained by impregnation, and consequently the specific surface areas of the FMT-SG samples is ca. twice that of the samples prepared by impregnation. A different behaviour can be observed, however, for sample FMT-SG3; in this case, the calculated specific surface area is of the same order as for the impregnated samples. Although the origin of this difference is not clear at all, it is also true that the nature of the precursor used to prepare this sample has an important influence on the type of the pores developed in the sample and also on its surface area. Table 2. Summary of textural properties X* Sample --FM FMT I 1.44 FMT IS 1 1.44 FMT SG1 1.44 FMT SG2 1.38"* FMT SG3 1.44 *percentage molar fraction of FezMo3012 ** percentage molar fraction of MoO3 =4.14
SBEX (mZg-1) 10 47 47 99 111 47
1111 Nitrogen adsorption-desorption isotherms of the the mixed oxides obtained by the impregnation and sol-gel methods exhibit very different shapes, showing the great influence of the preparation method on their textural properties (Figs. 1 and 2).
16s0
A |
O)
o
, g
0
E
:=L .,_4
o.
"0
~=
f-
-FMT-ISI_ ,,1:) .0,, ~
,= ,, . . . . .
,~ ~,
.o.o.
o
o
O ~=0,.
o,=
FMT-I 6) .o= ~
FM
~.p. 0
...... ~ '=e''~
~, p ~ . p . 0.2
~ o
0"
.o- ~'
~
= p ~.?. 0.4
= f,.o.p 0.6
,, ~ - o - j 0.8
c~
p/pO
1.0
Fig. 1. Nitrogen adsorption-desorption isotherms a t - 1 9 6 ~ impregnation
of samples prepared by
For comparison the isotherm of FM is also shown in Fig. 1. The isotherm for bulk iron molybdate is completely reversible, but the adsorption capacity is extremely low, the specific surface area being about 10 m2 g-1. The isotherms for samples prepared by impregnation (both simultaneous and consecutive) belong to type II in the IUPAC classification [15]. In all cases, a type H3 hysteresis loop, originated by particle aggregates forming slit-like pores, is observed. Samples FMT-SG1 and FMT-SG2 (not shown) exhibit, however, type IV isotherms, with type H2 hysteresis loops (Fig. 2). A different behaviour can be observed, however, for sample FMT-SG3; in this case, the shape of the isotherm again corresponds to type II, with a type H3 hysteresis loop. It should be also noticed that micropores do not develop in any of the samples, as concluded from the t-plots (not shown), which correspond to straight lines passing through the origin upon extrapolation. Pore size distribution curves for the samples prepared by impregnation shows the major presence of pores with diameters ranging from 3 to 4 nm, while larger pores (ca. 4 nm diameter) should be present in samples prepared by the sol-gel method (Figs. 3 and 4).
1112
l soo o
o o. I-' '
FMT-SG3
A 01
~.o.q
~' , p . O .
.o.o.. o
.o.O.
m
0
E
6
o
o
o
o
o
9
9 ~
9~
o
o
F M T - S G 1~
0
"
,o , o 9G o
i
I
I
I
0,2
0.4
I
I
0.6
0.8:
1.0
p/p0
Fig. 2. Nitrogen adsorption-desorption isotherms at -196~ of samples prepared by sol-gel
13 E r o
E
"o
l
~
o ~,Lg, . t l
~,~ 9
~--~ ' , ~ . . . . . . . . . . . N
-"N
ffl
FMT-IS1
~'~~r-~,
.",'--,'--~
,
FMT-I
FM . . . . . . . . "71"1
"= -
~N"
d (nm) Fig.3. Pore size distributions of samples prepared by impregnation
1113
[10
E c
E