New Developments in Selective Oxidation by Heterogeneous Catalysis: Proceedings (Studies in Surface Science and Catalysis)

Studies in Surface Science and Catalysis 72 NEW DEVELOPMENTS IN SELECTIVE OXIDATION BY HETEROGENEOUS CATALYSIS This ...

Author: P. Ruiz

62 downloads 1250 Views 21MB Size Report

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

DOWNLOAD PDF

Studies in Surface Science and Catalysis 72

NEW DEVELOPMENTS IN SELECTIVE OXIDATION BY HETEROGENEOUS CATALYSIS

This Page Intentionally Left Blank

Studies in Surface Science and Catalysis Advisory Editors: B. Delrnon and J.T. Yates Vol. 72

NEW DEVELOPMENTS IN SELECTIVE OXIDATION BY HETEROGENEOUS CATALYSIS Proceedingsof the Third European Workshop Meeting on New Developments in Selective Oxidation by Heterogeneous Catalysis Louvain-la-Neuve, Belgium, April 8-1 0, 1 9 9 1

Editors

P. Ruiz and B. Delmon Unite de Catalyse et Chimie des Materiaux Divises, Universite Catholique de

Louvain, Louvain-la-Neuve, Belgium

ELSEVIER

Amsterdam - London - New York - Tokyo

1992

ELSEVIER SCIENCE PUBLISHERS B.V. Sara Burgerhartstraat 25 P.O. Box 21 1, 1000 AE Amsterdam, The Netherlands Distributors for the United States and Canada:

ELSEVIER SCIENCE PUBLISHING COMPANY INC. 655, Avenue of the Americas New York, NY 10010, USA

ISBN 0-444-89466-7

0 1992 Elsevier Science Publishers B V All rights reserved No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science Publishers B V , Copyright and Permissions Department, P 0 Box 52 1, 1000 A M Amsterdam, The Netherlands Special regulations for readers in the USA -This publication has been registered with the Copyright Clearance Center Inc (CCC), Salem, Massachusetts Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA All other copyright questions, including photocopying outside of the USA, should be referred to the copyright owner, Elsevier Science Publishers 6 V , unless otherwise specified

No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this publication does not constitute a guarantee or endorsement of the quality or value of such product or of the claims made of it by its manufacturer pp. 117-122, 191-201,231-246,

279-304,353-362,

This book is printed on acid-free paper Printed in The Netherlands

435-441. copyright not transferred

V

PREFACE This volume contains the invited papers and communications presented at the Third European Workshop Meeting "New Developments in Selective Oxidation", held in Louvain-la-Neuve, Belgium, April 8-10, 1991. The First European Workshop Meeting took place in Louvain-la-Neuve in 1986 (Catalysis Today, Volume 1, Numbers 1 and 2, 1987). The First International Conference on "New Developments in Selective Oxidation", held in Rimini, Italy, September 18-22, 1989 (Studies in Surface Science and Catalysis, Volume 55, 1990) constituted simultaneously the Second European Workshop. The Third European Workshop was organised by the Unit6 (Laboratory) de Catalyse et Chimie des MatCriaux DivisCs of the UniversitC Catholique de Louvain in Louvainla-Neuve, under the auspices of the National Science Foundation of Belgium (FNRS-NFWO), the UniversitC Catholique de Louvain and the Commission of the European Communities (Brite-Euram Programme, Directorate General for Science, Research and Development). We acknowledge their support with gratitude. It is also with pleasure that we thank the companies that contributed financially to the organization of this meeting: ATO-CHEM, Paris (France), DOW Benelux N.V., Terneuzen (The Netherlands), DSM Research, Geleen (The Netherlands), REPSOLPETROLEO S.A., Madrid (Spain) and UNICAT, Brussels (Belgium). The meeting was attended by over 150 researchers from 20 countries. About 50% of the participants came from industrial companies. The programme of the meeting consisted of three invited lectures, ten extended communications, 16 communications and 13 posters. The organization of the topics in this volume is very similar to that in Volume No. 55 of the same series. We hope that this will give the reader an idea of current trends in the field of catalytic oxidation. We owe much to the members of the Organizing Committee and the Scientific Committee. Their help and advice were crucial for the realization of this meeting and in enabling it to reach a high scientific level. We are very grateful to them for their help. We also thank all the chairmen of the sessions for efficiently leading the discussions. Our thanks further go to all the members of the "Unit6 de Catalyse et Chimie des MatCriaux DivisCs" who worked very hard to ensure the success of this meeting. Special thanks are addressed to Mrs. Nathalie Blangenois and Mrs. Marianne Saenen. P. Ruiz and B. Delmon


VII

CONTENTS Preface Organization

V

XI11

Section 1 PROSPECTIVES IN SELECTIVE LIQUID-PHASE OXIDATION

Noble metal catalyzed oxidation of carbohydrates and carbohydrate derivatives (invited lecture) P. Vinke, D. de Wit, A.T.J.W. de Goede and H. van Bekkum

1

Alkane oxygenations by H,O, on titanium silicalite D.R.C. Huybrechts, Ph. Buskens and P.A. Jacobs

21

Selective oxidation of hydrogen to hydrogen peroxide L. Fu, K.T. Chuang and R. Fiedorow

33

The selective oxidation of methyl-a-D-glucoside on a carbon supported Pt catalyst Y. Schuurman, B.F.M. Kuster, K. van der Wiele and G.B. Marin

43

Section 2 OTHER HETEROGENEOUS SELECTIVE OXIDATION REACTIONS

Catalytic gas-phase oxidation of fluorene, anthracene and phenanthrene to quinones and dicarboxylic anhydrides M. Baerns, H. Borchert, R. Kalthoff, P. Kassner, F. Majunke, S. Trautmann and A. Zein

57

The influence of water on the oxydehydrogenation of isobutyric acid over heteropolyacid catalysts 0. Watzenberger and G. Emig

71

Stabilization of heteropolyacids by various supports M.J. Bartoli, L. Monceaux, E. Bordes, G. Hecquet and P. Courtine

81

VIII

New compounds of the vanadium-molybdenum oxide system. In situ investigation of the mechanism of acrolein oxidation to acrylic acid. The role of the structure and bond energy of the intermediate compounds T.V. Andrushkevich, V.M. Bondareva, G.Ya. Popova and L.M. Plyasova

91

Reaction of methyl acetate with methylal in the presence of oxygen M. Ai

101

On the bifunctional nature of gas-phase cyclohexanone ammoximation catalyst D.P. Dreoni, D. Pinelli and F. Trifiro’

109

Selective catalytic oxidation of N-, 0- and S-methyl-heterocyclic compounds L. Leitis, R. Skolmeistere, I. Iovel, Yu. Goldberg, M. Shymanska and E. Lukevics

117

Selective oxidation of hydrogen sulfide to elemental sulfur by supported iron sulfate catalysts P.J. van den Brink, R.J.A.M. Terorde, J.H. Moors, A.J. van Dillen and J.W. Geus

123

Selective oxidation of ammonia to nitrogen over silica supported molybdena catalysts. A structure-selectivity relationship M. de Boer, A.J. van Dillen, D.C. Koningsberger, F.J.J.G. Janssen, T. Koerts and J.W. Geus

133

High performance of vanadia catalysts supported on Ti0,-coated silica for selective oxidation of ethanol N.E. Quaranta, V. CortCs CorberAn and J.L.G. Fierro

147

Section 3 ADVANCES IN C,-C, ALKANE SELECTIVE TRANSFORMATION Methane, ethane and propane

Oxidative conversion of light alkanes on silver catalysts A.G. Anshits, S.N. Vereshchagin, A.N. Shigapov and H.D. Gesser

155

Catalytic properties of promoted vanadium oxide in the oxidation of ethane in acetic acid M. Merzouki, B. Taouk, L. Monceaux, E. Bordes and P. Courtine

165

Microcalorimetric studies of the oxidative dehydrogenation of ethane over vanadium pentoxide catalysts J. Le Bars, A. Aurow, J.C. Vedrine and M. Baerns

181

IX

Oxidative dehydrogenation of ethane on chromium modified zirconium phosphates M. Loukah, G. Coudurier and J.C. Vedrine

191

Nature of surface sites in the selective oxide hydrogenation of propane over V-Mg-O catalysts A. Guerrero-Ruiz, I. Rodriguez-Ramos, J.L.G. Fierro, V. Soenen, J.M. Herrmann and J.C. Volta

203

Oxidative dehydrogenation of propane over supported-vanadium oxide catalysts A. Corma, J.M. Upez-Nieto, N. Paredes, M. PLrez, Y. Shen, H. Cao and S.L. Suib

213

The selective oxidative dehydrogenation of propane on catalysts derived from niobium pentoxide: preparation, characterisation and properties R.H.H. Smits, K. Seshan and J.R.H. Ross

22 1

Butane and pentane

Problems and outlook for the selective heterogeneous oxidation of C, alkanes (invited lecture) G. Centi, J.T. Gleaves, G. Golinelli and F. Trifiro’

23 1

Vanadyl pyrophosphate as a selective oxidation catalyst

I. Matsuura

247

Butane oxidation to maleic anhydride on VPO catalysts: the importance of the preparation of the precursor on the control of the local superficial structure N. Guilhaume, M. RoulIet, G. Pajonk, B. Grzybowska and J.C.Volta

255

Synergetic effects in phosphorus vanadia catalysts Ph. Bastians, M. Genet, L. Daza, D. Acosta, P. Ruiz and B. Delmon

267

Section 4 NEW ASPECTS OF THE MECHANISM AND SURFACE REACTIVITY OF SELECTIVE OXIDATION CATALYSTS

Catalytic oxidation: State of the art and prospects (invited lecture) J. Haber

279

Kinetics of the reoxidation of propylene-reduced y-bismuth molybdate: a TAP reactor study D.R. Coulson, P.L. Mills, K. Kourtakis, J.J. Lerou and L.E. Manzer

305

X

TAP investigations of selective o-xylene oxidation F.-D. Kopinke, G. Creten and G.F. Froment

3 17

Temperature programmed desorption of oxygen on bismuth molybdates and reactivity for olefin oxidation M. Farinha-Portela, C. Pinheiro and M. Oliveira

325

An infrared spectroscopic study of the interaction of olefins on

vanadia-titania and PdC1,-vanadia-titania selective oxidation catalysts V. Sanchez Escribano, G. Busca, V. Lorenzelli and C. Marcel

335

Kinetic problems of selectivity in oxidation catalysis S.L. Kiperman

345

Site isolation in vanadium phosphorus oxide alkane oxidation M.R. Thompson and J.R. Ebner

353

Synergy effects in selective oxidation catalysis U.S. Ozkan, M.R. Smith and S.A. Driscoll

363

Role of oxide catalysts basicity in selective oxidation E.A. Mamedov, V.P. Vislovskii, R.M. Talyshinskii and R.G. Rizayev

379

Strong evidence of synergetic effects between cobalt, iron and bismuth molybdates in propene oxidation to acrolein 0. Legendre, Ph. Jaeger and J.P. Brunelle

387

Classification of the roles of oxides as catalysts for selective oxidation of olefins L.T. Weng, P. Ruiz and B. Delmon

399

Section 5 NEW ASPECTS ON THE PREPARATION OF OXIDE CATALYSTS AND THE APPLICATION OF CHARACTERIZATION TECHNIQUES

Oxidation catalysts obtained by supporting molybdena on silica, alumina and titania C. Martin, M.J. Martin and V. Rives

4 15

Preparation and characterization of M/TiO, catalysts (M =Pt, Ru, Rh) using metal acetylacetonate complexes J.A. Navio, M. Macias, F.J. Marchena and C. Real

423

Oxidation catalysis: Electrophoretic study of Sn-Sb and Mo-Sb oxides P.J. Gil Llambias and M. Escudey

435

XI

New preparation method of ox-red catalysts via topological heterogenization of metallocomplexes B.V. Romanovsky and A.G. Gabrielov

443

New preparation methods of multicomponent oxide vanadium systems for oxidative dehydrogenation of alkanes, alkylaromatic and alkylheterocyclic compounds I.P. Belomestnykh, E.A. Skrigan, N.N. Rozhdestvenskaya and G.V. Isaguliants

453

Immobilized hemin catalyst in oxidation processes. 111. Oxidation of cysteine Yu.L. Zub, T.N. Yakubovich and G.P. Potapov

46 1

Author index

469

Subject index

47 1


XI11

ORGANIZATION Organized by Unit6 de Catalyse et Chirnie des MatCriaux DivisCs UniversitC Catholique de Louvain Place Croix du Sud, 2/17 B-1348, Louvain-la-Neuve (Belgium) (Prof. B. Delmon, Dr. P. Ruiz) Supported by UniversitC Catholique de Louvain Fonds National de la Recherche Scientifique, FNRS, Belgium

Commission of the European Communities Brite-Euram Programme Directorate General for Science, Research and Development International Advisory Board M. Baerns (Germany) B. Delmon (Belgium) J. Haber (Poland) G. Hecquet (France) R.A. Sheldon (The Netherlands) F. Trifiro’(Ita1y) Organizing Committee G. Centi (Bologna, Italy) P. Ruiz (Louvain-la-Neuve, Belgium) J.C. Volta (Villeurbanne, France) Sponsoring DSM Research (Geleen, The Netherlands) DOW Benelux, NV (Terneuzen, The Netherlands) UNICAT (Brussels, Belgium) REPSOL-PETROLEO SA (Madrid, Spain) ATO-CHEM (Paria, France)


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Sutface Science and Catalysis, Vol. 12, pp. 1-20 Q 1992 Elsevier Science Publishers B.V. All rights reserved.

1

NOBLE METAL CATALYZED OXIDATION OF CARBOHYDRATES AND CARBOHYDRATE DERIVATIVES

P. Vinke', D. de Wit, A.T.J.W. de Goede, and H. van Bekkum, Laboratory for Organic Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands. present address: Shell Research, P.O. Box 3003, 1003 AA Amsterdam, The Netherlands.

SUMMARY The oxidation of carbohydrates over noble metals provides an important route towards various interesting compounds. Besides some general information on noble metal catalyzed oxidations an overview of the current state of the art on carbohydrate oxidations is presented. The oxidation of several classes of carbohydrates and derivatives is described and discussed. Also, attention is paid to the mechanism of the formation of sideproducts during the reaction. INTRODUCTION The agricultural surpluses in Europe and the need for new crops have given a strong impulse to carbohydrate based research in the European Community. The growing interest is reflected in several national and European programs, sponsoring innovative and promising research projects in the carbohydrate field [l]. The research is focussing primarily on non-food applications of carbohydrates and the cultivation of new crops with a high yield in biomass and carbohydrate content. Examples combining the efforts in both fields are the inulin containing plants chicory and Jerusalem artichoke. Here, the growth of a new promising industrial harvest is studied as a supplement to the crop rotation scheme, whereas inulin (a 13(2-l)-polyfructoside, starting with a (1-a) bound glucose unit) is being explored as a raw material in several chemical and food applications [2]. Another advantage of the use of carbohydrates is their good biocompatibility. Many carbohydrate based products are or are expected to be biodegradable too [3]. Especially C, oxidized products with their natural counterparts alginate and pectinate, are known to be degraded microbially rather easily.

2

The actual use of carbohydrates as chemical feedstock is, however, rather limited so far, compared to the agricultural production. In Figure 1 the industrial production of the most important carbohydrate feedstocks -(purified) cellulose, sucrose and starch- is shown [4,5]. The total volume of 260 million tons per year, which includes food as well as non-food applications, is only 2% of the annual world carbohydrate formation. Cellulose is mainly utilized in the production of paper and board, but other applications are found in fibers (e.g. rayon) and in the manufacture ofhydrocolloids. Starch is used in thickeners, in the paper industry and as a source for hydrolysates like maltodextrins, glucose and high fructose corn syrups (HFCS) [6]. Alkyl polyglycosides (AF’G’s), made from glucose, form a new class of surface active agents [7]. Sucrose is used as raw material in biotechnology [8], although the sucrose fatty acid esters are gaining interest, as indicated by the large number of patents in this field [9]. Two classes of esters can be distinguished, i.e. the mono-esters, used as detergent or emulsifier and the polyesters (ca. 6 alkyl chains) applied as substitute fats. production

applications (XI

(Mtons) cellulose 6.4

starch

5.8

cellulose 12

food starch

24

sucrose 14

EC

sucrose 110

world

non-food 50

starch

non-food

0 food 98

sucrose

food

non-food 98

cellulose

Figure 1. Annual production of the 3 main carbohydrate feedstocks and their use divided in food and non-food applications. In many cases, chemical modification of the carbohydrates is required in order to obtain the desired chemical and physical properties. Several large scale modification techniques are employed at the moment. An important class of modified carbohydrates are the ethers like hydroxyethyl and carboxymethyl cellulose. Oxidation

is another important technique to adjust the product specifications (see for example the review by one of the authors [lo]). Up to now, mainly stochiometric oxidation procedures are used in industry. For example, starch can be oxidized with sodium hypochlorite [ 111 or periodate [ 121, yielding dicarboxy- or dialdehyde starch,

3

respectively. Oxidation with hydrogen peroxide, sometimes catalyzed by M”’, predominantly results in fragmentation of starch into lower molecular weight products [13]. Another commercial process is the production of oxalic acid from starch, cellulose, or molasses by oxidation with nitric acid [14,15]. Selective oxidation at the 6-position of 1,4-glucans like cellulose can be achieved by the use of N,O, [16]. Also a few examples can be mentioned for mono- and disaccharides: Gluconic acid is manufactured by enzymatic oxidation of glucose, whereas citric acid and lactic acid

Table 1. Chemical structure and bulk prices of some carbohydrate feedstockr and carbohydrate based products.

k

structure

applications

price ($/kg)

starch

food, paper additive, adhesives, drilling

0.60

cellulose

paper, fibers, hydrocolloids

0.75

alginate

gelling agent, thickener

12.00

pectinate

binder, emulsifier

6.00

sucrose

food, bio-feedstock

0.30

citric acid

detergent builder, metal cleaning, beverages

1.80

L-lactic acid

food, baking industry, pharmaceuticals

2.10

ascorbic acid

vitamin, antioxidant

12.75

oxalic acid

metal and fabric cleaning, dyeing

1.20

c-cwu

I

HO-C-CWH

b-COO”

COOH HO-C

I

I

0

1 I HO--F\/c\F

HOOc-c I HO-C

C-OH

F””” COOH

4

catalytic

advantages

stoichiometric

heterogeneous

- often highly

- easy separation of

-

disadvantages

selective suitable for specific reactions

- often byproducts (salts) - expensive

-

dissolved products catalyst recycling not difficult

- not applicable to -

solid substrates heat transfer problems may be encountered

I homogeneous - mild conditions - applicable to solid substrates

- good heat transfer - difficult separation of dissolved products - no continuous processing possible

are obtained by fermentation of e.g. sucrose [14]. In Table 1 the chemical structure together with the bulk prices in the USA (1990) of a number of carbohydrates and carbohydrate based products is shown [17].In some cases, for instance glycol cleavage oxidation, the use of stochiometric reagents is inevitable, because no catalytic process is available at the moment. On the other hand the use of stochiometric reagents for the oxidation of carbohydrates, or in other applications, has a number of disadvantages compared to catalytic processes [ 181. First of all, stochiometric reagents often give a large amount of byproducts in the form of salts, in some cases as much as the desired product (cf. hypochlorite oxidation of starch). In many cases it is in principle possible to regenerate the oxidant, for instance electrochemically, but often such regeneration seems economically not interesting. Apart from the waste problem, the oxidants used are sometimes expensive (e.g. periodate), in contrast to catalytic oxidation, where oxygen or hydrogen peroxide can be used. In the oxidation

of polymeric substrates like starch, which are generally insoluble in water, heterogeneous catalysts cannot be applied. In these cases the use of soluble or even better gaseous stochiometric reagents is advantageous, although homogeneous catalysts can in principle be applied too. In Table 2 some advantages and disadvantages of stochiometric and catalytic oxidation are summarized. Two catalyst systems seem to have much potential in the oxidation of carbohydrates. One is the metal ion catalyzed glycol cleavage oxidation with hydroperoxides. Metals like iron, titanium, vanadium, and tungsten might be used. The oxidation system is able to work homogeneously (Fe"', W"') as well as heterogeneously (Vv, Ti").

5

This paper will focus on the second catalyst system, namely the heterogeneous noble metal catalyzed oxidation, with molecular oxygen as the oxidant. Attention will be paid to the principles of the oxidation reaction as well as to a number of interesting applications.

NOBLE METAL CATALYSTS The first publication concerning the use of platinum as an oxidation catalysts dates back to 1845 [19] when Dobereiner described the oxidation of ethanol towards CO, and water. Later, noble metal oxidation catalysts appeared to be well suited for the conversion of primary hydroxyl or aldehyde groups towards carboxylic acids. In most cases the reaction proceeds selectively, the carbon backbone remaining intact. The reaction temperature can be kept low, which enables the application of thermolabile substrates like carbohydrates. Air oxygen is generally used, although other hydrogen acceptors like quinone have been applied also in order to study the reaction mechanism [20]. The oxidation can be described as an oxidative dehydrogenation, which implies that the substrate is dehydrogenated by the noble metal, followed by oxidation of the adsorbed hydrogen atoms [21]. This mechanism is also confirmed by the dehydrogenation of glucose towards gluconic acid at high pH in the absence of oxygen. In these experiments, described by De Wit et al. [22], gaseous hydrogen was formed under very mild conditions. In Scheme 1 the general reaction mechanism of the oxidative dehydrogenation is shown.

As to the first step of dehydrogenation of the substrate Dijkgraaf et al. [23] propose deprotonation of a hydroxyl group under the mild basic conditions (pH 8-10) applied during the reaction, followed by hydride transfer from the carbon atom towards the noble metal surface in a second step. The influence of the pH on the rate of reaction, found in the oxidation of many substrates (e.g. glucose, ethanol), supports this model. On the other hand, deprotonation of the substrate in aqueous solutions at pH 8-10 will take place to a very small extend, considering the pK, values of the systems (e.g. 12.4 for glucose and 13.3 for mannitol [24]). Moreover, we were able to perform alcohol oxidation at pH free sub-surface hydrogen adsorption site.

Unfortunately, noble metal catalysts are rather sensitive towards oxygen. In many cases the catalyst is poisoned when the oxygen concentration in the liquid phase is too high. The oxygen tolerance depends on the type of noble metal used and on several reaction conditions (for instance temperature and type of substrate). As shown by Van Dam et al. [25] the oxygen tolerance of the noble metals in the oxidation of methanol as model substrate for primary alcohols increases in the foIlowing order: Ru, Rh < Pd < Ir < Pt. Upon oxidation of the aromatic compound 5-hydroxymethylfurfural, however, all metals show a high tolerance towards oxygen [26]. The deactivation of the catalyst by oxygen proceeds in two or three stages. First the noble metal surface is occupied by chemisorbed oxygen with a surface coverage of

0.25 [27]. Although the oxygen is chemisorbed dissociatively, it is quite easy to reactivate the catalyst; In many cases, removal of oxygen from the gas phase is sufficient. After prolonged exposure to oxygen, the surface oxygen migrates into the noble metal lattice, thus forming a noble metal oxide layer. Ultimately, especially with highly dispersed catalysts, the noble metal particles are oxidized completely towards noble metal oxide crystallites. When this stage is reached, reactivation of the catalyst

7

can be performed only with strongly reducing agents like hydrogen gas or formaldehyde. In this paper we will discuss the work on noble metal catalyzed oxidations performed during the last few years by other research groups and in our laboratory. Three classes of carbohydrates and derivatives will be dealt with, namely: (i) Oxidation of monosaccharides. The oxidation of unprotected as well as of some

protected monosaccharides will be discussed. Main themes will be the selectivity of the oxidation with regard to the aldehyde or primary alcohol group, respectively, and side reactions due to C-C bond cleavage.

(ii) Oxidation of di- and oligosaccharides. Besides the unwanted side reaction of C-C bond cleavage also the problem of regioselectivity arises. In many cases several primary alcohol groups are present, which can exhibit differences in reactivity. Furthermore glycosidic bond cleavage towards smaller carbohydrate fragments can occur. (iii) Oxidation of 5-hydroxymethyl’&ral (HMF). HMF is formed by acid catalyzed dehydration of carbohydrates and is a potential key chemical in carbohydrate based chemistry. A number of compounds and polymeric materials can be prepared starting with HMF, provided the oxidation products are readily accessible.

OXIDATION OF MONOSACCHARIDES Three classes of monosaccarides can be distinguished: (i) the unprotected aldoses, of which the anomeric center is mainly in a cyclic acetal form. (ii) the (1-0)-protected monosaccharides where the anomeric center is ’blocked’ by an alkyl or phosphate group. In this case the aim is to oxidize the primary alcohol group at C6 selectively.

(iii) ketoses, which have a keto group at C, and in the case of hexoses possess two primary alcohol groups in the cyclic furanose forms and one in the pyranose forms. Unprotected sugars. Important work on the oxidation of glucose and other monosaccharides over platinum black catalysts was done by Heyns and coworkers and is reviewed in [28]. Heyns et al. were able to formulate the following sequence of oxidation reactivity on Pt for aldehyde and primary and secondary hydroxyl groups in cyclitols and carbohydrates:

R-CH=O > R-C-0-CHR’-OH > R-CH,-OH > RR’-CH-OH, > RR’-CH-OH,,

8

The noble metal catalyzed oxidation of glucose has been studied by a number a research groups. For a recent review on glucose oxidation, see Roper [29]. Especially variation of the catalyst showed to have a profound effect on the oxidation product composition. In Figure 2 some oxidation products of glucose are shown.

4

H o I O H

OH

OH OH OH

-1

3 -

-5

-6

Figure 2. Oxidation products of glucose obtained by noble metal catalyzed oxidations. I glucose, gluconic acid, 3 glucaric acid, 4 2-keto-gluconic acid, 5: guluronic acid, and 6 fructose

The oxidation towards gluconic acid can be performed with high selectivity and in high yields [30]. So, Hattori [31] found that Pd/C catalysts were very active and selective in this reaction. Apparently, they performed better than Pt/C catalysts. Promotion of the palladium catalyst with for example Se or Bi gave a further improvement in selectivity. Despeyrow et al. [32] describe the use of a 4% Pd, 1% Pt, 5% Bi on carbon catalyst with which they reach a selectivity of 98% towards gluconic acid at quantitative conversion. These high yields and selectivities seem to make the noble metal catalyzed oxidation route fully competitive with the biocatalytic process. In comparison with Pd, Pt catalysts show a lower selectivity for the oxidation reaction of glucose towards gluconic acid. This can be understood by the higher oxygen

9

tolerance of Pt. Oxidation of a primary alcohol proceeds much slower than oxidation of an aldehyde. Therefore, at the end of the oxidation of glucose towards gluconic acid, the noble metal surface will be covered with oxygen, causing a sharp decrease in activity for Pd. Platinum however, is still active in the 'oxidized' state and will be able to oxidize gluconic acid also. A second effect on the activity of the catalyst is the interaction between the substrate and the noble metal. In general, nonionic substrates will have a stronger interaction with the noble metal than anionic substrates. As a consequence, it will be easier to oxidize the primary alcohol of a neutral protected sugar than of gluconic acid. Apparently, Pd is able to oxidize primary alcohol groups of a neutral substrate, whereas Pt can oxidize these groups also in a monocarboxylate. Therefore, the oxidation of glucose or gluconic acid towards glucaric acid (3)is only possible with Pt catalysts [23]. The reaction over Pt proceeds with moderate selectivity (= 70%) due to the formation of byproducts like oxalic acid. During the reaction guluronic acid (5) is formed as an intermediate product in max. 30% yield [33]. Promotion of the noble metal catalyst with Pb can have a profound influence on the selectivity of the reaction. The oxidation of glucose or gluconic acid over Pt, Pb/C gives 2-ketogluconic acid (4) in good yields [34, 351. This change in selectivity is explained by assuming a strong bidentate interaction of the lead ions with the carboxylate and neighbouring hydroxyl group. Hydrogen abstraction is thought to be easier in this complex, leading to the 2-ketogluconate. Oxidation of fructose towards the 2-ketoacid with the Pt/Pb catalyst was unsuccessful, yielding oxalic acid as main product [34]. Apparently, selective oxidation of a primary alcohol in the presence of an a-keto group towards an a-ketoacid is very difficult, whereas oxidation of the

Q-

hydroxyacid towards the same product proceeds in good yields. Probably, the intermediate 2-ketoaldose in the fructose oxidation decomposes, yielding smaller hydroxyacids. Carbohydrates which were also tested with this catalyst system include galactose, mannose, and xylose [31]. In general, the results are comparable with those for glucose, showing that the configuration of the 2-hydroxyl group is not of major importance. Several oxidations shown in Figure 2 (e.g. towards 1 and 4) are possible via biochemical routes also [29]. A special case in carbohydrate oxidations is the Pt or Rh catalyzed dehydrogenation of glucose in the presence of fructose as hydrogen acceptor, which was found in our laboratory some years ago [36] (Figure 3). The reaction proceeds at p H > 12,

10

yielding gluconic acid and a mixture of mannitol and sorbitol. In this 'combi-process' there is a good balance between glucose dehydrogenation and fructose adsorption and reduction. Probably, dehydrogenation in the absence of oxygen only takes place when the substrate is deprotonated, comparing the optimum pH for this reaction with the pK, values of carbohydrates. Recently, these reactions were also performed using enzymes, making it even possible to produce sorbitol or mannitol selectively (371.

gluconic acid

glucose \

fructose

pH 13

+

sorbitol mannitol

Figure 3. Dehydrogenation of glucose towards gluconic acid with concurrent reduction of fructose to sorbitol and mannitol over Pt or Rh catalysts.

Protected sugars. The aim is oxidation of the C, primary alcohol group in protected sugars, while leaving the anomeric center intact. In the case of 2-0 protected ketoses in the furanose form even two primary alcohol groups can be oxidized. Here also, the oxidation of glucosides cover the main part of the literature. The oxidation of 1-0methyl glucopyranoside towards 1-0-methyl glucuronic acid is described by Easty [38]. In general the selectivities were moderate (= 70%), mainly due to total oxidation of the substrate towards bicarbonate. Schuurman et al. [39] have studied

Pt/C as the catalyst for this conversion. We have studied the oxidation of 1-0-all@ glucosides ( alkyl= C,, C,,,

and C12)

towards 1-0-alkyl glucuronides (Figure 4) over platinum and palladium catalysts [40]. These products have potential use as specialty anionic-nonionic surface active

agents by analogy with alcohol-ethercarboxylates. The oxidation proceeds with selectivities of ca. 85% at total conversion. Main byproducts include the corresponding alkanoic acid, bicarbonate, and 1-0-alkyl ketoglucuronic acids, which were identified using HPLC/MS analysis. Several catalyst systems were tested, of which the Pt catalysts appeared to be most selective. Carbon supported catalysts with high

11

dispersion (e.g. 0.40 for Pt/C) showed lower selectivities at the end of the reaction. It is not clear whether this decrease is caused by the noble metal dispersion or by the catalyst support. When using 5% Pd/Al,O, the formation of 1-0-alkyl ketoglucuronic acids is considerable (ca. 30%). The keto acids formed are stable during the reaction. Generally, they decompose towards smaller hydroxy acids, but due to the alkyl chain they remain intact. Ir appears to be also a selective catalyst in this oxidation, but had a low activity. Rh and Ru catalysts are inactive for this oxidation. Thiem et al. [41] reported the oxidation of 1-0-tetradecyl-a-glucopyranoside over Adams' catalyst (Pt black), although no yields and selectivities were stated. In a patent [42], claiming the same reaction with different alkyl chains and catalyst systems, yields ranging from 70 to 80% were reported.

-

P t , Pd HO&

0

w

O,,pH

9,60 "C

-

0

Figure 4. Oxidation of I -0-octyl-a-D-glucopyranosidetowards the corresponding glucuronic acid.

The oxidation of glucose 1-phosphate was studied in our laboratory by Van Dam [43]. Apart from selectivity problems, the Pt/C catalyst suffers from oxygen poisoning during reaction [44]. This deactivation was mainly due to the strong anionic character of the substrate, decreasing the rate of reaction. The problem was solved by applying 'diffusion stabilized catalysts' [45], which catalysts consist of activated carbon extrudates with a diameter of 0.8 mm, uniformly loaded with 5% Pt. The outer shell of these catalysts serves as diffusion barrier for oxygen and is inactive in the oxidation reaction. The core of the catalyst particles is only exposed to very low oxygen concentrations and remains active during the oxidation. In Figure 5 the reaction rate v is schematically shown as a function of the distance from the outer surface and the oxygen concentration inside a catalyst particle.

12

Figure 5. Correlation between local oxygen concentration in the liquid phase [ O j and the activity (v) of a noble metal catalyst in a large catalyst particle. R = radius of catalyst particle, r = distance from center of particle, C, = bulk concentration oxygen. The selectivity of the oxidation reaction could be improved by modification of the activated carbon carrier [46]. When the carbon carrier was oxidized with nitric acid, thus increasing the number of surface functional groups and the surface charge, the selectivity towards glucuronic acid 1-phosphate increased from 73% to 87%. This effect was attributed to the electrostatic interaction between the substrate and the support, retarding the consecutive reaction of the anionic product. Furthermore, the negative charge of the catalyst surface may direct the primary hydroxyl group towards the catalyst surface. The noble metal dispersion did not affect the selectivity of the reaction. Concluding this part, one can say that the oxidation of protected or unprotected monosaccharides is reasonably well investigated. Main problems are encountered in unwanted oxidation of the substrate towards small hydroxy acids and in catalyst deactivation. Work has to be done to achieve further enhancements in e.g. oxidation selectivity and catalyst stability.

OXIDATION OF DI- AND OLIGOSACCHARIDES As well as the monosaccharides, di- and oligosaccharides can be divided in reducing (unprotected aldoses) and non-reducing (protected or keto-) sugars. The oxidation behaviour of reducing disaccharides is similar to that of reducing monosaccharides as far as the C, position is concerned. As an example the oxidation reactions of lactose will be discussed. Examples of non-reducing systems which are dealt with here include the disaccharides sucrose and trehalose and the cyclic oligomer B-cyclodextrin.

13

Disaccharides. Hendriks et al. have used several catalyst systems in the oxidation of lactose towards different products. When using a commercial Pd/C catalyst, in situ promoted with Bi, lactobionic acid was obtained with 100% selectivity and 95% conversion [47]. Other aldoses could be oxidized with similar selectivities.

AMS/H,O, (anthraquinone monosulfonate) system, selective decarbonylation occurs, forming galacto-arabinonate [48]. Glucose and galactose also gave the corresponding C, sugar acid with this catalyst system. In Figure 6 the oxidation reactions of lactose are shown. With

the

Figure 6. Oxidation of lactose towards (a) lactobionic acid and (b) galacto-arabinonic acid. The oxidation of sucrose towards mono- and dicarboxylic acids is described in several patents. Recently, Fritsche-Lang et al. published the results of sucrose oxidation towards the tricarboxylic acid [49] over a 5% Pt/C catalyst. The maximum yield was 40% tricarboxylic acid, dicarboxylic acid and oxalic acid being the main byproducts. The reaction times were quite long (up to 84 h), probably due to catalyst deactivation

during reaction.

In our laboratory the platinum catalyzed oxidation of trehalose, a glucose dimer, a, acoupled by the anomeric centers, was explored. A few publications deal with the oxidation of trehalose, using e.g. perchlorate and Ce(1V) [50] or bromine [51] as the oxidant, resulting in glycol cleavage oxidation and keto group formation, respectively. The Pt catalyzed oxidation of trehalose at 60 "C and pH 9 was fast towards the C6-monocarboxylic acid, but the catalyst deactivated upon further

14

oxidation. The yield in monocarboxylic acid was about 60% as determined by HPLC. 13C NMR measurements revealed the formation of oxalic acid as a byproduct. In Figure 7 the reaction is shown.

Figure 7. Oxidation of trehalose towards the monocarboxylate.

During the oxidation of a-alkyl maltosides with alkyl chains ranging from 8 to 12 carbon atoms, several reaction products are observed. The main reaction is oxidation of a primary alcohol towards a carboxyl group. We found by HPLC/MS analysis that the c6 of the terminal glucose unit (C;) was oxidized exclusively (see Figure 8). Thiem et al. [41] also came to this conclusion, although no experimental proof was given. Selettive oxidation of the C,' hydroxyl can be understood by assuming oxidation of alkyl maltoside in a micelle structure, where the C,'-primary alcohol in the terminal glucose unit is accessible to the Pt surface, whereas the c6 in the first glucose unit cannot be reached by the noble metal. We did not observe oxidation towards the dicarboxylic acid, probably due to catalyst deactivation after oxidation of the first primary alcohol. The second reaction, occurring with alkyl maltosides, is oxidative splitting of the terminal glucose unit, yielding alkyl glucoside and several fragmentation products. The alkyl glucoside is oxidized further towards alkyl glucuronide. A tentative model for this oxidative splitting reaction is given in the next section on 8-CD oxidation. The different oxidation products, shown in Figure 8, were analyzed using a Novapak C18 reversed phase column. This column can be used for the separation of all alkyl mono- and disaccharides with alkyl chains ranging from c6 to C14 The eluent is a mixture of methanol and water, buffered at pH 3 with ammonium formate. MeOH/H20 ratio is 50/50 (v/v) for the shorter chains to 80/20 for the longer ones. In Figure 9 a chromatogram of the oxidation mixture of decyl maltoside is shown as an example.

15

fragmentation products

O

w

Figure 8. Oxidation of 0-octyl maltoside over a 5% Pt/A1203 catahst.

0

10

20

30

40

t (min) Figure 9. HPLC chromatogram of an oxidation product mixture of decyl maltoside. 1 inorganic salts and free sugars, 2 oxidized deql maltoside, 3 oxidized decyl glucoside, 4 decyl maltoside, 5 decyl glucoside and 6 internal standard.

16

Oligosaccharides. Upon oxidation of B-cyclodextrin (B-CD) at the C6 position, the systems obtained can be of interest as cation complexing agent, as drug carrier and as enzyme model. Excellent cation coordinating abilities were observed for B-CD subjected to glycol cleavage oxidation with periodate and chlorite/hydrogen peroxide [12] yielding a polycarboxylate. Casu et al. have mentioned the preparation of mono-oxidized B-CD [52] by oxidation over a Pt catalyst, without giving any characterization of the products. In our hands the oxidation of B-CD over different catalyst systems did not result in selective formation of the mono- or dicarboxylic acid [53]. A number of side products are formed, which could be partly identified. At the start of the reaction, two products were present in relatively large amounts. By comparison with the reaction mixture of the oxidation of maltoheptaose these compounds appeared to be linear maltohexaose (MD6) and C1-oxidized maltohexaose (MD6ox). Upon prolonged oxidation, smaller oxidized maltodextrins were found also. The formation of linear MD’s is not caused by a normal hydrolysis reaction of B-CD or oxidative splitting of a glycosidic bond, possibly catalyzed by the noble metal, because MD7 and MD7ox would be expected to be present then instead of or in addition to MD6 and MD6ox. Furthermore, a blank experiment, using 13-CD and a Pt catalyst under standard conditions, but without oxygen, did not show any formation of MD7. The selectivity of the oxidation of 0-CD is pH and temperature dependent. At higher pH values ( > 10) the formation of MD6ox and smaller MDox species increases. At lower temperatures the formation of MD6ox decreases, compared to oxidation of C6 primary alcohol. These results indicate that different reaction mechanisms are involved for primary alcohol oxidation on one hand and keto group formation, followed by ring opening and degradation on the other hand. In Figure 10 the major oxidation products of 13-CD are shown. The oxidation of B-CD was followed using a Dionex Carbopac column. The separation is based on ion chromatography in alkaline medium (pH 13), using a pulsed electrochemical detector. The system is able to separate mono-, di- and oligosaccharides and their derivatives up to DP 40, when a sodium acetate gradient is applied.

17

fragmentation products

+

M&COOH

n= 1-5

Figure 10. Main reaction pathways of B-CD oxidation over noble metal catalysts.

OXIDATION OF 5-HYDROXYMETHYLFURFURAL (HMF). HMF can be readily obtained by acid catalyzed dehydration of carbohydrates like fructose or inulin [54, 551. Although HMF itself is not used for large scale applications, its different oxidation products -shown in Scheme 2- can be applied for several chemical products. HFCA and FDCA can be used in the production of e.g. polyesters [56], whereas FDC can be applied in photochromic materials and conducting polymers [57]. The oxidation of HMF over different catalysts is described by several authors. HFCA can be obtained in high yields over a Ag,O/CuO catalyst using oxygen as the oxidant [58].

OH HMF

FDC

HFCA

OH FFCA

OH

OH

FDCA

Scheme 2. Oxidation products derived from HMF: 2,5-furandicarboxaldehyde (FDC), 5hydroxymethyl-2-furancarboxylic acid (HFCA), 5-formyl-2-furancarboxylicacid (FFCA) and 2,5-furandicarboxylicacid (FDCA).

18

-H,O

1 H,O

OH

O Q w a @

Scheme 3. Resonance structures and equilibrium hydration of HMF.

Another surprinsing characteristic of HMF is the ability to be oxidized by all noble metals, including Rh and Ru, without poisoning of the catalyst. This is thought to be due to the relatively strong adsorption of the aromatic furan nucleus, which interacts with the noble metal surface, thus counteracting the adsorption of oxygen onto the surface.

19

CONCLUSIONS A wide range of carbohydrates and derivatives can be oxidized using noble metal catalysts, Various reactions, like oxidation of aldehyde and primary alcohol groups can be performed with moderate to high selectivities. Although no large scale processes based on noble metal catalyzed liquid phase oxidation are operated at the moment, the increasing number of patents on this subject shows the industrial interest herein.

ACKNOWLEDGEMENTS We would like to thank Johnson Matthey for providing Pt salts and Suddeutsche Zucker A.G. for a sample of HMF. The authors gratefully acknowledge financial support by the Netherlands Organization for Scientific Research (NWO/SON) and by Unichema Chemie B.V.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18. 19. 20. 21. 22.

For example ECLAIR in the EC and IOP carbohydrates in The Netherlands. A. Fuchs, Starch 39 (1987) 335. H. Roper and H. Koch, Starch 42 (1990) 123. H.U. Woelk, Proc. symp. 'Towards a carbohydrate-based chemistry', 23-26 October 1989, Amiens. M. Yalpani and PA. Sandford, in Progress in biotechnology Vol. 3, M. Yalpani (ed.), Elsevier Science Publishers, Amsterdam, 1987. H.Koch and H. Roper, Slarch 40 (1988) 121. A.D. Urfer, A.H. Malik, H.L. Kickle, G.M. Howell and N.F. Borys, Proc. 2nd World Con& Defergenfs,A.R. Baldwin (Ed.), AM. Oil Chem. SOC.(1987) p. 268. H. Schiweck, K. Rapp and M. Vogel, Chern. Ind. (1988) 228. C.E. James, L. Hough and R. Khan, Prog. Chein. OR. Nut. Prod. 55 (1989) 117. H. van Bekkum, in Carbohydrates as Organic Raw Materials, F.W. Lichtenthaler (Ed.), VCH, Weinheim (1991) p. 289. M. Floor, A.P.G. Kieboom and H. van Bekkum, Starch 41 (1989) 348. M. Floor, 'Glycol cleavage oxidation of polysaccharides and model compounds', thesis Delft University of Technology (1989). H.S. Isbell and P. Czubarow, Carbohydr. Res. 203 (1990) 287. Kirk-Otlimer Encyclopedia of Chemical Technology, Vol. 16, 3rd Edition, John Wiley, New York (1982) p. 618. A. Sattar, D. Muhammad, M. Ashraf, SA. Khan and M.K. Bhatty, Pak. J. Sci. Znd. Res. 3 1 (1988) 745. S.D. Dimitrijevich, M. Tatarko, R.W. Gracy, C.B. Linsky and C. Olsen, Carbohydr. Res. 195 (1990) 247. Chemical Marketing Reportec January 1990. R.A. Sheldon, Proc. symp. 'Heterogeneous catalysis and tine chemicals', 2-5 October 1990, Poitiers, France. J.W. Dobereiner, Ann. 53 (1845) 145. H. Wieland, Ber. 45 (1912) 484. K. Heyns and H. Paulsen, Angew. Clwm. 69 (1957) 600. G. de Wit, J.J. de Vlieger, A.C. Kock-van Dalen, R. Heus, R. Laroy, A.J. van Hengstum, A.P.G. Kieboom and H. van Bekkum, Carbohydr. Res. 91 (1981) 125.

20 23. P.J.M. Dijkgraaf, 'Oxidation of glucose to glucaric acid by Pt/C catalysts', thesis Eindhoven University of Technology (1989).

24. IUPAC Chemical data series no. 23, E.P. Serjeant and B. Dempsey (ed.), Pergamon Press, Oxford (1979). 25. H.E. van Dam, L.J. Wisse and H. van Bekkum, Appl. Cutul. 61 (1990) 187. 26. P. Vinke, W. van der Poel and H. van Bekkum, Proc. symp. 'Heterogeneous catalysis and tine chemicals', 2-5 October 1990, Poitiers, France. 27. J.P. Hoare, 1.Electrochem. SOC. 132 (1985) 301. 28. K. Heyns and H. Paulsen, Adv. Curbohydr. Chem. 17 (1%2) 169. 29. H. Roper, in Curbohydrufes us Organic Raw Materials, F.W. Lichtenthaler (Ed.), VCH, Weinheim (1991) p. 267. 30. J.M.H. D i r k and H.S. van der Baan, 1. Culul. 67 (1981) 1. 31. K. Hattori, Jpn. Kokia JP 7840713 (1978) to Kawaken Fine Chemicals Ltd. 32. K. Deller, H. Krause, E. Peldszus and B. Despeyroux, German Patent DE 3823301 (1989) to Degussa A.G. 33. P.C.C Smits, B.F.M. Kuster, K. van der Wiele and H.S. van der Baan, Curbohydr. Res. 153 (1986) 227. 34. P.C.C. Smits, 'The selective catalytic oxidation of D-gluconic acid to 2-keto-D-gluconic acid or Dglucaric acid', thesis Eindhoven University of Technology (1984). 35. P.C.C. Smits, Eur. Pat. EP 8500063722 (1985) to AKZO corp. 36. A.J. van Hengstum, A.P.G. Kieboom and H. van Bekkum, Starch 36 (1984) 317. 37. K.D. Kulbe, I. Haug, HA. Scholze and K. Schmidt, Proc. int. congress 'Food and non-food applications of inulin and inulin-containig crops', 18-21 February 1991, Wageningen, The Netherlands. 38. D.B. Easty, I. 0%.Client. 27 (1962) 2102. 39. Y. Schuurman et al., this volume. 40. A.T.J.W. de Goede, P. Vinke, F. van Rantwijk and H. van Bekkum, manuscript in preparation. 41. Th. Bocker and J. Thiem, Tenside Surf. Def. 26 (1989) 318. 42. N. Ripke, J. Thiem and Th. Bocker, Eur. Pat. EP 0326673 (1988) to Hiils A.G. 43. H.E. van Dam, 'Carbon supported noble metal catalysts in the oxidation of glucose-1-phosphate and related alcohols', thesis Delft University of Technology (1989). 44. H.E. van Dam, A.P.G. Kieboom and H. van Bekkum, Appl. Cuful. 33 (1987) 361. 45. H.E. van Dam, P. Duijverman, A.P.G. Kieboom and H. van Bekkum, Appl. Cuful.33 (1987) 373. 46. H.E. van Dam, A.P.G. Kieboom and H. van Bekkum, Recl. Truv. Chim. Pays-Bus 108 (1989) 404. 47. H.E.J. Hendriks, B.F.M. Kuster and G.B. Marin, Curbohydr. Res. 204 (1990) 121. 48. H.E.J. Hendriks, B.F.M. Kuster and G.B. Marin, Curbohydr. Res., in press. 49. E.I. Leupold, M. Wiesner and W. Fritsche-Lang, Proc. Eurocarb V, Prague, 1989, p. D-1; W. Fritsche-Lang, E.I. Leupold and M. Schlingmann, German Patent DE-OS 3535720 to Hoechst A.G. 50. K.C. Nand, Cient. Cult. (Sao Paulo) 36 (1984) 442. 51. E. Scholander, Carbohydr. Res. 73 (1979) 302. 52. B. Casu, G. Scovenna, A.J. Cifonelli and A.S. Perlin, Curbohydr. Res. 63 (1978) 13. 53. P. Vinke, D. de Wit and H. van Bekkum, manuscript in preparation. 54. P. Vinke and H. van Bekkum, Proc. 3rd Seminar on Inulin, Wageningen, The Netherlands, March 1989 (published 1990), p. 55. 55. B.F.M. Kuster, Sfurch 42 (1990) 314. 56. H. Hirai, f. Mucroniol. Sci.-Chem. A21 (1984) 1165. 57. J. Daub, J. Salbeck, T. Knoechel, C. Fischer, H. Kunkely and K.M. Rapp, Angew. Chem. 101 (1989) 1541; Angew. Chem. I f i f . Ed. Eng. 28 (1989) 1494. 58. B.W. Lew, US Patent 3326944 (1967) to Atlas Chemical Industries, Inc. 59. P. Vinke, H.E. van Dam and H. van Bekkum, Sfud. Surf: Sci. Cutul. 55 (1990) 147. 60. E.I. Leupold, M. Wiesner, M. Schlingmann and K. Rapp, German Patent DE 3826073 (1988) to Hoechst A.G.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science and Catalysis, Vol. 72, pp. 21-31 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

21

ALKANE OXYGENATIONS BY H 2 0 2 ON TITANIUM SILICALITE

D.R.C. Huybrechts, Ph.L. Buskens and P.A. Jacobs K.U. Leuven, De t. Biotechnische Wetenschappen, Centrum voor Oppervlaktechemie en Katalyse, Karinaal Mercierlaan 92, B-3001 Heverlee (Leuven), Belgium ABSTRACT

The Titanium-Silicalite-1 (TS-1) catalyzed oxidation of n-hexane by aqueous H202 with formation of 2- and 3-hexanols and the corresponding hexanones is carried out in presence or absence of solvents, and in stirred and non stirred reaction systems. The observed conversions and ketone/alcohol ratios are explained by the relative rates of the different reaction steps: liquid to liquid diffusion, liquid to solid diffusion, intraporous diffusion and catalytic oxidation.

INTRODUCTION Titanium-Silicalite-1 (TS-1) is, like the aluminosilicate ZSM-5, a molecular sieve of the MFI structure type [l-21. Unlike the latter material TS-1 is virtually free of aluminium and contains titanium in concentrations up to 4 mole %, which converts the material from an acid to an oxidation catalyst [3-41. Using aqueous hydrogen peroxide as oxidant, TS-1 catalyzes the (mono)epoxidation of (di)olefins [5-61, the ring hydroxylation of aromatics [7], the dehydrogenation of primary and secondary alcohols and the ammoximation of ketones in the presence of NH3 [8]. More recently it was reported independently by Tominaga et al. [9] and by us [lo], that TS-1 is also an efficient catalyst for the oxygenation of paraffinic hydrocarbons to a mixture of secondary and/or tertiary alcohols and ketones. The reaction scheme consists of two consecutive steps: hydroxylation of a secondary or tertiary C-H bond in the alkane and dehydrogenation of the formed secondary alcohols to the corresponding ketones. In the oxygenation of branched alkanes, tertiary C-H bonds are selectively oxidized over secondary ones, while oxidation at primary C-H positions is not observed under typical reaction conditions. The oxidation thus follows the 'normal' reactivity order: tertiary CH > secondary C-H > > > primary C-H. In the oxygenation of n-alkanes, the hydroxylation step occurs with a slight regioselectivity for hydroxylation at the C2position compared to the more inner carbon positions. The subsequent dehydrogenation, however, occurs with a very pronounced substrate selectivity for 2-

22

alcohols compared to other secondary alcohols. These remarkable selectivities are ascribed to shape selective effects exerted by the TS-1 lattice. In the present study the conversions and selectivities in the oxygenation of nhexane by aqueous H202 are determined in solvent containing and solvent free, stirred and non stirred reaction systems. The influence of these experimental conditions on the features of the reaction are rationalized in view of the different mass transfer and catalytic steps involved in the reaction. EXPERIMENTAL

TS-I was synthesized according to the method described by Taramasso et al. [l]. Its IR spectrum contains a band at ? 960 cm-l, and no extrazeolitic crystalline or amorphous phases are detected by XRD or SEM. n-Hexane oxygenations are performed at 100°C in a stainless steal batch reactor with a volume of 300 ml (PARR Instr. Corp.), previously flushed with nitrogen. After 1 hour reaction time, the reaction mixture is homogenized with an excess of acetone. Organic reaction products are analyzed by GC on a 50 m CPSil-88 capillary column (Chrompack), using toluene as internal standard. Within the accuracy of the GC analysis, yields of hexanols and hexanones add up to 100% on a carbon basis. H202 conversions are determined by a potentiometric titration with Ce(S04)2*4H20. DESCRIPTION OF THE TRIPHASIC REACTION SYSTEM

The reaction medium for the oxygenation of alkanes by aqueous H202 on TS-1 consists of three phases: an organic liquid phase (Lo), containing most of the alkane, an aqueous liquid phase (La), containing most of the H202, and the solid catalyst (S). In the absence of a solvent, the two liquid phases are almost completely insoluble in one another. The addition of a polar organic solvent, such as acetone, which is divided over the two phases, improves the mutual solubility, but does not result in the formation of a homogeneous liquid phase under typical reaction conditions. When the reaction mixture is vigorously stirred, an apparently homogeneous emulsion is obtained, which segregates very rapidly into two liquid phases when the agitation ceases. Segregation occurs by formation of organic bubbles in the emulsion which move upwards to form the Lo phase, indicating that the emulsion consists of dispersed particles of the Lo in the La phase. After (and during) the segregation of the two liquid phases, the catalyst is present as a relatively stable suspension in the La

23

phase (or emulsion), which needs several hours for complete precipitation. The Lo phase on the other hand is perfectly transparent and thus free from catalyst. Based on these observations, the reaction media for alkane oxygenation by aqueous H202 on TS-1 are schematically represented in Figure 1, both in the absence and presence of an organic polar solvent and in the absence and presence of mechanical agitation. In these representations the Lo phase contains most of the alkane and (in cases C and D) part of the solvent, whereas the La phase contains most of the H202 and H20, suspended TS-1 particles and (in cases C and D) the remaining part of the solvent.

Fig. 1: Schematic representation of the reaction media for alkane oxygenation by aqueous H202 on TS-1 in the absence of solvent and without mechanical agitation (A); in the absence of solvent and with mechanical agitation (B); in the presence of solvent and without mechanical agitation (C); in the presence of solvent and with mechanical agitation (D). : Lo phase;EZB La phase.

Due to the triphasic reactions conditions, the overall reaction between H202 and an alkane on TS-1 requires different transfer processes next to the catalytic reaction. The following steps, which are schematically represented in Figure 2, are involved: 123456-

78-

transfer of H202 from the aqueous phase La to the external surface of TS-1; transfer of H202 inside the pore volume of TS-1; transfer of the alkane from the organic phase Lo to the interphase between Lo and La; transfer of the alkane from the interphase to the aqueous phase La; mixing and diffusion of the alkane in the La phase; transfer of the alkane from the La phase to the external surface of TS-1; transfer of the alkane inside the pore volume of TS-1; catalytic reaction (adsorption, oxygen transfer, desorption).

24

Obviously, opposite transfer processes must be considered for the reaction products: H 2 0 is transferred from the catalyst to the La phase, and the organic reaction products (alcohols and ketones) will be divided over the La and Lo phase according to their partition coefficients. t I t I I t I

! I I t 1 t

I

! t I t

La

S

Fig. 2: Reaction steps involved in the oxygenation of alkanes by aqueous H202 on TS1.

The overall kinetics of the alkane oxygenation are determined by the physical kinetics of the different transfer processes and by the kinetics of the catalytic reaction itself. The liquid to liquid transfer rate (steps 3 to 5) will increase with increasing interface area between Lo and La, and with increasing solubility of the alkane in the La phase. Therefore, this reaction step is expected to be enhanced by mechanical agitation and by addition of a polar solvent. Its rate is, however, independent of the catalyst concentration. The rate of all other reaction steps will increase with increasing catalyst concentration: the liquid to solid transfer processes (steps 1 and 6) are enhanced by increasing external surface of the catalyst, and the intraporous transfer and catalytic processes (steps 2, 7 and 8) are enhanced by increasing catalyst mass. Therefore, the presence of liquid to liquid diffusion limitation can be ruled out if the reaction rate is dependant on the catalyst concentration. Furthermore, the rate of all external transfer phenomena (steps 1 and 3 to 6) is expected to be enhanced by mechanical agitation of the reaction mixture. This is not the case for the rates of the intraporous transfer steps and of the catalytic reaction (steps 2,7 and 8). If no dependance of oxygenation rate on agitation rate is observed, it can be concluded that no rate limitation by external transfer processes occurs.

25

RESULTS

The oxygenation of n-hexane by aqueous H 2 0 2 on TS-1 was performed in the four reaction systems depicted in Figure 1. The composition of the reaction mixtures and the agitation rates are listed in Table 1. Table 1: Reaction mixture composition and agitation rate in the oxygenation of nhexane by H202 (35% in H20) on TS-1. System

A B C D

n-hexane (mmole)

115 115 115 115

H202 (mmole)

acetone (ml)

agitation rate (RPM)

240 240 240 240

0 0 45 45

0 1000 0 1000

The influence of catalyst concentration on the conversion observed after 1 hour reaction time was investigated for each system. The influence of the agitation rate and of the nature and concentration of the solvent was examined for system D. Due to the presence of the two liquid phases, it was experimentally impossible to take representative samples of the reaction mixture during reaction. Therefore, no kinetic data could be collected and the n-hexane conversions observed after 1 hour reaction time were taken as semi-quantitative measures of the oxygenation rates. This approximation seems acceptable in view of the monotonic increase of the n-hexane conversion against reaction time up to conversions of about 70-80% [ 111. For system A, the n-hexane conversions observed in the presence of 0.5 and 1 g TS-1 after 1 hour reaction at 100°C are less than 1%. The n-hexane conversions obtained after 1 hour of reaction in systems B, C and D, are plotted against the amount of TS-1 in Figures 3 , 4 and 5, respectively.

For system B, the observed n-hexane conversion increases in a less than proportional way with increasing concentration of TS-I, as is shown in Figure 3 by the deviation between the dotted line, which represents a speculative linear correlation at low conversions, and the experimental points. For the acetone containing systems C and D on the other hand, a stronger correlation between n-hexane conversion and catalyst concentration is observed. The n-hexane conversion levels off at high catalyst concentrations in system D. At this moment, the reagents are exhausted as is also observed when lower catalyst concentrations are used combined with long reaction times [ll].

26 100

-

2 hexanone 80 n

8

W

d 0

60

.rl

u1

ec

*d

4)

40

6 20

0 0

250

500

7 50

1000

Amount of TS-1 (mg) Fig.3: Conversion in the oxygenation of n-hexane by H 0 2 (35% in H20) on TS-1 after 1 hour at 100°C against amount of TS-1, reaction syskrn B. 100

80 h

8

W

d 0

60

*rl

u1

ec

40

U

6 20

0 0

250

500

750

1000

Amount of TS-1 (mg) Fig.4: Conversion in the oxygenation of n-hexane by H 0 2 (35% in H20) on TS-1 after 1 hour at 100°C against amount of TS-1, reaction syshm C.

21

0

250

500

750

1000

Amount of TS-1 (mg) Fig. 5: Conversion in the oxygenation of n-hexane by H 0 2 (35% in H20) on TS-1 after 1 hour at 100°C against amount of TS-1, reaction syskm D.

Comparison of the n-hexane conversions measured in systems A and C with those of systems B and D, respectively, shows that mechanical agitation improves the oxygenation rate. This effect is more pronounced in the absence than in the presence of solvent. The hexanone/hexanol ratio of the product mixtures of all reaction systems increases with increasing n-hexane conversion, confirming that formation of hexanols and hexanones occurs in two consecutive processes. However, in the solvent free reaction systems A and B the ratios are systematically higher than in the acetone containing systems C and D. The conversions and product selectivities for the n-hexane oxidation reaction in system D at different agitation rates and in the presence of different solvents are summarized in Tables 2 and 3, respectively. It is again seen that the application of mechanical agitation, which corresponds to the transition from reaction system C to D, has a positive influence on the n-hexane conversion (Table 2). Neither the n-hexane conversion, nor the product selectivities are however significantly influenced when the agitation rate is varied within the range from 500 to 1000 RPM. Table 3 shows that the n-hexane conversion is of comparable magnitude in the presence of acetone, methanol or t-butanol as solvent. When methanol is used as a solvent, the formation of

28

formaldehyde by methanol oxidation is below detection limits, indicating that oxidation of methanol is much less rapid than that of n-hexane and of 2- and 3-hexanol. This is in agreement with the reported low reactivity of methanol compared to other alcohols [4]. Table2: ratea. Agitation rate (RPM)

0 500 700 1000

n-Hexane oxygenation by aqueous H202 on TS-1 against the agitation n-hexane conv. (%) 34 58 59 57

3-hexanone 19 14 15 14

Product selectivity (%) 2-hexanone 3-hexanol 49 51 49 54

19 23 24 22

2-hexanol 13 12 12 10

a; Reaction conditions: 500 mg of TS-1, 115 mmole of n-hexane, 240 mmole of H 2 0 2 (35% in HzO), 45 ml of acetone, 100°C, 1 hr. Table 3: solventa. Solvent

acetone methanol t.-butanol

n-Hexane oxygenation by aqueous H202 on TS-1 against nature of the n-hexane conv. (%) 63 57 58

3-hexanone 15 18 15

Product selectivity (%) 2-hexanone 3-hexanol 48 48 50

25 22 24

2-hexanol 12 12 11

a; Reaction conditions: 500 mg of TS-1, 115 mmole of n-hexane, 240 rnmole of H202 (35% in H20), 45 ml of solvent, 100°C, 700 RPM. The product selectivities obtained in n-hexane oxygenation by H 2 0 2 on TS-1 in the presence of different amounts of acetone are listed in Table 4. At low acetone concentrations, the n-hexane conversion increases substantially with increasing acetone concentration. It reaches an optimum at about 20 ml of acetone, and decreases slowly when the acetone concentration is further increased. The hexanone/hexanol ratio of the oxygenation products decreases continuously as the amount of acetone is increased. For acetone concentrations above 20 ml, this effect is in line with the decreasing nhexane conversions, which are expected to be accompanied by a decrease of the hexanone/hexanol ratios, since formation of hexanones and hexanols are consecutive reactions. For the lower acetone concentrations, however, the correlation between hexanone/hexanol ratios and n-hexane conversions is opposite to that obtained at high concentrations.

29

n-Hexane oxygenation by aqueous H202 on TS-1 against acetone Table4 concentration. amount of acetone (ml) 0 10 20 30 45 60 90

n-hexane conv. (%) 40 61 68 63 63 58 49

3-hexanone

42 34 28 20 15 11 12

Product selectivity (%) 2-hexanone 3-hexanol 44 53 56 56 48 47 49

10 9 12 18 25 27 24

2-hexanol 4 3 4 6 12 15 15

a; Reaction conditions: 500 mg of TS-1, 115 mmole of n-hexane, 240 mmole of H202 (35% in H20), 100°C, 700 RPM, 1 hr. DISCUSSION

The observation of a better correlation between n-hexane conversion and TS-1 concentration in the solvent containing reaction systems C and D (Figure 5 ) compared to that found in the solvent free reaction systems A and B (Figure 3), suggests that in the former cases, the overall n-hexane oxygenation is rate limited by liquid to liquid phase transfer phenomena, which become independent on catalyst concentration as discussed above. The slow, rate limiting transfer of n-hexane to the aqueous phase, is related to the low solubility of n-hexane in the aqueous phase, which can be considered as the driving force for the phase transfer. When the solvent free reaction mixture is not mechanically agitated (system A), virtually no reaction occurs, indicating that the rate of the liquid to liquid phase transfer is extremely low in this case. However, when the solvent free reaction mixture is vigorously stirred (system B), the n-hexane phase is emulsified in the aqueous phase, and the oxygenation rate increases considerably. This may be due to an increase of the liquid to liquid transfer rate by the increased interphase area of the two liquid phases, or even to direct diffusion of n-hexane from the emulsified phase to the TS-1 particles. The latter ones are suspended in the aqueous phase, and may contact the n-hexane particles under vigorous stirring. The low solubility of n-hexane in the aqueous phase is also responsible for the high hexanone/hexanol ratios which are observed in the solvent free reaction system B. Indeed, due to the low n-hexane concentration in the aqueous phase and thus in the catalyst pores, oxidation of formed hexanols will be highly competitive with n-hexane oxidation, and the majority of formed hexanols will be further converted to hexanones.

30

The acetone containing systems C and D on the other hand seem to be free from liquid to liquid diffusion limitations, as is evidenced by the strong correlation between the observed n-hexane conversions and the TS-1 concentration (Figures 4 and 5). The beneficial effect of acetone on the liquid to liquid transfer rate is related to the increase of the n-hexane solubility in the aqueous phase. Addition of acetone to the reaction mixture results, however, also in a higher degree of dilution of n-hexane and H202, which has a negative effect on the overall reaction rate. The presence of these two opposite effects explains the observation of an optimum acetone concentration for n-hexane oxygenation (Table 4). The observed increase of the hexanone/hexanol ratios upon addition of acetone is due to the increased n-hexane concentration in the aqueous phase and thus in the catalyst pores, which renders oxidation of n-hexane more competitive with hexanol oxidation. The successful replacement of acetone by other polar solvents such as methanol or tertiary butanol (Table 3) confirms that the role of the solvent mainly consists of facilitating the physical transfer phenomena rather than intervening directly in the catalytic reaction. As in the absence of solvent, the application of mechanical agitation to the solvent containing reaction mixture has a positive effect on the observed n-hexane conversions (Table 2). This indicates that external liquid to solid diffusion must be rate limiting when the reaction mixture is not stirred. Increase of the stirring rate beyond 500 RPM has, however, no additional positive effect on the conversion, which evidences that in the presence of acetone and under vigorous mechanical agitation, the oxygenation of n-hexane is not rate limited by external diffusion (liquid/liquid or liquid/solid) processes, but either by intraporous diffusion or by the catalytic reaction itself. The decrease of oxygenation rate of alkanes with increasing bulkiness [9-101, is in agreement with a reaction scheme which is rate controlled by intraporous diffusion rather than by intrinsic catalytic kinetics. CONCLUSIONS

The alkane oxidation by aqueous H202 on TS-1 occurs under liquid/liquid/solid three phasic conditions. In the absence of a polar solvent, mass transfer of the alkane from the organic liquid phase to the aqueous liquid phase, in which the catalyst is suspended, seems to be the rate limiting step of the overall reaction. This mass transfer is extremely slow, unless the reaction mixture is vigorously stirred so that an emulsion of the alkane in the aqueous phase is obtained. Under these conditions, the ketone/alcohol ratio of the oxygenation products is high even at low conversions. The addition of a polar solvent, such as acetone, methanol or tertiary butanol to the reaction system of the alkane oxygenation improves the liquid to liquid phase transfer, which results in an increased oxygenation rate and a decreased ketone/alcohol ratio.

31

High solvent concentrations have, however, a negative effect on oxygenation rates since they cause an increased dilution of the substrates in the reaction mixture. In a non stirred solvent containing reaction mixture, the apparent kinetics of the alkane oxygenation are limited by liquid to solid mass transfer processes. Under mechanical agitation, however, the external diffusion is accelerated and intraporous phenomena become rate limiting. This is in agreement with the observed decrease of oxygenation rate with increasing bulkiness of the alkane, which points to a rate limiting intraporous diffusion. ACKNOWLEDGEMENTS

D.R.C.H. is grateful to the National Fund for Scientific Research (Belgium) (NFWO) for a grant as Research Assistant. The authors acknowledge support from the ministery of Science Policy for a grant in the frame of a concerted action on catalysis and the NFWO for sponsoring. REFERENCES

M. Taramasso, G. Perego and B. Notari, U.S. Pat. 4,410,501 (1983), example 2. G. Pere 0, G. Bellussi, C. Corno, M. Taramasso, F. Buonomo and A. Esposito, Stud. Surf Sci. Catal. (1986) 129. 3. B. Notari, Stud. Surf, Sci. Catal. 37 (1987) 413. 4. U. Romano, A. Esposito, F. Maspero, C. Neri and M.G. Clerici, Stud. Surf. Sci. Catal. 55 (1990) 33. C. Neri, B. Anfossi, A. Esposito and F. Buonomo, Eur. Pat. 0 100 119 (1984). 5. 6. F. Maspero and U. Romano, Eur. Pat. 0 190 609 (1986). 7. A. Esposito, M. Taramasso and C. Neri, Deutsches Pat. 31 35 559 (1982). 8. P. Roffia, M. Padovan, E. Moretti and G. De Alberti, Eur. Pat. 0 208 311 (1987). 9. T. Tatsumi, M. Nakamura, S. Negishi and H. Tominaga, J. Chem. SOC.,Chem. Commun. (1990) 476. 10. D.R.C. Huybrechts, L. De Bruycker and P.A. Jacobs, Nature 345 (1990) 240. 11. R.F. Parton, D.R.C. Huybrechts, Ph. Buskens and P.A. Jacobs, to be published in Stud. Surf. Sci. Catal. 1. 2.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science and Catalysis, Vol. 72, pp. 33-41 @ 1992 Elsevier

33

Science Publishers B.V. All rights reserved.

Selective Oxidation of Hydrogen to Hydrogen Peroxide L. Fu', K.T. Chuang", and R. Fiedorow' I

Department of Chemical Engineering, University of Alberta, Edmonton, Alberta, Canada. T6G 2G6 2

Department of Chemistry, A. Mickiewicz University, 60-780 Poznan, Poland.

*

Author to whom the correspondence should be addressed.

Abstract The synthesis of hydrogen peroxide (H202) using H,-O, was studied at temperatures from -10°C to 25°C and pressures from 0.50 MPa to 2.3 MPa. Active catalysts included platinum group metals attached to hydrophobic carbon, supported on a mixture of hydrophobic and hydrophilic materials. Only palladium catalysts supported on a hydrophobic component were found to be selective towards hydrogen peroxide formation. The ratio of hydrophobic-to-hydrophiliccatalyst properties was important in determining the H202 yield. The effects of metal particle size and surface area appeared not to be important in determining H,O, yield. A simple way of characterizing catalyst hydrophobicity is proposed. Test results indicate that higher pressure, lower temperature and the addition of a stabilizer all have a beneficial effect on the final H202concentration.

1. INTRODUCTION Research on hydrogen peroxide (H202) production processes can be divided into two categories: to make incremental improvement on the existing technology (the autoxidation of an anthraquinone) and to develop a process based on the direct combination of hydrogen and oxygen in the presence of a catalyst. The latter offers considerable potential, as several patents indicate [2-41. In the patents, hydrogen and oxygen are reacted over noble metal catalysts in an acidic aqueous solution, some with an additional organic solvent. To date, the catalyst activity is too low for the process to become commercially viable. Because hydrogen and oxygen are only sparsely soluble in aqueous solutions. Thus the rate of reaction is limited by the mass transfer of reactants to the catalyst sites. The combination of low solubility and low liquid-phase diffusion coefficients of the reactants results in low apparent catalyst activities. Noble metals are the most active catalysts for H,-0, reactions at ambient temperature. If the reaction is carried out in the vapour phase, water is normally the only product. If the

34

reaction is carried out in acidic aqueous solutions, there is a change in selectivity; i.e., the reaction produces mainly H,O, instead of H,O. To overcome the mass transfer limitations, a hydrophobic catalyst was prepared by depositing a noble metal on a hydrophobic carbon. The resulting catalyst was then loaded into an autoclave reactor, where hydrogen and oxygen were introduced. Hydrophobicity prevents the metal catalyst from becoming covered permanently by the liquid solution, the reaction can thus proceed through gas-phase diffusion, and the product H,O, can be quickly transferred into liquid. 2. EXPERIMENTAL

In the direct synthesis of H,O, several reactions may occur simultaneously in the reactor: H2 + 0 2 --> H202

2H2 + O2 --> 2H20

+ H2 --> 2H20 2H202 --> 2H20 + O2

H202

Therefore, the catalyst should possess high selectivity as well as high activity. High H,O, concentrations in the reactor are also important, to avoid excessive downstream concentration costs. The following definitions are used throughout the text: (a) catalyst activity: the fraction of hydrogen consumed in the reactor. (b) catalyst selectivity: the fraction of consumed hydrogen resulting in the formation of H,O,. (c) product yield: the weight percentage of H,O, in the product solution. 2.1. Catalysts Palladium nitrate (Aldrich) dissolved in anhydrous methanol (Molincrockdt, AR) was used to prepare fluorinated carbon-supported catalysts and the solution of palladium nitrate in 5% HN0,-for impregnation of amorphous fumed silica Cab-0-Sil (Cabot, grade HS-5) and Silicalite S-115 (Union Carbide, Linde Division). Surface areas of the supports are given in Table 3. The palladium loading on the above supports, except for a few cases indicated in the text, was 5 wt%. Hydrophobic catalyst supports were prepared by fluorinating activated carbon to contain various amounts of fluorine. The higher the fluorine content, the higher the hydrophobicity of the support. The suspensions of supports in impregnating solutions were dried with an infrared lamp in a rotary evaporator and subsequently reduced in a H,/N,

stream at 300°C for 16 hours. Reduced catalysts were mixed with other components [e.g., Cab-0-Sil and Teflon suspension] dried at 7OoC for 24 h, cured in flowing nitrogen at 360°C for 10 min. and pulverized in an ultra-centrifugal mill (Retsch, Germany). The catalysts were labelled as follows: 5 Pdlcarbon 65, 5 Pdlcarbon 8, 5 Pdlcarbon 5, and 5 Pdlcarbon 3; they contained 5 wt% Pd on fluorinated carbon of 65, 8.2, 4.8 and 3.1 wt% F, respectively.

35

Catalysts containing 5 wt% Pd on Cab-0-Sil and 5 wt% Pd on silicalite were designated as 5 Pd/Cab and 5 Pd/Sil, respectively. Catalyst 5 Pd/carbon 65+5 Pd/Sil was prepared by mixing the two components in a ratio of 1:l by weight and adding 15 m L of Teflon suspension; catalyst 5 Pd/carbon 65+5 Pd/Cab was prepared in the same way, but 5 wt% Pd on Cab-0-Sil was used instead of 5 wt% Pd/silicalite. Catalyst 5 Pd/carbon 65+Cab 1.0 was made by mixing 5 wt% Pd/carbon 65 with Cab-0-Sil containing no palladium in a 1:l ratio; catalyst 5 Pd/carbon 65+Cab 0.5 was made by mixing the two components in a 2:l ratio. Increasing the silica content in the last catalyst by 25% made a catalyst designated 5 Pd/carbon 65+Cab 0.75. Mixing 0.5 wt% Pd+0.15 wt% Pd/carbon 65 with pure Cab-0-Sil (weight ratio of 1:l) made a catalyst designated 0.5 Pt+0.15 Pd/carbon 65+Cab. Teflon suspension was added to all mixed catalysts mentioned above. Moreover, a catalyst prepared by adding Teflon suspension to 2.5 wt% Pd/carbon 65, labelled 2.5 Pd/carbon 65+Teflon, was also used in the study.

2.2. Methods A stirred Parr autoclave of 450 mL capacity was loaded with 50 m L of 10% H,SO, and 1 g of catalyst. The acid concentration was chosen to obtain reproducible results. The flow rate of feed mixture (H2-02) was 250 mL/min. To avoid the risk of explosion, the hydrogen concentration in the feed gas was kept at 4.4 ~ 0 1 % The . gas mixture at the reactor outlet was analyzed at room temperature in a column packed with Porapak QS, using a Hewlett-Packard 5710A gas chromatograph. An OMNISORP 360 sorptometer [l] was used to measure surface areas and pore volumes at liquid nitrogen temperature. A Philips X-ray diffractometer with a copper tube and graphite monochromator, operated in the step-scan mode (0.02" 20 per step and counting for 100 s/step) was used to obtain XRD patterns. The scanning was performed from 36 to 44" 28 to cover the range around Pd(ll1) line. The proportional detector was interfaced with a computer that printed the XRD results. With the computer, it was easy to perform subtraction, integration and plotting of XRD patterns. Infrared spectra of carbon samples, diluted with KJ3r in the weight ratio of 1:100, were recorded on a Nicolet 740SX FTIR spectrometer. The resolution was 4 cm-I. 1600 scans were collected and averaged to get a single low-noise spectrum.

3. RESULTS AND DISCUSSION 3.1. Catalysts supported on fluorinated carbon In the patent literature [2-41 activated carbon is used as a support for hydrogen peroxide synthesis. However, non-modified activated carbon would not be a good choice, because the presence of oxygen-containing groups results in a wettable surface [5,6]. An activated carbon, therefore, adsorbs large quantities of water in the pores and the rate of H,O, decomposition on these catalysts is high. An attempt was made to employ fluorinated carbon (65% F)as a support. This carbon, distinguished by very high hydrophobicity, floats on the surface of the aqueous reaction medium. Because of the poor contact with the solution, the palladium catalyst deposited on this support showed high activity for H2 conversion but very little selectivity and thus low H,O, concentrations (see Table 1). Carbons with lower fluorine

36

contents were then prepared. The fluorine concentration can be characterized by a broad band of C-F stretching vibrations (Figure 1) at about 1220 cm-' [7]. Thus infrared spectroscopy seems to be a good tool for quick evaluation of hydrophobicity of fluorinated carbon supports.

Table 1 Activity of fluorinated carbon-supported palladium catalysts for hydrogen peroxide synthesis at 25°C under 1.34 MPa pressure Catalyst

5 Pdfcarbon 65 5 Pdfcarbon 8 5 Pdfcarbon 5 5 Pdfcarbon 3

F content, wt% in support 65.0

8.2 4.s 3.1

H, Conversion H,O, concentration, %

wt% *

82 40

0.05 0.20 0.02 0.005

33 29

Concentration of H,O, in post reaction mixture after 12 h of experiment. Although the use of palladium supported on fluorinated carbon resulted in low H,O, yields, a clear trend is visible in their behavior: namely, there is a certain fluorine content at which catalytic activity for H,O, formation passes through a maximum (Table 1). On the other hand, H, conversion increases with increasing F content in the support. The high catalyst activity at high F content indicates a reduced mass transfer resistance for H, and 0, to reach the catalyst sites on hydrophobic supports. However, the products, H,O,, is not very volatile, thus it is easier for it to move away from the catalyst sites via liquid rather than gas route. If the catalyst is too hydrophobic, the H,O, formed on the catalyst may be converted to H,O through reactions(3) and (4). It may he concluded that an optimum hydrophobicity of a support is required to obtain an efficient catalyst. A series of catalysts was prepared by mixing carbon 65 with a hydrophilic component to achieve a desired hydrophobic-tohydrophilic ratio.

3.2. Carbon-supported palladium mixed with other materials The first set of mixed-support catalysts consisted of platinum group metal deposited on fluorinated carbon (10 wt% of a metal on carbon 65). A thin layer of the above mixture was mounted onto 114'' (6 mm) ceramic rings, resulting in 0.2 wt% of metal on the combined support. Catalysts containing 0.2% Pd, 0.2% Pt, 0.2% Ru and 0.15% Pd + 0.05% Pt were tested, all of them showed poor selectivity toward H,O, formation. Large bubbles of gas were observed at all catalyst surfaces. Another form of combined-support catalyst was prepared for further experiments. It consisted of 5% Pd on carbon 65, which was then mixed with hydrophilic support, which contained Teflon but no palladium. Data presented in Table 2 indicate that the catalyst corresponding to the carbon 65:silica ratio of 2:l has low activity, producing only 0.01%

37

H,O,. Increasing the content of the hydrophilic component raises the activity to 0.016 and 0.70% H,O,. Also, the use of the catalyst prepared on the base of carbon 65 and Teflon results in a poor H,O, yield, most likely because of excessive hydrophobicity (it contains no silica) and small surface area (Table 3) of the Teflon-coated carbon particles. All highly hydrophobic and moderately hydrophobic catalysts float on the surface of an aqueous reaction medium (Table 3), but only the latter catalysts (e.g., 5 Pdlcarbon 65+Cab 1.0) show satisfactory catalyst activity. Thus, no prediction of catalyst behavior can be made on the grounds that a catalyst floats. If, however, catalyst particles sink, one can be sure that the catalyst will be inactive for H,O, synthesis.

Table 2 Hydrogen peroxide synthesis on mixed-support palladium catalysts at 2S0C under 1.34 MPa pressure

H202 concentration, wt%

Catalyst

5 Pdlcarbon 65 + Cab 0.5 5 Pdlcarbon 65 + Cab 0.75 5 Pdlcarbon 65 + Cab 1.0 2.5 Pdlcarbon 65 + Teflon

*

*

0.01 0.16 0.70 0.03

After 12 h of experiment

Table 3 Characterization of supports and selected catalysts Support or

Surface area

catalyst

m2/g

Carbon 65 Carbon 5 Silicalite Cab-0-Sil 5 Pdlcarbon 65 + Cab 1.0 2.5 Pdlcarbon 65 + Teflon 5 Pdlcarbon 65 + 5 Pd/Cab

419 208 380 325 27 5 95

* Pores above 10A. * * A part of particles

floats, the other sinks.

Total pore 3 volume, cm /g *

0.30 0.27 0.04 0.06 0.008

Aqueous solution behavior

floating fp + sp * * sinking sinking floating floating sinking

38

3.3. The effect of metal particle size on different supports A comparison of data given in Tables 1 and 4 as well as in Fig. 2 for carbonsupported catalysts shows clearly that the catalyst with the smallest palladium crystallites (5 Pd/carbon 8) achieves the highest H,O, yield. Other Pd on fluorinated carbon catalysts, which have an average metal crystallite size of about 13 nm, are several times less active (see Table 1). Although smaller palladium particles seem to be preferred in catalysts for H,O, synthesis, the metal crystallite size is certainly not the only important factor determining catalyst activity. Such a situation occurs for palladium supported on small particles of silica where the Pd crystallite size is 7.7 nm (Table 4) and, in spite of this, the H202yield is zero. Table 4 Palladium particle size on different supports Catalyst

5 Pd/carbon 65 5 Pd/carbon 8 5 Pd/carbon 5 5 Pd/carbon 3 5 Pd/Cab 5 Pd/Sil

Pd particle size, nm

13.9 10.4 13.0 13.0 7.7 14.0

Similarly, surface area (Table 3) is not a dominant factor controlling hydrogen peroxide yield. Catalysts with a surface area of 5 m2/g showed very poor activity, which may indicate that a higher surface area is recommended (as in 5 Pd/carbon 65 + Cab 1.0). However, when a catalyst sinks (excessive hydrophilic properties), then even a much higher surface area does not show good catalyst activity (as in 5 Pd/carbon 65 + 5 Pd/Cab). In summary, although metal particle size and catalyst surface area seem to affect catalyst performance in H,O, synthesis, they are not the dominant factors controlling catalyst activity. It appears that the level of hydrophobicity is a more important factor, and its optimum value should be established.

3.4. Decomposition of hydrogen peroxide Experiments for Hz02decomposition were carried out in the same reactor as for H202 synthesis in the presence of 0, only. Results, shown in Table 5, indicate that Pd deposited on the hydrophilic supports was very efficient for catalytic hydrogen peroxide decomposition. If Pd was present on the hydrophobic support and then mixed with a small amount of palladium-free hydrophilic component, H,O, was bound to decompose slowly, but such a catalyst showed poor activity for H,O, synthesis (see 5 Pd/carbon 65+Cab 0.5 in Tables 2 and 5). An increase in the amount of silica (as in 5 Pd/carbon 65+Cab 1.0) leads to some rise

39

in the rate of H,O, decomposition, but it also results in a strong increase in activity for H,O, synthesis. These results stress the importance of the appropriate proportion between hydrophobic and hydrophilic components. Table 5 Decomposition of H,O, over different catalysts at 2S°C in a stirred autoclave under 1.34 MPa of oxygen Catalyst

Concentration of H,O, after 12 h, wt% *

5 Pdlcarbon 65 + Cab 0.5 5 Pdjcarbon 65 + Cab 1.0 5 Pdlcarbon 65 + 5 PdJCab 5 Pd/Sil 5 Pdjcarbon 65 + 5 PdjSil No catalyst *

1.87 1.41 0 ** 0 *' 0.02 2.01

Initial concentration of H,Oz: 2.15 wt% occurred within a few minutes

* * Complete decomposition

3.5. The influence of pressure and stabilizer An increase in the H,O, mixture pressure has a beneficial effect on the hydrogen peroxide yield, though it is not very remarkable in the range of 0.65 to 2.14 MPa (Table 6). Table 6 Synthesis of hydrogen peroxide on 5 Pdfcarbon 65 t Cab 1.0 catalyst under different pressures Pressure, MPa

0.65 1.34 2.14 *

H,O, concentration wt% *

Hydrogen conversion %

Selectivity to H202, %

0.56 0.70 0.77

41 46 67

8.7

7 7

After 12 hours of experiment.

From the data listed in Table 6, it is expected that pressures higher than those used in this study can result in a higher hydrogen peroxide concentration. An improved H,O, yield was also observed with the addition of stabilizer. In an experiment, a reaction pressure of 1.34 MPa, the hydrogen peroxide concentration was 1.8 wt%, compared with 1.2 wt% without the presence of the stabilizer. The results are shown in Table 7.

40

Table 7 Synthesis of hydrogen peroxide on 5 Pdlcarbon 8

Pressure, MPa

Stabilizer

1.34 1.34 1.34 2.14 2.30

no

no Yes Yes Yes

Temperature

H,O, Concentration

OC

wt%

25 -10 -10 -10 -10

0.2 1.2 1.8 3.0 5.0

4. CONCLUSIONS 1. Palladium mounted on a hydrophobic support is a good catalyst for hydrogen peroxide synthesis. Its selectivity towards H,O, surpasses that of platinum. The catalyst was found to have a long lifetime. 2. Fluorinated carbon is a very suitable support for Pd. Its hydrophobicity can be controlled by changing the degree of fluorination. An IR spectrum in the range of C-F stretching vibrations provides a quick measure of carbon hydrophobicity. 3. The hydrophobicity of the catalyst is the decisive factor in determining selectivity. Factors of less importance are metal particle size and metal area. 4. All palladium must be placed on the hydrophobic component; otherwise, H,O, decomposes rapidly. 5 . Increased pressure and the addition of stabilizer to the reaction medium result in higher yields of hydrogen peroxide. Reduction in reaction temperature causes an increase in selectivity.

5. REFERENCES W.J.M. Pieters and A.F. Venero, Studies in Surf. Sci. and Catal. (Elsevier) 19 (1984) 155-163. L.W. Gosser, U.S. Pat. 4681751 (1987). W.F. Brill, U.S. Pat. 4661337 (1987). A. I. Dalton, Eur. Pat. Appl. 81107742.9 (1981). H.P. Boehm, Adv. Catal. 16 (1966) 179-274. Th. Van der Plas, in "Physical and Chemical Aspects of Adsorbents and Catalysts" (B.G. Linsen, Editor), Academic Press, London and New York, (1970) 425-469 7. J.J. Wu, L. Fu and K.T. Chuang, preprints, 11th Canadian Symp. on Catal. (J. Monier, Editor), Halifax, (1990) 300-308. 1

41

300

.

.

1450 1400 1350 1300 1250 1200 1150 1100 1050 --I

1

Diffraction Angle, "28

Wavenumber. cm

Figure 1. IR Spectra of fluorinated carbon samples. 1: carbon 65, 2: carbon 8, 3: carbon 5, 4: carbon 3, 5: non-fluorinated carbon (for comparison purposes). The spectra are offset for clarity.

Figure 2. XRD patterns for palladium on different supports. (Support pattern was subtracted.) 1: 5 Pdlcarbon 65, 2: 5 Pdlcarbon 8, 3: 5 Pdlcarbon 5 , 4: 5 Pdlcarbon 3, 5: 5 PdISil, 6: 5 Pd/Ci?b. Patterns are offset for clarity.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science arid Catalysis, Vol. 12, pp. 43-55 0 1992 Elsevier Science Publishers B.V. All rights reserved.

43

THE SELECTIVE OXIDATION OF METHYL-a-D-GLUCOSIDE ON A CARBON SUPPORTED Pt CATALYST Y Schuurman, BFM Kuster, K van der Wiele and GB Marin' Laboratoriwn voor Chembche Technologie Eindhoven University of Technology, The Netherlands P.O. Box 513 5600 MB Eindhoven, The Netherlands Abstract The selective oxidation of methyl-a-D-glucoside to methyl-a-D-glucosiduronateby oxygen using active carbon supported Pt was studied in a three phase stirred tank reactor. The temperature was varied from 293-313 K, the pH from 6-10, the oxygen partial pressure from 5.0 Id to 1.0 16 Pa at a constant pressure of 1 16 Pa, the methyl-a-D-glucoside concentration from 50 to 1000 mol m-' and the catalyst concentration from 1-5 kg m-3. The conversion ranged from 0.02 to 0.10. At this conversion range methy1-a-Dglucosiduronate was obtained with a selectivity of 100%. At the investigated reaction conditions the initial reaction rates were free from m a s and heat transfer limitations. Methyl-a-D-glucosedialdehydewas found to be a reactive intermediate. The data could be described by a reaction rate equation of the Langmuir-Hinshelwood type, corresponding to a rate-determining step on the Pt surface involving two active sites. No distinction was made between the sites of chemisorption for oxygen and methyl-a-Dglucoside. The rate equation was based on a reaction sequence with two reaction paths, the first involving the adsorbed methyl-a-D-glucoside and dominating at low pH, the second involving the methyl-a-D-glucoside anion and dominating at high pH. The adsorption of methyl-a-D-glucosedialdehydeand of methyl-a-D-glucosiduronate was found to be negligible. The surface was mainly covered with oxygen at low pH and with both oxygen and the methyl-a-D-glucosideanion at high pH. INTRODUCTION

The interest in processes with carbohydrates as chemical feedstock is growing considerably [1-31. A first step towards valuable chemicals from carbohydrates can be the selective oxidation of one of the functional groups, in particular the primary hydroxyl function. The platinum catalyzed oxidation with molecular oxygen is sufficiently active for the oxidation of primary or secondary hydroxyls [4-81. Primary hydroxyls are oxidized preferentially in the presence of secondary hydroxyls [4]. The catalytic oxidative dehydrogenation of alcohols to aldehydes in aqueous solutions is believed to proceed via abstraction of H atoms by platinum followed by the oxidation of the latter. Electrochemical potential studies indicate that, during the oxidation of various alcohols,

44

the platinum surface is covered with H atoms [9,10]. Furthermore, it is possible to replace 0, by other acceptors of H atoms such as benzoquinone [ l l ] or methylene blue [12]. In non aqueous media the reaction stops at the aldehyde [9]. In water aldehydes are in equilibrium with their corresponding hydrates and easily oxidized to carboxylic acids. Rottenberg and Baertschi showed with '*O-tracing that the oxygen atom incorporated in ethanol during oxidation to acetic acid, originates from H,O rather than from 0,[13], indicating that the oxidation of aldehydes to carboxylic acids occurs via their corresponding hydrates and is analogous to the oxidation of alcohols to aldehydes. Methyl-a-D-glucoside was chosen as a model substance to study the kinetics of the platinum catalyzed oxidation of a primary hydroxyl function in the presence of secondary alcohol functions. The aim of this study was to obtain insight in the reaction sequence by detailed kinetic modelling, based upon statistical regression of intrinsic kinetic data.

4 Y 3 Y

I

0.0 I

0

400

800

1200

Zoo0

1600 t

Figure 1 Reactor set-up. 1 reactor, 2 liquid supply section, 3 gas supply section, 4 liquid outlet & sampling section, 5 gas exhaust, 6 pH indicator and control, 7 oxygen electrode, 8 temperature indicator and control

/

S

Figure 2 Reaction rate versus time. 100 mol mJ,Po,=8.0 104 Pa, CRCHmH= T=298 K, pH=8, X=O.O6, C,=3 kg m.3

EXPERIMENTAL Equipment, procedure and conditions The oxidation reactions were performed in a three phase stirred tank reactor. The set-up is shown in Figure 1.The reactor was operated at atmospheric pressure and continuously fed by an oxygen /nitrogen mixture. Solutions of sodium hydroxide and methy1-a-Dglucoside were simultaneously added in a constant ratio. The sodium hydroxide was added to neutralize the sugar acids produced during reaction, in order to maintain a constant pH. The temperature was measured with a Pt probe and kept constant within 0.4 K by a waterbath. An oxygen electrode detected the amount of oxygen dissolved, expressed as a percentage of the oxygen solubility in water at 1 16 Pa of pure oxygen. A membrane filter at the bottom of the reactor allowed removal of the aqueous solution, while retaining the catalyst in the reactor.

45

Before reaction the catalyst was reduced with pure H, in situ in distilled water at 363 K. Next the water was removed under a nitrogen atmosphere and the reactor was cooled down to reaction temperature. Because the measurements were performed at a given conversion the reactor was filled with a solution containing the appropriate amount of sodium methyl-a-D-glucosiduronateand methyl-a-D-glucoside and saturatedwith nitrogen. The nitrogen stream was replaced by a given oxygen/nitrogen mixture and the stirrer was switched on to start an experiment. During reaction the catalyst is deactivating. The inlet liquid flow rate was adapted to the rate of sugar acid production by a feedback control based on a pH measurement of the reaction mixture. Provided the catalyst deactivation does not result in selectivity changes, this results in a constant methyl-a-D-glucosiduronate and methyl-a-D-glucoside concentrations and justifies the extrapolation of the experimental data to obtain initial rates of reaction free from deactivation, viz. Figure 2. The investigated range of reaction conditions is listed in Table 1. Within this range 57 initial reaction rates were obtained. Table 1 Investigated range of reaction conditions. ~~

~~

0.05-1.0 krnol ma

catalyst concentration

oxygen partial pressure

5.0 Id-1.0Id Pa

temperature

293-313 K

PH

6-10

conversion of me-glucoside

0.02-0.10

me-glucoside concentration

1-5 kg rn-3

Catalyst The catalyst used throughout this kinetic study was a commercial 5% Pt on activated carbon F196 RA/W from Degussa. The fraction platinum atoms exposed was 0.59, as determined by CO-pulse chemisorption with the assumption of 1:l stoichiometry. The corresponding platinum surface area amounted to 6.5 103 mz (kg cat)-', calculated with the assumption of 1.4 lOI9 (Pt atoms) rn-, (131. Transmission electron microscopy showed an average platinum particle size of d =2.2 nm with a standard deviation of 1 nm. The BET surface area amounted to 8.5 m2 (kg cat)-'. The catalyst was crushed in a ball mill to reduce the particle diameter below 20 pm in order to avoid transport limitations.

Id

Mass and heat transfer limitations In order to obtain intrinsic reaction rates the mass transfer and heat transfer phenomena occurring in a three phase reactor were examined. The concentration and temperature gradients were calculated using physical constants and correlations for transfer constants from literature. The calculations showed that the resistances to transfer of oxygen and in particular the transfer of oxygen from the gas phase to the liquid phase were critical. Heat transfer resistances were negligible. The investigated reaction conditions were such that the initial reaction rates observed were free from mass and heat transfer limitations. Chemical analysis Analysis of the aqueous reaction mixture was performed off-line. Two different hplc configurations were used for the analysis of methyl-a-D-glucoside and sugar acid anions. Two columns in series, an anion-exchanger (Benson BA-XS) in the acetate form and a

46

cation-exchanger (Benson BC-X8) in the proton form with water as eluent and refractive index detection, allowed the analysis of methyl-a-D-glucoside. An anion-exchange column (Benson BA-X8) in the sulphate form with (NH4),S04 as eluent and refractive index and UV-212 nm detection in series, allowed the analysis of sugar acid anions. An external standardization method was used for quantification. The only sugar acid anion detected was sodium methyl-a-D-glucosiduronate. Net production rate The conversion of methyl-a-D-glucoside was calculated from:

x = 1- CRCH~OH CR~H~OH,~

with the assumption of a constant liquid density. The net production rate of sodium methyl-a-D-glucosiduronate follows from the corresponding continuity equation:

Parameter estimation and model discrimination The data analysis was performed as outlined by Froment and Hosten 1141. Maximum likelihood parameter estimates b were obtained by application of the least square criterion to the observed and calculated initial production rates, i.e. by minimizing the residual sum of squares:

c (R;-Rio,)2 n

S(b) =

+. .

MZN

(3)

j-1

This minimization was performed with a single response Marquardt algorithm [151. Model discrimination based on statistical testing of the significance of the kinetic parameters and of the global regression was performed. The parameter estimates were tested for significance by means of their approximate individual t values. The significance of the global regression was expressed by means of the ratio of the mean regression sum of squares to the mean residual sum of squares, which is distributed according to F[16]. A high value of the F ratio corresponds to a high significance of the global regression, i.e. the rate equations describe the experimental data satisfactorily over the whole range of investigated conditions. To facilitate the estimation of activation energies a reparametrization was applied:

E E l 1 =Ai,exp[--(- -11 RT R T T,,,

k =A,exp(--)

with T, the average temperature of the experiments.

(4)

41

0.0

1.O

05

t I ks

X

+ :me-glucoside, A :me-glucosedialdehyde,o :sodium me,, = 500 mol m-3,pH = 7, glucosiduronate. ,,C Po,= 1 16 Pa, T=308 K, C,,= 15 kg m”. Batch experiment. Figure 3 Concentration of

Figure 4 Selectivity versus conversion. C,,,,, = 5oO mol m 3 Po2= 1 I@ pa, pH=7, T=325K, C,=U, kg m-3. Batch

experiment.

REACTION NETWORK

In Figure 3 the concentrationsof methyl-a-D-glucoside,methyl-a-D-glucosedialdehydeand sodium methyl-a-D-glucosiduronate are plotted versus time for a batch oxidation. The aldehyde is a reactive intermediate i.e. it disappears as soon as it is formed. The ratio of the pseudo first order reaction rate coefficients for the oxidation of the alcohol to the aldehyde and of the aldehyde to the acid is smaller than 0.01. Figure 4 shows the selectivity for methyl-a-D-glucosiduronate versus the conversion of methyl-a-D-glucosidefor a batch oxidation. Consecutive reactions such as the oxidation of the secondary hydroxyl functions cause a cleavage of the C-C bonds. The cleavage of the C-C bond between C3 and C4 is considered to be most favourable [17]. Several acids containing 1-4 carbon atoms are produced at conversions higher than 25%.

GIsa W ~-(--&, ’ Na+o-

0

Pt/C

Ho

OH

OH

+ H20

+

H20

Figure 5 The oxidation of methyl-a-D-glucoside via methyl-a-D-glucosedialdehyde to sodium methyl-a-Dglucosiduronate.

48

This paper, however, deals only with the data obtained at conversion low enough to obtain sodium methyl-a-D-glucosiduronate with a selectivity of 100%. Figure 5 shows the reaction network considered.

._

O W

025

050

0.75

P ,

1.00

tOaPa

Figure 6 Initial reaction rate versus oxygen partial pressure. C,,,,, = 100 mol mJ, pH=6, T=303 K, X=O.O6. P :exp data, line :rate equation (6) with parameters estimates obtained by regression of the complete set of experimental data.

,,C

I mol rn-'

Figure 7 Initial reaction rate versus me, P 2.6 10' Pa, pH = 6, glucoside concentration. = T=303K, X=O.O6. o :exp data, line :rate equation (6) with parameters estimates obtained by regression of the complete set of experimental data.

X

Figure 8 Initial reaction rate versus pH. CRCH,,=200 mol rnj, Po,=2.6 104 Pa, T=303 K,X=O.O6. 0 :exp data, line :rate equation (6) with parameters estimates obtained by regression of the complete set of experimental data.

Figure 9 Initial reaction rate versus conversion. CRCHmH = 100 mol mJ, Poz=2.6 104 Pa, pH = 6, T=303K. A :exp data, Line :rate equation (6) with parameters estimates obtained by regression of the complete set of experimental data.

49

KINETIC ANALYSIS

Effect of the reaction conditions on the initial reaction rate Figures 6 to 9 show the initial reaction rate versus the Po,, CRCHIOH, pH and conversion at 303 K. The dependence of the initial reaction rate on the oxygen partial pressure shows a maximum at oxygen partial pressure of 2.5 lo4 Pa. The initial reaction rate increases with increasing methyl-a-D-glucosideconcentration. At a oxygen partial pressure of 2.6 lo4 Pa, the rate increase per concentration unit levels off for methyl-a-D-glucoside concentrations higher than 200 mol m-3. A variation of the conversion from 0.02 to 0.10 has no effect on the initial reaction rate. The above dependencies are found at every pH investigated. In the pH range from 7-8.5 no effect of the pH on the initial reaction rate is observed. In the pH range 8.5-10 the initial reaction rate increases strongly with increasing pH. Table 2 Reaction sequences for the oxidative dehydrogenation of methyl-a-D-glucoside =I

+ 2*

1

0,

2 3

RCH,OH+ * RCH,OH* + O* RCH,OH + OH-

4

5 6

#

RCH,O- + * RCH,O-* + O* 2 RCH,OH

+

0,

208

+ RCH,OH* + RCHO + H,O + 2* RCH,O- + H 2 0 + RCH,O-* -, RCHO + OH- + 2* -,

2 RCHO

411

1

1

2

0

2

0

0 0

2 2

0

2

+ 2 H,O

Reaction sequences and rate equations In order to understand and at the same time describe quantitatively the observations summarized in Figures 6 to 9, several a priori possible reaction sequences starting from methyl-a-D-glucoside and oxygen and leading to the selective oxidation product were considered. Table 2 represents schematically in a format popularized by Temkin [18] two such sequences. The stoichiometric numbers, u, indicate the multiplicity of the corresponding steps in the closed sequence leading to the global reaction of which the kinetics are studied. Actually, only sequences leading to methyl-a-D-glucosedialdehydehad to be taken into account as the latter can be considered as a reaction intermediate and as there is no effect of the methyl-a-D-glucosiduronate concentration on the initial reaction rates, viz. Figures 3 and 9. The global reaction considered in Table 2 is assumed to be irreversible, as was observed for the oxidation of 2-propanol at similar conditions by Dicosimo and Whiteside [9]. Hence, the levelling off of the initial rate increase with increasing methyl-a-D-glucoside concentration shown in Figure 7 indicates a significant coverage of active sites caused by either an associative, step 2 of Table 2, or a dissociative chemisorption of methy1-a-D-

50

glucoside, e.g. forming an alkoxide species and a H atom. Considering associatively adsorbed methyl-a-D-glucosiderather than dissociatively adsorbed methyl-a-D-glucoside resulted in a more adequate description of the experimental initial reaction rates. The maximum shown by the initial rate in Figure 6 , indicates that the oxygen is chemisorbed, step 1 of Table 2, and moreover that two surface species chemisorbed on the same type of site are involved in the rate-determining step, step 3 of Table 2. Several reaction sequences involving irreversible oxygen chemisorption were considered and the regression results with the corresponding rate equations were compared to those based on the reversible oxygen chemisorption. None of the results came close to those based on sequence I and 11. The discrimination between the different modes of chemisorption of the reactants was based on the regression of the kinetic data obtained at 303 K and a pH lower than 8.5. At these conditions the best description of the data was obtained by rate equation (5). This equation corresponds to sequence I of Table 2 with the surface reaction, step 3, as rate-determining step and describing the adsorption equilibria, steps 1 and 2, by Langmuir isotherms. Estimates of the parameters featuring in rate equation ( 5 ) are given in the first part of Table 3. As expected from Figure 9, the coefficient corresponding to the adsorption equilibrium of methyl-a-D-glucosiduronate was found not to be significantly different from zero and, hence, does not appear in rate equation (9, nor in sequence I.

Obviously step 3 in Table 2, is not an elementary step but rather a combination of elementary steps involving several intermediates. One possibility can be presented as: 3.1-1

RCH20H* t *

3.1-2

RCH,O*

3.1-3

H* t O*

+ HO* t *

3.1-4

HO* t H*

p

H 2 0 + 2*

3

RCH20H* + O*

-,

RCHO t H 2 0 t 2*

+*

+ RCH20* t H* +

RCHO

+ H* t *

This reaction sequence involves two successive hydrogen abstractions from the adsorbed alcohol by a vacant platinum site followed by the oxidation of the hydrogen atoms. Electrochemical potential studies indeed indicate that the platinum surface is covered with H atoms during oxidation reactions at low oxygen concentration [9,10]. Another possibility is given by:

51

3.2-1

RCH20H* + O*

3.2-2

RCH,O*

+ HO* RCH20H* + O*

3

+ HO* + H2O + 2* + H 2 0 + 2*

+ RCH,O* 4

RCHO

+

RCHO

This reaction sequence involves two successive hydrogen abstractions from the adsorbed alcohol by chemisorbed oxygen the latter acting as a base towards the alcohol [19]. Both the above detailed reaction sequences have in common that the 0 - H bond rather than the C-H bond is broken first, but no full consensus is found in the literature on this issue [9,20]. Dicosimo and Whiteside [9] measured an isotope effect of k,/k,=3.2 for the competitive oxidation of 2-propanol-doand 2-propanol-d,, indicating that the C-H bond breaking is involved in the rate-determining step, thus steps 3.1-2 and 3.2-2. The estimates for the adsorption coefficients appearing in the denominator of the rate equations corresponding to both reaction sequences 3.1 and 3.2 were found not to be significantly different from zero, with the exception of the adsorption coefficients for methy1-a-Dglucoside and oxygen. Steady state kinetics alone do not allow to discriminate between these reaction sequences. Table 3 Parameter estimates with their amroximate individual 95% confidence intervals. parameter

pH 6-8.5, T=M3 K'

pH 6-10, T=303 Kz

pH 6-10, T 293-313 K3

K, (lo4 Pa-')

2.52

f

0.70

2.54 f 0.64

2.49 f 0.50

K2 (lo5 m3 mot')

3.41

f

0.68

3.32 f 0.61

3.36 f 0.45

k,

0.87

f

0.33

0.88 f 0.32

(kg cat mot' s-')

%K5/K,,, (m3 moil)

54.0

k6

(kg cat mol" s.')

1.79 f 0.23

EP:P

(W mol")

33.5 2 6.7

E$P

(W moI')

76.5 f 15.3

f

12.7

53.0 f 10.9

&(3) (10 ' kg cat mot' s-I)

6.0 f 1.9

&(6) (10" kg cat moll s.')

1.0

f

0.3

' Obtaincd by regression with rate equation (5) of the data at pH 6-8.5 and T=303 K.

* Obtained by regression with rate equation (6) of all data at T=303 K.

Obtained by regression with rate equation (6) and (4) for k3 and of all data. Confidence interval on the reparametrized preexponential factors

Following the regression of the data at low pH, a regression analysis of all the data obtained at 303 K, i.e. over the full pH range, was performed.

52

Rate equation (5) does not depend on the hydroxyl ion concentration and, hence, cannot describe the experimental data above pH 8.5, viz. Figure 8. In aqueous media methyl-a-D-glucoside is in equilibrium with its corresponding anion. The dissociation equilibrium coefficient amounts to 2 lo-” mol me3[211. Although a t a pH of 9 this results in a methyl-a-D-glucosideanion to methyl-a-D-glucoside ratio of 2 lo5, it is possible that two reaction paths have to be considered, one involving methyl-aD-glucoside and dominating at low pH, sequence I of Table 2, the other involving the methyl-a-D-glucosideanion and dominating at high pH, sequence I1 of Table 2. Equation ( 6 ) is the rate equation corresponding to the occurrence of two parallel reaction paths.

The estimates of the parameters featuring in equation (6) obtained by regression of the experimental data at the full pH range and at T=303 K are given in the second part of Table 3 together with their approximate individual confidence intervals. As expected the estimates for the Langmuir equilibrium coefficients K, and K2 and for the reaction rate coefficient k, featuring in equation ( 6 ) are quite close to those featuring in equation (5). Rate determining step 6 is an elementary reaction step, however, it still can be a combination of steps for reasons similar to those discussed concerning step 3. Based on the estimates reported in Table 3, the ratio of the Langmuir equilibrium coefficients for the methyl-a-D-glucosideanion, K,, and methyl-a-D-glucoside, K,, is calculated to be 8 Id. Such a high ratio is expected on the basis of the usually higher heat of adsorption for anions [22]. This explains the considerable contribution of reaction path I1 to the overall initial reaction rate, despite a very low methyl-a-D-glucosideanion concentration. The ratio of the Langmuir equilibrium coefficients for oxygen and methyl-a-D-glucoside is calculated to be 7 lo3, with the assumption of H = l 16 Pa m3 mol-’. With the equilibrium adsorption coefficients from Table 3 it can be calculated that the surface is mainly covered with oxygen at low pH and with both oxygen and the methy1-a-Dglucoside anion at high pH. Arrhenius parameters To estimate the activation energy and the heats of adsorption, experiments were performed between 293 K and 313 K for each process variable, except the conversion. The upper limit for the temperature range was dictated by the demand for intrinsic kinetics. The adsorption enthalpies for oxygen, methyl-a-D-glucoside and methyl-a-Dglucoside anion were found not to be significantly different from zero. Only the Arrhenius temperature dependence of the reaction rate coefficients could be accounted for. The apparent activation energies and the apparent preexponential factors, obtained by regression of the complete set of 57 experimental data, are given in the last part of Table 3.

53

The confidence intervals of the Arrhenius parameter estimates are rather large. Apparently the residual sum of squares is not very sensitive to the individual parameter estimates in the neighbourhood of the minimum. The parameter estimates are not strongly correlated, however. The highest binary correlation coefficient amounts to 0.90. Hence, an assessment of the individual parameter estimates with respect to values expected from rate theory is statistically meaningful. The fact that the adsorption enthalpies were not taken into account makes the physical interpretation of the apparent activation energies and preexponential factors difficult, however. The rather low Eapp for reaction step 3 is probably due to the fact that k, = Klzd, in which K stands for the adsorption equilibrium coefficient of an exothermic reaction e.g. steps 3.1-1 and 3.2-1. This pre-equilibrium is also the reason for the low preexponential factor. Nicoletti and Whiteside [8] reported for the oxidation of 2propanol in water at a pH of 7 over a Pt/C catalyst at the temperature range of 282-327 K an apparent activation energy of 38.1 50.8 W mol-' which is close to the value of Elpp. The values of the apparent activation energy and the preexponential factor for step 6 are more realistic. The estimate for the preexponential factor for step 6 corresponds to 1 lo-' m2 s-l a value lower than 1 10" m2 s-l expected from transition state theory if the standard entropy of activation can be neglected [23]. Agreement between experimental and calculated Figures 6,7,8 and 9 show the initial reaction rate, calculated according to equation (6) with the parameter estimates listed in the third column of table 3, and the experimental data versus the oxygen partial pressure, methyl-a-Dglucoside concentration, p H and conversion. The good agreement between the calculated initial reaction rate and the experimental data is also illustrated by the parity diagram shown in Figure 10. The corresponding F ratio amounts to 2500. No systematic deviations are observed. CONCLUSIONS

initial reaction rates 0006 7 -

5 2 g

. 0

0005 0034

o m OM*

000t

2' 0003

0001

0002

0003

OW4

OW5

0006

calc / m u kg cat)-' s-'

Figure 10 Experimental initial reaction rate versus initial reaction rate calculated with rate equation (6). Range of conditions : Table 1

A kinetic analysis of the intrinsic initial reaction rates has provided a better understanding of the platinum catalyzed oxidation of methyl-a-D-glucoside.The reaction kinetics can be described adequately over a broad range of reaction conditions by a relatively simple rate equation based on two reaction paths one involving adsorbed methyl-a-D-glucoside,the other involving the adsorbed methyl-a-D-glucosideanion. Both reaction paths contain a rate-determining step consisting of a surface reaction between these species and chemisorbed oxygen. The ambiguity of the steady state kinetics does not allow to come to more detailed conclusions about the rate-determining step.

54

NOTATION

preexponential factor of the reaction rate coefficient, m2 s-l b vector of parameters CoH- concentration of hydroxyl ions, mol m-3 concentration of methyl-a-D-glucoside, mol m 3 CRCHZOll E activation energy, kJ mol-' Fvc volumetric liquid flow rate, m3 s-' H Henry coefficient, Pa m3 mot' k reaction rate coefficient, (kg cat) mol-' s-l K equilibrium adsorption coefficient, equilibrium rate coefficient, Pa-' or m3 mol-' K, equilibrium dissociation coefficient of methyl-a-D-glucoside, m3 mol-' K, equilibrium dissociation coefficient of water, m6 mo1-2 L, surface concentration of active sites, mol (kg cat)-' Po, oxygen partial pressure, Pa gas coefficient, kJ mol-' K-' R R, reaction rate, mol (kg cat)-' s-l S selectivity S(b) objective function, mol (kg cat)-' s-l T temperature, K T,,, average temperature of all experiments, K t time, s W catalyst mass, kg cat X conversion * active site subscripts 0 at reactor inlet i step number j experiment number rds rate-determining step s surface superscript 0 at time=O s, standard app apparent n total number of experiments ' reparametrized A,,

REFERENCES 1. Fuchs A., Starch, 10, (1987), 335-343 2. Hickson J.L. (ed.), Sucrochemistry, ACS Symposium Series 41, Am. Chem. SOC., Washington D.C., (1977) 3. Schiweck H., Rapp K. and Vogel M., Chem. Ind., 4, (1988), 228-234 4. Heyns K., Paulsen H., Adv. Carbohydr. Chem., 17, (1962), 169-221 5. Van Dam H.E., Kieboom A.P.G. and van Bekkum H., Appl. Catal., 33, (1987), 361-

55

373 6. Haines A.H., Adv. Carbohydr. Chem., 33, (1976), 11-109 7. D i r k J.M.H., van der Baan H.S. and van den Broek J.M.A.J.J., Carbohydr. Res., 59, (1977), 63-72 8. Nicoletti J.W. and Whiteside G.M., J. Phys. Chem., 93, (1989), 759-767 9. DiCosimo R. and Whiteside G.M., J. Phys. Chem., 93, (1989), 768-775 10. Muller E. and Schwabe K., Z. Elektrochem., 34, (1928), 170-185 11. Wieland H., Chem. Ber., 54, (1921), 2353-2376 12. Rottenberg M. and Baertschi P., Helv. Chim. Acta, 39, (1956), 1973-1975 13. Scholten J.J.F., Pijpers A.P. and Hustings A.M.L., Catal. Rev. Sci. Eng., 27, 1,(1985), 151-206 14. Froment G.F. and Hosten L.H., in "Catalysis Science and Technology", Eds. Anderson J.R. and Boudart M., Springer Verlag, Berlin, (1981), Chap. 3 15. Marquardt D.W., J. SOC.Indust. Appl. Math., 11, (1963), 431-441 16. Draper N.R. and Smith H., Applied Regression Analysis, Wiley, New York, (1966) 17. Ogata Y., Sawaki Y. and Shiroyama M., J. Org. Chern., 42, (1977), 4061 18. Temkin M.I., Int. Chem. Eng., 11, (1971), 709 19. Davis J.L. and Barteau M.A., Surf. Sci., 197, (1988), 123-152 20. Razaq A. and Pletcher D., J. Electrochem. SOC.,(1984), 957-958 21. Vesala A., Kappi R. and Lijnnberg H., Carbohydr. Res., 119, (1983), 25-30 22. Anson F.C., Accounts Chem. Res., 8, 12, (1975), 400-407 23. Boudart M. and DjCga-Mariadassou G., "Kinetics of Heterogeneous Catalytic Reactions", Princeton University Press, Princeton, (1984)


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in SurJace Science and Catalysis, Vol. 12, pp. 51-10 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

CATALYTIC GAS -PHASE OXIDATION OF FLUORENE, ANTHRACENE AND PHENANTHRENE TO QUJNONES AND DICARBOXYLIC ANHYDRIDES

M. Baerns, H. Borchert, R. Kalthoff, P. KaDner, F. Majunke, S. Trautmann, A. Zein Ruhr -University Bochum P.O. Box 10 21 48, D-4630 Bochum, Germany

Abstract The title reactions were studied using supported and unsupported V 2 0 catalysts modified by MOO, and Fe,O, respective1 ; in some instances the catal$sts were also doped with alkali compounds. The a dition of molybdena to variadia resulted in a marked decrease in activity, selectivity, however, stayed nearly constant. When Fe,O, was added to vanadia the selectivity of inner -ring -oxidation products of all feed hydrocarbons increased, but activity decreased. The highest selectivities were obtained with a V:Fe atomic ratio of 1:1.4. By adding alkali the product distribution was markedly affected. Non -selective reaction steps were reduced, while the selectivity to inner -ring -oxidation products was further increased; with increasing atomic mass of the alkaIi metal this effect was enhanced. Some of the catalysts were characterized with respect to phase composition (XRD), surface com osition (XPS), oxidation sites (LTOC), and surface acidity (Calvet calorimetry, RIFTS and Transmission FTIRS). Relating the results of these measurements to catalytic performance, showed clearly that selectivity is de creased by surface acidity. As a conclusion recommendations for improved catalyst design in the selective oxidation of polycyclic hydrocarbons are derived.

J

!A

1. INTRODUCTION

Catalytic gas -phase oxidation of polycyclic aromatic hydrocarbons has received only little attention, except for naphthalene, in the scientific literature although such reactions being of continued industrial interest are already known for ,a long time /I/. Selective oxidations of fluorene, anthracene and phenanthrene being the subject of the present paper lead to quinones and dicarboxylic anhydrides. Besides these products also further oxidation to carbon oxides occurs. In this contribution, which is aimed to be a com arative study for the three different hydrocarbons, catalysts based on V,O, moified by the addition of MOO,, Fe,O,, and alkali metal compounds, i.e. Li,SO,, K,SO,, and Cs,SO, were applied. From the catalytic results obtained when applying the various solids as catal sts conclusions were d e rived for the underlying chemistry and the schemes o the oxidation reactions. Furthermore, an attem t was made to relate the catalytic performances to the physical and physico-c emical properties of the solids. Based upon the results obtained some suggestions are made for the design of improved catalysts.

K

r

58

2. EXPERIMENTAL 2.1 Preparation of Catalysts Vanadium oxide was the base component for the catalysts, ammoniumvanadat being the precursor; it was modified by ferrous oxide, alkali sulfates or molybdenum oxide and phosphoric acid. The catalysts were pre ared either by co re-Al,O,: 9 5 m2 g - l ; SiO,: 14 m2 cipitation or wet impregnation when supports 8-1; TiO,: 51 m2 g-1; all from Degussa) were used (for details see [2-4/). A compilation of the catalysts applied is given in Table 1 along with their compositions, calcination temperatures, BET surface areas and heats of ammonia adsorption on them.

x

2.2 Equipment for Catalytic Experiments Polycyclic hydrocarbons were generally oxidized with air in an electrically heated fixed-bed quartz reactor (d=0.8 cm, 1=30 cm) which was described previously / 5 / . The product distribution obtained in the oxidation experiments was determined by GC and HPLC analysis. Most of the condensable substances were separated on an OV-1 capillary column (Sichromat 2, Siemens, FID) and analysed by GC; the carbon oxides were analysed by GC using a TCD and two columns packed with molecularsieve 5-A and porapack Q for separation (Delsi 11 series, TCD). The HPLC analysis was ap lied for separation of quinones and anh drides (WATERS 712, UV array detector!; details have been reported elsewhere

/yg/.

2 3 Catalyst characterisation Catalysts were characterized with respect to their BET surface area, surface composition (XPS), bulk -phase composition (XRD), surface acidity (IRS), oxi dation sites (low tem erature oxy en chemisorption, LTOC), and initial heats of adsorption (Calvet ca orimetry). e essentials of the applied methods are d e scribed below. The specific surface area was determined by the 1-point BET method by low temperature (77 K) adsorption of N, after the virgin catalyst samples had been calcinated in air. In -sku IR -transmission experiments were performed to characterize surface adsorbates of the catalysts and to relate these findings to selectivities obtained. The equipment as well as the experimental conditions have been described else where /2/. Furthermore, the DRIFT -method (Spectra Tech,model 0030 -102) was applied for the same catalyst samples after outgassing them for two hours in nitrogen at 773 K to determine the kind of acidic sites on the catalyst surfaces by adsorption of pyridine at 470 K. 100 scans at a resolution of 4 cm-l were recorded for achieving a high signal-to-noise ratio. Powdered KBr was used as reference. The reflectance spectra were transformed to Kubelka -Munk units. The same interferometer (Perkin Elmer, model 1710) was used in both, transmission and DRIFT mode. The strength of surface acidity of some fresh catalysts after heating for 2 h at 573 K in vacuum was determined by the heat of adsorption of ammonia at.353 K. A volumetric adsor tion apparatus linked to a micro calorimeter of the TianSetaram) was used /7/. Calvet type (C80 and LTOC /8/ was carried out at 195 K; prior to adsorption the catalysts were reduced for ca 15 h with hydrogen at 623 K and then quenched to 195 K in an is0 -propanol/solid -CO, mixture (for experimental details see /9/).

P

fh,

?a

59

The surface compositions and the valence states of some key cations (V, Fe, Mo, 0, K, Cs) of selected catalysts were determined by XPS using an Al cathode (Leybold -Heraeus AG, LHS 10 spectrometer). The measurements were carried out after calcination of the samples in air and in a few cases also after reaction. The spectra were fitted with convolution of Gauss and Lorentz curves after nonlinear background substraction. The sensitivity factors of Wagner et al. /lo/ were used. Bulk compositions of powdered catalysts samples were obtained by XRD using Cu -K, -radiation. 3. RESULTS AND DISCUSSION

Firstly, the physical and physic0 -chemical properties of the various catalysts are communicated. Secondly, the catalytic results are described and the catalytic chemistry of the oxidation of the three hydrocarbons is explained. Finally, the effects of catalyst composition as well as of surface acidity determined by ammonia adsorption and IR studies are discussed and related to catalytic performance. Table 1 Properties of catalysts (bulk composition, BET surface area and initial heats of NH, adsorption), hydrocarbon applied (anthracene: A, fluorene: F, phenanthrene: P) and initial reaction rate for the oxidation of phenanthrene at 623 K (i.r.r.) Catalysts ComDosition atom- ratio

m?

V : MO = 1 : 0.33 V : MO : P = 1 : 0.35 : 0.1 V : MO = 1 : 0.05 v20,

V V V V V V V V V V

v l)

5, 7,

: Fe = 1 : 0.13

T,,,,

Qadinitia,l)

aromatic

i.r.r

K 773 773 773 773 773 773 623 623 623 623 623 623 623 773 773

kJ mol-1 88

feed P A P F, P A, F, P A, F, P F, P P A, P A, F, P P

wows

S,,

:Fe= 1 : l : Fe = 1 : 1.4 : Fe : Liz) = 1 : 1.4 : 0.06 : Fe : K3) = 1 : 1.4 : 0.06 : Fe : C S ~=) 1 : 1.4 : 0.06 : Fe : Cs/Al,O5) : Fe : Cs/SiO$ : Fe : Cs/TiO:) : Mo : P/Al,O6) : MO : P/si026j

g-'

5.8 0.8 8.4 4.0 8.0 2.0 1.5 1.6 1.8 1.2 98 149 51 74 128

ammonia adsorption at 353 K 20 wt.%; V:Fe:Cs = 1 : 1.4 : 0.06 extrapolated to 623 K

2) 6)

-

35 120 -

61 -

8 13 182 153 -

-

€9

0.17 -

0.29 0.31 0.42 0.067) 0.10 0.12 0.08 0.037)

P

-

P A A

-

-

to 4 alkali used as sulfates 10 wt.%; V:Mo:P = 1:1.2:0.1

3.1 Catalyst Properties The BET surface areas of the catalysts are presented in Table 1. The unsupported catalysts had low surface areas of less than 10 m'g-l. Especially low sur -

60

-

-

SiOz I

I

1

compounds. Heats of initial ammonia adsorption being a measure of surface acidity strength are given in Table 1 too. The

Table 2 Intensity of Lewis- @=1450 cm-1) and Bronsted-sites (V=:1540 cm-l) by adsorption of pyridine at 470 K of V/Mo/P-supported catalysts (V:Mo:P = 1:1.2:0.1) and pure su port materials in relation to the selectivity to the formation of carbon oxides at 6 3 K with 80% conversion of anthracene

!

Catalyst V/Mo/P -A1703 V/Mo/P 4 0 ,

Lewis type a. u. 1.77 0.79 2.13 .O

Bronsted type a. u. 0.94 0.05 =O =O

S(C0 ) [mol&io] 70.1 42.3 not measured not measured

61

XPS analyses carried out for selected catalysts showed only high -valence state cations, i.e. VS+(BE(2p3/?) = 517eV), Mo6+(BE(3d5/') = 232.3eV) and Fe3+(BE(2 3 4 = 711eV) and two different oxygen species, i.e. BE(O1s) = 530eV, BE! (01s) = 532.5eV (see Table 3); the binding energies were, however, independent of the catalyst composition indicating that no change in the oxidation state of the cations occured. OxyGen species with BE(O1s) = 530 eV represent the lattice oxygen of all the transition metals, while oxygen s ecies with BE(O1s) = 532.5 eV could be ascribed to oxygen containing species [-OH, H,O,,, -CO) on the surface. XPS data revealed that the surface Mo-contents of the V,O,MOO, -catalysts were higher than those of the bulk. In the vanadia -ironoxide catalysts the vanadia contents of the catalyst surfaces were higher than in the bulk exce t the catalyst with the atomic ratio V:Fe = 1:0.125. This may be related to the ower melting point of vanadia compared to iron oxide. In case of a high iron content a surface enrichment of vanadia was observed. In contrary the formation of mixed vanadia -iron -oxides (see XRD results, Table 4) besides pure V,O, results in an enrichment of iron.

P

Table 3 XPS -derived surface composition without accounting for carbon; (for bulk composition of catalysts see Table 4) Mn+: Mo6+, Fe3+ Ox: oxygen BE( 1s) = 530 eV; O##: oxygen BE( 1s) = 532.5 eV (V/M), V5+ Mn+ Catalyst composition atom % atom ratio 2.4 11.1 4.7 V : Mo1) = 1 : 0.33 1.6 1.4 0.9 V : M o ~ )= 1 : 0.33 V : Mol) = 1 : 0.05 6.6 13.6 2.0 6.2 2.5 0.4 V : Mo?)= 1 : 0.05 - 10.1 V*OS V : Fell = 1 : 0.13 4.9 17.3 3.5 V : Fe') = 1 : 0.13 4.7 24.0 5.1 V : Fe1) = 1 : 1 2.0 12.2 6.2 V : Fel) = 1 : 1.4 1.2 15.3 13.2 V : Fell = 1 : 1.4 1.0 10.6 10.4 V : Fe : K1) = 1 : 1.4 : 0.06 0.5 6.7 16.8 0.7 10.6 15.6 V : Fe : KZ) = 1 : 1.4 : 0.06 V : Fe : C S ~ . = ~ ) 1 : 1.4 : 0.06 1.2 12.7 10.4 V : Fe : CS?.~) = 1 : 1.4 : 0.06 1.8 13.4 7.4 l) fresh calcined catalyst after reaction 3) Fe(2s) -peak for determination

O#

O##

35.8 7.3 33.3 7.0 25.8 41.0 48.5 41.8 39.4 45.7 51.6 46.7 60.7 59.6

8.8 13.8 10.5 16.7 15.1 7.2 15.0 7.1 6.6 12.9 17.7 24.0 7.2 14.2

Alkali

-

-

-

-

1.5 2.0 3.3 3.9

The surfaces of the alkali doped catalysts were enriched with alkali (K and Cs respectively). The increase in the concentration of the oxygen species with a BE=530 eV caused by doping of pure V,O, with molybdena or iron oxide as well as alkali can be ascribed to the change in stochiometry as a first approximation. After carryin out the oxidation reaction the amount of the oxygen species at a BE=532.5 e $ increased.

62

The oxygen uptake determined by LTOC on reduced catalysts was affected by the calcination temperature for the catalyst with the atom ratio V : Fe : Cs = 1 : 1.4 : 0.06; the oxygen uptake decreased from 67 to 50 pmol/m’-,,, when calcination temperature increased from 623 to 773 K. No such effect was observed for the catalyst without alkali; the uptake amounted to 52 pmol/m2,,,. Phase composition of the bulk was determined by XRD (see Table 4). In all cases crystalline phases of V,O, and mixed oxides were detected. The amount of V20, decreased while adding secondary components. For the catalyst with the atomic ratio V:Fe = k1.4 with and without potassium the main crystalline phase was FeVO,. Table 4 Phase composition of the bulk determined by XRD Catalyst phase composition composition (%) atom ratio

pretreatment

V : Mo = 1 : 0.33

V,Os (33), MoVQ (67) v205 (25), MoV;OB (50), Mo,V,jO, (25)

calcination reaction

V : Fe = 1 : 0.13

V,O, (85), Fe,V,O, ( < 10) V20, (70), FeV,O, PO), Fe,V,O, (< 10)

calcination reaction

V : Fe = 1 : 1.4

V,O, ( < lo), FeVO, (80), Fe,V,O, ( < 10)

calcination

V : Fe : K = 1 : 1.4 : 0.06

calcination V,O, ( < l o ) , FeV,O, (65), Fe,V,0,,(20), Fe,V 0, ( < 10) V205 ( is prepared by mixing Ce(N03)4,6H20 and Y(N03)3,5H20 with ZrO2 and firing at llOO°C during 4 hours. c) Alkaline salts : Na3PMo12040, K3PMo12040 and Cs3PMo12040 are prepared according to Tsigdinos 1131. d) Commercial Sic is used (carborundum from Aldrich). The active phase is deposited by the incipient-wetness method (25 % wt). After calcination at 325OC under air the powder is compacted into pellets (3 mm diam.).

Catalytic testing Catalytic oxidation of IBA is performed in a stainless steel flow reactor at atmospheric pressure and 325OC. IBA (Merck synthesis grade) is injected into a nitrogen and oxygen mixture at 16OOC. The contact time is varied from 0.5 to 1 see. The partial pressure of IBA is 1.9 % atm and the molar ratio of H20/IBA is 2.4 and of 02/IBA is 2. The products are analysed on three chromatographs on-line with the reactor.

Characterization X-Ray diffraction experiments (XRD) are performed on a Rigaku Miniflex Diffractometer with the monochromatized CuK radiation (1 = 1.5418A). Powder samples are manually compacted on windowed aluminum holders. Transmission infrared spectroscopy is performed on a Perkin Elmer 577 spectrophotometer in the range 300 cm-1-1,400 cm-1. Samples are conventionally compacted with KBr. Setaram TG92 is used for thermal analysis (DTA) experiments.

RESULTS Catalysis When H ~ P M o ~ ~ V Ois~used, I - J the IBA conversion begins to decline after 3 days and 2 % of yield are lost after 9 days, which is sufficient to make this catalyst unsuitable for industrial application (fig 1). Concerning the alkaline salts of HPA supported catalyst, K3PMoiz040 seems to be the only compound on which selectivity in MAA of QPMollV040 alone is found, but the stability is not satisfactory (fig. 2).

84

Figures 1 to 5 IBA conversion (dark marks) and MAA selectivity (void marks) versus time for H4PMollV040

The performances of H4PMoilV040 when it is deposited on fluorine type like oxides are rather good, but decrease continually with time (fig. 3). In the case of the support Sic, the IBA conversion remains stable during a week, and the MAA selectivity, after a short increase, seems to be constant but is always too low (fig. 4). The results obtained with the different kinds of silica used as supports are shown on figure 5. SiO2 spherosil supported catalyst exhibits poor performance. It may be important to note that unlike the other silica type of support, Si02 spherosil when tested alone has its own reactivity toward IBA (table I). Though SiO2 (cristobalite fired without potassium) seems to be an efficient stabilizer during the first hours of testing, the

Fig. 1 : alone

" I

90

I

I a. \

I

I

0

1

2

Time (days)

Eg-2 : on Cs3PM0120~(squares), K3PM012040ftriangles), Na3PMo12040 (circles)

40t I

0

1

2

I Time (days)

Fig. 3 : on Ce02-Zr02 (squares) and dopped CeO2-ZrO2 (triangles)

85

90

n-n-n-8-

.

---.+---a$==%-*-*-*-

ao -

-A_-_

80

60 70

-

A-A-A-A

i \ -

5040 -

AyA-A-A-c

60.

conversion decreases after 3 days. When silica is fired in the'presence of K+ ions to obtain well cristallized silica, the selectivity in MAA is the same as the one of H4PMollV040 alone and the IBA conversion remains stable for more than 8 days. Table 1 Time 2 hours

Conversion 12 %

MAA 8,5 %

Selectivity c3H6 Aceton 38 % 1,5 %

co/cQ? 52 %

Characterization X-Ray diffraction patterns of the various supported or unsupported catalysts are recorded immediately after testing in order to avoid rehydration. Generally, it is very difficult to distinguish the characteristic peaks of the active phase (an example is given on figure 6 , where H4PMollV040 is supported on Cs3PMo12040). The main diffraction peaks of orthorhombic Moo3 are present in the pattern of the SiO2,S supported catalyst showing that a decomposition of the active phase occurred during testing. When Si02,K is used it appears, o n l y under operating or slightly reducing conditions, a well cristallized phase which is cubic and isotypic with K3PMo12040 (fig.7.a and b). DTA shows that in most cases the decomposition temperature of supported H4PMoilV040 is lower than for unsupported H ~ P M o I ~ V O (about ~ O 3OoC lower).

86

Figure 6 : XRD pattern of Cs3PMo12040 supported catalyst (Stars show the characteristic peaks of H 4 P M o l l V 0 4 0 )

10

20

30

40

28

Figure 7.a : XRD pattern of K3PMo12040

10

20

30

40

29

FiPure 7.b : XRD pattern of Si02,K supported catalyst (Stars show the characteristic peaks of K3PMo12040 relafed cubic phase)

Except for the SiO2,S supported compound, the 4 characteristic bands of Keggin structure heteropolyanion (between 1400 and 700 cm-1) are present on the IR spectra. This indicates that, though the catalyst looses a part of its activity, the bulk HPA structure is conserved. DISCUSSION

The use of the fluorine like oxides (Ce02-Zr02, Bi2O3) as supports of the active phase H4PMollV040 leads to an improvement of the stability of this HPA for the oxydehydrogenation of IBA. According to the fact that the formation of propylene may be correlated to the reduction of the catalyst [141, a decrease of the selectivity of propylene is consistent with an oxidizing effect of such supports. Similarly, the behaviour of the Sic supported catalyst is rather satisfactory and further investigations are in progress to obtain more complete informations. Concerning the effect of the alkaline salts, it may be noted that contrarily to what is reported for other reactions [8, 91, no stabilization is observed. As the support is catalytically active and, therefore may itself undergo a reduction when it is under operating conditions, it seems reasonable to think that the active phase instability is induced by the lack of stability of the support. Anyway, the best results are obtained when using silica containing potassium (tridymite or cristobalite).First of all it is important to note that several authors have tried to use various kinds of silica to support heteropolyacids and the (x = 0 to 3) is reported results are contradictory. Thus, when H3+xPM~12-xVx040 deposited on silica Aerosil 200 [12] its thermal stability decreases, moreover, the lower the amont of HPA, the lower the stability. On the other hand, silica Davison-Grace grade 400 enhances HPA thermal stability 115-171 when the coverage is low. Our catalytic testing results have shown that high specific areas silica are unsuitable because they degrade IBA and probably MAA, and, therefore, it is likely that the local hot spots which are generated that way, induce the active phase decomposition. On the contrary, low specific areas silica do not present this disadvantage. Moreover, a stabilizing effect is observed when potassium is added to silica. This may be related to the presence on such silica of the cubic phase isotypic with K3PMo12040. A similar phenomenon has been already mentioned in literature, particularly by Ueshima and al. who synthesized a structurally related HPA using pyridine and derivatives [18], and Goodenough [81 and later Haber 191 who report a stabilization of H3PMo12040 and H4PMollV040 by epitaxy on K3PMo12040. In our case, the potassium ions which are necessary to the generation of the cubic phase are located within the cavities formed by the framework of the SiO4 corner sharing tetrahedra of cristobalite or tridymite (fig. 8). Assuming that an important part of the alkali ions migrates to the silica surface, as it is often reported [19], it may be considered that the heteropolyanions are anchored on the silica surface by formation of ionic bonds between K+ ions and the heteropolyanionic units in order to create a K3PMo12040

88

a

b

Figure 8 : Base structure of : a) ideal cristobalite ;b) ideal tridymite structure base on which an isotypic acid phase could grow. As our experiments show, this phase is in a slightly reduced state but the Keggin structure is conserved. As a matter of fact, as it is shown on figure 9, the location of the K+ ions on the silica surface at the cavities entrance corresponds to the crystallographic sites occupied by K+ ions in K3PMo12040 with a low misfit. It may be thought that the silica structural modifications occurring during the various preparation steps of the catalyst (calcination, cooling, and evolution to the steady state) make this fitting easier.

Figure 9 : Possible crystallographic fit between high cristobalite ((111)plane) and K3PMo12040 ((100) plane)

89

On the other hand, as this cubic phase is generated under working conditions and anchored on a well adapted support from a crystallographic point of view, favourable conditions are combined to obtain a stable phase. Though its presence seems to stabilize H#MoI 1V040, the exact constitution of the interfacial zone between the external surface of the catalyst and silica is not yet well known. However, two hypothesis may be envisaged : i) either a few layers of K3HPMollV040 are formed on the silica and, in fact, the acid phase is supported on K3PMo12040 structurally related compound ;ii) or a concentration gradient of K+ ions, coming from silica, is set up throughout the catalyst bulk. However this last assumption is questionable taking into account the low diffusion rate of K+ ion owing to its large ionic radius. CONCLUSION In order to stabilize &PMollV040 in the oxidative dehydrogenati.on of IBA to MAA different supports have been employed. Well cristallized silica containing potassium led to the best results which may be relevant to the formation of a K3PMo12040 structurally related compound generated at the surface of silica under testing conditions. Nevertheless, the real constitution of the catalyst is not yet completely known. Then, further investigations are needed, noticely on the influence of nature and size of the foreign ions inserted in silica, for a better understanding of the observed stabilization phenomenon. BIBLIOGRAPHY

111 M. Misono ;Cata. Rev. Sci. Eng., 29 (2,3), 269 (1987) V. Ernst, Y. Barbaux, P. Courtine ;Catalysis Today, 1,167 (1987) M. Akimoto, Y. Tsukhida, K. Sato, E. Echigoya ;J. Catal., 83 (1981) M. Akimoto, K. Shima, H. Ikeda, E. Echigoya ;J. Catal., sf?l 173 (1984) M. Akimoto, H. Ikeda, A. Okabe, E. Echigoya ;J. Catal., 83 (1985) 0. Watzenberger, Th. Haeberle, D.T. Lynch, G. Emig ;New Developpements in selective oxidation, Rimini ; G. Centi and F. Trifiro Eds., p. 843, (1989) P. Coutine ;ACS Symposium Series : “Solid State Chemistry in Catalysis” ;5 37 (1985) ;R. K. Grasseli, J. F. Bradzil Eds. (a) J.B. Black, N.J. Clayden, P.L. Gai, J.D. Scott, E.M. Serwicka, J.B. Goodenough ;J. Catal., 106.1 (1987) (b) J.B. Black, J.D. Scott, E.M. Serwicka, J.B. Goodenough ;J. Catal., 106,16 (1987) 191 K. Bruckman, J. Haber, E. Lalik, E.M. Serwicka ;Catal. Letters, L 35, (1988) [I01 M.J. Bartoli ;Thesis, Compiegne, France (1990) [21 131 141 151 161

a a

90

C. Feumi Jantou ;Thesis, Paris VI (1989) H.G. Jerschewitz, E. Alsdorf, H. Fichtner, W. Hanke, H. Jancke, G. Ohlmann ;Z. Anorg. Allg. Chem., 526,73 (1985) G.A. Tsigdinos ;Ind. Eng. Chem. Prod. Res. Develop., 13 (41,267 (1974) 0.Watzenberger, G. Emig, D.T. Lynch ;J. Catal., 124 (l),247 (1990) S.Kasztelan, J.B. Moffat ;J. Catal., 106,512 (1987) J.B. Moffat, S. Kasztelan ;J. Catal., 109.206 (1988) S. Kasztelan, E. Payen, J.B. Moffat ;J. Catal., 125.45 (1990) E. P. 0 043 100, Nippon Shokubai R.K. Iler ;The surface chemistry of silica, Wiley Interscience N.Y. (1979)

91

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science and Catalysis, Vol. 12, pp. 91-100 0 1992 Elsevier Science Publishers B.V. All rights reserved.

NEW COMPOUNDS OF THE VANADIUM-MOLYBDENUM OXIDE SYSTEM. I N SITU INVESTIGATION OF THE MECHANISM OF ACROLEIN OXIDATION TO ACRYLIC ACID. THE ROLE OF THE STRUCTURE AND BOND ENERGY OF THE INTERMEDIATE COMPOUNDS T.V. Andrushkevich, V.M.

Bondareva, G . Y a .

Popova and L.M.

Plyasova

I n s t i t u t e of C a t a l y s i s , P r o s p e k t Akademika L a v r e n t i e v a 5, N o v o s i b i r s k 630090, USSR Abstract I n V-Mo o x i d e s y s t e m t h e c o m p o s i t i o n r a n g e o f 3-30 mol.% V 0 - 97-70 mol.% MOO i s a c t i v e and s e l e c t i v e i n t h e r e a c t i o n o f a c r o l e i n o x i d 2a t 4 i. o n t o ac3 r y l i c a c i d . T h i s r a n g e c o r r e s p o n d s t o a number of p h a s e s w i t h v a r i a b l e comp o s i t i o n , where VMo3O11 h a s t h e b e s t c a t a l y t i c p r o p e r t i e s . The s u r f a c e compl e x e s formed a t t h e a d s o r p t i o n o f a c r o l e i n on VMo30 h a v e been i d e n t i f i e d by I R s p e c t r o s c o p y , and t h e k i n e t i c s o f t h e i r t r a n s i o r m a t i o n t o a c r y l i c a c i d have been i n s i t u e s t a b l i s h e d , The bond e n e r g y o f t h e l a t t i c e oxygen h a s been d e t e r m i n e d by calorimetry i n s i t u . The q u a n t i t a t i v e dependence o f t h e r a t e o f t h e a c r y l i c a c i d f o r m a t i o n on t h e bond e n e r g y o f oxygen, a c r y l a t e and t h e q u a n t i t y o f V4+ i o n s h a s been shown.

1. INTRODUCTION Vanadium-molybdenum o x i d e s y s t e m i s d e s c r i b e d i n d e t a i l i n numerous publ i c a t i o n s s i n c e i t is t h e b a s e o f many c a t a l y s t s u s e d i n v a r i o u s p r o c e s s e s of p a r t i a l o x i d a t i o n 11’21

, including the process

of a c r o l e i n o x i d a t i o n

[I 3

The p r e s e n c e of e l e m e n t s w i t h v a r i a b l e v a l e n c y i n t h i s s y s t e m d e t e r m i n e s t h e v a r i e t y o f c h e m i c a l compounds a n d t h e i r f o r m a t i o n dependence upon t h e medium of t r e a t m e n t . Vanadium-rich r a n g e of c o m p o s i t i o n s h a s b e e n s t u d i e d most t h o r o u g h l y . Chemical compounds V Moo8, V9M06040, V6M04025 as w e l l a s s o l i d so2 i n V 2 0 5 have been d e s c r i b e d i n d e p t h . According t o numerous 3 p a t e n t s , i n t h e r e a c t i o n of a c r o l e i n o x i d a t i o n molybdenum-rich c o m p o s i t i o n s

l u t i o n o f MOO

are most a c t i v e . I n t h i s r a n g e o f h e a t t r e a t i n g new compounds VMo3011, [4,5]

a n d s o l i d s o l u t i o n of V204 i n Moo3

[6 ] have

been shown t o

form i n m i l d l y r e d u c i n g c o n d i t i o n s . Heat t r e a t m e n t a t e l e v a t e d t e m p e r a t u r e and t h e a c r o l e i n o x i d a t i o n medium are t h e o p t i m a l c o n d i t i o n s f o r t h e format i o n of t h e s e compounds. I n p r e s e n t p a p e r we r e p o r t t h e r e s u l t s of t h e i n v e s t i g a t i o n o f t h e a c t i v e component i n V-Mo o x i d e s y s t e m w i t h r e s p e c t t o t h e r e a c t i o n of a c r o l e i n o x i -

92

dation to acrylic acid, and the reaction mechanism.

2. EXPERIMENTAL The unsupported samples were prepared by evaporating an aqueous solution of ammonium paramolybdate and metavanadate in a spray dryer. The silica-supported samples were prepared by spray-drying the aerosil suspension (S = n

170 mL ) in aqueous solution of the salts mentioned above. The samples with additives of Cu, Cs, Ti and P were prepared by adding Cu(N0 3) 2’ CsN03,TiC14 or NH H PO into suspension. Powders were pressed into pellets, then ground 4 2 4 to 0.25-1.00 mm fraction. The samples were calcined in air at 3OO0C and activated in a reaction mixture (3-4% C3H40, 8-10% 02, 40% H20, balance - N2). Catalytic properties were studied in a flow circulation system at 300°C using a gaseous mixture of the above composition. After these experiments we have carried out the physical and chemical investigations. V4+ content was determined by ESR with a JES-3Q spectrometer and by a chemical method. Xray diffraction patterns were obtained with HZY-4B diffractometer with CuK, radiation used for phase analysis and with non-doublet CuKd radiation for the structural study. The installation for IR studies in situ and the method for determination of extinction coefficients are described in detail in [7] Adsorption heats of acrolein and acrylic acid were determined in a flow installation using Tian-Kalvet calorimeter [ 8 ]

. Lattice oxygen bond energy

was measured in a flow-impulse system, permitting to measure simultaneously both heat and rate of reagent interactions with the sample surface [ 9 ] 3.

.

RESULTS AND DISCUSSION Catalytic properties of supported V-Mo oxide system are illustrated in

Fig. 1. One can distinguish three regions essentially different in activity and selectivity:

(1) Moo3 and binary compositions with V 204 content up to 3 mol.% characterized by low activity and selectivity increasing with the rise of vanadium content: (2) compositions containing 7-30 mol.% V 204 active and highly selective in acrylic acid formation;

(3) vanadium-rich compositions (>50 mol.% V 2 0 4 ) with medium activity in complete oxidation of acrolein. Region 1 is a solid solution of V 0 in Moo3 of rhombic or hexagonal mo2 4 dification. Both modifications have close catalytic properties, but in the reaction conditions hexagonal modification is unstable and converts into a

93

-x

:q & 7

F i g . 1. C a t a l y t i c p r o p e r t i e s

;'IF;\

-6 4Do

....

.

and p h a s e c o m p o s i t i o n c h a r a c -

t e r i s t i c s of V-Mo o x i d e s y s tem. 1 - a c t i v i t y , 2

e

-

selec-

t i v i t y t o acrylic a c i d , 3 -

-

si ne tl e nc st i tvyi t of y t or e f CO+C02, l e x w i t h 4d-=

L Q)

-

0

0

v)

N

'5 0

0, d

0

1 E

2'

0

mole

4.00-4.07 A , 5 - i n t e n s i t y of 0

r e f l e x w i t h d = 3.25 A (Moog),

6 - p o s i t i o n of t h e most i n t e n s i v e d i f f r a c t i o n maximum o f V-Mo c h e m i c a l compound. The a c t i v e mass

r

Yo VZO,

- 30

wt.%.

F i g . 2. X-ray d i f f r a c t i o n p a t t e r n s o f vanadium-molybdenum

- 5 V2: 9 5 Mo, 2 - 10 V2 : 90 Mo; 3 -

compounds. 1

14.5 V2 : 85.5 Mo; 4 80 Mo; 5

- 30 V2

50 V2 : 50 Mo. ( x

- 20 V2

:

: 70 Mo; 6 -

-

reflexes

of Moo3).

rhombic o n e . The c o m p o s i t i o n r a n g e 7-14 mol.% V204 i s char a c t e r i z e d by t h e h i g h e s t act i v i t y a n d s e l e c t i v i t y . Its components are Moo3 and 0

18

16

14

12

10

8

6

-e

4

VMo3011 , t h e s y s t e m b e i n g mo-

nophase (VMo3011) a t t h e c o m p o s i t i o n 14.,5%V204 : 85.5% Moo3. The compound

94

VMo208-xwas i d e n t i f i e d i n t h e c o m p o s i t i o n r a n g e 15-30% V204 : 85-70% Moo3 w i t h d i f f r a c t i o n p a t t e r n somewhat d i f f e r e n t from t h a t o f VMo3OI1. A t V204 c o n t e n t i n s a m p l e s o v e r 30 mol.% V6M04025 p h a s e a n d s o l i d s o l u t i o n

of

MOO i n vanadium o x i d e c a n b e d e t e c t e d . 3 The comparison o f X-ray d i f f r a c t i o n p a t t e r n s o f t h e s a m p l e s i n t h e comp o s i t i o n r a n g e 5-50 mo1.X ( F i g . 2) shows on t h e whole t h e same s e t o f r e f l e x e s which o n l y s l i g h t l y d i f f e r i n t h e p o s i t i o n s and i n t e n s i t i e s . Two ref0

0

l e x e s ( d = 4.00-4.10 A and d = 3.62-3.53 A) are t h e most d i s t i n c t i v e , t h e i r positions s h i f t i n g i n opposite directions a t

v a r y i n g V/Mo r a t i o . I t c a n b e

c o n c l u d e d t h a t i n t h i s case a series o f compounds o f v a r i a b l e c o m p o s i t i o n are formed. S t r u c t u r a l a n a l y s i s based on powder X-ray d a t a f o r t h e sample with composition V

0.95M00.9705 [lo]

has shown t h a t t h e s e are a number of

s t r u c t u r a l l y r e l a t e d compounds V6M04025, V0.95M~0.9705,

VMo208

and VMo3011

of V 0 s t r u c t u r a l t y p e .

25

Certain r e g u l a r i t i e s of lattice parameters

c h a n g e d e p e n d i n g on V/Mo and

V4'/5

r a t i o a r e a p p a r e n t ( T a b l e 1). P a r a m e t e r s a a n d c i n c r e a s e w i t h 44i n c r e a s i n g molybdenum c o n t e n t and V s h a r e , and b p a r a m e t e r d e c r e a s e s .

T h i s r e s u l t s i n a g r e a t e r u n i f o r m i t y o f i n t e r a t o m i c d i s t a n c e s . The compound

VMo3011 c a n b e c o n s i d e r e d a s t e r m i n a l i n t h e series, s i n c e i t has t h e larg e s t v a l u e of c and t h e lowest v a l u e of b and M-0 bonds i n o c t a h e d r o n have t h e most u n i f o r m d i s t a n c e s . T a b l e 1. S t r u c t u r a l c h a r a c t e r i s t i c s of vanadium-molybdenum Formula

5'2' (V0.7M00.3)205 'gMo4O25 ' 0 . 95M00.5'79 VMo208-x vM03011

a

(1)

b

(i)

c

o x i d e compounds

(i)V4+/V4+ + V5+

11.51

4.37

3.56

0

11 .a0

4.17

3.65

30

11.99 6.33~2

4.09~2 4.05

3.36~2 3.13

66 60

4.04

3.73

75

4.00

3.76

98

6.34~2

-

(W)

- 100

It i s s e e n from X-ray d i f f r a c t i o n p a t t e r n o f t h e series s t u d i e d t h a t ref l e x (100) is b r o a d e n i n g w i t h i n c r e a s i n g molybdenum c o n t e n t a n d p r a c t i c a l l y

95

d i s a p p e a r s a t V/Mo = 1:3. According t o [lo]

t h i s may be due t o t h e weaken-

i n g bonds between p a c k e t s ( a l o n g w i t h t h e s t r e n g t h e n i n g of bond i n o c t a h e d -

. The-

r o n s ) up t o t h e i r r u p t u r e i n t o s e p a r a t e p a c k e t s i n VMo3011 compound

r e f o r e r e f l e x e s w i t h i n d e x h broaden a n d t h e n d i s a p p e a r , e . g . p e a k s (100)

(101), (001) c h a n g e

and (101), and r e l a t i v e i n t e n s i t i e s of p e a k s ( O l O ) , w i t h i n c r e a s i n g molybdenum c o n t e n t .

F i g u r e 1 r e p r e s e n t s t h e i n t e n s i t y of t h e most c h a r a c t e r i s t i c r e f l e x w i t h 0

d = 4.00-4.07 A showing t h e c o n t e n t i n V-Mo s y s t e m of t h e compounds belongi n g t o t h e series d e s c r i b e d . C a t a l y t i c a c t i v i t y i n a c r o l e i n o x i d a t i o n changes according

t o t h e c o n t e n t of t h e s e compounds. A t t h e same t i m e some d i f -

f e r e n c e s i n s e l e c t i v i t y are a p p a r e n t : t h e most s e l e c t i v e s a m p l e s h a v e comp o s i t i o n i n t h e r a n g e 3-14.5 mol.% V 0 - 97-85.5 mol.% MOO

2 4

VMo301

3

and c o n t a i n

phase.

Two main d i s t i n c t i v e f e a t u r e s c h a r a c t e r i z e t h i s compound: (1) i t h a s vanadium a l m o s t c o m p l e t e l y r e d u c e d t o 4 v a l e n t s t a t e and (2) i t h a s a l o o s e l a y of b u l k oxygen [9]

er s t r u c t u r e t h a t e n s u r e s h i g h m o b i l i t y

.

The i m p o r t a n c e o f vanadium o x i d a t i o n s t a t e f o r t h e s e l e c t i v i t y of acrol e i n o x i d a t i o n i s i l l u s t r a t e d i n T a b l e 2. S o l i d s o l u t i o n s of V204 i n Moo3 and V205 i n Moo3 h a v e comparable a c t i v i t y , b u t t h e f o r m e r i s h i g h l y select i v e , and t h e l a t t e r c a t a l y z e s o n l y c o m p l e t e o x i d a t i o n . C o r r e l a t i o n between s e l e c t i v i t y and V

4+ c o n t e n t i s a l s o o b s e r v e d i n a s e r i e s of c h e m i c a l com-

pounds ( s a m p l e s 1-4). T a b l e 2. C a t a l y t i c p r o p e r t i e s o f u n s u p p o r t e d V-Mo o x i d e compounds i n r e s p e c t t o acrolein oxidation Compound

VMo3011 vM0208-x

'gMo4O25 V2Mo08 0.03V204-0.97M00;

v4+ v4+ -v5+

w-Io-'~ molec

X(%>

m2. s

100

70 70 73 29 65

39.0 35.0 3.7 3.2 0.3

0

5

0.8

98 75 67 0

Selectivity (w)

C3H402 97.5 95.0 25.9 3.0 92.8

CO+C02

1.7 4.0 64.7 97.0 6.5

C2H402 0.7 1.5 9.4 0 0.7

solid solution

0.03V205.0.97MOO3

0

100

0

solid solution

:*

s u p p o r t e d on Si02, a c t i v e mass c o n t e n t i s 30%, X

W - rate of a c r o l e i n t o t a l conversion.

-

acrolein conversion,

96

The clue for understanding the important role of vanadium oxidation state is the reaction mechanism. Reaction mechanism In the conditions of proceeding the reaction of acrolein oxidation to acrylic acid in the temperature range 75-300°C three surface intermediate complexes: coordinatively bonded 6 -complex of acrolein SI-I ( 3 c=o = 1660 cm-I), surface acrylate SI-I11 ( 3 c=c = 1640, ~asCOo-= 1540, scoo- = 1420 cm-1 ) and molecular form of acrylic acid adsorbed on Si02 -1 SI-IV ( 1 c=o = 1750 cm ) have been identified by IR spectroscopy in situ.

b

that SI-I converts into SI-I11 with a high

We have shown previously [ll]

rate at 25-100°C through carbonyl bonded acrolein (SI-11). Centers on which 4+ SI-I is stabilized are Mo6+ ions, stabilization centers for SI-I11 are V

.

ions [ll]

The comparison of the rates of the reaction products accumula-

tion with those of SI decomposition in helium flow permitted to conclude that acrylic acid is formed during the step of surface acrylate (SI-111) destruction. The participation of the lattice oxygen in the reaction has been

. The sequence of SI transformation

established experimentally in

[12]

was established (see Scheme).

-

CH2=CH-CH0

K4

+1/202

t

TK1

(SI-I1

/H \\

K2

CH =CH-C

(SI-11)

CH2=CH-C

do

I

+

2e h rl

-0

(SI-111)

CH2=CHCOOH

Kg

4------

CH2

=

CH

.g m

-C ojp\\

0

Scheme K-l = 1.8*102 exp(-10000/RT) l/s; K2 = 4.5.106 exp(-19000/RT) l/s;

Kq = 3.104 exp (-8000/RT) l/atm.s.

K1 K3

=

46 l/atm*s;

=

2.1.103 exp(-13000/RT)

l/s;

91

%

V d .

Fig. 3. Kinetic curves of surface intermediates transformation and accumulation of the reaction products and acrolein in gas phase at temperatures: a - 125, b - 230, c - 26O0C. , 8 , A - concentrations of SI-I, SI-I11 and SI-IV. o , A , concentrations of acrolein and acrylic acid. Lines - calculated values, points - experimental values. Rate constants of the individual steps were estimated from the analysis of the kinetic curves of SI transformation and accumulation of acrolein and products in the gas phase (see

. a1 0

W

20

10

C, min

Fig. 3). The slow reaction step is the decomposition of surface acryla-

late. In accordance with the Bronsted-Polyani equation, acrylic acid formation rate can be as follows:

and q are adsorption heats of acrylic acid and oxygen, ac 02 the number of surface acrylate stabilization centers.

where q

v4+

is

Rate equations for CO and CO formation from acrolein and acrylic acid 2 can be written similarly, taking into account the slow step loosing of lattice oxygen

[I31

. Then

the selectivities for competitive and consecutive

routes can be expressed as

98

Scans

=

C'.exp( Y'qac - Y1'q

)/RT.f'"(Ci,Bi)

(4)

O2

- a d s o r p t i o n h e a t of a c r o l e i n , C. - c o n c e n t r a t i o n s , 8 . - s u r f a c e acr c o v e r a g e s by r e s p e c t i v e i n t e r m e d i a t e s .

where q

A d s o r p t i o n h e a t o f a c r o l e i n c h a r a c t e r i z e s t h e bond s t r e n g t h of lex (SI-I),

and a d s o r p t i o n h e a t of a c r y l i c a c i d - t h e bond s t r e n g t h

r y l a t e (SI-111).

6 -compof ac-

E q u a t i o n s 1-4 show c l e a r l y t h e dependence of a c t i v i t y a n d

s e l e c t i v i t y on t h e bond e n e r g y of oxygen and on t h e s t r e n g t h o f s u r f a c e i n t e r m e d i a t e s . R e q u i r e m e n t s t o a h i g h a c t i v i t y and s e l e c t i v i t y t o a c r y l i c acid d i c t a t e t h e c o n d i t i o n of optimum bond s t r e n g t h o f a l l t h e s u r f a c e i n t e r m e and t h e lower qaCr, - t h e O2. g r e a t e r c a t a l y s t s e l e c t i v i t y i s , i t s a c t i v i t y b e i n g l o w e r . A t t h e same time,

diates

[14]

: i t is o b v i o u s t h a t t h e h i g h e r q

t h e h i g h e r is qac,

t h e greater is t h e d e g r e e of a c r y l i c a c i d a f t e r o x i d a t i -

on. Optimum v a l u e of t h e bond s t r e n g t h i n

6 -complex o f a c r o l e i n ( S I - I )

must p r o v i d e t h e p o s s i b i l i t y of a c t i v a t i o n

C-H

bond i n a l d e h y d e group

w i t h o u t s i g n i f i c a n t p e r t u r b a t i o n of C=C and C-C bonds, which i s n e c e s s a r y f o r s e l e c t i v e o x i d a t i o n o f a l d e h y d e g r o u p t o c a r b o x y l g r o u p . Optimum bond s t r e n g t h of a c r y l a t e (SI-111) must e n s u r e S I - I 1 1 d e c o m p o s i t i o n w i t h o u t dest r u c t i o n o f t h e c a r b o n framework. T a b l e 3. C a t a l y t i c a n d thermodynamic p r o p e r t i e s of s u p p o r t e d V-Mo promoted c a t a l y s t s 3,102 W*10-l6 qacr qac '02 (molec/mZs) (kcal/mol)

Catalyst

S,"$Z)

V204' 9 Moo3

96.5

1.1

1.8

20

18

60

V204' 9Mo03. CuO

98.3

0.5

2.1

20

18

55

V204' 9Mo03. 0. 2Cs20

97.0

5.8

0.08

20

23

78

V204' 9MoOi 0. O6P2O5

88.0

1.4

1.1

18

80

0.5V 0 * 9Mo03. 0.5Ti02 2 4

97.0

0.7

1.8

28 -

18

65

B- a c r y l i c a c i d a f t e r o x i d a t i o n c o n s t a n t . The f o r m a t i o n of a c r o l e i n 6 -complex o c c u r s owing t o t h e e x i s t e n c e

of

l o n e p a i r o f e l e c t r o n s i n c a r b o n y l oxygen t h a t d e t e r m i n e s b a s i c p r o p e r t i e s

.-.

N

40.

E

95%), the yield falls due to the sequential oxidation (2) of sulfur to S02. At low conversions the selectivity asymptotically approaches a level of about 96%. The small quantities of SO2 produced at these low conversions strongly affect the maximum yield that the catalyst can achieve, as is apparent from figure 1. In order to optimize the performance of the catalyst, e.g., by addition of promoters, it is desirable to elucidate the mechanism of the catalytic reactions. Knowledge of the mechanism of the reactions also facilitates the set up of a kinetic model. -00

153

473

491

513

u)

UJ

Temperature("C. K)

figure 1

573

503

613

4U

473

103

513

u)

553

573

501

113

Temperature("C, K)

Dependence of the performance of the catalyst on the temperature P H2S = 1 kPa, P 0 2 = 5 kPa, P H 2 0 = 30 !@a, total flow = 200 ml/min, 0.40 g catalyst a) -0- conversion, -Bselectivity b) -A-yield

As compared to selective oxidation of hydrocarbons, the number of publications dealing with the selective oxidation of hydrogen sulfide is small [9-191. Most authors reporting on mechanistic studies on oxidation of H2S, used catalysts with narrow pores, such as, zeolites or active carbon [9-171. According to Steijns et.a1.[9-12] the narrow pores result in capillary condensation of elemental sulfur, that catalyzes the oxidation of H2S. Steijns et al. [lo] concluded to a reduction-oxidation mechanism with dissociatively adsorbed H2S. Other authors

125

consider @radicals to be important in the oxidation [ 171. The references [9-191 show a large variety of the orders of the reaction with respect to the partial pressures of 0 2 and H2S. The scatter of the values published for the activation energy of the catalytic oxidation of H2S is large; values ranging from about 10 to 60 kJ/mole are mentioned. It is likely that the true value of the activation energy is about 60 kJ/mole and that the lower values are due to diffusion limitation either within the catalyst bodies leading to an apparent activation energy of 30 kJ/mole or to the surface of the catalyst bodies resulting in an apparent activation energy of about 10 kJ/mole. Only one report attempts to set up a mechanism for an iron oxide catalyst containing wide pores [ 191. Unfortunately, the validity of the proposed mechanism was not demonstrated. Although the above mentioned literature proposes mechanisms and derives rate equations for the conversion of H2S, a mechanism accounting for the formation of SO2 has not been given. Only Steijns et.al. [lo] have measured an activation energy of the oxidation of adsorbed sulfur to SO2of 125 kJ/mole, which is near the activation energy of the oxidation of liquid sulfur (120 kJ/mole) [20]. Because the sulfur yield is strongly determined by the production of S02, we also studied the rate and the mechanism of the formation of S02. First we established the order of the total reaction with respect to H2S and 0 2 . The effect of the partial pressures of H2S and 0 2 on the selectivity to elemental sulfur and on the apparent activation energy was also determined. Steijns et al. published evidence that elemental sulfur is the catalytically active agent in the oxidation of H2S; both (dissociative) adsorption of H2S and 0 2 proceeds on adsorbed sulfur. With an iron species as the active component, the adsorption of oxygen presumably involves oxidation of iron(I1) to iron(II1). To demonstrate the effects of iron species on the catalytic reaction, we separately oxidized the surface of the catalyst at high temperatures with molecular oxygen and exposed the catalyst oxidized at different temperatures to H2S at room temperature. These results can indicate whether adsorbed elemental sulfur significantly affects the adsorption of oxygen.

2. EXPERIMENTAL 2.1. Preparation of the catalyst. The catalyst was prepared by impregnation of a preshaped silica support (Aerosil 0 x 5 0 ) using an ammonium iron citrate solution. After drying and firing at 773 K (500°C) the catalyst contains highly dispersed iron oxide particles (2-5 nm), that homogeneously cover the silica support. A detailed description of the preparation method has already been reported IS]. 2.2. Catalytic performance under steady state operation. The activity and selectivity were measured in a continuous microflow apparatus at atmospheric pressure. To obtain differential (conversion < 10%) and isothermal conditions, the catalyst was diluted (1:3) by crushing and thoroughly mixing it with silica powder. The thus obtained mixture was tableted (100 MPa) and a sieve fraction (0.42-0.63 mm) was made. The sieve fraction (400mg, about 1 ml) was placed into a quartz reactor (I.D. 10 mm). The high bed length to particle diameter ratio ensured that plug flow conditions in the catalyst bed are met. To transform the iron oxide rapidly into iron sulfate, which exhibits a stable performance, the catalyst was subjected to the following procedure [3]. Under typical reaction conditions (1% H2S, 5% 0 2 and 30% H20 in He) the temperature of the reactor was raised stepwise (10 K every 18 minutes) from 453 K (180°C) to 593 K (320°C) and cooled down to 453 K. This cycle was performed three times, totally taking 27 hours.

126

To investigate the influence of the partial pressures of both reactants and products, we used feeds containing different concentrations of H2S, 02, and H2O. Reducing atmospheres can arise when the ratio 02:H2S is substoichiometric (~0.5).Because reducing conditions can change the active material into an iron sulfide with a totally different performance, care was taken to avoid these conditions. The total flow rate was 200 m'(s'P)/min. The 02, H2S and SO2 content of the effluent was analysed with a gas chromatograph (Car10 Erba 6000) containing a 25 m Poraplot Q and 7 m Poraplot U column. At each gas composition a temperature-programmed procedure was performed. With this procedure the temperature of the reactor was raised stepwise (5 K) from 493 K (220°C) to 453 K (180OC). After each temperature step the effluent gas was analyzed three times at an interval of 9.0 minutes. Subsequently the composition of the gasfeed was changed and the same procedure was repeated.

2.3. Separate oxidation and reduction experiments. In order to investigate the behaviour of the catalyst under non steady state conditions, the H2S and the S0.L concentration was monitored continuously by use of an UV-Vis spectrophotonieter. The SO2 concentration and the H2S concentration was monitored at 280 nm and 232 nm, respectively. The adsorption at 232 nm was corrected for the contribution of S02. The 100 mm flow-through gas cell had an internal volume of only 7 ml to minimize the response time. The sample used for the experiment consisted of a sieve fraction (0.2-0.4 mm) of the undiluted catalyst. The catalyst sample was given almost the same pretreatment as mentioned above; only no water vapor was added. After this pretreatment the catalyst was cooled down in He to room temperature. Then three block-pulses of 1.2 % H2S in He (each 5 minutes long) were passed through the catalyst bed. Subsequently the temperature was raised (2 K/min) to 443 K (170OC) in a 1.2 % H2S in He flow. Next the desired oxidation temperature was set under He, followed by a reoxidation treatment of 10 minutes in a 2.0 % 0 f i e flow. The atmosphere was switched to He again, the reactor was cooled down to room temperature, and the three block-pulses and the temperature program were repeated. The area of first block-pulse was subtracted from one of the following two block-pulses to calculate the H2S consumption due to the reaction of H2S with the catalyst. It was assumed that the H2S reacted with the reactive oxygen species of the catalyst. 3. RESULTS

3.1. Catalytic performance under steady state operation. First the catalytic behaviour of the catalyst was measured as a function of the water partial pressure. Although the influence of the water pressure was measured at a range of temperatures, figure 2 only shows the influence at one temperature (473 K, 200°C). The deactivation by water is already obvious at relatively low water partial pressures. At water partial pressures higher than 5 kPa, however, no large change is observed. It must be noted that the effect of water is reversible; when the water feed is interrupted the higher activity is regained. This indicates that reversible adsorption of water at the surface of the catalyst impedes the catalytic reaction. This experiment indicates that the formation of water during the oxidation of H2S can affect the performance if the feed does not contain H20. We therefore used a feed of a high water pressure (30 Wa) to assess the effects of the partial pressures of 0 2 and H2S (fig. 3).

127 I0

tm

I

t\

1

d7RK

,

I

figure 2 The activity and selectivity versus the water partial pressure at 473 K (200°C). P H2S = 1.0 P a , P 0 2 = 5.0 kPa

Figure 3 shows the activity at a range of temperatures as a function of the 0 2 and of the H2S partial pressures. Assuming that the power rate law can describe the results adequately, the orders of the reaction have been calculated from the data. The order with respect to oxygen was found to be fairly high, viz., 0.63 to 0.78. The order with respect to H2S turned out to be lower, viz., it never exceeds a value of 0.5. At high H2S partial pressures the order with respect to H2S even approaches zero. The lower order with respect to H2S and the higher order with respect to 02 agree with values found by other authors [10,16]. The maximum order of 0.5 of H2S indicates dissociative adsorption of H2S to proceed on our catalysts also. The fact, that orders with respect to both reactants are fractional and also vary with temperature and partial pressure, indicates that adsorption of both hydrogen sulfide and oxygen determines the rate of the reaction. At high partial pressures no negative order with respect to both reactants is observed. This indicates that the adsorption of H2S and 0 2 presumably proceeds on different sites; competition for the same site does not occur. It is likely that oxidation and reducrion takes place subsequently, pointing to a Mars v. Krevelen [21] mechanism. The quite limited number of data points does not justify a fit to rate equations that are derived from this mechanism.

figure 3 The influence of the partial pressure on the activity (ml H2S/min gr) at different tcmperaturcs.

128

The selectivity to elemental sulfur and the apparent activation energy have also been determined as a function of the partial pressures of H2S and 0 2 . In figure 4 the selectivity is represented and in figure 5 the activation energy. It is remarkable that the selectivity and the activation energy do not vary significantly with the oxygen partial pressure.

&p -30 kPa

figure 4 The influence of the partial pressure on the selectivity

The activation energy of about 65 kJ/mole agrees well with the values published in literature for catalysts not displaying diffusion limitations. Interestingly the selectivity decreases strongly at lower partial pressures of H2S. The activation energy substantially increases and approaches the value mentioned in the literature for the oxidation of adsorbed or liquid sulfur to S02, viz., 120 kJ/mole. It can be concluded that at lower H2S partial pressures the rate of deep oxidation relative to the rate of mild oxidation increases and, therefore, the activation energy of the oxidation of adsorbed sulfur to SO2 contributes increasingly to the apparent activation energy.

It is significant that the selectivity to SO2 is hardly affected by the oxygen partial pressure, whereas that to elemental sulfur rises with the H2S partial pressure. The effect of the partial pressure of H2S is relatively strong: At a partial pressure of 0.057 kPa the selectivity to elemental sulfur is only 64 %, at a partial pressure of 4 kPa the selectivity is 98.5 %. One can therefore conclude that not only the concentration of the adsorbed oxygen is of importance for the selectivity, but that the concentration of another species, which is determined by the hydrogen sulfide partial pressure, has an even higher influence on the relative contribution of deep oxidation.

129

3.2. Separate oxidation and reduction experiments In figure 6 a typical H2S uptake peak as derived from the difference between the first and the third pulse is shown. The catalyst sample had been pretreated at 423 K (150°C) in an oxygen atmosphere. Catalysts pretreated at other temperatures show a similar peak. During the hydrogen sulfide uptake no formation of sulfur dioxide was observed. 1

I ,

lime (Me)

figure 6 H2S pulses admitted to a catalyst after a pretreatment in 2 kPa 0 2 in helium at 423 K (150OC).

Table 1 shows that after pretreatment at 523 K the catalyst can accommodate oxygen. At 523 K the adsorbed amount of elemental sulfur is readily oxidized to sulfur dioxide and is no longer present at the catalyst. Oxygen adsorbed after this pretreatment can not be accommodated by elemental sulfur, but is most probably adsorbed on iron sites. This indicates that iron(I1) sulfate can adsorb and activate oxygen species that already react with hydrogen sulfide at room temperature. Figure 7a shows that during the subsequent temperature program no additional uptake of H2S takes place. Table 1. The hydrogen sulfide consumption after 10 minutes of oxidation at different temperatures. 2.0 kPa 0 2 in He a) after reaction conditions b, after 100 minutes oxidation

Oxidation temperature

0

-a) 323 323b) 373 423 473 523

H2S

consumption (pmol/g catalyst)

2.8 6.8 17.9 11.8 24.0 12.4 30.8

However, as can be seen in figure 7b, a catalyst that had been pretreated at 623 K (350°C) and higher temperatures showed in addition to an increased H2S uptake at room temperature a significant uptake at 393 K (12OOC). Also during the uptake some sulfur dioxide was evolved. It is noteworthy that after this pretreatment the total amount of reactive oxygen present in the catalyst, as calculated from the total H2S consumption and SO2 production, equals the amount of oxygen necessary to oxidize all iron present in the catalyst from the bivalent to the trivalent state. Literature provides many publications about the oxidation of iron(I1) sulfate. Many authors

130

loTarnparalure (‘C)

figure7

TernparalumCC)

H2S consumption as a function of the temperature of a catalyst oxidized at different temperatures. a) 423 K, b) 623 K. Heating rate: 10 K/min.

mention the formation of an iron oxy sulfate (5)[22-251. However, some authors contradict this statement and claim that only iron(II1) oxide and iron(II1) sulfate is formed (6) [26-281.

4 FeSO4 + 0 2 -+ 2 Fe2O(SO4)2 12 FeS04 + 3 0 2 + 4 Fe2(S04)3 + 2 Fez03 Both reactions consume the same amount of oxygen, viz., the amount necessary to oxidize all bivalent iron to trivalent iron. To obtain some indication which reaction proceeds, high temperature X-ray diffraction [29] was performed while heating anhydrous iron(I1) sulfate in an oxygen atmosphere. Below 623 K no other phases but iron(I1) sulfate were seen. Above 623 K only iron(II1) oxide and iron(II1) sulfate were observed. The hydrogen sulfide consumed at temperatures above 393 K (120°C) can therefore be attributed to the reaction with a bulk iron (111) compound. The low temperature consumption of H2S is a reaction with surface oxygen only. This is also confirmed by the fact that the amount of oxygen calculated from the consumed amount of H2S is relative low compared to the amount of iron present in the catalyst. Rough calculations with iron(I1) sulfate with a particle diameter of 8 nm reveal that about 30% of the bivalent iron ions is at the surface and can react with oxygen to form trivalent iron ions. The amount of reducible iron after oxidation, as determined from the hydrogen sulfide consumptions at low temperatures, never exceeds 10% of the total amount of iron present in the catalyst. The surface oxygen shows a high reactivity towards hydrogen sulfide, even at room temperature. However, reoxidation of the surface is much slower, even at high temperatures (Table 1). The slow (re-)oxidation of the catalyst makes it very likely that under reaction conditions the reoxidation of the surfwe also will be a relative slow elementary step. This is confirmed by the fact that the order with respect to oxygen is high. A third piece of evidence that reoxidation mainly determines the overall rate is given by the finding that the active phase mainly consists of iron(I1) sulfate. Thermodynamic calculations predict reaction to only iron(II1) sulfate. The fact that most of the iron is present in the bivalent state indicates that the oxidation to the trivalent state proceeds more slowly than the reduction to iron(I1). An effect of the valency of an active transition metal ion is also seen in the selective oxidation of butane to maleic anhydride. The material exhibiting the highest selectivity is also in the reduced state, viz.,(V(IV)0)2P207 [30]. If the reoxidation step is slower than the reaction with hydrogen sulfide, a low surface coverage of oxygen must be observable under the conditions of the catalytic reaction. This is also confirmed by the pulse experiment after the catalytic reaction at 433 K (160°C). A very small H2S consumption was measured, which indicates a low surface concentration of oxygen.

131

4. DISCUSSION The maximum order with respect to H2S is 0.5, which points to dissociative adsorption of H2S. It is therefore likely that H2S is adsorbed on zero-valent sulfur species present on the surface. As proposed by Steijns et al. [lo], the adsorption leads to:

Because the reaction temperature is relatively low, the vapor pressure of sulfur is within the order of magnitude of the pressure of sulfur formed during the catalytic reaction. This suggests that elemental sulfur is physically adsorbed on the surface of the catalyst. Therefore reaction of adsorbed H2S with adsorbeg-elemental sulfur is likely. The activation of oxygen on the contrary is brought about by iron sites and not by elemental sulfur as found on active carbon by Steijns et.al.. The oxidation and reduction experiments have shown unambiguously that after high temperature oxidation in the absence of sulfur the surface accommodates highly reactive oxygen. It therefore is likely that the adsorption of oxygen proceeds on iron sites. This is also confirmed by the finding of Steijns [9], who obtained a more active catalyst by addition of iron oxide to active carbon. The separate oxidation reduction experiments have also shown that the (re-)oxidation of the surface by molecular oxygen is the rate-determining step of the selective reaction. Oxygen adsorbed on iron sites, subsequently reacts with the adsorbed S,H species. It is probable that the adsorbed SxH species is mobile. As can be derived from figure lb, the surface i s saturated at high H2S partial pressures with S,H groups causing the low order with respect to H2S. If the partial pressure of hydrogen sulfide is low, the S,H concentration decreases relatively to the S, concentration. Now the relatively slow reaction of adsorbed sulfur (S,) with adsorbed oxygen to sulfur dioxide can proceed. At high H2S partial pressures, a high coverage of S,H species is established. These species are highly reactive towards adsorbed oxygen, thus decreasing the number of adsorbed oxygen atoms. At the same time the amount of S, atoms will be low. Reaction of the small number of Sx species with the adsorbed oxygen atoms leading LO So;! will now be very limited, resulting in a high selectivity. The small influence of the oxygen partial pressure on the selectivity can also be understood. Because reoxidation of the catalyst is a rate-determining step, increase of the oxygen partial pressure enhances the overall reaction rate, without a strong increase of the oxygen surface concentration and without a marked change in the Sx concentration.

5. CONCLUSIONS From the results it is concluded that the activity of the catalyst for the selective oxidation of hydrogen sulfide is mainly determined by the reoxidation of the catalyst. The relative slow oxidation step causes a low surface coverage of reactive oxygen under reaction conditions. Due to this deep oxidation to SO2 is suppressed and a high selectivity to sulfur is obtained. The selectivity is mainly determined by the hydrogen sulfide partial pressure. Because hydrogen sulfide reacts with elementary sulfur to produce hydrogenated sulfur species a high partial pressure of H2S suppresses the reaction between adsorbed sulfur and adsorbed oxygen.

132

6. REFERENCES 1. P.H.Berben, A.Scholten, M.K.Titulaer, N.Brahma, W.J.J. van der Wal and J.W.Geus, in Stud. Surf. Sci. Catal. (Catalyst Deactivation), 34, 303-3 16, (1987). 2 P.H.Berben and J.W.Geus, in Proceedings of the 9th international congress on catalysis, M.J.Phillips and M.Ternan (Eds.), The chem. inst. of Canada, Ottawa, 284-291, (1988). 3 P.J.van den Brink, AScholten, A.J.van Dillen and J.W.Geus, in Stud. Surf. Sci. Catal. (Catalyst Deactivation), C.Bartholomew (Eds.), Elsevier, Amsterdam, 68,5 15-522, (1991) 4 J.A.Lagas, J.Borsboom, P.H.Berben, "SUPERCLAUS - The answer to Claus plant limitations", 38th Canadian chemical engineering conference, Edmonton, Canada, (1988). 5 C.N. Satterfield, Mass transfer in heterogeneous catalysis, M.1.T Press, Cambridge, (1970). 6 H.G.Karge, 1.G.Dalla Lana, S.Trevizan de Suarez, Y.Zhang, Proc. of the 8th Tnt. Congress on Catalysis, Berlin, (1984). 7 Z.M.George, Adv. Chern. Ser., 139,75-92, (1975). 8 P.J.van den Brink, A.Scholten, A.van Wageningen, M.D.A.Lamers, A.J.van Dillen and J.W.Geus, Stud.Surf.Sci.Catal., (Preparation of Catalysts V), G. Poncelet, P.A. Jacobs, P. Grange and B. Delmon (Eds.), Elsevier, Amsterdam, 63, 527-536, (1990). 9 M.Steijns and P. Mars, J.Catal., 35, 11-17, (1974). 10 M.Steijns, F.Derks, A.Verloop, P.Mars, J.Catal., 42, 87-95, (1976). 11 M.Steijns, P.Koopman, N.Nieuwehuijse, P.Mars, J.Catal., 42, 96-106, (1976). 12 M.Steijns and P. Mars, Tnd.Eng.Chem., Prod.Res.Dev. 16 (l), 35-41, (1977). 13 P.Zhenglu, H.Weng, J.M.Smith, Am.1nst.Chem.Eng.J. 30 (6),1021-1025, (1984). 14 1.Coskun and E.L.Tollefson, Can.J.Chem.Eng., 58, 72-76, (1980). 15 T.K.Ghosh and E.L.Tollefson, Energy Proc.Can., 77 (3,16-25, (1985). 16 M.Prettre, R.Sion, Z.Elektochem., 63, 100, (1959). 23, 699-707, (1975). 17 Z.Dudzik and M.Bilska-Zidek, Bull.Acad.Polon.Sci.SCr.Sci.Chirn., 18 YJwasawa, S.Ogasawara, J.Catal., 46, 132-142, (1977). 19 T.G.Alkhasov and N.S.Amirgulyan, Kinet.Katal., 23(5),962-966, (1982). 20 T.K.Wiewiorowski, B.L.Slaten, J.Phys.Chem., 71, 3014-3019, (1976). 21 P.Mars and D.W. van Krevelen, Chem.Eng.Sci.Specia1 Suppl., 3,41, (1949). 22 A.Bristoti, J.I.Kunrath, P.J.Viccaro, J.Inorg.Nucl.Chem., 37, 1149-1151, (1975). 23 A.H.Kame1, ZSawires, H.Khalifa, A.A.Saleh, A.M.Abdallah, J.Appl.Chern.Biotechno1, 22, 591-589,(1972). 24 M.S.R.Swamy, T.P.Prasad, B.R.Sant, J.Therm.Ana1, 15, 307-314, (1979). 25 P.K.Gallagher, D.W.Johnson, F.Schrey, J.Am.Ceram.Soc., 53( 12), 666-670, (1970). 26 E.V.Margulis, MM.Sokarev, L.A.Savchenko, N.I.Kopylov, L.I.Beisekeeva, Russ.J.Inorg. Chem. (English), 16(3), 392-395, (1971). 27 N.N.Kii, A.K.Zapol'skii, A.A.Mil'ner, G.S.Shameko, J.Appl.Chem.USSR, 61, 636639, (1988). 28 V.N.Turlakov, A.I.Sheinkman, S.D.Stanovnov, G.V.Kleshchev, J.Appl.Chem.USSR, 49(5), 1005-1008, (1976). 29 P.J. van den Brink, to be published. 30 F.Cavani, G.Centi, A.Riva, F.Trifiro, Catal.Today, 1, 17-26, (1987).

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shrdies in Surface Science and Catalysis, Vol. 72, pp. 133-145 0 1992 Elsevier Science Publishers B.V. All rights reserved.

133

Selective oxidation of ammonia to nitrogen over silica supported molybdena catalysts A structure-selectivity relationship Mark de Boera, A. Jos van Dillena, Diederick Janssenb, Tys Koertsc and John W. Geusa

C. Koningsbergera, Frans J.J.G.

a Department of

Inorganic Chemistry, University of Utrecht, P.O. Box 80083, Sorbonnelaan 16,3508 TB Utrecht, the Netherlands

b N.V. C

KEMA, P.O. Box 9035,6800 ET Amhem, the Netherlands

Department of Inorganic Chemistry, Eindhoven University of Technology, Den Dolech 2, P.O. Box 513,5600 MB, the Netherlands

Abstract The selective catalytic oxidation of NH3 to N2 over unsupported and silica-supported Mo03 catalysts has been investigated. The defect structure of the catalysts greatly affects the selectivity towards N20. The performance of the silica supported catalysts is controlled by the thermal pretreatment and the structure, which is installed by the preparation procedure. Water dramatically decreases the selectivity to N2 of Mo03-on-Si02 catalysts with low loadings. The selectivity to N2 highly depends on the ability of the catalyst to decompose released N20, which proceeds on oxygen vacancies.

1. INTRODUCTION The emission of ammonia substantially contributes to the present air pollution. About 94 % of the ammonia emitted in, e.g., the Netherlands originates from agricultural sources [l]. In areas with intensive stock breeding emission of ammonia leads to acidification of the atmosphere. Large volumes of ammonia are additionally produced in coal gasification and hydrotreating plants. If the NH3 produced is subsequently converted to nitrogen oxides, the environments will be comparably adversely affected. Many studies have been dedicated to the Selective Catalytic Reduction (SCR) of NO, with NH3. Silica- and titania-supported V2O5 catalysts have proven to perform adequately in the abatement of NO, [2,3]. A severe difficulty of the SCR is the stoichiometric amount of NH3 with respect to NO, that must be injected and thoroughly mixed into the gas stream to avoid slip of either NH3 or NO,. The Selective Catalytic Oxidation (SCO) of NH3 can meet with the removal of NH3 from stack gases. Molecular oxygen is used for the selective oxidation of NH3 to N2 and H 2 0

134

(reaction 1). However, besides the desired products, i.e., N2 and H20, also N20 and NO can result (reactions 2-3). A number of catalysts can be used for the SCO [4]. Some -unsupportedtransition metal oxides, such as, Moog, V2O5, Biz03 and PbO, exhibit a sufficient selectivity towards N2. Thermodynamically, molecular nitrogen is the most stable reaction product [5]. A sequential oxidation path to the deepest oxidation product with N2 and N2O as intermediate products (reaction 4) can therefore be excluded at the temperatures of interest (below 700 K).

4NH3 + 3 0 2

+ 2N2 +

6H2O

AG'(298 K)= -156.1 kJ/mol

(1)

2 NH3

+

202

+ N20 + 3 H20

AG'(298 K)= -131.2 kJ/mol

(2)

4NH3

+

502

+ 4N0 +

AG'(298 K)= -144.7 kJ/mol

(3)

6H2O

The catalytic oxidation of NH3 is presumed to proceed by a reduction-oxidation mechanism [6], in which the reduction-oxidation behaviour of the catalyst affects the catalytic performance. According to Golodets [4], the bond energy of lattice oxygen within the bulk oxides determines mainly the selectivity ratio (N2/N20+NO), if a parallel reaction scheme is assumed (reaction 13).

NH3

catalyst

> N2

catalyst

x+ N 2 0

catalyst

x-i NO

(4)

Baiker et al. [7] found that the selectivity of the SCR reaction over annealed Moo3 samples highly depends on the grain morphology and the exposed lattice planes. Depending on the preparation procedure, samples with different distributions of exposed lattice planes can be produced. A sample with predominantly exposed (010) planes appeared to be less selective for the formation of N20 in the SCR reaction than a sample exposing more (IOO), (001) and (101) faces. Gandhi et al. [8] investigated the influence of H20 on the performance of copper(I1) molybdate catalysts in the selective oxidation of NH3 to N2. Addition of H20 reduces the activity of these catalysts due to competitive adsorption of H20 on sites active for NH3 adsorption. The selectivity for N2 decreases as well. Thus, Baiker proved that, in addition to its chemical characteristics, the catalytic properties of a compound strongly depend on its surface texture. This simply explains why the preparation conditions affect the catalytic perfomance. It may be expected that with supported catalysts this effect is even more pronounced, because the texture of the supported particles of the catalytically active component strongly depends on the preparation conditions. A number of additional features controls on the performance in the selective oxidation of NH3 to N2 as well. The structure of the catalyst particles, the reduction behaviour, and thermal pretreatment are important parameters. Supported catalysts, such as, WO3flTiO2 and Fe203/Si02, have appeared to be fairly selective in the oxidation of NH3 to N2, but exhibit a poor activity. In this paper, the catalytic performance of a number of silica-supported Moo3 catalysts will be dealt with. The influence on the selectivity of the structure of the catalysts as determined with various characterisation techniques, will be discussed and compared with that of some unsupported Moo3 catalysts.

135

2. EXPERIMENTAL

2.1. Preparation of the Catalysts The MoOg-on-SiO2 catalysts were prepared by deposition of a molybdenum precursor from homogeneous solution onto Si02 (aerosil 200V, Degussa) as described by Geus [9]. To establish a sufficiently high positive interaction between the active phase and the support a precursor of a lower valence was used (MdnCli-), prepared by electrochemical reduction of H2MoO4 in concentrated HCl. It has been well documented that the preparation of Mo03/Si02 catalysts from a hexavalent ammonium heptamolybdate (AHM) precursor leads to the formation of fairly large MoO3 crystallites [lo-141 due to its poor interaction with silica. The merely anionic, dissolved (hexavalent) molybdenum species (Mo70& and MosO& [IS]) have no interaction with the SiO- groups on the surface of the support. Therefore, thus prepared catalysts usually exhibit the catalytic features of bulk MoO3. The interaction of trivalent molybdenum with SiO2 is much better, because of the lower acidity of this precursor. Precipitation of the active phase was brought about by slow injection of a 5% NH3-solution into a vessel containing MdnCli- in an aqueous Si02 suspension. Homogeneity was ensured by vigorous stimng and the presence of baffles in the vessel. When the pH had reached the level of 7.0, the injection was stopped, the slurry was filtered off, and washed with demineralised water. The catalysts were dried in air at 393 K for 16 hours, and calcined in air at 723 K for 72 hours. Unsupported Moo3 catalysts were obtained by precipitation of MoIrlC1i- from a homogeneous solution without suspended Si02 (type I :precipitated) or purchased from Cerac (ultrapure quality) (type 2: annealed). Both unsupported catalysts were calcined at 723 K. 2.2. Characterisation of the catalysts The catalysts were characterised by various techniques: Thermal Analysis (TA), X-ray Photoelectron Spectroscopy (XPS), Raman Spectroscopy, and Extended X-ray Absorption Fine Structure (EXAFS). A short description of the experimental procedures is given. Thermal Analysis experiments were performed within a Mettler TA-2 balance in a 10% HdAr flow (100 mllmin.). A sample of typically 50 mg of the powder was subjected to a temperature program (rate: 5 Wmin.). The weight loss between 673 and 1073 K was used to calculate the degree of reduction of the samples. Raman experiments were executed in a mplemate Spex (1877 model) spectrometer, coupled to an optical multichannel analyzer (Princeton Applied Research, model 1463) equipped with a intensified photodiode array detector. The 514.5 nm line of an Argon laser (Spectra Physics) was used as an excitation source. The sample was pressed into KBr and spun at 2000 rpm. The experiments were done under ambient conditions at a laser power of 10-50 mW. The XPS-experiments were accomplished in a VG-Scientific Escalab MK I1 with a Mg Ka source (1253.6 eV). The position of the peak maxima could be determined with an accuracy of f 0.1 eV. The spectra were corrected for static charging effects with the C (Is) peak as an internal reference. The X-ray absorption spectra in the EXAFS region (Mo K-edge at 19,999 eV) were measured at station 9.2 of the S.R.S. laboratory in Daresbury (U.K.). The storage ring was handled at 2.0 GeV with a ring current of 150-200 mA. A Si(220) crystal was used as the monochromator. Samples were crushed and pressed into a sample holder. Spectra were recorded at 290 K in helium atmosphere.

136

2.3. Kinetic Measurements After calcination the catalysts were pelleted at 400 MPa for two minutes and subsequently crushed and sieved into the fraction range of 0.25-0.50 mm. The amount of catalyst was typically 100 mg. The catalysts were treated in sifu at 673 K in a flow of 25 % 0 2 in helium for two hours prior to the kinetic measurements. Mixtures of NH3/He, NO/He, O N e , and highpurity helium were purchased from Air Products and used without further purification. The experiments were carried out at atmospheric pressure in a fixed bed reactor made of quartz. A Leybolt Q 200 mass spectrometer was used for detection of the reactants and products. The detection limit for the various products was 1 ppm. In some experiments NO was added to the feed to investigate the interference of this compound with the SCO reaction. The performance of the MoO3/SiO2 catalysts was also tested under non-stationary conditions. The catalysts, after prolonged exposure to ambient atmosphere (i.e., hydrated conditions), were submitted to a temperature program of 10 Wmin in a flow of 5000 ppm NH3,2 % 0 2 and 97.5 % He. The LHSV was 12,000 hr-1. 3. RESULTS

3.1. Preparation and characterisation of the catalysts A number of unsupported and silica-supported Mo03 catalysts was prepared according to the procedures described in the ext;aiytal section. Table 1 presents the codes and the metal oxide loadings (defined as w t MpO sio, .loo% ) of the samples, as determined by Inductively Coupled Plasma analysis (ICP). +

Table 1 Properties of the catalysts code Mo (prec.) Mo (ann.) M06 Moll Mo26

catalyst Moo3 Ma3 Mo03/Si@ Mo03/Si@ MwSi@

preparation

HDP' MoO3annealed HDPl HDP' HDP'

loading (%)

-

5.6 11.3 26.0

Homogeneous Deposition Precipitation

The reduction profiles as obtained from TA-experiments are shown i n figure 1. A significant difference in onset temperature of reduction of the two bulk-Mdj samples (d and e) is observed. Although these samples are chemically identical, their reduction behaviour is quite different. X.R.D.-analysis gave no elucidation of a possible difference in exposed lattice planes in the bulk samples. The explanation of the different reduction characteristic can be found in the

137

heterogeneity of the surface of the Moo3 samples. Van den Berg et al. [ 161 have shown that the activity for CO- and Hpoxidation on V2O5 catalysts and vanadium bronzes strongly depends on the concentration of defects in the V205-lattice (usually oxygen-vacancies). The catalytic reaction at high temperature proceeds via a reduction-oxidation mechanism, in which oxygen vacancies are the active species. Thus, it is feasible that a high concentration of defects in an oxidic lattice like V2O5 or MOO3 locally destabilises the lattice and thus enhances the activity for an oxidation reaction. Since the TA-experiment in H2-atmosphere can be envisioned to be an oxidation of H2 by the catalyst, the concentration of surface defects will control the onset temperature of the reduction of the catalyst. The lower onset temperature of Mo(prec.) can thus be explained by its higher content of defects due to the preparation procedure: precipitation of molybdenum (hydr)oxide from an aqueous solution yields a porous, badly crystallised solid even after calcination. The heterogeneity of Mo(prec.) is considerably higher than that of Mo(ann.), which contains annealed, well crystallised, stoichiometric Mo03 particles.

I

I

I

I

I

I

l

l

I

I

I

I

I

1

I

I

313

413

513

613

113

613

913

1013

T (K)

Figure 1 Themogravimetric analysis of a) Mo6, b) Moll, c) Mo26, d) Mo (prec.), and e) Mo (ann.) in a 10 % HdAr flow.

138

The onset temperatures of Mo6, Moll, and Mo26 (a-c) are equal, hut the shape of the peaks and the integrated weight loss deviate strongly. M o l l and M026 both exhibit a second peak at high temperature, which corresponds to deep reduction to metallic molybdenum, as evidenced by High Temperature XRD experiments. Mo6, however, has no peak at high temperatures and can be reduced less profoundly. with XPS, it was checked , that in all samples molybdenum is initially present in 6+ oxidation state. The Mo (3d5I2) peak is located at 232.8 eV. The reference values of the binding energy of the photoelectrons in Moo3 and Moo2 are 232.5 and 229.2 eV respectively. From the TG and XPS results we have calculated that Mo6 can be reduced only for 35%, whereas Moll and Mo26 can almost completely be reduced. The Raman spectra of Mo6, Moll, M026, and Mo (ann.) are represented in figure 2. The main bands of Mo (ann.) are positioned at i j = 996, 821, 668, 285 and 159 cm-l; This corresponds with the literature values of bulk Moo3 [17]. Obviously M o l l and Mo26 contain crystalline M003. Mo6, however, has a distinctly different structure: the bands at V= 944 , 880 and 220 cm-1 indicate the presence of oligomeric molybdenum oxide clusters (comparable to aqueous hepta/ octamolybdate) [18]. The Raman bands at 480 and 370 cm-l are due to SiO2 and a surface molybdenum oxide respectively.

x

,

1200

lm,

,w,

SIX

tnu

4 0

2Mi

Wavenumbers (cm-*)

Figure 2 Raman spectra of a) Mo (am.), b) Mo26, c) Moll, and d) Mo6, recorded at ambient conditions

139

The results of a preliminary EXAFS study reveal analogous results. The amplitude of a k1 fourier transform of the X(k) is shown in figure 3. The Mo-Mo shell in Mo(ann.) is positioned at approx. 3.5 A and originates from a strong Mo-Mo contribution, diagnostic for the existence of larger molybdenum oxide particles with long range order. The Mo-0 peak at 1.5 - 2.2 A is ascribed to more than one Mo-0 distance due to distortion of the octahedral environment of the absorber. The same pattern is observed for Moll and Mo26: The Fourier transform of Mo6, however, deviates dramatically from the bulk reference. It reveals hardly any long range order, as evidenced by the decrease of the Mo-Mo contribution at -3.5 A. The combined results of the characterisation of the catalysts show that M o l l and M026 contain crystalline molybdenum(V1) oxide particles, highly dispersed onto the Si02 support, whereas M06 consists of very small molybdenum oxide clusters strongly interacting with the support. The different reduction behaviour of Mo6 is presumably caused by the strong interaction with the support: the tiny molybdenum oxide particle cannot be reduced to metallic molybdenum. Consequently, the reduction behaviour, i.e., the onset temperature and the maximum degree of reduction depend on the structure of the catalyst.

Figure 3 EXAFS kl fourier transforms of X(k) (3.45A-1 < k < 12.9 A-') of a) Mo (ann.), b) Mo26, c) Mo 11, and d) Mo6

140

3.2. Kinetic experiments Ammonia oxidation experiments were performed over Mo (ann.) and Mo (prec.) in the presence of NO. The concentration profiles as a function of the temperature are shown in figure

4.

a NO h

,a

500-1

500

400-

400

300-

300

zoo:

200

100-

100

a

.-

2 58

I

V

O I

0

473

573

613

Figure 4 Concentration profiles of NH3, NO, N2, N2O and H20 in the NHg-oxidation over a) Mo (ann.), and b) Mo (prec.) Despite the chemical equivalence of the two samples, their catalytic performance is quite different. The NH3-conversion at 673 K of Mo (ann.) is somewhat higher than Mo (prec.). The selectivity towards N2, however, is almost 100% up to 673 K for Mo (prec.), whereas Mo (ann.) produces a considerable amount of N2O. Moreover, NO is reactive with Mo (ann.) and remains unaffected with Mo (prec.). It is interesting to note that the minimum for NO in Mo(ann.) is associated with a maximum in N2. Apparently the formation of NO and N2 from NH3 is non-independent. The results of the catalytic performance tests of the MoO3/Si02 catalysts are presented in figure 5. The selectivity of Mo6 strongly deviates from M o l l and Mo26. Mo6 produces a substantial amount of nitrous oxide (33% of the amount of NH3 converted), whereas M o l l and Mo26 produce only 8 and 5% N20 respectively. With Mo6 formation of NO proceeds at temperatures above 773 K, which is accompanied by a decrease of the selectivity for N2O. This observation suggests that the formation of N2, N20, and N2 is interdependent, and that the reaction mechanisms for the formation of these products do not run completely parallel.

141

b

a

la C

.-

f

8 I

413

C

j LoJ; I

1

I

I

I

I

I

1

513

613

413

513

613

413

513

613

T (K)

T (K)

-

T (K)

Fig. 5 formation of N2, N2O and NO over a) Mo6, b) Moll, and c) Mo26 These measurements were performed on fresh, calcined samples and, thus, contain adsorbed H20 due to prolonged exposure to the ambient atmosphere. The selectivity of the catalysts drastically increases after a thermal pretreatment in 0 2 at 673 K. Table 2 shows the selectivities of the catalysts for the production of N2 at 673 K before and after thermal treatment.

Table 2 Selectivities to N2 at 673 K catalyst

M06 Moll Mo26

fresh 67 92 95

selectivity after pretreatment >98 >98 >98

The origin of the low selectivity of the fresh samples will next be discussed. 4. DISCUSSION

The reduction behaviour of Mo (ann.) and Mo (prec.) exhibits large differences. It has been mentioned that the onset temperature for the reduction of the samples strongly depends on the

142

concentration of lattice defects present at the surface of catalyst particles and the ability of the oxide to accommodate vacancies. The different catalytic performance of the samples in the oxidation of NH3 can be explained along the same lines. The formation of each N2-molecule requires two nitrogen atoms from different NHg-molecules, which implies that a NHj-molecule has to adsorb dissociatively in close proximity of another NH3. The mobility of adsorbed Nspecies is low, as reported for V2O5JTiQ catalysts [19]. The recombination probability of two adsorbed nitrogen atoms to form an N g bond is thus determined by the surface density of active sites. Active sites for this reaction are oxygen atoms capable of accepting the hydrogen atoms from NH3, i.e., reducible under the conditions of the reaction. Mo (prec.) contains a high number of active sites for the SCO reaction, as evidenced by the low onset temperature of reduction in the TA-profile (figure 1). The selectivity to N2 is optimal, because of the high probability of recombination of two nitrogen atoms to form N2. The formation of nitrous oxide in Mo (ann.) is not caused by the sequential oxidation of evolved N2 to N20 as discussed before (reaction 4).Golodets et al. [4]explain the formation of N20 by assuming the reaction between two HNO species to form N2O and H20. A number of species is schematically postulated in the model in figure 6. Each unit on the surface ('cube') represents a surface oxygen atom, without taking specific lattice planes into account. An oxygen vacancy is symbolised by a pit. The adsorption of a NHymolecule in (I), or remote (2) from an oxygen vacancy is depicted in this model. We assume that an NHg-molecule is adsorbed dissociatively on the Moo3 surface ('stripped') and tends to occupy an oxygen vacancy, if present. Adjacent surface oxygen atoms are transformed into OH-groups. Two vicinal OH groups ( 3 ) can form an oxygen vacancy under evolution of HzO (4). The lattice oxygen can be replenished by gas phase 0 2 . Species 5 visualises the situation of Mo (prec.): two NH3-molecules are adsorbed and stripped in proximity (hydrogen atoms not shown) and are evolved as N2 (6). When the concentration of active sites is low, then the nitrogen atoms remain isolated, either on the surface of Moo3 ( 2 ) or in a vacancy ( I ) . In this case the recombination probability of nitrogen atoms is low and development of NO becomes likely (7). Two pathways for evolved NO are conceivable. It can react with another nitrogen species to form N20 (7-8)or it is swept out of the reactor (at high temperatures). The high selectivity for N20 of Mo (ann.) can thus be explained, because Mo (ann.) contains few, remote vacancies. Implicitly, we have postulated a reaction path opposite to reaction 4,as presented in reaction 5. NH3

(O)

NO) -

(N)

N20

(

)

N2

+ (0)

(5)

The last step of reaction 5 is the subsequent decomposition of N 2 0 occurring preferentially on oxygen vacancies. The oxygen atom from N20 can be conveniently accommodated in the bond can be formed. Shelef et al. [20] report MoOj-lattice, while the energy-favourable on the formation and decomposition of N20 on chromia catalysts. At high space velocities the evolved N 2 0 cannot decompose and is swept out of the reactor. Keenan et al. [21] and Vorotyntsev et al. [22] confirm that the decomposition of N20 takes place on surface vacancies.

143

Figure 6 Schematic representation of surface species

The performance of the MoO3/SiO2 catalysts is strongly affected by the presence of adsorbed H20. The molybdenum oxide phase and the silica are both hydroxylated. Since H20 is competes with NH3 for the adsorption on active sites [8],a smaller number NH3 molecules can be adsorbed and stripped in the presence of H20. Water leading to adsorbed hydroxyl

144

groups thus diminishes the concentration of adsorbed nitrogen species, which decreases the recombination probability of nitrogen atoms and enhances the selectivity towards N20. This effect is far more pronounced with Mo6 than with M o l l and MoZ6, because of the different structure of the catalysts. The oligomeric molybdenum oxide clusters in Mo6 can only be reduced for approx. 35%. Only a limited number of NH3-molecules can therefore be stripped, since the adsorption of each NH3 molecule requires the reduction of at least three oxygen atoms (formation of OH-groups). The recombination probability of nitrogen atoms and the rate of decomposition of N 2 0 are much higher on M o l l and Mo26 under hydrated conditions as compared to Mo6. The larger Moo3 crystallites in M o l l and Mo26 can more easily accommodate oxygen vacancies, because of their higher reducibility.

5 . CONCLUSIONS The activity and, more importantly, the selectivity of (un)supported Moo3 in the selective oxidation of NH3 to N2 appears to depend on the defect structure of the catalysts. The catalytic performance of the two unsupported Moo3 samples is controlled by the preparation procedure, which installs the reduction-oxidation properties and the selectivity to N2. It is postulated that the selectivity for N2 is influenced by the recombination probability of adsorbed nitrogen atoms. The probability is high for catalyst particles with a high surface density of active sites. An active site for the selective oxidation of NH3 is an ensemble of reducible surface oxygen atoms in the vicinity of an oxygen vacancy. The oxygen atoms must strip the hydrogen atoms from NH3 to form OH groups. A vacancy is created when two vicinal OH groups desorb under release of H20. Surface oxygen is replenished by gas phase 0 2 . An adsorbed and stripped NH3-molecule can react with another stripped nitrogen atom or immediately with a gas phase NHj-molecule. Formation of the energetically favourable N g bond is the driving force for the recombination of two adsorbed nitrogen species. When no other nitrogen atom is in close proximity, NO can desorb. N 2 0 results when a NO molecule reacts with another isolated nitrogen atom. The participation of NO is demonstrated by addition of NO to the feed. The Mo ( a m ) sample exhibits SCR behaviour. With isotopic experiments Janssen [19] proved that in the SRC reaction over vanadia catalysts the nitrogen atoms in N2O originate from NO and NH3. Formation of the undesired N2O does not present problems, provided it consecutively decomposes under formation of N2 and an adsorbed oxygen atom. We believe that the selectivity of these catalysts is partly due to the capability to decompose undesired N 2 0 , which proceeds preferably at oxygen vacancies. Since NH3 competes with H20 for the active sites [S], the catalytic performance of hydrated samples is worse than that of thermally pretreated samples. The selectivity is significantly enhanced by a thermal pretreatment. The increase is most prominent for Mo6. Due to the small dimensions and poor reducibility of the oligomeric molybdenum oxide clusters only a limited number of NH3-molecules can be stripped at the surface, thus decreasing the recombination probability. Decomposition of evolved N2O cannot proceed because the required vacancies are occupied by water. The larger Moo3 crystallites in M o l l and Mo26 can adsorb and strip more NH3 molecules and accommodate more vacancies for the decomposition of released N20.

145

6. ACKNOWLEDGEMENT We thank I.E. Wachs and M. Vuurman for the performance of the Raman experiments.

7. REFERENCES 1. 2. 3.

4. 5.

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.

E.H.T.M. Nijpels, G.J.M. Braks, VROM 9031918-89 6974/118 (1989) 1 H. Bosch, F.J.J.G. Janssen, Catal. Today, 2 (1988) 369 E.T.C. Vogt, A. Boot, A.J. van Dillen, J.W. Geus, F.J.J.G. Janssen, F.M.G. van den Kerkhof, J. Catal., 114 (1988) 313 G.I. Golodets Heterogeneous catalytic reactions involving molecular oxygen in: Studies in Surface Science and Catalysis (J.R. Ross ed.), Amsterdam, Elsevier 1983 p.312 I. Barin, 0. Knacke, 0. Kubaschewski, Thermochemical properties of inorganic substances, Berlin, Springer Verlag 1977 P. Mars, D.W. van Krevelen, Spec. suppl. to Chem. Engin. Sci., 3 (1954) 41 A. Baiker, P. Dollenmeier, A. Reller, J. Catal., 103 (1987) 394 H.S. Gandhi, M. Shelef, J. Catal., 40 (1975) 312 J.W. Geus, Production and thermal pretreatment of catalysts in: Studies in surface science and catalysis 16 (G. Poncelet, P. Grange, P.A. Jacobs (eds.)) Amsterdam, Elsevier 1983 p.1 T. Ono, M. Anpo, Y. Kubokawa, J. Phys. Chem., 90 (1986) 4780 M. Anpo, M. Kondo, Y. Kubokawa, C. Louis, M. Che, J. Chem. SOC., Faraday Trans. I., 84 (8) (1988; 2771 N. Kakuta, K. Tohji, Y. Udagawa, J. Phys. Chem., 92 (1988) 2583 A. Latef, R. Elamrani, L. Gengembre, C.F. Aissi, S. Kasztelan, Y. Barbaux, M. Guelton, Zeitschr. fur Physikal. Chem. Neue Folge, 152 (1987) 93 T-C. Liu, M. Forissier, G. Coudurier, J.C. Vtdrine, J. Chem. SOC.,Faraday Trans. I, 85 (7) (1989) 1607 C.F. Baes, R.E. Mesmer, The hydrolysis of cations, New York, J. Wiley & Sons 1976 J. van den Berg, A.J. van Dillen, J. van der Meyden, J.W. Geus in Surface Properties and Catalysis by Non-Metals (J.P. Bonnelle et al. (eds)) (1983) H.M. Ismail, C.R.Theochxis, D.N. Waters, M.I. Zaki, R.B. Fahim, J. Chem. SOC.,Faraday Trans. 1,83 (1987) 1601 J. Aveston, E.W. Anacker, J.S. Johnson, Inorg. Chem., 3 (1964) 735 F.J.J.G. Janssen, F.M.G. van den Kerkhof, H. Bosch, J.R.H. Ross, J. Phys. Chem., 91 (1987) 5921 M. Shelef, K. Otto, H. Gandhi, J. Catal., 12 (1968) 361 A.G. Keenan, R.D. Iyengar, J. Catal., 5 (1966) 301 V.M. Vorotyntsev, V.A. Shvets, V.B. Kazanskii, Kinetika i Kataliz, 12 ( 5 ) (1971) 1249


P. Ruiz and B. Delmon (Eds.)

147

New Developments in Sclcctive Oxidation by Heterogeneous Catalysis Studies in Surface Science arid Catalysis, Vol. 72,pp. 147-154 0 1992 Elscvier Scicnce Publishers B.V. All rights reserved.

HIGH PERFORMANCE OF VANADIA CATALY!iWS SUPPORTED ON Ti0,-COATED SILICA FOR SELECI'IW OXmATION OF ETHANOL

N.E. Quaranta', V. CortCs Corbergn and J.L.G. Fierro Instituto d e Catalisis y Petroleoquimica, C.S.I.C. Campus de la U.A.M., Cantoblanco, 28049 Madrid, Spain Phone ( t34-1) 5852626, Fax ( t34-1) 5852614.

' On leave from CIC, CINDECA, La Plata, Argentina ABSTRACT The effect of coating the SiO, support with TiO, on the properties of V,O,-SO, catalysts, and the use of vanadia catalysts supported on Ti0,-coated SiO, for the selective oxidation of ethanol to acetaldehyde has been studied. SiO, samples were coated with TiO, by homogeneous deposition-precipitation, for which X-ray photoelectron spectra showed that the most effective titanium dispersion was reached for the low TiO, coverages. The catalysts were prepared by depositing vanadia, in quantity equivalent to a monolayer, onto SO,, TiO,, or Ti0,-coated SO,, by wet impregnation with ethanolic solution of vanadyl acetylacetonate, drying at 348 K, and calcining at 773 K in air. This impregnation decreased the specific BET surface by a 20%. TPR profiles of binary and ternary samples were similar showing only one peak, sugesting the presence of just one vanadium species. Catalytic activity tests for ethanol selective oxidation at 400-700 K, showed that Ti0,-coating produced a sensible improvement on both activity and selectivity of V,O,-SiO, catalysts. Presence of water in the feed increased both activity and selectivity to acetaldehyde of the ternary sample. 1. INTRODUCTION

The interaction between catalytically active metal oxides particles and oxidic supports greatly influences their structure and size. This makes the deposition of monolayer(s) of an active phase on a carrier an attractive technique for the tailoring of catalytic properties. It offers the advantage of an increased exposure of the active phase, and allows to modify its structural and catalytic properties by interaction with the carrier. Vanadia catalysts constitute a relevant example of the influence of this interaction. Several authors [ 1-31 have compared the properties of vanadia supported on different carriers (SiO,, Al,O,, TiO,, MgO, ZrO,) and have concluded that the nature of the dispersed surface metal oxide phase depends on the specific supported metal oxide/support system. This may explain why, depending on which support is used, vanadia becomes an effective catalyst for selective oxidation of aromatics [4-61, olefins 171, and alcohols [S-lo], as well as for selective catalytic reduction (SCR) of NO, by NH, [ll]. Both V,O,-TiO, [8,9] and V,O,-SO, [lo] catalysts have been studied for selective oxidation of alcohols. Titania (anatase) interacts strongly with the first immobilized vanadia layer, which allows to generate a molecular dispersion of V,O, oxide layer [12,13], but suffers of limited

148

specific area and low resistance to sintering. Conversely to TiO,, interaction of V,O, with SiO, is weak and, therefore, the properties of VO, species over SiO, are modified to a lesser extent, showing a higher tendency for thermally induced aggregation, leading to a low dispersion of the active phase [14,15]. But the use of SiO, as a support of vanadia has the advantage of a higher specific area and a bigger resistance to sintering than those of TiO,. A way to obtain a titania surface with high, thermostable surface area and good mechanical properties is to apply TiO, onto silica [ 161. Vanadia catalysts supported on mixed compounds TiO,/SiO,, either coprecipited [17] or having the TiO, supported onto the silica [18,19] have been developped because of their selectivity as catalysts for SCR of nitrogen oxides. On this basis, the present work has focused mainly on the preparation and characterization of catalysts V-Ti-Si-0, with the aim to understand the interrelations between TiO, and SiO, that can be important at low titanium content, and their influence on the catalytic properties of supported V,O,, as a way to improve its perfomance for the selective oxidation of ethanol. This reaction has deserved recently an increasing technological interest as an important step in the use of biomass as a chemical resource. A number of oxides a r e active for this reaction [20-221, but the main product depends on the specific system involved. V,O, [20] and V,O,/SiO, [ 101 a r e very active and selective to acetaldehyde. 2. EXPERIMENTAL 2.1. Preparation of the TiO,/SiO, supports Ti-Si-0 supports were prepared by homogeneous precipitation, as described by Geuss et al. [23]. Silica (Aerosil MOX 80 from DEGUSSA, specific area: 86 m’/g) was suspended in deionized water, and acidified with HCI to p H < 1. An appropriate amount (see below) of TiCl, (MERCK, 15% solution in HCI) was added to the suspension with vigorous stirring to ensure a good homogeneity during the precipitation. Then, the suspension was neutralized up to a pH ca. 8 by slow addition of a solution of ammonium hydroxide (MERCK, 20% in NH,) at a neutralization rate of 0.002 mol OH-/min. The precipitate so obtained was throughfully washed twice with deionized water and dried in air at 373 K for 48 h. The solid was finally calcined at 823 K in air for 2 h. This procedure allowed to deposit hidrated Ti(TT1) oxide, and, after the thermal treatments to obtain TiO, dispersed on the silica surface. Four supports with TiO, amounts equivalent to 0.6, 1.0, 1.5, and 2.0 theoretical monolayers (denoted hereafter as O.GTS, lTS, 1.5TS and 2TS, respectively) were prepared. The monolayer of TiO, was considered as the complete recovering of the silica surface by a film of TiO, 0.38 nm thick, which corresponds to the longest axis of the rutile cell. 23. Preparation of catalysts Catalysts were prepared using three different supports: SiO,, TiO, and 1TS; the later was selected according to XPS results which showed the most effective dispersion for low titania coverages (see below). V,O, was deposited on them by wet impregnation, by using the specific reaction of the surface hydroxyls with the vanadyl groups of the vanadium(1V) acetylacetonate complex [24-251. The impregnation was made by adding an ethanolic solution of the metallic acetylacetonate to the support particles. The suspension was evaporated at 348 K under continuous stirring. The impregnates were washed repeatedly with pure ethanol, dried again, and finally calcined a t 773 K in air for 2 h. The added amount of vanadium was that calculated for a complete monolayer, taking into account the specific surface areas. In the case of TiO, and 1TS a value of 0.166 nm2 per center was considered, assuming that one vanadium oxide species is deposited on each center [24]. In the case of SO, the needed amount was calculated for a monolayer thickness of 0.234 nm [4] and the density of V,O, of 3.357 g The catalysts will

149 be denoted henceforth by their component elements (V-Ti, V-Si, V-Ti-Si). 23. Characterization of catalysts and supports

a) SDecific areas: Specific surface areas were determined by the B.E.T. method from the adsorption isotherms of nitrogen at 77 K, taking a value of 0.164 nm2 for the cross-sectional area of the adsorbed nitrogen molecule. b) X-ray photoelectron spectroscopy (XPS): XP spectra were obtained with a Leybold LHS 10 spectrometer provided with a hemispherical electron analyzer and a Mg anode X-ray excitating source (MgKa = 1253.6 eV). Samples were pumped to 10” Torr (1 Torr = 133.33 Nm”) before moving them into the analysis chamber. Pressure in this turbo-pumped main vacuum chamber was maintained below 7x10” Torr during data acquisition. Each spectral region was signal averaged for a number of scans to obtain good signal-to-noise ratios. Accurate binding energies (BE) were determined by reference to the T i 2p3,2 and Si2p lines to which arbitrary BE values of 458.5 and 103.4 eV were assigned [26]. These references gave BE values consistent with those calculated respect to Cls line at 284.6 eV. c) Temperature Dromammed reduction (TPR): The TPR profiles were obtained with a Cahn 2000 microbalance (sensitivity = 1 pg). Prior to the experiment, the samples were heated in a He flow (1.67~10”I/s) at 0.067 K/s up to 773 K in order to clean the surface. After cooling in He flow down to 373 K, He was substituted by H, ( 5 ~ 1 0I/s) . ~ and the experiment began, heating the sample at 0.067 K/s up to 800 K, while continuously recording the weight changes. d) Chemical analvsis: Quantitative analysis of vanadium of catalyst samples was made by atomic absorption spectrometry. Samples weighing ca. 0.1 g were dissolved in 10 ml of HF and heated on a sand bath until complete solution, and then diluted to 100 ml with deionized water. 2.4. Catalytic activity Catalytic activity tests were made in a tubular fixed bed flow reactor at nearly atmospheric pressure in the temperature range 400-700 K, with residence time W/F= 45 g cat.h /mol ethanol, and reacting mixtures ethanol-oxygen-helium, with or without added water, having compositions (in mole%): ethanol 1.4, oxygen 27.5, water 0 or 9.3, and helium balance. Ethanol (Prolabo p.a., 99.85 vol%), oxygen (SEO, 99.99%) and helium (SEO, 99.99%) were used as reactants and dilutant, respectively. Catalyst samples (ca. 250 mg, particle size 30-40 mesh) were diluted with S i c tips up to a bed volume of 5 ml. Reactants and products were analyzed on-line by G C using two columns: 13X molecular sieve for 0, and CO, and Porapak Q for the rest of compounds. C and 0 mass balances of 10025% were obtained. Conversion and selectivity to products were calculated on a carbon atom basis, expressed as mole% of ethanol transformed to ethanol fed, and of ethanol transformed to each product to total ethanol transformed, respectively. 3. RESULTS

3.1. Characterization of supports and catalysts Table 1 shows the specific surface areas of supports and catalysts samples. As it can be seen, the surface area of SiO, support increased when TiO, was deposited on it. Prior to the incorporation of vanadia, the Ti0,-coated silica carriers were examined by XPS. The binding energies (BE) of 01s and Ti2p peaks (532.9 and 464.5-458.7 eV, respectively) remained

150

TABLE 1 Characterization of catalysts Sample

sBET

(m’/g>

-

~. ~

V-Ti V-Ti-Si V-Si

V content (wt%) Theoretical Analysis

50 (51) (*I 92 (112) 77 (86)

2.56 5.17 3.39

2.41 20.01 4.19+ 0.01 2.53 2 0.01

TPR Weight loss (%) 0.77 1.35 0.62

T, (K) 665 663 688

___ (*)

S,,

of the corresponding support.

essentially unchanged for all samples. However, the titanium-to-silicon intensity ratios calculated from XP spectra increased progressively with increasing Ti0,-loading. As it can be seen in Fig. 1, the titania dispersion was indeed very high for the 0.6TS and 1TS samples, fitting well with that expected for the theoretical monolayer [27] (Fig. 1, dashed line), while it decreased for the 1.5TS and 2TS samples. In agreement with these results, the ITS sample was selected as a carrier for the vanadium oxide (catalyst sample V-Ti-Si). Incorporation of a vanadia monolayer caused a decrease of the specific surface areas of every support between 0.5 to 20% (Table 1). Chemical analysis of the catalysts (Table 1) showed that the amount of vanadium retained in sample V-Ti was close of that calculated for a V,O, monolayer, while only a 74% of this amount remained actually on the V-Si sample, and a slightly higher amount (82%) in sample V-Ti-Si. The three catalyst samples were also examined by XPS, and the respective BE of V2p, Ti2p, and/or Si2p peaks are compiled in Table 2 for

Fig. 1.- Titanium-to-silicon intensities surface ratio of Ti0,-coated silica supports vs. bulk composition (dashed line corresponds to the theoretical monolayer). Fig. 2.- TF’R profiles of supported vanadia catalysts: a) bulk V,O,, b) V-Si, c) V-Ti, d) V-Ti-Si.

151

TABLE 2 XPS Data of Vanadia-Containing Catalysts BE (eV) Sample

2P

V-Ti V-Ti-Si V-Si

516.5 517.5 518.2

Ti 2P3/2

lV/(lTi+lSi)

458.5 458.5

0.197 0.293 0.288

Bulk V/(Ti+Si) 0.0395 0.0555 0.031 1

comparative purposes. From these data it is clear that the BE of V2p3/*peak markedly depends on the type of carrier. The decrease of 0.7 eV of the V-Ti-Si catalyst respect to that of the V-Si counterpart agrees well with literature findings [28,29]. This has been interpreted as due to the electron withdrawing effect of the silica carrier and the subsequent increase of the electrostatic character of the surface bonded V,O, layer. Such an effect seems to be inhibited when a Ti0,-coating covers the silica substrate. Following this reasoning one would expect similar BE values for the V2p3,, peak in V-Ti-Si and V-Ti, and lower than in V-Si catalyst; however, the V-Ti catalyst showed a shift of 1.0 eV towards lower BE. This can be due to a larger extent of reduction of surface V5' ions upon exposure to X-ray radiation. This hypothesis is reinforced by t h e strong grey bluish color of V-Ti catalyst after XPS analysis, which contrasts with the yellow and pale yellow colour of V-Si and V-Ti-Si samples, respectively. Table 2 also shows that V,O, dispersion depended on the carrier. The V/(TitSi) intensity ratio is high, and almost equal, for V-Ti-Si and V-Si catalysts while it is lower by ca. 33% for V-Ti. TPR profiles of the catalysts showed only one reduction peak, alike bulk V,O, (Fig. 2). The observed differences between the temperature of maximum reduction rate (T,) of bulk V,O, and supported on SiO, and TiO, have been previously reported [17]. T, of the ternary sample was similar to that of V-Ti sample. Weight loss in the reduction step of V-Ti and V-Ti-Si samples corresponded to reduction to V,O,, as in bulk V,O,, while that of V-Si sample was intermediate between those corresponding to reduction to the lower oxides V,O, and V,O,. These findings agree well with those of Baiker et al. [17] and Rajadhyaksha et al. [19] who reported a higher reduction degree of vanadia when supported on Ti0,-SiO, than on SO2. 3.2. Catalytic activity Acetaldehyde was the main product of ethanol oxidation on each catalyst, with acetic acid, ethene, carbon oxides and ethyl acetate as minor products. The variation of activity of the catalysts with temperature and added water is shown in Fig. 3. In the conditions used, the activity of V-Ti-Si catalyst was rather superior to that of V-Si sample, and close to that of V-Ti sample. Arrhenius plots for V-Ti and V-Si samples produced single straight lines with apparent activation energies of 20 and 56 kJ/mol, respectively, while with V-Ti-Si two lines, breaking at ca. 500 K, were obtained: the apparent activation energy was 51 kJ/mol below 500 K, and 17 kJ/mol above 500 K. Addition of water to the feed, which caused a decrease in the activity of V-Ti sample, had little or no effect on the total conversion over V-Si. V-Ti-Si behaviour was similar to that of V-Si, although a small increase of conversion was observed at high temperatures. But the most interesting differences caused by water addition appeared in the evolution of selectivity with total conversion (Fig. 4). When pure ethanol was fed, selectivity of V-Ti-Si was similar to that of V-Si, while that of V-Ti was lower, specially at medium and high conversions, due to greater formation

152

T (K) Fig. 3.- Ethanol oxidation on supported catalysts in presence (empty symbols) o r absence (full symbols) of water in the feed. Symbols: A , V-Ti; 0 , V-Si; 0 , V-Ti-Si; 0 , homogeneous reaction; dashed line corresponds to data of 9.8 % V,O,/SiO, catalyst from Ref. [lo]. Experimental conditions in text. I

1

Fig. 4.- Selectivity to acetaldehyde as a function of total ethanol conversion. Symbols and experimental conditions as in Fig. 3.

153

of the main by-products, acetic acid and carbon oxides. Addition of water had no effect on the selectivity of V-Ti and V-Si, but caused a strong selectivity improvement on V-Ti-Si, allowing to reach yields as high as 60% at conditions where only yields ca. 40% were reached without water. This effect is due to inhibition of acetic acid formation, which is absent among the products when water is fed. 4. DISCUSSION

The increase of the surface area of silica, as well as XPS quantitative measurements, indicate that TiO, was deposited essentially in monolayer on sample 1 s . The decrease of the specific surface areas of SiO, and 1TS supports by deposition of V,O, indicates that a fraction of vanadia particles were blocking the pores remaining in the Ti0,-free SiO, surface. Nevertheless, T, in TPR and XPS results agree to point out that VzO, interacts mainly with the TiO, surface, thus making V,O, on 1TS to behave like on bulk TiO,. This can explain the close similarity between total activity of V-Ti and V-Ti-Si. Presence of only one TPR peak suggests the presence of a unique vanadium species, but T, values indicate differences in its reducibility depending on the support. In line with this, the change in the apparent activation energy of V-Ti-Si, from that of V-Si below 500 K to that observed for V-Ti above 500 K, suggests again a different interaction degree of vanadia with the supports. This agrees with previous findings of Vogt et al. [30] who observed that the activation energies for CO oxidation on supported vanadia catalysts is a good analytical device for monitoring the interaction between the active vanadia species and the support. It is noteworthy that most of the works in the literature report the oxidation of pure ethanol, and little or no attention has been paid to the influence of water, although biomass-based processes would deal with ethanol in dilute aqueous solutions. Only Oyama et al. [lo] reported the activity of V,O,/SiO, catalysts in the presence of water. Our results with V-Si agree weU with those of Oyama, taking into account a possible effect of heat transport for the particle size here used. The present results confirm that Ti0,-coating of SiO, support surface for supported VzO, catalysts allows to improve not only the cost of the catalyst but also its performance in the presence of water, which makes vanadia supported on Ti0,-coated SiO, to be specially suited for selective oxidation of ethanol in diluted aqueous solutions. 5. REFERENCES 1

2 3 4

5 6 7

Y. Murakami, M. Inomata, K. Mori, T. Ui, K. Suzuki, A. Miyamoto and T. Hattori, in " Preparation of Catalysts 111" (G. Poncelet, P. Grange and P.A. Jacobs Eds.), Elsevier, Amsterdam 1983, p. 531. J. Kijenski, A. Baiker, M. Glinski, P. Dollenmeier and A. Wokaun, J. Catal., 101 (1986) 1. I.E. Wachs and F.D. Hardcastle, Proc. 9th Int. Congr. Catal., Calgary 1988 (Ed. M. J. Phillips and M. Ternan), Vol. 3, p. 1449. B. Jonson, B. Rebenstorf, R. Larsson, S.L.T. Andersson, J. Chem. SOC.,Faraday Trans. I, 84 (1988) 1897. G.C. Bond and K. Briickman, Faraday Disc. Chem. SOC.,72 (1981) 235; G.C. Bond and P. Konig, J. Catal., 77 (1982) 309. K. Mori, M. Inomata, A. Miyamoto and Y. Murakami, J. Phys. Chem., 87 (1983) 4560. J.M. Ldpez Nieto, G. Kremenic and J.L.G. Fierro, Appl. Catal., 61 (1990) 235.

154

8

9 10

11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

A. van Hengstum, J.G. van Ommen, H. Bosch and P.J. Gellings, Proc. 8th Int. Cong. Catal., Berlin 1984, Vol. 4, p. 297. G.C. Bond and S . Flamerz, Appl. Catal., 33 (1987) 219. S.T. Oyama, K.B. Lewis, A.M. Carr and G.A. Somorjai, Proc. 9th Int. Congr. Catal., Calgary 1988 (Ed. M. J. Phillips and M. Ternan), Vol. 3, p. 1489. H. Bosch and F. Janssen, Catal. Today, 2 (1987) 369. D.J. Cole, C.F. Cullis and D.J. Hucknall, J. Chem. SOC.,Faraday 1, 72 (1976) 2744. M. Gasior, I. Gasior and B. Grzybowska, Appl. Catal., 10 (1984) 87. F. Roozeboom, M.C. Mittelmeijer-Hazeleger, J.A. Moulijn, J. Medema, V.H.J. de Beer and P.J. Gellings, J. Phys. Chem., 84 (1980) 2783. M. Takagi, M.Soma, T.Onishi and K. Tamaru, Can. J. Chem., 58 (1980) 2132. E.T.C. Vogt, A. Boot, A.J. Van Dillen, J.W. Geus, F.J.J.G. Janssen, and F.M.G. van der Kerkhof, J. Catal., 114 (1988) 313. A. Baiker, P. Dollenmeier, M.Glinski and A. Reller, Appl. Catal., 35 (1987) 365. M.G. Reichmann and A.T. Bell, Langmuir, 3 (1987) 111; idem., Appl. Catal., 32 (1987) 315. R.A. Rajadhyaksha, G. Hausinger, H. Zeilinger, A. Ramstetter, H. Schmelz and H. Knozinger, Appl. Catal., 5 1 (1989) 67. L. Wang, K. Eguchi, H. Arai and T. Seiyama, Chem. Lett., (1986) 1173. M. Hino and K. Arata, J. Chem. SOC.,Chem. Comm., (1988) 1168. T. Nakajima, K. Tanabe, T. Yamaguchi, I. Matsuzaki and S. Mishima, Appl. Catal., 52 (1989) 237. J.W. Geus, in " Preparation of Catalysts 111" (G. Poncelet, P. Grange and P.A. Jacobs Eds.), Elsevier, Amsterdam 1983, p. 1. J.G. van Ommen, K. Hoving, H. Bosch, A.J. van Hengstum and P.J. Gellings, Z. Phys. Chem. Neue Folge, 134 (1983) 99. A.J. van Hengstum, J.G van Ommen, H. Bosch and P.J. Gellings, Appl. Catal., 5 (1983) 207. C.D. Wagner, W.M. Rigs, L.E. Davis, J.F. Moulder and G.E. Muilenberg, in "Handbook of X-ray Photoelectron Spectrsocopy", Perkin-Elmer Co., Eden Prairie, Minnesota, 1978. F.P.J.M. Kerkhof and J.A. Moulijn, J. Phys. Chem., 83 (1979) 1612. J.L.G. Fierro, L.A. Gambaro, T.A. Cooper and G. Kremenic, Appl. Catal., 6 (1983) 363. B. Horwath, J. Strutz, J. Geyer-Lippmann and E.G. Horwath, Z. Anorg. Allg. Chem., 483 (1981) 181. E.T.C. Vogt, M. de Boer, A.J. van Dillen and J.W. Geus, Appl. Catal., 40 (1988) 255.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science arid Catalysis, Vol. 12, pp. 155-163 0 1992 Elsevier Science Publishers B.V. All rights reserved.

155

OXIDATIVE CONVERSION OF LIGHT ALKANES ON SILVER CATALYSTS A . G . ANSHITSa, S

. N . VEHESHCHAGINa,

’lnstitute of Chemistry of Krasnoyarsk 6 6 0 0 4 9 , USSR tJ

A . N . SHIGAPOVa and H . D . GESSER’

Natural Orpanir

Materials.

Chcinistry Department, IJniversi ty of Manitoba, Winnipeg,

Meni t,oba, Canada RR’I’ 2N2 .A B X ’ r R ACT

Otidativc conversion o f light alkanes in olknne-oxygen m i x t n n s i l v e r and p i i r e alkerin reaction with oxidized silver ltave b e e n studied. T h o r o s u l ts vbtainrd indicate that relativ e l y weakly bound atomic s u r f ’ s r e oxygen is active in total o x oxidation reaction of light a1k.anes. Tho strongly bound atomic ygen species or sub-surface oxygen is responsible for selecv e oxidation reaction. It is discussed that dimerization p r oducts and olefins (in part) can be produced by alkane reaction with oxygen atoms emitted to the g a s phase from a silver surface while a strongly adsorbed surface oxygen species is active in alkane dehydrogenation. iir .I4 4J

60 rl Q)

vl

40

20

0

500

600

700

800

900

T,K Fig. 1 . Products selectivity versus temperature. Reaction mix3 ture of 99.2% CH4-0.8% 0 2 , feed rate 0.125 cm / s . Silver foil S=80 cm2. 0 -C2; 0 - C 0 2 . Reaction mixture of 99% C2H6-1% 0 2 , 3 feed rate 0.33 cm / s . Silver powder 0.289. A -C2H4; A -C02

158

1) i s s o c ia t i on 20

O2

0

strongly bound

atomic oxygen in the

Dif f u s j on Desorption

gas phase

mo 1e cII 1a r oxygen

-O-I

Fig. 2 . Possible transformations of oxygen species o n silver.

For methane conversion

in methane-oxygen mixtures on silver

The selectivity to C 2 hydrocarbons increased with increasing of oxygen and methane

catalyst

an unusual

conversion [ I ] . We

effect was found.

believe this effect can be

attributed

to

t h e two procnr;t;es proceeding o n silver catalyst with different

and C 2 selectivities. The first process has a high rate

rates

and leads mainly t o CO

2

on weakly bound atomic oxygen. The se-

cond is lower rate process, and leads to C 2 formation.This may be a gas phase

process, but special

only traces o f

that oxygen

mixture

C

hydrocarbons

blank experiments

2 ( 9 9 . 2 % CH4 - 0 . 8 % 0 2 )

are formed at 9 3 3 K

reactor. The contribution of gas phase reaction is

show

in methanein

quartz

negligible

tinder these conditions. Therefore we suggest that the process is methane reaction

with strongly bound

oxygen

high C 2 selectivity. This reaction can proceed without i n feed, i t

C

R

~ lead

oxygen conversions. pure

methane and

to

further jncrease of C 2 yield at

second with oxygen high

To test this suggestion, the reaction

methane-oxygen mixture with oxidized

was investigated ( p u l s e experiments, Table I).

of

silver

159 TABLE 2 .

The products composition under CH4 and mixture of 99.2% CH

4

-

0.8% O2 pulses on oxidized silver foil (after treatment in an 2 oxygen flow at 523 K for 2 h.). S=13 cm , pulse volume 0.18 3 3 -1 cm , helium flow rate F = 4 0 cm min.

T,K

CH4:02 - pulse

CH4 - pulse

C2H6

753

C2H4

c,L

Concen t ra t ion,10- 3%

Concentration,

se 1ec tyvity,%

C02

C2H6

C2H4

“2

-

-

-

-

4

-

-

100

3.7

853

1.0

-

-

2.1

91 3

3.6

-

-

14.1

0.1

250

10.0

933

7.7

0.1

34.0

0.6

290

19.0

0.1

Under pure methane pulses low.

mic

the selectivity to

C02 is

very

A t these high temperatures the weakly bound surface atooxygen is not

present

high C2 selectivity. This

on silver surface, result also

bound oxygen is active in C2 C2 formation

that explains

confirms that strongly

formation on silver. The rate of

in this case is

considerably lower than

in the

membrane process, owing to a lower rate of oxygen diffusion.

A t the same t i m e high C 0 2 concentration under methane-oxygen pulses shows

that weakly bound

atomic oxygen is

responsible

for C 0 2 formation.

I t is important to note that detectable amounts of dimeriz ~ t i o nproduct--ethane were found at temperatures above 8 5 0 K which j . s higher than desorption temperature (750 K) of stronK I T bound oxygen, f 3 1 , set? Table 1 . The same result was obta-

ined for the membrane process [ l ] . Recently i t has been reported [ 5 , 6 1 about emission surface (see Fig. 2 ) .

of oxygen

S o we suppose

atoms from that methane

the silver can react

with oxygen atoms after the desorption of strongly bound oxy,yen into

the gas phase.

Oxygen atoms in the gas

phase

known to be most active species in reaction of hydrogen

are atom

160 abslraction from methane in comparison to other

oxygen

ies (7l.Strongly bound species is obviously unable

specactiv-

to

ate methane because C p formation was not observed

at

temper-

atures lower than desorption temperature of this species. Weakly bound oxygen can also produce oxygen g a s phase, (see Fig. 2 ) .

atoms

I t allows to explain

concentration under methane-oxygen

pulses

the

as

in

higher

compared

the c2 with

methane pulses (Table 1). F o r the membrane process and for the reaction of methane with

oxidized

strongly bound species being

the

silver only

catalyst,

one

with

present

on

a

the

silver we could expect high C concentration

selectivity owing to low oxygen 2 o n the silver surface. However the reaction in

methane--oxygen mixture with weakly bound

oxygen

high

favours

surface

total

concentration

oxidation

of

At

reaction.

higher temperature, we observe the higher contribution of gasphase reaction owing to increasing of rates of oxygen

species

diffusion and desorption, while oxygen concentration decreases on the increase

silver surface. o f C2

The above said allows

selectivity with

to explain the

rising temperature

for the

membrane process and partly for the reaction in methane-oxygen mixture. Thus, we suggest that

dimerization product-C2H6

by methane reaction with oxygen atoms which are the silver surface into the gas

phase,while

is formed

emitted

total

from

oxidation

reactions occur on surface weakly bound oxygen species. Ethane and propane conversion o n silver catalyst. The ethane

conversion

in

ethane-oxygen

reaction mixture

starts at the relatively low temperatures in methane conversion.

Carbon

dioxide

was

comparison

the

prodnct. Ethylene becomes the dominant product ti1

though

ethylene

temperature 523 K ,

formation

was

found

at

(Fig. 1). At temperatures

main

with

reaction

above

K,

973

relatively above

600

low

K

compl.ete oxygen conversion was observed while the reaction the

case

of methane-oxygen mixture was not found at

temperature.

Unlike

methane,

ethane

was

the

same

essentially

reactive to total oxidation and converted to C02

when

a in

more

weakly

161

bound oxygen species is present on

the

silver

main product of selective oxidation was

surface.

The

be C2H4 ' explained by the possibility o f hydrogen atom abstraction from

ethyl radical unlike methyl radical The dependence

which

can

181.

of tho selectivity o f products formation on

ethane conversion is analogous to methane conversion under similar conditions-the selectivity to products of partial oxidation increases with the increase o f ethane conversion(Tab1e 2 ) TAH1.E 2 ,

Ethane conversion on silver

99%

mixture 3 C 2 H 6 -1% 0 2 , T = 9 2 3 K , silver powder o f 0 . 2 8 g . ( O . 2 cm ) , Feed rate,

catalyst.

Conver.

Reaction

Selectivity, %

n

cmJ/s

C 2H6%

C2H4

C02

CO

n-C 4 H 1 0

CH4

H2

~~~

~

4

0.87

35

63

-

0.6

0.5

1.1

0.33

1.22

63

31

2.4

1.5

1.8

17

0.02

3.02

83

11

2.5

1.3

1.6

43

Rut blank experiments with empty reactor

indicate

phase reaction occurs under the same conditions. ethane

dehydrogenation

without

oxidant

In

proceeds C02

extent. Ethane conversion leads mainly to rate on silver apparently o n oxygen species adsorbed, When feed rate decreases, the

that

at

which

gas-

additjon to

some

high

feed

is

weakly

transition to

unoxi-

dative ethane conversion was observed. It is supported by high

H,2 concentratinn at low feed rates,(Table

2).

contribution of the second process with the

But the possible participation

strongly bound oxygen is not clear here owing

to

of

unoxidative

ethane dehydrogenation with high C 2 selectivity. Pulse experiments (Table 3 ) show that ethane is converted to ethylene as a dominant product on oxidized silver catalyst reaction in bound oxygen conversion.

ethane-oxygen mixture. I t is

active

in

processes

in

proves of

contrast that

selective

to

strongly ethane

162 When ethane i s converted on the surface of catalyst the gas phase according to

181 the

following

and

reactions

in can

'I'ARLE 3 .

I~roductscomposition under

C,H

pulses

2 6

on

oxidized

silver

catalyst. pretreated in an oxygen flow at 643 K for 2 h.Silver 3

powder 0 . 2 8 ~ (.0 . 2 cm 1 , pulse volume - 0 . 1 1 cm

3

.

~

Conver. C 2 H 6 , %

Selectivity, X

T.K

686

0.6

51

48

-

0.7

718

0.7

54

45

0.1

0.8

803

1.1

44

54

0.3

1.5

835

1.2

32

66

0.8

2.2

855

3.4

19

78

1.4

3.0

875

1.3

11

86

1.6

4.7

89 3

3.9

8

88

1.8

6.9

I n addition ethime can react with

oxygen

atoms

in

the

gas

p h a s e after desorption from silver surface:

T l i e low

temperatures of ethylene formation on silver indicate

( F i g . 1 , Table 3) that oxidative dehydrogenation of ethane can

proceed on

surface oxygen

species which is strongly adsorbed

Lefferts et ~ 1 [.4 ] also consider that

oxidative dehydrogena-

to CH 20 takes place on strongly bound oxygen. W e believe that dimerization product n-butane is formed mainly t i o n o f methanoJ

ir, the gas phase (like C H

2 6

for methane conversion). The incr-

163 ,?a?? o i ' nvbutane selr(:tivity with rising temperature

shows the higher contribution of alkane reaction atoms in the g a s phase. This oxygen

species

possibly

with

oxygen

be

active

must

also in ethylene formation. Preliminary results were obtained

using propane

oxidized silver catalyst. At 838 K propylene was dominant product, but at

K

887

-

ethylene

pulses found

becomes

to

as

the

a

main

product apparently owing to the reaction:

C3H7

C2H4 t CH3 In the

Dirnerization products C 6 hydrocarbons were observed.

case o f propane conversi.on propyl and iso-propyl radicals

are

formed. Their recombination produces different C 6 products. Summari z i n g ,

the

results

obtaj-ned for

light

alkanes

conversion in alkane-oxygen mixtures show that ethane is reactive in comparison with

methane.

which

on

is

weakly

adsovbed

oxidation reaction of

light

oxidized silver catalyst,

Mobile

silver

alkanes.

when

only

is

surface active

Alkane strongly

in

total

reaction bound

oxygen is involved in the reaction leads mainly

to

products

that

dimerization

with

atomic

oxidation. I t is

suggested

more

oxygen

selective and

partly olefins are formed by alkane reaction with oxygen atoms emitted from

R

silver surface,

while

olefins

formation

can

take pLace on a strongly bound surface atomic species.

REFER.ENC ES 1. 2.

3. 4.

5. 6.

7. 8.

A.G.Anshits. A.N.Shi,gapov,S.N.Vereshchagin, V.N.Shevnin, Catal. Today, 6 (1990) 593. S . A . T R ~R.. H . G r a n t . R.M.Lambert, J.Catal., 100 (1986) 383. Het,amoso Rohas M . , L.F.Pavlova and V.D.Yagodovskii, Z.Fiz. Khim. 6 3 (1989) 1012. L.L,offerts. Ph.D.Thesis, Enschede, The Netherlands, 1987. S.A.%avyalov. I.A.Myasnikov, Z.Fiz.Khirn. 62 ( 1 9 8 8 1 2786. V . M .Cryaznov, S . G .Gu Lianova, E . N . Kolosov and N . I .Starkovskii. Dokl. Acad. Nauk. 293 ( 1 9 8 7 ) 872. S.D.Hazumovskii. Oxygen-clementary species and properties. Khimiya. Moskva, 1979. E.Morales. J . H . L u n s f o r d ,J . Catal. 118 (1989) 255.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Sludies in Sritface Science and Catalysis, Vol. 12, pp. 165-179 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

165

CATALYTIC PROPERTIES OF PROMOTED VANADIUM OXIDE IN THE OXIDATION OF ETHANE IN ACETIC ACID M. Merzouki, B. Taouk, L. Monceaux, E. Bordes and P. Courtine

UniversitC de Technologie de Compikgne, DCpartement de GCnie Chimique, B.P. 649, 60206 Compikgne Cedex, France

Abstract Re03-like oxides belonging to various systems such as : [V-01, [V-P-01 and [Mo-V-Nb], have been prepared, characterized and studied in the mild oxidation catalysis of ethane. Pure and Pd doped (V0)2P207 are found to be very selective catalysts for the direct oxidation of ethane in acetic acid and anhydride as low as 250°C, whereas oxides having the composition [Mo0.73V0.18Nb0~09] catalyze selectively either the oxidative dehydrogenation of ethane in ethylene near 35OoC, or its oxidation to acetic acid at lower temperatures, according to the mode of preparation.

1. INTRODUCTION It is now well known that the difficulties to perform active and selective mild oxidation of linear alkanes arise from their chemical inertness. The most numerous investigations have been made on methane coupling and n-butane oxidation in maleic anhydride, while very few papers appeared on C2 and C, oxidation. Several patents claim satisfying results in ammoxidation of propane in acrylonitrile on [V-Sb-0] or [Bi-V-Mo-0] systems (1-3), which are partly due to the beneficial action of ammonia on the catalyst. In mild oxidation catalysis, n-butane (and also n-pentane) is the only alkane which can be actively (up to 100% conversion of n-butane) and selectively oxidized (up to 75 mol % of maleic anhydride) at a moderate temperature on [V-P-01 catalyst (4-6), while it can be oxidatively dehydrogenated in butadiene on CoMo04 (7). Apart from studies using N20 as an oxidizing agent (8), very few have succeeded in the oxidation of ethane which is generally limited to the formation of ethylene. After an extensive screening, Thorsteinson et al. have found that in the Mo-V-Nb-0 system, an

166

optimal composition Moo~73V0~18Nb0~090x gives 100 mol. % of ethylene at 10 mol. % conversion of ethane, acetic acid beginning to form at 300°C under 20.4 atm only (9). Mc Cain claimed formation of ethylene with the same system including promoters (10). A selectivity of 60-80 mol.% of ethylene at C2 conversion of 50-70 mol.% was found at 400°C when using a [V-P-01 catalyst (11). At last, supported boric acid was used by Morikawa et al., leading to 53 mol. % ethylene at 38 mol. % conversion (12). According to experimental results found in C1-C4 alcane reactions, and by comparison with mild oxidation of corresponding olefins, some criteria (14) can be used to find active and selective catalysts for alkane mild oxidation. An oversimplified view consists in considering that two kinds (at least) of lattice oxygen are necessary, the first one to catch hydrogen from (R-C)-H (activation of alkane) and the second to be incorporated in the molecule, yielding (R-C)-0 (oxygenation of the intermediate complex) (13). Studying the conditions of activation of methane and n-butane on known catalysts lead to conclude to the necessary presence of surface lattice oxygen (hard base) linked to hard acid Mn+ cations. This is the case of, e.g. 0-Mg2+, 0-Ba2+, O-La3+,... in methane coupling, or of O-V4+ in butane oxidation. From the above mentionned papers on ethane oxidation it is seen that such oxygens, which will evolve later in the form of water, are found on any surface since ethylene is formed quite easily. The oxygenation step, which needs lattice oxygens to be selectively incorporated in the molecule via a redox mechanism, seems by far more difficult. In searching for a selective catalyst in ethane oxidation, we can also remark that vanadium-based catalyst are selective for the oxidation of C2,, hydrocarbons (n = 1-7) whereas molybdates are better for C3 and branched isomers (13). This means that the surface crystal field of [V-M-0] catalysts (M = second element), which is determined by a geometric and energetic set of active sites, lays down symmetry rules : as a result, e.g. for C4 or c8, furan, maleic anhydride or phthalic anhydride are obtained respectively. A model has been proposed to account for the selective oxidation of n-butane in maleic anhydride on (100) faces of (VO)2PzO7, showing that selectivity is related to a specific "cluster" of V-0 and P-0 sites (15,16). For these reasons, we have undertaken the study of vanadium-based catalysts in the mild oxidation of ethane in acetic acid, and first of V02(B) which meets with several of the above criteria. The influence of promoter (Pd) and support (Ti02) on its catalytic and structural properties will be presented. The behavior of two other catalysts, (VO)2P207 and the ternary oxide system M0o,73V0.1pJJb0,09O~, will be also studied in the same way.

2. EXPERIMENTAL METHODS 2.1. Preparation of catalysts Ten samples were prepared in different ways.

167

2.1.1. V02(B)-based compounds - A sample was obtained by reduction of V2O5 in a flow of hydrogen (320"C, 12 hrs). The resulting yellow green solid was ground and the same treatement repeated for at least 20 hrs until pure V02(B) is checked by XRD. - B sample is V02(B) promoted by Pd (0.15 wt.%). An aliquot of A sample was ground with PdC12, and the mixture calcined under nitrogen up to 400°C. - C and D samples are 10 wt % V02(B) supported by Ti02 anatase and Ti02(B) respectively. V02(B) was ground in ethanol with Ti02. This suspension was stirred at 40°C until total evaporation. The resulting solid was dried at 120°C under nitrogen and finally up to 400°C under the same atmosphere. 2.1.2. (V0)2P207-based compounds - E sample is prepared with H3PO4 85% and V2OS (P/V=1.15) according to the alcohol procedure (18) slightly modified. The mixture was refluxed with stirring in 2-butanol for 24 hrs; after cooling, the suspension was filtered and washed several times with 2-butanol. The resulting light-blue solid was dried in vacuo for 8 hrs. The obtained precursor (VOHP04, 0.5 H2O) was heated under nitrogen up to 450°C for 24 hrs yields pseudomorphic crystals of (VO)2P2O7 (17). The resulting solid was put in a water-ethyleneglycol solution; the suspension was further evaporated and calcined at 400°C. - F sample is (V0)2P207 promoted by Pd (0.15 wt.%) : the same method is repeated except that PdC12 is dissolved in 2-butanol. The step of reduction of VSi to V4+ in the slurry is considerably shortened, since 3 hrs are sufficient instead of 24 hrs as above. 2.1.3. Mo0.,3V0.18Nb0.0gOx catalysts - G and H samples are made by solid-solid decomposition of mechanical mixtures in stoichiometric amounts, while I and J are prepared in aqueoues medium. The first step involves the reduction of NH4V03 by oxalic and hydrochloric acids respectively. - G sample : (NH4)6M070241.5 H20, NH4V03 and Nb2OS, were ground and heated up to 400°C under nitrogen overnight. A gray solid was obtained. - H sample : Moo3, V02 and Nb205 were mixed and ground, and then heated in a sealed evacuated silica tube at 700°C for 72hrs. The solid obtained was green. - I sample : NH4V03 (1.634 g) was dissolved with 16 g of oxalic acid in 250 ml of water at 85°C. The color of the solution turned immediately from yellow-orange to blue. A slurry made up of Nb205 (0.928 g) in oxalic acid solution was added to the first one. After partial evaporation of the resulting solution, ammonium heptamolybdate (10 g) was added while stirring until dryness. After grinding the resulting solid was heated in N2 up to 400°C. - J sample : NH4V03 (1.634 g) was dissolved in 37% hydrochloric acid (250 ml), while stirring at 80°C. A slurry made up of 1.5 mole of oxalic acid per mole of niobium and ammonium heptamolybdate, was added to the first solution. After partial evaporation of the resulting solution, the violet-grey solid obtained is dried at 120°C overnight and then,

168

heated up to 250°C under N2 two times in 24 hrs, and then overnight at 400°C.

2.2. Catalytic test The catalytic properties of catalyst pellets were examined in a stainless steel fixed bed flow reactor (14 cm in length and 1.5 cm diameter) at P = 1 atm. Different compositions of C2H6, 02,N2 were used. The reactor inlet and outlet gases were analyzed on-line by gas chromatography : 02,N2, CO and COz were separated on 5 A Linde molecular sieves, and C2H6 and C2H4 on Porapak Q (801100 mesh) at 65°C; acetic acid, acetic anhydride, acetaldehyde were separated on LAC 446 modified with H3P04 using a FID. When the conversion is lower than 5 % , it is calculated as the ratio between the total formed products and inlet ethane, and when higher than 5 %, conversion is calculated as the ratio of converted ethane to total ethane (mol). The selectivity in a given product is taken as the ratio between the formed product and the conversion.

3. RESULTS 3.1. The ( V - 0 ) system (A to D) Table I summarizes the main results on ethane conversion and selectivities, which were obtained on V02 catalysts at different temperatures (between 190°C and 430°C). Generally, the conversion is very low (0.2-2.6%). At low temperature (193°C) acetic acid is formed exclusively. When temperature increases, the selectivity in ethylene increases at the expense of the selectivity in acetic acid. At higher temperature C02 begins to form significantly. These observations are valid for all A - D samples. The behavior of V02(B)/Ti02 anatase (C sample) is completely different. At 193"C, the conversion reaches a maximum. With increasing temperature, the conversion drastically decreases while selectivity in acetic acid decreases slightly, but remains greater than 90 mol. %. When O2 / C2H6 > 1 / 5, the catalyst deactivates rapidly. IR spectra of used C shows that a band appears at 1025 cm-', as in pure V205 (fig. 1). In the case of Pd-V02 (B) a true steady state is reached, with a partial reoxidation of V 0 2 (B) into V409 and V6OI3 as shown by XRD after reaction.

3.2. ( V-P-0 ) catalysts (E, F) When using (V0)2P207 the conversion is very small at 300°C (1.5%), and increases with temperature (13% at 430°C). In this case acetic anhydride and acetic acid are both detected by chromatography. The total selectivity in acetic acid and acetic anhydride is 100% at 270"C, but decreases strongly while selectivity in ethylene increases with temperature.

169

Catalysts

Temperature Conversion

SCH~COOH

("(3 A

B

D

SC2H4

sc02,co

(%)

(%)

193 293 343 43 1 193 266

0.2 0.4 0.7 2.5 0.2 0.4

100 40 12 1 100 55

0 55 80 58 0 37

0 5 8 41 0 8

99 97 90 100 26 3

0 1 5 0 52 55

1 2

405 193 317 405

1.8 1.2 0.4 0.2 0.4 2.1

5 0 22 42

Selectivity in COz remains low even at higher temperature. Palladium in (V0)2P207 (F) improves the conversion (10% at 380°C). Moreover, in a large range of temperature, the total selectivity (acetic acid + acetic anhydride) is stabilized, with a maximum of acetic anhydride at 300°C. Contrary to the preceding case, the production of ethylene decreases when temperature increases (25 % of CO, at 400°C) (Fig. 2).

3.3. (Mo-V-Nb-0 ) catalysts (G to I)

Fig. 1 : Comparison between IR spectra of supported and non supported V02(B) catalysts. (a and b : after and befor testing).

While starting with the same initial composition ( M O ~ J ~IVs N~b.~ . ~ g ) ,this system leads to very different results according to the method of preparation. Results are summarized in Table 2. As a general rule, ethane conversion increases almost linearly with contact time. On the contrary, CO, C02 formation slightly increases at the expense of selectivity in

170

Catalysts

G

H

Temperature Contact Conversion SCH3coO~ SC2,, ("C) Time(sec) (%) (%I (%) 0 81 300 3.5 0.2 0.5 0 79 3.5 350 0 69 400 3.5 1.4 450 3.5 2.3 0 62 3.5 0.1 0 75 300 0 67 350 3.5 0.6 0 60 400 3.5 1.8 I 450 3.5 2.1 0 42 200 3.5 0.7 81 15 4.7 1.2 62 33 250 3.5 1 50 45 4.1 2.7 25 68 300 3.5 1.7 16 78 4.7 5.8 11 84 350 3.5 4.5 1 93 4.7 12.4 1 89 0 91 400 3.5 6 0 87 4.7 12.4 I 200 3.5 1.5 100 0 4.7 2.3 100 0 225 3.5 3.3 98 0 250 3.5 3.6 97 1 4.7 5.3 97 1 275 3.5 3.4 92 6 300 3.5 2.5 77 20 4.7 3.2 71 23 350 3.5 3.4 30 58 4.7 4.7 27 54

I

I I

J

ScO2,co

(%I 19 21 31 38 25 33 40 58 4 7 5 7 6 8 6 9 9 13 0 0 2 2 2 2 3 6 12 19

acetic acid. In the case of I catalyst (preparation with oxalic acid) acetic acidand ethylene are formed since 200°C with comparable selectivities (Fig 3a). Ethane conversion increases up to 13.2% with temperature. The ethylene formation is greatly enhanced as compared with acetic acid which becomes very small at 350°C. XRD patterns have been taken on catalyst samples several times during the study of the reaction. Lines of a MoI80s2-1ike phase are observed to increase as the steady state is reached (Fig 4). X-ray refinement program gives the following unit cell parameters : a = 7.977 A, b= 12.148 A, c= 19.143 A, a = 96.32", b = 90.32", c = 109.25". The J sample (prepared by HCl) exhibits a different behavior with increasing temperature. Ethane conversion passes through a maximum near 250"C, then through a minimum near

171

80

I 300 350 400 TEMPERATURE ("CI

A CH3COOH ;

+ C2H4 ;

C02 ;

C2H6 conversion

Fig. 2: Conversion of ethane and selectivity in various products vs temperature on : a) E sample : (V0)2P207.b) F sample : Pd-(VO),P207. Contact time : 3.8 sec; Partial pressures (atm) : C2H6 0.05, 0 2 0.025 and N2 0.925.

I

a

200

250

300

350

T

400 TEMPERATURE (OC)

TEMPERATURE I°C)

A CH3COOH;

+C2H4;

C02;

C2H6 conversion

Fig. 3: Conversion of ethane and selectivity in various products vs temperature on : a) I sample [Moo~73V0~18Nb0~090x] system prepared via oxalic acid. b) J sample [M00.73Vo.18Nb0,090,] system prepared via hydrochloric acid. Contact time : 6 sec ; Partial pressures (atm) : c2H6 0.09, 0 2 0.06 and N2 0.85.

172

320°C and finally increases again (fig 3b). This suprising behavior is not due to an artifact, since it was observed reproducibly, by increasing or decreasing temperature, and for various contact times (table 2). Acetic acid, alone, is produced at temperature lower than 270°C. C2H4, CO and C02 increase when t > 300°C. Orthorhombic and hexagonal Moo3 forms, (Nb0.0gM00.91)02,8and (M00.67V0.33)02 are characterized by XRD . H and G samples prepared by solid-solid reactions give very low performances. CO and C02 are present even at low temperature. Substituted Mo5OI4-phases and Moo3 are characterized by XRD in G sample, whereas only a mixture of the starting oxides is identified in H sample.

Rad : Cu K a ,

A : 1.540598

~a-+a b

Fig. 4: XRD pattern of I catalyst, showing Moo3 and M o ~ lines ~ O(expanded ~ ~ scale): a) before catalysis ; b) after catalysis. The shift of MoI8Os2 lines accounts for partial substitution of (V, Nb).

4. DISCUSSION Obviously, the results obtained in this first prospective study do not yet allow an exhaustive interpretation of the catalytic mechanism at an atomic level. Nevertheless, the experimental results on the activity and selectivity in ethylene and / or acetic acid found for the ten samples, throw light, at least, on the degree of validity of the few criteria formulated in the introduction (14). First of all, these (mixed) oxides belong to Re03-like structural family, exhibiting important common features such as a lamellar morphology, reduction mechanism by CS

173

planes and extended defects. In used catalysts, the presence of V2O5 in C sample, or of reduced oxides (Mo, V) in I, J samples shows that the own reactivity of the solid is an important factor. For the moment, and in the absence of surface characterization, we can use these informations in order to propose an interpretation. All these features mean therefore that the oxidation of ethane, like for other hydrocarbons, is a structure sensitive reaction. Secondly, the need for ethane to find oxygens linked to hard acid cations (V4+ and/or Mo5+) in order to be activated (and eventuelly dehydrogenated), is also a valuable criterion since the activity falls when V5+ and/or Mo6+ are present in such a large amout that V205 or Moo3 are detected by XRD. Oxygens linked to Vs+ and/or Mo6+ are nevertheless necessary to allow the oxygenation of the intermediate specie in acetic acid.

4.1. VOz (B) catalysts These oxides are selective in acetic acid at low and in ethylene at higher temperature respectively, but the conversion is very low, even for supported V02 (B). In the case of C sample, the unusual decrease of ethane conversion when temperature increases can be related to the rapid oxidation of V4+ in V5+ with formation of V205 (Fig. 1). Consequently the number of active sites 0-V4+ decreases and no ethylene desorbs. Correlatively the selectivity in acetic acid is remarquably constant. It was expected however that V4+ could be stabilised by Ti4+. The synergetic effect already observed in the case of V2O5 or V6OI3 supported on anatase in the oxidation of o-xylene in phthalic anhydride (19,20) does not occur in the present case. With Pd-V02(B), a true steady state is found at higher temperature with a noticeable selectivity in ethylene and also in acetic acid; in this case V6OI3 and V4O9 are detected by XRD. This result is not so surprising since Pd-V205 is known to oxidize selectively ethylene in acetaldehyde and acetic acid (21,22). This performance has been correlated with the presence of V4O9 in the used catalyst. However this last reaction occurs at a lower temperature (250°C). Ethylene is indeed a soft Pearson base (29) and does require neither hard acid sites, nor higher temperatures. 4.2. ( V-P-0 ) system

Like the preceeding A-D samples, (VO)2P,O7 is able to activate ethane, owing to the presence of hard acid V4+ ions. A-D and E, F samples are also able to give acetic acid (and / or ethylene) from ethane. It is interesting to note that (VO)2P207 and V409 (formed after reaction in B sample) are structurally related. According to Grymonprez et al. (23), V409 is a superstructure of V2O5 with ordered oxygen vacancies. Its framework can be consequently described as made up of columns of edge-sharing octahedra, corner-shared to square bipyramids (Fig.5). Let us recall that in the case of (VO)2P,O7 the same columns of edge-sharing octahedra are connected by means of (PO4) tetrahedra (6). As a result, one can consider that on the surface of the layers of these two compounds the same arrangement

174

is nearly found. The main difference is that in (VO)2P207 only V4+ exist (at least theoretically) whereas in V409 a V4+/V5+ couple is found in edge-sharing octahedra. Even if the actual structure of V4O9 is similar to that of V6O13 as suggested in (30), and then to V205 itself (14), these pairs are retained, which could account for the lower activity and selectivity of VO, (B) as compared to (VO)2P2O7. However, (VO)2P207 catalyst is the only one with which production of acetic anhydride is observed (Fig.2). Let us recall that this compound is able to oxidize selectively n-butane into maleic anhydride. This means that the surface of (VO)2P,O7 has a special quality to allow the formation of products presenting an even axis of symmetry CzV, as noted in the introduction (6,15,16). As a result, the dimer and anhydrous forms of acetic acid are (partly) recovered.

Fig. 5: Structural comparison between a) (VO)2P,O7 and b) V409. Arrows indicate the position of the double edge-sharing octahedra. Surface cleavage planes are indicated.

4.3. The ( Mo-V-Nb ) system After a careful investigation, the problem in this case seems to be less complex than apparent. Each of the components (MoO3, V205, Nb2O5) considered alone or put together by a simple mechanical mixing with the optimal composition (M00.73VO.18Nb0.09)(9), is neither active nor selective (G sample). According to Thorsteinson et al., molybdenum

175

should be active, vanadium allowing its reoxidation and niobium stabilizing the whole structure (9). It is not yet known wether vanadium participates or not as an active site, but preliminary experiments done on a M04011 - M ~ 0 catalyst 3 showed that only acetaldehyde in few amounts was obtained, whereas V02 alone is able to give acetic acid. Moo2 itself is very active but unselective since it yields only C 0 2 . For the same reasons as above our interpretation will involve structural and reactive properties of the phases which are present in used catalysts. During the preparation and calcination of I and J samples, we paid attention to have V and/or Mo in a reduced state in order to get more active samples. Ethylene is mainly obtained while I and J give acetic acid at atmospheric pressure, in discrepancy with Thorsteinson et al.6). Therefore it is necessary to discuss separately the properties of these samples I and J . 4.3.1. I sample XRD data obtained on fresh catalyst reveal the presence of Moo3 and of slightly displaced lines corresponding to This could be related to the formation of a solid solution including V, Nb or perhaps both. XRD patterns show that the intensity of (V,Nb)M018O52 lines increases during the establishment of the steady state. Fig.6 describes the relations between the active phases produced during the last step of the preparation, the formation of which is improved during the transitory state. The Mo18052 structure (24,25) is made up from zigzag rows of octahedra, found in Moo3 itself (Fig. 6a), and accommodated by means of C . S . planes. These layers are connected by means of (Moo4) tetrahedra. It is probable that V and Nb can enter both octahedra and tetrahedra. It is also interesting to note that the amount of Nb which gives the optimum activity (9) corresponds exactly to the tetrahedra / octahedra ratio. By comparison with Moo3, the vacant sites which exist on the surface after incorporation of 0 in the organic molecule, can be accommodated by increased edge-sharing along lines, occurring at regular intervals. Fig. 6b illustrates the net idealized result of this cooperative and easy rearrangement leading to Mols052-like phase. This oxide satisfies the first criterion (activation of alcane), since it exhibits Mo5+,which are distributed among 18 octahedra along the shear plane per mole of M 0 ~ ~ 0 Probably 5~. also vanadium atoms in the solid solution are in the V4+ state. As a result, ethylene at least can be obtained.

4.3.2. J sample i- The use of hydrochloric acid is responsible for the temporary formation of blue molybdenyl chlorides in which the valence state of molybdenum is 5 + . Drying and calcination steps of the preparation lead to a catalyst containing more numerous hard acid centers and much more cations ( Mo5+ and V4+) in a more reduced state than in the I sample. This could account for the deeper catalytic oxidation of ethane in acetic acid. XRD patterns reveal three kinds of oxide in the used catalyst : residual Moo3 oxide, (V,Nb) substituted M05014 (the so-called &phases) and substituted Moo2 oxides (30). This suggests that an improvement of the preparation would consist in avoiding the MoO2-like phases formation.

176

Fig. 6 a) Moo3 layer (idealized octahedra) divided into strips occurring in M 0 ~ 8 0 5 ~ . Arrows indicate on zigzag row formed by 18 Moo6 octahedra (after (24)). b) Connection by means of crystallographic shear planes (CS) between Moo3-type strips as above, leading to the structure of M018052.

On the other hand the bidimensional Mo5Ol4-like oxides which are assumed to be active and selective, are derived again from Re03 structure. They should contain a higher density of active sites than in I sample, which may be located at the coherent interfaces between microdomains of Mo50i4 in larger domains of excess Moo3 (14), as suggested by XRD characterization. TEM studies are in progress to confirm this view. ii- As far as the role of oxidation potential of vanadium molybdenum and niobium is concerned, Thorsteinson et al. (9) using EPR have proposed that vanadium and niobium help to reoxidize molybdenum in the active redox cycle. This can be possible in two cases : - The three transition metals are present in the same phase ( solid solution). But, up to now, only the binary systems [Mo0,-Ti02], [Mo03-Nb205] and [Mo03-V205] are known (26-28).

Fig. 7 : hfodel showing how a unit cell of O-Phase (M05014) C a n be intergrown in a Re03-like oxide (here Moo3). Small deformations of octahedra easily accommodate for distortion.

- Otherwise, coherent interfaces between neighboring microdomains containing one or two of these three elements, should be present to allow vibrational and electronic transfer between these ions. If this last point is verified, it could validate somewhat the authors' argument. iii- The facts that, by increasing temperature, the ethane conversion passes through a maximum at 250°C with the exclusive formation of acetic acid, and then by a minimum near 320°C with further increasing C2H4, CO, C02 formation, are difficult to explain without a kinetic study. At the present time the alternative assumption consists to suggest two different mechanisms: the first one at low temperature with radical or peroxidic surface species, and the second at higher temperature with a Mars and Van Krevelen mechanism.

5. CONCLUSION We have shown in this first prospective study that [V-P-01 and [Mo-V-Nb-0] systems containing Re03-like oxides account fairly well for activity and selectivity criteria in mild oxidation catalysis of ethane. These criteria were based essentially on the presence of oxygens linked to hard acid sites, such as vanadyl groups in (V0)2P207, which catalyze the direct oxidation of ethane in acetic anhydride at low temperature, and on the presence of either [V,Nb] substituted MolsOn-like phases in ethylene production, or [V,Nb] substituted 8-Mo5OI4phases in the direct mild oxidation of ethane in acetic acid. However, more work is still necessary, in the one hand, to confirm our models through kinetic studies as well as spectroscopic and TEM characterizations, and in the other hand, to improve activity.

We gratefully acknowledge Professor M . Toumoux for his assistance on X Ray reJnement program.

178

6 . REFERENCES 1 - US Patent 4,746,641 , (17/04/85) and 4,760,159 , (13/03/87)(StandardOil). 2 - G. Centi, R.K. Grasselli, E. Patane, F. Trifiro, "New Developments in Selective Oxidation", G. Centi and F. Trifiro Ms., Stud. Surf. Sci. Catal, 55 , 515 (1990). 3 - Y-C. Kim, W. Ueda, Y. Moro-Oka, "New Developments in Selective Oxidation", G.Centi and F.Trifiro Eds., Stud. Surf. Sci. Catal., 55 , 491 (1990). 4 - Papers in "Selective Catalytic Oxidation of C-4 Hydrocarbons to maleic Anhydride", Catal Today, 1 (1987). 5 - B.K. Hodnett, Cata. Rev. Sci. Eng., 22 , 373 (1985). 6 - E. Bordes, Catal. Today, 1. , 499 ; ibid, 3 , 163 (1988). 7 - J.S. Jung, E. Bordes and P. Courtine, Adsorption and Catalysis on Oxide Surfaces, Stud. Surf. Sci. Catal., Che and Bond Eds., 2 , 345 (1985). 8 - J.H. Lunsford, L. Mendelovici, J. Catal, 9 ,37-50(1985). 9 - E.M. Thorsteinson, T.P. Wilson, F.G.Young, P.H. Kasai, J. Catal, 2, 116-132 (1978). 10 - US Patent 4,524,236 (18/06/84) and 4,596,787 (1986)(Union Carbide Corporation).

11 - US Patents 4,410,752 (1983)(The Standard Oil Company). 12 - A. Morikawa, Y. Wada, K. Otsuka, Y. Murakami, Chem. Letters, The Chem. SOC. Japan, 535-538(1989). 13 - E. Bordes, American Chemical Society, Annual Meeting, Petroleum Chemistry Div , Boston, April 1990. 14 - P. Courtine, ACS Symp. Series, 279 , 37 (1985), R.X.Grasselli, J.F. Brazdil Eds. 15 - J. Ziolkowski, E. Bordes and P. Courtine, J. Catal, 122, 126 (1990). 16 - J. Ziolkowski, E.Bordes and P.Courtine,"New Developments in Selective Oxidation" G. Centi and F.Trifiro Eds., Stud. Surf. Sci. Catal., 55 , 747 (1990). 17 - J.W. Johnson, E. Bordes, P. Courtine, J. Sol. State Chem., 55 , 270 (1984). 18 - US Patent 4,172,084 (1979). 19 - A. Vejux, P. Courtine, (a) J. Sol. State Chem., 2,93-103, (1978). (b) "Atomic

20 21 22 23 24 25 -

Structure and Properties of Small Particules", Wickenburg, Arizona State University (Ariz) (1986). J. Papachryssanthou, E. Bordes, P. Courtine, R. Marchand and M. Tournoux, Catal. Today, 1,219-228(1987). J.L.Seoane, P. Boutry, R. Montarnal, J. Catal, 6 3 , 182 -190(1980). J.L.Seoane, P. Boutry, R. Montarnal, J. Catal, 6 3 , 191-200(1980) G. Grymonprez, L. Fiermans and J. Vennik, Acta. Cryst., , 834 (1977). L. Khilborg, "The Crystal Chemistry of Molybdenum Oxides" in "Non Stoichio- . metric Compounds", Adv. Chem. Series 3 , R.F. Gould. Eds, pp. 37-45(1973). J.C. Volta, 0. Bertrand, N. Floquet, J. Chem. SOC,Chem. Comm., 19, 1283-1285

(1985). 26 - T. Ekstrom, Acta. Chem. Scand., 2 5 , 2591-2595(1971). 27 - T. Ekstrom, Acta. Chem. Scand., 2 6 , 1843-1846(1972). 28 - T. Ekstrom, M. Nygren, Acta. Chem. Scand., 2 6 , 1827-1835(1972).

179

29 - Benchmark Papers in Inorganic chemistry, "Hard and soft Acids and Bases", R.P. Pearson Eds (1973). 30 - G. Calbet et al ., Mat. Res. Bull., 16 , 1107 (1981).


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Scierice arid Catalysis, Vol. 12, pp. 181-189 0 1992 Elsevier Science Publishers B.V. All rights reserved.

181

Microcalorimetric studies of the oxidative dehydrogenation of ethane over vanadium pentoxide catalysts J. LE BARS, A. AUROUX, J.C. VEDRINE Institut de Recherches sur la Catalyse, CNRS, 2 avenue Albert Einstein, F-69626 Villeurbanne CMex M. BAERNS Lehrstuhl fur Technische Chemie, Ruhr-Universitiit Bochum, Postfach 102148, D-4630 Bochum

Abstract Bulk V205 and V205/Si02 catalysts have been studied in the ethane oxidative dehydrogenation reaction. The surface characterization and the reactivity of these catalysts have been investigated using microcalorimetry linked to other techniques, such as volumetry or thermogravimetry. The number of acid sites of the supported catalysts and the initial heats of ammonia adsorption were found to increase with vanadium loading. The interaction with the support was enhanced after catalytic reaction as evidenced by the detection of new strong Lewis acid sites by DRIFT spectroscopy measurements and by higher heats of ammonia adsorption. INTRODUCTION The catalytic oxidative dehydrogenation of alkanes has become of major importance since the last ten years both in industrial and fundamental catalysis. Light alkanes such as methane and to a lesser extent ethane are more reluctant to activation than heavier alkanes as butane or pentane due to their higher dissociation energy of the C-H bond. An efficient oxidative dehydrogenation of ethane requires a selective catalyst avoiding total oxidation although partial oxidation may also be of great interest. Oxygen or air have so far been prefered as the oxidant rather than N 2 0 because they are cheap and readily available. Although the vanadium is known to catalyze oxidation of hydrocarbons, the selective transformation of ethane over V205 has been studied only by a few groups (1-6) and the mechanism for the partial oxidation of ethane involving 0 2 is still unknown. On one hand, both acidity and basicity of catalysts are known to be important factors for partial oxidation reactions. Moreover strong acidity can reduce the selectivity by carbon-carbon bond breaking and by promoting by the production of C02. Acid sites are cations which exhibit either a low oxidation state or an unsaturated coordination. On the other hand, redox-properties are also known to play an important role (6) and to be related to Mn+$ M(n-l)fequilibrium constant and to lattice 02-ion lability.

182

In order to try to clarify the different types of mechanisms involving either redox cycles or acid-base properties, a study of the surface chemistry of bulk vanadium oxide and supported vanadium-silica samples was performed using mainly microcalorimetry. The techniques allowing heat transfer measurements have not been yet widely applied to the study of the surface characterization and reactivity of these metallic oxides. However calorimetry can provide very informative data on the thermodynamics of solid-gas interactions. The redox and acidic features of V205/Si02 have been compared to those of bulk V203 in order to try to explain the differences observed in their catalytic activity for ethane oxidative dehydrogenation. Therefore, on one hand, the acidic features and their contribution to the studied reaction on V2O5 and V205/Si02 were investigated by means of incremental adsorption microcalorimetry of a base probe molecule. On the other hand, the reducibility in ethane gas flow of V2O5 and V205/Si02 was studied using differential scanning microcalorimetry linked to thermogravimetry (TG-DSC).

EXPERIMENTAL Materials : Catalysts were prepared in either the neat or the supported form. Pure unsupported V2O5 was prepared by stirring overnight vanadyl oxalate in methanol up to complete dissolution. Hydrolysis of the complex was then performed by addition of water. The solvent was then evacuated under hypercritical conditions of temperature and pressure for one hour. The supported V2O5 catalysts were prepared by impregnating Si02 (Aerosil, 380m2/g) with aqueous ammonium vanadate solutions. All the solids were dried at 550°C overnight under air flow. Methods : The catalytic reactions were carried out in a fixed-bed continuous-flow reactor. The reacting gas mixture consisted of 8 % ethane, 1% oxygen and helium as dilutant. The products were analyzed by gas chromatography with detection by thermal conductivity for COF and flame ionization for hydrocarbons. Calorimetric experiments were performed in a differential heat flow calorimeter from SETARAM linked to a volumetric line. The adsorption of ammonia was carried out by introducing successive doses of known amounts of the adsorbate onto the samples. The calorimeter was maintained at 80°C in order to avoid physical adsorption. The samples were outgassed at 400°C for 2 hours before any adsorption. The experimental set-up was described elsewhere (7). Acidic properties of the catalysts were also investigated by means of DRIFT spectroscopy. Successive ammonia pulses were applied onto the samples at room temperature. Nitrogen was used as carrier gas to remove physisorbed species. Temperature-programmed reduction and oxidation experiments were performed in a TG-DSC 111 apparatus from SETARAM allowing simultaneous monitoring both heat and mass changes. The reductive gas was pure ethane and the catalyst was reoxidized by pure oxygen prior to any other reduction. Details of the technique have be n described elsewhere (8). The catalysts were heated at a linear rate of 5"C.minPf to a final temperature of 620"C.

RESULTS AND DISCUSSION Vanadium oxide concentrations are reported in table 1 as weight percent. Surface areas (BET) were measured. The surface area of V205/Si02 catalysts decreased with

183

vanadium content. Assuming that the cross-reactional area of a molecule of supported V 05 is 0.201 nm2 (9), a monolayer of vanadium oxide completely covering the surface o the support should need 4.98 x 1OI8 molec. V2O5 m-2. The range of loading investigated is below the theoretical monolayer : 2.4, 12, 26 and 73%respectively for our samples. Table 1 reports the surface compositions defined as the number of sites measured by irreversible ammonia adsorption or determined from the relative intensities of the XPS signals corrected by the relative sensibilities. The table reveals clearly that vanadium as determined by XPS is at lower concentration than that determined by chemical analysis. This indicates that one has probably large V205 crystallites near small silico-spheres (14 nm in size for 380 m2.gm1 material). It is known from previous spectroscopic studies (1013) that vanadia does not spread widely on the silica support. V205 crystallites are often detected for low vanadium loading, besides other surface species which have not been yet clearly established although some suggestions have been postulated (14). Figure 1 displays the differential heats of ammonia adsorption as a function of the ammonia uptake and of the vanadia content in a three-dimensional diagram. It appears that the amounts of ammonia adsorbed and the corresponding heats evolved increase with vanadium content of the catalysts. The amounts irreversibly adsorbed increase almost linearly with the vanadium content but the integral heats tend to a plateau. Bulk vanadium oxide displays much lower heats of ammonia adsorption with an initial heat of adsorption of less than 70 kl/mol (8) (figure 4) which confirms that the acidic properties of silicasupported catalysts are considerably different from those of pure vanadium oxide.

?

Table 1

SAMPLES

Si02 V205/Si02

V205

V205 SURFACE AREA lwt% V205) lm2.g-11

SURFACE SITES f (micromol.m-2l

1V:Si at. ratio1 '

XPS ana.

Chem.ana.

-

305

0.46

-

0: 1

1.09 4.82 10.0 19.1

308 270 256 173

0.57 0.91 1.20 2.01

1:460

i:89 1: 45 1:25

1:138 1: 30 1: 14 1:6

100

12.8

4.95

1:0

l:o

184

NH3 Volume micrornol.rn-2 Figure 1. Differential heat of adsorption of ammonia at 80°C on different V205/Si02 F samples. I 10 [rf&-l.g-l V2051 10 R31C --__.--___ __

T

0

1

10

20

30

40

W

Cio

I% wt V205I -.--cai/r ..*.. c(J

70

00

90

100

Figure 2. Rate of formation of products as a function of vanadium pentoxide concentration. lOOmg of sample - T = 530°C - Fv = 50 cm3/min P (C2Hd = 61 torr - P(O2) = 8 Torr. Figure 2 displays the rate of products formation as a function of vanadium pentoxide content. The main products are ethylene and carbon monoxide. It appears that the highest activity for ethylene is observed at very low vanadium content and decreases further while that for CO increases and reaches a maximum around 20% V205. These results suggest :

185

i: ii :

that ethylene and CO are formed on different sites. well dispersed V cations form oxidehydrogenation reaction which correspond to specific V ion environment different from those existing in bulk V2O5.

Figure 3 presents the differential heat of adsorption of ammonia on a V205/Si02 sample (4.82 wt% V2O5) before and after ethane oxidative dehydrogenation. Before the calorimetric measurement the catalyst was treated with the reacting mixture at the reaction temperature (550°C) until a steady state activity was obtained. Surprisingly, the initial heat increased of about 50 kJ/mol on the used catalyst although the amount of ammonia irreversibly adsorbed decreased slightly (-7%). This behaviour was confirmed by DRIFT spectroscopic measurements of ammonia adsorption (figure 3). After reaction very strong Lewis acid sites, responsible for ammonia coordinative chemisorption, are present on the vanadium oxide, while Bronsted acid sites are decreasing. The Lewis acid sites are identified as coordinatively unsaturated vanadyls (15). The fact that ethane reacts with the fresh catalyst to give ethene and consequently more acidic surface sites indicates that the reaction increases the number of reduced cus species. Figure 4 shows the same kind of studies performed on bulk vanadium oxide before and after the catalytic test. In that case, very strong Lewis acid sites were not evidenced and on the contrary Bronsted sites increased at the expense of the Lewis sites. The calorimetric curve displays much lower initial heat of adsorption on the used catalyst than on the fresh one and the strongest sites were observed to decrease rapidly with time on stream, although the catalyst is still active. It is worth noticing that no chimisorbed hydrocarbon species were detected by DRIFT spectroscopy on the used bulk or supported samples at room temperature.

130 -

V205/Si02 (4.82 wt %)

vmellD’m

110-

90-

Bonsted

+..,

70-

”,.+

@lore reaction

+

I After reaction

......

..........

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

50 30 1 , 0 0.2 0.4 0.6 0.8 1 NH3 Volume (micromol.m-2) Figure 3. Differential heat of adsorption of ammonia at 80°C on V205/Si02 (4.82 wt% V2O5) : (+) fresh catalyst, (4 after catalytic run.

186

0

-----

F

4 NH3 Volume (rnicromol.m-2

2

0

8

Figure 4. Differential heat of adsorption of ammonia at 80°C on V2O5 (.. .) fresh catalyst, (+) after catalytic run. The redox properties of the samples were studied by temperature-programmed reduction and reoxidation experiments which were carried out in a TG-DSC apparatus. It allows a simultaneous determination of both heat and mass changes. The reducibility level and the heat of reduction by H , C2H4 and C2Hg and reoxidation of the materials were determined on the bulk oxide and on the 19.1 wt% V205/Si02 sample. The catalysts were flushed with helium containing 25% C2Hg and the weight loss was measured by thermogravimetry.

(i)

i

HEAT FLOW (niW!

I

30 20 10

0

-0.15

DTG (mg.mln-I )

Figure 5. Heats of reduction ( - ) in ethane and associated derivative of the thermogravimetric curves ( -- ) of V2O5 (x:) and V2Os/SiO2 (19,l wt% V2O5) (n@

187

Figure 5 shows the heats of reduction (heat flow signal) which are exothermic and the derivative of the thermogravimetric curves (DTG) which are negative as associated to a weight loss observed during the reduction by ethane of the bulk oxide and the silica supported oxide. After each reduction the catalyst was reoxidized in a flow of oxygen and helium at the same heating rate. Figure 6 displays the corresponding heat curves of reoxidation (exothermic) together with the DTG signals which are positive as corresponding to a weight gain.

Figure 6 . Heats of reoxidation i n oxygen and associated derivative of the thermogravimetric curve of V2O5 (:) and V205/Si02 (m (19.1 wt% V2O5), prereduced in ethane. All the curves are given for an initial mass of 15 mg of catalyst. The results are summarized in table 2 which reports the total heats of reduction and reoxidation, the associated mass variations and the temperatures of the maxima of the peaks. Reduction in ethane of bulk vanadium oxide started at about 350°C and increased more rapidly at higher temperatures with a maximum at 521°C. The extent of bulk catalyst reduction corresponds to about 16% weight loss and the redox cycle can be repeated. The conclusion is that lattice oxygen is very labile and oxygen deep within the bulk can participate in reduction and reoxidation at the surface.

188

Table 2 NEOXIUATION RY 02

UFUUCTION BY C 2 t 6

SAMPLES

Am

bti

Tm

I% wt lossl [kJ/g V205l I Cl V205

V205/Sl02 19 Iwt% V205

am I%w t gain1

AH [kJ/g V2051

Tm

1 C1

16.5

-025

521

ti7.i

-2.2

398

-14 3 w

-10

504

+I21 1

-10

356

-

The amount of oxygen sorbed by the catalyst and that calculated from the products agreed well with the formation of V203 as suboxide. The reduction of supported vanadium pentoxide started at temperatures lower than those of unsupported V2O5 and peaked at 504°C. The weight loss or gain values for V205/Si02 reduction and reoxidation are observed to be slightly lower than for the bulk catalyst. The weight-gain in reoxidation by oxygen is not completely reversible probably because of ethylene oligomerization occuring at a high rate above 550°C. However the catalyst is highly yellow colored and is reoxidized somewhat near the fully oxidized state. It clearly appears from the reduction heat expressed per g of V2O5 (column 3) that reduction of V205/SiO~evolves more energy than that of unsupported V2O5. V205/Si02 is more easily reduced and this can be related to its higher activity in oxidative dehydrogenation of ethane.

CONCLUSION The experimental results allow to draw some conclusions :

+ +

V2O5 and V205/Si02 catalysts are active for ethane oxidative dehydrogenation into ethene. On bulk V2O5 only redox-properties, i.e. electronic transfers between vanadium cations, and lattice oxygen lability are responsible for the catalytic activity. The activity and selectivity to ethene are enhanced on V2051Si02 especially for very low vanadium content, corresponding to presumably isolated vanadium cations, which are more easily reduced in ethane as shown by differential scanning microcalorimetry. The acidic species generated by the deposition of vanadia onto silica as evidenced by ammonia adsorption microcalorimetry, do not yield higher activity and selectivity at high vanadium content. However such an interaction between vanadia and silica is enhanced under catalytic reaction. It may correspond to peculiar unsaturated VOSi species not yet identified.

+

+

189

REFERENCES 1. 2.

3. 4.

5. 6. 7.

8. 9.

10. 11. 12. 13. 14. 15.

AMENOMIYA, Y., BIRSS, V.I., GOLEDZINOWSKI, M., GALUSZKA, J., SANGER, A.R., Catal. Rev., Sci. Eng. 32, 163 (1990) THORSTEINSON, E.M., WILSON, T.P., YOUNG, F.G., KASAI, P.H., J. Catal. 52, 116 (1978) IWAMOTO, M., TAGA, T., KAGAWA, S . , Chem. Lett. 1469 (1982) IWAMATSU, E., AIKA, K., ONISHI, T., Bull. Chem. SOC. Japan 59, 1665 (1986). ERDOHELYI, A., SOLYMOSI, F., J. Catal. 123, 31 (1990). OYAMA, S.T., SOMORJAI, G.A., J. Phys. Chem. 94, 5022 (1990) and 94, 5029 (1990). AUROUX, A., VEDRINE, J.C., in "Catalysis by acids and bases", Stud. Surf. Sc. Catal., Imelik B. et al., Editors, Elsevier Sci. Pub. Amsterdam, 20, 311 (1985). LE BARS, J., AUROUX, A., VEDRINE, J.C., POMMIER, B., PAJONK, G.M., accepted in J. Phys. Chem. LOPEZ NIETO, J.M., KREMENIC, G., FIERRO, J.L.G., Appl. Catal. 61, 235 (1990). HABER, J., KOZLOWSKA, A., KOZLOWSKI, R., J. Catal. 102, 52 (1986). ROOZEBOOM, F., MITTELMEIJER-HAZELEGER, M.C., MOULIJN, J. A., MEDENA, J., DE BEER, V.H.J., GELLINGS, P.J., J. Phys. Chem. 84, 2783 (1983). JONSON, B., REBENSTORF, B., LARSSON, R., ANDERSSON, S.L.T., J. Chem. SOC.Faraday Trans. I 84, 1987 (1988). BOND, G.C., FLAMERZ-TAHIR, S . , Appl. Catal. 71, l(l991). VOROB'EV, L.N., BADALOVA, I.K., RAZIKOV, K. Kh., Kinet. Catal. (Engl. Transl.) 23, 119, (1982). BUSCA, G., Langmuir 2, 577 (1986).


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shcdies in Surface Science arid Catalysis, Vol. 72, pp. 191-201 1092 Elsevier Science Publishers B.V. All rights reserved.

191

Oxidative dehydrogenation of ethane on chromium modiped zirconium phosphates Mostafa LOUKAH, Gisble COUDURIER and Jacques C. VEDRINE Institut de Recherches sur la Catalyse, CNRS, 2 avenue Albert Einstein 69626 Villeurbanne, France.

Abstract Layered zirconium hydrogenophosphates such as (Y and I3 Zr(HP04)2 have been studied with the idea to introduce redox-type transition metal cations as chromium either in the interlayer void space (d=0.76 and 0.95 nm, respectively) by cationic exchange or deposited on the crystallites by impregnation. Pure CrPO4 and C r 0 3 and Cr2O3 impregnated on different supports as SiO Zr02, ZrP207 have a s o been studied for comparison. It has been observed or the oxidative dehydrogenation of ethane into ethylene at 550°C that C r impregnated on (Y Zr(HP04)2 calcined at 500°C and CrP04 are the most efficient catalysts with 20 to 30% conversion and 50 to 60% selectivity in ethylene. All other exchanged and/or impregnated samples were shown to be less active and to yield total oxidation. A comparative an complementary study by UV-vjs and ESR techniques indicates that isolated C>+ cations and/or small Cr3 + cluqters are responsible for the better catalytic behavior while Cr 0 as bulk material o r as large crystallites deposited on any of the supports stu ie lead much more to total oxidation or at least to COX. This is a new example of the sensitivity of partial oxidation reactions to the structure of the oxide catalyst.

?

?

22

1. INTRODUCTION

Zirconium hydrogenophosphates Zr(HP04)2, nH 0 are layered materials with interlayer distances different depending on severa parameters as hydration extent, calcination temperature, nature and amount of exchanged cations. The more common phases are a, B and y characterized by an interlayer distance equal to 0.756, 0.947 and 1.22 nm respectively. The protons located within the interlayers void space are exchangeable by mono-, di- or trivalent cations resulting in potential catalysts with either acidic or redox-type catalytic properties (1). However such structures are known to be thermally unstable and to yield pyrophosphate groups by dehydration of phosphate groups resulting in amorphous pyrophosphates at 450500°C and well crystallized pyrophosphates above 750°C.

f

192

Catalytic properties of hydrogenophosphates and exchanged or impregnated samples with transition metal ions have been recently reviewed by A. Clearfield et a1 (1). Chromium phosphate (2) and Cr ions supported on AlP04 (3) have been studied for ethylene polymerization reaction. Chromium oxide and chromium oxide supported on silica, alumina or zirconia have been studied for various catalytic reactions such as hydrogenation, dehydrogenation, polymerization. Cr/Zr02 have been reported for ethane oxidehydrogenation reaction (4) and various phosphates have been shown to be active for propane oxidative dehydrogenation and partial oxidation (5). The purpose of the present work was to study Cr3+ exchanged and Cr3+ impregnated a! and B zirconium hydrogenophosphates in the ethane oxidehydroge ation reaction. Their catalytic and physical properties are compared to those of C j + impregnated on supports as ZrO2, ZrP207 and Si02.

2. EXPERIMENTAL PART 2.1. Sample preparations 2.1.1. Pure phases The a phase was pr pared following Clearfield et al' method (1). 300 cm3 of 3 M H3PO4 and 100 cm of 0.5 M ZrOC12, 8H 0 aqueous solution were mixed and stirred. The gel was stirred for 3 h at 7J"C under reflux, washed with centrifugation with 0.2 M H3PO4 solution and further with distilled ater. The gel was then dried in air at 40°C and the solid recrystallized in 300 cm?3 of a 4.5 M H PO4 solution for 48 h (sample designated aZrP1) or 114 h (sample designated a!&P2), washed with distilled water and dried in air at 100°C. T h e y phase was prepared according to Yamanaka et a1 (6) proc dure. 100 cm3 of 1 M ZrOC12 8H2O solution was added dropwise to 300 cm of 6 M NaHzP04 solution under stirring and reflux at 70°C. After one hour and a half the solution was transferred to an autoclave and heated at 180°C for one week. The material was then washed with centrifugation with a 2 M HCl solution and further with distilled water. The precipitate was dried in air at 100°C. Chromium phosphate was prepared by addi dropwise under stirring 500 cm3 of a 0.19 M Na2HP04 solution to 500 cmy of a 0.1 M CrCl , 6H20 solution. The pH was maintained equal to 4.2 with 500 cm of 0.1 h$sodium acetate solution. After stirring for 3 h at room temperature, the precipitate was washed with centrifugation with distilled water and dried in air at 100°C

f

e,

3

2.1.2. Cr exchanged samples 3g of a! or R phases were treated with 300 cm3 of a 1.5 M Cr(N03)3 9H20 solution under stirring and reflux heating at 70°C for 48 h. The product was then washed with distilled water, dried at 100°C overnight and calcined at 500°C under air flow for 4 h. The samples are designated cYZrCrP1 and BZrCrP.

193

2.1.3. Cr impregnated samples 3g of a or B phases were treated with 50 cm3 of Cr(N03)3 9H20 solution containing the amount desired of Cr ; water was evaporated under stirring by heating at 80°C and the material was then dried at 100°C overnight and calcined under air flow at 500°C for 4 h. The samples are designated Cr/aZrP and Cr/BZrP for the hydrogenophosphates and Cr/SiO , Cr/ZrO and CrlZrP28 for samples prepared using Si02 (aerosil from degussa), %r02 (preparea by precipitation from ZrOC12, 8H2O solution in basic pH and calcination at 500°C for 4 h) and ZrP207 (prepared by calcining aZrP1 at 830°C for 10 h). 2.2. Techniques used X-ray diffraction patterns were recorded with a diffractometer Philips PW1710 using K a emission from Cu. The interlayer distance is taken as the basal (002) peak. UV-vis spectroscopy studies were carried out in diffuse reflectance mode with a Perkin Elmer Lambda 9 spectrophotometer. Bas04 was taken for reference. ESR spectroscopy measurements were performed using a Varian El00 Line at room or liquid nitrogen temperatures using X or Q bands klystrons. XPS experiments were carried out using a HP 5950A spectrometer with AlKa as the anode source. The samples were deposited on an In foil and pressed to attach them to the foil and analyzed at room temperature without any treatment, excepted the vacuum. Binding energy values are referred to carbon impurity taken as C l s = 284.5 eV. EDX-STEM analyses were performed with a VG HB501 high resolution electron microscope equipped with a field emission gun and at magnification 20 k to 2 M. Catalytic experiments of ethane oxidation were carried out with a flow microrea tor wi h 100 mg sample at atmospheric pressure at 550°C with flow rate of 60 cm3.min-i and the gas mixture composition of 6:3:91 for C2 : 0 2 : He. The catal sts were heated to 500°C under the reaction mixture at a linear rate of 4.6"C min-I. Products were analyzed on line with two gas chromatographs. 3. EXPERIMENTAL RESULTS AND DISCUSSION The main characteristics of the samples are given in tables 1 and 2. It may be noted that when performed in the same conditions protons in the B phase are more exchangeable than in the a phase. This is probably due to its larger interlayer space.

194

Table 1. Some characteristics of Cr exchanged Zr(HP04)z samples after calcination at 500°C. Samples

Chem. Formulae

Exchange level %

Cr/Zr atoms

BET Surf ce area/m2g-B pure after phases exchange

Table 2. Some characteristics of Cr impregnated Zr(HP04)z samples after calcination at 500°C. Samples

Cr/aZrP2( 1) Cr/aZrP2(2) Cr/aZrP2(3) Cr/BZrP Cr/Zr02 Cr/ZrP207 Cr/Si02

Cr

Cr/Zr(or Si)

wt%

atoms

4 7.7 10.6 4.8 6.8 7.2 5.3 68.4 20.6

0.23 0.47 0.68 0.28 0.13 0.41 0.062

BET surface m2gBefore After Impregnation 20 20 20 33 41

34 340

39 54 48 25 36 34 267

10 16

Moreover impregnated Q samples exhibit higher surface area values than the starting material. This may be due to a morphological change due to the impregnation and/or to the presence of very finely dispersed chromium oxide deposited on the surface.

195

3.1. X-Ray diffraction data : Uncalcined a, B and yZrP phases exhibit X-ray diffraction patterns similar to those described in the literature (6-9) for aZr(HP0 ) 1H20, RZr(HP04)2 and yZr(HP0 )2, 2HzO. Note that P2 is better cristallize t an P1 which is probably due to the onger time of cristallization (114 instead of 48h). Calcination of the samples results in a strong amorphisation in the 400-600°C range while the corresponding pyrophosphates are well crystallized at temperatures as high as 850°C. In the intermediate region of temperatures, pyrophosphate bridges were evidenced by IR spectroscopy. Uncalcined BZrP XRD pattern corresponds to the superposition of B and y phases described by Clearfield et al (9) and Yamanaka et al. (6) (fig.la). By calcination at 300°C pure B phase was obtained in agreement with Clearfield. At 400°C the X-ray diffraction pattern changes with intense peaks at d = 0.827, 0.516, 0.387 and 0.330 nm (fig.lb). This phase stable up to 500°C is similar to the "ZrPH2" 550°C phase described by La Ginestra et a1 (10). Cr/aZrP and Cr/RZrP impregnated samples calcined at 500°C are either amorphous or badly crystallized. For the latter samples the 'nterlayer distance was preserved (do02 = 0.9416 nrn for B for instance). For Crj' exchanged samples ( < l o % ) the structure of the CrZrP phosphate was unchanged while a large increase in this distance was for nZrP (see fig. 1, Table 3). In the latter case the whole XRD pattern was changed indicating structural changes in the dense layers. This structure was thermally stable. Upon calcination at 500°C the interlayer distance was slightly decreased while amorphisation started. Pure CrPO4 sample calcined at 500°C was amorphous in the XRD sense.

4

3%

evidences

CPS

CPS

i

a

C

800

10

20

30

40

50

Figure 1. : X-ray diffraction patterns for : uncalcined 100°C dried RZrP a, 500°C calcined BZrP b, 300°C calcined BCrZrP c and 500°C calcined BZrCrP d samples.

196

All samples were analyzed by XRD before and after catalytic reaction. For all samples whatever impregnated or exchanged, (Y or B phases amorphous materials were obtained during activation under reactants (= 20 min). Some samples still exhibit a broad and very small peak near the d peak. For chromium oxide deposited on supports as Si02, Zr02, ZrP207 an??& some of Cr impregnated Zr hydrogenophosphate samples the presence of (Y Cr2O3 particles was evidenced by XRD before and after catalytic reaction (table 4). Table 3. Evolution of the interlayer distance for several samples before, after calcination and after 16h under catalytic reaction conditions. Samples

aZrP1 aZrP2 BZrP aZrCrP BZrCrP Cr/aZrP2(2) Cr/BZrP

Interlayer distances dOo2in nm reaction 500°C calcin. uncalcined" 0.7603 0.7616 0.9461** 1.0631***

0.615 0.8296 0.7384 1.0001 0.7468 0.9416

after catalytic

0.6341 0.9872 0.6773 0.9481

* uncalcined, dried at 100°C in air in an oven and kept in air. ** y and B phases were present *** calcined at 300°C to get B phase alone 3.2. Catalytic data They are summarized for all samples in table 4, the values being taken at steady state after one hour. Only a small deactivation has occurred depending on the catalyst (for instance 4 % for Cr/aZrP2(2) sample). Ethylene and carbon oxides are the only products, no other oxygenates were detected in sufficient amounts. It is all samples exhibiting Cr2O3 detected by XRD worthwhile noting that aCr2O3 and have a very low selectivity i n C2-, i.e. are total oxidation type catalysts. At variance CrPO4 and Cr/aZrP samples exhibit both high activity and 42 to 50% selectivity in ethylene.

197

Table 4. Catalytic result for e ane oxidation at 550°C with C2 : 0 2 : He = 6 : 3 : 91, flow rate 60 cm9.min

-P

Samples

Conversion

Selectivity

c2

aZrCrP1 BZrCrP Cr/aZrP2( 1) Cr/aZrP2(2) Cr/aZrP2(3) Cr/BZrP Cr/Zr02 Cr/ZrP207 Cr / Si02 aCr 0 3 cr~84

2.4 2.5 9 24 20 0.9 12.8 7 15.8 14.8 30

Rate of C, conv.*

(%)

(%)

%=

02

5.8 10 84 79 2.5 86 87 92 69

45 25 31 48 42 50 14 34 12 7 60

cox 55 75 50 52 58 50 86 34 88 93 40

0.6 1.2 50 7.7 7.4 0.6 3.6 66 1 2.6 33

4

C'2 O ( w Before After reaction

0 0 4.1 0 ne

0 0 O? 2.8 ne

0 3.7 13 100

? 99 17 100 E

* 106mol.min-lm-2 estimated taking 100% for cr-Cr2O3 at d=0.2664 and 0.2481 nm. n.e: means present but not estimated. -

3.3. UV-visible spectroscopy data UV-visible spectr Cr3+ ions impregnated on SiO , Zr02 or ZrP207 correspond mainly to Cr4+ ions in octahedral environment (1 )

1

4A2g (F) --> 4T2g (F) at 600 nm 4A2g (F) -- > +lg (F) at 460 nm 4A2g (F) --> 'klg (P) at 370 nm These absorption bands do not change after catalysis and are similar to those of bulk a-Cr20 which supports the presence of Cr2O3 crystallites on these supports as descn ed above. The spectra of Cr exchanged or impregnated aZrP samp es before catalysis exhibit an intense charge tr nsfer band near 360nm due to C&+ cations (fig.2). Such a band overlaps the C$+ d-d transition bands which appear only as shoulders at 660 and 440 nm. After catalysis the charge transfer band near 360 nm disappears while peaks at 40, 500 and 725 nm do appear. These bands correspond to d-d transitions of C j f in square pyramidal environment (12).

%

198

200

I

.

. ; 500

,

, , 800

,

-

Wavelength (nm)

Figure 2. UV-vis spectra of Cr/aZrP2(2) sample before reaction Lz and of CrPO4 sample after catalytic reaction c.

a, and after catalytic

Chromium phosphate before catalysis and calcined at 5 0°C and after catalysis exhibits a very similar spectrum (fig.2~).At variance C J + impregnated or exchanged on RZrP samples do not exhibit high absorption 360 nm, before catalysis. Only bands at 720 and 450 nm assigned as above to C$+ ions in square pyramidal environment, were observed. After catalysis no noticeable changes were detected. For Cr exchanged or impregnated aZrP samples Cr6+ ions are present be ore catalysis. Un er catalytic reaction conditions these ions are reduced into Cfj+. The latter C$+ i ns are in square pyramidal environment very different from that observed for Crg+ in Cr2O3. The same type of environment was also observed fo Cr impregnated or exchanged BZrP samples calcined or after catalysis but such C>+ ions should be unaccessible to reactants and thus were not oxidized during calcination in air. 3.4. EPR data Before catalytic reaction the ESR spectra of impregnated or exchanged samples are composed of two overlapping signals at g = 1.97 f 0.01, one being rather narrow (AH = 50 G) and the second much broader with a line width depending on the CPconcentration and on the Zr phosphate (table 5). CrP04 gives only a broad peak (AHpp = 840 G) with g = 1.95. After catalytic reaction for several hours the ESR either do not change (exchanged samples, Cr/BZrP for instance) or are broadened resulting in only one peak. This holds true for all Crla ZrP samples and for Cr PO sample with a line width in the range 1200-1500 G. These spectra are dift ult to 'nterpret. If one refers to the literature it seems that they correspond to C>+ (dl) ions in two different environments. After catalytic reaction the broad pea s were observed previously (13, 14, 15) and were interpreted in terms of small C$+ clusters designated also as R cluster phase. It is important to determine if dipolar or exchange interaction is responsible for the line width. It is known that the former one results in gaussian-type peak and the latter

199

one in Lorentzian-type. A detailed analyses of the peaks for impregnated aZrP samples after catalysis and CrPO4 shows that dipolar interaction is involved. At variance for CrP04 before catalytic reaction one deals with exchange interaction. Note that pure a-Cx-203 does not give an ESR spectrum due to the too strong exchange interaction between Cr ions resulting in a too short relaxation time and thus a too broad line width beyond detection. For Cr 0 3 impregnated on supports as Si02, Zr02, ZrP207 as detected by XRD, no 3etectable ESR signals were observed after catalytic reaction. It is worthwhile noting that broad and dipolar type spectra were observed only for catalysts which exhibit the better catalytic properties for ethane oxidative dehydrogenation. Table 5. ESR line width in Gauss of the different samples calcined at 500°C

Catalytic reaction

Samples Before aZrCrP1 RZrCrP Cr/aZrP2(1) Cr/aZrP2(2) Cr/aZrP2(3) Cr/RZrP CrP04

a =150 800 650 550 130 840

a =70 50 50 50 53

After 650 130

53

1425 1160 110 1540

-a: axial-type spectum with g l = 1.976 and 811 = 1.952 3.5. XPS and EDX-STEM analyses

The main results concerning the quantitative data are given in table 6 before and after cat ytic reaction. Two main conclusions may be rawn (i) Cr ions are present as C$+ (60%) (B.E. value of 577.6 eV) and as Cr8+ (40%) (B.E. value of 579.9 eV) (16). The amount of the second oxidation state 6+ decreases by roughly one half after catalytic reaction (ii) Cr ions are well dispersed within the materials, even for impregnated samples, since Cr/Zr atomic ratio values are close to chemical analysis values although one should expect a much higher Cr/Zr atomic ratio value for impregnated samples. Even more, after catalytic reaction part of surface Cr ions have entered the crystallites.

200

Table 6. XPS and EXD-STEM data for a/CrZrP2(2) sample expressed in atomic ratio.

Samples ~~~~~

Before catalysis

After catalysis

~

Chem. analysis XPS EDX-STEM

ZrIP 0.455 0.39 0.47

Cr/Zr 0.47 0.56 0.56

Cr/P 0.21 0.24 0.26

Zr/P 0.455 0.37 0.51

Cr/Zr 0.47 0.35 0.50

Cr/P 0.21 0.13 0.26

4. CONCLUSIONS Introduction of Cr3+ ions by exchange with protons turns out to be easier for B than for a zirconium hydrogenophosphate. This is c rtainly due to the larger interlayer spacing in the former case. However such C J + exchanged materials does not exhibit interesting catal tic properties for ethane oxidation reaction at 550°C, probably because the C 3 + ions are unaccessible to the reactants. At variance pure CrPO4 phase and chromium deposited on CyZrP phase exhibit much higher catalytic activity and mainly higher selectivity in oxidative dehydrogenation of ethane in ethylene. Conjunct analyses by XRD, UV-visible and ESR techniques in cate that the best catalysts for ethane oxidative dehydrogenation contain specific Crds+ ions environments either in chains of CrO6 octahedra parallel to V04 tetrahedra linking the chains (17) as * CrPO4 or aggregated in very tiny chromium oxide clusters. In these two cases C T + ions are in square ramidal environment. in Oh environment At variance when crystallites of CrzO3 are present i.e. Cr the catalysts exhibit mainly total oxidation features. Note also that as all samples exhibit excess P at the surface of the crystallites, it is possible that CrPO4 may be formed at the surface at the surface by interaction with Cr but was not detected by our techniques. These results allow to suggest that peculiar Cr3+ ions environments are active and selective sites for ethane oxidative dehydrogenation. This constitutes a new example of structure sensitivity of oxidation reactions on oxides.

Y+

5. REFERENCES 1. 2. 3. 4.

5. 6.

A. Clearfield and D.S. Thakur, Appl. Catal., 26 (1986) 1. T. Kagiya, T. Shimizu, T. Sano and K. Fukui, Kogyo Kagaku Zasshi, 66 (1963) 841. M.P. MC Daniel, Adv. Cata., 33 (1985) 47. S. Cheng and S.Y. Cheng, J. Catal., 122 (1990) 1. Y. Takita, H. Yamashita and K. Moritaka, Chem. Lett., (1989) 1733. S. Yamanaka and Tanaka, J. Inorg. Nucl. Chem., 41 (1979) 45.

201

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

A. Clearfield and S.P. Pack, J. Inorg. Nucl. Chem., 37 (1975) 1283. R. Dushin and V. Krylov, Inorg. Mater. (USSR), 14 (1978) 216. A. Clearfield, R.H. Blessing and J.A. Stynes, J. Inorg. Nucl. Chem., 30 (1968) 2249. A. La Ginestra and M.A. Massucci, Thermochimica Acta, 32 (1979) 241. F.S. Stone and J.C. Vickerman, Trans. Faraday SOC.,67 (1971) 316. V.B. Kazanskii, Kin. i Katal., 8 (1967) 1125. D.E. O’Reilly and D.S. Mac h e r , J. Phys. Chem., 66, (1962), 276. C. P. Poole, W.L. Kehl and D.S. Mac Iver, J. Catal., 1 (1962) 407. A. Cimino, D. Cordishi, S. de Rossi, G. Ferraris, D. Gazzoli, V. Indovina, M. Occhiuzzi and M. Valigi, J. Catal. 127 (1991) 761. S.A. Best, R.G. Squires and R.A. Walton, J. Catal., 47 (1977) 292. B.C. Frazer and P.J. Brown, Phys. Rev., 125 (1962) 1283.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shidies in Surface Science and Catalysis, Vol. 12, pp. 203-212

0 1W2 Elsevier Science Publishers B.V. All rights reserved.

NATURE OF SURFACE SITES IN THE SELECTIVE OXIDE HYDROGENATION OF PROPANE OVER V-Mg-0 CATALYSTS A. GUERRERO-RUIZ~, I. RODRIGUEZ-RAMOS~,J.L.G. FIERRO~, V.SOENEN3, J.M. HERRMANN2 and J.C. VOLTA3 1 Instituto de Catalisis y Petroleoquimica, CSIC, Serrano, 119, 28006, Madrid, Spain. 2 U.R.A.- CNRS - Photocatalyse, Catalyse et Environnement, Ecole Centrale de Lyon, BP 163, 69131, Ecully, CCdex, France. 3 Institut de Recherches sur la Catalyse - CNRS - 2 Avenue A. Einstein, 69626, Villeurbanne, CCdex, France

SUMMARY In order to explain the difference in catalytic behavior between the three VMgO reference phases, (orthovanadate Mg3V208, pyrovanadate a-MgzV207, metavanadate pMgV@f,), in propane oxidehydrogenation, temperature-programmed desorption (TPD) of the probe NO, X-ray photoelectron spectroscopy (XPS) of the catalysts pretreated "in situ", Electron Spin Resonnance and Electrical Conductivity measurements, under static conditions, have been performed. All these techniques can explain the highest selectivity into propene for the magnesium pyrovanadate by the stabilization of V4+ ions observed for this phase which was associated to the formation of oxygen vacancies.

INTRODUCTION The oxidative dehydrogenation of alkanes is the first important step in the general route to transform them into valuable chemicals. Up to now, little information is available concerning the nature of the catalytic sites which are responsible for this reaction, and how the reaction proceeds. Recently, the VMgO system has been shown to have a high efficiency in the activation of both propane and butane (1,2). The reactivity was considered as depending on the difference in the densities and the nature of the surface V - 0 bonds. Three different crystalline phases were identified and their respective catalytic properties have been reported. Working on the pure orthovanadate ( Mg3V20g), pyrovanadate (a-Mg2V207) and metavanadate (p-MgV206) phases, we have found (3), at variance with other authors (4), that the pyrovanadate is specific for the oxidehydrogenation of propane into propene. The three phases differ by the local environment of V5+ sites with oxygen : while Mg3V2Og consists of isolated V 0 4 tetrahedra, a- M g 2 V 2 q is built with corner sharing V04 tetrahedra and b M g V 2 0 6 has a VO6 octahedral structure. Nothing is known about the surface properties of these materials which should explain their catalytic specificities. This information was considered as necessary to understand the role of each VMgO phase (3).

203

204

The use of probe molecules has been the subject of several studies in order to evaluate the nature and the number of catalytic sites (5). Chemisorption of nitrogen monoxide (NO) has been shown to be a powerful tool to characterize metal oxide surfaces (6). Most studies were carried out by following the NO-solid interaction by infrared spectroscopy (6,7). Chemisorption of NO can take place to give monomeric or dimeric entities (8). Moreover, depending on the state of the oxide surface, chemisorbed NO can generate nitrite species or interact with oxygen vacancies. However, the investigation is frequently perturbed by the low transparency of the samples and/or their low surface areas. This problem can be overcome by the use of alternative techniques like thermcdesorption. In this case, the analysis of the products evolved should be indicative of the presence of oxygen vacancies and their eventual participation in the oxidation mechanism. The electrical conductivity is another technique which permits the detection and evaluation of oxygen vacancies. In order to explain the difference in the catalytic behavior of the three VMgO phases for propane oxydehydrogenation, extensive characterization of their surface properties has been carried out. The three magnesium vanadates were differentiated by hydrogen thermoreduction, NO-therrnodesorption (TPD), Electron Spin Resonnance (ESR), X-ray Photoelectron Spectroscopy (XPS) after interaction with the state mixture and by Electrical Conductivity measurements after contact with propane at the temperature of reaction.

EXPERIMENTAL The preparation of the three magnesium vanadates has been described elsewhere (3). They were obtained by precipitation from Mg(OH)2 and NH4VO3 and by subsequent calcination in air. Their purity was controlled by X-ray analysis, Infrared Spectroscopy and 5 l V MAS NMR. They were compared with pure V2O5 (Merck ; S= 0.3 m2.g-l) and MgO (UCB ; S=24 m2.g-l). TPR experiments were carried out in a Cahn microbalance. Samples (20-40 mg) were first heated at a rate of 4 OC.min-1 in dry air from room temperature up to 500°C for 1 hour, and then cooled to room tern erature. The samples were subsequently contacted with hydrogen (100 cm3.min- ) and heated at a rate of 4 OC.min-1 to a final temperature of 550°C. This temperature was maintained for 0.5 h. XPS spectra were obtained from a Leybold Heraeus LHS 10 spectrometer interfaced with a data system, which allowed the accumulation of spectra. The spectrometer was equipped with a magnesium anode operating at 12kV and 1OmA. The powdered samples were pressed into a cylindrical stainless steel holder. They were pretreated under a propane/oxygen mixture (1/8), p= 90 torr at 500"C, in the absence of air, for 1 hour, and subsequently transferred into the analysis chamber. The E.S.R. measurements were performed on a Varian E9 spectrometer operating in the X-band mode. DPPH was used as a standard for g value determinations using a dual cavity. Samples were pretreated under the same conditions as for the XPS study. Spectra were recorded at 77K using a silica dewar filled with liquid nitrogen. TPD (NO) was performed in a glass vacuum system connected to a Balzer mass quadrupole spectrometer. 500 mg of each sample was pretreated under a C3H8 + 0 2 mixture (1/8) (p= 90 torr) at 550°C for 2 hours. After cooling to 20O0C, the sample was outgassed for 0.5 hour and subsequently cooled down to room temperature. 40 t o r of NO was introduced and left in contact with the sample for 1 hour. After evacuation, the samples were heated to 6OOOC at 10"C.min-1. Variation of m/e values characteristic of NO ( d e = 301, N20 (m/e=44) and NO2 ( d e = 46) were recorded during the heating treatment. In order to avoid the eventual presence of C 0 2 (m/e= 44) on the catalyst surfaces which could have been produced during the propane/oxygen treatment, all

P

205

experiments were repeated with samples which had been heated under vacuum at 55OoC for 2 hours prior to treatment with NO. The electrical conductivity measurements were canied out in a conductivity cell of the static type described in ref.(9). The powder was slightly compressed (p =lo5 Pa) between two platinum electrodes soldered to two thermocouples which enable the measurement of the temperature and, when short-circuited, of the electrical resistance. The electrical conductivity o ( in ohm-1.cm-1) was varied as a function of temperature and oxygen pressure.

RESULTS AND DISCUSSION The H2-reduction profiles of the three vanadates and of the V2O5 reference are presented in Figure 1. According our experimental conditions,they show the quantitative reduction of V5+ to V3+ in a single step, without any transition through V4+. The onset temperature for reduction is lower for a-Mg2V207 than for V2O5 and the other vanadates. It thus appears that the structure of a-Mg2V207, consisting of rows of V207 groups with long V- 0 bridges within these groups, is more easily reduced than the two other vanadates (10). Differences in reduction can also be revealed from photoelectron spectra. Figure 2a shows the V2p3/2 spectra for the three vanadates after treatment under reaction conditions. The main feature corresponds to the BE value (517.3 eV) for V5+. A shoulder at lower BE is observed for a-Mg2V207 around 515.5 eV which is indicative of the presence at the surface of V4+ for this solid only(l1). This result is in agreement with the easier reducibility of this phase as previously observed. Besides that, a surface magnesium enrichment has been observed on the three vanadates, since the M g N XPS ratios were slightly higher than the stoichiometric ones (Fig. 2b), though less pronounced than previously observed on the original materials (3) subjected to the treatment with the reaction mixture (Figure 2b). ESR examination of the three vanadates does not show any V4+ signal on Mg3V20g and p MgV206 before and after the reaction treatment ,in contrast with a-Mg2V207. Figure 3 shows the presence of V4+ on this original material ( 3 4 which is increased by treatment with the reaction mixture (3b). These entities are stable in air at room temperature (3c), but disappear after further reoxidation at 55OoC(3d). The TPD-NO profiles (m/e = 30) for the pretreated samples are presented in Figure 4. The NO evolution from MgO at 180OC can be ascribed to NO chemisorbed on superficial structural defects of this oxide (6) (Figure 4a). The same interpretation can be given for the desorption of NO from V2O5. However, the higher temperature of evolution which corresponds to a stronger bond can be explained by the back-bonding from the d-orbitals of vanadium ions to the n*-orbital of the NO molecule. The three vanadates differ by the temperature of desorption of NO. The position of the TPD NO peak follows the order of reducibility as previously observed.It is worthnoting that the a-Mg2V207 shows also a second TPD peak at ca. 470OC. The appearance of such a peak could be explained by the interaction of NO with V4+ sites which have been independently detected by XPS and ESR. TPD spectra of N 2 0 (m/e = 44) is also helpful for the understanding of the surfaces sites involved in NO adsorption. As these spectra can be masked for desorption of C 0 2 produced from the reaction mixture and eventually chemisorbed on the solid, which could be confused with N 2 0 (both with d e = 44), experiments after pretreatment under vacuum were performed. In Figure 5

206

are shown the profile mJe = 44 corresponding to the a-Mg2V2Q sample after pretratment under reaction mixture with (Fig. 5a) and without NO chemisorption (Fig.5~).In Figure 5b, the same profile is given after vacuum pretreatment and NO chemisorption. The comparison of the three curves shows the contribution of C 0 2 (Fig. 5c) and that of N 2 0 (Fig. 5b) to the general profile (Fig. 5a). N 2 0 desorption is indicative of the presence of oxygen vacancies, which interact with chemisorbed NO and reoxidize the surface of the solid (Fig. 6). This oxidizing capacity was further evidenced by electrical conductivity measurement. As deduced from the N2O/NO ratio (compare Fig. 4a and 6), it appears that V2O5 presents a higher oxygen vacancy density than MgO. The relative positions of the N 2 0 maxima of desorption give information on the relative facility of reoxidation of the oxygen vacancies. The N 2 0 profile of k M g V 2 0 6 is similar to that of V2O5 in agreement with the similarity of both structures. The shou!der observed for Mg3V208 at high temperature (560'C) is due to C 0 2 as demonstrated by experiments after vacuum pretreatment in which it does not appear. In that case, the N 2 0 maximum occurs at 500'C. Figure 7 shows the Arrhenius plot of the electrical conductivity o of the three VMgO reference phases. The activation energy of conduction Ec was determined: Ec = 114.5, 105.8 and 108.7 kJ/mol. for ortho, pyro and metavanadate, respectively. Figure 8 indicates that the electrical conductivity is quite independant of the oxygen pressure. This means that the three samples behave as intrinsic semiconductors (60/6PO2 = 0). In this case, the theory demonstrates that the band-gap energy Eg is equal to twice the activation energy of conduction E, (Eg =2 Ec). The values for the three reference samples are of the same order of magnitude, with a slightly higher value for Mg3V208 (Eg =2.38,2.20 and 2.26 eV for ortho, pyro and metavanadate, respectively). These values are in excellent agreement with the values determined by UV absorbance spectroscopy (12). Insofar as the three reference phases are intrinsic semiconductors, the oxygen interacting with propane will be necesseraly the 02-surface anions in all cases. In order to evidence the formation of anionic vacancies Vo2- induced by the reaction, kinetic measurements (G= f(t)) have been carried out in the presence of propane under a partial presslire equal to that used in catalysis (3). Figure 9 shows that, as soon as propane is introduced, G increases b several orders of magnitude. This means that propane consumes surface anions 0 surf.,thus creating anionic vacancies according to the reaction :

3-

The pyrovanadate exhibits the largest initial rate and the highest level of reduction, in complete agreement with TPR results. This can be correlated with the highest selectivity to propene as previously observed (3). In contrast, the orthovanadate reduces much more slowly, which indicates that reaction (i) is much less easy than on pyrovanadate. Since, from our experiments (3), this solid is known to give mainly total oxidation into C02, it can be deduced that the propene produced remains much longer adsorbed on the surface and thus total oxidation occurs. The metavanadate presents an intermediate behaviour in agreement with the selectivity pattem(3). The three curves appear to be regularly distributed according to the number of C-C bond ruptures per reacting molecule.

207

CONCLUSIONS The three magnesium vanadates have been compared concerning their reducibility, their surface properties (XPS, ESR, Electrical Conductivity) and their interaction with NO after contacting them with propane and propane + air mixtures in static conditions. Profound differences have been observed between the three reference phases. This information can be used to explain the catalytic results in propane oxidehydrogenation (3).We have observed that a-Mg2V2Q presents the highest selectivity for oxidehydrogenation to propene. In the same conditions, e M g V 2 0 6 and Mg3V20g produce carbon oxides (mainly CO for metavanadate and C 0 2 for orthovanadate). From the results which are presently described, the stabilization of V4+ ions in a-Mg2V207, associated with oxygen vacancies, appears to be responsible for its highest selectivity for propane oxydehydrogenation into propene. This conclusion is supported by the easier reducibility of this phase, the NO-TPD experiments, the XPS and ESR studies and the electrical conductivity measurements. It is noteworthy that the corner- sharing V 0 4 tetrahedra structure (a-Mg2V2q) seems to favor the oxygen atom extraction in comparison with the isolated V 0 4 tetrahedral (Mg3V208) and the V06 octahedral (fbMgV206) structures. This could explain the specificity of this phase in propane oxidehydrogenation.

ACKNOWLEDGMENTS Authors acknowledge the support from the Cooperation Program Spain-France and from the ATOCHEM Co.

REFERENCES 1. Chaar, M.A., Patel, D., Kung, M.C. and Kung, H.H., J. C a d . , 105,483, (1987). 2. Chaar, M.A., Patel, D. and Kung, H.H., J. Catal., 109,463, (1988). 3. Siew Hew Sam, D., Soenen, V. and Volta, J.C., J . Catal., 123,417, (1990). 4.Patel, D., Kung, M.C. and Kung, H.H., in "Proceedings, 9thInrernutional Congress on Catalysis, Calgary, I988", (M.J. Phillips and M. Ternan, Eds.), p.1554, Chem. Institute of Canada, 1988. 5. Fierro, J.L.G. and Garcia de la Banda J.F., Catal. Rev. Sci. Eng., 28,265, (1986). 6. Escalona Platero, E., Spoto, G. and Zecchina, A., J. Chem. Soc. Faraduy, Trans I , 81, 1283, (1985). 7. Matsomoto, A. and Kaneko, K., Langrnuir, 6, 1202,(1990). 8. Caceres, C.V., Fierro, J.L.G., Lopez Agudo, A., Blanco, M.N. and Thomas, H.J., J. Catal., 95, 501, (1985). 9. Herrmann, J.M., in "Techniques Physiques d'Etude des Carulyseurs ",B.Imelik et J.C. Vtdrine, Ed., Edition Technip, Paris, p 753, 1988. 10. Clark, E. and Morley, R., J. SolidState Chern., 16,429, (1976). 11. Blaauw, C., Leenhouts, F., Van Der Woode, F. and Sawatzky, G.A., J. Phys. Chern., 8,459, (1976). 12. Soenen, V., Herrmann, J.M. and J.C. Volta, J . Cazul., submitted for publication.

208

degree

-v4+

v5+

,

I

-.......... a: P-MgV206 b: a-Mg2V20, c: Mg3V208

---d: V2O5

Figure 1 : H2-reduction profiles of the three magnesium vanadates and of V2O5

Ratio 525

520

515

Figure 2 : XPS results

0

0.5

1.0

1.5

209

DDPH J

Figure 3 : E.S.R. spectra of a-Mg2V207 at 77K : a : initial solid b : after reaction by propane + 0 2 mixture at 550°C

c : after reoxidation by 0 2 at room temperature d : after reoxidation by 0 2 at 550°C.

210

T("C) Figure 4a

Figure 4 : TPD-NO profiles for V2O5 and MgO (Fig.4a) and the three magnesium vanadates (Fig.4b).

211

-

I

I

,-, \

\

a;

\

I

100 -

' /' I / /

50 -

,\

\ \

/-'\ # , \

0 '

':\'

\

, / \'\

/'

,

.and phthalic anhydride (B) production in TAP reaction multi-pulse tests feeding a mixture of n-pentane in Argon (1:l)a t 497°Con a preoxidized PVO catalyst.

6.2 Spectroscopic Evidence.

Fourier-transform Infrared (FT-IR) studies in a flow reactor cell [231 of the time-on-stream evolution of adsorbed species on (VO),P,O, surface during the reaction of n-pentane/Oz provide useful informations on the role and dynamics of

240

Fig. 6 n-Pentane outlet (A) and furan (B), maleic anhydride (C)and phthalic anhydride (D) normalized formation at 427°C in single-pulse TAP experiments feeding an 02h-pentane (4:l)mixture on preoxidized PVO.

surface species during the catalytic reaction. Some of the principal FT-IR results are summarized in Figure 4.A base PVO catalyst equilibrated in long-run catalytic tests during alkane oxidation was used for these studies in order to have a representative situation of the active surface including the presence of strongly adsorbed species. This sample was pretreated in two ways before being put in contact with a n-pentane/02 stream t o follow by IR the formation and evolution of adsorbed species: (A) an evacuation treatment a t 400°C or (B) a heat treatment a t 450-500°Cin a flowing 0 2 atmosphere (Fig. 4 A and B, respectively). The interaction of an n-pentane/02 flow with the equilibrated catalyst surface (Fig. 4A) leads to the formation of a n adsorbed species characterized by a sharp band a t 1780 cm-' with a shoulder at 1850 cm-'. These bands are characteristic of v C=O modes of cyclic anhydrides such as maleic and phthalic anhydride [lll.The intensity of these bands does not change greatly with time-on-stream, but additional bands grow later centered a t 1720 cm-l and 1610 cm-'. The latter bands could be attributed, by comparison with the spectra of olefins, dihydrofuran and furan adsorbed on PVO [lll,to a partially oxidized intermediate such as a n unsaturated lactone. A very different situation is encountered when PVO is pretreated (Fig. 4B)a t high temperature in the presence of gaseous 0,. A broad absorption band in the 1500-1720cm-l is removed by the treatment which gives rise to the creation of a weak band a t 1780 cm-' due t o an adsorbed anhydride. No changes, on the contrary, are observed in the 1850-2150cm" region where the bands due t o overtones of

241

+20*

0

Pcntane Oxidation

0

Fig. 7 Proposed reaction network of maleic anhydride and phthalic anhydride formation on PVO [24] and corresponding kinetic model of reaction network which explicity takes into account the role and reactivity of surface adsorbed species after the rate determining step of alkane activation [171. fundamental skeletal vibrations of (VO),P,O, are present. This indicates that the structure of the catalyst does not change upon this oxidation treatment. The removal of the broad absorption band may be attributed to the partial combustion of very strongly adsorbed specieshntermediates formed on the catalyst surface during the catalytic tests and which are still present on the surface even after the evacuation a t 400°C. Thermogravimetric tests [241are in agreement with this indication. When the catalyst, after this cleaning procedure is later put in contact a t 400°C with the n-pentane/Oz flow (Fig. 4B), the broad absorption band immediately forms

242

100

rate n-pentane depletion, micromol/m2.s

Selectivity, %

10

SelTot.MA

0.05

5

0.5 o'/vsup

50

(Yo)

Fig. 8 Selectivity to MA and PA and rate of n-pentane depletion as a function of surface the oxidation state of the PVO, calculated on the basis of a surface kinetic model of the reaction 1171. again. In particular, two bands centered a t 1460 and 1650 cm-l with a weaker band at 1780 are created immediately. The latter band is due t o an adsorbed anhydride, whereas the two former bands resemble those observed by anaerobic interaction at low temperature of pent-1-ene with PVO and attributed to adsorbed C, olefins [ 111. With increasing time-on-stream, the band due to anhydride grows and the s ectrum in the 1400- 1750 cm-I region is modified: a band centered at 1720 cm' forms and the main band at 1650 cm'l shifts to 1620 cm-l and also becomes more intense. Similar changes were observed after treatment of a PVO sample with 0, after anaerobic interaction with pen-1-ene at 320°C and attributed to the formation of partially oxidized intermediates [lll. The mechanism for the creation of strongly adsorbed species on the clean PVO surface, therefore, can be summarized as follows: (1)Immediately a pentene- or pentadiene-like species forms which interacts strongly with surface sites; this species does not desorb and reacts slowly as confirmed by the low reactivity in oxygen (Fig. 4A). (2) Later, partially oxidized intermediates, together with the unsaturated hydrocarbon species, are present on the catalyst surface. (3) Finally, the anhydrides also remain adsorbed on the surface. Evacuation at 400°C of the PVO sample discharged after long-term catalytic tests removes only the anhydrides, which apparently are the only species present

P

243 U L the surface after this pretreatment procedure (Fig. 4A). However, the oxidizing pretreatment procedure (Fig. 4B)shows that the surface of PVO remains largely covered by adsorbed species (unsaturated hydrocarbons and partially oxidized molecules). In conclusion, these FT-IR data clearly indicate that during the catalytic reaction of n-pentane oxidation the surface of vanadyl pyrophosphate is largely covered by various intermediates, especially unsaturated hydrocarbons and products of intermediate oxidation, besides the final anhydride products.

6.3 Transient Studies. In order to clarify better the role of adsorbed intermediates in the mechanism of reaction, transient studies in unsteady- state conditions are necessary. In particular, Temporal Analysis of Products (TAP) reactor studies give unique information on these aspects. The TAP system is a new device for studying the reaction dynamics of solidstate-catalyzed vapour-phase reactions [3,18,251 and allows the study with sub- millisecond time resolution of the formatioddesorption of products from the catalyst surface during transients generated by micro-pulses of reagents. Reported in Figure 5 is a typical TAP experiment in which multi-pulses of pentane are fed on a clean PVO surface (a sample pretreated with oxygen a t high temperature as in FT-IR experiments), and MA formation a t mass 98 (A) and PA formation at mass 105 (B) is monitored. Similar trends are obtained when only n-pentane or n-pentane/02 mixtures are used as the feed, besides to the higher rate of surface deactivation using only n-pentane. This suggests that besides to a consumption of surface oxygen species for the anhydride synthesis, the specific catalyst activity decreases also due to the formation of strongly adsorbed species which hindered the reactivity of the PVO surface. A decrease in the formation of both anhydrides after several pulses is thus observed. However, MA formation decreases esponentially whereas PA formation passes through a maximum in correspondence to an increased formation of adsorbed species on the surface, in particular of unsaturated hydrocarbons according t o IR data. This is evidence fortheir role in the mechanism of PA formation. The concentration of these adsorbed species depends also on the rate of their consecutive oxidation to form MA, for example. In fact, the specific oxygen insertion functionality of the vanadyl pyrophosphate can be inhibited by doping with K [24], while the previous steps of alkane activation to form adsorbed unsaturated hydrocarbons are less influenced. The doping induces a lowering of the selectivity due t o inhibition of anhydride formation, but modifies the surface concentration of unsaturated hydrocarbons which, in turn, causes an increase in the ratio of PA to MA selectivities. The control of surface concentration of adsorbed intermediates and of the rate of their consecutive transformation are thus both key factors to modify the relative formation of PA versus MA. However, as shown in TAP experiments [18,23],the increase in surface concentration of adsorbed species, besides deactivating surface reactivity, induces a lowering of the global selectivity, because these strongly adsorbed species are also preferential sources of carbon oxide formation. The role of strongly adsorbed unsaturated hydrocarbons in the mechanism of PA formation is confirmed also by single-pulse TAP experiments using

244

n-pentane/02 mixtures. Reported in Figure 6 are the normalized formation of furan, MA and PA in this type of experiments. It should be noted that the TAP curves in these experiments reflect both the rate of formation and desorption of the product as well as the rate of formation of intermediates and thus a direct analysis is not simple. However, the presence of an induction time in the formation of PA (Fig. 6) shows that the synthesis of this product requires the formation of a different, higher concentration of adsorbed intermediates (reasonably unsaturated hydrocarbons) as compared to the synthesis of MA. Further useful information is given by this type of experimens: the rate of PA desorption is much slower than the rate of MA desorption from PVO surfaces, probably due to the presence of the aromatic ring. This explains the effect observed in kinetic experiments [lo] of a higher instability of PA toward consecutive oxidation t o carbon oxides as compared t o MA. A possible means of increasing the PA formation and selectivity, therefore, is t o modify the catalyst surface properties in order t o increase the rate of desorption.

7. A SURFACE KINETIC MODEL OF PA SYNTHESIS All the observations made can be rationalized in the mechanism of n-pentane oxidation on PVO shown schematically in Figure 7. Due to strong analogies, it is reasonable t o hypothesize a very similar initial mechanism of transformation for n-butane and n-pentane. For the former it has been shown [3,14,181that the initial steps of the n-butane to maleic anhydride pathway involve the intermediate formation of adsorbed butenes and butadiene. Similarly, the formation of adsorbed pentadiene from n-pentane can be indicated. However, a t this stage a main difference distinguishes the two reaction mechanisms of n- butane and n-pentane oxidation, namely the presence of additional allylic H- atoms in pentadiene as compared to butadiene and thus the possibility of further easy H-abstraction to form adsorbed cyclopentadiene. A template addition between two adsorbed cyclopentadiene molecules may give rise to a hydrocarburic precursor that evolves to phthalic anhydride. Due to the absence of desorption of these molecules from the catalyst surface and to the difficulty in obtaining clear spectroscopic evidence on the nature of these intermediates, the above discussed steps of the mechanism can be only hypothesized. However, it should be noted that in preliminary TAP scanning experiments at low reaction temperature fragments a t mass higher than that of MA or of methyl-MA can be found; these mass fragments are in favour of this hypothesis, but more detailed studies are necessary. In addition, the selective formation of phthalic anhydride from decalin [ 131is a n indirect indication that the bicyclo molecules, if formed, may easily evolve to PA. The higher reactivity of cyclopentane as compared t o n- pentane is also in agreement with this interpretation, but it should be noted that due t o the strong Lewis and B r ~ n s t e d acidity of vanadyl pyrophosphate [261 opening of the ring is very easy. The key step in the PA to MA synthesis from the common intermediate (adsorbed pentadiene) is thus the competition between 0-insertion to form the precursor ofMA formation and H-abstraction to form the precursor of PA formation as well as the surface concentration of near-lying unsaturated C, hydrocarbons.

245 111adsorbed butadiene the reaction of H- abstraction is more difficult due to the absence of reactive H, explaining the differences found between n-butane and n-pentane oxidation on the same catalyst. The formation of more dehydrogenate and, possibly, condensed molecules (called briefly i n Figure 7 Surface Carbon-containingResidues, SCR) also competes with the selective pathways of MA and PA formation. The pathway to SCR can be assumed not-selective, because these strongly adsorbed molecules do not evolve selectively to anhydrides, but are only slowly oxidized to carbon oxides as suggested by TAP tests of the reactivity of these species to gaseous oxygen [18,231. This surface reaction pattern can be modelled in a corresponding kinetic model which explicity takes into account the role and reactivity of surface adsorbed species after the rate determining step of alkane activation [ 171. In particular, the kinetic model can be utilized to determine theoretically the change in the selectivity to PA and MA and in the rate of alkane depletion as a function of the surface oxidation state of the catalyst (O*Nsup) (Fig. 8), an indicator of the relative rates of 0-insertion and H-abstraction. The calculations indicate that this parameter has a considerable effect on the ratio of PA and MA selectivities, but the ratio improves only when a large fraction of catalyst surface is deactivated from the presence of strongly adsorbed species. This suggests that a possible way to improve the performances in PA formation is to realize vanadyl pyrophosphate catalysts with a higher surface area (in order t o have a good reactivity even in the presence of a large fraction of surface deactivation) and working in oxygen controlled conditions in order to enhance the surface concentration of unsaturated adsorbed hydrocarbons, limiting their direct oxidation to MA as well as limiting the consecutive oxidation ofPA to carbon oxides. In fact, as discussed above, PA desorbs very slowly and thus it is necessary to avoid its consecutive interaction with gaseous oxygen to form carbon oxides. From this point of view, probably the riser reactor is the preferable reactor configuration because longer surface life-times of products are possible due to the separate stages of hydrocarbon and oxygen interaction.

8. CONCLUSIONS

Economic considerations of the reaction of n-pentane oxidation to maleic and phthalic anhydrides on vanadyl pyrophosphate suggest that when the ratio of the price of o-xylene to that of n-pentane is relatively high, interest in this reaction will be, but the target objective must be the improvment of PA formation as compared t o MA formation. This can be realized both using appropriate reaction conditions and reactor configurations as indicated by kinetic studies and by a suitable tuning of catalyst surface properties as discussed above. The reaction mechanism of PA synthesis depends considerably on the presence of strongly adsorbed unsaturated hydrocarbons and on the competition between their direct oxidation or template addition. Control of these factor is of critical importance to improve the catalytic behavior.

246 9. REFERENCES

[l]S.C. Arnold, G.D. Suciu, L. Verde, A. Neri Hydroc. Process., 9 (1985) 123. [2] R.M. Contractor, A.W. Sleight, Catal. Today, l(1987) 587. [3] G. Centi, F. Trifirb, J.R. Ebner, V. Franchetti, Chem. Rev., 28 (1989) 400. [4] G. Centi, V. Lena, F. Trifirb, D. Ghoussoub, C.F. Aissi, M. Guelton, J.P. Bonnelle, J. Chem. SOC.Faraday, 86 (1990) 2775. [5] J.S. Jung, E. Bordes, P. Courtine, i n Adsorption and Catalysis on Oxide Surface, M. Che, G.C. Bond Eds., Elsevier Science Pub.: Amsterdam 1985, p. 345. [6] R. Catani, G. Centi, R.K. Grasselli, F. Trifiro’, Ind. Eng. Chem. Research, in press (1992). [7] G. Centi, R.K. Grasselli, E. PatanB, F. Trifirb, in New Developments inSelective

Oxidation, G. Centi and F. Trifirb Eds., Elsevier Science Pub.: Amsterdam 1990, p. 515. [81 R.V. Porcelli, B. Juran, Hydroc. Process., 3 (1986) 37. [9] G. Centi, M. Burattini, F. Trifirb,AppZ.Catal., 32 (1987) 353. [lo] G. Centi, J. Lopez Nieto, D. Pinelli, F. Trifirb, Ind. Eng. Chem. Research, 28 (1989) 400. 1113 G. Busca, G. Centi, J.Am. Chem. SOC.,111(1989) 46. [12] G. Centi, J. Lopez Nieto, D. Pinelli, F. Trifirb, F. Ungarelli, i n New Developments in Selective Oxidation, G. Centi and F. Trifiro’ Eds., Elsevier Science Pub.: Amsterdam 1990, p. 635. [13] G. Centi, J. Lopez Nieto, F. Ungarelli, F. Trifirb, Catal. Letters, 4 (1990) 309. [14] G. Centi, F. Trifirb, Catal. Today, 3 (1988) 151. [151 M.C. Hoare, Hydroc. Process., 5 (1990) 116-B. [161 N. Harris, M.W. Tuck, Hydroc. Process., 5 (1990) 79. [171 G. Centi, F. Trifirb, Chem. Eng. Science, 45 (1990) 2589. [181 G. Centi, F. Trifirb, G. Busca, J . Ebner, J. Gleaves, Faraday Discuss. Chem. SOC.,87 (1989) 215. [191 L. Verde, A. Neri, Hydroc. Process., 11(1984) 83. [201 D. Honicke, K. Griesbaum, Y. Yang, Chern.-Ing. Techn., 59 (1987) 222. [2lI G. Centi, J. Lopez Nieto, C. Iapalucci, K, Bruckmann, E.M. Servicka, Appl. Catal., 46 (1989) 197. [221 J.C. Volta, A, Aguero, R.P.A. Sneeden, in Heterogeneous Catalysis and Fine Chemicals, M. Guisnet et al. Eds.; Elsevier Science Pub.: Amsterdam 1988; p. 353. [23] G. Centi, J. Gleaves, G. Golinelli, S. Perathoner, F. Trifirb, in Catalyst Deactivation V, J.B. Butt, C.H. Bartholomew Eds, Elsevier Science Pub.: Amsterdam 1991; p. 449. [241 G. Centi, G. Golinelli, G. Busca, J. Phys. Chem., 94 (1990) 6813. [251 J.T. Gleaves, J.R. Ebner, T.C. Kuechler, CataZ.Rev.- Sci.Eng., 30 (1988) 49. [261 G. Busca, G. Centi, F. Trifirb, V. Lorenzelli, J. Phys. Chem., 90 (1986) 1337.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shtdies in Surface Science arid Catalysis, Vol. 72, pp. 247-254 @I 1992 Elsevier Science Publishers B.V. All rights reserved.

247

VANADYL PYROPHOSPHATE AS A SELECTIVE OXIDATION CATALYST I. Matsuura Faculty of Science, Toyama University, Toyama 930, Japan

Abstract Vanadyl pyrophosphate i s a c a t a l y s t having both a c i d i c and o x i d i c powers. It i s extremely e f f e c t i v e c a t a l y s t f o r oxidative dehydrogenation o f aldehydes, carboxyl i c acids and ketones and f o r forming carboxi I i c acids by oxidation of aldehydes l i k e heteropoly compounds. Also i t acts as a c a t a l y s t f o r ammox i dat ion o f m e t hy I p y r i d i ne t o cyanopyr i d i ne.

1. INTRODUCT ION Butane oxidation t o maleic anhydride i s known t o be catalyzed by ( V O ) Z P Z O ~ as an oxidation catalyst, the reaction i s due t o the combined a c t i o n o f oxidative dehydgenataion and oxygenation on (100) surface plane o f ( V O ) Z P Z O ~ . As shown i n Fig.1, the vanadyl dimer formed by vanadium ion coordinated t o s i x oxygens l i e s on (100) plane which consists o f fundamental s t r u c t u r e o f ( V O ) Z P Z O ~ and the Then, the vanadyl oxygen ion double bonded i n positions trans t o V ions (V=O). dimer on (100) plane gets support from pyrophosphate ion on both sides o f t h i s plane.

Figure 1. Basic (100) plane of ( V O ) Z P Z O ~ . T r i f i r o and coworkers [l] reported t h a t p o l a r i z a t i o n was induced i n V-0-P linkages by bonding vanadly dimer on (100) plane o f ( V O ) Z P Z O ~ t o phosphorus ion w i t h strong electronegativity. thus surfase V ion took on a c i d i c character which was increased by large d i s t o r t i o n of V-0-P linkages. They thought, as a result, butane was activated on the a c i d i c s i t e of V ion and oxidized by the Different oxygen double bonded V ion (V=O) o f c a t a l y s t t o maleic anhydride.

248 d i s t o r t i o n o f V-0-P linkages i n (VO)2P207 prepared by d i f f e r e n t methods was observed 121. Some r e s u i t s of oxidation reaction used very e f f e c t i v e l y a c i d i c and o x i d i z i n g function of ( V O ) Z P Z O ~ catalyst are reported: (1) Oxidative dehydrogenation of caboxi I i c acids (CHJ) 2CHOOH CHA(CH3)COOH CH z=C (CH3 ) COOCH 3 (CH 3) z CHCOOCH3 (2) Oxidat ion of aldehydes CHz=C(CHj)CHO CHz=C(CH3)COOH CH3CH(CHj)COOH CH z=C(CH3)COOH (3) Ammoxidation of methylpyrideine CH 3 4 5 H4 N CN-C 5 H 4 N

-

-

2. EXPERMENTAL Vanadyl pyrophosphate, ( V O ) 2 P 2 O 7 , used i n t h i s study was proposed according t o the following methods: Method (A) Hydroxyamine hydrochloride (14.3 g> and 21.1 g o f 85 wt.g orthophosphoric acid added together i n t o 200 m l o f d i s t i l l e d water were dissolved by heating up t o 70C, and 18.4 g o f V205 added i n t o the s o l u t i o n reacted w i t h them f o r 1 h a t 9OC w i t h s t i r r i n g . Then excess water was f i l t e r e d , washed enough with d i s t i l l e d water and d r i e d a t 130C t o blue solid. Method (B) Hot iso-butyric alchol (80 m l ) contained 10 g of V Z O ~ was blowed i n t o w i t h hydrochloric gas f o r 5 m i n i t e s w i t h s t i r r i n g . The 98 wt.% orthophosphoric acid (11.3 g) was added i n t o the s o l u t i o n and the reaction was carried out under reflex. Toluene (100 m l ) was poured i n t o the mixture and the iso-butyric alchol was d i s t i l l e d out from the mixture by heating. P r e c i p i t a t e f i l t e r e d from the toluene s l u r r y was added i n t o 200 m l o f d i s t i l l e d water and boiled t o dissolve excess (phosphorus acid) and impurity. The p r e c i p i t a t e was f i l t e r e d , washed enough w i t h d i s t i l l e d water and d r i e d t o b l u i s h gray solid. Method (C) V205 (10 g) and 54 m l of 85 w t . % orthophosphoric acid reacted i n 100 ml o f d i s t i l l e d water f o r 1 6 h under r e f r u x a t lOOC and then the The obtained p r e c i p i t a t e was f i l t e r e d and washed enough w i t h acetone. voPo4 2H20 (10 g) treated 2-butyric alchol (200 m l ) f o r 6 h under r e f l u x was filtered. The f i l t r a t e was boiled i n 200 ml of d i s t i l l e d water, washed enough w i t h d i s t i l l e d water and then d r i e d t o b l u i s h white solid. A l l of three d i f f e r e n t color s o l i d from d i f f e r e n t method was confirmed t o be VOHP04 0.5H20 was dehydrated a t 500C f o r 6 h i n a stream o f He gas t o ( V O ) Z P Z O ~ respectively. The r a t i o o f r e l a t i v e i n t e n s i t y of ( v o ) 2 P & , l042/1200, taken from XRD measurement showed i n Table 1. Oxidative dehydrogenation of iso-butyl i c a c i d (IBA) and oxidat ion of The IBA or metacrolein (MAL) were conducted w i t h ordinary flow type reactor. MAL : O z : Steam : He 1 : 2 : 2 : 20 mixed gas was reacted a t the feed r a t e of 100 ml/min over 3 g o f catalyst. For the ammoxidation of 4-methylpyridine (MPy) , MPy :02 : NH3 : Steam : He = 1 :3 :10 :10 : 10 mixed gas o f 100 ml/min over 3 g o f catalyst.

249

A c i d i t y o f (V0)2P207 was measured of amount o f adsorbed pyridine. The amount of p y r i d i n e held by the catalysts a f t e r evacuation a t 150C. The r e s u l t s of a c i d i t y were l i s t e d i n Table 1. For a reference t o a c i d i c strength of catalyst, it abi I i t y of dehydration f o r 2-propanol was measured. I t s dehydration reaction was conducted by folwing He gas containing 4% of 2-propanol a t the feed o f 100 ml/min over 3 g of catalyst. Table 1 Some physico-chemical properties o f ( V O ) Z P Z O ~

(VO)ZPZO~

Surface Area

1042/1200

(m /g)

A-Type 8-Type C-Type

0.09 1.2 0.4

Acidity (pmo I/g) 18.4 21.2 19.5

12.1

12.3 10.7

Preparation Met hod Method (A) Method (B) Method (C)

3. RESULTS Fig.2 shows the r e s u l t of oxidation of I B A w i t h A-(V0)2P~07. The oxidation A t f i r s t acetone (A) was formed and then MAA was s t a r t e d from about 240C. formed by oxidative dehydrogenation. Propylene also was formed i n p a r a l l e l w i t h t h i s riaction. The conversion of MAL was 100 (II and the s e l e c t i v i t y o f MA 72 % a t 340C.

" 240

..

280

*-lo-

320

360

Temp. (C) Figure 2.

Oxidation of IBA over A-(V0)zPZ07.

Table 2 shows the r e s u l t of oxidation of IBA a t the reaction temperature o f optimum MAA y i e l d over each (VO)2P207. This shows that the s e l e c t i v i t y f o r MA of (V0)2P207 on oxidation of IBA i s A > C > B i n order. The formaton o f This propylene and carbon monoxide from IBA i s found t o C > B > A i n order. reaction due t o acid s i t e s on ( V O ) Z P Z O ~ .

250

Table 2 Oxidation o f IBA t o MA over (V0)2P207 Cat a Iys t

(C)

IBA Conv.(%)

Selectivity MAA A

340 280 320

100 100 96

73.2 52.7 56.4

React. Temp.

A-(VO)zP207 B-

C-

(I) C3H6

14.9 0.5 15.4 22.0 13.5 25.5

COX

MAA Yield (%)

11.4 7.9 4.6

73.3 52.1 54.1

The r e s u l t o f formation o f MAA by o x i d a t i v e dehydrogenation of IBA i s shown may be i n Fig. 3. This r e s u l t suggests t h a t the o x i d i c power o f ( V O ) Z P Z O ~ B > C > > A i n order.

100

-2

-2

50

240

280

320

36 0 Temp. (C>

Figure 3. Formation of MAA by oxidative dehydrogenation of IBA over (VO)2P207. Fig. 4 shows the r e s u l t o f oxidation o f MAL over C-(VO)ZPZO~ i n p a r a l e l l and About MAA, an object sharply progressed t o complete oxidation over 360C. o x i d i z i n g product, i t s MAL conversion was 45 % and i t s s e l e c t i v i t y 71 % a t

340 C. Table 3 Oxidation o f MAL t o MAA over (V0)2P207 Cat a Iyst

React. Temp.

(C> A-(VO)pPz07 BC-

330 340 340

MAL Conv.(%>

Selectivity MAA A

6.5 86.4 45.6

60.0 13.5 71.1

($1 AcOH

27.7 12.3 0.0 0.0 11.0 0.0

COX

MAA Yield

0.0 86.5 18.9

3.9 11.7 32.7

(96)

251

7

I

MAL

> c 0

v

240

280

320

360 Temp. (C)

Figure 4. Oxidation of MAL over C-(VO)zPz07. Table 3 shows the r e s u l t o f oxidation o f MA1 a t the reaction temperature o f optimum MAA y i e l d over each (V0)2P~07. This shows that the s e l e c t i v i t y f o r MAA of ( V O ) Z P Z O ~ decreased C > B > Ain order. Fig. 5 shows the r e s u l t o f ammoxidation o f 4-methylpyridine (MPY) over (V0)zP207. The s e l e c t i v i t y f o r 4-cyanopyridine (CPY) i s higher as 95 % over The c a t a l y t i c a c t i v i t y f o r ammoxidation o f every type of (VO)2P207. methylpyridine t o cyanopyridine dependes on B > C > A.

100

0

LO 0

500

Temp. ( C ) Figure 5. Ammoxidation of 4-methylpyridine over (V0)zPz07. Fig. 6 shows the r e s u l t of dehydration o f 2-propanol carried out as reference t o f i n d the a c i d i c strength of each (V0)2PZO7. The a c t i v i t y dehydration of 2-propanol decreases C > B > A i n order. As the r e s u l t i s same as that of formation o f propylene and CO y i e l d from I6A by reaction. i s clear that the a c i d i c strength o f (VO)2P207 i s C > B > A i n order.

a of as it

252

1 A

be

v

c

0

.r(

200

100

300 Temp. (C)

Figure 6 . Dehydration of 2-propanol over ( Y O ) Z P Z O ~ . 4. D ISCUSS ION Vanadyl pyrophosphate i s the we1 I-known c a t a l y s t f o r synthesis of maleic As above ment i oned T r i f i r o 111 emphas i zes anhydr i de f rom butane by ox i dat ion. He that strong Lewis acids on ( V O ) Z P Z O ~ are necessary t o a c t i v a t e butane. a considered that the surface V ions on the (100) plane of ( V O ) ~ P Z O ~ have nature o f Lewis acid due t o p o l a r i z a t i o n among V-0-P bonding caused by bonding between vanadyl ions and phosphorus ion w i t h strong electronegativity. I n fact, Puttock and Rochester [a] reported that the existence of Lewis a c i d with IRs i t e s and Brgnsted acid s i t e s w i t h the r a t i o o f 2 : 1 on (V0)2P207 spectrometer measurnents and i t s r a t i o turned t o 1 : 2 by steam treament of them. Furthermore, we assumed that P- ions come out on the (100) surface plane, forming unsaturated coordination, adsorbs Ht0 t o be B r ~ n s t e dacid. have three types of Bordes and Courtine [4] suggested t h a t (VO),P207 s t r u c t u r a l isomer. We reported that the configuration of pyrophosphoric ion is sustanding the (100) plane involving vanadyl dimer forming (VO)2PzO7 d i f f e r e n t among three types of s t r u c t u r a l isomer [2]. The r e s u l t of dehydration of 2-propanol over (V0)zPz07 shows t h a t the surface a c i d i t y of (V0)zPz07 great depends on the configuration of pyrophosphoric ion. Selective p a r t i a l oxidation o f aldehydes, carboxi 1 i c acids or ketones are oxidat ion property of catalyst. Over heteropoly compounds, Misono [ S l reported that oxidative dehydrogenation involving hydration w i t h proton a c i d progresses i n the oxidation of WL t o MAA as follows: RCHO

=f

RCH(OM)> -44RCOOM

1 (Add) (M

(Redox) 3.

-

RCOOH

-I

= H or Ma)

but oxidative dehydrogenation t o abstruction o f progresses i n oxidat ion of IBA t o MAA as follows:

hydrogen d i r e c t l y from

IBA

253

I n general, the surface face o f (VO)2P207 i s thought t o be (100) plane, the trans-type o f vanadyl dimer appeared on the clevage of which has two V-ions, one binds w i t h a double bonding oxygen having oxidation property and the other coorinative unsaturated one has a s i m i l a r property o f Lewis acid. Furthermore, the cooordinative unsaturated P-ion extruding from the (100) face has a tendency t o adsorb H20 t o form protonic acid. Therefore, each oxidat ion reaction goes forward on (100) surface plane provided w i t h such a c i d i c and o x i d i c sites. I n the oxidative dehydrogenation of IBA t o MAA, i t i s no need of the protonic acid c l e a r l y from such reaction scheme as shown here. Cll.

0

of

oxidative

dehydrogenation,

ci s - 2- b u t e n e tr an s - 2 - b u t e n e C@ (kJ mol-l\ 96.1 104 71.5 54.8

39.5 98.2

40.3 101

90.7

331

8 2 .,.is

0 553K V

70

Lo

0 573K

82 -trans p + p Phase

V

/

60

0,

20 10

0

Fig. 5 .

1

kPa

2

0

1

kPa

2

Effect of 1-butene partial pressure on products formation rates (mixed phase)

DISCUSSIONAND CONCLUSIONS As referred the samples after TPD experiments became gray. An experiment in the same conditions with the high temperature B i z M o o g ?-modification had showed in the XRD spectrum the presence of metallic bismuth. Such modification exhibits in the TPD chromatogram a very strong and predominant peak starting at 633K assigned to loss of Bi3+ bound 0 2 - as 02(10). On the other hand it was

found after temperature programmed heating of the Y-phase in helium up to 633K in a very sensible electrobalance, that the catalyst recovered the lost weight after subsequent identical treatment in dried air. As mentioned the catalysts were treated with dried air for 2h at 423K before the experiments. Such facts evidence that the peaks observed in the chromatograms correspond to removed oxygen.

332

TABLE 3 Selectivities for low partial pressures of 1-butene and low conversions Catalyst

Temperature

Selectivities (%)

(K) Butadiene ~~~

Bi2M03012(a)

Bi2Mo209(l3)

Bi2MoOg(Y)

Mixed phase

(a +%

Cis-2-Butene Trans-2-Butene

~~

~

553 573 593 623 573 593 623 653 673 553 573 593 623 553 573 593 623

32.0 38.8 45.7 55.1 48.6 52.5 59.8 70.8 74.8 41.7 53.3 63.4 74.7 50.7 55.7 60.3 66.3

33.2 29.5 25.7 20.7 43.8 39.2 30.6 17.6 12.8 33.9 25.6 18.6 11.1 29.2 25.9 23.0 19.3

29.4 27.7 25.6 22.3 7.6 8.3 9.6 11.5 12.4 18.0 13.6 10.0 6.0 20.2 18.4 16.7 14.5

C02 ~

5.4 4.0 3.0 2.0 0 0 0 0 0 6.4 7.4 8.0 8.3 0 0 0 0

Dadyburjor and Ruckenstein (1 1) found that for bismuth molYbdate (Bi2MoO6) the energy barrier for loss of an intermediate layer 0 2 - ion as 0 2 is less than that for a Mo6L bound 0 2 - ion as 0 2 which is less than that for a Bi3+ - bound 02-ion as 0 2 . On this basis, for the Y-phase, the TPD peak observed at 483K would be assignable to oxygen from intermediate layer 0 2 - ions, the peak at 593K to oxygen from Mo6L - bound 0 2 - ions, and peaks starting at 633K to oxygen from Bi3+ - bound 0 2 - ions. For the other catalysts, that have not layer structures, we are of opinion that similar assignments may be done according the predominant feature of the involved 02- ions, as confirmed by the afore mentioned XRD spectrum of Y-phase.

333

The sizes of the peaks starting at 633K (Fig.1) follow the sequence (8 + r) > 0 > Y a. This is the sequence followed by the rate of the 1butene reaction over the tested modifications (Table 1). This result is consistent with rate determining step involving the initial abstractions of hydrogen by an oxygen atom associated with bismuth. In connection with the second abstraction of hydrogen, butadiene

-

Selectivities follow the sequence >(8 + r) > 8 > a (Table 3). This is apparently also the sequence of the sizes of the peaks in Fig.1 assigned to oxygen from M o 6 L bound 0 2 - ions. Such sequence parallels the change of molybdenum coordination from octahedral (Y-phase) to tetrahedral (a-phase). Due to the differential conversions of operation, C 0 2 was only measurable in the case of a a n d Y phases that exhibit the strongest peaks assigned to oxygen from intermediate layer 0 2 - ions. It is noteworthy that the rates of formation of butadiene and 2-butenes have first order dependence on olefin only for the 8-phase that exhibits the weakest peak assigned to oxygen from intermediate layer 0 2 - ions. The high temperature ?-modification has an identical behaviour with a very weak peak of the same type (10). The other modifications show half-order dependence. The apparent activation energies for the butadiene formation 0 identical to the sequence of the sizes follow a sequence a>@+Y)y)>Y> of the peaks assigned to oxygen from intermediate layer 0 2 - ions. Sequences for cis-2-butene and trans-2-butene are somewhat different. The activation energies obtained for such differential conversions are different of the activation energies found by Burban, Schuit et al. (12).

1 B. Grzybowska, J. Haber and J. Jonas, J. Catal., 49, 150, 1977. 2 H. Miura, T. Otsubo, T. Shirasaki and Y. Morikawa, J. Catal., 56, 84, 1979. 3 R.K. Grasselli, Appl. Catal., 15, 127, 1985. 4 A.B. Anderson, D.W. Ewing, Y. Kim, R.K. Grasselli, J.D. Burrington and J.F. Brazdil, J. Catal., 96, 222, 1985. 5 L.C. Glaeser, J.F. Brazdil, M.A. Hazle, M. Mehicic and R.K. Grasselli, J. Chem. SOC., Faraday Trans. 1, 81, 2903, 1985.

334

6 7 8 9 10

11 12

K. Bruckman, J. Haber and T. Wiltowski, J. Catal., 106, 188, 1987. T.P. Snyder and C.G. Hill, Jr., Catal. Rev. - Sci. Eng., 31 (182), 43, 1989. B.C. Gates, J.R. Katzer and G.C.A. Schuit, Chemistry of Catalytic Processes. McGraw-Hill Book Company, New York, 1979. M.J. Pires, M.F. Portela, M. Oliveira, A. Saraiva and T. Miranda, in: Proceedings of the 7th Iberoamerican Symposium on Catalysis, La Plata (Argentina), 1980, 189. M. Farinha Portela, C. Pinheiro, C. Dias and M.J. Pires. in: R.K. Grasselli and A.W. Sleight (Eds.), Structure - Activity Relationships in Heterogeneous Catalysis. Proceedings of ACS Symposium, Boston, Ma, April 23-27, 1990. D.B. Dadyburjor and E. Ruckenstein, J. Catal., 63, 383, 1980. P.M. Burban, G.C.A. Schuit, T.A. Koch and K.B. Bischoff, J. Catal., 126, 317 (1990).

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science and Catalysis, Vol. 12, pp. 335-343 Q 1992 Elsevier Science Publishers B.V. All rights reserved.

335

An infrared spectroscopic study of the interaction of olefins on vanadia-titania and PdCI,-vanadia-titania selective oxidation catalysts V. SANCHEZ ESCRIBANO, G. BUSCA, V. LORENZELLI and C. MARCEL lstituto di Chimica, Facolta di Ingegneria, Universita P.le Kennedy, I - 16129 GENOVA (Italy) SUMMARY The adsorption and oxidation of the simple olefins ethylene, propylene and nbutenes on the surface of vanadia-titania and PdCI,-containing vanadia titania have been investigated by FT-IR spectroscopy. Two oxidation paths have been observed on vanadia-titania, one producing species functionalized at C, (acetaldehyde from ethylene, acetone from propylene, methyl-ethyl-ketone and acetic acid from nbutenes), and the other giving compounds functionalized at C, and C,/C, (acrolein and acrylic acid from propylene, butadiene, furan and maleic anhydride from nbutenes). On PdCI,-containing vanadia-titania olefins are oxidized much faster to carbonylic compounds, that are also more stable with respect to overoxidation, without the intermediacy of alkoxide species. A mechanism involving the formation of Pdalkylidene species and their oxidation by V5+is proposed.

INTRODUCTION Vanadium-titanium mixed or supported oxides constitute well-known catalytic systems for the selective oxidation (1) and ammoxidation (2) of alkyl aromatics. They have also been tested in the selective oxidation of olefins (3-8), but they have been found to be poorly performant in it. Nevertheless, significant selectivities at low conversions have been reported in the production of either acetic acid (3) or maleic anhydride (4) from butene oxidation. Starting from propene, mixtures of acrolein, acetone and acetic acid have been obtained (5-7). The industrial production of carbonylic compounds from olefins (acetaldehyde from ethylene, acetone from propylene and methyl-ethyl ketone from butene) is or can be performed with high yields by liquid-phase processes using Pd2+-Cu2+ homogeneous catalyst systems (Wacker-Hoechst processes (9)). In recent years,

336

solid catalysts derived from the Wacker type homogeneous systems have been tested with the aim to develop heterogeneously-catalyzed gas-phase processes for direct oxidation of olefins to carbonyl compounds (10-13). Following very recent indications (12) PdCI,-vanadia-titania should be very promising ones. The present paper summarizes our data concerning the surface reactions of Pdfree and PdCI,-doped vanadia-titania with olefin gases in different conditions. The aim is to obtain information on the mechanism of olefin oxidation on the two catalyst systems and, in particular, on the role of PdCI, in the activation of vanadia-based catalysts for selective olefin oxidation. EXPERIMENTAL Vanadia-titania catalysts (9.6 % V,O, w/w) have been prepared by impregnation of TiO, (from Degussa, 50 mYg, predominantly anatase) with NH,VO, boiling water solutions, followed by drying at 400 K and calcination in air at 723 K for 3 h. PdCI, doped vanadia-titania (Pd/V atomic ratio 0.21 ) has been prepared by successive impregnation of vanadia-titania by PdCI, solution in diluted HCI. Calcination has been again performed at 723 K in air. Pressed disks of the pure catalyst powders have been outgassed at 723 K for 1 h before adsorption experiments. FT-IR spectra have been recorded by a Nicolet MX-1 spectrometer using homemade liquid nitrogen cooled heatable-evacuable cells (NaCI windows). Adsorbate gases (CO and olefins) where pure products from SIO (Milano, Italy). RESULTS a) Surface cata lvst characterization.

Characterization of the vanadia-titania catalyst has been performed using several different techniques and has been reported previously (14-18). The IR spectrum of a pressed disk after activation in vacuum presents a very weak broadish absorption near 3650 cm-I, due to vOH of free surface OH groups, a weak band at 2045 cm-' and a strong band, evident with difficulty near the cut-off limit, at 1035 cm-l. These last features are due to V=O stretchings (first overtone and fundamental, respectively) of surface vanadyl species (14-18). This spectrum is weakly perturbed by the coimpregnation of PdCI,. The overtone absorption is now rather complex with two evident components at 2038 cm-I, with a shoulder at higher frequency (near 2050 cm-'), and 1970 cm-'. A very weak band is detected at 1822 cm-' while the fundamental V=O stretching can be seen with difficulty near 1030 crn-I. Consequen-

337

tly, the most evident perturbation is the formation of the component near 1970 cm-', in the overtone region. This component can be assigned to a vanadyl group in an higher coordination state or, alternatively, in a lower oxidation state. Carbon monoxide adsorption at r.t. in the PdCI,-containing catalyst results in the formation of a reather evident band whose main maximum is at 2142 cm-I, sharp, but having a shoulder at 2160 cm-'. Moreover, the weak band at 1820 cm-' shifts to 1790 cm-I and a band also grows near 1620 cm-I. The last feature is also observed on pure vanadia-titania (1 1) and is assigned to surface carbonate species. Instead the features in the region 2100-2200 cm-' are not found on vanadia-titanias (10) and are certainly due to surface carbonyls on oxidized Pd centers (19). A comparison with the data reported recently by Choi and Vannice (20) strongly supports the identification of such centers as Pd". b) Interaction of olefins at the vanadia-titania surface. The interaction of the simple olefins ethylene, propylene and n-butenes as well as of butadiene has been the object of previous IR studies (21-24). Simple olefins except ethylene are adsorbed reactively on vanadia-titania already at room and lower temperatures. At r.t. the spectroscopically more evident adsorption products are alkoxy species, namely isopropoxide from propylene and sec-butoxide from the three n-butene isomers. Ethoxy-species are formed together with other oxidized species from ethylene adsorption starting from near 373 K. The nature of the products (secondary alcoholates from C, and C, olefins) as well as the reactivity scale butenes > propylene > ethylene (related to the electron density on the C=C bond) strongly suggest that an electrophilic attack occurs from a weakly acidic VOH group to the C=C double bond, as the first step. Such alkoxy species are easily oxidized to the corresponding carbonyl compounds (acetaldehyde, acetone and methyl-ethyl-ketone) at the expense of oxidized surface ions, in the temperature range 300-373 K in the absence of dioxygen. However, all such carbonylic compounds easily give surface enolate species in very mild conditions, probably via a hydrogen abstraction from the carbon atom in alpha-position with respect to the carbonyl group. By heating, further oxidation reactions occur. All carbonyl compounds undergo breaking of the C-C bond adjacent to the carbonyl group, giving carboxylate species corresponding to the oxidative cleavage of the C=C double bond of the starting C, and C, olefin. In the case of butenes, the C,-C, bond is oxidatively cleaved, giving mainly acetates. Acetaldehyde produced by ethylene also undergoes oxidation at the carbonyl group, as usual for aldehydes, giving again acetates.

338

CH -C-CH2-CII C l i -CH=CH-CH

3

3 8

I

.1 HC‘I ‘&

->

4

MEK

(

HC-CII

furan

~

‘d J

HC-CII

o=d/

maleic anhydride

‘01

Scheme I. Oxidation pathways of n-butenes on vanadia-titania.

1

1

I

I

C

I

I

D)

U C

m

n

L 0 Lo

n

m

I 1800

1700

1600

1500

1400

1300

1200

wavenumbers

iiO0

cm-1

Fig. 1. FT-IR spectra of the adsorbed species arising from cis-2-butene adsorption at r.t. and following evacuation at r.t. (a), 373 K (b), 423 K (c) and 473 K (d).

339

" m C

n a 0 L

nm m

1900

1800

1700

1600

1500

1400

1300

wavenumbers cm-1

Fig. 2 . FT-IR spectra of the adsorbed species arising from adsorption of cis-2-butene at 150K and warming ander evacuation to r.t. (a), 373 K (b), 473 K (c) and 523 K (d). M = typical bands of maleic anhydride (C=O stretchings). This pathway is very evident from the spectra of the adsorbed species arising from cis-2-butene adsorption, reported in Fig. 1, where features of sec-butoxides (A), methyl-ethyl-ketone (K), and acetate ions (C), produced from one another, are well evident. Such a common pathway justifies some of the selective oxidation products obtained by flow reactor olefin selective oxidation (3-8), such as acetaldehyde and acetic acid from ethylene, acetone and acetic acid from propylene, methylethyl-ketone and, again, acetic acid from butenes. However, according to the very evident lability of such selective oxidation products on the catalyst surface at temperatures similar to those of the catalytic reaction, as well as to the tendency of carboxylate species to remain strongly bonded on the catalyst and to undergo overoxidation to carbon oxides, such a path is probably the main one of the total oxidation of oiefins on the poorly selective vanadia-titania catalysts (scheme I). A second path is evidenced by low temperature adsorption of propylene and nbutenes, followed by warming and heating under vacuum. Neither alkoxy-groups nor ketons are formed following this procedure. Intermediates, identified as allyland l-methyl-ally1 species produced by allylic hydrogen abstraction can in fact be isolated by this procedure, and their evolution evidenced. Both these species produce by warming different intermediates finally giving a spectroscopically easily identifiable compound, maleic anhydride. This is very evident starting from 1butene (24), but can also be observed with 2-butenes (see Fig. 2 for the cisisomer). By comparing the spectra obtained using this procedure with those obtained by adsorption of butadiene, furan and maleic anhydride (21), the reaction scheme I appears very likely. From our data the first step of such a hydrogenabstraction pathway, leading to products functionalized at C, and C,/C, is faster

340

than the OH addition to the C=C double bond, although the active sites are perhaps less abundant and the products spectroscopically less evident. c) Interaction of olefins at the surface of PdClicontainincl catalvsts. The above results, obtained on vanadia-titania have been compared with those obtained on PdCI,-containing vanadia-titania. The presence of PdCI, strongly enhances the reactivity of vanadia-titania towards olefins. Ethylene, unreactive towards V,O,-TiO, at r.t., is adsorbed reactively at r.t. on PdCI,containing catalysts. The spectrum observed (Fig. 3) shows a relatively strong band at 1670 cm-l, certainly a C=O stretching, and a weak band at 1355 cm-I, that can be assigned to the aldehydic CH deformation mode of acetaldehyde. Relatively strong bands are also detected at 1458,1440 cm-1(CH, asymmetric deformation), 1375 crn(CH, symmetric deformation) and 1332 cm-1(very likely a CH deformation). These bands, mainly because of their .intensity relative to the preceeding ones, cannot be assigned to acetaldehyde, but to another organic species, not containing oxygen. A tentative although reasonable assignment o.n spectroscopic bases, as discussed elsewhere (25), is to an ethylidene species, likely bonded to Pd (CH,CH=Pd). No traces are found of ethoxy-groups, the most evident species when ethylene reacts (at higher temperatures) on the PdCI,-free catalyst. By heating to 373 K the features of the aldehyde (K) grow while those of the intermediate (I) decrease. Simultaneously strong bands due to acetate species (C) appear, and become predominant at 423 K. In Fig. 4 the spectra obtained by adsorption of propylene at r.t. and following , PdCI,-V,O,-TiO, and PdCI,-TiO, are heating under evacuation on V,O,-TiO, compared. While on V,O,-TiO, the predominant species is constituted by isopropoxy groups (A), on PdCI,-V,O,-TiO, both acetone (K) and isopropxy groups are detected at room temperature. On PdCI,-TiO, acetone is detected in big quantities, while isopropoxy groups are not found at all. By further heating acetone is overoxidized to acetate species (C). This transformation occurs in the range 373473 K on PdCI,-TiO,, in the range 423-473 K on V,O,-TiO, and in the range 473-573 K for PdCI,-V,O,-TiO,. Moreover, the amount of carboxylic groups produced by acetone oxidation on the Pd-free catalyst is relatively much greater than on Pdcontaining materials. This is probably related to the easier desorption of acetone from Pd than from Vanadium centers. It is evident that the presence of PdCI, induces the formation of acetone from propylene already at r.t., probably without the intermediacy of isopropoxy-groups. Low temperature experiments evidence that propylene forms intermediately an hydrocarbon species whose spectrum is consistent with that of an isopropylidene

341

(u

0

c n m

L

0 u)

5 (0

1 1800

I

1700

1600

1500

1400

1300

1200

wavenumbers

1100 cm-1

Fig. 3. FT-IR spectra of the adsorbed species arising from ethylene adsorption at r.t. on PdCI, -vanadia-titania and following evacuation at r.t. a), 373 K b), 423 K c), 473 K d).

1800

1700

1500

1500

1400

1300

1200

1100

1800

1700

1600

1500

1400

1300

1200

*aYenumbers

1100 cm-1

Fig. 4. FT-IR spectraof the adsorbed species arising from propylene adsorption at r.t. on vanadia-titania (I), PdCI,- TiO, (11) and PdCI,- vanadia-titania (Ill),following evacuation at r.t. (full lines), 373 K (broken lines), 423 K (point lines) and 473 K (dashed lines).

342

species (CH,)C=Pd (25). It is important to note that by further CO adsorption experiments we concluded that Pd is reduced to metal by propylene on the PdCI,TiO, surface, while it is not on the PdCI,-V,O,-TiO, catalyst. CONCLUSIONS The results described above evidence that two different oxidation pathways leading to Wacker-type olefin oxidation products are found on V,O,-TiO, and on PdCI,-containing V,O,-TiO,, one at vanadium, with the intermediacy of alkoxygroups, and the second one at Pd, probably with the intermediacy of alkylidene species. The second mechanism is faster that the former, occurring also in the case of ethylene at r.t.. The Pd-bonded alkylidene intermediates would transform rapidly at r.t. or even lower temperatures into carbonylic compounds by insertion of an oxygen possibly arising from a Pd-O-V bridge, with the consequent reduction of V5+near species. When vanadium is absent, Pd reduces to metal. Moreover, from our data, the overoxidation of the selective oxidation product acetone is significantly less efficient on the PdCI,-V,O,-TiO, catalyst than on the other ones. This can be attributed to its more easy desorption from Pd than from V sites. AKN0WLE DG EMENTS This work has been supported by CNR, progetto finalizzato Chimica Fine II. The collaboration of G. Oliveri is also'aknowledged. REFERENCES 1. M.S. Wainwright, N.R. Forster, Catal. Rev. Sci. Eng., 19 (1979) 21 1. 2. F. Cavani, F. Trifiro, Chim. Ind. (Milan), 70 (1988) 58. 3. W.E. Slinkard, P.B. Degroot, J. Catal. 68 (1981 ) 423. 4. M. Ai, Bull. Chem. SOC.Japan, 36 (1976) 1328. 5. A. Doulov, M. Forissier, M. Noguerol Perez, P. Vergnon, Bull. SOC. Chim. France, part I, (1979) 129. 6. T. Ono, Y. Nagakawa, H. Miyata, Y. Kubokawa, Bull. Chem. SOC.Japan, 54 (1984) 1205. 7. C. Martin, V. Rives, J. Mol. Catal., 48 (1988) 381. 8. J.L. Garcia Fierro, L.A. Arrua, J.M. Lopez Nieto, G. Kremenic, Appl. Catal. 37 (1988) 323. 9. K. Weissermel, H.J. Arpe, Industrial Organic Chemistry, Verlag Chemie, Weinheim, 1978. 10. A.B. Evnin, J.A. Rabo, P.H. Kasai, J.Catal., 30 (1973) 109. 11. L. Forni, G. Terzoni, Ind. Eng. Chem. Proc. Res. Dev. 16 (1977) 288. 12. E. Van der Heide, M. de Wind, A.W. Gerritsen, J.J.F. Scholten, proc. 91CC, Calgary, (1988), p. 1648. 13. E. Van der Heide, J.A.M. Ammerlaan, A.W. Gerritsen, J.J.F. Scholten, J. Mol.

343

Catal. 55 (1989) 320. G. Busca, J.C. Lavalley, Spectrochim. Acta 42 A (1986) 443. G. Busca, G. Centi, L. Marchetti, F. Trifiro, Langmuir 2 (1986) 568. G. Busca, Langmuir, 2 (1986) 577. C. Cristiani, P. Forzatti, G. Busca, J. Catal. 116 (1989) 586. G. Rarnis, C. Cristiani, P. Forzatti, G. Busca, J. Catal. 124 (1990) 574. N. Sheppard, T.T. Nguyen, Adv. infrared Rarnan Spectr., 5 (1978) 67. K.J. Choi, M.A. Vannice, J. Catal. 127 (1991) 465. G. Busca, G. Ramis, V. Lorenzelli, J. Mol. Catal., 55 (1989) 1. V. Sanchez Escribano, G. Busca and V. Lorenzelli, J. Phys. Chem. 94 (1990) 8939. 23. V. Sanchez Escribano, G. Busca and V. Lorenzelli, J. Phys. Chern. 94 (1990) 8945. 24. V. Sanchez Escribano, G. Busca and V. Lorenzelli, J. Phys. Chem. 95 (1991), in press 25. G. Busca, V. Lorenzelli, V. Sanchez Escribano and G. Ramis, Mater. Chem. Phys., in press

14. 15. 16. 17. 18. 19. 20. 21. 22.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science and Catalysis, Vol. 72, pp. 345-352 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

345

KINETIC PROBLEMS OF SELECTIVITY IN OXIDATION CATALYSIS S.

L.Kiperman

N.D.Zelinsky Institute of Organic Chemistry, USSR Acad. Sci.,117913 Moscow, Leninsky Prospect 47

Abstract Kinetic aspects of selectivity in heterogeneous catalytic oxidation are discussed. The correlations between characteristics of selectivity for various reaction schemes and the properties of kinetic models are considered. Kinetic descriptions of the selectivity of different processes are analyzed.

1. INTRODUCTION

Kinetic problems of selectivity in oxidation catalysis are of great importance. This is conditioned by the wide variety of intermediates exhibiting different reactivities. Some problems of selectivity connected with the kinetics of catalytic oxidation processes will be discussed in this paper as a development of our previous analysis [l-41.

2.KINETIC DESCRIPTION OF SELECTIVITY Let us consider a simple case of oxidation process described by a parallel-consecutive scheme considering AO, to be the aim

product:

346

+ 02

A

AO,

02 ---+

I11

I

02

CO,

+ +

H,O

I1 The selectivity of this process will be:

-

s = I

+

rIII rII

where r I , r I Iand r I I Iare the rates of the reactions I, I1 and I11 under fixed conditions. For parallel and consecutive schemes we have respectively: 1

'consec.

= 1

-

rIII/rI

(4)

The selectivity corresponding to more complicated cases can be described by similar equations. The kinetic description of oxidation processes should correspond to the scheme (1) and the relation of selectivity to various factors. On extrapolating the experimental data to the initial value of conversion x=O, one obtains that the selectivity Sx = o =1 only for the case of consecutive scheme. As about the other selectivity schemes, the condition Sx = o 1 is possible. Such possibility is discussed in [16].

352 4.

CONCLUSION

The kinetic approach to selectivity problems is quite fruitfull. Processes exhibiting complicated mutual influence are very interesting and their detailed kinetic studies as well as mechanistic investigations are desirable. Another problem which is of interest may be the kinetic interpretation of promotion effects and their influence on selectivity in oxidation catalysis. 5. REFERENCES

1. S.L.Kiperman. Foundations of Chemical Kinetics in Heterogeneous Catalysis. MOSCOW, llKhimijal', 1979 (russ.). ' 2. S.L.Kiperman, Kinet. Katal., 22 (1981) 30. 3. S.L.Kiperman. In: Problems of Kinetics and Catalysis, 18 (1981) 14 (russ.). 4. S.L.Kiperman. Kinetic Problems in Oxidation Heterogeneous

Catalysis. MOSCOW, VINITI Publ., 1979 (russ.). 5. A.KH.Mamedov, M.S.Kharson, N.M.Guseinov, V.S.Aliev and S.L.Kiperman, Azerb.Khim.Zhurna1, N 3 (1978) 28. 6. M.S.Kharson, A.Kh.Mamedov and S.L.Kiperman,

Kinet. Katal., 25 (1984) 107, 353. 7. G.V.Shachnovich, I.P.Belomestnich, N.V.Nekrasov and S. L.Kiperman, Appl.Catalysis, 12 (1984) 23. 8. T.Yu.Sergeeva, N.V.Nekrasov, A.S.Drjachlov and S.L.Kiperman, Khim.Promyshlennost, 9 (1981) 532. 9. A.S.Drjachlov and S.L.Kiperman. Kinet.Kata1. 18 (1977) 861. lO.A.S.Drjachlov, B.E.Ulybin, L.I.Kalinkina, V.M.Kisarov, V.S.Beskov and S.L.Kiperman, 1zv.Akad.Nauk SSSR, Ser.Khim. N 4 (1982) 861 ll.A.S.Drjachlov, V.M.Kisarov, V.S.Beskov and S.L.Kiperman, Kinet.Kata1. 24 (1983) 104. 12.S.L.Kiperman. In: Theoretical Problems in Kinetics of Catalytic Reactions. Chernogolovka. 1984, p.12 (russ). 13.A.S.Drjachlov and S. L.Kiperman, Dokl.Akad.Nauk . SSSR. 258 (1981) 931.

14.A.S.DrjachlovI Yu.S.Burkin, A.A.Frontinskii, V.M.Kisarov, V.S.Beskov and S.L.Kiperman, Kinet.Katal., 24 (1983) 1406. 15.A.S.DrjachlovI N.V.Zhdanovich, L.I.Kalinkina, G.A.Foksha, V.M.Kisarov and S.L.Kiperman, In: Catalytic Purification of effluent Gases (russ.) I11 All Union Conference, Part 1. Novosibirsk, 1981, p.65. 16.Yu.I.Pjatnitskii an2 O.P.Nesterova, Kinet.Kata1. , 30 (1989) 1401.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Sludies in Surface Science and Calalysis, Vol. 72, pp. 353-362 1992 Elsevier Science Publishers B.V. All rights reserved.

353

Site Isolation in Vanadium Phosphorus Oxide Alkane Oxidation Michael R. Thompsona and Jerry R. EbneP aThe Molecular Sciences Research Center, Pacific Northwest Laboratory', Battellc Blvd, Richland, Washington, USA 99352 bMonsanto Company, 800 North Lindbcrgh Avenue, Saint Louis, Missouri, USA 631 14

Abstract Single crystal X-ray diffraction studies of vanadyl pyrophosphatc indicate that at least two polytypical structures cxist for this active and sclcctive alkanc oxidation catalyst. The crystal structures of these materials diffcr with rcspcct to thc symmctry and dircction of columns of vanadyl groups within the unit cell. Single crystals of vanadyl pyrophosphate have bccn gencrated at extreme temperatures not often cxpcricnced by microcrystallinecatalysts. The crystallography of the system suggests that othcr crystallinc modifications or disordcrcd phascs might also cxist. Zeroth-order models of crystal surfacc tcrmination of vanadyl pyrophosphatc havc bccn constructed which conceptually illustrate the ability of vanadyl pyrophosphate to accomrnodatc varying amounts of surface phosphorus parallcl to (1 ,O,O), (0,1,0) and (0,2,4). Pyrophosphatc termination of surfaces parallel to (1,O.O) likely results in the isolation of clusters of reactivc centers and limits overoxidation of the alkanc substratc.

1. INTRODUCTION

The vanadium phosphorus oxide (VPO) catalyst systcm performs the fourteen-electron oxidation of n-butane to maleic anhydride with exccptionally high selectivity [2]. The phase which is identified with optimal catalytic performance is vanadyl pyrophosphate, (VO)2P2O7 [3]. However, the literature is anything but clcar conccming the composition and structure of this phase. The solid-state chemistry of the VPO system is typified by the facile interconversion of numerous V4+ and V5+ phases, leading to considcrablc confusion in the interprctation of the experimental data. In an attempt to clarify the slructural chemistry of these materials, our work has recently focused on a careful and exhaustive re-investigation of the crystal structure of vanadyl pyrophosphate. These crystallographic studies indicate that the differences observed in X-ray powder patterns of catalysts derived from differing syntheses [4] arc consistent with the prcscnce of polytypical structures of vanadyl pyrophosphate. Our attempt to understand the relationship between variable metal atom order and catalytic performance has led us to postulate "zeroth-order" models of crystal surface termination. Regardless of the dubious nature of such models, these studies raise some intriguing questions related to the evolution of both bulk and surface structure during catalyst preparation and bum-in.

354

2. DISCUSSION

The literature contains numcrous citations conccming all aspects of thc catalytic chemistry of the vanadium phosphorus oxidcs [ 5 ] . Clearly, controvcrsy exists on several important structural issucs, including: (i) the cxact idcntity and structure of thc activc/sclcctivc phasc, (ii) thc rolc of V5+ specics rclcvant to butanc convcrsion to malcic anhydride, and (iii) the rclationship bctwccn surfacc atomic P:V ratios and catalyst sclcctivity. Thc intcnt of this papcr is to cxpound on thc idca that the solid-statc structure of vanadyl pyrophosphatc may bc highly variablc, cspccially with rcspect to metal atom ordcr and apparcnt phosphorus composition. Wc havc isolatcd and dctcrmincd the crystal structures for two polytypical forms of vanadyl pyrophosphatc dcrived from ncar solidificd mclts of mature microcrystallinc catalyst powders. Whilc thcsc studies clarify thc poorly dctcrmincd structure previously rcportcd by Lindc and Gorbunova [6], thcy also suggest that other forms of vanadyl pyrophosphatc may cxist. As for thc role of V5+in butanc oxidation to maleic anhydride, Trifiro ct al. [7] havc cstablishcd that sclcctivc VPO catalysts possess some limited and controllcd numbcr of V5+ sitcs and that the sclcctivity to malcic anhydride passcs through a maximum for a well-dcfincd value of dcgrcc of surfacc oxidation. Scvcral authors bclicvc that V5+ exists at the catalyst surfacc as p-voP04 or a structurally related amorphous state [8]. Howcver, in our cxpcricncc, dctcctablc quantitics of VOPO4 phascs in commcrcial catalysts can be shown to be associatcd with pcrformancc loss. Possibly the most controversial issuc relaling to the composition of thc catalyst conccms the prcsencc of cxccss phosphorus associated with catalyst surfaccs, and its rolc in stabilizing V4+ spccics and dctcrmining the sclcctive properties of the catalyst. Garbassi ct al. [91 havc found a value of surface atomic P:V ratios in thc range of 2.0-2.8, while that for bulk lies in the range of 1.0-1.4. Hodnctt and Dclmon [ 101 report that surface P V ratios similar to those reported by Garbassi. Recent work by Okuhara et al. [ 1I] supports the lattcr result, yiclding surfacc atomic P:V ratios of 1.10 k 0.04. As will bc shown below, it is not difficult to conccivc of structures bascd solcly on vanadyl pyrophosphate which support both a controllcd number of surface oxygcn ( and V5+) sitcs and accommodation of varying amounts of non-stoichiomctric surfacc phosphorus.

2.1

Crystallography of Single Crystals of Vanadyl Pyrophosphate.

We have reported that carefully controllcd recrystallizationof mature commercial catalysts at temperatures in excess of 9700C results in the formation of large single crystals of vanadyl pyrophosphate [12]. Spccimcns harvcstcd from thcse ncar solidificd mclts are variable in color, ranging from emerald-grccn to gray, and brown to rcd-brown. Emerald-grccn and rcd-brown crystals of vanadyl pyrophosphate crystallize in the non-ccntrosymmetric spacc group Peal [13]. AU crystals studied exhibit some disordcr of thc vanadium atom sites. Thc disordered positions lic approximately 0.325A above or below the distorted octahedral basal plane of the vanadium coordination sphere as illustrated in Figure 1. Emerald-grecn crystals of vanadyl pyrophosphatc

355

are composed of the structure earlier reported by Linde and Gorbunova. However, the structure exhibits vanadium atom disorder associated with only half of the metal atom sites. Addition of the disordered sites in cycles of least-squares refinement of the crystallographicmodel result in residual values in the range of R1=0.035. Bonding interactions for all metal-oxygen and phosphorusoxygen bonds fall within expected values including the vanadyl moieties, V=O, which average 1.604 (17) A, and are all within two esds of the average value. The disorder is a consequence of the coexistence within the crystal of both enantiomorphs of the Linde and Gorbunova structure.

0

0

II

II

Figure. Vanadium site disorder across the pyramidally distorted octahedral basal plane. Crystallographic models of the two enantiomorphs can be superimposed with respect to the pyrophosphate network, however, the resulting structure exhibits disorder for the vanadium atoms which lie in chains parallel to the c-axis and y=1/2, as illustrated in Figure 2. A similar interpretation of the disorder within the crystal structure of the red-brown material would indicate that it consists of a polytype of the Linde and Gorbunova structure in which the direction of the

A model of the superimposed pyrophosphate networks of the Linde and Gorbunova structure of vanadyl pyrophosphate and its enantiomorph. Note that only half of the V-sites are disordered.

vanadyl columns parallel to the crystallographic c-axis at approximately y=O, are reversed in direction with respect to the a-axis. Superimposing the pyrophosphate network of this second

356

polytype with its enantiomorph yiclds a model in which all vanadium atom sites are disordered across the octahedral basal planc. This typc of disordcr. common in othcr transition mctal crystal structures, is termed "linear displasive disordcr" and is noted for producing diffraction streak cffccts similar to hose reportcd by Bordcs [I41in electron diffraction studies of microcrystallinc catalysts. The differences between thc two proposcd polytypical vanadyl pyrophosphatcs can be understood in terms of the symmetry of thc eight columns of vanadyl groups that exist within the unit cell. Thesc differences are schematically illustratcd in Figure 3. The dircclion of the vanadyl columns rcprcscntcd for thc first polytypc corrcspond to thc symmctry rcportcd by Lindc and Gorbunova. Thc direction of the vanadyl columns along the cdgc of the ccll, parallel to thc c-axis, have bcen rcvcrscd in the sccond polytypc. Thc distancc rclationships bctwccn all vanadium, phosphorus and oxygen atoms within Ihc idcalizcd modcls of thcsc two structurcs are idcntical. b=9J70A

Polytype I

Polytype I I

Figure. Schematic representation of the vanadyl column symmetry for two polytypical vanadyl pyrophosphatc crystal structurcs.

and as a result, we would expect the crystal energies of thcsc spccics to be nearly idcntical. No phosphorus atom disordcr is observed for eithcr sct of single crystals. While the structurcs of both polytypes would exhibit the same rigorous space group extinctions (Pcazl), thc symmetry relationships between thc vanadium and phosphorus atoms within the structure arc not idcntical. This obscrvation is also consistcnt with thc diffcring Raman spectra exhibited by the two crystalline spccics [15]. It is clear that the topotaxy which transforms the intercalated orthophosphate precursors into the nctworked pyrophosphate structure 1161 occurs with considerable rcorganization of thc long-range ordcr of both thc phosphate and vanadyl networks. Extreme broadening of scveral classes of reflections in X-ray powdcr diffraction patterns, apparent even in mature microcrystallinc catalysts, supports the conclusion that crystalline ordcr in the system is highly variablc. Wc can conccivc of numcmus altcrnativc symmctrics for thc structure of vanadyl pyrophosphate which retain identical bonding shcll configurations, but would differ in second- and further-near ncighbor environments. We would cxpcct these structures to

357

exhibit only nominal energetic differences associate with the coulombic interactions between differing metal-metal octahedral hole occupancy, and strain energy associated with the ability of each structure to pack efficiently (minimization of cell volume). The obvious question which we find ourselves asking relative to the structures of the single crystal materials and the microcrystalline catalysts, relates to kinetic vs. thermodynamic control in the evolution of thc structure in the transformation from catalyst precursors to the vanadyl pyrophosphatc phase. Is it possible that the acentric structures observed for the single crystals studied by Linde and Gorbunova, Middlemiss [173 and ourselves are simply a set of thermodynamically stable species, or do these materials form as a result of the synthesis procedures and only slowly, if at all, interconvert? The importance of this question can be realized when considering that the direction and symmetry of the vanadyl moieties within the structure have an cffcct on any proposed surface topology at the (1,0,0) surface of vanadyl pyrophosphatc.

2.2

Zeroth-Order Models of Vanadyi Pyrophosphate Surface Termination

During the past decade there has been a remarkable increase in the detail of information available concerning the atomic geometry, bonding and electronic structure of surfaces. Much of this work, both theoretical and experimental, has concentrated on surface reconstruction of cleavage faces of tetrahedrally-coordinarcdcompound semiconductors [181 and simple oxide materials [19]. The theoretical research, primarily based on the empirical Tight-Binding model [20], has been instrumental in guiding the intcrprctalion or experimental data and has lead to mechanistic schemes which explain the reconstruction process in tcnns of surface bond rehybridization. Higher levels of theory will be necessary to properly describe the bulk and surface structures of octahedral mctal-oxidcs and systems such as vanadyl pyrophosphatc. However, the necessary prerequisite to theoretical studies involves the construction of "zeroth-order" models of the bulk and surface structures. Zeroth-order surface models are simply based on a rational termination of the crystal structure parallel to the desired surface, in a manner which preserves maximum bond valance for each atom under the constraint of generating a neutral surface. Needless to say, these models are a matter of conjccture, but they often help rationalize less than obvious features of the surface chemistry. We would like to advance scveral simple concepts with respcct to models of surface termination for VPO phases. (i) Surface vanadium centers are expected to be minimally five-coordinate. The sixth-coordination site could be unoccupied for the case where the vanadyl oxygen is inward directed, or coordinated by a weakly interacting adduct. Alternatively, the sixth coordination site could be occupied by an outwardly directed surface vanadyl oxygen. (ii) Since the amount of phosphorus associated with the surfams of W O phases is clearly a matter of interest, our models will maximize the use of phosphorus by fulling the coordination sphercs of all metal atoms with shared pyrophosphate oxygen. We think this is a physically reasonable assumption since the active/selective phase is synthesized in an excess of phosphate. "he effect generated in this manner

358

will illustrate the structure that would exist in a phosphorus saturated vanadyl pyrophosphate phase. (i) Neutral surface termination is most easily accommodated by appropriately protonating the dangling phosphate groups (c.f., as in the intcrcalated structures of the vanadyl hydrogenphosphate system [21]). Figurc 4 illustrates one such model of crystal surfacc tcrminadon for vanadyl pyrophosphatc. The perspective shown in Figurc 4 is in projcction of the (1.0.0) surfacc, with crystal "clcavagc" parallel to the (0,l.O) and (0,2,4) planes (hydrogen termination has bccn excludcd for thc purpose of clarity). Examination of thc composition of the "surfacc layer" parallcl to cithcr (0,1,0) or

An illustration of phosphorus-rich crystal tcrmination for vanadyl pyrophosphate, parallel to (0,1,0) and (0,2,4), and in projcction of (1,O.O).

(0,2,4) reveals that each would posscss somc degree of excess phosphorus, the exact proportion of P:V being dependent on the sampling dcpth. Fiyrc 5 further illustrates two possible surface terminations for (l,O,O). The surfacc at thc top of the figure illustrates pvrophosphatc surracc termination parallel to (1,O.O) a:ld would posscss one-half of onc equivalcnt of nonstoichiometric (excess) phosphorus. Thc surfacc depictcd at the botlom of F i y r c 5, which is terminated with surface orthophosphatc groups, would possess stoichiomctnc quantities of phosphorus.

2.3

Site Isolatlon, Phosphorus and Metal Atom Symmetry and Active Oxygen

Several key features of the bulk and surface structures of vanadyl pyrophosphate rclatc to thc non-centrosymmctric naturc of the pyrophosphate network. Unlike the pyrophosphatc, thc crystal structures of the precursor phascs exhibit centrosymmctric structures. Six phosphatc groups

359

e=16.600h

b=9.570h

1=7.710A

Illustration of two possitilc cascs of surfacc tcrmination parallcl to thc (1,O.O) surface of vanadyl pyrophosphacc : phosphorus-rich pyrophosphatc termination (top of figure), and stoichiometric orthophosphatetermination (bottom of figure). sumund each dimeric pair of vanadyl centers of the layered VOHP04. 112 H20 structure. Of the six hydroxyl moieties associated with the phosphate groups, four are oriented above the basal plane, and two below (or vise vcrsa) as illustralcd in Figure 6a. For thc pyrophosphatc phasc, three pymphosphatc groups bridge laycrs in eithcr direction but do so wilh a maximum of 2-fold symmetry as shown in Figure 6b. This non-ccntrosymmctric stmcturc, as reported by ourselves and previous authors, cannot bc constructcd with fcwcr than four indcpcndcnt phosphorus atoms rcgardlcss of thc symmctry of thc vanadium occupancy (i.c., thc assignmcnt of the space group symmetry is m a x i u that of Pca21). As we havc previously dcscribcd [12], terminating the idealized (1,0,0) surfacc of (VO)2P2O7 with pcndcnt pyrophosphatc groups sterically isolates vanadium centers in cavitics or clcfts. The degrce of isolation of thcsc centers, and the symmetry in the cavity, is influenced significantly by the oncntation of the vanadyl columns within thc structure, a factor which wc have notcd to be different for each of thc poly-

OH

6H

Figure. Idealizcd polyhedral structures for (a) VOHP04.112 H20, and (b) (V0)2P207.

360

typical vanadyl pyrophosphates. Howcvcr, the surface cavitation itsclf is a consequence of the pyrophosphate symmetry and would bc lcss pronounced for a highly symmctric structurc. Thcsc models of surface topology provide a means for active site isolation, an important general property for Selective oxidation catalysts. First dcscribcd by Grasselli [22], thc sitc isolation principlc requires that active oxygen be distributed in an arrangement that providcs for limitation of numbcrs of active oxygen in various isolated locations so as to restrict overoxidation. But what might be the role played by V5+ Centers within the context of this structure, and what possible forms of active oxygcn are associated with the surface cleft? The single crystal X-ray studies c o n f i i the transconformation of the vanadyl moieties across the vanadium ccntcred dimcr of vanadyl pyrophosphate. Consider that two adjacent dimeric units at the (1,0,0) surface of vanadyl pyrophosphate possess two open coordination sites and a potential to donate four clcctrons to a surface adsorbed 0 2 molecule. It is not difficult to conceive that if the (1,0,0) surface is reprcscnted as composed of coordinatively saturated vanadium centers, that the valance of all surfacc laycr mctal atoms must have a formal oxidation statc of +5.

3. CONCLUSIONS

The isolation of large single crystals of vanadyl pyrophosphate have indeed led to a resolution of the controversial stmcture carlicr reported for this matcrial. Howcvcr, as has bccn reported by Volta [8], Bordes [23] and others, thc X-ray powder patterns of microcrystalline VPO catalysts often possess additional pcaks of uncertain origin. We wish to point out that thc entire odd-odd-odd panty group suffers from an order-of-magnitude broadening in X-ray powdcr patterns of the microcrystalline catalysts as comparcd to the single crystal materials. Many of the often cited peak intensity differences and small shifts in spacing for reflections in powdcr patterns are fully consistent with changes in the crystal structure of vanadyl pyrophosphate. Secondly, surface termination in VPO phases can occur with apparent addition of excess amounts of phosphorus. These @yro)phosphates would be expected to stabilize the V4+ oxidation state by coordinating with vanadium, and blocking access to the (0,1,0) and (0,2,4) surfaccs. Full oxidation of the vanadium centers associatcd with surfaces parallcl to (1,0,0) would rcsult in coordinative saturation of all surface vanadium atoms. We believe that the addition of onc half of one equivalent of excess phosphorus to (1,O,O) would provide site isolation of thc active oxygen at that surface. Future experimental work will center on anglc-resolved surface XPS rncasurements of macroscopic single crystal materials, EXAFS, and variable tcmperaturc/pressure Rarnan spectroscopy to probe metal and phosphorus atom order in singfe crystal and microcrystalline materials.

361 4.

REFERENCES

1. Operated by BattelIe Memorial Institute for the United States Department of Energy under

contract DE-AC06-76RLO-1830. Support for this research is provided by the Office of Conservation and Renewable Energy, Advanced Industrial Concepts. 2. Cavani. F.; Centi, G.; Trifiro, F.; Grasselli, R.K.; Preprints, ACS Symposium of the Division of Petroleum Chemistry, "HydrocarbonOxidation", New Orleans Meeting, Sept. 1987. 3. Bordes,E.; Courtine, P.; J. Chem. Soc., Chem. Commun., (1985) 294; Wenig, R.W.; Schrader, G.L.; Ind. Eng. Chem. Fundam., (1986) 2, 612; Cavani, F.; Centi, G.; Trifiro, F.; Appl. Catal., (1984) 9, 191.

4. Centi, G.; Trifiro, F.; Busca, G.; Ebner, J.R.; Gleaves, J.T., Faraday Discuss. Chem. SOC., (1989) 215. 5 . Ebner, J.R.; Franchetti, V.; Centi G.; Trifiro, F., Chem. Rev., (1988) 88, 55.

6. Linde, S.A.; Gorbunova, E., Dolk. Akad. Nauk, SSSR (English Trans), (1979)

x, 584.

7. Cavani, F.; Centi, G.; Trifiro, F.; Vaccari, A., in "Adsorption and Catalysis on Oxide Surfaces" ; Che, M., Bond, G.C., Eds.; Elsevier. Amsterdam, (1985) 287. 8. Bergeret, G.; David, M.; Broyer, J.P.; Volta, J.C.; Hecquet, G., Catal. Today, (1987) 1,37; Berget, G.; Broyer, J.P.; David, M.; Gallezot, P.; Volta, J.C.; Hecquet, G., J. ChemSoc., Chem. Commun., (1986) 825. 9. Garbassi, F.; Bart, J.C.; Tassinari, R.; Vlaic, G.; Lagarde, P., J. Catalysis, (1986) B,317. 10. Hodnett, B.K.; Permanne, Ph.; Delmon, B., Appl. Catal.. (1983) 4, 231; Hodnett, B.K.; Delmon, B., ibid., (1984) 2 , 4 6 5 . 11. Okuhara, T.; Nakama, T.; Misono, M., Chem. Letters, (1990) 1941.

12. Thompson, M.R.; Ebner, J.R., in "Studies in Surface Science and Catalysis, Vol. 56: Structure-Activity Relationships in Heterogeneous Catalysis", Sleight, A.W., Grasselli, R.K., Eds., Elsevier, Amsterdam, 1990. 13. The space group Pcasl is !he standard setting of that reported by Linde and Gorbunova (reference 6). Lattice constants for this setting for emerald green crystals: a= 7.710(2)A, b=9.569(2)A c=16.548(3)& Those for red-brown crystals: a=7.746(2)A, b=9.606(2)& and C= 16.598(3)A. 14. Bordes, E.; Courtine, P., in: "Proceedings, 10th International Symposium on Reactivity of Solids," Dejon, France, Sept. 1984, p. 512; Bordes, E.; Johnson, J.W.; Material Science Monogr., (1985) m,887. 15. Freeman, J., Internal Monsanto Report.

16. Bordes, E.; Courtine, P.; Johnson, J.W., J. Solid State Chem., (1984)

s,270.

17. Middlemiss, N.E., doctoral dissertation, Department of Chemistry, University, Hamilton, Ontario, Canada, (1978). 18. Duke, C.B., in: "SurfaceProperties of Electronic Materials," King, D.A.; Woodruff, D.P., Eds., Elsevier, Amsterdam, (1987) Chpt 3.

362

19. Duke, C.B.; Thompson, M.R., in: "Proceedingsof Materials Research Society, Fall 1989" , Symposium C, Boston, MA. 20. Chadi, D.J., Phys. Rev. B.. (1979) 2, 2074 21. Torardi, C.C.; Calabrese, J.C., Inorg. Chem., (1984) 23, 1308. 22. Grasselli, R.K., in: "Surface Properties and Catalysis by Non-Metals", Nonnelle, J.; Derouane, E., Eds., Elsevier, Amsterdam, (1983) p. 273. 23. Bordes, E.; Courtine,P.; Johnson, J.W., J. Solid State Chem., (1984) 55, 270.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Sutface Science arid Catalysis, Vol. 12, pp. 363-317 0 1992 Elsevier Science Publishers B.V. All rights reserved.

363

Synergy Effects in Selective Oxidation Catalysis Umit S. Ozkan, Marianne R. Smith, and Sharon A. Driscoll The Ohio State University, Department of Chemical Engineering, Columbus, Ohio 43210, USA

Abstract Studies performed over multi-phase selective oxidation catalysts such as MnMo04/Mo03 and CdMo04/Mo03 for the partial oxidation of C, hydrocarbons have shown the presence of strong synergy effects. This phenomenon was investigated through detailed catalyst characterization and reaction kinetics studies. Pure phases as well as multi-phase catalysts were characterized using techniques such as BET surface area, X-ray diffraction, scanning electron microscopy, energy dispersive X-ray analysis, laser Raman spectroscopy and Xray photoelectron spectroscopy. Steady-state reaction experiments examining the effect of oxygen partial pressure, transient response studies using isotopic labeling, and oxygen chemisorption experiments were performed to investigate the catalytic job distribution among different components of the active catalyst. More recent studies combining in-situ laser Raman spectroscopy with isotopic labeling technique provided further evidence of the mechanism of synergy in these model multi-phase catalysts. Using a specially designed in-situ cell, samples were reduced by 1,3-butadiene and then reoxidized by labeled oxygen, thus allowing observation of Raman band shifts which could be identified with catalytic sites responsible for specific steps in the oxidation scheme. INTRODUCTION Partial oxidation reactions play an important role in the chemical industry, producing a large number of products and intermediates that include alcohols, aldehydes, ketones, and acids. The successful partial oxidation catalysts are often complex, multi-phase metal oxides with catalytic properties decidedly different than those of the individual constituents. The catalytic job distribution among different components of the catalyst is often quite involved, with multiple sites taking part in the reaction scheme [1-3]. One important application of selective oxidation reactions is the formation of maleic anhydride from C, hydrocarbons. In the last decade, in addition to widespread industrial applications, there have also been several fundamental studies investigating various catalytic phenomena involved in these reactions. Although most of these studies have focussed on the vanadium-phosphorus-oxide (V-P-0) catalysts [4-lo], some studies have also been published reporting high activity and selectivity levels over molybdenum oxide-simple molybdate based catalysts [l 1161. It is generally accepted that 1,3-butadiene and furan are two of the

364

intermediates in the reaction network leading to the formation of maleic anhydride from l-butene. However, there are still unanswered questions about the oxygen insertion mechanism and the role of lattice and gas phase oxygen in the overall reaction scheme. Our earlier studies revealed the presence of strong synergy effects in multi-phase catalysts that contained a molybdenum oxide phase in close contact with a simple molybdate phase [15-191. Transient and steady state kinetic investigations found the two-phase catalysts (MnMo04/Mo03, CdMo04/Mo03) to be more selective for maleic anhydride than either pure phase. The two pure phases were found to have markedly different responses to the presence or absence of gas phase oxygen and its concentration. Oxygen chemisorption experiments and transient isotopic labeling studies, in which 1802was used as the gas phase oxidant, further revealed the source of oxygen for MOO, to be the catalyst lattice, while the molybdate phase was found to adsorb and utilize gas phase oxygen much more effectively. A catalytic job distribution was postulated to explain the synergy observed for the two-phase catalyst. According to this scheme, the active sites for the formation of the selective oxidation products are located on the MOO, phase. MOO, is responsible for selective oxidation of the hydrocarbon to furan or maleic anhydride through use of its lattice oxygen while the MnMo04 phase contains the sites that can chemisorb the gas phase oxygen and allow it to migrate to the reduced sites on MOO, surfaces in an activated form through a spillover mechanism. Evidence for the existence of such a spillover phenomenon in multi-phase catalysts in selective oxidation reactions forming acrolein from propylene and methacrolein from isobutene have also been reported by Delmon and his coworkers [20-221. Our recent studies combining in-situ laser Raman spectroscopy and isotopic labeling techniques have provided further evidence of the catalytic job distribution that was suggested through our earlier studies. These combined techniques offer a powerful tool that can be especially valuable in identifying catalytic sites and their role in the catalytic scheme. Although the effectiveness of this technique has been demonstrated previously [23,24] the number of applications in catalysis research is still rather limited. EXPERIMENTAL METHODS Catalyst Preparation Pure phase MOO, was used as received (Aldrich). Pure phase MnMo04 was prepared by precipitation from aqueous solutions of manganese chloride (MnCI24*H2O) and ammonium heptamolybdate ((NH4)6M07024). For cadmium molybdate a solution of cadmium nitrate (Cd(NO,),*4H2O) was used. The twophase catalyst was prepared using a stepwise "wet impregnation" procedure in which molybdenum trioxide was soaked in an aqueous suspension of manganese or cadmium molybdate. These procedures are described in detail elsewhere [15, 161.

MOO, catalysts were also prepared with varying side-to-basal crystal plane ratios using tempererature-programmed calcination and recrystallization techniques. Temperature-programmed calcination of MOO, gave crystals which

365

were thick and round (Mo03-C), while melting the MOO, followed by recrystallization via rapid cooling resulted in long, thin, ribbon-like crystals (Moo3R). This technique has been described previously [25]. Catalyst Characterization All catalysts used in these studies have been characterized by a combination of techniques. Physical characterization of each catalyst included measurement of the surface area by the BET technique with a Micromeritics 2100E Accusorb instrument, using both nitrogen and krypton as the adsorbing gas. X-ray diffraction (XRD) patterns were obtained using a Scintag PAD V diffractorneter with Cu K, radiation as the incident X-ray source. Scanning electron microscopy was performed using a Hitachi S-510 scanning electron microscope. Threedimensional imaging technique was used in conjunction with stereo pairs of micrographs to obtain accurate areas for the calculation of side to basal plane ratios of Mo03-C and Moo3-R. Stereo images of the catalyst samples were digitized on a VAX 8550,which then performed calculation of the various crystal plane areas.

Compositional analyses of the catalyst samples were carried out using energy dispersive X-ray analysis on an EDAX 9100. X-ray photoelectron spectroscopy was performed using a V.G. Scientific X-ray photoelectron spectrometer. The oxygen uptake was measured over all three catalysts in both fresh and reduced form using a static adsorption system equipped with high-temperature chemisorption furnaces (Micromeritics 21OOE Accusorb) as described previously ~91. Laser Raman spectroscopy (LRS) has also been used to characterize each catalyst. Spectra were collected in the back-scattering mode using a Spex 1403 laser Raman spectrometer equipped with a Datamate microprocessor for data collection and processing. The 514.5-nm line of a 5-W Ar ion laser (Spectra Physics) was used as the excitation source. The laser power was 30 mW, with a scanning rate of 1.O sec/cm-l. An in-situ technique coupled with isotopic labelling was also developed for the laser Raman spectrometer which allowed for collection of spectra at high temperatures and with a variable atmosphere [26]. A quartz cell with optically clear windows was used to hold the sample. The cell allowed gases to pass through the catalyst, and exit to a vent or to analysis. For the in-situ experiments, the laser power was set at 100 mW. Each of the catalysts, the two-phase and each pure phase component, was first reduced under a stream of butadiene/N2 mixture. The sample was then reoxidized with 1802. Reaction Experiments Steady-state selective oxidation experiments were carried out using a fixedbed, integral reactor. Gas chromatographs with thermal conductivity and flame ionization detectors were used for analysis. Feed compositions, reaction

366

parameters, and the reactor and analytical systems have been described in detail elsewhere [15, 181. Transient response studies using isotopic labelling technique were performed using a pulse microreactor. The reactor was connected directly to the carrier gas line of a gas chromatograph-mass spectrometer system (Finnigan 4000)for analysis [19]. RESULTS

Our earlier studies clearly showed that the two-phase catalysts (MnMoO,/MoO,, CdMoO,/MoO,) were more selective for the formation of maleic anhydride than their pure phase constituents [15, 161. Comparison of selectivities at equal conversion levels reiterated this observation. It was also observed that the difference in selectivities became more pronounced as one moved from 1-butene to 1,3 butadiene and finally to furan as the starting material. As for the behavior of the single phase catalysts, it was seen that MOO, was capable of forming maleic anhydride, although the overall activity was low and the yield of complete oxidation products (CO,CO,) was higher. Manganese molybdate, on the other hand, showed no selectivity for maleic anhydride, but proved to be very active as a complete oxidation catalyst. Careful characterization of the multi-phase and singlephase catalysts using X-ray diffraction, X-ray photoelectron spectroscopy, and laser Raman spectroscopy showed no evidence of a new crystallographic phase being formed on the two-phase catalysts [15]. Scanning electron microscopy and laser Raman spectroscopy clearly exhibited the coexistence of the two phases in close contact [15], leading us to postulate that there was a strong synergy in operation between the two phases. Our later studies focussed on assessing the utilization of oxygen by the pure phases. Some of the results from those studies, which have been published previously [18, 271, are summarized in Figures 1 and 2. Figure 1 compares MOO, and MnMo0, phases in their response to changes in oxygen concentration in the feed gas in 1-butene oxidation, while Figure 2 shows this comparison for 1,3butadiene oxidation. When the slopes of the total conversion curves are compared for the two catalysts, we see that MnMo0, is much more sensitive to the concentration of gas phase oxygen, with its activity increasing very rapidly with increasing oxygen concentration. The molybdenum oxide phase, however, did not seem to be affected as much by the changes in oxygen concentration. High temperature chemisorption studies were performed over both fresh catalysts and catalysts reduced with hydrogen to compare the oxygen uptake capacities of the two pure phases [19], and the results are presented in Table 1. When fresh samples were used, MOO, was seen to chemisorb the smallest amount of oxygen. MnMoO,, on the other hand, appeared to be much more adsorbent towards oxygen, adsorbing close to five times more oxygen per unit surface area than did pure MOO,. Oxygen uptake measurements were also performed on catalyst samples which were reduced with hydrogen at 400OC. Since part of the oxygen taken up by each catalyst is used to reoxidize the reduced site, the oxygen uptake value also reflects the degree of reduction for each catalyst. After reduction,

21

1

I

I

1 - Butene Oxidation V Overall Conversion

30

Q)

0 A Yield of Maleic CO, Anhydride

'(a)

Yield of 1,3 -Butadiene Yield of Furan

0

F25

E

1

367

! . 3.90

7.80

i 1

0 1

11.71

15.61

19.51

0, Concentration (mole %) I

I

I

I

I

I

A Yield of Maleic Anhydride

-

I

3.90

7.80

*

-

d

11.71

15.61

-

I 19.51

0, Concentration (mole %) Figure 1. Variation of conversion and yield with 0, concentration in oxidation of 1-butene over a)Mo03; b)MnMo04. Reprinted from J. Catal., 122 (1990) p. 454.

368 30 I

2

a,

'F b

I

I

I

I

I

I, 3 - Butadiene Oxidotion V Overall Conversion A Yield of Maleic Anhydride H Yield of Furan 0 Yield of Acrolein 0 Yield of COX

I 50

-

40

-

30

-

20

20-

c

.-

2 a,

15-

>

s C

10-

8 - 10

0d

3.9

7.8

11.7

15.6

0

19.5

O2Concentration (mole %) I

I

-

I

I

I

I

I , 3 - Butadiene Oxidation V Overall Conversion A Yield of Maleic Anhydride H Yield of Furon 0 Yield of Acrolein 0 Yield of COX

I100 -

90

-

80

- 70

C

-

60

-

50

- 40

c 0 0

- 30

& -1:

8

0

3.9

7.8

11.7

15.6

19.5

0

O2Concentration (mole%) Figure 2. Variation of conversion and yield with 0, concentration in oxidation of 1,3-butadiene over a)Mo03; b)MnMo04. Reprinted from J. Catal., 123 (1 990) p. 175, 176.

369

MOO, was seen to chemisorb 3.3 times more oxygen than MnMoO,, larger degree of reduction than that of MnMoO,.

indicating a

Table 1 Oxygen uptake over single-phase catalysts (pmol/m2) Fresh Catalyst MOO, MnMo04

0.3348 1.429

Reduced Catalyst 43.75 13.30

Isotopic labeling studies provided more definitive evidence of the role of lattice and gas phase oxygen in the complete and selective oxidation over the two components of the two-phase catalyst [I 91. Figures 3 and 4 show the distribution of furan and CO, isotopes over the pure phases in transient oxidation of 1,3butadiene with lag2in the gas phase. It is seen that the Moo3 catalyst utilized almost exclusively lattice oxygen in the formation of furan. The involvement of the lattice oxygen did not show a rapid decline with the pulse number. In contrast to the MOO, catalyst, the gas phase oxygen was seen to be incorporated into the hydrocarbon molecule more readily over the MnMo0, catalyst. In the first pulse about 25% of all the oxygen incorporated into furan was derived from the gas phase. The contribution of the gas phase oxygen rose to about 35% by the fifth pulse. Upon examination of the CO, isotope distributions, pure MOO, did not show any Cl8O,. However, the use of gas phase oxygen was much more substantial in the formation of CO, than it was in the formation of furan, with the relative amount of C180160being around 30%. Over MnMoO,, both Ci8Oi60 and Ci8O, relative percentages increased rapidly with the pulse number, ranging from 22 to 39% for C180i60 and from 2 to 8% for Ciao,. In transient response experiments, MnMoO,/MoO, was the only catalyst to yield substantial quantities of maleic anhydride. Over this catalyst, no maleic anhydride was detected which had all three oxygen atoms labeled. In the first pulse the relative percentages of C4H21603, C4H21801602,and C4H21802160 were 65, 31, and 4%, respectively. For the fifth pulse, these percentages were 61, 34, and 5%. These numbers showed that close to 90% of all of the oxygen incorporated into the hydrocarbon molecules to form maleic anhydride came from the crystal lattice of the catalyst and the increase of the gas phase oxygen contribution was very gradual.

370

1, 3 - Butadiene Oxidation with I8O2over MOO,

c 0 .+

mw=68

-

2

rnw=70

e 100 .-v)

a,

80

Q

0

60

-

v)

c 40

E

2

20 n "

1

3

2

4

5

Pulse Number

1, 3 - Butadiene Oxidation with

C 0

mw=68

1 8 0 2 over

MnMoO,

mw=70

13. -i 100 4.-

cn .-_

a,

80

Q

3

0

60

-

v)

c 40

=

LL

20 n v

1

2

3

4

5

Pulse Number

Figure 3. Distribution of furan isotopes in transient oxidation of 1,3-butadiene. Reprinted from J. Catal., 124 (1990) 187. In order to further examine the role of lattice and gas-phase oxygen in the synergy effect observed over these two-phase catalysts, the transient activity of the catalysts for 1,3-butadiene conversion were compared in the absence and in the

371

1 1, 3 - Butadiene Oxidation with "0, over MOO,

n

c L

100

fn

6

80

a,

0"

c

60

t

0

rnw=44 rnw=46 rnw=48

0

fn -

0" 0

40 20

0 1

2

3

4

5

4

5

Pulse Number

1

2

3

Pulse Number

Figure 4. Distribution of CO, isotopes in transient oxidation of 1,3-butadiene. Reprinted from J. Catal., 124, (1990) 189. presence of gas-phase oxygen (Table 2). Catalyst samples were first degassed for all of the experiments. The most dramatic feature of these results was the drastic change that took place in the activity of the pure-phase catalysts when the feed was

372

depleted of oxygen. In the presence of oxygen, the manganese molybdate catalyst seemed much more active than molybdenum trioxide. In the absence of oxygen, however, MOO, became more active than MnMo04. These experiments provided further evidence of the dependency of MnMo0, on gas-phase oxygen while demonstrating the ability of MOO, to utilize its lattice oxygen. Table 2 Conversion levels of 1,3-butadiene Feed with Excess Oxygen MOO, MnMo0,

16% 31%

Oxygen-Free Feed 43% 7%

Our more recent studies focussed on in-situ laser Raman characterization technique coupled with isotopic labeling and provided further evidence of the mechanism of synergy, while providing hints about the catalytic sites responsible for complete and selective oxidation. In these studies, catalysts were characterized using a specially designed in-situ cell described previously [26]. Catalyst samples were reduced with 1,3-butadiene and re-oxidized with labeled oxygen (1802) allowing the observation of Raman band shifts due to the isotope effect. To facilitate comparison , all parameters were kept constant for reduction and reoxidation processes for all catalysts. The spectra obtained from the reduced catalysts showed the intensity of the MOO, spectra decreased more than the intensity of MnMoO, under the same reducing conditions. The spectrum obtained over the two-phase catalyst showed that the intensity loss of the 815 cm-l and 991 cm-l bands that are associated with MOO, was much larger than the intensity loss of the 926 cm-1 band which is associated with MnMo04. The spectra obtained from MnMo04 in fresh form and after were very similar, showing no detectable shifts in reduction/reoxidation with 1802 the band positions due to isotope effect. Figure 5 shows the comparison of spectra obtained from the fresh MOO, to that obtained after reduction and reoxidation with 1802, Very distinct shifts are seen in the 815 and 991 cm-l bands. Spectra obtained from the two phase catalyst in fresh form and after reduction/reoxidation with 1802 again showed distinct shifts in the 815 cm-1 and the 991 cm-1 bands, while no shifts were observed in the bands associated with MnMoO,.

0 0 N

-

0

0 0

0

0 43

0 0 (0

E 0 W

373

g 0

c

5

374

DISCUSSION The studies outlined above clearly showed a major difference in the way oxygen was used by the two constituents of the MnMo04/Mo03 catalyst. These results suggest a possible catalytic job distribution between the two phases of the active catalyst where MOO, is responsible for introducing oxygen to the hydrocarbon from its lattice, forming the selective oxidation product. Molybdenum trioxide, however, is not very efficient in utilizing gas phase oxygen. The role of the second component (simple molybdate phase) involves t h e chemisorption/activation of the gas phase oxygen and facilitating its transfer to the MOO, phase through an oxygen spillover mechanism. The more recent in-situ Raman studies have proved useful in gaining more insight about the synergism found in this system. Comparison of fresh and reduced catalysts indicated that Moo3 reduced more readily than MnMoO,, both as a pure phase, and as a component in the two-phase catalyst, further supporting the chemisorption results. A comparison of the fresh and reduced two-phase catalyst provided another indication that the lattice oxygen was derived from the MOO, component, as the bands associated with MOO, exhibited a greater intensity loss upon reduction than did those associated with MnMo04. Reduction/reoxidation experiments also proved invaluable in providing evidence for the identification of active sites for partial oxidation. As noted above, the spectrum of the MnMo04 catalyst showed no significant change after reduction and reoxidation with raO,. The MOO, catalyst, however, did exhibit marked shifts after reduction and reoxidation with l8O2. An interesting feature about these shifts is the fact that the 815 cm-1 band is completely replaced by the shifted band at 792 cm'l whereas the band at 991 cm-l does not disappear completely. Instead, it is seen to reduce in intensity while a shifted band appears at 938 crn-l. This observation is significant since the band at 815 cm'' is associated with the symmetric stretching vibration of the bridging oxygen bonds (Mo-O-Mo) and the band at 991 cm-l is attributed to the Mo=O stretching vibrations [28]. Since the major reaction product over Moo3 is CO,, this suggests that the oxygen insertion that results in complete oxidation takes place on the Mo-O-Mo sites, although a secondary mechanism where adsorbed oxygen is used in complete oxidation may also be in operation. Our earlier studies on the structural specificity of Moo3 [25]provide further clues about the catalytic sites present on molybdenum oxide crystals. In the abovementioned study, MOO, crystallites were grown with preferred orientation such that samples with varying basal (010)-to-side (100) area ratios were obtained. While catalytic activity measurements over these crystals showed that samples with a higher basal-to-side plane ratio were more inclined to completely oxidize the hydrocarbon (1-butene, 1,3-butadiene, furan), the laser Raman spectra of these samples exhibited a pronounced difference in the relative intensities of the 815 and 991 cm-l bands. The intensity of the 815 cm-l band relative to the 991 cm-I band

375

was much higher in the samples that gave a higher yield of complete oxidation products than it was in the MOO, samples which showed a lower selectivity towards CO and CO,. These results provide further evidence that complete oxidation is more likely to occur on the Mo-O-Mo sites which are associated with the 815 cm-1 band, while selective oxidation is more likely to occur on the Mo=O sites associated with the 991 cm-1 band. CONCLUSIONS

In this study, single phase catalysts (MnMo04, MOO,) and the two-phase catalyst (MnMo04/Mo03) have been characterized by the in-situ laser Raman spectroscopy technique using a specially designed controlled-atmosphere cell. The samples were reduced with 1,3-butadiene and then reoxidized with labeled thus allowing the observation of Raman band shifts as well as band oxygen (1802), intensity changes which can be identified with the catalytic sites responsible for specific steps in the oxidation scheme. These observations have been combined with earlier results obtained from studies that focussed on the structural specificity of molybdenum trioxide [25]. These combined observations provide further evidence and a better understanding of the catalytic job distribution and the mechanism of synergy in these model multi-phase catalysts for selective oxidation reactions. One possible explanation of the observed findings is that complete oxidation over the MOO, phase is associated with Mo-O-Mo sites. Selective oxidation sites, on the other hand, are more likely to be associated with terminal oxygen sites (Mo=O). The role of the molybdate phase is adsorbing/activating the gas-phase oxygen and facilitating its migration to the reduced MOO, sites through a spillover mechanism. ACKNOWLEDGMENTS

The financial support from ACS Petroleum Research Fund, from AMAX Foundation, and from National Science Foundation in the form of an Equipment Grant (CBT-8705-124) is gratefully acknowledged. REFERENCES

1

K. Bruckman, J. Haber, and T. Wiltowski, J. Catal., 106 (1987) 188.

2

J.F. Brazdit, L.C. Glaeser, and R.K.Grasselli, J. Phys. Chem., 87 (1983) 5485.

3

G.I. Straguzzi, K.B. Bischoff, T.A. Koch, and G.C.A. Schuit, J. Catal., 104 (1987) 47.

4

M. Ai., Bull. Chem. SOC.Jap., 44 (1971) 761.

5

P. Sunderland, Ind. Eng. Chem. Prod. Res. Dev., 15 (1976) 90.

376

6

F. Cavani, G. Centi, I. Manenti, A. Riva, and F. Trifiro, Ind. Eng. Chem. Prod. Res. Dev., 22 (1983) 565.

7

G. Centi, G. Fornasari, and F. Trifiro, Ind. Eng. Chem. Prod. Res. Dev., 24 (1985) 32.

8

T.P. Moser, and G.L. Schrader, J. Catal., 92 (1985) 216.

9

E. Bordes, Preprints, Symposium on Structure-Activity Relationships in Heterogeneous Catalysis, American Chemical Society, Division of Petroleum Chemistry, Inc., Boston, MA April 22-27, 1990, 35(1), 22.

10 J. Ziolkowski, E. Bordes, and P. Courtine, J. Catal., 122 (1990) 126. 11 F. Trifiro, G. Caputo, and P.L. Villa, J. Less -Common Met., 36 (1974) 305. 12 F. Trifiro, G. Caputo, and P. Forzatti, Ind. Eng. Chem. Prod. Res. Dev., 14(1) (1975)22. 13 U.S. Ozkan, and G.L. Schrader, J. Catal., 95 (1985) 120. 14 U S . Ozkan, and G.L. Schrader, Appl. Catal., 23 (1986) 327. 15 U.S. Ozkan, R.C. Gill, and M.R. Smith, J. Catal., 116 (1989) 171. 16 U.S. Ozkan, R.C. Gill, and M.R. Smith, Appl. Catal., 62 (1990) 105. 17 U.S. Ozkan, E. Moctezuma, and S.A. Driscoll, Appl. Catal., 58 (1990) 305. 18 U S . Ozkan, M.R., Smith, and S.A. Driscoll, J. Catal., 123 (1990) 173. 19 U.S. Ozkan, S.A. Driscoll, L. Zhang, and K. Ault, J. Catal., 124 (1990) 183. 20

P. Ruiz, B. Zhou, M. Remy, T. Machef, F. Aoun, B. Domain, and B. Delmon, Catal. Today, 1 (1987) 181.

21

L.T. Weng, P. Ruiz, B. Delmon, and D. Duprez, J. Mol. Catal., 52 (1989) 349.

22

L.T. Weng, S.Y. Ma, P. Ruiz, and B. Delmon, J. Molec. Catal., 61 (1990) 99.

23

L. Glaeser, J. Brazdil, M. Hazle, M. Mehicic, and R. Grasselli, J. Chem. SOC., Faraday Trans., 1(79) (1985) 2903.

24

G.L. Schrader, T.P. Moser, and M.E. Lashier, Proc., Int. Cong. Catal, 9th, 4 (1988) 1624.

25

R.A. Hernandez, and US. Ozkan, Ind. Eng. Chem. Res., 29(7) (1990) 1454.

377

26 U.S. Ozkan, M.R. Smith, and S.A. Driscoll, J. Catal., accepted for publication.

27 R.C. Gill, and U S . Ozkan, J. Catal., 122 (1990)452. 28 I.R. Beattie, and T.R. Gilson, J.Chem.Soc (A) (1969)2322.


P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science arid Catalysis, Vol. 72, pp. 379-386 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

379

Role of oxide catalysts basicity in selective oxidation E. A.Mamedov, V.P.Vislovskii, R.M.Talyshinskii and R.G.Rizayev Catalysis Department, Institute of Inorganic and Physical Chemistry, 29 Narimanov Avenue, 370143 Baku, USSR

Abstract Rates of oxidative coupling and ammoxidation of some hydrocarbons are found to increase linearly with increasing the amounts of weak and moderate basic sites on a surface of oxide catalyst. Such ii correlation indicates that hydrocarbon activation occurs via the C-H bond heterolytic dissociation under the attack of catalyst basic site resulting in formation of surface carbmion. The rate of this step controls the rate of overall catalytic processes studied. T o accelerate it, two ways for catalyst modifying with basic additive have been used: (i) insertion of promotor into the catalyst body, and (ii) gas-phase modifying when the additive has been brought in the reaction mixture.

I. INTRODUCTION

Many oxide catnlysts contain at the surface a nucleophilic oxygen ions that are able to react iis ii charged species without breaking the oxygen-catalyst bonds, i.e. they can take part in heterolytic processes. Some hydrocarbons, such iis olefins, alkylaromatics and other ones, possess the C-H bonds polarized under the effect of superconjugation with unsaturated bonds or under the influence of electronegative substituents. These factors may facilitate the hydrogen abstraction in a protonic form upon the hydrocarbon interaction to catalyst nucleophilic oxygen according to the mechanism proposed in [l]:

H I

R-C-H I

H

It is assumed that the hydrocarbon activation leads to the formation of carbanion stabilized near the metal cation. Then by means of electron transfer from carbanion to catalyst, it turns into radical that can be dimerized or be transformed into a carbocation by way of one more electron transfer, followed by the interaction to catalyst electrophilic oxygen with the formation of the oxygenated product. In the presence of ammonia, radical

380

can also react to its adsorbed species, producing nitrile. So, the represented mechanism of hydrocarbon activation may be the primary step of a number of mild oxidation reactions, and their direction will depend on catalyst nature and reaction conditions. Stage 1 occurs without changing the oxidation state of surface metal, and therefore can be regarded as a n acid-base one. The heat of this stage includes as a constituent the energy of proton interaction to basic site 02- [2]. If such a mechanism of hydrocarbon activation takes place, the rate of selective oxidation should depend on the catalyst basicity, passing through the maximum upon its wide variation. In the region of low basicity, one may expect the reaction rate to raise up with increasing the catalyst oxygen nucleophilicity. In this case, it will be limited by the rate of hydrocarbon activation while the formation of reaction products proceeds rapidly. But the oxygen nucleophilicity should not be too high because it may render difficult the step of surface dehydroxylation. Besides that, if salt-like acidic intermediates iire arised, the strengthening of their binding with catalyst surface may take place [2]. As ii result of these factors, the increase of catalyst basicity over the certain value will decrease the rate of hydrocarbon selective oxidation. In the latter case, the rate of catalytic process will be controlled by the rate of one of the hydrocarbon surface transformations or by the rate of product desorption. T h e considerations stated a r e applicable for reactions of oxidative coupling, ammoxidation and oxidative dehydrogenation of hydrocarbons. In the case of olefins and alkylaromatics, they occur by the way of cleavage of the C-H bonds polarized under the conjugation with unsaturated bonds. The rate of this stage seems to determine the rate of catalytic reaction [3-51. Many effective catalysts for these processes show basic properties. As following from this, we measured the rates of indicated reactions and compared them to the concentriition as well as to the strength of basic sites for a series of oxide catalysts. 2. EXPERIMENTAL 2.1. Preparation of catalysts.

Supported V-Sb-Bi and V-Sb-Ni oxide catalysts were prepared by impregnation of y-Alz03 with solutions of ammonium metavanadate, antimony chloride, and bismuth o r nickel nitrate, followed by evaporation and calcination in a n air stream at 450 or 600°C. The preparation procedure is described in detail elsewhere [6,7]. Alkali and alkaline earth elements were added to the catalyst as hydroxides during the impregnation stage. Binary oxide catalysts, containing tin (Sn-Pb, Sn-In, Sn-Ti, Sn-Mo) and bismuth (Bi-Zn, Bi-Gi1, Bi-Pb, Bi-Sn), were prepared by thermal decomposition at 600°C of hydroxides coprecipitated from an aqueous solution of metal nitrates o r chlorides by ammonia. The Bi-Sn catalyst was also prepared by precipitation of bismuth hydroxide on the dispersed stannic oxide obtained by solving the tin in nitric acid. T h e atomic ratio of metals in binary oxides was equal to 1.

2.2. Characterization of catalysts. Surface areas of catalysts were measured by thermal desorption of argon. The integrated basicity of the catalysts was characterized by the amount of benzoic acid adsorbed on the surface from solution (water-free benzene). T h e quantity of benzoic acid,

381

remained in solution after adsorption, was determined by titration with the potassium hydroxide solution. Contents of the basic sites of various strength were found from experiments on adsorption and stepwise thermal desorption of carbon dioxide. The number of C02 molecules, adsorbed on 1 m2 of the catalyst surface, was used as a concentration of basic sites. The strength of sites was characterized by the temperature at which catalyst was able to keep the adsorbed gas. The higher temperature was the stronger sites were. The nucleophilicity of surface oxygen was characterized by the extent of charge localization on it estimated from positions of photo and Auger lines of oxygen in X-ray photoelectron spectrum. Spectra were recorded on a VG ESCA-3 electron spectrometer using AIK,, radiation. The energy of oxygen bond with the catalyst surface was estimated by means of high-temperature microcalorimetry of the heat of reaction of CO with the oxidized surface as well as of the heat of oxygen sorption on the reduced surface. Values of the oxygen bond energy, obtained by these two methods, coincided for the samples studied.

2.3. Catalytic activity. Oxidative coupling, ammoxidation and oxidative dehydrogenation of hydrocarbons were carried out in a flow apparatus equipped with a gradientless reactor with a vibro-fluidized bed of catalyst. The rates of product formation, determined after the catalyst activity stationary state had been reached, were compared for similar conversions (15-20”/,).

3. RESULTS A N D DISCUSSION In Fig. 1 rates of toluene and m-xylene ammoxidation over V-Sb-Bi oxide catalysts, modified with alkali and alkaline earth elements, are plotted against the surface concentration of basic sites determined by titration of benzoic acid. One can see that there is a linear correlation between these characteristics of catalyst.The same data on catalytic activity are compared in Fig. 2 to the concentrations of basic sites of various strength determined by C02 adsorption-desorption. Linear correlation to the reaction rate is observed for weak (CO?:desorption temperatures 50-200°C) and moderate ( ( 2 0 2 desorption temperatures 200-4OO’C) basic sites, and not for strong ones (CO2 is desorbed at temperatures higher than 400°C). Similar result has been obtained for oxidative coupling of toluene over bismuth containing oxide catalysts. For a series of SnO2 - Me,O, oxides, it has been also established that the increase of catalysts basicity enhances their activity in oxidative couplings of propylene and isobutylene. All these results testify that the selective oxidation of hydrocarbons proceeds over the catalysts studied with the participance of basic sites of moderate strength. As for the strong sites, there are data [8,9] indicating their responsibility for the hydrocarbon total oxidation. But in a steady state of oxidation reactions mainly carried out at 200-4OO0C,most of them are expected to be bound with carbon dioxide that is being produced, as usual, in a large amount during the initial non-stationary period of catalytic reaction. This phenomenon along with the reduction of a catalyst surface may be responsible for its high selective operation under the steadystate conditions.

382

2

I

3

Basicity o( eqv

4

C6HsCOOH/m2)

Figure 1 . Rates of ( a ) toluene and (b) m-xylene ammoxidation at 360°C as functions of basicity of V-Sb-Bi catalysts ( 1 ) without additive and containing (2) LizO, (3)NazO, (4)K 2 0 , W M g O , (6)CaO, and 17) BaO.

* *

2 -

*

/* :

10

IS

20

25

30

2

1 -

3

4

5

6

0.2

0.3 0.4

Concentration of basic sites (N*lO-LS/mZ)

Figure 2. Dependence of the rates of m-xylene arnmoxidation on the concentration of (a) weak, (b) moderate and (c) strong basic sites. Numbers of catalysts see Fig. 1.

The functions of basic sites can be performed by the surface species having the affinity to it proton. I t ciin be done by the nucleophilic ions of oxygen. Such a conclusion follows

383

from the data of Table 1 where the rate of propylene allylic oxidation is compared to the difference between energies of oxygen photo and Auger electrons used as a measure of the oxygen nucleophilicity. Both these characteristics increase if one moves from tin dioxide down to Sn-Mo oxide system. Moreover, catalysts for propylene oxidative coupling to diallyl and for propylene partial oxidation to acrolein compile the same linear correlation between the activity (In r) and the nucleophilicity (&,-Ed. This fact supports the idea that mechanism of propylene activation is the same for both reactions.

Table 1 Energies of oxygen photo and Auger electrons for oxide tin containing catalysts and the rates of propylene allylic oxidation at 500°C

SnO2 Sn-Pb-0 Sn-Bi-0 ( I , from hydroxides) Sn-Bi-O(ll, from me ta I) Sn-Mo-0

530.7 530.5

511.2 511.3

19.5 19.2

traces 3.20

530.6

512.0

18.6

4.27

530.3 530.2

512.5 512.8

17.8 17.4

3.4

11.7 19.7

The role of basic sites can be also played by the species adsorbed on the catalyst surface. Such a property is inherent in ammonia. For instance, ammonia brought in the tolueneoxygen mixture increases the rates of both partial oxidation and overall conversion of hydrocarbon as shown in Table 2. This effect is stipulated by the formation of supplementary basic sites on the catalyst surface. The functions of these sites are assumed to be performed by the partly dehydrogenated species of ammonia like NH2- and NH2-. These species, coordinated to the surface metal cations [ l o ] , possess the affinity to a proton, and therefore can assist in a heterolytic cleavage of the hydrocarbon C-H bonds. Relation between the rate of hydrocarbon selective oxidation and the catalyst basicity enables the catalytic activity to be regulated. From this point of view, modifying the catalyst with basic additives can be effective. Here two ways are possible: (i) insertion of promotor into the bulk or onto the surface of catalyst during its preparation, and (ii) gas phase modifying when the additive is brought in the reation mixture. Validation of the first way has been shown by modifying the V-Sb-Bi oxide system with smikll amounts of alkali and alkeline earth metal oxides. As it is seen from Fig. 1, addition of 0.2 wt');, of BaO to the catalyst enhances by two to three times its activity for toluene and m-xylene amrnoxidation. Such a promoting effect cannot be ascribed to an increase in the mobility of catalyst oxygen because there is no notable difference in the energies of the surface oxygen bond for the unmodified and modified samples (Fig.3). According to the

384

Table 2 Effect of ammonia concentration in the reaction mixture on basicity and activity of V-Sb-Bi oxide catalyst in toluene ammoxidation at 400°C

NH3 conc. (T,)

r* lo-'' (molec. C,Hs/m2 s) total

0.0 2.0 5.7 10.0 15.0

15.7 19.0 21.7 23.1 23.6

Basicity (u eqv C6HsCOOH/m2)

benzonitrile

1.60* 7.50 19.7 21.6 22.4

1.89 2.21 2.74 3.26 3.89

rn

* - rate of

benzaldehyde production

XPS data, upon the catalyst modifying with KzO or BaO neither the oxidation states nor the relative contents of metal cations on the support surface do not change too. At the same time, catalyst basicity significantly increases (Fig. 1). These results allowed us to interpret the promoting effect under consideration as an increase in the amount of surface basic sites that participate in the catalysis. The same effect of alkaline additive has been established for Pb-Sn oxide catalyst. From the data presented in Fig.4, one can see that addition of 0.5 wt'x of K z 0 to the catalyst enhances by two times its activity for toluene oxidative coupling. Larger amounts of additive reduce the rate of this reaction. Similar relation between the activity and the content of alkaline additive is observed for the V-Sb-Bi ammoxidation catalyst. It appears that small amounts of additive, being uniformly dissolved in a catalyst, change mainly its acid-base properties according to the principle of electronegativity equalization [ 1 I ] . In this case, the additive efficacy will depend on its basicity as well as on the metal ionic radius. When the quantity of additive exceeds the optimal one, its homogeneous distribution seems to be broken, and new chemicnl compounds or phases are possible to be arisen. In such case, not only the acid-base properties but also catalyst structure, energy of the oxygen-catalyst bond and other characteristics may change. All these together may lead to the decrease in catalytic activity . Compounds able to form a n adsorbed basic species on the catalyst surface can be used as ii gas phase promotors. As shown in Table 2, ammonia brought in the hydrocarbonoxygen mixture increases both the basicity and the activity of V-Sb-Bi oxide catalyst. Another example for the ammonia influence on a catalyst properties is represented in Table 3. I t is seen that after the ammonia has been brought in the reaction mixture, the extent of n-butane conversion slightly decreases whereas the selectivity with respect to the dehydrogenation products essentially increases.

385

C

.e 2

21

24

8 21

b*

0

-g I8

15

12

97$+;*:* 6 0

20

40

60

80

100

,

0 0

1

2

3

Extent of s u r f a c e reduction (%)

Figure 3. Energies of oxygen bond for V-Sb-Bi catalysts ( 1) unmodified and modified with ( 2 ) K 2 0 and (3) BaO vs the extent of catalyst surface reduction.

4

5

,;

,

6

7

1

K 2 0 c o n t e n t (%)

Figure 4. Effect of KzO content in Pb-Sn oxide catalyst on oxidative conversion of toluene to ( 1 ) stilbene, (2) benzene and (3) C02 at 535°C.

Table 3 Effect of ammonia on activity m d selectivity of the supported V-Sb-Ni oxide catalyst in oxidative dehydrogenation of n-butane at 630°C.

1:0.8:20:0 I :0.8:19.5:0.5 1:0.8:19:1 1:0.8:18:2 1 :0.8:17:3 1:0.8:15:5

30.0 28.5 27.3 26.5 26.0 25.6

25.3 26.1 29.0 30.5 3 1.4 32.7

40.2 42.0 48.7 53.5 56.4 57.9

27 .O 24.8 15.6 9.7 6.0 5.2

The haloidorganic compounds are often used as a gas phase rnodificators for the selective oxidation catalysts [ 121. Their activating effect can be also interpreted as the appearance of supplementary nucleophilic species in the form of haloid ions resulted from the dissociative adsorption of additive on a catalyst surface.

386 4. CONCLUSIONS

From the data discussed in this paper, it is seen that basic sites may play an important role in selective oxidation of hydrocarbons. This circumstance allows to regulate a catalyst activity by way of adding the basic compound and varying its content in the catalytic system as well iis in the reaction mixture. This approach is expected to be fruitful if the hydrocarbon mild activation occurs under the influence of catalyst nucleophilic site, and involves a carbanion formation. However, other mechanisms of hydrocarbon activation are possible [ 131. For instance, it may proceed via the heterolytic dissociation of the C-H bond with the formation of a carbocation. Another way is homolytic C-H cleavage leading to the direct generation of corresponding radical. In these cases, other properties of catalysts, such a s acid, redox, et al., will play the role of first importance. According to this, different methods for regulation of catalysts activity should be used.

5. REFERENCES 1 2 3 4 5

10 11 12

13

V.D.Sokolovskii and N.N.Bulgakov, React.Kinet.Catal.Lett., 6 (1977) 65. G.I.Golndets, i n 0.V.Krylov (ed), Partial Oxidation of Organic Compounds (Russ.Pmbl.Kinet.C~Ital,,vol.19). Nauka, MOSCOW,1985, p.28. E.A.Marnedov, Kinet.Catal., 25 (1984) 868. R. K.Grasselli ,J. D.Burrington and .I.F. Brazdil, Advan.Catal., 30 (198 1) 133. T.G.Alkhaznv and A.E. Lisovskii, Oxidative Dehydrogenation of Hydrocarbons, Khimiyii, Moscow, 1980. R.G. Rizayev et al., BRD Patent No. 2 632 628 (1978). V.S.Aliyev, R.G.Rizayev et nl., US Patent No. 4 198 586 (1980). M.Ai, J.Catal., 54 (1978) 223. O.V.Krylov, in 0.V.Krylov and M.D.Shibanova (eds), Deep Catalytic Oxidation of Hydrocarbons (Russ.Probl.Kinet.Catal.,vol. 18). Nauka,Moscow, 1981, p.5. A.B.Azimov, A.A.Davydov, V.P.Vislovskii, E.A.Mamedov and R.G.Rizayev, Kinet.Catal., 32 (1991) 109. T.Seiyamn, M.Egsshira, T.Sakamoto and I.Aso, J.Catal., 24 (1972) 76. L.Ys.Margolis, Oxidation of Hydrocarbons over Heterogeneous Catalysts. Khimiya, Moscow, 1977, p.226. V.D.Sokolovskii, in 0.V.Krylov (ed.), Pnrtial Oxidation of Organic Compounds (Russ.Probl.Kinet.Catal., vol. 19). Nauka, Moscow, 1985, p.99.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shrdies in Surface Science and Catalysis, Vol. 72, pp. 387-398 0 1992 Elsevier Science Publishers B.V. All rights reserved.

387

Strong evidence of synergetic effects between cobalt, iron and bismuth molybdates in propene oxidation to acrolein 0. Legendre 1 , Ph. Jaeger 1 and J.P. Brunelle 2 Centre de Recherche d'Aubervilliers, Rhbne Poulenc Recherche, 52, rue de la Haie Coq, 93308 Aubervilliers Cedex - France 2

RhGne Poulenc, 25, quai paul Doumer, 92408 Courbevoie Cedex - France

Abstsact Different model catalysts with 1, 2 or 3 molybdate phases and with varying interactions between molybdates (no or weak interaction by mechanical mixture or strong interaction by co-evaporation) were prepared. These model catalysts were subjected to XRD to determine crystallographic phases, and to the picnometric method to determine surface area. Catalytic measurements were made with propene, air and water mixtures. The results point out large improvements in oxidation activity and acrolein yield. The higher the number of molybdate phases and the degree of interaction between them, the higher the strength of synergetic effects. Some tentative explanations are proposed. 1. INTRODUCTION

Acrolein synthesis by propene mild oxidation has become more economically attractive since the discovery by Standard Oil of Ohio (Sohio) of the Bismuth Phosphomolybdate catalyst system in 1957 [ll. During the last thirty years, improvements in this catalytic system have been claimed in a very large patent literature. Most of the patented catalysts are based on Molybdena and Bismuth, but also contain several other elements. The most widely used elements are Iron, Cobalt and Nickel, but Phosphorus, Tungsten, K, Mg, Cr, T1, Sn and B are also very often present. A survey of these modern multicomponent catalysts is available [2]. These catalysts are very poorly defined, except on patents calling on peculiar crystallographic phases 131. On the other hand, "academic" studies generally deal with simpler and more closely defined catalysts. As bismuth molybdate is the oldest and best studied catalyst for the oxidation of propene to acrolein, a lot of structural, catalytic and mechanistic work has been reported on a Bi2Mo3012 141 or y BizMoOs [5]over the last few decades. But few studies have been published on the defined influence of a third or fourth element upon the catalytic activity of bismuth molybdate. For example, addition of Vanadium in the scheelite structure of a Bi2Mo3012 leads to an increase in catalytic activity and in

388

selectivity to acrolein, as reported by MORO-OKA et al. [SI. In this case, the increase in activity could be attributed to the mobility of lattice oxide ions in the Bil-~/3Vi-~Mo~O4 structure. Somes synergetic effects were also reported for a mixture of Moo3 with y BizMoO6 17-91 and for a mixture of two crystallographic phases, a Bi2Mo3012 and yBi2MoO6 110-111.Very recently, some evidence of the improvement in catalytic activity and/or selectivity to acrolein on a supported bismuth molybdate has been reported 1121; the positive influence of Iron in the cobalt molybdate used as support is pointed out. Unfortunately, major parts of these catalytic experiments were conducted in conditions of low propene conversion and without water vapor in the reactants, and thus quite far from actual industrial conditions. Moreover, interpretation of these synergetic effects is not unique. There are two basic interpretation of this kind of synergy : 1) "remote control", which means spillover of oxygen from one phase to the other : it was proposed in the case Of MOO3 with y BizMoO6 191 and in the case of a Bi2Mo3012 and y BisMoO6 1111, and 2) the "coherent interface" which involves enhancement of the oxygen anion and electronic conduction between the two phases a t their interface : it was first proposed for the system V205/Ti02 in ortho-xylene oxidation [131. A third interpretation is the elimination of excess Bi (or Mo) at the surface of y (or a)phase as impurities [lo]. It is thus the aim of this paper to report studies of the observed effects of varying the number and degree of interaction of molybdate phases in the oxidation of propene to acrolein with reactants containing water wapor and at a high level of propene conversion. Another goal is to try to interpret these effects. 2. CATALYST PREPARATIONAND CHARACTERISATION 2.1. co-evaporatedmixhrresof molybdate phases :

The strongest interactions are postulated to be those of the molybdate phases crystallising simultaneously from the same precursor containing all the molybenum and metal ions homogeneously mixed a t the highest temperature. They are obtained by using the method described in example No1 of our first patent [31. The first part of this method is the one most widely used for this kind of solid, namely evaporating an aqueous slurry containing all the elements, drying and calcining. In our experiments, the total amounts of each calcined solid correspond within 4% t o the theoretical amounts expected if the incorporation yields were loo%, thus the chemical composition of these products, which we called "Co-evaporated Mixtures" in this paper, are perfectly known, which is what we are looking for. This will not happen if we have used the more industrial methods of coprecipitation without evaporation described in others of our patents [14]. The calcination temperature is 480 "C in order to be sure that the interactions between molybdate phases in such solids are the same as in the final catalyst calcined twice at 480 "C, as stated in our patents. Samples of all these solids calcined at 480 "C were used for structural determination of molybdate crystalline phases by XRD on PHILIPS 50 kV/40 mA diffractometer (CuKa radiation). As expected, all these co-evaporated

389

mixtures contain defined molybdate phases. They are thus referenced by the mnemonic code "xlMeM l-SI-x2MeM2" where xi are the calculated molar amount of each metal molybdate phase MeMi strongly interacting together in the solid. Some examples of these structural determinations are given in table 1below. Table 1 : Examples of structural determination of Co-evaporated mixtures References of some Co-evaporated Mixtures

Chemical compositions

More Crystalline visible phases Intereticular distance(s) (A)

Relative Intensity

Two Molybdate phases Strongly Interacting : SCM-SI-0.5FM

~ 0 9 ~ ~ 1 ~ 0 1 0 . 63.37 ~42

3.14 2.96

p coMoo4 01 CoMoO4 Fe2Mo3012

strong weak weak

Three Molybdate phases Strongly Interacting : 9CM-SI-0.5BM-SI-O.5FMCogBilFelMol20a 3.37 3.14 2.88 2.96

p~

strong weak 01 CoMoO4 a B i 2 M o 3 0 ~ weak Fe2Mo3012 weak 0 ~ 0 0 4

We have prepared co-evaporated mixtures containing : - each of the three single molybdate : CoMo04 (CM), FeaMoaO12 (FM) and Bi2Mo3012 (BM) - some mixtures of two molybdate phases containing cobalt, the major component of the industrial multimolybdate catalysts : SCM-SI0.5FM, SCM-SI-0.5BM, 4.5CM-SI-0.5FM and 4.5CM-SI-0.5BM - a mixture of the three molybdate phase in relative amounts close to those of industrial multimolybdate catalysts : 9CM-SI-0.5BM-SI0.5FM 2 5 Mechanical-

ofmolybdate phases : No or weak interactions between specific molybdate phases are obtained by mechanically mixing together these previously co-evaporated mixture t o obtain the active catalytic phase. This active catalytic phase is then coated onto the inert carrier to obtain the final catalyst in conventional industrial form. In order to ensure no or only weak interactions between the initial co-evaporated mixtures containing: more than one molvbdate phase, the final catalvsts are

390

calcined at a lower temperature : 400 "C instead of 480 "C for co-evaporated mixture. The catalysts prepared with these mechanical mixtures are thus XI (CoEM)1 referenced by the mnemonic code x2 (CoEM)2 where Xi are the molar amount

of each co-evaporated mixture (C0EM)i used. They contain the three molybdate phases in the same ratio as in the co-evaporated mixture of the three molybdates. Their overall chemical compositions are identical : CogBilFelM012048.Their descriptions are listed in table 2 below. Table 2 : Catalysts containing mechanical mixtures of Co-evaporated Mixtures (same chemical composition : CogBilFelMo12048) References of the catalysts

9 (CM)

0.5 (FM) 0.5 (BM)

Calcination References temperature of the Coof each Coevaporated evaporated Mixture used Mixture 480°C

1(9CM-SI-0.5FM) 0.5 (BM)

480 "C

1(9CM-SI-O.5BM)

480 "C

0.5 (FM) 1(4.5CM-SI-0.5FM) 1(4.5CM-SI-OABM)

480 "C

CM FM BM 9CM-SI-0.5FM BM 9CM-SI-O.5BM FM 4.5CM-SI-0.5FM 4.5CM-SI-0.5BM

Molar amounts of each Coevaporated Mixture used 9 0.5 0.5 1 0.5 1 0.5 1 1

Final calcination temperature of the catalyst 400 "C 400 "C 400 "C 400 "C

23. Model catalysts fmally obtained : The models catalysts are purposely prepared with an identical weight % of active phase for each of them, near 19 %. All the experimental values of % weight of active phase are very close to 19.0, with a standard deviation of 0.3.As expected, this parameter cannot effect the final catalysts. These active phases are either the Co-evaporated Mixtures alone or the mechanical mixtures of these Co-evaporated Mixtures. A more complete characterisation of these catalysts include textural measurements. The picnometric method by mercury penetration up to 1500 bar makes it possible to precisely measure the porous distribution of each pore size of the final model catalysts. Surface areas are then calculated from these values of porous volumes versus pore diameters, assuming the pores are cylindrical. Results are listed in table 3 below, where catalysts are referenced with the same code as explained previously for mechanical mixtures :

391

Table 3 : Textural determination of the final model catalysts References of the model catalysts

Overall chemical compositions of the active phase

(CM) (FM) (BM)

COMOO4 Fe2Mo3012 Bia03012

Total Specific Porous area volume (m2/g) (ml/g) 0.057 0.054 0.055

Specific area expected in simple mixture

1.4 0.4 0.1

Calculated PcFe > PcRu > PcNi E PcOs. From Table 3 it is also seen that molecular hydrogen is more efficient reductive agent towards NO than carbon monoxide. This result is in good agreement with the data published earlier [10,11]. Besides, the reactivities of carbon monoxide towards NzO and NO differ substantially, so that the extent of reduction of NO is greater than that of N2O. Dependence of nitrogen-containing product yields on total NO conversion is shown in Fig.6. It is to note that the experimental points obtained at both different temperatures and various contact times fit the same curves. The yield of NzO

449

passes through a maximum which is characteristic intermediate rather than of a final product.

of

Table 3 Degree of NOx conversion (or,mol.%) and composition nitrogen-containing products (mol.%) at 275OC ~

an

of

~

NO

+

Hz

NO

+ CO

NzO

+ CO

Catalyst PCCOY PcFeY PcNiY PcRuY PCOSY

100 68

22

-

19 48 5

-

-

25

34 5

-

56 18 90

66

-

-

31

80 65

9

40

15 9

67 68

20 35 60 33 32

9

19 8 8 4

4.DISCUSSION 4.1.Distribution of metallocomplexes in the zeolite matrices The sizes of both metallocenes and mononuclear carbonyl molecules allow them to penetrate into zeolite cavities where formation of PcM's occurs. In contrast, the penetration of large molecules such as octacarbonyldicobalt and dodecacarbonyltriosmium is hardly possible to occur because of of space restrictions. Nevertheless, the formation corresponding Pc complexes has been proved to proceed inside the zeolite bulk when we start with preadsorbed Coz(C0)e or Os3 (CO)12.

The most probable explanation for this fact is that the large complexes being adsorbed on the outer surface of zeolite crystallites dissociate then into small mononuclear fragments which can penetrate through the cage openings. Such subcarbonyl species could be stabilized by coordination to zeolite core or to exchangeable cations [12]. This dissociative mechanism seems to be operative at least in the case of bulky OS3(c0)12. Dodecacarbonyltriosmium clusters were shown to dissociate into mononuclear Os(C0)x fragments when adsorbed on SiOz or A1203. Such a dissociation occurs on heating the samples in vacuo but, none the less, is accompanied by oxidation of subcarbonyl particles with structural OH groups [13]. This conclusion is strongly supported by the XPS results obtained in present work. As seen from Fig.4, an increase of the Eb value for 0s 4f7/2 upon adsorbing OSa(C0) 12 is indicative for the rising positive charge on 0s atoms. Hence, the dissociative adsorption of oS3(co)12 might lead to corresponding mononuclear species. Their redistribution between the surface and the bulk of zeolite crystal is favorized by high adsorption potentials mentioned above. These enable the matrix synthesis of PcM molecules to proceed inside zeolite cages, and thereby topologically held species could be obtained. The question now arises on whether enrichment of outer layers of zeolite crystals takes place during the PcM

450

synthesis. It can be answered by evaluating of metal-to-silicon ratios from a XP spectra and by determining of the total metal contents in catalysts using a chemical analysis (CA). The surface concentrations as measured by XPS technique appeared to be very close to that for the bulk, as seen from Table 3 . Therefore it is suggested that the matrix-synthesized PcM complexes do not show any preference to reside on outer surface of zeolite support. 4.2.Valency state of metal The spectroscopy (XP or IR) shows little difference between the properties of unsupported Pc complexes of Co, Nil Fe and of included ones. Then, it was natural to assume that the zeolite environment does not affect the state of central atom. Thus, the PcM molecules obtained as a result of matrix synthesis might be suggested to contain bivalent atoms of Co, Ni and Fe. It implies that neither covalent nor ionic interaction between guest molecule and host lattice is involved. So that the holding forces are topological in nature, both axial coordination sites of Pc complex being vacant. In contrast, the structure of chelate center of zeolite-included PcOs and PcRu is quite different from that of PcCo, PcNi and PcFe. The E b value of 0 s 4f7/2 for PcOsY was found to be c.a.53.5 eV which is characteristic of trivalent state of central atom [14]. Similar result was obtained in the case of PcRu. The binding energy of Ru 3d5/2 was of 2 8 4 . 6 eV for PcRuY. Such magnitude of E b clearly shows that the metal is oxidized upon binding into Pc complex [15]. It is to note that individual analogs of Pc with trivalent central atoms possess always an extra ligand in 5th axial position [7]. Apparently, the zeolite OH groups may be playing a role of such extra ligand when zeolite lattice hosts the PcRu or PcOs guest species. 4.3.NOx reduction over zeolite-included PcM's The data obtained indicate that the nature of metal, its valency and coordination state affects significantly catalytic properties of supported PcM complexes. Among PcM's with free both 5th and 6th coordination sites, Co and Fe complexes are more active than Ni one. Such a variation in activity can be rationalized in terms of the electronic structure of these. Indeeda the out-of-plane atomic orbitals of Co2+and Fe2+(dz -AO) are unfilled. That enables them to participate in the substrate activation. In contrast, this A0 for Ni2+ is fully occupied. The different ability of Co2+ and Fe2+atoms for back-donation onto the antibonding n-MO of NO appears to account for their different behavior in catalysis. On the other hand, the low catalytic activity displayed by the Ru and 0 s complexes is obvious to be connected with their coordination state. It should be pointed that activity sequence for pure PcM's as well as for metal tetraphenylporphyrins as reported by Mochida et al.[11,16], is at variance with our observation. Thus, Ni complexes were found [11,16] to be the most active catalysts in NO+H2 reaction. This difference can be related to

451

matrix effect. Indeed, a generation of adsorbed nitrogen atoms from NO is suggested by Mochida et al. , and it can occur when at least two adjacent catalytic sites are involved. These requirements are met by crystal porphines where interplanar distances are about 0.3 nm. However, it is not the case with PcM's included within zeolite matrix where they are separated by distance of 1.2 nm. Since the zeolite matrix provides molecular one-to-one distribution of PcM's per cage, the activation of NO may be caused only by single active site, and thereby the coordination state of metal atoms in chelate molecule seems to play a key role. This suggestion is supported by the fact that the activity of PcM's in NO reduction is substantially higher than that in N2O reduction which couples with very poor ability of the latter to coordinate [17]. On the other hand, both NzO and Nz are resulted from nitrogen monoxide transformation, and the curves of NzO yield vs. NO conversion (see Fig.6) are typical of sequential reactions. In other words, dual-site mechanism as proposed for dinitrogen species formation on oxides and metals is not operative when NO reduction occurs over PcM's dispersed molecularly within zeolite matrix. 5.CONCLUSION Thus, one can conclude that the zeolite matrices when hosting PcM molecules may represent true inclusion compounds where bulky species generated in situ are topologically included in the matrix cavities owing to space restrictions. From the behavior of some zeolite-included PcM's towards the catalytic reduction of NO, these inclusion compounds could be considered as prospective ox-red catalysts. 6.REFERENCES 1 V.Yu.Zakharov 2 3 4 5 6

and B.V.Romanovsky, Vestnik Mosk. Univ., Ser. Chim., 18 (1977) 142 [Eng. transl. in Sov. Mosc. Univ. Chem. Bull. , 32 (1977) 161. G.Meyer, D.Wohrle, M.Moh1 and G.Schulz-Ekloff, Zeolites, 4 (1984) 30. N.Herron, S.A.Tolman and G.D.Stucky, Abstr. XXIII Intern. Conf. on Coord. Chem., Boulder, CO (1984) 111. T.Kimura, A.Fukuoka and M.Ichikawa, Catal.Lett.,4 (1990) 279 A.N.Zakharov and B.V.Romanovsky, J.Inclusion Phenomena, 3 (1985) 389. J.I.Landau and E.E.Petersen, J.Chromatogr.Sci., 12 (1974) 362.

7 A.B.P.Lever, Adv.Inorg.Chem.Radiochem., 7 (1965) 27. 8 E.S.Shpiro, G.V.Antoshin, O.P.Tkachenko, S.V.Gudkov, B.V.Romanovsky and Kh.M.Minachev, Stud.Surf.Sci.Catal., 18 (1984) 31. 9 P.H.Citrin, J.Amer.Chem.Soc., 95 (1973) 6472. 10 F.Steinbach and H.-J.Joswig, J.Catal., 3 (1978) 272. 11 K.Tsuji, H.Fujitsu, K.Takeshita and I.Mochida, J.Mol.Catal., 9 (1980) 389.

452

12 R.L.Schneider, R.F.Howe and K.L.Watters, J.Inorg.Chem., 13

14 15 16 17

23 (1984) 4593. J.Zwart and R.Sne1, J.Mol.Catal., 30 (1985) 305. V.I.Nefedov, 'IRentgenoelectron spectroscopy of chemical compounds"I Chimia, MOSCOW, 1984 (Russ.) O.P.Tkachenko, G.V.Antoshin, E.S.Shpiro and Ch.M.Minachev, 1zv.Akad.Nauk USSR, 6 (1980) 1249 (Russ.) I.Mochida, K.Takeyoshi, H.Fujitsu and K.Takeshita, J.Mol. Catal., 3 (1978) 417. R.Eisenberg and D.E.Hendricsen, In:Adv.Catal.,28 (1979) 79.

.

.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Shidies in Surface Sciecice and Catalysis, Val. 72, pp. 453-460 0 1992 Elsevier Science Publishers B.V. All rights reserved.

453

NEW PREPARATION METHODS OF MULTICOMPONENT OXIDE VANADIUM SYSTEMS FOR OXIDATIVE DEHYDREENATION OF ALKANES, ATsKWAROMATIC AND -CXXCLIC CaMpOuNDS

I.P. BelcPllestnykha, E.A. Skriganb, N.N. Rozhdestvenskayaa and G.V. Isaguliantsa %.D. Zelinski Institute of Organic Chemistry, USSR Academy of Sciences, Moscow, USSR

bInstitute of Physical Organic Chemistry, gyelorussian AcadenTy of Sciences, Minsk, USSR Abstract The present paper reports the results of complex investigation devoted to phase canpsition and texture formation of the new oxide vanadium catalysts for oxidative dehydrogenation of alkanes, alkylarmtic and alkylheterocyclic ccpnpounds. In the production of olefines, vinylarmtic and vinylheterocyclic ccmpounds the sugg ted st#cture of an active centre of V-containing catalysts comprises and V ions in octahedral configuration grouped in clusters. The valent and coordination states of V, the size of clusters (2-13 ions of V) and associated activity, selectivity and stability were shown to depend on the methcds of preparation, the content of vanadim in catalyst samples, the nature of salts in the active component, Maifying agents and supprts. At low vanadia content (2-5%) there are only isolated tetrahedral and octahedral surface vanadium species in the samples. These species are responsible for mplete oxidation. The content 7-12% is optimum for oxidative dehydrogenation. There are associated vanadium species. The nature of the acceptors (C H NO SO2, COS, 02) used has been found to affect the reactivity of comp%~&s.~'

2'

INTRODUCTION The selective convertion of paraffins into olefins, alkylarmtic and alkylheterocyclic into vinyl ccanpounds belongs to the challenging problems of the petrochemical industry. In catalytic processes high selectivity to vinyl ccmpounds is found only at rrpdest temperatures where the dehydrogenation equilibrium is unfavourable: U p n oxidative dehydrogenation the conversion of initial compouns and selectivity are increased, thenrodynamic limitations are eliminated and considerable energy saving is provided /1,2/. h n g catalysts currently employed in the petrochemical industry, vanadim-containing oxide systems are playing a leading part /3,4/. Despite nmrous investigation of the canpsition and properties of vanadium catalysts, the nature of catalytically active phase could be established in a limited nunker of instances only. A great deal of attention is, therefore, given to studies on correlation between physico-chemical and catalytic properties of the systems in question / 5 , 6 / .

454

The present paper reports the investigation of V-containing different preparation methods and their influence on active surface formation in oxidative dehydrogenation of organic comp3unds. EXPERIMENTAL

Catalyst preparation. Supported vanadium oxide catalysts were prepared by four methods. 1.Dry mixture of a support and a salt of vanadium. 2. Wet impregnation supports with a solution of vanadium salt in distilled water, drying at 12OoC and calcinating in air for 3 hours at 55OoC. The m u n t of vanadium salt used corresponds to that necessary to have a final canpsition in the calcined samples of about 2.5, 5.0, 7.0, 9.0, 12.0, 16.0 and 25.0% of V 05.3. Mechanmhmical method of preparation consists of s u p port (Mgo, Al , Ti02) activation for fomtion defective structure in desintegrater2a2 8 W - 1 5 0 rotation/min with mechanical energy 3.l, 8.8, 17.7 kw h/t. 4. Electro-chemical methcd - simultaneous precipitation b@ (OH) from =l2 and mixture with a vanadium salt solution in electrolyzer (A = 0.01.10- m / A h), anode is graphite, cathode is steel. To prepare the samples, m n i u m , sodium, potassium vanadate and potassium decavanadate were used. Ccmertial Mgo, Al 03,2ZrG2,Ti0 , CaO have been used as supprts "4th 110, 120, 2 0 , 8 and $0 m /g surfacg areaoand electrochemical bkj0 (110m /g) All samples were dehydroxylized at 120 C, then calcined inoair stream under the conditions of gradual temperature elevation to 550 C. sample characterization. Specific surface areas of the samples were determined by the BET method using N The characterization of pore stdcture of supports and the catalyst samples prepared was performed by high-pressure mrcury porosimetry utilizing a comnertial porosirwter (Car10 Erba Strum?ntazione). The diffuse reflectance spectra were recorded on SP-26 LOMO using barium sulphate as a reference. The resolution of spectrwter under the conditions of m e a s m n t was 0.2-0.6 nm. Catalytic tests. The reaction was conducted in an integral flow type quartz reactor over a fixed catalyst bed (2-40 m l ) , the process pxawters being varied in a broad r ge: temperature, from 320 to 5oOoC; space velocity, frorn 0.25 to 1.5 h- Y, and dilution with steam, 1:3-15, molar ratio ccmpound:acceptor = 1:l-3. Liquid and gaseous samples were taken every 15 minutes, and experiment duration equalled 3 to 8 hours. Reaction products were analysed chrmtqraphically, while the m u n t of products deposited on the catalyst surface was determined by the derivatographic technique. Use was m d e of analysis and mterial balance data to calculate the degrees of convertion into vinyl compounds, aniline (X1 and X2), and selectivities S1 and S2, respectively.

6

.

.

RESULTS AND DISCUSSION The dependence of the degree of conversion of hydrocarbon to vinyl cornpound (X,) and of nitrobenzene to aniline (X ) on V 0 /Mgo was found to have a maXirrmm of activity corresponding to $he V O2 ntent of 9-12% (Table 1, Fig. 1) . As have been shown with use of2 NMR / 7 / , at low vanadia content (2-5%) there are only isolated tetrahedral and octahedral surface vanadium species in samples, and activity and selectivity are small-

% '

455

Table 1 Effect of V 0 content in oxidative dehydrogenation of hydrocarbons. as sup&& Hydr0carbn:C H NO : H 0 = 1:3:9-12, space velocity 0.25-0150 h-’ 42OoC Ethy1benzene:g ? H20 = 21:1:7-10, space velocity 0.5-1 .O h- , f8Ook Ethylbenzene: 2S022: H 0 = 1:0.4:10, space velocity 0.5-1.0 h- , 5Oo0C 2 /V205/ Diphenylethane %

mass

Acceptor:

Ethylbenzene

lSt and p83methcds 0.20 0.97 0.26 0.24 0.96 0.30 0.99 0.34 0.98 0.35 0.35 0.97 0.32 0.95 0.23 0.28 0.95 0.10 0.90 0.60 0.65 O.7Oa 0.75 0.98

0.28 0.32 0.35 0.37 0.38 0.36 0.30 0.15 0.70 0.72 0.72

0.07

0.03

0.02

0

Prepared by 2.5 5.0 0.18 7.0 9.0 12.0 0.27 16.O 25.0 0.15 103.0 12.~3-ZnO 12.DtK SO 12.0 20.45a multicomponent Prepared by 12.0 Prepared by 3.0

02

‘gHgN02

0.35 0.40 0.42 0.42 0.43 0.42 0.40 0.30

0.27 0.35 0.43 0.49 0.43 0.39 0.38 0.15 0.60 0.65 0.72 0.72

0.30 0.39 0.48 0.50 0.55 0.48 0.40 0.07

0.85 0.88 0.92 0.95 0.94 0.85 0.89

0.90 0.80

0.97

0.82 0.82 0.85 0.90 0.88

0.82 0.84 0.80

the 3rd method 0.40 0.98 0.42 0.56 the 4th method 0.45 0.98 0.48 0.45 0.50 0.92 0.56 0.97

a S = 0.82, S 1 2

=

0.68.

0.6 8

s 2al

2 8

0.4

0.2

20 40 60 80 103 V205, % mss. Figure 1. Effect of V 0 content in vanadia-magnesia catalysts on catalytic activity (styrene - O2 aniline - @ . Acceptor: 1 - SO2, 2 - C6H5N02, 3 - 0 2

?

456

er. The n-r of ions in octahedral coordination grows with the vanadia concentration increase (up to 9-12%). There are associated vanadium species /7/. Further increase in the content of vanadia (16%and upmrds) results in the formation of the magnesium madate phase having a regular structure, wherein the vanadium ions are in tetrahedral coordination and fail to undergo reduction /8/. An additional phase, viz., free V 05, appears as the content of vanadia continues to grow (up to 25% and hiher), and such samples are close to -idividualV 0 in terms of their catalytic activity and selectivity. The absence of2tl?e V205 phase in the catalysts with a lower content of va7)adia is presufiably caused by the formation of vanadia-mqnesia structures, in which vanadium ions are in tetrahedral and octahedral coordinations, the latter coordination being predarrcinant. Increasing the mntent of vanadia apparently results in the transition f r m the strongly defective to the regular structure which is typical of vanadates that display low reducibility and adsorptive capacity and, hence, low activity in the process discussed here. The electrochemical mthod permit to prepare the active and selective samples with associated vanadium species in octahedral mrdination at lower contents of active ccerrponent - 2-5% (Table 1). The conditions of heat treamnt exert a mrked effect on the properties of catalysts (Table 2 ) . The most active catalysts are those subjected to heat treatment in air stream under the conditions of gradual temperature elevation to 55OoC, this activation d e resulting in the formation of a catalyst texture with a porous structure and a large surface area that favour the process. Elevating the activatAon temperature of all samples prepared by different methods up to 750-850 C, leads to the complete dehydroxylation of the catalyst, increased pore size, diminution of the surface area, and the loss of catalytic activity. V20 was found to react with m g nesia and yield regular structures of the IvQ3?V0,) type. The role of ions in octahedral coordination for the process is illustrated by the data on the catalytic properties of catalysts, prepared fram different vanadium salts deposited onto supprts of various nature (Table 3 ) . Samples 1 and 2, prepared frm NH VO and H@, are the m s t active in the pgpess. Mifying agents (sample 2f d e conductive to stabilization of V ions. It is noteworthy in the W-spectra of the reduced samples 1yhand 1 2 the band 3 2 0 nm after the effect of reaction d i um and even "rigid" H 2 ) . The oxidative dehydrqenation in presence of O2 proceeds without regeneration and activity decline, in the presence of nitrobenzene - cycle 0.5 h and regeneration for 1 h. W spectra of samples 6,lO shows band edg broadening not observed in others and corresponding to absorption by According to the data reprted in literature / -11/, this finding pint!? t the formation of clusters comprising 10-12 ions, while in sample 2 ions , perhaps, are present in the f o m of -f!i!er clusters ( 2 or 3 vanadium ions). The resistance of the clusters formed f r m 10-12 of vanadium ions to phase fomtion an a high degree of dispersion are respnsible for a w k e d stability of t m r d s reduction to lower oxidation states (sample 13) and ensure a greater stability in time - 1.5-8 h without regeneration. The investigation reprted here made it possible to correlate the catalytic properties of vanadium containing oxide systems with their structural acter' tics. The results obtained provided grounds for assuring that p a n d V" ions in octahedral coordination are responsible for the activity of catalysts. These findings served as a basis for the developnt of catalysts having high activity and selectivity in oxidative dehydrogenation of organic compounds.

g'

4+

.

5;

2

457

Table 2 Effect of heat treatment of V205/Mg0 Hydrocarbon : C6H5IW2 : H20 = 1:1:12, Heat treatment technique

Surf ace

area,

space velocity 0.25 h-',

Pore

volmf m2/g

3 QTI 19

Process characteristics Ethylbenzene

x1 Drying a t 12OoC Gradual temperature elevation t o 55OoC, 100 holding 3-4 h Gradual tempraturg elevation t o 850 C 60 From Mgo preheated a t 850° + V205 and gradual temperature elevation 52 t o 55OoC I@O + V O5 mixture a t 850°, 40 holddg 8 h Mechanochemical method, gradual temprature elevation t o 55OoC 100 Electrochemical mthod , gradual t-rature elevation t o 55OoC 110

40-42OoC

x2

Isobutane

x1

x2

Methane

x1

x2

0.30

0.12

0.10

0.70

0.35

0.38

0.30

0.33

0.08

0.05

0.60

0.07

0.09

0.05

0.09

0.02

0.20

0.40

0.08

0.08

0.35

0.07

0.06

0.85

0.40

0.42

0.80

0.45

0.48

0.40

0.38

0.10

0.70

The oxidative dehydrcgenation of ccxnpunds having various structures over or COS, atmospheric 2 oxygen as acceptors form the following sequences depending on the acceptor used: with C H NO - ethylbenzene =-diphenylethanezisopropylbenzene =isobutane =-k?h&e; with SO2 or COS - isobutane ;.-diphenylethanewethylbenzene zisopropylbenzene; with 0 - ethylbenzene pethylpyridine P d i e t h y l 2 zisopropylbenzene-ethyltoluene benzene Pdiphenylethane 7ethylthiophene -methane (Table 4 ) . A comparison of relative reactivity of alkylpyridines and alkylthiophenes i n oxidative dehydrcgenation shows the following sequences: 2-ethyl-z- 4-ethyl- -2-methyl-5-ethyl=-2-methyl-5-buthyl2,5-diethyl- >isopropyl-s- n-propyl- wbutylpyridines; 2-ethyl- =isopropyl- wn-propyl- wbutylthiophenes. The oxidative dehydrogenation of a l k y l a r m t i c and alkylheterocyclic ccmpounds was found t o proceed a consecutive-parallel pathway. carbon dioxide i s the product of oxidation of starting compounds and also results fran their dehydrcgenation or p a r t i a l oxidation (Scheme). There are small oxygen containing compounds ( t o 1%) i n reaction products. V205/Ivkj€) catalyst i n the presence of C6H5N02, SO

=-

458

Table 3 Influence of starting salt and supprt nature Ethylbenzene: C H No : H 0 = 1:3:9-12, space velocity p.25-0.50 h-I, 42OoC 2 Ethylbenzene: 0; HgO= 1:1:7-10, space velocity 0.5 h- , 48OoC

?

Starting salt

w

Support

. . . . . .Acceptor .................

spectraa,

nm 1. NH4v03 2. NH vo3 mlticcmponent 3. 4. 5. 6.

‘sHsN02

280, 320, 600, 880 weak 280, 320 sign.,600 sign., 880 weak 280 280 sign, 320, 400 280, 320, 400 sign 280 int, 320 weak, 4C0 Weak, 680-740 280, 680-740 sign 280, 340weak, 680-740 280, 320 weak, 880 280, 320, 4C0, 520, 720

NH4v03 NH4v03 NH4v03 NH4V03

7. NaV03 8. KVO,

11.NH vo3 r&ed

280, 603

12.m

280, 320weak, 540-640

vo

O2

x1

x2

x1

0.35 0.75

0.38 0.72

0.43 0.72

0.08 0.09 0.08 0.23

0.02 0.60 0.55 0.29

0.10 0.10

0.15 0.18 0.20 0.28

0.18 0.20 0.53 0.60

0.10 0.40 0,30 0.32 0.30

mfti2mponent reduced 13.K V 0 6 10&8

M203

reduc

a

$4

- 280

nm, V

280, 320, 400, 520, 720

- 340-360 ss (w)

z - 320 nm, V4+

V 4 + ( a 0 ) - 520 nm, V z C w ) - 600 nm, VTd(A1203) 4+ Oh 2 3 V4’(~)

m

-

m, V4+ (Al 0

ss

2 3

- 720 nm,

800-880 nm /5,6/.

Schem of oxidative dehydrqenation of alkylaromatic c m p n d s Ar-cH2-a3

I

I

-

Ar-CH=cH2

1,

Ar-cHGH-cH

Ar-CH-CH 0 1 ‘’

I

- 400

nm,

459

Table 4 Oxidative dehydrogenation of organic compounds v 0 /fipo, dticomp3nent C&&und:C H5N02 r_H20 = 1:3:10, 42OoC. cCanp0und:SO2:H20 = $:0.4:10, 500OC. CcPTIpouna: H20 - l:l:lO, 48OoC. Space velocity 0 , 5 h-

8,:

ccmpouna

Acceptor

Process characteristic

x1

x2

s2

0.90 0.95 0.98

0.72

0.72

0.76 0.70 0.72

0.65

0.60

0.70

0.68

0.70

0.90

0.40

0.75

S1

~

Ethylbenzene

O2

so2,a s Isopropylknzene

‘SHSNo2 O2

so2,cos Diphenylethane

‘gHgN02 O2 s02 COS

Diethylknzene Ethyltoluene Ethylpyridine Ethylthiophene Methane Isobutane

‘SHSN02 O2 O2 O2 O2 O2 ‘gHgN02

so2’ ‘gHgN02

0.72 0.80 0.75 0.20 0.77 0.35

0.50 0.85

0.87 0.70

0.85 0.75

0.96 0.82

0.60

0.90

0.15

0.70

0.60 0.30 0.05

0.80 0.70

0.90

-

0.65

0.45

0.95 0.96

CONCLUSIONS

The valent,andcoordination states of vanadium ions, t h e size of vanadium clusters (2-12 ions of V) and associated activity, selectivity and stability were sham to depend on the content of vanadium in catalysts samples, the nature of salts in the component, msdifying agents, and supprts-(Mgo, Al 0 2 3‘ TiO,, ZrO,, CaO) . .. h e sugeested structure of an active centre comprises V>+ and V4’ ions i n octahedral configuration, grouped in clusters and ensures high activity and selectivity in the production of vinylarmtic and vinylheterocyclic ccnnpounds and aniline. The nature of the acceptor used has been found to affect the reactivity

460

of the studied ccmpounds in oxidative dehydrogenation. The electro- and Iraechanochdcal methods of preparation permit to regulate the pore structure, termic stability, the content of active components, m k e it possible to prepare the samples with associated species at lower contents of active canponent and hence to regulate the activity and selectivity. The above studies made it possible to formulate the scientifically preparation and application of vanadium containing oxide system , to obtain high activity and selectivity in the process.

3 4 5 6 7 8

9 10 11

G.E Vrieland, J.Catal.lll(1988) 1. F.M. Bautista, J.M. Campelo, A. Garcia, D. Luna and J. M. Marinas, J.Catal., 116 (1989) 338. M. Shimanska, L. Leitis, R. Skolmeistere, I. Iovel, L. Golender. Vanadia catalysts for the oxidation of heterocyclic caqmmds. Riga "Zinkitne" Publishers, 1990. 256 p. J. Haber, A. Kozlowska, K. Kozlowski, J.Catal., 102 (1986) 52. G. Busca, G. Centi, L. Marchetti, F. Trifiro, Langmir. No 2 (1986) 568. W. Hanke, K. Heise, H . 4 . Jerschkewitz, G. Lischke, G. Ohlmann, B. Parlitz, Z.Anorg.Allgem.Chem., 438 (1978) 176. O.B. Lapina, A.V. Simakov, V.M. Mastikhin, S.A. Veniaminov, A.A. Shubin, J.MOl.Catal., 50 (1989) 55. A.V. Sirc.lakov,N.N. Sazonova, S.A. Veniaminov, I.P. Belomestnykh, N.N. Rozhdestvenskaya, G.V. Isaguliants, Kinetika and Kataliz, 30 (1989) 684. w. Hanke, R. Bienert, H . 4 . Jerschkewitz, Z.Anorg.Allgm.Chem., 44 (1975) 109. G. Lischke, W. Hanke, H . 4 . Jerschkewitz, G. Ohlmnn, J.Catal., 91 (1985) 54. E.G. Klhchuk, B.N. Shelirrpv, V.B. Kazanski, Kinetica and Kataliz, 26 (1985) 396.

P. Ruiz and B. Delmon (Eds.) New Developments in Selective Oxidation by Heterogeneous Catalysis Studies in Surface Science arid Catalysis, Vol. 72, pp. 461-468 @ 1992 Elsevier Science Publishers B.V. All rights reserved.

461

Immobilized hemin catalyst in oxidation processes 111. Oxidation of cysteine Yu.L. Zuba, T.N. Yakubovicha and G.P. Potapovb a

Institute of Surface Chemistry, pr. Nauki 31,252650 Kiev, USSR

b Department of Organic and Biological Chemistry, University of Siktivkar, pr.

Oktjabrsky 55,167001 Siktivkar, USSR

Abstract It is shown that 3-aminopropylpolysiloxane prepared by the hydrolytic polycondensation of Si(OEt)4 and (Et0)3Si(CH2)3NH2 is a space-crosslinked polymer with functional amino groups on its surface. Hemin (a complex of Fe(II1) with protoporphyrin IX) was attached t o the new matrix with participation of the latter. The resulting catalyst had a high efficiency in the reaction of cysteine with molecular oxygen. 1. INTRODUCTION

The selection of optimal pairs of metallocomplexes and their supports is one of the major problems in heterogeneous catalysis [l-41.This selection can result in the development of highly efficient and selective catalysts, and very often some form of silica is used as the support [51. In general, almost all forms of silica have silanol groups on their surface, and this allows the surface to be modified. However, many types of silica have very few surface silanol groups and this can limit the extent of any modification. This can also mean t h a t any grafted ligands may shear off in use [61. The use of polyorganosiloxane matrices with different functional groups can avoid some of these problems. The main method for the preparation of functional polysiloxanes is the hydrolytic polycondensation of alkoxysilanes according to the following scheme: Si(OR),

+ (R0)3Si(CH2),R

+H20

$0

-ROH

pO-Si(CH2)nR' +O

---------- >

1

Organosilicon compounds of this type (where R = -SH and n = 1) were first obtained by Finn et al. [7]. The samples thus made were then used in the study of metal sorption [8-121. Later, many reports appeared i n the literature describing the use of functional polysiloxane matrices i n catalysis [13-211. Herein, we report the synthesis of 3-aminopropyl-polysiloxanesupport (SAP) and the fixation of hemin t o its surface [Fe(III) complex with protoporphyrin M (Fig. l)] as well a s its catalytic properties in the oxidation of cysteine.

462

H

CH2

HO-c

\

\\

0

0Rc-OH

Figure 1. Hemin b. 2. EXPERIMENTAL 2.1 Materials

Hemin ("pure"), distilled 3-aminopropyltriethoxysilane and tetraethoxysilane, L-cysteine (Reanal), DMFA ("pure"), and carbonate-bicarbonate buffer were used in this study. The SAP matrix was prepared by the technique described in Ref. [181, keeping the reagent ratio (EtO)4Si:(EtO)3Si(CH2)3NH2 = 2:l. The amino group content was evaluated by the number of H+ ions bound to the matrix (after 72 h, back titration gave C S N H ~= 2.2 mmoVg and S = 130 m2/g). 2 2 Catalyst preparation SAP (2 g) was added to 50 ml of a hemin solution in DMFA (C = (1.0-1.5)x 10-3 mmol/g) and the resulting suspension was mixed in a magnetic stirrer for 2 -h. The content of fixed hemin was determined by the decrease in optical density in the electronic absorption spectrum of the mixture. For SAP it gave a hemin content of CS = 2.47 x 10-2 and 3.82 x 10-2mmoVg.

463

2.3 spectralstudies Infrared spectra were recorded with a UR-20 spectrophotometer (in Nujol and CCl4). Electronic absorption spectra were recorded with a Specord M-40 (UV-VIS) spectrophotometer. ESR spectra were measured with a S E E 2543 spectrometer a t 96 K. 2.4 Kinetic studies The kinetics of the oxidation of cysteine were investigated by measuring the initial rates of 0 2 absorption (Warburg apparatus, kinetic regime at 295 K). The catalyst concentration was in the range 3.2-24.9 x 10-4 molfl. The cysteine concentration varied from 0.025 to 0.1 moM. The 0 2 pressure was 19.6 Wa and the pH was kept constant at 9.6.

3.1. Infrared spectra In the region above 1000 cm-1 two strong, broad absorption bands a t 1060 and 1160 cm-1 were observed in the spectrum of the SAP matrix. Additionally, two weak absorption bands were detected a t 3300 and 3360 cm-1 in the background of an intense and very broad absorption band. 3.2. Electronic absorptionspectra The hemin solution in DMFA gave four absorption bands a t 390, 510, 545, and 640 nm. The fixation of hemin onto the matrix resulted in the shift of these absorption bands to 410, -490, -530, and 610 nm, respectively (see Fig. 2).

3.3. ESR analysis Figure 3 shows the ESR spectra and parameters for hemin in the polycrystalline state and for hemin supported by the SAP substrate.

3.4. Kinetic data Figures 4 and 5 show the kinetic curves of cysteine oxidation and they demonstrate the reaction rate dependence on substrate and catalyst concentration. 4. DISCUSSION The observed absorption bands a t 1060 and 1160 cm-l in the IR spectrum of the SAP matrix are indicative of space-crosslinked polyorganosiloxane fragments [221. The existence of the three-dimensional polymeric skeleton is confirmed by the insolubility and non-swellability of the support in organic solvents. The absorption bands beyond 3000 cm-1 can be assigned t o the symmetric and asymmetric stretching vibrations of NH2 involved in hydrogen bonding. The coordination of the hemin bond t o the matrix can be confirmed by the hypsochromic shift of the absorption bands in the 450-750 nm range as a similar effect was observed for hemin solutions (or analogues) in the presence

464

340

420

500

580

660

A,nm

Figure 2. Absorption spectra of hemin in DMFA. (1)Chemin = 9.85 x mol/l. (1')Chemin = 1.97 x 10-4 molfl. (2) Hemin supported by SAP matrix: Chemjn = 1.06 x 10-3molA.

465

Figure 3. ESR spectra of: (1)hemin; and (2) SAP supported hemin, g l = 6.15 and gl! = 1.99.

of a pyridine-nitrogen containing ligand [23]. The presence of the -NH2+Fe bond is also confirmed by the ESR data, which are consistent with the results of a study, using the same method, of a chloro-hemin complex with pyridine in the truns-position 1241. It should be mentioned that hemin also appears to be linked to the surface through carboxyl groups, taking into account the proportion of surface amino groups and the amount of fixed hemin. The curve in Fig. 5 shows a complex dependence of the reaction rate on the catalyst concentration. It cannot be discounted that this is due to the different character of the active sites on the catalyst surface. The latter may result from the different environments of these sites, i.e., on one hand, an excess of aminopropyl groups (relative to the amount of fixed hemin), and on the other, the porosity of the matrix itself. A similar dependence of the reaction rate on the catalyst concentration was also observed in the case of cysteine oxidation when Fe(I1) supported by ionite was present [25].

466

1

Figure 4. Initial rate dependence of cysteine oxidation on cysteine concentration: Ccatalyst = 4.78 x 10-4moM; pH = 9.6.

1

0

2

4

6

8

10

12

14 16

Figure 5 . Initial rate dependence of cysteine oxidation on catalyst concentration: Ccysteine = 0.05 moM; pH = 9.6.

467

5. ACKNOWLEDGMENT

Yu.L.Z. and T.N.Y. express sincere gratitude t o Professor A.A. Chuiko for his interest in this work.

6.REFERENCES 1 B.C. Gates, L. Guczi and H. Knozinger (eds.), Metal Clusters in Catalysis, Elsevier, Amsterdam, 1986. 2 F.R. Hartley, Supported Metal Complexes, D. Reidel Publ. Co., Dordrecht, 1985. 3 G.V. Lisichkin and A.Ju. Juffa, Heterogeneous Metallocomplex Catalysts, Khimija, Moscow, 1981. 4 Yu.1. Yermakov, V.A. Zakharov and B.N. Kuznetsov, Fixed Complexes on Oxide Supports in Catalysis, Nauka, Novosibirsk, 1980. 5 G.V. Lisichkin, G.V. Kudrjavtsev, A.A. Serdan, S.M. Staroverov and A.Ju Juffa, Modified Silicas i n Sorption, Catalysis and Chromatography, Khimija, Moscow, 1986. 6 R.V. Parish and M.I. Vania, J. Organomet. Chem., 263(1984)139. 7 L.P. Finn, I.B. Slinjakova, M.G. Voronkov, N.N. Vlasova, F.P. Kletsko, A.I. Kirillov and T.N. Shkl'ar, Russ. Reports AS USSR, 236(1977)1426. 8 N.N. Vlasova, L.M. Stanevish, S.A. Bolshakova and M.G. Voronkov, Russ. J. Appl. Chem., 60(1987)1479. 9 N.N. Vlasova, M.G. Voronkov, S.A. Bolshakova, Ju.N. Pozhydaev and A.I. Kirillov, Russ. J. Gener. Chem., 54(1984)2306. 10 M.G. Voronkov, N.N. Vlasova, M.Ju. Adamovich, Ju.N. Pozhydaev and A.I. Kirillov, Russ. J. Gener. Chem., 54(1984)865. 11 A.I. Kirillov, O.V. Zemljanushnova, N.N. Vlasova, M.G. Voronkov, I.B. Slinjakova and L.P. Finn, Russ. Anal. Chem., 37(1982)1201. 12 O.V. Zemljanushnova, A.I. Kirillov, I.P. Golentovskaja and N.N. Vlasova, Russ. Higher Educ. Instit., Chem. and Techn., 25(1982)568. 13 F.G. Younf, Ger. Patent No 2 330 308 (1974). 14 S. Suzuki, K. Tohmori and Y. Ono, J. Mol. Catal., 43(1987)41. 15 S. Suzuki, Y. Ono, S. Nakata and S. Asaoka, J. Phys. Chem., 91(1987)1659. 16 U. Shubert, K. Rose and H. Schmidt, J. Non-Cryst. Solids, 105(1988)165. 17 U. Schubert and K. Rose, Transit. Metal Chem., 14(1989)291. 18 I.S. Khatib and R.V. Parish, J. Organomet. Chem., 369(1989)9. 19 R.V. Parish, D. Habibi and V. Mohammadi, J. Organomet. Chem., 369(1989)17. 23 H.S. Hilal, A. Rabah, I.S. Khatib and A.F. Schreiner, J . Mol. Catal., 61(1990)1. 21 H.S. Hilal, C. Kim, M.L. Sit0 and A.F. Schreiner, J. Mol. Catal., 64(1991)133. 22 I.B. Slinjakova and T.I. Denisova, Organo-Silicon Adsorbents : Preparation, Properties, Application, Naukova Dumka, Kiev, 1988. 23 G.B. Eichorn (ed.), Inorganic Biochemistry, Elsevier, Amsterdam, Vol. 1 (1973);Vol. 2 (1975). 24 T.H. Moss, A.J. Bearden and W.S. Caughey, J. Chem. Phys., 51(1969)2624. 25 A.N. Astanina, Vesi Masisi, V.D. Kopylova, G.A. Smirnova, A.P. Rudenko, E.Ya. Frumkina and G.A. Artjushin, Dep. VINITI, No 5659-81 (1981).


469

AUTHOR INDEX Acosta. D . Ai. M . Andrushkevich. T.V. Anshits. A.G. .... Auroux. A .

267 101 91 155 181

Baerns. M . Bartoli. M.J. Bastians. Ph ...... Belomestnykh. 1.P Borchert. H . C . Bordes. E ........ Brunelle. J.P. Busca. G . . . . Buskens. Ph .

57 81 267 453 .... 57 81. 165 387 335 21

Cao. H . . Centi. G . Chuang. K.T. .. Corma. A ........ Cortes Corberin. V . Coudurier. G . . Coulson. D.R Courtine. P . Creten. G .

213 23 1 23 213 147 191 305 81. 165 ..... 317

Daza. L . Delmon. B . de Boer. M . . de Goede. A.T.J.W. de Wit. D . Dreoni. D.P. . Driscoll. S.A.

267 267. 399 133 1 1 109 363

Ebner. J.R. Emig. G . . . . Escudey. M . Farinha-Portela. M . Fiedorow. R . Fierro. J.L.G. Froment. G.F. Fu. L . Gabrielov. A.G. Genet. M . . Gesser. H.D. Geus. J.W. Gil Llambias. P.J. . Gleaves . G . (Invited lecture) Golinelli. G . ‘(Invited lecture)

353 . 71 435 325

....... 23

147. 203 317 23 443 267 155 123. 133 . 435 23 1 23 1

Grzybowska. B . Guerrero.Ruiz. A . Guilhaume. N . . . .

255 203 255

Haber. J ........... Hecquet. G ........ Herrmann. J.M. . Huybrechts. D.R.C.

279 81 203 . 2 1

..

Iovel. I ............. Isaguliants. G.V.

117 453

Jacobs. P.A. Jaeger. Ph . . Janssen. F . .

21 387 133

KaBner. P . . Kalthoff. R . Kiperman. S.L. .... Koningsberger. D.C Kopinke. F.-D. ..... Kourtakis. K ........ Kuster B.F.M. ....

.

Le Bars. J . Legendre. 0. Leitis. L . . . . Lerou. J.J. ........... Lopez.Nieto. J.M. . Lorenzelli. V ........ Loukah. M .......... Lukevics. E .........

57

. 57 345 133 317 305 43 181 387 117 305 213 335 191 117

Macias. M . . . . . Mamedov. E.A. Manzer. L.E. .. Marcel C ....... Marchena. F.J. Marin G.B. .... Martin. C . ...... Martin. M.J. ... Matsuura. I . . . . Merzouki. M . . Mills. P.L. ..... Monceaux. L . .

423 379 305 335 423 43 415 415 247 165 ..... 305 81. 165

Navio . J.A. ...

..... 423

Oliveira. M . . . . Ozkan. U.S. ...

.... 325 ..... 363

470

Pajonk, G. Paredes, N. Perez, M. Pinelli, D. Pinheiro, C. Plyasova, L.M. Potapov, G.P.

255 213 213 109 325 91 46 1

Quaranta, N.E.

147

Real, C. Rives, V. Rizayev, R.G. Rodriguez-Ramos, I. Romanovsky, B.V. Ross, J.R.H. Roullet, M. Rozhdestvenskaya, N.N Ruiz, P. Sanchez Escribano, V. Scholten, A., Schurrman, Y. Seshan, K. Shigapov, A.N. Shimanska, M. Skrigan, E.A. . Smith, M.S. ...

423 415 379 203 443 22 1 255 45 3 267. 399 335 123 . 43 22 1 155 117 453 363

Smits, R.H.H Sne, Y. .._ Soenen, V. Suib, S.L.

22 1 213 203 213

Talyshinskii, R.M. Taouk, B. Thompson, M.R. Trautmann, S. Trifirb F. .

379 165 353 57 109, 231

van Bekkum, H van den Brink, P.J van der Wiele, K. van Dillen, A.J. Vedrine, J . Vereshchagin, S . N Vinke, P. Vislovskii, V.P. Volta J.C.

1 123 ..... 43 123, 133 181, 191 155 1 379 203, 255

Watzenberger, 0. Weng, L.T.

71 399

Yakubovich, T.N.

46 1

Zein, A. Zub, Yu.L.

57 46 1

471

SUBJECT INDEX Acetylacetonate complexes Acrdlein Acrolein Acrylic acid Alkane Alkylaromatic Alkylheterocyclic Alumina Ammonia Ammoximation

423 91 387 91 21 453 45 3 415 133 109

Basicity Bifunctional Bismuth molybdate Butane oxidation

379 109 305, 325, 387 .. 247

C5 alkanes Carbohydrates Carbon supported Catalytic oxidation Chromium Classification Cobalt Cylo hexane Cysteine

23 1 1 43 279 191 399 387 109 461

I

Dicarboxylic anhydrides

57

...

435 123 165, 191 147

Electrophoretic study Elemental sulfur Ethane . Ethanol Fluorene

57

H202 Hemin Heterocyclic compounds Heteropolyacid catalysts Hydrogen.. .......... Hydrogen peroxide Hydrogen sulfide

21 461 117 71 33 33 123

Immobilized .......... In situ investigation Infrared spectroscopy Iron Iron sulfate catalysts Isobutyric acid

46 1 91 335 387 123 71

Kinetics of reoxidation

305

Kinetic problems of selectivity

345

Light alkanes Local superficial structurt

155 255

M/TiO2 Malleic anhydride MetallocompIexes Methyl-a-D-glucoside Methy la1 Methyl acetatr. Microcalori metrics Mo-Sb Moly bdena Molybdena catalysts Multicomponen t

423 255 443 43 101 101 181 435 415 133 443

Niobium pentoxide Nitrogen Noble metal

22 1 133

..

1

O-xylene 317 Olefin 325 Oxidative dehydrogenation .181,213, 221 Oxydehydrogenation 71 Oxydeh ydrogenation 203 PdC12 , Phenanthrene Phosphorous vanadia Propane Propene Propylene Pt catalysts

335 57 267 203, 213, 221 387

305 43

Quinones Roles

57 379, 399

Selective heterogeneous oxidation Selective oxidation Silica Silver catalysts Site isolatiop Sn-Sb Stabilization of heteropolyacids. Structure-selectivity Surface sites Synergetic effects Synergy effects

23 1 43 133 155 353 435 81 133 203 267 363

472

TAP ............................... 305. Temperature programmed desorption Ti02-coated silica ...................... Titania ................................... Titanium silicalite ................. 21. Topological heterogenization .........

317 325 147 415 453 443

Vanadium-molybdenum ................. 91 165 Vanadium oxide ........................ Vanadium pentoxide ................... 181 Vanadium phosphorus ................ 353 Vanadyl pyrophosphate .............. 247 255 VPO .....................................

V-Mg-0 ................................ Vanadia-titania ......................... Vanadia catalysts .......................

203 335 147

Water ....................................

71

Zirconium phosphates ................ 191

473

STUDIES IN SURFACE SCIENCE AND CATALYSIS Advisory Editors: B Delmon, Universite Catholique de Louvain, Louvain-la-Neuve, Belgium J T Yates, University of Pittsburgh, Pittsburgh, PA, U S A

Volume

1

Volume 2

Volume 3

Volume 4

Volume 5

Volume 6

Volume 7

Volume 8 Volume 9

Volume 10

Volume 1 1

Volume 12

Volume 13 Volume 14

Preparation of Catalysts I. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the First International Symposium, Brussels, October 1417, 1975 edited by B. Delmon, P.A. Jacobs and G. Poncelet The Control of t h e Reactivity of Solids. A Critical Survey of the Factors that Influence the Reactivity of Solids, with Special Emphasis on the Control of the Chemical Processes in Relation to Practical Applications by V.V. Boldyrev, M. Bulens and B. Delmon Preparation of Catalysts 11. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Second International Symposium, Louvain-la-Neuve, September 4-7, 1978 edited by B. Delmon, P. Grange, P. Jacobs and G. Poncelet Growth and Properties of M e t a l Clusters. Applications to Catalysis and the Photographic Process. Proceedings of the 32nd International Meeting of the Societe de Chimie Physique, Villeurbanne, September 24-28, 1979 edited by J. Bourdon Catalysis by Zeolites. Proceedings of an International Symposium, Ecully (Lyon), September 9-1 1, 1980 edited by B. Imelik, C. Naccache. Y . Ben Taarit, J.C. Vedrine, G. Coudurier and H. Praliaud Catalyst Deactivation. Proceedings of an International Symposium. Antwerp, October 13-15, 1980 edited by B. Delmon and G.F. Froment N e w Horizons in Catalysis. Proceedings of the 7th International Congress on Catalysis, Tokyo, June 30-July 4, 1980. Parts A and B edited by T. Seiyama and K. Tanabe Catalysis by Supported Complexes by Yu.1. Yermakov, B.N. Kuznetsov and V.A. Zakharov Physics of Solid Surfaces. Proceedings of a Symposium, Bechyiie. September 29-October 3, 1980 edited by M . LazniEka Adsorption a t the Gas-Solid and Liquid-Solid Interface. Proceedings of an International Symposium, Aix-en-Provence, September 2 1-23, 198 1 edited by J. Rouquerol and K.S.W. Sing Metal-Support and Metal-Additive Effects in Catalysis. Proceedings of an International Symposium. Ecully (Lyon), September 14-1 6, 1982 edited by B. Imelik, C. Naccache. G.Coudurier, H. Praliaud, P. Meriaudeau, P. Gallezot, G.A. M a r t i n and J.C. Vedrine M e t a l Microstructures in Zeolites. Preparation - Properties - Applications. Proceedings of a Workshop, Bremen, September 22-24, 1982 edited by P.A. Jacobs, N.I. Jaeger, P. JirG and G. Schulz-Ekloff Adsorption on M e t a l Surfaces. An Integrated Approach edited by J. Benard Vibrations at Surfaces. Proceedings of the Third International Conference Asilomar, CA, September 1-4, 1982 edited by C.R. Brundle and H. Morawitz

474 Volume 15 Volume 16

Volume 17

Volume 18

Volume 19

Volume 2 0

Volume 2 1

Volume 22 Volume 23 Volume 2 4

Volume 25

Volume 2 6

Volume 27 Volume 28

Volume 2 9 Volume 3 0

Volume 3 1

Volume 32 Volume 3 3 Volume 3 4

Heterogeneous Catalytic Reactions Involving Molecular Oxygen by G.I. Golodets Preparation of Catalysts Ill. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Third International Symposium, Louvain-la-Neuve, September 6-9, 1982 edited by G. Poncelet, P. Grange and P.A. Jacobs Spillover of Adsorbed Species. Proceedings of an International Symposium, LyonVilleurbanne, September 12-1 6, 1983 edited by G.M. Pajonk, S.J. Teichner and J.E. Germain Structure and Reactivity of Modified Zeolites. Proceedings of an International Conference, Prague, July 9-1 3, 1984 edited by P.A. Jacobs, N.I. Jaeger. P. JirG, V.B. Kazansky and G. Schulz-Ekloff Catalysis on the Energy Scene. Proceedings of the 9th Canadian Symposium on Catalysis, Quebec, P.Q., September 30-October 3, 1984 edited by S.Kaliaguine and A. Mahay Catalysis by Acids and Bases. Proceedings of an International Symposium, Villeurbanne (Lyon), September 25-27, 1984 edited by B. Imelik, C. Naccache, G. Coudurier, Y. Ben Taarit and J.C. Vedrine Adsorption and Catalysis on Oxide Surfaces. Proceedings of a Symposium, Uxbridge, June 28-29, 1984 edited by M. Che and G.C. Bond Unsteady Processes in Catalytic Reactors by Yu.Sh. Matros Physics of Solid Surfaces 1984 edited by J. Koukal Zeolites: Synthesis, Structure, Technology and Application. Proceedings of an International Symposium, Portoroi-Portorose, September 3-8, 1984 edited by B. Driaj, S.HoEevar and S.Pejovnik Catalytic Polymerization of Olefins. Proceedings of the International Symposium on Future Aspects of Olefin Polymerization, Tokyo, July 4-6, 1985 edited by T. Keii and K. Soga Vibrations at Surfaces 1985. Proceedings of the Fourth International Conference, Bowness-on-windermere, September 15-19, 1985 edited by D.A. King, N.V. Richardson and S.Holloway Catalytic Hydrogenation edited by L. Cervenq New Developments in Zeolite Science and Technology. Proceedings of the 7th International Zeolite Conference, Tokyo, August 17-22, 1986 edited by Y. Murakami, A. lijima and J.W. Ward Metal Clusters in Catalysis edited by B.C. Gates, L. Guczi and H. Knozinger Catalysis and Automotive Pollution Control. Proceedings of the First International Symposium, Brussels, September 8-1 1, 1986 edited by A. Crucq and A. Frennet Preparation of Catalysts IV. Scientific Bases for the Preparation of Heterogeneous Catalysts. Proceedings of the Fourth International Symposium, Louvain-la-Neuve, September 1-4, 1986 edited by B. Delmon, P. Grange, P.A. Jacobs and G. Poncelet Thin Metal Films and Gas Chemisorption edited by P. Wissmann Synthesis of High-silica Aluminosilicate Zeolites by P.A. Jacobs and J.A. Martens Catalyst Deactivation 1987. Proceedings of the 4th international Symposium, Antwerp, September 29-October 1, 1987 edited by B. Delmon and G.F. Froment

475 Volume 35 Volume 3 6

Volume 37

Volume 3 8

Volume 3 9

Volume 4 0

Volume 4 1

Volume 42 Volume 43 Volume 4 4

Volume 45 Volume 46

Volume 47 Volume 48

Volume 49

Volume 5 0

Volume 5 1 Volume 52

Volume 53

Keynotes in Energy-Related Catalysis edited by S . Kaliaguine Methane Conversion. Proceedings of a Symposium on the Production of Fuels and Chemicals from Natural Gas, Auckland, April 27-30, 1987 edited by D.M. Bibby, C.D. Chang, R.F. Howe and S.Yurchak Innovation in Zeolite Materials Science. Proceedings of an International Symposium, Nieuwpoort, September 13-1 7, 1987 edited by P.J. Grobet, W.J. Mortier, E.F. Vansant and G. Schulz-Ekloff Catalysis 1 9 8 7 . Proceedings of the 10th North American Meeting of the Catalysis Society, San Diego, CA, May 17-22, 1987 edited by J.W. Ward Characterization of Porous Solids. Proceedings of the IUPAC Symposium (COPS I), Bad Soden a. Ts., April 26-29, 1987 edited by K.K. Unger, J. Rouquerol, K.S.W. Sing and H.Kral Physics of Solid Surfaces 1 9 8 7 . Proceedings of the Fourth Symposium on Surface Physics, Bechyne Castle, September 7-1 1, 1987 edited by J. Koukal Heterogeneous Catalysis and Fine Chemicals. Proceedings of an International Symposium, Poitiers, March 15- 17, 1988 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, C. Montassier and G. Perot Laboratory Studies of Heterogeneous Catalytic Processes by E.G. Christoffel, revised and edited by 2 . Paal Catalytic Processes under Unsteady-State Conditions by Yu. Sh. Matros Successful Design of Catalysts. Future Requirements and Development. Proceedings of the Worldwide Catalysis Seminars, July, 1988, on the Occasion of the 30th Anniversary of the Catalysis Society of Japan edited by T. lnui Transition Metal Oxides. Surface Chemistry and Catalysis by H.H. Kung Zeolites as Catalysts, Sorbents and Detergent Builders. Applications and Innovations. Proceedings of an International Symposium, Wurzburg, September 48, 1988 edited by H.G. Karge and J. Weitkamp Photochemistry on Solid Surfaces edited by M. Anpo and T. Matsuura Structure and Reactivity of Surfaces. Proceedings of a European Conference, Trieste, September 13- 16, 1988 edited by C. Morterra, A. Zecchina and G. Costa Zeolites: Facts, Figures, Future. Proceedings of the 8th International Zeolite Conference. Amsterdam, July 10-14, 1989. Parts A and B edited by P.A. Jacobs and R.A. van Santen Hydrotreating Catalysts. Preparation, Characterization and Performance. Proceedings of the Annual International AlChE Meeting, Washington, DC, November 27-December 2, 1988 edited by M.L. Occelli and R.G. Anthony New Solid Acids and Bases. Their Catalytic Properties by K. Tanabe. M. Misono, Y. Ono and H. Hattori Recent Advances in Zeolite Science. Proceedings of the 1989 Meeting of the British Zeolite Association, Cambridge, April 17-19, 1989 edited by J. Klinowski and P.J. Barrie Catalyst in Petroleum Refining 1 9 8 9 . Proceedings of the First International Conference on Catalysts in Petroleum Refining, Kuwait, March 5-8, 1989 edited by D.L. Trimm. S . Akashah, M. Absi-Halabi and A. Bishara

476

Future Opportunities in Catalytic and Separation Technology edited by M. Misono, Y. Moro-oka and S.Kimura Volume 55 N e w Developments in Selective Oxidation. Proceedings of an International Symposium, Rimini, Italy, September 18-22, 1989 edited by G. Centi and F. Trifiro Volume 56 Olefin Polymerization Catalysts. Proceedings of the International Symposium on Recent Developments in Olefin Polymerization Catalysts, Tokyo, October 23-25, 1989 edited by T. Keii and K. Soga Volume 57A Spectroscopic Analysis of Heterogeneous Catalysts. Part A: Methods of Surface Analysis edited by J.L.G. Fierro Volume 578 Spectroscopic Analysis of Heterogeneous Catalysts. Part 6: Chernisorption of Probe Molecules edited by J.L.G. Fierro Volume 58 Introduction t o Zeolite Science and Practice edited by H. van Bekkum, E.M. Flanigen and J.C. Jansen Volume 59 Heterogeneous Catalysis and Fine Chemicals II. Proceedings of the 2nd International Symposium, Poitiers, October 2-6, 1990 edited by M. Guisnet, J. Barrault, C. Bouchoule, D. Duprez, G. Perot, R. Maurel and C. Montassier Volume 6 0 Chemistry of Microporous Crystals. Proceedings of the International Symposium on Chemistry of Microporous Crystals, Tokyo, June 26-29, 1990 edited by T. Inui, S. Namba and T. Tatsumi Volume 6 1 Natural Gas Conversion. Proceedings of the Symposium on Natural Gas Conversion, Oslo, August 12-1 7, 1990 edited by A. Holmen, K.-J. Jens and S.Kolboe Volume 62 Characterization of Porous Solids II. Proceedings of the IUPAC Symposium (COPS II), Alicante, May 6-9, 1990 edited by F. Rodriguez-Reinoso, J. Rouquerol, K.S.W. Sing and K.K. Unger Preparation of Catalysts V. Proceedings of the Fifth International Symposium on Volume 63 the Scientific Bases for the Preparation of Heterogeneous Catalysts, Louvain-laNeuve, September 3-6, 1990 edited by G. Poncelet, P.A. Jacobs. P. Grange and 6. Delmon Volume 6 4 New Trends in CO Activation edited by L. Guczi Volume 65 Catalysis and Adsorption by Zeolites. Proceedings of ZEOCAT 90, Leipzig, August 20-23, 1990 edited by G. Ohlmann, H. Pfeifer and R. Fricke Volume 66 Dioxygen Activation and Homogeneous Catalytic Oxidation. Proceedings of the fourth International Symposium on Dioxygen Activation and Homogeneous Catalytic Oxidation, Balatonfured, September 10-1 4, 1990 edited by L.I. Simandi Volume 67 Structure-Activity and Selectivity Relationships in Heterogeneous Catalysis. Proceedings of the ACS Symposium on Structure-Activity Relationships in Heterogeneous Catalysis, Boston, MA, April 22-27, 1990 edited by R.K. Grasselli and A.W. Sleight Volume 6 8 Catalyst Deactivation 1991 . Proceedings of the fifth International Symposium. Evanston, IL, June 24-26, 1991 edited by C.H. Bartholomew and J.B. Butt Volume 6 9 Zeolite Chemistry and Catalysis. Proceedings of an International Symposium, Prague, Czechoslovakia, September 8-13, 199 1 edited by P.A. Jacobs, N.I. Jaeger, L. Kubelkova and B. Wichterlova Volume 5 4

477 Volume 7 0

Volume 7 1

Volume 72

Poisoning and Promotion in Catalysis based on Surface Science Concepts and Experiments by M. Kiskinova Catalysis and Automotive Pollution Control II. Proceedings of the 2nd International Symposium (CAPoC 2). Brussels, Belgium, September 10-1 3, 1990 edited by A. Crucq N e w Developments in Selective Oxidation by Heterogeneous Catalysts. Proceedings of the Third European Workshop Meeting, Louvain-la-Neuve, Belgium, April 8-10, 1991 edited by P. Ruiz and B. Delmon


New Developments in Selective Oxidation by Heterogeneous Catalysis: Proceedings (Studies in Surface Science and Catalysis)

New Developments in Selective Oxidation (Studies in Surface Science and Catalysis vol55)

Catalysis by Zeolites: International Symposium Proceedings (Studies in surface science and catalysis)

New Developments in Catalysis Research

Studies in Surface Science and Catalysis 110. 3rd World Congress on Oxidation Catalysis

Isotopes in Heterogeneous Catalysis

Isotopes In Heterogeneous Catalysis

New Frontiers in Catalysis, Parts A-C (Studies in Surface Science and Catalysis)

Catalysis: An Integrated Approach to Homogenous, Heterogenous and Industrial Catalysis (Studies in Surface Science and Catalysis)

Catalysis by Metals and Alloys (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis in Organic Chemistry

Catalysis by Acids and Bases (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis in Industrial Practice

Mass Transfer in Heterogeneous Catalysis

Heterogeneous Catalysis in Organic Chemistry

Catalysts for Upgrading Heavy Petroleum Feeds, Volume 169 (Studies in Surface Science and Catalysis) (Studies in Surface Science and Catalysis)

Past and Present in DeNOx Catalysis: From Molecular Modelling to Chemical Engineering, Volume 171 (Studies in Surface Science and Catalysis) (Studies in Surface Science and Catalysis)

Catalytic Polymerization of Olefins (Studies in Surface Science and Catalysis)

Photochemistry on Solid Surfaces (Studies in Surface Science and Catalysis)

Fluid Catalytic Cracking V (Studies in Surface Science and Catalysis)

Laboratory Studies of Heterogeneous Catalytic Processes (Studies in Surface Science and Catalysis)

Fluid Catalytic Cracking V (Studies in Surface Science and Catalysis)

Preparation of Catalysts VII (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis Research Progress

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Surface and nanomolecular catalysis

Organometallic Oxidation Catalysis

Surface and Nanomolecular Catalysis

Catalysis and Automotive Pollution Control (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis Research Progress

New Developments in Selective Oxidation by Heterogeneous Catalysis: Proceedings (Studies in Surface Science and Catalysis)

New Developments in Selective Oxidation (Studies in Surface Science and Catalysis vol55)

Catalysis by Zeolites: International Symposium Proceedings (Studies in surface science and catalysis)

New Developments in Catalysis Research

Studies in Surface Science and Catalysis 110. 3rd World Congress on Oxidation Catalysis

Isotopes in Heterogeneous Catalysis

Isotopes In Heterogeneous Catalysis

New Frontiers in Catalysis, Parts A-C (Studies in Surface Science and Catalysis)

Catalysis: An Integrated Approach to Homogenous, Heterogenous and Industrial Catalysis (Studies in Surface Science and Catalysis)

Catalysis by Metals and Alloys (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis in Organic Chemistry

Catalysis by Acids and Bases (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis in Industrial Practice

Mass Transfer in Heterogeneous Catalysis

Heterogeneous Catalysis in Organic Chemistry

Catalysts for Upgrading Heavy Petroleum Feeds, Volume 169 (Studies in Surface Science and Catalysis) (Studies in Surface Science and Catalysis)

Past and Present in DeNOx Catalysis: From Molecular Modelling to Chemical Engineering, Volume 171 (Studies in Surface Science and Catalysis) (Studies in Surface Science and Catalysis)

Catalytic Polymerization of Olefins (Studies in Surface Science and Catalysis)

Photochemistry on Solid Surfaces (Studies in Surface Science and Catalysis)

Fluid Catalytic Cracking V (Studies in Surface Science and Catalysis)

Laboratory Studies of Heterogeneous Catalytic Processes (Studies in Surface Science and Catalysis)

Fluid Catalytic Cracking V (Studies in Surface Science and Catalysis)

Preparation of Catalysts VII (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis Research Progress

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis

Surface and nanomolecular catalysis

Organometallic Oxidation Catalysis

Surface and Nanomolecular Catalysis

Catalysis and Automotive Pollution Control (Studies in Surface Science and Catalysis)

Heterogeneous Catalysis Research Progress

Recommend Documents