Large pore molecular sieves

Large pore molecular sieves

CatalysisToday 19 (1994) 107-150 Chapter 5 Catalytic test reactions for probing the pore width of large and super-large pore molecular sieves Jens ...

3MB Sizes 2 Downloads 50 Views

CatalysisToday

19 (1994) 107-150

Chapter 5

Catalytic test reactions for probing the pore width of large and super-large pore molecular sieves Jens Weitkamp, Stefan Ernst Institute of Chemical Technology I, University of Stuttgart, P 7OSSOStuttgart, Germany

1. IntroductIoIl

The use of catalytic test reactions is a well known and versatile tool for characterizing the pore width of molecular sieve catalysts [ l-61. There are two possible objectives for applying such catalytic tests: in the early days, the main incentive was to collect information on the approximate crystallographic pore size (810- or 12-membered ring pore openings) of zeolites with unknown structure. hith the advent of more sophisticated and efficient crystallographic methods (e.g., the combination of model building, simulation of X-ray powder diffraction patterns and Rietveld refinement [ 7,8], modern diffraction techniques including those using synchrotron radiation [ 8- 10 ] and MAS-Nh4R spectroscopy [ 11,12 ] ) , a relatively rapid determination even of complex new structures became feasible. Hence, the initial incentive for the application of catalytic test reactions has shifted. Nowadays, the main purpose is to characterize the efsective pore width of molecular sieves under catalytically relevant conditions, so as to provide guidelines for the selection of a suitable zeolite catalyst for a given shape selective reaction. The principal experimental approach for applying test reactions is to measure the selectivities of a given catalytic reaction over a variety of zeolites with known pore structures. Once the method is calibrated in this way, the selectivities of the same reaction observed on a zeolite with unknown structure give valuable information on its effective pore width or the pore architecture in general. Of course, the probe molecule has to be selected in such a manner that shape selectivity effects can be expected to occur and that they can be unambiguously correlated with the pore size and/or architecture of the zeolite rather than with other factors like, e.g., strength and/or density of acid sites, crystallite size of the molecular sieve etc. Shape selectivity effects can only be expected if the pore size of the *Corresponding

author.

0920-5861/94/$26.00 0 1994 Elsevier Science B.V. All rights reserved SSDIO920-5861(93)E0061-Q

108

J. Weitkzmp, S. Ernst /Catalysis Today 19 (1994) 107-150

molecular sieve catalyst and the sizes of reactant molecules, transition states, intermediates and/or product molecules are similar. As a corollary, it can by no means be expected that a single test reaction or a single probe molecule are suitable for characterizing the whole range of pore sizes and cage diameters (viz. from ca. 0.3 to 3.0 nm, [ 131) currently covered by zeolites and related microporous solids. For many years, shape selective catalysis in zeolites was restricted to the conversion of alkanes or mononuclear aromatics in small (8-membered ring) and medium pore ( 1O-membered ring) molecular sieve catalysts [ 14,15 1. More recently, however, the attractiveness of large pore ( 12-membered ring) zeolites was demonstrated for the shape selective conversion of bulkier molecules, e.g., with the naphthalene [ 16-181 or biphenyl [ 19,201 skeleton. Consequently, there is now an increasing incentive for probing the effective pore width of large pore and even super-large pore molecular sieves by catalytic test reactions as well. The purpose of this paper is a critical evaluation of the catalytic test reactions proposed so far for probing the effective pore width of microporous catalysts. Special emphasis will be payed to their suitability for the characterization of large pore and super-large pore materials and to their sensitivity with respect to subtle differences in the size and shape of the pores and cavities. 2. Fundamentals of shape selective catalysis The phenomenon of shape selective catalysis has been explained in the simplest manner by a combination of the molecular sieve effect and a catalytic reao tion. In many instances, however, this is an oversimplified and unsatisfactory definition. Alternatively, it has been stated that shape selective catalysis may occur in microporous solids, if their pore width and the dimensions of the reactant molecules, transition states, intermediates and/or product molecules, which take part in the catalytic reaction, are similar. The most useful definition is, perhaps, as follows: the term shape selective catalysis encompasses all effects in which the selectivity of a reaction depends, in an unambiguous manner, on the pore width or pore architecture of the microporous solid. The first examples for shape selective catalysis in zeolites were described more than 30 years ago by Weisz and coworkers [ 2 l-23 ] and by Csicsery [ 24-261. All these early findings could be reasonably classified into three types of shape selectivity which have been referred to as (i) re~tunt shape selectivity, (ii) product shape selectivity and (iii) restricted transition state shape selectivity [26-291. Examples for each type are shown in Fig. 1. Reactant shape selectivity, illustrated in Fig. 1 by competitive cracking of noctane and 2,2,4&methylpentane can indeed be understood as molecular sieving combined with catalytic conversion: one type of molecules (2,2,4-trimethylpentane) out of the mixture of reactants is too bulky to enter the pores of the zeolite. These molecules are, therefore, hindered from reaching the catalytically active (acid) sites inside the pores. They can only be converted at catalytic sites

109

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

Reactant C”s C", C",

‘CC‘4‘

2

2

Shape

Selectivity

C".

x-

ic(

2

-C

0”

2 23

2

Product

+

Jt

CH, =

Shape

Selectivity

&H-

C”,

C&- C”, -Ti--r -&---

0”

-

///,,,/,//fl Restricted meta xylene

Transition

State Shape

Selectivity para xylene C”3

C”, b 1: C”3

no toluene and no trimethyl benzene8

Fig. 1. Examples for the three classical types of shape selectivity after Weisz [ 271 and Csicsery [26,28,29].

110

J. Weitkump, S. Ernst /Catalysis Today 19 (1994) 107-150

located on the external surface of the zeolite crystallites (not shown in Fig. 1) or leave the reactor without being converted at all. By contrast the small or slim molecules (n-octane) do have access to the pores and the active sites at the internal zeolite surface, where they are readily converted catalytically. The net effect measurable at the reactor exit is a selective conversion of the small or slim reactant molecules. Product shape selectivity, demonstrated in Fig. 1 by the example of acid catalyzed ethylation of toluene, may be looked upon as the reverse of reactant shape selectivity: the reactants are small enough to enter the zeolite pores, but out of the potential products (ortho-, metu- and puru-ethyltoluene), only one species (the puru-isomer) is small enough to leave the pore system. The bulkier product molecules, although they may have been formed in relatively spacious intracrystalline cages or at channel intersections are unable to escape from the pores and do not occur in the reactor effluent. Ultimately, these entrapped products may be transformed into smaller ones which are able to escape from the pores or into coke deposits. A case of restricted transition state shape selectivity is sketched in the bottom part of Fig. 1. Under the influence of an acid site, metuxylene can undergo isomerization into puru-xylene (and orthexylene, omitted in Fig. 1) and transalkylation into toluene and a trimethylbenzene. It is evident that transalkylation is a bimolecular reaction and as such it necessarily proceeds via bulkier transition states and intermediates than the monomolecular isomerization. In a zeolite with the appropriate pore width, there will be just enough room for the accommodation of the transition states and intermediates for the monomolecular reaction, but no room for the formation of the bulky transition states and intermediates of the bimolecular reaction, the net effect being a complete suppression of the latter reaction. Reactant and product shape selectivity have the same principal origin, viz. hindered diffusion of bulky reactant or product molecules in the pores of the zeolite. The complete exclusion of bulky reactant molecules as well as the complete encapsulation of bulky product molecules are, of course, limiting cases. In many instances, the hindered diffusion just results in a rate of disappearance of a reactant or a rate of formation of a product which are lower than they would be in the absence of mass transfer limitations, but which are not zero. In these instances, the observed rates of reaction limited by diffusional effects, and hence the observed selectivities, will be dependent, under otherwise constant conditions, on the length of the intracrystalline diffision paths, i.e., the crystallite size of the zeolite catalyst. By contrast, restricted transition state shape selectivity is due to intrinsic chemical effects emerging from the limited space around the intracrystalline active sites, hence the resulting selectivities will not depend on the crystallite size. For a given shape selectivity effect, a clear-cut discrimination between the two possible origins (mass transfer effects versus intrinsic chemical effects) may be difficult. One method for distinguishing between both effects consists of a separate determination of the effective diffusion coefficients for all reactants or all products in the zeolite under consideration. Apart from the fact, however, that there is, up till now, no reasonably simple, yet reliable technique for measuring

J. Weithmp, S. Ernst /Catalysis Today 19 (1994) 107-150

111

such diffusion coefficients, this approach is necessarily restricted to subcatalytic temperatures. In another approach, the length of the intracrystalline diffision paths is varied in catalytic experiments by using zeolite catalysts of the same structure, chemical composition and intrinsic activity but significantly (at least one order of magnitude) different crystallite size. A prerequisite for this more elegant technique is, of course, that synthesis procedures for the required zeolite samples are available. By now, such sophisticated syntheses are restricted to a few types of molecular sieves, the most prominent example being zeolite ZSM-5.

3. Requirementsan ideal test reaction should fulfill Ideally, a catalytic test reaction for probing the effective pore width of molecular sieve materials should fulfill a number of requirements: (i) Its selectivity should be dependent in a pronounced and well defined manner on the pore width. The range of pore widths in which such a dependency exists should be as broad as possible. (ii) The selectivity should not be dependent on the crystallite size. In other words, the chosen selectivity effect should be due to restricted transition state shape selectivity rather than to reactant or product shape selectivity. In such cases where either of the two latter mass transfer effects is governing the selectivity, care must be taken to compare zeolites of approximately the same crystallite size only. (iii) The influence of the pore width on the selectivity of the test reaction should be expressed in a quantitative manner by means of a suitably defined selectivitycriterionor index.Examples are the Constraint Index (CI), the Modified or Refined Constraint Index (Cl*) or the Spaciousness Index (SI) which will be discussed in detail in the subsequent chapters. (iv ) It is desirable that the mechanism of the catalytic reaction be understood in considerable detail, at least in the absence of spatial constraints inside the pores, so that the observed shape selectivity effects can be rationalized. (v) Usually, the selectivity of a reaction depends on many variables, such as the conversion, the reaction temperature or the chemical properties of the catalyst. It is particularly useful if the influence of such parameters on the selectivity happens to be small or negligible compared to the influence of the pore width since this makes the practical application of the test reaction much less cumbersome and/or its results more meaningful. (vi) Catalyst deactivation should be absent or very slow because it is generally difficult and time consuming to collect meaningful kinetic data on a deactivating catalyst. (vii) Catalyst preparation and pretreatment, the apparatus for the catalytic experiments and the procedures for product sampling and analysis should be as simple and cheap as possible.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

112

4. Test reactions for monofunctional acidic molecularsieve catalysts Mobil researchers introduced the technique of probing the pore width of molecular sieve materials by a catalytic test reaction. They proposed [ 301 the Constraint Index which is based on the competitive cracking of n-hexane and 3-methylpentane on an acid form of the zeolite catalyst. Later, a variety of other shape selective hydrocarbon reactions were proposed which proceed on monofunctional acidic catalysts as well. The vast majority of these reactions are isomerizations and/or transalkylations of alkyl aromatics, especially of m-xylene [ 1,3 l341, ethylbenzene [ 35,361 and 1-ethyl-2-methylbenzene [ 371. In the following discussion, emphasis will be placed on the cracking of n-hexane/3_methylpentane mixtures and the conversion of m-xylene because a considerable body of experimental data covering a wide range of molecular sieves and pore sizes is available for these test reactions. 4.1. Competitivecrackingof n-hexane and 3-methylpentane(the Constraint Index, CI) It has been known for a long time that, on acidic catalysts, branched alkanes are cracked at a higher rate than their unbranched isomers. Mobil researchers found that this salient feature of catalytic cracking is only given as long as the pores of the catalyst are sufficiently large. They discovered that in medium pore zeolites, such as HZSM-5, the opposite holds for a pair of paraffinic reactants like n-hexane and 3-methylpentane. Based on this shape selectivity effect, the Constraint Index was defined at Mobil as the ratio of first order rate constants: CI= k_ r-J&-M-Pn. The Constraint Index has been used routinely in Mobil’s patents for about two decades. In the scientific literature, precise experimental conditions for its determination were given by Frilette et al. [ 301: the gaseous feed mixture consisting of 10 mole-% n-hexane, 10 mole-% 3-methylpentane and 80 mole-% helium is passed continuously through the fixed bed reactor at atmospheric pressure. The mass of catalyst is 1 g or less and the liquid hourly space velocity (LHSV) between 0.1 and 1 h- ‘. A reaction temperature between 290 and 5 10°C may be chosen such as to achieve an appropriate overall conversion between 10 and 60%. After a time on stream of 20 minutes the reactor effluent is analyzed, preferentially by gas chromatography. This gives the conversions of the two feed hydrocarbons, Xn_nxand X3_M_Pn from which the Constraint Index can be calculated using the performance equation for integral fixed bed reactors [ 381 with the boundary condition pn_Hx =p3_M_p,, at the reactor inlet:

CIA== k 3-M-Pn

log( 1- xl-l-xx1 h(

1 - X3-M-Pn

1’

Results from cracking of equimolar mixtures of n-hexane and 3-methylpentane over an extended time-on-stream on three acidic zeolites (LaNaY-72 is a Y-type

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

113

zeolite in which 72% of the Na+ cations were exchanged by La’+) are shown in Fig. 2 [ 391. A comparison of the data for LaNaY-72 and HZSM-5 clearly shows the basic shape selectivity effect, i.e., the reversed order of the relative rates of cracking in the large pore zeolite Y and the medium pore zeolite ZSM-5. Mordenite has an intermediate position with both cracking rates being approximately equal. It is also seen from Fig. 2 that HZSM-5 is virtually free from deactivation, whereas the 12-membered ring zeolites deactivate considerably while on-stream. Mordenite is an example for a zeolite which, under the conditions applied in this experiment, deactivates so rapidly that a reliable determination of the Constraint Index becomes difficult. It is evident that the numerical value of the Constriant Index increases as the pore width of the zeolite decreases. According to Mobil [ 301, the Constraint Index allows a classification into large pore ( 1Zmembered ring), medium pore ( lo-membered ring) and small pore (8-membered ring) molecular sieve materials: CI< 1: large pore materials; 1 < CI< 12: medium pore materials; 12 < CI: small pore materials. It might be tempting to interpret the underlying catalytic effect in terms of reactant shape selectivity, assuming that the diffision of the bulkier 3-methylpentane inside the pores is progressively hindered, as the pores are getting narrower. However, in a most impressive study with HZSM-5 samples of equal concentration of acidic sites but different crystal sizes (0.05 to 2.7 w) Haag et al. [ 401 convincingly demonstrated that neither the measurable rate of cracking of n-hexane nor that of the bulkier 3-methylpentane depend on the length of the intracrystalline diffusion paths. From this finding, the selectivity effects encoun100

E ;

I

L+! 80

I



-

LaNaY-72

.--

HMordenite

I

:

x‘

0

0

10

20 TIME

0 ON STREAM,

10

20

h

Fig. 2. Catalytic cracking of an equimolar mixture of n-hexane and 3-methylpentane on three zeolites in their acidic forms at T= 300°C.

114

J. Weithmp, S. Ernst /Catalysis Today 19 (1994) 107-150

tered in the competitive cracking of n-hexane and 3-methylpentane can be unambiguously attributed to intrinsic chemical effects, i.e., restricted transition state shape selectivity. Haag et al. [ 401 suggested that the rate controlling step in the complex chain type mechanism of acid catalyzed alkane cracking via carbocations is the chain propagating hydride transfer between a cracked carbenium ion (e.g., a secondary propyl cation) and a feed molecule, and this hydride transfer step necessarily proceeds via a bimolecular transition complex. Obviously, the bulkier the feed alkane, the more space is required for this transition complex, and in a host with a sufficiently narrow pore system, such as zeolite ZSM-5, this results in a significant hindrance of cracking of 3-methylpentane. The principle findings and conclusions of Haag et al. were repeatedly confirmed by other groups [ 41,421. By extending the cracking experiments to HZSM-5 samples with considerably larger crystallite sizes, Voogd and van Bekkum were able to show that cracking of n-hexane and 3-methylpentane in HZSM-5 is free from mass transfer limitations as long as the crystal size does not exceed 40 pm [ 421. Using the conversion data in Fig. 2, Constraint Indices of 4.6, 1.O and 0.2 are calculated for HZSM-5, H-mordenite and LaNaY-72, respectively. Taking into account the crystallographic channel dimensions (0.5 1 nm to 0.56 nm for ZSM5, 0.65 nm to 0.70 nm for mordenite and 0.74 nm for faujasite, according to [ 13 ] ) , these Constraint Index values rank the pore widths of these three zeolites correctly. Constraint Indices taken from the patent literature for a broader variety of zeolites are listed in Table 1. By and large, a correct classification of eight-, ten- and twelve-membered ring materials can be achieved with the Constraint Index values. An exception is zeoTable 1 Contraint Indices (CZ) for selected zeolites. Data from ref. [ 431; the values in parentheses are from lur laboratory [ 5,441 ZEOLITE

Cl

STRUCTURE TYPE

CLASSIFICATION AFTER Cl

RING SIZE

Erionite ERI 38 narrow pores 8 - ring .____~~~_____~_~~______~~___~~_~~~~~___~~~~~~~___~~~~~~~_~~~~. ZSM-23 MTT 9.1 10 - ring ZSM-22

( 7.4 1

TON

7.3

6 - 8.3

( 4.6 1

ZSM-5

MFI

ZSM-11

MEL

5 - 8.7

ZSM-50

EUO

2.1

MTW ZSM-12 3.0 ______________________--________--_____-___-. Mordenite MOR 0.5 ( 1.0 1 Beta

10 - ring

“BEA

0.6 - 2.0

ZSM-20

FAU I EMT

0.5

X or Y

FAU

10 - ring medium

pores

10 - ring

10 - ring ,________________ II 12 - ring 12 - ring 12 - ring large pores

0.4

1 0.2 1

12 - ring 12 - ring

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

115

lite ZSM-12 which has 12-membered ring channels but, on the basis of its Constraint Index, would be classified as a material with lo-membered ring pores. The major shortcoming of the Constraint Index is its pronounced temperature dependence. For example, Frilette et al. demonstrated [ 301 that for one and the same HZSM-5 sample, the Constraint Index decreased from 11 to 1.5 as the temperature of the catalytic cracking experiment was raised from 290’ C to 5 10’ C. An analogous temperature dependence of the Constraint Index has been reported for zeolite HZSM-22 [ 5,441. Haag et al. [ 45,461 looked in much detail into the reasons for this pronounced temperature sensitivity of Constraint Index and, hence, of the selectivity of acid catalyzed cracking of alkanes. They invoked two basically different cracking mechanisms, viz. the classical chain type mechanism, involving tri-coordinated alkyl carbenium ions and a non-classical mechanism in which an alkane molecule reacts directly with the proton of a Brransted acid site to give a penta-coordinated alkyl carbonium ion which then decomposes into moieties which are entirely different from the cracked products typically formed through the classical chain type reaction. A pronounced influence of the pore width on the relative rates of cracking of n-hexane and 3-methylpentane is only given, if the classical mechanism is operative. Its contribution, however, diminishes as the reaction temperature is increased (i.e., the cracking via the non-classical mechanism has a higher activation energy), hence the gradual loss of shape selectivity and decrease of the Constraint Index with increasing temperature. The strong temperature dependence of Constraint Index renders it difficult to compare microporous catalysts which differ significantly in their concentrations and/or strengths of acid sites: to compensate for the large activity differences such catalyst may have, one would normally vary the reaction temperature, but this measure is forbidden if temperature itself has such a decisive influence on the selectivity of the test reaction. For large pore zeolites, the Constraint Index suffers from additional disadvantages: according to Table 1, the numerical range of the Constraint Index (ca. 0.4 to 1) for such materials is so small that virtually no meaningful ranking of the effective pore widths of large or super-large pore molecular sieves can be accomplished on the basis of the Constraint Index. Another inconveniency may be encountered in its experimental determination: cracking of alkanes in large pore zeolites is often accompanied by severe coking and deactivation (cf. right-hand side of Fig. 2, especially the data for H-mordenite). Coke depositing inside the channel system may be expected to modify the effective pore width, and this is, of course, not tolerable for a method meant as a tool for measuring this pore width. It has been demonstrated [ 461 for hexane cracking in ultrastable zeolite Y at high temperature (538°C) that deactivation can be avoided if the partial pressure of the hydrocarbon is sufficiently low (ca. 1 kPa), but it is not yet clear whether the same holds for aluminum rich forms of large pore zeolites. For all these reasons, acid catalyzed cracking of n-hexane/3_methylpentane is of limited value as a test reaction for ranking the effective pore width of large pore molecular sieves. The Constraint Index was, however, the first example for a quantitative criterion derived from catalytic data and meant to rank micropo-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

116

rous materials according to their pore widths. The work by Haag et al. no doubt fostered the search for alternative and improved catalytic test reactions, the most powerful and promising of which will be discussed in the subsequent chapters. 4.2. Isomerization and disproportionation of meta-xylene The isomerization of xylenes on acid zeolites, in particular on zeolite HZSM5, is of considerable industrial importance. In the industrial processes, the objective is to maximize the selectivity for para-xylene which is required for the manufacture of terephthalic acid, a starting material for the production of plastics 1141. Besides, xylene conversion has been widely used as a test reaction for probing the effective pore width of microporous materials. For this purpose, the preferred feed isomer is meta-xylene. On acidic catalysts, m-xylene can undergo two principle reactions, viz. isomerization into ortho- andpara-xylene and disproportionation (or transalkylation) into trimethylbenzenes and toluene, as shown in Fig. 3. While it is obvious that disproportionation is necessarily a bimolecular reaG tion, there is some ambiguity as to whether the acid catalyzed isomerization of xylenes proceeds via a monomolecular or a bimolecular pathway. Csicsery [ 47 ] was probably the first author who pointed out that these two mechanisms may be operative in the isomerization of dialkylbenzenes. He moreover found that the monomolecular mechanism, i.e., an intramolecular 1,Zshift of an alkyl group, is favored at higher temperatures, especially above 300°C. Converseley, intermolecular isomerization via transalkylated intermediates predominates at temperatures below 200°C. Csicsery was led to such conclusions from experiments with 1-ethyl-2-methylbenzene as a reactant and amorphous silica-alumina and acid forms of zeolite Y as catalysts. Recently (and, surprisingly, without quoting Csic-

-

lsomerization [email protected] b

C’-‘3

C’43

=‘I

e

:’

C’43

or b:’

b3 I

CH3

CH3

-

Disproportionation C’43

2

b3

‘J+3

C”3

Fig. 3. Principal reactions of meta-xylene over acidic catalysts.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

117

sexy’s relevant work published more than 20 years earlier), Corma and Sastre [ 48 ] presented evidence for a contribution of the bimolecular pathway in the isomerization of para-xylene on zeolite HY as well. They furthermore concluded that the relative contributions of the monomolecular and the bimolecular isomerization mechanisms depend not only on the reaction temperature, but also on the Si/Al ratio of the faujasite. It was another most interesting finding in this investigation that in zeolites with slightly narrower pores, such as mordenite or beta, the contribution of the spatially more demanding bimolecular mechanism is significantly lower or zero [ 48 1. Gnep et al. [ 3 1 ] suggested for the first time to use me&-xylene conversion as a test reaction for characterizing the pore width of microporous catalysts. This suggestion was based on their results obtained with acid forms of the zeolites ZSM-5, mordenite and Gnep et al. identified three selectivity criteria which may furnish valuable information on the effective pore width: (i) the relative rates of formation of ortho- and para-xylene, (ii) the ratio of rates of disproportionation and isomerization and, (iii) the distribution of the trimethylbenzene isomers (cf. Fig. 3) formed in the disproportionation reaction. Criterion (i ) is based on the finding [ 3 1 ] that on zeolite HY where shape selectivity effects are supposed to be absent, orfho- and paru-xylene are formed at virtually the same rate (at 350°C; unfortunately, the conversion is not given in ref. [ 3 1] ). In H-mordenite, para-xylene is formed 1.5 times as fast as o&o-xylene, and this preferred formation of puru-xylene is still more pronounced in HZSM-5 (ratio of rates of formation: 2.4). In other words, the narrower the pores, the more pronounced is the preferred formation of the puru-isomer. This effect is best interpreted in terms of product shape selectivity, i.e., progressively hindered diffusion of the bulkier ortho-xylene molecules, as the zeolite pores are getting narrower. Criterion (ii) by Gnep et al. [ 3 1 ] is a quantitative expression of the observation that with decreasing pore width, isomerization of me&z-xylene is more and more favored over its disproportionation. At 350°C (again no conversions are given in ref. [ 3 1 ] ) , the reported ratios of rates of disproportionation and isomerization are 0.75,0.06 and 0 for HY, H-mordenite and HZSM-5, respectively. In other words, the bimolecular disproportionation is completely suppressed in the medium-pore zeolite ZSM-5 whereas isomerization (which is likely to proceed via the monomolecular mechanism at these high temperatures) does not experience significant hindrance. Similar results were published slightly later by Olson and Haag [ 49 ] who studied the conversion of pure me&-xylene and a mixture of ortho- and me&-xylene in acidic forms of four zeolites, viz. faujasite, mordenite, ZSM-4 (mazzite topology) and ZSM-5. These authors demonstrated that a linear relationship results if the ratios of rate constants of disproportionation and isomerization are plotted versus the dimensions of the largest zeolite cavities. One of the conclusions of this study was that the suppression of disproportionation is due to restricted transition state selectivity rather than to mass transfer effects. Finally, criterion (iii) was deduced from the observation [ 3 1 ] that the disproportionation of m-xylene in the spacious cages of zeolite Y and in the much nar-

118

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

rower channels of mordenite resulted in significantly different distributions of the trimethylbenzene isomers: in the product obtained on H-mordenite, about 95% of the trimethylbenzene consisted of the 1,2,4-isomer as compared to 24% of the 1,3,5-isomer beside 74% of the 1,2,4-isomer in the product from HY (on both zeolites, only negligible amounts of 1,2,3&imethylbenzene were formed, and this was attributed to its low concentration in thermodynamic equilibrium). To interpret the hindered formation of 1,3,5trimethylbenzene from m-xylene in mordenite, the authors mainly invoked restricted transition state shape selectivity. meta-Xylene isomerization as a catalytic test reaction was later adopted by a number of other groups and the criteria proposed by Gnep et al. were refined and applied to a broader structural variety of molecular sieves. Dewing [ 1 ] introduced the ratio of rate constants observed experimentally for the formation of ortho- and para-xylene from me&-xylene under conditions of diffusional limitations. By applying Thiele’s concept for catalysis and mass transport in porous solids, Dewing deduced that this ratio, referred to as the R value, is only dependent on the ratio of diffusion coefficients of ortho- and puru-xylene inside the pores. Later, Joensen et al. [ 341 coined the term Shape Selectivity Index (SSI). It is defined as the molar product ratio of puru- and o&o-xylene observed on a shape selective zeolite and extrapolated to zero conversion minus the same ratio obtained under identical experimental conditions on a catalyst with sufficiently large pores, i.e. under the complete absence of shape selectivity in me&z-xylene isomerization. Neither Dewing’s R value nor the Shape Selectivity Index proposed by Joensen et al. have received much attention and credit by other groups. Since they are of limited value for characterizing the pore width of large pore molecular sieves, they will not be considered any further in this chapter. Martens et al. [ 321 undertook a thorough investigation with the aim to explore the full potential of me&z-xyleneconversion as a test reaction for probing the void structure of crystalline microporous materials. Acid forms of about ten different aluminosilicate catalysts with both ten- and twelve-membered ring pores were employed, and among these were ZSM-5,ZSM-48,ZSM-12, offretite, mordenite, omega, L, beta and faujasite. Extensive use of this same test reaction was made by Ratnasamy, Kumar and their coworkers [ 50-541 for characterizing the materials synthesized in their group. Pertinent results published by these two latter groups will be reviewed in the subsequent paragraphs to illustrate the potential and limitations of the me&z-xylenetest. Fig. 4 shows selectivity ratios of paw- and or&xylene in the conversion of meta-xylene. The data were collected at temperatures around 3 50 ’ C,and the residence time was chosen in such a manner that a [email protected] conversion of 5 to 10% was achieved. The selectivity ratios are plotted against the crystallographic pore or window diameter as compiled in ref. [ 131, For structures with elliptical ten- and twelve-membered ring openings, the geometric mean between the smallest and the largest dimension of the pore opening was used. It is evident from Fig. 4 that Spxylcne /So-xylene is considerably higher for the zeolites with ten-membered ring pores as compared to the zeolites with twelve-membered ring pores or windows. This is an expected result if one interprets the sur-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150 4

V”, “, ZSM-23 ;

I

,

I

,

119

I

I I I I

3ZSM-5’

i I

cn” 2. t

ZSM-48’

i I I

ZSM-500

E

I I I

OFF’

;

I MORob PZSM-12

1 -

II

PAZ (4 Pm) -MM

\:.3yrni

I I

lo-MR 0

-0.4

i I

0.5

12-MR

I

,,I,

0.8

I

I

0.7

WINDOW SIZE,

I 0.8

I

0.9

nm

Fig. 4. Sekctivity ratio ofpara- and ortho-xylene in the conversion of meta-xylene on acidic zeolites. Data for ZSM-23 and ZSMdO from ref. [SO], all other data from ref. [ 321.

plus of para-xylene in terms of product shape selectivity, i.e., mass transfer limitations, in agreement with a more recent investigation by Richter et al. [ 33 1. Even ZSM-50 with the topology of EU- 1 [ 131, i.e., ten-membered ring channels with very spacious side pockets [ 551, is correctly identified as a medium pore zeolite. As a whole, the pura/ortho selectivity in me&z-xylene isomerization allows a safe discrimination between the groups of ten- and twelve-membered ring molecular sieves. At the same time, Fig. 4 reveals that the criterion is less useful for ranking the effective pore width within the group of ten-membered ring zeolites or within the group of twelve-membered ring materials. A more detailed for the medium pore zeolites (left-hand part of Fig. discussion of~PXYlcnel~+xy~ene 4) is beyond the scope of this article. The data in the right-hand part of Fig. 4 /So-XYlcne for all twelve-membered ring zeolites scatter clearly show that Spxylene between 1 and 1.5 and do by no means allow a ranking of the window or pore size. Likewise, a comparison of the rates of isomerization and disproportionation, if applied quantitatively to a larger number of zeolites, does not allow a reliable ranking of large pore molecular sieves. This is evident from the right-hand part of Fig. 5 in which the ratio of rates of isomerization and disproportionation of me&-xylene are plotted versus the crystallographic window or pore size. To account for the deactivation of the catalysts due to coking and their different decay rates, the rates of isomerization and disproportionation were measured in depen-

120

J. Weitkamp, S. Ernst/Catalysis

I

40

,

Today 19 (1994) 107-150

I. ,

8

I

,

,

,

,

: ,

ZSM-23

- W) ZSM-5’ 30 -

10-w

d

I

12-MR

I

L .

j I I I I

I

20-

I I

P L

I

ZSM-48’ 10

’ I

-

ZSM-ifi

0 0.4



I 0.5

!

OFF0

OMM

(0.3 p-d

OMAZ

&M-l 2 I MOFio I-; ‘-I ’ PI 0.6 0.7

(4Pm) I 0.8

-

‘. 0.9

WINDOW SIZE, nm Fig. 5. Ratio of rates (extrapolated to zero time-on-stream) of isomerization and disproportionation of meta-xylene on acidic zeolites. Data for ZSM-23 and ZSM-50 from ref. [ 501, all other data from ref. [32].

dence of time-on-stream and extrapolated to zero time-on-stream [ 321. The ratio of initial rates obtained in this way is plotted on the ordinate of Fig. 5. No correlation at all exists between rrm./rnis. and the window or pore size. According to Martens et al. [ 321 there seems to be, however, valuable information on the pore architecture in the distribution of the trimethylbenzenes formed from metu-xylene in large pore zeolites. The reaction path of alkylbenzene transalkylation in sufficiently spacious pores is nowadays envisaged [ 56581 to proceed via carbocations with a diphenylmethane backbone (an exception being, perhaps, the transfer of alkyl groups with two or more carbon atoms at sufficiently high temperatures in medium pore zeolites which has been claimed to proceed via dealkylation and re-alkylation of another aromatic ring by the free alkene [ 591). Starting from metu-xylene, there are three possible diphenylmethane-type carbocations, and these intermediates of the disproportionation reaction are visualized in Fig. 6. It is clearly seen from Fig. 6 that each intermediate gives a particular trimethylbenzene isomer. If one assumes that the relative coverage of the acid sites with the three diphenylmethane-type cations depends on the size and geometry of the voids around these sites, then one expects the measurable distribution of the trimethylbenzenes to depend on the pore architecture as well. Martens et al. [ 32 ] carefully examined the distribution of these isomers as they

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

121

Fig. 6. Diphenylmethane-type carbocations as intermediates in the acid catalyzed disproportionation of meta-xylene.

were formed on a variety of zeolites at low conversion of meta-xylene. Pertinent results are listed in Table 2. Following, in part, the interpretation of Martens et al., one can conclude from these data: if disproportionation occurs at all in tenmembered ring zeolites, it leads to 1,2,4_trimethylbenzene exclusively, probably because the diphenylmethane-type carbocation shown on the bottom of Fig. 6 can be best accommodated in their narrow pores. An exception is zeolite ZSM50 (topology of EU-1) in which 1,2,3- and 1,3,5trimethylbenzene are formed as well. This can be rationalized if one assumes that the disproportionation takes place, at least in part, in the spacious side pockets running perpendicular to the ten-membered ring channels. For the zeolites with the largest pores or cages, the distribution of the trimethylbenzenes is close to equilibrium, hence mechanistic, i.e., kinetic conclusions from these data should be drawn with great care. For the rest of the twelve-membered ring zeolites, i.e., offretite, mordenite, omega and beta, the observed distributions of the trimethylbenzene isomers could indeed contain valuable information on the pore size and even on their shape, as claimed by Martens et al. [ 32 1. They went so far to suggest that adjacent cages or lobes, as they exist, e.g., in the pores of zeolite beta, favor the formation of 1,3,5-trimethylbenzene, whereas straight channels and side pockets at regular distances, as they exist, e.g., in the pores of mordenite, offretite or zeolite mega, are favorable for the formation of 1,2,3_trimethylbenzene. If this latter interpretation is correct, the disproportionation of metu-xylene is a remarkably powerful test reaction for probing the void geometry of twelve-membered ring molecular sieves up to a pore width of roughly 0.70 nm. The effect of the nsi/nA ratio on the catalytic performance in the meta-xylene conversion has been investigated for zeolites Y and beta by Corma et al. [ 601 and Perez-Pariente [ 6 1 ] et al., respectively. At low me&-xylene conversion, the faujasites described in ref. [ 601 with an nsi/nA ratio between 8 and 66 all gave a

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

122

Table 2 Distribution of trimethylbenzene isomers formed by disproportionation of meta-xylene at low conversion (X,,,,, = 5 to 10%). Data for zeolites ZSM-23 and ZSM-50 from ref. [SO], all other data from ref. [ 321 xi,2,3

Zeolite

x1.2.3

'1.2,4

X1,3.5

ZSM-23

0

100

0

ZSM-5

0

100

0

ZSM-48

0

100

0

ZSM-50

7.7

ZSM-12

0

Offretite

4.7

Mordenite Omega

84.6 100

7.7

/

X1.3.5

1.0

0

-

95.3

0

00

9.1

71.7

19.2

0.47

9.0

77.4

13.6

0.66

Beta

4.2

73.2

22.6

0.19

L

6.3

67.5

26.2

0.24

Y

7.9

64.0

28.1

0.28

8.0

68.0

24.0

0.33

Thermodyn. equil.

(623

K)

distribution of the trimethylbenzene isomers which was close to the one indicated in Table 2 for the non-dealuminated zeolite Y. Since this is virtually the equilibrium distribution, it remains, in our opinion, doubtful whether any information concerning the space inside the pores can be deduced from these data. The samples of zeolite beta described in ref. [ 6 1 ] covered an nsi/nA ratio of 11 to 42. The main effect of an increase of this ratio was an increase in the selectivity ratio Sr,./ SDis.from ca. 4 to 12. Following the line of arguing in ref. [ 601, two possible explanations for this observation were put forward in ref. [ 6 11, the more convincing one envisaging a surface concentration of the reactant on the zeolite and a decrease of this concentration effect upon dealumination, with a concomitant reduction of the likelihood for the bimolecular disproportionation reaction. This pronounced influence of the nsi/nAl ratio on the selectivity ratio &./&is. for one and the same topology illustrates again how questionable the criterion really is for the purpose of characterizing the pore width. There was almost no influence of the nsi/nd ratio on the distribution of the trimethylbenzene isomers formed by me&-xylene disproportionation (except for a slight increase in the content of both the 1,2,3- and the 1,3,5-isomer with increasing nsi/n,).

J. Weitkamp, S. Ernst1 Catalysis Today19 (1994)107-150

123

The main conclusions concerning the suitability of meta-xylene conversion for probing the pore width of microporous materials are: this test reaction has found widespread application, probably because it is easy to handle, e.g., the number of products which have to be considered is small. Criteria (i) and (ii) identified in the original work by Gnep et al. [ 3 11, i.e., the relative rates of formation of orthe and para-xylene and the relative rates of isomerization versus disproportionation may be useful for a coarse discrimination of ten- and twelve-membered ring molecular sieves. Their applicability and usefulness for a more subtle differentiation of the pore widths within the group of large pore zeolites is, however, doubtful. It is only criterion (iii) by Gnep et al. [ 3 11, i.e., the isomer distribution of the trimethylbenzenes, which seems to furnish meaningful information on the pore width of such materials. However, care has to be taken in order not to overinterpret these data, partly because this distribution tends to be in the vicinity of equilibrium whenever meta-xylene disproportionation is performed in a twelvemembered ring molecular sieve. 4.3. Isomerization and transalkylation of I-ethyl-2-methylbenzene As early as 1970, Csicsery [ 241 recognized pronounced shape selectivity effects during the reaction of 1-ethyl-2-methylbenzene in H-mordenite. One of the most striking features on this catalyst was that, in the products of the transalkylation reaction, the 1,3,5-trisubstituted isomers, viz. 1-ethyL3,5dimethylbenzene and 1,3diethyl-5_methylbenzene, were practically absent. By contrast, when the same reactant was converted on amorphous Si02-A1203 or an acid form of faujasite, these same 1,3,5-isomers were among the predominant components in the trialkylbenzene product fraction. Later, Csicsery ascribed the lack of formation of the symmetrical isomers in mordenite to restricted transition state shape selectivity [ 25,261, rather than to product shape selectivity. In the late 60s and early 7Os, Csicsery had at his disposal only a very limited number of zeolite structures. More recently [ 371, he extended the 1-ethyl-2methylbenzene test to a broader variety of zeolites such as ZSM-5, offretite and beta. As discussed most comprehensively by Csicsery in ref. [ 371, a wealth of information on the pore size of the catalyst can be deduced from the product distributions. Of course, all three criteria advanced for meta-xylene conversion (cf. section 4.2) are applicable to Csicsery’s test as well. The latter furnishes, however, additional information, inter alia because two types of alkyl groups of different bulkiness can be transferred. Minor disadvantages are the higher cost of the reactant and the somewhat more cumbersome product analysis. Of particular importance in the context of the present chapter is the fact that I-ethyl-2-methylbenzene is a bulkier molecule than meta-xylene and so are the intermediates, transition states and products involved in Csicsery’s test. As a whole, therefore, it might be much more appropriate for probing the pores of twelve-membered ring molecular sieves than the meta-xylene test. We recommend that more groups adopt Csicsery’s test so that its full potential and its true limitations for probing the pore width of large pore zeolites become clearer.

124

J. Weitkump, S. Ernst /Catalysis Today 19 (1994) 107-150

As demonstrated recently by Nishi and Moffat [ 621, this catalytic test is similarly attractive for other types of microporous materials, such as heteropoly oxometallates. A variant of the 1-ethyl-2-methylbenzene test has been proposed to characterize bifimctional catalysts carrying a noble metal on an acidic carrier [ 63,641. This test, in which the carrier gas is hydrogen, is not primarily directed towards collecting information on the pore width or geometry. 4.4. Disproportionation of ethylbenzene The disproportionation of ethylbenzene into benzene and the isomeric diethylbenzenes was originally proposed by Karge et al. as a test reaction to collect rapid information on the number of strong Bronsted acid sites in mordenites [ 65 ] and faujasites [ 661. The authors discovered that, with these large pore zeolites, the reaction exhibits an induction period during which the ethylbenzene conversion increases monotonously. It is only after a certain time-on-stream that the conversion reaches a plateau or declines very slowly. Karge et al. were able to show that a linear correlation exists between the rate of ethylbenzene conversion during this stationary or quasi-stationary stage and the number of strong Bronsted sites. It was a straightforward idea to explore whether the reaction is suitable as well for probing the pore width: the dimensions of the diethylbenzene isomers formed and, possibly, of the diphenylmethane-type cations leading to them, are different. From comparative catalytic experiments with a variety of ten- and twelve-membered ring zeolites [ 36,671, the following criteria were established: (i) Twelvemembered ring zeolites exhibit an induction period whilst ten-membered ring zeolites do not. (ii) After the induction period, there is very little or no deactivation in twelve-membered ring zeolites, whereas in all ten-membered ring zeolites, there is considerable deactivation from the very beginning of the catalytic experiment. (iii) On twelve-membered ring zeolites, the distribution of the diethylbenzenes is ca. 5 mol-% ortho-, 62 mol-% meta- and 33 mol-96 para-isomer; on ten-membered ring zeolites, the ortho-isomer is often completely absent or it appears as a very minor component ( 1 to 2 mol-% of the diethylbenzenes), especially at the onset of the catalytic experiment. The combined application of these criteria turned out to allow a safe discrimination between ten- and twelve-membered ring zeolites as demonstrated, for example, in refs. [ 681 and [69] for zeolites EU-1 and ZSM-57, respectively. So far, however, the method has not been relined to such an extent that it allows the twelve-membered ring zeolites to be ranked according to their effective pore widths. Ethylbenzene disproportionation will, therefore, not be discussed in more detail in this chapter. 4.5. Isomerization of meta-diisopropylbenzene and its alkylation withpropene

Very recently, Kim et al. [ 701 proposed a test reaction which seems to be particularly designed for probing the pore width of large and, perhaps, super-large pore molecular sieves. They chose a bulky reactant, viz. meta-diisopropylben-

J. Weitkump, S. Ernst /Catalysis Today 19 (1994) 107-150

125

zene, which has no access to ten-membered ring pores. From the fact that the authors did not report results on zeolite ZSM-12, we conclude that the reactant hydrocarbon may experience significant diffisional hindrance in strongly puckered twelve-membered ring pores as well. Rim et al. converted meta-diisopropylbenzene at 190” C with a sevenfold molar excess of propene on acidic forms of various molecular sieves. Under these conditions, isomerization and alkylation take place, the products being para-diisopropylbenze as well as 1,2,4- and 1,3,5triisopropylbenzene, respectively (no or&diisopropylbenzene and only traces of 1,2,3&isopropylbenzene are formed). Besides, there is some undesired cracking. It is the rationale in the design of this catalytic test that 1,3,5-triisopropylbenzene is a bulkier product than its 1,2,4-isomer. This can be inferred from (i) adsorption experiments (e.g., 1,2,4&isopropylbenzene is readily adsorbed in NaY zeolite at room temperature whereas its 1,3,5-isomer is not; under the same conditions, VPI-5 adsorbs both isomers) and (ii) the general experience acquired in related test reactions, such as the me&-xylene disproportionation (vide supra, paragraph 4.2) or the 1-ethyl-2-methylbenzene transalkylation (vide supra, paragraph 4.3). A logical criterion which can be expected to correlate with the space inside the pores is the selectivity ratio (or yield ratio, the terms yield and selectivity are confused in ref. [ 701) of 1,3,5- and 1,2,4-triisopropylbenzene. The second criterion defined by the authors is the selectivity (or yield) ratio of alkylated and isomerized products. One problem in this test reaction stems from the direct exposure of the catalyst to an alkene under a substantial partial pressure. (It is difficult to understand why such a large excess of propene was employed). It is generally known [ 7 1,72 ] that alkenes are the most potent class of hydrocarbons for coking and a concomitant deactivation of monofunctional acidic catalysts. To cope with this problem, the authors (i) took their product samples after a standardized time-on-stream, viz. 11 to 13 minutes, and (ii) adjusted the flow rate of the feed mixture to the activity of the catalyst under investigation such as to arrive at a conversion of ca. 25% after the time-on-stream chosen for sampling. While this is probably the best experimental strategy applicable in a severely deactivating system, it does not necessarily account for different quantities of coke deposited in various zeolites, and a potential influence of this coke on the effective pore width. In Table 3, the major results of Rim et al. [ 701 are summarized for various twelve-membered ring molecular sieves and compared with the Spaciousness Indices published earlier [ 73,741 for the same materials. The Spaciousness Index is determined by a different test reaction, viz. hydrocracking of butylcyclohexane, on a bifunctional form of the zeolite (vide infra). Roughly, the figures in the three columns parallel each other. Unexpected deviations, such as the surprisingly low value for EMT in column 3 or the unexpectedly high value for zeolite L in column 2 need experimental confirmation. As pointed out by the authors, a potential advantage of the new test is its ability to discriminate between the pores of the most spacious twelve-membered ring zeolite ( faujasite) and amorphous Si02-A1203. From this, they deduce a chance that the new test could turn out to be appropriate for probing super-large pore

126

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

Table 3 Selectivity ratios obtained in the alkylation and isomerization of metadiiso-propylbenzene on acidic forms of various microporous materials after Kim et al. [ 701 and Spaciousness Indices (SZ) for the ame materials after refs. [ 73,741. S 1.3.5-TIP-62

Catalyst

S 1.3.5-

+ 1.2.4-TIP-Br

SI

S 1.2.4-TIP-Bs

S pwa-DIP-52

0.2

0.15

4.0

Offretite

1.1

0.83

5.0

Mordenite

1.5

0.54

7.5 18.0

SAPO-5

Beta

1.8

0.71

EMT

2.3

1.20

FAU

2.5

3.80

21.0

L

2.9

3.02

17.0

SAPO-37

3.0

3.37

-

SiO*-AI,O,

3.5

4.47

-

-

molecular sieves, such as acid forms of VPI-5. On the basis of the data published in ref. [ 701 this is, however, somewhat dubious since so large differences between faujasite and the isostructural SAPO-37 are detected. The authors tentatively attribute these large differences to the residual sodium ions in the acid form of faujasite which was prepared via exchange with ammonium ions (unfortunately, no exchange degrees are given). We find it difficult to conceive such a large reduction of the free space in the faujasite supercages by residual Na+ ions. It is also desirable to find out the nature of the selectivity effects, i.e., product versus restricted transition state shape selectivity, on which the metu-diisopropylbenzene test is based. 4.6. Alkylationof biphenylwithpropene The alkylation of biphenyl with propene has recently been suggested as a test reaction for probing the pore size of pillared clays, large pore zeolites and related microporous materials [ 201. The major selectivity criterion discussed so far is the content of ortho-isopropylbiphenyl in the monoalkylated product fraction. The reaction could have a potential for characterizing large and even super-large pore molecular sieves, but more systematic work on the influence of the pore size and geometry of the catalyst on the distribution of the mono-, di- and, desirably, trialkylated products is needed.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

5. Test reactions for biictional

127

mokcular sieve catalysts

The potential of catalytic reactions occurring on bifunctional forms of zeolites (i.e., a Bronsted acid form modified by small amounts of a hydrogenation/dehydrogenation component, typically platinum or palladium) for probing the pore width and architecture of zeolites has been extensively explored since the early 80s. Particular emphasis was placed on the isomerization and hydrocracking of long chain n-alkanes, such as n-decane [ 75-801, and to hydrocracking of naphthenes, e.g., butylcyclohexane [ 73,741. All these reactions are carried out under hydrogen which is activated by the metallic catalyst component. One very favorable repercussion of the presence of activated hydrogen is that coke formation and the concomitant catalyst deactivation are absent or very slow. This does not only make the experiments much easier (neither the conversion nor the selectivities vary with time-on-stream), but also eliminates the risk of progressively narrowing the pores by the deposition of carbonaceous residues formed in the test reaction. Almost always, the amount of Pt or Pd in a bifunctional zeolite catalyst is between 0.1 and 1 wt.-%. It is often desirable to bring the noble metal as close to the acid sites as possible, i.e., to deposit small metal clusters inside the zeolite pores, in order to fulfill the so-called intimacy requirement defined by Weisz [ 8 11. From time to time, it has been argued that this could bring about an undesired modification of the pore width, in a similar manner as coke deposits can modify the effective pore width during hydrocarbon conversion on monofunctional catalysts. Simple mass balances, however, immediately demonstrate that no such risks exist, at least not for the typical metal loadings indicated above and zeolites with a three-dimensional pore system. For example, in an HY zeolite with 0.5 wt.-% of Pt or 0.27 wt.-% of Pd, one out of ca. 1000 supercages is filled with a metal cluster (assuming that such a cluster consists of 35 metal atoms), all other supercages being completely free and available for molecular diffusion. In agreement with these mass balance considerations, it has been demonstrated experimentally for zeolite EU-1 that the amount of noble metal introduced into the pores does not influence significantly the selectivity of shape selective hydrocracking of butylcyclohexane [ 74 1. In the following paragraph, the present mechanistic views on skeletal isomerization and hydrocracking of alkanes in the absence of shape selectivity effects will be outlined. Based on these concepts, the shape selective test reactions proposed for probing the pore width of bifunctional zeolite catalysts will subsequently be discussed. At this point we briefly address one argument which has been repeatedly raised. The whole mechanistic concept of bifunctional catalysis provides an intermediacy of alkenes and carbocations. However, under almost all conditions, these intermediates cannot be directly detected. It is, perhaps, for this reason that a fundamentally different mechanism has been invoked from time to time: from nothing else than the well known fact that alkanes can be isomerized and hydrocracked on platinum surfaces, i.e., on a monofunctional metallic catalyst, it has

128

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

been extrapolated that such platinum catalyzed hydrocarbon reactions might be operative on bifunctional catalysts as well. We definitely reject such arguments for the following reasons: (i) The selectivities on bifunctional catalysts do not depend on the nature of the hydrogenation/dehydrogenation component. If hydrocarbon conversion occurred on the metal, such a dependency would be expected. (ii) Within certain limits, the rate and selectivity of hydrocarbon conversion is independent of the amount of noble metal on the catalyst. While this can readily be accounted for in terms of the bifunctional mechanism, it is not consistent with hydrocarbon conversion on the metal. (iii) Platinum is the sole metal which is able to isomerize alkanes. Many examples for bifunctional catalysts have by now been described which contain other metals, e.g., palladium, and are excellent isomerization catalysts. (iv) Even on monofunctional platinum catalysts, skeletal isomerization is inevitably (and at all conversions) accompanied by hydrocracking on the metal (often referred to as hydrogenolysis), whereas on many bifunctional catalysts, isomerization is completely free from hydrocracking, at least at low to moderate conversion. (v) Hydrocracking on metals, i.e., hydrogenolysis, gives always significant amounts of methane and ethane. By contrast, these cracked products do not occur at all in bifunctional hydrocracking via carbocations, and this has to do with ,the high energy content of the methyl and ethyl cation. Under certain reaction conditions, especially at excessive metal loadings of the catalyst and at high temperatures, metal catalyzed reactions may, of course, interfere on bifunctional catalysts. By using the above and other criteria, the experienced experimentalist will easily recognize the occurrence of such undesired side reactions and, in almost all cases, find measures to suppress such reactions. 5.1. Mechanisticconceptsfor isomerizationand hydrocrackingof long chain nalkanesin the absence of shape selectivity During the past 30 years, the conversion of long chain n-alkanes on bifunctional catalysts with sufficiently large pores has been investigated in much detail [ 774361. Examples for catalysts, which have been extensively used for such studies, are 0.5 wt.-% Pt/CaY, 0.27 wt.-% Pd/LaY or 1.Owt.-% Pt/ultrastable Y. The reaction steps occurring on such catalysts can be visualized by the now generally accepted network depicted in Fig. 7. Two types of reaction take place, viz. skeletal isomerization into iso-alkanes and hydrocracking into alkanes with 3 to m - 3 carbon atoms, where m is the carbon number of the feed. These two reactions occur in series. The feed n-alkane is first dehydrogenated on the metal whereby the equilibrium mixture of n-alkenes with the same carbon number is formed. These alkenes move to Brransted acid sites where they are protonated to secondary alkylcarbenium ions. These alkylcarbenium ions, while chemisorbed at an acid site, can undergo two principle reactions, viz. skeletal rearrangement into a branched isomer or /&cission into a smaller alkylcarbenium ion and an alkene. fiScission of unbranched carbenium ions, referred to in Fig. 7 as type D /3-scission is so unfa-

129

J. Weitkump, S. Ernst /Catalysis Today 19 (1994) 107-150 -Hz +H’ n - CmH2m+2

N-’ -H” +H2

8 n - GJ%,+l

TYPOD ---#+ IS-scission

type B

8 Cp H2p+1

+

CqH2q

both unbranched

rearr.

i - Gdh+2 monobranched

-Ha +H” 4_ -H” +Hp

Type C

i - C,,,HF,,,+,

R-scissio;;

8 C, H2p+1

+

C,

H2q

both unbranched

monobranched type B rearr.

-Hz +H”

i-

CmH2m+2

dibranched

Type B

_-

i-C.HTm+t m

-Hs +Hp

dibranched

-

R-scission

8 CpHsp+t

+

C,Hsq

one fragment monobranched

typeB

t1

rearr.

-Hz +H” i - CmH2m+2 tribranched

JC [email protected] +H2

0 i - CmHzm+l tribranched

Type A F O-scission

8 CpH2p+1

+

CqH2q

both monobranched

Fig. 7. Reaction network for isomerization and hydrocracking of alkanes on bifimctional catalysts afierrefs. [4] and [82].

vorable that it does not play any significant role in hydrocracking of long chain n-alkanes. For a deeper understanding of the mechanism of hydrocracking, a classification of ionic fiscissions as shown in Fig. 8 has been proposed [ 871. According to this classification which is now widely accepted: - type A j?-scissions start from a tertiary and lead again to a tertiary carbenium ion (which requires three branchings arranged in a, cy, y-positions and a minimum of 8 carbon atoms), - type B jI-scissions start from a secondary and lead to a tertiary carbenium ion or vice versa (which requires two branchings arranged in LX,LX-and a, y-positions, respectively, and a minimum of 7 carbon atoms), - type C pscissions start from a secondary and lead again to a secondary carbenium ion (which requires at least one branching and a minimum of 6 carbon atoms ) whereas - type D pscissions start from a secondary and lead to a primary carbenium ion. The relative rates decrease drastically from type A to type D /3-scissions, and this has its main origin in the enthalpy contents of primary, secondary and tertiary alkylcarbenium ions: a tertiary carbenium ion is by roughly 60 kJ/mol more stable than its secondary isomers and their enthalpies, in turn, are by ca. 100 kJ/ mol lower than the ones of primary carbenium ions with the same carbon skeleton. Much earlier, the skeletal rearrangements of alkylcarbenium ions have been classified as shown in Fig. 9: in type A rearrangements, the number of branchings

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

130

A

rn3 8

rn8 7

5

ma7

b

C

ma

6

D

ma

4

Fig. 8. Classification ofjhcissions of alkylcarbenium ions after ref. [ 871.

TYPE A: The number of branching9 remains constant

e.g.:

h

MECHANISM:

r

vL_w

Via classical hydride and alkyl shifts

TYPE B: The number of branchings increases

or decreases

MECHANISM: Via non-classical protonated cyclopropanes ( PCPs )

Fig. 9. Classification of skeletal rearrangement reactions of alkylcarbenium ions.

remains constant, whereas in type B rearrangements the number of branchings increases or decreases. It is generally accepted that type A rearrangements proceed via a sequence of intramolecular hydride and alkyl shifts, while type B rearrangements proceed via protonated cyclopropanes. Almost always, the rate of type A rearrangements is considerably higher than the one of type B rearrangements. These few principles of carbocation chemistry almost suffice to understand the essential features in the complex mechanism of hydrocracking and isomerization of long chain alkanes. Just one more point has to be introduced, namely the important role of competitive sorption/desorption at the acid sites of the bifunc-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

131

tional catalyst [ 77-79,881: with strong hydrogenation/dehydrogenation components, especially with platinum or palladium, the formation of n-alkenes from the n-alkane is so rapid that the n-C,H2,+2/n-C,H2,+H2 equilibrium is virtually fully established. If, in addition, mass transport limitations for the alkenes between the metal and acid sites are absent (or, expressed in other words, the Weisz intimacy criterion [ 8 1 ] is fulfilled), then the highest possible alkene concentration, viz. near equilibrium concentration, exists in the vicinity of all carbenium ions adsorbed on acid sites. As proposed by Coonradt and Garwood in their most readable and valuable publication [ 881, this brings about a very efftcient and fast desorption of all chemisorbed carbenium ions from the acid sites. The term ideal bifirnctionalcataZysis has been coined [ 891 for this kinetic situation. It is a most important feature of ideal bifunctional catalysis that - at utmost variance to catalytic cracking over monofunctional acidic catalysts - the primary products of carbocation conversion are rapidly desorbed from the acid sites on account of the competitive sorption/desorption mechanism with alkenes. It is only under these conditions that the primary hydrocarbon products can be observed in the reactor emuent, and for these reasons it is highly desirable that ideal bifunctional conditions exist, if the catalytic reaction is to be exploited for probing the pore width. Returning to Fig. 7, it is evident from the above discussion that, at low conversions, the monobranched iso-C,H,,+ 1 cations will be readily desorbed from the acid sites as iso-alkenes, which are hydrogenated at the metal sites to monobranched iso-alkanes and these appear as the sole product. Since the rate controlling step in the overall sequence is the type B rearrangement of alkylcarbenium ions, the measurable distribution of i-alkanes furnishes direct insight into the rearrangement chemistry of long chain alkylcarbenium ions [ 831. In principle, the monobranched alkylcarbenium ions could undergo type C /3+cissions (cf. Figs. 7 and 8), but as a matter of experience, these are too slow so that, upon raising the conversion, the monobranched alkylcarbenium ions prefer to undergo another type B rearrangement. The resulting dibranched alkylcarbenium ions desorb and appear as dibranched iso-alkanes in the product; upon further increasing the conversion, the dibranched cations rearrange once more into tribranched ones. These can undergo the very rapid type A j%scission. Indeed, the mechanism of hydrocracking of long chain n-alkanes in the absence of shape selectivity goes via this route, i.e., severalfold skeletal type B rearrangement followed by type A /3-scission; besides, there is a smaller contribution of type B j?-scission of dibranched carbenium ions. Under ideal hydrocracking conditions, the primary products from the &scission steps are again efficiently desorbed through competitive sorption/desorption with long chain alkenes. One repercussion is a completely symmetrical molar distribution of the hydrocracked products starting with propane and ending with Cm_-3HZm_-4, m being again the carbon number of the feed. An individual analysis of these hydrocracked products can readily be achieved with modem gaschromatographic techniques and has furnished a large body of information on the mechanism of hydrocracking [ 77-80,85,86].

J. Weitkamp, S. Ernst /Catalysis Today 19(1994) 107-150

132

5.2. Isomerizationand hydrocrackingof long chain n-alkanesunder shape selectivityconstraints(theRejned or iUodi$ed ConstraintIndex, CP) It follows immediately from Fig. 7 that hydrocracking of n-alkanes proceeds via highly branched and, hence, relatively bulky aIkylcarbenium ions. Under the steric constraints in, e.g., a ten-membered ring zeolite pore, the tribranched precursor ions of the favorable type A /3-scission cannot form. The system is then forced into alternative routes involving less bulky intermediates, i.e., the narrower the pores, the higher will be the contributions of type B and, eventually, type C or even type D j%scissions. This shift in the hydrocracking mechanism with decreasing pore width brings about a large number of selectivity changes in the hydrocracked product. Once all these shape selectivity effects have been identified for a sufficiently large number of zeolites with knoti pore system, the most convenient ones can be sorted out for the routine characterization of microporous materials with unknown structure. Striking shape selectivity effects do not only occur in hydrocracking, but also in the skeletal isomerization which precedes hydrocracking. For example, if ndecane is converted on a bifunctional catalyst with sufficiently spacious pores, such as Pt/CaY zeolite [ 831, the product mixture obtained at low conversion consists of all possible iso-decanes with one branching. In Fig. 10, these isomers are arranged in such a way that the bulkiness increases from left to right. One would predict that, as the catalyst pores are getting narrower, the bulkier isomers cannot be formed any more inside the pores or cannot escape from there. In agreement with such predictions, isomers with a propyl side chain are lacking in the product formed in catalysts like Pt/H-EU- 1 or Pd/HZSM12 (these experiments were conducted with n-dodecane as reactant [ 681)) and upon further diminishing the pore size, e.g., in Pt/HZSM-5 [ 87,901 as well as in Pt/HZSM-22 and Pt/HZSM-23 [ 9 1 ] neither the propyl nor the ethyl isomers

E

J_ 3 - Ethyloctane

3 - Methylnonane

&&/

4 - Propy’heptane

4 - Ethyloctane

5 - Methylnonane

Fig. 10. Possible iso-decanes with one branching.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

133

form. This shape selectivity effect described here in a qualitative manner has been translated by Martens et al. [ 75,761 for n-decane isomerization into two quantitative criteria, namely Yethyloctanes/ Yiso-dccancs at yk0-d~~~~w 5% and Y4-propylheptane/ YmonObranched iso-decanes at

&-de-e

w

75%

where X and Y denote the conversion and yield, respectively. Another quantitative criterion proposed in [ 751 is the Refined or Modified Constraint Index (Cl*), defined as

in the isomerization of n-decane. The underlying shape selectivity effect was discovered fortuitously [ 921 and described in detail elsewhere [ 87,90,93]. It relates to the rates of formation of the individual methylnonanes (see left-hand part of Fig. 10) from n-decane. As already pointed out, the rate contolling step of this reaction under conditions of ideal bifunctional catalysis is the type B rearrangement of unbranched decyl cations (cf. Fig. 7 ). Modern mechanistic knowledge tells that this branching rearrangement proceeds via protonated cyclopropanes. It has been shown in ref. [ 831 that, if this branching mechanism is combined with a few straightforward statistical assumptions, it is expected to give an isomer distribution of 1:2:2:1 for 2-M-No:3-M-No:4-M-No:5-M-No, as opposed to 2:2:2: 1 in thermodynamic equilibrium. The left-hand part of Fig. 11 confirms that these two distributions Pt I CaY

F’t / HZSM-5

50

I”“““‘1

0

I

0

20

40

60

80

100

CONVERSION Fig. 11. Distribution of the methylnonanes cay’ and Pt/HZSM-5 after ref. [ 871.

0



20

I



40

I

80



I

80



100

OF n - DECANE , %

formed by isomerization

of n-decane in the zeolites Pt/

134

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

are nicely approximated at low conversion, i.e., in the kinetic regime, and at high conversion, respectively. This is, however, only the case if the isomerization of the long chain n-alkane is carried out in the absence of spatial constraints, e.g., in zeolite Y. If, however, the same isomerization reaction is carried out in a tenmembered ring zeolite, such as Pt/HZSM-5, an entirely different isomer distribution is observed at low conversion (see right-hand part of Fig. 11) : clearly, 2methylnonane is now the kinetically preferred isomer. At high conversion, equilibrium between the isomeric methylnonanes is again attained. From the data in Fig. 11, Modified Constraint Indices of ca. 1 and 4.5 are calculated for zeolites Y and ZSM-5, respectively. CI* values taken from the literature are presented in Fig. 12 for a number of additional zeolites. Like the Constraint Index (cf. paragraph 4.1), CI* increases with decreasing pore width. It is evident from Fig. 12 that the CI* values for ten-membered ring zeolites extend over a relatively broad range, viz. from ca. 2.3 to 15, hence this is the range where the Modified Constraint Index is particularly useful. On the other hand, only a very narrow range, namely from ca. 1 to 2.3, is available for twelve-membered ring zeolites. Furthermore, a comparison of the CI* value for ZSM-5 (7.0) with the one derived from the data in Fig. 11 (4.5 ) furnishes an example for the limited accuracy of indices derived from catalytic tests. While the Modified Constraint Index is now widely employed for characterizing ten-membered ring molecular sieves, the precise origin and nature of the shape selectivity effect on which it is based have not yet been fully elucidated. From the finding that the diffusion coefficients of the four isomeric methylnonanes in HZSM-5 and HZSM- 11 at 80 ’ C, i.e., a subcatalytic temperature, were practically identical [901, the strongly preferred formation of 2-methylnonane was attributed to restricted transition state shape selectivity rather than to product shape selectivity. Essentially the same conclusion was drawn by Martens and Jacobs [ 761 form n-decane isomerization studies with Pt/HZSbk5 of different crystal Definition:

YP-mNlyl”onane

cl* 5

( at Y ,somers = 5 % 1 Y5-methylnonane

in isomerization of n-decane

14

12

10

10 - MEMBERED

I ZSM-22

8

RING

Ferrlerlte

4

ZSM-35

2

0

ZEOLITES

ZSM-5

I ZSM-23

6

ZSM-11 I ZSM-4s

Y I ZSM-12

Fig. 12. Refined or Modified Constraint Indices for various zeolites. Data from ref. [ 3 1.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

135

size. Increasing the crystal size from 1 to ca. 15 m brought about only a minor increase in Cp, namely from 7.5 to 9.4, and this was considered by the authors to be essentially consistent with their earlier interpretation [ 901 of the favored formation of 2-methylnonane in ten-membered ring zeolites. Recently, by using molecular graphics as an ‘auxiliary tool for interpreting the selectivities of n-decane isomerization in medium pore zeolites, Martens et al. [ 94,95 ] were led to conclude that the high rate of formation of 2-methylnonane is a result of pore mouth catalysis, at least for zeolite Pt/HZSM-22. If this is correct, the question remains as to why catalysts like Pt/HZSM-5 (cf. Fig. 11) or Pt/HZSM-1 1 give so high selectivities for 2-methylnonane as welk molecular graphics suggested [ 941 that on these catalysts, isomerization of n-decane is likely to occur insidethe pore system. So far, the discussion was restricted to the first reaction of the network shown in Fig. 7, i.e., skeletal rearrangement of the long chain n-alkane into its monobranched isomers. As already mentioned, many additional shape selectivity effects are encountered in the consecutive reactions, i.e., the formation of dibranched isomers and hydrocracking. After a careful inspection of all these effects, Martens et al. [ 75,761 defined live quantitative criteria which are listed in Table 4 along with the three criteria explained in more detail above. An in-depth discussion of all these criteria and their underlying shape selectivity effects is beyond the scope of this review. In the subsequent paragraphs, we will focus on those criteria which, from our point of view, are best suited for the characterization of large pore materials. Criterion 4 in Table 4 makes use of the formation of Cpropylheptane from ndecane. As pointed out by Martens and Jacobs [ 761, this is one of the bulkiest hydrocarbons (perhaps the bulkiest ) in the products. Two different pathways for the formation of the corresponding carbocation (the 4-propylheptyl cation) have been envisaged [ 831, namely (i) a series of classical hydride and alkyl shifts starting from methylnonyl cations formed, in turn, via protonated cyclopropanes from unbranched decyl cations or (ii) its direct formation from n-decyl cations via non-classical protonated cyclopentanes. Martens and Jacobs [ 96,971 recently argued that the second route seems to be more likely. Data for the propylheptane criterion taken from the original publication (ref. [ 761) are depicted in Fig. 13. The following trends emerge: (i) The yield ratio increases with increasing pore width. (ii) On Pt/HZSM-12 with its strongly puckered twelve-membered ring pores, no Cpropylheptane at all is formed. (iii) For faujasites, the yield ratio may vary substantially (2.10 for a bifunctional form of zeolite X the precise nature of which was not disclosed in ref. [ 76 1, and 1.70 for Pt/HY) . If these latter differences are meaningful they suggest that, in a given structure, the selectivity for 4-propylheptane formation is influenced to a considerable extent by the nsi/nA ratio. Indeed, by extending the data to faujasites dealuminated via various techniques to different extents, Martens and Jacobs [ 761 found a linear relationship (see Fig. 14) between the yield ratio (referred to in ) and the faujasite unit cell constant which is known to Fig. 14 as s’ 4-propyhcptane correlate with the framework nsi/nN ratio.

136

J. Weitkamp, S. Ernst/Catalysis Today 19 (1994) 107-150

Table 4 Criteria based on shape selectivity effects observed in the hydroconversion et al. [ 75,761

Y2-msthylnonsne

Cl* ( at V,,,. = 5 % 1

z ‘.

of n-decane after Martens

Y3-methylnc”s”e

Y ethyloctanss ( at *.

=5%)

L.

ytcts, iso-dscane*

Y3-ethyloctane 3.

Y4-sthyloctane

( at Yb.

=5%]

Xn-dscrne=

75

YQ-~rc~ylhsptane 4. Ynwnobr,nch.AIso-drcsner

Ydibranched 5’ ‘t0hI

( at

Isc-decanss

at maximum

yield of iso - decanes

iso-dsosnss

Y2,7-dimethyloctsne

at Ydlbr,isom,= 5 % )

( 6.

‘total

% )

dibrsnched isomers

‘iso-pentsne

formed

at YCrack= 35 % 1

( n-dscsne. hydrocracked

6.

li,-

ilc7

and

y4

hc4-

hcs

I

at

‘Crack

=

35 %

- propylheplane

Y monobranched

0

A

1.0

0.5

I

I

I

A

I

1.5

I

AA

I

ZSM-47

I

A Y L

Offretite

Mordenite

2.0

I

A Beta

ZSM-12

ZSM-25

( at X n _ decane = 75 % 1

isc - decanes

X

Omega

SAPO-5

Fig. 13. Yield ratio of 4-propylheptane and total monobranched iso-decanes at an n-decane conversion of 75% for twelve-membered ring molecular sieves. Data from ref. [ 761.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150 I

2.5

8

I

137

I

2.0 Ls ?z

! F d kl

deep bed calcinedY ukastabilizedY 1.5-

SiCI, - dealuminatedY 1.0

2.40

2.45

2.50

UNIT CELL CONSTANT a, , nm Fig. 14. Yield ratio of 4-propylheptane and total monobranched iso-decanes at an n-decane conver. sion of 75% in dependence of the unit cell constant of faujasites. Data from ref. [ 761.

Such a pronounced influence of the nsi/nN ratio on the yield ratio is a severe disadvantage for a technique meant to probe the effective pore width. It is hard to understand why a faujasite dealuminated via the SiCL method could have an effective pore width which is significantly below the one of zeolite L and SAPO5. Also, there is some confusion in the literature as to the conversion at which the yield ratio should be determined: in the original publication [ 761, an n-decane conversion of 75% was recommended. Later, the same authors [ 98 ] used the criterion at a yield of iso-decanes of 5% (which we expect to be tantamount to a n-decane conversion of 5% because in ideal bifunctional catalysis, isomerization is the sole reaction at low conversion), and so did Olken and Games [ 991 in their attempt to collect information on the pore sytem of dealuminated mordenites. In summary, the 4-propylheptane criterion (Table 4, No. 4) may have a potential for probing the pore width of large pore molecular sieves. However, no safe assessment can be made on the basis of the scarce data published so far. Criterion 3 in Table 4, i.e., the yield ratio of 3- and 4-ethyloctane at a total yield of iso-decanes of 5% has been shown in ref. [ 751 to depend on the pore width of twelve-membered ring zeolites. So far, however, too little has been published on the origin of this shape selectivity effect and the reliability of the criterion. An assessment can, therefore, not be undertaken. Criterion 8 in Table 4 addresses the symmetry of the molar distribution of the hydrocracked products. As pointed out at the end of paragraph 5.1, one salient feature of ideal hydrocracking is a molar product distribution which starts at C3 and ends at Cm_3 and which is completely symmetrical around Cmiz. Such distribution curves for a number of feed n-alkanes including n-decane ( m = 10) have been reported, e.g., in refs. [ 781 and [ 791 for Pt/CaY as catalyst. The symmetry of these molar distribution curves indicates a pure primary cracking, i.e., desorption of both moieties of the first &scission step from the acid site is rapid compared to a consecutive second fiscission.

138

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

In non-ideal hydrocracking and much more so in catalytic cracking over monofunctional acidic catalysts, no such symmetrical product distributions are obtained. Instead, more C3 than Cm_-3,more C4 than Cm_-4,etc. are formed even at moderate or low yield of cracked products, and the molar distribution curves are skewed towards the light hydrocarbons, especially towards C3 and C+ Using a broad variety of zeolites in their acidic forms and loaded with 1 wt.-% of Pt, Martens and Jacobs [ 761 found that the symmetry in the molar product distributions (at moderate yields of hydrocracked products, viz. 35%) is no longer attained, if the zeolite contains unidimensional, non-intersecting pores: hydrocracking of n-decane gave unsymmetrical distributions for Pt/HL, Pt/H-omega, Pt/H-mordenite, Pt/H-offretite, Pt/HZSM- 12 and Pt/HSAPO-5. By contrast, the symmetry in the molar product distributions was given for Pt/HZSM-3, Pt/ H-beta and Pt/H-phi, in addition to Pt/HY. From these data, the authors concluded that (i) the lack of symmetry in these distributions is indicative of a unidimensional pore system and (ii) zeolite phi possesses three-dimensional intersecting pores. With n-decane as feed hydrocarbon, the deviations from symmetry can be expressed quantitatively by ( ricJ- AC,)/ACIO,cracLed or (tic.,- &,) /tic,,, cracled,where jr denotes the molar flux in the continuously operated reactor, and these data are to be determined at a yield of cracked products of 35% [ 98,991. (In the original publication [ 761, Martens and Jacobs proposed ( ric, - ric,) /2 and ( Ac4- lice) at a yield of cracked products of 20 to 45%). We consider this a dangerous criterion which, if applied without sufficient care and without a sound knowledge on the fundamentals of bifunctional catalysis [ 8 1,88 ] may lead to erroneous conclusions. Firstly, unsymmetrical product distributions have been observed at moderate yields of cracked products even on noble metal loaded faujasites [ 77,89 1, and the decisive properties which make a faujasite an ideal bifunctional catalyst are in our opinion still obscure. Secondly, the impossibility to make an ideal bifunctional catalyst from a zeolite with unidimensional pores has by no means been proven. Thirdly, whereas hydrocracking of n-decane suggested that zeolite phi contains intersecting twelve-membered ring pores, evidence has recently been published by Lobo et al. that this zeolite is a mixture of chabazite and offretite [ 1001. As a whole, the isomerization and hydrocracking of n-decane with its countless shape selectivity effects was a landmark in probing pores of microporous materials, i.e., pores of molecular dimensions, by catalytic reactions. It is the merit of Jacobs and Martens to have the suitability of bifinctionalcatalysis with its numerous practical advantages recognized as a tool for probing pore widths. Those who plan to take over isomerization and hydrocracking of long chain naIkanes as a test for probing pore widths are adviced to carefully select the optimum chain length for their purposes. While n-decane may be a reasonable compromise, n-nonane or even n-octane could suffice in some instances, which has the benefit of easier product analyses (we generally discourage, however, potential users to use n-heptane or even smaller n-alkanes ) . Conversely, it may be ad-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

139

vantageous or even necessary to employ an n-alkane feed with more than ten carbon atoms. For example, an attempt has been undertaken [ 1011 to detect differences between the pore systems of zeolites Y and ZSM-20. By that time it was not yet known that zeolite ZSM-20 is an intergrowth of what is frequently designated cubic (FAU ) and hexagonal (EMT) faujasite [ 102,103 1. However, even though the structure was unknown, it was clear that ZSM-20 possesses twelve-membered ring pores; moreover, evidence was available for a very open three-dimensional pore system. On the basis of this knowledge, n-tetradecane was selected as feed hydrocarbon with the rationale that there are three isomers with a propyl and one with a butyl side chain among the monobranched isomers and that such bulky isomers are probably required to detect differences in the pore systems. Indeed, such differences were found [ 1011: not only was 5-butyldecane lacking in the product obtained on Pd/HZSM-20 at low conversion (Xn-tetrade_e=2%, this product did appear at the same conversion with Pt/CaY [ 831)) but the three isomeric propyhmdecanes appeared in significantly different distributions on both bifunctional catalysts. Based on earlier studies [ 1041 on isomerization and hydrocracking of long chain n-alkanes with 9 to 17 carbon atoms, Martens et al. [ 105 ] recently suggested to use n-heptadecane as a probe molecule. They characterized several materials from the FAU/EMT family. In ref. [ 105 1, unlike in the test with n-tetradecane suggested by Weitkamp et al. [ 10 11, the information is acquired from the selectivity of hydrocracking, rather than from feed isomers of different bulkiness. 5.3. Hydrocrackingof Cl, naphthenessuch as butylcyclohexane(the Spaciousness Index, SI) All mechanistic concepts sketched in paragraph 5.2 for isomerization and hydrocracking of alkanes in the absence of shape selectivity can be applied to naphthenes (or even aromatics which are hydrogenated to naphthenes) as well. In particular, the feed hydrocarbon undergoes several steps of skeletal isomerization until a structure is reached from which the very rapid type A /&scission can start. In comparison with alkanes, there are, however, some mechanistic peculiarities [ 1061 which, for the most part, pertain to the hydrocracking reaction. The most pronounced differences between hydrocracking of alkanes and naphthenes have their origin in the sluggishness of carbon-carbon bond rupture insidethe naphthenic ring. The most convincing explanation for the much lower rate of cationic fiscission of endocyclic as compared to exocyclic bonds was put forward by Brouwer and Hogeveen [ 107,108 ] : they argued in terms of the orientation of the a-bond to be cleaved and the vacant porbital at the positively charged carbon atom (see Fig. 15 ) . For a smooth /3-scission, this vacant p-orbital and the /3-bond must be oriented in a coplanar manner since this enables a maximum overlap in the transition state of /3-scission. It is easily seen from the lefthand part of Fig. 15 that this coplanarity requirement is ideally fulfilled in the most stable conformation of an alkylcarbenium ion, and this favorable confor-

140

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) X07-150

ALIPHATICS

ALICYCLICS

Fig. 15. Orbital orientation in aliphatic and alicyclic alkylcarbenium ions [ 107,108 1.

mation can easily be reached through the free rotation around the at-carbon-carbon bond. In a cycloalkylcarbenium ion, by contrast, the /I-carbon-carbon bond which forms a rigid part of the ring is necessarily oriented perpendicularly (fivemembered rings) or near perpendicularly (six-membered rings) with respect to the vacant porbital at the carbon atom which carries the positive charge. This does not allow an efficient orbital overlap in the transition state of fiscission, hence its low rate. It is one repercussion of the slow rupture of endocyclic bonds that hydrocracking of Cl0 naphthenes, irrespective of which isomer (e.g., butylcyclohexane, diethylcyclohexanes or tetramethylcyclohexanes) is used, gives entirely different carbon number distributions than hydrocracking of a Cl0 alkane. Moreover, the C4 fraction in hydrocracking of Cl0 naphthenes consists to a surprisingly large extent (ca. 95 mol-%) of iso-butane, beside small amounts (5 mol-%) of n-butane. All these facts were discovered decades ago by Chevron researchers [ 1091111, but no satisfactory interpretation for the unique selectivity of hydrocracking of Cl0 naphthenes could be advanced by that time. On the basis of the reaction network shown in Fig. 7 combined with an exclusion of endocyclic carbon-carbon bond rupture, one can readily account for the observed selectivities: starting, e.g., from butylcyclohexane, the scheme in Fig. 7 is entered from the position of a monobranched feed hydrocarbon (this time iC,H2, instead of i-CmHzm+z ) . Dehydrogenation followed by protonation of the cycloalkenes gives monobranched cycloalkylcarbenium ions, i-C,H& 1. These undergo type B rearrangements (and repeated ring contraction and enlargements steps, which do not violate the orbital orientation principle) until an (Y,cy, ytribranched precursor cation is reached which can undergo exocyclic type A /3scission. It can easily by demonstrated that, with a total of ten carbon atoms and allowing for normal ring sizes only (i.e., five- and six-membered rings), there exist three such carbocations. It happens that fiscission of all these cations, which are shown in Fig. 16, results in iso-butane and methylcyclopentane as the hydrocracked products. The experimental finding that some minor amounts of by-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

w

TYPE

p

+ A

w

NPE

m

+ h

Fig. 16. All possible exocyclic type A jkissions pentane and isobutane.

B

---)

141

M-CPn

+

i-Bu

M-CPn

+

i-Bu

of CIO-cycloalkylcarbenium ions give methylcyclo-

products (propane, n-butane, pentanes and naphthenes other than methylcyclopentane) are formed can readily be accounted for by invoking some contribution of exocyclic type B j?-scissions, and this is in complete agreement with the role of type B &scissions in hydrocracking of alkanes. The Chevron researchers discovered the unique selectivities in hydrocracking of Cl,, naphthenes with non-zeolitic catalysts such as NiS/SiOz-A1203 [ 109,l IO]. Much later, exactly the same selectivities were encountered in hydrocracking of butylcyclohexane in sufficiently large pore zeolites, such as Pd/LaY [ 1121. By that time, a broad assortment of zeolites with different pore widths were available, so the question arose as to which selectivity changes would occur if catalysts with progressively narrower pores are employed. Molecular models suggested that the tribranched precursor cations (Fig. 16) for the favorable type A j?-scissionare rather bulky. It was a straightforward assumption hat, upon making the pores narrower, the contribution of type B and, ultimately, type C &scissions would increase. Exocyclic type B &scissions starting from dibranched cycloalkylcarbenium ions atoms with ten carbon atoms are shown in Figs. 17a and 17b. A much broader product slate is predicted for hydrocracking of Cl0 naphthenes under shape selectivity constraints: beside iso-butane and methylcyclopentane, one expects significant amounts of propane, n-butane, iso-pentane, cyclopentane, cyclohexane, ethylcyclopentane and methylcyclohexane. All these products were indeed detected when hydrocracking of butylcyclohexane was performed under shape selective conditions [ 73,74,112,113 1. Typical product distributions are depicted in Fig. 18. It is seen that hydrocracking of butylcyclohexane is a peculiar reaction in that it proceeds much less selectively (i.e., it gives many more products) if it is carried out under spatial constraints. The next step was the search for a quantitative selectivity criterion which is related to the pore width. For a number of reasons, the yield ratio of iso-butane and n-butane was chosen [ 73,741. Since its numerical value increases with in-

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

142

“D&/-B

“w

+

N

+

M-CPn

+ n-Bu

=

D

+

&

+

CPn

w

=

B

+

N

+

M-CPn + n-Bu

w

=

+

CHx

0

+

A

+i-Pn

+ i-Bu

Fig. 17. (a) Exocyclic type B, /kissions of dibranched CIO-cycloalkylcarbenium ions and the products resulting from these /kissions. (b) Exocyclic type Bz /3-scissions of dibranched Cl,,-cycloalkylcarbenium ions and the products resulting from these jI-scissions.

creasing space inside the pores, this ratio was named the Spaciousness Index (SZ) [731. SI values for a number of microporous materials are given in Fig. 19. It is evi-

dent that SZ is not appropriate for ranking the ten-membered ring zeolites, they all have Spaciousness Indices around or below 1. On the other hand, the SZvalues for twelve-membered ring molecular sieves cover a wide range from ca. 3 (ZSM12) to more than 20 (faujasite). Hydrocracking of a Cl0 naphthene is, to our knowledge, the best test reaction proposed so far for probing the pore width of large pore molecular sieves. Any Cl,, naphthene can be used as reactant for the determination of SZ. In most of our experiments, butylcyclohexane was employed because (i) it is readily available in good purity at low cost and (ii) the molecule is relatively small so

J. Weitkamp, S. Ernst /Catalysis Today 19(1994) 107-150 w

143

,,,,,,,,, 027 Pdl HZSM-22

.J

123455789

123455759

123458799

123458759

20 0

CARBON NUMBER C, OF CRACKED PRODUCTS Fi8.18. Hydrocrackin8 of butylcyclohexane on four btictional

catalysts with different pore width.

Y I-butane

Definition: SI

q Yn-butane

in hydrocracking

ZSM-23 Fig.

ZSM-12

butylcyclohexane

EU-1

Oftretlte

or pentylcyclopentane.

Mordenlte

L

ZSM-20

19. Spaciousness Indices for various zeolites. Data from ref. [ 741.

that undesired mass transfer limitations for the feed are unlikely. Propylcyclopentane is, perhaps, an even better choice, but this hydrocarbon is not readily available. Even cyclodecane can be used, reported yield ratios of iso-butane and n-butane measured with this naphthene by Jacobs et al. [ 3,114] are in good agreement with the SZ values for the same microporous materials. In the routine application, the Spaciousness Index offers several advantages: (i) As in all hydrocarbon reactions carried out on bifunctional catalysts, there is no deactivation. (ii) The analysis for iso-butane and n-butane is very easy and can be done quickly. (iii) SZ is, in a very broad range, independent of the butyl-

144

J. Weitkamp, S. Ernst /Catalysis

25

I

Today 19 (1994) 107-150

I

I

Y 0 20

0

/

b

0 0 -

O_ 0

/

Beta

l

15 -

10 )

~

/_

Mordenite

00

Offretite

50 0) 250

Cl

‘/ ZSM-5 s I”O, 275

I

0 O

z

w/o

-5

300

/ iTj

0

-

t I 325

350

Temperature, “C Fig. 20. The Spaciousness Index is, over a broad range, independent of the conditions (especially the conversion) in the catalytic experiment. Data from ref. [ 1131.

cyclohexane conversion or the yield of hydrocracked products. Hence, there is no prescribed conversion or yield, and a tedious search for the appropriate experimental conditions is superfluous. Fig. 20 demonstrates for a number of microporous materials the independence of SI of the experimental conditions. In our opinion, SI is based on restricted transition state shape selectivity rather than on mass transfer effects, i.e., hindered diffusion of iso-butane, at least for twelve-membered ring zeolites including ZSM- 12. By using ball-shaped ZSM- 12 crystallites with a diameter of ca. 0.5 pm and rod-like crystallites of the same zeolite with an approximate size of 11 ,um x 1.5 pm, no significant influence on SI could be detected [ 741. Nor was there any significant influence on SI when the nsi/nA ratio was varied from 70 to 300. The independence of SI of the conversion implies that there is no interconversion of iso-butane and n-butane under the conditions applied for catalytic hydrocracking of the Cl0 naphthene. This was confirmed experimentally for two zeolites, viz. Pd/HZSM-5 and Pd/HL [ 113 1. Since, according to our experience, the risk of undesired hydrogenolysis reactions on the metal is somewhat higher for Pt than for Pd, we prefer to use the latter hydrogenation/dehydrogenation metal, usually in an amount of 0.27 wt.-% on the catalyst. The scientific literature reveals that an ever increasing number of groups is getting aware of the potential of the Spaciousness Index for characterizing large pore molecular sieves [ 70,115,116]. 6. Conclusions As a whole, the technique of probing the effective pore width of microporous materials by catalytic test reactions has now reached a certain level of maturity.

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

145

We perceive four types of reaction which have been widely employed, viz. competitive cracking of n-hexane/3-methylheptane mixtures, isomerization and disproportionation of m-xylene, isomerization and hydrocracking of long chain nalkanes, and hydrocraclcing of butylcyclohexane. All these reactions have their specific advantages and disadvantages, as shown in the present article. For large pore molecular sieves, hydrocracking of butylcyclohexane is, in our opinion and for the reasons discussed in the preceding paragraphs, the method of the choice: the carbocation intermediates which govern the selectivity of hydrocracking of Cl0 naphthenes seem to be ideally suited for exploring the space available in the whole range of twelve-membered ring materials. Novel test reactions for super-large pore materials are emerging. Examples are isomerization and hydrocracking of n-heptadecane or alkylation and isomerization of meta-diisopropylbenzene. So far, only hydrocarbon reactions have been proposed for probing the pore width, probably because their mechanisms are known in detail. Reactions of other classes of organic compounds could, however, turn out in the future to be suitable as well.

7. Acknowledgements Financial support by the German Science Foundation (Deutsche Forschungsgemeinschaft ) , Fonds der Chemischen Industrie and Max Buchner-Forschungsstiftung is gratefully acknowledged.

8. References 1 2

3 4 5

6 7 8 9

10

J. Dewing, J. Mol. Catal., 27 (1984) 25-33. F.R. Ribeiro, F. Lemos, G. Perot and M. Guisnet, in R. Setton (Editors), Chemical Reactions in Organic and Inorganic Constrained Systems, D. Reidel, Dordrecht, The Netherlands, 1986, pp. 141-150. P.A. Jacobs and J.A. Martens, Pure Appl. Chem., 58 (1986) 1329-1338. J. Weitkamp and S. Ernst, in J.W. Ward (Editor), Catalysis 1987, (Studies in Surface Science and Catalysis, Vol. 38), Elsevier, Amsterdam, 1988, pp. 367-382. S. Ernst, R. Kumar, M. Neuber and J. Weitkamp, in K.K. Unger, J. Rouquerol, K.S.W. Sing and H. Kral (Editors), Characterization of Porous Solids, (Studies in Surface Science and Catalysis, Vol. 39), Elsevier, Amsterdam, 1988, pp. 531-540. J. Weitkamp and S. Ernst, Catal. Today, 3 ( 1988) 45 l-457. G.T. Kokotailo and J.L. Schlenker, Adv. X-Ray Anal., 24 ( 198 1) 49-6 1. C. Baerlocher, Zeolites, 6 (1986) 325-333. J.M. Newsam and D.E.W. Vaughan, in B. Drzaj, S. Hocevar and S. Pejovnik (Editors), Zeolites: Synthesis, Structure, Technology and Applications, (Studies in Surface Science and Catalysis, Vol. 24), Elsevier, Amsterdam, 1985, pp. 239-248. J.M. Newsam, in R. van Ballmoos, J.B. Higgins and M.M.J. Treaty (Editors), Proceedings from the Ninth International Zeolite Conference, Vol. 1, Butterworth-Heinemann, Stoneham, 1993, pp. 127-141.

146

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

11 G.T. Kokotailo, C.A. Fyfe, G.J. Kennedy, G.C. Gobbi, H. Strobl, C.T. Pasztor, G.E. Barlow and S. Bradley, in Y. Murakami, A. Iijima and. J.W. Ward (Editors), New Developments in Zeolite Science and Technology, Proc. 7th Intern. Zeolite Conf., Kodansha-Elsevier, Tokyo-Amsterdam, 1986, pp. 361-368. 12 C.A. Fyfe, H. Gies, G.T. Kokotailo, Y. Feng, H. Strobl, B. Marler and D.E. Cox, in P.A. Jacobs and R.A. van Santen (Editors), Zeolites: Facts, Figures, Future, (Studies in Surface Science and Catalysis, Vol. 49A), Elsevier, Amsterdam, 1989, pp. 545-557. 13 W.M. Meier and D.H. Olson, Atlas of Zeolite Structure Types, 3rd Edition, Butterworth-Heinemann, London, 1992. 14 N.Y. Chen, W.E. Garwood and F.G. Dwyer, Shape Selective Catalysis in Industrial Applications, Marcel Dekker, New York, Basel, 1989. 15 J. Weitkamp, S. Ernst, H. Dauns and E. Gallei, Chem.-Ing.-Tech., 58 ( 1986) 623-632. 16 D. Fraenkel, M. Chemiavsky, B. Ittah and M. Levy, J. Catal., 101 (1986) 273-283. 17 J. We& and M. Neuber, in T. Inui, S. Namba and T. Tatsumi (Editors), Chemistry of Microporous Crystals, (Studies in Surface Science and Catalysis, Vol. 60), Kodansha-Elsevier, Tokyo-Amsterdam, 1991, pp. 291-301. 18 A. Katayama, M. Toba, G. Takeuchi, F. Mizukami, S. Niwa and S. Mitamura, J. Chem. Sot., Chem. Commun., ( 1991) 39-40. 19 G.S. Lee, J.J. Maj, S.C. Rocke and J.M. Garces, Catal. Lett., 2 (1989) 243-248. 20 J.-R. Butruille and T.J. Pinnavaia, Catal. Lett., 12 ( 1992) 187-192. 21 P.B. Weisz and V.J. Frilette, J. Phys. Chem., 64 (1960) 382. 22 P.B. Weisz, V.J. Frilette, R.W. Maatman and E.B. Mower, J. Catal., 1 ( 1962) 307-312. 23 N.Y. Chen and P.B. Weisz, in P.B. Weisz and WK. Hall (Editors), Kinetics and Catalysis, Chem. Eng. Progr. Symp. Series, Vol. 63 (No. 73), AIChE, New York, 1967, pp. 86-89. 24 S.M. Csicsery, J. Catal., 19 (1970) 394-397. 25 S.M. Csicsery, J. Catal., (1971) 124-130. 26 S.M. Csicsery, in J.A. Rabo (Editor), Zeolite Chemistry and Catalysis, Am. Chem. Sot. Monograph Vol. 171, Am. Chem. Sot., Washington, DC, 1976, pp. 680-713. 27 P.B. Weisz, Pure Appl. Chem., 52 (1980) 2091-2103. 28 S.M. Csicsery, Zeolites, 4 ( 1984) 202-2 13. 29 S.M. Csicsery, Pure Appl. Chem., 58 (1986) 841-856. 30 V.J. Frilette, W.O. HaagandR.M. Iago, J. Catal., (1981) 218-222. 31 N.S. Gnep, J. Tejada and M. Guisnet, Bull. Sot. Chim. Fr., I ( 1982) 5-l 1. 32 J.A. Martens, J. Perez-Pariente, E. Sastre, A. Corma and P.A. Jacobs, Appl. Catal., 45 ( 1988) 85-101. 33 M. Richter, W. Fiebig, H.-G. Jerschkewitz, G. Lischke and G. ohlmann, Zeolites, 9 ( 1989) 238246. 34 F. Joensen, N. Blom, N.J. Tapp, E.G. Derouane and C. Femandez, in P.A. Jacobs and R.A. van Santen (Editors), Zeolites: Facts, Figures, Future, (Studies in Surface Science and Catalysis, Vol. 49B), Elsevier, Amsterdam, 1989, pp. 1131-l 140. 35 H.G. Karge, Y. Wada, J. Weitkamp, S. Ernst, U. Girrbach and H.K. Beyer, in S. Kaliaguine and A. Mahay (Editors), Catalysis on the Energy Scene, (Studies in Surface Science and Catalysis, Vol. 19), Elsevier, Amsterdam, 1984, pp. 101-l 11. 36 J. Weitkamp, S. Ernst, P.A. Jacobs and H.G. Karge, Erdbl, Kohle-Erdgas-Petrochem., 39 ( 1986) 13-18. 37 S.M. Csicsery, J. Catal., 108 (1987) 433-443. 38 0. Levenspiel, Chemical Reaction Engineering, 2nd edition, John Wiley, New York, 1972, pp. 484-485. 39 J. Weitkamp, S. Ernst and R. Gerber, unpublished results (Diploma Thesis R. Gerber, University of Karlsruhe, 1984). 40 W.O. Haag, R.M. Lago and P.B. Weisz, Faraday Discuss. Chem. Sot., 72 ( 1982) 317-330. 41 W. Holderich and L. Riekert, Chem.-Ing.-Tech., 58 (1986) 412-414. 42 P. Voogdand H. van Bekkum, Appl. Catal., 59 (1990) 31 l-331. 43 R.A. Morrison, US Patent 4 686 3 16, assigned to Mobil Oil Corp. ( 1987).

J. Weitkamp, S. Ernst /Catalysis

Today 19 (1994) 107-150

147

44 S. Ernst, J. Weitkamp, J.A. Martensand P.A. Jacobs, Appl. Catal., 48 (1989) 137-148. 45 W.O. Haag and R.M. Dessau, in Proc. 8th Intern. Congress on Catalysis, Vol. 2, Verlag Chemie, Weinheim, Deerfield Beach, Basel, 1984, pp. II-305/B-3 16. 46 W.O. Haag, R.M. Dessau and R.M. Lago, in T. Inui, S. Namba and T. Tatsumi (Editors), Chemistry of Microporous Crystals, (Studies in Surface Science and Catalysis, Vol. 60), Elsevier, Amsterdam, 1991, pp. 255-265. 47 S.M. Csicsery, J. Org. Chem., 34 (1969) 3338-3342. 48 A. Cormaand E. Sastre, J. Catal., 129 (1991) 177-185. 49 D.H. Olson and W.O. Haag, in T.E. Whyte, Jr., R.A. Dalla Betta, E.G. Derouane and R.T.K. Baker (Editors), Catalytic Materials: Relationship Between Structure and Reactivity, Am. Chem. Sot. Symp. Ser., Vol. 248, Am. Chem. Sot., Washington, DC, 1984, pp. 275-307. 50 R. Kumar, G.N. Rao and P. Ratnasamy, in P.A. Jacobs and R.A. van Santen (Editors), Zeolites: Facts, Figures, Future, (Studies in Surface Science and Catalysis, Vol. 49B), Elsevier, Amsterdam, 1989, pp. 1141-l 150. 5 1 G.N. Rao, R. Kumar and P. Ratnasamy, Appl. Catal., 49 ( 1989) 307-3 18. 52 R. Kumar and P. Ratnasamy, J. Catal., 116 (1989) 440-448. 53 R. Kumarand P. Ratnasamy, J. Catal., 118 (1989) 68-78. 54 P. Ratnasamy, R.N. Bhat, S.K. Pokhriyal, S.G. Hegde and R. Kumar, J. Catal., 119 ( 1989) 6570. 55 N.A. Briscoe, D.W. Johnson, M.D. Shannon, G.T. Kokotailo and L.B. McCusker, Zeolites, 8 (1988) 74-76. 56 A. Streitwieser, Jr., and L. Reif, J. Am. Chem. Sot., 82 ( 1960) 5003-5005. 57 M.A.L.anewalaandA.P.Bolton, J.Org. Chem.,34 (1969) 3107-3112. 58 P.A. Jacobs and J.A. Martens, in H. van Bekkum, E.M. Flanigen and J.C. Jansen (Editors), Introduction to Zeolite Science and Practice, (Studies in Surface Science and Catalysis, Vol. 58) Elsevier, Amsterdam, 1991, pp. 445-496. 59 D.S. Santilli, J. Catal., 99 (1986) 327-334. 60 A. Corma, V. Fames, J. Perez-Pariente, E. Sastre, J.A. Martens and P.A. Jacobs, in W.H. Flank and T.E. Whyte, Jr. (Editors), Perspectives in Molecular Sieve Science, Am. Chem. Sot. Symp. Ser., Vol. 368, Am. Chem. Sot., Washington, DC, 1988, pp. 555-568. 61 J. Perez-Pariente, E. Sastre, V. Fames, J.A. Martens, P.A. Jacobs and A. Corma, Appl. Catal., 69 (1991) 125-137. 62 H. Nishi and J.B. Moffat, J. Mol. Catal., 5 1 ( 1989) 193-207. 63 R.C. Sosa, H.K. Beyer and P.A. Jacobs, in Proc. 9th Iberoamerican Symp. on Catalysis, Vol. 2, Lisbon, Portugal, July 16-21,1984, pp. 1426- 1434. 64 S.M. Csicsery, J. Catal., 110 (1988) 348-353. 65 H.G. Karge, J. Ladebeck, Z. Sarbak and K. Hatada, Zeolites, 2 ( 1982) 94-102. 66 H.G. Karge, K. Hatada, Y. Zhang and R. Fiedorow, Zeolites, 3 ( 1983) 13-2 1. 67 H.G. Karge, Y. Wada, J. Weitkamp, S. Ernst, U. Girrbach and H.K. Beyer, in S. Kaliaguine and A. Mahay (Editors), Catalysis on the Energy Scene, (Studies in Surface Science and Catalysis, Vol. 19), Elsevier, Amsterdam, 1984, pp. 101-l 11. 68 R. Kumar, S. Ernst, G.T. Kokotailo and J. Weitkamp, in P.J. Grobet, W.J. Mortier, E.F. Vansant and G. Schulz-Ekloff (Editors), Innovation in Zeolite Materials Science, (Studies in Surface Science and Catalysis, Vol. 37), Elsevier, Amsterdam, 1988, pp. 451-459. 69 S. Ernst and J. Weitkamp, in G. Ghlmann, H. Pfeifer and R. Fricke (Editors), Catalysis and Adsorption by Zeolites, (Studies in Surface Science and Catalysis, Vol. 65), Elsevier, Amsterdam, 1991, pp. 645-652. 70 M.-H. Kim, C.-Y. Chen and M.E. Davis, in M.E. Davis and S.L. Suib (Editors), Selectivity in Catalysis, Am. Chem. Sot. Symp. Ser., Vol. 517, Am. Chem. Sot., Washington, DC, 1993, pp. 222-232. 71 H.G. Karge and J. Ladebeck, in C.H. Amberg and J.F. Kelly (Editors), Proc. 6th Canadian Symp. Catalysis, Ottawa, Ontario, Canada, August 19-2 1,1979, pp. 140-l 5 1. 72 M. Guisnet and P. Magnoux, Appl. Catal., 54 ( 1989) l-27.

148

J. Weitkump, S. Ernst /Catalysis Today 19 (1994) 107-150

73 J. Weitkamp, S. Ernst and R. Kumar, Appl. Catal., 27 (1986) 207-210. 74 J. Weitkamp, S. Ernst and C.Y. Chen, in P.A. Jacobs and R.A. van Santen (Editors), Zeolites: Facts, Figures, Future, (Studies in Surface Science and Catalysis, Vol. 49B), Elsevier, Amsterdam, 1989, pp. 1115-l 129. 75 J.A. Martens, M. Tielen, P.A. Jacobs and J. Weitkamp, Zeolites, 4 (1984) 98-107. 76 J.A. Martens and P.A. Jacobs, Zeolites, 6 (1986) 334-348. 77 H. Schulz and J. Weitkamp, Ind. Eng. Chem., Prod. Res. Dev., 11 ( 1972) 46-53. 78 J. Weitkamp, in J.W. Ward and S.A. Qader (Editors), Hydrocracking and Hydrotreating, Am. Chem. Sot. Symp. Ser., Vol. 20, Am. Chem. Sot., Washington, DC, 1975, pp. l-27. 79 J. Weitkamp, Erdiil, Kohle-Erdgas-Petrochem., 3 1 ( 1978 ) 13-2 1. 80 M. Steijns, G. Froment, P.A. Jacobs, J. Uytterhoven and J. Weitkamp, Ind. Eng. Chem., Prod. Res. Dev., 20 (1981) 654-660. 81 P.B. Weisz,Adv. Catal., 13 (1962) 137-190. 82 M. Steijns and G.F. Froment, Ind. Eng. Chem., Prod. Res. Dev., 20 ( 1981) 660-668. 83 J. Weitkamp, Ind. Eng. Chem., Prod. Res. Dev., 21 (1982) 550-558. 84 J. Weitkamp, W. Gerhardt and P.A. Jacobs, in: Proc. Intern. Symposium on Zeolite Catalysis, Si6fok, Hungary, May 13-16,1985, pp. 261-270. 85 J.A. Martens, P.A. Jacobs and J. Weitkamp, Appl. Catal., 20 (1986) 239-281. 86 J.A. Martens, P.A. Jacobs and J. Weitkamp, Appl. Catal., 20 (1986) 283-303. 87 J. Weitkamp, P.A. Jacobs and J.A. Martens, Appl. Catal., 8 (1983) 123-141. 88 H.L. Coonradt and W.E. Garwood, Ind. Eng. Chem., Proc. Des. Dev., 3 ( 1964) 38-45. 89 H. Pichler, H. Schulz, H.O. Reitemeyer and J. Weitkamp, Erdiil, Kohle-Erdgas-Petrochem., 25 (1972) 494-505. 90 P.A. Jacobs, J.A. Martens, J. We& and H.K. Beyer, Faraday Discuss. Chem. Sot., 72 (1982) 353-369. 91 S. Ernst, G.T. KokotaiIo, R. Kumar and J. Weitkamp, in M.J. Phillips and M. Teman (Editors), Proc. 9th Intern. Congress on Catalysis, Vol. 1, The Chemical Institute of Canada, Ottawa, 1988, pp. 388-395. 92 P.A. Jacobs, J.B. Uytterhoeven, M. Steijns, G. Froment and J. Weitkamp, in L.V.C. Rees (Editor ) , Proc. 5th International Conference on Zeolites, Heyden, Philadelphia, 1980, pp. 607-6 15. 93 J. Weitkamp and P.A. Jacobs, Preprints, Div. Fuel Chem., Am. Chem. Sot., 28 (No. 2) ( 1983) 153-157. 94 J.A. Martens, R. Parton, L. Uytterhoven, P.A. Jacobs and G.F. Froment, Appl. Catal., 76 ( 1991) 95-l 16. 95 J.A. Martens and P.A. Jacobs, in E.G. Derouane, F. Lemos, C. Naccache and F.R. Ribeiro (Editors), Zeolite Microporous Solids: Synthesis, Structure and Reactivity, Kluwer, Dordrecht, 1992, pp. 5 1 l-529. 96 J.A. Martens and P.A. Jacobs, in J.B. Moffat (Editor), Theoretical Aspects of Heterogeneous Catalysis, Van Nostrand-Reinhold, New York, 1990, pp. 52-109. 97 J.A. Martens and P.A. Jacobs, J. Catal., 124 (1990) 357-366. 98 J.A. Martens, P.A. Jacobs and S. Cartlidge, Zeolites, 9 ( 1989) 425-427. 99 M.M. Olken and J.M. Games, in R. von Balhnoos, J.B. Higgins and M.M.J. Treaty (Editors), Proceedings from the Ninth International Zeolite Conference, Vol. 2, Butterworth-Heinemann, Stoneham, 1993, pp. 559-566. 100 R.F. Lobo, M.J. Annen and M.E. Davis, J. Chem. Sot., Faraday Trans., 88 (1992) 2791-2795. 101 J. Weitkamp, S. Ernst, V. Corms-Corberan and G.T. Kokotailo, 7th Intern. Zeolite Conf., Tokyo, Japan, Aug. 17-20, 1986, Preprints of Poster Papers, Japan Association of Zeolite, Tokyo, 1986, pp. 239-240. 102 V. Ftllop, G. Borbely, H.K. Beyer, S. Ernst and J. Weitkamp, J. Chem. Sot., Faraday Trans. I, 85 (1989) 2127-2139. 103 J.M. Newsam, M.M.J. Treaty, D.E.W. Vaughan, KG. Strohmaier and W.J. Mortier, J. Chem. Sot., Chem. Commun., (1989) 493-495

J. Weitkamp, S. Ernst /Catalysis Today 19 (1994) 107-150

149

104 J.A. Martens, M. Tielen and P.A. Jacobs, in H.G. Karge and J. Weitkamp (Editors), Zeolites as Catalysts, Sorbents and Detergent Builders-Applications and Innovations, (Studies in Surface Science and Catalysis, Vol. 46), Elsevier, Amsterdam, 1989, pp. 49-60. 105 J.A. Martens, G. Vanbutsele and P.A. Jacobs, in R. von Balhnoos, J.B. Higgins and M.M.J. Treaty, Proceedings from the Ninth International Zeolite Conference, Vol. 2, Butterworth-Heinemann, Stoneham, 1993, pp. 355-362. 106 J. We&, S. Ernst and H.G. Karge, Erdiil und Kohle-Erdgas-Petrochem., 37 (1984) 457462. 107 D.M. Brouwer and H. Hogeveen, Rec. Trav. Chim., 89 ( 1970) 21 l-224. 108 D.M. Brouwer, in R. Prins and G.C.A. Schuit (Editors), Chemistry and Chemical Engineering of Catalytic Processes, Sijthoff and Noordhoff, Alphen aan den Rijn, The Netherlands, 1980, pp. 137-160. 109 C.J. Egan, G.E. Langl0isandR.J. White, J. Am. Chem. Sot., 84 (1962) 1204-1212. 110 G.E. Langlois and R.F. Sullivan, in L.J. Spillane and H.P. Leftin (Editors), Adv. Chem. Ser. 97, Am. Chem. Sot., Washington, DC, 1970, pp. 38-67. 111 R.F. Sullivan and J.W. Scott, in B.H. Davis and W.P. Hettinger, Jr. (Editors), Selected American Histories, Am. Chem. Sot., Washington, DC, 1983, pp. 293-313. 112 S. Ernst and J. We&, in: Proc. Intern. Symposium on Zeolite Catalysis, Siofok, Hungary, May 13-16,1985, pp. 457-466. 113 J. Weitkamp, C.Y. Chen and S. Ernst, in T. Inui (Editor), Successful Design of Catalysts, (Studies in Surface Science and Catalysis, Vol. 44), Elsevier, Amsterdam, 1988, pp. 343-350. 114 P.A. Jacobs and M. Tielen, in: Proc. 8th Intern. Congr. Catalysis, Vol. 4, Verlag Chemie, Weinheim, 1984, pp. IV-357/IV-369. 115 M. Otake, J. Catal., 142 (1993) 303-311. 116 L. Fomi, S. Amarilli, G. Bellussi, C. Perego and A. Carati, Appl. Catal., 103 ( 1993) 173-l 82.