Chapter 1: Relation between Acidic and Catalytic Properties of Zeolites

Chapter 1: Relation between Acidic and Catalytic Properties of Zeolites

5 Chapter 1 RELATION BETWEEN ACIDIC AND CATALYTIC PROPERTIES OF ZEOLITES J Dwyer and P J O'Malley Chemistry Department, UMIST, Manchester M60 lQD, U...

3MB Sizes 9 Downloads 14 Views

5 Chapter 1

RELATION BETWEEN ACIDIC AND CATALYTIC PROPERTIES OF ZEOLITES J Dwyer and P J O'Malley

Chemistry Department, UMIST, Manchester M60 lQD, UK

1.1

ACIDITY DEFINITION AND MEASUREMENT The acidity of solids is usually interpreted in terms of proton donor

capacity

-

Bronsted acidity (ref.1) or as electron acceptor capacity - Lewis

acidity ( 2 ) .

1.1.1

Basic sites in solids can also be utilised in catalysis.

Bronsted Acidity and Acidity Functions (H 0' HR) In homogeneous systems (aqueous and non-aqueous) a quantitative measure of

protic acidity may be provided by an acidity parameter ( 3 ) . For a weak base 'B' the parameter H

is defined as follows.

B + H+

BH+F

(1)

KBH+ = aB%+/aBH+

=

[B]/[

BH.1

( aH+fB/fBH+)

where ai is activity, fi activity coefficient and[

1' /[ 1

K + BH BH

[

and

pKBH+

Ho

-

B

-

-

(2)

]is concentration.

= aH+fB/fBHi

(3)

log([BH+]/[B])

log (aH+ f /f + ) B BH

In dilute solution (f /f + B BH

=

(4)

1) and H

= pH.

o-

+ and, since BH for a weak base the equilibrium in (1) is on the product side, pK + is a BH negative number (Ho< 1). Extensive protonation of B gives H < pK BH+* Some typical values of Ho for homogeneous solutions are given in Table 1.1 For equal concentrations o f free and protonated base H

= pK

6 Table 1.1

HAMMETT FUNCTIONS(a) -(Ho)

Acid

-H

Classification

From ref 3

(a)

<

Typically (4,5) acids with -H with -H these

5 are classified as weak acids.

= 8.2-12.8 as strong acids and for superacids -Ho>

divisions, especially

that

for classification of

Those

16.8, although

superacidity, are

arbitrary. Superacids were initially, and arbitrarily, taken as acids with H than that for 100% sulphuric acid (4).

less

Superacids are usually produced by

adding a stronger acid to an initially strong acid or more commonly by adding a strong

Lewis

acid

to

a

strong

Bronsted

acid

in

order

to

shift

the

autoprotonation equilibrium ( 5 ) . 2HA

+ L ==?

H2Af + LA-

In such cases there is a substantial increase in acidity (decrease in H ) over that of pure HA.

1.1.2Hammett indicators and the H

scale

For two bases B1 and B reacting with acid solutions having the same 2 strength (from eqn ( 2 ) )

7

and is constant at fixed temperature. Hammett (3) assumed that, when B1 and B2 are sufficiently similar in structure, the ratio of the activity coefficients in a given solution is constant so that the ratio of K values can be obtained from experimentally measured concentration ratios.

If the strongest base in the

series (B1) can be protonated in pure water, where f. Z 1, then KBH+ can be calculated from equn (2).

The base B1 and a weaker base B2 can then be

+

protonated in a stronger acid (eg dilute H SO ) where f. 1 but the ratio of 2 4 activity coefficients in equn ( 5 ) is unity. From measured values of and [BiHf], KB H+ can be calculated from KBiH+. 1

determine K

[Bi]

The weaker base B2 can be used to

H+ for a weaker base, B3, using a stronger acid solution.

In this

B3

way a range of Hammett indicators have been developed which facilitate the measurement of Ho in homogeneous systems using equation (3) which requires that [ B H ' ]

and [B]

should be measurable.

The initial Hammett indicators were

substituted primary aromatic amines

R. N H ~+ H+

R~NH~+

but others have been developed. The vi idity of Ho functions determine in this way depends upon the assumption that

the

activity

coefficient ratio

for

different bases is constant in a given acid solution.

This assumption implies

that a plot of log [BH+]/[B]

-1.0) for the series of

bases.

vs Ho is linear (slope

=

Other acidity functions have since been developed based on other

indicators, for example alkyl carbinols (6) which dissociate as, ROH + H+

R+ + H20

and generate ( 7 ) the function H R

An extensive discussion of acidity scales is available (8).

a 1.1.3Lewis Acids Whereas Bronsted acid-base interactions involve a common proton-transfer process there is no corresponding common process to provide a unique basis for comparison of the strengths of Lewis acids. coordinating

power

of

a

Lewis

acid

Consequently the strength, or

cannot

clearly

be

defined

experimentally, its strength depends on the particular Lewis base. preferential

interactions

are

found

of

the

kind

incorporated

and,

Specific into

the

classification (9) of donors and acceptors as type 'a' or 'b' and subsequently to the more general classification (10) into

'hard' and

'soft' with

the

development of the principle of hard and soft acids and bases (HSAB principle). The HSAB principle states that hard acids 'prefer' to bind to hard bases and soft acids 'prefer' soft bases.

In very general terms soft acid-soft base

interactions involve a significant covalent contribution whereas hard acid-hard base interactions involve less covalent and more electrostatic interactions. Consequently, in aqueous solution, A13+ (hard acid) will bind to F- (hard base) in preference to triphenyl phosphine (soft base) whereas Ag+ (soft acid) will preferentially bind to the phosphine. Theoretical approaches to HASB interactions have been made (11) which, although they are not completely satisfactory, do result in representation of the interaction energy in terms of an electrostatic and a covalent term.

Large

values for the electrostatic term are associated with pronounced hard-hard interaction and large values for the covalent term are expected for strong soft-soft interactions.

A semi-quantitative approach to such interactions

utilizes the enthalpies (AH) of formation of

donor-acceptor molecules

to

estimate parameters which may be used to estimate contributions to A H of mainly electrostatic (Ei) and mainly covalent (Ci) terms (12).

Attempts to quantify

complexation of Lewis acids and bases via the use of donor numbers (DN) and acceptor numbers (AN) have also been made (13). As with Bronsted acids, superacids of the Lewis type are known.

These are

taken arbitrarily as those Lewis acids stronger than AlCl (5). The difficulty, 3 outlined above, in defining the strength of a Lewis acid means that a unique scale cannot be generated. SbF5, AsF

5

However, it is evident that some Lewis acids (eg

have a greater ability to ionize alkyl halides than, for instance

A1C13, and consequently are much more effective catalytically. Additionally, in spite of the above difficulties, a wide range of experimental investigations have resulted in an order of relative Lewis acidity for the MFn type acids (5).

9

1.1.4Solid Acids The proton donating ability of a solid, reflecting its ability to protonate a base 'B', can, by analogy with solutions, be expressed in the equilibrium, H+ + B

HB+ such that,

H

= -log (aH+fB/fBH+ ) = p

~

-

~log ~(

(6)

+

fB etc refer to concentrations and activity coefficients on the catalyst surface. However, lack of knowledge concerning activity coefficients where [B], for

solid

species means

that

acidity

functions

are

not

well

defined

thermodynamically. Moreover, for solid acids, there is generally a distribution of acid site strengths. Consequently, surface acidity requires specification of the relative numbers of acid sites with given strength. Because surfaces can be heterogeneous so that differing types of acid (or basic) site can coexist it is common, in spite of the difficulty in defining strengths of Lewis sites, to extend the approach above to include Lewis acid sites using the equilibrium AB

A

+

B:

KAB = (aA[B]/[AB]

)(fB/fAB)

[ I is the concentration of a

where aA is the activity of the Lewis acid site, B

CI

neutral base ' B ' which interacts with the Lewis site and AB is the concentration of the addition product.

Scales based on both Hammett type

indicators,

typically involving amine or ketonic species, and aryl alcohol indicators (HR) are utilised.

The Hammett type indicators tend to interact with both Bronsted

and Lewis sites on the solid, whereas the arylcarbinol indicators tend to react only with Bronsted sites ROH + H+

R+ + H20

10 For solid acids the availability of sites to molecular probes, especially

large indicator molecules must always be considered.

Table 2 gives a typical

classification of solid acids (14).

Table 1.2

CLASSIFICATION OF MAJOR SOLID ACIDS ( 1 4 )

Zeolites and related materials Metal oxides and sulphides (eg A1203, V203,

CdS, ZnS)

Mixed oxides (eg Si02/A1203, Si02/Ti0 ) 2 Natural clays and pillared interlayer clays Supported acids (eg H3P04 on Si02) Metal salts (eg A12S04, MgS04) Heteropoly acids Cation exchange resins (eg Amberlyst Series, Nafion) Treated graphites

It is also evident that the acid function of many solids can be enhanced, for example halogenation increases the acidity of many oxides.

In some cases

enhancement can be sufficient to generate superacidity. Solid superacids may also be generated by deposition or intercalation of appropriate species (15).

1.1.5Zeolites An important class of solid acid catalyst is based on zeolites.

A brief

introduction to zeolite structures is now given and the acid catalytic activity of these materials is then discussed in some detail. Zeolites are crystalline aluminosilicates. The unit cell contents may be represented simply as M (A102)x(Si02)y w.H20 x/n

11 where the material within the brackets refers to the composition of the framework and the material outside refers to the charge compensating cations

(M) with valence 'n' and the intercrystalline water.

Both A1 and Si are

tetrahedrally linked to oxygens and each tetrahedral unit shares each of its oxygens with another tetrahedral atom ('TI atom) so that each IT' atom is linked via oxygens to four others in a three dimensional framework structure. A

more comprehensive representation has recently been proposed (17).

usual to represent zeolite frameworks by line diagrams.

It is

The intersection of

lines represents a 'T' atom and the line represents an oxygen bridge.

Because

of the tetrahedral angles the oxygens may be above or below the plane of any ring of 'T' atoms.

Rings containing four, five, six, eight, ten and twelve

tetrahedra are known in zeolites.

Zeolites are microporous and have, when

dehydrated, considerable void volume which is accessed by "windows" or .rings of tetrahedra.

The size of window restricting entrance to the micropore

volume of some commercially interesting zeolites is given in Table 1.3 along with pore dimensionality and cavity size. zeolites are shown in Fig 1.1.

The structures of some of these

Zeolites A and X (or Y ) have large cavities,

entered by eight and twelve ring windows respectively, the linking of which generates a 3D pore system.

The smaller cavities associated with the sodalite

units (six-ring windows) are, generally, not of catalytic significance. In mordenite the elliptical twelve-ring windows provide access to the single pores on which there are small side "pockets" entered via elliptical eight rings.

Void volumes in the pentasils are made up from two sets of

interconnecting ten-ring pore systems.

Mordenite and the pentasils do not

possess a cavity system but at the pore intersection of the pentasils there is significantly increased space. A proportion of the intersections is slightly larger in ZSM-11 than in ZSM-5. The introduction of each aluminium into the framework introduces a negative charge which must be compensated by extra framework cations. manipulation of cation types and of the framework ratio Si/A1 (O
+

The y)<

0.5) (18) allows for considerable variation in the acidic properties of

zeolites. The windows into zeolite crystals are of molecular dimensions (Table 1.3) and window size can be used in selective sorption (molecular sieving). Snrbates entering the crystals are subjected to considerable electric field gradients, associated with the charges on cations and framework, and may be extensively polarised. Heats of sorption, which can be large, depend upon the

12

CHARACTEWSTIC PENTASIL

LAYER

MORDENITE hannb 6 l x 5 6 A )

6.4 x5.6A)

THETA 1.

Fig.1.

13 structure of the sorbate, particularly polarisability and dipole/quadrupole moments, and upon the composition of the catalyst.

Since zeolites are

microporous they show type 1 isotherms thereby concentrating sorbates within the crystals relative to concentration in wider pore (amorphous) catalysts. Given that most of the reactive sites are present within the crystals, shape-selective catalysis can arise when there is preferential sorption of one reactant from a mixture (reactant selectivity), preferential diffusion and egress of one product in a product mixture (product selectivity) or selective discrimination, on the basis of size, against a particular intermediate to preclude a reaction path (transition-state selectivity). Excellent reviews on shape-selective catalysis are available (19)

WINDOWS AND PORE STRUCTURES OF ZEOLITES

Table 1.3

Zeolite Name

Code

Window

Peripheral in absence of 0-atom

Linde A

LTA

Free Dimensions/A Dimensionality Types of

8

of pore system Cavity

cations

and size/A 4.1

3

Sodalite(6.6) 26 hedron(ll.4)

Erionite

ERI

8

3.6 x 5.2

3

23 hedron (6.3 x 13)

ZSM-11

MEL

10

5.4 x 5.6

3

ZSM-5

MFI

10

5.4 x 5.6

3

23 hedron

5.1 x 5.5 Mordenite MOR

12

6.7 x 7.0

2

Linde X,Y FAU

12

7.4

3

sodalite(6.6) 26 hedron(ll.8)

14 1.2

MEASUREMENT OF ACIDITY IN SOLIDS The measurement of acidity is more difficult for solid acids than for

solutions. This is due, principally, to the heterogeneous nature of the acid sites in most solid acids and to the availability of non-acidic interactive sites for basic probe molecules. An additional problem arises, particularly in micropnrous materials, when

access of basic

molecules to

acid

sites is

restricted. Difficulties also exist in the definition of appropriate acidity functions (section 1.1). The methods commonly used to test the acidity of solids include the following.

-

Aqueous titration Non-aqueous titration and use of indicators Sorption/desorption energetics or rate processes

- Spectroscopic methods -

Catalytic rate measurements

1.2.1 Aqueous Methods

Typically, a suspension of the solid in aqueous solution is titrated with a base, for example NaOH, to a neutral end point.

Such methods are not generally

recommended (20) but can provide useful information particularly if combined with other techniques (21). 1.2.2

Non Aqueous Methods Hammett, and other indicators (eg aryl carbinols), are widely used in non

aqueous solution, usually benzene or iso-octane, to determine surface acidity. When the colour of a sorbed indicator (B) corresponds to that of its acid form then Ho for the solid is equal to or is lower than pKBH+ for the indicator (equations (3)and ( 6 ) ) . Consequently, a series of indicators having a range of pK values can be used to bracket the strength of surface acidity.

Examples of

indicators and procedures are given in references (15) and ( 2 2 ) .

Surface

acidity can sometimes be related to the strength of a sulphuric acid solution which corresponds to the mid-point of each indicator transition (the pKBH+ value).

Typical values for surface acidity are given in Table 2.1 (15a).

15 Table 2.1

SURFACE A C I D I T Y OF SOME S O L I D S (15a)

Solid Acid

-H

Kaolinite

3.0

-

5.6

Hydrogen Kaolinite

5.6 - 8.2

Silica-Alumina

8.2

H PO /silica qel 3 4

1.5 - 3.0

Relative numbers of

acid sites with

strength between given values,

determined by indicators, can be established by titration procedures using an amine in a non aqueous medium. titration.

Typically n-butylamine in benzene is used for

Successive amounts of amine are added to the pretreated catalyst;

the system is allowed to equilibrate and the indicator is then added.

The

amount of amine to discharge the acid colour of the indicator is representative of the number of surface acid sites with strength greater than pKBH+ for the indicator.

A distribution of acid site strengths for a particular solid can

then be established.

Such distributions are frequently expressed as fractions

of sites with acidity between given concentrations of sulphuric acid. In practice, problems arise in the use of indicators involving;

(i) the

correct interpretation of colour changes; (ii) the establishment of equilibrium between base and solid; (iii) the interaction of base with non-acidic sites; (iv) the

interaction of non-aqueous solvents with

accessibility of acid sites to base and to indicator.

acid

sites;

(v) the

Where possible, colour

changes should be observed spectroscopically (23) and the size of indicator molecules is particularly important in microporous solids such as zeolites (24). Moreover the concentration of indicator should be sufficiently low so that the equilibrium between surface sites and amine molecules is undisturbed by the indicator. The problems directly related to the indicator can be eliminated by using calorimetric, titration with non aqueous solutions of organic bases (25). However, problems associated with accessibility of sites, establishment of equilibria and interaction of base with non-acid sites (eg cations) remain.

A

16

further problem concerning all indicator and titration procedures is that they usually involve measurement under conditions far removed from those typical of catalytic processing.

1.2.3 Sorption/Desorption Methods

The adsorption of appropriate basic molecules can be used to probe surface acidity.

Interaction enthalpies can be measured directly by calorimetry (26) or

indirectly

from

relationship.

equilibrium

isotherms

(27) using

the

Clausius

Clapeyron

In both cases enthalpies may be represented as a function of

surface coverage and hence provide an acid site distribution.

Generally the

largest enthalpy changes occur at low surface coverage where the base adsorbed on the more active sites.

is

Problems associated with site accessibility

and establishment of equilibrium can arise and, as with indicator methods, it must be remembered that strong interaction between a base sorbate and a solid does not in all cases imply strong catalytic activity.

For example, the

isosteric enthalpy of sorption of ammonia is higher on A1203 than on Si02/A1203 but the former is unreactive in isomerisation of o-xylene whereas silica-alumina catalysts

are

effective

(28).

Differential

heats

of

sorption

measured

incrementally over a range of surface coverage can be used to develop acidity spectra, discussed later in this section. Enthalpies of sorption under conditions closer

to

those

applying

in

catalysis can be made chromatographically (29). Typically, the solid acid is made up into a chromatographic column and the corrected retention volume of suitable probe molecules is determined as a function of temperature and of pulse size.

By determining the sorption enthalpy of a weak base (eg benzene) at

different surface coverages of a strong and well-held base (eg pyridine) a spectrum of acidity may be generated. Currently, a technique based on the temperature programmed desorption (TPD) of basic molecules is widely used.

Typically a small fixed bed of the solid

acid is pretreated in dry carrier gas.

The acid sites are then saturated with a

volatile base, for example ammonia, at

a

pre-chosen

temperature is thenraised at a selected rate.

temperature and

the

As the temperature increases

ammonia is desorbed into the carrier gas stream which is continually monitored. In principle it is possible to estimate the amount of base desorbed up to any particular temperature.

The number o f peaks in the temperature profile can be

an indication of the number of types of acid site, providing resolution is adequate, and the higher the peak temperature the more strongly acid is the type

17

000501 K r l As l l 2 St M -1 5 5

00025 0

IGclZSM-5

0

250

500

750

d 1

TEMPERATURV'C TPD curves for B,Fe,Ga and Al forms of ZSM-5.

I

1

100

200

300 400 TEMPERATURU'C Relationship between TPD temperature max. and GO".

Fig.2.1. Isomorphous substitution of A V S i in HZSM-5. (Ref. 31).

3800 3400

3800 3400 @cm-1 Of Wdine dosed Imol'unit cel'l Fig.23.1.R hydroxyl stretching regkn for a).HY, b).NaHY-600, Fig. 22. Isomerisation of cy&Epane. Poisoning c). NaHY -45.6, and d)CeY. Rf.34b). -bygyridine.(UMIST),

-"'

--

18

of site.

Moreover, by making appropriate assumptions (30) it is possible to

determine activation energies associated with desorption processes. Some recent TPD results (31) for ammonia sorbed on pentasil (ZSM-5) type structures containing B, Fe, Ga, or A1 in the framework are shown in Fig 2.l(a) and acidity as reflected in peak maximum is compared with the stretching frequency of the acidic hydroxyl in Fig 2.l(b). Using TPD methods it is possible to compare catalysts reasonably quickly but it should be noted that results from TPD studies can vary somewhat with heating and carrier gas flow rates.

Moreover, difficulties can arise in

completely eliminating the effect of physisorbed species and it is important to know exactly what is being desorbed.

Measurement of pressure changes in static

TPD experiments or of physical changes in the carrier gas stream are unreliable and can result in spurious descriptions of desorption by

mass

acidity.

spectrometry eliminates most

of

Monitoring of these problems

the

since

different species including any decomposition products can be recognised.

As

with other sorption/desorption processes TPD does not define the types of acid site.

A modification of the sorption technique involves the use of sorbed bases in catalytic titration (32).

Incremental additions of a base are made to a

catalyst and the activity is measured following each additionuntil activity is effectively eliminated.

In this way the number of catalytically active sites,

under selected conditions, may be estimated.

Fig 2.2

shows typical results

(32).

.2.4 Spectroscopic Methods

The previous methods based on titration using indicators or calorimetry or on sorption give information about the interaction of bases with solid acids but provide no information on the chemical nature of the acid sites.

Spectroscopic

studies can provide this information. 1.2.4.1 Infrared Spectroscopy

The infrared technique is one of the most widely used to determine the surface acidity of solids.

The strength of Bronsted acidity, within a series of

similar solid acids, varies inversely with the stretching frequency of the acidic surface hydroxyl groups.

In simple terms this may be related to the

decrease in bond force constant with decrease in stretching frequency.

Both

Bronsted and Lewis sites, in solids may be probed using sorbed bases. An excellent and comprehensive review of the use of infrared spectroscopy

19 in the study of acid sites in zeolites is available (33).

In what follows only

the principal features of the technique, together with

some more

recent

developments not covered in ref (33), are outlined. (a) Infrared Studies of Hydroxyl Groups on Solid Acids F o r amorphous silica only one OH stretching band, at approx 3740 cm-l, is

observed and it is assigned to the ESi-OH group terminating polymer chains. In addition to this terminal ;Si-OH

group, other bands are observed in zeolites

(and in silica-alumina), the number and frequency of vibration depending upon zeolite structure, pretreatment and composition. The types of hydroxyl found in zeolites are discussed subsequently (section 1.4) in detail.

It is sufficient,

here, to say that infrared studies result in the classification of zeolite framework hydroxyls as either bridging.

terminal

(terminating siloxane

chains) or

The bridging hydroxyls are in the vicinity of a framework aluminium

atom, viz.

Y+

H+

The location of framework hydroxyl groups within zeolite structures can also be investigated, using infrared spectroscopy. Evidence for location, in the small pore region of zeolite HY, of hydroxyls giving rise to the low-frequency band and evidence for location of the high-frequency hydroxyls in the supercage can be obtained in this way ( 3 3 ) .

Typical infrared spectra of the hydroxyl region

of HY and M/H-Y zeolites dehydrated at 4OOOC are shown in Fig 2.3. Overlapping peaks in the infrared spectra of the hydroxyl region make it difficult to distinguish between sites which differ slightly in resonance frequency.

Deconvolution techniques used to separate out both "low frequency"

(34a.3and, more recently, "high frequency" (35) components of bands in HY

zeolite, have been helpful in this regard.

Enhanced resolution increases the

capability of the infrared technique. Diffuse reflectance infrared spectroscopy (36) has recently been shown to provide additional information, compared with previous transmission work, om surface hydroxyls

in silica, silica-alumina

and

zeolites.

The

diffuse

reflectance technique is particularly appropriate when the concentration of hydroxyls is low, as in the high-silica zeolites.

Moreover, extension into the

near-infrared permits the examination of overtones and combined bending and

20

stretching vibrations as well as fundamental transitions (36). (b) Infrared Studies of Sorbed Species The infrared spectrum of an adsorbed molecule, for example a base, is frequently modified by the nature and degree of its interaction with a surface site.

For example, the infrared spectrum of sorbed pyridine depends upon

whether the pyridine molecule is hydrogen bonded or is protonated by surface hydrogens or whether it is coordinated via nitrogen to a Lewis acid site (33). The infrared characteristics of pyridine in different adsorbed states are given in Table 2.2 and

typical IR spectra of pyridine sorbed on zeolites are given

in ref (33). The interaction of a weak base such as benzene with surface hydroxyls can result in broadening of the infrared hydroxyl band and a shift to lower

-

The measured "bathochromic" shift in frequency (AV ) OH is observed (38) to vary directly with the Bronsted acidity of zeolite hydroxyls absorption frequencies.

(see Fig 4.8, Section 1 . 4 ) .

The stretching frequency of molecular hydrogen

adsorbed at 77 K on solid acids also shows a dependence on the particular type of adsorption site (36).

The assignment of H2 stretching frequencies (Table

2.3) is of particular interest since it suggests that different types of Lewis

site may be distinguished.

However, in view of recent NMR studies (39) the

Lewis acid species may need reconsideration. In general the infrared technique can be used for quantitative studies of sorbed species, as well as for .identification, especially relative values.

in considering

However, a detailed knowledge of extinction coefficients is

not available and values cannot be assumed to be constant for a similar species sorbed on similar solid acids.

21 Table 2 . 2 ( 3 7 ) INFRARED SPECTRUM OF PYRIDINE

Sorbed s p e c i e s Pyridine

LP

GEL(^)

ADSORBED ON ALUMINOSILICA

(b) MP

BP

Assignment

3260

NH

7a

3188 3083

( 3147)

(3147)

3054

(3114)

3114)

3065

20b

CH

lb

CH

3054

( 3087 )

3087)

1580

1620

1638

3043

7b

CH

1614

8a

1572

1577

CC(N)

1620

1593

8b

CC(N)

1482 1439

1482

1490

1490

19a

CC(N)

1450

1545

1438

19b

CC(N)

( a ) R e s u l t s Wavenumber/crnfh)

1.P i n d i c a t e -

nvridinp

1

rnnrdinatplv

hnndpd

tn

a

Tewis

cite.

p r o t o n a t e d p y r i d i n e and MP r e f e r s t o m o l e c u l a r l y adsorbed hydrogen bonding t o s u r f a c e h y d r o x y l s .

RP

indirites

p y r i d i n e with

22

Table 2.3 CHARACTERISATION OF SOLID ACID SITES IN ZEOLITES USING H2 SORPTION AT 77 K

Site

Infrared Frequency (

Terminal 0-H

5/cm-'

)

4105

Extra framework aluminium hydroxy species

4060

Bronsted Acidity due to bridging 0-H

4125

Lewis Acidity (a) Trigonal Si

4035

( b ) Trigonal A1

4010

1.2.4.2 Nuclear Magnetic Resonance Spectroscopy

Recent developments including superconducting magnets, magic angle spinning (MAS) and improved computerisation have resulted in the application of nuclear magnetic resonance (NMR) spectroscopy to the study of solid acids (41). (a) The Quantity and acid strength of surface hydroxyls The total hydroxyl group concentration of a solid acid simply involves measuring the area under the proton magnetic resonance signal and relating this to a standard of known H'

content (eg adamantane).

Table 2.4 gives results for

NaY and HY where the expected decrease in concentration of hydroxyl groups with increasing pretreatment temperature is observed.

23

Table 2.4 (41) TOTAL NUMBER OF HYDROXYL GROUPS PER CAVITY FOR ZEOLITES NaY AND HY FOLLOWING HEAT TREATMENT BETWEEN 300'C

(90 HY

-

300) AND 700°C (90 HY

Sample

-

700)

Number of hydroxyls

NaY

0.3

90 HY

-

300

7.0

400

6.7

450

6.5

500

1.5

600

0.7

700

0.3

However, in order to distinguish between sites having differing acidity, which show different chemical shift values, it is necessary to use narrowing

techniques.

Magic-angle-spinning,

contributions to the H '

which

eliminates

line

anisotropic

resonance signal, is particularly effective in this

respect and leads to the desired resolution.

The enhanced resolution allows

recognition of sites of differing acidity in zeolites which are separated by only a few ppm (Fig 2.4). Based on the premise that, as the acid strength of a hydroxyl increases, the chemical shift 6

H

of the H '

proton magnetic resonance increases, the bands

(a) (b),(c) and (d) in Fig 2.4 have been assigned to terminal hydroxyls (associated with amorphous material and defect sites in zeolites) (a), to bridging hydroxyls (b) and (c) and to hydrogens in NH4+ ions (d). Consequently, NMR

provides

a

classification similar

to

that

observed

using

infrared

spectroscopy but allows for a more accurate quantification of the numbers of hydrogens giving rise to a particular signal.

24

Bo

c

7 45 2 hPPm Fig.2-4. Effect of MASNMR on the hydrogen band envelope. (Ref.41b).

4.75

n

3680

cc

-5 ?Ym 5 3660z

%2! 4.0 .u c

4.25 -

-3640f

>)

& t

A

r

~

> b

-3620

25 (b) Adsorption characteristics of basic probe molecules As

with infrared spectroscopy, suitable sorbate molecules may be used to

probe surface acidity.

The terminal hydroxyls giving rise to band (a) in Fig

2.4, which are present in silica and silica-alumina, as well as in zeolites, are

shifted by 6-8 ppm on adsorption nf pyridine.

This shift arises from hydrogen

bonding between the surface ZSi-OH group and the sorbed pyridine molecule. Interaction of pyridine with the bridging hydroxyls (giving bands (b) and (c) in Fig 2.4) results in severe line broadening of the resonance, corresponding to formation of the pyridinium ion. Chemical shifts in the I5N resonance are also used (39) to probe Lewis adsorption sites and some recent results are given in Table 2.5.

Table 2.5 MASNMR CHEMICAL SHIFTS VERSUS LIQUID F'YRIDINE

IN THE 15N MASNMR SPECTRA OF

PYRIDINE SORBED ON ZEOLITES (39)

Adsorption site

Observed Chemical Shift/ppm

Na'

-26 2 1

Bronsted

-115

Lewis

-50 to -80

2 2

1.2.4.3 Other Spectroscopic methods Electron spin resonance (esr) has been used, particularly to examine Lewis acidity (40). An early explanation for superacidity in zeolites was also based

on esr results (41) but confirmation for the species proposed is lacking.

More

recently (42) external and internal pore sites have been probed using esr signals associated with specific radicals formed by interaction of sorbates and surface sites.

Visible/uv

spectroscopy provides a means to probe charge

transfer complexes formed between appropriate sorbate molecules and surface

26

sites.

The OIs binding energy of metal oxides, determined by XPS, is a l s o

proposed as a measure of the basic strength of oxides, and XPS studies of sorbed nyridine(NIS) can be used to recognise Bronsted and Lewis acid sites (42).

1.2.5 Sites for Acid Catalysis in Zeolites

Catalytic sites in zeolites may

involve separate phases as discussed

subsequently. In the absence of other phases zeolites can act as both basic and acidic catalysts and free radical processes also occur in zeolites.

However,

acidic catalysis is the most widely studied aspect of zeolite catalysts. The hydroxyls within the channels provide the Bronsted sites which are usually generated from the ammonium exchanged zeolites by calcination. Na(Z) + NH +(as)4

NH4(Z) + Na+(aq)

Hydrogen forms of zeolites (HZ) may also be made by direct exchange of Na+ for H+ using mineral acids, providing the structure is stable in acid, but some

dealumination is to be expected especially at surfaces (47). Brnnsted sites are also generated by hydrolytic processes involving water coordinated to polyvalent counter ions

and by appropriate reduction (18). In true hydrogen forms of zeolites these hydroxyls may be regarded as protonic bridges bound to negatively charged framework oxygens associated with AIOq-

tetrahedra. A t higher temperatures (T>200

mobile moving between sites and at T>550

O C

O C )

Bronsted hydrogens can be

they may be lost by dehydroxylation

which is accompanied by an increase in Lewis acidity.

27

b

Bronsted Sites

o

b

0

h

h

h

h

h

h

h

h

A

h

A

0

Lewis Sites

"True" 0

Lewis

(AtW 0 0 0 0 \si/ \i/ \si/ h h h h

\si/

Sites

c

A first stage in dehydroxylation was assumed to result in a vacancy with tricoordinated framework A1 and Si+ species (48a). However, direct evidence for these tricoordinated framework atoms is not available and, in view of the fact that dehydroxylation appears to occur with dislodgement of aluminium from the framework, the current view is that dislodged aluminium species provide the source of Lewis acidity, a point made initially on the basis of x-ray fluorescence results (48b) and subsequently supported by infrared results (48c).

However, the situation is more complicated than is implied by

(C)

above. Additionally, there is some evidence for the existence of a small number of very strong acid sites (superacids). Early investigations using epr (44) lead to the suggestion that superacid sites arise from inductive effects on Bronsted sites generated by neighbouring Lewis sites.

Superacid O\ Site

,O\ Si h

?\,

,O\

A1 I Si h i h

A! hrCI

However, doubt about the existence of

FA1

h and

A

h

-

+Sic

has lead to

suggestions that superacidity arises (49) from synergism between dislodged aluminium species and framework hydroxyls.

TO, (OH),.

. . (A103p

28

The above classification is somewhat simplistic particularly with regard to the acid sites associated with extraframework aluminium species and with defects. Further discussion is reserved to later sections.

1.2.6 Zeolite Acidity - Relationship to Composition and Structure 1.2.6.1 General Considerations

The acid strength of zeolites can be measured using the techniques outlined earlier.

Infrared studies, using self-supporting discs reveal at least two

types of surface hydroxyl in the hydrogen f o r m s of zeolites. A band around 3740 -1 cm , which is also observed in silica and silicates, is assigned to terminal hydroxyls located at the outer surfaces of crystals or at defects where the framework is interrupted or when non-lattice hydrated silicieous material is present.

Defect sites where missing tetrahedra are replaced by hydroxyl nests

are particularly evident in silica-rich pentasils ( 5 0 ) and in some dealuminated zeolites ( 5 1 ) .

Bands in the region 3710-3740 cm-l have been assigned to defect

hydroxyls vibrating freely and a broad band centred around 3,400 cm-' is assigned to hydrogen bonded internal silanols ( 5 2 ) .

in H-ZSM-5

Hydroxyl stretches in

the region 3600 to 3650 cm-l, depending on zeolite composition are assigned to the bridged structures

$Al-(OH)-Si<

described earlier.

Bridged hydroxyls

show lower vibration frequencies when they are located in smaller pores ( 6 or 8 rings) presumably due to electrostatic perturbation ( 5 3 ) .

For example in

zeolite Y the bridged hydroxyl located in the small cages gives

5OH

-

3550

ern-'

and that in the large cages, which XRD ( 5 4 ) and neutron diffraction ( 5 5 ) locate

-

mainly on O1 oxygens, vibrates with

vo,

=

3640 cm-'.

Bronsted acidity in

zeolites is largely associated with accessible bridging hydroxyls although some The lower

defect hydroxyls show weak acidity as revealed by ion exchange ( 5 6 ) .

vibrational frequencies of bridged hydroxyls reflects their weaker 0-H bonds and stronger acidity.

Typical infrared spectra of the hydroxyl region of zeolite Y

are shown in Fig 2.3.

-

The decrease in

increase in silicon content of the increased acidity as Si/Al increases.

voH for

the bridged hydroxyl with

zeolite framework

(Fig 2 . 5 )

reflects

Results correlate well with chemical

shifts for lH MASNMR which also reflect acidity (Fig 2 . 5 ) .

Since acidity

strictly involves interaction with a base this point is more evident in the bathochromic shifts arising from interaction of bridged hydroxyls with the weak base benzene

(Fig 4.8, Section

1.4).

Increased bathochromic

shifts, as

intermediate electronegativity increases, reflect increased interaction between

29 zeolitic hydroxyls and benzene resulting from weakening of 0-H bonds and stronger zeolite acidity. Patterns in Figs 2.4 and 4.8 are consistent; acidity of bridged hydroxyls tends to increase with increase in Si/Al.

Many approaches

have been made to explain this observation and these are more fully discussed in Sectionl.4. At this point it iS convenient to recall that the effective acidity of Bronsted sites in faujasitic zeolites can be interpreted simplistically in terms of the proximity of aluminium atoms within the framework. Accepting this point a simple localised model (57) for H-forms of zeolites is,

tl

I

0

(TjO)3Si / \ A1(OSi)3 where T. = T1, T2, T may be, according to Lowensteins rule, either A1 or Si. 3 J This model predicts four types of bridged hydroxyl involving 1, 2, or 3 second neighbour aluminiums, the number of each type depending upon Si/Al.

Recently

(35a), hydroxyl stretch assignments have been made to at least four types of supercage hydroxyls in zeolite H-Y by numerical deconvolution of the high frequency bridged hydroxyl band (Fig 2.6a) and similar results are obtained by deconvolution (35b) using resolution enchancement (Fig 2.6b).

A consequence of this simple localised model is that, for an ordered three dimensional framework structure, there is only one kind of hydroxyl unit at Si/A1 2 7 .

Consequently the properties of zeolitic Bronsted sites should become

independent of composition at higher values of Si/Al.

It is of course important

to consider the distribution of A1 as well as the ratio Si/A1 but the notion that the acidic properties of H-forms of zeolites reach a limiting value as Si/A1 increases, although controversial theoretically as discussed subsequently, is supported by much experimental evidence (58).

results support this view (41a).

Recent infrared and H'

MASNMR

Although the effect is exaggerated in Fig 2.5,

and similar plots where Si/A1 is used as the variable, it seems that a limiting and is reached when S i / A l Z 10. value for H Structural differences between zeolites having the same framework

vo,

6

composition may also affect acidity. Preferential siting of A1 atoms may differ between structures so that A1-A1 distances are not the same at a given Si/AI. Additionally the hydroxyl stretch for bridged hydroxyls depends upon the bond angle /SiOAl which can vary within and between structures. Consequently the is not likely to be limiting ratio Si/A1 for zeolite properties, including

voH

30

y 0.80 z a

m CY

g 048

rn 4

0.16 I

3700

3600

3500

3400

3680

I

I

3640

WAVENUMBERArn-'

I

3600

I

I

I

I

I

3520 WAVENUMBER /cm-l

3560

Fig.26. Infrared spectra of hydroxyl region of H-Y zeolite. (a1Numerical deconvolut ion. (Ref.35a). (b1Deconvolution by signal enhancement. (Ref. UMISTI.

0

200

400

600

800

TEMPERATUREC'/

Fig.2.7. TPDA from NHI, mordenite degassed at 2OoC; heating rate 4°C/min.(ref.UMIST).

31

the same for all framework structures. Both Bronsted and Lewis sites in zeolites have been probed by infrared studies (33) of sorbed bases. 1450 cm-'

Pyridine is widely used, and bands at 1540 and

which are assigned to pyridinium ions and coordinatively bound

pyridine respectively (Table 2.2) are used to monitor Bronsted and Lewis sites. It is reported (59) that, over a range of composition (Si/Al), only Bronsted sites are present in carefully prepared H-ZSM-5.

However, preparation of

silica-rich faujasites by chemical or hydrothermal methods can result in the presence of non-framework ions or dislodged aluminium species which could act as centres of Lewis acidity.

Sorbed bases are also used in conjunction with I5N

NMR to examine Bronsted and Lewis sites in zeolites.

The number of pyridinium

ions per cavity can be determined using I5N NMR of sorbed pyridine (41b) and measurements using acetonitrile show that the electron acceptor strength of hydrothermally treated Y zeolite increases with temperature of activation and water pressure (ie severity of the hydrothermal process)(4lb). In general the dependence of acidity on framework composition of zeolites as revealed by spectroscopic investigations (eg IR and H ' sorption/desorption studies using basic sorbates.

NMR) is supported by

The temperature programmed

desorption of ammonia TPDA is widely used in work on zeolites.

Ammonia is

preferred because of its stability during TPD and because of its size.

Other

bases can be used in TPD but difficulties associated with decomposition during heating and with accessibility can arise (60). Even with ammonia it is important to use a detector which detects ammonia rather than changes in pressure, thermal conductivity or weight loss etc.

Fig 2.7 shows a TPDA profile for hydrated

NH -mordenite.

Mass 16 gives NH3 in the presence of H 0. Reports based on 4 2 indirect methods sometimes wrongly assign the high temperature dehydroxylation

peak to very strong acid sites (61). When the zeolite (NH4- or H-

form) is first outgassed at

temperatures and then cooled to sorb NH3 prior to the TPDA temperature

peak

due

to

quantification difficult.

physisorbed

and

weakly

chemisorbed

elevated

run, a NH3

low

makes

This low temperature peak is minimised by sorbing

ammonia at elevated temperatures, usually around 1 0 0 O C . forms of faujasites the sorption of ammonia at T >lOO°C

However, in cation

can lead to trapping of

NH3 in small cages which is then desorbed at higher temperatures along with ammonia from acid sites. The broad, frequently asymmetric peaks, associated with chemisorbed ammonia are not readily deconvoluted

so

as to provide clear evidence for different types

of Bronsted site in zeolites (62).

However, one report (63) presents results

32 for zeolite Y which give good agreement between the acid site distribution

determined from TPDA and that predicted by calculation based on a random siting of Al, subject to Lowenstein's rule (Fig 2.8). The pattern observed is broadly consistent with that obtained by deconvolution of the infrared spectra ( 3 5 ) discussed earlier.

The distribution of A1 in faujasitic frameworks is discussed

in Section 1.4. Considerable effort has been put into TPDA studies of ( N H /H)-ZSM-5. 4

Differences in reported TPDA profiles presumably

represent differences

in

experimental technique and in sample composition or quality. A high temperature peak

( w

400 OC) is assigned to ammonia chemisorbed on acid

sites, mainly

Bronsted sites in ZSM-5, and lower temperature peaks to ammonia sorbed on weakly acid internal silanols (64) or on cations or, at T < 15OoC, to specifically" sorbed ammonia (65).

Measurements on NH -2SM-5 4

"less

rather than on

HZSM-5 on which ammonia has subsequently been adsorbed show clear evidence for

only one TPDA peak at approx.400

O C

(66).

TPDA peaks associated with NH3 chemisorbed on acidic sites span a wide temperature range indicating an acidity spectrum, in agreement with measures of acidity using microcalorimetry ( 6 7 ) , conventional sorption ( 2 7 ) and titration procedures (68). Additionally, activation energies calculated from TPDA results on ZSM-5

(65)

enthalpies.

agree well with

microcalometric measurements

of

desorption

TPDA has been used to demonstrate the effect of framework Si/Al on

the acidity spectrum ( 6 9 ) .

It should be emphasised, however, that TPD studies

although useful in characterisation of zeolite acidity do not differentiate Bronsted and Lewis acidity.

A combination of TPDA and infrared, for example,

can resolve this point. Energy changes associated with the interaction of bases with zeolite surfaces may be estimated from calorimetric measurements of differential heats of adsorption (q) as a function of surface coverage (70). On the acid form of mordenite (H-M)

and on H-M

dealuminated with H C 1 very strong Lewis sites ( q =

170 kJ mol-I), on which ammonia is sorbed dissociatively, and weaker Lewis sites (q = 100 kJ mol-l) where ammonia is undissociated are observed.

Four Bronsted

sites are assigned to q values of 120 2 5, 130 2 5, 140 2 5 and 150 2 5 kJ mol-l.

The strongest Bronsted sites (q

= 150

5 kJ m o 1 - l )

are found on acid

leached mordenite (Si/A1 = 10 and 27.5) or on synthetic mordenite (Si/Al

=

10).

Since only the acid leached samples have the strongest Lewis sites and the synthetic materials had no strong Lewis sites it appears that the strongest Bronsted sites observed donotrequirethe presence of strong Lewis sites,

33

The above discussion covers some spectroscopic and TPDA results in relation to framework Si/Al.

It should be clear that the dependence of acidity on

framework Si/A1 is not always straightforward even within a given structural type.

For example, in the mordenite and ZSM-5 frameworks there is evidence of

preferential siting of A1 which might be expected to influence acidity so that acidity per site could change as Si/A1 is reduced even though Si/Al

3

7.

Moreover, dislodgement of aluminium should be avoided during any pretreatment prior to acidity measurement since the loss of framework aluminium can result in a reduced number of acid sites and changes in acidity (49). It is abundantly clear, therefore, that the acidity spectrum in zeolites is related to framework composition.

The distribution of aluminium within the

framework and the presence of other species, ions, dislodged aluminiums etc.can also influence acidity.

Consequently it is necessary to discuss some of these

more common modifications.

1.2.6.2 Modification of Zeolite Catalysts

Typical modifications to zeolite catalysts include ion-exchange; thermal/ hydrothermal treatment; dealumination by extraction of aluminium with acids or complexing agents; modification by secondary synthesis in which framework 'T' atoms are replaced by other 'T' atoms (isomorphnus substitution); incorporation of additional phases, usually metals, metal oxides or salts.

In what follows

consideration will be given only to modifications which affect acidity. include

ion

exchange;

dealumination;

hydrothermal

treatment;

These

secondary

synthesis; incorporation of oxides or halides.

(a) Modification by Ion Exchange In well-crystalline zeolites containing no other phases the effect of counter ions on Bronsted acidity is well established.

Ion exchange of Na+ for

polyvalent ions results in generation of Bronsted sites by a hydrolytic process, for example

M*+(H~o)-

M+(OH) + H+

The H+ neutralising a framework A104- tecrahedra constitutes the Bronsted site. Acidity is enhanoed by cation polarisation, proportional to e/r for the cation ( 3 3 ) , and by increases in the framework ratio Si/A1.

34

5 0.6W

N

-

e . u

$0.4E.

z

0

-

2 0.2I-

<

z

0

-

t

r <

OO

200

400 TEtlPERATURE/*C

Fig.2.8.Schematic TPDA profile of NHLY zeolite (!iOAl/UC) and estimated numbers of types of Al atom (Nil.(ref.63).

(060 .dnlatliu N spckr

C

416 L blZS N# L acid-OH

Fig-210-Diffuse ref'ectance I study of thermally modified HZSM-5. (Ref. 36).

Fig.Z.PLa).n-kxane cra&ii Over tiUfl-5 steamed at W fa 2.5hwrs.lRef.771. o----oFOmin on stream. n-----oZC-Omin. on stream. 0 parent zeolite. *-----raM d dislodged Al to framewrk kl. parent zedik.

, /

C

,

1 2 3 L A H W T ff 16HASS PEAU AT 4COY larb unihl

Fig29(b).Rate of n-kxaw cracking (at 28SoC1 Over steamed H-ZSM-5 (steam treatment at 600°C for 25hoursl. (Ref.771.

35 For the alkali metal ions acidity increases in the order Cs-Li

higher for Y than X zeolites (Si/Al(Y)

> Si/Al(X)).

and is

However, alkali metal

zeolites are much less acidic than H-forms or polyvalent cationic forms and in fact the high aluminium content X zeolites counterbalanced by heavier Ia cations can function as basic catalysts (71). The acidic forms of zeolites are usually produced by ion exchange of sodium forms, for example exchange with NH4+ is used for H- forms.

In zeolites

there is a distribution of acid sites and there is evidence, mainly catalytic, that the stronger acid sites retain Na+ ions more strongly than weaker sites. For example, in zeolite Y evidence for strong acidity was not obtained (72) until 30% of the Na+ ions were exchanged for H+ ions.

Hence residual Na+ ions

in H+ forms of zeolites can have a disproportionate effect on catalysis of demanding reactions. Recent work (70) assigns three types of Bronsted site in synthetic H-M (Si/Al

=

5).

Increasing ion exchange from 40 to 90% is reported

to result in increases in both the stronger and the weaker sites and a reduction in the number of intermediate sites. This implies that there may be some redistribution of Na+ ions as exchange progresses.

A further point concerning cations in zeolites is that they are inherently electron acceptor sites and can function as sorption centres and as Lewis acid sites (18).

(b) Modification by Dealumination Dealumination

is

common

to

several processes including hydrothermal

treatment and secondary synthesis, which are discussed subsequently.

First we

consider the extraction of aluminium under conditions where clear evidence for extensive replacement of A1 by

Si is not forthcoming.

Several chemical

treatments are used but extraction of A1 by mineral acids o r complexation with EDTA is most widely reported.

Framework A1 can be extracted by an aqueous

solution of EDTA (73).

If this extraction is made slowly it appears that some replacement of aluminium by silicon takes place since up to 80% of the aluminium can be removed

with

70%

retention

of

crystallinity

(73b)

and

there

is

XRD

crystallographic evidence that occupancy factors remain close to unity after 53% dealumination and 29Si NMR supports some replacement of A1 by Si (51). However, other XRD results suggest that occupancy factors are less than unity so that substitution of Si for A1 on treatment with EDTA is not extensive.

Recent MASNMR, FTIR and sorption results (51) suggest that, as dealumination

36

proceeds

amorphous

silicious material

and

defect

sites are

increasingly

generated and extensive replacement of A1 by Si is not observed and there are no reports of complete dealumination, with retention of crystallinity, using this method (EDTA).

Nevertheless, some "healing" is achieved possibly from

incorporation of soluble silica species or perhaps by migration of defects although this must be slow at the temperatures involved.

Furthermore, it is

clear that EDTA preferentially attacks the outer crystalline surface, even in large pore

(Y) zeolites, resulting in compositional inhomogeneity

However, 27Al

(47).

NMR shows that non framework octahedral aluminium, which can

influence catalytic properties is removed by EDTA (51). Accepting some uncertainity in the extent of healing and the composition of frameworks dealurninated by EDTA it is interesting to consider acidity studies of dealuminated X and Y zeolites. A measure Bronsted acid sites is very

(& )

of the efficiency of

nicely established, using titration/indicator

methods, and shown (68) to depend linearly on the number of aluminium atoms per unit cell (NAl)

a. = 1.40 - 1.45

x

lo-'

NA 1

The effective acidity is zero when NAl

= 96 (Si/A1 = 1) and is unity when N A l= 28 (Si/Al = 6). This is taken to imply that the effective acidity of Bronsted

sites in faujasitic zeolites is constant for Si/Al>6.

Explanations for this

pattern, based on the distribution of aluminiums in zeolite frameworks, have been proposed and are discussed in Section 1.4.

(c) Thermal/Hydrothermal Treatment Heating zeolites in the H- or NH

-

4

forms at temperatures around 45OoC and

higher can result in loss of structural water by dehydroxylation of surface This may be seen for NH -M as a mass 18 peak around 5OO0C 4 (Fig 2.7). Dehydroxylation of the H/NH - zeolite in a shallow bed (SB) with 4 effective removal of water results eventually in structure collapse. On the

hydroxyls.

basis of infrared studies it was proposed that dehydroxylation of H-Y resulted in generation of

positively

charged

silicon and

tricoordinate framework

aluminium which was associated with observed Lewis acidity as discussed above. However, this scheme

requires that an initial ratio of bridged hydroxyls to

four coordinated framework A 1 of unity should decrease as dehydroxylation proceeds since two hydroxyls are lost for the destruction of one tetrahedral A 1

37

framework site

(see 1.2.5 above).

Confirmation for

this scheme was

' forthcoming from X-ray fluorescent analysis (48b) nor from H

not

and 27Al MASNMR

results (73). During SB treatment the number of bridged hydroxyls remained in approximate

1:l

correspondence

with

framework

tetrahedral

A1

(73).

Consequently, it appears that, during SB dehydroxylation, A1 is continually removed from the framework which eventually results in structural collapse. Heating the H/NH -zeolite in a deep bed (DB) without effective removal of 4

water or heating in a gas stream containing sufficient water also results in release of aluminium from the framework, a fact first recognised in the generation of ultrastable zeolite Y

Aluminium

(74).

is

dislodged

by

hydrothermal attack and, at suitably high temperatures, silica is sufficiently soluble and mobile in the presence of steam, to substitute in the vacant 'TI site (74d) resulting in structure healing.

5'

7

P Si-OH

HO-Si

Si(OHI4

P

-

0

I

S i -0-9-0-9

Si

0

Si

The healing process can be aided by migration of vacancies in a 'T' jump mechanism

(74d).

Substitution of silicon

for

aluminium

in

the

zeolite

framework reduces the number of bridged Bronsted sites and in the case of zeolites initially rich in aluminium (eg Y ) , increases their effective acidity (cf EDTA dealumination). The above is, however, oversimplistic particularly in representing the dislodged aluminium as A1(OH)3. steaming Na/H or Na/NH4

-

Investigations using 27Al MASNMR show that

Y zeolites results in dislodgement of some aluminium

octahedrally coordinated to oxygen (chemical shift approx 0 ppm from A1

3+ (aq).

More extensive steaming results in a broad 27Al signal around 50 ppm assigned to low-symmetry oxyaluminium species probably polymeric at least some of which can be readily complexed by acetyl acetone (75a). reported that broad peak around 50 ppm from A1

3+

More recently (76) it is

(as) is due to tetrahedral A 1

with a peak shift of 30 ppm due to quadrupole effects.

2 7 ~ 1 NMR results

suggest that aluminium species five coordinated to oxygen can also be produced during steaming (76~).

Detailed X-ray

radial distribution analyses are

interpreted in terms of half unit cells of a boehmite phase which are said to be generated in faujasitic supercages during extensive steaming ( 74e ) .

38

Additionally there is clear evidence that at around 6OO0C the aluminium is dislodged by attack throughout the crystals in a homogeneous manner but that at elevated temperatures, in the presence of steam, migration of aluminium to outer crystalline surfaces takes place (47). available

evidence

it

is

clear

that

the

Taking into account all the nature

and

distribution

of

non-framework aluminium (within and between crystals) depends upon the initial composition (particularly Si/Al, Na/NH ) and structure of the zeolite and the 4

severity of the hydrothermal process (T, t, pH20) (77). aluminium species can modify dCidity.

Moreover, dislodged

In particular there is evidence from

both catalytic and ammonia desorption studies that increased acidity and enhanced catalytic activity can be generated by the presence of appropriate oxyaluminium species interacting with framework Bronsted sites (49)(77)(78) (Fig 2.9).

However, the nature of sites having enhanced activity and acidity

is not yet clear (Section 1 . 4 and Chapter 71.

Infrared studies of hydrothermally treated zeolites show additional bands around 3600 and 3700 cm-'

which are weakly or non-acidic and are assigned to

A1-0-H units associated with dislodged A1 species.

A line in the H '

NMR

spectrum at 2.6 ppm (from TMS) is assigned to these A1-OH units (79). Infrared investigations using sorbed hydrogen have provided a new look at active sites in zeolites and other acid catalysts (36). Comparison of bands in the hydroxyl region (3000

-

3800 cm-')

with bands due to sorbed hydrogen in the near

infrared and observing the effects of heat treatment suggests that bands due to interaction (Table 2.3).

of

molecular

hydrogen

and

Lewis

sites

Typical results are shown in Fig 2.10.

can

be

identified

After heat treatment of

HZSM-5,in addition to those assigned to interaction with bridged hydroxyls (4105 cm-')

and silanols (4125 cm-l), three bands due to sorbed hydrogen are

observed at 4060, 4010 and 4030 cm-'.

The band at 4060 cm-'

is seen only after

heating the zeolite at 1270 K when a band in the hydroxyl region is clearly seen at 3680 cm-' aluminium.

and assigned to AlOH species associated with dislodged

The band at 4060 cm-'

Bands at 4010 cm-'

and 4030 cm-'

Lewis acids ?A1

and

the framework.

+Sif

is, therefore, assigned to H2 sorbed on AlOH.

having comparable intensities are assigned to associated with tricoordinate A1 and Si atoms in

Deep bed calcination of Y zeolite followed by heating and

evacuation to achieve complete dehydroxylation gives bands at 4035, 4060 and 4125 cm-l.

The band at 4060 cm-l is now major and since extensive nnn

framework aluminium is present this result is taken as confirmation that the 4060 cm-'

band arises from the interaction of hydrogen and AlOH species and

bands at 4035 cm-l and 4010 cm-l arise from interactions of hydrogen with

39 tricoordinated silicon and aluminium. The presence of a band at 4035 cm-' hydrogen sorbed on

silica gel, previously

assignment of this band to

SSi'

heated

at

for

97OoC, results

and the band at 4010 cm-'

in

to hydrogen

interacting with S A l . In view of the previous discussion concerning H ' these assignments can probably be challenged. aluminium

in

zeolites

is

not

known

in

thermal/hydrothermal treatments frequently amorphous silica. available

in

and 27Al MASNMR results

The nature of the dislodged

all

cases

result

in

(see

earlier)

generation

of

and some

Consequently it is not yet certain what kinds of site are

dislodged

interaction with hydrogen.

material,

particularly

aluminium

species,

for

Currently NMR evidence is quoted for the existence

of aluminium in at least 3 types of coordination to oxygen (four, five and six) so that more than one type of Lewis site associated with dislodged material

must be a possibility and interpretation may not require the existence of framework tricoordinated A1 and Si.

A further point in relation to pentasils

is the fact that internal silanol groups might provide sites for H2 (weak Bronsted interaction). However, it has to be recognised that MASNMR is largely a bulk technique so that the absence of evidence for trigonal aluminium and silicon at oxygen defect sites cannot be taken to preclude such sites in small concentration.

In fact the hydrogen sorption work does suggest that only a few

percent of the total A1 in H-Y is in tricoordinate framework sites.

The

suggestion that such sites are major in dehydroxylated H-ZSM-5 and H-mordenite (36) is most interesting but remains to be established by other techniques.

Dehydroxylation of both of these zeolites results, as for other zeolites, in

loss of framework aluminium and it is not immediately clear why framework sites should be stabilised in these zeolites.

trigonal

Moreover there is no

doubt that under steaming conditions aluminium is dislodged from the framework of both

H-ZSM-5

and

H-M

and, in

the

presence

of

steam, migrates

to

outersurfaces as in the case of H-Y ( 4 7 ) ( 8 0 ) . At this time the detailed nature of Lewis sites and their function in hydrothermally treated zeolites is incompletely understood. (d) Secondary Synthesis Secondary synthesis involves the replacement of framework 'TI atoms by 'TI atoms from a separate source.

A multiplicity of T atom substitutions is

reported (81) but here we consider, briefly, the substitution of Si for A1 to generate silica-rich zeolites and the incorporation of A 1 to activate zeolites rich in silica.

Most reports concern the modification of faujasitic zeolites

40 using SiFg2-(aq) (82) or SiC14(g) ( 8 3 ) . (a) NH4 A102(Si0 ) 2 x (b) M l I n A102(Si02)x

+ (NH

) SiF6 = 4 2

+ SiC14

=

(NH4)3A1F6 +

(Si02)x

l/nMCln + A1C13 +

+

(Si02)x

+

Both of these processes have received much attention recently.

Evidence,

mostly based on MASNMR, IR and sorption studies demonstrates that isomorphous replacement of A1 by Si in faujasitic frameworks can take place with generation of few defect sites. A1F6 3-)

Contamination of the zeolite by aluminium fluorides (eg

species may occur in process

catalytic properties

(51).

(a) and such species can

However, they

can be

affect

completely removed

by

effective washing (51). Similarly, occluded aluminium species can result from reaction with SiC14 as evidenced by 27Al

MASNMR (84) but the quality of the

product depends, as in (a), on process conditions (85).

In principle these

secondary syntheses techniques do provide the possibility

for generating

zeolites with framework compositions outside the normal range of primary synthesis and with more control over non-framework species, hydroxyl nests and secondary pores than is the case with typical hydrothermal treatments. The reverse of these processes, substituting A1

for Si

in

zeolite

frameworks in order to increase the number of acid sites is also widely reported (16 ) .

In some instances the process largely involves condensation of

hydrated aluminium species into nested hydroxyl vacancies. (e) Modification by Incorporation of oxides, metal salts or by halogen treatment A series of papers and patents demonstrate that the catalytic activity and selectivity of zeolites can be modified by incorporation of oxides, usually by impregnation from solution (84). In particular oxides of B, P and Mg have been used to change acidity and diffusivity, and for example in the case of phosphorus oxide modification of Bronsted acidity is presumed to involve interaction of bridged framework hydroxyls and hydrated oxide species. This is reflected in a decrease in the number of strong Bronsted sites and an increase in the number of weak

sites

.

(88).

Depending upon mode of

incorporation the phosphorus can be concentrated onto outer dispersed more evenly (47c).

surfaces or

A wide range of techniques has been

monitor changes in acidity due to incorporation of oxides.

used to

41

The incorporation of metal halides into solid acids is reported to generate superacid sites (89) and the acidity of zeolites is enhanced by incorporation of aluminium fluorides (90)(51) or by treatment with fluorine (91).

Superacidity is established by

indicator methods, by

TPDA and by

catalytic testing for modified H- mordenite (92). Vapour phase deposition of volatile compounds of silica such as SiH4 (93) a Si(OR)4

(94) or of volatile halides ( 8 6 ) is also used to modify sorptive

and/or acidic properties of zeolites.

1.3

ACID CATALYSIS IN HYDROCARBON TRANSFORMATIONS

1.3.1 Carbocation reaction intermediates in solution

In the case of acid oxides generally, and particularly in the case of zeolites the bulk of the experimental evidence, most noticeably that based on reaction

product

intermediates.

distribution,

supports

reactions

via

carbocation

The initial steps leading to carbocation formation are not in

all cases the same nor is there always a consensus view about this.

These

initial steps are discussed subsequently. 2

The main hydrocarbon transformations proceeding via carbeniurn ions (sp hybrid bonding and empty p orbital at the charged carbon centre) include the following (SCHEME 1).

SCHEME 1

2

Hydride Abstraction

RIH + R 2 +;=R1+

+ R2H

(E)

42 3

Rearrangements

a ) Classical 1,2 shifts of Alkyl or H involving non-branching rearrangement

C

C + I

-Me:

I +

c-c-c-c

c-c-c-c -H:

C

==

I +

c-c-c-c I 1 H H

C H

I I

c-c-c-c + I

H

b ) Branching rearrangement.

-

postulated

The cyclopropane intermediate is frequently

particularly to avoid reaction via primary carbenium ions.

c c

C I

* c-c-c-c I

c-c-c-c-c

I

C C H+ \ fY

c-c-c

/

C C 1eavage

4

- c-c+cf 1:

+

4

I'

'P

a 5

Addition to olefins (repetition leads to polymerization)

C I

C-C-C

+

C

+

I

C=C-R

c

c

I 1 C-C-C-C-R

' +

C

and similar addition to arenes

F

43 These reactions are the basis for the typical hydrocarbon transformations; isomerization, cracking, polymerization and alkylation which take place over solid acids.

For example, conjunct

Several steps may occur in sequence.

polymerization involves, polymerization, isomerization, cyclization and hydrogen transfer. The properties of carbenium ions have been established largely by studies in homogeneous acidic media.

Stabilities of carbenium ions decrease in the

order: tertiary ie

> secondary > primary

R3C+

R~CH+

with approximate energy 1 changes (AElkcal mol- )

RCH~+

11-15

20

These differences in stability are reflected in reaction rates and hence in product distributions. An example (95b) is given is SCHEME 2.

SCHEME 2

a)

c c

CH3 :

I I

==

C-C-C-C

+ C' C H C I l l

b)

C-C-C-C-C

;A+

n:

=

c c I I

C-C-C-C

2..

C H C I I I C-C-C-C-C

k/s-'

=

lo7

8

- 10 at

-120OC

c c c

CH3:

I l l

it

C-C-C-C-C

k/s-'

= 5 x

at - 8 8 ° C

+ n~ "

I

LA+H

In reaction (a) the initial and final ions are tertiary whereas in (b) the interconversion of 2 , 4 , 4 and 2 , 3 , 4 trimethylpentyl ions proceeds via a secondary carbenium ion.

The difference in stability between the secondary and tertiary

ions results in a high activation energy and much lower rate.

For reactions in acidic solutions, extensive work has established broad correlations between acid strength and the type of reaction which may be effectively catalysed. Typical examples are included in Table 3.1 (95a). Variation in temperature and in reactant type can result in a effective function of weaker acids.

more

The effect of temperature is particularly

appropriate to reactions over solid acids.

Weaker solid acids can be effective

44

catalysts, in part, because reactions can be made at higher temperatures. From Table 3.1 it is clear that product distributions from acid-catalysed hydrocarbon transformations depend upon acid strength.

For example in the

isomerization of butenes (SCHEME 3) the double bond shift, involving only proton transfers is catalysed by weak acids whereas methide shift requires stronger

In the case of butene isomerization the methide shift requires a change

acids.

in the nature of the carbenium ion resulting in an activation energy increased by the corresponding changes in carbenium ion stability.

SCHEME 3

demanding

c

L

t The demanding rearrangement to isobutene appears to proceed via a cyclopropane intermediate ion (cf scheme 1, 3bl.

Use of a stronger acid to promote methide

shifts can, of course, promote other reactions, for example polymerization of butenes so that product distribution will be a function of acid strength and also of reaction conditions.

Selectivity in acid catalysis is therefore

dependent, among other factors, on acid strength.

An example

illustrates this for alkylation of m-xylene (95a).

SCHEME 4

ACID

A1C13/HC1

CATALYST

(STRONG)

BF3/H3P04

(WEAK)

99% 30%

1%

70%

(SCHEME

4)

45 Superacids provide a direct route to carbenium ions via carbonium ions which are presumed (96) to be intermediates or transition states possessing two-electron three-centre bonds (SCHEME 5)

SCHEME 5

a)

Protolytic Attack on C-H

(2)

;c+ +

b)

H2

Protolytic Attack on C-C

Reactions proceeding via (a(2))

result in hydrogen generation, the reverse

reaction being reduction of a carbenium ion with molecular hydrogen.

Reactions

proceeding via route (b) produce a carbenium ion and a smaller alkane.

The

reverse reaction is similar to that in SCHEME 1 (reaction 2) but with formation of a new C-C bond.

In the presence of superacids, alkanes including CH4 and C2H6 are reactive (20)

+ -

0

-C-C= /

+

C H

2 6

----c

However, factors other

nC4H10

than acid

strength

also

influence selectivity in

hydrocarbon transformations catalysed by acidic solutions. F o r example, product isomers from the alkylation of alkyl benzenes in solution depend upon:

(i) the

46

strength and amount of acid catalyst; (ii) the temperature and duration of the reaction; (iii) the type of alkyl group(s) on the benzene ring

and

reactivity of the substituting alkyl group;

(95a).

Reactivity

correlations

(97) have

been

(iv) steric established

factors for

the

electrophilic

substitution of alkyl benzenes in solution and, in the absence of soatial constraints these also hold in solid acids and zeolites (98). The factors, discussed above for solutions, also affect activity and selectivity in hydrocarbon transformations over solid acids.

Catalytic activity

and selectivity are dependent upon reaction conditions and can be influenced by acid

site

distribution

(type, strength, concentration

and

proximity).

Additionally, especially in microporous solids, spatial restrictions can modify activity/selectivity patterns, producing shape-selective catalysis.

In all

cases thermodynamic considerations must be recognised when evaluating catalyst selectivities.

For example production of isobutene (SCHEME 3) is not favoured

under equilibrium conditions at higher temperatures.

Table 3.1

Acid

Strength

HCOOH

weak

-n 2.2

Reactions Catalysed

Olefin reactions ( e g double bond shifts)

H2S04

strong

11.0

Olefin

reactions;

rearrangements hydride

of

shifts,

skeletal alkanes; conjunct

polymerization; alkylation

SbF5/HF

superacid

24.0

Catalyses the above reactions and acidity is

sufficiently

strong

to

attack

alkanes

to

form

ions

stable carbonium

47 1 . 3 . 2 Factors Affecting the Activity of Zeolites in Acid Catalysis

Zeolite catalysts have been very

extensively investigated, and what

follows is an attempt to provide some insight into the subject by use of selected studies rather than to provide a comprehensive review.

Several

excellent reviews are available (18)(19)(99)(100). Catalysis is generally considered in terms of total feed conversion and of product distribution.

In hydrocarbon transformations over zeolites these

parameters depend upon the chemical composition of the feed, the structure and framework composition of the zeolite, the nature of any non-framework species and the catalytic process conditions.

Since zeolite catalysts frequently tend

to deactivate readily, in association with coke build up, process conditions must include time-on-stream, in addition to reaction temperature, pressure, contact time, mixing regime and catalyst activation. studies activation procedures should be considered.

In any comparative

Fig 3.1 shows the effect

of temperature of activation on the type of acid site (Bronsted vs Lewis) and of catalytic activity.

These results suggest that xylene isomerization takes

place largely on zeolitic Bronsted sites so that rates are reduced after activation procedures which cause dehydroxylation.

0 4 1

16 -

- 30

12 Bronsted acidity

8,

. -20 Iwiq

I

I

I

I

II

600 700 800 CALCINATION TEMP ,OC Fig. 3.1. Acidity and catalytic activity for 0-xylene isomerisation of magnesium decationised Y as a function of calcination temperature. (ref. 110).

300

400

500

48

Hydrocarbon transformations over acid forms of zeolites, that is those containing bridged hydrogens, result in product distributions which, in many cases are consistent with mechanisms involving carbocation intermediates. Reactions which proceed via protonated intermediates have been widely used as test reactions to assess factors affecting catalytic activity.

Butenes are

readily isomerized by Bronsted acids, via proton addition, with first order kinetics. Over solid acids the intermediate is presumed to lie on the surface with one methyl group pointing away from the surface.

Since the probabillties

associated with loss of hydrogens HA and HB are expected to be equal then,

isomerization of 1-butene should give a 1:l ratio of cis- to trans-2-butene. This is observed for solid acids and also for acidic solutions.

However, a

detailed kinetic study of butenes isomerization over metal sulphates (Mg and

A l ) demonstrates that the stronger acid (A1 2 ( S O4 ) 3 ) stabilises the carbocation more than the weaker acid and also favours conversion, from the common carbocation intermediate, to 2-butenes relative trans-butene

will

isomerise to

give

a

tn

ratio

increases with the acidity of the catalyst.

1-butene.

This means that

cis-2-butene/l-butene

which

This is observed (100) over

zeolites (Fig 3 . 2 ) where acidity is seen to increase with framework S i / A 1 as Bronsted sites become more acidic (and \3 OH decreases), and is also observed in acid solutions. Fig 3.2(a) presents a schematic representation (18) of kinetic features showing the increase in stabilisation of the carbenium ion ( A ( A H ) ) (where 4 H may be thought of as the enthalpy of formation of the carbenium ion) on increasing acid strength.

A simple thermodynamic cycle is proposed

(107)(18) to represent the

enthalpy of formation ( A H ) of a carbocation in a zeolite as

n

3

4 x

s o

m w

I I

I

I

Ib

P

In

I

RELATIVE RATE AT 1509

SURFACE ACTIVATION ENERGY/k Jmol-'

0

L

c--

W-

# -

PRODUCT RATIO Cis-/ l-

energy /

W

Ip

50 where DOH is the heterolytic dissociation energy of the bridged OH bond, P is 2 the proton affini-ty of the (organic) reactant, e /477Eor is the coulombic energy of the carbocation/zeolite ion-pair and

the

final

summation

term

represents the coulombic interaction of the carbocation/zeolite ion pair with other ions within the zeolite cavities at a distance R. and possessing an effective charge

2..

The above equation is incomplete in that other factors

can influence the stability of ions as guests in zeolites and site geometries are not the same in all structures s o that reorganization energies may differ. Nevertheless, for a given reactant in a given zeolite structure with similar cations the equation probably includes the major effects.

Moreover, it is

clear from Fig 3.2(a) that relative changes in the stability of carbenium ions ( A ( A H ) ) can be associated with changes in activation energies via a typical

linear free energy relationship (102) such that

An

attempt

to

correlate

carbenium

ion

stabilities

composition of

faujasitic

zeolites

(Si/Al) is

with

reported

the

framework

(58b)(103).

The

isomerization of cyclopropane is taken as a test reaction and framework composition is varied by steam/acid leaching. so

Cyclopropane is a small molecule

that intercrystalline diffusion is rapid and the isomerization is a first

order

facile

temperatures.

reaction

catalysed

by

Bronsted

sites

at

relatively

low

Utilising results with mass balances better than 90% it is

observed that the activation energy for the surface reaction decreases rapidly as Si/Al increases to around 6 and thereafter remains constant at least until very low aluminium content when there is evidence for an increase (58b)(103) perhaps reflecting a changed mechanism.

Pyridine sorption and

pyridine

poisoning experiments enable turnover numbers to be obtained and these appear to be constant at Si/A1 greater than about 6. A constant turnover number with decreasing number of sites as Si/A1 increases beyond 6 2 1 leads to a linear decrease in rate with decrease in aluminium content for siliceous faujasites, and the rapid decrease in activation energy with increased Si/A1 below Si/Al = 6 results in a rapid increase in rate with decrease in aluminium content for aluminium rich faujasites.

The consequence is a volcano curve for rate as a

function of the aluminium content of faujasites (Fig 3 . 3 ) . This

pattern,

isomerization

also

for holds

changes for

in

more

activation demanding

energy

reactions

in such

cyclopropane as

xylene

isomerization which is also Bronsted catalysed (77) and the volcano curves for

-

L

v)

,

2 0-

7 \

._ 3 -1.->

Cracking of c5,c6 over Y zeolites treated with lNH&)* SiF6

n-Hexane lT=350°C? r

_r..-r

__/._

..-

&

,*’

Y

-- -221

/

1

I

50

1M

RELATlVE INTMSlTY 0 F ” A I - M R

SIGNAL

Fig34Dependence of catalytic a c t i v e on intensity of 2ksigml for Al in tetrahedral cdinah’onlref.661.

1600 1201(0

c

,

,

,,

lT=430°C)

w,‘

1

2

3

4

5 6 fluoride content/w t%

Fg.3.6. lref.511.

70

-.I

100

60

LO

30

activity region

z

8 u

0

&OO

INITIAL Al CONTENT/Wt %

P’

0

0

,“ n-Pentane

.‘,

.-

;-3-

/ -

/.*

0

TOTAL ALUMINIUM IONS /UNIT CELL

Fig.35. Comparison of activity enhancement on steaming (a1 and No of (paired] aluminiums distributed. initially in doubly occupied 4-rings (b) in H-ZSM-5 zeoMes.(R.M. Lago et al.Tokyo 1986). (ref.781.

reactivity ,with linear dependence of rnte on aluminium content for framework cornpositions iMit.h Si/Al>

6 is also observed in n-hexane cracking (58d).

where increased rates are not observed

as

Even

Si/Al increases from 2.5 to 5 or 6

(58e) the linear decrease with aluminium content in more siliceous faujasites is observed. An optimum ratio Si/Al is also observed in other zeolites.

For example

the rate of isomerization of n-pentane over acid extracted mordenites is said (104) to be optimal around Si/A1 = 8.5.

In the above work, and in similar studies, enrichment of silicon is achieved by chemical or hydrothermal treatment which can lead to changes other However

than those associated with framework composition (sections 1.2 a n d 1 . 4 ) .

ZSM-5 can be synthesised over a wide range of compositions having framework

Si/Al from around 15 to

OQ

.

Results for hexane conversion show that, in

carefully prepared and activated H-ZSM-5, rates of cracking are directly proportional to framework aluminium content at least up to Si/A1

=

8.

Good

linear correlation is observed ( 6 6 ) between reaction rate and the intensity of the 27Al MASNMR signal associated with tetrahedral aluminium (Fig 3.4). All

of

these results suggest that the catalytic acitivity of sites

associated with aluminium atoms in zenlitic frameworks reaches a limiting value as Si/Al increases.

At lower levels of framework aluminium the catalytic

activity is directly proportional to the framework aluminium content

so

that

turnover numbers remain constant with further decrease in aluminium content. This is generally true providing that sites are not activated by dislodged aluminium species and providing no other phases, contributing active sites, are present. However, the situation at very low levels of aluminium may not be clear in all cases.

For H-ZSM-5 synthesised with very low levels of aluminium

the linear correlation between catalytic activity and aluminium content appears to hold down to ppm levels of Al.

However, in faujasitic zeolites where

aluminium content must be reduced by dealuminatinn procedures there is some evidence that the correlation may not hold as A1-a 0.

In the isomerization of

cyclopropane activation energies tended to increase at very low levels of aluminium

(103).

This

result, which

requires confirmation using

other

reactions, suggests that either structural features or, perhaps more plausibly, preparation procedures which might result in different types or extent of defect sites may be responsible for differences in reactivity observed at very

For example steadacid leaching can

low levels of aluminium in faujasites. result

in

the

charge

compensation

of

framework

aluminium

non-framework aluminium which would eliminate Bronsted sites.

by

cationic

The theoretical

53 basis for the general dependence of activity on aluminium content is discussed in section 4. The stability of carbenium ions within zeolites, as in solution or gas phase

reactions is of prime importance in acid catalysis, as discussed above.

Recently, carbenium ion stabilites have been correlated with dehydration of alcohols over H-ZSM-5

the ease of

as revealed by TPD studies (105). Results

are used to establish a scale of relative stabilities for oxonium and carbenium ions for C1 to C4 alcohols and the olefinic products of dehydration.

As the

chain length of the alcohol increases the stability of the carbenium ion increases and temperature programmed desorption tends to favour dehydration rather than desorption of the alcohol. The discussion of acid catalysis and composition has so far centred on the H-forms

of

crystalline

zeolites.

Typical

modifications

for

example

incorporation of countervalent metal ions, or of other oxide or metal phases or treatment with steam, etc, can affect acidic properties (section 1.2.5.2) and catalytic properties (section 1 . 3 . 3 . 1 ) . Ion-exchange of Na+ for NH

4

+

creates, on activation, more Bronsted sites

However, in zeolite Y strong acidity and associated

and increases activity.

catalytic activity is not detected until around 30% of the sodium ions are exchanged ( 7 2 ) . effectively

Stronger acid sites are presumed to retain sodium ions more

than

weak

acid

sites

and,

in

demanding

reactions

such

as

hydrocarbon cracking, quite small residues of sodium can affect results (106). Ion exchange of sodium for multivalent ions can also result in generation of Bronsted sites by hydrolytic processes eg

The extent of this, or similar, hydrolytic reactions and the location of the cation and liberated proton are major factors in acid catalysis rather than the simple degree of ion exchange. 2+

around Ca

,

For example removal of the hydration sheath

ions with associated location of Ca2+ in site 1 (hexagonal prims)

of zeolite Y, without generation of CaOH' sites (107).

or H+ species does not produce active

The effect of ion siting is also evident in instances where

catalytic activity is found to increase rapidly after a limiting level of exchange of Na'

for Mn+.

For example the rate of isomerization of butenes

increases after about 25%-30% exchange of Na+ for Co2+ (108),and the rate of isomerization of cyclopropane after about 30% exchange of Na+ for Co2+ in NaX zeolite (109).

54

From a combination of catalytic and sorption studies it appears that the fraction of supercage protons which is catalytically active in faujasitic zeolites depends upon the reaction (18). However, for more demanding reactions such as skeletal isomerization and hydrocarbon cracking it appears that only a relatively small number of the acid sites available in H-Y are involved (110). The lack of a more direct correspondence between the concentration of Bronsted sites and the rate of isomerization of xylene has been explained on this basis (Fig 3.1). The emphasis on hydrolytic generation of Bronsted sites, discussed above, does not preclude a more direct role for cations. enhanced activity with multivalent

In fact, explanations for

ions initially emphasized the lack of

shielding of Mn+ ions, due to the distances between charge compensating A10 4

units in zeolites, which could clearly result in electron acceptor (cation) sites.

Such sites have the capacity for electron pair bonding (Lewis acidity)

or may, in the case of transition metal ions, act as redox centres.

Cations

can, of course act as sorption centres (111) in overall catalytic processes and the charge distribution within zeolites causes electric field gradients which can strongly polarise sorbed molecules (27) thereby facilitating proton or electron transfer processes. The effect of hydrothermal treament of zeolites depends upon the process conditions and on the composition of the initial zeolite.

In

general,

aluminium is dislodged from the framework and may be replaced by silicon. Consequently, the framework is enriched in silicon and this can lead to the generation of stronger acid ("isolated" Al) sites (section 1 . 4 ) when the initial zeolite is aluminium rich, and in all cases results in a reduced number of framework sites.

Additionally, it appears that, in some instances dislodged

aluminium may be involved in the formation of sites of enhanced catalytic activity.

This may be seem from Fig 3.5 which shows results (77) for the mild

steaming of H-ZSM-5.

The maximum in the rate of cracking of n-hexane is

explained by assuming that steaming decreases the number of framework sites but dislodged aluminium proximate to a framework site can enhance its acidity (section 1 . 4 ) .

Hence, there is an initial rise in activity followed by

a

decrease as more extensive steaming depletes framework sites and as dislodged aluminium agglomerates and migrates to outer surface sites.

This pattern for

H-ZSM-5 has been examined in some detail over a range of compositions and results interpreted qualitatively in relation to the siting of aluminium in paired or single sites in four rings ( 7 8 ) .

A role for non-framework aluminium

55 in the generation of stronger acid sites was earlier proposed for faujasitic zeolites ( 4 9 ) . Modification of acid catalysis by incorporation of oxides and salts is discussed briefly, in relation to alkane conversions in section 1.3.3. The role of

oxides

or

metals

which

introduces

a

second

function, for

example

hydrogenation/dehydrogenation are considered in Chapters 6 and 7.

1.3.3Hydrocarbon Transformations over acidic zeolite catalysts Evidence for the role of carbenium ions in hydrocarbon conversion (and related reactions) over acidic zeolites is available from a consideration of product distributions and

there is some more direct evidence for sorbed

carbenium ions. Butenyl carbenium ions are generated from butene sorbed on Si0,/A1,03

and uv studies also show (112) that cyclopentenyl and cyclohexenyl

carbocations are formed from CH OH and olefins on H-ZSM-5.

Subsequently these

3

cyclic carbocations are converted to protonated aromatic and then polyalkyl and polyaromatic cations.

In H-M and in Na/H-M similar cyclic carbenium ions are

produced from allene, cyclopropane, propene, isopropyl alcohol and acetone (113). The rate of formation and oligomerization of carbenium ions is strongly affected by zeolite Bronsted acidity.

Sorption/reaction studies of 2-propanol

and propene are also readily interpreted by assuming that, under appropriate conditions, both sorbates can give a common propenyl carbenium ion. results of a TPD study of a series of alcohols sorbed in H-ZSM-5

Similarly,

are explained

in terms of the stabilities of carbenium ion intermediates which are formed during reaction ( 1 0 5 ) . Among the hydrocarbon transformations, cracking reactions continue to be of

particular

interest

because

of

their

commercial

significance.

temperatures up to around 5OOOC the products from conversion of C

At

alkanes (n>

5) over acidic zeolites are generally consistent with reaction via carbenium ion

intermediates.

Consequently,

product

distributions

can

usually

be

explained, taking into account thermodynamic and shape-selective effects, on the basis of the reaction types listed in section 1 . 3 . 1

(Schemes 1 to 5 ) .

However, there is a continued interest in the generation o f the initial carbenium ions.

Traces of olefins in an alkane feed can be readily protonated

to carbenium ions

56

and the addition of traces of olefins can increase reaction rates.

Lewis acids (L) may also produce carbenium ions by hydride abstraction

Additionally, in so far as zeolites can behave as solid superacids there is the capacity

for

direct

measurements of H

protonation

(section 1.3.1).

Despite

the

fact

that

and similar acidity parameters do not place unmodified

zeolites in the superacid category there is growing evidence that they may behave as such at elevated temperatures.

For example, in a reaction sequence

involving only an initial (p-scission) cracking of a carbeniurn (R')

generated

either from an olefin or via a Lewis site (a,b above) the products consist of

- -

an olefin and a smaller carbenium ion (R,')

which can abstract a hydride from

the alkane to continue the process.

(c) monomolecular

8

-Q

R1Z

(d) bimolecular

-+-

@

RH

RlZ

-

-+-

RZ

+

olefin

Q +

alkane

This sequence suggests that the initial ratio of alkane/olefin is unity. Alternatively, if the initial carbenium ion is generatcd by direct protonation the sequence may be written as follows, where attention is focussed on a

57

particular bond C-R' ( R '

=

H or R =

(e)monomolecular 3C-R'

+

alkyl) in the alkane

HZ

-C-R')

(

-

a3

carbonium ion

+ R'H

alkane

;

alkyl

carbenium ion

@-scission

?C+

3

R= [F; R=H

[smaller) carbenium ion+olefin

The (smaller) carbenium ion, which does not crack readily, is available to continue the sequence by typical bimolecular hydride abstraction (reaction (d)) to produce a carbenium ion from the reactant alkane and a smaller alkane. Consequently,

reaction

alkane/olefin >l.

via

(e) and

(d)

produces,

initially,

a

ratio

At temperatures in the range 673-743 K, results suggest

that n-heptane is cracked over LaHY zeolite to give a primary product ratio 1 and since no hydrogen is detected a protolytic attack on a C-C

alkane/olefin

bond is proposed as the first step (114).

MIND0 calculations suggest t'ne

formation of a hydrogen bridged intermediate.

It is then energetically

favourable for this intermediate to crack directly (activation energy 12-18 kcal mol-l) rather than convert to other structures (114).

Whereas

Lewis

sites

are

available

in

LaHY

and

may

influence

product

distribution their involvement in conversion over H-ZSM-5 is less likely since pyridine sorption shows no evidence for Lewis-sites ( 5 9 ) on well-crystalline H-ZSM-5 (although it must be admitted these may be generated during activation procedures).

The reaction of C6 alkanes over H-ZSM-5 at T in the range 623-823

K adds support to a direct protonation mechanism.

As temperature increases,

product distributions suggest that there is an increasing contribution from the monomolecular reaction (e) over the bimolecular reaction (d).

In this work

p r o t o n a t i o n of C-H, used

to

cracking

explain rate

the

t o produce H

a s a p r o d u c t , i s proposed and r e s u l t s a r e 2 d e c r e a s e i n c o n s t r a i n t index ( r a t i o of t h e

observed

constants

for

n-hexane

and

3-Me-pentane)

with

increase

in

t e m p e r a t u r e (115). S i m i l a r l y , i n t h e i n i t i a l s t a g e s 05 r e a c t i o n , a t low c o n v e r s i o n , propane

is s a i d t o react o v e r H-ZSM-5

by p r o t o n a t i o n of t h e C-C

bond t o g i v e methane

and a carbenium i o n a l t h o u g h e n e r g e t i c a l l y t h i s must be u n f a v o u r a b l e ( 1 1 6 ) .

A t h i g h e r l e v e l s o f c o n v e r s i o n i r ; i s proposed t h a t propane i s a c t i v a t e d v i a

h y d r i d e a b s t r a c t i o n by s o r b e d carbenium i o n s ( ( d ) a b o v e ) . Energetic parameters associated with -1-ene

and 2-methylpent-2-ene

associated

with

o v e r H-USY

rearrangements

of

the

the

isomei-ization o f

2-methylpent

a r e found (117) t o be similar t o t h o s e same

(postulated)

carbocations

in

super-acid s o l u t i o n s (Table 3 . 2 ) .

TABLE 3 . 2

I s o m e r i z a t i o n o f Methyl P e n t e n e s (117)

A c t i v a t i o n Energy/k c a l mol-I Rearrangement

Superacid s o l u t i o n s

H-USY

10-14

9-

4

3-4

S i m i l a r l y , t h e primary p r o d u c t s from c r a c k i n g o f i s o b u t a n e on H-M a r e p r o p a n e , n-butane pentane r a t i o - 0

and i s o p e n t a n e (118).

and t h e propane/pentane

(Si/A1 = 8.5)

A t v e r y low c o n v e r s i o n t h e n / i s o

r a t i o >1.

Additionally,

carbeniurl! i o n p r o c e s s a c c o u n t s f o r a l l t h e r e s u l t s o b t a i n e d u s i n g I 3 C

classical labelled

i s o b u t a n e (119). A l k y l a t i o n of i s o b u t e n e by t h e t e r t i a r y b u t y l c a r b e n i u m i o n t o p r o d u c e t h e

59 C

8

carbenium ion is rate determining and all labelled products can be accounted

for by subsequent rearrangements of the C8

ion and by

scission steps.

However, the bimolecular disproportionation process above usually occxrs only on superacids, again providing support for zeolite superacidity. In view of the previous discussion it is not surprising that product distributions are frequently interpreted in terms of types of acid site. Product distributions from alkane cracking have been used to estimate the relative effects of Bronsted and Lewis sites (120) and the oligomerization of ethylene over HY and H-ZSM-5

at 300 K is said to produce largely branched

structures in the presence of Lewis acidity and largely linear structures in the case of strong Bronsted acidity. sites

are

presumed

to

be

Si' ,

In this latter study (121) the Lewis generated

by

thermal

treatments

and

characterised by infrared including the spectra associated with hydrogen sorbed at

77 K.

However, as mentioned previously, there remains some uncertaintv

about the nature of Lewis sites generated by thermal treatment of zeolites. Moreover, the authors suggest that, in the presence of strong Bronsted acidity, the surface species is closer to a sorbed ether than to a carbenium ion, accounting more readily for the lack of chain branching.

1.3.3.1 Alkane Conversions over Modified Zeolite Catalysts Typical modifications

(Chapter 7

and section 1.2.5.2) to zeolite acid

catalysts include modification of composition by hydrothermal treatment or by dealumination or secondary synthesis and modification by

incorporation of

separate phases such as oxides, halides or metals. The reaction/regeneration cycle associated with hydrocarbon cracking to produce gasoline involves treatment with steam at elevated temperatures. Consequently, hydrothermal treatment of zeolites is

widely

studied.

At

elevated temperatures, in the presence of steam, aluminium may be dislodged from framework sites and replaced to a greater or lesser extent by silicon. The shorter Si-0 compared to the A1-0 bond results in a shrinkage of the unit cell and a secondary pore system is developed (122)(74d) which is associated with intracrystalline voids developed by localised collapse of crystals and with generation of amorphous material. The increase in framework Si/Al, following hydrothermal treatment, results in changes in acidity as already outlined.

Additionally, there is evidence for

the influence of nnn-framework aluminium species

in

catalytic processes.

Hydrothermal treat.ment results in an increase in t.he number of Lewis sites and a reduction in the number @f Bronsted sites and there is.some evidence for the generation, under appropriate conditions, of superacid sites (49)(77)(78). Mild steaming of H-ZSM-5 can result in enhanced activity in the cracking of n-hexane (77)(78) and this is associated with increased acidity as measured by

TPDA (77). Prolonged steaming causes more extensive dealumination and results in reduced activity.

These catalytic features are explained in terms of a

synergistic effect of dislodged A1 species on bridged hydroxyls associated with tetrahedral framework aluminium.

If this

necessary that dislodged aluminium should be

explanation

is

correct

it

is

in the appropriate form and

sufficiently proximate to a tetrahedral framework aluminium.

In the case of

H-ZSM-5 theoretical calculations (section 1.4) suggest that aluminiums show preferential siting in TI2 and T2 sites. the four-rings which may be

doubly or

A separate approach implies siting in

singly occupied, in which partial

hydrolysis of one aluminium in a doubly occupied ring could result in a "dislodged A l "

still attached by one or more links to the framework but

representing termination of the -A1-0-Si chain such that it does not occupy a framework 'T' site.

In such a situation the dislodged A1 would be proximate to

the framework 'T' site aluminium present in the original four-ring.

Evidence

for partial hydrolysis of aluminiums is available from recent infrared studies (123) but at this time the nature of any superacid sites generated in H-ZSM-5

by hydrothermal treatment is speculative.

Simple formulations of these sites

are used (49)

H

to imply interaction between Lewis sites (A10')

and Bronsted sites to enhance

Bronsted acidity by electron withdrawal from the bridging oxygen. detailed chemical structure of the dislodged A 1

Whatever the

associated with

enhanced

acidity there is attraction in the identification of superacid sites as Bronsted sites having enhanced activity due to synergistic interaction with Lewis sites both because of analogies with superacid solutions (96) and because there is some supporting evidence for this (49)(77)(78). Because ZSM-5 has interconnecting channels but no definite cage system any aluminium dislodged will tend, initially, to be in the main channel system.

61 However, in H-Y zeolite dislodged aluminium may occupy small and/or large cage regions, depending upon treatment s o that the product of steaming is likely to be more dependent upon process conditions. showing enhancement of acidity and of

Nevertheless, similar results,

catalytic activity

in hydrocarbon

transformations are reported over hydrothermally treated Y zeolite (49). The role of hydrothermal treatment of zeolite Y in connection with alkane conversion reactions is of particular interest because of the commercial scale of

cat-cracking.

Currently,

the

H-

or

rare

earth

steam-stabilised "ultrastable Y" (USY) are widely used.

forms

Y

of

or

During operation these

catalysts are subjected to hydrothermal processes which cause changes in framework composition, secondary pore structures and amount of non-framework aluminium, as previously discussed, and associated reduction in cell parameter a

.

The cell parameter reflects, largely, the framework ratio Si/Al which

determines the density of framework A 1 sites and the distribution of types of A 1 framework site having different acid strength (Section 1 . 4 ) .

This single

parameter (ao) is found to correlate well with changes in both catalytic activity and selectivity in the cracking of gas oil over hydrothermally 0

modified faujasitic zeolites (106) for a o s 24.32 A where most of the sites correspond to those of stronger acidity (ie a framework A 1 possibly one A 1 in its second coordination sphere).

with zero or

In this region activity

tends to decrease and selectivity (octane number) tends to increase with decrease in a

(increase in framework Si/Al).

Increased selectivity is

attributed to reduced hydrogen transfer, which gives a more olefinic naphtha and

more

severe

correspondingly

cracking,

more

on

catalysts

isolated framework A 1

with

low

sites.

values

of

Proximate

a

and

sites are

considered to make for proximity between sorbed carbenium ions and sorbed hydrocarbons resulting in more facile hydride abstraction.

On a more isolated

site the carbenium ion more readily cracks to produce an olefin and a sorbed carbenium ion. However, it is not clear in all cases, that the role of extraframework species, for example dislodged aluminium, nor that of the secondary pore structure generated by steaming can be eliminated from discussions of the activity and selectivity of commercial cracking catalysts.

Currently there is

considerable interest in comparisons of FCC catalysts based on USY and on zeolite Y enriched in silicon by reaction with SiF62-

(124), and it is

suggested that steaming of these initially different A1 content zeolites to the same value does not produce identical catalytic behaviour (124).

As .a

is

reduced by steaming the number of framework aluminiums is much reduced and the

62

amount of extraframework aluminium is increased.

Moreover, in the presence of

steam, aluminium migrates to the outer surface of the zeolite forming a separate phase (47).

Consequently, the catalytic activity of the zeolitic

aluminium is diminished and the role of any enhanced.

extraframework material

is

Extraframework aluminium readily leads to the generation of Lewis

sites, perhaps similar to those found in active alumina, which can be effective in hydrogen transfer reactions.

Therefore, and especially in the cracking of

large alkane molecules which are likely

to

react

initially at

external

crystalline surfaces, it is to be expected that active sites associated with extraframework aluminium will become more and more a feature of the reaction as a

is reduced.

Consequently, catalysts starting from very different framework

compositions (Si/Al) may be dealuminated to the same values of a not have the same amounts of non-framework aluminium.

but they will

Moreover, the relative

numbers of the different types of aluminium generated during dealumination by steaming are likely to depend upon the initial composition as is the nature of the hydrothermally generated secondary pore system (122). These features could well account for the differences reported for USY zeolites made from a starting material with Si/Al

2.5 and catalysts prepared from a modified Y zeolite

with a silicon-rich framework (124).

Differences in Lewis activity for these

catalysts may explain the differences in coke and olefins observed at very low a

values. Treatment of oxide catalysts with halides can enhance their activity in

acid catalysed reactions (125) and similar effects are observed in zeolites. Fluoride treatment of H-M increases activity in cumene cracking and increases acidity as measured by the use of indicators or by calorimetric measurements of heats of sorption of ammonia (92). Enhanced catalytic activity in hexane and pentane cracking (Fig 3 . 6 ) is observed in faujasites containing occluded A1F species (51).

Infrared investigations using pyridine (126) show increased

numbers of Lewis sites in the presence of A1F involved in enhanced activity.

species which are presumably

Similarly the treatment of HAlY with fluoride

can enhance catalytic activity in cumene cracking, a reaction which takes place readily at Bronsted sites.

The conditions reported for treatment of zeolites

with fluorides are varied so that products must differ and a single explanation for enhanced activity may not be adequate.

Nevertheless, the evidence, to

date, suggests that in many instances enhanced catalytic activity can arise from strong Lewis acidity and Bronsted acidity which may, at least in some treatments, be involved cooperatively.

However, in a more general sense the

relative role of Bronsted and Lewis sites will depend upon their numbers,

63

A role for Lewis sites is

accessibility and upon reaction type and conditions.

well illustrated by studies of modified oxide catalysts. containing

A1C13

Lewis

acids

are

proposed

For example in A1 0 2 3 active centres for the

as

isomerization and decomposition of light alkanes at low temperatures (127) and appear to be involved in butane conversion over metal oxides containing SbF 5 (126). Zeolite catalysts are also frequently modified by oxides in some instances to reduce acidity (Mg,B,P oxides) or to add a second catalytic function ( V 0 2 5’ In the case of medium pore zeolites such as ZSM-5 these oxides

Moo3, etc).

(Mg,B,P) can also modify shape-selective effects.

For example the selectivity

in para-xylene production from methanol and toluene depends on the relative rates of acid catalysed alkylation and isomerization and on the relative diffusivities of the xylenes.

Both of these factors are favourably modified by

incorporation of oxides ( 8 7 ) .

Moreover, since active sites at external crystal

surfaces are non-shape selective, there can be advantages in surface enrichment of oxides (47c).

1.3.3.2 The role of free radicals and radical ions

Over non-acidic Na-forms of zeolites, hydrocarbon transformations occur, at relatively high temperatures to give product distributions anticipated from reactions via radical intermediates ( 1 2 8 ) . in acid forms of zeolites.

Radical processes are also observed

Early reports suggest that radicals may be

generated at sites associated with transition-metal impurities (eg Fe(II1)) but recent reports emphasise the role of “defects” in the zeolite structure (129). It is clear that radical species are readily formed, even temperatures, on hydrogen forms of zeolites. olefins form radicals at -78OC, sorbed. implying

at

low

On hydrogen mordenite (H-M),

their nature depending upon the type of olefin

However, at 2OoC a series of olefins give rise to the same signal a

common

hydrocarbon

fragment

produced

by

isomerisation

and

oligomerisation (130). The influence of ammonia demonstrates that, although Bronsted sites are not essential for radical/radical ion formation they can be involved in stabilisation of radical species. olefins, to

3OOOC

modifies

the esr

signal

Heating H-M, containing sorbed such

that

various

sorbates

(acetylene, ally1 chloride, butadiene, n-hexane, methanol) form a radical having a 1-,2-dialkylbenzene structure.

common

These higher temperature

transformations are strongly inhibited by ammonia implying the participation of Bronsted acid sites. The rate of generation of signal intensity, due to

64

o-xylene sorbed at 20°,

depends upon Bronsted sites as does the steady state

concentration of radicals. Radicals are also detected in the effluent stream from the conversion of methanol over H-ZSM-5 at T

> 444

K, and a radical

mechanism is proposed for formation of the first C-C bond in the methanol to hydrocarbons conversion. Product distributions are also used in support of radical intermediates during hydrocarbon conversion over acidic zeolites.

For example the major

isobutane conversion products to be expected from the tertiary carbenium-ion intermediate differ from those expected from a radical/radical ion intermediate (131).

Experimentally it is observed that products from conversion of isobutane at 700-920 K over amorphous solid acids differ from those over H-USY and it is suggested that, in both types of catalyst, the initial reaction involves an electron acceptor site ( E A ) . iC H

i-C4H10 + EA

/ 0 4 8 4

i-C4Hloo

+ H2 + EA

EA

C H

3 6

+ CH4 + EA

Differences in product distribution are attributed to subsequent transformation of olefins into carbenium ions by the acidic H-USY zeolite which results in consumption of hydrogen, increased amounts of paraffins and C5+ and reduced CH4.

However, i-butane conversion is also interpreted in terms of direct

protonation by H-M

(118) to produce carbonium ion intermediates.

There are

difficulties in resolving the extent of radical involvement by reference to product distributions over strongly acid catalysts.

Secondary reactions

involving carbenium ions mask the initial product distribution and at very low alkane conversions where primary products may be detected, protolytic attack and radical reactions can, in many instances, give a similar product slate. further and general problem associated with

reactions at

A

low conversion

concerns increased difficulty in analysis and increased distortion by any trace impurities.

Procedures involving extensive experimentation are reported to

give identification of primary products (132).

Unambiguous identification

could clarify mechanisms in some cases. There remains the question of the nature of the electron acceptor sites. One proposal relates their generation to modification of Bronsted sites in H-ZSM-5 during calcination processes ( 1 2 9 ) .

65

H

I

(a)

I I

%02

-Si-0-Al-

i

I

* t

I

4 - 0 -Al-

---

i

I

+

y H20

I

Electron acceptor (EA) (solid defect site)

Bronsted s i t e

The electron acceptor site can then react according to

I *+ I I I

(b)

+ R

-Si-0 -Al-

-

I

I

+

-Si-0-Al-

EA

i

Re'

I (EA-)

The charge transfer process in (b), generating the species R t ,

is said to occur

readily for organic molecules with ionisation potential less than the adiabatic ionisation potential (12.61 eV) of the nonbonding lone pair orbital in oxygen. Subsequent transformation of the radical ion may result in carbenium ions. For example, (129)(131),

CH -C-CH=CH2

3 1

+

I -+ I

-Si-0 - A l -

I

I

-

CH3

the

CH3 H

I

I

CH -C-C-CH

3 1

CH3

alkene radical

cation may

2-

carbenium ion +

EA

+* EA-

isomerise as

above.

aromatisation of alkyne radical ions is available (129).

Esr

evidence

for

It is a l s o suggested

(130) that strained T-0-T links may be the source of EA sites.

Strained links

may be a feature of distorted four rings in H-M which are stabilised by Na+ ions in NaM.

Strained four rings may also arise from elimination of water from

hydroxyl nests (132).

In H-M it is suggested, based on esr studies in the

presence of ammonia (133), that the radical/radical ion is formed at an oxidising site (EA) and transformed at a Bronsted site. The previous discussion reviews some current thinking on hydrocarbon transformations. Table 3.3 provides a summary of possible routes to carbenium ions in acidic zeolite catalysts. In the absence of dehydrogenation centres alkanes convert to carbenium ions via carbonium ions or radicals or directly in

66

the presence of strong Lewis acid sites.

The relative roles of direct

protonation, hydride abstraction or radical/radical ion fo:nation

will depend

on the type (Bronsted vs Lewis) number and strength of the acid sites and on the availability of electron acceptor sites.

In turn these depend upon the

structure and composition of the zeolite and on particular pretreatments.

The

influence of these sites will a l s o depend upon the chemical nature of the hydrocarbon and on process conditions.

Once the surface is "covered" with

carbenium ions products could be expected to follow by typical carbenium ion processes (section 1.3.l)subject to thermodynamic and soatial restraints.

TABLE 3 . 3 Summary: Hydrocarbon Transformations over Acidic Zeolite Routes to Carbenium Ions

CARBONIUM ION

+He strong site

-H2 or-alkane CARBENIUM ION

-

PRODUCTS

67

1.4

Theoretical Aspects of Zeolite Acidity Several theoretical approaches have been made to explain the properties of

zeolites. In this section emphasis is placed on molecular orbital calculations using cluster models and on calculations associated with distribution of T atoms with'in zeolite frameworks. Other approaches are considered briefly.

1.4.1 Quantum Mechanical Calculations on Clusters 1.4.1.1 Ab Initio Methods Techniques of quantum chemistry can provide powerful tools for elucidating microscopic details of structure and bonding (134). It has recently been shown that molecular orbital calculations can provide useful information concerning the intrinsic properties of acid sites in solids.

A variety of molecular

orbital calculations are available varying in complexity from essentially empirical to completely correlated wave function analysis.

Early quantum

chemical investigations of solid acid sites utilized semi empirical methods such as CNDO, I N D O and more recently MIND0 (135), and these are discussed subsequently. Recent developments in computer technology plus the availability of efficient small gaussian basis sets have rendered feasible ab-initio calculations which can in many cases give structural data with an accuracy close to that of experimental observation.

Some detailed accounts of the

earlier semi-empirical literature with reference to available (135) and is discussed subsequently. focussed

on

the

recent

ab-initin

solid

acid

sites is

This sectinn will mainly be

calculations,

while

application

of

semi-empirical methods are discussed in the following section. For feasible molecular orbital calculations it is necessary to choose a suitable model, (Fig 4.1) which should be sufficient for description of the intrinsic properties of the hydroxyl group representative of the Bronsted acid site in zeolites.

After selection of the model it is necessary to choose an

appropriate basis set to perform the quantum chemical calculation.

The choice

is limited by the computer cost and whereas use of an extended basis set together with

the

inclusion of

configuration

interaction

would

be

the

calculation of choice, such an extensive calculation is not feasible at present with the units of Fig 4.1.

Although many Calculations have been performed

using the STO 3G basis set (134), in recent years the 3-21 G basis set has been developed where the outer valence shell is split to allow for a more diffuse gaussian function in the valence shell.

In most molecules this effect leads to

68

an improved description of bonding characteristics. Hydroxyl groups in zeolites can be broadly classified into terminal and bridged forms (136). These are referred to previously in Sections 1.2.4.1a in discussion of experimental evidence which shows that the terminal forms exhibit a much lower acidity.

Using models (a) and (b) of Fig 4.1 to represent

terminal and bridged hydroxyl groups respectively, ab-initio calculations using the 3-21 G basis set have been performed in the hope of gaining some insight into their acidic behaviour.

Full geometry optimisation is performed for these

units (under the constraint of Cs symmetry) and the resulting geometries are

A Mulliken population analysis and an OH stretching

outlined in Table 4.1.

frequency calculation have also been performed and the results obtained are outlined in ,Table 4.2. It is apparent that the OH distance, the OH stretching frequency, the qH and 1q q I values are all predicted to be larger for the O H This is in agreement with the experimentally observed

bridged hydroxyl group.

greater acidity of the bridged hydroxyl group. calculated frequencies (64 2 20 cm-') difference of approx 80 cm-'. as charge distribution

9

The difference between the

is also close to the experimental

Whilst the calculation of static properties such and

OH bond strength do give an insight into the

potential acidity, the true Bronsted acidity can only be assessed by measuring the ability of the OH group to protonate a suitable base (137).

A more direct

measure of Bronsted acidity therefore is the weakening of the OH bond brought about by interaction with a suitable base.

We have referred to experimental

characterisation of acidity in this wayin section 1.2.4 possibility of

predicting

interactions of CO and NH

3

such

effects

using

and now investigate the

theoretical

methods.

The

with models (a) and (b) of Fig 4.1 are investigated

by monitoring their effects on the charge distribution and strength of the OH bond.

The results obtained are presented in Table 4.3.

These data indicate

that the perturbation of the bridged form is greater than that observed for the terminal form. NH

3

For example, whereas a

&G

value of 440 cm-'

is observed for the terminal OH form, a much greater

is observed for the bridged aluminium form.

on interaction with value of 1150 cm-1

In this respect it is useful to

note that the difference in shift between the bridged and terminal form is much greater than the difference observed in the free OH frequency (approx 60 cm-l). This

indicates

that

the

bridged

form

is

more

sensitive

to

external

perturbations, a conclusion which is in agreement with many experimental observations. Experiment shows that the acidity of zeolite solid acids is dependent on the framework composition (Si/Al).

As discussed in previous sections, acidity

69

IS

"Y ry

I

36 4.0

Fig.4.1. Model units used for Ma calculat'm.

I

4.4

ELECTRONEGATIVITY Fig.4.2.ta)Expt. A3 value on C6H6 adsorption. (b)Theoretkal A3 for H20 interaction as a function of electronegfity (ref.138).

//

' ,A'

H

l

Mcy d d N t W * , pH - I ~ dihr

38 w M YHOERYH IMERMDIATL ELECTROHGATIYITI

3.6

36

Fig.43. Ab-initio Mo calculation i3-21G ; SCF). Symmetrical units X,SiiOH)Al X, iX=H,F).

Fig.44. Model units used to monita acidity variation as a function of nearest Eighbour Al (ref.121).

70

TABLE 4.1

Moleculw Geometry Calculated for Units ( a ) and ( b ) of Figure 1 using 3-219 Basis Set (from ref 36)

SYSTEM

H3SiOH (Cs)

H SiOHA1H3(Cs)

rSiH rSiH /OSiH /OSiH /RSiH r(Si0) r(0H) /SiOH

147.6 149.0 106.9 112.4 108.9 167.4 95.9 127.8

147.9 147.1 104.5 108.5 111.2 173.4 96.7 119.6 130.7 192.7 162.2 161.1 96.8 102.7 117.3

r(A1H) /OA~H -/OAlH /HAlH

Bond distances in pm

TABLE 4.2 OH Bond Properties for Terminal and Bridged Hydroxyl Groups (refs 136,139,2451

OH Bond Characteristics

H3SiOHA1H3

H3SiOHBH3

0.4156

0.4727

0.4670

0.4710

0.3682

0.4415

0.4040

0.4357

(c)

0.959

0.967

0.963

0.965

(a)

6891.6

863.2

880.7

867.1

3995

3931

3970

3940

qH

IqHqO I OH fOH

H3SiOH

H3SiOHGaH3

71

increases with Si/A1 ratio.

One interpretation of this effect is that

replacement of aluminium by the more electronegative Si atom leads to increased electron withdrawal from the OH bond which results in increased acidity. Ion-exchange effects on acidity have also been explained in terms of the Using the OH stretching frequency as a

electrnnegativity principle (38).

measure of acidity a linear correlation between the OH stretching frequency and the intermediate Sanderson Electronegativity is observed (38). Deviations from this linear relationship are attributed to electrostatic or crystal field effects on OH groups in small cages.

The effect of increased electronegativity

on the OH properties has been investigated theoretically using the models of Fig 4.lb by progressively replacing the hydrogen atoms by fluorines (138). The effect of

Representative results obtained are presented in Table 4.4.

increased electronegativity on the change in hydroxyl frequency ( A D O H ) for interactinn with

H20

is illustrated

in

Fig

4.2.

For

comparison

experimental trend observed on adsorption of benzene is also given.

the Both

results indicate that increased electronegativity does lead to increased acidic properties and the theoretical trend of Fig 4.2

does mirror that observed

experimentally. Recent work shows that "zeolites" containing elements other than A1 or Si in framework positions can be synthesised.

The range of elements incorporated

into the framework continues to grow and experimentally it is shown that quite large variations in acidity can be achieved by such substitutions. interest, therefore, to investigate the aprinri prediction by

It is of

theoretical

calculations, of the acidic properties to be expected in these materials.

In

our laboratory we have recently instigated such a study (139) using the unit of Fig 4.lb where A1 is replaced by B and Ga.

Experimentally it is shown that the acidity varies in

collected in Table 4.2. the order A1

> Ga.> B

The data relevant to acidity are

(

31

).

The theoretically calculated OH stretching

frequency, and charge distribution for the OH bond, are in agreement with the experimental

sequence.

The

interaction of

CO

and

NH3

with

the boron

substituted form has also been investigated and the results are presented in Table 4.3.

It can also be seen that the perturbation of the properties of the

OH group for the boron unit is much less than that for the bridged aluminium unit, and is similar to that calculated for the terminal OH model.

This again

is in agreement with experimental acidity measurements of boralite which indicate a Bronsted acidity value close to that of silica (31).

72

m

d

o

% d

P P

c l m

m m m m

m m m

i +

P N

m +

N w

m

0,

W W

d

X

m z

+

+

z

m In 0

d

N m P

m

m

+

W

0,

m m

0

0

0 V

3:

m

0

V

+

m z

+

0

X

+

+

I:

3:

z

z

X

z

V

0

m

9

0

m

m

N

m

N

m m

P r ' m

m m

z

S

d d

0

0

m

D

m

m

2

.?I 0

"'m

"'m

73

TABLE 4.4

Free OH Group Properties in Model Compounds of Figure 4.lb and its Fluorinated Derivatives (from ref 138)

Compound

S

(090)

3.44 3.79 4.17

(191)

(2,2)

S

=

VOH

3920 3911 3901

-1

H'O'

0.4449 0.4757 0.5048

Sanderson Electronegativity

(1,l) Corresponds to one fluorine atom replacing a H atom on each side of the bridging OH group

TABLE 4.5

Effect of /SiOA1 angle on OH group properties

SiOAl ANGLE

re/A

126 140 150 160 170

0.966 0.970 0.974 0.978 0.981

VOH

3931 3870 3828 3773 3717

+ qH

0.472 0.465 0.456 0.445 0.432

74

Recently the effect of geometric structure on the properties of the bridging OH group has been investigated (140). The variation in properties of the OH group with increasing SiOAl angle are presented in Table 4.5.

In

general it appears that increasing SiOAl angle leads to a decrease in OH bond strength.

The ionicity of the bond is also decreased.

High-silica zeolites

(ZSM series, mordenite, THETA-1 etc) generally have more acidic Bronsted sites than low-silica structures. Si/A1 ratio. angles.

This has been attributed mainly to the higher

However, they are also characterised by much larger TOT bridging

(Mean

/TOT

in ZSM5

154O; Mean

=

/TOT in faujasite

143O).

=

The

above results suggest that it may be important to take into account structural properties as well as compositional, a point made previously, when attempting to explain acidic behaviour of zeolite catalysts. hydrogen charge on both

electronegativity

and

The dependence of the bond

angle

is

depicted

graphically in Fig 4.3.

1.4.1.2

Semi-Empirical Methods

As with the ab-initio applications discussed previously, most approaches in this field make use of cluster models which are taken to represent a specific unit and active site of the relevant polymorph or crystalline structure.

The cluster model is usually terminated by hydrogen atoms to mimic

the lattice environment although other approaches such as pseudoatoms with variable orbital ionisation potential (141) or electrnnegativity (142) have also been used. In

general

calculations are Calculations.

the

geometric

inferior

to

properties

obtained

from

similar quantities obtained

semiempirical via

ab-initio

This is principally due to the extensive parameterization and

approximations used in semi-empirical type methods.

Most investigations

utilise the parameterisatinn given in the original formulation (143) however it has been recently shown (144) that reparameterisation for third row atoms can lead to more accurate geometric results.

Absolute values of charges

calculated by CNDO methods have been shown to be too low accurate

ab-initio

results

(145), however, relative

in comparison with trends

distribution and energy can in general be accurately reproduced.

in

charge

For most

applications to solid acid sites one is primarily interested in the relative charge distribution or bond strength and in such cases the results of CNDO or similar semi-empirical methods can be of use. Extensive application of CND0/2 in the characterisatinn of solid acid

sites (zeolites in particular) has been carried out by Beran and co-workers (13,146). In these studies the standard version of the CND0/2 method (143) and an

s, p

basis for the Si and A1 atoms was used.

Inclusion of d orbital

basis sets dbes lead to a considerable reduction in the charges calculated, however, no noticeable change is observed in the relative charge distribution (145). A

distinct advantage of

semi

empirical

methods

over

ab-initio

calculations is the ability to perform calculations on large units. using the CNDO/2 method studied

This is

Beran (135)

due to the much less prohibitive computer costs of the former.

the electronic structure of HX and HY

zeolites using 4-ring cluster models such as Si A 1 0 H(OH)8-, Si A 1 0 H (OH)8, 2 2 4 2 2 4 2 Si A10 H(OH)8 and Si A 1 0 H(OHl8Na. Formation of hydroxyl groups at 01 and 03 3 4 2 2 4 was favoured. The calculations also indicated that the hydroxyl groups of HX zeolite are less acidic than the corresponding HY hydroxyls.

In general it

was found that the presence of two A 1 atoms in the four-ring decreased the acidity of the hydroxyl group. This was taken as quantum chemical support for the previously

proposed

effect of nearest neighbour A 1

characteristics of the hydroxyl groups.

on

the acidic

This is discussed later in this

section. The location of an Na’ ion in the SI position was shown to exclude formation of OH groups of the 03 type.

The acidic characteristics of HZSM-5

were also studied using the CND0/2 method (146). modelled by clusters of the T505(OH),o

The ZSM-5 structure was

and T 0 (OH)12 type (T 6 6

=

Si or A l ) .

The affinity of the skeletal oxygen atoms for the protons was found to be lower than in faujasites and decreased with increasing TOT angle.

The larger

TOT angles of ZSM-5 compared with faujasite can, therefore, be expected to give rise to an increased acidity.

However, analysis of the charge on the

hydroxyl hydrogen reveals that this parameter decreases with increasing angle. These trends have been confirmed by ab-initio calculations (see above) and suggest that the location of protons near the large TOT-angle sites is improbable and that the protons migrate to more normal positions with deeper potential wells.

As

described previously for ab-initio studies a better

description of the Bronsted acidity of surface OH groups can be obtained by investigation of their perturbation upon adsorption of basic molecules.

Beran

et a1 (146) have investigated the interaction of ethylene with the hydroxyl group of faujasites and HZSM-5 zeolites.

It was shown that interaction

leading to the formation of a stable n-complex

76

'c' C

I \ is energetically most favoured and results in a weakening of the OH and C=C

bonds.

The interaction with hydroxyl groups of HZSM-5 was shown to be

energetically more favourable than the interaction wi-th faujasitic hydroxyls and was attributed to the higher acid strength of HZSM-5. faujasitic type zeolites with A1

3+

Models

, A1(OH)2+ and A1(OH)2+ cations located

of in

the SII and SI cationic positions were studied by the CND0/2 method (146~). All these species were shown to exhibit strong electron acceptor capabilities with the Lewis acidity varying in the order A13+)

Al(OH)2+>

A1(OH)2+.

The

OH groups bound to the A1 ions were shown to be less acidic (Bronsted) than the skeletal hydroxyls. Dehydroxylated forms of faujasitic zeolites were also investigated by the CND0/2 method

(146d).

It was demonstrated that the

tricoordinated A1 produced by dehydroxylation is a very weak Lewis acid in contrast to tricoordinated 5Si+ whose electron acceptor ability is much greater and comparable with that of the A1(OH)2f species coordinated in the cation posi-tions in the zeolite.

Kazanski et a1 (147) have used the CNDO

parameterisation

Whitehead

of

Boys

and

characteristics of silica and zeolites.

to

investigate

the

acidic

The cluster models were bonded by

univalent hydrogen-like atoms (A) with variable valence orbital ionization potentials (VOIP) that allowed one to modify the acidic properties of the OH group.

Calculated shapes of -the potential energy curves for the (OA) SiOH 3

fragment coincided well with experimental curves calculated from near-infrared data.

These calculations provided support for experimental observations,

using overtones, which suggested that neither the shape of the potential curves for surface OH groups nor indicators of acidity.

the

frequency of vibration

are good

More reasonable characteristics were proposed to be

the protonic charge and the heterolytic dissociation energy of the OH group. The effect of increasing Si/A1 ratio on acidity o f zeolites has been investigated by

Kazanski and

coworkers

combinations as shown in Fig 4.4

(36c).

are used.

The

different

cluster

The charges on the protons

calculated for each unit are presented in Table 4.6.

Increasing the Si

content was found to lead to an increased charge on the bridging proton which leads to an increased bond polarisation and results in enhanced acidic behaviour on the interaction with a

basic

substrate molecule.

Whilst

increased protonic charge with increasing Si/Al ratio would appear to be

I7

similar to the proposal of Jacobs (38) concerning the effect of increasing electronegativity it is,important to note that the model of Kazanski ( 3 6 c ) is concerned with local effects on a particular bridged unit.

In the correlation

proposed by Jacobs (38), in addition to the local aluminium content, the influence of all the aluminium atoms in the crystal is taken into account. This would result in a continuous spectrum of acidity with the hydroxyl stretching frequency being dependent on the bulk electronegativity which should give rise to only one type of OH group. The local model proposed by Kazansky however predicts that four different types of OH group should be present, corresponding to the four units of Fig 4.3, the relative quantity of each one being dependent on the Si/A1 ratio of the particular investigated.

zeolite

1 all four types of OH groups should be

For 7>Si/A1)

present, whereas for high-silica zeolites ie S i / A Y 7 only one form should be observed and its acidity should remain constant for values above 7.

The

available experimental data does provide support for a localised model which explains (a) the simultaneous occurrence of several hydroxyl groups with different acidity (Section 1 . 3 . 2 1 and (b) the constancy in acidity at higher values of Si/Al (Section 1.3.21.

However, longer range effects presumably

generate an acidity spectrum within each particular type of hydroxyl site resulting in the more complex acidity spectra seen by TPDA. Moreover, this model is over simplistic in that it takes no account of structure nor of any factors influencing the distribution of aluminiums within the zeolite framework.

As mentioned

previously there

is evidence for

preferred siting in some zeolites and distribution may be more or less ordered.

Consequently the local composition may not reflect the global

framework composition so that the value of Si/A1 = 7 should not be accepted as uniquely defining all the acid sites in any zeolite with that framework composition, and this is discussed subsequently. A modified CND0/2 study was carried out by Chen et a1 (142) and Geerlings et a1 (137a) into the acidity of silica

-

alumina units.

To account for the

effects of framework termination, the cluster models were terminated by hypothetical hydrogen-like atoms, L, with variable electronegativity.

This

allowed for cluster models in which the negative oxygen charge was the same as the positive charge on Si thereby simulating fragments with a large number of tetrahedra.

Variation of the electronegativity of the L atoms demonstrated

theoretically the greater sensitivity to structural composition of bridged type hydroxyl

groups over

their

terminal

counterparts.

Non-empirical

molecular orbital calculations have been reported by Kazansky (148) on the

78

electronic structure of the protonated intermediate formed by the transfer of a proton from an acidic centre of an aluminnsilicate surface and from a molecule of sulphuric acid to an ethylene molecule in the absence of solvation effects.

In both cases the most stable state was found to be an ester with a Evidence for carbonium ion

moderate positive charge on the ethyl group. formation was not present.

This result was presented as evidence that inn

pair formation arises as a result of secondary solvation effects in the liquid state.

The authors conclude that the absence of such effects in the gas phase

reaction on

aluminosilicate surfaces casts

ion-type catalytic mechanisms.

doubt on proposed

in favour of carbenium ions this is somewhat surprising. remembered

that

factors

carbenium

However, in view of the experimental evidence

other

than

ion

pair

stabilisation of ions in zeolites (section 1 . 3 . 2 1 .

formation

It must be influence

the

MINDO/3, INDO and CND0/2

methods were used by Gorb et a1 (149) to calculate the stabilisation energies associated with the interaction of the bridged and open forms of protonated ethylene and benzene with a charged surface centre or with a polar solvent. The bridging forms were shown to be more strongly stabilised on a charged surface and the open ones in a polar solvent. In conclusion it can be said

that

the application of

approximate

molecular orbital methods to the investigation of solid acid sites has proven useful.

Computer cost is much

calculations,

and

where

relative

less

than for similar ab-initio

values

or

trends

only

calculations of this type are adequate. TABLE 4.6 Calculated Protonic Charge (q,)

OH I OH I1 OH I11 OH IV SiOH

for Units of Fig 4.l(b)

No of surrounding A1 atoms

Si:Al ratio

1 2 3

4-7

4

0

3 1.7 1

-

qH

0.404 0.392 0.384 0.376

0.317

are

type

required

79 1.4.2

Acidity and the siting of Aluminiums in Zeolite Frameworks Early studies showed that acidity increased with increase in framework

Si/A1 in the order NaHX(NaHY
Consequently, the dilution of the

aluminium concentration in the framework was recognised as a factor leading to increased acidity per site.

This was emphasised by careful work on the

dealumination of faujasitic zeolites which established (68)(72) that: a typical Y zeolite ( -

(i) in

56 Al/u.c.) strong acidity was detected after about 30%

exchange of Na for H ions, indicating that about 30% of the aluminiums were associated with weaker acid sites; (ii) the effective acidity

(a

)

of each

acid site increased with dealumination up to a ratio S i / A l c z 6 (iii) about 30% of the aluminium was more easily removed during dealumination, than the remainder and (iv) only strong acidity was observed with faujasite frameworks containing 35 or less aluminiums per unit cell. that

acidity

depended

on

geometrical

These results which suggested

arrangements

of

aluminiums

were

interpreted (150) by the recognition that, subject to Lowenstein's rule, the closest

approach

between

aluminiums

in

neighbours diagonally opposed in 4-rings.

faujasite

occurs

between second

Weaker acid sites were presumed to

be associated with aluminiums within the 4-ring system having more than one diagonal neighbouring aluminium. Proximity of aluminiums rendered them less strongly bound, electrostatically, within the framework and facilitated their removal during dealumination.

Similary hydrogens associated with aluminiums

proximate to other aluminiums were predicted to be weaker in acidity because of their increased interaction with surface sites. Discussion of aluminium distribution in faujasites is clarified by the recognition that the crystallographically identical 'T' sites are in three 4-rings, two 6-rings and one 12-ring (Fig 4.5). The first 'TI atom coordination sphere for any A1 atoms must consist of four Si atoms (Lowenstein rule).

The second IT' atom coordination sphere

contains nine T atoms (sites 1 to 9 in Fig 4.5) which may be either A1 or Si. Aluminiums in the 4-ring second neighbour sites (1,5,8) are at a shorter distance than aluminiums in the remaining second neighbnur sites (2,3,4,6,7,9) and presumably interact more strongly with the central aluminium. the 4-ring system

Considering

, therefore, results in four types of site for an

80

Fw 4

' Ia,b,c,d) and second (1 to 9) cwrdition spheres of a framewwk'T' a h . (Ref. 152).

MWINIW A M Y U

Fg.4dDistributim of types of Al atoms in fwjasites. (Ref. 52). unique structures of Cambridge group. d e e d unit ceIl(26).

TYPES OFAl ATOH

dd&& 1

0

2

-

-

3

-\"

"0

01

-----

P

0

02

0.3

Al/W*Si)

Fjk7.BAluminium fraction from ?Si NMR ?9Si NMR intensities of dealuminated Y zeolite. ( m t h curves calcdated for a random distribution of All

81 aluminium atom which may have 0 , 1, 2 or 3 diagonal neighbour aluminiums (Fig 4.5).

Weaker acid sites, which are observed at higher levels o f

aluminium, are associated with increased aluminium site density. An

acidity "spectrum" for faujasites was obtained (151) by calculating

the relative numbers of sites Ni (i

=

0, 1, 2, 3 ) , for a range of framework

compositions (Si/Al), assuming a random distribution of aluminiums (subject to Lowenstein's rule) over 'T' sites.

Population densities for the four types of

acid site were in agreement with thermal analysis for a Y zeolite with about 50 A 1 per unit cell (151) (Fig. 2.8). This approach was

extended to

include other models

for aluminium

distribution and further consideration was given to end effects to reduce

A separate approach utilized the siting of aluminiums in

possible bias (152).

independent 6-rings to investigate the first coordination sphere of silicons in faujasite frameworks (153). The relative numbers of types of aluminium site, based on several models are readily calculated from published data (153).

If the probability of

occupancy by aluminium is the same for the three diagonal 4-ring sites (1, 5, 8) then this is given by N

(i)

=

NAl 3 ! / ( 3

-

i)!i!(l

-

3-i i P) P

where N(i) (i = 0, 1, 2, 3) refers to the type of site, NAl is the total number of aluminiums per unit cell

and

P is the probability that a diagonally opposed site is occupied by Al.

In general fractional occupancies of the three diagonally opposed sites (1 and 5 or 8) may not be equal in which case the above expression must be extended

as appropriate. and gives results for the N at (i) (i) different framework compositions (Si/Al), for two extreme model distributions Fig4.6 shows the types of site N

of aluminiums in zeolites.

One model is based on a completely disordered and

the other on a completely ordered structure. Strong acidity has been associated, variously, with the No sites which, since they have no aluminiums as second 'T' neighbours in the 4-ring system, may be regarded as isolated sites o r with both the No and N1 sites.

However,

faujasite frameworks cannot be synthesised at low aluminium levels so that comparison of computed and experimental results over a range of Si/A1 is limited.

The aluminium content of zeolite Y can be reduced by dealumination

82

or secondary synthesis procedures (Section 1.2.5.2 and Chapter 7 ) but the effect of dealumination on the relative numbers of sites depends on the initial distribution and the relative rates of aluminium removal from the various sites. A theoretical treatment of changes in Ni on dealumination based upon initially specified values and on specified values of (first order) rate constants for dealumination is available (151) but

progression of

this

treatment awaits experimental determination of rates of dealumination. However, estimates of different types of silicons are available from 29Si solid state NMR.

These data (Fig 4.7) which reflect different treatments, operators

etc, are well-scattered, particularly for the Si(lA1) signal which may reflect some contribution to intensity from >Si-OH centres. /

with

a model

Nevertheless, agreement

for random placement of aluminiums

especially at lower A1 levels.

is reasonable

Of more significance is the fact that the

Si(2A1) peak persists to relatively low levels of aluminium which means that some aluminiums have a second aluminium within the full (9 atom) second 'T' atom coordination shell, and consequently are not isolated, in the range of global composition where catalytic results suggest there is only one type of site.

This suggests contrary to present opinion, that at higher values of

Si/A1, aluminiums linked through 0-Si-0 to a second aluminium may be as active as isolated aluminiums.

However, it must be accepted that NMR is a bulk

technique and results of the kind shown in Fig 4.7 may reflect inhomogeneity in samples so that this point requires further investigation. Although topological termination is taken at the 4-ring 'T' sites on the second coordination sphere for the specification of No, N1, etc, it is clear that particular sites could be modified if a more extended topology is considered. For instance taking the full second coordination sphere of 9 IT' atoms into account it seems reasonable that although aluminiums in these sites are farther from the central aluminium than those in the 4-ring system they may have some influence.

In which case aluminiums of type 'i' could vary

somewhat due to perturbation by different arrangements of aluminiums in the extended second or even higher coordination spheres. In applying the results of calculations of N. to acid catalysis it is necessary to make assumptions regarding their involvement. Only the strongest sites (N ) may be considered, or, a weighted average could be used depending 0 upon the particular reaction. An attempt to provide an answer to this type of question leads to the development of a parameter 'P' based on the 4-ring system and a parameter Q based on the full second coordination sphere of 'TI atoms (58b)(152).

Parameter P is the average occupancy by aluminium of one of

83 the three diagonal second neighbour sites (1, 5, 8) for an aluminium in the 4-ring system.

The parameter Q extends this over the nine second neighbour

IT' positions.

Both

parameters

depend

upon

the

faujasitic

framework

composition (Si/Al) and on the distribution of aluminiums over 'T' sites. the aluminium content of the framework tends to zero P(Q) experimental values for effective acidity in zeolites by (1

-

P ) or (1

-

(a)

-

0.

As The

can be approximated

Q ) as can some catalytic properties

(152)(77).These

parameters, therefore, reflect the self-shielding of aluminiums in

the

faujasite structure ranging from a value of zero at zero aluminium to unity at the maximum loading allowed by Lowenstein's rule (Si/Al = 1). Considerations of average acidity have utilised the average distance of separation of aluminiums and the topological density of aluminiums (58e) which is the product of the aluminium fraction and the topological density of 'TI sites taken over the second to the fifth coordination layers of tetrahedra around a 'T' site.

In this latter study (58e) relative rates of reactions

involving strong acid sites and relative effective acidities are presented as a function of framework composition for a series of zeolites (FAU, MOR, LTL and MOR).

The framework composition(Si/Al) below which aluminiums should have

no aluminiums in the second layer of T atoms is given for these zeolites based on a value for faujasite estimated by extrapolation of the experimental effective acidity of faujasitic zeolites ( E ) to a value of unity where Si/A1 = 5.8.

An earlier topological model dealt with acidity in zeolites and included a consideration of compensating cations (154).

Results were compared with

computations from a model based on a binomial distribution of Si and A1 over second neighbour

'TI atoms and correlated with acidity measurements and

catalytic results on faujasite and mordenite. However, there are some difficulties in relating strong acidity to 'isolated'or other types of site without reference to structure. In general, the IT' sites in zeolites are not all equivalent crystallographically,as they are in faujasite, and detailed structural features for example bond angles and lengths can affect acidity at a given site.

Moreover, there is some evidence

that aluminium shows preference for particular sites in some zeolites. Preferential siting of A1 in the 4-ring system of H-ZSM-5

is used to explain

(78)the observed enhanced activity consequent upon mild steaming ( 7 8 ) ( 7 7 ) .

Hydrothermal treatment is presumed to dislodge one aluminium from a 4-ring site containing two aluminiums. The proximity of the resulting framework and dislodged A1 atoms is presumed to give rise to an enhanced activity, in hexane

84 cracking, for the A1 remaining in the framework IT' site. Since, in the unmodified zeolites, catalytic activity constant

per

site

is

over a wide range of aluminium levels (58a)(66) up to compositions

where extensive pairing is predicted this means that, in H-ZSM-5,

aluminiums

with either zero (isolated) or one second neighbour aluminium contribute equally to catalytic activity.

Consequently we should not expect to find

correlations with the number of "isolated" sites.

It might be argued that

pairing in four-rings of H-ZSM-5 is exaggerated and that, in a general case, providing the number of preferred sites in a zeolite structure sufficiently exceeds the number of aluminiums then preferred siting may not distort results too much.

Additionally the mobility of hydrogens in silica-rich zeolites may

smooth differences resulting in an "averaged" effect. possibility

of pairing must

be

taken

discussion of the acidity of pentasils.

into

Nonetheless, the

account in

any

theoretical

MO calculations (155,156) do not and sorption

suggest extensive pairing of aluminiums in 4-rings of ZSM-5,

results suggest that not all the aluminiums are confined to 4-rings in mordenite.

Moreover, the * ' S i

NMR results (Fig 4.6) suggest that aluminiums

may not be "isolated" at Si/A1

>6

in the faujasite structure, a conclusion

also suggested in the case of H-ZSM-5 and mordenite by 29Si NMR (157,158). It must be concluded that a constancy of turnover number etc. over a range aluminium levels, which implies that sites have equal activity, does not prove that aluminiums are, in all cases isolated and this should not be presumed.

1.4.3 Electronegativity Concepts and Zeolite Acidity The

enhancement

in

Bronsted

acidity

due

to

increases

electronegativity of neighbouring atoms is well documented.

in

the

In particular,

the Sanderson intermediate electronegativity ( S ) is found to correlate well with many zeolitic properties, including acidic and catalytic properties (38). According

to

the principle of

equalization of

electronegativity

(159),

different atoms, when combined in a molecule, achieve a common (intermediate) electronegativity

by

charge

redistribution.

The

intermediate

electronegativity is given by the geometric mean of the electronegativities of the constituent atoms.

For example, the hydrogen forms of dehydrated

zeolites, with framework aluminium fraction x/(l HxAlxSil-x02

+ x), may be written (58c) as

with intermediate electronegativity (SHz) given by

85

The charge

(6.)

on any particular atom with atomic electronegativity ( S . ) is

calculated, taking as a reference NaF (assumed to be 75% ionic), from the intermediate value ( S HZ) as (ji= ( SHZ - Si)/2.08

&,

Note, since example

6.is a

linear function of SHz the charge on particular atoms, for

hydrogen, and

the

intermediate

electronegativity

equivalent in correlations involving experimental data.

are

formally

Moreover, since the

electronegativity of silicon exceeds that of aluminium electronegativity ( S ) and the charge on the hydrogens HZ silicon replaces aluminium in the framework.

the

intermediate

(6,) increase

as

Additionally, this formulation allows, at least in principle, for the effects of cations and of 'T' atoms other than A1 or Si in the framework,

so

that a single parameter can be obtained for effective correlation (Fig 4.8). However, since the intermediate electronegativity is determined by composition it can take no account of properties which are strongly structure dependent. Preferential siting of aluminiums or local compositional o r differences at active sites cannot be structurally

homologous

systems,

dealt with.

predictions

geometrical

Moreover,

based

on

even

in

intermediate

electronegativity do not, in many cases, appear to reflect experimental results at low levels of aluminium.

For example, in well prepared H-ZSM-5

zeolite the turnover number for n-hexane cracking is constant over aluminium loading down to very low levels of aluminium (66).

However, if turnover

number is a linear function of electronegativity it should continue to increase as A 1 4 0 as indeed should the charge on the hydrogens.

Moreover,

the rate of increase of intermediate electronegativity (SHz) with decrease in aluminium content reaches it maximum value (for 0 zeolites) when A1+0


0.5, typical of

as may be seen (58c) by differentiation of equation ( 7 )

with respect to x (mol fraction of aluminium).

Results for other zeolites

also' suggest that, at low levels of aluminium, properties tend to become linearly dependent upon aluminium content rather than electronegativity and, although discontinuities may arise in very highly dealuminated zeolites as A 1 4 0 (Section 3.2) there are no reports of maximal turnover as A 1 4 0 in any zeolite structure. However, intermediate electronegativity is clearly a feature in the physical/ chemical properties of zeolites and related materials and excellent correlations over a range-of

composition testify to the utility of the

concept. Recent theoretical formulations (160) confirm the electronegativity

86

c

‘E

5

J

CHA

1;5” HOR

Moo

a) Acid hydmxyl stretching frequencies.

1

1

blShift in OH frequc cy sorption of benzene.

on

>

I cl Catalytic

acfiity CH3 CH (OHlCH3 =CH, 423K -CH,CH Integral reactor.

I

38

4.0

I S

I

J

4.2

SANDERSON UECTRONEGATIVTTY

Fig.4-8.Dependence of some zeolite Properties on intermediate electmgativity. - (38).

equalisation to a common intermediate value as proposed by Sanderson and describe conditions sufficient for the

validity

of the

geometric mean

principle. The concept of orbital electronegativity has also been utilised in zeolite

chemistry

(161).

In

this

approach

the

equalisation

of

electronegativity is maintained but partial charges on crystallographically distinct atoms of the same element are distinguished due to differences in bond lengths and angles.

Electronegativity (160a,162) may be defined more

generally in terms of the chemical potential of the electrons

where E is the total energy of the system, N the number of electrons and the electronegativity. is

Na Np, 9

is

If the number of electrons on the atoms in a molecule

etc then the total energy of the molecule may be written as a

function of: charges Zcr Zfi...

(i) the number of electrons on each atom; (ii) the number of etc; and (iii) the internuclear distances

(Rap). For

the

ground state energy of the molecule the equalization of electronegativity requires that

x,=

XIS etc so

that

Using this approach (163) a charge dependent "effective" electronegativity

(XJ of

where

an atom in a molecule is represented as

xk

represents the isolated atom electronegativity, n>

the hardness,

- ,,N ) andbxkand A"& are corrections to the atomic charge, (4, = 2, q, isolated atom electronegativities and hardness due to the confinement of the atom in a molecule.

Confinement in a molecule or crystal results in an

external potential generated by surrounding charges (q

) at distances R

*P

*

This potential is represented by the summation term ( fqk /Rap ) which, in a solid, may be evaluated by a Madelung-type summation. This approach is still in development but has already produced some

88

interesting results on

the

electronic

including those with zeolite frameworks.

properties

of

silica

polymorphs

Moreover, it demonstrates clearly

the importance of structure type in determining charge distribution and average electronegativity. Differences in the properties of silica polymorphs (Si02) are,

of

course,

not

predicted

by

the

Sanderson

intermediate

electronegativity.

1.4.4 A Qualitative Approach to Zeolite Acidity

Zeolite acidity is also discussed in terms of the principle of "Hard and Soft Acids and Bases" (HSAB) due to Pearson (lOa)(lOb).

A theoretical basis

for this principle may be provided by calculating the energy of interaction between an electron acceptor (A) and an electron donor (B) on forming an adduct (AB).

If the interaction energy (AE) is not too large the wave

functions for the product ( of the reactants ('PA, theory.

'PAB )

w,) as

may be approximated taking the wavefunctions starting points and utilizing perturbation

Both valence bond and MO

theory give similar results.

In a

simplified case, considering interactions involving only the highest occupied molecular orbital on an atom of the donor or base (H(B)) and the lowest unoccupied molecular orbital on an atom of the acid (L(A)) the interaction energy can be approximated as

where

QB, QA are the charges on donor and acceptor atoms RAB is the interaction distance is the dielectric constant CH(B) is the MO coefficient of the AO(H(R)) on the donor atom CL(A) i s the MO coefficient of the AO(L(A)) on the acceptor atom EH(B) is the energy of the orbital H(B) and E is the energy of the L(A) (The orbitals H(B) and L(A) are frequently referred to

orbital L(A)

as the frontier orbitals.

For the aluminosilicate ion the oxygen 2p

orbital may be taken as H(B).)

AB is the interaction integral

89

The interaction energy is represented as the sum of two terms, the first representing purely electrostatic interactions and the second representing orbital contributions. Although this separation represents the nature of the wave functions initially chosen rather than physical

"reality"

it has,

nevertheless, been found useful in qualitative discussions of acidity to regard the first term, containing classical electrostatic parameters, as an approximation of hard-hard interactions, and to regard the second term, which represents the perturbation energy and depends upon orbital overlap and energy separation, as a measure of soft-soft (covalent) interactions. Catalytic properties have recently been used

(98c) to evaluate the

relative hardness or softness of the anion in aluminosilicate frameworks

Both the acid strength of the site, which determines the base strength of the anion, and the softness of the anion are assumed to determine the final interaction with the incoming base (or carbocation). The reaction of toluene with electrophiles (X

+

) may be considered as

The pattern established in reactions with different electrophiles is that the harder electrophiles tend to give the ortho- and softer ones the para-isomer. If, for a given electrophile (eg CH30H2+) it is accepted that its hardness will be increased the harder the anion with which it is in contact then the ratio o / p xylene can be used to generate a scale of hardness for various zeolite frameworks.

Of course, the influence of other factors on o / p ,

for

example, shape selective effects must be eliminated. The following is proposed for increasing softness of zeolite framework anions. Na-M, Na-Y, Ca N -Y, H-M(Si/Al = 150), H-M(Si/Al = 40). H-Y(Si/Al = 175!, 0.15 0.7 Ca0.3Na0.4-Y, HOe3Naom7-Y, H0.7Na0.3-Y, H-Y(Si/Al = 8.5), H-ZSM-5(Si/Al = 36.5) 2+

This suggests that exchange of Na+ for either Ca anion.

Increasing Al/Si

or H+ produces a softer

gives softer anions and softness is structure

90

dependent (M < Y <

ZSM-5).

selective effects might

However, particularly in the case of H-ZSM-5 shape modify

results.

Brown's reactivity-selectivity

relationship is found to hold for the Y-type zeolites as reported previously (98a) and also in a series of papers (98b) involving a detailed study of ion-exchanged Y zeolites. It is suggested (98c) that increased softness in the aluminosilicate species (I) favours orbital controlled reactions and, over harder surface anions, charge-controlled reactions are preferred.

In the

methylation of toluene, orbital control tends to give para- and charge control tends to give orthoxylene. The use of HSAB and linear free energy relationships has been

very

effective in correlating a wide range of chemical reactions and, as zeolites become more widely used in organic chemical reactions, it might be that this approach will be effective in correlating different types of reaction. Immediately it would be helpful to use MO theory on simple clusters to generate a "theoretical" scale of softness for comparison with experimental results.

Acknowledgements We thank BP Research Sunbury-on-Thames for an EMRA for research into zeolite chemistry and catalysis.

JD would like to thank research students and

associates, past and present, and the chemicals industry particularly BP Research Centre, Laporte Inorganics, and CrosfieldAJnilever and Shell BV, for support.

91

1

J.N.Bronsted, Recl.Trav.Chim.Pays-Bas,42

2

G.N.Lewis,"Valency and Structure of Atoms and Molecules" ACS Monograph, Chem.Catalog.Co., N.Y., 1923

3

L.P. Hammett and A.J. Deyrup, J.Amer.Chem.Soc., 54 (1932) 2721.

4

R.J.Gillespie and T.E.Pee1, Adv.Phys.Org.Chem., 9 (1972) 1.

5

G.A.Olah, G.K Surya Prakash, J.Sommer, Superacids, John Wiley & Sons, 1985.

6

A.E.Hirschler, J. Catal

7

N.C.Deno, H.E.Berkheimer, W.L.Evans and H.J.Peterson, J.Amer.Chem.Soc., 81 (1959) 2344.

a

C.H.Rochester, Acidity Functions, Academic Press, N.Y., 1970

9

S.

10

(a) R.G.Pearson, J.Amer.Chem.Soc., 85 (1963) 3533. (b) R.G.Pearson (Editor) in Hard and Soft Acids and Bases, Dowden, Hutchinson and Ross, Stroudsberg Pa. (1973).

11

(a) G.Klopman, J.Amer.Chem.Soc., 90 (1968) 223. (b) G.A.Klopman (Editor),in Chemical Reactivity and Reaction Paths, Wiley, N.Y., 1974 ( c ) R.G.Pearson, Proc.Natl.Acad.Sci., USA, 83 (1986) 8440.

(1923) 719.

.

.,

2 (1963), 428.

.

Ahrland, J.Chatt and N.P.Davis, Quart. Rev.Chem.Soc., 12 (1958) 265.

.

12

(a) R.S.Drago and B.B. Wayland, J.Amer.Chem.Soc., 87 (1965) 3571. (b) A.P.Marks and R.S.Drago, ibid., 97 (1975) 3324.

13

(a) V.Gutmann, Coord Chem.Rev., 18 (1976) 225. (b) A. J.Parker, U.Meyer, R.Schmid and V.Gutmann, J.Org. Chem., 43 (1978) 1843. ( c ) W.B.Jensen, Chem.Rev., 78 (1978) 1

14

K.Tanabe, Solid Acids and Bases; Their Catalytic Properties, Academic Press, 1970.

15

(a) H.A.Benesi and B.H.Winginst, Adv. in Catalysis, 57 (1978) 97 (b) A.K.Ghosh and R.A.Kydd, Catal. Rev.Sci.Eng., 27(4) (1985) 539. (c) H.Hattori, O.Jahahashi, M.Takagi and K.Tanabe, J.Catal., 68 (1981) 132.

16

D.W.Breck, Zeolite Molecular Sieves, Wiley, New York, 1974

17

W.M.Meier,in

Y.Murakami et al,(eds),New

.

Developments in Zeolite

92

Science and Technology, Elsevier, (1987) 13. 18

P.A.Jacobs, Carboniogenic Activit? Sci.Publ.Co., (1977)

19

(a)

of Zeolites,,Elsevier

S.M. Csicsery, in J.Rabo,(Editor), Zeolite Chemistry and Catalysis ACS Monograph 171, ACS,Washington,DC, 1976 (b) E.G.Derouane in Intercalation Chemistry, W.M.Stanley and A. J.Jacobson (Eds),Academic Press, N.Y., 1982 101. (c) N.Y.Chen and W.E.Garwood, Catal.Rev.Sci.Eng., 28 (1986)

.

185. 20

H.A.Benesi and B.H.Linquist, Adv. in Catal., 27 (1978) 97.

21

G.W.Skeels and W.H.Flank, in Intrazeolite Chemistry, ACS Sym. Ser,218 (1983) 369.

22

Reference 18 section 1.5

23

H.V.Drushe1 and A.L.Sommers, Anal.Chem.

24

(a) W.J.Kladnig, J.Phys.Chem., 83 ( 979) 765 (b) D.Barthomeuf, J.Phys.Chem., 83 1979) 766, M.W.Anderson and J.Klinowski, Zeolites, 6 (1986) 150.

25

Y.Okamata, T Imanaka and S Teranishi Bull.Chem.Soc., Japan, C13

2

(1966) 1723.

(1970) 3353. 26

H.S.Thuong, B.V.Romanovskii, K.V.Toochieva and L.I.Piguzova, Kin.Katal., 8 (1967) 691.

27

R.M.Barrer, Zeolites as Clay Minerals as Sorbants and Molecular Sieves, Academic Press, 1978.

28

A.Clark, V.C.F.Holm and D.M.Blackburn, J. Catal., 1 (1962) 2461.

29

Hou Shing Thouang. B.V.Romanmovskii, K.V.Topchieva and L.I.Pinnuzova, Kin.Katal., 8 (1967) 691.

30

(a)

(b)

R.J.Cvetanovic and Y.Amenomiya,in Advances in Catalysis and Related Subjects, Academic P r e s s , l 7 (1967) 103. D.Menze1, in R.Gomer (Editor), Interactions on Metal (1975) 101. Surfaces, Springer,

31

C.T. -W.Chu and C.D.Chang, J.Phys.Chem., 89 (1985) 1569.

32

J.Dwyer, unpublished data.

33

J.W.Ward, in J.A.Rabo (Editor) Catalysis by Zeolites, ACS monograph, 171 (1977) 118

34

(a)

P.A.Jacobs and J.B.Uytterhoevet, J.C.S. Faraday 1, 69

(b)

P.J.O'Malley, PhD Thesis, National University of Ireland,

(1973) 359. 1982

.

93

35

(a) D.Dombrowsky, J.Hoffman, J.Fruwert and T.Stock, J.C.S. Farad.l, 81 (1985) 2257. (b) J.Dwyer and P.J.O'Malley, unpublished work.

36

(a)

37

A.V.Kiselev and V.I.Lygin, Infrared Spectra of Surface Compounds and Adsorbed molecules, N a u k a , Moscow (1972).

38

P.A.Jacobs, Cat.Rev.Sci.

39

D.Freude, T.Frohlich, M.Hunger, H.Pfeifer and G.Scheler, Chem.Phys.Letters,98 (1983) 263.

40

M.W.Anderson and J.Klinowski, Zeolites, ( 986).

41

(a) D.Freude, M.Hunger, H.Pfeifer, Chem. Phys. Letts., 128 (1986) 62. (b) H.Pfeifer, D.Freude and M.Hunger, Zeolites,5 (1985) 274. (c) H Pfeifer, W Meilar and D Deininger, NMR of organic NMR compounds Adsorbed on Porous Solids, Ann.Rep. Spectroscopy, Academic Press, London, 15 (1983).

42

J.H.Lunsford, C.W.Zingery and M.P.Rosynek, J Cataly., 38 (1975) 179.

43

J.H.Lunsford, J.Phys.Chem., 72, 12, (1968) 4163.

44

O.E.KLebedeva, E.V.Lunina, 1.L.Aleksandrova and V.L.Baiburskii, Russ.J.Phys.Chem, 60, 1 (1986) 10.

45

C.Defosse and P. Canneson , J.C.S. Faraday.1 (19761, 2565.

47

(a)

V.B.Kazansky in P.A.Jacobs et al. (Editors), Structure and Reactivity of Modified Zeolites, Studies in Surface Science and Catalysis, 18, Elsevier (1984) 61 (b) V. B. Kazansky, V.Yu. Borokov, C .M.Kustov, Proc.7th Int.Conf.Cataly., I11 (1984) 3.

(b)

(c)

Eng. 24 (1982) 415.

J.Dwyer, F.R.Fitch, F. Machado, Guanlin Qin, S.M.Smyth and J. C.Vickerman, J.Chem.Comun., (1981) 433. J.Dwyer, F.R.Fitch, Guanlin Qin and J.C.Vickerman, J.Phys.Chem., 86 (1982) 4574 A.Ashton, J.Dwyer, I.S.Elliott, Y.R.Fitch, G.Qin, M. Greenwood and J.Speakman, in D.Olson and A.Bisio (Editors), Proc.6th 1nt.Zeolite Conf., Butterworths,(1984) 704.

.

J.B.Uytterhoeven, L.G.Christner and W.K.Hal1, J.Phys Chem. , 69 (1965) 2117. (b) G.H.Kuh1, in J.B.Uytterhoeven (Editor), Molecular Sieves, Lueven Univ. Press,(1973) 227. ( c ) P.A.Jacobs and H.K.Beyer, J.Phys.Chem.,83 (1979) 1174.

48

(a)

49

C.Mirodatos and D.Barthomeuf, J.C.S. Chem. Commun., (1981) 39.

50

R.M.Dessau, K.D.Schmitt, G.T.Kerr, G.L.Woolery and L.B.Allemany, 104 (1987) 484. J.Catal

.

94

51

D.Akporiaye, A.P.Chapple, D.M.Clark, J.Dwyer, 1.S.Elliott and D. J.Rawlence in Y.Murakami, A.Iijma and J.W.Ward (Editors), Proc.7th Int. Zeolite Conf., Kodansha-Elsevier, 1986 351.

52

(a) G.O.Brunner, Zeolites,7 (1987) 9 (b) G.L.Woolery, L.B.Alernany, R.M.Dessau and A. W.Chester, Zeolites,6 (1986) 14.

53

P.A.Jacobs and W.J.Mortier, Zeolites, 2 (1982) 226

54

(a) D.H.Olson and E.Dempsey, J.Cataly., 13 (1969) 221. (b) W.J.Mortier, J.Pluth and J.V.Smith, J.Cata1 45 (1976) 367

55

(a) V.Bosacek, S.Beran and Z.Jirak, J. Phys.Chem., 85 (1981) 3856. (b) Z.Jirak, S.Vratislav, J.Zajicek and V.Basacek, J.Cata1 , 49 (1977) 112.

.,

.

56

A.W.Chester, Y.F.Chu, R.M.Dessau, G.T.Kerr and C.T.Kresge, J.C.S.Chem.Comn., (1985) 289.

57

V.B.Kazansky, Proc. 4th Nat.Symp. on Catalysis, Catalysis SOC. of India, Bombay,(1978) 14.

58

(a) D.H.Olson, W.O.Haag and R.M.Lago, J.Cataly., 61 (1980) 390. (b) S.H.Abbas, T.K.Al-Dawood, J.Dwyer, F.R.Fitch, A. Georgopoulos, F.Machado and S.M.Smyth, in B.Imelik et al. (Editors), Catalysis by Zeolites, Elsevier (1980) 127. (c) J.Dwyer, F.R.Fitch and E.E.Nkang, J.Phys.Chem., 87 (1983) 5402. (d) J.R.Sohn, S . J.De Canio, P.O.Fritz and J.H.Lunsford, J.Phys.Chem., 90 (1986) 4847. (e) D.Barthomeuf, Mat.Chem. and Phys., 17 (1987) 49.

59

P.A.Jacobs and R.von Ballmoos, J.Phys.Chem., 86 (1982) 3050.

60

R.H.Meinhold, L.M.Parker and D.M.Bibby, Zeolites, 6 (1986) 491.

61

G.P. Babu, S.G.Hegde, S.B.Kulkarni and P.Ratnasamy, J.Cata1 (1981) 471

62

J.G.Post and J.H.C.van Hooff, Zeolites) 4 (1984) 9.

63

G.T.Kerr and A.W.Chester, Therm.Acta., 3 (1971) 113.

64

A.W.Chester, Y.F.Chu, R.M.Dessau, G.T.Kerr and C.T.Kresge, J.C.S.Chem. Commun., (1985) 289.

65

N.Y.Topsoe, K.Pedersen, E.G.Derouane, J.Cata1

66

W.O.Haag, in D.Olson and A.Bisio (Editors) Proc. Sixth 1nt.Zeol.Conf., (Reno, 1983), Butterworths,(1984) 466, and references therein.

67

A.Auroux, V.Bolis, P.Wierzchowski, P.C.Gravelle and J.C.Vedrine,

., 81

., 81

(1981) 471.

95

J.C.S. Farad. 1, 75 (1979) 2544. 68

R.Beaumont and D.Barthomeuf, J.Cata1.,26 (1972) 218.

69

N.R.Meshram, S.G.Hedge and S.B.Kulharni, Zeolites, 6 (1986) 434.

70

A.L.Klyachko, G.I.Kapustin, T.R.Brueva and A.M.Rubinstein, Zeolites,7 (1987) 119.

71

Y.Ono, in B.Imelik et al.(Editors), Catalysis by Zeolites. Studies in Surface Science and Catalysis 5, Elsevier (1980) 19.

72

R.Beaumont and D.Barthomeuf, J.Catal., 27 (1972) 45; J.Catal., 30 (1973) 288.

73

(a) V.Bosacek, D.Freude, T.Frohlich, H.Pfeiffer-and H. Schmiedel, J.Coll.Int.Sci., 85 (1982) 502. (b) D.Freude, T.Frohlich, M.Hunger, H.Pfeiffer and G.Scheler, Chem.Phys.Lett., 98 (1983) 263. (c) D.Freude, Adv.Coll.Int.Sci., 23 (1985) 21.

74

(a) C.V.McDanie1 and P.K.Maher, Conf .Mol.Sieves. SOC. of Chem. Ind., London,(1968) 186. (b) P.K.Maher. F.D.Hunter and J . Schemer, Ad.Chem. Ser 101 (1971) 266. (c) G.T.Kerr, J. Catal ., 15 (1969) 200; J.Phys.Chem.,72 (1968) 2594; Ad.Chem.Ser., 101 (1973) 219. (d) J.Scherzer, in T.E.Whyte (Editor) Catalytic Materials, ACS Symp.Ser., 128 (1984), 157. ( e ) R.D.Shannon, K.H.Gardner, R.H.Staley, G.Bergeret, P. Gallezot and A.A.Aurow, J.Phys.Chem., 89 (1985) 4778.

.,

75

(a) V.Bosacek, D.Freude, T.Frohlich, H.Pfeifer and E.Schmeide1, J.Coll.Int.Sci., 85 (1982) 502. ( b ) D.Freude, M.Hunger and H.Pfeifer, Z.Phys.Chem., in press (1986). (c) J.P.Gilson, G.C.Edwards, A.W.Peters, K.Rajagopalau, R.F. Worsbecher, T.G.Roberie and M.P.Shatlock, J.C.S. Chem.Comun., (1987) 91.

76

G.Engelhardt, U.Lohse, V.Patzelova, M.Magi and E.Lippma, Zeolites, 3 (1983) 233.

77

A.G.Ashton, S.Batmanian, D.M.Clark, J.Dwyer, F.R.Fitch, A. Hinchliffe and F.J.Machado, in B.Imelik et al.(Editors), Catalysis by Acids and Bases, Studies in Surface Science and Catalysis 20, Elsevier (1985) 101.

78

R.M.Lago, W.O.Haag, R.J.Mikowsky, D.H.Olson, S.D.Hellring, K.D. Scmitt, G.T.Kerr in Y. Murakarni, A. Iiijima and J.W.Ward (Editors) Proc. 7th Int. Zeolite Conf., Kodansha-Elsevier, (1986) 677.

79

U.Lohse, E.Loffler, M.Hunger, J.Stockner and V.Patzelova, Zeolites, 7 (1983) 11.

96

80

(a) B.L.Meyers, T.H.Fleisch and C.L.Marshal1, Appl. Surf. Sci., 26 (1986) 503. (b) T.H.Fleisch, B.L.Meyers, H.J.Ray, J.B.Hal1 and C.L. Marshall, J.Cata1 99 (1986) 117.

.,

81

E.Moretti, S.Contessa and M.Padovan, Chim.Ind., 67 (1985) 21.

82

G.W.Skeels and D.W.Breck,in D.Olson and A.Bisio (Editors) Proc.Sixth 1nt.Zeolite Conference, Butterworths,1984.

83

H.K.Beyer and I.Belenykao,in B.Imelik et al. (Editors), Catalysis by Zeolites; Studies in Surface Science and Catalysis, 5, Elsevier (1980).

84

(a) M.W.Anderson and J.Klinowski, J.C.S.Farad. 1, 82 (1986) 1449. (b) J.Klinocisk, C.A.Fyffe and G.C.Gobb; J.C.S.Farad. 1, 81 (1985) 3003.

85

P.J.Grobe1, P.A.Jacobs and H.K.Beyer, Zeolites, 6 (1986) 47.

86

C.D.Chang, C.T-W.Chu, J.N.Miale, R.F.Bridges and R.B.Calvert, J.Amer.Chem.Soc., 106 (1984) 8143.

87

(a) N.Y.Chen, W.W.Kaeding and F.G.Dwyer, J.Amer.Chem.Soc., 101 (1979) 6783. (b) W.W.Kaeding , C.Chu, L.B .Young, B.Weinstein and S .A.Butter, J.Cata1 67 (1981) 159.

.,

88

J.A.Lercher, G.Rumplmayr and H.Noller,Proc.Int.Symp. on Zeolite Catalysis. Siofok (Hungary) (1985) 71.

89

H.Hattori, O.Takahashi, M.Takagi and K.Tanabe, J.Catal., 68 (1981) 132.

90

K.A.Becker, S.Kowalak, J.C.S.Farad. 1, 83 (1987) 535.

91

(a) B.M.Lok, F.P.Gartsema, C.A.Messina, H.Rastelli, T.B.J.Izod, in G.D.Stucky and F.G.Dwyer (Editors), Intrazeolite Chemistry, ACS Symp. Ser.218 (1983) 41. (b) B.M.Lok, T.B.J.Izod, Zeolites, 2 (1982) 6 6 . (c) A.K.Ghosh and R.A.Kydd, Catal.Rev.Sci.Eng., 27 (1985) 539.

92

(a)

93

P.De.Hulsters, J.Verbiest, J.Philippaerts, G.Peeters and E.F. Vansant,in Y.Marakami, A.Iijima and J.W.Ward (Editors),Proc.7th Int. Zeol.Conf. (Tokyo), KodansharElsevier,(l986) 305 (see references therein).

94

(a) M.Niwa, S.Kato, T.Hattori and Y.Murakami, J.C.S.Farad. 1, 80 (1984) 3135. (b) M.Niwa, Y.Kawashima and Y.Murokami, ibid, 81 (1985) 2757.

95

(a)

A.Becker and S.Kowalak, J.C.S. Farad.1,82 (1986) 2151; Ibid, J.C.S. Farad.1, 83 (1987) 535.

H.Pines, The Chemistry of Catalytic Hydrocarbon

97

Conversions, Academic Press (1981). D.M.Brower, in R.Ribier (Editor), NATO AS1 Ser., Ser. E, 1984, 80, Zeolites: Sci.Techno1, Portugal,(l983). (c) R.E.Banks and P.J.King, in G.Haklor (Editor), Modern Petroleum Technology Part 1, John Wiley, 279 (1984). (b)

96

(a) G.A.Olah, G.K.S.Prakash, J.Sommer, Science 206 (1979) 13. (b) M.L.Poutsma in J.A.Rabo (Editor), Catalysis by Zeolites, ACS Symp. Ser., 171 (1977), 437.

97

L.M.Stock and H.C.Brown, J.Amer.Chem.Soc., 81 (1959) 3323.

98

(a)

P.B.Venuto, J Org Chem, 32 (1967) 1273. B.Coughlan, W.M.Carrol1 and J.Nunan, J.C.S. Farad. Trans.1, 79 (1983) 281; 79 (1983) 297; 79 (1983) 311; 79 (1983) 327. (c) K.P.Wendlandt and H.Bremer, Proc. 7th 1nt.Cong.Cat. (Berlin), IV (1984) 507. (b)

99

J.A.Rabo (Editor), Zeolite Chemistry and Catalysis, ACS Monograph 171, 1976

.

100 G.Poncelet, M.L.Dubru and P.A.Jacobs, Washington, ACS, 40 (1977) 606. 101 M.Zhavoronkov, Kin.Katal., 24, (1973) 512. 102 0.A.Leffler and F.Grunwald, Rates and Equilbria of Organic Reactions, Wiley, New York (1963). 103 S.M.Smyth, P M Thesis, UMIST, 1984. 104 (a) J.R.Kiovsky, W.J.Goyette and T.M.Noterman, J.Catal., 52 (1978) 25 (b) P.B.Koradia, J.Kiousky and M.Y.Asim, J.Cata1 66 (1980) 290.

.,

105 M.T.Aronson, R. J.Gorte and W.E.Farneth, J.Cata1 434. 106 L.A.Pine,P. J.Maher and W.A.Wachter, J. Catal

., 98 (1986)

., 85 (1984) 466.

107 K.V.Topchieva and Ho Chi T'hoang, Kinet.Katal., 14 (1973) 756. 108 H.Hoser and S.Krzyanowski, J.Cata1

., 38 (1975) 366.

109 F.R.Benn, J.Dwyer and A.A.A.Esfahani, Inorg.Chim.Acta., 31 (1978) 101. 110 J.W.Ward, J.Cata1

.,

11 (1968) 251.

111 J.Galuszka, A.Baranski and S.Ceckiewicz, J.Chem.Soc., Farad. 1, (1977) 146. 112 J.C.Vedrine, P.Dejaifve, E.D.Garbowski and E.G.Derouane, in B. Imelik et al. (Editors), Catalysis by Zeolites. Studies in Surf.

98

Sci. and Catalysis, Elsevier, 5 (1980) 29. 113 H.Forster, J.Seebode, P.Fejes and I.Kiricsi, J.Chem.Soc.Farad.1, 83 (1987) 1109.

114 A.Corma, J.Planelles, J.Sanchez-Martin and F.Tomas, J.Catal., 93 (1985) 30. 115 W.O.Haag and R.M.Dessau, Proc. 7th 1nt.Cong. Cat. (1984) 305. 116 H.Kitagawa, Y.Sendoda and Y.Ono, J.Catal.,

(Berlin), I1

101 (1986) 12.

117 G.M.Kramer, G.B.McVicker and J.J.Ziemiak, J.Catal., 355.

92 (1985)

118 C.Bearez, F.Chevalier and M.Guisnet, Reac. Kinet.Catal.Lett., 22 (1983) 405. 119 C.Bearez, F.Avendano, F.Chevalier and M.Guisnet, Bull.Soc.Chim., 3 (1985) 346. 120 A.Corma and V.Fornes in B.Imelik et al. (Editors) Catalysis by Acids and Bases; Studies in Surface Science and Catalysis 20, Elsevier (1985) 409. 121 L.M.Kustov, V.Yu.Borokov and V.B.Kazansky in P.A.Jacobs et al. (Editors) Structure and Reactivity of Modified Zeolites; Studies in Surface Science and Catalysis 18, Elsevier (1984) 241. 122 U.Lohse and M.Mildebrath, Z.Anorg.Allg.Chem., 476 (1981) 126. 123 G.D.Chukin, A.S.Kulikov and S.A.Sergienko, React.Kinet.Catal.Lett., 27, 2 (1985) 287. J.E.Magnusson and F.Valeri, paper presented to Katalistik's 7th Annual Symposium, Venice, Italy, May 1986 (b) G.N.Long, R.L.Chiang, R. J.Pellet and J.A.Rabo, paper presented to Katalystik's 7th Annual Symposium, Venice, Italy, May (1986). (c) J.A.Rabo, R. J.Pellet, J.S.Magee, B.Mitchel1, J.W.Moore, L.L.Upson, J.E.Magnusson. Paper presented at NPRA meeting, Los Angeles, California, March 1986.

124 (a)

.

125 H.Hattori, O.Takahashi, M.Takagi and K.Tanabe, J Catal (1981) 132.

., 68

126 J.Dwyer, K.A.Karim unpublished work; K.A.Karim, MSc Thesis, UMIST, to be submitted 1987. 127 M.Marczewski, J.C.S., Farad. 1, 82 (1986) 1687. 128 J.A.Rabo, Rev.Sci.Eng., 129 S.Shih, J.Cata1

23 (1981) 293.

., 79 (1983) 390.

99 130 G.B.McVicker, G.M.Kramer and J.J.Ziemaiak, J Catal., 83 286.

(1983)

131 A.V.Katcherov and A.A.Slinkin,in P.A.Jacobs et al. (Editors), Structure and Reactivity of Modified Zeolites, Studies in Surface Science and Catalysis, Elsevier (1984), 77. 132 D.A.Best and B.W.Wo.iciechwski

, J. Cata! .,47 (1977)

11.

133 G.K.Beier, I.M.Belen'kaya, M.M.Dubinuin and F.Khange, Izv. Akad. Nauk, SSSR, Ser.Khim., 7 (1982) 1457. 134 W.J.Hehre, L.Radom, P.V.R.Schleyer and J.A.Pople, Ab initio Molecular Orbital Theory., John Wiley and Sons, New York, 1986. 135

(a) S.Beran, J Molecular Catalysis, 10 1981, 177. (b) S.Beran and R.Zahradnik, Kinet.Katal., 18 (1977) 321.

136 W.J.Mortier, J.L.Samer, J.A.Lercher, and H.Noller, J.Phys.Chem., 88 (1984) 905. 137

(a) (b)

P.Geerlings, N.Tarie1, A.Botre1, R.Lissilour and W.J. Mortier, J.Phys. Chem., 88, 1984, 5752. P.J.O'Malley and J.Dwyer, unpublished data.

138 J.Datka, P.Geerlings, W.J.Mortier and P.A.Jacobs, J.Phys.Chem., 84 (1985) 3488. 139 P.J.O'Malley and J.Dwyer, JCS Chem.Commun., (1987) 47. 140 P.J.O'Malley and J.Dwyer, J.Phys.Chem., publication.

submitted for

141 I.D.Mikheikin, I.A.Abronin, G.M.Zhidomirov and V.B.Kazansky, J, Molecular Catalysis, 3 (19771781, 435. 142 Z.Chen, Z.Wang, R.Hong, J.Catalysis, 79

(1983), 271.

143 M.S.Gordon, M.D.Bjorke, F.J.Marsh, M.S.Korth, J.Amer.Chem.Soc., 100 (1978) 2670. 144 A.C.Lasaga, Phys.Chem.Mineral.,8

(1982) 36.

145 P.J.O'Malley and J.Dwyer, unpublished data. 146 (a) (b) (c)

(d)

S.Beran, Z.Phys.Chem. N.F., 137 (1983) 89. S.Beran, J. Molecular Catalysis, 30 (1985) 95. S.Beran, P.Jiru and B.Wichterlova, J. .Phys. Chem. ,85 (1981) 1951. S.Beran, J.Phys.Chem., 85 (1981) 1951.

147 V.B.Kazanzsky, A.M.Gritskov9 V.M.Andreev Molecular Catalysis, 4 (19781, 135.

148 V.B.Kazansky, Russ.J.Phys.Chem, 59

and G.M.Zhidomirov, J.

(1985) 621.

100

149 L.G.Gorb, 1.A.Abroni.n and V.V.Goncharuk, Russ.J.Phys.Chem., 59 (1985) 703. 150 E.Dempsey, J.Catal., 33 (1974) 497; J.Catal., 49 (1976) 170.

.

151 P.J.Mikovsky and J.F.Marshal1, J.Cata1 , 44 (1976) 170; P.J. Mikowsky, J.F.Marshal1 and W.P.Burgess, J.Cata1 ., 58 (1979) 489. 152 B.Beagley, J.Dwyer, F.R.Fitch. R.Mann and J.Walters, J.Phys.Chem., 84 (1984) 1744. 153 A.W.Peters, Book of Abstracts, 183rd Nat.Meeting of the ACS, Las Vegas, N.V., March, 1982, ACS, Washington, DC, (1982) 482. 154 W.A. Wachter in D.Olson and A.Bisio (Editors), Proc.6th Int.Zeolite Conf., Reno (1983), Butterworths,(1984) 141. 155 (a) E.G.Derouane and J.G.Fripiat, Zeolites,5 (1985) 165. (b) J.G.Fripiat, F.Berger-Andre, J.Marie-Andre and E.G. Derouane, Zeolites, 3 (1983) 306. 156 P.J.O'Malley and J.Dwyer, Zeolites, submitted. 157 C.A.Fyffe, G.C.Fobbi and G.J.Kennedy, Chem.Lett., (1983) 1551. 158 K.Itabashi, T.Okada and K.Igawa,in Y.Murakami, A.Iijima and J.W. Ward (Editors), Proc. 7th Int. Zeol.Conf., Tokyo, 1986, KodanshaElsevier (1986) 369. 159 R.T.Sanderson, Chemical Bonds and Bond Energy, 2nd Ed. Academic Press, NY, 1976

.

160 (a) R.G. Parr, R.A.Donnelly, M.Levy and W.E.Polke, J.Chem. Phys., 68 (1978) 3801. (b) R.A.Donnelly and R.G.Parr, ibid, 69 (1978) 4431. 161 T.N.Kyoung, C.Hakze, R.Talkyne, J.S.Mu, J.Phys.Chem., 85 (1981) 2065. 162 R.P.Iczowski and J.L.Margrave, J.Amer.Chem.Soc. 83 (1961) 3547. 163 W.J.Mortier, in Y.Murakami, A.Iiijima and 1.W.Ward (Editors) Proc. 7th Int.Zeo1. Conf., Kodansha-Elsevier,(1986) 423.