Vous êtes sur la page 1sur 30


The Design and Preparation of

Supported Catalysts


1 Introduction
In this Report of catalyst-preparation technology we have placed particular
emphasis on catalyst design as opposed to preparation. A properly designed
catalyst should have the essential attributes of activity, stability, selectivity,
and regenerability. These can be related to the physical and chemical proper-
ties of the catalyst, which in turn can be related to the variable parameters
inherent in the method used for the preparation of the catalyst. In the past
much of the literature on supported catalysts has not included this infor-
mation. In part this was due to the lack of techniques for physically and
chemically characterizing supported catalysts. Many advances have been
made in recent years in this area, as described in Chapier 2, so that the design
of supported catalysts has become a feasible activity.
In addition to a wide range of techniques for the preparation of supported
catalysts a substantial number of supports are available for such systems. In
this Chapter we highlight the technology of catalyst preparation and the role
of the support in its application. In Table 1 are listed the total U.S. sales of
catalyst support materials for 1977.
The predominence of alumina and zeolites is reflected in the literature on
the preparation of supported catalysts and hence in the contents of this
2 General Methods of Preparation for Supported Catalyst
The principal catalyst-preparation technique involves two stages. First,
rendering a metal-salt component into a finely divided form on a support
and secondly; conversion of the supported metal salt to a metallic or oxide
The first stage is known as dispersion and is achieved by impregnation,
adsorption from solution, co-precipation, or deposition, while the second
stage is variously called calcination or reduction.’ It is brought about by a
thermal treatment in either an inert atmosphere or an active atmosphere
of either oxygen or hydrogen. When the active atmosphere is hydrogen the
process is known as reduction. Although calcination/reduction can cause
major problems in catalyst preparation on a large scale, it is a generalization
to say that once the metal species has been bound to the support surface its
B. Delmon, P. Grange, P. A. Jacobs, and G. Poncelet, ed., ‘Preparation of Catalysts 11’,
Elsevier, Amsterdam, 1979.
2 Catalysis
Table 1 US. sales of catalyst supports in 1977*
Catalyst support products U.S. saZes (% wt.)
Alumina 56.9
Zeolites 22.2
Monolith cordierite 7.9
Diatomaceous earth 7.5
Activated carbon 3.0
Silica gel 1.3
Silica/alumina 1.2
Total 100.0
* Adapted from Chem. Tech., 1979, April

degree of dispersion and location will be retained during subsequent treat-

ments. This Chapter therefore concentrates on the dispersion stage of catalyst
preparation rather than the thermal treatment stage, although where this is
known to cause difficulty it is discussed.
The primary aim of applying a catalytically active component to a support
is to obtain the catalyst in a highly dispersed form and hence in a highly
active form when expressed as a function of the weight of the active com-
ponent. This feature of supported catalysts is especially important with
regard to precious-metal catalysts, because it allows more effective utilization
of the metal than can be achieved in bulk-metal systems. However, in the
case of base-metal catalysts the use of the support is often primarily aimed at
improving the catalyst stability. This can be achieved by suitable interaction
between the active material and the support. For example: unsupported
copper oxide is a very active oxidation catalyst but suffers from thermal
instability at high temperatures. However, when copper oxide is supported
on a high-surface-area alumina, its thermal stability is improved.
A wide range of techniques has been employed for the incorporation of
a catalytically active species onto a support material. A summary of the
most widely used techniques is given below as an introduction t o later
Sections in this Chapter, which describe the more important chemical and
physical factors involved in the dispersion of metal saIts onto supports
and their influence on the activity, selectivity, and durability of the catalyst
Impregnation. - Impregnation as a means of supported catalyst preparation
is achieved by filling the pores of a support with a solution of the metal salt
from which the solvent is subsequently evaporated. The catalyst is prepared
either by spraying the support with a solution of the metal compound or by
adding the support material to a solution of a suitable metal salt, such that
the required weight of the active component is incorporated into the support
without the use of excess of solution. This is then followed by drying and
subsequent decomposition of the salt at an elevated temperature, either by
thermal decomposition or reduction. With this method of preparation it is
essential to have an understanding of both chemical and physical properties
J. F. Roth and J. W. Cambell, SOC. Auto. Engineers, 1973, Paper 730277.
Design and Preparation of Supported Catalysts 3
of the support and the chemistry of the impregnating solution in order to
control the physical properties of the finished catalyst. Comment on these
factors is reserved for discussion in a later Section of this Chapter. When
used for the preparation of mixed metal catalysts, care has to be taken to
confirm that a component in an impregnating solution of metal salts is not
selectively adsorbed, resulting in an unexpectedly different and undesirable
concentration of metals in a mixed-metal catalyst. This technique has been
widely used for the preparation of small amounts of catalyst for basic studies,
Adsorption from Solution. - Adsorption is defined as the selective removal
of metal salts or metal ion species from their solution by a process of either
physisorption or chemical bonding with active sites on the support. Depending
upon the strength of adsorption of the adsorbing species, the concentration
of the active material through the catalyst particle may be varied and con-
trolled. This technique is widely used in the preparation of industrial catalysts
as it permits a greater degree of control over the dispersion and distribution
of the active species on the support. In some systems, however, the weight
of the active component that can be incorporated into the support is limited.
Although multiple adsorption is often possible it is not recommended when
close control of physical parameters is required.
Co-precipitation. - The preparation of supported catalysts by the co-
precipitation of metal ions with the support ions usually produces an intimate
mixing of catalysts and support. An example of this technique is the co-
precipitation of metal ions with aluminium ions to produce a precipitated
alumina gel containing the metal hydroxide. This precipitate when calcined
produces a refractory support with active component dispersed throughout
the bulk as well as at the surface. However, in the preparation of multi-
component catalysts, it is possible under improper conditions to obtain a
heterogeneous product because of the different solubility products of the
constituents. Care should be taken therefore to avoid this undesirable situ-
ation by appropriate forethought.
Deposition. - Deposition, as used in preparing supported catalysts, is the
laying down or placing of the active components on the exterior surface of
a support. One means by which this may be achieved is the preparation of
catalysts by sputtering, which involves condensing the metal vapour onto an
agitated finely dispersed support. However, as this process is performed under
a high vacuum, the technique is probably only useful for the preparation of
‘model’ catalysts. Alternatively, the process may be performed in the liquid
phase by the deposition of a metal sol onto a suspended support.
Chemical Vapour Deposition (CVD). - Another example of deposition is the
vapour plating of the support with a volatile inorganic or organometallic
compound. The process requires only a moderate vacuum and is currently
one of the methods under research in industry3 as a means of preparing
catalysts with a purely surface deposition.
Also included in this preparation category is the addition of a precipitating
A. J . Bird, B.P., 1 5 7 8 1 2 3 .
4 Catalysis
agent for the metal ion to a suspension of the support in an impregnating
solution. A layer of precipitated metal ion adheres to the support material,
which can be thermally decomposed as before.
In the case of vapour-phase processes for metal deposition on the support,
only limited control of dispersion and distribution of the metal crystallites
is possible. In the case of liquid-phase systems, they do not provide as wide
a range of catalysts as is possible with techniques based on adsorption from
solution. However, the technique does provide a means of preparing well
characterized surface-impregnated supports.
3 Catalyst Design Parameters
For catalyst design purposes it is first necessary to translate the catalyst
performance parameters into a physical picture of catalyst structure. As we
shall see, different performance parameters can give rise t o different struc-
tural features and so a compromise is generally required. For example it is
commonly found in industrial applications that initial catalyst activity may
be sacrificed in favour of improved catalyst stability, since a lower activity
and a prolonged operating catalyst life is in general preferable to a higher
initial activity that decays rapidly. First, we should therefore discuss some
of the relationships between the catalyst performance parameters and
physical structure.
Activity. - In general activity arises from maximizing both the dispersion and
availability of the active catalytic material. Ideally, from an activity view-
point, the catalyst material should be highly dispersed and concentrated on
the external surface of the support. Already, however, there is an inherent
conflict as high concentrations of active material become progressively more
difficult to disperse.
Stability. - By stability we refer to the loss in activity with time. This is due
to one or several of four main causes; fouling of the active surface with
involatile reaction by-products, sintering or crystal growth of the active
material, poisoning of the active surface by feed impurities, and blockage of
the support pore structure.
Sintering during catalyst use is usually not a problem if catalysts are
properly designed for their end use, although it is perhaps an important
problem during catalyst preparation, activation, and reduction if the impreg-
nated metal is not bound to the support surface. It also becomes an important
factor under the more severe conditions imposed during catalyst regeneration.
Fouling of the active surface by reaction by-products is a real problem,
which typically can be partially met by selective poisoning of the active
ingredient. In a general sense the use of bimetallic supported catalysts would
also commonly fall into this category, since selective poisoning implies a close
control over the ratio of poison to active material. In this case a severe
constraint is imposed upon catalyst design in that both active and moderating
components should ideally be in a constant ratio throughout the catalyst
support, that is to say, the placement of both should be the same.
Poisoning of the catalyst by impurities introduced with the reactants can
Design and Preparation of Supported Catalysts 5
often be minimized by placing the active material deep within the catalyst
support structure, and since most catalyst supports are also good absorbents,
poisons frequently can be selectively removed by such absorption before
reaching the active surface. An example would be the removal of traces of
lead and phosphorous from a car exhaust by the surface of the catalyst
support. A catalyst design modification of this same technique would be the
deposition of a poison-resistant catalyst component close to the surface and
a poison-sensitive component deep within the support. This technique can
be taken even further; an inert material can be used as a poison trap close
to the support’s external surface. In this way each catalyst support particle
can be viewed as coming complete with its own catalyst guard bed. Once
again for poison resistance the location of the active component becomes
a critical factor in proper catalyst design.
Finally, blockage of the support-pore structure is critically dependent
upon the pore-size distribution of the support. Normally a correct balance
of large and small pores is required; the former to aid reactant transport
and the latter to provide the large surface necessary for the optimal disper-
sion of the active components. Whereas one might intuitively expect that
small pores would block more readily, an important exception has only
recently been recognized in the case of ZSM 5 type zeolites. In these the
structure is small enough to prevent the formation of the high molecular
weight involatile by-products that normally are the pore blocking agents,
and yet is still large enough to allow for the transport of reactants and pro-
ducts to and away from the catalytically active sites.4
Selectivity. - Catalyst selectivity can change due either to physical or chemical
reasons. For sequential reactions diffusivity and mass transport through the
pore structure can lead to apparent loss in selectivity in the formation of
intermediate products. Location of active ingredients and pore-size distri-
butions are therefore again of importance. Changes in selectivity can also
arise from changes in intrinsic chemical activity of the active component.
Typically this can be affected by use of multicomponent catalysts in which
case, as we saw earlier for stability improvement, the location of the differ-
ence components ideally should be the same. A specific example of this type
of selectivity arises in the case of multifunctional catalysts in which a hydro-
genation function is combined with an acid function. Since the latter is
typically provided by the support and the former by the impregnated
material, a uniform impregnation is required.
Regenerability. - Regenerability refers to the reactivation of a catalyst, which
typically will involve an air calcination followed in some cases by a redis-
persion of the active components. From the catalyst design viewpoint this
will generally imply enhanced thermal-hydrothermal stability of the support
itself, combined with stability of the active components under the high
temperature oxidizing environments required for the oxidation of the de-
activating carbonaceous deposits. It is now generally recognized that many
J. R. Anderson, K. Foger, T. Mole, R. A. Rajadhyaksha, and J . V. Saunders, J. Catal.,
1979, 58, 114.
6 Catalysis
metals sinter more readily under oxidizing conditions and in extreme cases
may even dissolve in the underlying support and become effectively removed
from the reaction system. A further complication arises with multicomponent
catalysts in which the combination ratio is all important, since such combin-
ations frequently are destroyed under oxidizing conditions.
Summarizing this Section, the activity, stability, and selectivity are deter-
mined by the correct dispersion and location of the active ingredients. Dis-
persion, location, and regenerability are each in their turn determined by
the interaction of the active components with the support surface and with
each other during preparation, activation, use, and regeneration. It is the
purpose of this Report to examine in greater detail the extent to which our
knowledge of these matters has progressed in the past few years. It is our
thesis that substantial progress has been made and although much remains
to be done, the present status is such as to justify our claim that we can
now.talk of catalyst design rather than catalyst preparation.
4 The Control of Metal Dispersion and Location during
Catalyst Preparation
It is not our purpose in this Section to give an extended list of catalyst-
preparation recipes. Neither is it our intent to give an exhaustive and com-
plete review of all the published papers on catalysis in recent years that may
have some aspect of catalysis preparation. The former has been amply covered
in recent reviewssp6and the proceedings of two recent symposia are devoted
to this subject.’*’ The latter would be a virtual impossibility. Rather we
intend to try and identify those factors that contribute most to the prepar-
ation of viable catalysts and the recent papers that exemplify these require-
ments and contribute to what we feel have been some of the major significant
advances in this field.
Techniques used for Characterization. - As is true in other fields of scientific
endeavour, much of this advance has been due to the introduction and wider
use of new analytical techniques for catalyst characterization. These are
discussed elsewhere in this Volume, but since these form an integral part of
much of the work to be discussed, some recapitulation is appropriate. This
recent period has been particularly fortunate in the introduction and dis-
semination of these newer techniques, and this has done much to put a
firmer foundation to catalyst preparation and augers well for the immediate
The use of electron probe microanalysis (EPMA)6 is rapidly becoming
standard practice. This gives us a relatively rapid and unambiguous method
for investigating the location, not only of active materials but also potential
catalyst poisons. The use of ESCA for characterizing surface-metal concen-

J. R. Anderson, in ‘Structure of Metallic Catalysts’, Academic Press, New York,

R. B. Anderson and P. T. Dawson, ed., in ‘Experimental Methods in Catalytic
Research’, Academic Press, New York, 1976.
B. Delmon, P. Jacobs, and G . Poncelet, ed., ‘Preparation of Catalysts I,, Elsevier,
Amsterdam, 1976.
Design and Preparation of Supported Catalysts 7

tration has also been reported.8 The use of temperature-programmed reduction

and oxidation (TPR,TPO),8-’3is rapidly becoming standard for the charac-
terization of supported catalytic species, particularly with regard to metal-
support and metal-metal interactions. On an atomic scale the use of X-ray
absorption edge fine structure (EXAFS) is rapidly growing and one is now able
to obtain a detailed picture of atomic environment and co-ordination. The
application of techniques, which likewise determine metal co-ordination from
radial electronic distributions, have also been r e ~ 0 r t e d . ll5~ ’These, together
with, should we say, the classical although recently improved techniques of
electron microscopy and gas adsorption for measuring metal dispersions, now
allow for a great deal of sophistication in catalyst characterization.
Dispersion of Active Ingredients. - While we shall see that it is incorrect to
separate dispersion from metal location completely, and indeed this forms
the basic theme of this polemic, for convenience the two subjects are best
discussed separately. The art of catalyst preparation is, has been, and prob-
ably always will be the art of preparing a material or suitable combination
of materials in a highly dispersed form, and in maintaining that dispersion
and combination against every natural thermodynamic tendence towards
agglomeration. All the techniques of catalyst preparation discussed earlier
reduce to two simple steps, dispersing the active material in a liquid or
gaseous form and immobilizing this dispersed material as it is reconverted
to an insoluble solid form. Broadly speaking there are two ways of doing
this. Either one converts the material itself to an insoluble form or one
allows it to react with an insoluble support surface. If one chooses the latter,
it is necessary that the support surface should be reactive with the soluble
or gaseous active material. The majority of commercial catalysts of interest
are typified by this second approach for the reason that in this way an
atomic dispersion is in principle possible and in practice more closely
approached. For low-surface area generally unreactive supports, however,
the former approach must necessarily be used. With many of the compounds
used for catalyst impregnation hydrolysis giving colloidal solutions is always
a possibility and may not always be recognized. This is no doubt the reason
for some of the cases of surface impregnation observed, which could arise
by filtering out or coagulation and precipitation of a colloid during impreg-
For the reasons given above and the fact that most of the published
catalyst informatioh is characterized by the use of high-surface area reactive
supports (primarily alumina and the zeolites and to a less extent silica and

G. J. K. Acres, PlatinumMet. Rev., 1980,24, 14.

V. C. F. Holm and A. Clark, J. Catal., 1968, 11, 305.
J. W.Jenkins, B. D. McNicol, and S . D. Robertson, Chem. Tech., 1977,7 , 317.
J . W. Jenkins, in ‘Preprints 6th Canadian Symposium on Catalysis’, 1979.
P. A. Jacobs, J. P. Linart, H. Nijs, J. B. Uytterhoeven, and H. V. Beyer, J. Chem.
SOC.,Faraday Trans. I , 1977,7 3 , 1745.
l3 P. A. Jacobs, H. Nijs, J . Verdonck, E. G. Derouane, J. -P. Gilson, and A. J . Sirnoens,
J. Chem. SOC.,Faraday Trans. I , 1979,7 5 , 1196.
l4 P. Ratnasamy and A. J. Leonard, Catal. Rev., 1972,6 , 293.
’’ P. Gallezot, A. Bienenstock, and M. Boudart, Nouv. J. Chim., 1978,2 , 263.
8 Catalysis
activated carbon) most of the following discussion will pertain to this second
type of system.
Interaction Between an Active Ingredient and the Support Surface. - For
simplicity in this Section we will consider a limited number of illustrative
interactions important in catalyst preparation. By far the most important
of these in terms of our objective of immobilizing the active material on
the support surface is either the cationic or anionic exchange with protons
or hydroxyl groups on the surface, respectively [e.g., reactions (1) and (2)] .
In general C and A will represent complex cations or anions. An even stronger
interaction may result particularly upon drying and calcination from the
decomposition of these complex ions with the result that a ligand is dis-
placed and the support itself therefore becomes a ligand [e.g.,reaction (3)] .
This process may proceed further with the displacement of a second ligand
and the support then becoming, in effect, a bidentate ligand [e.g., reaction
( 4 ~ .
Cationic Exchange ,
SOH' + C' --+ SOC' H+
+ (1)
Anionic Exchange ,S(OH)- + A- --+ :SA- + (OH)- (2)

We postulate that binding to the surface can be either direct, via metal-
metal bond, or indirect through a bridging ligand oxide ion or possibly a
bridging chloride or hydrogen atom. In general several of these surface
compounds will coexist and, in favourable circumstances, these can now be
identified and quantified by TPR.ll
In certain cases the surface of the support may be pre-modified to either
increase or decrease its absorptive capacity. Techniques for the former have
been thoroughly explored in the area of anchoring homogeneous catalyst
complexes and metal clusters. Since this subject has been amply reviewed16
we will not discuss it further. These techniques primarily involve pre-treatment
of the support surface with a compound that can serve as a bridging ligand.
Techniques for decreasing the absorptive capacity are also of importance and
these will be covered later in greater detail when we come to consider metal
location on a catalyst support.
A completely different type of interaction from the above, in all of which
the metal ends up attached to the support surface, is that in which the
support acts as a reactant to precipitate the active material in a highly dis-
persed form within its pore structure but not attached to the support.
Examples of this latter type of interaction would be the precipitation of a
metal hydroxide from an acidic impregnating solution by a strongly basic
support, e.g., a sodium X zeolite, and the precipitation by reduction to the
J. M. Barret and A. K. Smith, in 'Fundamental Research in Homogenous Catalysis',
Plenum Press, New York and London, 1976,
Design and Preparation of Supported Catalysts 9
metal by strongly reducing supports, e.g., the aldehyde surface groups on
some activated carbons. Very little systematic work has been reported on this
type of interaction, but practically these types of interactions can be very
important. Most of the remainder of this part of our discussion will concern
ionic exchange behaviour during catalyst impregnation.
Thermal Treatments and Reduction of Adsorbed Species. - After impregna-
tion many catalysts require a thermal treatment and/or a reduction stage to
render the adsorbed metal species active. Thermal treatments may take the
form of low-temperature drying operations (up to 150 "C) simply to remove
water, although some decomposition of species such as chloroplatinic acid is
known to occur within this temperature range particularly on relatively
non-reactive supports such as silica gels [reaction (5)]. Treatments at tem-
peratures between 150 and 500°C are principally used to decompose the
adsorbed species to the metal or metal oxide [reactions (6), (7), and (S)] .

[Pd(NH3)4]+++ O2 -
HZPtCl, -PtC14

- PdO
+ 2HC1 (5)
R~Cl3.5H20 - Pt
Thermal treatment alone to form the metal is not always desirable and

can lead to crystallite and in extreme cases metal re10cation.l~

Similar effects can occur, particularly in zeolites on reduction with hydrogen.
For example, if a Pt(NH,)2+-CaY zeolite catalyst is heated in air the complex
decomposes at about 25OoC, and on subsequent reduction in hydrogen at
400°C a highly dispersed platinum catalyst is obtained.'' However, if the
heat treatment and reduction stages are carried out simultaneously at about
3OO0C, then the decomposition process gives rise to a platinum hydride
species, Pt(NM3)2H2, which being uncharged is highly mobile and leads to
agglomeration of the metal. When Pt(NH,)?-NaY catalyst is heated in vacuo
at 350°C a singly charged platinum hydride is formed within the zeolite
pores and cavities, and on reduction with hydrogen at 400°C provides a
highly dispersed metal catalyst [reaction (9)] . Temperature-programmed
reduction and oxidation can provide valuable guides to the thermal treatment

[Pt(NH3)4]2+ -
and reduction stages of catalyst p r e p a r a t i ~ n . ~ - l ~
PtH+ + H+ etc. (9)

5 Ion-exchangeBehaviour
Two factors are of importance and are best discussed separately. First, we
need to examine the nature and ionic composition of the metal solutions
used to impregnate the support and secondly, the nature and concentration
T. A. Dorling, B. W. J. Lynch, and R. L. MOSS,
J. Cutul., 1971, 20, 190.
C. H. Bartholomew and M. Boudart, J. Cutul., 1972, 25, 173.
I. R. Leith, CHEMSA, 1978,4, 72.
10 Catalysis

of the exchange sites on the catalyst surface. It would seem trite and un-
necessary to observe that one would expect poor results from trying to
cation exchange an anionic support and vice versa, if this were not in fact
what is so frequently attempted. Similarly one would not normally expect
to attempt cation exchange of protons at low pH or anion exchange of
hydroxyl ions at high pH and yet this again appears to be what is frequently
attempted. (One can see immediatly, however, that there are going to be
problems when one tries to co-impregnate two metals, one in a cationic form
and the other in an anionic form!)
For the purposes of the ensuing discussion we would indicate that of the
commonly used catalyst supports: (a) zeolites are strong cation exchangers;
( b ) silica is a weak cation exchanger; (c) alumina is a weak anion and cation
exchanger; ( d ) magnesia is a stronger anion exchanger; (e) carbon has peculiar
and variable properties tending to form charge-transfer complexes as an
electron donor, but it also can act as a weak cation exchanger by virtue of
acidic surface oxide groups.
Let us therefore review in general terms what is known about ionic absorp-
tion on surfaces.
Simple Cation Adsorption from Solution. - The relevant equation here is
(10). This equilibrium is largely determined by the pK, of the surface acidic
group and shifts to the right at a high solution pH. The affinity of the func-
tional group for the metal ion determines the relative adsorption strength
of the different metals and this is largely a function of the charge to radius

in the same order as the cation charge C4+> C* > C2+> C -

ratio of the cation. Hence, the affinity of a given cation adsorber is generally
' H+ and a high
valent cation, e.g., Cr3+,would displace that of a lower valency, e.g. , Cu2+.
SOH+ + C+ (
= SOC'
, + H+ (10)
Simple Anionic Adsorption from Solution. - Here the relevant equation
would be (1 l), and again the equilibrium would be pH dependent but in this
case shifts to the left at a high pH. It is also dependent upon the ionic affinity
which increases with anionic polarizability and ionic charge, e.g., I- > Br- >
C1- > F- and SO:- > Cl-.
S(0H)- + A- SA- + (OH)- (1 1)
Complex Ionic Adsorption from Solution. - Much of the metal adsorption
that normally is of importance in catalyst preparative chemistry concerns the
interaction with complex ions. Obviously, for example, all metal anionic
adsorption must of necessity involve the metal as a complex ion. This ten-
dency to form complex ions can cause substantial deviations from what one
might normally expect from considerations of simple ion-exchange equilibria.
This is true in the case of ferric chloride, for example reactions (12)-(15).
Fe3+ C1- % [FeClJ2+ (12)
[FeCl] 2+ + Cl- [FeC12]+ (1 3)
[FeC12]++ Cl- * FeCl, (14)
Design and Preparation of Supported Catalysts 11
Table 2 Ionic adsorption of groups VIII and IB metals*
Key :
C Cationic
A Anionic
N Neutral
W Weak Anionic Absorption
I Moderate Anionic Absorption
S Strong Anionic Absorption
M Molarity of Solution

* Adapted from D. C. Kennedy, Chem. Eng., 1980, 12, 186 and reproduced by per-

FeC13 + Cl- * [FeClJ (1 5 )

Similar equations can be written in terms of the hydroxide ion, i.e., pH,
giving a series of hydrolysed cations of different valence, even as seen above
to the extent of converting a nominal cation into a complex anion. These
complex anions tend to be very strongly adsorbed because of their symmetry
and high polarizability.
A very useful review of the subject of metal adsorption has recently
appeared,20 which gives tables of the relative strengths of anionic and cationic
adsorption for each of the metals in the periodic table, with separate tables
for each of the different anion environments, i.e., F+, Cl-, NO,, CN-, and
SO:-. Table 2 shows some of the results from this reference that pertain to
the catalytically important Group VIII and IB metals. There would seem to
be ample scope for more detailed work along these lines for systems of more
particular catalytic interest, e.g., those containing ammonia as a complexing
ligand. Some more qualitative work has appeared in several papers, which
will be discussed later.
It is important to know how strongly a metal is adsorbed since it is this

D. C, Kennedy, Chem. Eng., 1980, 87, 106.

12 Catalysis
parameter that determines both the dispersion and location of the metal on
the catalyst surface. When impregnating more than one metal these same
factors will determine whether both metals are separated or in fact are in
close proximity. From a catalyst preparatory point of view the above con-
cepts can guide, if not dictate, which salts to use and which supports are
likely to be most effective.
Nature of the Catalyst Support. - Of equal importance to the ionic character
of the impregnating solution is the ionic exchange type and capacity of the
catalyst-support surface. This is, of course, directly related to the chemical
structure of the support surface. To say that information in the catalyst
literature on the chemical structure of support surfaces is sparse is almost
an understatement. Evaluations of this important catalyst-preparation para-
meter are almost without exception entirely overlooked in the published
literature on catalyst preparations. A very limited amount of information
can be found in a review of 1970.21 This review is, however, primarily con-
cerned with the physical rather than the chemical structure of catalyst
supports. The chemical reactivity of an oxide support surface would appear
t o depend upon the extent of its hydroxylation. This in turn depends upon
the chemical type of the support, the way it was made, and particularly
upon its previous thermal history. A few generalizations can be made, as
Silica Supports. The active exchange sites on a silica surface are the silanol
groups, which are weakly acidic. Cationic exchange takes place at pH values
greater than 5.22 The surface concentration of silanol groups appears to
depend upon the pH of the solution from which the silica is formed, decreas-
ing as the pH increases. Hence acid precipitation (pH 2) gives an almost fully
hydroxylated surface (and a high surface-area, low pore-volume gel), whereas
base precipitation (PH6) gives a surface perhaps only 40% hydroxylated
(and a medium surface-area, high pore-volume gel).’” 24 Crystalline silica
(silicalite) prepared at a very high pH is reported2’ to be hydrophobic and
devoid of surface hydroxyls. The concentration of active silanol groups can
be reduced by thermal dehyroxylation above 400 K, being virtually complete
at 900 K?’ Silicas made by flame hydrolysis of SiC14 will in general have low
exchange capacity because of the high temperature of formation. Similarly
diatomaceous earths will have low capacity because of calcination during
preparation in addition to their inherently low surface areas and possibly
their organic odgin. The siloxane bonds resulting from thermal dehydroxyla-
tion are resistant to subsequent hydrolysis.
The silanol groups can also be deactivated chemically, although this would
not be normal commercial catalyst-preparation practice. Possibilities would

21 B. G. Linsen, ed., ‘Physical and Chemical Aspects of Adsorbents and Catalysts’,

Academic Press, New York, 1970.
H. A. Benesi, R. M. Curtis, and H. P. Studer, J. Cutul., 1968, 10, 328.
E. G. Acker, J. Colloid Interface Sci., 1970, 32,41.
F. Yu. Sheinfain and I. E. Neimark, Kinet. Kutul., 1967, 8, 443.
E. M. Flanigan, J. M. Bennett, R. W. Grose, J. P. Cohen, R. L. Patton, R. Kirchner,
and J. V. Smith,Nuture (London), 1978, 271, 512.
Design and Preparation of Supported Catalysts 13
be ligand replacement by, for example, fluoride or esterification, e.g., by
benzyl alcohol and silanation. These silanol groups can also interact very
weakly with complex anions either by ion pairing or hydrogen bonding. Such
complexes are readily hydrolysed however.
Aluminas. Again the active exchange sites are the surface hydroxyl groups,
which now have a more basic character and will also exchange or react with
anions. These surface hydroxyl groups can again be removed by thermal
dehydroxylation or be ‘deactivated’ by anionic replacement by anions of
both minerals and organic acids. The dehydroxylated surface is more readily
rehydrated and is fairly readily hydrolysed by mineral acids,
Aluminas have a weak cationic exchange capacity above pH 7, i.e., weaker
than that of silica:’ which possibly arises from protonation associated with
more acidic anions, e.g., F-, SO:-, NO, on the surface. Although a less effec-
tive cation exchanger than silica at the same pH, uptake may be greater for
a given salt solution since the alumina itself is more basic than silica and will
shift the pH to higher values.
Silica-Aluminas. Amorphous. These are generally considered t o be cationic
exchangers owing to their protonic character. The adsorption character may
also vary depending upon whether they were prepared by ‘acid side’ pre-
cipitation (e.g., by adding sodium silicate to aluminium sulphate in which
case the alumina tends t o precipitate last giving more of an alumina type of
surface) or whether conversely if prepared by a ‘base side’ precipitate (in
which case the surface tends towards that of silica). There is no published
information that differentiates what effects, if any, this may have.
Crystalline (Zeolites). Again the zeolites because oftheir protonic character
are excellent cationic exchangers at least for simple cations. Amongst all the
usual catalyst supports used these are perhaps the only ones in which the
physical structure can have a dominating effect on catalyst preparation. The
generally small pore size gives rise to a substantial molecular-sieve effect,
preventing the access of many of the more common complex cations, i.e.,
those of the platinum group metals. Despite a very large and rapidly growing
literature on the use of zeolite catalyst supports, no work has yet been
reported that explicitly deals with this very real problem.
Magnesia. Magnesia is not a particularly useful support in itself since it
hydrates and dehydrates too readily leading to a loss in physical integrity
of the catalyst granule. It is, however, a not infrequent component of mixed
catalyst where because of its greater basicity it can confer a useful anionic
exchange capacity.
Carbon. The uses of carbon as a catalyst support have recently been dis-
cussed.’6 Four broad classes of carbon support must be differentiated,
activated charcoals, chemically activated charcoals, carbon blacks, and
(a) Activated charcoals. These are produced by steam/air (or COz) acti-
26 C. Moreno-Castilla, 0. P, Mahajan, P. L. Walker, jun., H. -J. Jung, and M. A. Vannice,
Carbon, 1980, 18, 271.
14 Catalysis
vation of carbonaceous matter such as coal, wood, or peat. Typically metal
adsorption is more physical in nature and the active ingredients have to be
immobilized by subsequent reduction. Many charcoals contain appreciable
quantities of silica and potash. The former can contribute to the adsorption
properties and also incidentally the attrition resistance, and the latter to
base-catalysed hydrolysis within the pore structure. The high potash content
of organically derived charcoals under the oxygen-deficient activation pro-
cedures that produce CO, can also lead to the presence of appreciable con-
centrations of potassium formate, which makes some carbons very good
reducing agents. (Surface aldehyde or quinhydrone groups may also contri-
bute to the reducing properties.)
Although normal activated carbons have very little ionic-exchange activity
this can be increased by chemical and electrochemical ~xidation.~'This
treatment generates weakly acidic (carboxylic acid?) oxygenated surface
groups, having cationic-exchange properties. In this case it was shown that
metal uptake followed the sequence that suggests this cationic exchange
capacity, i.e., Pt(NH3)2(N02)2 > Pt(NH&(OH), > Pt(NH&C12 > H2PtCl,.
( b ) CStlemically activated charcoals. These typically are activated at much
lower temperatures and have a high concentration of surface functional
oxygenated groups. Typically they will also have a high residual phosphate
content. Both of these can contribute to high cationic-exchange capacity.
Reducing activity would not however be expected.
(c) Carbon blacks. These have a lower surface area and a much higher
oxygen content than the activated charcoals. Again, the high oxygen content
suggests the possible presence of surface acidic oxides and weak cationic
exchange activity might be expected.
( d ) Graphite. These supports typically have a very low surface area and an
unreactive surface. Metal adsorption appears to be associated with disloca-
tions and discontinuities in the basal graphite planes, also possibly with
partial intercalation between the graphite planes. A recent report28 claims
that better resistance to subsequent metal sintering is obtained if the graphite
is pre-oxidized to give etch pits within which the metal crystallites can lodge
on the surface of these graphites.
Monolith Catalysts. Monoliths are ceramic or metallic structures made up of
many parallel channels of circular, square, triangular, hexagonal, or sinusoidal
shape. By controlling channel diameter (cell geometry) and wall thickness,
monoliths of varying cell density can be produced and this in turn determines
the important parameters of geometric surface area and hydraulic diameter.
Newer forms of metallic monoliths are becoming available made by the com-
pression of wire or knitmesh materials to varying degrees.
The microstructure of monoliths is important particularly with ceramic
monoliths when the chemical nature and structure of the crystalline and
glass phases, together with the pore structure, determine the thermal expan-
sion, thermal conductivity, melting point, surface area, and strength of the
D. R. Lowde, J. 0. William, P. A. Attwood, R. J. Bird, B. D. McNicol, and R. T.
Short, J. Chem. Soc., Faraday Trans. 1 , 1979, 7 5 , 2312.
H. R. K u n z , U . S . P . , 4 0 2 8 2 7 4 .
Design and Preparation of Supported Catalysts 15

product. The relationship between these parameters is considered in some

detail by DeLuca and Campbell?’
Four basic methods of catalyst manufacture are used with monoliths.
(i) Incorporation of catalyst into the monolith during manufacture
(base-metal catalysts only).
(ii) Coating of the monolith directly (base-metal and precious-metal
catalysts) .
(iii)Washcoating monoliths with an oxide coat then catalysing (base-
metal and precious-metal catalysts).
(iv) Preforming the catalyst and incorporating it into a washcoat (base-
metal and precious-metal oxide catalysts).
Each method has specific advantages and disadvantages depending on the
(i) Incorporation of the catalyst into the monolith during manufacture.
This method always leads to poor utilization of the catalytic material; some
is buried within the crystalline matrix and is not accessible to reactants. As
the catalytic material modifies the monolith’s physical properties, either a
limit has to be placed on the amount of catalytic material incorporated or
some deterioration in monolith performance has to be accepted. This method
of manufacture is reserved for the preparation of base-metal catalysts, par-
ticularly nickel and manganese catalyst^.^.^^ Care has to be taken to avoid
the formation of spinels, etc., although in some cases the formation of a
complex oxide structure is actively
(ii) Coating of the monoliths directly. Coating monoliths directly has been
accomplished on a number of occasion^^^-^^ and several techniques used.
With ruthenium and platinum catalysts the monolith is usually soaked in a
dilute solution of the salt (RuC13 or H2PtC16)dissolved in water or a simple
organic solvent, e.g. ethanol or acetone. Very little science is involved in the

catalysing procedure and in general poor catalysts result. The main difficulty
is that the maximum metal dispersion is controlled by the surface area of
the monolith, which is extremely low (0.1-1 m2 g-’). Sintering takes place
rapidly because of the high crystallite density per unit. By multiple dipping,
base-metal oxide coats of greater dispersion can be obtained but adherence
of these coats is usually low.
By far the greatest effort with monolith catalyst preparation has been
devoted to methods (iii)and (iv).
(iii) Washcoating then catalysing. The washcoating of the monolith first
followed by a catalysing step is a direct application of the usual principle
of precious-metal pelleted-catalytic techniques, but (iv), the incorporation
of the catalyst into the washcoat medium either as a preformed catalyst or a
J . J . Burton and R. L. Garten, ed., ‘Adv. Mat. Catal.’, Academic Press, 1977.
30 J. J. Benbow and L. W. Lord, U.S.P., 3 824 196.
31 H. Berger, U.S.P., 3 533 753.
32 T. P. Koblynski and B. Taylor, U.S.P., 3 840 389.
33 D. J. Cassidy, M. Esper, and D. J . Ronnie, Ger. P.,DT 2 349472.
34 C. A. Leech and L. E. Campbell, in ‘Spinel Solid Solution Catalysts for Automotive
Applications’, Am. Chem. SOC. Spring Meeting, Los Angeles, California, 1974.
35 M. Shelef and H. S. Gandhi, Platinum Met. Rev., 1974, 18,2.
16 Catalysis
soluble catalyst component, is taken from base-metal catalyst-manufacturing
practice. Washcoats usually consist of high area (50-250 m2 g-') oxides, e.g.,
A1203, Si02, Ti02, Zr02, etc., together with bonding agents such as hydrous
alumina, silica, etc. Much of the practice of washcoat manufacture is devoted
to ensuring that the hydrous oxide forms a strong bond between the wash-
coat components. Monolith porosity and the physical and chemical nature of
the monolith surface play a large part in ensuring washcoat a d h e r e n ~ e . ~ ~
Washcoats are not simple dispersions of insoluble oxides in water, but are
complex mixtures whose chemical and physical properties are modified by
the addition of acids, etc., milled for prolonged periods to obtain slips that
have sharply defined rheological properties. Once the monolith has been
coated, usually by a dipping technique, it is drained, dried at 100 O C , and
then fired to between 400 and 750°C to develop the bonds essential to its
adherence. Washcoating can be accomplished by dipping monoliths into
solutions of metal salts in water or organic liquids, then firing to develop the
oxide. This procedure is slow in that it requires multiple dipping and is not
generally used industrially. Mixed coats of successive oxide layers, e.g., Al2O3
and Si02, can be built up by this method, however, the process is not easy
to control .37-39
Methods of catalysing the washcoat can be very simple ones of dipping the
washcoated monolith in solutions of metal ions, but great care has to be
taken to ensure that the bonding properties of the washcoat are not destroyed
by the treatment. By far the best practice is to use the exchange properties of
the washcoat material with suitable metal salt solution^.^*^^ Once impreg-
nated it is usually necessary to employ a thermallreductive treatment to
convert any precious-metal species to the metallic state.
(iv) Incorporating catalysts into the washcoat. The method of incorpor-
ating the catalyst into the washcoat inevitably leads to poor utilization of
the catalytic material as some of it will be buried deep within the washcoat,
and some of it will be covered by bonding agents. The process of adjusting
the rheological properties of the washcoat can give rise to dissolution and
relocation of catalytic components. The method has, however, been used
with complex barium ruthenate ,35 and copper chr0mite.4~
6 Catalyst Case Studies
It is the purpose of this Section to review some of the more significant
recent papers on catalyst preparation particularly with regard to some of
C. Keith, P. Kenan, and D. Bair, U.S.P., 3 565 830.
37 G. F. Foster and H, E. Meissner, Ger. P., DT 2 317 560.
E. W. Albers and G. C. Edwards, U.S.P., 3 730910.
39 M. W. Pepper, J. P. DeLuca, R. P. Rhodes, and L. S. Bernstein, SOC.A u t o . Engineers,
1975, Paper 750684.
40 J. C. Summers and S. A. Ausen, J. Catal., 1978, 52,445.
B. Harrison, J. Heffer, and F. King, Preprints 7th Int. Congr. on Catalysis, B 2 Tokyo,
42 P. K. Gallagher, D. W. Johnson, J . P. Remika, F. Schrey, L. E. Trimble, E. M. Vogel,
and R. J. H. Voorhoeve, Mater. Res. Bull, 1975, 10, 529.
43 R. D. Shoup, K. E. Hoekstra, and R. J . Farrauto, 'Thermal Stability o f Copper
Chromite Auto Exhaust Catalyst', Am. Cer. SOC.Annual Meeting, Chicago, Illinois,
April 29, 1974.
Design and Preparation of Supported Catalysts 17
the points raised in our earlier discussion. We have already commented upon
the fact that none of these workers give details or evidence of surface charac-
terization of the supports that they used. This does not militate against the
usefulness of many of the results, as in most cases the same support was used
and the studies concerned changes in impregnating solution. A further criti-
cism of much of this published work is that in many cases no conscious effort
was made to control or note the pH of the impregnating solutions. This is
certainly a most important preparation variable and one that in any industrial
catalyst preparation must be very carefully controlled if reproducible cata-
lysts are to result. The initial reaction that pH measurement is impossible
when using a dry impregnation based upon the more usual incipient wetness
technique is a specious argument, as the titratable acidity or basicity of both
the impregnating solution and support used are generally measurable. Even
details of the pH of the impregnating solutions used and their possible vari-
ation would be useful information.
Alumina-supported Catalysts. - Because of the long standing commercial
importance of platinum/alumina catalyst much work has been done in this
area. Many workers have been concerned with varying the dispersion and
metal placement by using chloroplatinic acid in conjunction with a competing
anion. Hence the use of HNO,, HF, and HCl," citric HNO, and
H O A C , ~and HCl and ethanolamine4' have all been reported. Aboul-Gheit4'
used an ion-exchange technique with a three-fold excess of solution. In such
a case the metal salt has to diffuse into the catalyst pellet from the excess of
solution; the results are reproduced in Figures 1 and 2. In these Figures the
metal dispersion was measured by hydrogen chemisorption and a strong
correlation was noted between metal dispersion and metal distribution. For
the case of HCl additions the metal penetrated more deeply into the support
structure and became better dispersed as the chloride concentration increased.
Without chloride the metal was very poorly dispersed and was found on the
pellet's exterior surface. Despite this poor dispersion the ready accessibility
gave quite an active catalyst.
For ethanolamine, the dispersion, distribution, and activity all increase but
at the higher pH the metal is completely displaced towards the pellet interior.
This leaves a marginal zone devoid of metal and a marked decrease occurs in
both activity and dispersion, as diffusional resistance becomes more impor-
tant and as the surface concentration of metal in the pellet interior increases.
These effects may readily be explained by the high pH of the external solu-
tion, which would tend t o reduce the anionic-exchange capacity of the
Cationic exchange of platinum as F't(NH3)4(OH), with alumina has also
been reported.22 Here, in contrast to the above results, metal adsorption
only occurred at a high pH, greater than 9. A particularly interesting aspect
R. W. Maatman, Ind. Eng. Chem., 1959, 51, 913.
45 E. Michalko, U.S.P., 3259454, 3259 589; J. C. Summers and L. L. Hegedus, J.
CataL, 1978,51, 185.
G. N. Maslyanskii and B. B. Zharkov,Kinet. Catal., 1971, 12, 699.
A. K. Aboul-Gheit, J. Chem. Tech. Biotechnol., 1979,29, 480.
18 ciztalysis
100 - 100
~8 0 - 80
ul h

g6 0 - 60 e?
7 ul
1io ;

20 g
8 a,

of this work was the comparison with NHB absorption on the same alumina,
which occurred readily at a lower pH, i.e., above 7. Apparently the cationic
affinity, generally taken as a function of the ionic charge density, is less for
the more bulky platinum ion than for the ammonium ion despite the higher
cationic charge of the former.
A very wide range of co-ingredients for chloroplatinic acid impregnation

100 100

.-5 80 . 80
L h
G) 0
ul v

;60 $
60 .-
40 40
r u
2 20
I 20 5
0 I
I I I I I I I 0
2 3 4 5 6 7 8 9
pH of impregnating solution
Figure 2 Effect of monoethanolamine concentration on the distribution and dispersion
of platinum in alumina and on the hydrogenation activity of the prepared
(Reproduced by permission from J. Chem Tech. Biotechnol., 1979, 29,480)
Design and Reparation of Supported Catalysts 19

0 1 0 1 0 1
Radial P o s i t i o n f r o m Center r / R

Figure 3 Types of Pt profiles obtained in co-impregnation experiments. Type (l),outer

shell, sharply defined; (2), outer shell, diffuse; (3), outer shell, diffuse to centre;
(4),inner shell, sharply defined; ( 5 ) , inner shell, diffuse; (6),inner shell, diffuse
to centre; (7), core, sharply defined; (8), core, diffuse; (9), linearly increasing
from centre
(Reproduced by permission from J. Catal., 1980,63,425)

has recently been reported.48 These workers investigated metal location

only and not metal dispersion and certainly some of the co-ingredients used
would be expected to poison any catalytic reactions. The results, which
are reproduced in Figure 3 and Table 3, are interesting insomuch as they
show the truly remarkable range in metal placement that can be achieved
by suitable choice of co-ingredient. The results of Table 3 can readily be
explained and interpreted in terms of the general principles, which we have
already enunciated for anionic exchange. Some of the details, however,
further help to quantify some of the relative comparisons. Hence in the
sequence NaBr, NaCl, NaF it is only the bromide anion that can successfully
compete and appears comparable in strength to the chloroplatinate anion,
whereas the more highly charged phosphate is much more strongly adsorbed.
For the organic anions the order of adsorption appears to be tartrate>
citrate > acetate > PtCl, >benzoate.
Other workersm have given results for the adsorption of a wide range of
platinum-group metal complexes and have related the rapid rate of adsorption
to the extent of surface impregnation. They attempt t o correlate the rate

Y.-S. Shyr and W. R. Ernst, J. Catal., 1980, 63,425.


Table 3 Properties of co-impregnated catalysts4'

Acid or salt Impremation time, 1h Impregnation time, 2 2 h
- Y r

Conc. (M) Pt (wt.%) Pt band r/R Conc. (M) Pt ban?

Begins at Ends at 5Pe
None - 0.5 1 1 0.77 1 - 0.5 1 2
Alc1, 0.01 0.50 1 0.56 1 0.01 0.5 1 2
HCl 0.0 1 0.51 1 0.67 1 0.01 0.50 2
NaCl 0.01 0.49 1 0.81 1 0.01 0.5 1 2
HF 0.01 0.50 - - 5 0.01 0.50 5
NaF 0.01 0.51 1 0.83 1 0.01 0.5 1 2
NaBr 0.01 0.50 1 0.77 1 0.01 0.50 9
HNO, 0.01 0.50 1 0.50 1 0.01 0.5 1 3
NaNO 0.008 0.49 1 0.79 1 0.008 0.5 1 2
Na,POz 0.01 0.31 0.83 0 7 0.01 0.10 8
Na benzoate 0.01 0.48 1 0.78 1 0.01 0.50 2
Acetic acid 0.01 0.5 1 - 3 0.01 0.50 uniform
Citric acid 0.01 0.49 0.84 0.44 4 0.01 0.48 6
Na Citrate* 0.02 0.19 0.33 0 7 0.01 0.14 8
Tartaric acid 0.01 0.49 6 0.01 0.5 1 7
* The 2 2 h sample was prepared from a fresh solution o f Na,PO, (or Na citrate) and H,PtCl,, while the 1 h sample was prepared by using a
portion o f that same solution after it had been stored for 2 or 3 days
(Adapted from Y.-S. Shyr and W. R. Ernst, J. Catal., 1980, 63, 4 2 5 )
Table 4 Uptake of metal complexes by aluminam (Percentageof metal absorbed after 60 min)
Ru Rh Pd Ir Pt
H,PtBr, (96.7)

[Pd(NH,),] Cl,(36.4) (NH4),1rCl6 (28.8)

[ Rh(NH,),CIl a,(27.0)

(Adapted from J. C. Summers, and S. A. Ausen, J. Cutul., 1978, 52,445)

22 Ca t a b sis
Table 5 Rhodhm depth profiles on y-A1203
pel1ets4l (pellet diameter 3 500 pm)*
Maximum dep th
* B. Harrison, J. Heffer, and F. King, Preprints 7th
Int. Congr. on Catalysis, B2, T o k y o .

of adsorption with the rate of ligand displacement from the complex anion
rather than with simple anionic exchange. Qualitatively the trends to be
expected would be similar. Their results are given in Table 4, which shows the
percentage extent of metal adsorption after 60 min.
Results for a more extended series of rhodium complexes have also been
given recently41 (see Table 5). These workers also discuss the poor metal
dispersion that can result from a too rapid adsorption and the build up of
high localized metal surface concentrations.
Uniform impregnation of ruthenium can be obtained47 with ruthenium
chloride in the presence of citric acid; the acid is competitive with the very
strongly adsorbed complex ruthenium ion although as we saw earlier& citric
acid will completely displace the chloroplatinate anion.
Very few results have been published on other metals, but again those
that have been are in good agreement with our overall general picture. Hence
in a study of nickel catalysts prepared from NiC12 and Ni(N03)2 metal
uptake increased rapidly above pH 5, being independent of the anion.49
This is indicative of simple cation exchange and metal placement could be
varied by suitable pH adjustment with either N b O H , giving a surface impreg-
nation owing to strong absorption, or H N 0 3 , giving a core impregnation.
A recent paperm on the co-impregnation of copper and chromium from
their nitrate solutions again follows the pattern of simple cation exchange
- with the higher valent Cr3+ ion being much more strongly adsorbed and
indeed capable of displacing the Cu2+.The competitive situation was such as
to exclude the close proximity of these two metals. These results could have
important ramifications in the preparation of uniform bimetallic catalysts.
Silica Catalysts. - Little problem is to be expected from depositing metals
by simple cation exchange onto the surface of silica gel with appropriate
control of pH to control the depth of metal penetration. Many high-surface-
49 M. Komiyama, R. P. Merrill, and H. F. Harnsberger, J. Catal., 1980, 6 3 , 3 5 .
50 H. -C. Chen, G. C. Gillies, and R. B. Anderson, J. Catal., 1980, 6 2 , 3 6 7 .
Design and Preparation of Supported Catalysts 23

100 -
Ion Exchanged

Ea 80 -

'E 60
0 40 -

f 20,

r, 5
Platinum, wt%

Figure 4a Hydrogen chemisorption by platinum silica gel catalysts

(Reproduced by permission from J. Catal., 1968, 10, 328)

area silica gels are so hydrophillic, however, that large support particles will
tend to fracture under the mechanical and thermal stresses generated by
sudden wetting. For this reason, and in order to improve control of metal
placement, the gel is often chosen to be in a partially dehydroxylated form
made either by thermal pretreatment or in one of the other ways used for
deactivating the surface, as discussed earlier.
Silica is a poor support on which to attempt the dispersion of metals in an
anionic form. This was clearly shown22 some time ago (1968) in work in
which a catalyst prepared by cation exchange using [Pt(NH&I2+ was com-
pared with one made by simple impregnation with H2PtC16. Similar results
have been reported more recently (1976)51 for which Keller's complex was
used for cationic exchange. In both cases, as shown in Figure 4, high disper-
sions resulted from the cation-exchange procedure and the more simple
impregnation procedure with platinum in its common anionic form was
ineffective, particularly at the higher metal loadings.
One way of overcoming this limitation has recently been described." In
this procedure a dried chloroplatinic-acid-impregnated silica gel is thermally
pyrolysed at 625 K. It was shown that the divalent platinum halide formed
under these conditions was sufficiently volatile and reactive as to react subse-
quently with and become dispersed on the support surface.

J. -P. Brunelle, A. Sugier, and J . F. Le Page, J. Catal., 1976, 43, 273.
24 Catalysis

2 ‘t 8
Platinum content (wt%)

Figure 4b Variation of the carbon monoxide chemisorption vs. metal content of P t S i O ,

catalysts; 0,cation-exchangemethod; V, impregnation method
(Reproduced by permission from J. Catal., 1976,43,273)

Zeolite Catalysts. - Crystalline silica-aluminates or zeolitic supports differ in

three important respects from the alumina and silica supports discussed
above. First, they are very strong cation exchangers and tend to stabilize low
valent cations, making them more difficult to reduce completely. Secondly,
they in general have a pore structure commensurate with atomic dimensions.
This microporous structure is such as to restrict the adsorption or exclude
altogether the more bulky ions. Together with the highly polar surface
this small pore size also makes them very difficult to dehydrate at low tem-
peratures. Finally, the catalyst structure itself tends to be thermally and
hydrolytically unstable, particularly at a low pH.
Simple Ciztionic Exchange. The cationic-exchange capacity of a zeolite is
generally taken to be directly related to the aluminium content. In an ideal
zeolite structure each aluminium atom will contribute one protonic-exchange
site. Simple cationic exchange is normally fairly straightforward with dilute
nitrate or chloride solutions (less than 0.1 M), which are used in large excess
and allowed to equilibrate. Typically several exchange cycles would be used
starting with fresh metal solution each time. For the less hydrolytically stable
zeolites with a high aluminium content, it may be necessary to increase this
solution pH by suitable buffering or by using salts of a weaker acid, e.g.,
acetates. Conversely, in order to prevent base catalysed hydrolysis of the
cation and possible precipitation, it may be necessary to use the zeolite in
its acid or preferably the ammonium exchanged form.
Design and Preparation of Supported Catalysts 25
Procedures typical of those used for simple cation exchange using a
mordenite substrate have been reported for the alkali and alkaline-earth
metalss2 and certain hydrated transition metalss3 (Mn2+, Cu2+, Ni2+, Co2+,
Zn2+). The latter article also notes that the maximum extent of cationic
exchange for these transition metals is only half that for the Group IA and
IIA metals and this is ascribed to steric hindrance in the case of the larger
hydrated transition metal cations. This limits their exchange to only those
sites that are exposed in the main zeolite channels. In agreement with this
concept they further noted that additional exchange was possible subsequent
to drying the samples, which they ascribed to the dehydrated cations now
being able to enter to some extent the more restrictive sodalite cages. It must
be noted however, that this re-distribution of metal during drying can more
usually lead to a loss in atomic dispersion with the separation of a bulk oxide
phase either within the zeolite or even on the exterior of the zeolite particles.
The destruction of the cationic complexes is probably owing to hydrolysis
under the relatively severe conditions required t o dehydrate these micro-
porous zeolitic supports. We shall return to this problem later.
Reducibility. The strong acidity of the zeolites results in a very strong ionic
interaction, which can markedly stabilize the univalent cation and make
complete reduction difficult if not impossible. Therefore, if we write the
reduction equation as an equilibrium:
ZeM+ 4 H2 f=f ZeH+ + Mo
the Ze + M ' interaction is so much stronger than the Ze H+ interaction+
as to displace this reaction to the left. Reduction is generally facilitated by
the presence of moisture, which may serve t o shift the equilibrium to the
right due to the relative hydration energies of the cations. More likely it
would seem as though this effect is one of hydrolysing the metal cationic
complex since reduction under these conditions or hydrothermal treatment
frequently leads to a loss in atomic dispersion and the formation of sintered
metal crystallites. Such effects have been suggested in the case of Ag,54955
Ni,55956 and ~ o p p e r . ~Such
~ ~ ~effects
' are also important from a catalytic
point of view, either if on the one hand, one desires a highly dispersed metal
or on the other hand, to stabilize a metal in a lower valent form. It may be
noted that reduction of copper (and probably nickel) on the much weaker
cationic exchnger, silica gel, or as free oxide, results in a clean reduction
directly from the divalent to the metallic state with no evidence of a mono-
valent intermediate.58
R . M. Barrer and J. Klinowski, J. Chem. SOC.,Faraday Trans. I , 1974, 70, 2362.
R. M. Barrer and R. P. Townshend, J. Chem. SOC., Faraday Trans. 1 , 1976, 7 2 , 661.
P. A. Jacobs and J . B. Uytterhoeven, J. Chem. SOC., Faraday Trans. I , 1977, 73,
P. A. Jacobs, J . -P, Linart, H. Nijs, and J . B. Uytterhoeven, J. Chem. Soc., Faraday
Trans. I , 1977, 73, 1745.
P. A. Jacobs, H. Nijs, and J . Verdonck, J. Chem. SOC.,Faraday Trans. 1 , 1979, 75,
S. J. Gentry, N. W. Hurst, and A. Jones, J. Chem. SOC.,Faraday Trans. I , 1979, 75,
S. D. Robertson, B. D. McNicol, J. H. deBaas, S. C. Kloet, and J. W. Jenkins, J.
Catal., 1975, 37, 424.
26 Catalysis
Table 6 Ion exchange of [Ru(H#)6] **
Catalyst Exchange % Exchange p H Average charge/Ru species
Ru A 43 6.5 3.5
Ru L 33 6.5 3 .O
Ru M 41 1.0 3.O
Ru X 52 6.5 1.2
Ru Y 27 6.5 2.1
* Adapted from B. Coughlan and W. A. McCann, J. Chem. SOC., Faraday Trans. I ,
1979, 75,1969

The redox behaviour of these zeolite supported metals has been used to
good advantage” as a very quick and simple test to characterize the metal
dispersion and hence the location of the metal. Three extreme cases have
been differentiated, atomically dispersed metal in the pore structure, larger
aggregates within defects of the crystalline lattice, and very large aggregates
on the external surface of the zeolite crystals. These are characterized in
terms of their increasing ease in reducibility and decreasing ease of oxidiz-
ability in the order given.
Ruthenium-Zeolite Catalysts. - This is taken as an example of some of the
problems and difficulties that can be encountered and have to be overcome
in the preparation of a zeolite supported catalyst. There is at present very
much interest in using highly dispersed ruthenium catalysts for Fischer-
Tropsch synthesis. It has recently been found that the maximum chain
growth in the synthesis reaction is strongly affected by the crystallite size
of the ruthenium metal. The smaller the crystallite size the lower in general
is the molecular weight of the synthesis product. Specifically it has been
found that maximum gasoline should result if the crystallite size can be
controlled in the range of 3-4 nm.” Conceivably this can be achieved with
the aid of zeolite supports.
Various authors have reported differing attempts to prepare ruthen-
ium zeolites by cationic exchange using, for example, [Ru(H20),]
N ~ ] and [ R u ( H N ~ )*
[ R u ( N H ~ ) ~ 2+,61 ~ ]62 cationic complexes. Results ob-
taineda using the hexa-aquo complex [ R U ( H ~ O ) ~are shown in Table 6 .
Exchange was carried out for 18h in solutions buffered at pH 6.5 except
for the more acid-resistant mordenite sample. Significant metal adsorption
was observed in all cases but this was accompanied by extensive hydrolysis,
which in effect reduced the cationic charge owing to the ligand replacement
of H 2 0 by OW. Spectroscopic examination indicated that in all cases the
spectra agreed with those to be expected from a Ru* ion in distorted octa-
hedral co-ordination. No attempt was made in this work to reduce the ruthen-
ium or to characterize the resulting metal dispersion.
59 P. A. Jacobs, in ‘Catalysis by Zeolites’, Proceedings of an International Symposium,
ed. B. Imelik, Elsevier, 1980.
6o B. Coughlan and W. A. McCann, J. Chern, SOC.,Faraday Trans. I , 1979,75, 1969.
61 C. A. Clausen and M. L. Good, Inorg. Chem., 1977,16, 816.
62 J . J. Verdonck, P. A. Jacobs, M. Genet, and G. Ponceiet, J. Chem. SOC., Faraday
Trans. 1 , 1980, 76,403.
Design and Preparation of Supported Catalysts 27
The RuY samples prepared from [Ru(NH,),] 3+ were thoroughly charac-
terized62 by XRD, TEM, i.r., and the combination TPO/TPR technique55
for characterizing crystallite sizes. Highly dispersed ruthenium within the
zeolite cavities was obtained even by heating in an inert atmosphere owing
to autoreduction by the ammonia released by decomposition of the complex.
This would appear to be a fairly general phenomenon when using ammine
complexes. Further heating in hydrogen resulted in crystallite growth, particu-
larly in the presence of water, to a size which was bigger than the crystallite-
cage size. These crystallites were apparently still contained within localized
defects within the zeolite lattice produced by localized hydrolytic decom-
position of the lattice structure. Some of the crystallite growth would appear
to be caused by hydrolysis of the zeolite surface complex and growth of small
oxide crystallites. Oxidation of the highly dispersed ruthenium above 400 K
led to the growth of very large crystallites on the external surface of the zeolite
crystals, which is due to the reactivity of the highly dispersed ruthenium and
the volatility of the ruthenium oxides.
Carbon-supported Catalysts. - Preparation of catalysts on carbon is very
dependent on the available surface area, and supports can be divided into two
broad categories:
(i) the low surface area supports with little or no pore structure such as
graphite carbon blacks, graphitized carbon blacks, and non-activated or low
activated carbons from carbonized synthetic fibres, e.g. , polyacrylonitrile,
etc. (surface area varies from 50-250 m28-l support material);
(ii) the high area supports such as activated charcoals and channel black
carbons, whose surface area varies from about 500-2500 m2 8-l support
material. In general, the high area arises from the development of a porous
structure within the carbon particles. As surface area rises there is a ten-
dency for the average pore diameter to fall and the supports with the highest
area are predominantly microporous (< 25 a).
This dependance on area arises from the density of functional groups on
the surface. For highly orientated surfaces6, such as graphite or the graphitized
carbon blacks, where surface functional groups occur only at the edges of
graphite planes, the density of functional groups is extremely low.
With less orientated carbons (turbostratic carbons) with a much higher
density of graphite edge planes and tetragonal carbon atoms, functional
group density is much higher, and in some carbons oxygen content can reach
as much as 12%. Catalysing carbons, if one is to believe the published litera-
ture, is still more of an art than a science, principally because few authors
have connected the surface chemistry of carbon with the dispersion of the
metal. Even where this is so, and for instance the ion-exchange properties
of a carbon have been used to obtain high dispersion of the metal ions,
little attention has been paid to the reduction stage and only moderate
metal dispersions have resulted.

63 P. L. Walker, jun., Am. Sci., 1962, 50, 259.

28 Catalysis
60 .l10

50 .1 0 0

40 . 90

'vl 30
a, m
Lo c\I


.-+Jz 20 7'1 ;
-am u-

0' :

lfl .6n


Figure 5 Effect of degree of bum-off on the surface area and degree of dispersion of a
1%platinum on graphitized carbon black

Catalysing Low-area Carbons. Activation of low-area graphitized carbons

increases the total surface area. However, the rate of formation of edge
sites increases far more rapidly than the total surface area and these edge
sites are also the sites at which surface functional oxygen complexes are
formed.64 On catalysing such carbons with chloroplatinic acid from an
ethanol/benzene solution to form a series of 1% platinum catalysts, the
connection between increasing surface area and dispersion can be seen.65
More importantly, by looking at the curves for increase in surface area
and metal dispersion with increasing burn-off shown in Figure 5, it can be
seen that while this surface area of the 48.9% burn-off sample increases by
a factor of two over that of the unactivated sample, the corresponding
increase in dispersion is almost 3.5 times that of the unactivated catalyst.
It is also interesting to note from Table 7, that the degree of dispersion
falls as the metal loading increases from 0.3 to 1.0%Pt with the unactivated
support, whereas the dispersion is constant at 55% for catalysts of the same
platinum loading range with the most highly activated sample (48.9% burn-
off). For the two catalysts with 1.0% loading there is a 12% increase in the
N. R. Laine, F. J . Vastola, and P. L. Walker, jun., J. Phys. Chem., 1963, 67, 2030.
P. Ehrburger, 0. P. Mahajan, and P. L. Walker, jun., J. Cutul., 1976, 43, 61.
Design and heparation of Supported Catalysts 29
Table 7 Degree of dispersion of platinum on a graphitized
carbon as a function of degree of bum-offand platinum
% Bum-off Metal loading Degree of dispersion
%Pt of Pt (%I
0.0 1.o 16
3.7 1 .o 17
8 .O 1.o 18
21.4 1 .o 26
48.9 1 .o 55
48.9 0.3 55
0.0 0 -5 26
0 .o 0.3 21
!Adapted from P. Ehrburger, 0. P. Mahajan, and P. L. Walker,
jun., J. Catal., 1976,43, 61)

number of surface platinum atoms on activation of the support. Changing

the solvent for the impregnation stage to water gave only a 24% dispersion
on the 48.9% burn-off carbon compared to 55% for the ethanol/benzene
solvent. This arises from the lack of anionic-exchange capacity on air-
activated carbons, which is necessary for the highest dispersion (PtCl,) ions
from aqueous solution.
Catalysing High-area Carbons. Although methods similar to those given
above for low-area carbons have been used with activated charcoals, more
frequently the carbons are allowed to exert their natural adsorbtive capacity
on a metallic species. The adsorbtive capacity can be modified by treating
the surface with oxidizing agents such as hydrogen peroxide, nitric acid, etc.
Pope et aZ.66 adsorbed chloropalladous acid from aqueous solution on to a
commercial charcoal and on to the same charcoal after treatment with nitric
acid. The resulting catalysts were reduced with alkaline formaldehyde and
examined by transmission electron microscopy. Both had a uniform distri-
bution of metal throughout the carbon structure, although the palladium
crystallite size of the nitric acid treated carbon catalyst was smaller than that
of the non-treated carbon catalyst. This was confirmed by CO chemisorption
when metal areas of 1.62 m2 g-' catalyst for the non-treated catalyst and
2.37 m2 8-l catalyst for the treated catalyst were obtained.
Increasing metal loading almost always leads to a decrease in dispersion.
Hence with a series of platinum on charcoal catalysts6' ranging from aloading
of 0.5% to 10%Pt the dispersion decreased from 47%for the 0.5%Pt catalyst
to 23% for the l O % P t catalyst. In other words, the total platinum surface
area increased by ten times for a platinum loading increase of twenty times.
Other Methods of Preparing Carbon-supported Catalysts. Some interest has
been shown in catalysing an organic material and then carbonizing the

D. Pope, D. S. Walker, and R . L. Moss, J. Cutal., 1973,28, 46.

C. E. Hunt, J. Catul., 1971,23, 93.
30 Catalysis
material to produce a metallized carbon. Two examples are discussed here,
first platinum on carbon catalysts from ion-exchange resins,@ and secondly
premetallizing peat.69
The preparation of platinum on carbon by pyrolysis of ion-exchange resin
consists basically of four stages; neutralization of the resin, exchange with
platinum ions, pyrolysis in an inert atmosphere, and washing and drying the
pyrolysis product. Weak acid cationic-exchange resins, a strong acid cationic-
exchange resin, and an anionic resin were examined as possible precursors.
Chloroplatinic acid was used t o impregnate the anionic resin, while diammino
platinum(I1) dinitrite was used for the cationic resins. As is to be expected,
the preparation of high platinum dispersions was dependant on the type of
resin, the ionic form of the resin, and the resin pyrolysis temperature. The
most highly dispersed catalysts were made on the calcium form of Amberlite
IRCSO and Amberlite CG50 by pyrolysis between 500 and 800°C. Metal
loadings of up to 12% can be achieved with crystallite sizes of less than 15 8.
Similarly, with platinum on carbon prepared by impregnation of peat with
chloroplatinic acid followed by subsequent pyrolysis, it was found necessary
to adjust the ion-exchange capacity of the peat by treatment with aqueous
ammonia. Only low loadings of platinum could be obtained with a weakly
humified sample of peat (0.25-0.34% Pt), but much higher loadings can be
obtained with well humified material, e.g., 1.3%Pt. However, the average
crystallite size increased from 9.4 A at the lower loading to 83.4 8 with 1.3%
platinum. This is to be expected, but does illustrate the possibilities of this
particular route.

68 K. F. Blurton, Carbon, 1972, 10, 3 0 5 .

69 J. C. Cavalier, E. Chornet, B. Beauregard, and G. Coquard, Carbon, 1978, 16, 21.