Sigma Xi, The Scientific Research Honor Society

THE EFFECT OF GRAIN SIZE ON CATALYST PERFORMANCE

Author(s): ERNEST W. THIELE
Source: American Scientist, Vol. 55, No. 2 (JUNE 1967), pp. 176-184
Published by: Sigma Xi, The Scientific Research Honor Society
Stable URL: https://www.jstor.org/stable/27836820
Accessed: 12-03-2020 22:34 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/27836820?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Sigma Xi, The Scientific Research Honor Society is collaborating with JSTOR to digitize,
preserve and extend access to American Scientist

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 american scientist, 55, 2, 1967

THE EFFECT OF GRAIN SIZE ON CATALYST

PERFORMANCE
By ERNEST W. THIELE

Prehistory

The recognition ofandthe

aday and Berzelius; action
we may of the
consider catalysts
year 1835, goes back to the days of Far
not as the
birth date, but so to speak as the date of baptism, since in that year
Berzelius [i] gave catalysis the name we still use. During the nineteenth
century, the number of catalytic reactions grew steadily, and by the end
of that century at least one process involving solid catalysts?the contact
process for sulfuric acid production?was in commercial operation.
Of necessity, a solid catalyst must act at its surface, and therefore it is
desirable, at least from an industrial point of view, that catalysts have as
much surface in a given volume as possible. Practically, this means that
catalysts should, if possible, have a large internal surface; they should be
porous. The advantages of porous catalysts were soon recognized.
However, it requires only a meager understanding of fluid flow to real
ize that the pressure drop in an ordinary bed of catalyst is not adequate
to force through the small pores in a grain of catalyst any perceptible
amount of fluid. If the reactants are to come in contact with the interior
surface, they must do so by diffusion. But diffusion itself is a relatively
slow process; and if the reaction is relatively fast, a perceptible, or even a
large part of the conversion may take place while the molecule of reactant
is entering and leaving the pores of the catalyst. In this case the interior
of the grain will be filled with more or less completely converted reac
tants. As conversion proceeds, the reaction rate goes down ; and the inte
rior of the catalyst will therefore be operating at a lower rate, and will pro
duce lesfe product than would be calculated on the basis of the concentra
tion in the bulk of the fluid bathing the grain. We may expect, therefore,
that very small grains of catalyst will be active throughout their volume,
while most of the interior of very large grains will be filled with com
pletely converted material, and contribute nothing to the activity of the
catalyst bed. In small grains, the catalytic action will be proportional to
the mass of catalyst; in large grains, to the external surface.
These qualitative considerations are not enough to specify what
"small" and "large" mean in this connection; but it is obvious that, for
each set of conditions, we must expect to find some grain size below which
the whole catalyst mass is effective, and another size above which the ex
ternal surface controls the activity, with an intermediate transition re
gion, the exact shape of which can be determined only by more quantita
tive considerations.
176

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 THE EFFECT OF GRAIN SIZE ON CATALYST PERFORMANCE 177
Pertinent Early Observations
It would not be expected that such ideas would completely escape for a
long period the attention of the many workers, scientific and industrial,
who studied catalysis. In fact, it is surprising that there are so few evi
dences of this in the literature.
In 1899, the anonymous author of a German patent specification [20 ]
proposed making catalyst grains with a thin layer of porous ceramic on
the surface, impregnated with platinum, and explained, "The platinum
taken up on the thin porous surface is fully active, since the gas circu
lating between the grains comes in contact with the whole surface and
diffuses into the small depth of the layer."
In 1922, Larson and Tour [8] studied ammonia synthesis on two sizes of
catalyst, and found activity proportional to mass. They did not give any
reason for making the experiment, or say what they expected to find.
In 1938, Juliard [7] studied the effect of grain size in hydrog?nation,
and he too found activity proportional to mass. He remarked that this
shows that the internal surface was active.

Fluid Film Diffusion

Of course, even for non-porous catalysts it is necessary for the reac
tants to pass from the bulk of the fluid to the surface of the catalyst grain,
but since the dimensions of the voids in a bed of grains are of the same
order of magnitude as the grains themselves, and the diffusion in the
grains is impeded by the solid, it follows that, for porous catalysts, the
intra-particle diffusion must predominate; if the flow of the fluid is turbu
lent, it predominates very greatly. We concern ourselves here only with
intra-particle diffusion.

First Quantitative Theory

The role of intra-particle diffusion, then, was appreciated qualitatively
long ago by a few people at least. It is a remarkable fact, however, that
for a century after Berzelius, paper nobody attempted (so far as the
records go) to give this idea a quantitative formulation. Then three
workers independently presented a quantitative theory?Damk?hler
[5], in Germany, in 1937; Thiele [io], in the United States, in 1939; and
ZePdovich [i8] in Russia in the same year. The approaches of the three
authors differed markedly, but they arrived at essentially the same re
sult : there is a modulus, an expression depending on the diffusivity of the
reactants, the rate of reaction, and the size of the grains, which deter
mines the effectiveness of the catalyst mass; if this modulus is low, all the
mass is effective; if it is high, only an outer layer is effective.
How did it happen that, after this long time, the subject became of in
terest to several people at once in different countries? The details of scien
tific advances are always somewhat a matter of chance, but at any rate

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 178 AMERICAN SCIENTIST

the following can be said. During all the period, say from 1899, chemistry
and chemical engineering were, of course, growing rapidly; but interest
in catalysis was growing even more rapidly. Some idea of the relative
rates can be obtained from a study of the decennial indexes of Chemical
A bstracts. For every thousand lines of index the number of lines devoted
to "catalysis" and "catalyst" were as follows:
1907-1916 1.18 lines
1917-4926 2.55 lines
1927-1936 3.14 lines
1937-1946 3.72 lines
Thus, the chance that someone would take up the matter was greatly
increasing, and it is perhaps not a very strange coincidence that three
workers should do so independently.
It might be thought that the development of improved methods for
determining pore characteristics would furnish an incentive for develop
ing a quantitative theory. This was probably not the case. The definitive
paper by Brunauer, Emmett and Teller }?] appeared in 1938?too late
to be influential. The fact seems to be that the BET method was itself a
product of the general increase in interest in catalysts, and was a par
allel rather than an antecedent development.
Why was it that precisely these three workers took up the subject? In
the case of Damk?hler, this was almost inevitable. He had worked inten
sively on reactor design, and also on the kinetics of absorption on porous
materials. His work appeared in the large German treatise on chemical
engineering edited by Eucken and Jakob, and it was natural that the
editors should choose him to prepare the relevant sections, and natural
that in doing so he should apply to reactions the ideas he had developed
for absorption. Thiele had worked previously on the kinetics of the
reaction of steam on porous carbons, and had been led to consider the
type of action involved, though he had not reduced it to a theory. When
his industrial assignments required him to deal with catalysts, he looked
for the corresponding theory in the literature, and not finding it, under
took to supply the lack. ZePdovich also had worked on porous cata
lysts for the oxidation of carbon monoxide. Thus, in all three cases there
was a previous history of interest in related matters.
The simultaneous publication of the same ideas by different workers
has given rise in the past to several celebrated scientific controversies.
Nothing of the sort happened in this case. The scientific world received
the new light with entire calm. To be sure, there were other things to
think about, as the second world war was just getting under way.
Whatever the reason, literature on the subject developed very slowly.
Of the pioneers, only Damk?hler pursued the subject further. Thiele
published very little of any kind; ZeFdovich went on to become an

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 THE EFFECT OF GRAIN SIZE ON CATALYST PERFORMANCE 179
astrophysicist. The new theory was duly incorporated in Hougen and
Watson's "Chemical Process Principles," and in the "Handbuch der
Katalyse/' but during the period up to 1950, papers in this area appeared
at the rate of only one or two a year.
Certainly, the original authors had left plenty to be done. None of
them offered any experimental data; their results were applicable only
to the isothermal case; they considered only single reactions, not con
secutive or parallel reactions; and they took no notice of the complex
kinetics typical of heterogeneous catalysis. The necessary diffusion rate
information was not available for catalysts. Starting in 1950, these and
other matters began to be considered at an increasing pace. It will be con
venient to discuss the various areas separately.

Diffusion Constant
It was obvious that progress in the area would require a study of the
diffusion of fluids into porous materials. At the time, not only was in
formation on this point completely lacking for catalysts, but there was
practically nothing on other materials, a single paper on diffusion in
soils, dated 1904 [3] being the only one to be found. There were two pos
sible approaches. On the one hand, experiments on diffusion in actual
materials (without reaction) could be made. Damk?hler and ZePdovich
assume data of this kind. On the other hand, models of the pore structure
of the catalyst could be assumed, and the diffusion properties calculated
on the basis of the known behavior of fluids in bulk. Thiele assumed a
straight cylindrical pore.
Both approaches have turned out to be quite attractive to chemical
engineers. The experiments could be made at reasonable temperatures,
and the technique was not too difficult; the mathematicians found an
outlet for the newly-developed computers. Wicke and Kallenbach in 1941
[17] were the first to make experimental measurements of this kind,
although they were interested more in the kinetics of absorption than in
catalysis. Since then, more than twenty experimental studies have de
termined the transport rate of fluids (mostly gases) through a variety
of porous substances, not all of them catalysts.
The original straight cylindrical pore of Thiele was obviously a dubious
substitute for the branching pores of non-circular and varying cross
sections which are certainly found in real catalysts. It was of interest to
compare the experimental diffusion data with various models, and this
line of study was aided by the development of methods for determining
pore size distribution. It soon became apparent that the discrepancy
between experiment and the results based on the simple model of a
straight pore could not be attributed merely to the crookedness of the
pores. Pilled catalysts are ordinarily composed of small grains with mi

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 180 AMERICAN SCIENTIST

nute pores; when pressed together, the interstices between the grains
form a system of larger pores. Models based oh this fact have been shown
to agree much better with the results of diffusion experiments. It was
also noted experimentally that diffusion of vapors was sometimes much
greater than would follow from the known pore volume and properties of
the vapor state. This is attributed to more rapid diffusion in a liquid
layer absorbed on the internal surface. This latter phenomenon, however,
appears to be rather rare under practical catalytic conditions.
Altogether, the needs of students of catalysis, together with those of
students of ion exchange and of physical adsorption, have led in thje last
quarter-century to the development of a minor branch of chemical
engineering. If activity continues, we may look forward to the issuing by
Julius Springer of a "Handbuch der Porenkunde" before long.

Activation Energy
If diffusion and reaction conditions are such as to render much of the
grain interior inactive, the effect of temperature on the observed reaction
rate will be different from the true temperature coefficient. For as the
temperature rises the rate increases, and the active fraction of the cata
lyst must decrease. Qualitatively, this was noted by Thiele, and ZeFdo
vich gave the quantitative result: the apparent activation energy
is half the true value. A few workers have used this result to define the
diffusion regimes?by working over a wide range of temperatures, they
noted the point at which the activation energy fell off, which is the point
at which part of the catalyst begins to be ineffective.

Temperature Gradients in the Grains

All chemical reactions are accompanied by heat effects; hence, as
material must be transported to or from the grain interior, so also must
heat, and there must therefore be a temperature gradient in the grains.
This was noted by Damk?hler [6] as early as 1943. He shows that there
is a simple relation between the concentration gradient and the tem
perature gradient. If the reaction is exothermic, we may expect a higher
internal temperature and greater activity than a simple isothermal theory
would indicate. For endothermic reactions the reverse is true.
A detailed study of the consequences of these facts did not become
popular as early as the study of diffusion. For one thing, thermal con
ductivities are not too easy to measure. For another, the effects cannot
be separated experimentally from the reaction itself. However, within the
last few years at least, the theory involved has attracted several authors.
They have shown that, under appropriate conditions, the extent of
reaction in large grains, far from being smaller than in an equivalent
mass of small grains, may (for exothermic reactions) be many times

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 THE EFFECT OF GRAIN SIZE ON CATALYST PERFORMANCE 181

greater. Cunningham, Carberry, and Smith [4] actually observed a rise

in temperature in the interior of a catalyst grain which was consistent
with theory. But their comparison of activity with theory was beset
with difficulties, and we do not yet have a clear-cut case where the
theoretical expectations were realized. It has been suggested, but not
demonstrated, that the effect is appreciable in some commercial opera
tions.

Complex Kinetics
Early theories?and most subsequent theories as well?have been con
fined to the case of simple first- or second-order kinetics?usually first
order. To these, Wheeler [is] added zero order in 1951. It is well known,
however, that heterogeneous catalytic reactions are particularly liable
to exhibit a variety of more complex relationships, which have been
the subject of many experimental and theoretical studies and are
well understood in only a few cases at best. Since the relevant formulae
greatly complicate the mathematics, which must also be restudied
for each individual case, it is reasonable that most workers have been
content to assume that the kinetics can be approximated by a simpler
form.. Recently, however, a few studies have been made on particular
Langmuir-Hinshelwood formulae, and in 1966 Roberts and Satterfield
[9] reported that, with such kinetics, unexpected results may be
found?increased rather than decreased reaction in large grains, even
under isothermal conditions.

Successive Reactions
So long as only a single reaction occurs in a system, the only effect
of diffusion would be to reduce the rate of reaction. If, however, more
than one reaction goes on, the effects of large grain size may be qualita
tive as well as quantitative. This was noted by Thiele, who did not give
any specific examples. In 1951, Wheeler pointed out the qualitative results
to be expected in several cases of successive and parallel reactions.
Perhaps the most important case is that in which the product proceeds to
form some other less desirable compound. At every point in the bed, the
effect of slow diffusion is to increase the quantity of product in contact
with the catalyst in any given system?this is in fact the basis of the
decreased activity. If the product decomposes, therefore, an increased
amount will decompose, while a smaller amount will be formed, and the
ultimate yield must suffer.
Experimental proofs of effects on selectivity are rare; probably the
first was that of Weisz and Swegler [i4] in 1955 who studied the dehydro
genation of cyclohexane and noted an increased amount of cyclohexene
when small-particle catalyst was used.

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 182 AMERICAN SCIENTIST

Experimental Verification
Most of the work already referred to has either been theoretical, or
in the case of diffusion studies, has been directed to the determination of
necessary constants. The actual amount of experimental work under
taken to substantiate the theories, though considerable, is small com
pared to the mass of paper and computer work. The reasons for this are
not far to seek. Satisfactory verification requires experiments with at
least two catalyst grain sizes of identical composition, along with in
formation regarding the pore structure of the catalyst (or its diffusion
properties) and the diffusion characteristics of the reaction mixture.
The accuracy of kinetic data is normally so low that verification of fine
points relating to different theories is not to be expected.
Nevertheless, a considerable number of experimental studies has
been made?roughly twenty! It is not so easy in this case to assign
priorities. On the one hand, as has already been noted, a few tests with
different grain sizes had been made before the theories were developed;
on the other, after they were developed, it was possible to take existing
experiments, estimate or determine catalyst diffusion characteristics, and
compute whether diffusion was controlling, and this was done quite
early?by Wagner in 1943 [12], for example. But this is not verification.
Perhaps Wicke and Br?tz, in 1949 [i6], were the first to make experiments
on two sizes of catalyst, with knowledge of the theory and an intention
to test it. As has been said, they were followed by a considerable number
of others.
How have the experiments compared with theory? In general, rather
well, considering the low accuracy of kinetic data, the admitted im
perfections of the theory, and the uncertainty of catalytic kinetics in
general. Agreement with expected results within 15 or 20% appears to be
typical. Such results will, of course, not permit the resolution of detailed
questions. There are a couple of cases, however, where the experimental
results differed widely from that expected on a simple theoretical basis.
It appears likely that there may be a variety of unexpected results which
depend on complex kinetics?such as were predicted by Roberts and
Satterfield.

Influence on Fluidization Processes

At very nearly the same time as the early theories of diffusion in
catalysts with fixed beds were developed, the fluid catalytic cracking
process, using catalyst powders agitated by the reacting materials was
taking shape in the petroleum industry. And this was the first commer
cial application of the procedure. It might be thought that improved
mass transfer was the advantage sought in the development of the proc
ess. This, however, is not the case. The flow processes in fixed beds and

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 THE EFFECT OF GRAIN SIZE ON CATALYST PERFORMANCE 183

fluid beds are so different that a comparison between them does not
have* much meaning. However, if we consider that the advantage of a
fine-grained fixed bed lies in the small proportion of product in contact
with the catalyst, it will be found that the proportion of product in the
fluid bed is much greater, the dense phase in the fluidized bed being
roughly comparable to a mass of very large grains. The advantage of
the fluidized bed lies in improved heat transfer and in ease of regener
ation, not in efficient catalyst use under given temperature and pressure
conditions.

Application
As things stand at present, it appears that approximate theories of the
phenomena have been worked out for many cases. Our ability to predict
behavior as affected by grain size is limited on the one hand by our
ignorance of the precise kinetics of catalytic reactions and on the other
by our imperfect understanding of the details of the structure and
activity of specific samples of catalyst?plus, no doubt, the limitations in
computer time.
The amount of experimental work in this area is much less than the
amount of theoretical work, which no doubt reflects the relative cost of
both approaches. Yet the attainable accuracy of experimental work is not
such as to make it likely that the data can be a guide to improved
theory. So it is unlikely that an extensive experimental program would
produce much improvement in our knowledge, except for the specific
catalysts studied.
We have, therefore, a very full demonstration that catalyst grain
size may affect both activity and selectivity of catalytic reactions, and
means for roughly quantitative calculation of the effect on activity and
selectivity. Indeed, as was first shown by Wheeler, the observed overall
activity of a single size of catalyst, combined with the diffusiyity of the
fluid, is sufficient to indicate approximately the area where the transition
from mass effectiveness to surface effectiveness occurs. Weisz [133 has
developed a convenient modulus for this purpose.
From a practical point of view, there are two areas in which this
knowledge is of importance. All laboratory studies of heterogeneous ca
talysis, or of specific catalysts or catalytic reactions, should include ex
periments or computations to show that diffusion does not markedly in
fluence the results. It is impossible to reason on adsorption, sites, or
mechanisms if the effects observed are due to something else. It is prob
able that today all chemical engineers are familiar with this fact. It is
not certain that all chemists who are interested in catalysts are aware of
the pitfalls. Many work in conjunction with chemical engineers, but
certainly not all. In general, chemists tend to be late in adopting chemical
engineering developments?the use of efficient distillation is an example.

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms
 184 AMERICAN SCIENTIST

The other area is the development and choice of commercial catalysts.

Here our knowledge of actual commercial procedures is practically non
existent, but reasonable guesses can be made. In developing new cata
lysts, or in choosing a catalyst for commercial use, the chemical engineer
may be faced with an economic balance. Large pellets may be less active,
and less selective; small pellets will be more costly and increase pressure
drop. However, the developer of a catalyst has other things to think
about. His catalyst must ri? thermally and mechanically stable; easy to
manufacture, and so forth. It is therefore unlikely that very precise in
formation, even if available, could be used effectively to "design" a
new catalyst. The developer will be guided by the principles of diffusion
mass transfer, but probably will find the existing state of knowledge
adequate for his purposes.

Geography
A review of the pertinent literature shows that a large proportion of the
work in this area was done in the United States. It is true that there may
be some bias, in that foreign work has not been so fully covered; and there
are also questions as to definition of the area. A rough figure, however,
might be that over 60% has been done in this country, which is surely
a higher percentage than for technical work as a whole.
Presumably this figure reflects mainly the predominance of the
United States in chemical engineering in general. Chance and fashion no
doubt have played a part?to these we may perhaps add the greater
availability of computers.
REFERENCES
1. Berzelius, J. J., Jahresberichte f?r Chemie, 13, 237 (1836).
2. Brunauer, S., P. H. Emmett, and E. Teller. J. Am. Chem. Soc, 60, 309
(1938).
3. Buckingham, E., U.S. Dept. Agr. Bur. Soils Bull, 25 (1904).
4. Cunningham, R. A., J. J. Carberry, and J. M. Smith, Am. Inst. Chem. Eng.
J., 11, 636 (1965).
5. Damk?hler, G., Der Chemie-Ingenieur, 3, 430 (1937).
6. Damk?hler, G., . Physik. Chem., A 193, 16 (1943).
7. Juliard, A., Bull. Soc. Chim. Belges, 46, 587 (1938).
8. Larson, A. T., and R. S. Tour, Chem. Met. Eng., 26, 647 (1922).
9. Roberts, G. W., and C. N. Satterfield, Ind. Eng. Chem. Fundamentals, 5,
317 (1966).
10. Thiele, E. W., Ind. Eng. Chem., 31, 916 (1939).
11. Thiele, E. W., and R. T. Haslam, Ind. Eng. Chem., 19, 882 (1927).
12. Wagner, C, . Physik. Chem., A 193, 1 (1943).
13. Weisz, . ., Z. Physik. Chem. (Frankfurt), 11, 1 (1957).
14. Weisz, . B., and E. W. Swegler, Phys. Chem., 59, 823 (1955).
15. Wheeler, A., Advances in Catalysis, 3, 250 (1951).
16. Wicke, E., and W. Br?tz, Chem. Ing. Tech., 21, 219 (1949).
17. Wicke, E., and R. Kallenbach, Kolhid-Z., 97, 135 (1941).
18. Zel'dovich, Ya. B., Acta Physicochim. URSS, 10, 583 (1939).
19. Zel'dovich, Ya. B., Acta Physicochim. URSS, 1, 449 (1934); 3, 529 (1935),
8, 527 (1938).
20. D.R.P. 119280 (1899).

This content downloaded from 194.167.201.133 on Thu, 12 Mar 2020 22:34:00 UTC
All use subject to https://about.jstor.org/terms