105

A Theory o f the Catalytic Surface.

By H u g h S tott T a y l o r ,D.Sc., Associate Professor of Physical Chemistry,
Princeton University.

(Communicated by Dr. E. F. Armstrong, F.R.S. Received 12th February, 1925.)

The problem of the mechanism of a catalytic action occurring at the surface

of a solid catalyst can be studied in several ways. Kinetic studies of such re­
actions supply important information with regard to the manner by which the
process is achieved. On the other hand, attention may be focussed, not on the
reactions which occur at the catalyst surface, but on the properties of the surface
itself, and, from the results of such studies, generalizations may be made with
respect to a variety of reactions which occur at such a surface. It is this latter
method of attack on the problem of contact catalysis which has been adopted in
the last few years in a comprehensive series of investigations on the properties of
solid catalytic surfaces conducted in Princeton by the author and his co-workers.*
Taken in conjunction with the results of other investigators in the field of
reaction kinetics, these studies have led to a concept of the catalytic surface
which is, perhaps, more comprehensive than earlier efforts and which leads to
interesting general conclusions with reference to matter in the solid state.
The theory of Bodenstein and Fink,! based on kinetic studies, assumed that
reaction velocity was determined by the rate of diffusion of reactant gases
through a layer of gas adsorbed on the catalyst surface. More recent studies,
especially those of Langmuir,! have led to the conclusion that multi-molecular
layers of gas do not occur on plane surfaces of an adsorbent. Langmuir’s
view is that the surface of a catalyst is to be regarded as an area containing a
definite number of elementary spaces in which some of the spaces are vacant,
while others are covered with adsorbed atoms or molecules. The activity of
the surface is conditioned by the nature of the arrangement, and the spacing
of the atoms, in the surface layer. The velocity of reaction on such surfaces
depends on the fraction of such surface which is covered by adsorbed substances.
Langmuir developed two cases in detail, one in which the surface was relatively
bare, the other in which the surface was nearly completely covered with adsorbed
gases.
* For a resume of this work see 3rd Report of the Committee on Contact Catalysis of
the National Research Council (‘ Journal of Physical Chemistry,’ vol. 28, p. 897, 1924).
t ‘ Zeits. Physik. Chem.,’ vol. 60, p. 1 (1907).
$ See especially, ‘ Trans. Farad. Soc.,’ vol. 17, p. 621 (1922).
106 H. S. Taylor.
In this last case Langmuir gave a convincing explanation of the kinetics of
the combination of sulphur dioxide and oxygen at platinum surfaces, and
considered similarly the combination of carbon monoxide and oxygen at
quartz—glass surfaces. The data of Bodenstein and his co-workers on these
reactions were used to show that the surface of the platinum was very nearly
covered with adsorbed sulphur trioxide, and the surface of the quartz—glass
nearly completely covered with carbon monoxide.
The experimental evidence obtained in the Princeton studies of the properties
of catalytic surfaces has led to the conclusion that some modification of the
Langmuir concept is necessary. Benton showed, for example, that the ad­
sorption of carbon monoxide on quartz—glass surfaces, even when the material
was reduced to an impalpable powder, was extremely small at room tem­
peratures, much smaller than could be anticipated on the assumption that
the surface was nearly completely covered by adsorbed carbon monoxide.
This suggested either that Langmuir’s assumption was wrong, or alternatively
that only a very small fraction of the surface was catalytically active, and th at
this small fraction was nearly completely covered with carbon monoxide.
This latter alternative, that only a small fraction of the surface is active, has
been confirmed by a number of other observations. It has also been shown that
a catalyst surface shows a varying capacity both to adsorb gas and to promote
catalytic change. It has been shown that active catalysts manifest an extra­
ordinary sensitivity to heat treatment and to the action of poisons. Both
heat treatment and poisoning alter adsorptive capacity and reactivity to
different degrees, the reduction in catalytic activity being much more pro­
nounced than the reduction in adsorptive power. The concept of the catalytic
surface to which this varying capacity of the surface to adsorb and to promote
catalytic change has led, may be set forth in reference to a metal catalyst,
such as nickel, capable of effecting catalytic hydrogenation.
The X-ray examination of metallic hydrogenation catalysts has shown that
they possess, even when prepared by low-temperature reduction processes,
the definite lattice structure of the crystalline material. A granule of catalyst
must therefore possess, in part, the ordered arrangement of the atoms found
in crystals of the metal. The sensitivity of the product to even' moderate
heat treatment, whereby marked sintering occurs, suggests that this completely
ordered arrangement of the atoms has not been attained, and that, here and
there, on the surface of a mainly crystalline granule, there are groups of atoms
in which the process of crystallization is not yet complete. The following
representation of a cross-section of a minute portion of such a granule of nickel
 Theory o f the Catalytic Surface. 107

may be employed as an illustration of such an incomplete stage in the crystalli­

zation process.*
Ni

Ni Gas'phase. Ni-
I
Ni-Ni
I
Ni-

Ni-Ni-Ni Ni-Ni-

I -Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-

II
M i i i I i I I I I I I
-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-Ni-
Granule P roper.
F ig . 1.—Diagrammatic representation of a portion of a Granule Surface
in Cross-section.

The atoms in the second layer II of the granule proper will be surrounded, on
six sides by other nickel atoms. The atoms in the surface layer will be sur­
rounded by neighbouring metal atoms in all directions except that towards
the gas phase. The degree of constraint or saturation imposed by this orderly
arrangement of atoms becomes progressively less and less as we proceed outward
from the granule proper (I and II) towards the most exposed metal atoms.
Thus, the uppermost nickel atoms in the above representation suffer only one
constraint by reason of their single attachment to the nickel atom immediately
below. Further, we may note that the atoms in the edge of a granule have
one less degree of saturation than those in the surface proper ; atoms at a
corner have two less than those in the surface and one less than those at an
edge. A surface of a granule may thus be regarded as composed of atoms in
varied degrees of saturation by neighbouring metal atoms, varying from those
one degree less saturated than interior atoms to those which are only held
to the solid surface by a single constraint. It is by this constraint alone that
these outermost atoms differ from gaseous metal atoms.
A gaseous atom of nickel can combine with four molecules of carbon monoxide
to form nickel carbonyl. It is not, therefore, improbable that exposed nickel
atoms, held by only a single valence to the solid surface, shall have a capacity
to attach to themselves or adsorb three molecules of carbon monoxide, or,
* The core of an active nickel catalyst prepared by reduction of an oxide granule will
be largely unreduced. The fact that the core is oxide in no way alters the ensuing argu­
ment, which is concerned only with the surface layers.
108 H. S. Taylor.
alternatively, three molecules the bonding between which and the nickel
atom is identical with that obtaining in nickel carbonyl. With the metal
attached to the granules by two constraints, two such molecules might be
adsorbed. This concept introduces a mechanism whereby both the con­
stituents of a hydrogenation process may be attached to one and the same
nickel atom. It obviates the necessity, inherent in the Langmuir concept, of
having the reactants adsorbed on adjacent elementary spaces. The activation
of such catalysts as platinum and silver gauzes when used in catalytic oxida­
tions would be attributable, not to a variation in the distances between or
spacing of the atoms as postulated by Langmuir, but to a production, by
disintegration of the plane-surfaced wire, of metal atoms to a large degree
unsaturated and detached from the normal crystal lattice of the metal and
capable of adsorbing several molecular reactants.
The concept of metal atoms detached to varying extents from the normal
crystal lattice is in accord with the observations recorded from the studies of
the physical properties of such catalyst surfaces. The atoms of metal loosely
held will have a greater freedom of motion, will be relatively sensitive to thermal
influence. They will show greater volatility and the thermal sensitivity which
we have already noted. The varied saturation of the metal atoms by
neighbouring atoms will explain the varying saturation capacity of the surface
for different adsorbed gases. This has been frequently noted in the Princeton
studies. It is well illustrated in the data of Pease on copper,* where practical
saturation of the surface is reached at relatively low pressures with both
carbon monoxide and hydrogen, but the amount of carbon monoxide adsorbed
at saturation is much greater than that of hydrogen adsorbed at saturation.
This means that a greater proportion of the surface atoms can adsorb carbon
monoxide under the experimental conditions. This concept of the surface is
also in agreement with the data of Gauger and Taylor j on adsorption of hydrogen
by nickel at varying temperatures. They found, for example, with a given
sample of nickel that, at 25° C., saturation was reached at 75 mm. pressure,
and that, at this temperature, a further tenfold increase in pressure did not
alter the adsorption (8*7 c.c.) by more than one per cent. At 305° C., saturation
was reached at 250 mm., and a further threefold increase in pressure did not
alter the adsorption by one per cent. But, in this case, the adsorption was
only 5*4 c.c. The spread in the values for the quantity adsorbed indicates
that, at the higher temperature, only the more unsaturated nickel atoms in
* ‘ J. Am. Chem. Soc.,’ vol. 45, pp. 1196, 2235 (1923).
t ‘ J. Am. Chem. Soc.,’ vol. 45, p. 920 (1923).
 Theory of the Catalytic Surface. 109

the surface are capable of holding hydrogen. At the lower temperatures

the more saturated nickel atoms are capable of retaining adsorbed
hydrogen.
It is the less saturated catalyst atoms in the surface which will be the
preferred positions of attachment of adsorbed catalyst poisons. As the quantity
of poison increases, more and more of the surface atoms will be covered with
poison. In the final issue the whole surface will be covered. Since, as was
observed by Pease,* minute amounts of carbon monoxide poison suppress the
hydrogenation of ethylene on copper—quantities of carbon monoxide even
smaller than the capacity of the surface for adsorbed hydrogen—we must
conclude that the varying degree of surface saturation of the catalyst atoms also
involves a varying catalytic capacity of the surface atoms. This varying
capacity for catalysis has already been demonstrated experimentally by Vavon
and Huson.f
They found* that the hydrogenation of propyl ketone by colloidal platinum
could be suppressed by the addition of a limited amount of carbon disulphide.
The poisoned catalyst would, however, still effect the hydrogenation of both
piperonal and nitrobenzene. A further addition of the poison in limited
amount suppressed the activity of the catalyst in the hydrogenation of piperonal,
but not in that of nitrobenzene. Finally, with a large amount of poison, the
remaining reaction with nitrobenzene could be stopped. This clearly demon­
strates the varying hydrogenation capacity of a catalyst material.
This evidence as to varying capacity of the surface to effect catalysis as
revealed by progressive poisoning indicates that the amount of surface which
is catalytically active is determined by the reaction catalyzed. There will
be all extremes between the case in which all the atoms in the surface are active
and that in which relatively few are so active. The former will occur in a
readily catalyzed process. The latter has been realized visually in one case.
Wood has shown J that in a clear glass tube the rate of recombination of
hydrogen atoms is relatively slow. If the inner surface of the tube be simply
scratched, however, rapid catalysis occurs at the scratch and is rendered visible
by the incandescence of the catalyst at that point, due to the energy of reaction.
It may be noted in passing that this is direct evidence that the unsaturation
of the surface atoms (due to the scratch) has produced the catalytic efficiency.
The same atoms in the vitreous state were ineffective. Activation consists in
* ‘ J. Am. Chem. Soc.,’ vol. 45, p. 2235 (1923), and unpublished data.
t ‘Compt. Rend.,’ vol. 175, p. 277 (1922).
X ‘ Roy. Soc. Proc.,’ A, vol. 102, p. 1 (1922).
110 Theory o f the Catalytic Surface.
increasing the amount of the surface atoms which are in a state of unsaturation
relative to the main body of the catalyst material.
It is evident from the preceding paragraphs that the theory of a catalytic
surface which has been put forward harmonizes with the facts of the
investigation of such surfaces to a considerable degree. Thus far, however, it
is an ad hoc hypothesis, set up to fit a number of experimental facts already
known. Experimental evidence that unsaturation of the catalyst surface,
such as has been postulated, does actually exist would strengthen considerably
its intrinsic probability. A recent research and some considerations based
thereon seem to provide such confirmation. Garner and Blench* have
recently published data on the heat of adsorption of oxygen on charcoal.
They have obtained a range of values varying from 60,000 calories, at low
temperatures and high adsorptions, to the astonishingly high value of
220.000 calories at high temperatures and low adsorptions. The astonishing
nature of this latter value is at once evident when it is remembered that the
heat of combustion of solid carbon to gaseous carbon dioxide is only 94,GOO-
97.000 calories. The heat of adsorption may, therefore, be more than twice
as great as the heat of combustion of the solid body.
On the basis of the theory of a catalytic surface which has been put forward
in the preceding paragraphs, this high value is to be expected. It has already
been pointed out that the atoms in the surface of a granule most completely
detached from the main bulk of the material are atoms most closely analogous
to gaseous atoms of the same material. Adsorption of oxygen at such isolated
carbon atoms would much more closely resemble the conversion of gaseous
carbon to carbon dioxide than the conversion of the solid to gas. But, the
combination of gaseous carbon and oxygen involves the abnormally high
heat effect of 380,000 caloriesf owing to the high heat of vaporisation of
carbon. The value obtained by Garner and Blench is intermediate between
that of the gaseous and solid oxidation processes, as would be predicted from
the theory.
This observation provides a thesis upon which a considerable body of
experimental investigation can be based. It is apparent that the thermal
data obtained with material which shows a high ratio of surface to mass should
be of greater magnitude than those obtained with material in the coarsely-
crystalline form. It has recently been shown by BackstromJ that finely-
* ‘ J. Chem. Soc.,’ vol. 125, p. 1288 (1924).
t Fajans, * Verh. d. Phys. Ges.,’ vol. 14, p. 324 (1912).
f Unpublished experiments.
 Study of Catalytic Actions at Solid Surfaces. Ill

ground Iceland spar lias a definitely higher heat of reaction with acid than the
coarsely-crystalline material. The greater solubility of finely-divided materials,
such as barium sulphate, indicates also a higher heat of solution of finely-
divided as compared with coarsely-crystalline materials. It suggests that
charcoal should have a higher heat of combustion than graphite, since the
former must be less completely crystalline, though graphitic in crystal form.
This fact is recorded in the literature, a value of 97,000 calories as opposed
to 94,000 calories for graphite being obtained. Uncertainty has attached to
the former figure owing to the difficulty of removing all hydrogen from the
samples of charcoal studied. But, on the basis of the theory here put forward,
a definitely higher value should be obtained for charcoal and for all other
materials in which the process of crystallization is not complete.
The theory of a catalytic surface here put forward may be extended to show
how the two important problems—namely, specificity of catalytic action and
promoter action—may be, in part, elucidated. These two problems will
receive more extended treatment in a subsequent communication, when the
opportunity will be taken to bring forward new experimental material bearing
on these two points.

A Study o f Catalytic Actions at Solid Surfaces. P art —

Some Observations relative to those Particles a Catalyst

which participate in Chemical Change.
By E. F. A rmstrong, D .S c., F.R.S., and T. P. H ilditch, D .S c .

(Received February 12, 1925.)

The publication of Taylor’s theory of the catalytic surface* renders it oppor­

tune to describe certain hitherto unpublished experiments which accord excel­
lently with the view which he has advanced.
Therefore, we give a brief outline of the work in question and adduce other
experimental evidence which points to the strong probability that the acting
(nickel) atom is, at the moment when catalytic change occurs, not merely
held to the rest of the solid surfaces by a single constraint, but can be actually,
momentarily, detached therefrom; in physico-chemical parlance, it has
become for the time being a gaseous (nickel) atom.
* Previous paper.