I 28 THE EXISTENCE OF ACTIVE CENTRES ON THE EXISTENCE OF ACTIVE CENTRES IN CHEMICAL ADSORPTION AND CONTACT CATALYSIS BY A. EUCKEN Received 25th January, 1950 The problem as to whether chemisorption and catalytic reaction occur over the whole available surface of a catalyst or only at active centres is discussed. The apparent heterogeneity arising from mutual interaction of adsorbed par- ticles and the possibility of different states of the adsorbate on the surface are considered in detail for the adsorption of hydrogen on highly activated nickel. It is concluded that the supposed effect of active defects should be rejected and that the essential part is played by the active intermediate states of the adsorbed reactant.A. EUCKEN A final answer has not yet been given to the question as to whether chemical adsorption or catalytic reaction occurs on the whole available surface of the solid adsorbent or catalyst, or only at certain places (active centres).It is, however, doubtful having regard to the complexity of this field, whether this question can ever be completely answered. It should therefore be noted that the conclusions obtained here depend on a limited number of facts, and that different conclusions might be obtained with different types of adsorption processes or catalytic reactions. I t is advisable to consider the existence of active centres in chemical adsorption separately from contact catalysis since, contrary to a view held earlier, it appears that the chemical adsorption of any particle is frequently, but by no means always, connected with an increase of its chemical reactivity (for this reason the term " activated adsorption ", which is frequently used, is not considered suitable).Chemical Adsorption.-Two cases of the inhomogeneity of solid surfaces in regard to chemical adsorption can a priori be distinguished. (i) The heat of adsorption and also the activation energy often varies considerably with coverage, and this has led to the conception of the existence of a continuous spectrum of adsorption centres of different activity, which might be identified with the lattice defects of the contact surface. (ii) On heating, the adsorption isobar frequently shows two maxima which, following H. S. Taylor, are considered as an indication of the existence of two (homogeneous) regions, whose adsorption capacity differs from that of the bulk.No details are given as to how this comes about. We can consider these two regions as crystal planes having different arrangements. Fundamentally, however, at least with the ionic type of adsorbent, it is possible to consider this difference as being determined by the individual lattice units, for example, the one as an adsorption on the cations, and the other on the anions. An alternative interpreta- tion of these two adsorption effects which will apply to monatomic ad- sorbents (pure metals) and in part also to salts (oxides, etc.), can be pro- posed. (i) THE SURFACE IS PERFECTLY HOMOGENEOUS IN REGARD TO ITS ADSORPTION PROPERTIES.-The apparent inhomogeneity (the existence of a spectrum of active centres) then arises from the influence of the ad- sorbed particles on each other, since a particle is less tightly held in the immediate environment of another adsorbed particle than on a completely clean surface.l If we assume this kind of mutual interaction2 it is clear that the heat of adsorption continually decreases with increasing coverage of the surface.(ii) DIFFERENT STATES OF THE ADSORBATE CAN EXIST ON THE HOMO- GENEOUS SURFACE, i.e. the steps in the adsorption isobars are not associ- ated with the regions of different activity of the surface. These two theories will be discussed in detail for the adsorption of hydrogen on highly active nickel. Two maxima occur with heating, the first being not strongly marked and only observed at relatively low pressures (approximately I O - ~ mm.Hg). I t can be shown thermodynamically for the adsorption equilibrium above 273' K that the hydrogen is essentially atomically bound to the surface. 'For example in the chemical adsorption of hydrogen atoms on a metal (e.g. Ni) the interaction between adsorbed hydrogen atoms occurs in the following way. As the metal surface possesses a certain electron affinity, the hydrogen atoms become positively polarized, and consequently there is coulombic repulsion between them. Furthermore as more hydrogen atoms become available as electron donors, the electron affinity of the metal surface becomes satisfied. At the same time the binding energy of the individual hydrogen atoms decreases. a Recently this point of view has been expressed by Beeck (Rev.Mod. Physics, 1945, 17, 61) ; cf. also Herington and Rideal (Trans. Faraday Soc., 1g44,40, 505). The shape of a typical adsorption isobar is shown in Fig. I. E1 30 THE EXISTENCE O F ACTIVE CENTRES Thus the first maximum of the adsorption isobar can be ascribed to an adsorption of molecules, similar in type to an atomic adsorption of a chem- ical nature, since physical (van der Waals') adsorption is only observed at much lower temperatures (in the region of liquid hydrogen). The heat of adsorption is estimated to be about 6 kcal.,/mole, i.e, it is relatively small, so that on heating, desorption occurs at about zooo K, and at 273' K the fraction of the surface covered by molecules falls to about I yo. However, one would have to assume, not only at this temperature but also at higher temperatures, that the hydrogen is primarily adsorbed as molecules.The adsorbed molecules subsequently dissociate more or less rapidly into atoms, which are initially closely packed but later diffuse FIG. I. Isobars of adsorption velocities (schematically). FIG. 2 . H, adsorption isobars at 10-2 mm. Hg on 28.8 g. Ni powder. Desorption graphs. Expt' T6)Adsorption at oo C on completely pure Ni surface. 0 Expt. 34 I A Expt. 35. 0 Expt. 33. Previous covering at 100' C after adsorption a t oo C . Adsorption at 100' C. Adsorption graphs. 0 Expt. 33 from 200 to 100' C. -0 Expt. 36 from 100' to oo C. apart (slowly at the lower temperatures and more rapidly at higher temperatures), since according to the above consideration a uniform distribution of the atoms over the whole surface corresponds to minimum potential energy.IA. EUCKEN The shape of the isobar on subsequent cooling is abnormal. Contrary to a number of other investigators, a smaller amount of material is ad- sorbed in the range 260' K to 360' K on subsequent cooling, than on heat- ing (Fig. I and 2 ) . The effect becomes clearer on comparing the two velocity isotherms at oo C (Fig. 3 ) . Here, the adsorption was in some cases on pure nickel and in others on a nickel preparation which had been previously covered with hydrogen at 100' and zooo C respectively, FIG. 3. 0 Expt. 28 O- Expt' 31 Adsorption without previous covering. 0 ' I EXPt. 37 A Expt. 29. Adsorption after previous covering at 100' C. 0 Expt. 32. Adsorption after previous covering at 200' C.Isotherms of adsorption velocities at oo C (23.0 g. Ni). This result can hardly be explained by the assumption of two separate regions (independent of each other), since in such circumstance there is no reason to expect previous covering to cause a decrease in the ad- sorption velocity. On the contrary, it can be regarded as a Poisoning of the total surface by the adsorbed atomic hydrogen. On the basis of the assumption previously made, we can visualise this effect as follows. Since the atoms adsorbed at rooo C are almost uniformly distributed over the surrace, then during cooling only those places with relatively small adsorption activity are available for adsorption. Moreover the dissociation and diffusion of the newly-formed atoms will be greatly hindered by atoms already present in stable positions, so that establish- ment of the final stable state, which is characterized by the most uniform distribution of hydrogen atoms over the total surface, is made much more difficult than on a surface initially completely unoccupied. One might expect to find similar effects during the adsorption of hydrogen on other metals ; in the experiments of Emmett and Harkness (1935) adsorption isobars with two maxima are obtained for the system Fe-H,.Unfortunately, in the isobars as yet determined for other metals and gases, the decreasing temperature branches are lacking in almost every case. The results obtained with the Ni-H, system will probably not be directly comparable with those obtained with salt-type catalysts, as con- ditions with these vary considerably from case to case.In Fig. 4 the full line shows an isobar recently obtained for the adsorption of hydrogen on WS,. The quantities of gas chemically adsorbed (near saturation) are approximately the same as for physical adsorption ; therefore the total surface is active in the chemical adsorption. Evidently two processes are superimposed which correspond to the dotted lines I and z in the graph. It is probable that the first graph corresponds to a chemical (atomic) adsorption and the second, to a solution of hydrogen. The1 32 THE EXISTENCE OF ACTIVE CENTRES heat of adsorption (approximately I 7 kcal. /mole) changes relatively little with coverage, at least within the region where measurements could be made.The influence of the adsorbed particles on each other is in this case smaller than for a metallic surface. A number of other salt-type solid surfaces show analogous behaviour to that of the metal and WS,, as here again the physically and chemically active surfaces practically coincide. One of the few exceptions is y-Al,O, for which only about 10 yo of the physically active surface is shown to be reactive in the adsorption FIG. 4. H, adsorption isobar of H, on WS, at TOO mm. Hg. of steam and alcohols. Here the lattice defects must be responsible for the preferred chemical adsorption. This can easily be understood as the lattice of y-Al,O, is known to be of the spinel type in which the Mg2+ ions are replaced by an equivalent number of A13+ ions ; to main- tain electroneutrality 33 yo of the original sites occupied by Mg2+ ions (a total of about 10 yo of the cation lattice sites of the spinel) must remain unoccupied. In the environment of such " holes " there are regions of excess negative charge, which can now interact with the hydrogen atoms of the hydroxyl groups of water and the alcohols, and thus cause strong adsorption.Apparently we are concerned here with a reinforced hydrogen- bridge binding. CONTACT CATALYsIs.-We shall confine ourselves to the discussion of a typical example of low temperature catalysis, since in high temper- ature catalysis (e.g. the catalytic combination of H atoms to molecules on metals) there is no doubt that the total surface is catalytically active. However, it may at first appear as though in low temperature catalysis, e .g . the hydrogenation of unsaturated hydrocarbons in the region of of oo C, the chemical change occurs only on a negligibly small fraction of the contact surface, i.e. as if the assumption of the existence of active centres cannot be avoided. As an example we may quote an experiment by Toyama (1937) and its interpretation. It deals with the hydrogenation of ethylene on the surface of a Ni powder at a pressure of about I mm. Hg. The reaction is zero order with respect to ethylene, and first order with respect to hydrogen. Taking the reaction velocity as -- dn -(--- no. of moles a value of the order of I x 10-l~ per cm., of the contact surface was ob- tained. Assuming that every H, molecule from the gas phase, colliding dt .p ~ ~ : ~ ~ ~ ) sec. mm. Hg. ern..)'A. EUCKEN I 3 3 with sufficient energy with the ethylene adsorbed on the surface leads t o a reaction, an activation energy of 5000 cal./mole and a value of 2.5 x 10-7 for dn/dt are obtained. The collision factor is approximately 4 x 10-6, which may be taken to mean that only a very small part of the total surface is catalytically active. This and similar results support the view that the reaction only occurs on a relatively few small centres of the surface ; but the problem can by no means be regarded as finally solved. Some experiments under- taken with a view to a solution of the problem showed first of all t h a t ethylene a t normal pressures has a strong poisoning effect on the reaction. Starting with a carefully cleaned surface and allowing the reactants to react simultaneously at low pressures (of about I O - ~ to 10-1 mm.Hg), values IOO times greater are obtained for the reaction velocity. In order t o remove the possibility of any poisoning effect remaining even at low pressures, an attempt was made to replace the ethylene by a hydrocarbon the poisoning effect of which would be much less. It was shown that cyclohexene served this purpose, and was also very suitable for use in other contact catalytic hydrogenation experiments. The real clue to the understanding of the processes occurring in the low-temperature hydrogenation on Ni was found in experiments with a, relatively large amount of solid, on which one of the reactants was adsorbed; the other was subsequently added and allowed to react from the gas phase, If cyclohexene was adsorbed, and then hydrogen allowed to react in this manner, only a negligibly small amount of C6Hl, was formed.If, however the hydrogen was adsorbed and the cyclohexene subsequently added, a considerable change was obtained. This surprisingly depended to a. great extent upon the time interval between adsorption of the hydrogen and addition of the C6H,,. It was also of importance whether the hydrogen was adsorbed a t oo C or 100' C. I TABLE HYDROGENATION EXPERIMENTS. SURFACE PREVIOUSLY COVERED WITH HYDROGEN I Ni powder (670 mg.) ; reaction temperature : oo C I 5 5 5 5 I5 5 5 5 2 Expt. No. I 2 3 4 5 6 7 8 9* Relative Covering of Surface with H, 0.200 0.172 0'143 0.404 0.400 0.286 0.286 0.306 0.310 Temp.of Previous Covering (" C) 0 0 0 0 0 I00 I00 0 -44 Time between Covering and Start of Reaction Duration of Reaction (min.) Yield of CIHlZ (%I 86 80 I 7 45 66 I9 35 68 I 2 * At - 46' C for reaction temperature. The results are summarized in Table I. It can be seen that the atomically adsorbed hydrogen, distributed over the whole surface, is apparently not capable of reaction. The fact that the reactivity of the hydrogen adsorbed a t oo C for a short duration of adsorption is rather large but decreases rapidly with time, may mean that the adsorbed molecules cause the reaction, prior to their dissociation and subsequent diffusion as atoms. A more detailed quantitative calculation of the reaction kinetics indicates that the adsorbed molecules as such most 9 Pease (1923) was the first t o point this out.I 3 4 THE EXISTENCE O F ACTIVE CENTRES probably do not determine the kinetics of the reaction.Finally we may suppose that a characteristic maximum of the reactivity is reached im- mediately after their dissociation i.e. in a state where two hydrogen atoms are still close together or form a H-atom pair. On the basis of this picture it is possible to calculate the course of the catalytic hydrogenation of C6HIo from the gas phase on Ni, if we assume that the change of coverage per unit time O,, is proportional to the number of H atom pairs and to the number per sec. colliding with C6H10 molecules from the gas phase. The adsorption velocity of the hydrogen also enters into the process, but for this a sufficiently accurate empirical expression is available.In the final expression for the reaction velocities we have to include the collision factor, the activation energy and the ratio A / v of the active contact surface to the reaction volume. Taking all the experiments into account, a collision factor of about 0-1 was obtained, which is of a reason- able order of magnitude. The activation energy was calculated in the usual manner from the temperature dependence of the reaction velocity. The ratio A / v can be estimated with an accuracy of at least & 50 yo, as the effective surface was always determined by physical adsorption experiments. The experiments were carried out with a series of different hTi surfaces (nickel-powder from carbonate and formate, nickel-foil, Raney-Nickel from NiA1, and Nisi,, evaporated nickel films) at pressures of I O - ~ to 10-1 mm.Hg. The observed variation of the reaction velocity with time was on the whole given satisfactorily by the final equation; as a rule, however, it was found that the reaction velocity decreased rather more rapidly within the individual series of experiments than demanded by the equation. This can be explained by the fact that a small amount of adsorbed hydrogen is not used up during the reaction but distributes itself over the surface in the atomic form and thus poisons it. It is remarkable to find that the final expression remained quantitatively valid for all the Ni preparations employed, provided they had not been previously heated above 280° C.l One gains the impression therefore, that on pure nickel planes irre- spective of the method of preparation, the same hydrogenation mechanism always occurs, and thus the supposed effect of active defects, which should necessarily show some difference in the compounds prepared differently, must be rejected.This must not, however, be taken to mean that at very instant the total surface is catalytically active. It seems more reasonable to regard the sites instantaneously occupied by H-atom pairs as temporary centres of catalytic activity, their position on the surface being subject to con- tinuous change. From the above, the function of the surface covered with these temporary centres, OH,, may easily be calculated ; in our experiments values between 5 x 10-3 and I x 10-1 (maximum) accord- ing to the magnitude of the ratios of pH,@,, were obtained. In fact, at any instant, only a relatively small fraction of the surface is taking part in the reaction, whereas the major part of the stably adsorbed reactants (C6HIo and H atoms) covering the surface, are not at all, or only to a very small extent, capable of reaction. It may be expected that similar conditions exist for other low temperature catalytic processes and that an essential part is played by the active intermediate states of the adsorbed reactants.* Institut fur Physikalische, Chemie der Universitat Gottingen, Gottingen, Biirgerstrasse 50, Germany. * The results communicated here are given in detail in Naturwiss., 1949, 36, 74 ; 2. Elektrochem., 1949, 53, 285, and 1950, 54, Heft 2.