5th SPIERS MEMORIAL LECTURE CATALYSIS : RETROSPECT AND PROSPECT BY HUGH S. TAYLOR The finest memorial to the first secretary of the Faraday Society is the maintenance of the General Discussioiis at the highest level of ex- cellence. Wherever physical chemistry is prosecuted, wherever students are acquiring physicochemical science, there the record of the discussions which Secretary Spiers initiated and organized for so many years will be found indispensable. There is a curious discrepancy in the archives of the Society concerning the General Discussions. In the index of Volume I11 for 1907 two general discussions are recorded, one on “ Osmotic Pressure,” one on “ Hydrates in Solution ”. In the index to the first twenty volumes of the Transactions the earliest recorded General Discussion is in Volume IV, on “ The Constitution of Water ”.I am at a loss to account for such a discrepancy unless it is a simple over- sight. The last General Discussion organized by Secretary Spiers was in October rg25 on “ Photochemical Reactions in Liquids and Gases ” and was held in Oxford. I was present on that occasion, which marked, as the General Discussions so frequently do, a milestone in the develop- ment of the particular phase of physical chemistry then under consider- ation. For such reasons, and with a peculiar sense of privilege and honour, I am happy on this occasion to introduce a general discussion on “ Hetero- geneous Catalysis’’ as the Spiers Memorial Lecturer, in my own Uni- versity, where, as a student in the departmental library, I first learned to appreciate how greatly these General Discussions can contribute to the definition of the present state of science and in what direction progress may develop.Sir Oliver Lodge was then President of the Society. The present Discussion is the third in a sequence which includes two famous predecessors. In 1922 the Discussion on Catalysis brought for- ward two basic concepts, that of Lindemann on the nature of unimolecular kinetics and that of Langmuir on reaction a.t surfaces betweep adjacently adsorbed reactants, with adsorption restricted to monolayers. The dis- cussion on “ Adsorption ” in 1932 was concerned, in the main, with slow processes of sorption and evoked, from our President, a definition of van der Waals’ and of chemisorption in terms of the potential energies between an impinging molecule and a surface.The diagram brought out clearly the manner in which activation energy of adsorption might be involved, and sharply differentiated physical and chemical processes of adsorption, the differences between which had been indistinct up to that time. It emerged that physical adsorption had little relevance to catalysis at surfaces ; the chemisorbed monolayers of Langmuir were the loci of such reactions. It is pertinent here to emphasize that the Langmuir formulation of surface kinetics was reFtricted to those surface reactions in which the velocity of interaction on the surface was the rate-determining process. This condition was indeed fulfilled in the classic researches of Langmuir and in further developments by Hinshelwood, Rideal, Schwab and others.A* 9I 0 RETROSPECT AYD PROSPECT It is useful, however, to recall an example which does not conform to this condition, the decomposition of ammonia on doubly-promoted iron synthetic ammonia catalysts as studied by Love and Emmett.1 They found a kinetic equation, d[NH3] [NH,]O dt LH,]O.~ -~ = k - which, on the Langmuir basis, would suggest a strong adsorption of hydro- gen on the iron surface and a moderate ammonia adsorption on the surface bare of hydrogen, with nitrogen an inert constituent. We now know that such a view is erroneous ; the slow step is the desorption of nitrogen from the surface, the slow sorption being rate-determining in synthesis. This example alone would have justified the emphasis on slow sorption in the monolager, first discussed by the Society in 1932.The availability of isotopic forms of a given molecular species has revealed the variation in temperatures at which these species will interact on surfaces which still further emphasizes the necessity for the concept of activation energy accompanying their chemisorption on catalyst surfaces. Some examples are cited in Table I. TABLE I.-INTERACTION OF ISOTOPIC MOLECULES ON SURFACES Catalyst Fe synthetic ammonia cata- lyst, doubly promoted . 2s * , ,, 8 Rhenium . NicGkl Osmium . , 9 , Isotope Reaction Temp. O C . for Measurable Rates - I95 + 25 +450 + 25 + I00 +500 - I95 +I50 + 250 1 Reference ?I Kummer and Emmett, Brookhaven Conference Report (Dec. 1948), p. I . Taylor and Jungers, J . Amer.Chem. SOL, 1935, 57, 660. McGeer, Thesis (Princeton, 1949). Sadek and Taylor, J . Amer. Chem. SOC., 1950, 72, 1168. Guyer, Joris and Taylor, J . Chem. Physics, 1941, 9, 287. 5 Wright and Taylor, Can. J . Bes. B, 1949, 27, 303. It is important for the further argument that these variations found with technical catalysts are also to be found with the idealized catalysts produced by the Beeck method of evaporating metals to form oriented or non-oriented films. IP these cases, also, where purity and cleanliness of surface have been raised to the highest attainable standards, similar variations in the temperature at which chemisorption occurs are found, although there will be differences between films and technical catalysts in the actual temperature range involved.Thus, on nickel films, Beeck has shown chemisorption of hydrogen even at 1 5 O K, but examines interaction between ethane and nickel films only in the temperature range 200-250° C. In contrast to Beeck's measurements on nickel films are those of Eucken and Hunsmann with reduced nickel from the oxide where the adsorption at zoo K shows a heat of van der Waals' adsorption. Chemi- sorption with 5 kcal. of heat of adsorption is found only at 50° K. While iron films chemisorb nitrogen (as molecules ?) at liquid-air temperatures Love and Emmett, J . Amer. Chem. SOC., 1941, 63, 3297. Eucken and Hunsmann, 2. physik. Chem. B, 1939, 44, 163.HUGH S. TAYLOR I 1 there is a slow activated adsorption of nitroga, with a much higher heat of adsorption, around room temperatures.This latter fact is to be con- trasted with the studies of Emmett and Brunauer with iron synthetic ammonia catalysts where the slow sorption of nitrogen was measured in the temperature interval from 224 to 449' C. This example takes us at once to the heart of a problem which it ought to be the objective of this Discussion finally to resolve. In 1925, a concept of the catalytic surface was formulated which emphasized heterogeneity or, as it came to be expressed, the concept of " active centres ". A variety of evidence on the properties of technical catalysts, which were the only catalysts then extensively studied, contributed to this concept of active centres. This evidence included observations on adsorption by catalysts both active and inactivated by heat treatment.It attempted to account for the great influence of poisons and promoters, present in t NITROGEN ISOTOPE EXCHANGE ON IRON CATALYST 9% I 1 1 1 ! ! 1 1 1 1 ! 1 1 1 1 ~ l 0 1 2 3 4 5 6 7 HOURS FIG. I minimal amounts, and invoked the existence of centres of high activity, very sensitive to poisoning. The quantitative measurements of poison- ing by Pease in the hydrogenation of ethylene and by Almquist and Black on the poisoning action of oxygen on water vapour irammonia synthesis on iron catalysts were conspicuous examples of such studies. Over the zs-year period, and largely due to the work of Balandin in Russia, of Roberts on the properties of a clean tungsten wire surface and of Beeck and his co-workers on evaporated films, a contrary view has emerged which has sought the interpretation of catalysis solely in terms cf the properties cf plane faces of crystalline materials, which, by specialized techniques, could be studied in a " clean " condition.It is my purpose to attempt a reconciliation of these two points of view in a more generalized and uni- fying concept. on the exchange reaction between light and heavy nitrogen on iron synthetic ammonia catalysis and on rhenium surfaces at 450°C provides a clue to such aa attempt. The velocity of Emmett and Brunauer, J . Amer. Chem. SOC., 1940, 62, 1732. McGeer, Thesis (Princeton, 1949). A recent research by McGeerI 2 RETROSPECT AND PROSPECT reaction was shown to be very sensitive to the reduction process t o which the catalyst was submitted prior to the velocity measurements.In Fig. I the slowest rate is that of an iron-alumina catalyst (No 954, 1-5 yo A1,0,) which had been prepared by reduction at 450' C with a fast stream of tank hydrogen containing 0-15 yo oxygen. The intermediate rate shows the rate of exchange when the same catalyst was subjected to re- duction in a stream of hydrogen from which oxygen and water vapour had been removed by the normal techniques of oxygen removal and drying of the gas stream. The residual water vapour was probably well below 0.01 yo. The fastest rate of exchange was obtained when excessive pre- cautions were taken to ensure dry, oxygen-free hydrogen for reduction. Similar findings resulted with the rhenium catalyst. The data suffice to show how very sensitive the exchange reaction is to the residual traces of oxygen that are left on an iron surface even with good reduction tech- niques.In the terminology of 20 years ago this is a typical example of " active centres " on a technical catalyst. It parallels entirely, as it should, the quantitative data obtained by Almquist and Black5 on an iron-alumina catalyst in ammonia synthesis in presence of water vapour as a poison. TABLE II.-AMMONIA YIELD AT A SPACE VELOCITY OF 25,000, 444°C AND The data in Table I1 recall some of these results. I ATM. PRESSURE Mg : 0, retained by catalyst . yo NH, produced The very marked effect of the first 5 mg. of oxygen retained is the mole striking since, from B.E.T. measurements of the surface area of this catalyst, only 10 to 15 yo of the total surface would be covered by this oxygen.Because the ammonia synthesis reaction is determined in rate by the slow step of chemisorption of nitrogen it is apparent that we must seek the interpretation of the " active centres " or quasi-heterogeneity in the effect of adsorbed oxygen on the surfaces of the iron crystallites. The presence of oxygen results in an incieased energy of activation of nitrogen adsorption on the iron surface. A generalization of this point of view for technical catalysts makes possible a reconciliation between the findings of those who have worked with clean tungsten wires and evaporated metal films and those whose attention has centred on the properties of technical catalysts. In each case one is concerned with the properties of one or more crystal faces of a particular catalytic species.In the case of technical catalysts, however, these properties may be profoundly modified by the presence of poisons as adsorbed oxygen or added ingredients such as alumina in the case just considered. Our knowledge in this area is being rapidly increased by reason of the studies of electron emission from hot wires and by studies of the properties of semi-conductors. One need only cite the beautiful studies of Miiller6 and of Jenkins relative to the emission of electrons from fine tungsten points as revealed by the field-emission projection electron microscope, and the influence of adsorbed oxygen, barium, thorium and sodium on the emission process. The varying activity of different crystal faces, the preferential adsorption of the poison on particular faces and the migration of poisons at definite temperatures can be visually demon- strated.Especially, however, the data accumulating on semi-conductors demonstrate how profoundly the conducting properties of a pure substance 5 Almquist and Black, J. Amer. Chem. SOC., 1926, 48, 2814. 6 Miiller, 2. physik., 1949. 126, 642. Jenkins, Reports Prog. Physics, 1943, 9, 177.HUGH S. TAYLOR 13 can be modified by traces of a prescribed impurity. One may cite in this respect the recent data of Pearson and Bardeen8 on the activation of silicon as a semi-conductor by additions of boron. The addition of 0.0013 atomic per cent of boron significantly changes the energy required to release an electron, whilst a concentration of 0.013 atomic per cent lowers this energy to zero. As Mott pointed out recently in the Kelvin Lecture @ impurity centres may be imagined as " gieat swollm ' atoms ' extending over 10-20 lattice parameters ',, a concept pertinent to the whole problem of interaction between adsorbed species on a surface.We must assume, in the case of the iron-synthetic ammonia catalyst already discussed, that the presence of impurity centres raises the activation energy of ad- sorption of nitrogen as oxygen would raise the activation energy ?f con- duction in zinc oxide semi-conductors. In this manner the findings of Beck on evaporated iron films can be correlated with those of Emmett and Brunauer and of McGeer with the technical iron catalysts. It is, on this basis, the impurity centre randomly distributed over the plane face of a crystal which would confer on such a crystal face quasi-heterogeneity.W7e can draw from the data on semi-conductors further analogies to problems of surface catalysis. VenVeyJ1O cited by Mott, has prepared semi-conducting NiO by dissolving in the lattice Li,O, the radii of the Li+ and Ni++ being practically identical. On heating in air, Li+ replaces Ni++ in the lattice and €01 each Li+ introduced a Ni+++ ion is formed. The latter are carriers of current by electron displacement. The magnetic measurements carried out by Selwood and his co-workers on valence induction in nickel and other oxides are illustrative of the same effect." Magnetic measurements, on nickel oxide impregnaked on y-alumina sup- ports, indicate that the nickel is present in the trivalent form.On mag- nesia, the isomorphous divalent oxide is present. On the rutile structure of titania the nickel assumes a valency of four, the oxide being bright yellow in colour. Manganese and iron oxides behave similady. The relation 01 valence induction to semi-conductors and to the whole problem of promoter action in catalysis is illuminated by such studies. Let us recall that the activation energy of adscrption of hydrogen by zinc oxide is markedly diminished by incorporating chromium oxide in the pre- paration. On the viewpoint here presented poisons and promoters become impurity centres in the normal lattice of the catalyst, the former tending to raise the activation energy of adsorption, the latter to lower it. A further factor which can be influenced by such impurity centres in a catalyst surface is the heat of adsorption.The data are scanty but what data are available tend to indicate that the measured heats of chemisorption are markedly less on technical catalysts than on the cor- responding evaporated films. The data of Beeck recorded for this Discussion on a variety of metal films for hydrogen, ethylene and nitro- gen are conspicuously higher heats of adsorption than have been recorded by Beebe, Eucken and others with technical catalysts. Indeed, the high heats of adsorption on evaporated films constitute a grave disadvantage of these films regarded as catalysts. They are, indeed, clean, but they " die " after brief experimentation because they are self-poisoned owing to the high heats of adsorption of one or more of the reactants.The technical cataljsts, though " dirty", at least " live ", largely because of the lower heats of binding to surfaces having " impurity centres ". From such circumstances it can result that a technical catalyst may have a higher activity than an evaporated film. has made one such Wright Pearson and Bardeen, Physic. Rev., 1949, 75, 865. Mott, Proc. Inst. Electr. Eng., 1949, 96, 253. 10 Verwey, Haayman and Romeyn, Chem. Weekblad, 1948, 4, 705. l1 Selwood, Bull. SOC. cham. France D, 1949, 489. l2 Wright, Thesis (Princeton, 1949).I 4 RETROSPECT AWD PROSPECT concrete comparison. With 27 mg. of a nickel-chromia (80 yo Ni) catalyst in a reaction volume of 350 cm.3, i.e. 0.077 mg. catalyst per ~ m . ~ gas, containing an equimolar hydrogen-ethylene mixture Wright found a half-life of tllz = 4 min.at -78’ C. From Beeck’s data and his temper- ature coefficient one computes a half-life of 45 min. at -78” C with 30 mg. nickel film in a reaction volume of 400 cm. or 0.075 mg. catalyst per ~ m . ~ of reacting gas. On this basis, the technical catalyst is 11 times as efficient as the evaporated film. The cause is obvious from an examin- ation of Beeck’s paper to this Discussion on the reactions of hydrocarbons. Most of his surface is covered with “ acetylenic residues ” by reason of the high heat of chemisorption of ethylene, thereby becoming ineffective for the catalytic interaction. Beeck’s data on the heat of adsorption of hydroen on clean tantalum, 45 kcal., and on a nitrided tantalum film, 27 kcal.are evidence for a lower heat of adsorption on metal surfaces covered in part with other constitnents. Another method of formulating the same idea is that, with technical catalysts, the areas responsible for the high initial heats of adsorption found with metal films are already occupied by impurity centres. Beeck’s data for heats of adsorption of hydrogen on his metal films vary directly with the change in d-band character of the metals. Boudart has recently l3 shown that Beeck’s measurements of activities in the hydrogenation of ethylene on metal films increase with increasing d-band character of the metallic bond as given by Pauling, with rhodium of maximum activity. This suggests that “ the lattice parameter is not to be considered solely as a cause but as an effect.The primary cause has to be sought in the electronic structure of the metal and a deeper insight into the latter may be obtained by means of Pauling’s theory.” From this point of view there is 9 notable discrepancy between the findings with films and technical catalysts in the case of copper. Beeck reports no hydrogen adsorption on films of copper whose d-band is filled. Technical copper catalysts have revealed chemisorption of hydrogen -from the eariiest studies of adsorption on catalysts, with a heat of adsorption of some 10 kcal. It is pertinent to ask whether this discrepancy is to be associated with the intrinsic nature of technical copper catalysts, to what extent it may be a function of a mixed Cu+-Cu++ structure deriving from impurity centres of a promoting character.Similar considerations, involving activation energy or heat of adsorption or both, animate the research work on catalysis by metal alloy systems. Schwab associates the catalysis in the decomposition of formic acid, on alloys of silver and of gold,l4 with a variety of metals giving both homo- geneous and heterogeneous phases, with an entry of protons into the interlattice planes and electrons dissolving in the electron gas of the metal, the required activation energy depending on the degree of com- pletion of Brillouin zones which define the allowed electron energies in the metallic stlucture. Similarly Couper and Eley l5 have associated the increase in activation energy of the hydrogen-deuterium exchange with the filling of the partly empty d-band of palladium by alloying with gold.Dowden l6 has treated the problem in detail theoretically and Reynolds l7 has applied the treatment to experimental findings on homo- geneous solid solution binary alloys of Group VIII metals with copper. In all such cases both activity and activation energy should be studied ; it is preferable also to study exchange reactions between isotopic mole- cules in order to minimize the influence of other factors such as displace- ment effects whcn two different molccular species are competing for a given surface. In this area there is a wealth of experimental opportunity. l3 Boudart, J . Amer. Chem. SOC., 1950, 72, 1040. l4 Schwab, Trans. Faraday SOC., 1946, 42, 689. l5 Couper and Eley, Nature, 1949, 164, 578.l6 Dowden, Chem. and Ind., 1949, 320 ; J. Chem. SOC.. 1950, 242. l7 Reynolds, Chem. and Ind., 1949, 320 ; J. Chem. SOC., 1950, 265.HUGH S. TAYLOR 1 5 Parallel studies with oxide systems will be equally interesting. In these, the properties of the catalytic oxide viewed as a semi-conductor will be useful aids to understanding. One example is copper oxide, recently examined by Garner, Gray and Stone,18 in terms of the electrical conductivities of thin films during formation, on reduction and after adsorption of gases. Oxygen enhances, hydrogen and carbon monoxide depress the conductivity of the Cu,O-CuO surface. The conductivity is interpreted in terms of a movement of electrons across an array of Cu+ and Cu++ ions. In contrast to cuprous oxide which is an oxidstion semi-conductor, zinc oxide is a reduction semi-conductor losing oxygen in a vacuum, the conductivity increasing and the excess zinc atoms enter- ing interstitially into the lattice as ions, the electron being held in the field of the positive charge.In this way the oxide acquires the characteristics of a metal, hydrogenating in character. Foreign ions can enter the lattice substitutionally and interstitially, electrons being trapped in their fields, the radius of the orbit of the trapped electron extending over several interatomic distances. The conductivity of the semi-conductors varies with the temperature and there is evidence that a whole spectrum of energies may be involved. Our knowledge of the properties of solid catalysts conferred by im- purity centres or admixtures of two or more constituents is still very much in the qualitative stage.It is known that the admixture of alu- minium oxide in magnetile increases considerably the difficulty of re- duction of the iron oxide. On the other hand the admixture of alumina has little influence on the dissociation pressure of copper oxide, but in- creases considerably the oxygen dissociation pressure with cerium oxide, CeO,. Chromium oxide admixture strongly increases the dissociation pressure of copper oxide. These findings on dissociation pressure are paralleled by the results of Rienacker l9 on the catalytic activity of such mixed oxides in the oxidation of carbon monoxide. The added oxide in- fluences both activity and activation energy of the basic catalyst, with marked lowering of the activation energy in those cases where admixture of the second oxide increases the oxygen dissociation pressure, and mith increase of activation when the added oxide decreases the reducibility of the basic catalyst.The old data of Kendall and Fuchs 2o on the influence of added oxides on oxygen release from barium peroxide, silver and mercuric oxides need to be recalled. The influence of added oxides on the slow sorption of hydrogen as illustrated by the accelerating influence of chromium oxide and the retarding effect of molybdenum oxide on the adsorption of hydrogen by zinc oxide is an example in another area. Such material should now be re-examined in the light of data revealed by Selwood on valence in- duction and with the newer techniques stemming from the study of semi- conductors.The catalyst itself, rather than the reactions which occur on it, seems to be the principal objective for future research in the coming years. This is not to deny the importance of a study of such reactions, since they, also, can reveal the nature and action of the catalyst surface. An excellent example in this respect is the recent work on the nature and functions of cracking catalysts, as exemplified by the recent con- clusions of Greensfelder, Voge and Good based on earlier work and newer investigations of the cracking of cetane, cetene and other hydro- carbons thermally and over catalysts such as activa.ted carbon, alumina, silica and alumina-silica commercial cracking catalysts. Two funda- mental types of cracking emerge, characteristic both as to the type of In the older literature there are many such examples.l8 Garner, Gray and Stone, Proc. Roy. SOC. A , 1949, 197, 294. 2o Kendall and Fuchs, J . Amer. Chem. SOC., 1921, 43, 2017. 21 Greensfelder, Voge and Good, Ind. Eng. Chem., 1949, 41, 2573. Rienacker, 2. anorg. Chem., 1949, 258, 280.16 RETROSPECT AXD PROSPECT primary and secondary reactions. Two types of mechanism are involved. The one involves free radical fragments and the pattern of cracking is described by the Rice-Kcssiakoff theory of cracking. This mechanism is characteristic of thermal cracking. The other mechanism is ionic in nature conforming closely to acid-activated, carbonium ion mechanism. Commercial acid -treated clays and synthetic silica-alumina catalysts are of the latter type.Activated carbon, an active, non-acidic catalyst gives a product distribution which can be interpreted as a free-radical type of cracking, quenched, as compared with thermal cracking, by reaction of free radicais with chemisorbed hydrogen on the catalyst to yield the corresponding normal paraffin, With silica-alumina, acidic- type catalysts, isomerization accompanies cracking to yield its particular spectrum of products. Such acidic surfaces are poisoned specifically by organic bases such as quinoline and by potash, ammonia and other alkalis, Oblad and his co-workers indicate that some 4 yo of the total surface is involved in such catdyses. Volkenstein 22 has recently analyzed the concepts of adsorption and of adsorption kinetics from the standpoint of the solid catalyst viewed as a structure containing a certain concentration of lattice defects, two-fold in nature.On the one hand, there are macro-defects, such as cracks, whose perturbations exceed those of the crystallographic unit. On the other hand, there are micro-defects which exercise perturbations of the order of the unit cell, the periodic structure being re-established at a distance of a few lattice parameters. In the terminology of semi- conductors, these defects include holes or vacant sites, interstitial atoms or ions, ions in a heteropolar lattice with normal position but anomalous charge, or foreign atoms in substitional or interstitial positions. Such defects deform a region of the lattice and it is the region which should be regarded as the defect.It is with the properties of such regions that catalysis may well be associated. Volkenstein introduces into his con- siderations the concept of mobility of micro-defects with an associated activation energy determined by the nature of the defect and the lattice and by the direction of migration. These micro-defects may react, attract or repel each other, dependent on the charges involved. Two defects may interact to produce a new defect with different properties, Volkenstein envisages the " disorder " in terms of the total number of defects which is small compared with the total number of unit cells. Eisorder may be " biographical ", that which is present at oo K and arising from the circumstances of the catalyst preparation or it may be " thermal " or that which results from the effect of temperature.The biographical disorder x, at oo K will increase progressively with temperature to a maximum y at T = 00, the thermal disorder a t any temperature varying from o at T = oo K to y - x at T = co. The relative importance of biographical and thermal disorder depends both on the temperature and the " biography ". In the older theories of adsorption one assumes (a) constancy' of ad- sorption centrts, (6) immobility of adsorption sites, and ( G ) invariance of sites with coverage. None of these assumptions enter necessarily into Volkenstein's treatment. Volkenstein assumes that adsorption occurs in the defect region. With one kind of defect the surface is energetically homogeneous.With differing defects there will be heterogeneity. Ad- sorption creates new centres. Increase of temperature increases the number of adsorption centres up to b u t not beyond the maximum, y. Volkenstein shows that the Langmuir isotherm results when the whole disorder is biographical. When, however, the defect is thermal involving an absorption of energy u, then, depending on the nature of the process producing the new defect whether unimolecular or by bimolecular inter- action of two defects, the mathematics shows that, with no heterogeneity 22 Volkenstein, Zhuw. Fiz. Khim., 1949, 23, 917.HUGH S. TAYLOR 17 of sites and without interaction between adsorbed atoms, the adsorption isotherm may conform to the Freundlich equation, uads = hpl/n, and the differential heat of adsorption can fall from an initial value q to a minimum q - u.The curve obtained is reverse sigmoid falling at first slowly, then rapidly and finally asymptotically to q - u at complete coverage. The adsorption centres on the crystal surface operate as a kind of plane gas the concentration of which increases on filling up and the change of energy of which is measured by the differential heat of adsorption. For the kinetics of adsorption the bimolecular production of sites can be shown to yield either a rapid Langmuir kinetics determined by the rate at which molecules strike the surface or an adsorption proportional to the square root of the time. For a unimolecular producticn of sites the kinetic expression for adsorption is that typical of the so-called activ- ated adsorption, k = k , exp (- E,,t/RT).In the theory of activated adsorption a potential bzrrier bntween adsorption sites and impinging molecule is assumed. In Volkenstein’s treatment it is the number of sites which increases with temperature and proportionally to exp (--u/RT). No potential barrier is assumed. An alternative method of stating the same is to say that the number of sites remains constant but the number of excited centres which are able to absorb increases as observed. There are several observations from the older literature on activated adsorption which can readily be interpreted on the bases of the Volkenstein concepts. In the General Discussion €or 1932 it was pointed out that the velocity of activated adsorption of hydrogen as measured on zinc oxide was very much smaller than the number of molecules striking the surface with the necessary activation energy.Our subsequent knowledge, from B.E.T. measurements, of surface area indicates that the discrepancy be- tween calculation and observation was of the order of IO*. A similar discrepancy was noted by Emmett and Brunauer in their careful measurements of the velocity and - activation energy of adsorption of nitrogen on synthetic ammonia catalysts. In this case the discrepancy was a t least of the order of IO~. From Volkenstein’s point of view these data would involve both the concentration of defect centres and the activation energy of adsorption of the molecule. In the case of doubly- promoted ammonia synthesis catalysts this might well be associated with the iron-potassium-oxide-aluminium oxide centres where the varying valency of iron, the univalent potassium and the trivalent aluminium ions suggest at once the possibility of lattice defects, as was pointed out to me by Boudart.Boudart has also emphasized in this regard the ob- servations of Pace and myself 23 on the identity in the velocities of adsorp- tion of hydrogen and deuterium on oxide catalysts which led to the con- clusion that “ the activation energy necessary is required by the solid adsorbent ’,. Other data on the influence of pressure on the velocity of chemisorption of hydrogen on oxides such as chromium oxide gel are also in agreement with the view that velocity is not determined by thz number of molecules striking the surface. To account for the slow sorptions ob- served it was earlier suggested that interaction must occur between the adsorbent and van der Waals’ adsorbed gas. The Volkenstein concept represents a mechanism by which such would be achieved. The apparent saturation capacity of an oxide surface for hydrogen adsorption at a given temperature and the large change to a new apparent saturation a t another temperature, facts familiar to a.11 who have studied the slow sorption processes on oxides, should be re-studied in reference to Volkenstein’s assumption that the sites available for adsorption, the “ thermal ” sites, vary with temperature. On this view the measure- ments of Shou-Chu Liang24 and the writer would gain new significance. In brief, measurements of slow sorption on oxide surfaces need to be 23 Pace and Taylor, J . Chem. Physics, 1934, 2, 578. 24 Taylor and Liang, J . Amer. Chem. SOL, 1947, 69, 1306.18 MOLECULE NEAR METAL SURFACE re-examined from the standpoint of the entropy as well as the energy of activation . In retrospect, three decades of scientific effort devoted to the science of catalytic phenomena have revealed a wealth of detail and understanding not available to the technologist in the empirical developments of the nineteenth and early twentieth centuries. The background of scientific theory and data that has been accumulated since Mr. Spiers first organized a General Discussion on Catalysis has led to a surer and swifter attack on any new catalytic problem that emerges. Out of the discussions which will ensue, here in Liverpool, we may anticipate a reconciliation of the several attitudes that sometimes have appeared to divide us, but are, in reality, a spur to further and continued effort towards the mastery of our science in an era which is of deep significance in all human affairs. In prospect, therefore, the future of our science is both challenging and bright. Chemistry Department, Princeton University, Princeton, N . J .