CHAPTER 1
The Reactions of Hydrocarbons on Multi metal lie Catalysts
BY D. A. DOWDEN
1 Introduction
Elemental metals have been used since the Industrial Revolution as active phases contributing their characteristic catalytic properties to both simple and complex catalysts. Until the late fifties most such catalysts contained only a single base metal, as for example nickel extended upon kieselguhr, copper stabilized with chromia or unsupported cobalt (for the selective hydrogenation of various functional groups), iron in the Haber synthesis of ammonia and the Fischer-Tropsch process, etc. Individual precious metals, notably platinum and palladium, were widely used in small quantities for specific hydrogenations on a small scale, and to a larger extent in processes involving oxidation, e.g. platinum in earlier sulphuric, nitric, and hydrocyanic acid plants, and silver for formaldehyde and ethylene oxide production. Indeed the best examples of the application of well-defined binary alloys in industrial catalysis were and are platinum-rhodium solid-solutions (wire gauzes for ammonia oxidation but also supported crystallites for the modern Andrussov HCN process).
The petroleum refining industry is currently a major user of heterogeneous catalysts and the energy crisis has speeded the search for catalysts and processes leading to greater efficiency and economy. Catalysis in petroleum refining developed rapidly in the period 1930 — 1945 initially because of the attempts to devise a viable process for the production of petrol by the hydrogenation of coal, and later because of the exigences of the Second World War. The large amounts of compounds of sulphur, nitrogen, and oxygen in the heavy oils from the liquefaction of coal were removed by hydrocracking over sulphide catalysts but a final stage, similar to the present day reforming of naphtha, involved a multifunctional catalyst comprising a base metal (e.g. iron) supported on an acidified natural alumino-silicate (e.g. fluorided Fullers earth). Petroleum fractions were also upgraded to higher octane numbers by dehydroaromatization over oxides of the metals of Group VI. However, the need to upgrade the greater yields of naphthas coming from catalytic cracking led to the emergence of new reforming catalysts in which active metals were combined with oxides of various acid strengths; typical examples were nickel on silica-aluminas and platinum on halogenated γ-alumina, among which those containing precious metals soon became dominant, as in the 'Platforming' process which emerged in the late fifties. The potential of alloys to affect the activity and the selectivity of the metallic components of catalysts had been known since 1950, but economic pressures were insufficient to compel the adoption of even more complicated catalysts until the end of the past decade when platinum reforming-catalysts began to be displaced by 'bimetallic' catalysts especially by the Pt-Re pair ('Rheniforming'). Subsequently the empirical exploration of multimetallic catalysts by industry and the cognate fundamental research in the establishments of higher education increased in all the major industrialized countries, more especially in connection with the catalytic reforming of hydrocarbons.
The first generation of bimetallic reforming catalysts possessed increased stability and enabled operating pressures to be decreased while maintaining times on line comparable to those for Pt catalysts at higher temperatures. The second generation of bimetallic and multimetallic catalysts gives further gains in selectivity while improving still further stability at low pressures. Current publications show that much relevant research continues world-wide, aimed at improving the activity, selectivity, and life of catalysts, and the efficiency of the processes as reflected in longer periods of use before re-activation (cycle times), less expensive recycling of hydrogen and hydrocarbons, and more effective re- activation of spent catalysts.
This Report will treat only those catalysts that contain at least two metallic elements in zero-valency states and their properties in the catalysis of reactions of hydrocarbons, with emphasis on industrial findings; other material is introduced only insofar as it is essential to an understanding of the main theme. The earlier sections outline the background information both to provide a framework and to minimize later interpolations. Readers are referred to an earlier review in this series.
2 The Metals
The large number of patents making claims for enhanced activity and selectivity in catalysis by combinations of two or more metals stems from empirical investigations, which seldom advance unequivocal evidence about the nature of the states in which the metals exist in situ. Current research shows clearly that such information is elusive where the concentrations of the metals in the catalyst and the sizes of their aggregates are small; the usual situation when at least one component is a precious metal. In this Report therefore a somewhat arbitrary selection has been made of catalysts which, in use, can reasonably be supposed to contain atoms of not less than two metals in their zero-valency state dispersed, alloyed, or juxtaposed in clusters.
Metals with Reducible Oxides. — The ubiquity of small concentrations of oxidants, especially water, adventitious or intrinsic, requires that the metals, if they are to be formed and to remain unoxidized throughout the life of the catalyst, must have oxides which are reducible in operation. Metals having oxides which are more or less readily reduced are shown in Table 1; the intrinsic semiconductors ('semi-metals') of Groups VB and VIB are also included. Reducibility increases with atomic number across the A-section but decreases in the B-sections of the long periods. Other oxides cannot be reduced to elemental metal under the conditions usually prevailing during the reactions of hydrocarbons, although it must be noted that some degree of reduction of even the very stable aluminium oxide (in the form of γ-alumina) has been claimed. The elements marked with an asterisk have a relatively volatile higher oxide. The table provides only a rough indication; detailed statements about the phases persisting under steady-state conditions require direct observation. The complications introduced into investigations of multicomponent catalysts formed from complex mixtures are illustrated by the following examples in which the precursors of the metals are represented by oxides, in accord with common preparation procedures. The oxide compounds and solid solutions may not extend throughout the mass of the precursor; not infrequently they are confined to a thin skin on one of the irreducible oxides forming the support; however, the distribution can be controlled by choice of appropriate methods of preparation and a uniform distribution is usually preferred.
Oxide Precursors. — Reduction of a Binary Oxide XOa. The extent of reduction depends upon the free energy change at the reduction temperature and upon the ratio of the hydrogen and water partial pressures. It may also depend upon the particle sizes of the solids because decreasing the radius of curvature increases the chemical potential of the phase. Under some conditions oxides such as ZnO pass only to a lower valency, non-stoicheiometric state as in reaction (1), whereas
XOa(s) + δH2(g) [right arrow] XOa-δ( (s) + δH2(O(g) (1)
XOa-δ(s) + (a - δH2(g) [right arrow] X(s) + (a - δ) H2(O(g) (2)
XOa(s) + δH2(g) [right arrow] X(s) + aH2(O(g) (3)
others such as CuO go more directly and rapidly to the metal as in reaction (3). The temperature and the pressures of water and hydrogen also affect the rates of reduction and sintering of the solids.
Reduction of a Mixture of Binary Oxides. In reaction (4) ZOc is irreducible (A12O3), YOb essentially irreducible (ZnO), and no alloy of or Y is stable,
XOa(s) + YOb(s) + ZOc(s) + (a + δb) H2(g)
[right arrow] [X(s) + Yδ(adsorbed)](s) + Y1-δOb(1-δ) (s)
+ ZOc(s) + (a + δb)H2O(g) (4)
XOa(s) + YOb(s) + ZOc(s) + (a + b)H2O(g) [right arrow]
X(s) + Y(s) + ZOc(s) + (a + b)H2O (g) (5)
whereas in reaction (5) only ZOc is irreducible, but again no alloys are stable. There is also a second case of reaction (5) in which the alloy XY is stable. In reactions (6) and (7) YOb, ZOc, and YZOb+c are irreducible and the ternary oxide is stable, e.g. NiO and MgO with Cr2O3 giving Ni with MgCr2O4.
XOa(s) + YOb(s) + ZOc(s) + aH2 (g)
[right arrow] X(s) + YOb(s) + ZOc(s) + aH2O(g) (6)
[right arrow] X(s) + YZOb+c(s) + aH2O(g) (7)
Reduction of a Ternary Oxide. The oxide may be a solid solution (MgO, NiO; CuO, NiO) or a compound (NiFe2O4, MgFe2O4, NiAl2O4). As alumina is frequently used as a support the spinels provide suitable illustrations as shown in reactions (8) — (10). The composition of a catalyst in its steady state of reaction
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (8)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (9)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (10)
should be close to that at equilibrium for the ambient partial pressures because of the finite, sometimes fast rates of diffusion of the lattice components, the small intraparticle distances, and the long times for which industrial catalysts remain on-line.
As the design and optimization of catalytic systems lead to closely denned catalysts and restricted operating variables the phases present in situ depend mostly upon the free energies of formation of the alloys and of the chemisorbed complexes and compounds formed between the oxidized states of the metals and the supporting oxides.
The 'alloys' may be solid solutions, intermetallic compounds, multimetallic clusters, or mixtures of isolated atoms. The asymmetry of interfaces can be considered to extend to the next-nearest neighbours of surface atoms; thus the adsorption surface and the immediate sub-surface layers will behave somewhat differently from the interior, and very small crystals may differ from large ones even in lattice structure and equilibrium phases.
The increment in the free energy of very small particles implies that the surfaces of their supports may carry in addition isolated atoms and small groups of atoms which have migrated from the metal at points of high radius of curvature to give, as it were, a dilute atmosphere adsorbed by the support. A model of the sintering of supported metals requires just such migrating atoms ; some of the isolated atoms and groups may ionize by donation of electrons to the support.
In subsequent sections it will be seen that questions of the foregoing kind are raised with regard to almost all industrial multimetallic catalysts. Therefore the systems reported for reducing atmospheres will be those drawn from the elements shown in Table 1 unless the partial pressures of water are appreciable (as in steam reforming) when the bracketed elements, if present, will be considered to remain unreduced.
The artificiality of such distinctions can be measured by conclusions recently presented. Whereas the presence of Pt-Al alloys has been demonstrated in Pt/α-Al2O3 catalysts such as are used in the Degussa process for the manufacture of hydrogen cyanide from methane and ammonia at temperatures ~ 1300 °C, their presence at lower temperatures has not hitherto been seriously suspected. New evidence, based on the conversion of hexane and the chemi-sorption of hydrogen by 0.8 wt % Pt on γ-Al2 O3 and n.m.r. measurements of the Pt Knight shift (5 wt % Pt on γ-Al2O3), suggests that treatment with hydrogen at 550 °C or 650 °C produces alloys between the highly dispersed platinum particles and aluminium from the support. The results are of wide implication as temperatures close to these are used in catalytic reforming and dehydrogenation.
3 Relevant Properties of the Metallic Elements
To make headway in comprehending the mass of empirical data it is necessary to recall some salient characteristics of the metals as active phases, and of their compounds, solid, liquid, and gaseous, which may form under reaction conditions.
Specific Area. — In the absence of mass transfer limitations the activity of unit volume of a phase is proportional to its accessible surface area. The particle size of the metal is therefore made and maintained as small as is compatible with other constraints; the higher the melting point of the metal (Table 1) the more easily this can be achieved, other things being equal. Especially noteworthy are the high melting points of the platinum metals and of rhenium and the catastrophic decrease to the right of Group VIII. The depression of the melting points of subgroup A metals by solid solution of subgroup B elements is correspondingly marked. As there exist rough correlations between the cohesive energy, surface energy, and melting points of the metals it is apparent that the Gibbs surface excess of the B-elements in A-B alloys will be considerable and the surfaces more mobile than might be expected from the average composition.
The melting points and volatilities of oxide and halide precursors or of halides and hydrides which may be formed during reduction, reaction, or reactivation are relevant because these too provide an indication of a mobility which can be either advantageous or disadvantageous according to circumstances. The low melting points of some of the oxides and chlorides, and their appreciable volatilities may be useful in reactivation (see later) or deleterious as is suggested by the accelerated surface diffusion of copper-carrying chemisorbed chlorine.
Specific Activity. — The characteristic activities of metals in the reactions of hydrocarbons derive from the facility with which they break C — C', C — H, and H — H bonds, and their relative efficacies from the different ratios of these rates to those of rearrangement and desorption. The catalytic properties are evident in isotope exchange, hydrogenation, dehydrogenation, hydrogenolysis, cracking, and isomerization (both of double bonds and skeletons) under conditions which in industry fall broadly in the ranges 25 — 1100 °C and 0.1 — 500 bar, depending upon the thermodynamics and the kinetics of the desired reactions.
The activity per unit area of the metals of Table 1 often shows a maximum in Group VIII in each long period. The fall in activity at Group IB for the typical kinds of activity is much sharper and deeper than that to the left at Group VIIA but because of a general tendency towards an inverse relationship between activity and selectivity the latter may increase as activity decreases. The addition of less active elements, in particular from the B subgroups, to the more active transitional metals may then result in selective inhibition, or ultimately to poisoning. The amount and the stability of the precursor(s) of a relatively inactive element of a B subgroup must be carefully controlled if flooding of the principal active metal (e.g. Pt) by a selective, low-melting inhibitor (e.g. Sn) is to be avoided. No doubt this explains the insistence in many patents upon the presence of the modifier in only high valency states (e.g. GeII, GeIV) and its introduction by special procedures. Some possibilities for the generation and migration of modifiers are alluded to, see pp. 6 and 7.
Activity also changes regularly within the subgroups. Hydrogenation activity in Group VIII generally increases with atomic number in each vertical triad but may pass through a maximum, as at rhodium (among Co, Rh, and Ir) for the saturation of ethylene." Rhodium is more active than iridium, and palladium more active than platinum for the isomerization of olefins at low pressures. Multimetallic catalysts therefore offer opportunities for the juxtaposition of different activities and selectivities, and for the development of synergy by exposing different metals, alloys, metal areas, and interfaces. The metals are themselves multifunctional in the strictly phenomenological sense because each effects more than one kind of catalysis (e.g. exchange, isomerization, and hydrogenation) albeit these may be related at a deeper level.
Whereas the elements of the B subgroups may be present in chemisorbed, alloyed, or clustered states, the inhibition by sulphur, a recognized poison and selective inhibitor, is more often due to chemisorbed species; it may be introduced as a residual anion during the catalyst preparation or into the reactants as a volatile inorganic or organic compound (H2S, CS2, mercaptans, etc.).
All catalysts in contact with hydrocarbons acquire, more or less rapidly, deposits of carbonaceous material on their surfaces. The deposits may be oligomers, polymers, species of lower H/C ratio down to 'graphitic' carbon or 'coke' and surface or bulk carbide. Under adverse conditions the catalyst may become encapsulated and poisoned; less seriously or even advantageously it may be only inhibited selectively. Such 'carbon' can obscure the patterns of activity noted above whereas, apparently, carbide may activate or deactivate depending upon the metal and the reaction. As will be seen, a major advantage of the new multimetallic catalysts is the control they offer over catalyst ageing caused by carbon deposition.
