Selectivity of Metals and Alloys in Hydrocarbon Reactions BY MARTIJN W. VOGELZANG, MARIAN J. P. BOTMAN AND VLADIMIR PONEC Gorlaeus Laboratoria, Rij ksuniversiteit Leiden, P.O. Box 9502, 2300 RA Leiden, The Netherlands Receiued 1 I th May, 198 1 A classification of selectivity phenomena and factors determining the selectivities is presented in brief. The alloying effects, as established, are also briefly reviewed. It is shown that with neo- hexane it is possible to test the selectivity for various reactions as well as the preference of metals and alloys for the formation of particular chemisorption complexes. The metals Pt, Pd, Ir and Ni as well as the alloys Ni-Cu, Pt-Au and Pt-Ag have been tested by neohexane react ions with hydrogen. It is shown in this paper that alloying with a Group Ib metal causes in most cases a shift from hydrogenolysis to isomerisation and, whereas aj3 complexes are preferred by a pure metal, an ay complex formation is relatively promoted by alloying.Carbon seems to have a similar effect. Several recent review ~ a p e r s l - ~ and Proceedings of conferences4p5 show that a certain progress has been achieved in studies on the selectivities of metals. Some factors determining the selectivity have been identified, or at least their possible role has been indicated. To summarize the present state of affairs and to place this paper into the context of other selectivity studies it is convenient to start with the following classification. Following Bond,6 one can conveniently distinguish three types of selectivities : I.Selectivity SI in the reactions of mixtures (e.g. simultaneous hydrogenation of various alkyl benzenes) or of molecules with several functional groups of comparable reactivities (e.g. hydrogenation of crotonaldehyde) (2). 11. Selectivity SII in consecutive reactions (e.g. C2H2 -+ C2H4 --f C2H6 or C6H6 --f C6H1, -+ C6HI2) (++). 111. Selectivity SIII in parallel reactions (skeletal reactions of hydrocarbons, Within these groups catalysts may also reveal a stereo-selectivity for certain products or starting compounds. Such a separation of various selectivities sometimes appears to be only an abstrac- tion; in reality the various types of selectivities have to operate simultaneously. For example, the hydrogenation of acetylene on Pd is probably a reaction governed by SII selectivity which in this case is determined by thermodynamic factors-the heat of adsorption of reaction components.6 On Ir, in contrast to Pd, a proportion of the ethane is most probably formed by a parallel pathway where the intermediates do not leave the surface before ethane is formed.Then this is rather a case of SIII type selectivity, whereby the degree of dehydrogenation upon adsorption of C2H, is likely to play a decisive ro1e.1*2 The literature on selectivity problems l-’ mentions the following factors as being responsible for the selectivity of metals. (1) Chemisorption bond strength (heat of adsorption). The importance of this factor is particularly well documented for reactions under SI and SII.6 (2) Formation of multiple bonds with the surface.’ (3) Degree of dissociation of the bonds in chemisorbed species.The importance of this oxidations of olefins, various reactions of alcohols, amines, etc.) (,). f34 SELECTIVITY OF METALS factor has been shown for some reactions under SII and SIII.1*2 (4) Number of bonds formed with the surface and the number of metal atoms involved upon chemisorption of reaction components (" ensemble size ") which factor is closely related to factor (3) and probably operating in all three types of selecti~ity.l-~ The last two factors appeared to be the key factors in determining the selectivity in the skeletal (reforming) reactions of hydrocarbons also studied in this paper. How can the selectivity be influenced by alloying? Certainly not to the same extent in the various reactions of different types (1-111 above).The electronic structure of metals changes only moderately on alloying [see ref. (1)-(5) for a review] and so does the heat of adsorption.* Where the thermodynamic factors are most essential (as studied in the SII reactions), not very much can be expected from alloying with respect to the selectivity changes. In this case an application of gas additives or of multi- component catalysts other than alloys may be a more useful tool in influencing the selectivity.8 On the other hand, alloying is a suitable way to modify the selectivity in reactions where factors (3) and (4) are most important. Namely, by alloying the size (alloy : active/inactive metal) and the composition (two active metals) the nature of the ensembles of active sites can be easily be varied.The first effect, the variation in ensemble size, is also the dominating effect with alloys (Pt-Ag, Pt-Au, Pd-Ag, Ni-Cu, up to ca. 60% Cu) studied in this Cu-rich Ni alloys may be an example of the second effect, variation in ensemble composition (whatever the function of Cu may be). All f a c t ~ r s l - ~ mentioned above are in one way or another related to the electronic structure of metals and they may also be influenced by the particle size of the metals (and alloys), since they are more or less all influenced by the coordination of metal atoms. For example, the selectivity in formation of 3-carbon and 5-carbon inter- mediates of hydrocarbon isomerisation illustrates this ~taternent,~'~~ being very sen- sitive to both alloying and particle size. Various reactions reveal a different sensitivity for alloying and for diminishing the ensemble The very " sensitive " reactions are e.g., methanation, Fischer- Tropsch synthesis, skeletal reactions of hydrocarbons, ether formation from alcohols, etc.The " insensitive " ones are exchange reactions with deuterium and hydro/ dehydrogenations on the C-H, C-0 and most probably also on the N-H, S-H, etc. bonds. In work described in this paper a few " sensitive " reactions have been studied. The above-mentioned division of alloying effects reveals a great similarity with Boudart's l3 classification of structure-sensitive and -insensitive reactions, although the parallelism does not hold for all the reactions. Cyclopropane hydrogenolysis and methanation are examples where this parallelism does not hold.14 The alloying effect may also be more subtle than those already mentioned.The data on Pt alloys reveal that some reactions, such as isomerisation and hydrogenolysis, may run in parallel via different intermediates and sometimes with very different rates but to the same Moreover, while hydrogenolysis by a 2C-ap-mechanismf. is apparently the most sensitive reaction for the ensemble size, isomerisation is less sensitive and within isomerisation one of the pathways, namely that employing a 3C- ay-intermediate, is more sensitive than is the route using a 5C-intermediate.9Jo In this paper we intend to demonstrate how the preference of metals for a certain overall * Even such a sensitive tool as i.r.absorption spectra of CO reveals only very modest changes in the electronic properties of adsorption sites due to alloying; the main effect of alloying on the fre- quency of the i.r. CO band is simply due to dilution of a CO layer (as has recently been shown by Toolenaar and Stoop from our laboratory) by not coupling dipoles (different frequency!) residing on the atoms of the second alloy component. t This notation means two carbon atoms in neighbouring (a, /?I positions are involved in the reaction in que~tion.'~M. W. VOGELZANG, M. J . P . BOTMAN A N D V . PONEC 35 reaction (hydrogenolysis, isomerisation) as well as for certain intermediates (2C-@, 3C-ay) can be varied by alloying, particle size and carbon deposition. A suitable molecule to test the preference of metals and alloys for complexes involving either two or three C atoms is neohexane 1 6 9 1 7 (2,2-dimethylbutane) : C6 I c1-c2-c3-c4. I CS Initial tell us product distributions of this reaction (conversions must be kept under 1-2%) immediately how the molecule is attached to the surface upon adsorption and on running the reaction (see Experimental section for details).In addition, the pro- duct patterns also reflect in the overall reaction network, the possible presence of multiple reactions during one single sojourn of molecules on the surface. Using neohexane reactions, such an analysis of the selectivity behaviour has been applied to some metals (Pt, Pd, Ni, Ir) and alloys in this paper. EXPERIMENTAL CATALYSTS The preparation and structural properties of Pt (and Pt-Cu) catalysts on Si02 have already been described in detai1.16-18 The Pd and Pd-Ag catalysts were prepared as follow^.'^ Pd (Drijfhout, Amsterdam) was dissolved in boiling HN03 (a droplet of HC1 is needed) and then mixed with AgN03 (Merck) in the required proportions.This solution was used for impregnation (in a slurry) of SiOz (Merck); the metal loading was 10 wt.%. Pd-Ag alloys were sintered overnight in hydrogen at 650 K and the alloying checked by X-ray diffraction. The Pt-Au and Pd-Ag alloys were described in our earlier paper.20 The same holds for Ni-Cu alloys prepared from mixed carbonates.2' The Ir catalysts were prepared in two forms: Ir in small particles on Si02 and Ir bulk metal. Ir/Si02 was prepared with 1% metal loading from (NH4)21rC16 (Drijfhout, Amsterdam) using Si02 (Merck) as a carrier, by drying a slurry of the impregnated carrier (average Ir particle size ca.1.5 nm). If neces- sary, this catalyst was diluted further (l/lO, 1/1000) by Si02 in order to be able to work at high temperatures. The same dilution technique was applied to the pure Ni catalysts. The Ir catalysts were also prepared by mechanical mixing of crystals of (NH4)21rC16 with inert SiOz (loading 1 wt.%) and reducing afterwards. This procedure led to Ir metal crystals of a larger size. RATE A N D SELECTIVITY MEASUREMENTS The reactions of neohexane (puriss. grade, Fluka, Switzerland) were followed in a tubular continuous flow fixed-bed reactor working under differential conditions already described.21 The neohexane partial pressure was ca.40 Torr and the H2/neohexane ratio was 18. The flow rate was 9 cm3 min- in most experiments. The product analysis (g.1.c.) and calcula- tion of the various parameters characterizing the selectivity were as in our previous papers.16-21 DATA EVALUATION The rates and selectivity parameters were defined in conformity with our previous papers.'6-21 The physical consequence of the appearance of various products is most easily derived when the molecules undergo only one reaction per single sojourn on the surface. This is the case with Pt, Pd, Ni and Ir at the lowest temperatures. Table 1 shows the products derived from complexes which are attached to the surface, always through two carbon atoms (@, ay, "7' complexes). These complexes involve either two (2C-@) or three (3C-ay,36 SELECTIVITY OF METALS TABLE 1 .-RELATION OF PRIMARY PRODUCTS TO STRUCTURE OF COMPLEXES attached as products of single-step conversions ~ isomerisat ion h ydrogenol ysis EY‘, 3 G C 2,3-dimethylbutane methane, 2-methylbutane 2-me t hylpen tane ethane, 2-methylpropane UY, 1c-5c 3 -met h y 1 pent ane methane, 2-met hylbutane U P , 3c-4c methane, neopentane examples of products of some multiple reactions in adsorbed state $3, my’, my’ repeated methane only my’, m y repeated methane > ethane > other products uy’ followed by UP (into 2,2-dimethylbutane much propane without tripod-like complex or ay, my’ repeated or propane, ethane, as the first step) ethane and methane 2-methylpropane, ethane methane 3C-ay’) carbon atoms and one or more metal atoms.However, it is in principle possible and also experimentally observed that in some cases molecules undergo a multiple reaction during one sojourn on the surface. For example, when propane and butane are formed as primary products (i.e. also formed at the lowest conversions and even at contact time z --f 0) this can be due either to consecutive multiple splitting of complexes bound to the surface through two carbon atoms, or to simultaneous or consecutive splitting of two bonds in tripod-like complexes. Similarly, when excessive methane is formed at low conversions, multiple (1C) splitting is evidently taking place during each sojourn of the molecules on the surface, etc. RESULTS Some results representative of the particular metals and their alloys are collected in table 2.It can be seen immediately that the selectivity of metals for isomerisation Siso follows the order Pt > Pd > Ir > Ni, which is also the order of the decreasing ratio 3C-(ay + ccy’)/2C-ap, characterizing the formation of various complexes. It may be further concluded that Siso is also sensitive to the particle size. In the form of very small particles, Pt and Ir catalyse increased hydrogenolysis and less isomerisation. From the metals studied by this paper Pd is most easily self-poisoned by the running reaction. With this metal it is also most clearly observable that self-poisoning increases Siso and the ay’/ay ratio; the same effect is achieved by addition of Ag. Self- poisoning is evidently due to carbon deposition and formation of highly dehydro- genated, firmly bound species; for the sake of simplicity we shall call it “ carbon ”.As already mentioned above, the results for Ir indicate a similar role for carbon. Small Ir particles, which are more difficult to poison, reveal almost pure hydrogenolysis, while in the form of large-particle powder Ir is more easily poisoned and reveals a rather high isomerisation selectivity: not as high as Pt, but nevertheless Ir reminds one of Pt (see table 2) more than of Ni. Also the results on Ni-Cu alloys supply support for the idea that “ carbon ” influences the selectivity in hydrocarbon reactions.M . W. VOGELZANG, M . J . P . BOTMAN A N D V . PONEC 37 TABLE 2.-TYPICAL PRODUCT DISTRIBUTIONS FOR METAL AND ALLOY CATALYSTS total metal types of adsorption complexes loading no.(%) catalyst T/K “Y’iso “Y’hydr “Y’tot “Yiso “Yhydr “Ytot statistical random (I contribution by complexes 0.43 0.43 0.14 metals 1 2 3 4 5 6 alloys 7 8 9 10 9 6 6 9 9 1 12 16 10 Pt/SiOz; impregnation 563 Pt/Si02 (Euro-cat) 503 same as 2 (stable performance) 523 Pd/SiO, (fresh catalyst) 610 same as 3 (after self- poisoning) 610 Ir/SiO, (diluted 1 : 7 by SiO,) 497 Ir( black)b/ SiO, (diluted 1 : 30 by SOz) Ir( black)/ SiOz (after self-poisoning) 623 Ni powder 493 same as 6 (diluted 1 : 14 by SiO,) 603 Pt (82.3%)/ Ag (17.7%) Si02 665 Pt (4%)/ Au (96%) Pd (75%)/ Ag (25%) Ni (65%)/ c u (35%) 598 SiOz 648 SiO, 613 663 carrier-free powder 604 0.52 0.16 0.68 0.22 0.08 0.30 0.02 0.37 0.25 0.62 0.17 0.18 0.35 0.03 0.28 0.23 0.51 0.14 0.22 0.36 0.13 0.33 0.02 0.35 0.13 0.51 0.64 0.02 0.50 0.02 0.52 0.19 0.27 0.46 0.02 0.06 0.06 0.94 0.04 0.56 0.60 0.03 0.04 0.07 0.23 0.13 0.36 0.49 0.11 0.06 0.17 0.26 0.09 0.09 0.12 0.12 0.76 extensive multiple reactions 0.94 0.01 0.95 0.04 0.01 0.05 0.98 (0.01 0.99 0.01 (0.01 0.02 0.46 0.01 0.47 0.25 0.25 0.50 0.01 0.01 0.49 0.50 0.03 0.01 0.04 0.38 a Calculated for the case that the number of C atoms in the respective positions is decisive; Particle size of Pt catalysts from the X-ray line broadening: the rest up to loo%, multiple reactions.no. 1, 5 f 0.5 nm, no. 2, 1.4 f 0.2 nm.38 SELECTIVITY OF METALS With regard to the effect of Ag or Au on Pt or of Ag on Pd, the conclusion is straightforward: addition of a Group Ib metal promotes isomerisation and ay' ad- sorption and correspondingly suppresses other modes of adsorption and other reac- tion pathways.We observe that multiple reactions (i.e. repeated or successive conversions in the adsorbed state) take place to a greater extent than with Pt, Pd and their alloys. However, the general trend observed many times in the past1-6*11*12 is again confirmed: by alloying the activity is quite strongly suppressed but not equally for all reactions. The selectivity in isomerisation increases from 0 for Ni to several percent for alloys between 50 and 60% Cu (fig. 1 and 2). The extent of isomerisation is low but easily measurable at this level. Fig. 3 shows that the changes in the overall selectivity as well as in other para- meters characterizing the catalytic behaviour of these alloys (see below) are accom- panied by changes in the apparent activation energy calculated from the temperature coefficient of the overall reaction rate under standard flow, pressure and gas composi- tion.With Ni-Cu alloys the situation is more complicated. at. % Cu FIG. 1.-Activity of Ni and Ni-Cu alloys as a function of alloy composition. Activity defined as rate under standard conditions, at 571 K, rate in arbitrary units. The existence of multiple reactions makes it more difficult to analyse the preference that Ni and Ni-Cu alloys (or Ir) have for the formation of various complexes at higher temperatures. Only at the lowest temperatures is the conclusion straightforward : Ni (and Ir) strongly prefer to split the molecule at the thermodynamically unfavour- able but evidently kinetically most favourable place between C(3) and C(4), most likely starting from the formation of 2C-olp type complexes.However, when Ni is self-poisoned or when Ni is alloyed with Cu, and measurable rates (still at very low overall conversions, a few percent with alloys and <2% with Pt and Pd) are achieved only at temperatures ca. 80 K higher than those at which pure Ni can be measured,M . W. VOGELZANG, M. J . P . BOTMAN A N D V . PONEC 39 - -\ \ 0 20 40 60 00 at. % Cu FIG. 2.-Selecti~ity'~J~ in hexane and neohexane reactions as a function of alloy composition. T = 573-603 K. n 200 2 150 -= r;l" f 100 ! 1 I I I 0 20 40 60 80 at. % Cu FIG. 3.-Apparent activation energy calculated from the temperature coefficient of the overall conversion as a function of alloy composition.product patterns clearly show that multiple conversion or adsorption takes place before the molecules leave the surface and we also observe that complexes other than ab are being formed. We attempted to characterize the catalytic behaviour under these conditions in the following way. First consider the data obtained with pure Ni for the lowest temperatures. In the low T limit almost equal amounts of methane and neopentane are formed, viz. 41% of each [see fig. 4(a)J. Comparable data for Ni at high T and for the various alloys show a similar low-T limit of equal concentrations of CH, and neo-C5, always around 42%. Let us define a parameter characterizing the deviations from this limit: [CH,] - 42 f= 42 - [neopentane]' If each missing neopentane molecule (due to multiple reactions at higher temperatures) supplied only one CH, and one isobutane molecule, f would be unity; the whole40 SELECTIVITY OF METALS multiple reaction would be under these conditions solely induced by a 3C-ay complex, which at these temperatures apparently reached a reactivity already comparable with that of the 2C-ap complexes.If, on the other hand, each molecule of neopentane which started to react beyond the adsorbed neo-C, stage were completely broken down to CH4, f would be 9 1. For example, if 10% of the neopentane molecules (42 -+ 32%) were all converted into methane,f = 2.4. If other molecules are formed (from T K FIG. 4.-Product patterns as a function of temperature. From left to right: (a) pure Ni measured at low temperatures; (b) pure Ni diluted by SiOz and self-poisoned by the running reactions (notice that methane + neopentane, i.e., multiple reactions are also running at the lowest conversions); ( c ) alloy of indicated composition (increase in methane < decrease in neopentane; this indicates that other than methane molecules are formed, i.e., the role of cry is larger here).For the sake of simplicity, experimental points are shown only for one case, neohexane) than methane and neopentane, then f < 1. In any case, the presence of extensive multiple reaction reveals that, at a certain stage and at least once, an ay type bonding occurs and, in particular, these complexes must be more abundant when We can also use the so-called multifission parameter Mf introduced earlier21 for the characterization of the catalytic behaviour of alloys.The conclusions are the same as withf, as can be seen immediately. Consider the case, first, that each step of the multiple reaction destroying adsorbed neopentane releases just one methane. This means that one may assume that the reaction goes like f < 1. neo-C, -+ neo-C5 + C1 --f isobutane + 2C1 --f propane + 3C1 -+ . . . employing a sequence of complexes like 2C-aP -+ 3C-ay’ -+ (either 2C-ap or 3C-a7 with isobutane) etc. . . . . Again, the mere existence of the multiple reaction in the adsorbed state indicates that the 3C-ay’ and 3C-ay complexes are being formed at a certain stage, in any case. A parameter Mf defined as 6 2 Ci(6 - i) M -- i = 2 - [CH,] (experimental) could be near to unity if the reaction were a pure stepwise splitting off of C1 units. If adsorbed neopentane were broken down to C, species, M , < 1 ; if molecules other than methane were formed by multiple reaction, Mf 9 1.Reactions leading to Mf > 1 like neo-C6 --+ isobutane and ethane, or isomerisation into 2-methylbutane followed by splitting into two propanes, etc., all comprise at least one formation of aM. W. VOGELZANG, M. J . P . BOTMAN AND V . PONEC 41 3C-cry complex. Fig. 5 shows that in the range from 0 to 60-70% Cu the parameter Mf increases; correspondingly f decreases, i.e. the role of ay complexes increases on alloying. At still higher concentrations of Cu, Mf decreases and f increases again. The increase in Mf with Cu% in the region 0-60% Cu has already been observed with hexane and methylcyclopentane.21 The decrease in Mf at Cu > 70% has already been observed with hexane.21 2 .o 0.5 V I I I 1 - 0 20 40 60 at.% Cu FIG. 5.-Fission parameters f and Mf (see text for definitions) as a function of alloy composition. The bars indicate the variations of parameters with temperature for a given alloy. These variations of Mf and f indicate an increasing role for a7 complexes in this region of concentration. DISCUSSION It is easy to see that metals differ in their behaviour in the hydrocarbon, or more specifically neohexane, reactions. It is tempting to correlate this with the position of the metal in the periodic system, but it is most difficult at the moment to identify definitely the reasons leading to the mentioned difference in the catalytic behaviour of metals.Let us postpone this question to the end of the discussion and start with some more obvious conclusions. Alloys of Ni and Cu reveal that alloying considerably suppresses the extent of multiple reactions and simultaneously the role of the my-type of bonding increases. The extent of multiple reactions is inter alia characterized by the divergence of the C, and neo-Cs lines in fig. 4. The analogous divergent lines, when compared with pure Ni, are shifted by alloying by ca. 80 K to higher temperatures. For pure Ni highly diluted by inert SiOz and self-poisoned by a running reaction, the divergent lines show a much wider gap at the low-temperature side. With pure Ni apparently even at the lowest measurable conversions the extent of multiple reactions is high, while it is low42 SELECTIVITY OF METALS on alloys (see fig.4). The data shown in fig. 4 also reveal another interesting feature: with increasing temperature, the contribution of the reaction leading to isobutane (3C-ccy) increases and it is always higher at comparable conversions than on pure Ni. This is another piece of supporting evidence for the conclusion drawn above, that suppression of the 2C-aP splitting by alloying gives more chance for a 3C-ay splitting to show up at temperatures where this becomes possible. The dramatic drop in activity and the selectivity shift towards isomerisation by alloying has already been explained by the role of the ensemble size.'-5 This ex- planation can be applied to the present results on neohexane as well.However, it is necessary to have an additional explanation for the fact that with hexane much higher isomerisation selectivities are reached than with neohexane. We believe that this difference is amongst other effects due to the extent of the carbon layer kept deposited on the surface under a steady-state reaction of, respectively, hexane or neohexane (see below for more support for this). We suggest that in the first case the extent of the carbon deposition is higher and so correspondingly is the value of Siso. Above a Cu concentration of ca. 60% Cu all trends in selectivity shifts (Siw, Mf) appear to turn back to the Ni-like behaviour, only the activity is much lower here, even considerably lower than with the alloys of 10-60% Cu.The catalysts with Cu > 60- 70% are also rather unstable and easily poisoned. Therefore, the points for Cu > 60- 70% were not trusted (see dotted line) in an earlier paper.21 However, we con- sistently find this behaviour with various reaction components and various alloys of Cu. Having now experience with Pt-Cu alloys in neohexane and other hydrocarbon reactions16*18 we are inclined to explain the behaviour of the catalysts in this range of percentage of Cu by assuming the presence of mixed Ni-Cu ensembles. They should isomerize less than the small isolated (by Cu and carbon) ensembles of Ni and thus favour splitting. They split hydrocarbons much more slowly than pure Ni ensembles, but they are active. The picture suggested is as follows.When ca. 10% Cu is added to Ni, the surface concentration of Ni drops to 10 & 5% Ni and varies only margi- nally between (say) 10 and 70% bulk CU.'~ However, Ni atoms in the surface form part of the three-dimensional clusters 1~23-26 decreasing in size with percentage of Cu so that the clusters project smaller and smaller two-dimensional ensembles into the surface. In the above-mentioned region of Cu concentration fewer and fewer Ni clusters make contact with each other, which leads to a decreasing ferromagnetic saturation moment of the Ni-Cu alloy^.^^-^^ At around 60% Cu the alloys become paramagnetic and remain so with further dilution. We suggest that somewhere between 60 and 80% Cu the frequency (and activity) of bigger pure Ni ensembles in the surface required for isomerization and hydrogenolysis are made negligible by that dilution and the mixed ensembles get a chance to manifest their own role in an overall reaction.Between 0 and 60-70% Cu in bulk the average ensemble size of Ni decreases, and isomerization and the role of 3C-ay complexes increase. The transition from a kinetically easier reaction (terminal ZC-oc/3 splitting) to reactions requiring more activation (internal 3C-ccy' splitting of the weakest bond) is reflected in an increasing activation energy, as shown in fig. 3. A similar picture apparently holds for other hydrocarbons. 21 Small particles of metals, as a rule less self-poisoned by running hydrocarbon reaction^,^' reveal a lower isomerization and higher hydrogenolysis selectivity at low temperatures than the same metals in a bulk massive form.Carbon deposited on the surface evidently increases the isomerization selectivity and decreases the selectivity in hydrogenolysis of neohexane. (Note that with this molecule no dehydrocyclization can occur in one reaction step: with hexane2' this is possible.) Carbon also causesM. W. VOGELZANG, M. J . P . BOTMAN AND V . PONEC 43 shifts in the preference for the formation of various complexes from neohexane and it is interesting to see that this is a quite general phenomenon.” Moreover, in a very suggestive way the effects of carbon remind one of the effects of alloying transition metals with a Group Ib metal. What could be the reason? It is known from the i.r. spectra of adsorbed CO that alloying suppresses delocalized binding of carbon (CO) with several transition metal atoms in the s ~ r f a c e ~ ~ - ~ ~ and favours binding to indivi- dual surface atoms.In other words, alloying with Group Ib metals suppresses bind- ing in the multi-coordinated “ valley ” positions and favours the “ summit ” position. Deposition of carbon30 or on Ni (or oxygen on Pt or Rh32) causes the same effect as alloying with Cu. This is summarized by fig. 6 . deactivation the valley position hydrogenolysis -+ isomerisation (dehydrocyclisation) ++ XY activity in multisite reactions + FIG. 6.-Deactivation of the “ valley ” positions leads to the indicated changes by alloying or de- positing carbon or sulphur. The results which we have just mentioned as well as those in table 2 led us to the formulation of the following conclusions.(a) If the structure, state (“ C ”) and composition (alloys) of the surface allow it, then with a molecule like neohexane, Ni and Ir always prefer the 2C-@ complex formation, and splitting of the C-C bond is easier when a valley position is available among Ni or Ir atoms. Alloying with a Group Ib metal invalidates the valley position and in this way promotes the reactions which occur (also) on top of the surface atoms. (b) Pd and Pt easily covered by carbon under the reaction conditions reveal much less of the 2C-ap splitting, although the formation of a/3 complexes as seen at a low temperature (100 “C or more below the temperature of measurable hydrogenolysis) by the exchange reaction with deuterium is possible with these metals.Therefore, the preference of Pd and Pt for 3C-ay complexes over 2C-@ complexes at higher tempera- tures is partially, or even mainly, due to the carbon on the surface of these metals. It may be expected that carbon blocks the sites, which are good for the q? splitting. We identify these as the “ valley ’’ positions. Pd and Pt also show the lowest activity in hydrogenolytic splitting (of all metals) and in other reactions, as pointed out by This fact is additional support for the last statement.44 SELECTIVITY OF METALS (c) Ir shows a high preferential 2C-@ activity with neohexane, but at the same time its Mf is high and with hexane it is still higher at low temperatures (Mf = 7.8-4.9 at 425-525 With this metal 2C-ccp and 3C-ccy complexes most probably are not sharply differentiated as far as the requirements with regard to the activation are concerned.V. Ponec, in Electronic Structure and Reactivity of Metal Surfaces, ed. E. G. Derouane and A. A. Lucas (Plenum Press, New York, 1976), p. 537. ’ V. Ponec, in Progress in Surface and Membrane Science, ed. D. A. Cadenhead and J. F. Daniel (Academic Press, New York, 1979), vol. 13, p. 2. J. K. A. Clarke and J. J. Rooney, Adv. Catal., 1976, 25, 125. Chemistry and Chemical Engineering of Catalytic Processes, Proc. NATO Adv. Studies Inst., ed. R. Prins and G. C. A. Schuit (Sijthoff and Noordhoff, Amsterdam, 1980), pp. 337-349. Proc. Meeting on Catalytic Reactions of Alkanes and Hydrogen on Metals, in Bull. SOC. Chim. Belg., 1979, 88. G. C. Bond, in Catalysis by Metals (Academic Press, London, 1962).C. Kemball, Catal. Reu., 1971, 5, 33. J. A. Don and J. J. F. Scholten, Faraday Discuss. Chem. SOC., 1981, 72, 145. J. M. Dartiques, A. Chambellan and F. G. Gault, J. Am. Chem. Soc., 1976,98, 856. V. Ponec, Int. J. Quantum Chem., 1977, 12, 1. V. Ponec, Surf. Sci., 1979, 80, 352. (The Chemical Society, London, 1976). lo H. C. de Jongste, V. Ponec and F. G. Gault, J. Catal., 1980, 63, 395. l3 M. Boudart, in Proc. 4th Int. Congr. Catal., ed. G. C. Bond, P. B. Wells and F. C. Tompkins l4 J. W. E. Coenen, W. M. T. M. Schats and R. Z. C. van Meerten, in ref. (3, p. 435. l5 H. C. de Jongste and V. Ponec, in ref. (4), p. 337 and ref, (9, p. 453. l6 M. J. P. Botman, H. C. de Jongste and V. Ponec, J. Catal., 1981, 68, 9. I7 V. Ponec, in The Chemical Physics of Solid Surfaces and Heterogeneous Catalysis, ed. D. A. I8 H. C. de Jongste, F. J. Kuijers and V. Ponec, in Proc. 6th Int. Congr. Catal. ed. G. C. Bond, l9 E. K. Poels, MSc Thesis (Rijksuniversiteit Leiden, 1979). ‘O H. C, de Jongste, F. J. Kuijers and V. Ponec, in Scientific Basis of the Preparation of Catalysts, ed. B. Delmon, P. A. Jacobs and G. Poncelet (Elsevier, Amsterdam, 1976), p. 207. V. Ponec and W. M. H. Sachtler, in Proc. 5th Int. Congr. Catal., ed. J. Hightower (Miami Beach, 1972), vol. 1, p. 645. King (Elsevier, Amsterdam, 1982), vol. 4, to be published. P. B. Wells and F. C. Tompkins (The Chemical Society, London, 1977), vol. 2, p. 915. ’’ F. J. Kuijers and V. Ponec, Surf. Sci., 1977, 68, 294. 23 J. Vrijen, Thesis (Rijksuniversiteit Utrecht, 1977). 24 J. P. Perrier, B. Tissier and R. Tournier, Phys. Rev. Lett., 1970, 24, 313. 25 C. G. Robbins, H. Claus and P. A. Beck, Phys. Rev. Lett., 1969, 22, 1307. 26 E. Vogt, Phys. Stat. Sol. (B), 1972, 50, 653. ’’ P. P. Lankhorst, H. C. de Jongste and V. Ponec, in Catalyst Deactivation, ed. B. Delmon and 28 Y. Soma-Noto and W. M. H. Sachtler, J. Catal., 1974, 32, 315. 29 Y. Soma-Noto and W. M. H. Sachtler, J. Catal., 1974, 34, 162. 30 W. L. van Dijk, J. A. Groenewegen and W. M. H. Sachtler, J. Catal., 1976, 45, 277. 31 P. T. Rewick and H. Wise, J. Phys. Chem., 1978, 82, 751; J. T. Yates and C. W. Garland, 32 E. K. Poels and F. C. J. M. Toolenaar, personal communication (Rijksuniversiteit Leiden, 33 G. C. Bond, in Mechanisms of Hydrocarbon Reactions, ed. F. Merta and D. Kallo (Elsevier, 34 C. T. J. Wreesman, MSc Thesis (Rijksuniversiteit Leiden, 1980). F. G, Froment (Elsevier, Amsterdam, 1980), p. 43. J. Phys. Chem., 1961,65, 617. 1981). Amsterdam, 1975), p. 49.