Metal-Catalysed Reactions of Hydrocarbons / 07-Hydrogenation of Alkenes and Related Processes

Figure 7.13. Isomerisation of 1-butene on Pd/Al2 O3 at 313 K181. Equilibrium amounts are indicated by +.

take place before desorption occurs. In line with this view, there is considerable exchange at C3 as well as at C2.176 It was however argued that the Horiuti-Polanyi mechanism could not account for all the observations, and that other processes involving dissociative chemisorption of the reacting butene were needed.

For supported or unsupported iron, cobalt and nickel catalysts made by the solvated metal atom route179 (Section 2.2), the values of ri/rh in the reaction of 1-butene with hydrogen vary greatly with preparative method; this broad-ranging but somewhat superficial study did not lead to any definitive explanations for the variations.

We must now consider how the butenes react with hydrogen and with deuterium on palladium catalysts.18,171,173,177,182−185 As with nickel there is very rapid isomerisation and extensive exchange at low temperatures,177,181 so that for example at 273 K, starting with 1-butene, the butenes attain their equilibrium proportions when only half of them have been hydrogenated181 (Figure 7.13). Under such conditions the butene contains a preponderance of the d0- and d1-species.177,181 However, unlike the situation with nickel, the activation energies for these processes are smaller than for addition, so they become less important as temperature increases.181 The opposite effect was found in the ethene-deuterium reaction on Pd/Al2O3. Values of Eh are notably higher than the norm for other noble metals of Groups 8 to 10 (viz. 35–45 kJ mol−1). Again the Z /E ratio for the 2-butenes arising from 1-butene is less than the thermodynamic ratio, being only 1.6 at 273 K, although this value is higher than that found with platinum. The isomerised 2-butenes are however extensively exchanged, to an extent that increases with temperature. There is another significant difference from nickel; microwave analysis shows that in exchanged 1-butene-d1 nearly all the label is at C1, and in Z -2-butene-d1 formed by isomerisation it is also mainly at C1.173 The ratio ri/rh shows particle-size

sensitivity; its value approaches 100 for particles of mean size 2.8 nm (Pd/α-Al2O3, 273 K),177 but falls to about unity for particles 4.5 nm in size.180

As well as the Horiuti-Polanyi mechanism, two others have been proposed to account for the features seen in the reactions of the butenes. (1) The vinylic dissociative mechanism supposes adsorption of the alkene by breaking a C––H bond at C1 or C2: such a mechanism was indeed suggested many years ago to explain the exchange of ethene with deuterium, and has been resurrected to explain some aspects of reactions on nickel176 and platinum.173 We shall meet it again shortly. (2) The allylic dissociative mechanism requires the loss of a hydrogen atom from C3, giving a delocalised π -bond over three carbon atoms (see Section 4.41). Such species are know to be favoured by palladium (and nickel), but the presence of gaseous hydrogen or deuterium is thought to prevent their formation,173 and reactions over palladium have been interpreted without their aid.181 This mechanism does not allow the direct E-Z isomerisation; this has to proceed through adsorbed 1-butene. We shall meet this mechanism again also. The unusual temperature coefficients of exchange and isomerisation of the butenes over palladium catalysts may be associated with change in the equilibrium between dissolved and surface hydrogen, the latter becoming more abundant as the β-hydride phase becomes unstable. In support of this view, hydrogen exchange is insignificant at low temperatures.

Rhodium catalysts also show quite peculiar behaviour.6,173,186 The value of ri/rh is very small at and below 273 K, but between 273 and 423 K it increases very rapidly, so that the products at low temperature resemble those given by platinum, while those at high temperature are like those that palladium gives; hydrogen exchange is however much more marked. Increasing temperature therefore favours alkene desorption over addition.

Ru/Al2O3 and Os/Al2O3 have also been investigated.108 They show striking similarities, both giving rapid isomerisation between 290 and 340 K, to extents that with the 2-butenes decrease slightly with increasing temperature. There is therefore a general resemblance to the character of palladium. At 523 K on Ru/SiO2,187 both 1-butene and Z -2-butene isomerise rapidly during their hydrogenation, but hydrogenolysis to lower alkanes and homologation to C5 alkenes occur to small extents. These two processes are linked, and involve (i) breaking of a butyl radical into a methylene group and a smaller alkyl radical; and (ii) insertion of the methylene into a M––C bond to give a larger alkyl radical. Mechanisms have been discussed in detail; homologation of 3,3-dimethyl-1-butene has also been observed on Ru/SiO2.188

While one has to admire the skill and hard work that has been applied in obtaining the results discussed in this section, it is also necessary to note that many of the conclusions are based on analysis of products obtained only under a single set of conditions (conversion, temperature, reactant ratio etc.); hydrogen exchange is often not monitored, and sometimes176,177 there are even no butane analyses given.

7.3.2. The Single Turnover Approach189–92

We have noted before that one of the principal goals of research in heterogeneous catalysis has been to pin down the identity of the active centre at which reaction occurs. For structure-sensitive reactions this has perhaps a definite physical identity, while for structure-insensitive reactions the problem has been to understand why no such specific site is needed, or why only homogeneous sites are available (Section 5.4). The enquiry demands some kind of conceptual model, and this has been provided by input from homogeneous catalysis and organometallic chemistry, from geometric models of small metal particles, and from an innovative method for performing reactions: this is the single turnover method, the virtues of which have been vigorously prosecuted by R. L. Augustine.

This method is based on the following reasoning. If a metal particle is coated with hydrogen atoms under defined conditions and a pulse of an alkene (usually a butene) in stoichiometric amount is passed over it, carried by an inert gas, the observed products will be those that have undergone only a single reaction step, i.e. a single turnover (STO), and thus will reflect directly on the number and nature of the surface sites. With 1-butene as reactant, there is observed some unreacted 1-butene, some Z- and E-2-butene formed at isomerisation sites and some n-butane formed at direct saturation sites: some butyl radicals remain, and are removed by a hydrogen pulse, giving butane that is formed at two-step saturation sites. The original state of the sample is then re-created. This procedure resembles somewhat the 1-pentene titration method58 for estimating the number of chemisorbed and spiltover hydrogen atoms on platinum catalysts, and indeed for a number of such catalysts the number of hydrogen atoms detected by the STO method correlates very well with that found by conventional selective chemisorption.

Identification of the sites at which these reactions occur is based on an imaginative scheme devised by Samuel Siegel,193,194 in which, inspired by a knowledge of the mechanisms of homogeneous catalytic reactions involving for example Wilkinson’s complex (PPh3)3RhICl, he assigned specific roles to atoms having a number x of coordinatively unsaturated positions. So for example atoms of type x M (x = 3 or 2) can react with a molecule of hydrogen to give two adsorbed atoms (processes 7.H and 7.I), and one of these can diffuse to an 1M atom at which

chemisorption cannot directly occur. This of course ignores the possibility that two 1M atoms will do the trick. Invoking then the Hartog-van Hardeveld models (Section 2.41), the 3M sites are reasonably denoted as apical (corner) atoms, 2M

as edge atoms, and 1M as planar atoms. Augustine has adopted this analysis, and has refined it by recognising two types of direct saturation site, designated 3 MI and 3 MR at which hydrogen atoms are adsorbed either strongly (i rreversibly) or weakly (r eversibly). Ledoux and colleagues have used a similar analysis.17

Space constraints do not allow a detailed account of how the observations are manipulated to give site densities; activity for hydrogenation and isomerisation is confined by definition to 3M, 3MH and 2M sites, 1M being assumed inactive. Examination of a number of Pt/SiO2 catalysts (including EUROPT-1) showed the fraction of saturation sites to vary between 27 and 83%, without showing the expected dependence on particle size, although this correlation was nicely established for a series of platinum catalyst having controlled-pore glass (CPG) as support. The problems of interpretation are however highlighted when results for platinum, rhodium and palladium are compared. Analysis of the first pulse of 1-butene showed products in about the same relative amounts as expected from reactions under corresponding conventional conditions, i.e. values of ri/rh decreased in the sequence Pd > Rh > Pt. The validity of the method depends critically on the assumption that only one type of reaction proceeds on a given site, and that therefore hydrogenation and isomerisation are independent processes. This is not what we have concluded from the extensive work reviewed in the last section, and it is hardly reasonable to suppose that the fraction of isomerisation sites should vary so much from one metal to another. The method has not been widely applied, and the occasional attempt to use it has not been especially fruitful.

A recurring theme throughout much of the work on reactions of hydrocarbons with hydrogen or deuterium has been the possible existence of two or more distinct forms of adsorbed hydrogen atom, distinguished by their strength of adsorption and reactivity.190,195 In the hydrogenation of 1-butene on platinum black, an attempt was made196 to identify the active state as classified by TPD spectroscopy by filling each state with a different isotope (and assuming they did not interconvert); the deuterium content of the resulting butane then pointed to the reactive state, which appeared to be the β-form, i.e. atoms adsorbed atop.

7.3.3. Isobutene

Isobutene is not nearly so interesting a molecule as the other butenes, as it can undergo only exchange or addition, although with a suitably carbon-labelled molecule the movement of the double bond could in principle be followed, as has been done with propene;36 such experiments have not however yet been made. Indeed there are very few studies on this molecule to report.18,197 It was hydrogenated on Pt(111) at 300 K with a PH/ PC ratio of 10 more slowly than 1-butene,175 with an activation energy of 49 kJ mol−1, the orders of reaction being 0.7 in hydrogen and -0.2 in isobutene. SFG spectra suggested198 that the reactive form at 295 K is the

π-state; tert-butyl groups were also seen, and were less reactive towards hydrogen than isobutyls; only small amounts of dehydrogenated species (e.g. isobutylidyne) were present.

Some old work18,19 showed that the product of its reaction with deuterium on Ni/kieselguhr at 196 K was almost entirely isobutane-d2, so presumably exchange, and movement of atoms between species, does not occur significantly under these conditions. On nickel and rhodium powders, and on Ir/SiO2, microwave spectroscopy has shown addition and exchange to give the mono-deutero species expected if primary and tertiary butyl radicals were involved, but on rhenium powder a π -butenylic species also appeared to contribute.36 Pd/CaAlO4 was reported to give rapid exchange with a random deuterium distribution, implying a

π-butenyl intermediate; Pt/Nb2O5 and Rh/SiO2 gave exchange in the methylene group via a vinylic species, while Ir/SiO2 gave little exchange of any kind.199 A comprehensive kinetic study200 using PtSn/SiO2 catalysts gave results that were successfully modelled by the Horiuti-Polanyi mechanism, a particular feature being the occurrence of both stepwise and multiple exchange of the isobutene, with a sharp cut-off after –d7.

7.3.4. Exchange Reactions between Alkenes

As was mentioned in Section 7.2.4, alkenes are able to undergo reactions between themselves in the apparent absence of hydrogen or deuterium by processes that must involve their dissociative chemisorption or intramolecular atom movement. Thus for example 1-butene is isomerised on a number of supported metals and on platinum black above 373 K.36 Intermolecular isotopic transfer was also introduced in Section 7.2.4, where work by Hirota and Naito was cited. Extensive studies have been made of exchange and isomerisation caused by the interaction of propene-d6 with various alkenes,36 especially 1-butene, the reaction of which has been followed over films of chromium, iron, nickel, rhodium, palladium, iridium and platinum.17,201 Exchange between ethene-d0 and ethene-d4 has been reported to occur rapidly on nickel/kieselguhr at 298 K202 (Z-E isomerisation of ethene- d2 was even faster), and on palladium single crystal surfaces,203 but no detailed studies have been reported. The vinylic dissociation mechanism was invoked in most cases, although allylic dissociation contributed over palladium. This work required careful microwave analysis of mono-deuterated products, and extended discussions of mechanism have been given. The relevance of the conclusions to the situation when molecular hydrogen or deuterium is present is however not quite clear.

Hydrogen atoms returning to the metal by reverse spillover can however initiate isomerisation between the butenes and isotopic exchange between labelled and unlabelled alkenes, after which the reaction is self-propagating. Since the amount of the initiator may be immeasurably small, it cannot necessarily be detected by

mass-balance measurements. The processes referred to above do not therefore necessarily proceed through vinylic intermediates.204

7.4.REACTIONS OF HIGHER ALKENES WITH HYDROGEN AND WITH DEUTERIUM205−207

The reactions of linear or branched alkenes containing more than four carbon atoms reveal no types of reaction not already met, but their lower volatility permits their study as liquids or in solution as well as in the vapour phase. Thus for example the relative isomerisation rate ri /rh of liquid 1-pentene over Pd/C at 290 K was independent of the conditions of agitation,208 showing there were no mass-transport effects attributable to the hydrocarbons. The presence of solvents was also without effect.6

The hydrogenation of 1-hexene in vapour phase on Pt/SiO2 catalysts was sizeinsensitive,209 and on EUROPT-1 (6.3% Pt/SiO2) at 603 K (quite a high temperature for these reactions) 1-hexene and the 2-hexenes all showed extensive doubleband migration, although equilibrium ratios were not attained:210 isomerisation was suppressed by increasing hydrogen pressure, so these results are in line with expectations based on the butenes at lower temperature. The use of alkene titration using 1-pentene to estimate chemisorbed hydrogen has already been noted.58 In the liquid phase, neither 1-pentene95,208 nor 1-hexene6,211 show much isomerisation on platinum at low conversions, but with the latter exchange is ten times faster than isomerisation. The competitive hydrogenation of 1-heptene and 1- and 2-decene on silicalite-coated Pt/TiO2 has also been investigated.32

At 378 K, exchange of 1-hexene occurs on Ni/SiO2, causing the value of M in the deuterohexane distribution (Figure 7.14A and B) to increase with conversion.212 On Raney nickel at 296 K with cyclohexane as solvent it rapidly isomerises, first to the 2-hexenes and afterwards to the 3-hexenes, although complete product separation was not achieved.213 Similar results were obtained using cyclohexane as the source of hydrogen. Extensive isomerisation is also shown with liquid 1- pentene on Pd/C, Rh/C and Rh/Al2O3, and with Ru/Al2O3, but not with Pt/C, Pt/Al2O3 or Ir/Al2O3, in keeping with what was found in the reaction of 1-butene on these metals.208,214 At 299 K, 3-methyl-1-butene isomerises on Pd/SiO2 to give the two 2-methyl-butene isomers in approximately their equilibrium amounts, the process being completed after about 40% addition has occurred.215 The equilibrium proportions at this temperature are 2-methyl-2-butene, 92.0%; 2-methyl-1-butene, 7.8%, 3-methyl-1-butene 0.2%.

Double-bond migration of terminal alkenes forms more of the Z-isomer than corresponds to thermodynamic equilibrium. A study of the exchange and isomerisation of 1-pentene-1,2-d2 on various types of nickel catalyst has shown that this is due to ‘crowding’ at the active centre, so that the pentyl radical on losing a

Figure 7.14. Distributions of deuterated products plotted semi-logarithmically.

(A)n-Hexanes from 1-hexene + D2 on nickel, PD / PC = 3.5.212

(B)The same, PD / PC = 70.212

(C)Cyclohexanes formed on Adams PtO2. 214

(D) and (E) Z - and E -decalins formed from 9,10 -octalin + D2 on Adams PtO2 .211

hydrogen atom preferentially adopts the form of 2-pentene that occupies the smaller volume. Catalysts in which the nickel atoms are widely separately by being in complexes give more of the E -isomer.99

The glyceride esters of the long-chain fatty acids are not strictly speaking ‘hydrocarbons’, and the catalytic chemistry and technology of fat hardening is therefore not treated in this work: however, the alkyl side-chains behave exactly as

if they were hydrocarbons, and the reactions of methyl oleate (where each chain contains one CC bond) with hydrogen and deuterium have been studied.216

We must remember that branched alkenes may contain centres of optical activity; an example is (−)3,7-dimethyl-1-octene, which when hydrogenated on platinum shows little racemisation, but on palladium this happens extensively, to an extent depending upon the form of catalyst and reaction conditions.205,217 Isomerisation of the double bond to the 2-position negates the optical activity, so that it may return to the terminal position in either the (+)- or (−) form. When tetra-substituted alkenes of the type RR CCCRR are hydrogenated, two centres of optical activity are created; the E -form gives the meso product, while the Z -form gives a racemic mixture.16

Gold has the undeserved reputation of being pretty useless as a catalyst, except for oxidising carbon monoxide. However, the specific activity of Au/SiO2 catalysts for 1-pentene hydrogenation in a large excess of hydrogen increased as the gold content (and hence perhaps particle size) was decreased; active catalysts were mauve in colour, but showed no double-bond migration.218

The susceptibility of the hydrogenation of alkenes to sulfur compounds has been reviewed;98the effect has been used as a way of estimating the active metal area.219

7.5. HYDROGENATION OF CYCLOALKENES

7.5.1. Cyclohexene

On metallic catalysts, cyclohexene can either be hydrogenated or dehydrogenated to benzene, depending on the temperature and hydrogen pressure;220,221 the first is favoured at low temperatures, but it can also disproportionate to give cyclohexane and benzene:222

It therefore participates in a somewhat complex set of reactions, which overlap and share common adsorbed intermediates. At near ambient temperatures in the presence of hydrogen, hydrogenation predominates, and the simplicity of product analysis due to the absence of competing reactions such as double-bond migration have made this reaction an attractive one to study.223,224 Its major drawback is however the speed with which activity is lost through formation of ‘carbonaceous residues’,62,220−225 which are probably just highly dehydrogenated forms of the reactant.220,226 These parasitic reactions are easier with higher alkenes than with ethene, because of the more favourable thermochemisty, and the formation of these

strongly-held species raises interesting but difficult questions about the role they may play in the target reaction.

We have already noted the suggestion made some years ago that reactions such as alkene hydrogenation might proceed on top of these species,43,62 or that they might at any rate act as bridges over which hydrogen atoms might migrate to the reactant adsorbed on the remaining free surface.227 Since they should occupy and inactivate the more reactive sites, or reside in places where their removal by hydrogenation is least easy, the few remaining sites might be fairly uniform in character, and this would account for the apparent structure insensitivity of this class of reaction.224 In support of this idea, it has been found that for a series of Pt/SiO2 catalysts, values of the TOF for cyclohexene hydrogenation vary much more when derived from initial rates than from the much lower final rates. The same conclusion was reached from UHV work on Pt(233), where at low pressure dehydrogenation on clean surface was structure-sensitive, but at high pressures, in the presence of a carbonaceous overlayer, hydrogenation was ‘insensitive’.221 Extensive measurements on supported nickel, palladium28 and platinum catalysts lend further support to the ‘insensitivity’ of the catalyst in the steady state. Turnover frequencies are also independent of size for rhodium particles over a very wide range,229 but for Ru/SiO2 catalysts they show a maximum at about 4 nm. The reaction of cyclohexene with deuterium gives the familiar semi-logarithmic distributions of deuterocyclohexane but there is no discernible break after the -d6 product (Figure 7.14C)212,214 showing that there is no barrier to exchange on both sides of the ring.

7.5.2. Other Cycloalkenes

Cycloalkenes other than cyclohexene do not have the same propensity to aromatise, so that the processes of addition and exchange can be followed without complications. The reaction of cyclopentene with deuterium has been studied using various partially deuterated solvents (CH3OD, CH3COOD etc.) to minimise dilution of the deuterium through exchange with the solvent.214 With platinum there is little alkene exchange, and there is a break in the distribution of the deuteroalkanes after cyclopentane-d5, which corresponds to the completion of one-side exchange: detailed understanding of the distribution however requires there to be at least three different types of site, varying in their hydrogen: deuterium content and each being responsible for a single aspect of the process. A similar break occurs after cycloheptane-d7 when cycloheptene is used, but not after cyclo-octane-d8 in the case of the C8 cycloalkene:214 this is because the larger ring is so flexible that through an eclipsed conformation the exchange can switch from one side of the ring to the other. These findings exactly parallel those obtained in the exchange of the cycloalkanes with deuterium, although the extent of exchange is much less because of the greater coverage of the surface by adsorbed hydrocarbon species.

The ring in cyclodecene is large enough to allow the existence of E - and Z -isomers having approximately equal stability, although with the E -isomer the ring stops access to one side of the double bond and this is therefore less reactive.230,231 Their interconversion in solvents has been followed over Pt/C and Pd/C catalysts.231 As with the n-butenes, isomerisation occurs freely with the latter but not the former; with carbon tetrachloride as solvent, addition is preferentially poisoned with Pd/C (not with Pt/C) due to partial chlorination of the surface. It then becomes clear that in this case the isomerised products formed in the reactions with deuterium contain mainly -d0 and -d1 molecules. When considering deuterium contents of isomerised alkenes, it must be remembered that the release of a single hydrogen atoms from an initially formed deuteroalkyl radical can start a chain reaction that will lead to many light isomerised molecules, if alkyl reversal and alkene desorption are easy, and providing the hydrocarbon species are close enough together for this to happen before the hydrogen atom leaves for the gas phase (see process 7.J).

Heats of hydrogenation decrease progressively as the size of the ring increases,232 presumably because the greater flexibility minimises the strain in the reactant (see Table 7.1).

7.5.3. Substituted Cycloalkenes: Stereochemical Factors189,217,233

Just as the employment of complex cyclic alkanes has proved beneficial in research on alkane-deuterium exchange – at least it has stimulated discussion if not the attainment of conclusions (Section 6.3) – so the hydrogenation of substituted cycloalkenes has served to reveal aspects of mechanism that lie hidden when simpler alkenes are used. The question rarely considered, however, is whether these features are also at work in the simple molecules, or whether they are just operating in the complex cases. Three types of molecules in particular have been used: (i) dialkylcycloalkenes where the double bond is within the ring (endo-), (ii) alkylmethylenecycloalkanes where it is adjacent to the ring (exo-), and (iii) fused-ring cycloalkenes. Names and structures of some of the compounds used are given in the accompanying Table 7.7.

The hydrogenation of these compounds has attracted much interest for the following reason: 1,2-dimethylcyclohexene (for example) is likely to chemisorb either in the π - or the di-σ form, with the two methyl groups pointing away from the