Olefin Metathesis and its Catalysis By R. J. Haines DEPARTMENT OF CHEMISTRY, UNIVERSITY OF CAPE TOWN, CAPE TOWN, REPUBLIC OF SOUTH AFRICA and G. J. Leigh SCHOOL OF MOLECULAR SCIENCES, UNIVERSITY OF SUS SEX, BRIOHTON, BN1 9QJ 1 Introduction The olefin metathesis reaction (1) is one of the most remarkable catalytic reactions discovered in recent years1* R’CH R’CH RICH =CHR* ” -k R*CH RICH=CHRS” + RICH A non-catalytic counterpart of this reaction was observed in 1931 by Schneider and Frohlich,2 who established that at 725°C propylene could be converted into ethylene and butene. However, it was not until the discovery of hetero- geneous and homogeneous catalysts, which can promote the reaction at much lower temperatures and minimize side-reactions, that the potential of the metathesis reaction could be realized. The reaction is essentially thermoneutral, involving just the making and breaking of carbon-carbon double bonds.Equilibrium can be reached from either side of reaction (l), and the distribution of products is then statistical. The thermal activation of this entropy-controlled reaction is symmetry forbidden according to the Woodward-Hoffman rules,a which is consistent with the high temperatures necessary for the reaction dis- covered by Schneider and Frohlich. The catalysis of the reaction was first reported by Banks and Bailey,l their discovery stemming from a study of the catalytic activity of activated molyb- denum hexacarbonyl supported on alumina. They found that over this system linear olefins of three to eight carbon atoms were readily converted into mixtures of homologues of lower and higher molecular weight, and in particular that propylene was converted into ethylene and but-2-ene according to equation (2).*This reaction has been variously called the ‘olefin disproportionation reactiony1 and the ’olefin dismutation reaction’* but because it involves the interchange of alkylidene radicals (transalkylidenation) we prefer the term ‘olefin metathesis’ originally proposed by Calder~n.~ R. L. Banks and G. C. Bailey, Znd. and Eng. Chem. (Product Res. and Development), 1964, 3, 170. a V. Schneider and P. K. Frohlich, Znd. and Eng. Chem., 1931,23, 1405. See, for example, R. Hoffmann and R. B. Woodward, Accounts Chem. Res., 1968,1, 17.C. P. C. Bradshaw, A. J. Howman, and L. Turner, J. Caralpis, 1967, 7, 269. 6 N. Calderon, H. Y, Chen. and K. W. Scott, Tetrahedron Letters, 1967, 3327. 155 Olefn Metathesis and its Catalysis 2 propylene + ethylene + but-2ene (2) They also observed the formation of cyclopropane and methylcyclopropane from ethylene, and this has still not been adequately explained. Temperatures of ca. 150 “Cand pressures of ca. 30 atmospheres were employed. Prior to Banks and Bailey’s discovery, the polymerization of cyclic olefins by both heterogeneous (e.g. molybdenum oxide supported on alumina)6 and homogeneous catalysts [e.g. tungsten(v1) chloride-triethylaluminium]7-9 had been reported, but it was not recognized at the time that the ring-opening implicit in the polymerization involved olefin metathesis.The proposed mechanism involved cleavage of a bond adjacent to the olefinic bond. The first homogeneous catalysis of olefin metathesis was reported by Calderon and co-workers in 1967.5J0 They used [WCh]-EtOH-EtAICl2 as a catalyst mixture, and were the first to recognize that olefin metathesis involves trans- alkylidenation, and that ring-opening polymerization employing molybdenum or tungsten systems as catalysts also involves this step.11 The object of this review is to describe the scope of the reaction, to enumerate the theories which have been advanced to explain its catalysis (with particular emphasis on the nature of the catalysts), to evaluate critically the available literature (there is undoubtedly much industrial research which has not been revealed), and to attempt to present a model consistent with all observations.We shall discuss homogeneous systems in more detail than heterogeneous systems. Several reviews and catalogues on, and related to, this topic have appeared in the literat~rel~-~~ but none adequately covers all the aspects dis- cussed in this review, and a number are unavailable to the average reader. H. S. Eleuterio, Ger. Offen., 1960, 1 072 811 (Chem. Abs., 1961, 55, 16 005 h).G. Natta, G. Dall’Asta, G. Mazzanti, and G. Motroni, Makromol. Chem., 1963, 69, 163.* G. Natta, G. Dall’Asta, and G. Mazzanti, Angew. Chem. Internat. Edn., 1964, 3, 723. O G. Natta, G. Dall’Asta, I.W. Bassi, and G. Carella, Makromol. Chem., 1966, 91, 87. lo N. Calderon, E. A. Ofstead, J. P. Ward, W. A. Judy, and K. W. Scott, J. Amer. Chem. SOC.,1968, 90, 4133. l1 K. W. Scott, N. Calderon, E. A. Ofstead, W. A. Judy, and J. P. Ward, Abstracts 155th NationalMeeting of the American Chemical Society, San Francisco, April 1968, paperL54; Advances in Chemistry Series, 1969, 91, 399. la G. C. Bailey, Catalysis Rev., 1969, 3, 37. la F. D. Mango, Adv. Catalysis, 1969, 20, 307. S. Yoshitomi, Sekiyu Gakkai Shi, 1970, 13, 92 (Chem. Abs., 1970, 73, 3386~); Sekiyu to Sekiyu Kagaku, 1971, 15, 89 (Chem. Abs., 1971, 75, 119 670g); Nenryo Kyokaishi, 1967, 46, 514 (Chem. Abs., 1968,68,29 168a). l6 J. Tsuji, Kagaku No Ryoiki, Zokan, 1970, 89, 169 (Chem.Abs., 1970,73, 35 770k). l6 C. Inoue and K. Hirota, Yuki Gosei Kagaku Kyokai Shi, 1970,28,744 (Chem. Abs., 1970, 73,98 312h). l7 M. L. Khidekel, A. D. Shebaldova, and I. V. Kalechits, Russ. Chem. Rev., 1971, 40, 669. l* F. D. Mango and J. H. Schachtschneider, ‘Transition Metals in Homogeneous Catalysis’, cd. G. N. Schrauzer, Marcel Deuer, New York, 1971, p. 223. loN. Calderon, J. Macromol. Sciencp, 1972, C7, 105. N. Calderon, Accounts Chem. Res,, 1972, 5, 127. a1 L. Marko, J. Hung. Chem. SOC.,1972,27,213. Ia D. J. Cardin, M. J. Doyle, and M. F. Lappert, Chem. SOC,Rev., 1973,2,99. O3 R. L. Banks, Fortschr. Chem. Forsch., 1972, 25, 39. a4 W. B. Hughes, Organometallic Chem. Synth., 1972, 1, 341, Haines and Leigh 2 Scope of the Reaction A.Acyclic Olefins.-(i) Mono-oZe$ns. Whereas acyclic unsaturated hydrocarbons containing internal olefinic bonds are generally inert towards homopolymeriza- tion involving addition across the double bond, such as Ziegler-Natta poly- merization,25 they participate almost as readily as alkenes containing terminal olefinic bonds in the olefin metathesis reaction. This reaction thus provides a valuable route to the synthesis of new alkenes from those which may be more readily available. Its potential for redistribution of olehic bonds in systems containing func- tional groups is considerable, although thus far little exploited. Isolated examples are the metathesis of methyl oleate employing [WCl6]-[Me4Sn] as catalyst to give octadec-9-ene and the dimethyl ester of octadec-9-enedioic acid,26 and the metathesis of acrylonitrile and propylene over a heterogeneous catalyst derived from ammonium tungstate, to give ethylene and l-~yanopropylene.~~ Some problems may arise using olefins of this type due to catalyst poisoning, for example, where halogenocarbons are used, but they should be surmountable by suitable choice of catalyst.Thus p-chloroviuylbenzene and pent-2-ene, and 5-bromopent-l-ene and pent-Zene, undergo metathesis over [WCh]-EtNC12 in benzene.28 The metathesis of simple alkenes, particularly employing hetero- geneous catalytic systems, has, in contrast, been extensively studied. For a comprehensive catalogue of these reactions the reader is referred to the review by Bailey,lZ as well as to those by Banks23 and Hughes.24 The rate of reaction of olefins in the metathesis reaction is sterically con- trolled, as demonstrated by the decrease along the series CH2= >RCH2CH= > R2CHCH= >RzC=.~O Substitution of vinylic hydrogens by chlorine has been shown to deactivate the double bond towards metathesis.20 The industrial potential of the olefh metathesis reaction is considerable and obvious.The first commercial plant to utilize it, using in particular the Phillips Triolefin process to convert propylene into polymerization-grade ethylene and high-purity butene, was in operation within three years of the first report of the catalysed olefin metathesis reaction.29 Numerous industrial processes have been patented in which olefin metathesis is an integral step,12J7 for instance, a new route to high-octane alkylate.The process involves the alkylation of ethylene and but-2-ene, obtained by disproportionation of propylene, to di-isopropyl and but-Zene alkylate.30 It has also been shown that the yield of ethylene from a naphtha cracker can be economically increased through the addition of a propylene met at hesis st age.31 There are numerous reviews on this topic: see, for example, C. E. H. Bawn and A. Ledwith, Quart. Rev., 1962,16, 361 ;J. Boor, Ind. and Eng. Chem. (Product Res. and Development), 1970, 9,437, and references therein. z6 P. B. van Dam, M. C. Mittelmeijer, and C. Boelhouwer,J.C.S. Chem. Comm., 1972, 1221. 87 G. Foster, Ger. Offen., 1971,2 063 150 (Chem.Abs., 1971,75,63 172b). a8 J. I. O'Hara and C. P. C. Bradshaw, B.P., 1972,l 283 348 (Chem. Abs., 1972,77,113 786u). as Anon., Chemical Week, July 23, 1966, p. 70. *O R. S. Logan and R. L. Banks, Oil Gas J., 1968, 66, 131; see also Hydrocarbon Process, 1968, 47, 135. *l R. E. Dixon, J. F. Hutto, R. T. Wilson, and R. L. Banks, Chem. Age, 1967, 2,49. Olefn Metathesis and its Catalysis The metathesis reaction has provided a valuable method for characterizing polymer structures. For example, the monomer sequence distribution in a number of styrene-butadiene copolymers was determined by treating the polymer with a high proportion of but-2-ene in the presence of a suitable catalyst. This degraded the polymer to low-molecular-weight species, which were analysed by g.l.~.~~ This method has also been used to determine the extent of double-bond migration during free-radical cross-linking of butadiene.33 (ii) Di-and tri-olefns.Acyclic unsaturated hydrocarbons containing more than one olefinic bond also undergo the olefin metathesis reaction, although yields may be somewhat low over heterogeneous catalysts owing to coke formation on the surface. They can react either intermolecularly or intramolecularly. Intermolecular metathesis is effected on treatment of buta-1,3-diene with a tungsten oxide-silica catalyst that has previously been treated with sodium carbonate, to give ethylene and cyclohexadiene.34 For simple metathesis, hexa- triene should be obtained as product, and this must have cyclized to cyclohexa- diene under the reaction conditions. Deca-195,9-triene is formed on metathesis of hexa-lY5-diene by [MO(NO)~(PP~~)~CI~]-[M~~N~CI~].~~However, octa-1,7- diene reacts intramolecularly on treatment with [Mo(NO)~(PP~)~C~~I-[M~~A~~-C13] to give cyclohexene as a major product.35 Other reported reactions involving dienes include metathesis of buta-1,3-diene with pr~pylene,~~ of buta-1,3-diene with isobutene,a4 and of ~enta-l+diene.~~ B.Cyclic Ole&.-Metathesis is not restricted to acyclic olefins. As mentioned above, early reports showed that polyakenamers of the general formula [-CH=CH(CH2)n-] (n = 2,3,5,6, or more) are produced by treatment of the appropriate cyclic mono-olefin with heterogeneous catalysts such as molyb-denum oxide supported on alumina or homogeneous catalytic systems such as [WCls]-AIEt3.6-g Although it was not appreciated at the time that the reaction involves olefin metathesis, it was realized that the polymerization involves ring opening and not an addition polymerization of the Ziegler-Natta type because the latter would have yielded polymers containing cyclic repeat units.Further evidence against an addition mechanism was that cyclic olefins in which the rings are not highly strained were found not to be polymerized by normal Ziegler- Natta catalysts. The polymers formed were assumed to be essentially linear, and a mechanism in which a carbon-rbon bond adjacent to the olefinic bond is cleaved was proposed.8 Calderon and co-workers subsequently made a detailed study of the poly- merization of cyclo-octene and cyclo-octa-1 ,S-diene using [WCleI-EtAICh as 32 L.Michajlov and H. J. Harwood, Amer. Chem. Soc., Div. Polym. Chem., Preprints, 1970, 11, 1197. 33 W. Ast and K. HLmmel, Kautschuk Gummi Kunststofe, 1971, 24, 220. 34 L. F. Heckelsberg, R. L. Banks, and G. C. Bailey, J. Catalysis, 1969, 13, 99. 36 E. A. Zuech, W. B. Hughes, D. H. Kubicek, and E. T. Kittleman, J. Amer. Chem. Soc., 1970, 92, 528. G. Doyle, Ger. Offen., 1971,2047 270 (Chem. Abs., 1971, 75, 5202a); J. Catalysis, 1973, 30,118. Haines and Leigh catalyst, and they established that the polymerization mixtures contained low- as well as high-molecular-weight species.11~37 By means of selective and fractional extraction they isolated the former, and by gas chromatography, mass spectro- metry, and n.m.r.spectroscopy showed that they were macrocycles and multi- plets of the parent compound or, in the case of the cyclo-c?cta-1,5-diene reaction, oligomers of formula (C~HI~)~C~H~.~~ They also showed that further treat- ment of the high-molecular-weight species with [WC16EEtAlC12 led to the formation of low-molecular-weight macrocycles. These results led to the con- clusion that the ring-opening polymerization of cyclic olefins by catalysts of the type [WC16]-Rz~Cb-z is a special case of the olefk metathesis reaction essen- tially involving macrocyclization of the parent compound (1) by consecutive metathesis of larger and larger ring systems as shown. Intramolecular metathesis leading to ring reduction will obviously occur concurrently, and this explains why low- as well as high-molecular-weight species are found at equilibrium.On this basis, cyclic olefins that can be considered as multiples of the same mono-olefb should afford the same equilibrium mixture, and this has been observed.37 Strong evidence for the involvement of olefin metathesis in ring-opening poly- merization has come from studies of the ozonolysis products of the random copolymer produced from cyclo-octene and [1 -14C]cyclopentene with a [WO-C14]-EtAlClz-benzoyl peroxide catalyst.38 The high-molecular-weight species formed in the product mixtures obtained by polymerization of cyclic olefins are suspected to be primarily linear and not macrocy~lic,~~ but their formation is readily explained in terms of trace amounts of acyclic olefins being present in the reaction mixture and participating in the polymerization.The polyalkenamers obtained by ring-opening polymerization of cyclic olefins range from amorphous elastomers to crystalline materials, depending on the structure of the repeat units and the configuration about the olefinic bonds.40 The crystallizability of these polymers in general is associated with their stereo- regularity. Significantly, polymerization of substituted unsaturated cyclic olefins of medium ring size by this method is a conventional route to perfectly alternating copolymers, provided that side-reactions do not occur during the polymerization process. For instance, ring-opening of 5-methylcyclo-octene by s7 N.Calderon, E. A. Ofstead, and W. A. Judy, J. Polymer Sci.(Polymer Chem.), 1967,5A, 2209. 38 G. Dall’Asta and G. Motroni, European Polymer J., 1971,7,707. 3s K.W. Scott, N. Calderon, E. A. Ofstead, W. A. Judy, and J. P. Ward,Rubber Chem. Technol., 1971,44, 1341. 40 N. Calderon and M. C. Morris, J. Polymer Sci.(Polymer Phys.), 1967, 5A, 1283. 159 6 Olefn Metathesis and its Catalysis [WCls)-EtAlC12 gives a polymer with the repeat unit (2), which is equivalent to the alternating copolymer of butadiene, ethylene, and propylene.37 Similarly, a polyalkenamer containing the repeat unit (3), and equivalent to the alternating [-(CH~-CH=CH--CH+(CHaCH=C--CHa)-]I de (3) copolymer of butadiene and isoprene, is obtained20 by ring opening of (4).It has been established that the yield of low-molecular-weight macrocycles in the ring-opening polymerization of olefins is considerably increased under certain reaction conditions, e.g. high dil~tions.~~s~~-~3 For instance, metathesis of cyclo-octene in dilute benzene solution by [WC16J-EtAlC12-EtOH gives a mixture which contains the three low-molecular-weight species (C8H14)n (n = 1, 2, or 3) in high yield.41 Ring opening of cyclo-olefins by metathesis thus provides a convenient method for the synthesis of macrocycles, which are useful intermediates in the preparation of perfume bases. Intramolecular metathesis of cyclic diolefins can lead to products not obtain- able by more conventional routes.For example, catenanes have been identified in mixtures obtained by catalytic metathesis of cyclododecene by [WClsI-EtAlCl~-EtOH.4,45 It has been proposed that these are formed by a 360" twist of the cyclic olefin prior to the metathesis reaction, as shown in Scheme 1. Although most alicyclic olefins possessing more than one double bond readily undergo ring-opening polymerization, they do so only if these bonds are not c0njugated.1~ Thus the metathesis of cyclopentadiene, cyclo-octa-l,3-diene, or cyclohepta-l,3,5-triene has not been effected. It is believed that this inertness results from catalyst deactivation. For instance, addition of small amounts of cyclo-octa-l,3-diene to a tungsten-aluminium-catalysed polymerization of 41 N.Calderon, U.S.P., 1969, 3 439 056 (Chem. Ah., 1969,71, 38 438c). 49 N. Calderon, US.P., 1969, 3 439 057 (Chem. Abs., 1969,71, 80 807~). 45 E. Wasserman, D. A. Ben-Efraim, and R. Wolovsky,J. Amer. Chem. Soc., 1968,90,3286. 44 R.Wolovsky, J. Amer. Chem. SOC.,1970, 92,2132. 46 D. A. Ben-Efraim, C. Batich, and E. Wasserman,J. Amer. Chem. SOC.,1970,92,2133. Haines and Leigh m Scheme 1 cyclo-octene drastically reduces the activity of the catalyst.19 For a comprehensive review on the polymerization of cyclic olefins, see the excellent article by Calder0n.1~ C. Metathesis of Cyclic Olefins with Acyclic Olefins.--Cross-metathesis of cyclic and acyclic olefins is a convenient method for the synthesis of polyene species.For instance, it has been shown that the interaction of ethylene with cyclopentene, cyclo-octene, or cyclohexene over [Mo(CO)6]-Al203 or cobalt molybdate on alumina gives hepta-1 ,6-dieneY deca-1 ,Pdiene, or octa-1 ,7-dieneY re~pectively.~~ Similarly, treatment of cyclo-octa-1,5-diene with ethylene in the presence of [Mo(NO)~(PP~~)~C~~I-[A~~M~~CI~]gives deca-lY5,9-triene.35 The acyclic triene C14H24, the tetraene ClQH32, and the pentaene C24H40, as well as the diene CSH16, have been isolated from the reaction of cyclopentene with ~ent-2-ene.~' D. Alkynes.-Metathesis of alkynes has also been observed, although the reaction has received far less attention than reactions involving alkenes. Pent- 2-yne has been converted into but-2-yne and hex-3-yne, employing tungsten oxide-silica as catalyst, and of the 44% of the initial pentyne undergoing reaction, 53% was converted into hexyne and butyne.48 Terminal alkynes, e.g.propyne, also participate in this type of reaction, but in all cases studied the selectivity is low.49 This is because these alkynes preferentially cyclotrimerize to yield derivatives of benzene under the conditions required for metathesis. I* G. C. Ray and D. L. Crain, Fr. P., 1968, 1 511 381 (Chem. Abs., 1969,70, 114 580s).J.-L. Herisson and Y.Chauvin, Mukromol. Chem., 1970, 141, 161. 4* F. Pennella, R. L. Banks, and G. C. Bailey, Chem. Comm., 1968, 1548. 49 J. A. Moulijn, H. J. Reitsma, and C. Boelhouwer, J. CuruZysis, 1972, 25,434.161 Olefin Metathesis and its Catalysis 3 Catalyst Systems and the Nature of the Catalytic Site Both heterogeneous and homogeneous systems have been employed as catalysts in the olefin metathesis rea~tion.l~e~~1~4 The former are normally composed of a ‘promoter’, such as molybdenum or tungsten oxide, and a refractory ‘support’ having a high surface area, such as alumina or silica. However, the functions of the components cannot be rigidly defined, and the terms ‘promoter’ and ‘support’ should be used with caution. The homogeneous catalysts are generally derived from a transition-metal complex, e.g. wCl6] and an organometak derivative or a Lewis acid such as EtAlCl2 or LiBu. A. Heterogeneous Catalysts.4f the heterogeneous systems thus far studied, those derived from oxides or carbonyls of molybdenum, tungsten, or rhenium exhibit the highest catalytic activity.l~~~-~~ The sulphides of molybdenum and tungsten and the oxides of iridium, lanthanum, niobium, osmium, ruthenium, rhodium, tantalum, tellurium, and tin have also been found to be effective promoters, but the catalytic activity of systems containing these species is con- siderably less.54-56 Various refractory materials have been employed as supports.These include magnesium silicate, magnesia-titania, alumina-titania, the oxides of metals such as zirconium and titanium, and the phosphates of aluminium and magnesium.54 Silica and, in particular, alumina have been found to be the most effective and are the most widely used.In fact, alumina has been observed to effect the metathesis of propene in the absence of any promoter, although its activity for such is low.57 Generally, heterogeneous catalyst systems require activation before use. Many of the metal oxide systems need to be heated for up to five hours in a stream of an inert gas, optionally containing dioxygen, to temperatures of 600 “C. It is believed that the high temperatures effect some chemical interaction between the promoter and the support, and also desorb polar material which may poison the catalyst surface. The activity of a number of catalysts based on metal oxides can be increased by modifying their preparation. For example, the alumina used as support in some systems has been pretreated with a strong inorganic or organic acid (hydrochloric or acetic acids) prior to impregnation with the oxide.58 Rhenium oxide has been sublimed directly on to the alumina support.5Q.60 Hydrochloric acid and chlorinated hydrocarbons which yield R. L.Banks, U.S.P., 1966, 3 261 879 (Chem. Abs., 1966, 65, 12 lose). 61 L. F. Heckelsberg, R. L. Banks, and G. C.Bailey, Ind. and Eng. Chem. (Product Res. and Development), 1968, 7, 29. British Petroleum Co. Ltd., Dutch P., 1966, 6 511 659 (Chem. Ah., 1966, 64, 19 408c). sa K. V. Williams and L. Turner, B. P., 1968, 1 116 243 (Chem. Abs., 1968, 69’29 085s).L. F. Heckelsberg, R. L. Banks, and G. C. Bailey, Ind. and Eng. Chem. (Product Res. and Development), 1969, 8, 259. 66 L. Turner and K.V. Williams, B. P., 1967, 1 096 200 (Chem. Abs., 1968, 68, 77 69111). 66 R. B. Regier, U.S.P., 1972, 3 652 703 (Chem. Abs., 1972, 76, 139 91Oj). G. V. Isagulyants and L. F. Rar, Zzvest. Akad. Nauk S.S.S.R., Ser. khim., 1969, 1362. 6* British Petroleum Co. Ltd., Dutch P., 1966, 6 610 196 (Chem. Abs., 1967, 67, 53 607t). 6s British Petroleum Co. Ltd., Dutch P., 1966, 6 605 329 (Chem. Abs., 1967, 66, 48 081s). 60 L. Turner and C. P. C. Bradshaw, B. P., 1968, I 103 976 (Chem. Abs., 1968,68,86 834~): 162 Haines and Leigh hydrogen chloride at the activation temperature also increase the catalytic activity of WOpSiO2 systems.61 The activity and selectivity of oxide-based catalysts can also be increased by controlled treatment after activation, with gases such as CO and H2,50151~62163 although prolonged treatment deactivates them.Certain reducing agents have also been observed to have a beneficial effect on the activity of heterogeneous catalytic systems. For instance, the conversion of pent-Zene by WO3-SiO2 is increased by the addition of tributylphosphine to the feed,64 and the activity of supported oxides of molybdenum, tungsten, rhodium, vanadium, niobium, or tantalum is increased by treating with tri- ethylaluminium.65 The catalysts derived from metal carbonyls normally require much lower activation temperatures (120-140 "C),and are deactivated by air.66 Thus they must be activated under a high vacuum or in an inert atmosphere. Kemball and co-workers have established, by i.r.spectroscopy, that during the activation of [MO(CO)6]-&03, 2 or 3 moles of carbon monoxide are evolved.66 However, it has been shown subsequently that Mo(C0)4 is not an active species, and probably the active sites are completely free of C06' and possibly oxidi~ed.~~.~~ The activity of [Mo(CO)6]-&03 is enhanced by treatment with halogeno- olefms before admission of reactanksg Allyls of molybdenum, tungsten, and rhenium are apparent exceptions in not requiring activation, since, when adsorbed on alumina, they are immedi- ately active catalysts. However, possibly they activate rapidly as metathesis com- mences.7O Heterogeneous catalysts can function over a wide range of temperatures, the optimum temperature depending both on the nature of the support and on the promoter employed.Catalysts containing molybdenum or tungsten oxide generally require temperatures that are higher by about 100 "C to effect metathe- sis than the corresponding catalysts derived from molybdenum or tungsten hexacarbonyl. A higher operating temperature by about 200 "C is also required for systems containing silica as support compared with those using alumina. Silica-supported catalysts have the advantage that by operating at high tempera- tures (ca. 400 "C51)they are considerably more resistant to poisoning by polar molecules. Catalysts derived from rhenium oxide and alumina are noted for F. Pennella, Belg. P., 1968, 713 187. Oa L. F. Heckelsberg, U.S.P., 1968, 3 365 513 (Chem. Abs., 1968, 68,61424s).63 E. A. Zuech Ger. Offen., 1971, 2 017 841 (Chem. Abs., 1971,74, 5300). O4 L. F. Heckelsberg, Belg. P., 1968, 713 185. 66 Shell Internationale Research Maatschaapij N.V., Dutch P., 1969, 6 814 835 (Chem.Abs., 1969, 71, 49 21 lz). 66 E. S. Davie, D. A. Whan, and C. Kemball, Chem. Comm., 1969, 1430;J. Catalysis, 1972, 24,272. R. F. Howe, D. E. Davidson, and D. A. Whan, J.C.S. Faraday I, 1972, 68, 2266. 8o D. A. Whan, M. Barber, and P. Swift, J.C.S. Chem. Comm., 1972, 198. E. S. Davie, D. A. man, and C. Kemball, Chem. Comm., 1971, 1202. 'O A. Morris, H. Thomas, and C. J. Attridge, Ger. Offen., 1972, 2 213 948 (Chem. Abs., 1973, 79, 65 769h); J. P. Candlin, A. H. Mawby, and H. Thomas, Ger. Offen., 1972, 2 213 947 (Chem. Abs., 1973, 79, 97 101e); I.C.I.Ltd., Fr. P., 1972, 2 120 509 (Chem. Ah., 1973,79, 115 104h). 163 OlefinMetathesis and its Catalysis their activity at reIatively low temperatures, as demonstrated by a 38 % conver-sion of a but-1-ene feed into ethylene and hexenes with a 95.5% selectivity at 25 "C, at atmospheric pressure and at a 1600 vol./vol. gas hourly space velocity.52 B.Homogeneous Catalysts.-The first homogeneous system reported to catalyse the ole& metathesis reaction was tungsten hexachloride-ethylaluminium dichlorideethan~l.~Equilibrium is attained within minutes for a W :A1 ratio of 1:4 and a W:olefin ratio of 1:lO OO0.lo Neither constituent of this catalytic mixture catalyses metathesis alone. Subsequent studies have revealed that a wide range of organometallic or hydridic derivatives of Main Groups I, 11, In, or IV may be employed as co-catalyst in conjunction with [WclS].These include LiR (R = Bun or RMgX (X = C1, R = Bun or n-C~H11;X = Br, R = Prn, Bun, or n-GH11; X = I, R = B~~),~~-75R2AICl,&AI (R = Et or Bui), EtAIC12,7s-78 SnR4 (R = Me or Bun),79 LalH4, and NaBK.80 There is, however, no firmdividing line between homogeneous and hetero- geneous catalysis. Thus [wCls]-EtAIc12 (a homogeneous catalyst) acts hetero- geneously when adsorbed on alumina.81 It was initially believed that the function of the organometallic species was to reduce [WclS] to tungsten tetrachloride since the [WCls]-LiBun system shows a maximumof activity for a Li:W ratio of 2:l. Complexation of the olefin would then give [WCl4(ole&)2], the actual catalyst, in which the olefins are bonded in cis-positions.This is supported by the fact that [WCla]-Et& exhibits maximum activity over only a narrow range of W :Almolar ratios, the optimum being ca. 0.54.6.With an excess of Et3Al the activity is decreased, possibly due to over- reduction, but it may be restored by treatment of the mixture with dioxygen.77 However, wc14 obtained by reduction of [WclS] with H2 at high temperatures or prepared in situ by treatment of [WclS] with reducing agents such as zinc, magnesium, or sodium amalgam is not a metathesis catalyst,20s80 although the addition of aluminium trichloride to wc14 so prepared gives a highly active catalyst.82 Significantly, [WclS] and [WBrs] also form active metathesis cata- 71 J.-L.Wang and H. R. Menapace, J. Org. Chem., 1968,33, 3794. 7a M.L.Khidekel, V. I. Marin, A. D. Shebaldova, T. A. Bolshinskova, and I. V. Kalechits, Izvest. Akad. Nauk S.S.S.R., Ser. khim., 1971,663. P. A. Raven and E. J. Wharton, Chem. andInd., 1972, 292. 74 T. Takagi, T. Hamaguchi, K. Fukuzumi, and M. Aoyamo, J.C.S. Chem. Comm., 1972,838. 76 V. I. Marin, A. Shebaldova, T. A. Bolshinskova, M. L. Khidekel, and I. V. Kalechits, Kinetika i Kataliz, 1973, 14, 613. 70 A. Uchida, Y. Mukai, Y.Hamano, and S. Matsudu, Znd. and Eng. Chem. (Product Res. and Development), 1971, 10, 369. Y. Uchida, M. Hidai, and T. Tatsumi, Bull. Chem. SOC.Japan, 1972,45, 1158. 7* N. Calderon and H.Y.Chen, B. P., 1968,1 125 529 (Chem. Abs., 1968,69, 105 851f). 7@ C.P.C. Bradshaw, B. P., 1970,1 208 068 (Chem. Ah., 1970,72, 12 047v). 80 J. Chatt, R. J. Haines, and G. J. Leigh, J.C.S. Chem. Comm., 1972, 1202; see also S. A. Matlin and R. G. Sarnmes, ibid., 1973, 174. C. P.C.Bradshaw and J. I. Blake, B. P., 1972,1266 340 (Chem.Abs., 1972,76,126 352d). D.M.Singleton, US,P., 1970, 3 530 196 (Gem, Abs., 1970,72, 89 7342). Haines and Leigh lysts with the non-reducing AlCh or AIBr320983984 but it is conceivable that olefins can insert into aluminium-halogen bonds similar to the well-known olefin-insertion into aluminium-hydrogen bonds,*5 forming aluminium chloro- alkyls, and it is unlikely that the final systems contain WV or WVI. The organometallic or hydridic species which interacts with the tungsten compound to form the catalyst may bond through chloride, hydride, or alkyl bridges and/or may function as a Lewis acid, giving rise to acid-base equilibria of the type: [WClz] + [AlCl3]+ [WCIz-lIf + [AlC14]-in the case of AlCl3, the removal of a chloride ion from the tungsten providing an additional vacant co-ordination site for the incoming olefin.20 The addition of MC13 or AlBr3 to [WC16]-2LiBun increases the catalytic activity of this system considerably, and this may also be explained by the aluminium halide acting as a Lewis acid and effecting rapid isomerization of trans-[WC14(olelin)~]to cis-[WCL(olefin)~] by the mechanism outlined in Scheme 2.86 AlClt A.ctrans-[ WCla(olefin)z] 4-AlC& cis-[ WC14(olefh~] Scheme 2 The ability of lithium alkyls, tin tetra-alkyls, lithium aluminium hydride, or sodium borohydride to produce active metathesis catalysts with [WClS] would imply that although acid-base equilibria of the type described above can lead to an increase in the rate of metathesis, it is not an essential part of the catalyst const it ution.The homogeneous catalyst mixture [WCle]--LiBu* appears to yield substantially the same product from the metathesis of but-2-ene throughout the range 60-130 "C, although the rate increases with temperature. Above 130 "C,the catalyst is deactivated or destroyed.87 Controlled treatment of [WClS] with protic solvents such as ethanol, phenol, water, or acetic acid prior to the addition of the organoaluminium compound increases the catalytic activity of systems [Wc16]-RzA1c~3-z.5~10~78~88The reaction of [WClS] with ethanol produces a stoicheiometric amount of HCl together with species such as [WC15(OEt)].20 It is therefore not suprising that compounds such as [WOc14], WOa, [ {o-(MeOzC)C6&0 }Web], w(oPh)6], 83 P.R. Marshall and B. J. Ridgewell, European PoZymer J., 1969, 5,29. 8p J.-L. Herisson, Y. Chauvin, N. H. Phung, and G. Lefebvre, Compt. rend., 1969, 269, C, 661. See, for example, R. Koster and P. Binger, Adv. Znorg.Radiochem., 1965, 7, 263. 86 J.-L. Wang, H. R. Menapace, and M. Brown, J. Catalysis, 1972, 26,455. J.-L. Wang and H. R. Menapace, J. Catalysis, 1973,28,300. D. Medema, W.Brunmayer-Schilt, and R. vaa Helden, B. P., 1970, 1 193 943 (Chem.Abs., 1970, 73, 76 637r). Olefin Metathesis and its Catalysis and [WO(OPh)4] are active metathesis catalysts in the presence of a suitable cocatalyst. 47~84~890 90 Group V donors reduce the activity of catalysts derived from [WCle]. Treat- ment of [WCh]-EtAICIz with pyridine inhibits metathesis, as also does tri- phenylphosphine (but less Pyridine reduces [WCle] to [WC14(pyridine)2],92 and the ligand may block possible co-ordination sites for the olefin. It may also be that the aluminium compound removes PPh3 more readily from [WC14- (PPh3)2] than it does pyridine from [WC14(pyridine)2].91 Direct comparison of the effects of the two ligands is not really justified, however, because although the reactions for both ligands employed a W:ligand ratio of 1:1, they used different W: A1 ratios.Furthermore, for W: A1 ratios greater than those employed above, [WCle(pyridine)z)-EtAIClz, as well as [WCh (CzH4(PPh2)2 )2)-EtAIC12, are effective met at hesis cat a1 yst s.93 Systems derived from high-valent molybdenum compounds are far less effective catalysts than those obtained from the tungsten analogues. For instance, [MoCI~I-E~~A~,~~[Mo(a~ac)3l-EtAl,~7 [M0Cl5]-PrnMgBr,~~ and [MoC15]-LiAlH4*O show very little or no activity, although the [MoC14]-AlBr382 system has been claimed to be effective. Catalytic activity for a number of molybdenum systems can be induced by treatment with nitric oxide prior to addition of an organoaluminium compound.This has been established in particular for [MOCI~L~] = PPh3, C5H5N, or (L PrnCN), [GH5Mo(C0)31], [MoC13(PhC02)2], [MoO2(acac)], [MoOCI~], and [MoClb] in the presence of ligands such as C5H5N, Ph3P0, Bu%PO, and OCtn3- PO.35 The enhancement of activity is ably demonstrated by the [MoCh(PhCO&I- [Me3Al2C13] system. Treatment of the molybdenum compound with nitric oxide for 0.5 h before the addition of the aluminium sesquichloride produces a catalyst which gives a 60% yield of oct-4-ene from pent-l-ene after 17 min. Without the NO treatment this catalyst effects only a 0.8% conversion of the pent-l-ene after 1 h.24 Interaction with nitric oxide should yield molybdenum nitrosyl species, and, consistent with this, derivatives of the type [Mo(N0)2L2X2] (X = C1, Br, or I etc.; L = PPh3, AsPh3, Ph3P0, C5H5N, or EtC5H4NY etc.) are highly effective catalysts in the presence of an organoaluminium halide.35e94 These systems have been studied in some detail, and a number of significant features have been noted.35 For instance, a wide range of Mo:Al ratios can be employed, but 1 :4-5 gives highest activity.Further, the catalyst requires a pre-formation period to reach maximum activity, this being 1 h in the case of [Mo(NO)Z(C~H~N)~C~~]-E~AICIZ.During this period the two nitrosyl stretching vibrations associated with [MO(NO)Z(C~H~N)ZCIZ] shift to higher frequency and eventually disappear.24~95 This is explained as initial complexation of two J.-L. Wang, M.Brown, and H. R. Menapace, Ger. Offen.,1972, 2 158 990 (Chem. Ah., 1972,77, 125 933c). 90 H. Knoche, Ger. Offen., 1970, 2 024 835 (Chem. Abs., 1971,74,44 118b). 91 M. Kothari and J. J. Tazuma, J. Org. Chem., 1971, 36, 2951. O3 R. E. McCarley and T. M. Brown, Znorg. Chem., 1964, 3, 1232. O3 L. Bencze and L. Marko, J. Organometallic Chem., 1971, 28, 271. O4 E. A. Zuech, Fr. P., 1969, 1 575 778 (Chem. Abs., 1970,72,999 882). Ob W. B. Hughes, J. Amer. Chem. SOC.,1970, 92, 532. Haines and Leigh equivalents of the organoaluminium reagent with the chlorides and a conse- quential increase in the NO stretching frequencies, followed by attack of two further equivalents of the organoaluminium compound on the oxygens of the nitrosyl groups, leading to the removal of these groups from the rn0lybdenum,~4 as outlined in Scheme 3.In this scheme it is apparent that a function of the aluminium compound is to produce two adjacent vacant co-ordination sites. PY PY DY NO EtC1,AlCI I Ci 1JY PY PY + 2NOAlEtCIa Scheme 3 The catalytic activity of systems derived from compounds of the type [Mo-(NO)abX2] depends on the nature of L and of the anion X, an increase in activity being observed along the series AsPhs < 4-EtC5H4N PhsPO and along the series I < Br < Cl.24 It was therefore suggested that the neutral and anionic ligands are not removed from the metal in the formation of the actual catalytic intermediate. Tungsten derivatives [W(NO)2L2X2] also form active metathesis catalysts but their activity is not as high as that of the corresponding molyb- denum systems.35 Carbon monoxide is also capable of increasing the activity of metathesis catalysts derived from Group VI metal halide complexes and organoaluminium compounds.93 It has been found that [WC~~(C~H~N)Z]-E~AICIZ under an atmos- phere of carbon monoxide exhibits a very different catalytic activity from the same system under argon.The interaction of CO with [WC14(C5H5N)2]-EtAlC12 gives tungsten carbonyl species, and this led to experiments demonstrating that [W(C0)3(PPh3)2C12] in the presence of EtAlCl2 is highly active in converting pent-Zene into but-Zene and he~-3-ene.~~ [WCh {CzH4(PPhz)z }2]-EtAlC12 is also a metathesis catalyst, but excessive exposure to CO can destroy its a~tivity,~3 owing to its conversion into ~(co)6].96 Tungsten carbonyl derivatives D.y(CO)5L] or [W(CO)4L2] in the presence of AlC13 or EtAlCl2 promote the metathesis reaction, provided that trace quantities of dioxygen are present.97 It has been suggested that, in the presence of the tungsten complex, oxygen inserts into a carbon-aluminium bond in [(EtAlC12)2], and the resultant dimer then rearranges to give the species (5).This is believed to bond to the tungsten at a site originally occupied by a carbonyl group. A ten-fold excess of EtAlClz was employed in these studies, however, and it is L.Bencze, J. Organometallic Chem., 1972, 37, C37. B’ L. Ramain and Y. Trambouze, Compt. rend., 1971, 273, C, 1409. 167 Olefin Metathesis and its Catalysis Et\Al 0c’\Al/ CA \lo’ La unlikely that all of the aluminium compound is associated with the tungsten, as must have been assumed in proposing this mechanism.The catalytic activity of [W(PMezPh)4CN2)2]-EtAlCl2 is also increased in the presence of dioxygen9* but this can be readily explained in terms of the creation of vacant or potentially vacant sites through the removal of dirnethylphenylphosphine as its oxide. Dioxygen also affects the kind of product obtained from olehs and wCl~]- EtAI in chlorobenzene.99 Dioxygen does not appear to be necessary to activate metathesis catalysts derived from zero-valent halogenocarbonyl anions of molybdenum or tungsten, [M(CO)5X]-(M = Mo or W; X = C1 or Br), and MeAlC1236 or Grignard reagents73 and from the related metal acyl species [M(C0)5(COR)]- (M = Mo or W; R = Me or Ph) and alkylaluminium dichlorides or sesquichlorides.lOOJO1 The activity of these systems is far less than that of most of the catalytic systems described above.The order of reactivity found for the disproportionation of pent-l-ene by MeAlClz-[M(C0)5COR]- was Mo > W and Ph > Me.100 The dinuclear carbonyl anions [(CO)5M-M’(C0)5In- (M = Mo or W; M’ = Mo, W, Mn, or Re; n = 1 or 2) and the metal carbene complexes [W(C0)5C(OMe)Et] and [W(CO)&(NMez)Me] also form active metathesis catalysts with alkyl- aluminium chlorides, but only in the presence of an excess of chloride ions.lo0J02 It is assumed that the function of the chloride is to produce [M(C0)5Cl]-.Homogeneous catalysts derived from rhenium salts are very much less active than the molybdenum and tungsten species. For instance, 48 h are required for [ReCls]-SnBu4 to convert pent-2-ene into its equilibrium mixture with but-2- ene and hex-3-ene.103 Reactions employing [ReCl4(PPh3)2]-EtAlCh and [ReOX3(PPh3)2l-EtAIClz (X = C1 01Br) also require long reaction times.lo4 At least one transition metal from each Periodic Group has been shown to exhibit homogeneous catalytic activity in the presence of an appropriate co- catalyst. Systems reported include ~iC14(C5H5N)2]-EtAlC12,105J06[(C5H5)2-TiCl]-[Me3A12Cl3],106 [Zr(aca~)4]-[Me3AlzCl3],~~~J~~[NbC15]-[MeA12C13] in the R. J. Haines, unpublished results. OaA.Uchida, K. Kobayashi, and S. Matsuda, Ind. and Eng. Chem. (Product Res. and Development), 1972, 11, 389. looW. R. Kroll and G. Doyle, Chem. Comm., 1971, 839. Iol W. R. Kroll and G. Doyle, U.S.P., 1972, 3 689 433 (Chem. Abs., 1972, 77, 151 431g). lo4 W. R. Kroll and G. Doyle, J. Catalysis, 1972, 24, 356. lo3J. A. Moulijn and C. Boelhouwer, Chem. Comm., 1971, 1170. Io4 E. T. Kittleman and E. A. Zuech, Fr. P., 1969, 1 561 025 (Chem. Ah., 1970,72, 31 193f). lo6D. H. Kubicek and E. A, Zuech, U.S.P., 1972,3 670 043 (Chern.Abs., 1972,77,100 716w). W. B. Hughes, E A. Zuech, E. T. Kittleman, and D. H. Kubicek, 23rd I.U.P.A.C. Con-gress, Boston, 1971, Abstract 566. Haines and Leigh presence of benzoic acid and nitric oxide,lo7 [(Fe(NO)2Cl>2]-[Me3Al~Cl~],106~108 [OSOC~(PP~~)~~[M~~~C~~],~~~~~~~[CoC12(4-vinylpyridine)ne)zl-[Me3AlzC13],106 [(Rh(NO)2Cl }z]-[M~sA~sC~~],~~~,~~~ [{(C3H&RhCl )2]-[Me3A12C13],106~10S [Rh-(NO)(PPh3)2C12]-[Me3A12C13],106Jo* [Ir(NO)(PPh3)2C12]-[Me3A12C13],106~107 PdBrzEtAlClz in the presence of PPh3,109 [C~(PPh3)3Cl]-[Me3Al2Cl3],~~~~~~~ [Cu2Clz(PPh3)2)-EtAlC12,111 [Ag(PPh3)Br]-EtAlC12,106J10[Au(PPh3)Cl]-[MeaAl~Cl3],~~69SmC13-[Me3A12C13],106,l1 ThC14-[ and UClr l10 [Me~A12Cl3].~06The activity of these systems is not as high as those derived from Group VI metal derivatives.For instance, the conversion of heptene was only 2.5 % after 90 h at room temperature employing SmC13-[MeA2C13] as catalyst and 0.2 % after 2 h at 80 "C using [Cu(PPh3)sClI-[Me3Al2Cl3]as the catalyst.The rhodium systems described above are fairly active as metathesis catalysts but, not surprisingly, they also effect rapid isomerization of the olefin.lO8 All of the systems described thus far require two components for catalytic activity. Lewandos and Pettit113 have reported a transition-metal species which does not require the presence of a co-catalyst to promote olefk metathesis. They have shown that for a W :olefin ratio of ca. 1:4 and at a temperature of 98 "C, toluene tungsten tricarbonyl effects a 28%conversion of non-4-ene into oct-4- ene and dec-Sene, in heptane as solvent, during 24 h. Metathesis was only observed when the reaction was carried out in a system where any liberated carbon monoxide could readily escape.No metathesis occurred in a closed system, or with large o1efin:W ratios, or when an excess of toluene was added. On the basis of these results and electron-counting considerations,l14 they suggested that the actual catalyst is [W(C0)2]. In view of the low activity of their catalyst under the experimental conditions employed, other possible mechanisms cannot be immediately discarded. Rhodium(1) complexes in the absence of any co-catalyst effect the metathesis of electron-rich 01efins.l~~ Thus [Rh(PPh&LCl] (L = CO or PPh) catalyses the reaction of olefin (6) with olefin (7) to give (8), the yield of (8) approaching Ph N Ph N C,H,MeN C,H,Me Ph CeH4Me N N N N N Ph Ph CeH4Me C,H4Me Ph C,H4Me (6) (7) (8) lo' W.B.Hughes and E.A. Zuech, U.S. P., 1971,3 562 178 (Chem. Abs., 1971,75,10 868b); US. P., 1972, 3 691 253 (Chem. Abs., 1973,78, 3657r). lo* W.B. Hughes and E. A. Zuech,U.S. P., 1971,3 558 517 (Chem. Abs., 1970,72,99 985w). loS P. H. Phung and G. Lefebvre, Fr. P., 1970, 1 594 582 (Chem. Abs., 1971,74, 87 316g). noD. H. Kubicek and E. A. Zuech, U.S. P., 1971,3 558 520 (Chem. Abs., 1971,74,99 441g). D. H. Kubicek and E. A. Zuech, U.S. P., 1972, 3 703 561 (Chem. Abs., 1973,78,71 378a). 11* E. T. Kittleman and E. A. Zuech, U.S. P., 1971, 3 554 924 (Chem. Abs., 1971,74,99 440 f); U.S. P., 1973, 3 708 551 (Chem. Abs., 1973, 78, 71 382x). llS G. S. Lewandos and R. Pettit, J. Amer. Chem. Soc., 1971, 93, 7087. 114 G. S. Lewandos and R. Pettit, Tetrahedron Letters, 1971, 789.ll6 D. J. Cardin, M. J. Doyle, and M. F. Lappert, J.C.S. Chem. Comm., 1972, 927. 169 Olefin Metathesis and its Catalysis the statistical value after 2 h in xylene at 140 "C.A mechanism involving a four-membered metallocycle as transition state has been proposed, but whether this can be applied generally to reactions involving normal olefins remains an open question. One other system, viz. [WOCl4], has been reported to catalyse olefin metathe- sis in the absence of a co-catalyst, but its activity is exceptionally low.84 The determination of the nature of the catalytic site for olefin metathesis is difficult. In most cases the catalyst is generated by interaction of a transition- metal compound with an organometallic species, and a wide range of products are possible.The catalyst may only be present in solution in low concentration. However, molybdenum and tungsten species form the most effective catalysts, a co-catalyst capable of bonding to the molybdenum or tungsten through halogen, alkyl, or hydride bridges being required. Thus, for [WCl6]-EtAICI&tOH, the catalyst is claimed to be a complex of the transition metal with an aluminium compound, with two tungsten-carbon bonds,ll6 and transition metal-aluminium complexes are formed by Ziegler-Natta catalysts. It is possible that more than one oxidation state is capable of promoting metathesis, but +II is most likely for Mo and W. In any case it is possible to over-reduce the catalyst, so that oxidation states greater than zero seem likely.C.Catalytic Selectivity.-The catalysts employed in olefin metathesis can also promote other types of reaction, including olefin isomerization, olefin oligo- merization/polymerization,and, in the case of homogeneous systems, alkylation of aromatic solvents. Double-bond migration is readily effected by supported metal oxide metathesis catalysts. This explains why Banks and Bailey observed the formation of a large number of unsaturated hydrocarbons in their original studies on but-1 -ene, pent-l-ene, and hex-l-ene.l It can be minimized by poisoning of the isomerization sites on the heterogeneous system with alkali-metal, alkaline-earth, or thallium(1) These presumably reduce surface acidity and inhibit isomerization occurring by a cationic mechanism.l17 The effectiveness increases along the series Na < K < Rb N TI1 < Cs, which is the order of the ionic radii.117 In some chemical processes, for instance the formation of detergent-range olefins from propylene, double-bond isomerization activity is essential.Magnesium oxide is a very selective catalyst for double-bond migration and has been em- ployed in conjunction with metathesis catalysts for obtaining mixtures containing a wide range of olefins.llg Isomerization has also been observed for most homogeneous catalytic systems. For instance, the metathesis of oct-l-ene by [Mo(NO)~(PP~~)~CI~]-E~AIC~~at ambient temperature affords, after 16 h, a mixture of all olefins from ethylene to pentadecene, and not just ethylene and tetradecene as expected.s5 Olefin isomeri- Il6 H.Hoecker and F. R. Jones, Makromol. Chem., 1972, 161, 251. 117 T. P. Kobylinski and H. E. Swift, J. Catalysis, 1972, 26, 416. 118 Shell Internationale Research Maatschaapij N.V., Dutch P., 1967, 6 607 427 (Chern. Abs., 1968,69, 43 389a). 11* R. L. Banks and J. R. Kenton, Belg. P., 1968, 713 190. Haines and Leigh zation is generally considerably slower than olefin metathesis, and can be mini- mized in metathesis reactions by performing the latter at low temperatures and for short periods. Thus little isomerization of pent-l-ene or oct-l-ene by [MO(NO)~(PP~~)~CI~]-E~AICI~is observed after I hour at 0-5 0C.35 Terminal olefins are isomerized more readily by metathesis catalysts than internal olefins.For example, whereas only bu t-2-eneY pent-Zene, hex-3-eneY and ethylene are found in the reaction mixture obtained by metathesis of pent-Zene by [ReC15F AIEt3-02, all olefins from ethylene to octene are formed in the corresponding metathesis of pent-l-ene.77 The extent of isomerization also depends on the non-transition-metal component of the catalyst. The catalytic system [Mo-(NO)2(PPh3)2Cl2]-EtAlC12 promotes isomerization more readily than [Mo-(N0)2(PPh3)2C12]-[Me3A12C13], suggesting that for these systems isomerization is effected by a cationic mechanism, as found for heterogeneous ~atalysts.~5 Oligomerization and/or polymerization of olefins has been reported for a number of homogeneous metathesis catalysts.Although definite attempts have not been made to identify the oligomers in the various reactions nor to establish the nature of the oligomerizations, it has been ascertained that the so-called oligomers formed on metathesis of pent-l-ene by [WCls]-Et&l or [MoC15]- EtsAl in chlorobenzene are not alkylated products of the solvent.77 Further, the small amounts of branched olefins formed in the disproportionation of pent-l- ene by [M~(NO)~(PP~~)~CI~]-[M~&~CIS]can only be explained by some dimeriza- tion process.35 Terminal olefins are oligomerized far more readily than internal olefins, as demonstrated by some reactions involving [WOC14]-EtAlC12 as catalyst.s4 Whereas oligomers in yields of 50 and 30%, respectively, were produced on metathesis of propylene and pent-l-ene employing this catalytic system, only disproportionation products were observed in the corresponding reactions involving pent-2-ene and hept-3-ene. These results are not particularly surprising when one considers that the metathesis catalysts are Ziegler-Natta in type.The oligomerization behaviour of [WCls]-Et&I, [MoC15]-Et3AlYor [ReC15]-Et3AI is destroyed by addition of an excess of EtsAl or PPh3.77 In fact, it has been found that the oligomerization behaviour of [WClsI-Etdl towards oct-l-ene in benzene increases rapidly as the W:AI ratio is varied from 0.25 to 1.O whereas the disproportionation activity decreases.99 These results suggest that the oligomerization is effected by a species of higher oxidation state than is active in the metathesis; a carbonium-ion process participating in oligomeriza- tion was eliminated because alkylbenzenes were not detected in the reaction of [WCls]-Et3Al with oct-l-ene in benzene.99 Heterogeneous catalytic systems for metathesis also readily promote the oligomerization of olefins, as demonstrated by the reaction of CoO-Mo03-Al203 with b~t-l-ene.~ Over SO% of the octenes formed in this metathesis reaction are branched, and this can only be attributed to dimerization of the butenes present in solution.A cationic polymerization mechanism120 has been proposed to account for this dimerization, and, consistent with this proposal, it is found that laoC.E. H. Bawn, Proc. Chem. SOC.,1962, 165.Olefin Metathesis and its Catalysis the oligomerization process can be eliminated by poisoning of the acid sites with sodium carb~nate.~ Addition of fluoride ions to oxide catalysts of molybdenum or rhenium also inhibits oligomerization.121 A sidereaction often encountered when aromatic solvents are employed in metathesis reactions utilizing homogeneous catalysts is the Friedel-Crafts alkylation of the solvent by the olefm. The mixture [WCI6]-EtAIC12 is a particu-larly effective alkylation catalyst under certain conditions. Thus, addition of pent-2-ene to a premixed solution of [WClS] and EtAlCl2 in toluene results in the almost exclusive alkylation of the solvent, and di- as well as mono-pentyl- toluenes are formed.g1 On the other hand, however, if the [WCls]-EtAIC12 catalyst is prepared in the presence of the pent-2-ene, rapid metathesis occurs.Q1 The tungsten species [WCls] promotes the Friedel-Crafts alkylation of benzene with propylene in the absence of any EtAIC12, although at temperatures up to 150 "C (much higher than those for [WC16]-EtAlC12).It has been suggested that the actual catalyst in the [WCls]-EtAlch system is a derivative containing the highly active Friedel-Crafts catalyst AICL, formed by the reaction of [wc16] with EtAIC12, co-ordinated to a reduced tungsten spe~ies.~lJ22 On this basis, it would be expected that [WCls]-Et3AI would be a less effective alkylation catalyst than [WCle]-EtA1C12 under comparable conditions, and this is found to be the case for both terminal and internal 0lefins.76,~~J~3 In fact, in the complete absence of water or dioxygen, [WCh]-Et3Al does not effect the alkylation of benzene with oct-1 -ene.99 The inability of [WCl6]-LiAlHg to promote alkylation of aromatic solvents is further indirect evidence for the suggested intermediate.80 Addition of pyridine and triphenylphosphine to [WCh]-EtAlC12 has been shown to inhibit the alkylation without completely curtailing the metathesis reaction.gf Like [WClS], [MoC15] and [ReC15] also promote the alkylation of aromatic solvents by but in the presence of SnBun4 the [ReC15] loses its alkylating ability while remaining an effective metathesis catalyst .lo3 4 The Mechanism of the Olefin Metathesis Reaction No single mechanism has been suggested to explain all olefm metathesis reactions and, indeed, it may proceed by more than one mechanism.Nor is it necessary that the reaction route should be the same for both homogeneous and hetero- geneous reactions. It is apparent that even different homogeneous catalysts use different mechanisms. A satisfactory mechanism should be extendable to hetero- geneous alkyne metathesis, which occurs in conditions very similar to those which promote olefin metathesi~.~~,~~ We first consider the general experimental evidence upon which the mechan- isms must be based. A. Factors to be Considered in Proposing a Mechanism.-(i) The Olefins. The lal R. Argankight, U.S.P., 1972, 3 697 613. la*J. R. Graham and L. H. Slaugh, Tetrahedron Letters, 1971, 787.lSs A. Uchida, Y. Hamano, Y. Mukai, and S. Matsuda, Ind. and Eng. Chem. (Product Res. and Development), 1971, 10, 372. lap J. Tsuji, T. Nogi, and M. Morikawa, Bull. Chem. SOC.Japan, 1966, 39, 714. Haines and Leigh metathesis of [2-14C]propene over a heterogeneous catalyst of rhenium oxide on alumina yields an ethylene sample free of 14C.125 2C3H6 + Ca + CHsCH=CHCH3 (cis-and trans-isomers) This excludes a linear mechanism, e.g. reaction (3). * * * * * 32c=c-c --t [C-c-c-c-c-C] c=c + c-c=c-c (3) On this same catalyst, methyl groups retain their integrity throughout the reaction,126J27 as shown by experiments with [1-14C]propene and [2-14C]propene. This precludes hydrogen transfer from olefin to metal and back again, and hence r-allylic intermediates.Similar studies over a cobalt molybdate-alumina catalyst yielded essentially the same results, but above 60 "Cisomerization of the propene occurs,127 and then an allylic intermediate cannot be excluded. An investigation of the homogeneous metathesis of deuteriated butenes also showed that hydrogen transfer to the metal does not occur, because no H-D exchange was ob~erved.~J~ This is also consistent with the results from the heterogeneous metathesis of ethylene and 2,3-dimethylbut-2-ene,128which show that if hydrogen does migrate to the metal it must always return to the carbon atom from which it came. This is unlikely. The observations cited are generally consistent with a transition state in which the four carbon atoms involved in the bond switching are equivalent.To account for this, Bradshaw et aL4proposed a 'quasi-cyclobu- tane' picture of the tiansition state, which has since been adopted by 0thers.1~~-~3~ An alternative model, which uses the same equivalence concept, pictures the four carbon atoms in the transition state as bound to a single metal ion essentially as met hylene fragment s.1139114 Kinetic studies of the classical type have done little to clarify the mechanism, because they are very difficult, and the results from different systems often con- flict. Various attempts have been made to rationalize the rates of metathesis of propylene over a Co0-Mo03-A.h03 catalyst in the temperature range 120-200 "C.The rates are consistent with a two-site mechanism, with an activation energy of about 8 kcal m0l+3~ The rates of olefin metathesis over the homogeneous catalysts derived from molybdenum(r1) dinitrosyls and alkylaluminium halides have been measured.95 At low o1efin:catalyst ratios the order of reaction with respect to olefin was found to be greater than unity, but it fell towards unity as the olefin concentration was increased.The order of reaction with respect to lS6 J. C. Mol, J. A. Moulijn, and C. Boelhouwer, Chem. Comm., 1968, 633; J. Catalysis, 1968, 11, 87. 18e F. L. Woody, M. J. Lewis, and G. B. Wills, J. Catalysis, 1969, 14, 389. la' A. Clark and C. Cook, J. Catalysis, 1969, 15, 420. D. L. Crain, J. Catalysis, 1969, 13, 110. 189 E. A. Zuech, Chem.Comm., 1968, 1182. 130 C. T. Adams and S. G. Brandenberger, J. Catalysis, 1969, 13, 360. lS1 R. Pettit, H. Sugahara, J. Wristers, and W. Merk, Discuss Faraday SOC.,1969, No. 47, p. 71. 13p M. J. Lewis and G. B. Wills, J. Catalysis, 1969, 15, 140; R. C. Luckner, G. E. McConocie, and G. B. Wills, ibid,, 1973, 28, 63; A. J. Moffat and A. Clark, ibid., 1970, 17, 264. 173 Olefn Metathesis and its Catalysis catalyst could not be determined. The results are consistent with rapid complexing and decomplexing of the olefin, the metathesis step being rate-determining. The activation energy for the process was found to be ca. 7 kcal mol-l, which is low, as expected for a reaction that is basically athermal and entropy-controlled.95 The stereochemistry of metathesis over the same homogeneous metal nitrosyl- aluminium alkyl catalysts has also been studied.133 At equilibrium of the metathesis of pent-Zene to but-2-ene and hex3-ene the cis and trans pairs of each olefin are in thermodynamic equilibrium, but in the early stages of the reaction the isomer ratios depend upon the starting pent-2-ene. Thus, cis-pent-2- ene yields, preferentially, cis-but-Zene and cis-hex-3-ene, and trans-pent-Zene yields, preferentially, the corresponding trans-isomers. The rates of isomerization and metathesis appear to be related, and the observations are explicable in terms of two olefin molecules bound cis to molybdenum, with their axes parallel, and undergoing metathesis via the ‘cyclobutane’ transition state.133 The isomer ratios from pent-2-ene in contact with [WC16]-EtAlC12 and ethanol (1 :4: 1) in a homogeneous system are in sharp contrast.10 Substantial amounts of cis-and trans-hex-3-ene are formed at the outset (i.e.after about 20 seconds at room temperature) regardless of which isomer of pent-Zene is used, but cis-isomers are present in slightly more than equilibrium proportions initially. The residual pent-Zene was also observed to approach slowly its equilibrium isomer content. These results are consistent with a random transalkylidenation coupled with a minimum of steric control.lO~71 It has also been inferred that the overall mechanism of the reaction over homogeneous tungsten catalysts is different from that over molybdenum nitrosyl derivatives.Thus, investigation of cyclo-octene polymerizations shows that high- molecular-weight polymers are formed during the early stages of the reaction, and this is unexpected if olefm exchange is more rapid than metathesis20 because otherwise the high-molecular-weight polymers would form more slowly. The explanation for these divergences may lie in the different structures of the catalysts, The generalizations above, complex as they are, may nevertheless be an oversimplification. Many of the reactions are accompanied by isomerization, even though this is generally a slower reaction. More interesting, ethylene over a heterogeneous molybdenum-alumina catalyst at 210 “C in a static reactor, rather than in the more usual flow systems, gives propene, butene, small amounts of hexenes and heptenes, and, as the major product, ~entenes.13~ The initial reaction would seem to be a reaction of ethylene to form propene, possibly via formation of methylene (carbene) and thence trimethylene (cyclopropane).However, a concerted reaction, such as depicted below, cannot be excluded. Reactions producing compounds with an odd number of carbon atoms have also been lSs W. B.Hughes, Chem. Comm., 1969, 431. lS4P.P.O’Neill and J. J. Rooney, J. Amer. Chem. SOC.,1972,94, 4383. Haines and Leigh observed in homogeneous systems,135 but these could arise by isomerization followed by metathesis. Molybdenum hexacarbonyl-alumina catalysts also catalyse the conversion of ethylene into cyclopropane and methylcyclopropane, although the reaction temperature (140 "C)is much lower.1 Cyclopropane itself rearranges to propene under very mild conditions, so that a common route for all these reactions seems possible.Further, olefins are converted quantitatively into the corresponding cyclopropanes by the action of carbene (derived from CH2N2) under the action of palladous salts.136 All this lends credence to the idea that carbenes and olefins may exist together on an appropriate metathesis catalyst. The metathesis reaction, whether homogeneous or heterogeneous, is rapid, isothermal, and entropy-controlled. The activation energy is less than 10 kcal mol-1. This makes the formation of free carbenes unlikely, but, at least where electron-rich olefins are involved, carbene-containing products have actually been isolated.Thus in the homogeneous metathesis of (9), catalysed by [RhCl-(Pph,)~],(10) has been identified as an i11termediate.~zJ15 It is not clear whether intermediates of this kind form in the general case, since the olefins involved are atypical. (ii) The Catalysts. There would seem to be no easy classification of the catalysts for either homogeneous or heterogeneous metathesis. The heterogeneous catalysts are generally compounds of molybdenum or tungsten, often supported on alumina,54 and they are not well-defined compounds. The homogeneous catalysts* are generally derived from high-oxidation-state halides and a reducing agent, generally a metal alkyl, such as ethylaluminium dichloride.5,20J~3 Although the proportion of the second metal to the primary metal can affect activity, and even the kind of products, neither the alkyl nor the second metal is necessary in every case; tungsten and [(n-CH3C6H5)W(C0)3] gives rise to a catalyst by loss of carbon monoxide and toluene.1'3 In the system [WC~~(P~~PCH~CH~PP~Z)~]-E~AICI~,carbon monoxide actually stimulates the catalyst.93 The alkyl-metal compound apparently reduces the tungsten halide. It has not yet been unequivocally demonstrated which oxidation state(s) of tungsten are responsible for the catalysis.It is believed that the catalytic species derived from [(PPh&Mo(N0)2C12] and organoaluminium halide contains molyb- * Note added in proof: it has been suggested that the [WClJ-metal alkyl system may give rise to catalyst systems that are wholly heterogeneous; unequivocal proof would seem difficult (E.L. Muetterties and M. A. Busch, J.C.S. Chem. Comm., 1974, 754). lS6 K. Hummel and W. Ast, Naturwiss., 1970, 57, 245. lS* R. Padisson, A. J. Hubert, and P. Teyssit, Tetrahedron Letters, 1972, 1465. Olefin Metathesis and its Catalysis denum(o).~5~QsIn one cases6'MO(C0)4' has been suggested as the active entity, although potentially oxidizing halogenated olefins stimulate reactivity.69 Later, the complete loss of CO from [Mo(CO)s] was postulated in catalyst formation.67 In another homogeneous system, a tungsten dicarbonyl fragment has Wn inferred as the catalyst.113 This suggests a d6 configuration for the transition metal, although substitution reactions at octahedral low-spin @-metal ions are usually slow.It is unlikely, however, that catalytic systems produced by re- ducing, say, Wch], contain Wo, and oxidation to above Wo is required for the preparation of some of the others. Thus [W(C0)5(PPhs)]-EtAlC1~~7requires dioxygen for catalytic activity, [WClsl--EtAIC12-EtOH presumably contains tungsten in an oxidation state of at least +x11,10 and there is also the range of various tungsten halide-aluminium halide mixtures, and the majority of the heterogeneous catalysts derived from metal oxides on supports such as alumina. The catalyst for the dismutation of electron-rich olefins contains rhodium(1) or rhodium(r~x).~~~ In summary, although active catalytic species may be derived from Moo or WO (d6)in some systems, the majority probably have fewer d-electrons and higher oxidation states.(iii) The Transition State. The synchronous making and breaking of double bonds as apparently required by the experimental evidence is thermally forbidden according to the Woodward-Hoffmann rules.3 The transition-metal catalyst makes the reaction possible, and in the absence of such a catalyst the reaction occurs only at ca. 725 "C.2 It is therefore of great theoretical interest to identify the transition state. Before considering the three current models it is worthwhile making two reservations. The first is that no mechanism yet proposed offers any role to the metal alkyl (or comparable reagent) other than that of a reducing agent or of a modifier of the base behaviour of the transition metal. Two of the models, which may apply only to specific systems, do not require a second metal compound at all.In the general case, a polynuclear metal catalysis has not been considered although the formation of compounds with halogen or hydrogen bridges between, say, tungsten and aluminium is very probable. The second reservation is that apparently synchronous transitions may be step-wise. Thus, the thermally forbidden conversion of quadricyclene (11)into norbornadiene (12) is catalysed by [Rhz(C0)4C12]. The conversion involves oxidative addition of Rhrto the quadricyclene, which is ratedetermining, and the thermally forbidden Haines and Leigh reaction is thus made possible.137J38 The olefin metathesis reaction could be facilitated similarly. However, if we assume synchronous bond-making and bond-breaking, a possible transition state is (1 3), the ‘quasi-cyclobutane’ transition state.4 R’z R~,c=cR’~ M+ C 3E R22 It should be emphasized that the fact that synchronous transformations of this kind are forbidden by the Woodward-Hoffmann rules simply means that they have excessively high activation energies for normal thermal activation.How a transition-metal ion might facilitate such forbidden reactions is discussed in the next section. The cyclobutane transition state cannot represent the whole picture, because under heterogeneous conditions even-chain olefins can undergo reactions that produce odd-chain pr0ducts.13~ These products could arise by isomerization, e.g.by but-l-ene isomerizing to but-Zene, which then undergoes metathesis with more but-l-ene to produce propene. However, the conversion of ethylene into propene must involve the complete fission of an ethylene molecule. Diazomethane decomposes, apparently at the sites on a heterogeneous catalyst that also catalyse the metathesis reaction, to yield dinitrogen and ethylene, and the latter presum- ably arises by dimerization of carbene.139 Thus, carbene is implicated to some degree in metathesis reactions involving ethylene, and carbene complexes can generate metathesis catalysts. 100 The heat of formation of two molecules of ethylene is very similar to that of cyclobutane.One would therefore expect at least some cyclobutane to be produced from ethylene and a heterogeneous metathesis catalyst, but the con- version over a heterogeneous oxide-based catalyst is less than 0.1 %. Conversely, the yield of ethylene from cyclobutane over the same catalyst is only 3 %. These results suggest that cyclobutane is not involved in the transition state,114 though one can make the reservation that ‘quasi-cyclobutane’ is in a condition far removed from that of cyc10butane.l~~ For these reasons, an alternative mechanism involving the breaking of both cr-and n--components of the carbon-carbon double bond of the olefin has been proposed.l13J14 The transition state (14) is best described as four methylene fragments bonded to a single transition-metal ion, the carbon atoms remaining sp3 hybridized.Evidence from homogeneous systems supports this interpretation,l13 for example [(~T-CHQC~H~)W(C~)~] is a metathesis catalyst for non-4-ene in heptane H. Hogeveen and H. C. Volger, J. Amer. Chem. SOC.,1967,89,2486. L. Cassar and J. Halpern, Chem. Comm., 1970, 1082. 1s9 P. P. O’Neill and J. J. Rooney, J.C.S. Chem. Comm., 1972, 104. Olefin Metathesis and its Catalysis :'a g, II +M+ II C C Raa R4a at 98 "Cwhen the system is open, but not when carbon monoxide is kept in the system. In these latter circumstances isomerization occurs. A large excess of olefin also suppresses metathesis relative to isomerization. An excess of toluene inhibits both reactions of the olefin.These data are consistent with a species [(olefin)2W(CO)z] being involved. The conversion of (15) into (16) involves a change in the number of electrons in the valence shell of tungsten from 14 to 18, and (16) thus obeys the inert-gas rule. For a corresponding alkyne metathesis, the analogue of (15) should only possess 10 electrons, which seems rather unlikely, but compounds are known in which fCR groups are bonded to transition-metal ions, e.g. [M(CR)X(C0)4] (M = Cr, Mo, or W; R = Me or Ph; X = C1, Br, or I).l*O Although the evidence for the tetramethyl intermediate in these particular circumstances is convincing, the system may not be typical. Reaction tempera- tures are high (N 100 "C), whereas homogeneous metathesis catalysis usually occurs at ca.25 "C, and the reaction is inhibited by an excess of olefin. Gen- erally, very high o1efin:transition metal ratios are used (ca. 500:1), but in the tetramethylene system the proportions used are comparable. A third mechanism has been suggested, stimulated by the observations that [WCls]-alkyl-lithium is most active for metathesis with a tungsten :lithium ratio of 1:2 and that rhodium-catalysed carbocyclic ring rearrangements proceed via metal-carbon o-bonded intermediates.141 A conceivable pathway is that shown in Scheme 4. To establish the feasibility of this scheme, the reaction of [WclS] with 1,4-dilithiobutane in benzene was investigated, and it was found to yield ethylene quantitatively, presumably from the breakdown of an intermediate (19) [cf.(17) and (18) in Scheme 41.Experiments with deuteriated dilithiobutane, 1,4-dilithio-2,3-dideuteriobutane,produced ethylenes in the proportions ldoE. 0. Fischer, G. Kreis, C. G. Kreiter, J. Muller, G. Huttner, and H. Lorenz, Angew.Chem. Internat. Edn., 1973,12,564. 141 R. H. Grubbs and T. K. Brunck, J. Amer. Chem. SOC.,1972,94,2538. Haines and Leigh H,C-CHSA 7 1' HRC, M,CRHC CHR RH H*C CHR 11-M --I1 H*C CHR Scheme 4 CzHzDz:CzH3D:CzH4 = 6:88 :6. Hence rearrangements or methathesis definitely occur, and subsequent experiments proved that there is stereoselectivity. Cl, This work is only in its preliminary stages, and leaves several questions unanswered.There appears to be no reason why the transformation (17) -+(18) occurs, or why it should be rapid. A subsequent paper, which describes the synthesis and structure of the complex (20), proposes two possible mechanisms, but without any supporting evidence.142 The metallocycles of type (17) are not normally formed by ethylene,* but tetrafluoroethylene readily does so, and norbornadiene and [Ir(CO)Cl(PPh&] yield a stable complex containing a * Note added in proof: there is persuasive evidence that ethylene can form such a metallo-cycle in a titanium complex (J. X. McDermott and G. M. Whitesides, J. Amer. Chern. SOC.,1974,96, 947). 14*C. G. Biefeld, H. A. Eick, and R.H. Grubbs, Inorg. Chem., 1973, 12, 2166. 179 Olefin Metathesis and its Catalysis metallocycle, which, with acetylacetone, yields the compound (2l).I43 Thus the identity of the five-membered metallocycle with the metathesis transition-state (21) is open to doubt.Thirdly, it is not yet proven that metal-hydrogen bond forma- tion is not involved in the transformation (17) -+ (18), whereas experiments in other systems have excluded this route. Fourthly, the deuteriated ethylenes are not formed in the equilibrium proportions typical of the metathesis reaction. In addition, the conversion of ethylene into propylene and cyclopropane is not immediately explicable on this model. Finally, it does not explain why buta- diene should be a poison for metathesis, nor how alkyne metathesis might occur. However, the transition state does overcome the activation-energy ob- jections to synchronous mechanisms, as do the two other mechanisms so far discussed.The most recently proposed mechanism115 has already been mentioned, and is shown in Scheme 5. The carbene complexes (22) and (23) have been isolated, -CR~~=CR*,\(22) (u) Scheme 5 14a A. R. Fraser, P. H. Bird, S. A. Beman, J. R. Shapley, R. White, and J. A. Osborn, J. Amer. Chem. SOL, 1973, 95, 597. Haines and Leigh and the metathesis has been found to take place in boiling xylene, but only when :CR21 and :CR2a are derived from electron-rich olefins. Although the evidence for the mechanism proposed in this particular caseis compelling, it is unclear how far it can be extended to olefin metathesis reactions in general.B. Theories of Catalysis.-These will be discussed in order, from the simplest to the most complete. The basic problem can be grasped by considering the re- arrangement of molecular orbitals involved in the overall transition (Scheme 6). Scheme 6 If a cyclobutane intermediate is involved, then the transformation can be repre- sented as the sum of two individual steps, viz. Scheme 7 and its converse. Con- CR*~--CR~~ Scheme 7 sidering only the p-orbitals involved in double-bond formation in the two ethylenes, then the two diagrams for correlation of orbitals and states shown in Figure 1may be drawn. From either of these it is evident that the transformation I\ I-\ II + II 0 II + It 0 X Figure 1 of the ground state of two ethylene molecules into a ground-state cyclobutane in the absence of catalyst involves the transfer of electrons from one orbital to another, a process which requires a high activation energy so that, in Woodward Olefn Metathesis and its Catalysis and Hoffmann’s terms, the thermally activated process is symmetry-forbidden. The catalysis has been rationalized as follo~s.~44 The two olefin molecules form a complex with a transition metal, with symmetry properties as indicated in (26).The simple correlation diagrams of Figure 1 must then be redrawn to in- clude transformations of the metal orbitals, and particularly the d-orbitals. The orbi tal-correla t ion diagram is now as shown in Figure 2. A ground-state Complex Figure 2 reaction path from cyclobutane to the bis-(olelin) complex is now available, provided that the transition-metal ion has between 2 and 8 d-electrons, and this is consistent with experiment. In theory, too, a path exists for a non-trans-ition-metal catalyst with only 2 available electrons, in an orbital that is not a &orbital but of the appropriate symmetry, and such catalysts, e.g.potassium in graphite, are known.145 The tetramethylene model has received a similar treatment.ll4 In this case, the metal-tetramethylene complex has a set of molecular orbitals built up from sp3 carbon hybrid orbitals and the appropriate metal orbitals. The orbital-correlation lQ4The most recent discussion is by: F. D. Mango and J. H. Schachtschneider in ‘Transition Metals in Homogeneous Catalysis’, ed. G.N. Schrauzer, Marcel Dekker, New York, 1971, p. 223. See also F. D. Mango and J. H. Schachtschneider, J. Amer. Chern. SOC., 1967,89,2484. F. B. Carleton, personal communication. Haines and Leigh diagram provides a ground-state pathway, and the metal apparently contributes 6 electrons to the bonding scheme. It will be recalled that the experimental data from which the tetramethylene model was developed were derived from a system containing tungsten(o) (d9.114 The function of the metal in a quasi-cyclobutane intermediate has been described144 as follows, The metal injects an initially non-bonding pair of d-electrons into a carbon-carbon antibonding orbital combination, such that bonding between C-1 and C-2 (and C-3 and C-4) is weakened [see (27)], but, as a consequence of the nodal pattern of this combination, bonding between C-2 and C-3 (and C-4 and C-1) is strengthened.At the same time, the metal with- draws an electron pair from an orbital bonding between C-1 and C-2 (and C-3 and C-4). Considerations of this kind suggest that a d2 ion should be a better catalyst than ions with more d-electrons. The metallocycle model has also been discussed in these terms.144 Three paths for the rearrangement have been suggested, the asterisks being used in Figure 3 to denote the identities of the individual carbon atoms. * Y (b$ **= M *-C%* /\ M 6, ,cM Figure 3 Pathway (a) involves reversible insertion of the metal into a cyclobutane ring, maintaining the cyclobutane-metal bond at all times, since cyclobutane is never Olefin Metathesis and its Catalysis evolved.This is not considered likely.* Pathway (6) is similarly not likely because it implies metal-carbon interactions across the ring, and these cannot be strong if the ring is planar. However, in (28) the ring is far from ~1anar.l~~ Pathway (c) apparently suffers from further symmetry restrictions, which a d2 metal could not overcome because it would become do in the metallocycle. Since a tetra- methylene intermediate is considered improbable on activation-energy grounds, then all these theoretical discussions suggest the quasi-cyclobutane model as the only reasonable one? This kind of approach has been criticized on the grounds that orbital-correla- tion diagrams are not adequate, and that state-correlation diagrams should be ~sed.1~6It is claimed that the ‘allowed-forbidden’ problem is quite distinct from another, which is that the activation energy is normally too high in reactions such as the metathesis reaction, for steric reasons.The lowering of the activation energy by transition-metal catalysis for certain ‘forbidden’ reactions may be as great as 15 kcal mol-l for some d1Oand d8systems (e.g.AgI and RhI). A proper consideration of the state-correlation diagram for the olefin metathesis reaction shows that all the states of the bis(o1efin) complex and of the cyclobutane-metal intermediate are completely symmetrical with respect to reflection in the appro- priate planes, but that the ground states of the two species do not correlate.This is, however. only true if the metal ion formally has an even number of electrons, and this may not occur if odd-electron ligands such as NO are also present. The role of the metal ion is to provide excited states on the ion which are of lower energy than those excited states of the organic molecules which would have to be occupied if the non-catalysed transformation were to occur.146 However, in the example selected, all the states of the organic moiety-metal ion complex are of the same symmetry because of the presence of the metal ion, so that one might equally argue that catalysis occurs because forbidden tran- sitions become allowed. In any case, it is not clear what ‘forbidden’ means, unless it is that the activation energy is very high.The most sophisticated analysis of the cyclobutane system develops the argu- ments concerning state-correlation diagrams one stage further.147 The additional consideration introduced is that the relative energies of the metal ion d-orbitals * Note added in proof: an analogous step for alkyne metathesis has been suggested on the basis of reaction products obtained from [Fe(CO),] and diynes (H. B. Chin and R. Bau, J. Amer. Chem. SOC.,1973, 95, 5068). 146 W. T. A. M. van der Lugt, Tetrahedron Letters, 1970, 2281. G. L. Caldow and R. A. Macgregor, J. Chem. SOC.(A), 1971, 1654. Haines and Leigh will change during the course of the reaction. Consider, for example, the usual bis-olefin to cyclobutane conversion (Scheme S), with the transition metal Scheme 8 initially square-planar.It is evident that the act of removing electrons from the olefinic bonds and placing electrons between the initially unbound carbon atoms means for the metal an isomerization from square-planar to tetrahedral. When this is included in the correlation argument, it becomes apparent147 that a square-planar-tetrahedral isomerization is unlikely to be involved in the meta- thesis catalysis. For an axial-equatorial trigonal-bipyramidal exchange (Scheme 9), however, catalysis does seem more likely, with low-spin d6 metal ions, and possibly with dl and d2. Scheme 9 It must be emphasized that none of these theoretical discussions is valid if the mechanisms postulated are eventually found to be inapplicable.The recent demonstration148 that ethylene and butadiene (a poison for metathesis catalysts) can form considerable quantities of vinylcyclobutane under the influence of a catalyst such as tetrabenzyltitanium, which in the transition state may not possess any d-electrons at all, suggests that there may be catalysts without transition- metal ions. It is not clear whether this system also promotes metathesis (the products of such metatheses are the same as the starting materials), but possibly unsymmetrical olefins may break orbital degeneracies and promote new reaction paths. The scope for future research seems very large.C. A Proposed Mechanism.-It is not proven whether there is a mechanism common to all the metathesis reactions discussed. In fact, they fall into two natural divisions, those which operate at room temperature and below (most of the homogeneous systems) and those which operate at about 100 "C and above 148 L.G. Cannell, J. Amer. Chern. Soc., 1972, 94, 6867. Olefin Metathesis and its Catalysis (most of the heterogeneous systems, the tetramethylene system, and the ‘electron- rich-olefin’ system). All the reactions in the second group are those in which there is evidence for participation of CR2 fragments (whether termed ‘carbene’ or ‘methylene’). The fact that heterogeneous metathesis of propylene and the decomposition of diazomethane to yield ethylene and dinitrogen are very fast over the same catalyst may not be relevant.However, although there is good evidence that the catalysts can split the carbon-carbon double bond, the CR2 fragments retain their integrity, and there is no evidence for hydrogen transfer to metal. The production of cyclopropane from ethylene could occur as shown in Scheme 10. However, only one carbene (methylene) fragment is necessary, so that a or Scheme 10 process involving three carbon atoms (Scheme 11) is also feasible. Various versions of this can be sketched. Under the reaction conditions, cyclopropane will almost certainly rearrange to give propylene, and a further possibility is the (29) Scheme 11 insertion of a carbene fragment into cyclopropane to give methylcyclopropane, and thence various butenes.Such reactions are, however, generally of minor significance, but the intermediate (29) is adequate to explain the whole metathesis reaction, any stereoselectivity being ascribed to the steric characteristics of the catalysts, as shown in Scheme 12. Scheme 12 Haines and Leigh A variation of this has already been pr~posed,l~~J~~ in which a metaIIocycle (30) is envisaged as the transition state. It is not possible to make a definitive judgement. (30) recalls the four-carbon metallocycles (17) and (18), but in that system metathesis has not been detected with certainty, although hydrogen scrambling has been demonstrated.124 However, neither (29) nor (30) presents any conceptual difficulty for the metathetical reactions. We prefer not to commit ourselves, and shall refer to (29) or (30) as the trimethylene model.Whether this can be extended to homogeneous systems is open to question. Generally the activation energies for homogeneous systems are low, and it seems doubtful whether the activation energies for metathesis via metallocycle (30) with normal olefins or via a tetramethylene complex are small enough. It may be, however, that one function of the EtAlC12 is to generate a carbene during breakdown of an aluminium-tungsten alkyl or of a tungsten alkyl, and aluminium alkyls are known to react with some transition-metal complexes to form carbene complexes.149 Additionally, whereas oxidative addition to RhI (to give RhnI) is eminently reasonable for the route via (30), the corresponding transformation in the tungsten-catalysed homogeneous reaction (possible WI1-WN) seems less likely.We have assumed that alkyne metathesis follows a path analogous to that of olefin metathesis. If this is so, the cyclobutane model seems most appropriate. The formation of complexed cyclobutadienes from acetylenes is relatively common,l50 and these, upon thermal decomposition, would undoubtedly give the impression of alkyne metathesis. The analogue of the tetramethylene model (tetramethyne) does not seem feasible. If the effective atomic number rule is to be obeyed, then the critical transformation must be that between (31) and (32), so that M must be effectively free of any other ligands if it is to be a Group VI metal, and this is improbable.A similar objection holds to the trimethyne variant, and although metallocycles such as (33; R = H) are ~XIOW~,~~~their chemistry is virtually unexplored. RC M'CR=fR~R-CR W. Petz, J. Organometallic Chern., 1973, 55, C42. lKoSee, for example, P. M. Maitlis, Adv. Organometallic Chem., 1966, 4, 95. ls1 See, for example, R. P. Dodge and V. Schomaker, J. Organometallic Chem., 1965, 3, 274. 187 Olefin Metathesis and its Catalysis Internal acetylenes C2Rz react with MoCh or WBr5 to yield complexes (X =~(C~RZ)] C1, M = Mo; X = Br, M = W;R = alkyl or aryl), which react with organic nitriles to yield the well-characterized [M&(nitrile)2]. The acetylene complexes are believed to contain metallocycles (33), but their thermal decomposition does not lead to metathesis products.153 5 Conclusions Our understanding of the olefin metathesis reaction is in its infancy.There is obviously a close relationship between catalysts which promote this reaction and those which aid transalkylation of arornatic~~~~J53 and Ziegler-Natta p0lymerization.16~An understanding of how these catalysts function and why minor constitutional changes endow them with such different capabilities will go far towards an important goal of organometallic chemistry, namely the design of specific catalysts for specific reactions. 16’ A. Greco, F. Pirinoli, and G. Dall’Asta, J. Organometallic Chem., 1973, 60, 115. lSa L. Hocks. A. J. Hubert,and Y.Teyssie, Tetrahedron Letters, 1972, 3687. lS4 0.Henrici-Olive and S. Olive, Angew. Chem. Internat. Edn., 1967, 6, 790; see also ref. 25.