J. MATER. CHEM., 1994, 4( l), 13-1 7 Mechanisms of Pyrolysis of Organometallic Deposition Precursors lain M. T. Davidson, Andrew M. Ellis, Graham P. Mills, Mark Pennington, Ian M. Povey, J. Barrie Raynor, Douglas K.Russell,*t Sinan Saydam and Andrew D. Workman Department of Chemistry, University of Leicester, Leicester, UK LEI 7RH The gas-phase pyrolysis mechanisms of a number of potential transition-metal deposition precursors have been investigated using the techniques of: (i)infrared laser powered homogeneous pyrolysis coupled with product identifi- cation by FTIR, NMR and GC-MS; (ii) stirred Flow Reactor kinetic measurements; (iii) EPR spectroscopy of matrix- isolated free radicals. Preliminary results are presented for: (a) MeMn(CO), and AcMn(CO),, both alone and in the presence of Me,SiH; (b)C,H,Mn(CO), and MeC,H,Mn(CO),; (c) C,H,Fe(CO),; and (d) (C,H&Fe, all of which provide clear evidence that purely homogeneous pathways can be very different from those of surface-catalysed decomposition.In contrast to the long-established use of main-group organometallic compounds,' the potential of volatile organo- transition metal compounds as precursors for the deposition of metals, metal oxides, metal silicides, and other compounds has come to be recognised only recently. Until the mid 1980s, deposition of materials containing metals such as W, Ta, Ti or Mo was achieved using volatile fluorides or chlorides. Halides are not always volatile or convenient, however, and frequently generate films heavily contaminated with the hal- ogen, as well as hazardous by-products.Although binary transition-metal carbonyls usually produce very clean films, these compounds bring their own problems of handling and toxicity. For these reasons, attention has turned to simple organic derivatives of carbonyls. For example, Nouhi and Stirn2 have shown that tricarbonyl(methylcyclopentadieny1)-manganese MeC,H,Mn(C0)3 (MCMT), has potential as a manganese precursor, and Pain et aL3 have discussed the advantages of pentacarbonyl(methy1)manganese(MMP) as a dopant source and in the production of magnetic materials such as manganese tellurides. Although some of these potential precursors have been the subject of some investigations of an empirical nature, rather little is known of the fundamental mechanisms involved in their thermal decomposition. As has been trenchantly demonstrated in main group systems, infor- mation of this sort can often lead to the rational design of new precursors of desirable physical, chemical and economic properties.We have shown that the adaptation of methods well estab- lished in physical chemistry can provide unique insights into the mechanisms of metal-organic chemical vapour deposition (MOCVD) and molecular beam epitaxy (MBE) processes. For example, the technique of infrared laser-powered homo- geneous pyrolysis (IR LPHP) has been put to very effective use in the elucidation of the pyrolysis mechanisms of main group precursors such as Me3A1,4 Et3Ga,5-7 Bu:Ga and Bu; Gag and Et,Zn.' The technique of determination of kinetic parameters using the stirred flow reactor (SFR) has proved invaluable in the study of the chemistry of organosilicon compounds." The EPR detection of free radicals produced in the pyrolysis of organic compounds," trapped by matrix isolation (MI), has been known for many years, and has recently been adapted to the study of organometallic com- pounds in our laboratory.In the present work, we describe the application of these techniques to a number of potential organo-transition-metal MOCVD precursors, and present some preliminary conclusions. t Present address: Department of Chemistry, University of Auckland, Private Bag 92019, Auckland, New Zealand. Experimental The experimental methods employed in the present investi- gations have been described in detail elsewhere, and hence only brief summaries of the more significant aspects are presented here.Except where noted below, analytical investi- gations (FTIR, NMR, GC-MS, elemental analysis, and EPR) were conducted using commercial instrumentation, in con- junction with comparison with authentic samples. In most of the studies described here, all three techniques have been applied. Infrared Laser-powered Homogeneous Pyrolysis The majority of qualitative studies were carried out using the technique of IR LPHP. The technical details and adkantages of this method have been described in an extensive recent review by one of the present authors,', and a wide range of A1 and Ga systems have been investigated with its use.13 Static pyrolysis is carried out in a cylindrical Pyrex cell (length 10 cm, diameter 3.8 cm) fitted with ZnSe windows. In compari- son with cheaper materials such as NaCl, ZnSe has ;i higher optical transmission at the IR laser wavelength of 1Opm.In addition, it has greater mechanical strength and is not hygro- scopic, a point of considerable importance in the investigation of moisture-sensitive materials. The cell is filled with up to 10Torr of the vapour of the material under study, Together with 10Torr of SF, (for compounds of moderate volatility, liquid or solid can be condensed into the cell; this does not alter the basic features of the IR LPHP method, but does introduce evaporation as a possible rate-limiting process).The contents of the cell are then exposed to the output of a free- running CW CO, IR laser at power levels of a few watts. The SF, strongly absorbs the laser radiation, which is then con- verted via rapid inter- and intra-molecular relaxation into heat. The low thermal conductivity of SF, ensures lhat this produces a highly inhomogeneous temperature profile, in which the centre of the cell may be heated up to 1500 K, but where the cell walls remain at room temperature. This has a number of advantages. The first is that pyrolysis is initiated unambiguously in the gas phase, eliminating the complications frequently introduced by competing surface reaction. The second is that the primary products of pyrolysis are ejected into cool regions of the cell, where they are protected from further reaction.In favourable cases, these products may be collected as less volatile liquids or even solids. On the other hand, the temperature of the pyrolysis is neither uniform nor easily measured, so that comparisons with conventional methods of pyrolysis must be made with care. Such indications as are available through studies of systems with well known kinetic parameters (e.g.CH3C0,CH3) suggest that the overall cell reaction is almost entirely that at the maximum tempera- ture. Reaction is monitored in the first instance through FTIR spectroscopy, with further product identification by NMR, GC-MS, or elemental analysis. Stirred Flow Reactor Kinetic Measurements The SFR consists of a spherical quartz vessel of volume 10cm3, at the centre of which is either a smaller perforated bulb or a simple jet inlet, which provides rapid mixing and thermal equilibration of the reactant.The reactor is housed in a conventional furnace capable of providing temperatures up to the softening temperature of the quartz. Rather than a continuous flow, reactant diluted in a carrier gas of N, or He is admitted in a pulse or batch mode; this method confers the benefit of economy as well as technical advantages. Reaction in the vessel competes with the sweeping out of reagents and products, so that a controllable proportion of conversion may be achieved. Unreacted starting material and products may be analysed directly via GC-MS, or accumulated in cold traps for subsequent investigation.The output of the flame ionis- ation or thermal conductivity detector of the GC may also be fed directly into a data-capture station for kinetic or other analysis. This kind of technique has been used extensively in the investigation of pyrolysis mechanisms of organosilicon compounds such as Me3Si-SiMe,.l4 Matrix Isolation EPR Spectroscopy Two approaches have been utilised for the matrix isolation of free radicals in pyrolysis reactions. In the first, reagents are pumped through a conventional resistively heated hot-wall quartz tube by means of a mercury diffusion and rotary pump, at pressures of much less than 1Torr. Pyrolysis products are condensed onto a finger cooled to 77 K by liquid nitrogen; the whole cold finger assembly is removable for the examin- ation of EPR or other spectra.Spectra were stored digitally on computer for subsequent manipulation. Radicals may be trapped in a matrix of unreacted starting material, or of a suitable host such as adamantane; the latter usually provides more easily interpreted isotropic EPR spectra, but does add another variable to the experimental arrangement. Matrices sufficient to produce very strong EPR spectra can be con- densed in 5-30 min, depending on flow rates and pressures, Recently, a second approach has been used, in which the conventional heater is replaced by a laser pyrolysis cell of the sort described above. This approach permits the differentiation of radicals produced in homogeneous and surface processes, and hence additional insights into the mechanisms of depos-ition and other reactions.Many of the EPR spectra obtained arose from two or more species; these could be distinguished by judicious variation of experimental conditions (temperature or pressure), followed by computer subtraction. The two techniques have been used in the identification of many radicals in main-group organometallic pyrolyses, for example Me radicals from Me,Ga or Et radicals from Et,In.15 Results We have applied the techniques described above to the study of a number of organo-transition-metal systems of current or potential use in MOCVD or MBE applications. In addition, we have made considerable use of the armoury of techniques widely employed in mechanistic investigations: isotope substi- tution, intermediate trapping, etc.Each of the systems will be J. MATER. CHEM., 1994, VOL. 4 described in turn, and some of the trends discernible at this stage will be discussed in the next section. MeMn(CO), and AcMn(CO), As mentioned above, pentacarbonylmethylmanganese (MMP) has been utilised as a source for Mn doping in CdMnTe epilayer~,~but little is known of its mechanism of decompo- sition. Furthermore, co-pyrolysis of MMP with organosilicons containing Si-H bonds, such as Me,SiH, has potential for the deposition of MnSi layers in applications such as intercon- nects. Finally, the mode of decomposition of acetyl manganese pentacarbonyl (AMP) is of considerable theoretical interest. We have therefore undertaken investigations of the pyrolysis of both MMP and AMP and their deuteriated derivatives, both alone and in the presence of Me,SiH, using the techniques described in the previous section.In addition, three of us have collaborated in spectroscopic investigations of the gas phase and crystal structure of MMP, using IR laser spectroscopy16 and neutron diffra~tion.’~ MMP was prepared by minor modifications of the literature method [conversion of Mn,(CO)lo to NaMn(CO),, followed by reaction with CH,I], and deuteriated MMP similarly, using CD,T in the last stage.’* AMP and deuterated AMP were prepared similarly, using CH,COCl or CD,COCl.SFR pyrolysis yielded methane and Mn,(CO),, as the only products detectable by GC-MS, in accord with the results of earlier photolysis experiments.” Initially, kinetic studies exhi- bited non-Arrhenius behaviour over the temperature range 15O-26O0C, a clear indication of the unpredictable effects of surface reaction. Subsequent studies with a clean vessel yielded first-order kinetic behaviour, with linear Arrhenius behaviour over the range 190-230 “C corresponding to an activation energy of 214(9) kJ mol-’ for the production of CH,. Co-pyrolysis of CH,Mn(C0)5 and CD,Mn(CO), using IR LPHP yielded CH,, CH,D, CD,H, and CD, as isotopic variants of methane; the lack of CH,D, is strong support evidence for the involvement of Me radicals.Using conven- tional MI, very strong EPR spectra identical to those assigned to Mn(CO), radicals by Symons and Sweany20 were obtained, as shown in Fig. 1; identical spectra were also produced by pyrolysis of Mn,(CO),,. No features arising from Me radicals were detected; that this absence was not simply an artefact of the design of the pyrolysis apparatus was confirmed by the observation of very strong Me radical spectra generated in the pyrolysis of Me,Ga.” I 1I I I 31 50 3275 3400 3525 3650 HIG Fig. 1 EPR spectrum of matrix-isolated Mn(CO), radicals produced in the hot-wall pyrolysis of MeMn(CO), J. MATER. CHEM., 1994, VOL. 4 The addition of Me,SiH produced both quantitative and qualitative changes in the reaction.IR LPHP of MMP-Me,SiH mixtures resulted in methane and Me,SiMn(CO), as the sole products; isotope labelling verified that the methane arose from the Mn-Me and the Si-H units. In kinetic SFR studies, addition of Me,SiH substantially increased the rate of pyrolysis, the activation energy falling to 187(13) kJ mol-’; paradoxically, the rate law was zero order in the pressure of the silane, although it remained first order in MMP. In addition, minor quantities of HMn(CO), and Me,Si were detectable using GC-MS. Co-pyrolysis of MMP with Me,SiH, was carried out in the hope of producing as an intermediate Me,Si(Mn(CO),} ’; in the event, only traces of this potential 2: 1 Mn-Si precursor of were detected, the major product being Me,SiHMn(CO),.AMP produced similar results, the initial step in pyrolysis alone being extrusion of CO to yield MMP, which then decomposed as above. In the MI/EPR study, AMP produced Mn(CO), radicals at a significantly lower temperature than MMP (95 ’C as opposed to 180 “C), consistent with the lower laser power required in the LPHP investigations; again, no CH,CO or CH, radicals were detected. Co-pyrolysis of AMP with Me,SiH produced a result significantly different from a previous investigation; in the present study, the only products were Me,SiMn(CO),, CO and CH,, whereas Me,SiMn(CO), and CH,CHO were produced in toluene solution.” C,H,Mn(CO), and MeCSH,Mn(CO), Tricarbonyl(cyclopentadieny1)manganese (CMP) and its methyl derivative (MCMT), frequently known as cymantrene, were among the earliest organo-transition-metal derivatives to be explored for use as MOCVD precursors.’ However, there are problems with their use, principally the incorporation of carbon, and it is therefore of interest to investigate their mode of decomposition.An SFR study of MCMT revealed that the major products of pyrolysis were CO and free methylcyclopentadiene. The kinetic data afforded a good straight-line plot over the tem- perature range 250-390 “C, with an activation energy of 208( 14) kJ mol-’. The hot-wall MI EPR spectra of both sets of pyrolysis products were complex, with at least two contribu- ting species. However, the relative proportions of these could be altered by simply varying the temperature; computer subtraction produced the spectra shown in Fig.2 in the case of CMT. Fig. 2(a) is clearly identifiable as arising from the cyclopentadienyl free radical,2’ but the highly anisotropic spectrum of Fig.2(b) has not yet been identified. MCMT produces a complex spectrum identifiable as arising from the methylcyclopentadienyl radical, overlapped by a contribution almost identical to that of Fig. 2(b).The anisotropic spectrum almost certainly does not contain the organic moiety, there- fore, and its marked shift in g-value from the free-spin value strongly suggests that it contains a metal, presumably Mn. In addition to cyclopentadienes, IR LPHP at laser powers of 1.5-2.0 W for both CMT and MCMT produced hydro- carbons not observed in the SFR system, principally ethyne.This observation was unexpected; although ethyne is often a thermodynamic ‘sink’ in many hydrocarbon reactions, its generation usually requires very much higher temperatures than those employed in the LPHP experiment (estimated at 450-500 K). It would appear that reaction of the cyclopen- tadienyl unit in situ may be responsible. If we assume that the origin of the observed cyclopentadiene is hydrogen abstraction from the parent compound by a released cyclopentadienyl radical, as suggested by the EPR observations, then the resultant Mn-C5H, unit may rearrange in an electrocyclic fashion to yield ethyne and perhaps the species responsible 3300 3350 3400 3450 35 3300 3350 3400 3450 3500 HIG Fig.2 EPR spectra of products of hot-wall pyrolysis of C,H,Mn(CO),: (a) cyclopentadienyl, (b) unidentified anisotropic species, possibly MnC for the anisotropic EPR signal.This hypothesis was tested by co-pyrolysing CMT with deuteriated MMP. The temperature required for release of a CD, radical from deuteriated MMP is lower than that required to initiate reaction of CMT, and CD, radicals are relatively good hydrogen abstracters; this system would therefore produce the conjectural Mn-C,H, centre via a different route. The co-pyrolysis did indeed produce ethyne at the lower temperature, as well as a quantity of CD,H, providing substantial support for this hypothesis. Butadienyltricarbonyliron To date, butadienyltricarbonyliron (BDFECO) has not found use as a MOCVD precursor.However, its mode of pq,rolysis is of fundamental interest, since it is a prototypical q-bonded diene compound of the kind suggested as intermediates in heterogeneous catalytic processes23. There has been one reported study of conventional pyrolysis of BDFEC0,’4 in which the major hydrocarbon products were free butadiene together with its oligomers and polymers (largely vinylcyclo- hexene). SFR experiments resulted in similar products, with a preliminary value for the activation energy of 136(6)kJ mo1-’, close to the estimated Fe-CO bond strength. A GC-MS trace of the results of pyrolysis at 500 K is shown in Fig. 3(a). IR LPHP was carried out at laser powers ranging between 0.5 and 1.5 W, corresponding to similar pyrolysis temperatures of 450-550 K.This resulted in a deposit analysed as approximately FeC,H, (presumably J. MATER. CHEM., 1994, VOL. 4 3 Q)0 \ 13 i4 1 2 3 4 5 6 timdmin Fig. 3 GC-MS traces of products of pyrolysis of C,H,Fe(CO), from (a)the stirred flow reactor, and (b) IR laser powered homogeneous pyrolysis; (1) starting material, (2) benzene, (3) butadiene, (4) toluene; unidentified peaks arise from higher aromatics iron plus butadiene polymers) and free CO. In addition, a range of hydrocarbon products was identified by combined use of FTIR, 'H NMR, and GC-MS, with molar ratios as follows: butadiene ( l.O), benzene (OS), but-1-ene (0.3), cis-but-2-ene (0.3), trans-but-2-ene (0.2), and ethene (O.l), together with traces of toluene and higher aromatics.A GC-MS trace of the results of pyrolysis at 500 K is given in Fig. 3(b)for comparison (note that relative proportions cannot be obtained directly from such spectra, because of differing sensitivities). It is very evident from the more complex spectrum of Fig. 3(b)that homogeneous pyrolysis results in a much greater range of hydrocarbon products than the conventional approach, underlining yet again the dominance of wall pro- cesses in the latter, even under conditions designed to encour- age homogeneous reaction. The mode of production of the additional products is currently under investigation through the use of selectively deuteriated and substituted butadiene complexes.25To date, no EPR spectra of the pyrolysis products of BDFECO have been obtained.Ferrocene Like BDFECO, ferrocene has found little use as an MOCVD precursor; its vapour pressure is rather low, and pyrolysis requires rather high temperatures. However, its mode of decomposition is of interest as a non-carbonyl compound, and as a paradigm for other cyclopentadienyl compounds. To date, only IR LPHP investigations have been conducted. Because of the low vapour pressure of the compound, solid ferrocene was sublimed into the pyrolysis cell, and experiments conducted with the cell maintained at a temperature of 100-150°C by means of resistive heating tape. IR LPHP required rather high laser powers of 8-10 W, corresponding to temperatures of 700-750 K. FTIR spectroscopy [see Fig.4(a)] indicated the production of ethene, ethyne and benzene, as well as free cyclopentadiene; 'H NMR spec-troscopy further revealed the production of considerable quantities of naphthalene [Fig. 4(b)], an observation con-firmed by GC-MS. Discussion and Conclusions The results presented above provide a range of examples of the way in which considerable insight into the mechanisms of 2 1 1 I I 1 I I I 1400 1200 1000 800 wavenumber/cm-' 33 I I I I I 1 7.60 7.20 6.80 6.40 6.00 5.60 5.20 6 Fig. 4 Products of IR laser-powered homogeneous pyrolysis of ferro-cene: (a) FTIR spectra before (lower) and after (upper), (1) cyclopen-tadiene, (2) benzene, (3) ethyne, unidentified peaks =starting material of SF,; (b) partial 'H NMR spectrum, (1) naphthalene, (2) benzene, (3) cyclopentadiene, (4) ethene, unidentified =residual protonated solvent pyrolysis of organo-transition-metal complexes can be obtained by a combination of experimental techniques.The differences found in the SFR and IR LPHP experiments confirm the widely held view that under conventional pyrolysis conditions (which would include MOCVD, and with even more force, MBE) reaction is likely to be dominated by surface processes. Nonetheless, there may be significant contri- butions from homogeneous reactions, which may well play a part in producing deposits of undesirable characteristics; this kind of process is thought to be the origin of the carbon found in films deposited from mixtures containing Al-Me precursors, for example.4J3 Although firm conclusions about the mechanisms of individ-ual complexes must await the results of further investigations, we may make some general observations at this stage. It is to be anticipated, of course, that the initial step in the pyrolysis of all the complexes investigated here will normally be homol- ysis of the weakest bond in the system, unless a lower energy intramolecular route is available.In the case of MMP, this implies breakage of the Me-Mn link, a contention supported by the EPR and IR LPHP observations; for AMP, expulsion of CO is the initial step, followed by a similar pathway. The kinetic results in the presence of Me,SiH suggest that isomeris- ation of MMP may be rate-determining, the isomer perhaps being the unsaturated acyltetracarbonyl proposed to account for the ready CO exchange exhibited by this compound, and tentatively identified in our study of its high-resolution IR spectrum.16 Our failure to detect organic radicals in either system is puzzling; it may be that Me radicals are lost at the wall in the predominantly heterogeneous reaction, and that J.MATER. CHEM., 1994, VOL. 4 Me (or even CH3CO) radicals may be observed using IR LPHP combined with MI/EPR. The primary loss of the organic moiety will in this case result in a deposited metal film largely free of deleterious carbon, in keeping with obser- ~ation.~For the more strongly q-bonded systems, loss of CO (where available) will be the more facile pathway, resulting in coordinatively unsaturated species of the sort widely observed in low-temperature photolyses of such systems;26 for ferrocene, initiation of reaction presumably requires the even higher energy loss of the q5-cyclopentadienyl unit, in keeping with the higher temperatures required.The subsequent reactions of resultant coordinatively unsaturated species are largely unknown, despite the very considerable interest in such inter- mediates through their role in catalysis.23 Our results would suggest that the q-bonded units may undergo in situ rearrange-ment, resulting in products more normally associated with photolytic processes involving the free moiety. In the case of BDFECO, for example, the major products may result from an intramolecular hydrogen transfer in a bisbutadiene complex ( butenes). Alternatively, an in situ electrocyclic rearrangement of butadiene in the mono-complex may result in the ethene and ethyne observed in the photolysis of free b~tadiene,~’ with the latter undergoing the familiar and facile Fe-catalysed cyclisation to benzene.For CMT and MCMT, electrocyclic rearrangement of the native compound would result in ethyne together with a manganese-carbon unit; for the hydrogen abstraction product C,H,Mn(CO),, this may result in manga- nese carbide itself. Catalysis of the cyclisation of ethyne by Mn is less facile, so that benzene is not observed in this case; manganese carbide may be the origin of the anisotropic EPR signal of Fig.2. The observation of naphthalene in the pyrol- ysis of ferrocene strongly suggests a rearrangement involving both cyclopentadienyl units, a hypothesis under investigation with the use of substituted ferrocene~.~’ As a general obser- vation, it would appear that strongly q-bonded complexes are less likely to be useful MOCV D precursors as a consequence of extensive carbon incorporation resulting from ligand rearrangement processes. On the other hand, a-bonded organo-transition-metal carbonyls have more promise, as has already been demonstrated. We thank the SERC for their extensive support for this work, including equipment grants (to D.K.R., J.B.R. and I.M.T.D.), post-doctoral fellowships (to M.P., I.M.P.and A.D.W.), and studentships (to G.P.M.). We also acknowledge support from the Government of Turkey through a studentship to S.S. References G. B. Stringfellow, Organometallic Vapor Phase Epitaxy-Theory and Practice, Academic Press, San Diego, 1989, and references therein. 2 A. Nouhi and R. J. Stirn, Appl. Phys. Lett., 1987,51, 2251. 3 G. N. Pain, N. Bharatula, G. I. Christiansz, M. H. Kibel, M. S. Kwietniak, C. Sandford, R. S. Dickson, R. S. Rowe, K. McGregor, G. B. Deacon, B. 0. West, S. R. Glanvill, D. G. Hay, C. J. Roussow and A. W. Stevenson, J. Cryst. GroMth, 1990, 101, 208; G. N. Pain, G. I. Christiansz, R. S. Drckinson, G. B. Deacon, B. 0. West, K. McGregor and R. 5. Rowe, Polyhedron, 1990,9,921. 4 G.A. Atiya, A. S. Grady, S. A. Jackson, N. Parker and D. K. Russell, J. Organomet. Chem., 1989,378,307. 5 A. S. Grady, A. L. Mapplebeck, D. K. Russell and M. G. Taylorson, J. Chem. SOC., Chem. Commun., 1990,929. 6 A. S. Grady, R. D. Markwell and D. K. Russell, J. Chcm. SOC., Chem. Commun., 1991,14. 7 A. S. Grady, R. E. Linney, R. D. Markwell and D. K. Russell, J. Muter. Chem., 1993,3,483. 8 A. S. Grady, R. E. Linney, R. D. Markwell, G. 1’. Mills, D. K. Russell, P. J. Williams and A. C. Jones, J. Muter. Chem., 1992, 2, 539. 9 R. E. Linney and D. K. Russell, J. Muter. Chem., 1993,3, 587. 10 A. C. Baldwin, I. M. T. Davidson and A. V. Howard, .I. Chem. SOC., Faraday Trans. I, 1975,71,972;I. M. T. Davidson, (3. Eaton and K. J. Hughes, J. Organomet. Chem., 1988,347,17.11 K. Mach, I. Novakova, V. Hanus and J. B. Raynor, Tetrahedron, 1989, 45, 843. 12 D. K. Russell, Chem. SOC.Rev., 1990, 19,407. 13 D. K. Russell, Coord. Chem. Rev., 1992, 112, 131. 14 I. M. T. Davidson and A. V. Howard, J. Chem. SOC., Faraday Trans. I, 1975,71,69. 15 G. P. Mills, J. B. Raynor, D. K. Russell and A. D. norkman, unpublished results. 16 J. Gang, M. Pennington, D. K. Russell, P. B. Davies, G. M. Hansford and N. A. Martin, J. Chem. Phys., 1992,97,3885. 17 M. Pennington, D. K. Russell, D. R. Russell and S. Saydam, work in progress. 18 R. B. King, in Organometallic Synthesis, ed. J. J. Eisch and R. B. King, Academic Press, San Diego, 1965, vol. 1. 19 T. E. Gismondi and M. D. Rausch, J. Organomet. Chem, 1985, 284, 59. 20 M. C. R. Symons and R. L. Sweany, Organometallics, 1982,1,834. 21 R. W. Wegman, Organometallics, 1986,5,707. 22 M. Kira, M. Watanabe and H. Sakurai, J. Am. Chem. Soc., 1980, 102,5202. 23 A. J. Pearson, in Comprehensive Organometallic Chemistry, ed. G. Wilkinson, F. G. A. Stone and E. W. Abel, Pergamon Press, Oxford, 1982, vol. 8, ch. 58; A. J. Deeming, in comprehensive Organometallic Chemistry, ed. G. Wilkinson, F. G. A. Stone and E. W. Abel, Pergamon Press, Oxford, 1982, vol. 4, ch. 31.3. 24 D. L. S. Brown, J. A. Connor, M. L. Leung, M. I. Paz-4ndrade and H. A. Skinner, J. Organomet. Chem., 1976,110,79. 25 D. K. Russell and S. Saydam, unpublished results. 26 S. P. Church, M. Poliakoff, J. A. Timney and J. J. Turner, J. Am. Chem. SOC.,1981,103,7515. 27 I. Haller and R. Srinivasan, J. Am. Chem. Soc., 1966, 138, 3694; W. J. Leigh, Chem. Rev., 1993,93,487, and references therein. Paper 3/04314B; Received 21st July, 1993