Mechanism of Thermolysis of Hexamethyldisilane and the Silicon-Silicon Bond Dissociation Energy BY IAIN M. T. DAVIDSON AND ANTHONY V. HOWARD Department of Chemistry, The University, Leicester LE1 7RH Received 25th April, 1974 A detailed investigation of the thermolysis of hexamethyldisilane in a novel stirred-flow system at temperatures between 770 and 872 K is described. Evidence is presented for a mechanism which accounts for the different behaviour of hexamethyldisilane on thermolysis under different conditions, and a new value of 337 kJ mol-I for D(Me3Si-SiMe3) is deduced. The silicon-silicon bond dissociation energy in hexamethyldisilane (HMDS) is a key quantity on which the values of bond dissociation energies in other organosilicon compounds have been based,l but attempts to measure it by electron impact and by gas kinetics 3 9 have led to values ranging from 201 to 359 kJ mol-l.This spread of values is partly due to the complex behaviour of HMDS on thermolysis, as revealed by several gas kinetic s t ~ d i e s . ~ - ~ At high pressures of HMDS the reaction is relatively simple, the isomeric trimethylsilyl(dimethylsilyl)methane being formed as the sole main product in a radical chain reaction of well-established mechani~m.~'~ At low pressures, however, other products predominate, the product composition and kinetic behaviour being strongly dependent on experimental condition^.^. 4 9 In a static system with mass spectrometric analysis of reaction products, kinetic data were obtained which were interpreted as relating to a radical non-chain mechanism, so that the observed Arrhenius parameters could be identified with those for the initial rupture of the silicon-silicon bond.4 The bond dissociation energy thus obtained was plausible and, in conjunction with electron impact data,g led to values for silicon- methyl and silicon-hydrogen bond dissociation energies which agreed well with those deduced from the thermolysis of trimethylsilane if that thermolysis was also inter- preted as non-chain.' O Although the overall picture was thus reasonably consistent,l some worrying discrepancies became apparent, notably a suspiciously low A factor for silicon-silicon bond rupture (particularly in view of rotating sector experiments on the recombination of trimethylsilyl radicals which gave normal values 9 2, and evidence for a chain mechanism for the formation of methane in the thermolysis of tetramethyl- silane,l in contrast to the supposed non-chain mechanism for the formation of meth- ane in the very similar compound, trimethylsilane. O The thermolysis of HMDS at low pressure therefore merits thorough re-investiga- tion.This paper, enlarging on our preliminary account,14 describes the results of this re-investigation and puts forward a detailed mechanism which accounts for the apparently divergent results described above. EXPERIMENTAL APPARATUS For this re-investigation, we have developed a " pulsed stirred-flow " technique which The technique uses a offers several advantages over conventional methods in gas kinetics. 6970 THERMOLYSIS OF HEXAMETHYLDISILANE quartz stirred-flow reactor, identical in design to that of Mulcahy and William~,’~ but substantially smaller with a total volume of 54.58 cm3.The reactor was housed in a smoothed electric furnace of conventional design. A dried and regulated stream of nitrogen flowed through a sample valve,16 through the reactor, and then through a gas chromatograph. The sample valve was connected to a simple greaseless vacuum line with provision for the storage of reactants and samples of products. This vacuum line was fitted with a pressure trans- ducer, enabling the pressure of vapour in the sampling volume of the valve to be measured. A pulse of reactant vapour of known quantity could thus be injected into the carrier gas stream and carried into the reactor where complete mixing and partial reaction occurred ; the reaction mixture then entered the gas chromatograph for analysis.Thus, one stream of nitrogen served as the kinetic and analytical carrier gas, the flow rate determining the reaction time. This apparatus is very much simpler than a conventional flow system for gas kinetics and eliminates problems due to minute leaks, since the reactor is always at above atmospheric pressure. The technique is of course very economical in reactant compared to the conven- tional continuous flow techniques. As we have found in previous flow experiments with organosilicon compounds,” rigorous purification of the carrier gas is necessary to avoid the formation of siloxanes. The nitrogen carrier gas was dried by passage through a molecular sieve column, then passed through traps containing molten sodium.The latter were subsequently replaced by an ‘‘ Oxy-Trap ’’ column. To analyse the wide range of products formed, the technique of fractional co-distillation was used, with a fore-column which could be cooled to 113 K followed by the main column, 3 m long and 0.64 cm 0.d. packed with 10 % wlw squalane on 60-100 mesh Embacel. This column and the gas density balance detector were housed in an oven at 354 K. The kinetic situation in a pulsed reactor appears to be greatly different from that in a continuous flow reactor, but similar kinetic equations may be derived provided that the rate of reaction is low compared to the rate of expulsion of the pulse from the reactor, i.e. provided that the degree of decomposition is small.Under these conditions, with perfect mixing in the reactor, the basic equation is for the formation of product P from reactant A in a reaction of order n and rate constant k. In the equation, (P) and (A) are the number of moles of P and A measured by the gas chromato- graph, Vis the volume of the reactor andz = V/u, where u is the volumetric flow rate through the reactor. z was about 40 s in most experiments, but was varied between 29 and 112 s. Perfect mixing in the reactor was verified by placing a thermal conductivity detector at the outlet and observing the shape of the response curve when a chemically inert pulse was in- jected. Details of this test and of the derivation of the kinetic equation for the reactor will be published separately.’* For a first order reaction, eqn (i) with n = 1 is valid for any degree of decomposition.MATERIALS HMDS was prepared by treating trimethylchlorosilane (a gift from Dow Corning) with potassium-sodium alloy in ethylbenzene, followed by fractionation and purification by preparative g.l.c6 The isomeric trimethylsilyl(dimethylsi1yl)methane was prepared from HMDS by thermolysis.6 Commercial rn-xylene (B.D.H. L.R. Grade) was purified by pre- parative g.1.c. lY1,3,3-tetramethyl-1 ,3-disilacyclobutane was a gift from I.C.I. Ltd. Other compounds were obtained commercially. The purity of all compounds was determined by mass spectrometry and by infra-red spectroscopy, n.m.r. and refractive index where appro- pr ia te . RESULTS The thermolysis of HMDS was studied at temperatures between 770 and 872 K in the pulsed flow reactor with nitrogen carrier gas above atmospheric pressure.Low pressure pulses of HMDS were used, corresponding to initial concentrations in theI . M. T . DAVIDSON A N D A . V. HOWARD 71 range 1.4 x to 5.5 x mol ~ m - ~ . The products were trimethylsilane (3MS) ; tetramethylsilane (4MS) ; trimethylsilyl(dimethylsilyl)methane (ISO), the isomerisa- tion product ; and 1,1,3,3-tetramethyl-l,3-disilacyclobutane (TMDS). A little meth- ane was also observed at the highest temperature. Excess of rn-xylene was added in a second series of experiments covering the same range of temperature and composition, substantially altering the kinetic behaviour and product composition. Throughout the work, the main kinetic emphasis was on the measurement of the rate of formation of 3MS, but kinetic data over the whole range of conditions were also obtained for the formation of 4MS and over restricted ranges with less accuracy for the other products.Without added rn-xylene, the thermolysis of HMDS alone was found to be considerably more complex kinetically than had previously been supposed. The formation of 3MS and TMDS followed the same rate law, which was about first order at low temperature, but became of higher order at higher temperature particularly at low concentration of HMDS ; the average order was about 1.3. At higher concen- trations and lower temperatures than were used in this work it is known that the isomerisation of HMDS has an order of 1.5. Whilst the order for isomerisation was about 1.5 in the present investigation around 770 K, it increased with increase in temperature, being about 1.8 at the upper end of the temperature range.These results will be analysed in more detail in the Discussion. The formation of 4MS on the other hand was kinetically simple, first order kinetics being obeyed under all conditions. First order rate constants were calculated for the formation of 3MS (and TMDS) and 1.5 order rate constants for isomerisation, to facilitate comparison with earlier results 4 9 although of course no quantitative deductions can be made from such " rate constants ". Arrhenius parameters are collected in table 1, error limits only being appropriate for the formation of 4MS. TABLE 1 .-ARRHENIUS PARAMETERS FOR THE FORMATION OF PRODUCTS IN THE THERMOLYSIS OF HMDS WITHOUT ADDED m-XYLENE product log A* E/kJ rnol-' apparent order t true order TMDS 3MS } 15.3 295 1 N 1.3 IS0 13.4 207 1.5 -1.8 4MS 13.75 0.7 282+ 12 1 1 .o * first order rate constants and A factors in s-' and second order in cm3 mol-' s-l.t see text. When excess rn-xylene (up to 300-fold) was added, the formation of TMDS was completely suppressed and 3MS was formed at a reduced rate in a first-order reaction (under our experimental conditions m-xylene was thermally stable). The rate of the isomerisation was also reduced and its order was about 1.5. No change was observed, however, in the kinetics of formation of 4MS. First order rate constants for the formation of 3MS in excess of m-xylene were given by log,, k/s-l = (17.53k0.25)- (336.654.0) kJ mol-'/2.303 RT.Although a substantial excess of m-xylene was usually present, relatively small amounts were also found to produce the above effects. DISCUSSION The simplest explanation of the above results is provided by schemes 1 and 2 ; the approximate Arrhenius parameters in scheme 1 will be estimated below. The formation of 4MS will be discussed separately later.72 THERMOLYSIS OF HEXAMETHYLDISILANE SCHEME 1 .-THERMOLYSIS OF HMDS WITHOUT ADDED W~-XYLENE log A EIkJ rno1-I Me3SiSiMe3-+2Me3Si- (1) 17.25 337+ Me3Si-+ Me3SiSiMe3-+Me3SiH+ Me5Si2cH2 (2) 13.2 72 Me3SiCH2SiMe2-+ Me3Si-+ Me2Si=CH2[-+Me2Si SiMe2] (4) 14.6 212 Me3SiCH2SiMez+ Me6Si2-+ Me3SiCH2Si(H)Me2+ Me5Si2cH2 (5) 13.2 80 0.3 4 Me5Si2cH2 -+ Me3SiCH2Si Me, (3) /\ v 2Me3Si--+ Me3SiSiMe3 (6a) 13 0 Me3Si.+ Me3SiCH2SiMe2+ Me3SiCH2Si2Me5 (6b) 12.4 0 2Me3SiCH2SiMe2 -+(Me3SiCH2SiMe2)2 (6c) 11 0 Arrhenius parameters without error limits are approximate estimates.Scheme 1 offers an explanation for the different product composition observed at high and low pressures of HMDS. At high pressures, reaction ( 5 ) is much faster than (4) and the dominant chain sequence is (l), (2), (3) and (5), as has previously been established,6 giving IS0 as the main product with only small quantities of 3MS. At low pressures of HMDS, reaction (4) can compete with (5), pro- pagating a chain reaction producing TMDS and increased quantities of 3MS (TMDS is assumed to result from the dimerisation of the double-bonded intermediate >- Me2Si=CH, 1,10.13,19 SCHEME 2.-THERMOLYSIS OF HMDS WITH ADDED W2-XYLENE (RH) Me3 SiSiMe3 -+2Me3Sie (1) Me3Si*+ RH+Me3SiH+ Re (7) Re+ Me3SiSiMe3-+RH+ Me5Si2CH2 (8) RH+ Me5Si2cH2+R*+ Me3SiyiMe3 (- 8) Me5Si2cHz-+ Me3SiCH2SiMe2 (3) (9) 2R*-+R2 (10) Me3SiCH2SiMe2 + RH-, Me3SiCH2Si(H)Me2 + Re With added m-xylene (scheme 2), chain isomerisation of HMDS continues, but by a different sequence involving xylene and xylyl radicals in reactions (7), (8), (9) and (lo), giving different kinetic behaviour.Formation of 3MS, however, is now non-chain, with the rate equalling 2kl[HMDS]. The experimental rate constant for the formation of 3MS in rn-xylene thus equals 2kl, and hence the Arrhenius parameters for reaction (1) in scheme 1 are obtained, the activation energy of 337 kJ mol-1 being identified with D(Me,Si-SiMe,). This value of E l , with those for E2 and E5 estimated previously,6 enabled the activation energy for the isomerisation of HMDS at higher concentration to be calculated as 249 kJ mol-I in good agreement with the experimental value of 251 & 8 kJ mol-I.Estimation of A factors is rather more difficult. The rate of combination of trimethylsilyl radicals has been measured in the gas phase and in solution. No activation energy was observed and log A6a was 14.26 in the gas phase and 12.74 in Whilst the former figure seems to be rather high, there is little doubt that trimethylsilyl radicals have a comparable recombination rate to methyl radicals. Log A l -log ACia has been put at 4.2 from estimates of the entropy of HMDS and the trimethylsilyl radical.ll Hence log A,, = 13.0 from our value of 17.2 for log A , .I .M. T. DAVIDSON AND A . V. HOWARD 73 There were indications from the kinetic study of the isomerisation of HMDS at higher concentration that the A factor for reaction (6c) should be as low as possible.6 A reasonable estimate in the light of current ideas on recombination would be log A 6c = 11, whence log A66 = 12.4 by the geometric mean rule. To reconcile the experi- mental results obtained in this work and in the isomerisation study at higher concen- tration with scheme 1 it is necessary to postulate that reactions (2) and (5) have rela- tively high A factors. There is no reliable experimental evidence on these,6 and we simply estimate log A2 = log A5 = 13.2. These estimates may well require revision as more information becomes available on abstraction and combination reactions of silicon-containing radicals.The relative rates of reactions (4) and (5) may be deduced from the relative yields of TMDS and I S 0 in our experiments. Since reaction (4) is first order and (5) is second order, it follows from eqn (i) that V(IS0) (ii) 5- - 2V( ISO) - k4 (Me,Si=CH,)(HMDS) - (TMDS)(HMDS) where the bracketed quantities are number of moles and not concentrations. The most consistent product analyses over a range of (HMDS) were found at 795 and 820 K, giving average values for k5/k4 of 1.894 x lo7 and 1.033 x lo7 respectively. These figures, with the Arrhenius parameters for reaction (5) estimated above, gave the values of E4 and log A4 shown in scheme 1.When m-xylene was added in excess, the rate of formation of 3MS was reduced by a factor varying from 4.5 at 770 K to 2.0 at 872 K, indicating that the formation of 3MS in the absence of m-xylene is a chain reaction of short chain length. For this chain reaction the approximate Arrhenius parameters in scheme 1 may be used to assess the relative importance of the three termination reactions from th? relative concentrations of the chain-carrying radicals, given by [Me,Si-]/[Me,SiCH, SiMe,] = k4/k2[HMDS], and hence to predict the order and Arrhenius parameters under different conditions. Over most of the range of temperature and concentration of HMDS, termination by (66) would be the most important, whence formation of 3MS would be first order, with rate constant (klk,k4/k6b)0-5, log A-16.3, and E-311 kJ mol-l. However, termination by (6a) would be more important at high temperature and low concentration, giving a 1.5 order reaction with rate constant (k,kg/k6a)0-5, log A z 15.3, and Ez241 kJ mol-l.The chain length of this 1.5 order reaction would drop below unity at the lowest concentrations of HMDS ; for example, the cut-off at 870 K would be below 3 x mol ~ m - ~ . Termination by (6c) is only important at low temperature and high concentration; this termination would give an order of 0.5, but the chain would be slower than reaction (1) under these conditions and this process can therefore be ignored. Formation of 3MS in the absence of m-xylene would thus be expected to have an order of either 1 or 1.5 in HMDS, depending on the experimental conditions.The extent to which the experimental results were consistent with this prediction was examined by comparing experimental points with lines of slope 1.5 or 1.0 and intercepts calculated from the appropriate Arrhenius parameters on a plot of log(3MS) against log(HMDS). For the regions of temperature and concentration where 1.5 order behaviour would be expected, experimental points were corrected by subtracting the amount of 3MS formed in reaction (I) from the total observed, to take account in a simple way of the cut-off discussed above. Results at three typical temperatures are given in fig. 1 , showing quite satisfactory agreement, and increasing confidence in the validity of the estimates of Arrhenius parameters for reactions (2), (4), (5) and (6).It is known that formation of I S 0 in the absence of m-xylene is a clean 1.5 order74 THERMOLYSIS OF HEXAMETHYLDISILANE reaction at lower temperature and higher concentration than were used in this work, with log A = 16.65 and E = 251 kJ mol-l. These kinetic characteristics would be observed in the present work also, but at high temperature and low concentration there would be a significant contribution from a sequence terminated by (6b), which would be second order in HMDS, with rate constant (klk2k-~/k4k6b)o*s, log A = 14.9, and E= 179 kJ mol-'. This accounts for the difference in order and Arrhenius parameters for isomerisation between this work (table 1) and the earlier.6 log (HMDS) FIG. 1 .-Thermolysis of HMDS. Comparison of experimental points with calculated lines.A, 861.3 K, n = 1.5; B, 820K, n = 1; C, 778.7K, n = 1.0. At 870 K with added rn-xylene some methane was observed, the molar ratio of methane to 3MS being 0.2, corresponding to a first order rate constant of 3.7 x s-l for the formation of methane at 870 K. Methane most probably results from the minor dissociation Me,SiSiMe3 -,Me3SiSiMe2 + Me- (la). Combining our estimate of k6, with the well established rate constant for the combination of methyl radicals we calculate by the geometric mean rule a value of 1013-6 cm3 mol-l s-1 for the A factor for the combination of methyl and trimethylsilyl radicals to form 4MS. The A factor for the reverse reaction, dissociation of 4MS, Me4Si+Me3Sib + Me., may then be estimated as 10 17*6 cm3 mol-l s-l from a recent value for the entropy change.ll If Al, is assumed to have about the same value then E l , = 350 kJ mol-l, which may be identified with the silicon-methyl bond dissociation energy in HMDS.The bond dissociation energies deduced from this work, and other recent values consistent with them are in table 2. The new value for D(Me,Si-SiMe,), with aI . M. T . DAVIDSON A N D A . V . HOWARD 75 recent literature value 2 o for AH"fMe,Si,),, gives AHi(Me,Si-), = - 11 kJ mol-'. These thermochemical quantities may be used to speculate upon the mode of forma- tion of 4MS in the thermolysis of HMDS. TABLE 2.-BOND DISSOCIATION ENERGIES bond D/kJ mol-' Me3Si-SiMe3 337 la Me5Si2CH2-H 400f Me3Si-H 368 b9c-d Me3SiCH2-H 406 Me,SiCH,Si(Me,)-H 360 f Me5Si2-Me 350 a a this work ; b ref.(14) ; C I. M. T. Davidson, M. Jones, and H. F. Tibbals, unpublished work ; dR. Walsh and J. M. Wells, Chem. Cumm., 1973, 513 ; e J. A. Kerr, A. Stephens and J. C. Young, Int. J. Chem. Kinetics, 1969, 1, 339 ;f ref. (6). 4MS was the only product formed in a first order reaction with Arrhenius para- meters unaffected by added m-xylene, which suggests that it may result from a uni- molecular elimination : Me,SiSiMe, +Me,Si + Me,Si:. (1 1) Dimethylsilylene would then react with HMDS, probably yielding Me,SiSiMe,CH,- SiMe,H by insertion into one of the eighteen carbon-hydrogen bonds. (Our g.1.c. apparatus would not have produced a well-defined peak for a product of such high molecular weight.) This explanation for the formation of 4MS could reconcile the results of our present and earlier investigations of the low-pressure thermolysis of HMDS.In the earlier work,4 concentrations of HMDS below lo-* mol ~ m - ~ were thermolysed between 796 and 828 K. Kinetic data were obtained by measuring the rate of increase of a distinctive peak at m/e 203 in the mass spectrum of the reaction mixture. This peak, due to Me5Si,CH,SiMe2+, was believed to come from Me5Si2CH,SiMe3, a radical combination product formed in the final step of a radical non-chain sequence. First order rate constants were given by log k/s-l = (1 3.5 &- 1 .O)- (281.6 k9.2) kJ mol-'/2.303 RT, the activation energy being identified with D(Me,Si-SiMe,) in accordance with the above interpretation. Over the entire temperature range, the above expression gives rate constants which are smaller than k , (scheme 1) by a factor of at least two; hence neither the rate of dissociation nor a chain reaction was being observed.However, the above Arrhenius parameters are essentially the same as those obtained in this work (table 1) for the formation of 4MS, and Me,Si,CH,SiMez would be a prominent peak in the mass spectrum of the in- sertion product Me5Si2CH2SiMezH. Hence, in the earlier in~estigation,~ kinetic measurements were probably made on a product resulting from reaction (11) and not reaction (1). The chain sequence shown in scheme 1 would have proceeded concurrently with the sequence initiated by reaction (1 1) ; 3MS was indeed observed but could not be measured mass spectro- metrically in the presence of so many other similar compounds, all with prominent peaks due to Me,Si+.At the very low pressures of these experiments no isomerisation would have been expected, but Me,Si=CH, should have been formed in reaction (4). The mass spectrum of TMDS is distinctive and could easily be detected ; TMDS and other cyclic compounds were in fact formed in preliminary experiments at higher pressure but were definitely not present in the main series of experiment^.^ Probably Me,Si=CH, diffused to the wall instead of dimerising in the gas phase. since the76 THERMOLYSIS OF HEXAMETHYLDISILANE pressure of HMDS was very low and there were no added gases, unlike the present study in a stream of nitrogen. If the earlier kinetic results do relate to reaction (1 I), then Ell = 282 kJ mol-l.The activation energy for the reverse reaction, (- 11) is not known but is unlikely to be negligible.,'. 22 From quantitative data 21 on the insertion reactions of silylene, could be as low as 21 kJmol-l, giving AH,, = 261 kJmol-l. Then, from published enthalpies of formation 2o AHf"(Me,Si), = 138 kJ mol-l. This gives the second bond dissociation energy in 4MS, D(Me-SiMe,) as 288 kJ mol-l, sub- stantially less than the first dissociation energy. However, the divalent state is relatively more stable for silicon than for carbon, and there is evidence that in silane the first and second bond dissociation energies are 398 and 249 kJ mol-1 respectively.21 The quantitative model developed here thus offers a coherent explanation for our two low-pressure investigations of the thermolysis of HMDS 4* I4 despite their widely differing results.The same model has been successfully applied to the kinetic results at high pressure of HMDS, where isomerisation occurs almost exclusively.6 Finally, it permits some comment on another mode of decomposition of organodi- silanes. Sommer and his co-workers 23 while studying the photolysis of phenyl- methyldisilanes obtained convincing evidence for an intramolecular hydrogen transfer which they suggested might be an important pathway in the thermolysis of organo- disilanes. The relevant reaction for HMDS would be : + Me,SiH + Me,Si=CH,. ( W The complete suppression of the formation of TMDS by added xylene is clear evidence against the occurrence of reaction (lb) in our thermolysis of HMDS.Whilst Elb cannot be estimated accurately at present, it would almost certainly be rather less than El. However, Alb would be several powers of ten lower than A , and hence reaction (lb) would not be expected to compete with (1) in thermolysis experiments. We thank the S.R.C. for financial support, Dow Corning (Barry) Ltd. and I.C.I. Ltd. for the gift of chemicals, and Dr. D. R. Deans of I.C.I. Heavy Organic Chemicals Division for his interest and encouragement in the development of the pulsed flow technique. I. M. T. Davidson Quart. Rev., 1971, 25, 111. G. G. Hess, F. W. Lampe and L. H. Sommer, J. Amer. Chem. Soc., 1965, 87, 5327 ; J. A. Connor, B. Finney, G. J. Leigh, R. N. Haszeldine, P. J. Robinson, R. D. Sedgwick and R. F. Simmons, Chem. Comm., 1966,178. J. A. Connor, R. N. Haszeldine, G. J. Leigh and R. D. Sedgwick, J. Chem. SOC. A , 1967, 768. I. M. T. Davidson and I. L. Stephenson, J. Chem. SOC. A, 1968,282. C. Eaborn and J. M. Simmie, Chem. Comm., 1968, 1426. I. M. T. Davisdon, C. Eaborn and J. M. Simmie, J.C.S. Faraday I, 1974, 70, 249. P. J. Robinson, personal communication. S. J. Band, I. M. T. Davidson and C. A. Lambert, J. Chem. SOC. A , 1968,2068 and references therein. ' N. Sakurai, A. Hosomi and M. Kumada, Chem. Comm., 1968, 930. lo I. M. T. Davidson and C. A. Lambert, J. Chem. SOC. A, 1971, 882. l1 P. Cadman, G. M. Tilsley and A. F. Trotman-Dickenson, J.C.S. Furaday I, 1972, 68, 1849. l2 G. B. Watts and K. U. Ingold, J. Amer. Chem. SOC., 1972,94,491. l3 R. P. Clifford, B. G. Gowenlock, C. A. F. Johnson and J. Stevenson, J. Organometallic Chem., l4 I. M. T. Davidson and A. V. Howard, Chem. Comm., 1973, 323. l S M. F. R. Mulcahy and D. J. Williams, Austral. J. Chem.. 1961, 14, 534. 1972, 34, 53.I . M. T . DAVIDSON AND A . V . HOWARD 77 l6 G. L. Pratt and J. H. Purnell, Anal. Chem., 1960,32,1213. l7 G. H. Cady and D. P. Seigwarth, Anal. Chem., 1959,31,618. l 8 A. C. Baldwin, I. M. T. Davidson and A. V. Howard, to be published. l9 M. C. Flowers and L. E. Gusel’nikov, J. Chem. SOC. By 1968,419. 2o J. B. Pedley and B. S. Iseard, Catch Tablesfor Silicon Compounds (University of Sussex, 1972). 21 P. John and J. H. Purnell, J.C.S. Faraday I, 1973,69, 1455. 22 I. M. T. Davidson, J. Organometallic Chem., 1970, 24,97. 23 P. Boudjouk, J. R. Roberts, C. M. Golino and L. H. Sommer, J. Amer. Chem. SOC., 1972, 94, 7926; J. R. Roberts, Ph.D. Thesis (University of California, 1970).