J. Chem. Soc., Perkin Trans. 2, 1997 953 Ring strain energy and enthalpy of formation of oxiranone: an ab initio theoretical determination Christopher F. Rodriquez and Ian H. Williams School of Chemistry, University of Bath, Bath, UK BA2 7AY The enthalpy of formation DHo f,298 for oxiranone is estimated as 2190 ± 10 kJ mol21 by means of ab initio molecular orbital calculations at the QCISD(T)]] full/6-311G(2df,p)//MP2]] full/6-311G(d,p) level of theory, corresponding to a conventional ring strain energy of 169 kJ mol21.The QCISD(T) calculated enthalpy of formation of cyclopropanone is 6.3 kJ mol21. The oxiranone ring is probably slightly less strained than the cyclopropanone ring. Introduction Oxiranones (a-lactones) are very unstable and highly reactive intermediates which are difficult to isolate; they polymerise readily to polyesters and undergo decarbonylations, even at low temperature.1 The parent oxiranone 1 was reported in a matrix isolation study of the reaction of methylene with carbon dioxide,2 and other species have recently been formed by carboxylation of substituted carbenes in low-temperature matrices.3,4 They may be generated by photodecarboxylation of malonyl peroxides 5 or by epoxidation of ketenes.6 Only two species, 2 and 3, with electron withdrawing7 and, additionally, bulky 8 substituents have any stability at room temperature.Oxiranones have been postulated as transient intermediates in a variety of reactions,9 perhaps most notably in the alkaline hydrolysis of a-halocarboxylates in dilute solution 10 and most recently in the gas-phase pyrolysis of a-chlorocarboxylic acids.11 It has been suggested 6 that the extreme reactivity of oxiranones is due to their ease of ring-opening to a zwitterionic form 4.Theoretical studies employing the semiempirical INDO12 and MINDO/313 molecular orbital (MO) methods both found the cyclic structure 1 to be of lower energy than ring-opened forms.However, no minimum-energy structure corresponding to a closed-shell acyclic zwitterion 4 was found at the HF/3-21G* level of ab initio MO theory; high-energy zwitterionic structures in which the methylene and carboxylate moieties were coplanar disintegrated spontaneously to CH2 and CO2, whereas those in which these groups were mutually perpendicular collapsed to the cyclic oxiranone.14 Configuration interaction (CI) calculations, involving single and double excitations selected by perturbation theory, and extrapolation to the full CI limit, for various electronic states of the acetoxyl diradical 5, in both planar and perpendicular geometries, were all found to be at least 147 kJ mol21 higher in energy than the oxiranone 1.14 Our interest in the thermochemistry of oxiranones was aroused by the suggestion put forward in a recent experimental study that the spontaneous and acid-catalysed hydrolyses of the p-nitrophenyl glycoside of N-acetylneuraminic acid occur by means of a transition state involving intramolecular nucleophilic assistance by the neighbouring carboxylate group, even though this mechanism might lead to formation of a highly strained a-lactone intermediate.15 The precedent for this suggestion is the classical work, following Ingold, on nucleophilic substitutions of a-halocarboxylates:16 the observations of retained stereochemical configuration and lack of dependence on hydroxide ion concentration in the reaction of a-bromopropionate anion with dilute sodium hydroxide may be explained by the intermediacy of an a-lactone.10 How strained is such a species, and how likely is it to be formed as a reaction intermediate? To answer the question ‘What is the strain energy of an alactone?’ requires knowledge of either the enthalpy of formation or the enthalpy change for a suitable isodesmic process involving a three-membered heterocycle of known strain energy.The only value available in the literature 17 appeared to be a bond-additivity estimate of the enthalpy of formation for the parent oxiranone 1 of dubious validity.The purpose of this paper is to present the results of high-level ab initio MO calculations for the enthalpy of formation and the ring strain energy of oxiranone; results for oxirane and cyclopropanone are also presented. First, the conventional ring strain energy of oxiranone is estimated by means of an isodesmic relation to oxirane 7.Secondly, the enthalpy of formation of oxiranone is estimated by means of isodesmic relations both to oxirane and to cyclopropanone 10. Thirdly, the enthalpies of formation of 1, 7 and 10 are each evaluated directly from computed atomization energies. Finally, comparison is made between the ring strain energies of oxiranone and cyclopropanone. These ring strain energies should be of value in mechanistic discussions involving possible a-lactone intermediates. Computational methods All calculations were carried out using the GAUSSIAN92 and 94 series of programs.18,19 Full geometry optimizations were performed at the second-order Møller–Plesset (MP2) level using the 6-311G(d,p) basis with all MOs being active (MP2]] full).Hartree–Fock [HF/6-31G(d,p)] harmonic vibrational frequency calculations were used to obtain zero-point and thermal energy corrections at 298 K. Single-point energies were evaluated using fourth-order Møller–Plesset (MP4SDTQ) theory including single, double, triple and quadruple excitations but with inactive core MOs (frozen-core, MP4]] fc) and also quadratic configuration interaction [QCISD(T)] including single and954 J.Chem. Soc., Perkin Trans. 2, 1997 Table 1 Calculated total energies, vibrational zero-point and thermal energies (Ezp and Eth/kJ mol21) and standard enthalpies of formation and of reaction (DHo f,298 and DHo r,298/kJ mol21) for the isodesmic reaction (1) HF/6-31G(d,p) MP2/6-311G(d,p) MP4/6-311G(2df,p) Total energy/ Ezp b/ Eth/ Total energy/ Total energy/ DHo f,298/ Structure hartreea kJ mol21 kJ mol21 hartree hartree kJ mol21 Expt. 1 2226.570 24 110 9 2227.365 21 2227.458 68 2210 6 2232.160 58 383 18 2233.110 87 2252.1 ± 0.8d 7 2152.874 07 163 8 2153.453 99 2153.522 78 262 252.6 ± 0.6e 8 2305.889 50 332 18 2307.045 98 2444.9 ± 0.7f DHo r,298 c/ kJ mol21 285 261 a 1 hartree = 2625 kJ mol21. b Unscaled vibrational zero-point energy. c Evaluated using zero-point energies scaled × 0.89.d Ref. 23. e Ref. 24. f Ref. 21. Table 2 Calculated total energies, vibrational zero-point and thermal energies (Ezp and Eth/kJ mol21) and standard enthalpies of formation and of reaction (DHo f,298 and DHo r,298/kJ mol21) for the isodesmic reaction (2) HF/6-31G(d,p) MP2/6-311G(d,p) MP4/6-311G(2df,p) Total energy/ Ezp b/ Eth/ Total energy/ Total energy/ DHo f,298/ Structure hartreea kJ mol21 kJ mol21 hartree hartree kJ mol21 Expt. 1 2226.570 24 110 9 2227.365 21 2227.458 68 2210 9 2231.010 91 315 17 2231.938 55 2240.8 ± 0.6d 10 2190.733 26 173 10 2191.465 39 2191.545 21 27 15.9 ± 4.2e 11 2266.846 53 253 15 2267.827 38 2410.0 ± 0.8d DHo r,298 c/ 4 29 kJ mol21 a 1 hartree = 2625 kJ mol21.b Unscaled vibrational zero-point energy. c Evaluated using zero-point energies scaled × 0.89. d Ref. 25. e Ref. 26. double excitations together with a perturbative treatment of triple excitations and with all orbitals active; these calculations employed the 6-311G(2df,p) basis.Results and discussion Oxiranone ring strain energy from isodesmic relation to oxirane The conventional ring strain energy of a cyclic compound is obtained as the difference between the enthalpy of atomization estimated from a group-additivity scheme for a strain-free molecule and the actual observed enthalpy of atomization for the compound.20,21 The accepted value 22 for the conventional ring strain energy of oxirane 7 is 114 kJ mol21, very similar to that for cyclopropane (115 kJ mol21).The strain energy of oxiranone 1 relative to that of oxirane 7 may be evaluated as the enthalpy change DHo r,298 for the isodesmic reaction (1), which may be quantified by use of the energetic contributions listed in Table 1. [An isodesmic relation is a formal equilibrium in which the number of each different type of bond is conserved. The energy change for eqn. (1), therefore, is simply a measure of the relative preference for the carbonyl group to be located in a cyclic or acyclic ester.] At the MP2/6-311G(d,p) level of theory this quantity is 261 kJ mol21.The conventional ring strain energy of oxiranone may therefore be estimated by this means as 114 1 61 = 175 kJ mol21. Oxiranone enthalpy of formation from isodesmic relations to oxirane and cyclopropanone Combining the computed value of DHo r,298 for the isodesmic reaction (1) with experimental enthalpies of formation for diethyl ether 6,23 oxirane 7 24 and ethyl acetate 821 allows the enthalpy of formation for oxiranone 1 to be evaluated.DHo f,298(1) =DHo f,298(7)1DHo f,298(8)2DHo f,298(6)2DHo r,298 = (252.6 ± 0.6) 1 (2444.9 ± 0.7) 2 (2252.1 ± 0.8) 2 (261.0) = 2184 kJ mol21 (1) The scheme used by Liebman and Greenberg 17 to estimate the enthalpy of formation of oxiranone (as 2130 kJ mol21) is equivalent to the isodesmic reaction (2), in which the molecular strain is considered relative to that of cyclopropanone 10.[The energy change for eqn. (2) is a measure of the relative preference for the ether oxygen to be located in a cyclic or acyclic ester.] The enthalpy change DHo r,298 for reaction (2) may be quantified by use of the energetic contributions listed in Table 2. At the MP2/6-311G(d,p) level of theory this quantity is 29 kJ mol21. Combining the computed value of DHo r,298 for the isodesmic reaction (2) with experimental enthalpies of formation for butan-2-one 9,25 cyclopropanone 10 26 and methyl acetate 1125 allows the enthalpy of formation for oxiranone 1 to be evaluated.DHo f,298(1) = DHo f,298(10) 1 DHo f,298(11) 2 DHo f,298(9) 2 DHo r,298 = (16 ± 4) 1 (2410.0 ± 0.8) 2 (2240.8 ± 0.6) 2 (29.2) = 2182 kJ mol21 (2) The two isodesmic schemes yield computed estimates for the enthalpy of formation of oxiranone which are identical within the errors of the experimental DHo f,298 values employed. It should be noted, however, that there remains a yet unquantified error associated with the choice of the MP2/6-311G(d,p) theoretical method.[The more reliable QCISD(T)]] full/6- 311G(2df,p) method (see below) could not be employed for the larger species 6, 8, 9 and 11 (containing more than four carbon or oxygen atoms) involved in the isodesmic relations (1) and (2), owing to lack of sufficient computational resources.] In Liebman and Greenberg’s scheme,17 the quantity which is here identified as DHo r,298(reaction 2) was replaced by a ‘p resonance energy’ whose value (246 kJ mol21) was taken to be equal to the p conjugation energy of acetamide.27 Since this value differs in both sign and magnitude from our computed value of DHo r,298(reaction 2) = 129 kJ mol21, it is not surprising thatJ.Chem. Soc., Perkin Trans. 2, 1997 955 Table 3 Calculated total energies and standard enthalpies of formation (kJ mol21) for selected species MP4SDTQ]] fc/6-311G(2df,p)// QCISD(T)]] full/6-311G(2df,p)// MP2]] full/6-31G(d,p) MP2]] full/6-31G(d,p) Expt.Total energy/ DHo f,298/ Total energy/ DHo f,298/ DHo f,298/ Structure hartreea kJ mol21 hartree kJ mol21 kJ mol21 1 2227.458 68 2210 2227.536 96 2190 7 2153.522 79 259.0 2153.585 19 254.4 252.6 ± 0.6b 10 2191.545 21 26.7 2191.627 30 6.3 16 ± 4c C 237.775 44 237.796 42 O 274.965 54 274.986 01 H 20.499 81 20.499 81 H2 21.167 73 21.168 32 a 1 hartree = 2625 kJ mol21. b Ref. 24. c Ref. 26. Liebman and Greenberg’s estimate for DHo f,298(1) is considerably different from our MP/6-311G(d,p) derived value of 2184 kJ mol21.Enthalpies of formation from computed atomization energies: oxirane, cyclopropanone and oxiranone Although it is not computationally feasible to evaluate the energy changes for the isodesmic reactions (1) and (2) using a theoretical method of any significantly greater accuracy than that used above, it is possible to perform more sophisticated calculations upon the individual species of interest 1, 7 and 10 (containing only three or four carbon or oxygen atoms), as follows.Oxirane 7 and cyclopropanone 10 are considered first, since these are the ‘standards’ used in the alternative isodesmic schemes. Oxirane. Enthalpies of formation may be calculated directly using a previously described procedure 28,29 which has been shown to be accurate to ca. ±10 kJ mol21. This method may be illustrated step by step using oxirane (for which the experimental value 24 of DHo f,298 = 252.6 ± 0.6 kJ mol21 is reliably established) as the working example.(i) The atomization energy is determined at the MP4(fc)/ 6-311G(2df,p)//MP2(full)/6-311G(d,p) level for reaction (3) (energies from Table 3). C2H4O 2C 1 O 1 4H (3) DEatom(7) = 1.007 13 hartree (ii) Since this reaction is non-isogyric (C and O atoms each have two unpaired electron spins and atomic H has one), 10H are added to the left-hand side and are balanced on the righthand side by 5H2, to yield the isogyric reaction (4).Use of the MP4(fc)/6-311G(2df,p) energies for C, O, H and H2 (Table 3) then leads to a corrected atomization energy. C2H4O 1 6H 2C 1 O 1 5H2 (4) corrected DEatom(7) = 0.166 57 hartree (iii) The exact dissociation energy 30 for 5H2 (5 × 0.174 47 hartree) is added, giving a value for SDe(7) = 1.038 92 hartree (2728 kJ mol21). (iv) The zero-point vibrational energy of C2H4O [HF/6-31G- (d,p), Table 1, scaled by 0.89] is subtracted to give SD0(7) = 2583 kJ mol21.(v) Using the experimental enthalpies of formation of atoms at 0 K (C, 711; O, 247; H, 216 kJ mol21),31 the standard enthalpy of formation of oxirane at 0 K is obtained from eqn. (5). DHo f,o(7) = DHo f,o(C) 1 DHo f,o(H) 1 DHo f,o(O) 2 SD0 (C2H4O) 2 × 711 1 4 × 216 1 247 2 2583 = 250 kJ mol21 (5) (vi) Thermal corrections are added, using standard heat capacities for the elements in their standard states 32 and theoretical values from Table 3 to give DHo f,298(7, MP4) = 259 kJ mol21.This MP4-derived value compares reasonably well with the experimental value 24 of 252.6 ± 0.6 kJ mol21, given the estimated error of ±10 kJ mol21 in enthalpies of formation computed in this manner.28 Use of the yet more reliable (but also more expensive!) QCISD(T)]] full/6-311G(2df,p)//MP2]] full/6- 311G(2df,p) energies 29 (Table 3) in place of the MP4 energies, within the procedure laid out in steps (i) to (iv) above, yields a still better calculated result of DHo f,298(7, QCI) = 254 kJ mol21.Cyclopropanone. The same procedures applied to cyclopropanone 10 yield MP4 and QCI calculated values for its enthalpy of formation: DHo f,298(10, MP4) = 26.7 kJ mol21 and DHo f,298- (10, QCI) = 16.3 kJ mol21. Unfortunately, it is not possible to choose between these two computed values simply by appeal to available experimental data, since published estimates for the enthalpy of formation of cyclopropanone vary between ca. 212 and 121 kJ mol21.At the lower end of this range are Liebman and Greenberg’s estimates 27 which, however, are unlikely to be reliable in view of their implicit assumption that the enthalpies of reaction for the isodesmic processes (6) and (7) are zero. The value of 16 ± 4 kJ mol21, adopted in the above discussion of isodesmic reaction (2), was obtained by Thomas and coworkers 26 from measurements of the appearance potential for the C2H4 1 ion formed from cyclopropanone by electron impact in a mass spectrometer.Evaluation of DHo f,298(10) by this method requires a known value for DHo f,298(C2H4 1 CO); use of more recent thermochemical data,31 together with the appearance potential (9.69 eV) of Thomas and co-workers,26 yields a value for DHo f,298(10) of 21 kJ mol21 rather than 16 kJ mol21. Note that any excess of internal energy might have yielded an overestimated appearance potential; however, correction for this possible source of experimental error would lead to a still larger value for DHo f,298(10).[It may also be noted that the accuracy of ±0.03 eV claimed for the appearance potential measurement seems optimistic for mid-1970s equipment; ±0.1 eV (±10 kJ mol21) may be more likely.] McLafferty and co-workers 33 calculated the enthalpy difference between cyclopropanone and propenal as 76 kJ mol21 at the MP4/6-31G(d)//HF/6-31G(d) level with a scaled HF/6- 31G(d) zero-point energy correction; taking the enthalpy of formation of propenal 31 as 277 kJ mol21 yielded a value for DHo f,298(10) of 21 kJ mol21.Our present MP4 calculations use956 J. Chem. Soc., Perkin Trans. 2, 1997 a considerably larger basis set than was employed by these workers, and should therefore lead to more reliable results. McKee and Radom34 employed the G2 theoretical procedure to compute DHo f,298(10) = 17.5 kJ mol21, but also calculated DHo f,298(propenal) = 267.6 kJ mol21, as compared with the literature value31 of 277 kJ mol21, the latter result suggests the possibility of larger errors for molecules of this size with G2 theory than the average error of less than 4 kJ mol21 claimed for atomization energies over its test set of smaller molecules.35 If the G2 value for DHo f,298(propenal) is overestimated by 9 or 10 kJ mol21, then our calculated value for DHo f,298(10, QCI) = 6.3 kJ mol21 may be quite close to the true enthalpy of formation for cyclopropanone.Use of the QCI calculated value for DHo f,298(10) = 6.3 kJ mol21 in the isodesmic relation (2) in place of the literature value of 16 kJ mol21 leads to an MP2 estimate for the enthalpy of formation of oxiranone equal to 2192 kJ mol21 (see below).Oxiranone. The same procedures applied to oxiranone 1 yields MP4 and QCI calculated values (Table 3) for its enthalpy of formation: DHo f,298(1, MP4) = 2210 kJ mol21 and DHo f,298- (1, QCI) = 2190 kJ mol21 As for oxirane and cyclopropane, the MP4 method also predicts a more negative enthalpy of formation for oxiranone than does the QCISD(T) method.Since the latter method leads to very good agreement with experiment for oxirane and, as argued above, may be rather close to the true value for cyclopropanone, we suggest a best estimate for DHo f,298(oxiranone) = 2190 ± 10 kJ mol21. This value is in fair agreement with those obtained from the isodesmic relations (1) and (2) at the MP2 level but contrasts with the bond-additivity estimate of 2130 kJ mol21 previously reported by Liebman and Greenberg. 17 Although it would be desirable to employ the QCISD(T) method with the isodesmic relations (1) and (2), in practice the computations involved at this level for diethyl ether 6, ethyl acetate 8, butan-2-one 9, and methyl acetate 11 are beyond our present resources.Nonetheless, the measure of agreement between the results obtained from these different approaches is such that we propose that our ab initio value for DHo f,298- (oxiranone) = 2190 ± 10 kJ mol21 should now be adopted in preference to the previous estimate in the literature.17 Comparison of ring strain energies for oxiranone and cyclopropanone The enthalpy change for isodesmic reaction (1) determined using the present QCISD(T) value for DHo f,298(oxiranone) together with the experimental values for 6–8 noted above, is 255 kJ mol21, the MP2 value noted above (261 kJ mol21) is in reasonable agreement.The negative of this quantity represents the additional strain introduced into the three-membered heterocyclic oxirane ring due to replacement of a methylene group (sp3 carbon) by a carbonyl group (sp2 carbon). The QCISD(T) estimate for the conventional ring strain energy of oxiranone is therefore 114 1 55 = 169 kJ mol21.How does this extra increment of strain compare with that arising from the corresponding replacement in the three-membered carbocyclic cyclopropane ring? This question may be answered by consideration of the enthalpy change for the isodesmic reaction (8), evaluated using the QCISD(T) value for DHo f,298(cyclopropanone) together with the experimental values 25 for cyclopropane 12, acetone 13 and propane 14: DHo r,298(reaction 8) = DHo f,298(10) 1 DHo f,298(14) 2 DHo f,298(12) 2 DHo f,298(13) = (6.3) 1 (2104.5 ± 0.3) 2 (53.3 ± 0.5) 2 (2217.2 ± 0.4) = 166 kJ mol21 (Use of pentan-3-one and pentane in place of acetone and propane, respectively, yields a very similar result, DHo r,298 = 165 kJ mol21.) Bearing in mind that the QCISD(T) value for the enthalpy of formation for cyclopropanone is lower by 10 kJ mol21 than the experimental value of Thomas and coworkers, 26 adoption of any higher value for this quantity in the isodesmic relation (8) would lead to a further increased estimate for the additional strain energy of cyclopropanone over cyclopropane. This would suggest that replacement of CH2 by C]] O in a three-membered ring involves an increase in strain energy for oxirane æÆ oxiranone which is certainly not larger—and is very likely to be smaller—than for cyclopropane æÆ cyclopropanone.Since the conventional ring strain energy of cyclopropane is 115 kJ mol21, that of cyclopropanone may therefore be estimated as 181 kJ mol21. Like oxiranone, cyclopropanone is a highly strained molecule and very reactive; it may be formed by reaction of diazomethane with ketene at low temperature, but it is not stable at room temperature.36,37 Cyclopropanones are known intermediates in, for example, Favorksii rearrangements.Oxiranones, although elusive species, would seem to be (if anything) slightly less strained than cyclopropanones and may therefore be no less likely to occur as reaction intermediates, albeit transiently. Conclusions Ab initio molecular orbital calculations at the QCISD(t)]] full/ 6-311G(2df,p)//MP2]] full/6-311G(d,p) level of theory provide a best estimate for the enthalpy of formation DHo f,298 for oxiranone of 2190 ± 10 kJ mol21, corresponding to a conventional ring strain energy of 169 kJ mol21.In the absence of an experimental determination, this computed value is to be preferred over the previous bond additivity estimate. The oxiranone ring is probably slightly less strained than the cyclopropanone ring. Acknowledgements We are grateful to the EPSRC for support of this work through research grant GR/J53218. References 1 G. L’abbe, Angew.Chem., Int. Ed. Eng., 1980, 19, 276. 2 D. E. Milligan and M. E. Jacox, J. Chem. Phys., 1962, 35, 2911. 3 W. W. Sander, J. Org. Chem., 1989, 54, 4265; J. E. Chateauneuf, Res. Chem. Intermediates, 1994, 20, 159. 4 S. Wierlacher, W. Sander and M. T. H. Liu, J. Org. Chem., 1992, 57, 1051. 5 O. L. Chapman, P. W. Wojtkowski, W. Adam, O. Rodriguez and R. Rucktäschel, J. Am. Chem. Soc., 1972, 94, 1365; W. Adam and R. Rucktäschel, J. Am. Chem. Soc., 1971, 93, 557. 6 R. Wheland and P.D. Bartlett, J. Am. Chem. Soc., 1970, 92, 6057. 7 W. Adam, J.-C. Liu and O. Rodriguez, J. Org. Chem., 1973, 38, 2269. 8 P. G. Cole, A. Sellars, J. C. Tatlow, G. Whittaker and H. C. Fielding, J. Chem. Soc., Chem. Commun., 1982, 362. 9 E.g. J. K. Crandall, S. A. Sojka and J. B. Kromin, J. Org. Chem., 1974, 39, 2172 and references cited therein. 10 L. P. Hammett, Physical Organic Chemistry, McGraw-Hill, New York, 1940. 11 (a) G. Chuchani, I. Martin, A. Rotinov, R. M. Dominguez and M.Perez I, J. Phys. Org. Chem., 1995, 8, 133; (b) G. Chuchani and A. Rotinov, Int. J. Chem. Kinet., 1989, 21, 367. 12 A. Liberles, A. Greenberg and K. Megerle, Tetrahedron, 1975, 31, 657. 13 C. S. C. Chung, J. Mol. Struct., 1976, 30, 189. 14 D. Antolovic, V. J. Shiner and E. R. Davidson, J. Am. Chem. Soc., 1988, 110, 1375. 15 M. Ashwell, X. Guo and M. L. Sinnott, J. Am. Chem. Soc., 1992, 114, 10 158. 16 W. A. Cowdrey, E. D. Hughes and C. K. Ingold, J. Chem. Soc., 1937, 1208. 17 J. F. Liebman and A. Greenberg, J. Org. Chem., 1974, 39, 123. 18 M. J. Frisch, G. W. Trucks, M. Head-Gordon, P. M. W. Gill, M. W. Wong, J. B. Foresman, B. G. Johnson, H. B. Schlegel, M. A. Robb,J. Chem. Soc., Perkin Trans. 2, 1997 957 E. S. Repogle, R. Gomperts, J. L. Andres, K. Raghavachari, J. S. Binkley, C. Gonzalez, R. L. Martin, D. J. Fox, D. J. Defrees, J. Baker, J. J. P. Stewart and J. A. Pople, GAUSSIAN 92, Revision C.4, Gaussian, Inc., Pittsburgh, PA, 1992. 19 M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G. Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. A. Petersson, J. A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G. Zakrzewski, J. V. Ortiz, J. B. Foresman, C. Y. Peng, P. Y. Ayala, W. Chen, M. W. Wong, J. L. Andres, E. S. Repogle, R. Gomperts, R. L. Martin, D. J. Fox, J. S. Binkley, D. J. Defrees, J. Baker, J. J. P. Stewart, M. Head-Gordon, C. Gonzalez and J. A. Pople, GAUSSIAN 94, Revision B.3, Gaussian, Inc., Pittsburgh, PA, 1995. 20 D. Cremer and E. Kraka, in Structure and Reactivity, eds. J. F. Liebman and A. Greenberg, VCH, New York, 1988. 21 J. D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds, Academic Press, London, 1970. 22 Ref. 21, p. 574. 23 G. Pilcher, A. E. Pope and H. A. Skinner, Trans. Faraday Soc., 1963, 59, 2233. 24 A. S. Pell and G. Pilcher, Trans. Faraday Soc., 1965, 61, 71. 25 J. B. Pedley and J. Rylance, Sussex-NPL Computer Analysed Thermochemical Data: Organic and Organometallic Compounds, University of Sussex, Brighton, 1977. 26 H. J. Rodriguez, J.-C. Chang and T. F. Thomas, J. Am. Chem. Soc., 1976, 98, 2027. 27 Ref. 17, footnote 27. 28 C. F. Rodriquez, D. K. Bohme and A. C. Hopkinson, J. Am. Chem. Soc., 1993, 115, 3263. 29 C. F. Rodriquez, D. K. Bohme and A. C. Hopkinson, J. Phys. Chem., 1996, 100, 2942. 30 W. Kolos and L. Wolniewicz, J. Chem. Phys., 1968, 49, 404. 31 S. G. Lias, J. E. Bartmess, J. F. Liebman, R. D. Levin and W. G. Mallard, J. Phys. Chem. Ref. Data, Suppl., 1988. 32 CODATA, J. Chem. Thermodyn., 1978, 10, 903. 33 F. Turecek, D. E. Drinkwater and F. W. McLafferty, J. Am. Chem. Soc., 1991, 113, 5950. 34 M. L. McKee and L. Radom, Org. Mass Spectrom., 1993, 28, 1238. 35 L. A. Curtiss, K. Raghavachari, G. W. Trucks and J. A. Pople, J. Chem. Phys., 1991, 94, 7221. 36 H. H. Wasserman, G. M. Clark and P. C. Turley, Top. Curr. Chem., 1974, 47, 73. Paper 6/06820K Received 4th November 1996 Accepted 20th January 1997