GENERAL DISCUSSION Dr S . Lunell (Uniuersity of Uppsala, Sweden) said: I would like to make a comment in connection with Dr Iwasaki's results for the cyclohexane cation. We have recently completed a series of theoretical calculations on different states of the c-C,Ht2 cation, using the a6 initio MO-LCAO-UHF method.' Our results for the symmetric-chair conformation ( C 2 h symmetry) do confirm that the elongated- chair (2A,) is slightly more stable (ca. 1 kcal mol-') than the compressed-chair ('B,) conformation, as suggested by Dr Iwasaki and collaborators. However, a further unsymmetrical distortion to C, symmetry, according to fig. 1, produces a 2A" ground state with an energy ca. 7 kcal mol-' lower than both of the abovementioned states for C2h symmetry. For this state the calculations predict an e.s.r.spectrum consisting of three sets of triplets with coupling constants 60 G(2H), 20 G(2H) and 9.4 G(2H). Apart from a uniform scale factor of ca. 1.5, these values are in excellent agreement with the experimental values 85 G(2H), 34 G(2H) and 14 G(2H) reported by Iwasaki et al. This suggests another possible explanation of the low-temperature e.s.r. spectrum of c-C,HT2 without changing the numerical values of the coupling constants reported by these authors, namely that the ion has an unsymmetric-chair conformation with C, rather than C 2 h symmetry, and that the largest set of couplings should be assigned to H1 and H2, the next largest to H3 and H4, and the smallest set of resolved couplings to H9 and H l l , which, although not symmetry-related, are predicted to have almost identical coupling constants by the calculations.I wonder if Dr Iwasaki would like to comment on these results? ' S. Lunell, M. B. Huang, 0. Claesson and A. Lund, J. Chem. Phys., to be published. 9 10 Fig. 1. Unsymmetrical distorted chair with C, symmetry.84 C E NERAL DISCUSSION Dr M. Iwasaki (Nagoya, Japan) said: Dr Lunell has assumed the C, distortion for c-C,Hf, for energy optimization to obtain three sets of coupling constants. However, the e, ring-deformation mode, which leads to the C, distortion, is not Jahn-Teller active for the e, electronic orbital of c-C6HT2 with D3d symmetry. Thus, such a deformation cannot be assumed because there is no matrix element for vibronic interaction. In addition, the C,$ distortion cannot explain the observed temperature change of the three coupling constants reported in the first paper.Dr S. Lunell (University of Uppsala, Sweden) replied: It is correct that e, is not a Jahn-Teller-active mode for the e, orbitals in D3d symmetry. However, once the D3d symmetry has been broken by one of the Jahn-Teller-active modes, the Jahn- Teller theorem no longer applies, and in principle any type of further distortion is possible. Thus in this respect the distorted-chair conformation of C, symmetry is perfectly allowed. Therefore, if the temperature dependence uniquely confirms a ’A, assignment, this implies that the cyclohexane cation exhibits a change of ground state owing to matrix interactions in these matrices, in the same way as the propane cation does in e.g.CFC13 or CF3CC13, which is itself an interesting result. Dr M. Iwasaki (Nagoya, Japan) said: It is hard to believe that c-C6Ht2 with C2, symmetry distorts to the asymmetric C, form by intramolecular interactions. With the exception of c-C,HT2, all of the Jahn-Teller-active cations studied so far exhibit the static distortion expected from the Jahn-Teller-active vibrational mode [ref. (3) of our paper]. Furthermore, in order to explain the averaging of the two sets of coupling constants (85 and 34 G) by C, distortion one has to assume a jumping between the two equivalent distorted structures which interchanges (HI, H2) with (H3, H4) and (€-I9, H l l ) with (Hlo, HI’). The dynamic process involving such interchanges cannot explain the observed temperature change. Because of no resolvable extra coupling from Hlo and HI2, the interchange of (H9, HI1) with (Hlo, HI2) must result in a decrease of the coupling to (H9, H I 1 ) with an increase of the coupling to (Hlo, HI2).However, this is not observed. Furthermore, it seems difficult to explain the continuity of the changes in the three coupling constants by the model without a dynamic average. For such a dynamical process the potential barrier at the symmetrical structure must be very high and equal to the distortion energy of the C, form. On the other hand, the dynamic process in the Jahn-Teller potential minimum is a two-dimensional oscillation. The radial oscillation must have a very high potential maximum at the symmetrical position, whereas the tangential oscillation can have a low potential barrier resulting in the pseudo-rotation at quite low temperature. Prof.M. Sevilla (Oakland University, Michigan) said: The static Jahn-Teller distortions of the cycloalkane cations described in Dr Iwasaki’s work produce sites of high charge localization in certain instances. Could it be that some form of matrix interaction with these sites tends to stabilize one state over another? Dr M. Iwasaki (Nagoya, Japan) replied: It is hard to believe that matrix interaction with a specific site having a high charge density preferentially stabilizes one state over another regardless of matrix. In addition to the variation of the matrix molecule, the distribution and the relative orientations of the surrounding molecules must be different from matrix to matrix, so that it is difficult to adduce that one of the two states is stabilized by a specific matrix interaction.For this reason we have suggested that matrix interaction assists stabilization of a givenGENERAL D I SC U SS ION 85 ground state, which is determined by the intramolecular interaction and is intrinsic to radical cations. We may have to consider solvation of matrix molecules similar to that of a self-digging trapped electron. One might suppose that complexing with a matrix molecule at some specific site having high charge density may play a role in stabilizing one state over another. However, the unpaired electron of radical cations of alkanes delocalizes over the C-C and C-H bonding orbitals. Even when the unpaired electron is more confined to one of the C-C bonds, it is a a-bonding orbital.Thus the ability for complex formation with a matrix is thought to be lower than for the T- and n-cation radicals, in which the unpaired electron orbital is directed outside the molecular frame of the cations. The complexing of cations with the matrix may be of less importance in alkane radical cations. Prof. L. Kevan (University of Houston, Texas) asked: Is there much predictive understanding of the differing dynamic behaviours between matrices? Dr Iwasaki suggests that the activation energies may be related to matrix polarity, but the low value for CFC1, does not seem to fit in with this. What about the role of asymmetry of the matrix molecule or of the matrix sites? Dr M. Iwasaki (Nagoya, Japan) replied: The temperature of the onset of dynamic averaging is generally lower in SF, in comparison with that in other halogenocarbon matrices, suggesting that the activation energy is lower in non-polar solvents.However, since the dipole moments of C-Cl and C-F bonds are not very different, halogenocarbons such as CFCI, and CFC12CF2C1 also have relatively small dipole moments. For example, the dipole moment of CFC1, is only 0.45 D. The total matrix polarity may not be a decisive factor, but a difference between the bond polarizabilities of C-F and C-C1 may be much more important. Since maxima and minima in the Jahn-Teller potential trough are determined by the second-order effect such as anharmonicity and a quadratic term, the contact of the radical cations with a cavity wall may be another important factor in determining the potential barrier to pseudorotation.Thus tight or loose contact with the surrounding matrix molecule may be related to the difference in the activation energy. In this sense the size of the matrix cavity may be an important factor. In general, the Jahn-Teller radical cations exhibit more rigid structure immediately after irradiation at 4 K. However, once the sample is annealed at 77 K, they exhibit a larger zero-point motional effect, suggesting that irreversible matrix relaxations give more space to cations. The molecular symmetry may play a role through interactions at the cavity wall of the substitutional site with a solute molecule. However, not only the spherical SF, molecule but also asymmetrical CFCI2CF2Cl give generally low activation energies. CFC13 and CF3CC13 form rather rigid matrices in which the onset tempcrature of the dynamics is usually higher than in SF, and We have also compared glassy and crystalline media.However, no decisive difference has been found. At present, the details of the matrix interaction are not known. However, what is most striking is that the sign of the distortion coordinate is independent of matrix. Recently, we have examined the effect of environment on benzene cations adsorbed on silica gels and on a variety of synthetic zeolites. Although the activation energy of the dynamic motion is lower in these environments than that in halogenocarbon matrices, the sign of the distortion coordinate is independent of environment, i.e.the compressed quinoidal structure is always the ground state. C FC12C F2Cl.86 GENERAL DISCUSSION Dr N. Klassen (N.R. C., Ottawa, Canada) said: In several hydrocarbon liquids, including cyclohexane, the cation is found to possess an unusually high mobility. Does Dr Iwasaki have any evidence that the initial cation of cyclohexane is different from other alkane cations, for example, is it more stable? Dr M. Iwasaki (Nagoya, Japan) replied: The cyclohexane radical cation does not seem to be especially stable'in halogenocarbon matrices under our experimental conditions. It is converted into the cyclohexyl radical in CFC12CF2CI. Experiments on the addition of electron scavengers to alkanes indicate that the alkyl radicals must be formed from primary cationic species.' In addition, pure n-alkanes (n-C,oH22-n-C19H,0) irradiated at 4 K indicate that the chain-end alkyl radicals are selectively formed regardless of the chain length.2 These results suggest that the primary event in alkyl-radical formation from the cationic species takes place at the chain end, and this is just what is expected from the unpaired electron distribution in the linear alkane radical cations with an extended structure in the crystal [ref.( 3 ) and (4) in our paper]. Taking hydrogen-atom formation in irradiated alkanes into c~nsideration,'~~ it is suggested that the unrelaxed primary radical cation decomposes into R+ and H: (RH+)* -+ RT+H (1) Rr+e- + RI (2) (3) In reactions (1)-(3) R, is the chain end, RII is the penaltimate alkyl radical and R,,, the interior alkyl radical.Another reaction path which has to be considered is the ion-molecule reactions of unrelaxed alkane radical cations, followed by a charge recombination to form H:3 followed by the neutralization reaction to form a chain-end alkyl radical: and by the hydrogen-abstraction reaction to form various types of alkyl radicals: H+RH --* R,, RI,, R,,,. (RH+)*+RH -+ R,+RH; (4) RH;+e- -+ RH+H ( 5 ) ( 6 ) H+ RH -+ R,, RII, R,,,. The suppression of radical-pair formation at 4 K by the addition'of a small amount of electron scavenger' indicates that R+ and RH; are mobile since hydrogen abstraction by the inmobilized hydrogen atom at 4 K would form radical pairs if R+ and RH; were not mobile before the neutralization reactions (2) and ( 5 ) .So, there are three candidates for the mobile positive charge in irradiated cyclohexane, namely unrelaxed RH+, R+ and RH;. K. Toriyama, H. Muto, K. Nunome, M. Fukaya and M. Iwasaki, Radiat. Phys. Chem., 198 1,18,104 1. ' M. Iwasaki, H. Muto, M. Fukaya, K. Toriyama and K. Nunome, 26th Symp. Radiat. Chem. (Osaka, Sept. 28, 1983, Abstract B207), p. 118. M. Iwasaki, K. Toriyama, H. Muto, K. Nunome and M. Fukaya, Radiat. Phys. Chem., 1981,17,304. Dr N. Klassen (N.R.C., Ottawa, Canada) said: A reaction of alkane cations which seems more and more likely to me is proton transfer to an alkane molecule: RH++RH + R*+RHl. Although reactions such as this do not seem to take place in the gas phase, published proton affinities suggest that many such reactions are exothermic.GENERAL DISCUSSION 87 Dr M.Iwasaki (Nagoya, Japan) replied: The fate of a detached proton cannot be detected because it does not form a paramagnetic species. In any case, the deprotonation reaction is bimolecular because of the solute concentration depen- dence. There are three possible mechanisms which can explain the concomitant decrease in cation radicals with an increase in alkyl radicals. The first is the ion-molecule reaction: RH+i + RH --+ Re? + RH;. (1) Although gas-phase energetics are unfavourable for this mechanism, in the solid phase there is a possibility that the potential-energy surface is different from that in the gas phase, so that this mechanism cannot be excluded. It is consistent with selective deprotonations at the C-H bond having the highest spin density.The second is the charge-neutralization reaction: RH+i+X- ---* R"/'+HX. (2) This mechanism can also explain selective deprotonation, because X- will attack the C-H bond with high positive-charge density. The third mechanism is positive-hole migration from the cation to X- forming X atoms followed by hydrogen abstraction by the X atom from RH. However, this mechanism is less probable because selective deprotonation cannot be explained. Now, the reactions of alkane radical cations are dependent upon matrix, and the deprotonation reaction is typical of SF6 and CFCl2CF2C1. Shida's group have reported that cycloalkane radical cations undergo deprotonation in CFC1, when the solute concentration is as high as 10% [ref. (15) in our paper].However, unimolecular detachment of H2 or CH4 is more common in CFC1, at low solute concentrations. Prof. M. C. R. Symons ( University of Leicester) (communicated) In our original study of the (Me,C-CMe,)+ cation we postulated that the SOMO was considerably confined to the central C-C u bond, which was therefore stretched and weakened, with some degree of flattening at the two Me3C- groups.' This has been confirmed by the extensive studies of Iwasaki and coworkers.2 For example, for the series of cations (Me$ H)+, (Me,C CH,)+ and (Me,C CMe3)+ the representation given appears to be reasonable. The cation (Me,C * H)+ has a 'H coupling of 251 G, showing the very large spin density in the unique C-H orbital. Probably the extent of flattening of the Me3C- units decreases in this series for steric reasons. ' I.G. Smith and M. C. R. Symons, J. Chem. Res. ( S ) , 1979, 382. * K. Nunome, K. Toriyama and M. Iwasaki, J. Chem. Phys., 1983,79, 2499. Dr M. Iwasaki and Dr K. Toriyama (Nagoya, Japan) (communicated) We totally agree with Prof. Symons' comment that the structure of branched-alkane radical cations are characterized in line with HME'.' Note that C2H6+ and other linear- alkane radical cations, on the other hand, are not characterized as an extension of HME+ as we have discussed in our earlier work.* In these cations the unpaired electron is rather delocalized over the in-plane u molecular orbital. ' M. C. R. Symons and I. G. Smith, J. Chem. Rex ( S ) , 1979, 382. ' K. Toriyama, K. Nunome and M. Iwasaki, J. Chem. Phys., 1982, 77, 5891.Dr M. Iwasaki and Dr K. Toriyama (Nagoya, Japan) (communicated) In his presentation at the Discussion, Dr Lunell pointed out the following two problems in relation to our paper at this Discussion and to ref. (3) therein: (1) the disagreement of the principal directions of the &proton coupling tensor calculated by their ab88 GENERAL DISCUSSION initio spin density for c-C3H; with the values used in our spectral simulation and (2) the possibility of 2A, propane cation radicals. ( 1 ) The ab initio directions seem to be incorrect because the two p protons possess essentially coaxial tensors which cannot be expected from the geometry. The directions of the intermediate and minimum elements are quite sensitive to the spin distribution, especially on the C( 1 ) atom.In our preliminary communication [ref. (7) of our paper at this Discussion] a small spin density on the C ( l ) atom is taken into consideration, whereas it was assumed to be zero in the present paper. The difference arises only in the extent of the deviation from the axial symmetry of the P-proton couplir,g tensor. In principle it is difficult to estimate a small deviation from axial symmetry with high reliability either experimentally or by theoretical calculation. Thus the comparison has little meaning. (2) We have, of course, considered the possibility of 2AI. However, we thought that it may give a smaller set of coupling constants as compared with a large experimental value because of a smaller overlap of H I , and C,, orbitals.Indeed, calculated values presented in Dr Lunell’s paper support our prediction, although their ab inftio values for both ,B2 and *A1 do not reproduce the experimental ratio of 2: 1 for the central and non-central CH,-proton couplings. Our INDO values gave better agreement in this respect [ref. (3) of our paper]. According to our comparisons over more than 35 alkane radical cations, the INDO values usually show reasonable agreement with experimental values. We stress that our experi- mental data made such an assessment possible. Of course, our data must be a guide for the refinement of more sophisticated ab initio molecular-orbital theory of (T- electron systems. Dr S . Lunell ( University of Uppsala, Sweden) (communicated) (1) The disagreement that Dr Iwasaki refers to is with the principal directions of the P-proton coupling tensor given in his preliminary communication [ref.(7) of his paper], which indeed differ significantly from ours. Dr Iwasaki’s revised results, published in these proceedings, which were not available to us when our paper was written, are in contrast rather close to our ab initio results, so that I do not think that any real controversy remains on this point. The puzzling coaxiality of the ab initio directions for the two P-proton coupling tensors might well be an artefact because of the very small magnitude of the coupling constants, which makes them sensitive even to small inaccuracies in the wavefunction. (2) I quite agree that experimental data should always constitute a guide for the development and refinement of theoretical methods.However, this applies to the primary data and not necessarily to their interpretation, which often relies on theoretical models or methods of much lower reliability than the experimental work. I should perhaps also add, although this is well known, that good agreement with experiment in itself does not guarantee the quality of a parametrized theory, since deficiencies and approximations in the theory may be masked by a suitable choice of parameter values. This applies to INDO as well as to any other semiempirical theory . Dr M. Iwasaki and Dr K. Toriyama (Nagoya, Japan) (communicated) We address Dr Lunell: In your answer, you still persist in comparing your ab initio directions with ours presented in our preliminary and Faraday papers, even though you admit the peculiarity of the ab initio directions.It is hard to believe that your ab initio results have sufficient reliability to judge which is closer to the true one, becauseGENERAL DISCUSSION 89 your ab initio directions show unrealistic coaxiality for the two P-proton coupling tensors. As is stated in our paper, further refinement is needed for our spectral simulations to reproduce the fine details of the substructures. Before we reproduce the observed substructures, we cannot decide which direction is closer to the true one. In the present stage, the comparison of the ab initio directions with our tentative ones is meaningless. Prof. F. Williams (Knoxville, Tennessee) said: In addition to the ring-closed cyclopropane radical cation described in the papers by Iwasaki et al.and Lunell et al., we have recently obtained experimental evidence for the ring-opened 'CH2CH2CH2+ species.' This isomer is formed from the cyclopropane radical cation in a thermal reaction which occurs at 80-84 K in the CFCl2CF2C1 matrix. The e.s.r. spectrum is well resolved at 108 IS and the parameters [a(2H,) = 22.4 G, a(2Hp) = 30.2 G, g = 2.00281 can be assigned to the trimethylene radical cation in which the two terminal CH2 groups are perpendicular to each other. Essentially, the electron spin is confined to one end of the molecule, the radical centre adopting a bisected conformation very similar to that of the n-propyl radical. We have similarly observed that the ring-closed form of the 1 , 1,2,2-tetramethyl- cyclopropane radical cation undergoes an irreversible monorotatory ring opening above 110 K in CFCl2CF2C1 to give an orthogonal ring-opened structure.2 In this case, however, the radical centre adopts an eclipsed conformation as expected for an a,&-dimethyl-substituted alkyl r a d i ~ a l .~ As far as we are aware, these observations are novel insofar as both the ring-closed and ring-opened forms of a radical cation have been characterized by e.s.r. spectros- copy. Note that these ring-opened radical cations do not isomerize by 172-hydrogen shifts to give olefin cations under these conditions. This conclusion is reinforced by the fact that attempts to generate the propylene radical cation4 directly from the olefin led only to allyl radicals, presumably by proton transfer from the propylene cation, at 108 K in the CFC12CF2C1 matrix, these allyl radicals being absent in the corresponding cyclopropane experiments.Moreover, the e.s.r. spectrum and hyper- fine parameters previously assigned to the propylene radical cation4 are quite different from those reported here for the trimethylene radical cation.' ' X-Z. Qin and F. Williams, Chem. Phys. Lett., 1984, 112, 79. * X-Z. Qin, L. D. Snow and F. Williams, J. Am. Chem. SOC., 1984, 106, 7640. K. S. Chen, D. Y. H. Tang, L. K. Montgomery and J. K. Kochi, J. Am. Chem. SOC., 1974,96,2201. K. Toriyama, K. Nunome and M. Iwasaki, Chem. Phys. Lett., 1984, 107,86 and references therein. Dr M. Iwasaki and Dr K. Toriyama (Nagoya, Japan) said: We have observed exactly the same spectral change and have interpreted the results by two models, one is the asymmetrical ring-opened structure, which is similar to that Prof.Williams has suggested, and the other is the twisted propylene radical cation, which may be formed by ring opening followed by intramolecular hydrogen shift. However, it is difficult to decide which is the case, so we have not published the result nor described the situation in our paper. Ring opening followed by intramolecular hydrogen transfer is also observed for c-C,Hi [see our paper and ref. (24) therein]. In this case the spectral change can be interpreted unequivocally.90 GENERAL DISCUSSION Dr M. R. Wasielewski (A.N.L., IZZinois) said: In the case of the radical cation of propane Dr Lunell proposes that C-H bond stretching may be responsible for the observed hyperfine splittings.I thus have two questions. ( I ) Has anyone used site-selective I3C enrichment to measure whether the C(1) or C(2) carbon atom shows enhanced hyperfine splitting? (2) The theoretical values of the I3C hyperfine interaction should be obtainable by default from Dr Lunell's calculations. What values of the "C hyperfine interaction do his calculations predict? Dr S . Lunell ( University of Uppsala, Sweden) replied: Let me first make a small clarification of my remark about the C-H bond stretching vis-u-vis observed proton hyperfine splittings. The point I wanted to make was that both the hyperfine splittings and the C--H bond-length variations can be understood from the shape and nodal pattern of the singly occupied molecular orbital in the different states of the propane cation.The C--H bond stretching does also affect the proton hyperfine splitting directly, but this effect is sufficiently small that it can be overlooked in the present context. In answer to Dr Wasielewski's questions, I do not know of any e.s.r. work on C-enriched propane to date. The calculated 13C splittings (table 1 ), however, indicate that such an experiment would be very illuminating, since the Cjl) and C(2) splittings are sufficiently different in the different electronic states to provide an independent check on the assignments obtained by other methods. We will, indeed, look into the possibilities of performing such an experiment in the near future. 13 Table 1. Isotropic I3C hyperfine coupling constants (in G) for the different carbon atoms in C,H: 2 ~ i aUHF -23.2 13.2 HFA A -7.2 11.3 -3.5 -14.0 -0.9 -4.7 2Ai aUHF -20.9 8.9 aUH FAA - 14.2 7.7 Dr W.Siebrand (N.R.C., Ottawa, Canada) said: My comment concerns the paper of Dr Lunell and also that of Dr Iwasaki. Dr Lunell calculates that the lowest three states in the propane cation are close in energy and argues on the basis of observed e.s.r. spectra that their ordering may be solvent-dependent. In cases such as this it is not immediately obvious that the Born-Oppenheimer approximation applies. It is possible that these states are subject to strong non-adiabatic mixing. Thus although this cation is not subject to a Jahn-Teller effect, similar to that discussed by Dr Iwasaki, it may be subject to a pseudo-Jahn-Teller effect if the coupling between two electronic states is larger than their separation.In my opinion this aspect deserves further investigation.GENERAL DISCUSSION 91 The breakdown of the Born-Oppenheimer approximation is of course manifest in the Jahn-Teller-active cation radicals studied by Dr Iwasaki. However, I am puzzled by his remark that the observed e.s.r. spectrum of the cyclopropane radical cation cannot be explained in terms of a thermal distribution over different levels but requires a model in which the electron jumps between different minima. I see no contradiction between these two pictures. In the Jahn-Teller system the lower vibronic levels represent states that are partially delocalized, their degree of localiz- ation depending on the depth of the trigonal minima induced by quadratic electronic- vibrational couplings.In principle, the e.s.r. spectrum can at all temperatures be expressed in terms of these eigenstates. It should be realized, however, that these states are not electronic but vibronic, since the Born-Oppenheimer approximation does not apply here. Dr M. Iwasaki (Nagoya, Japan) replied: (1) In Jahn-Teller-active species, generally, the energy difference Vo of the two distorted structures having an opposite sign of the distortion coordinate is believed to be very small as compared with the zero-point energy if the cations are free. However, in our case Vo is considerably higher than the zero-point level and is affected by the environment, so that a relatively high value of Vo is supposed to be due to a matrix interaction.In this sense we termed our case a ‘matrix-assisted Jahn-Teller distortion’. What is most surprising is that the direction of the distortion coordinate is determined uniquely regardless of environment. Thus we suggest that a matrix interaction assists in the stabilization of a given distorted structure, which is determined by the intramolecular interaction. In this sense the crossing of the level ordering observed for the propane radical cation is unusual. However, this is only one exceptional case among more than 35 alkane radical cations studied so far. The nature of the matrix interaction which causes this level crossing is not clear, and is not evident if one has to consider the propane radical cation as a pseudo-Jahn-Teller species.(2) Strictly speaking, the problem must be treated quantum-mechanically by solving Mathieu’s equation. However, I have treated the problem rather classically because our experimental results show that a relatively high potential barrier as compared with kT results in a localized oscillator in each minimum in the trough at 4.2 K. There seems to be misunderstanding concerning our statement that the temperature dependence of the spectrum cannot be explained by the change of the Boltzmann populations of the two near-degenerate states, as is often assumed in solution e.s.r. studies of Jahn-Teller-active species. The near-degenerate states mean 2Al and 2Bl for c-C,H; for example, the energy difference ( A E ) of which is usually thought to be very small.In such a case the method customarily used in solution e.s.r. is as follows: The temperature change of the coupling constant is assumed to be expressed by a(2Al)+ a(’&) exp( - A E / k T ) 1 +exp( - A E / k T ) a( T ) = where the *A, ground state is assumed and a stands for the coupling constant. In such a case, the hyperfine coupling constant shows a gradual change with increasing temperature. However, the observed spectra exhibit a sudden change, which is typical of jumping phenomena between the three sites, corresponding to the three Jahn-Teller potential minima. Of course, we have to take the zero-point oscillational92 GENERAL DISCUSSION effect for a low potential barrier, i.e. the mixing of the 2Al and 2Bl states. However, we have neglected such an effect, because strictly speaking we have to solve Mathieu's equation for the low-barrier case.Instead, we have handled the problem phenomenologically. This is also the way customarily used. Dr S. Lunell (University of Uppsala, Sweden) also replied to Dr Siebrand. You bring up a very important problem and, in principle, I fully agree with you that the strength of the vibronic interactions should be investigated before one can make any safe statements about geometry or electronic state of any molecular system. Since we have not done such a study, I really should not try to answer your comment. I do believe, however, that one can get a rough feeling for the importance of these interactions in the present case from rather qualitative considerations.At the equilibrium geometry of the neutral molecule, it is clear that one will have a Jahn-Teller-like situation upon vertical ionization, because of the extremely small spacing between the 4 b l , 2b2 and 6a, orbitals, which all crowd into an interval of ca. 0.2eV. The lines in the photoelectron spectrum of C3H81 indeed show a structure very similar to the Jahn-Teller-active eg levels of C2H6. For the e.s.r. spectra, however, I do not think that vibronic effects need to be quite as important, for the following reason. Assume that we are interested in the vibronic mixing between a certain electronic state, e.g. the 'B2 state, and the remaining states. Following Fulton and Gouterman,' we can in the harmonic approximation express the total wavefunction for the system approximately as Here q and Q are of the electronic state of interest the electronic and nuclear coordinates, $k( q, Qo) the eigenfunctions Hamiltonian for the equilibrium nuclear configuration Qo of the and the functions &( Q ) determine the extent of vibronic mixing between the different electronic states.Fulton and Gouterman showed that under certain sim- plifying assumptions, vibronic coupling between two electronic states (c-( q, Qo) and &( q, Qo) can be ignored if I v , k < Q > i < < I w/ - WkI (3) where and Q is in the range of nuclear vibrations. It is naturally not a trivial task to estimate the magnitude of yk(Q) without performing the actual calculations, but Herzberg' suggests as a rule of thumb that the possibility of vibronic mixing should be considered whcnever I - Wk( < ca.I eV. From fig. 3 of our paper one can see that, at the equilibrium geometry of the 282 state, the other two states are ca. 3 eV higher in energy which, according to this suggestion, should make it unnecessary to consider vibronic effects for this state. The situation for the other states is similar.GENERAL DISCUSSION 93 It is, of course, possible that a more careful theoretical treatment will change these conclusions. There are also certainly points on the energy hypersurfaces where the different states are much closer and even cross. The 2B2 and 2A1 surfaces should, for example, cross at a C-C distance intermediate between their equilibrium values 1.480 and 1.600 A. Judging from the same figure, however, it seems likely that the barrier separating these two minima should be rather high, which again would make a vibronic mixing between them less plausible.' He I Photoelectron Spectra of Organic Compounds (Monograph Series of the Research Institute of Applied Electricity, Hokkaido University, 1978), no. 25. R. L. Fulton and M. Gouterman, J. Chem. Phys., 1961, 35, 1059. G. Herzberg, Molecular Spectra and Molecular Structure (Van Nostrand, Princeton, 1966), vol. 111. Dr S. F. J. Cox (Rutherford Appleton Laboratory) said: In reply to various queries from the floor about the freedom of methyl groups to rotate within molecules, let me confirm that there is considerable variation between different materials. The barrier heights to rotation may be deduced from magnetic-resonance or neutron- scattering measurements (usually of the proton TI minimum or of the tunnel-splitting frequency at low temperature) and these are found to vary over at least three orders of magnitude.' The materials for which these barrier heights have been determined are for the most part diamagnetic and this enormous variety is usually interpreted simply in terms of different degrees of steric hindrance in the immediate vicinity of the methyl group.My question is to what extent the barrier may be altered in the paramagnetic derivative of each material, as a result of conjugation or hyperconjugation. Le., to what extent does electron release from a C-H bond into the singly occupied orbital (I) affect the barrier to rotation, and is the effect likely to be dominant, or only a small perturbation? An illustrative example is given below.It may be that the answer depends on whether the species is ionic or neutral. Thus Prof. Symons has demonstrated that methyl groups can be essentially frozen in radical cations (11) where the positive charge favours this electron release.2 On the other hand, in radicals formed by radiolytic ejection of hydrogen the tunnel splitting is always found to be larger, so that the barrier to rotation must be smaller, than in the original molecule. CH3-C (molecule) CH3-C (radical) ,sp2 orbital 400 K only barrier 200 MHz splitting (111) __* sp3 orbital 2000 K barrier 100 kHz splitting The numerical values in (111) are for methyl malonic acid, which exhibits this be haviou r .3 An elucidation of this question would be particularly valuable in interpreting the additional barriers to rotation which appear upon substitution of the methyl protons, for instance with heavier deuterons or lighter (positive) muons.4 Is this94 GENERAL DISCUSS ION isotope effect uniquely steric or dynamic in origin or is there, as Prof.Symons has ~uggested,~ an additional electronic contribution? ' S. Clough, A. Heidemann, A. J. Horsewill, J. D. Lewis and M. N. J. Paley, J. Phys. C, 1981,14, L525. M. C. R. Symons, personal communication. S. Clough, personal communication. M. J. Ramos, 0. McKenna, B. C. Webster and E. Roduner, J. Chem. Soc., Faraday Trans. 1, 1984, 80, 267. M. C. R. Symons, Hyperfine Interactions, 1984, 17-19, 793. Dr E. Roduner (University of Zurich, Switzerland) commented: I would like to respond on the questions posed by Dr Cox on barriers to internal rotation of methyl groups in radicals and on their isotope dependence.Apparently, the results for methyl rotation in diamagnetic molecules are usually understood well in terms of a potential barrier with threefold symmetry. In a radical, CA2X-CB,, all odd terms in the Fourier series expansion of the potential are zero as long as the radical centre is planar and the leading term is of twofold symmetry. The barriers are therefore not directly comparable. Furthermore, for X = A the twofold and fourfold terms drop out, and the leading term is of sixfold symmetry,' obviously associated with a very small barrier. Let us consider the ethyl radical and its isotopic derivatives (A, B, X = hydrogen isotopes Mu, H, D).An ab initio calculation by. Pacansky and Dupuis2 yields a sixfold potential of 630 J mol-' (75 K) for CH,-CH,. Calculations are performed for different rigid geometries and exclude dynamic effects. The resulting energy hypersurface is conventionally termed the electronic energy. Within the Born- Oppenheimer (B.O.) approximation it is the same for all isotopic species. The effects of internal rotation and vibration lead to isotope-dependent average nuclear coordin- ates and thus to different positions on this surface. Although isotopic substitution affects the average distribution of electron density the origin of the effect is dynamic. Experimental values for twofold barriers are obtained from the analysis of the temperature dependence of p- hyperfine coupling constants, assuming that they are a function of internal rotation only.For CHD2-CD2 V, = 385 J mol-' (46 K) was reported,' for CMuH,-CH, V, = 2963 J mol-' (356 K) was found.3 These drastic isotope dependences were explained as dynamic-steric effects. The lighter isotopes have higher zero-point vibrational amplitudes and appear bulkier (higher van der Waals radii). Furthermore, owing to anharmonicity, the average C-X bond length will be slightly larger for the lighter isotope. This certainly facilitates what Dr Cox calls electron release from the C-X bond into the singly occupied orbital. It is reflected in the isotopic dependence of the constants in the cos2 8 law.3 It should not be subsumed under electronic effects.Of course, the B.O. surface is always an approximation which leaves the option of an electronic effect. However, this was shown to be reasonably small, even for molecules containing MU.^ ' R. W. Fessenden, J. Chim. Phys., 1964, 1570. ' J. Pacansky and M. Dupuis, J. Am. Chem. SOC., 1982, 104, 415. M. J. Ramos, D. McKenna, B. C. Webster and E. Roduner, J. Chem. SOC., Faraday Trans. 1, 1984, 80, 267. D. McKenna and B. C. Webster, J. Chem. Soc., Faraday Trans. 2, 1984, 80, 589. Prof. M. C. R. Symons ( University of Leicester) (communicated) Dr Rpduner has raised the interesting question of the source of the isotope effect in R,C-CH,Mu radicals. There are two factors involved, one being the preferred orientation, which tries to maximise u-T overlap between the 2p, oribital on carbon and the C-Mu bond, and the other is any residual isotope effect once this has been allowed for.GENERAL DISCUSSION 95 The marked experimental enhancements detected are normally explained in terms of the ‘steric’ factor which makes Mu ‘larger’ than H, as mentioned by Dr Roduner.I have recently suggested that this may be less important than what I consider to be an ‘electronic effect’,’ despite the fact that Roduner says that the phenomenon ‘should not be subsumed under electronic effects’. We have moved forward in that there is now agreement that the phenomenon may exist, namely that hyperconjuga- tive electron release may be more efficient from C-Mu than from C-H bonds. This will contribute to both effects mentioned above. However, if this is the case, then it is, surely, ‘electronic’, since it involves electrons.We should also consider the possibility that, when in the preferred conformation, the muon itself moves as well as the electrons, as in (I), so as to increase the extent of the interaction. I M. C. R. Symons, in Muon Spin Rotation and Associated Problems, Part ZI, ed. T. Yamazaki and K. Nagamine (J. C. Baltzer AG, Basel, 1984), p. 771. Prof. T. J. Kemp ( University of Warwick) said: Concerning the doubts expressed by one questioner about the role of the protonated form of cyclopentadiene as the photoactive species leading to the cyclopentadiene radical cation, is it not the case that the presence of these precursor species in CF3C02H has been confirmed unequivocally by n.m.r.spectroscopy? Prof. A. G. Davies (University College, London) replied: Yes, the formation of the protonated species from most of the methylated cyclopentadienes has been established by n.m.r. and U.V. spectroscopy but usually in stronger acids such as H,SO, and FS03H [e.g. ref. (l)]. ’ N. C. Deno, H. G. Rickley, N. Friedman, J. D. Hodge, J. J. Houser and C. U. Piltman, J. Am. Chem. SOC., 1963,85,2991; N. C. Deno, N. Friedman, J. D. Hodge and J. J. Houser, J. Am. Chem. SOC., 1963, 85, 2995; N. C. Deno, J. Bollinger, N. Friedman, K. Hofer, J. D. Hodge and J. J. Houser, J. Am. Chem. SOC., 1963, 85, 2998. Prof. W. J. Albery (Imperial College, London) said: First, can Prof. Davies give us any estimate of the quantum efficiency of his systems? Secondly, it may be possible to follow the production of H2 in situ using the membrane electrode developed by Dr A.Mills while he was at the Royal Institution. Prof. A. G. Davies (University College, London) replied: (1) As yet, we cannot give any quantative estimate of the quantum efficiency. We have measured the rate of decay of the Me5C; radical and have found it to be diffusion controlled. Its e.s.r. spectrum, however, is more intense than we normally observe for such a short-lived radical, and the most we can say is that the quantum efficiency for the formation of Me&; from Me,C,H is greater than that of most of the reactions which we have studied. The radical cations are relatively very persistent. The high intensities of the spectra reflect the low rate of decay rather than rate of96 GENERAL DISCUSS ION formation, and we have no idea of the quantum efficiency.Although we believe we know the bare bones of the mechanisms involved, alternative mechanisms cannot be ruled out, and the systems warrant a thorough photochemical study. (2) We have used laser Raman spectroscopy to analyse the dihydrogen ( H2, HD and D2) evolved from isotopically labelled pentamethylcyclopentadiene.' This is a very convenient method, which can be carried out on the sealed e.s.r. tube. Mills' electrochemical method might be useful for following the kinetics of the reaction. ' A. G. Davies, E. Lusztyk, J. Lusztyk, V. P. J. Marti, R. J. H. Clark and M. J . Stead, J. Chern. Soc., Perkin Trans. 2, 1983, 669. Prof. M. C . R. Symons (University of Leicester) said: We now direct attention to Prof.Williams' paper. ( 1 ) With respect to your qualitative view of spin delocalisation onto S protons, this receives considerable support from our study of ( '3CH3)2CO'f cations. Our estimated spin density on l3C is only slightly less than that onto 'H for H,CO'+ cations, showing that, as you suggest, delocalisation into the u frame by normal hyperconjugation is extensive.' (2) The problem of the reversible loss of chlorine hyperfine coupling for CH,CHO+ and CH,CH,CHO+ cations is most interesting. Our suggestion was that this is due to a reversible breaking of the O-.-Cl v bond, postulated for these complexes,* whilst you favour the idea of a rotational averaging. Obviously, the former RCHOf-C1CFCl2 S RCHO" + CFC13 ( 1 ) will result in loss of chlorine coupling, but you suggest it should also result in a change in the e.s.r.parameters for the RCHO" group. On the other hand, the isotropic coupling for RCHO" should not alter if the complex rotates as a whole unit, whilst the chlorine coupling could average to near zero provided A, is negative and ca. half the value of All. Just this situation is observed for R'X- adducts (X=Cl, Br, I), where the isotropic coupling is very small, and A , (hal) must be n e g a t i ~ e . ~ - ~ However, it seems to me that such a rotation must involve the whole unit [RCHO+-- .G1CFCl2], which is perhaps surprising. Since the large proton coupling is averaged, the RCHO' unit must rotate, but if that involved the RCHO+ unit alone, that would require bond breakage.If the specific CFC13 molecule were able to rotate on its own, this again implies reversible bond-breakage. It is conceiv- able that both units librate so extensively that much of the anisotropy is lost, but this still requires that A, (Cl) be negative. If A, is negative, the estimated spin density on chlorine is ca. 19%, which is quite high. I have previously shown that for genuine u* complexes there is a good correlation between the s character (estimated from AiSJ and the p character (estimated from 2B).' Using a percentage s character of 0 and percentage p character of 19, the correlation is not obeyed (as indeed is the case for alkyl-radical halide-ion adducts). However, if A , is ca. 9.5 G and positive, we get a percentage s character of ca. 0.5 and a percentage p character of ca.9.5. These results fit nicely onto the line which correlates the results for many u* complexes. Regarding your case that for bond-breaking a sudden change should be observed, I am not sure that this is correct. What I expect, and what seems to me to be observed, is that, for dissociation, both species should be observed together over a range of temperatures. Given some broadening of the Cl hyperfine features caused by librational effects, that seems to be what both of us observe [Prof. Williams' fig. 2 and ref. (S)].GENERAL DISCUSSION 97 If A , is positive, as we suggest, the extent of delocalisation is about half that required on your interpretation. Thus the bond is weaker and reversible dissociation is more probable. Furthermore, for such a weak interaction I see no reason why there should be any major change in A('H) for RCHO' cations.Certainly for strongly bonded complexes, such as RS-SR- or R2S'SR2+, there are major changes, but can this argument be carried through to very weak complexes with < 10% delocalisation? It would be nice to have this issue settled since, if our hypothesis is correct, we can attempt to estimate the bond strength for these interesting complexes. P. J. Boon, L. Ham's, M. T. O h , J. L. Wyatt and M. C. R. Symons, Chem. Phys. Lett., 1984, 106, 408. M. C. R. Symons and P. J. Boon, Chem. Phys. Lett., 1983, 100, 203. M. C. R. Symons and I. G. Smith, J. Chem. Soc., Perkin Trans. 2, 1979, 1362. M. C . R. Symons and I. G. Smith, J. Chem. SOC., Perkzn Trans. 2, 1981, 1180.I. G. Smith and M. C. R. Symons, J. Chem. Soc., Faruday Trans. 1, 1985,81, 1095. M. C. R. Symons and I. G. Smith, J. Chem. SOC., Faraday Trans. I , 1981, 77, 2701. P. J. Boon, M. C. R. Symons, K. Ushida and T. Shida, J. Chem. Soc., Perkin Trans. 2, 1984, 1213. ' M. C. R. Symons, Chem. Phys. Lett., 1980, 72, 559. Prof. F. Williams (Knoxville, Tennessee) remarked: Although the results presen- ted in our paper establish that the strongly coupled hydrogens in the cyclohexanone and adamantan-2-one radical cations are in the S positions, it was pointed out that the assignment to the equatorial rather than the axial hydrogens depends on the adoption of the 'trans rule'. To the best of our knowledge, however, this rule has never been proved rigorously, at least in e.s.r.studies. It is therefore of interest to report that additional work carried out since submitting the manuscript has verified this rule through the direct proof of the 'H hyperfine coupling assignments for the adamantan-2-one cation. These studies were made possible through the synthesis of the stereospecifically labelled monodeuteroadamantan-2-ones 1 and 2 in high isotopic purity by Prof. S. F. Nelsen and his group at Wisconsin. 1 2 First, the e.s.r. spectrum of 1" shown in the upper half of fig. 2 demonstrates that the introduction of a deuterium atom into one of the two y positions of the adamantan-2-one radical cation has no effect on the main quintet pattern [A(4H) = 22.5 GI but changes the substructure from a triplet to a doublet [A( 1H) = 7.0 GI.This establishes that the 7 G triplet substructure in the spectrum of the fully protiated cation (fig. 5 of our paper) originates from coupling to the 2y hydrogens, as proposed. Secondly, the contrasting effect of deuterium substitution into one of the equatorial S positions is revealed by the e.s.r. spectrum of 2" shown in the lower half of fig. 2. The main pattern is now seen to consist of a quartet [A(3H) = 22.4 GI instead of a quintet, proving that the strongly coupled hydrogens are indeed in the 6 equatorial positions. The loss of resolution in the substructure is also explained by the additional 3.5 G coupling to the one deuterium ( I = 1) nucleus. These results clearly establish the 'trans rule' and will be reported in more detail elsewhere.' ' S. F. Nelsen, D.Kapp, L. D. Snow and F. Williams, to be published.98 GENERAL DISCUSSION 9273.0 MHz 9271.3 M H z 0 m r h 0 Fig. 2. First-derivative e.s.r. spectra of the monodeuteroadamantan-2-one radical cations 1" (upper spectrum) and 2*+ (lower spectrum) at 140K. The cations were generated by y- irradiation of solid solutions of the parent compound in CFC13 at 77 K. Dr M. R. Wasielewski (A.N.L., Illinois) said: Prof. Williams' long-range hyperfine splittings nicely illustrate the ability of the all-trans a-bond framework to transmit spin density over many bonds with good efficiency. As we know, this problem is related to that of the electronic coupling present between an electron donor-acceptor pair spaced by rigid hydrocarbon spacers. Since his 6 splittings are so large, has he tried, or does he contemplate trying, to measure longer-range splittings in e.g.a 2-keto decalin system or a 3-keto steroid molecule? Prof. A. Weller ( Max-Planck-lnstitut, Gottingen, West Germany) added: Would Prof. Williams agree that the strong equatorial hyperfine coupling constant aHh involves through-bond orbital interaction rather than through space? Prof. F. Williams (Knoxville, Tennessee) said: Yes. A through-bond orbital interaction is strongly supported by the stereospecificity of the 6- hydrogen coupling in the adamantan-2-one system.' Moreover, the coupling to the y hydrogens is about a factor of three smaller than that to the more distant equatorial 6 hydrogens, a result which is clearly incompatible with a through-space interaction.' F. Williams, Furuday Discuss. Chem. SOC., 1984, 78, 97; S. F. Nelsen, D. Kapp, L. D. Snow and F. Williams, to be published. Prof. W. Bernhard (University of Rochester, New York) said: Prof. Williams' analysis of long-range hyperfine interactions will most likely be helpful in solving a problem that has been puzzling me for some time. There are a fair number ofGENERAL DISCUSSION 99 alkoxy radicals that have been characterized using single-crystal e.s.r./ ENDOR. These studies provided the first measurements of the unusually large P - hyperfine couplings in oxy-centred radicals. However, in many of the alkoxy radicals there are additional weak hyperfine couplings that have been carefully measured using ENDOR.' Although these weak hyperfine interactions have been ascribed to y hydrogens, I have found it difficult to work out a self-consistent explanation based only on y-hyperfine interactions. From Prof.Williams' analysis it is now clear that we should go back and consider the rather likely possibility that at least some of these additional hyperfine couplings are due to hydrogens in S positions. ' H. C. Box, E. E. Budzinski and G. Potienko, J. Chem. Phys., 1980, 73, 2052. Dr M. Iwasaki and Dr K. Toriyama (Nagoya, Japan) said: Prof. Williams has observed extremely large &proton couplings in a variety of aldehyde and ketone radical cations. Although these range from 12.5 to 27.5 G, cyclic ketone radical cations seem to give a larger coupling than that of aliphatic aldehyde cations. Now, if one considers a spin density of ca.0.25 on the P-carbon atom, the conventional pB, cos2 8 rule for P-proton coupling with B2 = 60 G can give a 8-proton coupling of ca. 15 G, if 8 = 0 is assumed. The B2 value of neutral alkyl radicals is 58 G. The largest coupling of 27.5 G corresponds to B, = 110 G assuming 8 = 0. This value is fairly close to our B2 value of 120 G determined for a series of branched alkane cations [ref. ( 5 ) in our paper]. So, a large 6-proton coupling may not be surprising and the origin of the long-range coupling is in the high-spin density on the P-carbon atom, as Prof. Williams called it relay of spin transfer. Now the conformation with respect to the C,-C, bond in the cyclohexane radical cation, which gives the largest &proton coupling of 27.5 G, is gauche rather than trans.In the case of methyl-substituted butane radical cations, the INDO calculations show that the gauche methyl proton gives a considerably lower value than that of the trans methyl proton [ref. ( 5 ) in our paper]. Compared with this result, the higher coupling value in the cyclic-ketone radical cations is puzzling. Prof. F. Williams (Knoxville, Tennessee) said: We very much appreciate the comments and suggestions by Dr Wasielewski and Prof. Weller. In answer to the question, our work has not progressed beyond the systems described in our paper and in my supplementary remarks. As pointed out, through-bond orbital interactions can contribute significantly to the efficiency of long-range electron transfer between donor and acceptor groups separated by a rigid hydrocarbon framework, the impor- tant parameter being the electronic coupling J.In view of the recent elegant work on the rates of intramolecular electron transfer in radical anions possessing two rr-electron groups linked by steroidal spacers,' we are particularly interested in the investigation of long-range hyperfine interactions in related skeletal systems of the type suggested by Dr Wasielewski. One cautionary note should be added, however, in regard to the possibility of observing longer-range splittings. While it is certainly true that the &hydrogen splittings reported in table 1 of our paper are large enough (12.5-27.5 G) to suggest that coupling to even more remote hydrogens might be detectable in favourable circumstances, such expectations must be tempered by the fact that splittings of < 4 G are not generally resolved in these freon matrices.I am glad to learn from Prof. Bernhard that our analysis of long-range hyperfine interactions may be of some assistance in clarifying some of the assignments for the weakly coupled protons in alkoxyl radicals. Clearly it would be helpful to have a quantitative treatment of long-range hyperfine interactions comparable to that100 GENERAL DISCUSS I 0 N developed some time ago by Prof. Bernhard for P-proton couplings in oxygen- centred radicals,2 and the comments by Dr Iwasaki and Dr Toriyama are pertinent on this point. Using a pB2 cos2 8 relation, their first comment quantifies our proposal that spin delocalisation from the carbonyl group into the sigma-bonded carbon frame, thus producing p = 0.25 at our C,, allows spin d b i t y to be relayed to the S, hydrogens via the trans or alignment effect.The close connection between our results for cyclic ketone cations and Dr Iwasaki's work on branched alkane cations [ref. (5) in his paper] became apparent to us after submitting our manuscript, and we regard the similar B2 values for the two systems as highly supportive of the model for spin transmission. For the acyclic aldehydes and ketones, the couplings to the trans S hydrogens are smaller than those for the rigid cyclic ketones by a factor of ca. 2 [table 1 of our paper]. This apparent discrepancy could be due to torsional oscillation in the non-rigid systems resulting in a value of (cos2 8) of < 1. In response to the last comment by Dr Iwasaki and Dr Toriyama, there seems no reason to believe that the gauche conformation for the cyclohexanone cation about the C,-Cy bond should have any significant effect on spin delocalisation from the carbonyl group into this bond, and hence on the magnitude of the 8-hydrogen splittings.The conformation about the C,-Cy bond can be varied in an acyclic system, and MNDO calculations for the propionaldehyde radical cation3 show only a small dependence for the 6-hydrogen splittings on the OC,C,C, dihedral angle. Irrespective of the conformation about the C, -C, bond, however, the calculations show that the &hydrogen splitting is extremely sensitive to the C,C,C,H, dihedral angle, the maximum Hs splitting being obtained in each case when this dihedral angle is 180', as expected for the trans effect.Similarly in the methyl-substituted butane radical cations, the trans (8 = 180') and gauche (8 = 60") methyl-proton splittings are determined by the C2C3CM,H, dihedral angle, the spin density now being in the C2-C3 bond. L. T. Calcaterra, G. L. Closs and J. R. Miller, J. Am. Chem. Soc., 1983, 105, 670; 1984, 106, 3047. W. A. Bernhard, D. M. Close, J. Hiittermann and H. Zehner, J. Chem. Phys., 1977, 67, 1211. S. F. Nelsen, personal communication. Dr W. Siebrand (N.R.C., Ottawa, Canada) said: In his paper Prof. Williams presents convincing evidence that the methyl groups in cations of simple aldehydes start to rotate (on the timescale of the experiment) above 120 K. In our work on tunnelling in a methanol glass, to be presented later during this Discussion, we see evidence for rotation down to 15 K or lower (on, admittedly, a much longer timescale).Other estimates can be found in the literature; they cover a wide range of values. Is there any serious discrepancy between these results? Prof. F. Williams (KnoxviZZe, Tennessee) said: In our experience, the onset temperature of methyl-group rotation in radicals is extremely variable and has to do with structural factors which are not always easy to predict. Therefore, I see no serious underlying discrepancy between Dr Siebrand's results and ours. This is a general question on which I am sure others in the audience may also wish to comment. Dr M. Iwasaki (Nagoya, Japan) said: It is well known that the hindering potential barrier to the rotation of CH3 group is very low when the radical carbon atom has an sp2 planar structure, whereas it becomes higher when the radical site becomes non-planar. For example, the hindering potential barrier in CH3CH2 is essentially zero, whereas it is 2.2 kcal mol-' in a bent CH3CF2.'GENERAL DISCUSSION 101 In radical cations of alkanes, in which the radical carbon atom has bent structure, the hindering potential barrier is expected to be relatively high.’ K. S. Chen and J. K. Kochi, J. Am. Chem. SOC., 1974 96, 794. Dr M. Iwasaki (Nagoya, Japan) said: I address my remarks to Prof. Sevilla. I would like to make a comment on the rearrangement of radical cations, more specifically the intramolecular proton transfer in ester radical cations.In principle, gas-phase mass spectroscopy cannot give direct evidence for intramolecular rearrangement because there is no change in mass. However, e.s.r. spectroscopy can give direct evidence. Sevilla et al. have assumed intramolecular proton transfer based on solute concentration dependence. On the other hand, we have obtained direct e.s.r. evidence for intramolecular proton transfer from the ester alkyl group to the carbonyl oxygen atom. In addition, the out-of-plane conformation of the transferred proton is also determined by anisotropic spectral simulations [ref. (21) in Prof. Sevilla’s paper]. It is to be mentioned that the transfer to the ether oxygen atom is also excluded by the simulation. Prof. M. C . R. Symons (University of Leicester) said to Prof.Sevilla. In our own work on ester cations, which has run parallel with yours,’ we also supposed that the species derived from the ( HC02Me)+-C1CFC12 adduct on annealing was the rr cation, but, like you, we are inclined to agree that the new results of Iwasaki et aL2 are better interpreted in terms of the proton-transfer cation (I). I would like to stress that we were both misled by the fact that theoretical estimates of the ‘H coupling for the rr cation agree remarkably well with the experimental results now assigned to (I). I was also misled by the clear e.s.r. evidence for a non-rearranged cation obtained from various a m i d e ~ . ~ These cations are very similar to the ester n cations, except that the spin-density-on the -NR2 group is greater than that on the -OR group, whilst that on the carbonyl oxygen is corre- spondingly reduced.Problems remain, however. If, as is now ~uggested,~ the first-formed cation at 4 K is indeed the n cation rather than the a cation, why should the structure change to the a-cation adduct (11) on annealing, and why should this be formed uniquely on irradiation at 77 K? If this is correct, then it sheds interesting light on the charge-transfer mechanism for cation formation in this matrix. It also implies that the energies of the n and n cations must be very close. However, it is noteworthy that A(’H) for the unique proton for this complex is only ca. 17 G. Possibly this large reduction relative to the value for the aldehyde cation complexes (ca. 140 G) arises in part because the structure is not actually ‘planar’, as is implied in (11), but twisted, so that the orbital on the ester group is neither pure a nor pure n.102 GENERAL DISCUSSION I would like to stress that these ester cations and the lactone cations that our two groups are studying jointly, represent perhaps the most recalcitrant and contrary species that I have yet encountered in this field.Genuine ester cations which have neither undergone rearrangement nor formed bonds to solvent molecules seem to be rare indeed, in marked contrast with almost any other system so far studied. ' D. N. R. Rao, J. Rideout and M. C . R. Symons, J. Chem. SOC., Perkin Trans. 2, 1984, 1221. ' M. Iwasaki, H. Muto, K. Toriyama and K. Nunome, Chem. Phys. Lett., 1984, 105, 413. D. N. R. Rao and M.C. R. Symons, Chem. Phys. Lett., 1982,93, 495. Prof. M. D. Sevilla (Oakland University, Michigan) replied: There is an alterna- tive possibility to the discrepancy found between the 4 and 77 K results for methyl formate. Dr Iwasaki reports the cation at 4 K and suggests the lack of rotational motion at this low temperature does not allow for the u*-complex formation. Dr Iwasaki has shown that the uncomplexed cation is unstable at 77 K and undergoes proton transfer to produce 'CH20C(OH+)H. The e.s.r. parameters for this proton- transfer radical are very close to those reported by Dr Iwasaki at 4 K for the cation. As a consequence the 4 K species may also be a proton-transfer radical not the original cation. It is true that molecular-orbital calculations suggest that the n cation and the proton-transfer species should have similar couplings so this problem may be a difficult one to resolve.A possible solution may be 4 K experiments on deuterated methyl formate (CD,OCHO) which does not undergo deuteron transfer even at 77 K. Dr M. Iwasaki (Nagoya, Japan) said: I turn first to Prof. Sevilla. The evidence for the oxygen n cation of methyl formate is weak as compared with the firm evidence for the carbon-centred r-radical cations formed by McLafferty rearrangement. However, at 4 K methyl and ethyl formate radical cations exhibit essentially the same three-line spectra, the feature of which is more like arising from the two equivalent &proton couplings. Even if the spectra at 4 K could be attribu- table to the carbon-centred T radical having the two Q protons, the resemblance of the spectra obtained from methyl and ethyl formate may not be explained unless we assume that CH, group of ethyl formate once shifts to the carbonyl group to give a three-line spectrum and then CH2 returns to the radical carbon atom: O+ 0-CH, OH 0-CH2 +/ + / .CH3 + H-C --* H-C-CH2 // \ / \ H-C \ / 0-CH, O-CH2 If this sort of rearrangement can occur, the 0, ester cation might not exist. However, I would like to point out that the SOMO of this sort of molecule is well known to exhibit a u-T crossing by a slight change of the bond angle and other geometries. Prof. Sevilla has also shown this by INDO calculations and we have confirmed it. I turn now to Prof. Symons. Methyl formate in CFCl, gives a u* complex with a matrix C1 atom when irradiated at 77 K, whereas the complexing is suppressed when irradiated at 4 K.This seems to suggest that the local molecular reorientation is required for the cation to form a u* complex. Originally an electron may be ejected from the in-plane oxygen u orbital forming 0, cation radicals. However, it changes into more stable oxygen-centred n- cation radicals because the competitive u* formation is sup- pressed by retardation of molecular reorientation at 4 K, whereas at 77 K the 0,GENERAL DISCUSSION 103 cation radical can complex with CFC1, because the local molecular reorientation becomes possible at 77 K. The yield of 0, cation radicals at 4 K is relatively small, and upon warming to 77 K the u" complex is formed, with a decrease of the remaining matrix cation radicals.This also indicates that the complexing is possible when positive charge transfer to the solute occurs at elevated temperature. On the other hand, we have observed that complexing of acetaldehyde radical cations cannot be suppressed at 4 K (unpublished work). This is because CH,CHO'+ cannot be converted into an oxygen T cation. Thus we may have to take the two factors into consideration at 4 K: one is the retardation of the change of the local geometry, including the deformation of the radical cation itself, and the other is the change of the unpaired electron orbital from u to T. Prof. M. C. R. Symons ( University ofLeicester) (communicated) Whilst I accept the mechanism for ester cation formation proposed by Dr Iwasaki, and also the reservations made by Prof.Sevilla about the 4 K results, I think the point is not entirely answered. The following equations for the formation of ester (E) cations summarise the problem as I envisage it: (CFCl,)+ + E ---* (FClZCCl-. -E)+ (1) (CFCl,)++E -+ CFCl,+E; (2) (3) (4) (CFCI,) + E + CFC1, + EZ E: + CFC13 --* (FC12CC1- * *E)+ (where EfR is the rearranged final product). Although reaction (3) is possible, there is no direct evidence for it since EZ has not been detected. The solvent cr complex can be formed directly, as in reaction ( 1 ), provided the correct relative orientations can be achieved, rather than via charge transfer followed by adduct formation, as in reactions (3) and (4). Similarly, the T cation, E:, can be formed directly, as in reaction (2), there being no need to postulate reactions (3) and (5).Unfortunately, the irreversible step (6) apparently occurs so readily that it seriously interferes with attempts to learn more about this reaction. Dr W. Siebrand (N.R.C., Ottawa, Canada) said: The proton transfer in the ester cations which Prof. Sevilla described resembles the transfer between keto and enol tautomers which is known to proceed by hydrogen tunnelling.'.2 This, combined with the deuterium effect he observes, suggests that the present transfer also proceeds by tunnelling. Has this been confirmed? ' K-H. Grellmann, H. Weller and E. Tauer, Chem. Phys. Let?., 1983, 95, 195. W. Siebrand, T. A. Wildman and M. Z. Zgierski, J. Am. Chem. SOC., 1984, 106, 4089. Prof. M. D. Sevilla (Oakland University, Michigan) said: We have recently found that the (T* complex formed between [2H3]methyl formate and CFC1, is far more stable than that formed with methyl formate itself and it undergoes a different chemistry on dissociation.While methyl formate undergoes reaction ( 1):I04 GENERAL DISCUSSION X +- // 0' II OH+ II ( CT* complex) the deuterated compound undergoes loss of the formyl proton: X +- ,/ Of 0 where X = CFC1, and S represents a solute molecule. Williams' in other systems and are associated with tunnelling. Such 'all or none' isotope effects have been reported previously by Wang and I J. T. Wang and F. Williams, J. Am. Chem. SOC., 1972, 94, 2930. Dr M. Iwasaki (Nagoya, Japan) added: We have observed intramolecular proton transfer in a variety of ester radical cations even below 77 K and in some cases it takes place even at 4 K .The reaction at such low temperatures suggests that the tunnelling process may be involved. In addition, the steric factor is important for the occurrence of this proton transfer at low temperature, suggesting that the reaction rate depends upon the tunnelling distance. Prof. M. C. R. Symons ( University of Leicester) (communicated) Regarding the solvent adducts exhibiting hyperfine coupling to chlorine nuclei observed for certain aldehyde and ester cations, we now have several examples of such interaction. For example, RCl" and RBr" cations show this coupling, although RI" cations do not.''2 Also, as has also been observed by Williams and his coworkers, (MeO),PO'+ cations show a large coupling to ~ h l o r i n e .~ We have also sometimes observed a species having coupling to two equivalent chlorine nuclei [A11(35Cl) = 101 GI, which may be due to two solvent molecules sharing a 'hole'. [Neb. The isostructural Cl, ion has an identical Ai,(35C1) coupling.] Furthermore, the cation of Me,C-CN, which probably has an ionisation potential slightly greater than that of CFC13, actually forms a complex having A11(35C1) of 125 G, which is greater than that expected for 50% electron sharing4 We suggest that this adduct has structure (I), since the isotropic coupling to 14N (ca. 70 G) is very large. Me,C-C_N-Cl I c / I \ CI C1 F ' G. W. Eastland, D. N. R. Rao, J. Rideout, M. C. R. Symons and A. Hasegawa, J. Chem. Rex ( S ) , ' G. W. Eastland, S.P. Maj, M. C. R. Symons, A. Hasegawa, C . Glidewell, M. Hayashi and 1983, 258. T. Wakabayashi, J. Chem. SOC., Perkin Trans. 2, 1984, 1439. G. D. G. McConnachie and M. C. R. Symons, J. Chem. Rex ( S ) , 1985, 54. J. Rideout and M. C. R. Symons, J. Chem. Rex ( S ) , 1984, 268.GENERAL DISCUSSION 105 Prof. F. Williams (Knoxville, Tennessee) said: I should like to report e.s.r. evidence for a strong interaction between the trimethyl phosphate radical cation and the CFC13 matrix' which resembles Prof. Sevilla's results reported some time ago on the u* complex between the methyl formate cation and CFC13.* The e.s.r. spectrum of the trimethyl phosphate radical cation in CFC13 is characterised by a large anisotropic chlorine hyperfine interaction [All(35C1) = 85 GI and a 3'P doublet splitting of ca.26 G, the chlorine coupling being very close to that (84.4 G) observed for the methyl formate complex.2 Since the ionisation potentials of trimethyl phophate3 and methyl formate4 are both equal to 10.81 eV, the similarity in the chlorine hyperfine couplings supports the idea that strong cation-solvent interactions are facilitated when the ionisation potential of the solute is within ca. 1 eV of that for the solvent (ionisation potential of CFC13 = 11.78 ev).' Localisation of the spin density has also been suggested as a favourable factor contributing to complex formatioq2 and it is noteworthy that the unperturbed SOMO of these two cations satisfy this criterion. Thus in the methyl formate radical cation, the unpaired electron is largely localised in a 2p, orbital on the carbonyl oxygen, while in the trimethyl phosphate radical cation the unpaired electron is similarly in a non-bonding 2p orbital on the unique oxygen.It is also interesting that the u* complex between the trimethyl phosphate cation and CFC13 undergoes an irreversible thermal or photoinduced dissociation.' This dissociation is accompanied by formation of the (MeO),P+( 0H)OCH; radical formed by hydrogen-atom transfer from one of the methyl groups to the unique oxygen,' a rearrangement which again is similar to that recently described for the dissociation of the methyl formate cation-CFC1, u* ~ o m p l e x . ~ - ~ Contrary to the initial reports,2 the uncomplexed methyl formate cation is not observed following diss~ciation.~-~ It appears, therefore, that the nature of the bonding between the radical cation and the CFC13 solvent 'protects' these very reactive u radicals from undergoing the irreversible internal hydrogen-atom transfer reaction until dissoci- ation takes place, as depicted in scheme 1.1+' Scheme 1. Turning to the smaller chlorine hyperfine interactions between the aldehyde radical cations and CFC13 which are reported in our paper, we have presented reasons why the reversible loss of the 35Cl coupling with temperature in this case is more likely to be due to a motional averaging of the hyperfine anisotropy associatedI06 GENERAL DISCUSSION with the cation-solvent complex than to a dissociation of the complex. In the light of the results for the methyl formate5-' and trimethyl phosphate' complexes, an additional argument against dissociation is that internal hydrogen-atom transfer to oxygen is not observed for the propionaldehyde radical cation between 120 and 140 K despite the fact that the chlorine interaction is already lost and methyl group rotation sets in between these temperatures [see fig. 2 ( a ) and 3 ( c ) of our paper]. However, it must be admitted that this argument loses its force if the propionaldehyde cation adopts conformation 2, which is unfavourable for this hydrogen-atom transfer reaction. The possibility of dissociation to give an unreactive aldehyde radical cation can therefore not be ruled out at this time. Despite the fact that the substructure in the e.s.r. spectrum of the acetaldehyde cation in CFCI3 (fig. 1 of our paper) cannot be analysed in detail, we have confidently attributed this substructure to a matrix interaction resulting mainly from one 35Cl nucleus because deuteration of the methyl group produces only linewidth changes in the e.s.r. spectra of MeCHO'+ and MeCD0'+.8 This conclusion has a corollary, namely that the aforementioned substructure should be eliminated in the e.s.r. spectrum of the acetaldehyde cation in a neon matrix. Prof. L. B. Knight Jr has recently shown this to be the case.' The species was generated using an open-tube neon discharge lamp with 17 eV photons and then trapped in a neon matrix at 4 K, the technique being similar to that used in the previous investigation of the formal- dehyde radical cation." Small hyperfine couplings to the hydrogens of the methyl group, which were masked by the solvent interaction in the Freon matrix, are now clearly revealed together with I3C satellite lines in natural abundance for partially oriented CH3CHO'+. A detailed study is underway, this species being of particular interest because it is the first cation to be investigated by e.s.r. using both the Freon method emphasised in this Discussion and Knight's technique." ' X-Z. Qin, B. W. Walther and F. Williams, J. Chem. Soc., Ghem. Commun., in press. * ( a ) D. Becker, K. Plante and M. D. Sevilla, J. Phys. Chem., 1983, 87, 1648; ( b ) G. W. Eastland, D. N. R. Rao, J. Rideout, M. C. R. Symons and A. Hasegawa, J. Chem. Res. ( S ) , 1983, 258. V. I. Vovna, S. N. Lopatin, R. Pettsol'd and F. I. Vilesov, Khim. Vys. Energii, 1975, 9, 9. K. Watanabe, T. Nakayama and J. Mottl, J. Quant. Spectrosc. Radiar. Transfer, 1962, 2, 369. M. Iwasaki, H. Muto, K. Toriyama and K. Nunome, Chern. Phys. Lett., 1984, 105, 586. M. D. Sevilla, D. Becker, C. L. Sevilla, K. Plante and S. Swarts, Faraday Discuss. Chem. Soc., 1985, 78, 71 M. D. Sevilla, D. Becker, C. L. Sevilla and S. Swarts, J. Phys. Chem., in press. L. B. Knight Jr, personal communication. * L. D. Snow and F. Williams, Chem. Phys. Lett., 1983, 100, 198. 10 L. B. Knight Jr and J. Steadman, J. Chem. Phys., 1984, 80, 1018.