NUCLEAR MAGNETIC RESONANCE SPECTRUM AND MOLECULAR STRUCTURE OF ALUMINIUM BOROHYDRIDE BY RICHARD A. OGG, Jr. AND JAMES D. RAY Dept. of Chemistry, Stanford University, California, U.S.A. Received 3rd February, 1955 High resolution HI and B11 magnetic resonance spectra are presented for liquid ALB3H12 and highly deuterium-substituted derivatives. In the case of H1 spectra the technique of double resonance is also employed, with saturation of either the B11 or A127 resonance. It has been found, using such spectra for identification, that at moderate temperatures AlB3H12 undergoes reversible dissociation into B2H6 and a hitherto unrecognized com- pound, described as A12B4H18. The equilibrium is characterized by an extraordinarily large standard entropy change. The nuclear magnetic resonance spectra of the new substance have also been studied.It is concluded that AlB3H12 is characterized by a bridge bond structure, analogous to that of B2H6, but that a dynamic process renders bridge and terminal protons in- distinguishable. Evidence is offered in support of the view that a quantum-mechanical tunnel effect is involved. The structure assigned to A12B4H18 has similar features, but it is concluded that proton tunnelling and actual rotation of borohydride groups are in operation. In the course of an extensive programme of study of high resolution nuclear magnetic resonance spectra of boranes, the fist results of which have appeared,l the occasion arose to examine the remarkable compound aluminium borohydride, AlB3H12.2 The molecular structure of this undoubtedly covalent substance has been the subjcct of investigations of its vapour by electron diffraction 3 and by infra-red spectroscopy.4 These have established certain structural features, notably the location of the aluminium nucleus at the centre of an equilateral triangle formed by the three boron nuclei.The most probable arrangement of the protons 4 would appear to involve approximately tetrahedral configuration in groups of four about the respectivc boron nuclei, with two protons in terminal position and two in bridge position between boron and aluminium nuclei. This structure would lead one to expect nuclear magnetic resonance spectra somewhat similar to thosc observed 1 for diborane, B2fhj. In particular, the proton reson- ance spectrum should present the overlap of a sharply defined multiplet due to boron spin-spin interaction with terminal protons and of a probably ill-defined broad multiplet due to the bridge protons.(The A127 nucleus has spin quantum number 5/2. Since the predominant isotope B11 has spin quantum number 3/2, twenty-four lines are cxpected, with only accidental coincidences.) The first, and in many respects still the most remarkable result of the nuclear magnetic resonance studies, was the observation of a completely structureless proton resonance spectrum for liquid AIB3H12 in the entire temperature range practically available (melting point - 64" C to normal boiling point 4- 44" C). The brcadth and absolute position of this spectrum on the magnetic scale coincided roughly with those of borohydride on and diborane,l and were independent of temperature.In the course of experiments designed to elucidate the above phenomenon, it was observed that samples of AIB3H12 vapour heated in sealed glass vessels at -I- 80" C for periods of the order of 10 h, and then rapidly liquefied by cooling, displayed a strongly modified proton magnetic resonance spectrum. Such modified samples when allowed to remain at room temperature were observed 239 - -240 ALUMINIUM BOROHYDRIDE to revert slowly to the “normal” state (in the sense of the nuclear mabmetic resonance spectrum). The further observations that diborane was always present in the vcssels containing the “ abnormal ” aluminium borohydridc, and that artificial introduction of diboraxic suppressed the appearance of the “ abnormal ” form, led ultimately to the conclusion that AlB3H12 undergoes a reversible chemical changc into diborane and a hitherto unrecognized compound of aluminium, boron and hydrogen, whose formula would appear to be most probably A12B4H18.The programme of high rcsolution nuclear magnctic resonance studies whosc results are presented hcre is designed to elucidate the structural features of this ncw substance as well as those of AlB3H12, and to correlatc thcse withthe thermodynamic aspects of thc equilibrium involving the two substanccs. EXPERIMENTAL Several different samples of AIB3H12, synthesized by different methods, were used. There is no cause to believe that possible trace impurities are significantly responsible for the results obtained.Samples which had been stored in the liquid state at - 80” C or in the vapour state at room temperature yielded indistinguishable results. The conventional vacuum-line techniques were used for all transfers, and for preparing mixtures of B2H6 and AIB3H12. Partially deuterium-substituted samples, up to com- positions approximating AIB3H3D9, wcre prepared by heating mixtures of AIB3H12, .B2H6 and D2 in sealed glass vesscls provided with slender side arms for condensation of liquid. The deuterium gas was synthesized in the vacuum system by reaction of D2O and metallic sodium, and was dried by slow passage through liquid nitrogen cooled traps. The bulbs were heated at -I- 80” C for periods from 10 to 30 h. In the earlier studies the AIB3H12 samples were contained in sealed Pyrex vessels some 4 mm in diameter, with relatively small vapour space.The samples used for the observa- tion of the reversible chemical change were contained in vessels similar to those previously used for B2H6. In these the sample was largely in the vapour state at room temperature. Cooling of the slender tip caused condensation of the liquid for nuclear magnetic resonance study. Studies of nuclear magnetic resonance spectra were performed with the high resolution spectrometers of the research and development laboratories of Varian Associates, Palo Alto, California. Proton resonance was studied at 30.0013 Mc/s, and B11 resonance at 12.3 Mc/s. A vcry important aspect of the proton studies was the use of the double resonance technique, described in the paper by Dr.J. Shoolery. For B11 saturation a crystal controlled oscillator at 9.6257 Mc/s was used, final slight retuning being used for precise matching of resonance. A variable frequency oscillator with frequency counter was used for A127 saturation. The approximate frequency at which the best match was obtained was 7.8177 Mc/s. In most cases the shape of the containers made impractical the use of ‘‘ spinning ” of the sample to improve resolution, and hence great care was taken to search for the region of magnetic field giving maximum resolution. It is felt that no structural details of the spectra have been overlooked because of possible inferior adjustment of resolution. Calibrations were performed by the usual technique of modulation with an audio-frequency oscillator.RESULTS In fig. 1 are presented representative spectra selected from a very large number obtained for various samples of AlB3H12. For the proton spectra the calibration is with reference to protons in liquid water selected as the arbitrary zero on the magnetic scale. The numbers along the top of the record indicate displacements in parts per million, one part per million corresponding to 30c/s. For the boron spectra the zero was arbitrarily selected as the centre of the indicated band, the displacements again being in parts pcr million. At this frequency one part per million corresponds to 12-3 c/s. It is to be emphasized that the spectral details displayed in fig. 1 are practically inde- pendent of the temperature of the samples. Only with the proton resonance scanned whilc saturating the B11 resonance was there a noticeable change, in the sense that the ‘’ flat top ” appearance of the spectrum tended to round off as the temperature was raised. That presented corresponds to about - 60” C.Even for this the breadth at half maximum did not vary greatly.R . A. OGG, JR. AND J . D. RAY 241 For spectrum 1A the breadth at half maximum is about 300 CIS, while in 1B this has decreased to some 220 c/s. The separation of neighbouring peaks in spectrum 1C is some 87 c/s, a figurc in excellent agreement with that for the multiplet separation in fig. 1D. In fig. 2 arc presented spectra exactly comparable with those in fig. 1, but obtained with the most highly deuterium-substituted samplc obtained.Several samples of inter- mediate deuterium substitution, whose spectra are not presented here, showed intermediate spectra. Temperature effects are practically negligible, as in fig. 1. The relatively low fraction of protons leads to a rather unsatisfactory signal to noise ratio in A and 13, but as far as can be judgcd, the characteristic features are indistinguishable from those in the corresponding spectra in fig. 1. The definitely greater line breadth in fig. 2C, as compared to fig. lC, is a matter whose significance will be dealt with in the discussion. The highly complex appearance of fig. 2D is expected, since the sample is a mixture of several species, in most of which the 1311 resonance is split both by protons and deuterons. The small spacing compared to that in fig.1D is in satisfactory agreement with the ratio of gyromagnetic ratios for dcuterons and protons. In fig. 3 and 4 are presented spectra obtained for the ‘‘ abnormal ” substance obtained by heating the vapour of AlB3H12 (at a few hundred millimetres partial prcssure) at + 80” C for periods of the order of 10 h, and then condensing as rapidly as possible by cooling the vessel in liquid nitrogen. The details displayed here are observed to alter slowly as the sample is allowed to stand at room temperature, so that in most cases after a day or so the spectra are identical with those prcsented in fig. 1. In cases in which the sample was condensed with dry-ice in the side tip, and then sealed off, this ‘‘ reversion ” did not take place. It was also observed that a dark film (elementary boron) formed at the seal, and that the remaining relatively large bulb contained diborane, whose normal boiling point is considerably below dry-ice tempcrature.Samples to which diborane had been initially added before the above heating and condensation treatment showed a spectrum intermediate between those in fig. 1 and in fig. 3 and 4, i s . repression of the “ abnormal ” form. The samples which displayed most strongly the “ abnormal ” spectra were also obviously different in physical properties. Crude vapour pressure measurements were made by observing the fraction of liquid volatilized from the small side arm when main- tained at a known temperature, and without question the “abnormal” substance is less volatile than AlB3H12. The various observations are best explained by proposing the reversible reaction the equilibrium being slowly attained, with a value of the mass law constant very strongly temperature dependent, so that at 20” C the proportion of A12B4H18 is negligible, while at 80” C it is the dominant species.The above reaction would not cause any pressure change, and this is in accord with such observations as have been madc at clevated temperatures.5 For the purpose of subsequent discussion this intcrpretation is adopted, and the spectra in fig. 3 and 4 are hence presented as those of “A12B4H18”. (It is planned to subject this system to other physical chemical tcsts designed to test this hypothesis.) The most striking featurc of the simple proton resonance spectrum of A12B4H18 is its extraordinary sensitivity to temperature, as shown in fig.3A, 3B and 3C. At sufficiently low temperature its spectrum approaches that of AlB3H12. At these low temperatures simultaneous saturation of the A127 resonance is seen to have an effect very similar to that on the spectrum of AlB3H12. The absolute positions of the spectra on the magnetic scale are very similar to those for AlB3H12, any apparent displacement being probably within the calibration uncertainty. However, the multiplet separation in fig. 3D would appear to be 83 c/s (as compared with 87 in fig. 1C) and this difference is probably real. The proton rcsonance of A12B4H18 under B11 saturation is strikingly different from that of AlB3H12, as comparison of fig. 4A and 4B with fig. 1B demonstrates.This reson- ance is also temperature sensitive in the same scnse as the simple proton resonance, thc breadth at half maximum increasing from some 50c/s at room temperature to about 120 cycles at - 60” C. The B11 resonance of A12B4Hl8, fig. 4C, is strikingly similar to that of AIB3H12. fig. lD, except for a small decrease of multiplct separation (agreeing with the 83 c/s from fig. 3D). Unlike the proton resonance spectra of A12B4H18, the B11 resonance spectrum shows no marked alteration with temperature. Numerous experiments were performed with partially deuterium-substituted samples of AlzB4His. In fig. 4D is presented a representative spectrum, that of B11 in a specimen 2AlB3H12 A12B4H18 + B2H6,242 ALUMINIUM BOROHYDRIDE approximating A I ~ B ~ H I ~ D ~ .It appears to be the superposition of a spectrum like fig. 4C, and of a 1,3,3, 1 quartet, each meniber of which is a small triplet, due to deuteron splitting. The small spacings are similar to those in fig. 1D. DISCUSSION Interpretation of the nuclear magnetic resonance spectra of AlB3H12 leads unambiguously to the following conclusions regarding molecular structures. (i) All protons in a given molecule have identical electronic environment, i.e. are " chemically equivalent ". This is most simply shown by the spectrum in fig. lC, where the saturation of the A127 resonance has removed the spin-spin perturbing effect of this species. The resultant spectrum is virtually identical with that of simple borohydride,l the details resulting solely from spin-spin coupling with boron nuclei.Even the weak satellites due to the BlO species are partly resolved. (ii) All protons are spin-spin coupled to both aluminium and boron nuclei. This is shown by the spectrum in fig, lC, discussed above, and by that in fig. lB, where saturation of the B11 resonance has removed the spin-spin perturbing effect of this spccies. The spin-spin intcraction with A127 nuclei should produce a six- fold multiplct of equal intensities. The " flat-topped" spectrum in fig. 1B is interpretcd as this multiplet, the individual lines being sufficiently broadened by electric field gradient interactions with the large electric quadrupole moment of the A127 nucleus to give an apparent continuum. The separation of adjacent lines would appcar to be about 44 c/s.The appearance of thc simple proton spectrum in fig. 1A is rationally explained as the result of broadening each line in fig. 1C by the amount indicated in fig. 1B. (iii) All boron nuclei are spin-spin coupled, i.e. " chemically bonded " to four equivalent protons. This is demonstrated by the extraordinarily well-defined 1,4, 6,4, 1 quintet for B11 resonance displayed in fig. lD, again remarkably similar to that of free borohydride ion. Following the argumcnt given before,l the extreme sharpness of the resonanccs strongly suggests the tetrahedral configuration of the four protons. The reconciliation of the above conclusions with any strictly static model appears impossible. The model discussed in the introduction, in which half of the protons are bridge bonded and half are terminal, would appear to require some dynamic feature leading to exchange and hence time-average equivalence of the two typcs of protons.A conceivable mechanism of such exchange would involve bodily transfer of borohydride ion from one molecule to another, achieved eithcr by spontancous ionization, or by a displacement reaction catalyzed by adventitious impurity. Demonstration of such effects in quenching of multiplet structure has already been given.6 7 To explain the spectra presently observed, such chemical transfers would ncccssarily fall within a narrow frequency domain centred about the abovc given A127 multiplct separation of 44 c/s. An exchange much slower than this would lead to the observation of two types of protons, as discussed in the introduction.An exchange of much highcr frequency would lead to complctc quenching of the A127 spin-spin coupling, leaving only that due to boron spin-spin interaction. The invariance of the spectra over a very wide temperature range, and their independence of the source of the AlB3H12, would appear to exclude any such chemical transfer as a reasonable explanation. A mcchanisrn requiring internal rotation of borohydride groups, resulting in the breaking and reformation of the bridge bonds to the aluminium nucleus, would formally lead to interchange of bridge and terminal protons. Such an internal process is subject to the samc frequency restrictions discussed above. With any rcasonable assignment of the temperature independent factor, a first- order rate constant of some 44 sec-1 at, say, 300" K is associated with an activation energy of the order of 14 kcal/mole.The obscrved invariance over a 100-deg.R. A . OGG, JR. AND J . D . RAY 243 temperature range seems definitely to exclude such a process. It would appear that the potential barrier hindering internal rotation of the borohydride groups must be considerably higher than 14 kcal/mole. The only tenable remaining hypothesis would appear to call for a non-classical penetration of the proton system through the hindering potential barriers, i.e. the quantum-mechanical “tunnel effect”. The important feature of such a process is that it would leave the bonding electron system essentially invariant, i.e. it would not break the spin-spin coupling with the aluminium nucleus.Again a frequency criterion is involved, the essential quantity for comparison being the magnitude of the “chemical shift” between the bridge and terminal protons. In diborane 1 this chemical shift at 30 Mc/s amounts to some 130 c/s, and a rather smaller value might be expected for ALB3H12. Were the frequency of the “ tunnelling exchange ” small in comparison with such a figure, the observation of spectral detail resulting from distinguishable bridge and terminal protons would be expected. If the “ tunnelling ” frequency is sufficiently high, spectral identity of the protons results, with an electronic environment equivalent to the average of the values characteristic of bridge and terminal positions. The practical spectral identity of the protons is independent of the exact value of the “ tunnelling” frequency, provided only that the latter exceed a figure large in comparison with the chemical shift of bridge and terminal protons.A value in excess of some 103 c/s would certainly suffice to explain the present observations. Such a frequency restriction is in sharp contrast to the narrow range requirement for any process leading to rupture of the bridge bonds. The frequency of the proposed “ tunnelling ” process would depend in an inverse exponential fashion upon the total mass of the system of hydrogen nuclei undergoing concerted motion. It is to be expected that with a sufficiently extensive replacement of protons by deuterons (a minimum of two per borohydride group is indicated), the frequency could be sharply reduced. Such considerations dictated the experiments leading to the spectra displayed in fig.2. While in overall aspects these are very similar to those of ordinary AlB3H12, there appears to be a striking difference in ’the breadth of the proton resonance lines scanned while saturating the A127 resonance. Separate experiments (whose spectra are not presented here) suggest that in deuterium-substituted borohydridc ion any spin-spin interaction of protons and deuterons is too small to account for the observed broadening in deuterium-substituted AlB3H12. Thc broadening discussed above is in accord with the proposed “ tunnelling ” mechanism. Were the frequency of such a process to be negligibly small, the simple proton resonance spectrum corresponding to the bridge model would be the superposition of four narrow lines on a broad band, with some relative “ chemical shift ” of the respective centres of gravity.Saturation of A127 reson- ance should reduce the band to four narrow lines, not incident with the original four. If, now, conceptually the exchange frequency of bridge and terminal protons be monotonically increased, the eight lines must correspondingly broaden and merge pair-wise into a pattern of four lines which at the optimum exchange frequency havc a breadth of the magnitude of the chemical shift. Further fre- quency increase leads to progressive narrowing of each of these four lines. (Comparablc effects have been demonstrated7 for exchange due to a chemical mechanism.) It is suggested that in the highly deuterium-substituted AlB3H12 the exchange frequency is still higher than that of the chemical shift, but suf- ficiently reduced in comparison with ordinary AlB3H12 so that some residual broadcning is still to be observed.That even for the deuterium-substituted species the exchange fi-equcncy is so great as to lead to near environmental identity of the protons is in rational accord with the failure of the othcr spectra in fig. 2 to differ significantly from those in fig. 1. In short, the spectrum with saturation of the A127 resonance provides the most sensitive test of the expectcd effect.244 ALUMINIUM BOROHYDRIDE Rough quantum-mechanical calculations indicate that the appropriate " tunnelling " frequencies required to explain the above effects are compatible with a potential barrier structure of sinusoidal character with spacings determined by the known geometry of borohydride ion, and barrier heights in considerablc excess of 14 kcal/molc.The above evidence is suggested as strongly indicative of the essential correctness of the bridge model discussed in the introduction. From the standpoint of thermodynamic properties, at ordinary temperatures the possible internal rotation of borohydride groups is to be regarded as frozen out. The proposed " tunnel " effect does not of course result in any net angular momentum of the borohydride groups. With the magnitude of tunnclling frequencies indicated the splitting of torsional levels makes no significant differencc in the standard entropy. Before considering the structure of A12B4H18, it appears appropriate to discuss the thermodynamic aspects of the proposed equilibrium The nuclear magnetic resonance spectra may serve an analytical function, indicating the composition of a mixture, without regard to the detailed interpretation in terms of molecular structure.The spectrum in fig. 3A is interpreted as that of a mixture of AlB3H12 (responsible for the underlying continuum) and of A12B4H18. which in pure state would give four lincs going nearly down to the base. A graphical resolution of the observed superposition spectrum, and the comparison of the respective areas, allows in principle the estimation of the ratio of the two substances in the mixture. For the present it suffices to point out that clearly A12B4H18 is the more abundant, the ratio being at least three to one.Since this rcsult was obtained for several samples heated for different fairly long periods at 80" C, this rough ratio is taken to represent the composition of the equilibrium mixture, with a mass law constant of at least 10. Samples of AlB3H12 which have stood for weeks at 20" C show proton spectra with no trace of the four-fold structure, indicating a negligible contamination with &B4H18. However, the possibility of a trace of excess B2H6 (formed in the synthesis) must always be borne in mind. Weighing these facts, it would appear that the mass law constant at 20" C is almost certainly less than 10-1. These crude estimates make it highly probable that the equilibrium constant for the reaction in question increases at least a hundred-fold between 20" C and 80" C.This would indicate the reaction (left to right) to be endothermic by at least 16 kcal/mole &&&f18, with a standard entropy increase of at least 50 cal/deg. mole A12B4H18 at 80" C. It is to be emphasized that this rough estimate would appear to constitute a conservative lower limit. The accurate determination of the equilibrium constants at various temperatures may well yield a value of the standard entropy increase substantially larger than this figure. Such a magnitude of standard entropy increase for a gas-phase reaction in which the number of molecules remains constant appears to be without precedent. It should be noted that the change in translational entropy is relatively insignificant, and that such an effect must result from changes in rotational and vibrational entropy.The crudity of the data does not warrant a detailed discussion, but it would appear that only a marked increase of entropy of internal rotation could account for a figure of such a magnitude. The known thermodynamic properties of B2H6 force the conclusion that it is the species Al2B4H18 which is characterized by large entropy of internal rotation, and that AlB3H12 is normal. The above discussion is seen to lead independently to the conclusion that at moderate temperatures any internal rotation of borohydride groups in AlB3H12 molecules is frozen out. This is in gratifying agreement with the structural conclusions arrived at from the nuclear magnetic resonance studies. Conversely, the con- clusion that A12B4H18 is characterized by relatively free rotation of borohydride 2AlB3H12 + M2B4H18 3- B2H6.R.A . OGG, JR. AND J. D . RAY 245 groups should find confirmation from its nuclear magnetic resonance spectra. It will appear below that this is in fact the case. The structure suggested for A12B4H1g involves two hydrogen bridges between A1 centres, which are each bridge-bonded to two borohydride groups. The proton resonance of the AP-H-Al27 bridge structures in principle should exhibit 36 lines, expected to appear as a broad and relatively weak continuum. Super- imposed upon this should appear the borohydride structure. As was discussed previously, true rotation of the borohydride groups, leading to continuous rupture and reformation of their bridge bonds to aluminium, would at sufficiently high frequency result in practically complete quenching of any multiplet structure caused by spin-spin coupling with the A127 nucleus.This is seen to be the case in the spectrum displayed in fig. 3A. Saturating the B11 resonance under thcse conditions is seen to result in a single relatively narrow line, displayed in fig. 4A. Particularly interesting is the effect of lowering the temperature as seen in fig. 3B and 3C. A progressive broadening and ultimate nearly complete merging of the lines is seen to occur. This corresponds to a sharp decrease in the frequency of internal rotation, so that at the lowest temperature the spin-spin coupling with the A127 is scarcely disturbed. That it is this coupling which is responsible for the broad structure is shown clearly by the spectrum in fig.3D, where saturation of the A127 resonance has restored the four-fold structure. The temperature effects on the proton spectrum scanned while saturating B11 resonance (fig. 4A and 4B) arc in agreement. The broadening at low temperatures corresponds in sense to the broadening of the respective lines in fig. 3A, 3B and 3C. That the borohydride group protons in A12B4H18 are identical in environment as in AlB3H12, and are arranged tetrahedrally in groups of four around respective boron nuclei, is suggested by all of the nuclear magnetic resonance spectra. The clearest demonstration is offered by the B11 spectrum in fig. 4C, seen to be a well- defined 1,4, 6,4, 1 quintet, similar to that given by AlB3H12. Additional evidence is offered by the B11 resonance of partly deuterium-substituted A12B4H18, fig.4D. The maintenance of thesc details over a very wide temperature range, in contrast to the marked changes suffered by the proton spcctra, indicates that the identity of protons is achieved by a mechanism separate from the internal rotation of borohydride groups. This would logically appear to be the same type of " tunnelling " as that proposed for AIB3H12. As will appear below, an experi- mental test should be even more difficult with A12B4H1g. The models proposed for AlB3H12 and A12B4H18 are seen to have great structural similarities. In each case the rotations of borohydride groups leading to interchange of terminal and bridge protons are to be regarded as true chemical reactions, actually breaking the aluminium-hydrogen bridge bonds.The necessary energy requiremcnt corresponds to a potential barrier hindering rotation. The essential difference of the two species would appear to lie in the height of this barrier. In AIB3H12 this would appear to be so high that at ordinary temperatures practically all molecules occupy the lowest torsional level. It is from this level that the proposed tunnel exchange of protons occurs. In A12B4H1g the barrier would appear to be much lower, perhaps of the order of 10 kcal/mole, as the striking temperature coefficient of the proton spectra sho us. The relatively large internal entropy, above termed " rotational ", is more accurately to be regarded as associated with the relatively low frequency torsional oscillations. The frequency of tunnel effect exchange of protons through the relatively low barrier is expected to be extremely high. It is felt that the original objective of the investigation has been satisfactorily achieved. A strong case can be made for the bridge model for ALB3H12, with associated tunnel exchange of protons. It is of interest that the alternative behaviour of internal chemical reaction, rejected on logical grounds for AlB3H12, is apparently actually displaycd by the new compound discovered in the course of the work.246 GENERAL DISCUSSION The authors wish to acknowledge the generosity of Prof. A. B. Burg, who supplied the samples of aluminium borohydride and of diborane. The invaluable experimental assistance and stimulating suggestions of Dr. J. N. Shoolery of Varian Associates are gratefully acknowledged. Ogg, J. Chem. Physics, 1954,22, 1933. Schlesinger, Sanderson and Burg, J. Amer. Chem. Soc., 1940, 62, 3421. J. Amer. Chem. SOC., 1946, 68, 312. price, J. Chem. Physics, 1949, 17, 1044. egg, J. Chem. Physics, 1954,22, 560. 3Beach and Bauer, J. Amer. C/lem. SOC., 1940, 62, 3440. Silbiger and Bauer 5 Brokaw and Pease, J. Amer. Chem. SOC., 1952,74, 1590. 7 Ogg, Farady SOC. Discussions, 1954, 215.