GENERAL DISCUSSION Prof. A. J. Kresge (University of Toronto) said In our BEBO calculations we varied AV both by changing p and holding V constant and by changing V while holding p constant. We find however that more important than changes in either of these parameters is the inclusion or neglect of end-atom repulsion. With this effect included we obtain sigmoid dependences of a upon AV which are fairly linear over their mid-portions (a= 0.2 to 0.8) and the (calculated) data for these regions are easily fitted to the simple Marcus formulation. The intrinsic barriers which these fits produce however are invariably less than calculated values of AE at AV = 0 often by factors as much as two and the work terms obtained are therefore corres- pondingly too large.I would like to add that although Marcus keeps both d1 and A constant in his model this does not necessarily correspond to empirical practice. For example along a series of proton transfers from carbon acids such as carbonyl or nitro compounds to a single base or even to a series of bases changes in AGO are commonly obtained by making structural variations in the substrate usually by changing the R. P. BELL oo 2 n \ z -Y El I c -30 -20 -10 0 10 20 30 AGO FIG.1.-Correlation of kinetic isotope effects for carbonyl compound ionization according to eqn (1) ; data from ref (1). extent of charge delocalization in the carbanionic product. Charge delocalization however is one of the more important factors governing the rate of proton transfer to or from carbon and these structural changes are therefore likely to alter d for the identity reaction between the carbon acid and its conjugate base.89 GENERAL DISCUSSION There is some experimental evidence to support this idea from isotope effect correlations using simple Marcus theory. The theory predicts that kinetic hydrogen isotope effects will be largest at AGO = 0 and will decrease symmetrically on either I BORDWELL AND BOYLE 0 0 -1 0 0 +I0 +20 AGO FIG.2.-Correlation of kinetic isotope effects for nitroalkane ionization according to eqn (1) ; data from ref. (2). side of this maximum at a rate controlled by AG the intrinsic barrier for the system being correlated small values of AGZ will give rapid decrease of kH/kD with AGO whereas large values of AG will lead to more gradual changes (eqn (1)).ln(k,/kD) = In(k,/kD),,,[ 1-(AG0/4AGZ)*]. (1) Experimental data for a group of carbonyl compound ionization reactions (fig. 1)' and a series of nitroalkane ionizations (fig. 2)2 do fitthis relationship reasonably well. But there are systematic deviations from the best (least squares) fits to all of the data in each case the points to the left of AGO = 0 cluster above the correlation line and those to the right of AGO = 0 concentrate below it. This suggests that AGZ changes along these reaction series and the direction of the change decreasing AG with increasing AGO is just that expected on the basis of more delocalization in the carbanions derived from the stronger acids required to make AGO < 0 than those produced by the weaker acids for whose reactions AGO > 0.Dr. W. J. Albery (Oxford University) said This morning Bell using Morse curves showed that in certain cases one could obtain maxima and minima in the variation of a with ApK (see fig. 3). Is there a simple explanation of the shape of these curves? R. P. Bell The Proton in Chemistry (Chapman and Hall London 1973) p. 265. F. G. Bordwell and W. J. Boyle Jr. J. Amer. Chern. SOC.,1975 97 3447. GENERAL DISCUSSION Prof. L. Melander (Giiteborgs Universitet) said It is pleasing to find that Marcus represents the longitudinal motions in the C1. . . H . . . Br system by means of the skewed potential-energy surface which is a very useful device but seldom used.In the present energetically very unsymmetric reaction the reaction coordinate in the transition state is almost parallel to one of the interatomic-distance axes. This means that the motion along that coordinate comes close to a mere approach between C1 and HBr i.e. a relative motion between two heavy species. In such a case the effective mass must be considerably heavier than the proton and tunnelling corres- pondingly negligible. (Another reason for tunnelling being negligible is the low potential-energy level of the saddle point relative to the reactants of course). Appreciable tunnelling could arise only when the reaction coordinate in the transition state has such a direction that the heavy neighbours of the hydrogen stay at almost constant distance from one another i.e.heavy-atom motion is negligible. I would be glad to know Marcus’ answers to the following questions Jn general could energetically very unsymmetric reactions be expected to have potential-energy surfaces as unsymmetric as the one presented ? In discussions of the variation of the “ classical ” isotope effect with symmetry it has been argued frequently that the (valence-bond force field) force constants between hydrogen and its next neighbours could hardly be expected to differ from each other by more than a factor of about ten. From fig. 1 it seems rather likely that the C1-H force constant could be even negative. Are such force constants likely to be applicable to strongly unsymmetric systems in general? Prof.R. A. Marcus (University of Illinois) said In response to Melander’s questions the answer is probably “ yes ” to the first one the exothermicity itself implies a large asymmetry about the bisector of the acute angle in fig. 1 of my paper. The experimental observation of high vibrational excitation of the reaction product HI in the gas phase reaction implies a large potential energy drop along the angular coordinate (the protonic coordinate) in the transfer region. Because of this large drop the answer to the second question should also be “yes” for a sufficiently exothermic reaction. Prof. R. P. Bell (University of Stirling) said The form of the Marcus relation frequently employed in treating proton transfers is based on a picture of intersecting parabolas or other types of potential energy curves.While this is reasonable for electron transfers hydrogen atom transfers (3-electron systems) certainly require a different approach such as BEBO. It is not clear however how far this approach can be transferred to transfers of protons or hydride ions which are 4-electron and 2-electron systems respectively and involve a net transfer of charge possibly the intersecting curve model is more appropriate here ? The form of the curves relating Bronsted exponents to AG” shown in fig. 3 of my introductory lecture depends on properties of the Morse energy curves on which they are based. These have a point of inflection at an energy of tD,where D is the dissociation energy and the sign of dp/dAG” at AGO = 0 depends upon whether the two curves intersect above or below this point of inflection.These results are not to be taken too seriously but serve to emphasize the fact that the theoreticaldependence of /3 upon AGO (and hence also values of AG; w and w deduced from experiment) do depend critically upon the model adopted for the energy profile. The consequences K. G. Anlauf P. E. Charters D. S. Horne R. G. McDonald D. H. Maylotte J. C. Polanyi W. J. Skrlac D. C. Tardy and K. B. Woodall J. Chem. Phys. 1970 53 4091. GENERAL DISCUSSION in this respect of an electrostatic model of proton transfer are at present being investigated. For a reaction of the type X+HY + XH+Y (charges not specified) the accepted procedure is to plot the energy surface with skewed co-ordinates as in fig.1 of Marcus’ paper and when X and Y are heavy entities the angle between the axes becomes very small. This procedure is undoubtedly correct when the species concerned are moving freely in one dimension but is questionable if the proton is being transferred within a reaction complex in which the reactants are already in a suitable position orientation and state of solvation. Under these circumstances the situation may resemble the transfer of a proton or hydrogen atom between two fixed centres in a solid lattice for which the energy surface becomes a function of a single linear co-ordinate. Melander has suggested that the validity of this last approach depends on how “stiff” the reaction complex is with respect to motions of the reactants and solvent molecules it would be valuable to have Marcus’ views on this question.Prof. M. M. Kreevoy (University of Minnesota) said The values of W‘ which have been obtained certainly have substantial uncertainties. It’s also easy to conclude that W‘ may be a weak function of pKHA. Nevertheless the general conclusion that W’ is much larger than the free energy required to assemble an encounter pair seems quite secure. That conclusion does not depend on the detailed form of the function used to fit the Bronsted plot. In the work I described today and in several earlier studies acids of increasing strength have been studied until a limiting rate has been reached or closely approached. The existence of such limiting rates far short of the diffusion rate is sufficient to establish that large values of Wrare required.Dr. R. A. More O’Ferrall (University College Dublin) said It is perhaps worth noting that Marcus’s expression must underestimate the “intrinsic ” barrier to reaction AG*’ when AGi’ = 0. Plotting AG* against AGif for the proton transfer step of an ideal reaction family / I FIG. 1.-Full line (ad) idealised plot of AG*’ versus AG;’ for proton transfer. Dashed lines best quadratic approximation to ad at AG;‘ =O (bb’); Marcus’s quadratic expression (cc’). M. M. Kreevoy and S. Oh J. Amer. Chem. Soc. 1973,95,4805. GENERAL DISCUSSION for which AG*' varies continuously with AG;' and -$ 0 and AG;' as AG;' + -and +infinity respectively gives the line labelled aa' in the figure which might be repre- sented by a hyperbola (E.S. Lewis unpublished) or by Marcus's BEBO expression. Near AG;' = 0 the best quadratic approximation to the line is given by the parabola bb'. However bb' differs from the quadratic expression used by Marcus AG"' = (1/4)(1 +AG;'/A)' which carries the implication that AG*' = 0 when dAG*'/dAGi' = 0 with the result that the parabola is shifted along the AG*' axis to cc'. Marcus's expression does not lead to an inferior correlation of results because what is plotted in practice is not AG*' but AG* = AG*'+wr so that low values of AG*' are compensated by a high value of w'. Nevertheless it is clear that A/4 must underestimate the barrier to the thermoneutral reaction and indeed if aa' is a hyperbola A/4 is low by a factor of 2 while if aa' is given by the BEBO expression it is low by 1.4.On the other hand this does not alter the qualitative conclusion emphasised by Kreevoy that a high Bronsted curvature associated with large activation energies implies a major contribution from resolvation to the reaction barrier. Prof. R. A. Marcus (University of Illinois) (communicated) In response to O'Ferrall and Kresge I would note first that the results of Kreevoy in finding large W"S from studies of the limiting rate at very negative AGO'S (fig. 1 of his paper) substantiate his earlier finding of large w' using a quadratic plot and measurements of AG" against AGO over a more restricted AGO range. Incidentally if one takes the limiting kHA/p there to be -3 M-l s-l and writes it as Zexp (-w'/RT) taking 2 the collision frequency in solution to be ca.10" M-I s-l then w' is about 62 kJ mol-l. This is only about 10 kJ 11101-' less than that deduced from the quadratic fitting to the data in fig. 1. As O'Ferrall and Kresge note evaluation of AG* from a restricted AGO range by fitting could lead to a model-dependent w'. However in a region of Bronsted slopes of 0.2 to 0.8 the In cosh BEBO expression (eqn (2.8) of my paper) is reasonably well approximated by the quadratic one (eqn (2.6)). For example when A/4 for eqn (2.8) is x kJ mol-' use of eqn (2.6) for AG* yields a new A/4 of 0.8~ kJ mol-l and a w' that is correspondingly 2x kJ mol-' larger than before. This difference in the wr's of these models is relatively small and helps explain the agreement of w"s in Kreevoy's results.Typically when obtaining w' from data over a restricted AGO range one might consider using both eqn (2.6) and (2.8) to explore the sensitivity of the deduced w' to the model equation. The extent to which any arbitrary curvilinear (AG* AGO) plot however can be approximated by a quadratic one depends on the shape of the former Use of an arbitrarily large (d3AG*/dAGo3)at AGO = 0 causes the fitting to differ by an arbitrarily large amount. Thus it is best to use experimental or theoretically-derived plots when comparing them with a quadratic one since otherwise any comparison could become quite meaningless. The quadratic plot appears to agree better with the In cosh one than with the plot in the figure in O'Ferrall's query.Prof. R. A. Marcus (University of Illinois) said Calculations of potential energy barriers have now been performed using harmonic BEBO and/or Morse-like potential energy surfaces by Bell,' Koeppl and Kresge and by me.3 Yet our conclusions R. P. Bell this Symposium Spiers Memorial Lecture. 'G. W. Koeppl and A. J. Kresge J.C.S. Chem. Cornrn. 1973 371. R. A. Marcus paper at this Symposium. GENERAL DISCUSSION as indicated in Bell’s comment are apparently so very different The first two sets of authors find that the Bronsted slope can differ considerably from 0.5 at AGO = 0 while I find that when A is held constant the slope is close to 0.5 at AGO = 0. Even for a highly asymmetric reaction in the sense that the A’s of the two reactants are very different (6 l) the slope at AGO = 0 is still close to 0.5 namely 0.6 in the present paper.In response to Bell’s comment I believe therefore that the difference in these findings lies in my use of a constraint on A rather than in choice of the form of potential energy surfaces themselves. For example a BEBO surface was used throughout the calculations in my paper rather than intersecting parabolas. In a previous paper it was shown that if A varies in a series of reactants for which AGO is varied the Bronsted slope should indeed be very different from 0.5 at AGO = 0 even negative or greater than unity in some cases. Unless special precautions are taken ,JL will vary when potential energy surface parameters are varied to vary the net potential energy change for the reaction AU.In particular variation of AU in a BEBO model by varying onZy the dissociation energies D1causes A to change. Accordingly both D,and the bond order coefficients pi were altered simultaneously in the present paper to alter AU at fixed A. The question arises therefore as to whether one should perform the calculations in a constrained way or not. The answer depends on the problem one wishes to examine. For example in proton transfers between an acid (base) having a large A and a series of bases (acids) having a small A the net ;1 for the reaction which is the mean of the two l’s should not vary appreciably from one member of the series to the next. In such cases e.g. for certain carbon acids reacting with oxygen or nitrogen bases which do not suffer extensive electronic rearrangement the conditions are best simulated by surfaces in which AU is varied holding A fixed.The fact that the experimentally observed Bronsted slope is indeed close to 0.5 at AGO = 0 for such highly asymmetric systems also supports this conclusion. Dr. D. M. Goodall (York University) said Marcus alluded to the possibility of facilitating proton-t ransfer reactions in hydrogen bonded systems by supplying vibrational energy using short laser pulses. Greenhow and I have already reported such an e~periment,~ in which liquid water was photoionized using a Q-switched neodymium glass laser. The quantum yield for this process was evaluated from the transient conductivity increase due to the excess hydroxyl and hydrogen ions produced.With Knight we have recently extended these experiments to excitation at other wavelengths. Quantum yields at 328 K are 1 x lo-’ (0.69 pm) 5 x lo-’ (1.06 pm) < 2 x lo-* (1.41 pm). Only an upper limit can be given at 1.41 pm where the transient conductivity profile is indistinguishable from that for microwave- heated water.4 We have shown that photoionization is a one photon process and anticipate that a theoretical account of this data will provide valuable information concerning the role of vibrational excitation in the activation process for proton- transfer reactions in solution. Prof. R. A. Marcus (University of Illinois) (commimicated) It was good to hear of these experiments of Goodall and of the new results.Of particular interest too would be measurements of the quantum yield 4 at still shorter wavelengths the R. A. Marcus J. Amer. Chem. SOC.,1969 91 7224. R. A. Marcus paper at this Symposium. D. M. Goodall and R. C. Greenhow Chem. Phys. Letters 1971,9 583. G. Ertl and H. Gerischer Z. Elektrochem. 1961 65,629. GENERAL DISCUSSION pre-exponential factor of the thermal liquid phase proton transfer rate constant of 2H,O -+ H30++OH- lo7 s-l corresponds to a AS* of -28 cal mol-’ K-I. If the deactivation frequency of the vibrationally-excited sys tem described by Goodall is of the order of 10l3s-l a collision frequency the maximum 4 would be 107/1013 i.e. at sufficiently high energies if the range of 0.. . 0 distances and solvent orientations over which the proton can transfer is the same as in the lower energy thermal system.If instead the range of these coordinates permitting proton transfer is greater in the photochemical case then the maximum 4 would exceed even with a collision deactivation frequency of lOI3 s-l. The experiments do not directly distinguish between statistical and non-statistical regimes of the unimolecular process whereas the lower energy picosecond experiments proposed in my paper would probably be in the non-statistical regime. Both types of experiments thereby complement each other. Dr. W. J. Albery (Oxford University) said I would like to present a somewhat different analysis of Kreevoy’s results which reinforces the point he has made that his conclusions do not depend upon the detailed algebraic formulation of the Marcus theory.In comparing the effects of changing the catalyst (HA) while keeping the diazo compound (S) the same with changing the diazo component while keeping the catalyst the same it would be helpful if we knew the dissociation constant Kf of the protonated diazo compound Unfortunately the values of K$ cannot be measured because SH+ decomposes. However we can assume that KF will obey a Hammett Relation and since Q depends on the dissociation constants of Ar COOH we may expect that ps > 1 log KF = log K:+pp. Now we relate the Bronsted coefficient as for changing S and keeping HA constant to the Hammett p values by writing where AGFD is the standard Gibbs free energy charge for the whole reaction and AG* kLA,and pHAare as defined in Kreevoy’s paper.Ignoring terms connected with Wp and assuming J. constant we may also write and as is the usual Bronsted coefficient for keeping S constant and varying HA. If awr equals zero then as = as and the system would be “ well behaved ”. Comparison of eqn (1) and (2) shows that -pHAshould vary linearly with aB. This is the equivalent of Kreevoy’s eqn (19) which does indeed lead to the linear variation shown in his fig. 3. However -PHA moves in the opposite direction to aB as shown by the following values -PHA aB H C Clz COOH 1.43 0.3 3,4,5-C13C6HzOH 1.30 1 .O cf. D. M. Goodall and R. G. Greenhow Chem. Phys. Letters 1971 9 583. W. J. Albery A. N. Campbell-Crawford and J.S. Curran J.C.S.Perkin 11 1972 2206. GENERAL DISCUSSION This implies that aWyris greater than unity which in its turn from eqn (2) means that for small values of aB,as is greater than unity. In terms of Kreevoy's pw and pc we can show that Ps = -(Pw+Pc) = 1.3 awr = I+pC/ps = 1.15 and for HCC1,COOH as = 1.10. As expected the value of ps is greater than unity but perhaps is somewhat small when compared to values for similar ionisations e.g. p = 2.8 for the ionisation of Ar N+H3;'the neglect of Wp terms may have led to the apparent low value. FIG.1.-Free energy profile for aliphatic diazo compounds showing the change in Wr plotted along the S co-ordinate going into the paper and the proton transfer going across the paper.The various compounds are of the type RIRzCNz where Rl and R2 are 1 H,[COO-; 2 COO- COO-; 3 Me COOEt ; 4 Me COMe; 5 COOEt COO-. The nitro-compounds studied by Bordwell amongst others are a similar system to the diazo compounds and it is interesting that Kreevoy's work demonstrates that like the nitro compounds the diazo compounds also have a value of cis greater than unity. The fact that aWr is close to unity confirms the variation of free energy with solvation and with the degree of proton transfer deduced by us for other aliphatic diazo compounds from an analysis of the degree of proton transfer in the H30+ catalysed reaction. Fig. I shows the pattern for five different diazo compounds which span a range of over 6 orders of magnitude in their rate constants.Similar " Marcusian " behaviour is found for the proton transfer for each compound across the diagram while the position of each compound on the solvation co-ordinate running into the diagram corresponds to awr 2i 1. F. G. Bordwell and G. D. Cooper J. Amer. Chem. SOC.,1952,74 1058. F. G. Bordwell paper at this Symposium. W. J Albery A. N. Campbell-Crawford and J. S. Curran J.C.S.Perkin ZZ 1972 2206. W. J. Albery C. W. Conway and J. A. Hall J.C.S.Perkin IZ 1976 473. GENERAL DISCUSSION Prof. W. H. Saunders (University of Rochester) said Would Kreevoy please comment on whether the encounter complex and reaction complex should be regarded as real or virtual intermediates? Could one be real and the other virtual? Bell and Goodall suggested that kH/kDwould reach a maximum value at ApK (or AGO) = 0.According to Marcus theory the quantity that should be zero is AGO within the reaction complex rather than the overall AGO. But observed kH/kD maxima usually occur within one or two pM units of ApK = 0 insofar as one can locate the maximum withreasonable precision. This fact suggests Wp-Wr < 2-3kcal/ mol. Although Wp and Wr might meet this requirement by being large but nearly equal another possibility is that both are fairly small. Prof. M. M. Kreevoy (University of Minnesota) said 1.The partners in an encounter complex are in no way different from completely separated molecules except that they happen to be close to one another. The process of reseparating them is simply their diffusion apart.Since diffusion does have a free energy of activation (albeit small) the encounter complex is a real intermediate. The reaction complex has a structure partially determined by the requirements of the transition state. That is the reaction complex must have a structure such that the transition state can have a minimum energy in all its normal modes except the reaction coordinate (including those of the solvsnt). Depending on the specifics of the solvent and the structure this may make the reaction complex correspond to a region of potential energy hyperspace which is not a local minimum but one in which the character of the atomic motions leading toward the products changes. The minimum energy path calculated for the transfer of a proton from NH; to NH3 shows such a reaction complex with the relevant coordinate changing from f" to rNH.' Such a reaction complex would be a virtual rather than a real intermediate.There may be other cases where the reaction complex corresponds to a local minimum in potential energy hyperspace. Even in those cases however it will probably be a very shallow minimum. As soon as the reaction complex begins to revert to the encounter complex normal solvation forces-and probably structural forces as well-will begin to work to lower its energy so that the free energy of activation for that reversion will probably always be much less than that for diffusion. 2. I think the kH/kDmaxima do not occur as close to ApK = 0 as Saunders implies. The two shown by Kresge in this symposium seen to have maxima around -3 and -4 kcal mol-l respectively.Caldin has referred to other results leading to the same conclusion. Further W' and Wp have several factors in common and in many cases would not be expected to be too different. I believe these results are entirely consis tent with the postulated large values of W'. Prof. E. F. Caldin (University of Kent) said (a) Although in a series of proton- transfer reactions a maximum isotope effect kH/kDis often found when AGO or ApK is zero (cf. Professor Bell's introductory lecture above) this is not necessarily to be expected when the series includes different types of base if tunnelling corrections are large. Variations in AGO (or equilibrium constant K) are usually influenced by the enthalpy of reaction AH"; but variations of kH/kDare greatly affected by the shape of the energy-barrier which does not depend solely on AHo but also on the barrier width and appears to be sensitive to the type of base.It is therefore not surprising that in our results (fig. 2 p. 127) the largest value of kH/kD is not at log K = 0. Nor is there a smooth relation between kH/kDand K;unpublished work by Dr. Rogne has P. Merlet S. D. Peyerimhoff and R. J. Buenker J. Amer. Chem.Soc. 1972 94 8301. S 10-4 GENERAL DISCUSSION shown that n-butyldiethylamidine has nearly the same equilibrium constant as tetramethylguanidine in its reaction with 4-nitrophenylnitromethane but a markedly higher value of kH/kD. These two bases both contain the imine grouping N€i=C/ -\’ the larger effect for n-butyldiethylamidine appears to be due to a higher barrier.(b) The effects of tunnelling on the parameters of the Marcus theory have not yet been worked out. Formally the theory can be expressed in terms of a potential- energy diagram composed of intersecting parabolae. For reactions where the tunnelling correction is important the rate will be affected not only by the energy terms for proton transfer and solvent reorganisation but very markedly by the width (or strictly the curvature) of the energy-barrier for proton transfer. This barrier- width will undergo changes if the positions of the parabolae are shifted either vertically or laterally by changing substituents or the solvent. The effects on the tunnelling correction and hence on the rate have not yet been examined; it seems likely that they will be significant.Dr. W. J. Albery (Oxford University) said Would Margerum like to comment on the fact that he has calculated Marcus parameters from Bronsted plots (fig. 1 2 3 and 6 of his paper) which include points at each extreme for catalysis by H20 and H30+? These two catalysts often deviate from the Bronsted plots for other catalysts ; for instance see point 1 in fig. 1 of Kreevoy’s paper.l Thus it may be unwise to calculate Marcus parameters from Bronsted plots including these catalysts. On the other hand if these catalysts are excluded from Margerum’s plots then it would appear that the remaining points do not cover a sufficient pK range for the Marcus parameters to be calculated at all.Prof. D. W. Margerum (Purdue University)said :Insofar as fig. 1and 3are concerned there is no reason to omit the data points for H30+,H20and OH-from the Bronsted plots since the points fall on the plots. The same is true for acetylacetone reactions where the Marcus parameters have been calculated by Kreevoy (see table 5 of our paper). It is amusing that one now needs to defend the fact that these points fail to fall below the Bronsted plot that is the deviant behaviour is now considered normal. However I would suggest that it may be unwise to omit these points in calculations of the Marcus parameters. The failure of H30+ H20 and OH-to fall on linear Bronsted plots is expected from the Marcus theory and it is not clear that one should arbitrarily drop these points.In the reactions of acids of differing strength with a given base the necessity to desolvate the acid can be interpreted in terms of a large WR term as suggested by Kresge.2 Our data with the macrocyclic complexes fail to show an anomalous desolvation effect for H30+ and H20 compared to the other acids used. In these reactions the WR values are of moderate magnitude and are comparable to the A/4 values. Hence it is difficult to justify the suggestion that the H30+points with the metal-peptide reactions are anomalous due to specific solvation effects for the reactions in fig. 2 and 6. In these reactions we find that WRvalues are the same order of magnitude as those found for acetylacetone and the macrocycles but the 2 values are much smaller.Kreevoy’s reactions were carried out in 80 %DMSO where the W values are much larger and where specific solvation of H30+ may be more important. For the metal-peptide reactions it is unfortunate that a wider pK range A. I. Hassid M. M. Krevoy and T. Laing paper at this Symposium. A. J. Kresge Chem. SOC.Rev. 1973,2,475. GENERAL DISCUSSION of other acids cannot be used but the experimental system precludes this. Our preference is to explain the observed behaviour in terms of the relative magnitude of Wkand A/4 rather than in terms of an exceptional behaviour for H30+which is not found for other reactions with nitrogen bases. Prof. D. W. Margerum (Purdue Uniuersity) said In our paper we point out the possible compensating nature of the A and WRvalues where large values of WRtend to occur with small values of L and vice versa.This is not unlike the compensating nature of AH" and ASo values and of AH* and AS* values which Bell has discussed and has indicated a preference for correlations of AGO (equilibrium constants) and of AG* (rate constants). The question which arises is whether or not it is possible to predict the relative magnitudes of WR and A for various systems. Prof. M. M. Kreevoy (University of Minnesota) said Large values of W' can be obtained by an appropriate choice of solvent and/or by steric obstruction of the transfer site. Solvents which interact strongly with one of the reagents in such a way that the interaction must be substantially disrupted in order to form the transition state should lead to large values of IT".The DMSO component of the solvent mixture used for our work on diphenyldiazomethane was chosen partly because it was thought that it would increase the strength of the hydrogen bonds from the acids to the solvent. I believe this is one of the reasons for the unusually high value of W' observed. For reactions involving Hf anhydrous DMSO as solvent should give particularly large values of W' because in that solvent H+ probably has the structure \s-0 . . M . . . 0-s /+ * ,2 / \ and one of the strongly bound DMSO molecules has to be removed in order to allow a reaction to occur. In support of this idea the rate constant for protonation of tribenzylamine by H+ in anhydrous DMSO has been found to be only about 1 104 ~-s-1.3 R. P. Bell The Proton in Chemistry (Cornell University Press Ithaca New York 2nd Edn. 1973). p. 79. M. M. Kreevoy and J. M. Williams J. Arner. Chern. Soc. 1967 89 5499. Y. Wang unpublished results.