GENERAL DISCUSSIONH///’Ar+HR . C02Hs Ar I ’”\ H . . . 02C . R,’ 4 H . . . O///’ ‘\\ ,/+Ar c . R (1)\H . . . OH//’ArH + CH3 . C02H+Ar’ \\H . . . OzC.CH3 Haccording to the following line of reasoning.If the intermediate is supposed to be the same cation ArHi(1) for catalysis byhydrogen ions and by acetic acid then we can write the following four equilibria1 Streitwieser and van Sickle, J. Atner. Chem. SOC., 1962, 84, 254.2 Batts and Gold, J. Chem. Soc., 1964, 4284.9GENERAL DISCUSSION 95between reactants and intermediate, depending on which catalyst and which medium(H20 or D20) is used :1-1H,O++S+I+H,O K , (3)(4)HA + S $1 +A-' K2 ( 5 )DA + S +-Ii +K- K2 (6)_-D30++S+T'+ D20 Rlthe bar over a symbol designating solution in heavy water.On the basis of mechanism (1) and with some assumptions, it is possible to coni-bine rate measurements on various H-D-T exchange reactions to lead to thevalue 3.5 for the ratio K ~ K ~ I K I R ~ .In terms of equilibria (3)-(6)' this ratio ofconstants is the ratio of dissociation constants of acetic acid in ordinary and heavywater :[H30'][S] [%][S] [DA][D20]The directly determined ratio of these dissociation constants is 3.3. The satisfactoryagreement of this ratio with the value 3.5 supports the basis of the calculation andthat the same intermediate is formed in reactions (3) and (5).A similar treatment can be applied to the forward rate constants of the equilibria(3)-(6), or the quasi-equilibria between reactants and transition states :H30++S+TS, k ,D30+ +S+TS; ElDA+S+TS; k2___- - __HA+S+TS, k2 - - -The isotope effects on these reactionsandcan be combined to giveThe experimental value of this ratio is 1-85, very different from the value 3.3 of thefirst factor and indicative of the difference between transition states TS1 and TS2,the quotient [TS2][%;]/[TS;][TS,] having the value 0.55.The results of this investigation also allow us to give an answer to Mr.Bell'squestion about the possible consequences of neglecting secondary isotope effect96 GENERAL DISCUSSIONon the reversal step (- 1) of reaction (3) on the calculated isotope effects for the stepsof the exchange process. If a reasonable estimate of the secondary isotope effect,i.e., of the statistically corrected ratio of rate constants forandArHl +ArH + (H')ArHD'+ArD + (H')is taken, the primary isotope effect, expressed by the statistically corrected ratioof rate constants forArHz -+ArH + (H')andArHD' +ArH + (D'),has the value 8-05, whereas with neglect of the secondary isotope effect, the primaryisotope effect is 6.84.Prof.A. J. Kresge (Illinois Institute of Technology) said : In addition to supplyingvery useful data for aromatic hydrogen exchange in concentrated acids, Prof. Longhas convincingly argued for the existence of the conjugate acid of the aromatic asan intermediate in this reaction. However, the presence of this intermediate canbe deduced from two other pieces of kinetic data, viz., the reaction is subject togeneral acid catalysis and is not catalyzed by bases.For exchange to occur, the system must pass through a symmetrical configura-tion such as that in the species shown in eqn.(1) of Prof. Long's paper :L H 6 <*:This symmetrical species can be either an intermediate or a transition state; letus assume that it is the latter. Then, since the reaction shows general acid catalysis,this transition state must contain the conjugate base of the catalyzing acid attachedby a partial bond to one of the exchanging hydrogens. But if the species is to besymmetrical, both of the exchanging hydrogens have to be alike, and the otherhydrogen must be bonded to the same kind of base as well. This transition state,therefore, must contain HA and A- in addition to the aromatic, and, on this hypo-thesis, the reaction will be subject to base catalysis as well as to general acid catalysis.Since base catalysis is known to be absent, the symmetrical species cannot be a transi-tion state and must be an intermediate.The requirement that both of the exchanging hydrogens be attached to a basein a symmetrical transition state can be accommodated without base catalysis byinvoking a cyclic structure for the transition state such as ,that considered (and dis-missed) by Batts and Gold 1 :AIn exchange catalyzed by carboxylic acids, this structure would be strain-free becausedifferent oxygen atoms would be used for bonding to the different hydrogens.But1 Batts and Gold, J. Chem. SOC., 1964, 4248GENERAL DISCUSSION 97for catalysts such as the ammonium ion or the water molecule, strain-free con-figurations are not possible. This difference would be expected to result in ratesof reaction which are not correlated well by a single Bronsted relation.Since thereaction does obey a single Bronsted relation with considerable precision, such acyclic transition state can be eliminated from consideration.Prof. V. Gold (King's College, London) (communicated) : Prof. Kresge's commentthat a cyclic transition state for aromatic hydrogen exchange can be excludedsimply on the evidence of a single Bronsted relation presupposes that proton transfersdo not involve intermediate water molecules. If one were to write the reactionintermediate asH. .A. .H,' '.i/H-0 0-€3H LIstrain effects would no longer distinguish the different acids HA.This commenttherefore adds weight to the conclusion, suggested by our experiments on olefinhydration, that intermediate water molecules are not concerned in proton. transferto carbon.A carbon atom as a basic site is probably not significantly hydrogen-bonded tothe water-acid system, in distinction from lone-pair atoms (N, 0) commonly involvedin rapid proton transfer. A further example bearing on this question is the protontransfer between carbon and nitrogen which, in one direction, constitutes the measur-able rate-limiting step in the pyridine base-catalyzed halogenation of ketones and,in the reverse direction, represents a proton transfer from a pyridinium ion toolefinic (enolate) carbon.This reaction is characterized by considerable stericstrain in the transition state.1 One need not accept the unconventional and perhapsimprobable detailed suggestions that have been made about this transition state.1What is beyond dispute is the conclusion that this structure must be compact so that,in the reaction between 2,6-lutidine and pinacol, e.g., there is steric interferencebetween the methyl groups (Me) of the two moieties of the transition stateHMe-O /\ Me- -Me3C. CO . C . . . H . . . N / \HThis requirement of compactness would seem to preclude the interposition of a watermolecule between the participants.Prof. A. J. Kresge (Illinois Institute of Technology) said: Since our paper waswritten, we have extended our measurements to solutions of other strong mineralacids.I would like to present some preliminary data for nitric and phosphoricacids (fig. 4). Both monobasic acids give much the same rates of exchange whencomparison is made at the same concentration of HArH+, and these rates are1 Feather and Gold, J. Chem. Soc., 1965, 1752.98 GENERAL DISCUSSION- - . - - . - I - L . - - - - - - - I i- I 0 + Ib i o CWH+ICHA~FIG. 4.-Rates of loss of tritium from 1,3-dimethoxy-benzene-4-t compared to equilibrium pro-tonation of azulene.significantly less than those for exchange in the two dibasic acids, sulphuric andphosphoric.Mr. R. P. Bell (Oxford) said: The most certain way of establishing catalysisby bisulphate ions would be to study solutions of alkali bisulphates, with or withoutthe addition of sulphates, since in these solutions the hydrogen ion concentrationwould be low.I would like 10 enquire whether such experiments have been carriedout, or are feasible.Prof. A. J. Kresge (Illinois Institute of Technology) said: It is possible to measureexchange rates in bisulphate salt solutions, and we have done this, without addingsulphates, for another substrate. The difficulty here comes in sorting out thecontribution to the total rate made by the hydrogen ion. I am not sure that enoughRaman data for alkali bisulphate solutions exist to enable one to estimate the con-centrations of the various acidic species present. One could use the technique wedescribe in our paper, i.e., compare rates at the same Concentration of HArH+,and we have in mind to do this.The addition of sulphate salts might help, but here again the interpretation wouldnot be straightforward: bisulphate ion is a moderately strong acid and does notmake good buffers.In our work on 1,3,5-trirnethoxybenzene in fluoroacetic acidsolutions we had the trouble of changing pH with changing buffer concentration,and fluoroacetic acid is a slightly weaker acid than bisulphate ion.Dr. M. Spiro (Inzperial College) said: I would like to comment briefly on theresults shown on Prof. Kresge's fig. 4, which indicates that H3P04 and HzS04have much the same effect on the rate of the reaction. Since H2PO; has a muchsmaller dissociation constant than HSO,, the HzPOi ion would not be expectedto be as good a proton donor.However, we have recently deduced 1 from con-ductance and transference measurements that solutions of phosphoric acid containtriple ions of formula HzPO, . Hf . H2PO; which, as proton donors, would beintermediate in power between HzPO; and H3P04. It could well be that in thesolutions used by Prof. Kresge there was an appreciable concentration of such triple1 Selvaratnani and Spiro, Trans. Faraday Soc., 1965, 61, 360GENERAL DISCUSSION 99ions and that these contributed to the catalytic rate. The solutions may alsohave contained some dimeric molecules (H3PO& but it is only at high concentra-tions that these would be present in sufficient degree to affect the rate. I wouldlike to ask Prof. Kresge what concentrations of phosphoric acid were used in hiswork.Prof.A. J. Kresge (Illinois Institute of Technology) said: The experiments inphosphoric acid were done at stoichiometric acid concentrations of 1-5-53 M.We, also, find it surprising that the effects of phosphoric and sulphuric acid are sosimilar. Perhaps the reason for this is that phosphoric acid is weaker than sulphuricacid in its first dissociation as well as its second dissociation. In phosphoric acid,then, more of the excess rate could be due to catalysis by the completely undis-sociated acid, and this might make up for the smaller contribution from thedihydrogenphosphate ion.DP. M. Spiro (Imperial College) (partly communicated) : Some calculations havenow been carried out to estimate the relative contributions of the various species.The concentration chosen as an example was 4 M, which is inside the range usedby Prof.Kresge. The necessary stability constants were taken from ref. (1) andthe activity coefficients from Davies’ equation.1 This provides for the influence ofionic strength but makes no allowance for the undoubtedly big medium effects.The results are listed in the following table, in which &cid is the concentrationacid dissociation constant of the species named and in which the relative contri-butions to the rate u have been assumed equal to (cOncn.)(K~cid)+, as in the paperby Kresge et al. Statistical factors have not been taken into account.species HtPOZ H3m4 HzPOZ.H+ .H2POS (H3po4)2concn., M 0.06 2-6 0.5 0.14Kacid z x 10-7 0.013 0.004 ca.1V 3~ 10-5 0.3 0.03 ca. 0.1Thus undissociated H3P04 clearly has the greatest effect on the rate but tripleions do contribute significantly. Dimeric (H3P04)2, evidence for whose existenceis rather indirect, would appear to have an even larger influence at 4 M. Thesecalculations show how important it is to know which species are present beforzcatalytic efficiencies are assigned. Unfortunately our knowledge of the compositionof concentrated solutions of this sort is rather sketchy.Prof. F. A. Long (Cornell University) said: In our studies of azulene, we haveobserved that a number of the azulenes are unstable in strongly acidic solutions.Furthermore, the decomposition is greatest when there is an approximately equiv-alent amount of protonated and unprotonated species present.Azulene itselfdecomposes moderately and decomposition of some of the substituted azulenes issignificantly more. It is my impression that the rates of decomposition whichwe had observed were somewhat higher than those suggested by Kresge. Further-more, it is my recollection that the rate of decomposition was substantially greaterin sulphuric acid than in perchloric. This also, Dr. Kresge seems not to havefound. I wonder if he would comment on these statements.Prof. A. J. Kresge (Illinois Institute of Technology) said: We also observed thatazulene is unstable in strongly acidic solutions. Since the rate of decompositionis greatest when roughly comparable amounts of protonated and unprotonatedaromatic are present in the solution together, we believe that the reaction which occurs1 Davies, Ion Association (Butterworths, London, 1962), chap.3100 GENERAL DISCUSSIONis an electrophilic substitution of the cationic conjugate acid on the neutral molecule.There is some evidence to support this, e.g., in strongly acid solution, phloroglucinolis converted to “ phloroglucid ” which has a dimeric structure.1 Since this reactionis bimolecular, it depends on the square of the total azulene concentration, and,by working at high dilutions, we were able to reduce it to such an extent that it didnot interfere with our measurements.Prof. A. J. Kresge (Illinois Institute of Technology) (communicated) : The decom-position of azulene in acidic solutions has recently been found to be second orderin the aromatic, first order each in protonated and unprotonated forms, and therate constant reported is consistent with the rates of disappearance which we ob-served?Prof.F. A. Long (Cornell University) said: I would first comment that Prof.Gold is somewhat too modest about the significance of his results. For someyears, recognizing that the original Gross “ cubic ” formulation for deuteriumsolvent isotope effects was not the only way in which the data could be fitted, ithas been important to attempt to obtain data that explicitly showed behaviourwhich required a cubic or at least some high order of expression to give a fit. Asfar as I know, Gold’s fig. 1 which does show the behaviour expected from a cubicequation, is the first instance where this direct consequence of the Gross formulationhas been shown.This alone is a most significant result.I would make two different points about Gold’s results. A first point relatesto the formula for the solvated proton, specifically H3O+, or H90;t. In analyzingdata for the deuterium solvent isotope effects, we have noted also that one can obtainreasonable fractionation factors utilizing the I 3 9 0 model, values which are aboutas satisfactory as for H30+. Whether fractionation factors for related moleculeswill be as consistent under the H90,+ treatment as they now are under the H3O+treatment is unknown. In any case, I agree that other things being equal, the simplerH30+ treatment is to be preferred.The other comment concerns the possible role of medium effects.Prof. Salomaaof the University of Turku, Finland, when he was working at Cornell and sincethen, has been concerned with an interesting way to get values for the constant L.It consists of obtaining the ratio for the K1K2 product of a dibasic acid whichis added to the solution in anhydride form. Two examples with which Salomaahas worked are COZ and S02. Considering the first of these, the equations whichare relevant areC02+2H20 = H30++HC0,, K,HC0;+H20 = H30++COg- K2For equilibrium in this system, assuming that one works at constant activity,i.e., constant pressure of the anhydride C02, the degree of hydration of this speciesis immaterial and one obtains directly the equation :In other words, this gives a possible way of obtaining the value of L directly.Usingliterature data for the CO2 system, Salomaa reports that the resulting value of Lis about 14. His own experimental studies with sulphur dioxide have led to anL value of about 19. Both of these are large compared to the best experimentalvalue of L of 7-1 1. In a personal communication, Salomaa has suggested that this1 Beilsteiti, VI, 1099. 2 Myhre and Anderson, Tetrahedron Letters, 1965, 1497GENERAL DISCUSSION 101high L value results from the fact that there is a medium change in going from H20to D20, an effect which is for most situations absorbed into the fractionation factor,treated as a parameter. He further speculates that, because he has made his studiesof Kl and K2 under different medium conditions, the medium effect is here evident.This suggests that the assumption of ideality of the solvent, which most of theseanalyses of the H20+D20 system utilize, may be doubtful.Perhaps Prof. Goldwould comment.Prof. V. Gold (King’s College, London) said: The question whether protium-deuterium isotope effects depend on the isotopic composition of the medium ispartially answered by the results in table 1 of the paper by Gold and Kessick. Theisotope effect Y, which is an experimental quantity independent of the assumedreaction mechanism, shows no significant trend with the isotopic composition ofthe medium expressed by n. The same conclusion follows from the correspondingtritium product isotope effects in H20+D2O mixtures containing a trace amountof tritium.1Prof.Maurice M. Kreevoy (University of Minnesota) said : I would like to supportthe Gold and Kessick view on the invariance of Y under changes in the isotopiccomposition of the solvent. We have determined isotope effects both from pro-duct ratios and rate ratios for cleavage of vinylmercuric iodide 2 and allylmercuriciodide.3 I have described the mechanism of the latter reaction in my comment onthe paper of Caldin and Kasparian. The former is similar. Our results arequalitatively similar to those of Gold and Kessick. The table illustrates the in-variance of r under a change of a factor of 15 in the hydrogen to deuterium ratioin the solvent.(D/H) soh. 0.677 1-09 1.60 1.90 2.75 3-15 4.50 10-47r 7.16 7.34 7.37 7.18 7.24 7.44 6.83a 7-36 av. a 7-30f0.09(a) The average and the average deviation from the mean omit the value 6.83 which was theresult of the first measurement and varies from the mean by 5 average deviations.I also want to comment on the possibility of distinguishing between the directtransfer mechanism,A-@-+B,and the double transfer mechanism,HA--@-+O--@-+B,Ion the basis of isotope effects.I agree that this does not seem possible when HAis the hydronium ion. However, this can be done for monobasic acids. For thedirect transfer, with some reasonable assumptions, it can be predicted that RH/RD,the product ratio is a mixed solvent, should be given by [HA]~HA/[DA]~DA. Forthe double transfer mechanism RH/RD should be given by (H/D)so~v.(kHA/kDA) R,where R is the ratio of the following rates :1 Gold and Kessick, J . Chem. SOC., 1965, in press ; Pure and Appl. Chem., 1964, 8, 421.2 Kreevoy and Kretchmer, J . Amer. Chem. Soc., 1964, 86,2435.3 Kreevoy, Steinwand and Kayser, J. Amer. Chem. SOC., 1964, 86, 5013102 GENERAL DISCUSSIONMIA-@ + 0-0 -+ SMIA--@+O-@+S(M can be either hydrogen or deuterium but must be the same in both reactions.)There is some reason to believe that R should be not far from unity, and, in any event,there is no reason why (H/D)so~v.R should be consistently equal to [HA]/[DA]..We are now trying to evaluate the required quantities for allylmercuric iodidecleavage.Mr. Brian Case and Dr. Roger Parsons (University of Bristol) (communicated) :The equilibrium constant L (eqn.(1.2)) of the paper by Gold et al. can be evaluatedfrom L’ (eqn. (1.1)) without approximation from experimental data. The ratiois related to the difference in the standard real free energies a of solvation of thechloride ion in H20 and in D20 byThe difference in standard real free energies of solvation is a directly accessibleexperimental quantity since the standard e.m.f. of a cell such as 1L/L’ = [cl-]2/[cl-]2 (1)RT In (L/L’) = -2(a:;o-a:;o).Ag I AgCl I NaCl in HZO I air I NaCl in D,O I AgCl I Ag(2)(3)E*= - (a??- a$?)/F. (4)sat. sat.is given byThe potential of a cell with an air gap in which the field is zero can be measureddirectly using the Kenrick method, or indirectly using Kelvin’s method, or theradioactive method.*The calculations of Swain and Bader2 effectively replace the difference of realsolvation energy in (2) by the difference of chemical solvation energy as calculatedfrom the difference in the librational behaviour of the solvent in contact with theion. This is equivalent to the assumption that the surface potential x of H20 isequal to that of D20. In view of the importance of the libration of the solventmolecules in the isotope effect on solvation parameters it is possible that the sub-stitution of D20 for H20 in the surface may change x owing to a change in theaverage dipole orientation. While this effect is probably small because the valueof x itself is small ( N 100 mV for HzO) the effect on L may not be negligible since itis evident from (2) that L/L’ would be changed by a factor of 2 if ~D20- xH*0 were9 mV.We have recently made preliminary measurements of the e.m.f.of cell (3) byKenrick’s method using an apparatus similar to that described by Randles.4 Wefind a value of E* = -3.5f2 mV. This corresponds to a ratio L/L’ of 0-762&0.12and L = 13-9&2-0 which is somewhat higher than the value proposed by Gold andKessick. These results suggest that there is in fact a difference in x between waterand D20. Using Swain and Bader’s estimate of the difference in chemical solvationenergy for Cl-, this amounts to 6.6+2 mV.1 Parsons in Modern Aspects of Electrochemistry, I, ed. Bockris (Butterworths, London, 1954),4 Randles, Trans.Faradzy Soc., 1956, 52, 1573.pp. 115-127. 2 Swain and Bader, Tetrahedron, 1960, 10, 182GENERAL DISCUSSION 103Prof. M. Eigen (Gottirigen) said: Although there is no doubt that H3O+ in wateris further hydrated it is completely adequate-as Prof. Gold pointed out-to ascribehis results to the H30+ species. There may be some relation to the migrationmechanism of the proton in which the rate-limiting step is the incorporation of furtherH2O molecules in the outer hydration shell of H30+. Similarly, in proton transferreactions to H-bonded acceptors the partner must somehow become incorporatedin the hydration structure of H3O+. The proton will be transferred only if thenew acceptor finally becomes the centre of hydration; otherwise the proton willjump back into its original central position with a rate comparable to its valencevibration frequency (cf. Franck-Condon principle).Thus it is really the H3Of(i.e., the centre of the hydration complex) which has to reach the new acceptor,and the higher multiplicity of protons in the outer hydration sphere does not comeinto play. The situation might have been different for a mechanism in which aproton jump over the whole diameter of the hydration complex (i.e., from the originalcentre to a newly formed one) were the rate-limiting step.Prof. H. Zollinger (Swiss Federal Inst. Technology, Zurich) (communicated) : Theproblem of whether indirect proton transfers via a water molecule do or do not occurin reaction mechanisms has been investigated in connection with general base catalysisof diazo coupling reactions of 1-naphthol and 1 -naphthylamine derivatives in the2- and 4-positions.1 The effect of water molecules relative to the respective effectH H0..'* .. /f0'' 1, HCsoyIof acetate ions as proton acceptors for the intermediates in diazo coupling is 2.2times larger in the 2- than in the 4-position of 1-naphthol-3-sulphonic acid.Similarly, the effect of water relative to the respective effect of secondary phosphateions is 5.7 times larger for the diazo coupling intermediate in the 2-position of1 -naphthol-3-sulphonic acid than in the 2-position of 1 -naphthylamine-3-sulphonic.Together with the observation 2 that the activation entropy of diazo couplingsin o-positions are smaller (more negative) than that of the respective reactions inp-positions, the most feasible explanation of these results consists of a protontransfer (I) from the 2-position to the naphtholate oxygen.The water moleculeis fixed by a hydrogen bond to the oxygen and, therefore, is in a sterically favourableposition to accept a proton from the 2-position. In the 4-position such a transferis impossible. It is less probable with coupling at the 2-position of 1-naphthylaminebecause of the lower degree of solvation of the NH2 group in comparison of theO0 group of the naphtholate.Prof. V. Gold (King's College, London) (communicated) : The decision whethera particular item of experimental evidence is convincing is bound to be a subjectiveone. Prof. Long's generous remarks about our work should, therefore, not be al-1 Zollinger, Chem. and Ind., 1965, 885.2 Stamm and Zollinger, Helv. chirn. Acfa, 1955, 40, 0oO104 GENERAL DISCUSSIONlowed to pass without some mention of the fact that Purlee and Taft 1 studied thedependence of rate of olefin hydration on the isotopic composition of H20+DzOmixtures several years before us.Dr. A. Gandini and Dr. P. H. Plesch (University of Keele) said: Under strictlyanhydrous conditions the protonation of styrene and other phenylolefins by per-chloric acid in methylene dichloride is a relatively slow reaction which only goesto completion if the [HC104] is at least 102 [olefin]. We studied the reactionbetween styrene and perchloric acid with a high-vacuum spectroscopic device byfollowing the absorption at 427 mp due to the 1-phenylethyl ion. At roomtemperature (20-22") the rate law isd[CH,CHPh]/dt = k[HC104]a[C8H8],with k = (2-35&0.1) x 102 l.* mole-* sec-1 for [CgHg] in the range 10-4-10-3 Mand [HC104] at least a hundred times greater. The fractional power in acid con-centration indicates that the acid is associated in methylene dichloride. Only themost careful technique gave closely reproducible results since the carbonium ionsare very sensitive to traces of water.21 Purlee and Taft, J. Arner. Chem. SOC., 1956,78, 5807.2 Gandini and Plesch, J. Chem. SOC., in press