GENERAL DISCUSSION Dr. J. R. Jones (University of Surrey) (communicated) The reference to the decomposition of nitramide a reaction that has contributed greatly to the development of modern theories of acid-base catalysis prompts me to report some isotope effect data which I believe to be the first example of an isotope effect maximum for a nitrogen acid. The results in the table (kHrefers to ordinary nitramide in H20and in the presence of phenolate bases kD to deuteriated nitramide in D,O and also phenolate bases) represent the product of a solvent isotope effect (which probably lies between 2 and 3) and a primary isotope effect. The presence of the former does not however alter the fact that the primary isotope effect is at a maximum for catalysis by 2.4 6-trichlorophenolate and pentachlorophenolate anions and as the pK of nitramide is 6.48 this behaviour is similar to that observed for carbon acids.TABLE 1 .-ISOTOPE EFFECTS IN THE BASE-CATALYSED DECOMPOSITION OF NITRAMIDE AT 198.2 K base pK kH/1 mol-1 s-1 kHlkD water -1.74 8.5x 5.1 2,4-dini trophenol nitramide ion 4.09 6.48 0.0166 0.0368 3.5 5.4 pent achlorophenol 2,4,6-trichlorophenol2-nitrophenol 5.25 6.00 7.17 0.0320 0.449 1.84 10.2 9.0 6.3 This reaction also affords the unusual opportunity of studying catalysis by the anion of the substrate-that the isotope effect is lower than expected is probably related to the fact that the anion in aqueous solution is thought to possess the aci-form structure of nitramide. It is significant that the catalytic coefficient for the nitramide anion is much lower than that predicted from the Bronsted relationship established by phenolate bases.Prof. W. H. Saunders (University of Rochester) said How do the heats of solution of the neutral acids compare to those of the anions ? If the former are small compared to the latter it is easy to see why variations would be essentially negligible for the present purposes. If not changes in the energy of cavity formation with molecular size should be a significant factor. Prof. E. M. Arnett (University of Pittsburgh) said Because of electrostatic solvation the heats of solution of all the anions are much larger than those of the neutral acids. However in terms of the results presented in in this paper we are concerned with the relative effect of structural change on the anions compared to cyclopentadienyl anion and of the neutral acids compared to cyclopentadiene.In these terms again the calculated heats of solution of the anions are more sensitive to structure than are their neutral precursors-the range for the organic anions being about thirty kcal/mol and for their acids about six kcal/mol. The role of cavity size is indeed demonstrated in fig. 4 of our paper where a rough relationship between solvation energy and a crudely estimated reciprocal ionic radius is shown. I don’t believe that it is appropriate to interpret this correlation in terms of any particular solvation model since molar 50 GENERAL DISCUSSION volume terms play a significant role in the regular theory of solutions for non- electrolytes as well as in most electrostatic treatments of ions.The Born theory of ionic solvation does not work very precisely even for small spherical ions such as sodium or chloride. Its application to large complicated organic ions through correlations such as fig. 4,therefore represents the broad applicability of cavity terms which contribute both to solvation of the acid and its anion. Prof. El. P. Bell (University of Stirling) said The use of relative heats of solution in carbon tetrachloride in place of heats of vaporization is a reasonable procedure but I would like to ask whether this has been checked by using results for other non-polar solvents or by applying Trouton’s rule to the boiling points of the solutes.Prof. E. M. Arnett (University of Pittsburgh) said Yes indeed we have. A number of other non-polar solvents such as benzene and dichlorobenzene have been tried in our laboratory showing the same type of cancellation. A number of comparable measurements reported by Fuchs Drago Friedman and their students demonstrate the same type of cancellation. We admit to its vulnerability and believe that it will affect some of the results reported here but not very seriously. Prof. F. G. Bordwell (Northwestern University) said I would like to amplify and clarify the statement made in Arnett’s paper regarding “ acids. ..for which com- parable . . . pKs cannot be easily measured. ” The method for measuring pK’s outlined in our paper is applicable to most acids for which enthalpies can be deter- mined in DMSO.The one exception is alcohols. Here the formation of strongly hydrogen-bonded species e.g. RO- . . . H-OR causes interference. It is for such compounds that measurement of heats of deprotonation can provide supplemeiitary pK data. Prof. E. M. Arnett (University of Pittsburgh) said Bordwell is basically correct about his titration method with regard to which our claim sounds somewhat pejorative. When I made that statement I had in mind not only the problem of the alcohols but also the fact that all free energy measurements covering a broad spectrum of acidity require extrapolations with cumulative errors from compound to compound or from solvent to solvent. The particular value of our calorimetric procedure is that all acids both strong and weak are compared in exactly the same medium by the same method under the same conditions.Bordwell’s titration method is simple precise and elegant. Fig. 1 of our paper demonstrates the close parallel between his results and ours. Prof. E. F. Caldin (University of Kent) said For very fast proton-transfers in aprotic solvents such as those of substituted phenols considered by Crooks and Robinson the question is whether the rates and activation parameters can be explained by viscosity control of encounter or of rotation within the initial complex or whether we must consider also “chemical ” interactions such as desolvation breaking of internal hydrogen bonds and stabilisation of the initial complex by dispersion and dipole-dipole forces.When the experimental results are compared with those calculated by the simple Smoluchowski theory in which the reactant molecules are treated as spheres with uniformly reactive surfaces and translational diffusion alone is considered the rate constants k are smaller than predicted the slope of the plot of k-l against viscosity for a given reaction in a series of solvents is larger by a factor of the order of 10 than the predicted value 1/4RT and the activation enthalpy in a GENERAL DISCUSSION given solvent is often smaller than the value calculated from the viscosity one value being even negative. Crooks and Robinson find however that when they apply more sophisticated diffusion theory (due to solc and Stockmayer) in which only a fraction of the surface of each molecule is available for reaction the slope of the plot of k-' against viscosity can be explained if these fractions have values that appear not unreasonable (see their table 4).They note that the model ignores any effects of solvation and of internal hydrogen bonding. Recent work by Burfoot 'has added two more reactions to the list of those whose rates have been studied in a series of solvents and show anomalous viscosity- dependences (table I). An application of solc and Stockmayer's treatment (for which we are indebted to a personal communication from Dr. solc). in which the molecular surface area available for reaction was taken as the van der Waals area of the N or 0 atom gave predicted values for the slope of the plots of k-' against viscosity that were 2-10 times larger than the experimental values.The values for the reactions of 2,4,-dinitrophenol and picric acid differ by a factor of 5 although if internal hydrogen bonding is important it will affect the geometry of both acids ; and the value for picric aicd is about the same as for trichloroacetic with the same base. The variations of the enthalpy of activation of a given reaction from solvent to solvent and from reaction to reaction is a given solvent do not seem to be explicable by viscosity and geometrical factors alone. It appears that " chemical " factors such as solvation internal hydrogen bonding and complex-formation by dispersion forces must be considered as well. This can be done in terms of the rate constants for the individual steps of a three-step scheme such as given by Crooks and Robinson.More experimental work is needed to make clear the roles of translational and rotational diffusion. Dr. J. E. Crooks (Uttii*ersityof London) and Dr. B. H. Robinson (Utiirersitj of Kent) said We would not claim that the solc and Stockmayer equation gives exact numerical answers because even their model is highly simplified. However their treatment does explain from first principles why the plot of k-' against q has the observed form and why the slope is much greater than that predicted by the Smoluchowski equation. The values of AH for the reactions of Magenta E with a range of amine bases show a poor correlation with AH&c.but the average value of AH is equal to the average value of AH&.within the limits of experimental error. The negative value of AH for 2,4-dinitrophenol as acid may well be associated with the strong intra- molecular hydrogen bond. Prof. L. Melander (Giireborgs Utzirersitei) said Could the writers give some further comments on the tentative transition-state structures V and VI? As they stand it seems somewhat unnatural that the charge should be less delocalised in VI than in V in spite of the fact that the proton has been fully transferred to the nitrogen atom in VI while it is about half-way in V. If VI is the best picture according to the experimental facts is it the a-bond energy in the sultone ring that should be expected to offer resistance to the development of a delocalisation of the same kind as in V? Dr.J. E. Crooks (University of London) and Dr. B. H. Robinson (Utii~~ersiry of Kent) said Transition state V is intended to show a state of the system where the proton-transfer is occurring synchronously with ring opening. VI shows a transition ' K. Solc and W. H. Stockmayer Ifif. J. Cherri. Kirretics 1975 5 733. G. D. Burfoot and E.F. Caldin J.C.S. Furuduy I 1976 72 1O00. GENERAL DISCUSSION 53 state where the proton has transferred completely before ring opening has started i.e. the two processes are decoupled. This is the essential difference we are trying to make between V and VI. If the sultone ring is intact it is expected that most of the charge will be localised on the phenolate oxygen rather than in the benzenoid ring.In fact VI probably represents a highly energetic unstable intermediate and the transition state for the overall reaction will likely correspond to a species where rupture of the sultone ring has just begun. However it will be closest in structure to VI. The delocalization of charge in the particular case of BPB is slow for two reasons (a)As stated by Melander delocalization is slow because a C-0 0bond must be broken in the sultone ring for which some activation energy will be required. (When no 0bond rupture is needed for delocalization as in the case of proton transfer from Magenta E then this process will be very rapid.) (b) In a low-polarity medium the close proximity of the positive charge on the protonated amine will inhibit delocalization of negative charge.We conclude therefore that (i) Delocalization of charge (and by implication sultone-ring opening) is not synchronous with proton-transfer from BPB. (ii) The rate constant for ring-opening is still faster than that for proton transfer to weak bases. A value of lo6 s-' seems reasonable. (iii) Very slow ring-opening/delocalizationcan be observed when a sterically hindered acid is used. (See separate comment by Robinson and Parbhoo). Prof. E. F. Caldin (University of Kent) said The interpretation of a value for the activation volume AV* in a single solvent for a reaction producing an ion-pair is not simple. There are two contributions to AVF one from changes of bonding as the reactant molecules approach each other (AV,') the other from changes in the arrange- ment of solvent molecules (AV:).To an approximation these are additive AV* = AVT+AVZ. (1) If we identify AV; with the volume change AV& due to electrostriction produced by the charge-separation and calculate this for the formation of a point dipole (p*) in a cavity of radius r* in a medium of dielectric constant D,we obtain (putting q = (D-1)/(2D+ I)) This expression is evidently sensitive to the values of p* and r* which are not known. (With p* = 10 D and r * = 5& it gives AV& = -6 cm3 mol-'). In a series of solvents however AV* should be linear with (dq/dp), the intercept at (dqldp) = 0 will be AV and for any particular solvent AVZ can then be found as AV* -LIP':. For the Menschutkin reactions of pyridine with methyl iodide and triethylamine with ethyl iodide in chlorobenzene at 50°C this treatment gives AVZ = -7 cm3 mol-' for each reaction.The charge-development in the transition state of these reactions is probably comparable with that for the reaction of bromophenol blue with amines (the slope of the plot gives p* = 8 D) so we may seek to compare this value of AVF with one H. Hartmann H. D. Brauer H. Kelm and G. Rinck Zeit. phys. Chem. (Frankfurt) 1968 61 53. GENERAL DISCUSSION derived from Crooks and Robinson's value of A V* in chlorobenzene (-16 ~m~mo1-l). In the absence of experimental values in other solvents we estimate AV from a molecular model and use eqn (1). If we suppose that AVlf is the volume difference between OH.. N at the van der Waals distance (0. . N = 3.75 A) and the hydrogen- bonded distance (0.. N = 2.78 A),2 we find AVT = -17 cm3 mol-'. This value is comparable with the experimental value of AV* and would suggest that reorganisa- tion of solvent molecules is not responsible for any large contribution to AV*. The calculated value is however sensitive to the distances assumed. Dr. J. E. Crooks (University of London) and Dr. B. H. Robinson (University of Kent) said As stated by Caldin it is important to be able to dissect AVT into its separate contributions. In the terminology used in our paper the two contributions are A V;, the volume change on forming the intermediate hydrogen-bonded complex (analogous to AV:) and AV,. (AVg) the further volume change on forming the transition state which presumably is dominated by solvation changes in response to charge development (electrostriction).Then (cf. Caldin's eqn (l)) we have AV = AVY,+AV,',. The question is raised as to which of these terms dominates for our system and Caldin presents evidence in support of AV,",. In principle an estimate of AV& can be obtained from (2). However the calculation is sensitive to the size of the cavity. If p* = 3 D and Y* = 1.5 A (corresponding to proton-transfer along a hydrogen-bond) then A V2+-becomes -20 cm3 mol-' which is close to A Vf'. It would also be useful to make the proposed plot when more data become available but there must always be doubt as to whether conclusive evidence can be obtained when the analysis demands that macroscopic parameters are used for interactions in a microscopic environment of the system.(However viscosity works well). The calculation of AV; similarly must ignore any contribution due to changes in orientation and packing imposed by hydrogen-bond formation but these are very difficult to estimate. Direct experimental evidence on the magnitude of AV," is limited but several authors seem to favour a value of -5 cm3 m~l-'.~-' If this is the case then the large value of -AV would be associated mainly with solvent reorgan isat ion. However it is a relatively simple matter to measure AVr2 experimentally for the BPB-pyridine system using the MagE-pyridine system as a model (as used previously for the dissection of AH and AS:).All that is required is a visible spectrophotometer with an optical cell which can be pressurised to a few kbar. Then AYY2 is obtained from the pressure dependence of the equilibrium constant for the model system. It will then be possible to estimate with some confidence the absolute magnitudes of A V,02and A VA. AVf" has recently been measured for BPB and the more sterically hindered base 2t-butylpyridine in chlorobenzene. Although the initial H-bond formation is much weaker (KI2(2tBupy) = 0.66 M-l KI2(py) = 29 M-I) AV is large and negative (-21 cm3 mol-') and close to that observed for the pyridine system. C. D. Hubbard C. J. Wilson and E. F. Caldin J. Arner. Chem. Suc. 1976 98. S. N. Vinogradov and R. H. Linnell Hydrogen Bondin.9 (van Nostrand New York 1971).p. 178. W.J. Le Noble and T. Asano J. Ainer. Chem. Soc. 1975 97 1778. E. Whalley Adv. Phys. Org. Chem. 1964 2 93. E. Fishman and H. G. Drickamer J. Chem. Phys. 1956 24 548. T. Altinata B. H. Robinson and C. J. Wilson unpublished work. GENERAL DISCUSSION Dr. D. M. Parbhoo and Dr. B. H. Robinson (University of Kent) said Some preliminary experiments have been carried out on the system Bromothymol Blue (I) + Pyridine which is related to the Bromophenol Blue (I1 in paper) + Pyridine system. The two indicator acids differ only in the ring substituent groups. Bromothymol Blue (BTB) is more sterically hindered to rotation about the central carbon atom than Bromophenol Blue (BPB) due to the methyl group in the 3 position of the ring.Thermodynamic and kinetic parameters (by the stopped-flow method) have been measured for ion-pair formation from I. From consideration of equilibrium constant values the acids appear to behave similarly. However kinetic measurements indicate two discrete rate processes zf and z (7 > 5zl) for BTB in the stopped-flow time range. In contrast a single relaxation is observed for BPB under the same experimental conditions. For BTB zfis base concentration dependent but z is almost independent of base concentration. (C The results are consistent with the detailed mechanism proposed in the paper for BPB. It would appear that the faster rate process for BTB refers to the formation of an ion-pair species D’ resembling VI in the paper (i.e.proton transfer has been effected but the sultone ring is still intact). The slow process could then be associated with the opening of the sultone ring to form a more stable ion-pair. This involves rupture of a a-bond and delocalisation of charge from the phenolate ion to the sulphonate ion. D’ would have a lower extinction coefficient than D and so such a process can be observed spectrophotometrically. (The migration of the protonated amine cannot be observed in this way). It is possible that the ring-opening reaction is slower with BTB due to the effect of steric hindrance. It is perhaps also significant that BTB fails to form an ion-pair complex (A,, = 560nm)with the much stronger aliphatic amine bases due to the loss of the second proton. The results thus lend support to the idea of proton-transfer (step IV +.VI) decoupled from ring opening (VI -, 111). Prof. M. M. Kreevoy (University of Minnesota) said Many hydrogen bonded complexes in which the basicity of the two basic sites is about equal are mixtures of tautomers as required by the kinetic analysis given by Crooks and Robinson.’ However there are systems in which only a single intermediate structure can be observed. The u.-v. spectra of pyridine-1-oxide and its complexes with a series of acids of varying strength are shown in fig. 1.2 In fig. 2 is shown the spectrum of a solution containing a two-fold molar excess of pyridine- 1-oxide over trifluoro-methanesulphonic acid. The vibrational spectra and freezing point depressions of S. N.Vinogradov and R. H. Linneli Hydrogen Bonding (Van Nostrand Reinhold Co. New York 1971) pp. 163-169. ’K.-C. Chang Ph.D. Thesis (University of Minnesota 1975). GENERAL DISCUSSION such solutions show that simple one-to-one complexes are involved.'* This is confirmed by the X-ray diffraction pattern of a related solid which also shows that the oxygen-oxygen distance is only a little over 2.4A.3 These ultra-violet spectra are inconsistent with the existence of a tautomeric equilibrium in these complexes. The spectra in a series of tautomeric mixtures should show a steady decrease in the A A Inm FIG.1.-The u.-v. spectra of pyridine-1-oxidc and a series of its complexes with carboxylic and sulphonic acids in sulpholane solution. Curve A is for 0.081 M pyridine-l-oxide path length 5.01 pm ; curve B for 0.065 M complex with CH,COOH path length 6.15 pm; curve C for 0.076 M complex with CHCl,COOH path length 6.73 pm ; curve D for 0.101 M complex with CF,COOH path length 14.0 pm ; curve E 0.151 M complex with CH3S020H path length 8.87 pm ; curve F 0.093 M complex with CF3S020H path length 14.3 pm.The progressive blue shift of the charge transfer band is indicated by --; the location of the benzenoid band by---. Because both the concentration and the path length were variable the molar absorptivities A/cl are indicated at one or more points on each spectrum to make a comparison of intensities easier. intensity of the bands due to the free base with a concomitant increase in the bands due to the protonated base without much change in the wavelengths of maxiinurn absorption.If the spectra are superimposed one or more isosbestic points should be observed. Vinogradov and Grunwald and their coworkers have observed such patterns. In contrast each spectrum in the present series contains only one set of bands the location of which shifts continuously with the increasing strength of the acid from that characteristic of the free base to that characteristic of the protonated base. Thus no distinguishable tautomers can exist for more than about s (the K.-C. Chang Ph. D. Thesis (University of Minnesota 1975). J. Husar and M. M. Kreevoy J. Arner. Chenr. Soc. 1972 94 2902. L. GoliE and F. Lazarini Vestnik Sloc. Kern. Dr. 1974 21 17. 'S. N. Vinogradov and R. H.Linnell Hydrogen Bonding (Van Nostrand Reinhold Co. New York 1971) pp. 163-169. D. Eustace and E. Grunwald J. Anier. Chem. SOC.,1974 96 7171. GENERAL DISCUSSION characteristic time of an ultra-violet experiment). The homoconjugate complex of pyridine-1-oxide and its conjugate acid (fig. 2) and also that of 3,5-dinitrophenol and A 200 250 300 h/nm FIG.2.-The u.-v. spectrum generated by a solution of 0.148 M pyridine-1-oxide and 0.074 M CF3S020H,in sulpholane. The path length is 5.91 pm. The molar absorptivity is indicated. its conjugate base,” behave similarly but that of p-nitrophenol is a tautomeric mixture by this criterion.2 In the cases where tautomers do not exist the potential function for hydrogenic motion along the line connecting the two basic sites must have only a single minimum or if there are two the central maximum must not rise significantly above the lowest allowed vibrational It would be of interest to carry out the Crooks and Robinson experiments with substances of this sort since eqn (9) then simplifies to eqn (5).Prof. J. H. Fendler (Texas A & M University) said As part of our interest in the behaviour of surfactants in nonpolar solvents we have investigated the formation of hydrogen bonded ion-pair complexes between indicators and polyoxyethylene(6) nonylphenol in dry benzene as well as apparent dissociation constants in restricted volumes of water solubilized in benzene by surfactants. Bromophenol blue in dry benzene transfers protons to the fraction of polyoxyethylene(6) nonylphenol which is ionized to from two complexes with absorption maxima at 41 1-412 nm (complex I) and 580-590 nm (complex IX) respectively.The equilibrium constant for the for- mation of c~mplex I K, has been calculated at absorbances at 41 1-412 nm from Ao-A 1 log -= log -+log [polyoxyethylene (6) nonylphenol] (1) A-A KI where Ao A and A are absorbances of bromophenol blue in benzene in the absence of polyoxyethylene(6) nonylphenol that in its presence and that when all of bromo-phenol blue is in the form of complex I. The value obtained for K, 66.6 M-I is considerably smaller than those determined for the interaction of bromophenol blue with aliphatic amines in chlorobenzene by Crooks and coworkers. The slope of the line in the plot of the left hand side of eqn (1) against log [polyoxyethylene(6) nonylphenol] at relatively low surfactant concentration is 0.94 implying a 1 1 I.M. Kolthoff M. K. Chantooni Jr. and S. Bhowmik J. Anrer. Chem. SOC.,1966 88 5430. T.-M. Liang unpublished results. J. C. Speakman Structure and Bonding 1972 12 141. GENERAL DISCUSSION stoichiometry. At higher surfactant concentration however there is a significant deviation from this stoichiometry. This deviation is the consequence of the dynamic formation of surfactant aggregates and the interaction of these species with bromo- phenol blue. The spectrophotometric data treated in terms of monomer ~1dimer + trimer . . . +n-mer type association afford the calculation of the average aggregation number of polyoxyethylene(6) nonylphenol in benzene and the relative concentrations of each species.Significantly bromophenol blue assists the formation of higher aggregates. Apparent dissociation constants pKtPP values have been determined for malachite green bromophenol blue thymol blue and methyl orange in water pools solubilized by polyoxyethylene(6) nonylphenol in benzene by titration with HC104. The pKtpp values in this medium differ from the corresponding pK values in bulk water by as much as 7.0 units. For malachite green effects on pKtppincreases with increasing ratios of [polyoxyethylene(6) nonylphenol]/[H,O]. At given HClO concentrations ratios of unprotonated to protonated malachite green increase with increasing concentrations of the surfactant.In no case however is the deprotonation complete. The higher the HC104 concentration the less complete is the deprotonation. Water pools entrapped by polyoxyethylene(6) nonylphenol in benzene provide therefore a relatively basic medium. Dr. R. A. More O’Ferrall (University College Dublin) said In connection with the question of whether a concerted reaction can occur in conditions where it is not ‘‘enforced ” by the instability of an intermediate it has been claimed recently that in olefin forming eliminations of fluorene derivatives i.e. CH3O-/CH30H22 5’ RCHCH2X + RC = CH2+X-where RCH2 is fluorene reaction can occur by a concerted E2 mechanism even though the fluorenyl carbanion is quite stable. The principal evidence for a concerted reaction is the high bromide/chloride elimination rate ratio ; with two methyl substituents a to the leaving groups the ratio is 30.As calculated from the pK,’s of the fluorenes the free energies of the potential carbanion intermediates are some ten kcal lower than those of the E2 transition states. Prof. W. P. Jencks (Brandeis University) said If the leaving group becomes sufficiently good it will leave with no significant energy barrier the “intermediate ” will have too short a lifetime to exist and the reaction must become ‘‘concerted ” rather than step-wise. When it becomes concerted it is likely that there will be significant breaking of the bond to the leaving group in the transition state which in turn facilitates proton removal. Is it known whether there is a significant barrier for expulsion of good leaving groups such as bromide from a carbanion “inter- mediate ” in this reaction? Dr.R A. More O’Ferrall (University College Dublin) said :No that is not known. Prof. W. H. Saunders (University of Rochester) said I think it unlikely that the eliminations from fluorenylmethyl halides follow the E2 path solely because of instability of the carbanion intermediate that would be involved in a stepwise path. Taken to its logical conclusion this hypothesis would put the carbanion along the reaction path at some point after the transition state but make it unobservable as an intermediate because it would decompose to products without an activation R. A. More O’Ferrall and P. J. Warren Chem. Comm. 1975,484.GENERAL DISCUSSION energy. In such a case a sizable bromide/chloride rate ratio would not be observed because the bond to the leaving group would still be intact at the transition state. There may well be different requirements for concertedness for elimination reactions which involve proton transfers from carbon and the type of reactions Jencks deals with which usually involve proton transfers between oxygen and/or nitrogen atoms. In the latter case the proton transfers are diffusion-controlled so long as they are exothermic. Thus no energetic advantage is to be gained from concertedness unless it makes an otherwise endothermic proton transfer less endo- thermic. The proton transfers in elimination reactions are in contrast slow even when exothermic and a concerted process can reduce the activation energy under any circumstances except where a full carbanion on the P-carbon is required before expulsion of the leaving group can begin.This ready availability of a lower-energy path would seem a sufficient condition for concertedness without the necessity for any additional assumption about the instability of the carbanion.