Aust. J. Chem., 1979, 32, 1735-42

Steric Effects in Forward and Reverse Menschutkin Reactions of Some Pyridines, Quinolines and Thiazoles

Leslie W. Deady, Wayne L. Finlayson and Olga L. Korytsky

Organic Chemistry Department, La Trobe University, Bundoora, Vic. 3083.

Abstract

The effects of ortho alkyl substituents on the rates of methylation of pyridine and demethylation of N-methylpyridinium iodides are compared (some quinolines are included as 'substituted pyridines'). For monosubstituted compounds, substituent steric effects are twice as important in the methylation (forward) reaction as in the demethyIation reaction. In contrast, the ratio is 3 : 1 for disubstituted compounds, because the substituent effects are additive in the reverse reaction but greater than additive in the forward reaction. Some thiazole results are also reported and, here, steric effects on the forward reaction are small while no steric acceleration of the reverse reaction is evident.

Introduction

Of the numerous methods developed in the last decade for the dealkylation of quaternary salts of nitrogen heterocycles, reaction with triphenylphosphine in an aprotic solvent is probably the best general meth0d.l It is also well suited to quanti- tative study and one report dealing with steric and electronic effects in methylpyridin- ium salts has a p ~ e a r e d . ~ It has since been shown3 that the iodide counter ion rather than the triphenylphosphine is the dealkylating agent, which means that this de- quaternization (reverse) reaction is the reverse of the quaternization (forward) reaction with methyl iodide. This is important, in view of the continuing interest in the Menschutkin r e a ~ t i o n , ~ , ~ because it is now possible to approach the transition state for this classic SN2 reaction from either direction. The results previously reportedZ are still valid since all were obtained from iodide salts by a competition technique.

It was found2 that steric effects were more pronounced in the forward reaction. With the series of 2-alkylpyridines (Me, Et, Pri, But) of similar basicity, the steric effects were treated quantitatively and a linear correlation of forward and reverse rates was obtained. Methylation was hindered by the bulky substituents while de- methylation was enhanced (due to relief of steric strain) and the sensitivity to steric effects was twice as important in the forward reaction. This was then related t o the position of the transition state on the reaction coordinate.

Kutney, J. P., and Greenhouse, R., Synth. Commun., 1975, 5, 119. Berg, U., Gallo, R., and Metzger, J., J. Org. Chem., 1976, 41, 2621. Deady, L. W., and Korytsky, 0. L., Tetrahedron Lett., 1979, 451. Le Noble, W. J., and Asano, T., J. Am. Chem. Soc., 1975,97, 1778. Arnett, E. M., and Petro, C., J. Am. Chem. Soc., 1976, 98, 1468.

L. W. Deady, W. L. Finlayson and 0. L. Korytsky

We have been interested in steric hindrance in quaternization reactions6 and it seemed that this system was an ideal one for probing the steric accelerating effects of other ortho substituents on the reverse reaction. This work is concerned with the combined effect of two ortlzo substituents on azine reactivity.

We wished to restrict the ortho substituents to alkyl ones so that differences in electronic effects would be minimal. While it is not easy to obtain a suitable range of dialkylpyridines, if quinoline is regarded as benzopyridine then a greater range of compounds is readily accessible. Of the compounds (1)-(10) studied, (7)-(10) can be regarded as combinations of the substituents in (2)-(5). As methylation data were not available for all compounds, these also were obtained. For comparison purposes, the reactions of the five-membered ring compounds (1 1)-(14) were also studied.

M e

Results and Discussion Methylations

The compounds studied covered a wide range of reactivities and a combination of methods was required to get the necessary results. Relative reactivities for (1)-(4), (7) and (9) were known from previous s t u d i e ~ . ~ , ~ The reactivity of (8) was obtained by competition with (9) for methyl iodide in dimethyl sulfoxide at room temperature. The reactivities of (5) and (6) were obtained from absolute rate determinations in dimethyl sulfate. Reactions were too slow at room temperature and were measured at 82". In order to be able to relate these results to others obtained at a lower tempera- ture, the rate of methylation of isoxazole (judged to have comparable reactivity from the knowng result at 33") was also determined at 82" and relative reactivities could then be obtained. It was assumed that the relative reactivities would not be significantly affected by the change in temperature. The reactivity of (lo) could not be obtained by this method. Reaction was sufficiently slow that some decomposition of the dimethyl

Deady, L. W., and Stillman, D. C., Aust. J. Chem., 1976,29, 1745. ' Deady, L. W., and Zoltewicz, J. A., J. Oug. Chem., 1972,37, 603.

Zoltewicz, J. A., and Deady, L. W., J. Am. Chem. Soc., 1972, 94, 2765. Davis, M., Deady, L. W., and Homfeld, E., Aust. J. Chem., 1974, 27, 1221.

Menschutkin Reactions of Pyridines, Quinolines and Thiazoles

sulfate seemed to occur to give acidic by-products which affected the reactivity of the basic quinoline. For this compound, the required result was obtained by competition with (6) for methyl fluorosulfonate in sulfolane at room temperature. It has been foundlo that, in methylation reactions, steric effects are less pronounced for reactions with methyl fluorosulfonate than for methyl iodide by a factor of 0-69. The experi- mental value for (10) was therefore adjusted by this amount. From these results, by using known relative reactivity data, it was possible to calculate reactivities relative to pyridine. These are listed in Table 1. While a combination of methods was required to get these results and various approximations have had to be made, it should be noted that the reactivity differences are very large (2,8-dimethylquinoline is lo6 times less reactive than pyridine) and uncertainties in individual values will not affect the conclusions drawn below.

Methylation data for two of the thiazoles with methyl iodide in dimethyl sulfoxide were available1' and results for the other two were obtained in the same way. Reactivi- ties relative to pyridine are given in Table 1.

Table 1. Methylation (relative to pyridine) and demethylation (relative to 1-methylpyridinium iodide) reactivities

Hetero- log krei cycle Methylation Demethylation

2-Methylpyridine -0.42 0 2-Ethylpyridine -0.77 0.27 Quinoline -0.96 0.48 8-Methylquinoline -4.12 2.19 8-Ethylquinoline -4.54 2.45 2,6-Dimethylpyridine -1.64 0.16 2-Ethyl-6-methylpyridine -2.46 0.40 2-Methylquinoline -2.21 0.50 2,8-Dimethylquinoline -5.97 2.31 Thiazole -1.32 0.85 2,4-Dimethylthiazole -1.22 0.25 Benzothiazole -2.22 1.32 2-Methylbenzothiazole - 2.49 1 .04

The new results are qualitatively as expected in that considerable steric hindrance is evident in the azines. The 8-methylbenzo 'substituent' in (5) has a tenfold greater effect than does a t-butyl substituent [logk,,, -3.1 (ref.2)]. Steric hindrance also occurs in the thiazole reactions but the effects are much smaller.

In order to quantify the steric effects it is necessary to take account of substituent electronic effects and this can be done by considering the Bronsted plot," it being

* pK,values are taken from ref.12 except for 8-ethylquinoline (assumed same as for 8-methylquinolinej, 2-ethyl-6-methylpyridine (assumed same as for 2,6-dimethylpyridinej, 2,4-dirnethylthia~ole,'~ benzo- thiazole14 and 2-methylbenzothiazole (methyl group assumed, as in pyridine and quinoline, to add 0.8 to the pK, of benzothiazolej.

'O Berg, U., Gallo, R., Metzger, J., and Chanon, M., J. Am. Chem. Soc., 1976, 98, 1260. " Deady, L. W., Aust. J. Chem., 1973, 26, 1949. l2 Albert, A., in 'Physical Methods in Heterocyclic Chemistry' (Ed. A. R. Katritzky) Vol. 1, p. 1 (Academic Press: New York 1963). l 3 Phan-Tan-Luu, E., Surzur, J. M., Metzger, J., Aune, J. P., and Dupuy, C., Bull. Soc. Chim. Fr., 1967, 3274. l4 Albert, A,, Goldacre, R., and Phillips, J. M., J. Chem. Soc., 1948, 2240.

L . W. Deady, W. L. Finlayson and 0. L. Korytsky

reasonably assumed that basicity, as opposed to nucleophilicity, is not subject to significant steric effects. The solid line in Fig. l a is the correlation line for methylation of 3- and 4-substituted pyridines,l i.e., the substituent electronic effects in the methyla- tion reaction are proportional to their effects on basicity. The broken line is a parallel one drawn through thiazole, assumed here to be a different class of Bronsted base. Vertical deviations from these lines can then be taken as a measure of substituent steric hindrance effects.

Fig. 1. Bronsted plots for (a) methylation; (b) demethylation.

Demethylations

These reactions were all carried out on iodide salts, in an aprotic solvent in the presence of an excess of triphenylphosphine. The literature work2 used a competition method in dimethylformamide at 150" and this was found to be suitable for a number of compounds. For the most reactive compounds, demethylation caused by the solvent was significant and so a number of the reactivities were obtained by measuring absolute rates of demethylation in sulfolane solvent. Reactions at 150" were too fast for the most reactive quinolines and these were measured at 130". By measuring the de- methylation of I-methylquinolinium iodide at both temperatures it was possible to then relate all relative reactivities on the same scale. These are listed in Table 1.

The results for the azines show the expected trends, at least in qualitative terms, in that steric acceleration to demethylation is provided by bulky ortho substituents. How- ever, the range of reactivities is much less than in the methylation process.

As for the methylation results, the magnitude of the steric accelerating effects can be obtained from deviations from the Bronsted plot (Fig. lb). The solid line in Fig. Ib is that derived for reactions of 3- and 4-substituted methylpyridinium iodide^.^ It is of interest in this case that the points for all four thiazoles lie on a line parallel

'"oltewicz, J. A., and Deady, L. W., Adv. Heterocycl. Chem., 1978, 22, 71.

Menschutkin Reactions of Pyridines, Quinolines and Thiazoles

to the pyridine one, ie . , the reactivities are determined solely by substituent electronic effects and no steric acceleration is provided by these substituents.

Comparison of Methylation and Demethylation Results

In Table 2 are noted the deviations in log units of the experimental points from the respective Bronsted plots and a number of interesting points arise.

Table 2. Deviations of substituents from Bronsted plots In log units. Values in parenthesis are calculated assuming additivity of effects

-- -

Compounds Substituents Methylation Demethylation

Azines 2-methyl 2-ethyl benzo 8-methylbenzo 8-ethylbenzo 2,6-dimethyl 2-ethyl-6-methyl benzo-2-methyl 8-methylbenzo-2-methyl

Thiazoles 2,4-dimethyl benzo benzo-2-methyl

The situation in the thiazole compounds is quite uncomplicated. As noted in earlier discussion, steric hindrance by substituents to methylation is much less than in the corresponding azines. Since the reverse reaction is less susceptible to steric effects, it is probably not surprising that, for the thiazoles, no deviations are observed.

By contrast, the azine results are much more involved. For the monosubstituted compounds, steric effects occur in both reactions but are twice as important in the forward reaction. This observation was made previously for four 2-alkylpyridines2 and our results indicate it to be general for ortho substituents. The result was inter- preted to indicate that the strain in the transition state was about two-thirds of that in the products, a deduction previously made by Brown from other evidence.16 This in itself is interesting since the general consensus is that the transition state for the forward reaction is 'early'.4,5,17

The most intriguing results come from the disubstituted azines which, surprisingly, do not obey the same inverse relationship. Here, the ratio of hindrance in the forward reaction to acceleration in the reverse reaction is approximately 3 : 1. This arises because the effects of the two substituents are non-additive in the forward reaction but are close to additive in the reverse reaction.

The non-additivity effect on the forward reaction has long been known and the explanation suggested18 for 2,6-lutidine was that, with a single methyl group, the molecules could bend slightly so as to minimize the transition-state energy. With a methyl group in both ortho positions this strain-reduction mechanism is no longer

l6 Brown, H. C . , Gintis, D., and Domash, L., J. Am. Chem. Soc., 1956, 78, 5387. l7 Swain, C . G., and Hershey, N. D., J. Am. Chem. Soc., 1972,94, 1901. l8 Brown, H. C., Gintis, D., and Podall, H., J. Am. Chem. Soc., 1956, 78, 5375.

L. W. Deady, W. L. Finlayson and 0. L. Korytsky

available and the strain buildup is more than twice that observed for the mono- substituted compound. It is important to realize that the strain effects are also greater than additive in the quaternary products16 and we are therefore left with the situation that substituent effects are non-additive in the transition state and products, but addi- tive in the reverse reaction. The deduction from this is that the 'non-additivity factor' is the same in the transition state and products, i.e., though the strain increases from transition state to products, the effect giving rise to the non-additivity is already present in the transition state and does not increase further. If the strain in the transition state is close to that in the products this would be true. Indeed, it is tempting to interpret the 3 : 1 ratio for the disubstituted compounds as meaning that the transition state for these is closer to products and has a greater proportion of final strain (314) than for the monosubstituted compounds, but this is probably incorrect.

In this system, it is assumed that strain effects do not exist in the free amines and all are therefore of equal energy. The position of the transition state on the reaction coordinate will therefore, according to the Hammond postulate,4J9 be related to the activation energy, i.e., the slower the forward reaction, the closer the transition state is to products. It is therefore important to note that the 3 : 1 ratio of effects is a phenomenon of disubstitution and is not related to reactivity. For example, 8-methyl- quinoline (5), a 'mono compound', is methylated more slowly than is 2-methyl- quinoline (9), a 'di compound', and the inverse reactivity order applies to demethyla- tion. As a result, we are not convinced that the ratio of substituent effects on forward and reverse reactions does give a measure of the proportion of product strain that exists in the transition state. (The estimate of 213 deduced by Browni6 was for both 2,6-lutidine and 2-t-butylpyridine, which is not compatible with the results reported here.)

The foregoing discussion sets out the problem and we do not have an explanation for the different behaviour of mono- and di-substituted azines. It seems that, in addition to the steric effects of substituents on the methylating entity, there is at least one unknown factor which is more marked for disubstituted compounds and is more important in one or other of the two reactions.

A factor which has not been discussed but which is known to be important in the quaternization reaction is s ~ l v a t i o n . ~ For the reverse reaction, solvent effects are an obvious area for further study, and the solvation changes in proceeding to the transi- tion state may well be quite different from each direction.

The interesting results reported here show that a study of the demethylation reaction, whereby it is possible to approach the transition state for the Menschutkin reaction from the reverse direction, may be a powerful aid in increasing our knowledge of the detail of this formally simple substitution reaction.

Experimental Compounds

8-Ethyl-" and 2,8-dimethyl-quinoline2' (as for 2,7-dimethylq~inoIine~~) were prepared by Skraup syntheses. 2-Ethyl-6-methylpyridine, b.p. 160" (MZ3 160-161.5"), was prepared from 2,6-

l9 Hammond, G. S., J. Am. Chem. Soc., 1955, 77, 334. 'O Glenn, R. A., and Bailey, J. R., J. Am. Chem. Soc., 1941, 63, 639.

Karr, C., Estep, P. A., and Papa, A. J., J. Am. Chem. Soc., 1959, 81, 152. 22 Bowen, D. M., Belfit, R. W., and Walser, R. A., J. Am. Chem. Soc., 1953, 75, 4307. 23 Lochte, H. L., and Cheavens, I. H., J. Am. Chem. Soc., 1957, 79, 1667.

Menschutkin Reactions of Pyridines, Quinolines and Thiazoles

lutidine by reaction with phenyllithium and methyl iodide.24 The original aim was to prepare 2,6- diethylpyridine but alkylation of only one methyl group occurred in our experiment. The remaining azines and thiazoles were commercial samples whose identity was checked by n.m.r. before use.

Most of the required quaternary salts were prepared by reaction with methyl iodide in ethanol and had melting points in agreement with literature values. 6-Ethyl-1,2-dimethylpyridinium iodide had m.p. 148" (Found: C, 40.8; H, 5.3; N, 4.7. C9H141N requires C, 41.1; H, 5.3; N, 5.3%). For thiazole and 2,4-dimethylthiazole, reaction was carried out in ethyl acetate at room temperature and the precipitated salt was washed with ether. 8-Methyl-, 8-ethyl- and 2,8-dimethyl-quinoline were quaternized by reaction with methyl iodide in a sealed tube at 90" for 1-2 days. In this way the follow- ing salts were prepared. I,8-Dimethylquinolinium iodide, which rapidly formed the hydrate, m.p. 213" (from ethanol) (Found: C, 43.8; H, 4.3; N, 4.8. C11H14XN0 requires C, 43.6; H, 4.6; N, 4.6%). 8-Ethyl-I-methylquinolinium iodide, m.p. 171-172" (from ethanol) (Found: C, 48.2; H, 4.8; N, 4.5. Cl2HI4IN requires C, 48.2; H, 4.7; N, 4.7%). 1,2,8-Trimethylquinolinium iodide, m.p. 220-222" (from ethanol) (Found: C, 47.8; H, 4.8; N, 4.5. CI2Hl4IN requires C, 48.2; H, 4.7; N, 4.7%).

Methylation Kinetics

(A) Competition reactions.-A standard m e t h ~ d ~ , ~ ~ was used to obtain the relative reactivities listed in Table 3.

Table 3. Competition methylations and competition demethylations

Competition methylations Competition demethylations Competing pairs log krel Competing pairsA log krd

A All heterocycles were present as their N-methyl quaternary iodides. More upfield product signal of pair. Competing for Me1 in Me2S0. Competing for FS03Me in sulfolane. The experimental result has been corrected by the addition of 0.16 as the factorlo relating

10gk,,~(FS0~Me) to logk,,,(MeI). More upfield N-methyl signal of pair.

(B) Absolute rates.-Rates of reaction in dimethyl sulfate (treated with solid sodium bicarbonate before use) at 82" were measured for isoxazole, and 8-methyl- and 8-ethyl-quinoline by an n.m.r. technique." For isoxazole, all reactant and product ring proton signals could be distinguished. The spectra of the quinolines were characterized by two one-proton doublets (A and B) downfield (A further) from the main group of aromatic peaks. Formation of product was accompanied by the appearance of a two-proton signal (C) which partly overlapped A. Thus, product was given by +(A+ C-B) and an internal standard was not necessary because A+B+ C was constant. In these ways, pseudo-first-order rate constants of 26.0 x (isoxazole), 22.2 x (8-methylquinoline) and 8 . 4 ~ s-I (8-ethylquinoline) were obtained. These are average values from 3-5 runs with individual values being within 4% of the average.

In order to correct the above results to the common scale of reactivity listed in Table 1, the factors log k,,l(pyridine/isoxazole) 4.05,9 log k,,,(pyridine/2-methylquinoline) 2.21 ,7 log krel(pyridine/3- bromopyridine) 0.80 (obtained from Bronsted correlati~n'~) and l~gk,,~(pyridine/benzothiazole) 2.22 (ref.") were required.

24 Nakashima, T., Yakugaku Zasshi, 1958, 78, 666 (Chem. Abstr., 1958,52, 18400a). 25 Deady, L. W., and Zoltewicz, J. A., J. Am. Chem. Soc., 1971, 93, 5475.

L. W. Deady, W. L. Finlayson and 0. L. Korytsky

Demethylation Kinetics

(A) Competition reactions.-The published method,* dimethylformamide solvent being used at 150°, was followed and the results quoted in Table 3 are the average of 2-4 runs, with at least four k,,, determinations being made during each run. The t-butyl signal of p-t-butyltoluene was used as internal standard.

Table 4. Rate data for demethylation

Hetero- Temp. 104kj, Hetero- Temp. 1O4k+ cycleA ("C) W1) cycleA ("0 W1)

8-Methylq~inoline~,~ 130 8.12 2-ethyl-6-methylpyridineB3c 150 1.08 8-Ethylq~inol ine~,~ 130 15.0 thiazole" 150 3.04 2,8-Dimethylq~inoline~,~ 130 10.7 2,4-dimethylthia~ole~,~ 150 0.77 Q ~ i n o l i n e ~ ~ ~ 130 0.16 benzothiazoleE 150 8.97 Q ~ i n o l i n e ~ * ~ 150 1 .30 2-methylbenz0thiazole~~~ 150 4.70

A As the N-methyl quaternary iodide. Dibenzyl ether as internal reference. Disappearance of the N-methyl signal followed. Triphenylmethane as internal reference. H 2 in reactant and product followed. H 2 of benzothiazole as internal reference.

(B) Absolute rafes.-A solution of the appropriate quaternary iodide (0.15 M) and triphenyl- phosphine (1.4 M) in sulfolane was heated and the reaction was monitored by n.m.r. The pseudo- first-order rate constants were obtained by normal graphical means and the values quoted in Table 4 are the average of 2-4 runs.

Manuscript received 15 March 1979