The study of solvent effects in reactivity has been one of the cornerstones of physical organic chemistry, and still constitutes these days one of the most fascinating scientific challenges. The ongoing interest over these studies not any longer resides in the possibility of making predictions of rate constants for other solvents, but especially in its potential to assist us in the understanding of the true nature of substrate - solvent - solvent interactions, at a molecular level. In this context, there has been a continuing interest from the scientific community in the study of the solvolysis reactions of tertiary alkyl halides, which are still considered good model systems to monitor solvent and solvation effects [1–3]. One of the most used linear solvation energy relationships (LSER) to study solvent effects, in particular in aqueous organic media, has been the Grunwald-Winstein (G-W) equation, proposed for the first time in 1948 [4]. This equation, originally intended for use with S_N1 (and E₁) solvolyses, correlates the reaction rate, k, of an RX substrate in a given solvent with an empirical solvent parameter, Y (eq. 1):

where m represents the substrate sensitivity towards changes in the solvent ionizing power, Y, and the independent parameter c is the substrate rate constant in a reference solvent, usually 80% ethanol/water (v/v). The Y scale was initially derived for the solvolyses of t-butyl chloride, but it has been shown that this compound’s solvolyses occur with a non-negligible nucleophilic character [2, 5–7]. For this reason, the most commonly used Y scales are now based on the solvolyses of 1- and 2-adamantyl derivatives (originating different Y_x scales depending on the leaving group X), since these substrates are unable of undergoing either elimination or rearside nucleophilic attack from the solvent (or at least are strongly impeded from doing so) and seem therefore more appropriate model substrates to define scales of solvent ionizing power [2, 8] (during the peer review process of this manuscript, one of the Editors kindly drew these authors’ attention to a very recent paper on development and applications of the G-W equation [9]; in the referred paper, a very exhaustive review on the establishment and development of the several scales used in G-W approaches is made.).

In order to be able to apply eq. 1 to substrates reacting with the involvement of some nucleophilic solvation contribution, the introduction of an additional term to eq. 1 to explicitly consider the solvent’s nucleophilicity, N, was proposed (eq. 2) [10–12]:

In this extended Grunwald-Winstein equation, l is the sensitivity of the substrate to changes in solvent nucleophilicity. Throughout the years a number of solvent nucleophilicity scales have been proposed but the most frequently used are the N_OTs scale based on methyl tosylate solvolysis [7,13–15] and the N_T scale based on the solvolysis of S-methyldibenzothiophenium ion [7, 14–16]. As Mayr et al. [14] pointed out, in spite of the good correlation of these two scales with each other, as well as with some other solvent nucleophilicity scales, there is a still a vivid ongoing discussion on the role of nucleophilic solvent participation (NSP) in solvolysis reactions [17, 18]. This is certainly related to the indiscriminate use of the terms nucleophilic solvent participation and nucleophilic solvation (NS) as recently claimed by Richard et al. and other authors [3, 19].

Equations (1) and (2) have been successfully applied to substrates with localized charges [20, 21]. However, when the substrate’s cationic charge becomes resonance delocalized, the solvent stabilization effect can no longer be accounted for solely by the solvent parameter Y [22, 23]. Kevill et al. have shown that the introduction of a new solvent parameter, which they named as “aromatic ring parameter”, I, based on the solvolyses of (p-methoxybenzyl)dimethylsulfonium ion and 1-adamantyldimethylsulfonium ion (as stated by Kevill and co-workers, the I scale is based on differences in the nature of the variations of rate constants of solvolysis of the p- (methoxybenzyl)dimethylsulfonium ion and of the 1-adamantyldimethylsulfonium ion (no aromatic ring) as the solvent composition is altered: I = log (k/k₀)_{p-MeOC6H4CH2SMe2} -1.3 log (k/k₀)_1-AdS+_Me2, where k₀ is the rate of solvolysis in 80% ethanol [15]), corrected for the dispersion observed in Grunwald-Winstein plots for substrates with aromatic rings which could enter into conjugation with the developing positive charge at the α- carbon [15, 23]. These authors have thus proposed the inclusion of this new term in eq. (2) to comprise those situations for which there was cationic charge delocalization by resonance:

I measures the ability of the solvent to stabilize the delocalized cationic charge and h represents the substrate’s sensitivity to changes in the aromatic ring parameter, I, and should be related to the degree of cationic charge delocalization into the ring. The greater the degree of charge delocalization, the more stable the carbocation and therefore the larger the magnitude of h [22].

It was often observed that a significant hI contribution was usually accompanied by a negligible lN contribution and eq. (3) appears sometimes in the following truncated form [23, 24]:

Kevill et al. have illustrated rather profusely the statistical improvements in the G-W correlations for the solvolyses of several (secondary and) tertiary benzylic derivatives as a result of the inclusion of the hI term [15, 25]. However, these correlations are, in general, statistically comparable to those obtained for the same substrates by an alternative approach using appropriate similarity model scales based on the solvolysis of benzylic derivatives (Y_BnX, Y_BnOTs), as proposed by Liu and co-workers [26]. In fact, both authors have put forward a set of serious arguments to demonstrate the goodness of their methodologies. It is not our intention to enter this discussion but rather call the attention for a few other puzzling results, in the same line as those obtained by Kevill’s group when they showed, rather unexpectedly, that the best correlations obtained for the solvolysis of 4-chloro-2,2,4,6,6-pentamethyl-heptane and 3,3-dimethyl-1-neopentylbutyl mesylate included significant I contributions, even though these compounds have no aromatic rings or other conjugate multiple bonds [24]. These authors justify these apparent awkward results by saying that “for lack of a better term, what (they) label as the aromatic ring parameter could well be governed to some degree by perturbations due to variations in ion-pair return”, since, as they claim, it is difficult to distinguish between ion-pair return and aromatic ring solvation effects [23]. Bentley et al. [27], further state that predictions of solvent effects on solvolytic reactions can be attained through eq. 5, if one makes the appropriate choice of the similarity model to use and that this approach accounts for solvation effects adjacent to the reaction center in systems involving alkyl, alkenyl, alkynyl and aryl groups in various aqueous binary mixtures.

However, these authors emphasize that the reliability of these predictions depend in a great extent on the choice of the model and that therefore the proliferation of Y scales is not a minor problem.

The motivation for the research here presented came from our interest on the solvolysis reactions of tertiary alkyl halides [3, 28]. In the context of our systematic work on solvent effects, we came across the need to isolate and rationalize certain solvent contributions and for that reason we had to select some particular substrates. Among the various LSER equations available in the literature, the methodology used within our group to evaluate the solvent’s involvement in these reactions has been the TAKA model equation [29]. However, the kinetics of the newly addressed substrates turned out to be extremely slow in some pure solvents preventing the use of this approach in an appropriate set of solvents, both in number and diversity. An alternative way to obtain rate constants in a reasonable time scale and also some information on the processes appeared to us to be the use of solvent mixtures and Grunwald-Winstein plots, which we have shown in previous work [30] to have unsuspected affinities with the TAKA approach.

Hence, in the present work the Grunwald-Winstein extended equation, eq. (3), or one of its truncated versions, was used to probe the influence of carbon-carbon multiple bonds associated to tertiary carbons on the reactivity of a number of alkyl halides, namely, 3-chloro-3-methylbutyne (1), 2-chloro-2-phenylpropane (2), 2-bromo-2-methyl-1-phenylpropane (3), and also 4-chloro-4-methyl-2-pentyne (4) [6, 27].

For comparative purposes, we have also applied eq. 3 to a substrate already studied by Liu et al. [20] for which there is no possibility of charge delocalization, namely 2-chloro-2-methylbutane (5).

Rate constants for the solvolyses of substrates 1 to 3 were determined at 25.00 °C in several pure and mixed solvent mixtures (Table 1). Table 1 also presents rate constants from literature for the solvolyses of substrates 4 [6, 27] and 5 [20], and of substrate 2 in 80% acetone/water [31] and 3 in 40% trifluoroethanol/ethanol [32], as well as reported Y_Br, Y_Cl, N_OTS and I values for each solvent (for references, see footnote on Table 1).

2-Chloro-2phenylpropane and 2-bromo-2-methyl-1-phenylpropane were synthesized by standard methods [3, 34]. 3-Chloro-3-methylbutyne was obtained commercially. The three compounds were purified by column chromatography (Silica gel 60) and their purity was subsequently assessed by NMR (¹H-NMR spectra were recorded at 400 MHz in CDCl₃). The substrate concentration used in the kinetic experiments was 0.01 mol dm⁻³. Solvents were obtained commercially from Aldrich and Riedel-de Häen with purities above 98%. Kinetic measurements were made at 25.00 °C using an automated conductance bridge. Temperature control was always better than 0.01 °C. Reactions were followed to at least 90% of the apparent plateau. Conductance values were assumed to be linearly related to the acid concentration formed during the reaction. k values were derived using an Excel spreadsheet previously designed for this purpose [35]. Mean k values resulted from at least three different runs and showed a standard deviation better than 4%.

In this paper we report a comparative study of the influence of carbon-carbon multiple bonds on the solvolyses of 5 tertiary alkyl halides by using the Grunwald-Winstein extended equation. A careful correlation analysis was undertaken which lead to some unexpected results, that is, a negative coefficient associated to the “aromatic ring parameter”, I, for substrates without possibility of any charge delocalization, namely substrates, 1, 3 and 5. The reason for this is claimed to be related to the possibility of a non-negligible N_OTs character in the I parameter, due to the absence of any correction for differences in N_OTs in the two model compounds used to establish the I scale. On the other hand, the best correlations found for substrates 2 and 4 show that the solvent nucleophilicity contribution, as measured by N_OTs, is not statistically significant, in line with the increase in steric hindrance and/or charge stabilization in these compounds. Also, the analysis of c values gives evidence for a different behavior of the phenyl group in substrates 2 and 3. While in 3 the ring is acting only as an electron-withdrawing group, diminishing the intrinsic reactivity of the substrate, in 2 the stabilization of the positive charge through resonance is the prevailing effect, leading therefore to a higher c value.

Further studies of solvent effects on the reactions of other tertiary substrates with several degrees of branching and involving different types of carbon-carbon multiple bonds will certainly contribute to shed more light onto the understanding of the true role of the various solvent parameters involved in these G-W equations.