Correlations and predictions of solvent effects on reactivity: some limitations of multi-parameter equations and comparisons with similarity models based on one solvent parameter

Three recent publications on multi-parameter correlations of solvent effects on solvolytic reactivity are re-examined, by considering 'similarity' and/or 'analogy'. Systematic errors due to compensation effects and to comparisons between dissimilar processes are found. Models for solvent nucleophilicity involving dissimilar spectroscopic processes (e.g. or B parameters) give insensitive measures of low nucleophilicity. From qualitative considerations based on structural similarities, it is predicted that the sensitivities to changes to solvent polarity for solvolyses of chloroalkanes should be in the order: I-adamantyl (3) > 2-methyl-2-adamantyl (1) > t-butyl (2). The predictions are confirmed quantitatively by simple linear free-energy relationships and similarity models, involving correlations with Y-CI (based on solvolyses of I-chloroadamantane) or E-T(30) (based on solvatochromism). Multiparameter correlations, indicating that solvolyses of 1 show a low sensitivity to both solvent polarity and electrophilicity, and also a negative sensitivity to solvent nucleophilicity, are shown to be unreliable. Large errors are also evident in recent KOMPH2 calculations. Conclusions are supported by comparing several multi-parameter treatments of solvolyses of 4-methoxyneophyl tosylate, for which there is a reliable set of kinetic data and a generally accepted mechanism. Copyright (c) 2006 John Wiley & Sons, Ltd.