Collection of Czechoslovak Chemical Communications - digital archive 

Abstract

The paper presents a classification of 51 solvents based on clustering in three-dimensional space formed by the empirical scale of PAC, PBC, and PPC parameters designed for interpretation of solvent effect on a model with cross-terms. For the classification used are the clustering methods of the nearest neighbour, of the furthest neighbour, of average bond, and the centroid method. As a result, the solvents have been divided into 8 classes denoted as: I - nonpolar-inert solvents (aliphatic hydrocarbons), IIp - nonpolar-polarizable (aromatic hydrocarbons, tetrachloromethane, carbon disulphide), IIb - nonpolar-basic (ethers, triethylamine), IIIp - little polar-polarizable (aliphatic halogen derivatives, substituted benzenes with heteroatom-containing substituents), IIIb - little polar-basic (cyclic ethers, ketones, esters, pyridine), IVa - polar-aprotic (acetanhydride, dialkylamides, acetonitrile, nitromethane, dimethyl sulfoxide, sulfolane), IVp - polar-protic (alcohols, acetic acid), and V - exceptional solvents (water, formamide, glycol, hexamethylphosphoric triamide). The information content of the individual parameters used for the classification has been determined. The classification is based primarily on solvent polarity/acidity (PAC), less on polarity/basicity (PBC), and the least on polarity/polarizability (PPC). Causal relation between chemical structure of solvent and its effect on the process taking place therein has been established.