Solvent Parameters from Computational Chemistry: A Possible Alternative to Experimental Solvent Parameters?

Abstract Previous studies have shown that solvent parameters strongly correlate with molecular properties derived from computational chemistry. Thus, for example, Kamlet and Taft’s β , Gutmann’s donor number and Catalan’s SB show strong correlations with the partial charge on the most negative atom and the energy of the associated orbital only. These consistencies suggest the possibility that solvent parameters calculated from computer-generated molecular properties could be used, in place of experimental parameters, in linear free-energy-type relationships. In this study, the solubilities of thioxanthen-9-one in 60 mono solvents were correlated against experimental and calculated solvent parameters. It was found that the standard deviations between experimental and back-calculated solubilities were comparable. Moreover, the correlations against calculated parameters highlighted the fact that the solubilities in aromatic solvents and cyclic ethers are around seven times those predicted from the correlation against solubilities in aliphatic solvents.