Solvatochromic linear solvation energy relationships for gas-liquid partition coefficients

Linear solvation energy relationships for gas-liquid partition coefficients have been reevaluated and reinterpreted in light of a new set of data for six prototypical solutes in many common solvents. Multiple linear regression equations are developed based on the π*, α, and β solvatochromic parameters describing the solvent polarizability/dipolarity, hydrogen bond acidity, and hydrogen bond basicity, respectively. The coefficients that arise from these equations are correlated with the π*, a, and β values for each of the solutes. The use of an adaptive Kalman filter, an algorithm that can detect inconsistencies in linear regression models, indicates a hitherto overlooked contribution to the gas-liquid partition coefficients from excesssolute-induced solvent reorganization, relative to that caused by an alkane solute. Inclusion of a solvent reorganization parameter yields regression coefficients that correlate with known solute properties. These results show that previous approaches for linear solvation energy correlations are incomplete due to the absence of this reorganization term. Even after introducing this new parameter, the quality of the regression equations is still worse than the anticipated random experimental error in the measured partition coefficients. Reasons for this discrepancy are discussed.