ab initio solubility prediction of non-electrolytes in ternary solvents using a combination of Jouyban-Acree and Abraham models

Abstract

Applicability of a previously trained model derived from a combination of the Jouyban-Acree and Abraham models for predicting the solubility of non-electrolytes in ternary solvent mixtures was shown using 80 data sets of anthracene and pyrene. A sample program written in SPSS environment was presented in Appendix. The accuracy criterion was mean percentage deviation (MPD) and using experimental solubility data in neat solvents, i.e. three data points for each set, the overall MPDs (+/- SD) were 7.5 (+/- 5.5)% and 20.2 (+/- 13.0)%, respectively for water to solvent and gas to solvent parameter of Abraham model. The corresponding MPDs for full predictive versions of the models were 27.5 (+/- 18.2) % and 20.8 (+/- 19.5) %. The results showed that one can use the, proposed method to predict solubility of non-electrolytes in ternary solvents and the expected prediction errors are quite acceptable specially with employing experimental solubility of solutes in neat solvents as input data.