Calculation of drug-like molecules solubility using predictive activity coefficient models

Fátima L. Mota (1) , António J. Queimada (1) , Alfonsina E. Andreatta (2) , Simão P. Pinho (1) , Eugénia A. Macedo (1)

1 LSRE - Laboratory of Separation and Reaction Engineering
2 PLAPIQUI - Planta Piloto de Ingeniería Química [Bahía Blanca]

The A-UNIFAC, UNIFAC, and NRTL-SAC models are used to predict solubility in pure solvents of a set of drug-like molecules. To apply A-UNIFAC, a new set of residual interaction parameters between the ACOH group and six other groups had to be estimated. The solute model parameters of NRTL-SAC were also estimated for this set of molecules. NRTL-SAC showed better performance at 298.15 K, with an average absolute deviation of 37.6%. Solubility dependence with temperature was also studied: all models presented average deviations around 40%. In general, there is an improvement given by the A-UNIFAC over the UNIFAC in aqueous systems, proving the importance of taking association into account. The reference solvent approach was also applied improving the results. Solubility in pure solvents can now be predicted with an average deviation around 35.2%. This approach reduces differences previously found between the three models, being a powerful methodology.

Mots clés: UNIFAC; Solubility; A-UNIFAC; Complex chemicals; Modelling; NRTL-SAC

Domaines: Génie chimique

Fátima L. Mota, António J. Queimada, Alfonsina E. Andreatta, Simão P. Pinho, Eugénia A. Macedo. Calculation of drug-like molecules solubility using predictive activity coefficient models. Fluid Phase Equilibria, 2012, 322-323, pp.48 - 55. ⟨10.1016/j.fluid.2012.02.003⟩. ⟨hal-01637119⟩