(126c) Calculation of the Solubility of Pharmaceuticals in Pure and Mixed Solvents

The accurate knowledge of the solubility of pharmaceutical molecules in pure and mixed solvents is of outmost importance for the relevant process design (i.e., crystallization) and product formulation. Standard thermodynamic models developed primarily for the oil & gas and chemical structure often result in inaccurate predictions when applied to pharmaceutical molecules, primarily due to the significant complexity of the latter. In this work, an integrated approach is proposed that combines molecular descriptors with an equation of state with a strong theoretical basis. Molecular descriptors derived from the molecular structure of pharmaceutical molecules are used for the calculation of the pharmaceutical solubility in water (log S) and of the octanol / water partition coefficient (log P). Calculations for a wide range of pharmaceuticals (on the order of several hundreds) that vary in functionality and molecular structure result in root-mean-square deviation from experimental data of 0.44.

The solubilities obtained in pure solvents are used to tune interaction parameters in the Perturbed Chain-Statistical Associating Fluid Theory (PC-SAFT). PC-SAFT is used subsequently for the prediction of pharmaceutical solubility in a broad temperature range and for mixtures of solvents. Comparison with experimental data reveals that the approach provides reliable results.