Thermodynamic sensitivity and transferability limitations of Pharma Mod UNIFAC for aqueous solubility prediction of sulfonamide APIs

Abstract Predicting the aqueous solubility of active pharmaceutical ingredients (APIs) remains challenging because highly dilute systems are particularly sensitive to activity‐coefficient calculations and solid‐reference thermodynamic properties. In this work, we evaluated Pharma Mod UNIFAC for predicting the aqueous solubility of sulfonamide‐based APIs by combining new solubility and calorimetric measurements with traceable literature data. Before assessing parameter transferability, we verified the molecular representation, solid‐reference properties, and critical water–sulfonamide interaction pairs for each system. When applied directly, the original Pharma Mod UNIFAC parameters reproduced qualitative trends but were not uniformly transferable across the evaluated APIs. Local constant reparameterization of critical water–sulfonamide interactions substantially improved calculated solubilities in several systems; however, the fitted parameters varied among structurally related compounds and should therefore be interpreted as local effective corrections rather than universal replacements for the original database. Temperature‐dependent local regressions showed that different functional forms, such as and , can yield nearly indistinguishable effective trajectories over the temperature interval studied, revealing an identifiability limitation of the fitted coefficients. Directional sensitivity analysis showed that forward and reverse perturbations of the same interaction pair are not equivalent for all APIs. Overall, Pharma Mod UNIFAC remains a useful, structured framework for predicting aqueous solubility, but its application to highly dilute pharmaceutical systems requires careful local validation. The main contribution is a thermodynamic audit of parameter transferability, not a new universal interaction‐parameter matrix.