Solid–Liquid Phase Equilibrium Study of 2,5-Furandicarboxylic Acid in 14 Pure Solvents: Dissolution Mechanism and Solubility Prediction

Abstract Although 2,5-furandicarboxylic acid (FDCA) is a key bio-based platform chemical, its solid–liquid phase equilibrium behavior remains insufficiently characterized, limiting rational crystallization process development. Here, FDCA solubility in 14 pure solvents was measured by a dynamic method over 283.15–323.15 K. Molecular simulations were subsequently used to investigate the molecular basis of its dissolution behavior. The results indicate that strong hydrogen-bond acceptor ability, favorable solvation energy, and a small polarity difference promote FDCA dissolution. More importantly, a quality-guided TabPFN workflow for small-sample solubility prediction is proposed, which can predict the solubility of FDCA in unknown solvents with limited data. Compared with three traditional machine-learning models, this workflow demonstrates lower mean squared error (0.03), mean absolute error (0.16), and higher R2 (0.88) on the external validation set. These results provide mechanistic insight and practical guidance for the FDCA crystallization process design.