An Inverse Design Workflow for Solvent Mixtures Based on Hansen Solubility Parameters

Abstract

Abstract We present SolvPred, an open-source Python framework for inverse design of solvent mixtures based on Hansen solubility parameters (HSPs). SolvPred allows users to control the composition of solvent mixtures while setting essential experimental constraints. As a case study, we applied SolvPred to optimize a polyimide condensation as a model reaction, where the hydrogen-bonding HSP term was selectively tuned. SolvPred identified seven viable ternary mixtures from an otherwise intractable space of 171,871 possibilities. The best-performing solvent mixture achieved a fivefold improvement in the product yield at a milder temperature when compared with the conventional single-solvent condition. This work establishes SolvPred as a practical solution for rational solvent mixture design and experimental planning, as well as an efficient tool for property-guided solvent optimization within data-driven experimental workflows.