Machine Learning-Enhanced First Principles Inverse Design of Anion Exchange Membrane Morphology and Performance

Alkaline anion exchange membranes (AEMs) must balance ionic conductivity and mechanical strength, yet optimizing both typically requires extensive experimentation. We introduce a first-principles, multiscale inverse-design framework that connects atomic interactions to macroscopic performance. Starting from topology optimization of membrane morphology, we identify the structures that simultaneously maximize conductivity and mechanical robustness. Machine-learning exploration of the Flory–Huggins (χ)–block fraction parameter space, guided by self-consistent field theory (SCFT), then reveals which χ values and block fractions favor these optimal morphologies. Finally, density-functional theory and COSMO-RS calculations determine molecular chemistries that realize the target χ parameters without empirical fitting. Applied to ABC triblock terpolymer AEMs, the framework reproduces the experimental preference for gyroid over lamellar morphologies and uncovers new design regions with superior predicted performance. By coupling topology optimization, machine learning, and first-principles modeling across scales, this approach enables rapid in silico discovery of AEMs optimized for ion transport and mechanical integrity.