Element mapping-based Bayesian optimization framework enabling direct materials design: a case study on NASICON-type cathode materials

Bayesian optimization (BO) helps in efficiently navigating complex and high-dimensional design spaces. Recently, it has been applied to materials science to discover novel materials with high performances. However, the application of BO to material design has been hindered by the challenges in handling discrete input variables, such as elements. This study introduces a novel element mapping strategy that encodes elemental identities into chemically meaningful continuous values, enabling the creation of easy-to-predict chemical spaces. This new framework is used to design high-capacity Na₃V₂(PO₄)₂F₃ cathode materials for sodium-ion batteries, aiming to shift all working voltages into the desired operational voltage window (2.5–4.3 V). The proposed framework successfully suggested 16 optimal compositions within 50 iterations. The proposed approach can overcome the limitation of categorical input and broaden the applicability of BO to a wider range of material discoveries.