Oxide state aware ai generates more realistic crystal materials
OxiGen: Oxidation-State-Aware Crystal Generation
Machine Learning
Summary
Making new crystal materials in the lab is tricky because the materials need to balance electrical charges properly. The authors made a new AI model called OxiGen that pays close attention to these electrical charges, called oxidation states, while designing crystals. This approach helps the AI create crystals that are more stable, unique, and closer to real materials. This could speed up discovering new materials with useful properties by making the designs more realistic.
What this means in practice
- •For materials engineers: Design new inorganic crystal materials with improved chemical validity and stability using AI-generated oxidation-state-aware models.
- •For chemical manufacturing teams: Develop materials with predictable charge balance for more reliable synthesis outcomes guided by AI-driven crystal generation.
- •For battery technology developers: Create novel crystal materials with stable oxidation states to enhance battery component performance and durability.$Commercial implications: Enables development of better battery materials by generating realistic, stable crystal structures that manufacturers could commercialize.
Authors
Dylan John, Kim E. Jelfs, Alex M. Ganose, Eleonora Giunchiglia
Abstract
Generative models have the potential to accelerate inorganic materials discovery by enabling inverse design, but generating experimentally realisable crystals remains challenging. Oxidation states are widely used to assess the compositional validity of crystals and guide inorganic materials discovery. While existing generative models for crystals can generate materials with charge-neutral oxidation-state assignments, they poorly reproduce the distributions of oxidation states observed in synthesised materials. To address this limitation, we propose OxiGen, an oxidation-state-aware crystal diffusion model that explicitly represents oxidation states during generation. OxiGen enforces global charge neutrality by construction using a structured output layer with exact inference over a finite-state automaton. Empirically, OxiGen substantially improves oxidation-state fidelity, generates the highest rate of stable, unique, and novel crystals among evaluated methods, and maintains high compositional validity even under property conditioning.