Chemical filters for ultra-high-throughput materials screening and generation
2026-07-20 • Artificial Intelligence
Artificial IntelligenceMachine Learning
AI summaryⓘ
The authors developed a tool that uses chemical rules to help AI models design new materials more realistically. Their method checks if the materials' chemical compositions make sense, especially regarding oxidation states, which are important for chemical stability. They tested this on several AI models and found that while many get basic formulas right, they often miss realistic oxidation states. Their tool can filter out unlikely materials and even guide AI to design better ones by rewarding chemically sound options. This approach helps make AI-generated materials more trustworthy and easier to understand.
generative artificial intelligencematerials designchemical validityoxidation stateSMACTstoichiometryreinforcement learninglatent diffusion modelconvex hullchemical heuristics
Authors
Kinga O. Mastej, Panyalak Detrattanawichai, Hyunsoo Park, Anthony Onwuli, Masahiro Negishi, Aron Walsh
Abstract
Generative artificial intelligence is rapidly transforming materials design by enabling de novo exploration of immense chemical spaces. Yet a large proportion of AI-generated compositions remain implausible, violating established chemical principles, which limits the reliability and interpretability of generative materials design. Here, we introduce a chemical validity operator that recasts heuristic chemical rules as a configurable algorithmic prior for evaluating and guiding generative materials discovery. Built on the open-source SMACT package, a data-informed oxidation-state model exposes tunable thresholds, allowing users to interpolate continuously between permissive and conservative chemical constraints, while supporting both exploratory and conservative materials-design workflows. Benchmarking six state-of-the-art generative models for inorganic crystals shows that most reproduce stoichiometry but under-represent realistic oxidation-state combinations, and that filtering removes compositions reliant on rarely observed oxidation states while preserving low-energy compounds near the convex hull. Beyond screening, the same operator can also serve as a reinforcement-learning reward, steering a latent diffusion model towards chemically grounded compositions. By encoding chemical heuristics and observations, this work establishes a foundation for oxidation-state-aware generative models.