Chemical filters for ultra-high-throughput materials screening and generation

2026-07-20

AI Summary

This work addresses the frequent generation of chemically infeasible compositions in generative materials design—structures that violate oxidation state rules, thereby undermining reliability and interpretability. The authors propose a configurable oxidation-state-aware algorithmic prior that, for the first time, encodes heuristic chemical rules into tunable operators. This framework supports both post-hoc filtering and integration as a reward signal in reinforcement learning to guide the generative process. Built upon the open-source SMACT toolkit, the approach employs a data-driven oxidation state model coupled with an adjustable threshold mechanism, seamlessly integrating into latent-space diffusion and reinforcement learning architectures. Experiments across six state-of-the-art generative models demonstrate that the method effectively filters out candidates relying on rare oxidation states while preserving low-energy, convex-hull-proximal stable compounds, substantially enhancing the chemical plausibility of generated compositions.

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.

Problem

• chemical validity
• generative materials design
• oxidation states
• chemical constraints
• materials screening

Innovation

• chemical validity operator
• oxidation-state-aware
• generative materials design
• algorithmic prior
• latent diffusion model