Explainable Machine Learning for Oxygen Diffusion in Perovskites and Pyrochlores

2025-05-16

AI Summary

This study investigates the physical mechanisms governing oxygen diffusion activation energy (Eₐ) in perovskites and pyrochlores. Leveraging an experimental Eₐ database comprising over 100 materials, we employ feature grouping engineering, a seven-model ensemble learning framework, and SHAP-based interpretability analysis. We discover that weighted-average metallic properties—not conventional binary oxide descriptors—are the most predictive features for Eₐ, thereby challenging established descriptor paradigms. Further, we identify A-site ionicity and oxygen partial pressure as the dominant regulators in perovskites, and A-site s-electron count and B-site electronegativity as the key determinants in pyrochlores. These findings provide both a physically interpretable machine learning framework and fundamental structure–property insights, enabling high-throughput design of advanced oxygen-ion conductors.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Knowledge Representation and Reasoning: Description Logics
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Explainable machine learning can help to discover new physical relationships for material properties. To understand the material properties that govern the activation energy for oxygen diffusion in perovskites and pyrochlores, we build a database of experimental activation energies and apply a grouping algorithm to the material property features. These features are then used to fit seven different machine learning models. An ensemble consensus determines that the most important features for predicting the activation energy are the ionicity of the A-site bond and the partial pressure of oxygen for perovskites. For pyrochlores, the two most important features are the A-site $s$ valence electron count and the B-site electronegativity. The most important features are all constructed using the weighted averages of elemental metal properties, despite weighted averages of the constituent binary oxides being included in our feature set. This is surprising because the material properties of the constituent oxides are more similar to the experimentally measured properties of perovskites and pyrochlores than the features of the metals that are chosen. The easy-to-measure features identified in this work enable rapid screening for new materials with fast oxide-ion diffusivity.

Problem

• Identify key material properties affecting oxygen diffusion activation energy
• Apply explainable ML to perovskites and pyrochlores for feature importance
• Enable rapid screening of materials with high oxide-ion diffusivity

Innovation

• Explainable machine learning identifies key material features
• Grouping algorithm analyzes experimental activation energies
• Ensemble consensus determines most predictive features