Materials informatics across the length scales

Le résumé fourni par la source

Materials informatics is increasingly used to support modelling, analysis and design across the length scales of materials science, from atomistic simulations to mesoscale evolution, microstructural characterisation and continuum-level property prediction. Its impact, however, differs substantially between scales because the data, descriptors, governing physics and validation requirements change from one regime to another. Nanoscale problems require models that can treat local atomic environments, defects, interfaces, long-range interactions and uncertainty in atomically resolved data. Mesoscale problems often involve evolving microstructures, phase fields, coarse-grained descriptions and expensive simulations that motivate surrogate, reduced-order and operator-learning approaches. At the micro-to-continuum scale, the emphasis shifts towards extracting reliable features from microscopy and related characterisation data, and using them for segmentation, classification and structure–property prediction. This Review examines these data-driven methods across the three scales, while distinguishing capabilities that are already reliable from limitations that still restrict broader use, including data scarcity, inconsistent metadata, weak uncertainty quantification, limited interpretability and incomplete cross-scale consistency.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.