Topic: Large Language Models and Foundation Models for Materials Informatics and Discovery

Foundation models (FMs), including large language models (LLMs), are opening new opportunities in materials informatics and discovery. Large-scale pretraining and task-specific adaptation provide new opportunities for integrating information from scientific literature, materials databases, computational simulations, and experimental measurements, supported by appropriate data representations and external tools. Recent advances have demonstrated their potential in scientific knowledge extraction, materials representation learning, property prediction, materials design, and the development of AI-assisted research workflows.

This Special Topic welcomes advances in the development, evaluation, and application of LLMs and other FMs for materials research. Topics of interest include scientific knowledge extraction and literature mining; materials representation learning and property prediction; multimodal foundation models; generative foundation models for inverse materials design; retrieval-augmented generation, scientific question answering, and knowledge-grounded reasoning; integration with materials simulations and experimental data; and model-enabled agents for autonomous and closed-loop research workflows. Contributions on materials-specific pretraining and adaptation, data quality, benchmarking, uncertainty quantification, reliability, and interpretability are particularly encouraged. We also welcome applications that assess model predictions through computational or experimental validation and demonstrate their utility in materials discovery and design.

Large language models, foundation models, materials informatics, materials discovery, materials representation learning, multimodal learning, inverse materials design, retrieval-augmented generation, AI agents, autonomous experimentation