Coupling Language Models with Physics-based Simulation for Synthesis of Inorganic Materials

2026-05-29

AI Summary

This work addresses the challenge that generative models for inorganic crystalline materials often propose candidates lacking feasible synthesis pathways. To overcome this limitation, the authors introduce a hybrid framework that integrates the implicit prior knowledge of large language models (LLMs) with physics-based simulations. By incorporating thermodynamic databases, simplified kinetic models, and classical path-planning algorithms, the approach leverages LLMs—used here for the first time in inorganic materials synthesis planning—to transcend the constraints of conventional search strategies. Validation on the niobium–oxygen system demonstrates that synthesis routes generated by the LLM significantly outperform those from classical algorithms, highlighting the method’s viability and advantages for complex materials synthesis tasks.

Abstract

Modern generative machine learning (ML) models can propose novel inorganic crystalline materials with targeted properties; however, synthesis planning of these materials remains difficult due to the complexity of the associated physical processes and limited availability of computational tools. We introduce a novel hybrid framework to evaluate Large Language Models (LLMs) in inorganic synthesis planning by combining thermodynamic databases with simplified kinetics models to approximate realistic synthesis conditions. As a case study, we focus on the niobium-oxygen system, which features multiple industrially relevant oxide phases with well-characterized data. In computational simulations, we compare LLM-generated synthesis routes with classical path-planning algorithms, showing that the implicit priors in LLMs can yield more viable strategies. In our evaluation setting, classical search methods serve primarily as a foil rather than a direct competitor. This illustrates the relative complexity of the problem and highlights where the LLM's implicit priors add value.

Problem

• inorganic materials synthesis
• synthesis planning
• physics-based simulation
• Large Language Models
• thermodynamic databases

Innovation

• Large Language Models
• physics-based simulation
• inorganic synthesis planning
• thermodynamic databases
• kinetics modeling