LLM-Assisted Design of Ultralow- κ Chalcogenides

Accelerating the discovery of materials with extreme lattice thermal conductivity ( $\kappa$ ) is limited by the high cost of first-principles screening. Here, we develop an LLM-assisted low- $\kappa$ discovery workflow, where Bayesian optimization (BO) proposes candidates, an LLM performs evidence-conditioned re-ranking, and physics-based structure generation, stability validation, and $\kappa$ calculations serve as the acceptance criteria. This integration enables efficient candidate prioritization while maintaining rigorous physical validation. The workflow identifies dynamically stable Ag-based chalcogenides, including Ag₃SbSe₄, Ag₄TiSe₄, and Ag₄TiTe₄. Mechanistic analysis reveals that their ultralow $\kappa$ originates from strong phonon renormalization and coherence-dominated heat transport. Ag₃SbSe₄ is predicted to reach a maximum $ZT$ of 2.94 at 500 K due to its ultralow $\kappa$ of 0.11 W $m^{-1}$ $K^{-1}$ . Furthermore, automated high-throughput screening of ternary compounds containing 14 elements identifies 2742 dynamically stable materials with $\kappa \le 1.0$ W $m^{-1}$ $K^{-1}$ . These results demonstrate that evidence-conditioned LLM re-ranking effectively enhances BO-driven, physics-validated materials discovery.