Lattice-to-total thermal conductivity ratio: A phonon-glass electron-crystal descriptor for data-driven thermoelectric design

Thermoelectrics (TEs) are promising candidates for energy harvesting with performance quantified by figure of merit, Z T . To accelerate the discovery of high- Z T materials, efforts have focused on identifying compounds with low thermal conductivity κ . Using a curated dataset of 71,913 entries, we show that high- Z T materials reside not only in the low- κ regime but also cluster near a lattice-to-total thermal conductivity ratio ( κ L / κ ) of approximately 0.5. This empirically derived ratio provides a quantitative descriptor for the well-known phonon-glass electron-crystal (PGEC) design concept. Building on this insight, we construct a framework consisting of two machine learning models for the lattice and electronic components of thermal conductivity that jointly provide both κ and κ L / κ for screening and guiding the optimization of TE materials. By applying this framework to 104,567 inorganic compounds, we identify 2522 ultralow- κ candidates while simultaneously evaluating their proximity to the PGEC regime. A follow-up case study on chemical doping demonstrates how the framework can qualitatively provide optimization strategies that shift pristine materials toward the empirical design target of κ L / κ ≈ 0.5 while maintaining their low total κ . Ultimately, by integrating rapid screening with PGEC-guided optimization, our data-driven framework takes a critical step toward closing the gap between materials discovery and performance enhancement.