The High-Value Thermoelectric Properties of Ruby-Silver Ores Ag3XS3 (X = As, Sb): A Combined Density Functional Theory and Machine Learning Approach
ACS Applied Energy Materials · 2026-07-27 · 9 (14) · 9255-9265
DOI: 10.1021/acsaem.6c01192 ↗At a glance
Journal article in ACS Applied Energy Materials, published online on 2026-07-16. Written by Alfredo Ranes Coro III and 7 co-authors. 53 references. The full text is freely available (open access, CC BY 4.0). Funded by: National Center for Theoretical Sciences, Academia Sinica, National Science and Technology Council.
Abstract (original language)
Silver thio-compounds, Ag3AsS3 and Ag3SbS3, have emerged as highly promising candidates for thermoelectric energy conversion due to their intrinsically low thermal conductivity and favorable electronic transport properties. In this work, we present a comprehensive study of the thermoelectric performance of Ag3XS3 (X = As, Sb) using first-principles calculations and machine learning interatomic potentials (MLIPs). By training the MLIPs, we performed large-scale molecular dynamics simulations with density functional theory-level accuracy, enabling a rigorous assessment of the full lattice anharmonicity and its impact on thermal transport. Our results reveal that the inclusion of fourth-order phonon-phonon interactions is essential for accurately capturing the temperature dependence of lattice thermal conductivity, which exhibits a crossover behavior between the As- and Sb-based systems at elevated temperatures. We identify a transition from particle-like to wave-like transport, where a significant fraction of phonon modes enters the overdamped regime, surpassing the Ioffe-Regel limit. In this regime, heat is primarily conducted through coherent vibrational modes, with the wave-like contribution accounting for over 95% of the total lattice thermal conductivity. On the electronic side, the materials exhibit high power factors attributed to the large density-of-states effective masses of the carrier states, which sustain a high Seebeck coefficient even at high carrier concentrations. The synergistic effect of ultralow, glass-like thermal conductivity and optimized electronic transport leads to exceptional dimensionless figures of merit zT. Specifically, we predict a peak n-type zT of 3.91 at 500 K for Ag3SbS3 and p-type zT of 3.37 at 300 K for Ag3AsS3. These findings not only position Ag3XS3 as the premier candidates for efficient waste heat recovery but also demonstrate the transformative potential of MLIP-driven large-scale simulations in the discovery of next-generation energy materials.
Funding
- National Center for Theoretical Sciences
- Academia Sinica
- National Science and Technology Council
Cite
APA
III, A. R. C., You, H., Villaos, R. A. B., Huang, Z., Yao, L., Liang, G., Lin, H., & Chuang, F. (2026). The High-Value Thermoelectric Properties of Ruby-Silver Ores Ag3XS3 (X = As, Sb): A Combined Density Functional Theory and Machine Learning Approach. ACS Applied Energy Materials, 9(14), 9255-9265. https://doi.org/10.1021/acsaem.6c01192
Vancouver
III ARC, You H, Villaos RAB, Huang Z, Yao L, Liang G, et al. The High-Value Thermoelectric Properties of Ruby-Silver Ores Ag3XS3 (X = As, Sb): A Combined Density Functional Theory and Machine Learning Approach. ACS Applied Energy Materials. 2026;9(14):9255-9265. doi:10.1021/acsaem.6c01192
BibTeX
@article{III2026,
author = {III, Alfredo Ranes Coro and You, Hao-Jen and Villaos, Rovi Angelo Beloya and Huang, Zhi-Quan and Yao, Liang-Zi and Liang, Gengchiau and Lin, Hsin and Chuang, Feng-Chuan},
title = {The High-Value Thermoelectric Properties of Ruby-Silver Ores Ag3XS3 (X = As, Sb): A Combined Density Functional Theory and Machine Learning Approach},
journal = {ACS Applied Energy Materials},
year = {2026},
volume = {9},
number = {14},
pages = {9255--9265},
doi = {10.1021/acsaem.6c01192}
}
EUVIMED