Unraveling Thermal Transport Correlated with Atomistic Structures in Amorphous Gallium Oxide via Machine Learning Combined with Experiments

Aug 5, 2026

Upload by Patrick

Thermal transport properties of amorphous materials are essential for energy and electronic devices, yet understanding heat conduction in disordered systems is difficult due to computational limits and the scarcity of physically intuitive structural descriptors. This work integrates machine learning interatomic potentials with experimental evidence to model amorphous gallium oxide accurately using a self-guided, minimum-quantum-computation strategy. Atomistic simulations link density-driven short- and medium-range order changes to reduced localization modes and improved coherence contribution to heat transport. A physics-inspired descriptor enables linear prediction of thermal conductivity from structure, supporting accelerated discovery of mechanisms in disordered functional materials.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages27

Reading Time1 h 8 min

AccessFree to download

Introduction

Thermal transport importance in amorphous materials

Heat conduction frameworks and theoretical formalisms

Challenges in realistic modeling of disordered systems

Emergence of machine learning interatomic potentials

Unraveling Thermal Transport Correlated with Atomistic Structures in Amorphous Gallium Oxide via Machine Learning Combined with Experiments Yuanbin Liu 1, Huili Liang2,3, Lei Yang 1, Guang Yang 1, Hongao Yang 1, Shuang Song3, Zengxia Mei2,3, Gábor Csányi4,*, and Bingyang Cao1,* 1Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China 2Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China 3Songshan Lake Materials Laboratory, Dongguan, Guangdong 523808, China 4Engineering Laboratory, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, United Kingdom *[gc121@cam.ac.uk](gc121@cam.ac.uk);*[caoby@tsinghua.edu.cn](caoby@tsinghua.edu.cn); Abstract: Thermal transport properties of amorphous materials are crucial for their emerging applications in energy and electronic devices. However, understanding and controlling thermal transport in disordered materials remains an outstanding challenge, owing to the intrinsic limitations of computational techniques and the lack of physically-intuitive descriptors for complex atomistic structures. Here, we show how combining machine learning-based models and experimental observations can help to accurately describe realistic structures, thermal transport properties, and structure–property maps for disordered materials, which is illustrated by a practical application on gallium oxide. First, we report the experimental evidence to demonstrate that machine learning interatomic potentials, generated in a self-guided fashion with minimum quantum-mechanical computations, enable the accurate modeling of amorphous gallium oxide and its thermal transport properties. Our atomistic simulations then reveal the microscopic changes in the short-range and medium-range order with density and elucidate how these changes can reduce localization modes and enhance coherences’ contribution to heat transport. Finally, a physicsinspired structural descriptor for disordered phases is proposed, with which the underlying relationship between structures and thermal conductivities is predicted in a linear form. This work might shed light on the future accelerated exploration of novel thermal transport properties and mechanisms in disordered functional materials. 1. Introduction Thermal transport properties of amorphous materials are crucial for their emerging applications in thermoelectric devices[1], phase-change memory devices[2], ﬂexible electronics[3], radiation detectors[4], artificial intelligence chips[5], thermal barrier coatings[6], and batteries[7] . For instance, low vibrational thermal conductivity (􀁎) in amorphous solids is advantageous to improve heat-toelectricity conversion efficiency in thermoelectric applications or sensitivity in gamma-ray detectors, but on the contrary, it may lead to serious heat dissipation problems for next-generation electronics or batteries where a large amount of Joule heat is generated. Clearly, these various technological aspects have triggered immense interest in accurately describing, understanding, and ultimately controlling thermal transport in amorphous materials. Heat in nonmetallic crystals is mainly carried by phonons and thermal resistance arises from phonon scattering, while the absence of periodicity makes thermal behaviors in amorphous solids quite different by the strong localization of vibrational modes and the suppression of vibrational scattering length scales. To date, tremendous progress has been made in the theoretical formalism for describing heat conduction in disordered phases by explicitly considering off-diagonal terms of the heat current operator[8] . The landmark theoretical frameworks include the Allen-Feldman (AF) theory[9] and the more recently developed unified theory[8] as well as the quasi-harmonic GreenKubo method[10] . Their accuracy, thanks to the development of powerful computer simulation

FAQs

What challenge does the paper address in studying thermal transport in amorphous materials?

It targets the difficulty of accurately understanding and controlling heat conduction in disordered systems, driven by computational constraints and the lack of physically intuitive descriptors for complex atomistic structures.

How are machine learning interatomic potentials used in the study?

Machine learning-based models are combined with experiments to create interatomic potentials in a self-guided fashion using minimum quantum-mechanical computations, enabling accurate modeling of amorphous gallium oxide.

How does the work connect atomic structure to thermal conductivity?

Simulations reveal how density-related changes in short- and medium-range order affect localization and coherence contributions. A physics-inspired structural descriptor then predicts thermal conductivities in a linear form based on structure.

Aug 5, 2026

Upload by Patrick