Balázs Póta^1*, Paramvir Ahlawat¹, Gábor Csányi², Michele Simoncelli^1*

1 Theory of Condensed Matter Group of the Cavendish Laboratory, University of Cambridge, Cambridge, UK

2 Engineering Laboratory, University of Cambridge, Cambridge, UK

* Corresponding authors emails: bp443@cam.ac.uk, ms2855@cam.ac.uk

DOI10.24435/materialscloud:2d-4b [version v1]

Publication date: Oct 22, 2024

How to cite this record

Balázs Póta, Paramvir Ahlawat, Gábor Csányi, Michele Simoncelli, Thermal conductivity predictions with foundation atomistic models, Materials Cloud Archive 2024.171 (2024), https://doi.org/10.24435/materialscloud:2d-4b

Description

Advances in machine learning have led to the development of foundation models for atomistic materials chemistry, enabling quantum-accurate descriptions of interatomic forces across diverse compounds at reduced computational cost. Hitherto, these models have been benchmarked relying on descriptors based on atoms' interaction energies or harmonic vibrations; their accuracy and efficiency in predicting observable and technologically relevant heat-conduction properties remains unknown. Here, we introduce a framework that leverages foundation models and the Wigner formulation of heat transport to overcome the major bottlenecks of current methods for designing heat-management materials: high cost, limited transferability, or lack of physics awareness. We present the standards needed to achieve first-principles accuracy in conductivity predictions through model's fine-tuning, discussing benchmark metrics and precision/cost trade-offs. We apply our framework to a database of solids with diverse compositions and structures, demonstrating its potential to discover materials for next-gen technologies ranging from thermal insulation to neuromorphic computing.

Materials Cloud sections using this data

Files

File name	Size	Description
data.zip MD5md5:b55af517e53a86693990ab00f783d3e4	359.8 MiB	Dataset containing reference DFT thermal conductivity results, as well as scripts and data used for fine-tuning.

License

Files and data are licensed under the terms of the following license: Academic Software Licence ("ASL").
Metadata, except for email addresses, are licensed under the Creative Commons Attribution Share-Alike 4.0 International license.

External references

Keywords

thermal conductivity foundation machine learning potential wigner transport equation interatomic potential fine-tuning matbench-discovery