Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity
Abstract Lattice thermal conductivity is important for many applications, but experimental measurements or first principles calculations including three-phonon and four-phonon scattering are expensive or even unaffordable. Machine learning approaches that can achieve similar accuracy have been a lon...
Main Authors: | , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2023-06-01
|
Series: | npj Computational Materials |
Online Access: | https://doi.org/10.1038/s41524-023-01020-9 |
_version_ | 1797811390247862272 |
---|---|
author | Ziqi Guo Prabudhya Roy Chowdhury Zherui Han Yixuan Sun Dudong Feng Guang Lin Xiulin Ruan |
author_facet | Ziqi Guo Prabudhya Roy Chowdhury Zherui Han Yixuan Sun Dudong Feng Guang Lin Xiulin Ruan |
author_sort | Ziqi Guo |
collection | DOAJ |
description | Abstract Lattice thermal conductivity is important for many applications, but experimental measurements or first principles calculations including three-phonon and four-phonon scattering are expensive or even unaffordable. Machine learning approaches that can achieve similar accuracy have been a long-standing open question. Despite recent progress, machine learning models using structural information as descriptors fall short of experimental or first principles accuracy. This study presents a machine learning approach that predicts phonon scattering rates and thermal conductivity with experimental and first principles accuracy. The success of our approach is enabled by mitigating computational challenges associated with the high skewness of phonon scattering rates and their complex contributions to the total thermal resistance. Transfer learning between different orders of phonon scattering can further improve the model performance. Our surrogates offer up to two orders of magnitude acceleration compared to first principles calculations and would enable large-scale thermal transport informatics. |
first_indexed | 2024-03-13T07:21:55Z |
format | Article |
id | doaj.art-20a91f4c7dec49ab81e2236d4aeac064 |
institution | Directory Open Access Journal |
issn | 2057-3960 |
language | English |
last_indexed | 2024-03-13T07:21:55Z |
publishDate | 2023-06-01 |
publisher | Nature Portfolio |
record_format | Article |
series | npj Computational Materials |
spelling | doaj.art-20a91f4c7dec49ab81e2236d4aeac0642023-06-04T11:34:08ZengNature Portfolionpj Computational Materials2057-39602023-06-019111010.1038/s41524-023-01020-9Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivityZiqi Guo0Prabudhya Roy Chowdhury1Zherui Han2Yixuan Sun3Dudong Feng4Guang Lin5Xiulin Ruan6Department of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityDepartment of Mechanical Engineering, Purdue UniversityAbstract Lattice thermal conductivity is important for many applications, but experimental measurements or first principles calculations including three-phonon and four-phonon scattering are expensive or even unaffordable. Machine learning approaches that can achieve similar accuracy have been a long-standing open question. Despite recent progress, machine learning models using structural information as descriptors fall short of experimental or first principles accuracy. This study presents a machine learning approach that predicts phonon scattering rates and thermal conductivity with experimental and first principles accuracy. The success of our approach is enabled by mitigating computational challenges associated with the high skewness of phonon scattering rates and their complex contributions to the total thermal resistance. Transfer learning between different orders of phonon scattering can further improve the model performance. Our surrogates offer up to two orders of magnitude acceleration compared to first principles calculations and would enable large-scale thermal transport informatics.https://doi.org/10.1038/s41524-023-01020-9 |
spellingShingle | Ziqi Guo Prabudhya Roy Chowdhury Zherui Han Yixuan Sun Dudong Feng Guang Lin Xiulin Ruan Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity npj Computational Materials |
title | Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
title_full | Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
title_fullStr | Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
title_full_unstemmed | Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
title_short | Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
title_sort | fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity |
url | https://doi.org/10.1038/s41524-023-01020-9 |
work_keys_str_mv | AT ziqiguo fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT prabudhyaroychowdhury fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT zheruihan fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT yixuansun fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT dudongfeng fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT guanglin fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity AT xiulinruan fastandaccuratemachinelearningpredictionofphononscatteringratesandlatticethermalconductivity |