Thermal transport properties of silicon based on interatomic machine learning potentials

CAO Qingyuan; HOU Chaofeng; GE Wei

doi:10.11949/0438-1157.20260421

您当前的位置：

首页 >

文章列表页 >

Thermal transport properties of silicon based on interatomic machine learning potentials

更新时间：2026-06-09

- Thermal transport properties of silicon based on interatomic machine learning potentials
- CIESC Journal (2026)
- 作者机构：
  
  1.中国科学院过程工程研究所介科学与过程工程全国重点实验室，北京 100190
  2.中国科学院大学化学工程学院，北京 100049
- 作者简介：
- 基金信息：
- DOI：10.11949/0438-1157.20260421
  CLC： TQ127.2
- Received：28 March 2026，
  
  Revised：2026-06-08，
  
  Accepted：09 June 2026，
- 稿件说明：
移动端阅览
CAO Qingyuan, HOU Chaofeng, GE Wei. Thermal transport properties of silicon based on interatomic machine learning potentials[J/OL]. CIESC Journal, 2026.
DOI：

CAO Qingyuan, HOU Chaofeng, GE Wei. Thermal transport properties of silicon based on interatomic machine learning potentials[J/OL]. CIESC Journal, 2026. DOI： 10.11949/0438-1157.20260421.

摘要

当前，在硅热输运性质的计算中，纯硅的原子间机器学习作用势函数已经有研究报道，但N型半导体和P型半导体中As掺杂和B掺杂的原子间机器学习势函数还未见提出。机器学习势函数具有与量子力学模拟相当的精度，同时又避免了高昂的计算成本，可以克服硅热输运性质计算在大空间和长时间尺度上应用受限的问题。本文通过密度泛函理论分别构建了包含As掺杂和B掺杂硅结构数据集，使用深度神经网络训练得到了两个势函数，并对稳定性和准确性进行一系列测试。通过纯硅声子色散曲线和声子能态密度的计算，以及热导率的非平衡态分子动力学模拟，结果表明相较传统原子间经验力场势函数，机器学习势函数可以更好地描述晶硅的热声子特征，并且有限位移法比密度泛函微扰理论具有更高的声子色散计算精度，热导率的计算结果亦与已报道的工作接近。

Abstract

There have been several studies reporting the interatomic machine learning potential functions for the calculation of thermal transport properties of pure silicon

but the machine learning potential functions for N-type As-doped and P-type B-doped silicon have not been reported yet. Machine learning potential has shown excellent computational accuracy consistent with quantum mechanics

and avoids the expensive computational cost

which breaks the applicability limit of thermal transport calculation in large space and time scales. In this paper

based on the density functional theory energies and forces from the developed As-doped and B-doped silicon dataset and a public pure-silicon database

we trained two deep neural network potentials

and conducted a series of tests on the stability and accuracy of the potential functions. Furthermore

we evaluated the phonon properties and performed the non-equilibrium molecular dynamics simulation to calculate the thermal conductivity of the pure silicon systems. The results show that

compared with traditional empirical potential functions

the machine learning potential functions perform better in calculation of the thermal phonon characteristics

and the finite displacement method has higher accuracy in calculating the phonon dispersion curves than the density functional perturbation theory. And

the calculated thermal conductivities are consistent with the reported work.

关键词

Keywords

references

Oyake T , Feng L , Shiga T , et al . Ultimate confinement of phonon propagation in silicon nanocrystalline structure [J ] . Physical Review Letters , 2018 , 120 ( 4 ): 045901 .

Uma S , McConnell A D , Asheghi M , et al . Temperature-dependent thermal conductivity of undoped polycrystalline silicon layers [J ] . International Journal of Thermophysics , 2001 , 22 ( 2 ): 605 - 616 .

Schelling P K , Shi L , Goodson K E . Managing heat for electronics [J ] . Materials Today , 2005 , 8 ( 6 ): 30 - 35 .

Pop E . Energy dissipation and transport in nanoscale devices [J ] . Nano Research , 2010 , 3 ( 3 ): 147 - 169 .

Moore A L , Shi L . Emerging challenges and materials for thermal management of electronics [J ] . Materials Today , 2014 , 17 ( 4 ): 163 - 174 .

Liebchen R , Breitschädel O , Durmaz A R , et al . Thermal characterization of epitaxial grown polycrystalline silicon [J ] . Thin Solid Films , 2016 , 606 : 99 - 105 .

Lindsay L , Broido D A , Reinecke T L . Thermal conductivity and large isotope effect in GaN from first principles [J ] . Physical Review Letters , 2012 , 109 ( 9 ): 095901 .

Xu J , Liu J B , Li S N , et al . Self-healing properties of nanocrystalline materials: a first-principles analysis of the role of grain boundaries [J ] . Physical Chemistry Chemical Physics , 2016 , 18 ( 27 ): 17930 - 17940 .

Wang J , Cheng J T , Ma S Y , et al . Adsorption of self-assembled monolayer on Cu(111): First-principles study [J ] . Modern Physics Letters B , 2017 , 31 ( 24 ): 1750198 .

Qiu B , Wang Y , Zhao Q , et al . The effects of diameter and chirality in the thermal transport in free-standing and supported carbon-nanotubes [C ] //. ASME 2012 Third International Conference on Micro/Nanoscale Heat and Mass Transfer , March 3–6, 2012 , Atlanta, Georgia, USA . 2013 : 633 - 638 .

Zhu L L , Luo H N . Phonon properties and thermal conductivity of GaN nanofilm under prestress and surface/interface stress [J ] . Journal of Alloys and Compounds , 2016 , 685 : 619 - 625 .

Su R X , Zhang X . Size effect of thermal conductivity in monolayer graphene [J ] . Applied Thermal Engineering , 2018 , 144 : 488 - 494 .

Mortazavi B , Zhuang X Y , Rabczuk T , et al . Atomistic modeling of the mechanical properties: the rise of machine learning interatomic potentials [J ] . Materials Horizons , 2023 , 10 ( 6 ): 1956 - 1968 .

Unke O T , Stöhr M , Ganscha S , et al . Biomolecular dynamics with machine-learned quantum-mechanical force fields trained on diverse chemical fragments [J ] . Science Advances , 2024 , 10 ( 14 ): eadn4397 .

Zhang S H , Makoś M Z , Jadrich R B , et al . Exploring the frontiers of condensed-phase chemistry with a general reactive machine learning potential [J ] . Nature Chemistry , 2024 , 16 ( 5 ): 727 - 734 .

Pfau D , Axelrod S , Sutterud H , et al . Accurate computation of quantum excited states with neural networks [J ] . Science , 2024 , 385 ( 6711 ): eadn0137 .

Wei M Q , Zhang G Q , Zhou T T , et al . Atomic-scale insights into hydrogen transport at tungsten grain boundaries with machine learning potentials [J ] . International Journal of Hydrogen Energy , 2026 , 200 : 152835 .

Behler J , Parrinello M . Generalized neural-network representation of high-dimensional potential-energy surfaces [J ] . Physical Review Letters , 2007 , 98 ( 14 ): 146401 .

Artrith N , Urban A . An implementation of artificial neural-network potentials for atomistic materials simulations: Performance for TiO 2 [J ] . Computational Materials Science , 2016 , 114 : 135 - 150 .

Artrith N , Urban A , Ceder G . Efficient and accurate machine-learning interpolation of atomic energies in compositions with many species [J ] . Physical Review B , 2017 , 96 : 014112 .

Wang H , Zhang L F , Han J Q , et al . DeePMD-kit: a deep learning package for many-body potential energy representation and molecular dynamics [J ] . Computer Physics Communications , 2018 , 228 : 178 - 184 .

Wen M J , Afshar Y , Elliott R S , et al . KLIFF: a framework to develop physics-based and machine learning interatomic potentials [J ] . Computer Physics Communications , 2022 , 272 : 108218 .

Fan Z Y , Wang Y Z , Ying P H , et al . GPUMD: a package for constructing accurate machine-learned potentials and performing highly efficient atomistic simulations [J ] . The Journal of Chemical Physics , 2022 , 157 ( 11 ): 114801 .

Behler J . Four generations of high-dimensional neural network potentials [J ] . Chemical Reviews , 2021 , 121 ( 16 ): 10037 - 10072 .

Babaei H , Guo R Q , Hashemi A , et al . Machine-learning-based interatomic potential for phonon transport in perfect crystalline Si and crystalline Si with vacancies [J ] . Physical Review Materials , 2019 , 3 ( 7 ): 074603 .

Li R , Lee E , Luo T . A unified deep neural network potential capable of predicting thermal conductivity of silicon in different phases [J ] . Materials Today Physics , 2020 , 12 : 100181 .

Bartók A P , Kermode J , Bernstein N , et al . Machine learning a general-purpose interatomic potential for silicon [J ] . Physical Review X , 2018 , 8 ( 4 ): 041048 .

Blöchl P E . Projector augmented-wave method [J ] . Physical Review B , 1994 , 50 ( 24 ): 17953 - 17979 .

Kresse G , Furthmüller J . Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set [J ] . Physical Review B , 1996 , 54 ( 16 ): 11169 - 11186 .

Kresse G , Furthmüller J . Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set [J ] . Computational Materials Science , 1996 , 6 ( 1 ): 15 - 50 .

Perdew J P , Ernzerhof M , Burke K . Rationale for mixing exact exchange with density functional approximations [J ] . The Journal of Chemical Physics , 1996 , 105 ( 22 ): 9982 - 9985 .

Kresse G , Joubert D . From ultrasoft pseudopotentials to the projector augmented-wave method [J ] . Physical Review B , 1999 , 59 ( 3 ): 1758 - 1775 .

Togo A , Chaput L , Tadano T , et al . Implementation strategies in phonopy and phono3py [J ] . Journal of Physics: Condensed Matter , 2023 , 35 ( 35 ): 353001 .

Togo A . First-principles phonon calculations with phonopy and phono3py [J ] . Journal of the Physical Society of Japan , 2023 , 92 : 012001 .

Thompson A P , Aktulga H M , Berger R , et al . LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales [J ] . Computer Physics Communications , 2022 , 271 : 108171 .

Stillinger F H , Weber T A . Computer simulation of local order in condensed phases of silicon [J ] . Physical Review B , 1985 , 31 ( 8 ): 5262 - 5271 .

Tersoff J . New empirical approach for the structure and energy of covalent systems [J ] . Physical Review B , 1988 , 37 ( 12 ): 6991 - 7000 .

Abs da Cruz C , Termentzidis K , Chantrenne P , et al . Molecular dynamics simulations for the prediction of thermal conductivity of bulk silicon and silicon nanowires: Influence of interatomic potentials and boundary conditions [J ] . Journal of Applied Physics , 2011 , 110 ( 3 ): 034309 .

Flensburg C , Stewart R F . Lattice dynamical Debye-Waller factor for silicon [J ] . Physical Review B , 1999 , 60 ( 1 ): 284 - 291 .

Lee Y , Hwang G S . Force-matching-based parameterization of the Stillinger-Weber potential for thermal conduction in silicon [J ] . Physical Review B , 2012 , 85 ( 12 ): 125204 .

Glassbrenner C J , Slack G A . Thermal conductivity of silicon and germanium from 3°K to the melting point [J ] . Physical Review , 1964 , 134 ( 4A ): A1058 - A1069 .

Fulkerson W , Moore J P , Williams R K , et al . Thermal conductivity, electrical resistivity, and seebeck coefficient of silicon from 100 to 1300°K [J ] . Physical Review , 1968 , 167 ( 3 ): 765 - 782 .

Lee Y , Hwang G S . Mechanism of thermal conductivity suppression in doped silicon studied with nonequilibrium molecular dynamics [J ] . Physical Review B , 2012 , 86 ( 7 ): 075202 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Prediction of biomass pyrolysis product yield based on machine learning algorithms

SINDy for discovering the governing equations of noisy rough granular flow in the homogeneous cooling state

Machine learning-assisted high-throughput screening of high-performance zeolites for CO2 adsorption

Machine learning-driven performance prediction and reverse design of lignin-based carbon quantum dots

Machine learning-driven design and optimization of molecular sieve-based efficient CO adsorbents

Related Author

LIU Xiaoyu

LI Qichen

HUO Lili

ZHAO Lixin

YAO Zonglu

SUN Peihao

JIA Jixiu

ZHU Benhai

Related Institution

Institute of Agricultural Environment and Sustainable Development, Chinese Academy of Agricultural Sciences

Key Laboratory of Green and Low-carbon Agriculture in North China Plain, Ministry of Agriculture and Rural Affairs

College of Mechanical and Transportation Engineering, China University of Petroleum

State Key Laboratory of Heavy Oil Processing, China University of Petroleum

State Key Laboratory of Deep Geothermal Resources, China University of Petroleum

⁰