Abstract:The calculation of material phonon thermal conductivity from density functional theory calculations requires computationally expensive evaluation of anharmonic interatomic force constants and has remained a computational bottleneck in the high-throughput discovery of materials. In this work, we present a machine learning-assisted approach for the extraction of anharmonic force constants through local learning of the potential energy surface. We demonstrate our approach on a diverse collection of 220 ternary materials for which the total computational time for anharmonic force constants evaluation is reduced by more than an order of magnitude from 480,000 cpu-hours to less than 12,000 cpu-hours while preserving the thermal conductivity prediction accuracy to within 10%. Our approach removes a major hurdle in computational thermal conductivity evaluation and will pave the way forward for the high-throughput discovery of materials.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to significantly reduce the computational cost of anharmonic force constants (IFCs) extraction through machine - learning - assisted methods in the calculation of phonon thermal conductivity of materials, thereby accelerating the thermal conductivity prediction in the high - throughput material discovery process. Specifically, traditional density functional theory (DFT) - based calculations to determine the phonon thermal conductivity of materials require a large amount of computational resources, especially for the evaluation of anharmonic force constants, which has become a computational bottleneck. This paper proposes a machine - learning - assisted method to extract these anharmonic force constants by locally learning the potential energy surface, thereby reducing the computational time by more than an order of magnitude while maintaining the accuracy of thermal conductivity prediction, from 480,000 CPU hours to less than 12,000 CPU hours. ### Main contributions: 1. **Improvement of computational efficiency**: By replacing DFT calculations with a machine - learning model, the computational time required for anharmonic force constant extraction is greatly reduced, and the computational efficiency is improved. 2. **Maintenance of prediction accuracy**: Although the computational time is greatly reduced, this method can still maintain the accuracy of thermal conductivity prediction within 10%. 3. **Wide range of applicability**: This method is applicable not only to simple materials such as silicon, but also to 220 different types of ternary materials, demonstrating its wide applicability. 4. **Description of high - order thermal transport physics**: In addition to basic three - phonon scattering, this method can also effectively describe higher - order thermal transport physics, such as iterative solution of the Boltzmann transport equation (BTE), wavelike coherent transport channels and four - phonon scattering. ### Method overview: - **Dataset construction**: Use the finite difference method or thermal snapshot technology to generate a training dataset. - **Model training**: Adopt the SOAP - GAP machine - learning model and train the model to learn the local potential energy surface. - **Force constant extraction**: Use the trained model to predict forces and then extract anharmonic force constants. - **Performance evaluation**: Evaluate the prediction performance of the model by comparing it with the actual DFT calculation results. ### Conclusion: The method proposed in this paper successfully solves the computational bottleneck problem in the calculation of phonon thermal conductivity of materials and provides a powerful tool for high - throughput material discovery. Future work will further expand the application range of this method, including the treatment of harmonic force constants.

Accelerating Phonon Thermal Conductivity Prediction by an Order of Magnitude Through Machine Learning-Assisted Extraction of Anharmonic Force Constants

Accelerating thermal conductivity prediction through machine-learning: Two orders of magnitude reduction in phonon-phonon scattering rates calculation

Efficient Calculation of the Lattice Thermal Conductivity by Atomistic Simulations with Ab Initio Accuracy

Accelerating High-Throughput Phonon Calculations via Machine Learning Universal Potentials

Accelerating first-principles estimation of thermal conductivity by machine-learning interatomic potentials: A MTP/ShengBTE solution

Fast and accurate machine learning prediction of phonon scattering rates and lattice thermal conductivity

Accurate description of high-order phonon anharmonicity and lattice thermal conductivity from molecular dynamics simulations with machine learning potential

Thermal conductivity prediction by atomistic simulation methods: Recent advances and detailed comparison

High-Throughput Estimation of Phonon Thermal Conductivity from First-Principles Calculations of Elasticity.

Exploring Phononic Properties of Two-Dimensional Materials using Machine Learning Interatomic Potentials

End-to-end Material Thermal Conductivity Prediction through Machine Learning

Accelerating Evaluation of Converged Lattice Thermal Conductivity

Machine learning for predicting thermal transport properties of solids

Predicting lattice thermal conductivity from fundamental material properties using machine learning techniques

Accelerated molecular dynamics force evaluation on graphics processing units for thermal conductivity calculations.

Phonon properties and thermal conductivity from first principles, lattice dynamics, and the Boltzmann transport equation

Thermal Conductivity Predictions with Foundation Atomistic Models

Efficient machine-learning based interatomic potentialsfor exploring thermal conductivity in two-dimensional materials

Rationalizing Phonon Dispersion for Lattice Thermal Conductivity of Solids

Predicting thermal transport properties in phononic crystals via machine learning

A High-Throughput Framework for Lattice Dynamics