Machine Learning-Driven Analytical Models for Threshold Displacement Energy Prediction in Materials

TL;DR

This paper develops machine learning models to accurately predict the threshold displacement energy in materials, aiding radiation damage assessment with a focus on monoatomic substances and potential for broader applications.

Contribution

It introduces a novel application of the SISSO machine learning method to derive analytical models for Ed prediction based on fundamental material properties.

Findings

Models outperform traditional methods for monoatomic materials.

Cohesive energy and melting temperature are key predictors.

Challenges remain for polyatomic materials due to dataset complexity.

Abstract

Understanding the behavior of materials under irradiation is crucial for the design and safety of nuclear reactors, spacecraft, and other radiation environments. The threshold displacement energy (Ed) is a critical parameter for understanding radiation damage in materials, yet its determination often relies on costly experiments or simulations. This work leverages the machine learning-based Sure Independence Screening and Sparsifying Operator (SISSO) method to derive accurate, analytical models for predicting Ed using fundamental material properties. The models outperform traditional approaches for monoatomic materials, capturing key trends with high accuracy. While predictions for polyatomic materials highlight challenges due to dataset complexity, they reveal opportunities for improvement with expanded data. This study identifies cohesive energy and melting temperature as key factors influencing Ed, offering a robust framework for efficient, data-driven predictions of radiation damage in diverse materials.