Hardness Prediction of MoNbTaW Alloy Films Based on Machine Learning and Interpretability Analysis

Area of Science:

• Materials Science
• Computational Materials Science
• Alloy Design

Background:

• High-entropy alloys (HEAs) are advanced materials with unique properties.
• Predicting HEA performance, such as hardness, is crucial for their application.
• Current prediction methods can be time-consuming and expensive.

Purpose of the Study:

• To develop a machine learning (ML) framework for predicting the hardness of MoNbTaW HEA films.
• To identify the most influential physical features for hardness prediction in HEAs.
• To establish a rapid and cost-effective method for HEA design.

Main Methods:

• Implemented a ridge regression-based ML model.
• Screened 20 candidate physical features to select an optimal subset.
• Utilized a dataset of HEA films prepared via magnetron sputtering.
• Evaluated model performance using 10-fold cross-validation and a reserved validation set.

Main Results:

• An optimized feature set including δG, Λ, and Ω was identified.
• The ML model demonstrated strong predictive accuracy (R²=0.88, RMSE=0.37 GPa on validation set).
• The model revealed trends in how constituent elements and features influence hardness.

Conclusions:

• The developed ML framework effectively predicts HEA film hardness.
• This ML approach accelerates the discovery and design of new HEAs.
• The study offers a cost-effective alternative to traditional materials development.