Accelerating High-Entropy Alloy Design via Machine Learning: Predicting Yield Strength from Composition.

Area of Science:

• Materials Science
• Metallurgy
• Computational Materials Science

Background:

• High-entropy alloys (HEAs) offer exceptional properties but are developed inefficiently via trial-and-error.
• This hinders exploration, increases costs, and impacts sustainable production.

Purpose of the Study:

• To develop a machine learning (ML) methodology for predicting HEA yield strengths.
• To accelerate the design and optimization of novel HEA compositions.

Main Methods:

• Trained an ML model on 181 HEA composition data points.
• Achieved an R-squared (R²) score of 0.85 for yield strength prediction.
• Validated the model's generalization across diverse HEA categories (Cantor, refractory, eutectic).

Main Results:

• The ML model accurately predicted yield strength trends across various HEA types.
• Validation confirmed robust performance and reliability on external datasets.
• Demonstrated alignment between predicted and experimental yield strength data.

Conclusions:

• The ML approach facilitates efficient combinatorial design of HEAs.
• It enables rapid optimization of alloy compositions for desired properties.
• The methodology serves as a guideline for sustainable alloy design and environmentally conscious production.