Enhanced modeling of electronic structure – Mechanical property relationships in high-entropy alloys through data driven analysis

High-Entropy Alloys (HEAs) exhibit remarkable mechanical properties, making them promising candidates for next-generation structural materials. However, accurately predicting their yield strength and plasticity remains a significant challenge due to their complex multi-element interactions. In this study, we establish a data-driven framework that integrates Density of States (DOS)-derived electronic features with machine learning models to predict mechanical properties with higher accuracy than composition-only approaches. Using Principal Component Analysis (PCA), we decompose DOS features into dominant energy-dependent contributions and correlate them with mechanical behavior. Our results reveal that low-energy DOS strongly influence yield strength, with molybdenum (Mo) playing a key role in stabilizing metallic bonding. In contrast, high-energy states promote plasticity, with vanadium (V) contributing significantly to electronic flexibility and deformation mechanisms. Machine learning models incorporating DOS and compositional data outperform those based on composition alone, demonstrating that electronic structure provides critical insights into HEA strengthening mechanisms. Based on these insights, we propose two chemical design principles to extend the strength-plasticity tradeoff limit. These findings reveal a direct connection between electronic structure and mechanical properties, providing a predictive framework for alloy design.