High‐Throughput Calculation and Machine Learning‐Assisted Prediction of the Mechanical Properties of Refractory Multi‐Principal Element Alloys

Abstract Refractory multi‐principal element alloys (RMPEAs) are promising materials for high‐temperature applications due to their exceptional mechanical stability, yet their vast compositional space poses challenges for efficient design. To address this challenge, this study introduces a novel computational framework integrating high‐throughput Exact Muffin‐Tin Orbitals with Coherent Potential Approximation (EMTO‐CPA) calculations, Copula entropy‐based feature selection, and machine learning (ML) to accelerate RMPEA development. Focusing on V‐Nb‐Ta ternary alloys, a comprehensive dataset of elastic properties for 4485 different compositions is constructed using EMTO‐CPA, achieving accuracy comparable to traditional the special quasi‐random structure (SQS) methods with significantly reduced computational cost. A Random Forest ML model, trained on Copula entropy‐selected features, predicted elastic constants ( C 11 , C 12 , E ) with high accuracy (R 2 > 92%), and C 44 with good accuracy (R 2 ≈ 88%). Experimental synthesis and characterization of selected V‐Nb‐Ta alloys validated the predictions, confirming the trend of elastic modulus with varying Ta content. This integrated approach not only overcomes the limitations of conventional computational methods but also provides a scalable pipeline for designing RMPEAs tailored for extreme environments.