Data-driven prediction of thermoelastic and defect properties of Ti-V-Ta-W high-entropy alloys up to the melting point

The theoretical investigation of high-entropy alloys (HEAs) from 0 K to melting temperature is proposed through a workflow combining density functional theory (DFT), Monte Carlo simulations, and machine learning. This workflow provides accurate predictions for HEA phases and enables a detailed characterization of defects in HEAs. This study enables the incorporation of anharmonic vibrational free energy in the understanding of point defects in HEAs, providing a useful approach for understanding the complex energy landscape of HEAs. The chosen HEA is Ti-V-Ta-W, for which thousands of DFT calculations over a wide range of compositions were conducted to design a reliable machine-learning force field. Predicted properties such as lattice parameters, elastic constants, distributions of formation energies of vacancy and interstitial defects, and anharmonic free energy of monovacancy formation have been discussed. This paper opens the way for a systematic investigation of high-temperature HEA properties with DFT accuracy. The presented methodology can support systematic investigation of high-temperature HEA properties with DFT-level accuracy.