Exploring the relationship between lattice distortion and phase stability in a multi-principal element alloy system based on machine learning method

Abstract

Lattice distortion is a basic characteristic of multi-principal element alloys (MPEAs), or high entropy alloys (HEAs). The severe lattice distortion strategy is an effective way to improve the solid solution strengthening of HEAs, but the phase stability will decrease with the increase in lattice distortion. The trade-off between lattice distortion and phase stability is yet to be investigated. Herein, the complex relationship between lattice distortion and phase stability in the Co-Ni-Cr-Mo-V system has been established by using ab initio calculations, thermodynamic calculations and machine learning methods. The results show that electronegativity difference and valence electron concentration are the most important descriptors in the prediction models of lattice distortion and phase stability, respectively, and the support vector machine algorithm has the highest prediction accuracy in the above two groups of models. The establishment of the relationship between lattice distortion and phase stability in the whole composition space has two meanings. One is to point out that the lattice distortion strategy must consider the temperature factor, and the other is to guide the manufacturing and heat treatment process of HEAs with severe lattice distortion. In addition, the correlations between lattice distortion and constituent elements were discussed. It is pointed out that the V element plays an important role in increasing lattice distortion degree in the Co-Ni-Cr-Mo-V system. This study is helpful to understand the relationship between lattice distortion and phase stability and to promote the development of high entropy alloys with severe lattice distortion.