PyMPEALab Toolkit for Accelerating Phase Design in Multi-principal Element Alloys

Abstract

Multi-principal element alloys (MPEAs) occur at or nearby the centre of the multicomponent phase space, and they have the unique potential to be tailored with a blend of several desirable properties for the development of materials of future. The lack of universal phase diagrams for MPEAs has been a major challenge in the accelerated design of products with these materials. This study aims to solve this issue by employing data-driven approaches in phase prediction. A MPEA is first represented by numerical fingerprints (composition, atomic size difference , electronegativity , enthalpy of mixing , entropy of mixing , dimensionless Omega parameter, valence electron concentration and phase types ), and an artificial neural network (ANN) is developed upon the datasets of these numerical descriptors. A pyMPEALab GUI interface is developed on the top of this ANN model with a computational capability to associate composition features with remaining other input features. With the GUI interface, an user can predict the phase(s) of a MPEA by entering solely the information of composition. It is further explored on how the knowledge of phase(s) prediction in composition-varied hbox {Al}_xCrCoFeMnNi and hbox {CoCrNiNb}_x can help in understanding the mechanical behavior of these MPEAs.Graphic