Accelerated exploration of high-performance multi-principal element alloys: data-driven high-throughput calculations and active learning method

We propose an active learning guided density functional theory calculation framework for rapid screening of multi-principal element alloys (MPEAs) with superior mechanical properties. Using this framework, we fast construct the datasets of the bulk modulus (<i>B</i>), shear modulus (<i>G</i>), yield strength, and Pugh’s ratio of 12,698 noble metal MPEAs. These datasets were obtained with density functional theory prediction accuracy (<i>R</i>²= 0.98 and 0.96 for <i>B</i> and <i>G</i>, respectively) based on active learning guided 120 DFT calculated data. Analysis of the dataset shows that Ni and Au would enhance the yield strength and the toughness of these noble MPEAs, respectively. An active learning guided density functional theory calculation framework, which reduces computation time by three orders of magnitude, was proposed for rapidly screening multi-principal element alloys with superior mechanical properties.