|About this Abstract
||1st World Congress on High Entropy Alloys (HEA 2019)
||High Entropy Alloys 2019
||Screening of Multi-principal Element and High-entropy Alloys by Materials Informatics
||Jeffrey M. Rickman, Helen M. Chan, Martin P. Harmer, Joshua Smeltzer, Christopher J. Marvel, Ankit Roy, Ganesh Balasubramanian
|On-Site Speaker (Planned)
||Jeffrey M. Rickman
The identification of promising high-entropy alloys presents a daunting challenge given the vastness of the chemistry/composition space associated with these systems. This work describes a supervised learning strategy for the efficient screening of high-entropy alloys that combines two complementary tools, namely: (1) a multiple regression analysis and its generalization, a canonical-correlation analysis (CCA), connecting predictive metrics with material properties, and (2) a genetic algorithm (GA) that explores a multi-dimensional chemistry/composition space with a CCA-inspired fitness function. Starting with a database comprising 82 alloys for which reliable mechanical property information was available, this procedure was used to identify new alloy compositions having potentially high hardness values. These compositions were subsequently fabricated using arc-melting, and the microstructures characterized using SEM. The corresponding hardness values were measured experimentally. The results demonstrated that our informatics based methodology provides a convenient means to predict, at the 90% confidence level, complex alloy compositions with enhanced hardness.