|About this Abstract
||1st World Congress on High Entropy Alloys (HEA 2019)
||High Entropy Alloys 2019
||Optimization of MPEA Properties with Sequential Learning
||James Edward Saal, Chris Borg, Marie-Agathe Charpagne, Daniel Miracle, Tresa Pollock, Bryce Meredig
|On-Site Speaker (Planned)
||James Edward Saal
The combinatorial nature of the composition space multi-principle element alloys (MPEAs) inhabit requires novel techniques to discover and design novel MPEAs with enhanced performance. Trial-and-error experimentation to exhaustively search this space would be prohibitively costly and time-consuming, and current ICME approaches rely on models which have yet to be matured for MPEA-type compositions. Given the increasing data on MPEA compositions being published, the situation is ideal for data-driven materials informatics approaches. To that end, we employ the cloud-based informatics platform, Citrination, to train machine learning models for mechanical properties and single-phase stability for MPEAs from literature data. We then use the FUELS sequential learning approach to design the most efficient series of experiments to rapidly explore composition space for high-performance MPEAs. Candidate alloys were synthesized, characterized, and used to retrain the models, culminating in improved property models and the discovery of novel MPEA systems.