|About this Abstract
||2019 TMS Annual Meeting & Exhibition
||Computational Approaches for Big Data, Artificial Intelligence and Uncertainty Quantification in Computational Materials Science
||An Autonomous Characterization System for Limited-data Experimental Materials Screening: Composition Spread Thin Film Experiments
||Brian DeCost, Heshan Yu, Xiaohang Zhang, Seunghun Lee, Yangang Liang, Jason Hattrick-Simpers, Ichiro Takeuchi, Aaron Gilad Kusne
|On-Site Speaker (Planned)
Active learning, leveraging data-driven models with explicit confidence estimates that direct further experimentation, has made possible autonomous materials characterization systems that adaptively select and perform a sequence of measurements to efficiently reduce their uncertainty about a materials system of interest. We present an autonomous X-ray diffraction system that performs active cluster analysis to efficiently map structural phase diagrams using composition spread thin films. This system has mapped ternary FeGaPd and NiMnGe phase diagrams at room temperature, and composition-temperature phase diagrams for transition-metal- substituted VO<sub>2</sub> which displays a sharp metal-insulator transition. Our current focus is integrating computational modeling, synthesis, and measurement steps to obtain online experimental systems for mapping and optimizing functional properties of materials, especially in the data-limited regime. This approach is general, and we are extending this autonomous research platform to additional experimental techniques, including electrochemical and mechanical evaluations.
||Planned: Supplemental Proceedings volume