Machine Learning, Phase Stability, and Disorder with the Automatic Flow Framework for Materials Discovery
Traditional materials discovery approaches - relying primarily on laborious experiments - have controlled the pace of technology. Instead, computational approaches offer an accelerated path: high-throughput exploration and characterization of virtual structures. These ventures, performed by automated ab-initio frameworks, have rapidly expanded the volume of programmatically-accessible data, cultivating opportunities for data-driven approaches. Herein, a collection of robust characterization methods are presented, implemented within the Automatic Flow Framework for Materials Discovery (AFLOW), that leverages materials data for the prediction of phase diagrams and properties of disordered materials. These methods directly address the issue of materials synthesizability, bridging the gap between simulation and experiment. Powering these predictions is the AFLOW.org repository for inorganic crystals, the largest and most comprehensive database of its kind, containing more than 2 million compounds with about 100 different properties computed for each. As calculated with standardized parameter sets, the wealth of data also presents a favorable learning environment. Machine learning algorithms are employed for property prediction, descriptor development, design rule discovery, and the identification of candidate functional materials. When combined with physical models and intelligently formulated descriptors, the data becomes a powerful tool, facilitating the discovery of new materials for applications ranging from high-temperature superconductors to thermoelectrics. These methods have been validated by the synthesis of two new permanent magnets introduced herein - the first discovered by computational approaches.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations