Material Discovery using Evolutionary Algorithms: Finding the Missing Ruthenium-based Ternary Phases for Breakthrough Applications in Energy Technologies

Santanu Chaudhuri, University of Illinois at Urbana-Champaign

Usage Details

The ability to search and optimize crystal structure and chemical motif for identifying unique compounds with breakthrough electronic properties is important for the success of the materials genome initiatives in materials for energy applications. A focused use of petascale computing capabilities requires smart algorithms that are self-aware of the trend of the ongoing search, adjust the allocation of petascale resources judiciously, and interact with expert users. The project will enable testing and improvement of a one-of-a-kind framework in machine learning to discover new ternary phases of semiconducting, magnetic and high-temperature superconducting compounds. The project will leverage the existing funding from the DOE, NSF, and University of Illinois at Urbana–Champaign. A team of multi-disciplinary experts with background in first-principles based search methods, computing, material science, chemistry, x-ray crystallography, and condensed-matter physics will be able to provide real time inputs to the computational search and improve the algorithms to create an efficient pipeline for material discovery.