ECE Assistant Professor Joel Harley is embarking on a unique collaboration with Douglas E. Spearot from the Department of Mechanical & Aerospace Engineering (MAE)—using a UF Informatics Institute Seed Fund Program grant to better understand how a high entropy alloy (HEA) works on an atomic level. This objective supports a long-term vision of using computer simulation and computational tools to discover and design new HEA compositions and microstructures that have high strength and are resilient to fracture and fatigue deformations. Put another way—Harley is using AI to find out how these alloys work in order to design lighter and stronger combinations of materials for aerospace and military applications.
This is the type of multi-disciplinary collaboration many ECE researchers find themselves doing of late as AI tools become indispensable to research going on around the enterprise. Dr. Harley has extensive experience using machine learning to improve models of complex physical processes, often where experiments and models fail to align. Co-PI Spearot and his lab use atomistic and mesoscale simulation techniques to study the mechanical and thermodynamic properties of materials, with a focus on the behavior of material defects. Together, leveraging the massive AI infrastructure powered by UF’s partnership with NVIDIA, Harley and Spearot have set out to find out how HEAs operate on the atomic level, with an eye towards creating and designing new and promising structural materials.
High entropy alloys are solid, metallic materials in which multiple elements are mixed without a clear distinction of the dominant species. Think: disorganized metals. This contrasts with conventional metallic alloys, such as steel, where a single element has a dominant concentration. High entropy alloys can have high yield strengths and lower densities compared to conventional metals; thus, they are promising as structural materials in next-generation aerospace or automotive applications, and as high strength, light-weight armor in defense applications.
Creating HEAs with tailored mechanical properties remains an elusive goal because of the innumerable possible compositions that can be proposed from elements in the Periodic Table. Currently, extensive trial-and-error experimental testing is used to attempt to correlate HEA composition, processing, and yield strength. Given the numerous variables and chemical environment combinations, testing this hypothesis is impossible without artificial intelligence. Leveraging the computational power of AI & machine learning, the repetitive tasks of running simulations, storing results in a database, and re-testing are all automated.