Utilizing AI to Discover Faster Paths to New Materials

Richard Hennig, Ph.D.

Using various methods of advanced processing to manipulate the structure of materials is a universal approach in materials science. It’s how we discover the new materials essential to developing tomorrow’s technologies. In the past, researchers accomplished this through the use of time-intensive costly experimentation.

Richard Hennig, Ph.D., Alumni Professor of Materials Science & Engineering at the University of Florida’s Herbert Wertheim College of Engineering, is leading a team of investigators using Artificial Intelligence (AI) to create a guidebook of how to get there faster. 

“In a nutshell, we’ll be using AI as a route to the faster discovery and creation of new, useful and sufficiently stable materials and, in the process, create the design rules of how to do so,” said Dr. Hennig. “We’re hoping to take some of the guesswork out of getting from A to Z and speeding up the timeline from idea to product.” 

The $1.8 million National Science Foundation (NSF) Design Materials to Revolutionize and Engineer our Future (DMREF) award will enable Dr. Hennig and co-investigators James Hamlin, Ph.D., Peter Hirschfeld, Ph.D., and Gregory Stewart, Ph.D., from the UF Department of Physics to develop and apply innovative machine learning technologies that will help identify new methods to discover unique, stable materials. With complementary expertise in materials synthesis, characterization, theory and simulations, the group’s focus is on materials that could lead to new superconductors, highly textured magnets, or ultra-hard materials. 

Research and design-oriented experiments steered by theory have greatly accelerated materials discovery, leading to the synthesis of novel materials that often challenge chemistry and materials science’s conventional principles that promise significant advances for technological applications. However, one considerable hurdle when using high-pressure materials in applications is that they are often unstable under ambient conditions. 

“One of the biggest challenges in materials science and physics is the control and processing of matter away from equilibrium,” said Dr. Hennig. “This project aims to identify and document novel synthesis pathways for metastable materials that will dramatically expand the materials design space and build the fundamental knowledge base for future applications.” 

Dramatic advances in machine learning techniques allow entirely novel approaches to the design of synthesis routes for metastable materials, and the team has already predicted and synthesized novel superconducting and magnetic materials at both ambient and at high pressure. This project will build on those efforts and extend them into the new domain of metastable material synthesis. 

“We hope that our discoveries and research will eventually lead to new materials for applications such as particle accelerators, medical imaging devices, quantum information technologies, energy technologies and structural applications,” said Dr. Hennig. “We are extremely excited about the opportunity to develop a new direction of materials discovery and synthesis over the next few years.”