Your browser is unsupported

We recommend using the latest version of IE11, Edge, Chrome, Firefox or Safari.

Using deep learning to discover materials with exotic properties

Suvo Banik, a PhD candidate in mechanical and industrial engineering

Suvo Banik, a PhD candidate in mechanical and industrial engineering, won the best presentation award for his research at the 2022 Materials Research Society (MRS) Spring Meeting and Exhibit in Honolulu, Hawaii.

The MRS is a member-driven organization of more than 11,000 people from industry, academia and national labs from more than 90 countries around the world. Its goal is to advance materials and improve the quality of life.

Banik, who works under the direction of Associate Professor Subramanian Sankaranarayanan, presented on two topics during the event. The first presentation was titled “CEGAN: Crystal Edge Graph Attention Network for multiscale classification of materials environment,” and the second was “Exploring Polymer Degradation Pathways using Reinforcement Learning.”

His research focuses on the characterization of materials with unique properties, which is the core of data-driven material design and discovery.

“Compared to the anticipated potential diversity of the materials, a relatively small fraction of them have been characterized experimentally or with computational methods,” he said. “Given the surge in the development of materials databases in recent years, there is an urgent need for automated tools to analyze large amounts of structural data.”

By distinguishing the unique characteristics across different classes of materials, the research can provide key insights into learnable aspects, which is crucial for AI-guided material design and discovery.

“Data-driven machine learning methods have become the heart of discovery and an integral part of our daily life,” Banik said. “Designing new materials holds the key to a clean energy future where we need more efficient solar panels, wind turbines, and batteries made with cheap end efficient materials. My data-driven characterization method uses deep learning approaches to enhance the design and discovery process, which will lead to the rapid discovery of materials with exotic properties.”

His successful research comes from working in a positive and supportive environment with Sankaranarayanan at UIC and Argonne National Laboratory.

“Our group works at the forefront of machine learning and materials science. We are always encouraged to put our own thoughts, creatively formulate and explore our own solutions to the research problems, which have always led to a fruitful outcome,” Banik said. “He constantly uplifts your morale and encourages you even during constant failures. We have access to some of the best computational facilities in the world and we always get opportunities to work with collaborators who are pioneers in the field. All these contribute to a great learning experience”