BostonScience, Tech & Medicine

MIT Masters Material Marvel, New Method Revolutionizes Surface Science with Machine Learning Magic

MIT Masters Material Marvel, New Method Revolutionizes Surface Science with Machine Learning MagicSource: Massachusetts Institute of Technology Official Website
Benjamin Cortez
Published on December 07, 2023

MIT engineers have made a significant breakthrough in the field of material science by developing an innovative method that uses machine learning to determine how material surfaces behave. This novel technique stands to benefit industries relying on catalysts, semiconductors, and battery components.

The team's approach eliminates the need for guesswork traditionally associated with chemists' intuition, allowing instead for comprehensive automatic sampling of material surfaces. According to a report by MIT News, the system has already identified new atomic configurations on a material previously studied for three decades, suggesting a leap forward in efficiency and accuracy in surface characterization.

Detailed in the journal Nature Computational Science, the research led by graduate student Xiaochen Du, alongside professors Rafael Gómez-Bombarelli, Bilge Yildiz, and other experts from MIT, turns a corner from static observational methods. Recognizing the dynamic nature of material surfaces and the myriad ways they can change under different conditions, this new strategy predicts surface energies and atomic arrangements in a way that’s robust and dynamically relevant.

“We are looking at thermodynamics,” Du told MIT News, referring to the study of how surfaces behave under varying external conditions such as temperature, pressure, and chemical potential. By applying machine learning and a Monte Carlo algorithm, the research team effectively identified configurations of surface atoms that could significantly contribute to the development of catalysts for 'green' hydrogen and in capturing carbon dioxide emissions.

In an industry-first, the Automatic Surface Reconstruction Framework, or AutoSurfRecon, allows researchers to predict the behavior of a material’s surface with far fewer calculations than previously thought necessary. The tool has been made freely available, encouraging global collaboration in the quest for better-performing materials across various applications.

Professor Gómez-Bombarelli emphasized the discrepancy between the bulk material and the catalytic surface's behavior during operation, stating that "the entity responsible for the catalyst doing something is a few atoms exposed on the surface, so it really matters a lot what exactly the surface looks like at the moment.” The practical implications of this development are wide-ranging, from advancing hydrogen fuel production to improving the efficacy of carbon capture technologies, as per MIT News.

This pioneering work has been supported by entities, including the U.S. Air Force, the U.S. Department of Defense, and the U.S. National Science Foundation. With such backing and the potential of the AutoSurfRecon tool, MIT's latest scientific advancement could pave the way for significant progress in materials science and related industries.

BostonScience, Tech & Medicine