The new algorithm opens the door to studying interactions that occur on the surface of materials using artificial intelligence and machine learning.
Scientists and engineers are studying the atomic interactions that occur on the surface of materials to develop more energy-efficient batteries, capacitors, and other devices. However, accurately simulating these fundamental interactions requires immense computing power to capture geometric and chemical complexity in full, and the current method is simply scratching the surface.
“It's outrageous right now, and there's no supercomputer in the world that can do that kind of analysis,” says Siddharth Deshpande, an assistant professor at the University of Rochester's Department of Chemical Engineering. “We need to have clever ways to manage that large dataset, use our intuition to understand the most important interactions on the surface, and apply data-driven methods to reduce sample space.”
By assessing the structural similarities of various atomic structures, Deshpande and his students found that by analyzing just under 2% of the unique composition of surface interactions, they could obtain accurate pictures of related chemical processes and draw relevant conclusions. They developed an algorithm that reflected this insight. Chemistry Science.
In this study, the authors used algorithms to analyze the complexity of defective metal surfaces and how it affects the oxidation reaction of carbon monoxide for the first time. Deshpande says that the algorithm they developed SuperCharges density functional theory is a computational quantum machine modeling method they call “the mainstay” for studying material structures over the past decades.
“This new approach will be a building site to incorporate machine learning and artificial intelligence,” Deshpande said. “We want to take this into more challenging and challenging applications, such as understanding electrode electrolyte interference in batteries, solvent surface interactions for catalysts, and multicomponent materials such as alloys.”
