Scientists and engineers analyze atomic interactions on material surfaces to design more energy-efficient batteries, capacitors, and other devices. However, accurately simulating these interactions is extremely computationally intensive due to the complex geometry and chemistry involved current techniques only begin to address the challenge.

“There’s no supercomputer in the world that can fully perform this kind of analysis,” says Siddharth Deshpande, assistant professor of chemical engineering at the University of Rochester. We need smarter approaches to manage massive data sets, identify the most critical surface interactions, and use data driven techniques to narrow down the possibilities.

Deshpande and his students tackled this by evaluating structural similarities among atomic configurations.They found they could achieve reliable insights by analyzing just 2% or fewer of all possible surface configurations.This led them to develop a new algorithm,detailed in Chemical Science.

Using this algorithm,the team was the first to examine how defects on metal surfaces impact carbon monoxide oxidation an insight relevant to understanding energy losses in alcohol fuel cells.The algorithm enhances density functional theory, a cornerstone of materials modeling for decades.

This approach lays the foundation for integrating machine learning and AI into materials science, says Deshpande. Our goal is to apply it to more complex problems like electrode electrolyte interactions in batteries, solvent-surface dynamics in catalysis, and the behavior of multi component materials such as alloy.

Souce : https://www.rochester.edu/newscenter/new-method-to-study-catalysts-could-lead-to-better-batteries-657422/