Advanced algorithm to study catalysts on material surfaces could lead to better batteries

A new algorithm opens the door to the use of artificial intelligence and automatic learning to study the interactions that occur on the surface of materials.

Scientists and engineers study atomic interactions that occur on the surface of materials to develop batteries, capacitors and other more energy efficient devices. But the precision simulation of these fundamental interactions requires an immense computing power to fully capture the geometric and chemical subtleties involved, and the current methods only scrape the surface.

“Currently, it is prohibitive and there is no supercomputer in the world that can do an analysis like this,” explains Siddharth Deshpande, assistant professor in the Department of Chemical Engineering at the University of Rochester. “We need intelligent ways to manage this important data set, to use intuition to understand the most important interactions on the surface and apply data -based methods to reduce the space of the sample.”

By evaluating the structural similarity of different atomic structures, Deshpande and its students found that they could obtain a precise image of the chemical processes involved and draw the relevant conclusions by analyzing only two percent or less of the unique configurations of surface interactions. They developed an algorithm reflecting this information, which they described in a study published in Chemicals.

In the study, the authors used the algorithm to, for the first time, analyze the subtleties of a defective metallic surface and how it affects the oxidation reaction of carbon monoxide, which can, in turn, help to understand energy losses in an alcohol fuel cell.

Deshpande says that the algorithm that they have developed the functional theory of the density of the superfoods, a method of mechanical computer modeling computer that he calls the “battle horse” in recent decades to study the structure of materials.

“This new method becomes the construction field to incorporate automatic learning and artificial intelligence,” explains Deshpande.

“We want to bring this to more difficult and more difficult applications, such as understanding electrolery-electrolyte interference in batteries, solvent-surface interactions for catalysis and multi-component materials such as alloys.”