Predicting Unknown Substances Using Artificial Intelligence

AI Utilized to Reverse-Engineer New Promising Hydrogen Production Catalyst Materials
Expectations for Application in Various Material Fields

Developed composition-condition based solid crystal structure generation model.

[Asia Economy Reporter Junho Hwang] A technology has been developed that uses artificial intelligence (AI) to explore all compositions and crystal structures of inorganic compounds to discover new materials. The research team that developed this technology succeeded in predicting crystal structures that can be used as catalysts for hydrogen production. It is expected to be a technology that can be utilized to discover materials needed in various fields.

The research team led by Professor Yuseong Jeong of the Korea Advanced Institute of Science and Technology (KAIST) announced on the 27th that they succeeded in developing a technology that uses AI to explore hidden material spaces and discover new substances, and the related paper was published in the international journal ACS Central Science.

Professor Jeong’s team developed an AI programmed to find materials by setting the desired properties first (material inverse design). The goal of material research is to find the desired substance. However, considering all compositions and crystal structures of inorganic compounds means examining an almost infinite number of cases. By utilizing the AI developed by the research team, it is possible to efficiently find materials with the required properties.

The research team also discovered a new hydrogen production catalyst using the newly developed AI. They secured the crystal structure of a 'magnesium-manganese-oxide based photocatalytic material' that can be used as a catalyst for hydrogen production using light.

Professor Yuseong Jeong said, "The material design framework applied to the design of photocatalytic materials can be used not only for the chemical composition of compounds but also for inverse designing materials with specific properties desired by users," adding, "It is expected to be utilized in various material application fields."

Professor Jeong Yuseong (from the left), Kim Seongwon, PhD candidate, Noh Juhwan, PhD candidate

This content was produced with the assistance of AI translation services.