Development Time for New Materials Reduced... AI Predicts Synthesizability

Predicting Material Synthesis Feasibility with Artificial Intelligence
Expecting Reduced Time for New Material Discovery

[Asia Economy Reporter Junho Hwang] A domestic research team has developed a technology that can predict the synthesizability of materials up to 87% accuracy using deep learning. This technology allows for the pre-assessment of the synthesizability of new materials, which is expected to significantly shorten the development time of new materials.

The research team led by Professor Yuseong Jeong of the Department of Biological Sciences and Engineering at the Korea Advanced Institute of Science and Technology (KAIST) announced on the 22nd that they developed a model that predicts the synthesizability of new materials by learning the structural similarities of previously synthesized host materials using a graph convolutional network (GCN). The results of this study were recently published in the Journal of the American Chemical Society.

The team built the model using the 'Materials Project,' a materials-related database consisting of about 50,000 already synthesized substances and about 80,000 virtual substances. A team official explained, "As a result of utilizing this technology, we were able to predict the synthesizability of materials with about 87% accuracy."

Through this model, the research team also found that thermodynamic stability alone cannot predict the actual synthesizability of materials after analyzing the thermodynamic properties of already synthesized materials. Additionally, a literature survey was conducted on the 100 virtual materials with the highest synthesizability scores in the Materials Project (MP) database. Among these, 71 materials, whose synthesizability was not yet known in the MP database, were actually synthesized and reported in scientific papers.

Professor Yuseong Jeong said, "Although various material design frameworks exist for rapid discovery of new materials, the judgment of the synthesizability of designed materials remains in the realm of expert intuition. The synthesizability prediction model developed this time can help greatly shorten the development time of new materials by allowing the synthesizability to be assessed before experiments when designing new materials."

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