A joint research team of Professor Seungchul Lee (ME), Soo Young Lee (ME Ph.D. Integrated Candidate, advisor Prof. Seungchul Lee), Prof. Jin, Hyungyu (ME), Seokyeong Byun (ME Ph.D. Integrated Candidate, advisor Prof. Jin, Hyungyu), and Prof. Kim Hyoung Seop (MSE) has developed a phase prediction technology for high-entropy alloys using Artificial Intelligence (A.I.). The research was published in Volume 197 of Materials and Design.
Conventionally, metallic materials are made out of alloys by mixing a prime element that is most fitting with two or three auxiliary elements. In contrast, high-entropy alloys are mixed with five or more elements in equal or similar proportions without a prime element. Theoretically, the types of alloys that can be made through this method are almost infinite, and the alloys show excellent mechanical, thermal, physical, and chemical properties.
Designing new materials has always required excessive time and cost. Consequently, it has been extremely difficult to determine the phase, mechanical, and thermal characteristics of an alloy in advance.
The team focused on developing predictive models on high-entropy, high-performance, and explainable alloys by applying deep learning technology. In particular, the team aimed to build a data-enhancing model based on the Conditional Generative Adversary Network. This allowed A.I. models to take high-entropy alloys that have not yet been discovered into account, thus achieving rapid improvements in phase prediction accuracy.
Also, the research team developed a predicting model of high-entropy alloy phase based on “explainable” A.I. This opened the possibility of interpreting the black box deep learning model, while also providing guidance on important design factors to develop high-entropy alloys with specific phases.
Prof. Seungchul Lee said, “The research has drastically overcome the limitations of existing research by incorporating A.I. technology into high-entropy alloys that have been drawing attention recently,” and added that, “It is meaningful because the joint research team’s multidisciplinary collaboration has produced results that can promote the development of new A.I.-based materials.”
Meanwhile, the research was funded by the National Research Foundation of Korea, the Institute of Information & Communications Technology Planning & Evaluation, and the Korea Institute of Energy Technology Evaluation and Planning.