Computational Materials Science, Energy Materials, Machine Learning
Research Interests
COMPUTATIONAL MATERIALS SCEINCE
- Computational Materials Science (CMS) is one of R&D tools to understand the nature. CMS imitates the nature as a simple model and simulates the simplified model in the computer system. CMS 1) finds key factors for developing new materials of desired properties from the complex experimental effects using the simple model systems, 2) provides high-throughput virtual screening among the experimentally large number of candidate materials compared to the Edisonian approach (trial and error) of time- consuming and economically expensive experiment, 3) brings new insights into the development of new materials to guide experiments into a promising direction. Therefore, CMS can design new materials of desired properties through the key factors, high- throughput virtual screening and new insights.
ENERGY MATERIALS
- Environmental issues have generated global attention for the development of eco-friendly renewable energy and storage technologies. On the next-generation secondary battery technology roadmap with higher energy density, safety, durability, faster charging and lower cost, all-solid-state batteries will continue the next generation mainstream and metal- air batteries will be the last step in the visible future. The use of CMS in this next- generation secondary battery roadmap will play a very important role for the development of original material technology. Especially, we are focusing on developing new catalysts for several electrochemical reactions,HER/OER/ORR/NRR/CO2RR etc, and new solid-state electrolyte materials.
MACHINE LEARNING
- Materials scientists are constantly striving to advance their ability to understand, predict, and improve materials properties. Material scientists have increasingly relied on simulation and modeling methods to understand and predict materials properties a priori. Materials informatics (MI) is a resulting branch of materials science that utilizes high-throughput computation to analyze large databases of materials properties to gain unique insights. More recently, data-driven methods such as machine learning (ML) have been adopted in MI to study the wealth of existing experimental and computational data in materials science, leading to a paradigm shift in the way materials science research is conducted. We are focusing on developing Moment Tensor Potential (MTP) as a Machine Learned Interatomic Potential (MLIP) and applying Active Learning to develop new materials.
Home