Data-driven Materials Research with Experimental Research Data

Information	Photo
Name Hyunju Chang Affiliation Chemical Data-driven Research Center, Korea Research Institute of Chemical Technology E-mail hjchang@krict.re.kr

Abstract

Recently, many data-driven studies have been reported in materials science. Many data-driven studies of materials mainly utilize databases from first-principles calculations or databases obtained from the literature. However, machine learning prediction models from computational data tend to be confined to the limitations of computational properties. Moreover, the data from the literature are limited in the champion data from the experiments. Building a machine-learning prediction model based on experimental research data, including the “dark results”, is necessary to develop new materials efficiently. In this presentation, I will introduce our recent works that collect and utilize experimental data from the entire material development cycle. First, we built web-based platforms to collect the research data easily for various applications, including thermoelectric materials, perovskite solar cell materials, etc. Then we built and utilized a machine-learning prediction model from the collected data to find new materials. As an example, I will discuss the following topics. [1] SnSe-based thermoelectric material development, [2] Closed-loop optimization of catalyst for oxidative propane dehydrogenation with CO2

Biography

Dr. Hyunju Chang has worked as a research scientist at the Korea Research Institute of Chemical Technology (KRICT) in Korea since 1996. Now she is leading Chemical Data-driven Research Center at KRICT. She received B.S. and M.S. degrees in physic from Seoul National University (Korea) in 1986 and 1888, respectively, and a Ph.D. in condensed matter theory from Michigan State Univ. (USA ) in 1995. Before joining KRICT, she had worked as a post-doctor researcher at the dept. of Materials Sci. & Eng. of Northwestern Univ., USA (1995-1996). She has been working on computer-aided materials design for various materials, including oxides and nanomaterials. Her current research interests include materials informatics to utilize the materials data and to develop novel materials using machine learning from the materials database.