Seminar

Date 2019-01-28 

<MSE SPECIAL SEMINAR>

 

■ Title : Computational Discovery and Design of Materials for Energy Technologies

 

■ Speaker : Prof. Yongjin Lee (ShanghaiTech Univ.) 


■ Date and Time : 2019/1/28 (Mon) 16:00 

 

■ Venue : Applied Engineering Building (W1-1) Room #2427 

 

■ Abstract :

With the amazing advance in computer technology since the late 20th century, computational research is increasingly important particularly in the fields of materials science. In today’s materials science, a critical issue is to gain a deeper understanding about quantitative and qualitative relationship among their synthesis conditions, structures, and properties. In this regard, computational modeling can provide researchers with significant insights into atomic-level interactions and underlying fundamental theories. Furthermore, nowadays, while novel synthetic methodologies allow experimentalists to create billions of possible materials, however, in practice we can only synthesize and test a small fraction of them. Therefore, a key challenge is to develop rational strategies to find the best performing materials given these experimental constraints. Big data techniques and machine learning can contribute to solve this issue by uncovering complex relations between observed behavior and the underlying mechanisms.

In this talk, I will discuss our efforts into computational discovery and design of novel materials for energy related applications. In the first part, I will focus on development of silicon-based thermoelectric (TE) materials. To develop high performance silicon-based TE materials, a key issue has been how to further reduce thermal conductivity (k) beyond the so-called alloy limit which is the minimum k of silicon germanium (SiGe) alloy. I will first describe critical factors determining k of SiGe, and then discuss the possibility of lowering κ below the alloy limit with utilizing ternary alloying and poly-crystallizing. In the second part, I will present development of a descriptor to identify nanoporous materials using big-data analysis and show that our descriptor successfully discovered top-performing structures for methane storage applications from nanoporous materials genome. Furthermore, using big-data analysis it will be explained that top performing materials can be divided into topologically distinct classes and that each class requires different optimisation strategies.