top of page

Research

High-throughput first-principles calculations accelerated functional materials development

Our group has been developing a python-based framework for high-throughput first-principles calculations (https://github.com/bitsoal/VASP_HTC_framework), and we have employed this semi-automatic framework to accelerate the innovation of various functional materials. For instance, we have screened 9 promising 2D metallic electrocatalysts for hydrogen evolution reaction (HER) from nearly 2000 2D materials.

圖片1.png

Our recent focus is on extending and combining high-throughput first-principles calculations with machine learning technologies, aiming to speed up development of catalysts towards a broad spectrum of (photo-)electrochemical reactions, such as HER, OER, ORR, CO2RR and NRR.

 

Ref.:

 

Synergizing DFT calculations and experiments to reveal underlying mechanism

Another research focus in our group is on revealing and understanding novel properties of emerging materials such as 2D materials and their heterostructures, perovskite oxide interfaces, and topological insulators by strongly interacting in-depth DFT calculations and intensive experiments. Our group is also engaged in developing high-throughput experimental growth methods for electrocatalysts and physical vapor deposition (PVD) for 2D monolayers on a large scale.

圖片3.tif
錨點 1

Physics-informed machine learning for functional materials innovation

Our research interest also includes the integration of physics and chemistry domain knowledge into the design of machine learning framework to improve the model's efficiency, interpretability, and generalizability. For example, recently, we have included the information of short-range interaction and passivation of dangling bonds into the graph representation, which results in much improved performance as illustrated in the below figure.

bottom of page