Incoming Ph.D. candidate @ UChicago
I’m fascinated by the efficiency of using computational technics to numerically simulate or predict materials’ properties. Also, I really enjoy implementing or developing new algorithms to break through restrictions that are ubiquitous in the computational physics’ field. My passion for this subject has led me to pursue further studies and ultimately, a career in computational physics.
Experiences
Designed and implemented a state-of-the-art python package for predicting the most likely structure of organic interfaces.
- Designed a graph-theory algorithm for efficient slab generation, which outperformed all previous algorithms in conserving the atom numbers and structures as well as mechanical properties
- Implemented a lattice matching algorithm for optimizing the lattice parameters of new interfaces and the coordinates of every atom
- Created a Graphic Neural Network (GNN) model for predicting organic systems‘ non-bonded interactions, with the efficiency of empirical methods and the accuracy of DFT method
- Used three different workflows to optimize the structure of interfaces, and GNN exceeded others in accuracy and time cost (Errors < 3%, 10000 × faster than DFT)
Improved the efficiency of CO2 photoreduction by doping alkali metal element to engineer the electronic properties of the catalyst
- Found the most stable relaxed configuration of the potassium-doped g-C3N4 and predicted the most practicable doping density as a reference for the experiment
- Theoretically proved that potassium doping changes monolayer g-C3N4 from indirect gap to direct gap
- Calculated numerically the formation energy and Gibbs free energy (ΔG) of hydrogen atom adsorption on g-C3N4 and K-doped g-C3N4 , proved that the CO2 reduction with K-doped g-C3N4 is more efficient than that with ordinary pristine g-C3N4
- Used differential charge density to numerically prove that K-doping alters the charge distribution of g-C3N4 and inhibit the electron-hole pair recombination
Numerically proved that oxygen doping could enhance SnS2’s photocatalytic property in visible-light-driven CO2 reduction
- Analyzed the electronic structure and magnetic properties of O-doped SnS2 and Ni-O-doped SnS2 via DFT, proved that oxidation had impact on charges’ distribution
- Proved numerically that oxygen-doping made Sn the binding sites of the reaction and could decrease formation energy of a significant step that influenced the whole CO2 reduction process
- Calculated the optimized structure of different doped SnS2 to quickly predict whether the novel materials could be produced in experiment