(6bg) Scaling up First Principle Simulation in Realistic Environment: Solvent Effects and Excited State Properties in Computational Catalysis
AIChE Annual Meeting
2019
2019 AIChE Annual Meeting
Meet the Faculty and Post-Doc Candidates Poster Session -- Sponsored by the Education Division
Meet the Faculty and Post-Doc Candidates Poster Session
Sunday, November 10, 2019 - 1:00pm to 3:00pm
First principle simulations have been playing an increasingly important role in the study of molecular properties and reactions, especially in computational catalysis. First principle simulations have multiple aspects for researchers to explore: the level of electronic structure method, the system complexity, and the time scale of simulation. Due to computation bottlenecks, improvement of one aspect of the simulation is often achieved at the cost of compromising the others. My research has been focused on improving different aspects of first principle simulations simultaneously in order to understand reactions in the realistic complexes experimental environment both accurately and efficiently. Specifically, I have focused on (1) molecular properties and reactions in solution phase; (2) improving the accuracy of density functional theory (DFT) in predicting transition metal chemistry; (3) accurate and efficient description of excited state properties; and (4) Automated control and analysis of first principle simulations.
During my Ph.D. study, I developed the graphical processing units (GPUs) algorithms to accelerate the widely used implicit solvent model, the conductor-like polarizable continuum model (C-PCM), in first principle calculations to enable the evaluation of solution phase molecular properties and reactivity in both ground state and excited state. I also exploited the GPUs in developing the new ensemble DFT methods that are capable of describing the electronic structure of molecules in photochemistry reactions.
During my postdoctoral studies, I investigated the density delocalization errors in approximate DFT in the studies of transition metal chemistry. Transition metal containing molecules play essential roles in catalysis. However, they are challenging to study with conventional computational modeling using DFT. My works on benchmarking density the delocalization errors and spin splitting energy evaluations provide broad recommendations for computational modeling of open-shell transition metal containing systems.
Currently, I focus on building workflows to improve the automated control and analysis of transition metal chemistry simulations. A crucial step to obtain accurate results for transition metal containing systems is to choose between single-reference and multi-reference based methods. I develop an open-source module, MultirefPredict, to automate the calculation of widely used multi-reference diagnostics, which will then be embedded into the inorganic chemistry toolkit, molSimplify, to enable automated recommendation of first principle methods and computing resources.
Teaching Interests:
Based on my background in physical chemistry, I am pleased to teach kinetic, thermodynamics, numerical analysis, physical chemistry, and courses specialized in atomistic simulations, including electronic structure methods (e.g. DFT), molecular dynamics, etc.