(2am) Understanding Heterogenous Catalysis Via Multiscale Kinetic Simulation

Heterogeneous catalysis is one of the most fundamental solutions for climate change and net-zero emission technologies. However, the full potential of heterogeneous catalysts remains untapped due to the unknown synergetic effects between catalytic motifs and interfacial variables such as mass transport, redox mediation, and solvent influence. To investigate the synergies between multiple components in complex reaction systems, we need to develop multiscale kinetic models to better engineer the catalytic interface.

My PhD in chemistry in University of Oxford, UK under the supervision of Prof. Richard G. Compton focused on kinetic modeling of electrode-electrolyte interfaces, exploring the influence of mass transport, redox mediator, and double-layer effect on electrode reactions (Curr. Opin. Electrochem., 2019, 14, 186; Chem. Sci., 2017, 8, 6423-6432; J. Phys. Chem. Lett., 2016, 7, 1554). My research as an independent investigator in Chinese Academy of Sciences, China focused on engineering metal/metal oxide interfaces used for electrocatalytic reactions. I developed a novel multiscale model predicting the improvement in reactivity and selectivity for electrocatalytic reactions using nano-impact reactors where nano-catalysts were dispersed in electrolyte instead of immobilized at the electrode surface (Chem. Eng. J., 2021, 418, 129393). The prediction was then validated by our experiments on electrochemical nitrogen fixation at V₂O₅ nanodots (Chem. Eng. J., 2023, 139494) and H₂O₂ reduction at perovskite nanoparticles (Angew. Chem. Int. Ed., 2022, 61, 2022072). In a recent study on electrocatalytic CO­₂ reduction, we observed the enhancement of selectivity towards high-value chemical C₂H₄ with nano-impact reactors via suppressing competitive processes (paper in preparation). Compared to immobilized electrocatalysts, nano-impact reactors enable good selectivity of C₂₊ species under high CO₂ conversion rate.

My current research project in Nanyang Technological University under the supervision of Prof. Tej S. Choksi is to design bimetallic catalysts for selectively dehydrogenating hydrogen carriers like methylcyclohexane (MCH), collaborating with the Chiyoda Corporation, Japan. We propose a model simplification strategy via generalized Brønsted–Evans–Polanyi relationship for kinetic modeling and apply machine-learning (ML) approaches for rapid estimation of reaction energies. Taken together, these approaches reduce computational expenses by at least 90% when compared to Density Functional Theory calculations and microkinetic modeling of the entire reaction network. Our findings provide clear guidance to catalyst preparation and strengthen the understanding of the structure-activity relationship of bimetallic catalysts on MCH dehydrogenation.

In the future, I envision investigating CO₂ conversion towards value-add chemicals such as C₂H₄ and CH₃OH at metal/metal oxide catalysts. Practical challenges for CO₂ electrochemical reduction and CO₂ thermo-hydrogenation include catalyst degradation, coke formation, and low selectivity of high-value products. Multiple catalytic and interfacial parameters, including active site geometry, electrolyte composition, mass transport, local pH, adsorption to CO and H, etc., have been reported as potential solutions for these challenges, but a modelling paradigm clarifying the interplay between distinct operating levels is still missing. Through collaborations with experimentalists and industrial partners, multiscale kinetic models will be developed to understand the reaction mechanism of the CO₂ conversion network and optimize reaction efficiency via tuning the metal/metal oxide interface. Another research interest of mine is to develop new kinetic modeling methodologies for heterogeneous catalysis, in order to link microscale mechanisms with continuum phenomena. Given that the computing time increases exponentially when involving a new variable, model simplification approaches are required for simulating realist conditions. To save computation expenses, data-driven approaches such as graph-convolutional neural network and reinforcement learning will be applied with multiscale modeling to exploit the variable space and provide descriptors for reaction interface engineering.

Teaching Interests:

Based on my interdisciplinary research experiences, I would like to teach courses in chemistry, chemical engineering, and material science, such as physical chemistry, reaction kinetics, catalysis, solid-state physics, etc. I also envision opening new courses on numerical simulation which teaches approaches from matrix operations to machine-learning modeling and electrochemistry which introduces electrochemical theories, techniques, and applications.