ChE PhD Dissertation Defense: Su Sun

Name:
Su Sun

Title:
Toward Automated Reaction Mechanism Generation for Electrocatalytic CO2 Conversion, Boron Nitride CVD, and Beyond

Date:
11/24/2025

Time:
10:00:00 AM

Committee Members:
Prof. Richard West (Advisor)
Prof. Francisco Hung
Prof. Qing Zhao
Prof. Peter Schindler
Dr. Harsha K. Chelliah

Location:
157 Ryder Hall

Abstract:
The advancement of sustainable energy and advanced material deposition technologies relies on a deep mechanistic understanding of chemical reactions governing electrochemical and heterogeneous catalysis processes. Computational modeling offers a powerful means to accelerate the design and optimization of these systems. However, constructing detailed reaction mechanisms by manually enumerating all possible pathways is laborious and error-prone, as comprehensive models can involve hundreds of chemical species and thousands of elementary reactions. In this dissertation, I present advances toward automated reaction mechanism generation and simulation for complex chemical systems, focusing on (i) electrocatalytic carbon dioxide reduction (CO2RR), (ii) chemical vapor deposition (CVD) of boron nitride (BN), and (iii) the development of a modern MATLAB interface for Cantera that links automated mechanism generation with process-level modeling tools.
Building upon the Reaction Mechanism Generator (RMG) framework and prior methodological developments for electrochemical kinetics, I developed and modernized a prototype RMGElectrocat implementation that can automatically generate detailed mechanisms for CO2RR. I implemented a transport-coupled reactor interface, in collaboration with colleagues, to connect surface and solution-phase chemistry with mass transport at the electrode–electrolyte interface. I integrated ab initio thermochemical and kinetic parameters provided by collaborators into the RMG database, and I introduced the first bimetallic catalyst surface, Cu3Sn, into RMG. Using these capabilities, I constructed a comprehensive “Grand Model” for CO2RR on Ag, Cu, and Cu3Sn surfaces, which includes 296 unique species and 1,322 reactions, among them 79 protoncoupled electron-transfer (PCET) steps. I performed Faradaic efficiency (FE) analyses to evaluate model predictions against experimental data and to identify key mechanistic pathways leading to C1–C4 products. Through this work, I established a practical, extensible workflow for automated generation and quantitative evaluation of electrochemical reaction mechanisms.
To extend automated mechanism generation toward materials synthesis, I carried out preparatory work for BN CVD using RMG. I expanded the RMG database to include boron and Group III atom types and the associated functionalities required to describe B–N–H–Cl chemistry. I performed targeted ab initio calculations on key precursors and intermediates to benchmark quantumchemical methods for boron-containing species, and conducted a preliminary group-additivity-value (GAV) fitting into the RMG database. I then generated a prototype gas-phase BN mechanism with 29 species and 101 reactions and conducted a proof-of-concept CVD reactor simulation to demonstrate its viability. These efforts established a foundation for future automated construction of detailed BN CVD mechanisms and illustrated the adaptability of RMG to emerging thin-film materials systems.
Finally, I developed a modern MATLAB interface for Cantera, enabling robust and user-friendly access to advanced chemical-kinetics, thermodynamics, and transport models directly withinMATLAB. I built the interface usingMATLAB’s clibgen infrastructure to provide an objectoriented, memory-safe API that maps directly onto Cantera’s application binary interface (ABI). I implemented automated build workflows and created a comprehensive suite of MATLAB Live Script tutorials to support educational and research use. This interface connects mechanisms generated by RMG with detailed reactor and process modeling in Cantera, bridging automated mechanism generation with engineering analysis and simulation.