Graduation Semester and Year
Summer 2025
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Chemistry
Department
Chemistry and Biochemistry
First Advisor
Peter Kroll
Second Advisor
Kwangho Nam
Third Advisor
Robin Macaluso
Fourth Advisor
Frank Foss
Abstract
Accurate simulations and calculations using quantum chemical methods such as DFT provide invaluable insight into various chemical problems. Though very accurate, they are limited by their computational cost. For well-ordered crystals, small models are often good enough. However, for amorphous solids, namely SiCO ceramics, the model size must be large enough to capture their full expression. The sizes required are typically well beyond what is practical for DFT. Machine learning interatomic potentials (MLIPs) have emerged as a powerful alternative. Such potentials are fit to DFT-level accuracy and are capable of simulating very large systems.
This thesis is split into two parts: Part I deals with the application of MLIPs, based on the moment tensor potential (MTP) formalism, to polysiloxane pyrolysis. Chapter 1 gives details into the fitting of the potential as well as some preliminary simulations describing near-equilibrium vibrations and high-temperature reactions. Chapter 2 applies an improved MLIP to describe vibrations of some interface models of SiCO with graphitic carbon (Cg). Chapter 3 describes the use of yet another improved MLIP to simulate the thermal degradation of several polysiloxanes which vary in degree of cross-linking. Finally, Chapter 4 uses the same MLIP from Chapter 3 to simulate the genesis and growth of free carbon during polysiloxane pyrolysis. Part II focuses on a different problem entirely, which contains a single chapter devoted to the prediction of a novel anti-spinel Al4C3using first-principles methods.
Keywords
MLIP, DFT, Polymers, Ceramics, MTP, PDC, Machine Learning
Disciplines
Computational Chemistry
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Falgoust, Mitchell, "Machine Learning Interatomic Potentials for the Conversion of Polysiloxanes to SiCO Ceramics" (2025). Chemistry & Biochemistry Dissertations. 289.
https://mavmatrix.uta.edu/chemistry_dissertations/289