Mitchell Falgoust / Chemistry & Biochemistry / Faculty Mentor: Peter Kroll
Machine learning interatomic potentials (MLIPs) provide a means of scaling up quantum chemical calculations. Typical methods such as density functional theory are limited to system sizes of < 2,000 atoms. MLIPs on the other hand can be applied to systems containing millions of atoms. For the study, we describe the development of a MLIP to simulate polysiloxane pyrolysis. After demonstrating its efficacy with some preliminary results (low temperatures vibrations and high temperature reactions), we apply the MLIP to more complicated problems. Namely, we use the MLIP to study the impact of cross-linking on polysiloxane thermal stability as well as the genesis of free carbon during PDMS pyrolysis.
Leave a Reply