(110d) Automating the Generation of Detailed Kinetic Models for PFAS Thermal Destruction with Reaction Mechanism Generator

The disposal of poly- and perfluorinated alkyl substances (PFAS) raises both environmental and public health concerns. Particularly worrisome are perfluorinated carboxylic acids (PFCAs). Although incineration remains the most common PFAS remediation method, the complete combustion/pyrolysis mechanisms of PFAS are unknown. Reaction Mechanism Generator (RMG) is a widely used software that can automatically generate kinetic mechanisms. In principle, RMG can be used to develop PFAS degradation mechanisms and investigate reaction pathways in PFAS destruction; however, the existing RMG database lacks sufficient high-quality PFAS-specific chemical data for accurate mechanism generation. The present work aims to extend and re-train the current RMG database with high-accuracy thermodynamic and kinetic data for PFCAs and direct products of PFCA destruction.

Thermochemical properties for C1-C5 PFCAs, C1-C6 perfluoroalkyl ether carboxylic acids (PFAECAs) and relevant species were determined using computational quantum chemistry methods and parameterized with NASA 7-coefficient polynomials. Elementary rates for more than 230 reactions are computed using RRKM/ME theory. High-accuracy thermochemistry is used to re-train thermochemical group additivity values and hydrogen bond increment groups within RMG, while kinetic data is employed to re-train rate rule decision trees. In addition to extending training datasets for existing reaction families, this work also introduces new families (i.e. carboxylic acid to alpha lactone, CO/CO₂ elimination of alpha lactones and lactone ethers, HF elimination of PFCAs/PFAECAs) into RMG that are imperative for accurate modeling of PFAS degradation.

To evaluate how the database extension affects PFAS modeling, PFAS degradation mechanisms are constructed with the newly extended RMG database and compared to previously generated RMG mechanisms. The PFAS-specific database additions performed in this work improve RMG’s capability to automatically construct PFAS-related mechanisms, with generated models now including more accurate PFAS chemistry than before. As such, RMG can now be used to construct quality PFAS models that aid in understanding complex PFAS degradation behavior.