Prediction of a complete and self-consistent collision cross-section set of C5F10O by data-driven method

Abstract The lack of a complete and consistent set of electron–molecule collision cross sections for the new eco-friendly gas C 5 F 10 O hinders the study of its microscopic discharge mechanisms. In this paper, we first calculate the elastic collision and electronic excitation cross-sections of C 5 F 10 O based on R-matrix theory and estimate its total attachment cross-sections through pulsed Townsend measurement. By combining cross sections from other literature, we are able to compile an initial set of collision cross-sections for C 5 F 10 O. However, this initial set remains incomplete and requires refinement. Therefore, this work establishes a neutral network to solve the inverse swarm problem of deriving cross-sections from swarm parameters. The inversion model is trained using a substantial amount of collision cross-sections from the LXCat project. Using the model and the initial set for C 5 F 10 O, a complete and self-consistent collision cross-section set for the C 5 F 10 O gas is predicted for the first time, based on the measured electron swarm parameters of C 5 F 10 O/Ar mixtures. The proposed method is capable of predicting C 5 F 10 O’s unknown vibrational excitation cross-section without any prior knowledge, thus enhancing its completeness. The refined set can reproduce the electron swarm parameters within an acceptable range of uncertainty, thus verifying its self-consistency. The set will be made available in the LXCat database, which is expected to be significant for fundamental studies of its discharge mechanisms as well as applications of C 5 F 10 O.