Muccignat, Dale L., Boyle, Gregory J., Garland, Nathan A., and White, Ronald D. (2025) HFO-1234ze(E): electron-scattering cross section set through an iterative deep learning procedure. Plasma Sources Science and Technology, 34 (12). 125018.

PDF (Published Version) - Published Version Restricted to Repository staff only

Abstract

Recently, HFO-1234ze(E)-based gas mixtures have been proposed as an environmentally friendly replacement for SF<inf>6</inf> in medium voltage gas insulated switchgear. The prediction of electron transport coefficients and reaction rates is therefore crucial for the optimisation of gas mixtures to achieve the desired performance. These calculations rely on the determination of a complete and self-consistent electron scattering cross section set for HFO-1234ze(E). Therefore, in this study, we use a recently developed iterative deep learning method to determine a self-consistent set of electron scattering cross sections for HFO-1234ze(E). The network is trained using perturbations of cross section data available on the LXCat database and is applied to HFO-1234ze(E) transport data measured by the ETH Zurich High Voltage Laboratory group. Overall, the resulting cross-sectional regression achieves a reasonable reproduction of transport coefficients for the pure gas state along with mixtures of CO<inf>2</inf> and SF<inf>6</inf>. Critically, for its application as a dielectric insulator, the set of cross sections reproduces the positive synergistic effect observed in mixtures with SF<inf>6</inf>

Actions (Repository Staff Only)

Item Control Page