Prediction of gaseous medium insulation strength based on electrostatic potential on real space function isosurface.

CONTEXT

SF is widely used in electrical equipment due to its chemical stability and insulation strength, but it is a strong greenhouse gas and its use has been restricted internationally. In order to reduce the SF usage, it is needed to find a replacement gas for SF. Electrical breakdown test is always adopted to select potential substitutes, but it is resource and time intensive. Thus, a structure-activity relationship model is needed to effectively predict the gas insulation strength. In this work, we calculated the isosurface electrostatic potential of 68 gas molecules in case of electron probability density, Laplacian of electron density, electron localization function, and localized orbital function. The distribution characteristics of these four real space functions were analyzed. Furthermore, correlation between the electrostatic potential parameters and insulation strength was presented. Finally, a prediction model for insulation strength of gaseous medium was established. Using the electrostatic potential parameter on the localized orbital locator function with a threshold of 0.05 a.u., the prediction model achieved the best performance with a coefficient of determination of 0.860 and a mean squared error of 0.0663.

METHODS

The quantization calculation tool used in this work is the GAUSSIAN 16 software. The M06-2X method with the 6-311G++(d, p) basis set is used to optimize the molecular structure and output stable wavefunction files. Then the wavefunction analysis software Multiwfn is used to plot the contour map of the gas molecules and calculate the radial distribution patterns.