Data-driven surrogate modeling of hPIC ion energy-angle distributions for high-dimensional sensitivity analysis of plasma parameters’ uncertainty

The quantification of wall erosion due to plasma-surface interactions requires the determination of the energy-angle distribution of the ions accelerated through the plasma sheath and impacting on the material surfaces. However, such calculations are computationally expensive and thus a surrogate model is desired. Several challenges arise when attempting to construct a surrogate model for the ion energy-angle distribution (IEAD) computed using hPIC. Challenges include: high computational cost of particle-in-cell (PIC) codes; high dimensionality of physical parameters affecting the IEAD; and significant variation of IEAD support over the range of physical parameters. To effectively address these issues, a data-driven surrogate model strategy was developed. The strategy utilizes sparse grids in the parameter space and coordinate transformations in the energy-angle phase space. The surrogate model showed a significant reduction in computation time and was used to draw samples necessary to perform global sensitivity analysis (SA) for the ions’ energy and angle moments at the plasma-material interface. SA reveals a strong dependency of the moments on the electron-to-ion temperature ratio and intermediate dependency on the magnetic field inclination angle, whereas the dependencies on the magnetic field magnitude and plasma density are less significant.