Evaluation of computational models and cross sections used by MCNP6 for simulation of electron backscattering

This work evaluates the accuracy of the single-event (SE) and condensed-history (CH) models of electron transport in Monte Carlo simulations of electron backscattering from thick layers of Be, C, Al, Cu, Ag, Au and U at incident electron energies from 200 eV to 15 MeV. The CH method is used in simulations performed with MCNP6.1, and the SE method is used in simulations performed with an open-source single-event code MCNelectron written by the author of this paper. Both MCNP6.1 and MCNelectron use mainly ENDF/B-VI.8 library data, but MCNelectron allows replacing cross sections of certain types of interactions by alternative datasets from other sources. The SE method is evaluated both using only ENDF/B-VI.8 cross sections (the “SE-ENDF/B method”, which is equivalent to using MCNP6.1 in SE mode) and with an alternative set of elastic scattering cross sections obtained from relativistic (Dirac) partial-wave (DPW) calculations (the “SE-DPW method”). It is shown that at energies from 200 eV to 300 keV the estimates of the backscattering coefficients obtained using the SE-DPW method are typically within 10% of the experimental data, which is approximately the same accuracy that is achieved using MCNP6.1 in CH mode. At energies below 1 keV and above 300 keV, the SE-DPW method is much more accurate than the SE-ENDF/B method due to lack of angular distribution data in the ENDF/B library in those energy ranges. At energies from 500 keV to 15 MeV, the CH approximation is roughly twice more accurate than the SE-DPW method, with the average relative errors equal 7% and 14%, respectively. The energy probability density functions (PDFs) of backscattered electrons for Al and Cu, calculated using the SE method with DPW cross sections when energy of incident electrons is 20 keV, have an average absolute error as low as 4% of the average PDF. This error is approximately twice less than the error of the corresponding PDF calculated using the CH approximation. It is concluded that the SE-DPW method is sufficiently reliable to be used as an alternative of the CH method in MCNP6 at electron energies up to the order of 100 keV, including energies from 200 eV to 1 keV, where CH approximations are not applicable in principle.