Quantitative Test of the Evolution of Geant4 Electron Backscattering Simulation

TL;DR

This study evaluates how updates in Geant4 software impact electron backscattering simulations, analyzing compatibility with experimental data and assessing the performance of new scattering models across different versions.

Contribution

It provides a quantitative assessment of Geant4 evolution effects on electron backscattering simulation accuracy and performance, focusing on new scattering models and geometry improvements.

Findings

Goudsmit-Saunderson model performs variably across energies.

New Coulomb model is physics-flawed but computationally fast.

Geometry updates improved simulation accuracy.

Abstract

Evolutions of Geant4 code have affected the simulation of electron backscattering with respect to previously published results. Their effects are quantified by analyzing the compatibility of the simulated electron backscattering fraction with a large collection of experimental data for a wide set of physics configuration options available in Geant4. Special emphasis is placed on two electron scattering implementations first released in Geant4 version 10.2: the Goudsmit-Saunderson multiple scattering model and a single Coulomb scattering model based on Mott cross section calculation. The new Goudsmit-Saunderson multiple scattering model appears to perform equally or less accurately than the model implemented in previous Geant4 versions, depending on the electron energy. The new Coulomb scattering model was flawed from a physics point of view, but computationally fast in Geant4 version 10.2; the physics correction released in Geant4 version 10.2p01 severely degrades its computational performance. Evolutions in the Geant4 geometry domain have addressed physics problems observed in electron backscattering simulation in previous publications.