Geant4 simulations of the influence of contamination and roughness of the detector surface on background spectra in CRESST

CRESST is an experiment for the direct detection of dark matter, capable of detecting nuclearrecoils down to 10 eV, which results in an impressive sensitivity for sub-GeV dark matter particles.For a better understanding of the measured background a background model is developed. Thebackground components are considered via Geant4 simulations. At the current state, the CRESSTbackground model only considers bulk contaminations and treats all detector surfaces as perfectplains.This contribution presents potential effects of a surface contamination with radiogenic nuclides, incombination with the influence of the crystals surface roughness. Nuclide decays near the crystalsurface may lead to partial energy deposition inside the detector, potentially causing MeV energyevents to influence the background in the keV energy range. Since default Geant4 is not capableof simulating a rough surface, a new extension for simulating a rough surface is developed and theimpact of different roughness configurations is studied.