Benchmarking of Geant4 simulations for the COSI Anticoincidence System

TL;DR

This paper benchmarks Geant4 simulations of the COSI satellite's anticoincidence system, validating the models against laboratory data and developing correction matrices to improve gamma-ray detection accuracy.

Contribution

It introduces a detailed Geant4 simulation framework including optical physics, validated with laboratory measurements, for the COSI anticoincidence system, enhancing modeling accuracy.

Findings

Simulations replicate experimental energy resolution within 20%.

Light collection uniformity discrepancies are within 10%.

Position-dependent detection efficiency varies by up to 8%.

Abstract

The Compton Spectrometer and Imager (COSI) is an upcoming NASA Small Explorer satellite mission, designed for all-sky observations in the soft gamma-ray domain with the use of germanium detectors (GeDs). An active Anticoincidence System (ACS) of BGO scintillators surrounds the GeDs to reduce the background and contribute to the detection of transient events. Accurately modeling the ACS performance requires simulating the intricate scintillation processes within the shields, which significantly increases the computational cost. We have encoded these effects into a correction matrix derived from dedicated Geant4 simulations with the inclusion of the optical physics. For this purpose, we use laboratory measurements for the energy and spatial response of the ACS lateral wall to benchmark the simulation and define instrument parameters, including the BGO absorption length and the electronic noise. We demonstrate that the simulations replicate the experimental energy resolution and light collection uniformity along the BGO crystal, with maximum discrepancies of 20% and 10%, respectively. The validated simulations are then used to develop the correction matrix for the lateral wall, accounting for the light collection efficiency and energy resolution based on the position within the crystal. The gamma-ray quantum detection efficiency is also position-dependent via the inclusion of the optical physics. It is enhanced by $\sim$8% close to the SiPMs and suppressed by $\sim$2% in the adjacent corners with respect to the average value. Finally, we explore the energy threshold and resolution of the bottom ACS, considering the impact of its smaller crystals compared with the lateral walls.