Characterizing, correcting, and repairing the effects of radiation damage in the COSI germanium cross-strip detectors

TL;DR

This work addresses radiation damage in COSI's germanium cross-strip detectors by developing a physics-based trapping model centered on the product $[n\sigma]$, validating it with proton irradiations up to multi-year space-equivalent fluences, and implementing a depth-of-interaction–driven correction to restore spectral performance. It quantifies hole trapping as $[n\sigma]_h = (5.4 \pm 0.4)\times10^{-11} F_p$ and demonstrates that proton-induced damage can be corrected in data, while annealing at elevated temperature can repair damaged detectors, achieving substantial recovery in spectral resolution. The combination of DOI-based corrections and controlled annealing provides a path for maintaining COSI's spectral performance through its baseline and extended missions, with ongoing work to optimize corrections for moderate exposure levels and to characterize on-orbit trapping using SAA background lines. Overall, the study offers a practical framework for sustaining high-resolution gamma-ray spectroscopy in the presence of radiation damage in germanium detectors.

Abstract

The Compton Spectrometer and Imager (COSI) is a gamma-ray survey telescope utilizing a compact Compton imager design, enabled by an array of 16 high-resolution germanium cross-strip detectors. After its launch into an equatorial Low Earth Orbit (LEO) in 2027, COSI will experience radiation damage primarily due to energetic protons, with the proton fluence dominated by the passage of COSI through the edge of the South Atlantic Anomaly (SAA) for a few minutes each orbit. We have developed a comprehensive program focused on the modeling, characterization, data correction, and physical repair of radiation damage effects in the COSI detectors. We have performed energetic proton beam irradiations of a spare COSI detector at a proton synchrotron, with proton fluences consistent with multiple years of exposure to the COSI space radiation environment. These exposures allow us to characterize the relationship between proton fluence and induced charge trapping. We demonstrate our techniques to correct for trapping effects, as well as characterize the effectiveness of high-temperature annealing on correcting this damage, as characterized by the resulting spectral performance of the detector. We will present our efforts to characterize the effects of radiation damage in the COSI detectors, as well as our techniques for correcting these effects in the data analysis pipeline and ultimately repairing the detectors on orbit every few years through high-temperature annealing.

Figures (4)

• Figure 1: (left) Cutaway view of the COSI arraytomsick2023comptonspectrometerimager. (right) Diagram of the charge collection process in a COSI GeD induced by a photon interaction of energy Eboggs2023numerical. By measuring the time difference between collection of the free electrons and holes, we can measure the full 3D position of interactions within the cross-strip GeDs
• Figure 2: Charge Collection Efficiency (CCE) as a function of interaction depth measured with the COSI detector for electron-collecting strips (HV signal) and hole-collecting strips (LV signal) pike2025characterizing. (left) Pre-damage curves, demonstrating the dominance of electron-trapping on the HV signal. (middle) After damage with a proton fluence of $2.0\times10^8 p^+/cm^2$, demonstrating the dominance of hole trapping on the LV signal after irradiation. (right) After extensive damage with a total proton fluence of $5.0\times10^8 p^+/cm^2$. Shown for comparison (solid curves) are the corresponding best fit curves from our numerical charge trapping model boggs2023numerical, which does an excellent job reproducing the measured curves. These curves were measured utilizing a Cs-137 (661.7 keV) source.
• Figure 3: Cs-137 (661.7 keV) spectra for the COSI detector after undergoing radiation damage with an energetic proton fluence of $2.0\times10^8 p^+/cm^2$, corresponding to $\sim$4 years radiation exposure (worst case) in the COSI LEO radiation environment. (upper left) Electron-collection signal before correction, and (lower left) after correction. (upper right) Hole-collection signal before correction, and (lower right) after correction pike2025characterizing. Our correction technique allows us to restore the spectral resolution of the hole-collecting signal to near pre-damage levels for the COSI detectors, even for this extreme level of radiation exposure. (The red dashed lines show the corrected peak positions for comparison between the spectra.)
• Figure 4: The COSI LV (hole-collecting) signal Cs-137 (661.7 keV) spectra during different stages of high-temperature annealing, compared with the pre-radiation damage spectrum. Annealing was performed after the COSI detector was exposed to an energetic proton fluence of $5.0\times10^8 p^+/cm^2$, corresponding to $\sim$10 years radiation exposure (worst case) in the COSI LEO radiation environment. Anneals 1 & 2 were performed at 80° C for a total of 72 hours and Anneals 3-7 were performed at 100° C for a total of 552 hours. The detector was cooled to 80K between anneals to test the spectral performance. The detector survived 7 annealing cycles, and the spectral resolution was ultimately restored to within 37% of the pre-radiation resolutionHAIGHT2025170538.