In-orbit Spectral Calibration Prospects for the COSI Space Telescope

TL;DR

The paper tackles the challenge of maintaining COSI’s spectral precision in orbit by exploiting background line emissions generated by the space radiation environment. Using MEGAlib/Geant4 Monte Carlo simulations, the authors identify activation lines from germanium and aluminum that span COSI’s 0.2–5 MeV band and develop the t10 calibration metric to quantify how quickly these lines yield reliable photopeak positions, enabling regular in‑orbit recalibration without an onboard radioactive source. They show that instrument‑level calibration can be accomplished on short timescales (instrument‑level on the order of hours) while detector‑level calibration requires longer integrations (tens of days), with gains and radiation damage tracked via line shifts and broadenings. The study also discusses telemetry constraints, potential extension to multi‑site events, and uncertainties in the background model, ultimately providing a practical calibration framework for COSI’s mission. The approach promises robust spectral performance and informs telemetry and housekeeping needs, supporting precise measurements of nuclear lines and positron annihilation in the Milky Way.

Abstract

The Compton Spectrometer and Imager is an upcoming NASA space telescope in the MeV range. COSI's primary science goals include precisely mapping nuclear line and positron annihilation emission in the Milky Way galaxy through Compton imaging. This relies on our ability to maintain COSI's spectral performance over its mission lifetime. Changes to the detectors' gain characteristics over time will result in a non-linear stretching of the entire energy range. Moreover, observations from past MeV telescopes and proton-beam experiments have shown that radiation damage in space causes photopeak shifts and spectral line broadening. These necessitate a plan for regular, in-orbit calibration. In this study, we demonstrate a method to monitor and recalibrate the COSI detectors using background line emissions produced by the space radiation environment. We employ Monte Carlo simulations of particle background and show that strong background lines arise from nuclear excitation of COSI's detectors (germanium) and cryostat (aluminum) materials. These span COSI's entire bandwidth for single-site interactions and can be used to monitor the effects of radiation damage and gain shifts every eight hours at the full instrument level and every 24 days at the individual detector level. Methods developed by Pike et al. to correct the effects of hole trapping and gain characteristics can then be applied to recover the original spectral performance. These results inform COSI's telemetry requirements for calibration and housekeeping data, and rule out the need for an on-board radioactive calibration source which would have increased the complexity of the spacecraft.

Figures (7)

• Figure 1: A model of the COSI instrument. The GeD array is enclosed in a vacuum cryostat (olive-yellow) and surrounded by bismuth germanate shields (orange, see Appendix A). The cryostat and shields are surrounded by eight flex circuits (bright yellow), and placed on top of a hexagonal, payload interface plate (light blue). The spacecraft bus is not shown here but has been included in our simulations.
• Figure 2: Top (above) and side view (below) of instrument background for single-site interactions. Each layer in the x-y plane contains four 8 cm $\times$ 8 cm $\times$ 1.5 cm GeDs. The topmost layer is pointed to outer space, and receives the highest photon count rate (primarily continuum background components). The count rates of GeDs at any given layer are uniform within 5%, and the count rates decrease with depth as fewer photons penetrate that deep. The detector array is enclosed in an aluminum cryostat, which is a source of increased count rates on the edges of the detectors. As this study is focused on monitoring and calibrating the instrument at the level of an individual GeD, all subsequent analyses use data integrated over a GeD module.
• Figure 3: Single-site event rates for the simulated background models. Multiple activation lines are visible and are marked in red. The list of candidate lines selected from this study are marked in green and span the entire energy range. Most of these lines are the result of activating the GeD material or the aluminum cryostat. The spectrum is truncated at 1800 keV -- the saturation energy for single-site interactions used in our simulations.
• Figure 4: Fitting the 1369 keV, 7-day integrated data with a Gaussian+line model, along with their 99.7% confidence intervals. The peak and continuum counts on the top right and the SNR value on the top left are calculated using the counts within $\pm 1.4\sigma_e$ of the photopeak.
• Figure 5: The evolution of the SNR value with time for various activation lines is shown here, with photon data integrated over the entire instrument. The horizontal line denotes SNR=10 and the integration time required to cross this horizontal line is defined as the t10 value. The slope is $1/2$ as expected from Equation \ref{['eq:SNR']}.