New space-based tools for GRB detection in the MeV energy range

Gamma-ray bursts (GRBs) are among the most energetic transient phenomena in the universe. Despite their scientific importance, the MeV domain remains the least explored region of the electromagnetic spectrum. This thesis addresses this challenge through the performance evaluation of two compact space-based detectors: the CALOg calorimeter of the NUSES/Zirè instrument and the Crystal Eye all-sky gamma-ray monitor. Advanced inorganic scintillators are central to both instruments. The intrinsic radioactive background of cerium-doped LYSO, arising from the natural decay of 176Lu, is characterized through Geant4 Monte Carlo simulations, revealing a persistent internal background spectrum with a pronounced feature near 600 keV. A comparative effective area analysis demonstrates that GAGG and LYSO yield nearly equivalent detection performance across the full energy range, motivating the adoption of cerium-doped GAGG as the reference scintillator for CALOg on account of its intrinsic radiopurity and absence of self-activity. First, the performance of the CALOg sub-detector is evaluated through dedicated Geant4 Monte Carlo simulations. Effective area, orbital background spectra, and transient sensitivity are assessed primarily under a 98° orbital inclination. The performance of a sliding-window transient trigger algorithm is evaluated against a simulated average GRB flux through a signal-to-noise analysis applied to a population of 2300 Fermi-GBM GRBs. A boosted decision tree classifier is implemented to investigate the feasibility of separating astrophysical gamma-ray events from intrinsic LYSO background, confirming that material-driven background limitations cannot be reliably mitigated through machine learning alone. The Crystal Eye detector is evaluated across four geometrical configurations and three scintillator material compositions. The hybrid GAGG+LYSO configuration, with GAGG in the upper detector layer and LYSO in the lower layer, is identified as the optimal material choice, reducing the intrinsic background by nearly two orders of magnitude relative to the LYSO-only baseline while preserving detection efficiency. The optimized geometry achieves a minimum detectable flux of approximately 3 × 10⁻⁴ ph s⁻¹ cm⁻² keV⁻¹ in the 400–600 keV range and a source localization accuracy of r95 ≈ 2.75° using a template-based Kolmogorov–Smirnov reconstruction pipeline, with the goal to deliver rapid sky localizations suitable for multi-messenger follow-up.