Speaker
Description
Gamma-ray bursts (GRBs) are the most luminous explosions in the Universe, releasing an enormous amount of energy on the order of $10^{52–54}$ erg lasting from tens of milliseconds to several hundred seconds, and are regarded as one of the most powerful probes for exploring the early Universe. HiZ-GUNDAM is a proposed future satellite mission aimed at investigating the early Universe (z > 7) using GRBs as probes, and consists of the wide-field X-ray monitor EAGLE and the near-infrared telescope MONSTER. EAGLE is an X-ray detector integrating Lobster Eye optics with pnCCD imaging sensors. It surveys a wide field of view (0.53 sr) in the soft X-ray band (0.4 – 4 keV) and determines the positions of detected GRBs with a localization accuracy of a few arcminutes.
In this study, we developed in-house pnCCD driving and readout electronics intended for satellite deployment. Using a pnCCD imaging sensor with a pixel size of 132 µm, an active area of 9.6 × 19.2 mm$^2$, and a depletion depth of 450 µm, we conducted X-ray imaging demonstrations using an Fe-55 radioactive source. The system consists of four types of custom-developed electronic boards, is designed to be lightweight and compact for spacecraft integration. Furthermore, our system incorporates an on-board function for autonomous real-time X-ray event extraction. In orbit, however, charged particles trapped by the geomagnetic field can produce background events in the pnCCD image sensors, which may result in false GRB alerts. In areas of intense charged-particle flux, such as the polar areas and the South Atlantic Anomaly, observations may be interrupted; however, to maximize observational efficiency, it is critical to efficiently remove charged-particle events while staying within the computational resources available for on-board processing. We evaluate and compare the background rejection performance and on-board implementation feasibility of multiple removal algorithms using X-ray data from an Fe-55 source and electron-event data obtained from an Sr-90 β-source. Preliminary results show that, under current test conditions, applying a machine-learning-based approach reduces the residual background fraction to about 3 %, indicating an improvement compared to conventional methods.
