Beam Background Monitoring and Characterization of CMOS Pixel Detectors for Belle II at SuperKEKB

Sumitted to PubDB: 2026-07-07

Category: Phd Thesis

Tags: Upgrade

Principal Authors Yannik Buch
Date 2026-06-19
Belle II Number BELLE2-PTHESIS-2026-023
Abstract The SuperKEKB storage ring at KEK in Japan aims to deliver an instantaneous luminosity of up to $6 \times 10^{35} \, \mathrm{cm^{-2}\,s^{-1}}$. To achieve this, a redesign of the interaction region will be necessary. The projected beam backgrounds that the Belle II experiment will experience are substantial. The Belle II experiment plans an upgrade of its vertex detector to cope with the increasing hit rates and accommodate the changing geometry of the interaction region. The proposed vertex detector (VTX) consists of 5 fully pixelated barrel-shaped layers using the specifically developed OBELIX sensor. The OBELIX sensor is currently in the design stage and is derived from the TJ-Monopix2 sensor. Test beam results of irradiated TJ-Monopix2 chips in a temperature controlled environment to test the feasibility of operation at room temperature are presented in this thesis. The results will impact the thermal design and choice of front-end for the final design of OBELIX. Further, the results will provide a reference for the expected impact of irradiation on the performance of OBELIX. Another important task is the mitigation of beam backgrounds. To support this effort, a diagnostic beam background monitoring tool, named BGNet, was developed. It is a neural network for predicting the beam background decomposition of the hit rate of Belle II sub-detectors based on the state of SuperKEKB. The neural network is trained using $1\,$Hz time series of diagnostic variables from the SuperKEKB slow control system as input features and measured hit rates of Belle II sub-detectors as regression targets. This thesis presents performance evaluations of BGNet that were performed for its commissioning. Further, a real-time application and a chat bot were developed to increase the accessibility and applicability of BGNet predictions.
Institute Goettingen

Files