Characterisation and simulation of 65 nm CMOS imaging process Monolithic Active Pixel silicon Sensors for the ALICE ITS3 Upgrade

Abstract Part I The ALICE experiment at the CERN Large Hadron Collider (LHC) is designed to study strongly interacting matter, in particular the quark-gluon plasma (QGP), created at the extreme temperatures achieved in high-energy heavy-ion collisions. During the LHC Long Shutdown 2 (2018-2022), ALICE underwent a major upgrade of several of its detectors, enabling the reconstruction of rare physics channels that were previously inaccessible. The upgraded Inner Tracking System (ITS2) plays a central role in ALICE, providing the precise reconstruction of primary and secondary decay vertices and enabling accurate particle tracking close to the interaction point. The ITS2 consists of seven concentric layers of Monolithic Active Pixel Sensors (MAPS), with a total of roughly 12.5 billion pixels covering an active area of about $10~\mathrm{m^2}$, representing the largest application of MAPS ever realised. During the next LHC long shutdown (2026-2029), the ITS2 is scheduled to undergo an additional upgrade in preparation for LHC Run 4 (2029-2032), replacing its three innermost layers with a new fully cylindrical vertex detector, the ITS3. Such an innovative tracker will consist of two half-barrels, each featuring three self-supporting semi-cylindrical layers built from large-area, ultra-thin (50~$\mu$m) bent MAPS fabricated in a 65 nm CMOS process. By minimizing mechanical supports and eliminating cooling structures and material for power and data transmission, the ITS3 will reach an unprecedentedly low material budget of 0.09\%~X$_0$ per layer. Together with a reduced beam-pipe radius placing the innermost layer at 19~mm from the interaction point, this design will significantly improve pointing resolution and tracking efficiency, especially for low-momentum particles. This will enhance precision measurements in the heavy-flavour sector, enabling the reconstruction of B$^0_s$, $\Lambda^0_b$, non-prompt D$^+_s$ and $\Xi^+_c$ decays, improving the study of low-mass di-electrons by reducing background, and potentially allowing for the direct tracking of charged strange baryons to reconstruct weakly decaying strange particles with novel techniques. The wafer-scale sensors developed for ITS3 represents a major technological breakthrough in detector technology, made possible by stitched-circuit manufacturing and the 65~nm CMOS. This thesis is part of the extensive characterisation campaign which has been planned for the ITS3 sensor prototypes. Specifically it is focused on the full characterisation of the Digital Pixel Test Structure (DPTS), designed to validate the 65~nm CMOS process, and the MOnolithic Stitched Sensor (MOSS), developed to demonstrate the stitching technology and evaluate the performance of different pixel and logic solutions in view of the final sensor design. This thesis is structured as follows. The physics of heavy-ion collisions and the signatures of the quark-gluon plasma are first introduced, followed by a description of the ALICE apparatus, with particular emphasis on the ITS2 characteristics and the motivations for its upgrade. In particular, the expected performance improvements in the measurements of $\Lambda_c$ and thermal di-electrons are presented to illustrate the impact of the ITS3 upgrade. A description of the ITS3 layout follows, including a general overview of the MAPS sensor and a detailed presentation of the prototypes characterised in this work. The main characterisation tests are discussed, and the procedure used to determine the optimal operating conditions of the sensors is outlined. Finally, results from in-beam measurements are presented to demonstrate the radiation tolerance of the ITS3 prototypes.

Abstract Part II An extensive study of the noise sources in the prototypes was carried out to identify their origin and, where possible, reduce them in the final sensor. In fact, electronic noise, defined as the random output signal in the absence of incident particles, sets the lower bound for the detectable signal and affects both the detection efficiency and the energy resolution. Therefore, a thorough understanding of its underlying sources is crucial to minimise it and optimise the sensor performance. A thermal model was proposed to explain part of the observed noise, and additional frequency-dependent contributions were identified, one of which will be addressed in the final design. The energy response of the DPTS was investigated using X-rays of well-defined energy. The measured spectra were analysed by evaluating the peak characteristics, such as the mean value and the energy resolution. This comprehensive study, including energy calibration up to high energies, provides both a strengthened characterisation of the prototypes for the ITS3 and an evaluation of their potential for applications beyond their intended use. Finally, TCAD simulations were performed to model the electric field inside the pixels, analyse the charge collection properties of the ITS3 sensor, and explore its potential for energy-loss measurements using Garfield simulations. This approach also provided a validation of the results previously obtained in energy-loss studies.