ADVANCED CMOS SINGLE-PHOTON IMAGER ARCHITECTURES FOR QUANTUM AND SCIENTIFIC IMAGING APPLICATIONS

This thesis presents the design, modeling, and characterization of advanced Single- Photon Avalanche Diode (SPAD) imager architectures, specifically tailored for quantum and scientific imaging applications. While SPADs offer unmatched picosecond-scale tim- ing precision and single-photon sensitivity, their integration into large-scale arrays has traditionally been hindered by massive data bandwidth requirements and limited in-pixel functionality. To address these bottlenecks, this research introduces three primary innovations in three different fields targeting Quantum Ghost Imaging (QGI), high-speed scientific imaging, and Flash-LiDAR for space applications. First, for QGI, the "Looking Back" mechanism is proposed and implemented in the two sensor prototypes, namely Casper and Slimer sensors. By integrating in-pixel electrical delay lines and asynchronous cor- relation circuits, these sensors compensate for optical path delays and perform real-time coincidence detection, enabling a resolution increase of nearly two orders of magnitude while reducing acquisition times by over 10× compared to traditional scanning systems. Second, a hardware-friendly on-chip compression scheme is developed for high-speed imaging. Leveraging temporal and spatial redundancies through a shot-noise-based Dif- ferential Pulse Code Modulation (DPCM) technique and cluster-level suppression, the architecture achieves compression ratios up to 96% while maintaining high image fi- delity (PSNR 29 dB). Finally, two Flash-LiDAR sensor prototypes , namely Wallie64 and Wallie256 sensors are designed for space-grade applications, including landing, ren- dezvous and target approaching operations. These sensors incorporate reconfigurable Time-to-Digital Converters (TDCs) and a distributed digital Silicon PhotoMultiplier (d2SiPM) mechanism to ensure robust 3D imaging in harsh, high-ambient-light envi- ronments. Collectively, these contributions demonstrate a shift toward "smart" focal plane arrays that alleviate back-end computational burdens and enable next-generation scientific and autonomous applications.