The biomedical world is seeking imaging devices that are compact, operative at room-temperature and of high spatial resolution to enhance the localization and detection of living cells. This includes applications such as the detection of cancer cells among healthy tissue. To this end, CMOS chips containing arrays of single-photon avalanche diodes (SPADs) are a promising subject with excellent potential to drive advancements in the field of bio-sensing. In this work, we present a 1.32 x 1.32 mm imaging chip consisting of a 16 x 16 SPAD-pixel array and a 32 pin I/O ring. The SPAD-pixels are organized into rows, with each SPAD-row connected to a single I/O pin. To output all the data from each SPAD-pixel in a row serially, a parallel in, serial out (PISO) shift register is integrated into the SPAD-row. Each SPAD-pixel consists of a photon-detecting SPAD-cell, a voltage controlled resistor (VCR) and its complementary circuit. This circuitry includes an active quenching and recharge circuit, a counter-based externally configurable hold-off mechanism and a count register with a Schmitt-trigger capable of counting up to 1.0·10⁶ detected photons over an 1-s integration time. In addition, the SPAD-array architecture is scalable, requiring minimal adjustments to expand to a larger format. The chip is developed for applications in quantum diamond sensing, a promising method for high-resolution magnetic field imaging. This work was conducted in the context of the Bachelor Graduation Project at the TU Delft Faculty of Electrical Engineering, Mathematics and Computer Science, in collaboration with the Quantum Integration Technology (QIT) group.

Nitrogen-Vacancy (NV) color centers in diamond are a promising and powerful tech-
nology for sensing small magnetic fields in the order of μT. One of its applications is
the use in magnetic biosensing to identify cells marked with magnetic nanoparticles.
These nanoparticles can target cells linked to malignant tissue, which can subsequently
be identified with great spatial resolution using the NV centers. Therefore promising a
less invasive procedure and more accurate identification. Such a system using NV cen-
ters as detectors requires small and effective photon detection hardware in order to be
used in practice. In this bachelor graduation project Single Photon Avalanche Diodes
(SPADs) are characterized, simulated and validated, to be integrated in a SPAD imaging
sensor designed for magnetic bioimaging using NV centers.
A 3D model of the SPAD is created in TCAD Sentaurus to verify whether the designed
SPAD performs as intended. To this end, the I-V and C-V curves are simulated. These
graphs show that a proposed SPAD design works as intended whilst the previously im-
plemented design does not. Furthermore, the simulation results can later be compared
to the behavior of a physical SPAD. During the physical analysis of the stray SPADs on
a previous IC, a design flaw was identified. To verify this a test circuit was built and
the device was retested. The results confirm an error in the SPAD design. The SPAD is
missing an N-well resulting in standard diode behavior and preventing it from detecting
photons. Furthermore, to enhance single-photon detection, we explored a new analog-
to-digital converter (ADC) circuit that employs a Schmitt trigger in place of a standard
comparator. Our results show that this ADC delivers predictable behavior and achieves
a more precise and efficient conversion to a digital signal.

In this work, we aim to characterize the Rabi oscillations of nitrogen-vacancy (NV) centers in diamond by implementing and utilizing a tailored experimental setup, where optical pulses are produced by an acousto-optic modulator (AOM). A central focus of this thesis was to design and construct a custom AOM around a TeO2 crystal. This was a necessity, because the commercial AOM failed during one of the experiments. The work that has been done involved extensive characterization of precise and synchronized timing. High-quality Rabi oscillations could not be measured due to time limitations and the breaking down of the commercial AOM. However, the thesis shows full control over pulse-sequencing, optical microwaves and NV excitation. This work provides a foundation for further spin dynamic experiments of NV centers. During the thesis software for controlled timing has been built, and the designing and integration of custom made electronics has been done, which can provide practical guidance for controlling quantum instrumentation devices. We highlight the engineering challenges in building quantum instrumentation from the ground up.

Demand for high density, high bandwidth, and low power Integrated Circuits (ICs) is rapidly increasing even as Moore's Law is starting to plateau. Among the new wave of technologies that are meant to continue the prevalent trend of semiconductor device scaling, 3D System Integration promises many advantages over traditional single-planar designs. This technology enables designers to stack multiple dies and unlock new functionality by reducing the footprint of the final device package. Using 3D Integration, multiple different types of chiplets (Digital IC, Analog IC, MEMS) can be integrated into a single package, potentially bypassing an expensive move to a newer process node and unlocking even more functionality from the same package.

These downscaling issues are not limited to traditional CMOS technologies but extend to Quantum Computers. Due to the highly heterogeneous nature of Quantum ICs, and especially their requirement of low qubit decoherence noise, we require high-density connections from the Qubit layer to the Microelectronics Control layer while maintaining appropriate spacial separation between the two. In this project, the qubit layer is comprised of Diamond Colour Centers in a Photonic IC with other optical components such as SNSPDs, and MEMS switches. On the other hand, the microelectronics control layer is a Cryogenic CMOS chip.

The main goal of this project is to design an Interposer and related technologies like Through Silicon Vias (TSVs) and Microbumps to successfully integrate these two chiplets in a single 3D assembly. We conduct thermal analysis of multiple 3D assembly designs using Finite Element Modelling to derive optimum fabrication specifications. Subsequently, fabrication recipes are developed and optimised for TSV etching, coating, and patterning with superconductive sidewall liners. Recipes are also developed to fabricate Indium Microbumps using techniques like evaporation, liftoff, and atmospheric reflow. Utilizing these microbumps, we conduct cold-compression flip-chip experiments. Finally, we establish cryogenic measurement setups to electrically characterize these fabricated devices.

Diamond color centers have been increasingly gaining attention in research due to their desirable optical properties that make them suitable for applications such as quantum computing, quantum sensing and quantum communication. However, the main challenge has been on-chip integration of diamond color centers with high scalability and high yield. Many diamond integration methods have been researched, such as heterogeneous wafer bonding of diamond-on-insulator photonics. While this method offers high scalability and wafer level fabrication, optical performance is not guaranteed due to the non-ideal processing which results in lower yield. Another integration method is pick and place, which demonstrated high yield as the diamond color centers can be pre-selected before integration, and thus resulting in higher yield. Here, pick and place integration of diamond nano-photonic structures is presented. For that purpose, an adiabatic coupler is designed, with simulation results showing 0.36 dB insertion loss and 3dB misalignment tolerance of 180 nm and 5 um in x and z directions respectively. A mechanical support structure is designed such that optical losses at that interface are minimized. The simulated loss is reduced from 33 % to only 4 % and the cross-talk is -11 dB. Further optical measurement is needed to determine the achieved transmission of the integrated diamond chiplets. Finally, pick and place is performed with 6 different supporting structure designs. The yield is 25 % with minimum misalignment of 50 nm which is mainly dominated by random motion of the diamond chiplet as it gets closer to the receptor chip. This random motion is attributed to the surface roughness of the receptor chip and and lower surface of the diamond chiplet. In addition, residual charges in both chips results in Coulomb forces acting on the diamond chiplet and thus contributing to the misalignment. It is expected that the magnitude of the random motion, and consequently, the misalignment, will be improved by treating the receptor chip surface with HMDS and de-ionizing both the diamond chiplet and the receptor chip.