Events Calendar

ECE PhD Dissertation Defense: Deniz Onural

Title: CMOS Electronic-photonic Integrated Circuits For Efficient Data Links And High Precision Sensors

Presenter: Deniz Onural

Advisor: Professor Miloš Popović.

Chair: Professor Roscoe Giles

Committee: Professor Miloš Popović, Professor Rabia Yazıcıgil Kirby, Professor Abdoulaye Ndao, Professor Anna Swan

Google Scholar Link: https://scholar.google.com/citations?user=c3K7hA8AAAAJ&hl=en

Abstract: The growing demand for data-centric applications, such as deep learning and AI, is putting a strain on interconnects in high-performance computing systems. As Moore's Law approaches its limits, silicon photonic optical interconnects offer a scalable solution, leveraging advanced CMOS infrastructures. The high index contrast between silicon and silicon dioxide allows telecom wavelengths to develop micron-scale devices monolithically integrated with transistors, enabling complex electronic-photonic systems-on-a-chip (EPSOC). This thesis highlights high-performing EPSOC building blocks that enhance quantum supercomputers and millimeter-wave communication systems through advances in silicon photonics.

A key component of these systems is the electro-optic modulator. This dissertation employs 45 nm CMOS silicon-on-insulator (SOI) platforms to fabricate high-performance photonic devices, meeting stringent energy and bandwidth requirements by exploring modulator variations with high-swing and high-linearity drivers. A significant achievement is the improved sensitivity of a spoked-ring modulator, using a vertical p-n junction with shallow doping. For cryogenic egress systems, these modulators are assembled in a transmitter operating at 4 Kelvin, linking superconducting logic with single-flux quantum (SFQ) signals to DRAM memory at room temperature. Creating efficient optical cryogenic egress links is challenging due to low millivolt-level outputs from superconducting circuits, necessitating modulators with exceptional sensitivity. This work demonstrates the cryogenic egress link with monolithically integrated amplifiers less than 10 μm from a high-sensitivity p-n junction modulator, transducing millivolt-level NRZ data with minimal parasitic capacitance.

Acknowledging silicon's material limitations and the potential to surpass them in bandwidth density and energy per bit, this dissertation explores integrating high electro-optic nonlinearity materials with silicon. By designing and demonstrating silicon-hybrid devices using organic materials with the Pockels effect, significant modulator performance improvements are targeted. Methods are developed to integrate materials while preserving a full stack of back-end-of-the-line (BEOL) metal levels, enabling silicon-hybrid photonics with complex digital systems in CMOS. These systems can be packaged with high-density PCBs or organic substrates via flip-chip attachment. Demonstrations focus on using the Pockels effect for modulation, leveraging electro-optic polymers compatible with cryogenic temperatures, and a new class of electro-optic materials, ferroelectric nematic liquid crystals.

The photonic device advancements in this thesis, including low-loss waveguides, high-performance modulators, high Q-factor resonators, and narrow-bandwidth optical filters, enable EPSOCs with higher sensitivity, energy efficiency, communication bandwidth, and signal integrity for diverse applications. These device-level innovations address system-level challenges, showcasing the vast potential of silicon photonics in spectroscopic sensing, environmental monitoring, biosensing, and quantum information processing.