Interest in nanophotonics continues to grow as integrated optics provides an affordable platform for areas like telecommunications, quantum information processing, and biosensing. Designing and characterizing integrated photonics components and circuits, however, remains a major bottleneck. This is especially true when complex circuits or devices are required to study a particular phenomenon.To address this challenge, this work develops and experimentally validates a novel machine learning design framework for nanophotonic devices that is both practical and intuitive. As case studies, artificial neural networks are trained to model strip waveguides, integrated chirped Bragg gratings, and microring resonators using a small number of simple input and output parameters relevant to designers. Once trained, the models significantly decrease the computational cost relative to traditional design methodologies. To illustrate the power of the new design paradigm, both forward and inverse design tools enabled by the new design paradigm are demonstrated. These tools are directly used to design and fabricate several integrated Bragg grating devices and ring resonator filters. The method's predictions match the experimental measurements well and do not require any post-fabrication training adjustments.

Computing as a service is rapidly becoming the new normal for many sectors of the economy. The widespread availability of broadband internet has allowed an extensive range of services to be delivered on-demand from centralized computing systems known as ‘data centers’. These systems have evolved to be enormously complex. Optical-based communication is desired to increase data center capability and efficiency, however traditional optical technologies are not feasible due to cost and size. Silicon photonics aims to deliver optical communications on an integrated and affordable platform for use in data centers by leveraging the existing capabilities of complementary metal-oxide semiconductor manufacturing. This thesis contains a description of the development of monolithic silicon photodiodes for use in photonic integrated circuits in, and beyond, the current telecommunications wavelength windows. The focus is on methods which are compatible with standard silicon processing techniques. This is in contrast to the current approaches which rely on hybrid material systems that increase fabrication complexity. Chapter 1 and 2 provide background information to place this work into context. Chapter 3 presents an experimental study of resonant devices with lattice defects which determines the refractive index change in silicon-on-insulator waveguides. High-speed operation of resonant photodiodes is demonstrated and is found to be limited by resonance instability. Chapter 4 demonstrates high responsivity avalanche photodetectors using lattice defects. The detectors are shown to operate error-free at 10 Gbit/s, thus confirming their capability for optical interconnects. Chapter 5 presents photodiodes operating with absorption through surface-state defects. These detectors show fast operation (10 Gbit/s) and have an extremely simple fabrication process. Chapter 6 demonstrates photodiodes operating beyond the traditional telecommunications window. Operation at 20 Gbit/s, at a wavelength of 1.96 µm is demonstrated, offering potential for their use in the next generation of optical communication systems which will exploit the thulium doped fiber amplifier. / Thesis / Doctor of Philosophy (PhD) / This thesis describes photodiodes constructed on silicon optical waveguides. The photodiodes are notable for their high-speed performance and simple fabrication methods. Such devices may find use within chip-integrated optical transceivers, which are desired for optical interconnects within large-scale computing systems such as data centers.

No description available.

No description available.

No description available.

<p>The continuous reduction of feature size in silicon-based electronic integrated circuits (ICs) is accompanied by devastating propagation delay time and power consumption that have become known as the “Interconnect Bottleneck”. Optical interconnection is a proposed solution that is poised to revolutionize the data transmission both within and between ICs. By forming the optical transmission and functional elements from silicon, they can be monolithically incorporated with standard ICs using the established CMOS (Complementary Metal Oxide Semiconductor) infrastructure with minimal incremental cost. A key required functional element is the photodetector, which provides optical-toelectrical conversion of signals. In this thesis, a method of achieving such conversion is explored, which uses the optical absorption at 1550 nm wavelengths provided by lattice defects. The physics governing defect-enhanced silicon waveguide photodiode operation is described, and a device model is used to verify the posited detection process and propose design improvements. The model was used to design a novel photodetector structure using a waveguide formed by the LOCOS (LOCal Oxidation of Silicon) process with a poly-silicon self-aligned contact. The fabricated device exhibited a responsivity of 47 mA/W, providing an improvement over previous devices of similar dimensions, although were ultimately limited by the quality of the poly-silicon/silicon interface. A sub-micron waveguide photodiode fabrication process using electron-beam lithography was developed, which produced photodiodes with responsivities of 490 mA/W. This process was used to integrate photodiodes onto micro-ring resonators, which exhibit resonant enhanced photocurrent. The physics of this enhancement were explored, and found to produce a 50 μm long resonant photodiode of responsivity equal to that of a 3 mm long non-resonant photodiode. Lastly, the integration of such sub-micron photodiodes as functioning power monitors throughout photonic circuits was demonstrated as a means to characterize and tune micro-rings during operation.</p> / Doctor of Philosophy (PhD)

<p>The development of silicon photonics, driven by the increasing demand for bandwidth from data centre applications, is receiving growing attention. As a result of the indirect bandgap of Si material, it is more practical to heterogeneously incorporate the laser source than fabricate directly on Si. Of all the approaches, an external cavity laser (ECL) approach which consists of III-V gain material and Si photonic integrated circuit (SiPIC), is a flexible and cost effective solution. This thesis captures theoretical and experimental work on the design of SiPIC ECLs. In addition, a four wavelength laser source using an SiPIC ECL scheme is proposed and studied.</p> <p>The theoretical tool is first introduced on the traveling wave model (TWM) and it is numerically solved with FDTD in Matlab. A digital filter approach is used to describe the feedback from an SiPIC external cavity, where the phase delay of the digital filter is investigated and utilized to set the cavity length.</p> <p>The III-V gain chip and SiPIC are then examined separately for their characterization, along with the coupling and feedback requirements in an ECL design.</p> <p>Lastly, experiments are conducted to demonstrate the feasibility of four wavelength ECLs and SiPIC ECLs.</p> / Master of Applied Science (MASc)

The ever-increasing bandwidth requirements to support telecommunications infrastructure necessitates large-scale fabrication of low-cost and scalable silicon photonic integrated circuits. Wavelength-division multiplexing (WDM) schemes are fundamentally limited in the number of channels supported in long-haul transmission by the erbium doped fiber amplifier (EDFA). To address this, researchers have turned focus toward the thulium doped fiber amplifier (TDFA), which provides 3× more bandwidth. This thesis describes the development of high-speed silicon-on-insulator (SOI) micro-ring resonator (MRR) modulators optimized for wavelengths in the TDFA band. Chapter 2 presents a theoretical performance comparison between MRR modulators designed for optimized use at EDFA and TDFA wavelengths. Chapter 3 presents an experimental study of optical loss mechanisms at extended wavelengths which suggests reduced waveguide scattering and enhanced divacancy defect absorption as well as larger bending and substrate leakage losses when compared with shorter wavelengths. An electronic variable optical attenuator is characterized in Chapter 4 to experimentally verify the predicted 1.7× TDFA-band free-carrier effect enhancement over EDFA-band wavelengths. The first steady-state operation of an MMR modulator near a central wavelength of 1.97 µm is also demonstrated under the enhanced free-carrier effect. Chapter 5 demonstrates the first high-speed reverse bias operation of an MRR modulator with a measured bandwidth of 12.5 GHz, and an on-chip optical link consisting of a modulator followed by a defectmediated detector with open eye-diagrams up to data rates of 12.5 Gbps. Chapter 6 introduces an electrically-driven post-fabrication defect-assisted resonance trimming technique via local annealing for use in MRR devices. Chapter 7 presents a Monte Carlo simulation of resonance alignment in multi-MRR systems subjected to spatially-correlated wafer variation created through the Virtual Wafer Model process to predict thermal power consumption and power reduction through resonance trimming. / Thesis / Doctor of Philosophy (PhD)

This research introduces tellurium oxide (TeO2) glass doped with optically active erbium ions (Er3+) as an active oxide cladding material for silicon-on-insulator (SOI) waveguides for realization of a silicon-based erbium-doped waveguide amplifier (EDWA) for integrated optics. Optical amplification of this nature is enabled by energy transitions, such as stimulated absorption and emission, within the shielded 4f shell of the rare-earth atomic structure caused by excitation from photons incident on the system. Er3+ ions are doped into the TeO2 film during deposition onto the SOI waveguides using a reactive magnetron co-sputtering system operated by McMaster’s Centre for Emerging Device Technologies (CEDT). Prior to fabrication, the waveguides are designed using photonic CAD software packages, for optimization of the modal behaviour in the device, and Matlab, for characterization of the optical gain performance through numerical analysis of the rate and propagation equations of the Er3+-based energy system. Post fabrication, the waveguide loss and gain of the coated devices are experimentally measured. The fabricated waveguide amplifier produces a peak signal enhancement of 3.84 dB at 1533 nm wavelength for a 1.7 cm-long waveguide device. High measured waveguide losses (> 10 dB/cm) produce a negative internal net gain per unit length. However, the demonstration and implementation of an active rare-earth doped cladding material on a silicon waveguide is successful, which is a major step in developing integrated optical amplifiers for conventional silicon photonics platforms. / Thesis / Master of Applied Science (MASc)

Advancements in photonic integrated circuits, so-called PICs, have progressed fast in the last decades. More complex PICs are getting developed, which are promising in possibly offering advantages like low power consumption and high-performance computing. Re-programmable photonic FPGAs are one of these candidates. To make these PICs viable, fundamental building blocks based on photonics need to be developed. Some of those fundamental building blocks are tunable silicon ring resonators, which can be used to filter signals in the transmission of light through photonic circuits. Fabrication of PICs is developing and those components are getting smaller, which leads to a strong sensitivity of their behavior to nanometer-scale variations. That has created a need for active tuning of those devices to recuperate those variances. One promising way to tune silicon ring resonator devices is to integrate microelectromechanical systems (MEMS) into the tuning section of the devices, because of their local and low power actuation. They are prospective to eliminate drawbacks from usual actuation methods like thermal actuation, which comes with high power consumption and cross talk while heating the functional sections of the ring. In this thesis, we have measured and analyzed MEMS-tunable silicon ring resonators, featuring two different designs, being an all-pass ring resonator and an add-drop ring resonator. The MEMS in the design are used to change the gap between the waveguides in their directional coupler and phase shifter section to control the position and extinction ratios of the ring resonance dips, which has been successfully demonstrated for the all-pass ring resonator. For the add-drop ring resonators, we have obtained performance parameters of their resonances with an average Q-factor of 3000 over the measured wavelength ranged from 1460nm to 1580nm and the characteristic behavior of their transmission has been shown without actuation. Further investigation with MEMS actuation of add-drop ring resonators and passive measurements on all-pass ring resonators can be done for a better understanding of their behavior and functionality. This can be achieved by characterizing all-pass ring resonators in terms of obtained performance parameters and by active measurements on add-drop ring resonators, as we expect that their MEMS could enable similar functionalities as all-pass ring resonators. Our first characterization results confirm the potential of MEMS for ring resonator tuning and could enable future circuits based on ring resonators with low power consumption. / Framsteg inom fotoniska integrerade kretsar, så kallade PIC, har utvecklats snabbt under de senaste decennierna. Mer komplexa PIC utvecklas, vilket lovar att möjligen erbjuda fördelar som låg strömförbrukning och högpresterande datorer. Omprogrammerbara fotoniska FPGA är en av dessa kandidater. För att göra dessa PICs livskraftiga måste grundläggande byggstenar baserade på fotonik utvecklas. Några av dessa grundläggande byggstenar är avstämningsbara kiselringresonatorer, som kan användas för att filtrera signaler vid överföring av ljus genom fotoniska kretsar. Tillverkning av PIC utvecklas och dessa komponenter blir mindre, vilket leder till en stark känslighet för variationer, även på nanometer skala. Det har skapat ett behov av aktiv inställning av dessa enheter för att återhämta dessa avvikelser. Ett lovande sätt att ställa in kiselringresonatoranordningar är att integrera mikroelektromekaniska system (MEMS) i enhetens stämningsdel på grund av deras lokala och lågeffektaktivering. De kan eliminera nackdelar med vanliga manövreringsmetoder som termisk aktivering, som kommer med hög strömförbrukning och termisk överhöring. I denna avhandling har vi mätt och analyserat MEMS-avstämbara kiselringresonatorer, med två olika designer, som är en all-pass ringres-onator och en add-drop ringresonator. MEMS i konstruktionen används för att ändra gapet mellan vågledarna i deras kopplare och fasskiftarsektion för att styra positionen och djupet på ringresonaserna, vilket har visats framgångsrikt för allpassningsresonatorn. För add-dropringresonatorer har vi erhållit prestandaparametrar för deras resonanser med en genomsnittlig Q-faktor på 3000 över den uppmätta våglängden som varierar från 1460 nm till 1580 nm och det karakteristiska beteendet för deras överföring har visats utan aktivering. Ytterligare undersökning med MEMS-aktivering av add-drop-ringresonatorer och passiva mätningar på all-pass-ringresonatorer kan göras för en bättre förståelse av deras beteende och funktionalitet. Detta kan uppnås genom att karakterisera allpassningsresonatorer i termer av erhållna prestandaparametrar och genom aktiva mätningar på add-drop-ringresonatorer, eftersom vi förväntar oss att deras MEMS kan möjliggöra liknande funktioner som all-pass-ringresonatorer. Våra första karakteriseringsresultat bekräftar MEMS potential för ringresonatorinställning och kan möjliggöra framtida kretsar baserade på ringresonatorer med låg strömförbrukning.