Silicon has proven to be one of the materials of choice for many integrated photonic applications. However, silicon photonics is limited by certain material shortcomings. Two shortcomings addressed in this work are zero second-order optical nonlinearity, and the lack of methods available to achieve broadband polarization diversity. Heterogeneous integrated solutions for these shortcomings of silicon photonics are presented in this work. First, nonlinear frequency conversion is demonstrated with thin-film lithium niobate on silicon substrates. The method for reaching the highest-achieved second-harmonic generation conversion efficiency, using active monitoring during periodic poling, is discussed. Additionally, a cascaded approach for generating higher-order harmonics is presented, along with a theoretical model to extract conversion efficiencies from measurements performed with pulsed sources. Initial work to integrate second-order and third-order nonlinearities together using thin-film lithium niobate and chalcogenide is also presented. Second, a spatially-mapped anisotropic material platform that exhibits broadband polarization diversity is discussed. This platform currently demonstrates polarization beam splitters, and polarization-selective beam taps and microring resonators, whose results are presented. Also discussed is a method to include polarization rotators to demonstrate full polarization diversity, as well as designs and initial work to expand the platform to operate at longer wavelengths, specifically those in the telecom band.

As a promising next-generation display, augmented reality (AR) and virtual reality (VR) have shown attractive features and attracted broad interests from both academia and industry. Currently, these near-eye displays (NEDs) have enabled numerous applications, ranging from education, medical, entertainment, to engineering, with the help of compact and functional patterned liquid crystal (LC) devices. The interplay between LC patterns and NEDs stimulates the development of novel LC devices with unique surface alignments and volume structures, which in turn feedback to achieve more compact and versatile NEDs. This dissertation will focus on the patterned LC with applications in NEDs. Firstly, we propose and explain the working principles and generation of novel patterned LC devices, including LC configurations, surface alignment mechanism, polarization field generation, and fabrication process. Secondly, we theoretically analyze the optical properties of patterned LC devices, providing the optical efficiency, devices thickness, polarization selectivity, wavelength, and angular bandwidth. Based on the dimensions of the surface pattern, the LC devices can be divided into reflector, grating, and lens, respectively. Finally, we focus on the applications of these novel patterned LC devices to address some challenges in current NEDs. More specifically, achieving a high-resolution density in NEDs, especially for VR systems is an urgent issue. To enhance the resolution without introducing any extra burden to the system, we propose an elegant method with the combination of foveated view and polarization multiplexing, based on LC reflector. For LC grating, it shows a nearly 100% efficiency with a large diffraction angle, which is a perfect candidate for the waveguide-based AR systems. We propose and demonstrate the LC grating-based waveguide AR with benchtop demo and further performance optimization. For LC lens, it can achieve controllable power and large off-axis angle while maintaining high efficiency. These unique and attractive features give LC lenses the ability to achieve a glasses-like AR architecture while maintaining high optical efficiency. Based on this LC lens, we demonstrate a novel AR system design using polarization and time multiplexing methods to simultaneously obtain a double field of view and a glasses-like form factor. The proposed patterned LC devices for NED applications are validated by both optical simulation and experiment. Multiple tabletop demos are constructed to illustrate how these patterned LC devices can significantly improve the visual experiences of these next-generation NEDs.

Fluorescence microscopy has emerged as a popular and powerful tool within biology research, owing to its exceptional signal contrast, specificity, and the versatility of the various microscope designs. Fluorescence microscopy has been used to study samples across orders of magnitude in physical scale ranging from tissues to cells, down to single-molecules, and as such has led to breakthroughs and new knowledge in a wide variety of research areas. In particular, single-molecule techniques are somewhat unique in their ability to study biomolecules in their native state, which enables the visualization of short-lived interactions and rare events which can be highly relevant in clinical applications. For example, single-molecule real-time DNA sequencing has become a workhorse in genomics and personalized medicine. However, there have been few other analytical tools based on single-molecule fluorescence microscopy that have become popular in biomedical applications. This dissertation describes work performed in an effort to transition single-molecule techniques from a research setting to a clinical setting. There were two main goals throughout: to develop quantitative single-molecule assays for data-rich analysis, and to make those assays more user-friendly to facilitate their adoption as standardized techniques. An initial study demonstrated the practicality of single-molecule analysis as a diagnostic tool by measuring differences in protein content between healthy patients and patients with Parkinson's disease. From there, the assay was improved through various methods of beam shaping, which enabled more quantitative analysis of the detected biomolecules. A passivation scheme and sample preparation protocol were developed that reduce the time to perform a single-molecule assay by more than half while improving the assay sensitivity. Additionally, work performed to control the fluorescent labeling of the target protein is described, with a goal of determining the stoichiometry of protein complexes, which is highly relevant to the pathology of Parkinson's disease and other neurodegenerative diseases. The report concludes with prospective projects that could extend the work completed thus far. An alternative labeling approach is outlined that may achieve one-to-one labeling between the proteins and fluorophores, as well as a project that shifts away from fluorescence microscopy and moves to a label-free scattering-based microscope design.

Light does not interact with itself in linear optical materials. Such interactions occur only in non-linear optical (NLO) materials and typically require high intensity optical beams to be signifi-cant. The ever-increasing role of NLO, where intense light may change the properties of the me-dium, has created a pressing demand to invent materials for achieving more efficient light-light and light-matter interaction due to their potential capacity to augment and possibly replace cur-rent technologies with more efficient devices. There are numerous applications of NLO devices in fundamental science, technology, health, and defense such as all-optical computation and sig-nal processing, ultrashort laser technology, photodynamic cancer therapy, and quantum commu-nication and information. The main objective of my Ph.D. dissertation is to investigate the interaction of laser puls-es with an exciting class of material that has the dielectric constant close to zero, so-called epsilon-near-zero (ENZ). The ENZ materials and their scientific development have become a topic of interest owing to their fascinating nonlinear optical properties particularly in frequency ranges that the material is transitioning from dielectric to metal. The goal of my dissertation is the theo-retical and experimental study of transparent conducting oxides such as Indium Tin Oxide (ITO) as a candidate material exhibiting the ENZ condition and utilizing this effect for nonlinear opti-cal devices. The NLO effects in TCOs are dominated by carrier related nonlinearities. Additionally, this dissertation studies instantaneous third-order and non-instantaneous carrier nonlinearities in semiconductors such as GaAs and Silicon and fifth-order nonlinear absorption (three-photon absorption) in direct gap semiconductors. In this work, we first introduce the development of NLO spectroscopy systems for the characterization of the NLO properties. In particular, the Beam-Deflection (BD) technique, which allows us to simultaneously characterize the nonlinear refraction and absorption of the ma-terial. This technique enables us to characterize the sign, magnitude, temporal and polarization dependence of the nonlinearities, which all are of paramount importance in understanding the underlying physics of the material. We also extend BD to a cross-propagating geometry to measure off-diagonal nonlinear susceptibility matrix elements. Moreover, we employ BD to measure ultrafast and carrier-induced nonlinear absorption and refraction of common semiconductor materials and transparent conducting oxides. The ability of BD to measure the time-dynamics and polarization-dependence of the nonlinear phase shift becomes apparent in both subjects presented in this dissertation.

Machine learning (ML) has been flourishing in various fields, including image recognition, natural language processing, and even protein structure analysis. In recent years, it is getting attention in the optoelectronics field. Researchers not only use ML tools to help boost the research of optoelectronic devices but also try to invent new optoelectronic devices to build computers to help the application of ML in real life. In this dissertation, both directions are explored, including using ML to help design high-performing perovskite solar cells (PSCs) and synthesizing new materials to build new optoelectronic synapses for future neuromorphic computers for ML applications. First, ML is used to predict the bandgaps of perovskite materials and performances of PSCs, which shows that ML benefits the research of optoelectronic devices. Several promising findings are discussed based on ML's predictions to help guide the design of high-performing PSCs. Next, new optoelectronic synapses are fabricated, which can act as building blocks for neuromorphic computers. By applying heterogeneous nucleation principles to grow perovskite quantum dots (PQDs) on multi-wall carbon nanotubes (MWCNTs) and Graphene, new materials are synthesized and used to fabricate optoelectronic synapses. The potentiation of the synapses is realized by light pulses, and the depression is accomplished by electrical pulses. Using the properties of the device to do simulations, the ability of the new type of optoelectronic synapses to act as building blocks of optoelectronic neuromorphic computers is demonstrated. Finally, plasmonic OECTs are fabricated using a low-cost method called the nanoimprint method. Using glucose sensing as proof, this new type of OECT devices can significantly enhance the sensitivity of glucose sensing under light illumination. This new type of OECTs could be a new direction for optoelectronic synapses or work as building blocks for the human-machine interface.

High-power fiber laser systems enjoy a widespread use in manufacturing, medical, and defense applications as well as scientific research, due to their remarkable power scalability, high electrical to optical efficiency, compactness and ruggedness. However, single-mode fiber power scaling has stagnated in the past years, primarily due to the onset of nonlinear effects such as stimulated Brillouin/Raman scattering and transverse modal instabilities. This thesis addresses the analysis and mitigation of transverse modal instabilities in high-power fiber amplifiers. I describe the high-power fiber amplifier testbed that I set up to test fibers fabricated in house. I will show our results of a Yb-doped fiber amplifier with more than 2.2 kW signal power and beam quality of 1.1 M2. In consequence, I demonstrate mode-selective amplification in a large mode-area Yb-doped fiber using a 3-mode photonic lantern. All three modes were amplified to above 4 W with OSNRs higher than 16 dB. In addition, I show a novel high-speed beam analysis technique to study transverse modal instabilities. To guide fiber designs, I developed a GPU accelerated simulation suite to study the dynamics that occur in high-power fiber amplifiers. A 64 x 64 spatial grid, with 6000 time- and 20000 distance-steps can be solved at 2 min/meter on a GeForce GTX 1080 Ti. Based on these simulations, I will show dynamic transverse modal instability mitigation strategies that rely on mode modulation.

Liquid crystal planar optics (LCPO) with versatile functionalities is emerging as a promising candidate for overcoming various challenges in near-eye displays, like augmented reality (AR) and virtual reality (VR), while maintaining a small form factor. This type of novel optical element exhibits unique properties, such as high efficiency, large angular/spectral bandwidths, polarization selectivity, and dynamic modulation. The basic molecular configuration of these novel reflective LCPO is analyzed, based on the simulation of molecular dynamics. In contrast to previously assumed planar-twist structure, our analysis predicts a slanted helix structure, which agrees with the measured results. The optical simulation model is established by rigorous coupled-wave analysis (RCWA). With a higher precision and faster computation speed, the model comprehensively investigates the diffraction properties of various types of LCPOs. This fundamental study on LCPO paves the way for its further applications in AR/VR displays. Several approaches adopting LCPO to solve major challenges in AR/VR like insufficient resolution, limited field-of-view (FoV) and small exit pupil are presented. A foveated display system with doublet liquid crystal lenses is built to concentrate the resolution in the central FoV, corresponding to human eye's highest visual acuity. The proposed foveated display can improve the effective resolution with a fixed total resolution and is expected to alleviate the screen-door effect in VR caused by inadequate resolution. In addition, a new display system named scanning waveguide display is proposed to break the FoV limit (80°) of current AR waveguide displays. The system adopts an ultra-low f-number liquid crystal lens array and reaches a FoV of 100°. Finally, a pupil steering approach is proposed to effectively enlarge the exit pupil of retinal-scanning displays. One in a set of liquid crystal lenses is selectively turned on at each time to match the viewer's pupil location. In comparison with previous approaches, our pupil steering exhibits advantages like aberration-free, fast response time, and compact size.

The decision on whether to use a mid wave infrared (MWIR) or long wave infrared (LWIR) sensor for a given task can be a formidable verdict. The scope entails facts about the observable source, the atmospheric interactions, and the sensor parameters within the hardware device. Even when all the individual metrics are known, the combination ultimately determines whether a MWIR or LWIR sensor is more appropriate. Despite the vast number of variables at play, the reduction of inputs through focused studies can provide essential insight into MWIR and LWIR comparisons. This dissertation focuses on the roles of point source target detection, atmospheric scattering and absorption effects, and target identification has for MWIR vs LWIR performance. The point source analysis details the Pulse Visibility Factor (PVF) and how it affects the Signal to Noise (SNR) for Infrared Search and Track (IRST) tasks. The PVF is an essential parameter that not only depends upon camera system hardware but also the dynamics of the imaged point source target. The numerical predictions of the PVF show how the hardware transfer function spreads the point source object across the detector array. As a result, it is a critical aspect for MWIR vs LWIR IRST system performance. Atmospheric effects are another essential study for MWIR and LWIR imaging performance. Given the magnitude of atmospheric variables, the focus here is to reduce the atmospheric conditions with known particulates and concentrations to provide predictable results. The analysis details how a sparse aerosol medium can absorb and scatter incident light to produce a blur and compromise image quality. Predictions of the aerosol Modulation Transfer Function (MTF) detail the differences in MWIR vs LWIR performance due to aerosols. The MTFs are then added into the Night Vision Integrated Performance Model (NVIPM) to calculate the ability to identify a target at range for typical MWIR and LWIR sensors.

For two decades, extraordinary optical transmission (EOT) has amplified exploration into subwavelength systems. Researchers have previously suggested exploiting the spectrally selective electromagnetic field confinement of subwavelength cavities for multispectral detectors. Utilizing the finite-difference frequency domain (FDFD) method, we examine electromagnetic field confinement in both 2-dimensional and 3-dimensional scenarios from 2.5 to 6 microns (i.e., mid-wave infrared or MWIR). We explore the trade space of deep subwavelength cavities and its impact on resonant enhancement of the electromagnetic field. The studies provide fundamental understanding of the coupling mechanisms allowing for prediction of resonant spectral behavior based on cavity geometry and material properties. In addition to work on spectral response due to geometric parameters for subwavelength cavities, we investigate the spectral response with the inclusion of an absorber on the output in the mid-wave infrared. The placement of an absorbing layer causes a dramatic increase on the effective index within the subwavelength cavity while causing the cavity to become energetically leaky. We have found this broadens the spectral response of the cavity. To mitigate this undesired effect for spectral filter applications, we investigate modulation of the absorber-cavity field coupling by addition of an isolation layer; we show this layer decreases the spatial overlap of the cavity mode with the lossy absorber. In addition, we examine the effect of these layers on the quantum efficiency of the system. We also explore changing the material environment both within and surrounding the cavity to increase quality factors of designed cavities. We examine and quantify such systems by the trade-off that occurs between the quality factor and quantum efficiency. This trade-off occurs due to the spatial extent of fields in the propagating direction. The lateral spatial extent of the cavity is also examined by changing the lateral subwavelength spacing of a periodic array of cavities. The cavities are found to be sensitive to fields extending ~10x larger than the physical extent of the cavity. This phenomenon is indicative of the funneling effect. In addition, fabrication techniques were examined and found to be successful in creating the subwavelength features necessary for creation of such systems. The spectral response of fabricated devices was found to be in excellent agreement with simulation. This dissertation sets the groundwork for development of a novel multispectral detector in the MWIR by examining the spectral relationship of subwavelength cavities coupled to an absorber.

Transparent electrodes represent a critical component in a wide range of optoelectronic devices such as high-speed photodetectors and solar cells. Fundamentally, the presence of any conductive structures in the optical path leads to dissipation and reflection, which adversely affects device performance. Many different approaches have been attempted to minimize such shadowing losses, including the use of transparent conductive oxides (TCOs), metallic nanowire mesh grids, graphene-based contacts, and high-aspect ratio metallic wire arrays. In this dissertation I discuss a conceptually different approach to achieve transparent electrodes, which involves recapturing photons initially reflected by highly conductive electrode lines. To achieve this, light-redirecting metallic wires are embedded in a thin dielectric layer. Incident light is intentionally reflected toward large internal angles, which enables trapping of reflected photons through total internal reflection (TIR). Light trapping transparent electrodes could potentially reach the holy grail of transparent electrodes: the simultaneous achievement of high conductivity and near-complete optical transparency. We numerically and experimentally investigate several light trapping electrode structures. First, we study the spectral and angular optical transmission of embedded interdigitated metallic electrodes with inclined wire surfaces and demonstrate efficient broadband angle-insensitive polarization-independent light trapping. Proof-of-principle experiments are carried out, demonstrating several of the features observed in our numerical studies. Second, a novel type of grating-based light trapping transparent electrode is discussed. In this approach, diffraction from metal wires covered with nanoscale silicon gratings is used to achieve total internal reflection. We show that careful grating optimization achieves strong suppression of specular reflection, enabling a more than fivefold reduction of shadowing losses. The realization of a high light-trapping efficiency in a coplanar structure makes the design a promising candidate for integration in real-world optoelectronic devices. Finally, the transmission of high-index metasurfaces is investigated. Such structures may enable efficient light redirection around metallic contacts, if reflection losses by the metasurface can be suppressed. We demonstrate that the traditional anti-reflection coating approach fails for such structures, and present an improved design approach that reduces reflection losses over a broad range of structural parameters.