Model membranes are synthetic structures that mimic cell membranes. They may be used for basic research or as part of an engineered device. Phospholipids are typically used to make model membranes, but they may be replaced by amphiphilic block copolymers that are more physically and chemically stable, more tunable, and less costly. In this work, we explore the use of block copolymer membranes for biosensing applications. Specifically, we target the goal of nanopore DNA sequencing using the protein nanopore MspA in block copolymer membranes.

As a first step, we extended previous work on MspA in a block copolymer membrane by translating it to an array of apertures using a microfluidic painting technique. We found that the use of a volatile solvent in a microfluidic channel caused variations in membrane properties with corresponding undesirable variations in MspA behavior. Additionally, we used this method to demonstrate DNA translocation through MspA in block copolymer membranes. We selected droplet interface bilayers (DIBs) to investigate which membrane properties are responsible for variations in MspA behavior. However, we found that our aqueous block copolymer DIBs behave differently than any previously studied DIB, necessitating the use of new techniques to characterize their behavior. Key differences include reduced bilayer stability due to poor packing of the monolayer and slow equilibration kinetics. We addressed the poor bilayer stability by establishing a technique for manually packing monolayers prior to bilayer formation. We addressed the slow equilibration kinetics by establishing a new method for measuring monolayer and bilayer tensions in DIBs using droplet shape analysis. While droplet shape analysis is typically used for single droplets with axial symmetry, our method works even for DIBs with asymmetric volumes and tensions. Another advantage is that the method can determine monolayer and membrane tensions in real time, independent of the applied voltage. The method may be used for polymer and phospholipid DIBs alike. The work presented in this thesis contributes to the understanding of amphiphilic block copolymer behavior and will aid in the optimization of membrane properties for biosensing applications.

Self-referenced optical combs have proven pivotal for numerous metrology applications including precision navigation, LiDAR, and molecular spectroscopy. While plenty of research has improved and broadened the scope of this instrument, most implementations to date have been lab-scale setups that require kilowatts of power. The size, weight and power (SWaP) needs to shrink in order to fully realize the potential of this technology.

Silicon chip-based Photonic Integrated Circuits (PICs) provide a platform to exploit the same optical phenomena in a reduced SWaP. However, these miniature devices are inherently prone to fabrication variation and environmental fluctuations during operation. An associated issue is the coupling of precision and power. Where bench top implementations can circumvent many issues by increasing the laser power to mitigate downstream losses in optical elements, an integrated solution requires novel electronic signal estimation, detection and stabilization architectures to maintain precision under a low power budget.

This dissertation presents a mixed-signal controller designed to handle the challenges of achieving parts-per-trillion frequency stability in an integrated optoelectronic frequency synthesizer. I discuss the development of a heterodyne-based architecture and highlight experimental results from a PCB prototype using commercial-off-the-shelf electronics. The limitations present at the board-level are further mitigated by the development of a custom IC. Results from an application-specific integrated circuit (ASIC) designed in 55nm CMOS show the potential of integration in reducing the SWaP. Ultimately, this architecture achieves state-of-the-art performance, producing a 193 THz output with 5.6 mHz average deviation (2.9e-17 ADEV @ 1000s). The synthesizer is tunable >40 nm across the C-band with 745 mHz setpoint resolution, capable of full configurability in real-time via a custom Graphical User Interface (GUI).

Interfacing with the brain is a challenging problem. While many innovative methods for providing input and output to neural systems have been developed and demonstrated successfully in human patients, these invasive systems use less biologically compatible means than are realizable. Materials and mechanisms which are closer mimics to biological systems in their behaviors can lead to more stable and effective medical prosthetic and research devices.

Retinal prosthetics, as well as Cochlear implants, are neural implants which provide stimulation via electrical impulses. Currents passing across neurons trigger neurons to begin firing or generating action potentials down their axons, stimulating neurons with dendrites connected to those axons terminals. This scheme transduces the desired simulation into neural firing spikes, but the majority of applied current is shunted around neurons rather than contributing to stimulation. Excess current contributes to the power and thermal budgets of neural simulation devices, which are implanted in tissue, limiting their functionality. Additionally, the electrical contacts which provide a source and return for the stimulation currents are subject to degradation over time, as currents are repeatedly applied during stimulation events. Stimulation of neurons via local delivery of potassium in excess of available intercellular potassium can be used in place of direct electrical stimulation and promises to be a more biologically compatible method.

Several forms of neural recording devices have been developed and the widest know and highest density is the Utah Array, a 3D array of silicon spires, which can record from their tips when inserted into neural tissue. While this 3D array topology can access a field of neural activity, the stiffness of these silicon spires is very different than that of neural tissue, which can lead to an unwanted inflammatory response. Conductive polymer pillars made of softer materials that are a closer match to neural tissues, while mirroring the same density and insertion length as the Utah Array, may provide a better recording mechanism due to their improved mechanical compatibility with neural tissues.

This thesis investigates microfabrication schemes to produce biomimetic structures that can enable neural simulation and recording devices which feature greater biological stability and improve utility.