Phot1x Project Report

Abstract

This report describes the design, fabrication, and data analysis for a Mach-Zehdner Interferometer (MZI) implemented in a silicon on oxide wafer. The device was fabricated using electron beam lithography, and we explored variations in the path length difference of the two arms of the interferometer as well as varying the waveguide geometry. The designs were measured using an swept wavelength source, whereby the transmittance was shown to sinusoidally vary with respect to the input wavelength as expected through theory. We account as well as possible for the losses in coupling light onto and from the chip using grating couplers, and also for propegation losses through the strip waveguides. Finally, the match between the measurements and theory / simulation is evaluated, and we attempt to account for deviations in this match.

Introduction

For this course I designed a collection of Mach Zehnder Interferometers (MZI) to explore the process of designing, fabricating, and testing silicon integrated photonics devices. I was motivated to take this course because I hope to use the knowledge gained here for my work in industry. The MZI device illustrates many important aspects of silicon integrated photonics, making them a good candidate for study.

The devices for this course were fabricated using a 100 keV electron beam lithography process at either of the University of Washington Nanofabrication Facility or Applied Nanotools Inc., Canada. The substrate is a Silicon on insulator (SOI) wafer with 220 nm silicon thickness. The etch parameters are single full etch at an 82º sidewall angle, yielding a minimum feature size of 60 nm. The fabrication area is 605 x 410 μm.

The devices were tested using an automated optical probe station at the University of British Columbia, Canada.

Theory

To analyze our MZI devices we first have to model the individual components of the device. These include the strip waveguide, the bent waveguide, the y-branch splitter/combiner, and the surface grating coupler. Since we are evaluating our components' behaviour with respect to wavelength, we need to determine the parameters of these devices as a function of wavelength.

The strip waveguide is a rectangular silicon structure with a high index of refraction surrounded by a cladding region (either silicon dioxide or air) with a much lower index of refraction. This arrangement allows the waveguide to guide light using total internal reflection. We use the effective index method to determine the phase and group velocity of light propegating down the waveguide. The effective index method is a phenomenological approach to modeling, wherein the various parameters of the waveguide (geometry, material, etc.) and propegating mode (confinement, polarization, etc.) are combined into a single value, namely, the effective index neff. The effective index thus differs for modes where the polarization of the electric field is mainly parallel to the wafer, quasi-TE, vs. modes where the polarization of the electric field is manly perpendicular to the wafer, quasi-TM.

The effective index and group index are determined as a function of wavelength to determine the velocities of the phase fronts and guided energy, respectively.

Bent waveguides are a device in and of themselves, with distinct modes that differ from that of a straight waveguide. Because of this, there will be a mode-mismatch between a section of straight and a section of bent waveguide which is responsible for most of the loss when including a bent waveguide in one's design. Other sources of loss are regular propegation loss and radiated power, caused by coupling of light into higher order unguided modes.

The third component necessary to implement our MZI is the y-branch splitter / combiner. This device, when used as a splitter, divides the intensity of light on the input side equally into th