Modeling silicon photonics process variations using ring resonator devices

Abstract

Silicon photonics is an emerging industry that aims to implement photonic systems using extensions to traditional CMOS process technologies. With silicon technologies being used to transmit light, there exists a need by designers to understand the performance of standard silicon photonic elements and the deviations that may arise in actual production. Drop-in models for silicon-based photonic structures do exist and are used to make light-based systems. However, these models do not take manufacturing variations into account and are assumed to be ideal. This thesis establishes the use of models for photonic structures that take spatial and random variations from manufacturing into account. More specifically, we demonstrate two elements of a variation-aware photonic design and analysis methodology. First, we develop first-order models of passive ring resonators and its active counterpart, the ring modulator, to simulate and predict the behavior and impact of variation in these devices on large-scale designs. Both are small footprint devices capable of selective filtering, yet often deviate considerably from expected behavior due to process deviations. Second, we explore ways to use these basic photonic elements to build a larger system that can probe process deviations over a chip. Together, these methods contribute key steps toward a design for manufacturability methodology to achieve high yield and high performance silicon photonic systems.