DESIGN AND SIMULATION OF A RACETRACK PCM-BASED MICRORESONATOR FOR PHOTONIC OPTICAL SWITCHING IN NEUROMORPHIC SYSTEMS

Abstract

Compact and energy-efficient optical switching elements are important for scaling neuromorphic photonic systems and reconfigurable integrated photonics. Resonator-based devices integrated with phase-change materials (PCMs) offer strong light–matter interaction, nonvolatile programmability, and high transmission contrast.In this work, we study a PCM-integrated silicon racetrack microresonator as a geometry- optimized platform for nonvolatile optical switching, with a focus on material selection and PCM placement. A comparative material study is carried out by introducing a fixed PCM segment at the racetrack waveguide–resonator coupling region and evaluating Sb2Se3, Sb2S3, and Ge2Sb2Te5 (GST) in both amorphous and crystalline states. Using static refractive-index models based on experimentally reported optical constants [1], the materials are compared in terms of transmission contrast, resonance modulation, and linewidth variation. Among the materials considered, GST shows the strongest state-dependent response and is selected for further analysis. A spatial optimization study is then performed to understand the impact of PCM placement. Four angular positions along the racetrack, separated by 90 degrees, are first evaluated, with the coupling region identified as the most sensitive location. To further improve performance, a finer position sweep is carried out along the straight section of the coupling region, where the GST segment is shifted in 1-μm steps across eight positions. The configuration corresponding to placement P5, located just to the right of the baseline position, produces the highest contrast ratio and modulation depth. This comes at the cost of increased optical loss and reduced resonator quality, highlighting the trade-off between strong modulation and maintaining a high-Q resonance. All results are obtained using three-dimensional finite-difference time-domain simulations. While dynamic phase-transition effects and power-dependent nonlinearities are not included, the results provide a clear, geometry-driven design framework for PCM-integrated racetrack resonators as nonvolatile optical switches and building blocks for neuromorphic photonic systems. Portions of this work were accepted and presented as an oral paper at SPIE Photonics West 2026 and published in the SPIE Proceedings [2]. The fine-tuning position sweep presented in this thesis extends beyond the published work.