Adjoint-based particle defect yield modeling

Abstract

Silicon photonics, where photons instead of electrons are manipulated, shows promise for higher data rates, lower energy communication and information processing, biomedical sensing, and novel optically based applications such as wavefront engineering and beam steering of light. However, silicon photonics does not yet have mature process, device, and circuit variation models for the existing IC and photonic process steps; this lack presents a key challenge for design in this emerging industry. This thesis addresses analysis of the process variation impact of particle defects. Such particle defects can arise in photolithography, deposition, etch, and other processes, and can perturb the intended function of photonic devices and circuits. The adjoint method previously used in optimization is modified and implemented to facilitate the simulation of the impact of defects in silicon photonic devices. More specifically, we demonstrate the methodology to build both component-level and circuit-level models based on the adjoint method. For the component-level models, we show how S-parameters of the device components are impacted by different types of particle defects. For the circuit-level models, we show the impact on circuit output spectrum and performance features based on component-level models, and perform critical area extraction for yield estimation. The model and result will be used to help generate layout design rules, predicting, and optimizing yield of complex silicon photonic devices and circuits for tomorrow's silicon photonics designers.