A Silicon Migration Model Incorporating Anisotropic Surface Energy and Non-Uniform Diffusivity

Abstract

Silicon migration is a process that can produce seamless, single-crystal silicon cavities in a single mask layer, greatly simplifying process flows for forming various cavity- and channel-based structures. However, controlling the dimensions of the final cavity and membrane thickness is complex and difficult to predict. To address this challenge, we present an efficient and accurate method of simulating silicon migration, which advances the state of art in three ways. First, it accounts for anisotropic surface energy. Second, it uniquely models selective migration. Third, the method is implemented in Python, an open source and highly pervasive programming language. We validate the model using experimental results from the literature, and we show that incorporating anisotropy is critical for simulating high aspect ratio, high density etch hole arrays. We also find indications in the simulations of selective migration that anisotropy may be less significant for very large curvature (radius less than 100 nm) features. Lastly, we use this new model to predict the relationship between layout and annealing time for a given set of final objective cavity and membrane dimensions. [2022-0128]