Octree-search kinetic monte carlo algorithm for the simulation of complex 3D MEMS structures

This paper presents a kinetic Monte Carlo (KMC) method based on an octree search algorithm and data representation for the simulation of complex MEMS structures created by the propagation of a dynamic surface. We consider the case of wet and DRIE etching of silicon, including multi-masking by silicon dioxide and nitride layers, as well as sacrificial layers. The algorithm leads to a faster tree-based search, efficient updating and good modeling ability for dynamic surfaces. Typically it achieves speedup factors of 3-4 over a state-of-the-art technique. Meanwhile, it reduces the computational memory by 50-80% over the dynamic method, with this factor increasing for larger scale simulations.