Multiscale modeling of anisotropic wet chemical etching of crystalline silicon

– We combine ab initio and Monte Carlo simulations in multiscale modelling of anisotropic wet chemical etching of silicon. The anisotropy of the macroscopic etching patterns observed in the experiments is explained by two mechanisms at an atomistic scale: the weakening of backbonds following OH termination of surface sites and the existence of significant interaction between the surface-terminating species (H and OH). For the first time, we demonstrate that the H/OH and OH/OH interactions have an essential role, directly controlling the appearance of the fastest-etched planes in the macroscopic etching patterns. Anisotropic wet chemical etching of silicon in alkaline solutions [1] is one of the key techniques for the manufacture of microsystems. However, the mechanisms responsible for the strong crystalline anisotropy of the etch rates are not completely understood. It is generally thought that hydroxyl ions OH− present in the etching solution play a central role, catalyzing the removal of surface silicon atoms by weakening the backbonds after substituting the surface-terminating hydrogen atoms [2]. In this letter, we show how subtle features of the microscopic reaction environment lead to dramatic changes in the macroscopic etching patterns. In particular, we show for the first time that the geometrical restrictions imposed by the presence of a specific type of next-nearest neighbours are crucial for the appearance of such important features of the macroscopic etching patterns as the fastest-etched planes. This is done by combining atomistic total-energy calculations with large-scale Monte Carlo simulations for the etching kinetics. A number of techniques for the simulation of anisotropic wet chemical etching have been presented in the literature [3]. Of all possible approaches, only atomistic models enable the identification of the essential microscopic mechanisms which control the complicated electrochemical reactions [2] between the surface atoms and the molecules and ions of the etching solution. In these models, the probability of removal of a surface atom is “controlled” by its neighbourhood, since it determines the number and type of backbonds which need to be broken. The number of bonds is determined by the number of first neighbours, and their energy (i.e. the bond type), by the number of OH groups terminating both the surface atom and the first neighbours, as shown in fig. 1. This dependence of the removal of a surface atom on the