Predictive Control of Surface Roughness as a Function of Temperature in a SiNx Thin Film Deposition Process Using the Kinetic Monte Carlo Method

Abstract

Silicon nitride (SiN_x) films are crucial in microelectronic and photonic devices, where their interface affects performance and reliability. In this study the effect of deposition temperature on the surface roughness of amorphous SiN_x films on silicon (Si) substrates is investigated using our kinetic Monte Carlo algorithm (Silicon 15, 5209 (2023)). The low-pressure chemical vapor deposition process is modeled on a three-dimensional triangular lattice with disilane (Si₂H₆) and ammonia (NH₃) as precursor sources. The nanoscopic events included in this work are the nanoparticle adsorption and the Si adatom migration. A new growth model is adopted to control the obtained surface morphologies, i.e., size and density of amorphous Si clusters as well as film surface roughness. The deposition of Si and N atoms is carried out alternately to create a SiN compound characterized by small Si clusters and a rough surface. Both volume migration and surface migration are taken into account during the simulation, leading to the development of vacancies and pores. The formation of peaks and valleys is described by our kinetic Monte Carlo algorithm. Our analysis includes deposition simulations at temperature values ranging from 723 to 753 K, with a gas flow rate fixed at 0.3 and a deposition duration of 1 h. The surface roughness values of the deposited nanostructures are deduced from the simulation matrix. Numerical results indicate that an increase in process temperature leads to an increase in the size of Si clusters along with an increase in surface roughness. The deposition temperature largely determines whether the film surface is smooth or rough. This means that our growth model is able to accurately predict the evolution of the film nanostructure for a wide range of process conditions. The stoichiometry x (N/Si ratio) is determined based on all deposition parameters. The average distance between Si clusters is also calculated here. The acquired insights enable the refinement of thin film deposition simulation techniques, the improvement of surface morphology properties, and the support of the development of reliable SiN platforms for microelectronics and photonics.