Multiscale atomistic modelling of CVD: From gas-phase reactions to lattice defects

Growth from vapour/gas/plasma phases is a key process to produce high-quality nanostructures and thin films. The quest for high performances at low cost calls for the development of modelling strategies able to accurately predict growth rates and structure morphology under a variety of process conditions. In the semiconductor nanotechnology, Lattice Kinetic Monte Carlo (LKMC) is considered an advanced approach for simulating selective epitaxy of semiconductors by Chemical Vapor Deposition (CVD). However, state-of-the-art LKMC tools often neglect fundamental aspects such as lattice defects and chemical reactions, both in the vapor phase and around the evolving surface. We present a multiscale workflow for modelling CVD growth and etching processes also accounting for these critical phenomena. We implement it in the open-source KMC super-Lattice (KMCsL) code MULSKIPS, whose peculiar design allows for the generation and evolution of point-like and extended defects in tetrahedrally-bonded materials, such as Si, SiC or SiGe alloys. Gas-phase reactions at the meso-scale are considered by coupling with an external thermodynamic simulator, while surface reactions involving the equilibrium gas species are described by an analytical continuum model. We perform experiments to calibrate and validate the KMCsL model. We then apply the methodology to simulate nanoscale morphology modifications in planar, nanostructured and constrained geometries, unveiling the role of temperature, precursors’ pressures, surface coverage and defects kinetics in the CVD process.