MULTICOMPONENT TRANSPORT AND MIXED CONVECTION FLOWS IN CVD REACTORS

摘要

This paper is concerned with two aspects in Chemical Vapor Deposition (CVD) modeling: multicomponent diffusion of precursor species in the gas-phase and mixed convection flows inside the reactor. We first investigate the impact of accurate modeling of the species diffusion velocities on growth rate predictions, dealing with the Soret effect, cross-diffusion phenomena and mass versus mole fraction formulation of the diffusion driving forces. The impact is assessed quantitatively by comparing it to the influence of detailed gas-phase and surface chemistry on growth rate predictions. We investigate gallium arsenide growth for which fairly accurate chemical data is available. We next present Petrov-Galerkin finite element discretizations for flow and heat transfer inside CVD reactors. The adaptive mesh refinement strategy relies on a posteriori error estimates derived from the solution of a linearized dual problem which provides essential information on where the numerical error is generated.