A calculation method of deposition profiles in CVD reactors using real-coded genetic algorithms (2)

Abstract

Fast and accurate calculation of the predicted results of Chemical Vapor Depositions (CVD) is very helpful to the R&D of the CVD processes. Therefore, we have developed a novel calculation method, by which accurate calculations along with reduced computing cost were achieved, to reproduce deposition profiles in three types of reactors. Mass balance equations in the reactors with complicated reaction models were divided into simple equations with single deposition species, in order to identify a few basis functions, with which all the deposition profiles can be synthesized. Accordingly all the deposition profiles were calculated from the linear combinations which consisted of the basis functions. The coefficients of the linear combinations were optimized by Real-Coded Genetic Algorithms (RCGA). In addition, we investigated the influence of three methods of RCGA, that is, BLX-α (Blend Crossover) + tournament selection, REX(φ, n+k) (Real-Coded Ensemble Crossover) + JGG (Just Generation Gap) and REX^star + JGG on the calculations of the profiles. REX^star showed the best overall performance for the calculations.