Development and evaluation of automatic modeling for reaction systems using genetic algorithms

Abstract

The improvement of the Research and Development (R&D) for material formation processes both in the speed, and the cost is strongly demanded in the semiconductor industries. The automation of the R&D will be the candidate for it. However, intellectual works, for example, analyzing the experimental data, and modeling the processes, are formidable obstacles for the automation. In this paper, we showed the novel system, which automatically infers the reaction models from the experimental data using Genetic Algorithms (GA). In the system, the reaction models were presented by the hierarchical chromosomes, and determined chemical-kinetically using the experimental data. We applied the system to the Tetra EthOxy Silane (TEOS) thermal Chemical Vapor Deposition (CVD) processes in order to investigate the potential of the system. The system presented the simple models, which successfully reproduced the experimental data. In addition, the proposed models were in good agreement with the ones, which were manually estimated using another authorized procedure. Therefore, we concluded that the system has enough ability to contribute to the automation of the R&D for the material formation processes.