A Domain Decomposition Method: A Simulation Study

By: Carlo Cercignani, Irene Gamba, Joseph W. Jerome and Chi-Wang Shu

A mesoscopic/macroscopic high-field model was derived by Cercignani, Gamba, and Levermore in [4]. The model was summarized in relationship to semiconductors in [3]. In [2], a conceptual domain decomposition method was introduced, based upon use of the drift-diffusion model in highly-doped regions of the device, and use of the high-field model in the channel, which represents a (relatively) lightly-doped region. The hydrodynamic model was used to calibrate interior boundary conditions. This paper extends the approach of [3]. (1) Benchmark comparisons are described for a GaAs $n^+-n-n^+$ diode, via a global kinetic model. (2) An elementary global domain decomposition method is presented, wherein the optimal points of separation between drift-diffusion and the high field model are chosen by systematic sampling. The interdomain boundary conditions are implemented through the stencil overlap of the algorithms in both regions.
This paper appeared in the Proceedings of the 1998 Sixth International Workshop on Computational Electronics, Osaka University Press, 1998, pp. 174--177, and can be viewed in the following format: