Microscopic Analysis of Noise Behavior in Semiconductor Devices by the Cellular Automaton Method

Abstract

Noise is one of the crucial features in modern semiconductor devices. Nevertheless, only a few microscopic investigations of intrinsic noise behavior of semiconductor devices in the GHz regime have been performed in the past [1]. Although the dependency of intrinsic noise on the doping concentration, applied voltages and the influence of hot carrier effects is an important issue, systematic investigations are hampered by the fact that very long time sequences have to be simulated for the calculation of the relevant correlation functions. For semiconductor devices, such simulations are usually performed with the Monte Carlo method, self-consistently coupled to a Poisson equation. In [2], we have shown a highly efficient cellular automaton (CA) approach to be physically equivalent to the Monte Carlo technique for macroscopic transport quantities. However, our initial implementation of the CA method showed discrete lattice effects that led to a slightly enhanced carrier diffusion and hampered an accurate calculation of correlation functions. We have recently developed a new scheme for the CA that eliminates this problem.