Device simulation and multi-objective genetic algorithm-based optimization of Germanium metal-oxide-semiconductor structure

Germanium (Ge) and high-κ dielectric materials draw many attentions due to their fascinating electrical characteristics comparing with silicon (Si) material. However, in physical and electrical simulation, the physical model may have deviation to reality case due to the process condition and manufacturing technology. To computationally study the device with Ge material, it is necessary to optimize the theoretical result with experimental data. This paper originally provides a new method to examine the static characteristic of Ge metal-oxide-semiconductor field effect transistors (MOSFETs) with aluminum oxide (Al₂O₃) by integrating device simulation, multi-objective evolutionary algorithm (MOEA), and unified optimization framework (UOF). To deal with the realistic problem, especially for the steep change of capacitance, we consider not only residual sum of squares (RSS) (i.e. the sum of squares of residuals) function but also physically crucial points in the optimization problem. Comparing to single-objective genetic algorithm (GA) with a weighted fitness, the preliminary result of this study shows the method has great improvement to optimize the suitable parameters which not only minimize the RSS of capacitance but also agree the key capacitance values from physical view.