Recovering doping profiles in semiconductor devices with the Boltzmann-Poisson model (Q543757)

Parameter extraction for semiconductor devices is an important tool in modern semiconductor device design and performance specification. The aim of parameter extraction is to recover material parameters of devices from measurement of device characteristics such as e. g. current-to-voltage (I-V) data and voltage-to-capacitance (C-V) data. The parameters that are of interests are mainly the doping profile and carrier mobilities. In this paper, an inverse problem related to the Boltzmann-Poisson system of equations for transport of electrons in semiconductor devices is studied. Linear reconstructions algorithms as well as nonlinear algorithms of Newton type to recover numerically the doping profile function from measurement of device characteristics (I-V) are presented. To reduce the degree of ill-posedness of the inverse problems the unknown doping profile is parametrized to decrease the number of unknowns in the inverse problems. Although the transport of charges in semiconductor devices is best modeled by the Boltzmann-Poisson system (BP) of equations the majority of works on semiconductor inverse problems uses for the model of the charged particles transport the drift-diffusion-Poisson system (DDP) of equations. The main objective of the paper is to study numerically reconstruction problems for the more accurate BP model and characterize the difference between reconstructions obtained with the DDP model. It follows from detailed numerical reconstruction results based on synthetic data (generated by BP model) that two models can give sufficiently different results. As the size of the device is getting larger or the noise level in the data getting higher, the differences in reconstructions are indistinguishable.