| Due to the random nature of ion implantation, dopant diffusion, and other processes involved in the doping of silicon devices, the dopant density in shallow junctions and short channels is subject to stochastic variations, which translate directly into variations in device behavior. Dopant profilers with nanometer scale or even atomic scale resolution are needed to properly measure these dopant fluctuations. This thesis presents a new approach for two- and three-dimensional dopant profiling with atomic resolution on the Si(100) surface by scanning tunneling microscopy (STM).; The lack of surface states within the band gap of the perfect Si(100)2x1:H surface opens the way to STM studies of dopant distributions in Si(100). STM topographic images, dI/dV images and current image tunneling spectroscopy (CITS) were acquired across the lateral PN junctions of Si devices. Two-dimensional dopant (carrier) profiles were extracted from CITS data with 5Å resolution. Moreover, the N and P type dopant induced features were observed in filled state and empty state STM images. The donor (Arsenic) induced feature appears as a protrusion in both the filled and empty state images, while the acceptor (Boron) induced feature appears as a hillock in the filled state image and a depression in the empty state image. The bias dependence, depth dependence and dopant concentration dependence of the dopant induced features were investigated in detail. Based on scattering theory, the numerical calculation was performed to achieve a fundamental understanding of dopant induced features, and the calculation results were in qualitative agreement with the experimental observations. The potential application of this study for 3D dopant profiling with atomic resolution on both P and N type samples is discussed, and the optimal scanning conditions are also suggested. This work reveals the real physical picture of randomly distributed dopants and may be useful to verify and calibrate TCAD simulators. |
