Statistical analysis of semiconductor devices

A technique for the analysis of random dopant-induced effects in semiconduc tor devices is presented. It is based on the "small signal analysis" (pertu rbation) technique. It is computationally much more efficient than the exis ting purely "statistical" techniques, and it yields the information that ca n be directly used for the design of dopant fluctuation-resistant structure s of semiconductor devices. This technique requires only the knowledge of v ariances of fluctuating doping concentrations and in this sense, it is a "s econd-moment characterization" technique. This technique can be naturally e xtended to take into account random fluctuations of oxide thickness and oxi de charges in metal-oxide-semiconductor filed-effect transistor. The numeri cal implementation of this technique is discussed and numerous computationa l results are presented and compared with those previously published in the literature. (C) 2001 American Institute of Physics.