Impact of parameter extraction methodology on variances of extracted parameter values

Two general classes of strategies for extraction of compact model electrical and temperature scaling parameters are identified and compared in the present paper. Applying these to extract parameters of the p-n junction depletion capacitance, we show that the reproducibility of estimated parameter values can strongly depend on the extraction strategy applied in the nonlinear regression procedure. We present an approach to assess statistical properties of parameter extraction strategies and demonstrate the merits of such assessments.