Abstract
100nm CMOS technology has been characterized through Design of Experiment (DOE) and statistical modeling. Initially, a set of 21 process parameters (factors) have been identified to determine their impact on transistor performance metrics such as threshold voltage Vt, sub-threshold slope SS, drive current Iddrive, leakage current Idleak, both in saturation and linear region. Through first order linear modeling of Vt, SS, Iddrive, and Idleak, a subset of 10 most significant process parameters are picked using Plackett-Burman screening experiment for both NMOS and PMOS devices. Significant process parameters which impact the device characteristics are seen to be different, for NMOS and PMOS devices, inspite of a common process flow. Response surfaces (RS) have been built in terms of these 10 parameters for NMOS device. Statistical parameters of the device characteristics fluctuations like mean(?) and standard deviation (\sigma) for Vt, SS, Iddrive, Idleak and Gm (maximum transconductance), have been determined by Monte carlo (MC) analysis of these response surfaces. Application of the Transmission of Moment Technique (TMT) on these models is shown to be a simple means to determine ? and \sigma of the device characteristics, with simple mathematical calculations.