Microwave Modeling And Parameter Extraction For 90-nm Gatelength MOSFET Devices

The deep submicrometer metal oxide semiconductor field effect transistors?MOSFETs?are very attractive for radio frequency?RF?and microwave integrated circuits due to its excellent performance,low power,high integration density and low cost.The market demand for higher levels of integration and higher operating frequencies has inspired the tremendous advancements in complementary metal-oxide semiconductor?CMOS?technologies.In order to reduce the cost,shorten design cycle time and improve design success rate in RF integrated circuits,it is necessary to build an accurate model of MOSFET devices.The model and corresponding parameter extraction method are provided in this dissertation and the innovative research results achieved include:1)An novel extrinsic parasitic resistances extraction method for MOSFET devices in the modeling of small-signal equivalent circuit is proposed.This method overcomes the difficulty of ingoring the substrate elements and resulting in frequenc-dependent resistances in the traditional parameter extraction methods,and extracts the extrinsic gate,source and drain resistances in the intermediate frequency range through the measured S-parameters under cut-off bias condition;2)The extraction of device model parameters and model verification heavily depend on the instrument measured data,and the data usually have the uncertainties which will eventually lead to the inaccuracy of model and extracted parameters obtained.In this dissertation,the several sources of uncertainty of that contribute to S-parameter measurement are introduced in detail and the uncertainty is also modeled.The sensitivities in intrinsic parameters to S-parameter are deduced,and associated with microwave RF measurement uncertainties.Finally,the curves of model parameter uncertainty varying with the frequency are acquired,and the optimum frequency range for parameter extraction is obtained.Meanwhile,in order to provide reliable model parameter results for large-signal modeling,the relationship between the intrinsic parameter uncertainties and the bias voltage is also studied.3)The nonlinear equivalent circuit model for silicon-based MOSFET that incorporates DC/AC dispersion model is studied.A complete large-signal equivalent circuit model is established as a symbolically defined device?SDD?and implemented into Keysight Advanced Design System?ADS?to demonstrate the model's accuracy.The model is modified as follows:firstly,in order to improve the accuracy of DC model,a new drain-current model which modified the model parameters related to V_gss is proposed based on STATZ model.The model can accurately describe the current-voltage characteristics overall bias conditions;and secondly,based on our knowledge,for the first time,an improved RF large signal model for deep-submicron MOSFET that incorporates DC/AC dispersion model is proposed and the RF current source in parallel with the conductance are added to the traditional large-signal model topology to simulate the DC/AC dispersion effect of MOSFET devices;lastly,new equations for the nonlinear capacitance phenomenon in the saturation region are provided in this dissertation.4)The analytical expressions for the noise parameters of MOSFET devices including optimum noise coefficientF_min,optimal noise resistanceR_n,optimum source conductanceG_{op t} and optimum source conductanceB_{op t} are derived.Since the traditional expressions only consider the intrinsic gate-to-source capacitance and resistance,transconductance,output conductance and neglects the substrate losses and the extrinsic resistances,thus it will affect the noise characteristics of small size silicon-based MOSFET devices.Based on the Pospieszalski model,four noise parameters are derived in this dissertation.The parasitic series resistance and substrate effect are taken into account in the derivation process,and a set of new expressions are given without any assumptions and approximations.