Overview

The accuracy of the device model predictions of the device characteristics are fully dependent on the model parameter values being used. Most of the circuit models discussed in the previous chapters are semi-empirical analytical models. These models always contain some fitting parameters that do not have physically well defined values, and very often physical values of model parameters do not always give the best fit to the actual device characteristics. For this reason, device model parameters are determined from the device data obtained from electrical measurement on different length and width devices and under different bias conditions. Collecting measured data and processing these data to accurately determine model parameter values is an essential task for the complete characterization of a transistor model for use in the circuit simulator.

In this chapter we will first discuss experimental setups for measuring the device data that is required for extracting DC and AC model parameters. We then discuss different methods of determining basic MOSFET parameters such as substrate doping concentration N _b (or doping profile), threshold voltage V _th, carrier mobility ?, device effective or electrical channel length L and width W, gate oxide capacitance C _ox, etc., which are essential for the characterization of any MOSFET model. These basic parameters are also used routinely for electrical (E)-test measurements. The general model parameter extraction using a nonlinear optimization technique will be the topic of discussion for the next chapter.

Before discussing data acquisition and the measurement...