In state-of-the-art semiconductor development and manufacturing environments, competitive market pressures dictate the pace of technological change. To cope, process, device, and circuits are being designed simultaneously. This practice, normally referred to asco-design, places a lot of demands on accurate technology characterization from a limited amount of data available early in the design cycle. Data driven approaches [71] which are based on empirical and semi-empirical models have limited usefulness in this context because they require a complete set of experimental data for tuning the simulators to a specific VLSI process. By relying on more physically based models, TCAD can provide a solution to early technology characterization needs. The available data is used for full or incremental calibration of TCAD simulator models. This process will yield a virtual fabrication line that could be used to generate the required characterization data covering the whole range of processing conditions.
This particular role of TCAD is described and illustrated in this chapter. Empirical model building and Monte Carlo simulation task modules are used to determine key variable values of a VLSI process based on a statistical worst case methodology. These values are then used as input to TCAD simulators to generate necessary data for the extraction of analytical model parameter values for use in circuit simulation.
Another TCAD application in technology characterization involves a new simulation based method for MOSFET polysilicon gate length determination. This technique could be used to determine the range of intra-die variation which is becoming comparable to inter-die and inter-wafer variation. A meta-simulation polynomial model of the dependence of MOSFET gate capacitance on polysilicon doping and physical gate length is used in conjunction with experimental measurement as a fast, accurate and non-destructive method for gate length determination. These applications highlight the importance of TCAD in modern technology development.
model