The numerical simulation of the fabrication processes as well as the device performance has become an essential method required for the design of new semiconductor device generations. The frequently used term Technology Computer Aided Design (TCAD) is the utilization of computational methods by means of software tools, for the design and analysis of semiconductor devices and their fabrication processes.
For the simulation of complex device structures, for instance Very Large Scale Integration (VLSI) products, additional assistance of TCAD frameworks, for the automatic execution of the simulation flows, extraction of the results etc., is provided. But even with these simulation environments tasks like sensitivity analysis or optimization are difficult and very time consuming for the user. The handling of high level analysis and optimization tasks asks for optimization environments for TCAD applications.
This dissertation presents modules integrated into the SIESTA (Simulation Environment for Semiconductor Technology Analysis) TCAD environment to handle optimization tasks occuring during the development of semiconductor devices. With the two integrated optimizers, a least square optimizer and a nonlinear constrained optimizer, tasks like process tuning, tool calibration, technology development, and inverse modeling can be performed. For statistical analysis, Design of Experiments (DoE) and Response Surface Methodology (RSM) tools have been integrated into the framework.
The presented TCAD environment makes it possible to perform optimization tasks which can be applied on whole simulation flows or on generated response surfaces. It also enables the user to insert simulators and tools from different projects and vendors thus making SIESTA to an attractive optimization environment for various problems. For the efficiency of the optimization environment features for parallel execution on a cluster of workstations are implemented.
The applicability and usefulness of the framework and it's optimization modules is demonstrated by three examples. In the first example an optimization task using DoE and RSM is performed. The second example illustrates a performance optimization of a MOS transistor with analytical doping profiles. In the third example the inverse modeling of the doping profiles based on the measured electrical characteristics of the examined devices is shown.