In this text, the importance of a dedicated data level for a TCAD framework has been demonstrated. The tasks typically performed by process and device engineers in a TCAD framework have been analyzed, and the corresponding data and data flow have been investigated. The anticipated features of a data level have been derived from this investigation, and various architectures of data levels have been presented and compared. The importance of both an interchange format in binary and textual representations as well as a well-defined procedural interface with high-level functionality have been underlined. The VISTA data level architecture has been presented, and its implementation using computer-aided software engineering (CASE) methods has been discussed.
Two examples of CASE tools used in the VISTA data level design have been presented in more detail. As an example of a high-level library, the implementation of the grid support (GRS) library with the help of the VISTA Object-Oriented Programming System (VOOPS) has been discussed in detail.
On the example of the capacitance simulator VLSICAP, the integration of TCAD tools into the VISTA data level has been demonstrated. Using the FORTRAN bindings of the PAI library on the one hand, and the VISTA user interface calls on the task level on the other hand, a minimum-effort integration can be easily achieved. Selected examples are given, which demonstrate the ability of PIF to provide not only a flexible and reliable tool data interchange format, but also serves well as a TCAD framework database format.
Because of its homogeneity, the combination of LISP as the task level extension language and PIF as the data level interchange format has proven to be a flexible basis for a powerful TCAD framework. The following sections present a few directions, on how this basis could be extended and what will be the important issues in future data level designs.