2.3 Tool Data



next up previous contents
Next: 2.4 Task and Framework Up: 2 TCAD System Data Previous: 2.2 TCAD Users and

2.3 Tool Data

     

Conventional TCAD tools usually expect some kind of physical problem description, which includes:

The logical parameters are all non-physical (not directly related to the physical problem considered by the simulator) parameters used by the simulator like controlling which output should be produced, error margins for solver convergence, or flags selecting a specific model. A further distinction may be made between simulator-independent logical parameters and simulator-dependent logical parameters. In a tightly integrated TCAD simulator output control parameters should ideally be simulator-independent, since for example a doping concentration or an electric field are output of many simulators. Special error margins or model selection flags are inherently simulator-dependent and have to be handled separately for each simulator by the TCAD system.

The physical parameters may be further divided into those related to the geometry, like material type of certain segments or bias boundary conditions, and those describing geometry-independent properties of the physical process, like implantation dose, or environmental temperature and pressure. Also note that these physical parameters can be distributed quantities defined on a grid on the physical geometry (e.g. doping profiles), or discrete physical quantities as mentioned before.

On return, TCAD tools produce a physical state of a wafer region or a sequence of such states, which may contain:

Modified geometries are usually produced by process simulators simulating process steps like etching, deposition and oxidation. However, even other simulators may modify the simulated geometry, although not qualitatively but quantitatively, e.g. when inserting boundary points during grid refinement.

Distributed physical quantities can for example be doping profiles produced from an implantation or diffusion process simulator, as well as an electric field or current distribution produced by a device simulator.

Discrete physical quantities (lumped quantities, integral quantities) are for example produced by device simulators when reporting the total integral current flowing through a device contact.

The logical geometry will usually be stored as a hierarchically organized set of points, lines, faces and solids. This requires a basic reference mechanism on the TCAD data level. The physical geometry, consisting of simulation domain (or segment) and boundary descriptions, will reference some elements of the logical geometry, again relying on that mechanism.

Input parameters, be they logical or physical ones, can consist of just a primitive data value like a number or a string, or may be a (multidimensional) array of those values. Therefore the data level must support those arrays and indexing into those arrays. Since parameters are named, a naming mechanism must be provided too, through which objects can be identified and looked up by their (unique) name. Since a parameter is not always a real value, many different data types must be supported. Furthermore, if distributed quantities should be represented in a homogeneous way, a data type handling mechanism (i.e. providing means for an application to recognize and handle quantities with different data types) and a unit handling mechanism for the physical units of the respective distributed quantity should be available.



next up previous contents
Next: 2.4 Task and Framework Up: 2 TCAD System Data Previous: 2.2 TCAD Users and



Martin Stiftinger
Tue Nov 29 19:41:50 MET 1994