2. Data Model Aspects for TCAD Simulators

Many tools to simulate the various steps of a semiconductor fabrication process are available today. Some of those come from universities, others are developed and distributed commercially. As pointed out in the previous chapter, to date there is no accepted standard as to which data are crucial to carry out coupled TCAD simulations. As a matter of fact existing solutions are based on a file format instead of a data format. The difference between file format and data format lies in the algorithms that are associated with data structures. Per definition, a data structure is defined via its representation (data in memory, data on disk) and a set of operations that are valid on that representation. On the other hand, a file is only a means to persistently store data on a computer, thus it is just one representation of the data.

To make things worse, the existing file formats are usually not even semantically compatible to each other. As some of the file formats are optimized for a certain purpose like, e.g. device simulation, they are not intended to pass data among process simulators. As a consequence, there does not exist a rigorous solution to couple existing tools in a way such that one particular process simulation step could be exchanged by another. Thus, if interoperability of such tools is desired, the only solution for the user is to use data wrapping tools to achieve at least partial compatibility on the file level.

If we look at the needs from a TCAD user's perspective, there are several major issues to take into account. According to [22] three keys to a successful operation of TCAD are

I would like to add another point to the above list of issues, namely the importance of tool interoperability. There has been a strong user driven urge to support file wrappers that convert data between the file formats of different vendors.

The above sketched issues lead to the following requirements that must be met by a TCAD data model.