Integration at the data level is concerned with the exchange of simulation results between simulation tools. The two fundamental problems in this domain arise from different data formats and from different information models employed for representing simulation data. In general, the first problem is easily solved by a data format converter that maps a given data format onto another one. The second problem poses by far the greater challenge. Care has to be taken to correctly convey all information represented in one model to another one and vice versa. In general, this problem cannot be solved completely, and a certain loss of information needs to be reckoned with. However, in some situations, data bypassing can help to limit the loss of information to the position of a particular tool in a sequence of tool steps; after that step, information can be regained by appropriate merging operations (Figure 4.3).
Figure 4.3:
Different information models of tools A and
B lead to loss of information for tool B. Bypassing data
regains that loss for tools subsequent to B.
For the general case, it is more instructive to focus on the subsequent modifications of a wafer state model instead of thinking of chaining tools together. In Figure 4.4, the wafer model is modified from step to step by the operation of tools whose respective inputs and outputs are derived from, and merged with, the current contents of the model. According to this scheme, a tool's input may be completely unrelated to state of the model.
Figure 4.4:
Wafer model and tool input and output: The
state of the model is iteratively modified by the operation of tools.