2.3.1 Statistical Analysis

The variations of the characteristic properties of the devices which are included in an integrated circuit are critical for an acceptable yield of the fabrication process. Unless these variations are under control and can be kept within certain ranges, they will cause unacceptably large numbers of defects. Therefore, it is important to identify the variation margins of the device parameters which are caused by statistical variations during fabrication.

The model of an IC fabrication technology, as it has been discussed before, can be used to investigate the impact of statistical variations of fabrication parameters on the characteristic device parameters of an integrated circuit [1,75,77,83,100], as shown in Figure 2.6. By introducing artificial statistical variations (standard deviation $\sigma_{x,n}$) to the input parameters of this model, we are able to derive the resulting variation margins of the output quantities ($\sigma_{y,m}$). However, such an analysis represents a considerable computational effort since it is necessary to carry out a large number of model evaluations in order to achieve representative statistics.