Comparing results from defect-tolerant yield models

Abstract

To date, many models have been developed to predict the yield of defect-tolerant integrated circuits (ICs). In this paper, results obtained from several of these models are compared. Their sensitivity to various model parameters is also examined. These results lead one to conclude that, despite differences in the predicted amount of redundancy, it may be possible to obtain good solutions. The differences in the solutions come from the models as well as from the parameters used in these models, and solutions are said to be good when the resulting figures of merit are rather insensitive. Consequently, a simple method is proposed to select the number of spares to add in defect-tolerant ICs.<>