Cache RAM inductive fault analysis with fab defect modeling

This paper describes how an inductive fault analysis (IFA) method was used in determining the expected yield and test algorithm effectiveness in SRAMs. One portion of the paper describes the process of gathering and correlating defect data from different sources to get a meaningful and accurate defect distribution. This data is used in the IFA software to weight the probability of faults occurring. With the fault data from two SRAM cells the yield of those devices is estimated and compared with actual yield. The effectiveness of certain test patterns is also evaluated based on the probable faults for those circuits.