Comparing the effectiveness of deterministic bridge fault and multiple-detect stuck fault patterns for physical bridge defects: A simulation and silicon study

Abstract

Shrinking feature size and increased wire density increase the likelihood of occurrence of bridge related defects. N- detect based pattern sets are used commonly to improve the detection of bridge defects. However, achieving high bridge coverage requires deterministic bridge sites extraction from physical layout and bridge fault pattern generation. In this paper, we present a comprehensive comparative analysis about the effectiveness of deterministic bridge fault patterns and n-detect patterns for two large designs (90 and 65nm). We show that extracting different types of bridge faults is required as they represent different unique defect sites. Simulation results show that n-detect patterns have very poor bridge coverage performance and commonly used metric bridge coverage estimate (BCE) does not relate to the true bridge fault coverage. Finally, we discuss the DPPM impact for deterministic bridge fault and n-detect stuck-at patterns for the 90nm design.