Evaluation of metallization systems with test structures and yield modeling

Abstract

Comb test structures are used to evaluate three metal systems for IC use. Defects are analyzed based on density alone using a simple noncluster Poisson yield model and for clustering as well as using a negative binomial yield model. Results show that to accurately evaluate the highest yielding metal system at VLSI chip sizes, defects must be separated by type and evaluated in terms of clustering as well as defect density.<>