Manufacturability evaluation of deep submicron exposure tools using statistical metrology

Statistical Metrology has been proposed as a technique to extract variability components of an IC process sequence through the combined use of conventional metrology and statistical filtering. We have developed a methodology to decompose and categorize CD variability into individual equipment contributions, specifically the steppers and reticles used in 0.35 /spl mu/m polysilicon gate patterning. Spatial variability was sampled using a reticle designed to collect CD measurements over the exposure field and the wafer. Then, a series of statistical and physical filters were implemented to separate the reticle and stepper contributions to CD variability. Results have shown that CD variability has strong spatial and causal components. Decomposition results are applied towards: (1) Identifying bottlenecks in manufacturability by providing an accurate error budget analysis. (2) Using isolated equipment variability components as a manufacturability metric to benchmark exposure tools. (3) Quantifying the correlation between spatial variability components can be manipulated to improve net process manufacturability.