Multiple applications of an automatic defect review SEM in semiconductor manufacturing yield enhancement

An automatic defect review SEM (DRSEM) has many improvements over traditional manual review SEMs. These include high throughput, consistent image quality, reduced operator training, hands off operation, and automatic defect classification. These features have made the DRSEM useful in multiple aspects of a yield enhancement strategy. Optical classification has been replaced by automatic defect classification using the DRSEM at multiple baseline inspection steps. Improvements have been observed in both classification accuracy and consistency over manual classification. In addition, the topographical information obtained by the DRSEM has enabled the correct distinction of optically similar defects, such as partial height metal pattern and tungsten filled ILD holes. The DRSEM has also been successfully used for selective review of ambiguous or unknown defects seen during optical review of defect excursions. The high throughput of the DRSEM has made it practical to perform SEM review of full wafer lots in the evaluation of defect reduction experiments. Defect reduction process change experiments are typically evaluated based on defect density, which is susceptible to previous layer defects. The DRSEM has proven most useful in separating the defects of interest from previous layer defects, and speeding experimental analysis. Finally, the DRSEM has help reduce recipe setup time on defect inspection tools by assisting in the evaluation of true defects versus false or nuisance counts.