Intelligent defect screening and sampling method

Abstract

The invention discloses an intelligent defect screening and sampling method, which comprises the following steps of: previously processing a design layout into a plurality of pattern groups based on the layout; dividing the design layout into a plurality of crystal cells; overlapping the crystal cells which belong to the same pattern group; extracting a plurality of pieces of defect data of a plurality of defects on a wafer; mapping a plurality of defects to the overlapped pattern groups to establish a plurality of defect synthesis pattern groups based on the layout; performing layout pattern matching on the defect synthesis pattern groups; performing some defect sampling selection rules on each defect synthesis pattern group to judge a potential systematic defect priority order; classifying the defect synthesis pattern groups into different defect types according to the potential systematic defect priority order; inspecting the defect synthesis pattern groups of different sampling numbers to obtain a defect image file according to the defect types; and performing defect generation rate analysis on the defect image file to generate a defect pattern database or perform defect generation rate prediction. According to the intelligent defect screening and sampling method, the defection rate of systematic defects and defect inspection efficiency can be increased.

Description

technical field [0001] The invention relates to a defect analysis method, and in particular to a defect screening and sampling method. Background technique [0002] The manufacturing process of integrated circuits includes thin film deposition, photomask exposure, photolithography, etching, etc. During the manufacturing process, random particle defects and systematic defects inevitably occur. It will affect the yield rate of the product, and the yield rate of the product is related to the cost of the die. [0003] As the feature size of design layouts shrinks, yield-related defects become smaller. In order to extract all the killer defects on the wafer, the foundry must increase the sensitivity of its scanning and inspection equipment. Therefore, the number of detected defects will also increase, however, in fact, the percentage of non-fatal defects to all detected defects will also increase. In addition, in order to identify the real potential defects, the foundry will us...

AI Technical Summary

Problems solved by technology

However, foundry time is limited, and the performance of scanning electron microscopes used for inspection may limit the number of observations per wafer

Therefore, the current technology has difficulty in identifying important types of defects on the product, and there is a risk of missing the identification of fatal defects

[0004] Without a fast and innovative method to pre-identify systemic defects, foundries will face huge production declines and spend a lot of learning time in mass production

Images