Defect Segregation Using CAD Identifiers for IC Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing number of defects reported by wafer inspection tools due to higher integration densities makes it impractical to review large defects effectively, leading to high false alarm rates and difficulty in identifying real defect root causes, which slows down the manufacturing process and reduces yield.
Innovation Solution
The proposed solution involves segregating defects based on Computer-Aided Design (CAD) identifiers associated with the defects, using techniques such as machine learning clustering to group defects and perform additional processing, such as sampling and root cause analysis, to efficiently identify and filter real defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If wafer inspection tools are used to detect defects at higher integration densities, then defect detection capability is improved, but the number of reported defects increases sharply making review impractical
Solution Approach 1:
The patent segments the large set of defects into smaller defect groups based on CAD identifier characteristics. By dividing defects into groups sharing common CAD identifiers (such as cell instances, polygons, or design structures), the system makes the review process manageable by presenting only representative defects from each group rather than all defects individually.
Solution Approach 2:
The patent introduces CAD identifiers as an intermediary between defect detection and defect review. These identifiers serve as a filtering and categorization mechanism that bridges the gap between the large volume of detected defects and the practical needs of reviewers, enabling efficient segmentation and prioritization.
2Reliability
If all defects are reviewed individually, then comprehensive defect analysis is achieved, but resource consumption and time increase significantly
Solution Approach 1:
The patent merges defects that share common CAD identifiers into defect groups. By combining defects with identical or overlapping CAD characteristics into single groups, the system maintains comprehensive analysis coverage while reducing the total number of review items, thus saving time and resources.
Solution Approach 2:
The patent applies partial action by reviewing representative defects from each group rather than every single defect. This selective approach provides sufficient information for root cause analysis without the excessive time investment required for complete individual review of all defects.
3Productivity
If defect groups are created based on CAD identifiers, then defect segregation efficiency is improved, but system complexity increases due to additional processing steps
Solution Approach 1:
The patent performs preliminary action by pre-segmenting defects into groups based on CAD identifiers before the actual review process. This advance organization reduces the complexity during the review phase, as defects are already categorized and ready for efficient processing.
Solution Approach 2:
The patent uses CAD identifiers as a copying mechanism to replicate defect groupings across multiple defects. By copying the same CAD identifier information across related defects, the system efficiently creates consistent groups without requiring complex real-time analysis during review.
Data Source
AI summary
For each defect in a set of defects, the defect may be associated with a defect attribute constructed from a set of computer-aided design (CAD) identifiers associated with polygons in an integrated circuit (IC) design that overlap with a defect area of the defect. Next, the set of defects may be segregated into defect groups based on the associated defect attributes. The defect groups may be used to perform additional processing on the set of defects.