Repeater Defect Detection Through Multi-Die Image Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting repeater defects in semiconductor manufacturing are limited by sensitivity due to reliance on single die image analysis, which is prone to noise and nuisances, and current systems fail to effectively exploit spatial correlation across multiple die images.
Innovation Solution
A method that combines and fuses information from multiple die images using spatial correlation to enhance the signal-to-noise ratio, specifically through die fusion and statistical image processing, to improve defect detection sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If single die image analysis is used for repeater defect detection, then the inspection process is simple, but the sensitivity is limited due to noise and nuisances
Solution Approach 1:
The patent combines multiple die images from different inspection channels to form a composite image for repeater defect detection. This merging approach amplifies the signal from repeating defects while averaging out random noise and nuisances, thereby improving detection sensitivity without requiring fundamentally new inspection equipment
Solution Approach 2:
The patent transitions from analyzing single die images in isolation to analyzing multiple die images across different spatial dimensions and inspection channels. By exploiting the periodic spatial correlation of repeater defects across multiple dies, the system extracts defect patterns that are not visible in individual die images, effectively adding a dimensional aspect to the analysis
2Measurement precision
If multiple die images are fused using spatial correlation, then the signal-to-noise ratio is enhanced, but the processing complexity increases
Solution Approach 1:
The patent extracts the periodic spatial correlation pattern from multiple die images by analyzing the repeating defect signatures across different dies. This extraction process isolates the characteristic repeater defect signal from the background noise and non-repeating defects, enhancing the signal-to-noise ratio through statistical pattern recognition
Solution Approach 2:
The patent introduces an intermediate processing step that fuses information from multiple inspection channels before final defect detection. This intermediary fusion process acts as a bridge that combines the strengths of different inspection channels while canceling out their individual noise characteristics, achieving superior signal-to-noise ratio
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Repeater defects on a wafer can be detected by fusing multiple die images. In an instance, multiple die images are statistically fused to form a die-fused image. Each of the die images can be of a different die on a wafer. A presence of a repeater defect is detected in the die-fused image. The die images can be generated using a laser-scanning system or other systems.