Pixel-to-Design Alignment Target Selection for Semiconductor Hotspot Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor inspection systems face challenges in accurately detecting defects at smaller design rules due to limitations in pixel-to-design alignment (PDA) methods, which often miss critical hotspot areas, leading to increased care area expansion and reduced sensitivity.
Innovation Solution
The method involves biasing pattern-to-design alignment targets to include hotspot locations, evaluating these targets for repetitive patterns, and using adaptive Fourier filtering to select unique targets with maximum intensity, thereby improving defect location accuracy and reducing nuisance defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current PDA methods are used for target selection, then alignment can be performed, but defect location accuracy deteriorates and care area expansion increases
Solution Approach 1:
The patent applies local quality by differentiating target selection based on location characteristics. Hotspot regions (critical areas) are identified and given priority in target selection, while non-hotspot regions use conventional selection methods. This localized approach ensures that alignment targets are preferentially chosen from areas where accurate defect location is most critical, thereby improving defect location accuracy without uniformly expanding care areas across the entire wafer.
Solution Approach 2:
The patent introduces asymmetry in the target selection process by treating hotspot and non-hotspot regions differently. Instead of uniform target selection across the entire image frame, the system applies asymmetric weighting where hotspot regions contribute more significantly to target selection. This asymmetric approach allows the system to focus alignment resources on critical areas, improving measurement precision where it matters most while avoiding unnecessary care area expansion in less critical regions.
2Measurement precision
If more defects are detected to improve yield management, then sensitivity increases, but the number of nuisance defects increases
Solution Approach 1:
The patent applies local quality by differentiating defect evaluation based on location. Defects detected in hotspot regions are given higher priority and lower threshold for reporting, while defects in non-hotspot regions are evaluated with lower priority. This location-based differentiation allows the system to maintain high sensitivity for critical defects in hotspot areas while filtering out nuisance defects in less critical areas, thereby improving overall defect detection quality without being overwhelmed by false positives.
Solution Approach 2:
The patent inverts the conventional approach by not trying to detect all defects uniformly, but rather by strategically focusing detection sensitivity on hotspot regions. Instead of applying the same detection threshold across the entire wafer, the system inverts the priority structure by giving higher weight to defects in critical areas and lower weight to defects in non-critical areas. This inversion allows the system to maintain high sensitivity where it matters most while tolerating lower sensitivity in less critical regions, effectively reducing nuisance defects.
3Manufacturing precision
If conventional target selection is used, then alignment can be performed, but PDA quality is insufficient for logic care areas
Solution Approach 1:
The patent applies preliminary action by pre-identifying and marking hotspot regions before the target selection process. The system uses design information to predetermined which areas are critical (hotspots) and which are not. This preliminary classification allows the target selection algorithm to efficiently prioritize targets in hotspot regions without requiring complex real-time analysis during the alignment process itself, thereby improving PDA quality for logic care areas while maintaining operational simplicity.
Data Source
AI summary
Techniques and systems to achieve more accurate design alignment to an image by improved pixel-to-design alignment (PDA) target selection are disclosed. PDA targets in an image frame of a die can be biased to include a hotspot location in one of the PDA targets. The PDA targets can be evaluated for repetitive patterns by analyzing the uniqueness of the points used as the PDA targets.