Semiconductor Wafer Hotspot Detection via Defect Stacking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, detecting and addressing hotspots (defects) on wafers is crucial for quality and yield, but existing methods struggle with precision and efficiency, especially as device dimensions shrink, leading to potential device failures due to small defects.
Innovation Solution
A method and system for detecting hotspots on semiconductor wafers involve inspecting each die, stacking defect signals to create a 3D hotspot distribution, determining specific coordinates with higher peaks, identifying common patterns, and classifying them using a database to distinguish known from unknown patterns, allowing for in-line inspection and automatic yield enhancement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional inspection methods are used, then inspection process is simple, but detection precision is insufficient for small defects
Solution Approach 1:
The inspection system segments the wafer surface into multiple dies and further into multiple hotspot detection regions. Each region is inspected independently and defect signals are stacked corresponding to their spatial coordinates, enabling precise localization of hotspots even at the sub-die level. This segmentation allows the system to achieve high measurement precision for small defects while maintaining manageable system complexity through modular processing.
Solution Approach 2:
The patent transforms the inspection from a two-dimensional surface inspection to a three-dimensional hotspot distribution analysis. By stacking defect signals at their spatial coordinates and creating a 3D distribution map, the system adds a depth dimension that enables precise identification of hotspot locations and patterns, significantly improving detection precision for small defects without proportionally increasing system complexity.
2Measurement precision
If comprehensive hotspot detection is performed, then detection accuracy improves, but inspection time increases
Solution Approach 1:
The system performs preliminary actions by pre-defining multiple hotspot detection regions and pre-establishing the stacking methodology. During inspection, defect signals are automatically stacked at their spatial coordinates without requiring complex real-time analysis. This preliminary preparation enables comprehensive hotspot detection with high accuracy while minimizing inspection time through efficient signal processing.
Solution Approach 2:
The patent creates a virtual copy of the wafer surface in the form of a hotspot distribution map. Instead of physically analyzing every possible defect location, the system inspects each die, extracts defect signals, and creates a stacked distribution map that represents the entire wafer's hotspot pattern. This copying approach enables comprehensive detection accuracy while reducing inspection time by working with the simplified digital representation rather than physical re-inspection.
3Reliability
If unknown hotspot patterns are encountered, then detection completeness improves, but classification difficulty increases
Solution Approach 1:
The system employs self-service by automatically classifying hotspot patterns through comparison with stored known patterns. When unknown patterns are detected, the system uses the accumulated hotspot distribution information and pattern recognition algorithms to automatically identify and classify them, assigning them to appropriate categories without requiring manual intervention. This self-service mechanism improves detection completeness while managing classification complexity through automated decision-making.
Solution Approach 2:
The patent implements feedback by continuously comparing detected hotspot patterns against stored known patterns and using the results to refine future detections. The system learns from previously identified patterns and uses this knowledge to improve the classification of unknown patterns. This feedback loop enables the system to maintain high detection completeness for unknown patterns while reducing classification complexity through iterative improvement and pattern recognition.
Data Source
AI summary
Methods and system for detecting hotspots in semiconductor wafer are provided. At least one semiconductor wafer is inspected to detect a plurality of hotspots of each die in the semiconductor wafer, wherein each of the hotspots has defect coordinates in a layout of the die. The hotspots of the dies are stacked in the layout according to the defect coordinates of the hotspots. A common pattern is obtained according to the stacked hotspots corresponding to a location with specific coordinates in the layout. It is determined whether the common pattern is a known pattern having an individual identification (ID) code. A new ID code is assigned to the common pattern when the common pattern is an unknown pattern.