Sample Observation Device for Semiconductor Wafer Defect Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated defect review (ADR) systems in semiconductor wafer manufacturing face challenges in accurately detecting defects due to initial processing parameter settings, leading to reduced model accuracy and incomplete representation of defects in high-picture quality images.
Innovation Solution
A sample observation device that learns a high-picture quality image estimation model by acquiring and adjusting defect detection parameters, using a combination of low- and high-picture quality images to improve defect detection accuracy, allowing for precise imaging and parameter adjustment based on defect positions and features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple high-picture quality images are captured for each defect position, then defect detection accuracy is improved, but processing time and system complexity increase
Solution Approach 1:
The system performs preliminary defect detection using low-picture quality images to identify defect positions before capturing high-picture quality images. This preliminary action filters out non-defect areas, so that high-resolution imaging is only performed at locations where defects are suspected, significantly reducing the total number of high-picture quality images that need to be processed while maintaining high defect detection accuracy
Solution Approach 2:
The inspection process is segmented into two distinct stages: first, a wide-field low-magnification scan to detect defect positions; second, targeted high-magnification imaging only at the detected defect positions. This segmentation allows the system to balance between comprehensive defect detection and efficient resource utilization, avoiding the time cost of capturing high-picture quality images across the entire wafer surface
2Measurement precision
If multiple high-picture quality images are captured for each defect position, then defect detection accuracy is improved, but device complexity increases
Solution Approach 1:
The sample observation device is designed with multi-functionality, integrating both low-magnification wide-field imaging capability and high-magnification detailed imaging capability into a single system. This allows the device to perform both defect detection and defect verification functions without requiring separate equipment, thereby improving defect detection accuracy while controlling system complexity through functional integration
Solution Approach 2:
The system uses low-picture quality images as a preliminary screening tool to identify defect positions before proceeding to high-picture quality imaging. This two-stage approach simplifies the overall system architecture by using a coarse-to-fine strategy, where the first stage guides the second stage, reducing the burden on the high-resolution imaging system while maintaining high detection accuracy
3Productivity
If defect detection parameters are adjusted using initial processing parameter settings, then processing speed is maintained, but model accuracy decreases
Solution Approach 1:
The system implements a feedback mechanism where the results from low-picture quality image analysis are used to guide and adjust the parameters for high-picture quality image capture. The defect detection results from the first stage provide feedback that optimizes the imaging parameters (such as magnification, focus, and exposure) for the second stage, ensuring that high-picture quality images are captured with optimal settings tailored to the specific defect characteristics, thereby improving model accuracy without significantly impacting processing speed
Data Source
AI summary
In learning processing performed before sample observation processing (steps S705 to S708), the sample observation device acquires a low-picture quality learning image under a first imaging condition for each defect position indicated by defect position information, determines an imaging count of a plurality of high-picture quality learning images associated with the low-picture quality learning image for each defect position and a plurality of imaging points based on a set value of the imaging count, acquires the plurality of high-picture quality learning images under a second imaging condition (step S702), learns a high-picture quality image estimation model using the low-picture quality learning image and the plurality of high-picture quality learning images (step S703), and adjusts a parameter related to the defect detection in the sample observation processing using the high-picture quality image estimation model (step S704).