Semiconductor Defect Examination With ML-Generated Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection methods in semiconductor manufacturing are inefficient due to high image acquisition times and processing requirements, particularly in die-to-reference detection, which affects throughput and sensitivity.
Innovation Solution
A computerized system using a machine learning model, trained on pairs of defective and defect-free images, generates a synthetic reference image to perform defect examination, reducing the need for additional image acquisition and pre-processing, thereby improving detection throughput and sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional die-to-reference detection methods are used, then defect examination can be performed, but image acquisition time increases and throughput decreases
Solution Approach 1:
The system performs preliminary action by acquiring a reference image before the runtime inspection. This pre-acquired reference image is then used during defect examination, eliminating the need to acquire reference images in real-time and significantly reducing the total image acquisition time while maintaining detection sensitivity.
Solution Approach 2:
The patent uses a copy of the reference image (acquired in advance) instead of requiring a fresh reference image acquisition during each defect examination. This copied reference image enables rapid defect detection without the time-consuming real-time reference acquisition, thereby improving throughput while preserving measurement precision.
2Measurement precision
If additional image acquisition and pre-processing steps are performed, then defect examination accuracy improves, but processing time increases
Solution Approach 1:
The system performs all necessary image acquisition and pre-processing actions in advance before runtime defect examination. The reference image is acquired and processed beforehand, so that during actual defect examination, no additional time-consuming acquisition or pre-processing is needed, thus reducing processing time while maintaining accuracy.
Solution Approach 2:
The pre-acquired reference image serves itself by being directly utilized during defect examination without requiring additional processing steps. This self-service approach eliminates redundant processing operations and reduces overall processing time while preserving defect detection accuracy.
3Reliability
If runtime reference image acquisition is performed, then defect examination can be conducted, but acquisition time increases
Solution Approach 1:
The system performs the reference image acquisition in advance before runtime defect examination. This preliminary action ensures that the reference image is ready for immediate use during defect examination, eliminating the need for time-consuming real-time acquisition while maintaining the reliability of the examination process.
Data Source
AI summary
There is provided a system and method for defect examination on a semiconductor specimen. The method comprises obtaining a runtime image of the semiconductor specimen, generating a reference image based on the runtime image using a machine learning (ML) model, and performing defect examination on the runtime image using the generated reference image. The ML model is previously trained during setup using a training set comprising one or more pairs of training images, each pair including a defective image and a corresponding defect-free image. The training comprises, for each pair, processing the defective image by the ML model to obtain a predicted image, and optimizing the ML model to minimize a difference between the predicted image and the defect-free image.