Semiconductor Defect Review Using Synthesized Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic defect review (ADR) systems for semiconductor wafers face low throughput and reliability issues due to the need for taking reference images, especially when dealing with large numbers of defects and complex patterns, which can lead to erroneous detection and increased processing time.
Innovation Solution
The method involves synthesizing a reference image from the defect image itself, allowing for the omission of the reference image step, and using this synthesized image to detect defects, thereby improving throughput and preventing erroneous detection by determining the reliability of detected defects based on probability distributions and feature analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference images are taken and stored for each defect review, then defect detection accuracy is improved, but processing time increases and throughput decreases
Solution Approach 1:
The patent pre-calculates and stores feature amounts (such as area, perimeter, circularity) for all possible pattern types in a lookup table during system initialization. This preliminary action eliminates the need to calculate these features during actual defect review, enabling rapid comparison between defect images and reference patterns while maintaining high detection accuracy.
Solution Approach 2:
The patent creates simplified reference images by extracting only the essential pattern features (geometric characteristics, intensity distributions) from original wafer patterns, storing these simplified copies in a database. During defect review, only these compact reference copies are retrieved and compared, dramatically reducing data processing requirements while preserving the ability to accurately identify defects.
2Reliability
If multiple imaging steps are performed (reference image acquisition, defect image acquisition, comparison), then defect identification reliability is improved, but processing complexity and time increase
Solution Approach 1:
The patent merges the reference image acquisition and defect image comparison steps into a single processing operation. By pre-storing reference pattern data and directly comparing defect images against these references using automated feature extraction and matching algorithms, the system eliminates the need for separate manual reference image creation and comparison steps, reducing process complexity while maintaining reliable defect identification.
Solution Approach 2:
The system performs automated defect classification by having the defect review apparatus itself extract features, compare patterns, and classify defects without requiring external reference images or manual intervention. The apparatus uses its own stored pattern database and automated algorithms to independently identify and classify defects, eliminating dependency on external imaging steps.
3Measurement precision
If high magnification imaging is performed for all defects, then defect detail review quality is improved, but processing time and data volume increase
Solution Approach 1:
The patent applies different processing qualities to different defect types based on their characteristics. For defects that can be reliably identified using lower magnification and simplified feature comparison (such as obvious particles or large defects), the system uses rapid low-magnification review. For defects requiring detailed structural analysis, high magnification imaging is selectively applied. This localized quality adjustment maintains review quality for critical defects while reducing overall processing time.
Solution Approach 2:
The patent performs partial high-magnification imaging only for defects that cannot be confidently classified at lower magnification. The system first attempts rapid classification using low-magnification images and pre-stored pattern references, then selectively applies high-magnification imaging only to ambiguous or critical cases, avoiding the time cost of universal high-magnification imaging while ensuring adequate review quality for problematic defects.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables high-throughput automatic defect review with improved reliability by reducing unnecessary imaging steps and enhancing the accuracy of defect detection, allowing for real-time processing without the need for additional reference images.
Implementation Method 1
obtaining an image including a defect on the semiconductor device detected by a detection device using a scanning electron microscope
Data Source
AI summary
A method and apparatus for reviewing defects of a semiconductor device is provided which involves detecting a defect on a SEM image taken at low magnification, and reviewing the defect on a SEM image taken at high magnification, and which can review a lot of defects in a short period of time thereby to improve the efficiency of defect review. In the present invention, the method for reviewing defects of a semiconductor device includes the steps of obtaining an image including a defect on the semiconductor device detected by a detection device by use of a scanning electron microscope at a first magnification, making a reference image from the image including the defect obtained at the first magnification, detecting the defect by comparing the image including the defect obtained at the first magnification to the reference image made from the image including the defect at the first magnification, and taking an image of the detected defect at a second magnification that is larger than the first magnification.