A method for wafer color image consistency evaluation and defect detection

By performing consistency evaluation and differential processing on wafer color images, the problem of grayscale cameras being unable to identify wafer defects and hardware differences was solved, achieving high-precision defect detection and consistency of results between equipment, and simplifying the calculation process.

CN120634971BActive Publication Date: 2026-11-17KINGSEMI CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202510679146.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-05-26
Publication Date
2026-11-17
Estimated Expiration
2045-05-26

AI Technical Summary

Technical Problem

Existing grayscale cameras cannot accurately identify wafer defects, and hardware differences lead to inconsistent results between equipment. Traditional image consistency evaluation methods have high computational complexity or limited effectiveness, making it difficult to meet the semiconductor industry's demand for high-precision inspection.

Method used

By evaluating the consistency of wafer color images across three dimensions—image quality, physical properties, and pixel differences—the machine hardware parameters are adjusted, a standard template is created, and differential analysis is performed to achieve accurate location of defective areas.

Benefits of technology

It improves the accuracy of defect detection and the consistency of results between machines, simplifies the calculation process, and meets the semiconductor industry's demand for high-quality color image acquisition and inspection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120634971B_ABST
    Figure CN120634971B_ABST
Patent Text Reader

Abstract

This invention discloses a method for evaluating the consistency and detecting defects in wafer color images. First, the consistency of the wafer color image is evaluated from three dimensions: image quality, physical properties, and pixel difference. Based on this evaluation, the hardware of the monitoring equipment is adjusted and monitored to ensure the consistency and high quality of the wafer color image. During detection, the image of the wafer under test is compared with a standard template based on the smallest repeating unit of the die. Then, multi-channel weighted integration is used to finally determine the defect region. This method can improve the accuracy of defect detection and the consistency of results from different equipment. Furthermore, the implementation and calculation process are simple and easy to perform, further meeting the needs of the semiconductor industry for automated optical inspection.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention provides a method for evaluating the consistency and detecting defects in wafer color images, applicable to the field of automated optical inspection of semiconductors. Background Technology

[0002] In the semiconductor industry, wafer manufacturing is extremely complex and delicate, typically involving hundreds or even thousands of process steps. The production process inevitably suffers from issues such as raw material contamination, environmental particles, and process risks, leading to various defects on the wafer surface, including scratches, cracks, particles, and bubbles. These defects impact production efficiency, product performance, and final yield. The industry uses automated optical inspection technology to monitor macroscopic defects during wafer manufacturing; this involves acquiring images of the wafer surface and identifying defects through image processing.

[0003] Due to considerations such as image resolution, scanning speed, and cost-effectiveness, grayscale cameras are mostly used in the industrial vision field. However, to further observe defect morphology, identify defect types, and improve detection accuracy, automated optical inspection of wafers is transitioning to three-dimensional color imaging. This also means that factors such as machine lighting conditions, lens distortion, and shooting angle have a greater impact on image consistency. To eliminate hardware differences between machines and facilitate subsequent inspection, consistency evaluation and hardware parameter adjustments must be performed during the image acquisition stage to obtain consistent and high-quality wafer color images at the machine. Image consistency evaluation generally first calculates information such as the pixel distribution or key features of the image, and then quantifies differences based on statistical measures. Typical methods include histogram methods, mean squared error methods, and feature matching methods. Histogram methods directly compare the pixel histograms of two images, without considering the specific pixel distribution and failing to capture local information. Mean squared error methods directly calculate the squared mean of the differences between corresponding pixels, focusing primarily on pixel-level differences while ignoring structural and semantic information, and are highly sensitive to geometric transformations and noise. Feature matching methods assess image consistency by calculating the number and similarity of feature points between images, overcoming some of the shortcomings of the first two methods, but with high computational complexity and results often differing from visual perception. Commonly used image processing methods can be broadly categorized into traditional methods and deep learning. The latter has developed rapidly in recent years, achieving remarkable processing results, but requires significant manual annotation and computational resources, and struggles to handle small samples and unknown defects, thus its practicality in this field remains limited. For general applicability, current methods primarily identify defects by comparing the image to be tested with a template or between different units. Summary of the Invention

[0004] To address the problem of traditional grayscale cameras failing to accurately identify wafer defects and hardware differences leading to inconsistent results across different equipment, this invention proposes a method for wafer color image consistency evaluation and defect detection. First, the consistency of the wafer color image is evaluated from three dimensions: image quality, physical properties, and pixel difference. Based on this, camera parameters and equipment hardware conditions such as illumination are monitored and adjusted to ensure high-quality wafer color images. Then, using the die as the basic unit, a standard template is extracted, and the three channels are differentially analyzed to initially obtain difference information. Finally, through weighted integration of each channel, the defect region is determined. This method improves the accuracy of defect detection and the consistency of results across different equipment, and its implementation and calculation process are simple and easy to perform, further meeting the needs of the semiconductor industry for automated optical inspection.

[0005] The technical solution adopted in this invention is: a method for wafer color image consistency evaluation and defect detection, comprising the following steps:

[0006] Step 1: Combining various operating conditions, use wafer color image consistency evaluation to measure and monitor the hardware of the monitoring equipment to obtain the hardware parameters related to equipment consistency that need to be adjusted.

[0007] Step 2: For the batch of wafers to be tested, first create a standard template, then acquire a color image of the wafer to be tested, and perform differential integration with the standard template to obtain the defect area.

[0008] The various operating conditions include:

[0009] After the first device is installed and the hardware parameters are calibrated, a wafer sample image is acquired as a reference image. The new installation station initially acquires an image of the same wafer sample and evaluates its consistency with the reference image, thereby adjusting the hardware parameters until the score meets the threshold requirements.

[0010] b. For the images of the wafer samples or wafers in the same batch that are periodically collected during the operation of the machine, their consistency is evaluated with other machines or historical images, thereby monitoring the machine hardware parameters. If the parameters exceed the threshold, the hardware parameters are readjusted.

[0011] Step one, color image consistency evaluation, includes:

[0012] 1) Calculate the physical properties and image quality / clarity of the wafer image for both the image under test and the reference image. Pixel difference ;

[0013] The physical properties include: the center of the circle. ,radius Roundness with deflection angle ;

[0014] 2) First, calculate the image quality consistency score between the two images:

[0015] Among them, weight , , , These are image quality sharpness, signal-to-noise ratio, and structural similarity, respectively.

[0016] Then calculate the physical property consistency score between the two images:

[0017] Among them, weight ;

[0018] Finally, the mean squared error method was used to calculate the difference between the two images. Set weights Calculate the final consistency score: ;

[0019] 3) If S q If the image exceeds the set threshold, the camera focal length needs to be adjusted to improve image sharpness, and camera flat field correction needs to be applied to improve the signal-to-noise ratio.

[0020] If S p If the image size exceeds the set threshold, the imaging distance and camera pitch angle need to be adjusted to improve the image size and roundness.

[0021] If S q and S p If the conditions are met and the final S score is greater than the set threshold, the camera optical axis direction needs to be adjusted to improve the consistency between units.

[0022] The method for calculating the physical properties of the image includes:

[0023] Image preprocessing converts the acquired color image into a grayscale image, and extracts the wafer region using the Blob algorithm;

[0024] Create uniformly around the wafer circumference A measuring caliper is used to perform edge detection on the caliper area to obtain... The center of the circle is obtained by fitting a circle to each edge point using the least squares method and the RANSAC algorithm. ,radius ;

[0025] pass The wafer profile is obtained by fitting the edge points, and the roundness is calculated by the area integral between the wafer profile and the fitted circle. The integral of the product of the area between the wafer profile and the fitted circle and its corresponding angle is the deflection angle. ;

[0026] The method for calculating image quality includes:

[0027] Image preprocessing involves converting the acquired color image to a grayscale image, extracting the wafer region using the Blob algorithm, and calculating the width of the wafer region. With Gao ;

[0028] The Sobel operator is used to calculate the image gradient in the horizontal and vertical directions at each point in the wafer region. and Calculate the gradient of the composite image Calculate image sharpness ;

[0029] Calculate the signal-to-noise ratio of the entire image. ,in This is the proportionality coefficient. and These are the maximum and minimum values ​​of the local variance for all pixels in the image, respectively.

[0030] The method for calculating the pixel difference includes:

[0031] Image preprocessing converts the acquired color image into a grayscale image, extracts the wafer region using the Blob algorithm, translates and aligns the centroid, and rotates and aligns the wafer Notch port.

[0032] The three channels in the wafer region are differentially processed and then combined: ,in and These are the pixel values ​​of a point in the two images, respectively. Represents the number of channels.

[0033] The second defect detection step includes:

[0034] 1) Fix the hardware parameters related to the consistency of the fixed machine, adjust the illumination, acquire color images of defect-free wafer samples, preprocess and divide them into the smallest units, and fuse them to generate a standard template;

[0035] 2) Fix the hardware parameters related to the consistency of the fixed machine, adjust the illumination, acquire the color image of the wafer to be tested, preprocess and segment the smallest unit, and perform differential integration with the above template to obtain the defect area.

[0036] The adjustment of illuminance includes:

[0037] A color image of the wafer is acquired at a preset illumination level, converted into a grayscale image, and the number of pixels with a grayscale value greater than a threshold is counted. Less than the empirical threshold Increase illuminance greater than the empirical threshold Then reduce the illuminance ,in and This is an empirical coefficient;

[0038] After adjusting the illumination, acquire the wafer image again, and repeat the above steps until... If the above standard is not reached before the maximum number of adjustments is given, the illuminance adjustment will fail and the machine will issue an alarm.

[0039] The defect detection also includes:

[0040] a) Image preprocessing process: First, the acquired color image is converted into a grayscale image. Then, the wafer region is extracted from the grayscale image as the ROI (Region of Interest) using the Blob algorithm.

[0041] b) Convert the grayscale image of the above region to the frequency domain, analyze the maximum frequency component, determine the size of the smallest repeating unit by combining the wafer layout information, identify the boundary of the repeating unit by using the sliding window method, and accurately segment the smallest unit of the wafer.

[0042] c) Select several repeating units from the segmented color image of the defect-free wafer sample, covering different radii and positions of the wafer;

[0043] The average width and height of the above units are taken as the size of the standard template.

[0044] d) By using bilinear interpolation and shape correction, the smallest unit obtained from the segmentation of the color image of the wafer under test is unified to the size of the standard template, resulting in several test units of different sizes;

[0045] e) The three channels of the unit under test are differentially compared with the standard template, and the regions with differences and areas greater than the threshold are marked. The coordinates of the above regions in the wafer image are calculated. The image and coordinates of the region are output as the defect detection result.

[0046] The standard template is a series of standard cells of different sizes obtained from a pre-collected defect-free wafer through steps a)-d).

[0047] The present invention has the following beneficial effects and advantages:

[0048] To address the problem of traditional grayscale cameras failing to accurately identify wafer defects and hardware differences leading to inconsistent results across different equipment, this invention proposes a method for wafer color image consistency evaluation and defect detection. First, the consistency of the wafer color image is evaluated from three dimensions: image quality, physical properties, and pixel difference. Based on this, camera parameters and equipment hardware conditions such as illumination are monitored and adjusted to ensure high-quality wafer color images. Then, using the die as the basic unit, a standard template is extracted, and the three channels are differentially analyzed to initially obtain difference information. Finally, through weighted integration of each channel, the defect region is determined. This method improves the accuracy of defect detection and the consistency of results across different equipment, and its implementation and calculation process are simple and easy to perform, further meeting the needs of the semiconductor industry for automated optical inspection. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the hardware platform of the present invention.

[0050] Figure 2 This is a flowchart of the defect detection process of the present invention.

[0051] Figure 3 This is a comparison image of the template wafer and the test wafer of the present invention.

[0052] Figure 4 This is a schematic diagram illustrating the detection of local defects in a wafer according to the present invention.

[0053] In the figure, 1 is the wafer station, 2 is the two-dimensional sliding rail module, 3 is the reflector, 4 is the line scan camera, 5 is the light source, 6 is the wafer under test, A is the camera pitch angle, B is the optical axis direction, and d is the imaging distance. Detailed Implementation

[0054] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0056] like Figure 1 The diagram shown is a schematic of the hardware platform of this invention. The hardware platform includes a wafer workstation 1, a two-dimensional sliding rail module 2, a reflector 3, a line scan camera 4, and a light source 5. Among these,

[0057] Wafer station 1 is used to place the wafer to be tested 6, which is mounted on a two-dimensional slide rail module. The wafer station moves in a plane on the two-dimensional slide rail module 2, changing the local viewing angle area of ​​the wafer entering the camera, so that the line scan camera can scan the entire area of ​​the wafer to be tested; finally, a complete wafer image is stitched together.

[0058] Reflector 3 is placed above the wafer under test to shorten the optical path and project the wafer image onto the line scan camera;

[0059] Line scan camera 4 is used to capture partial projection images of the wafer under test and output complete wafer images.

[0060] Light source 5 is placed above the wafer under test to provide suitable illumination; a white strip light source is used to ensure the uniformity of the wafer surface imaging.

[0061] The present invention provides a wafer color image consistency evaluation and defect detection method, comprising:

[0062] Step 1: Measure and monitor the equipment hardware through wafer color image consistency evaluation, including the matching status between equipment (ensuring high consistency in parameters such as size, centroid, roundness, and sharpness of the acquired wafer images between equipment) and the stability of the equipment itself (ensuring uniformity of light source imaging, normal camera acquisition frequency, etc.); determine the hardware parameters related to equipment consistency that need to be adjusted, including camera focal length f, camera flat field correction C, imaging distance d, camera pitch angle A, and optical axis direction B;

[0063] Step 2: For the batch of wafers to be tested, first create a standard template, then acquire a color image of the wafer to be tested, and perform differential integration with the standard template to obtain the defect area.

[0064] In step one, after the first device is installed and the hardware parameters are calibrated, a wafer sample image is collected as a reference image. The newly installed device initially collects images of the same wafer sample and evaluates their consistency with the reference image. The hardware parameters are then adjusted until the score meets the threshold requirements. During the operation of the device, images of the wafer sample to be tested or wafers in the same batch are periodically collected and evaluated for consistency with other devices or historical images. The device hardware stability is monitored, and if the threshold is exceeded, the hardware parameters are readjusted.

[0065] The color image consistency evaluation in step one includes:

[0066] 1) For the acquired images, calculate the physical properties of the images (center of the circle). ,radius Roundness with deflection angle Image quality and sharpness Pixel difference ;

[0067] 1a) Methods for calculating the physical properties of images include:

[0068] Image preprocessing converts the acquired color image into a grayscale image, and extracts the wafer region using the Blob algorithm;

[0069] Create uniformly around the wafer circumference A measuring caliper is used to perform edge detection on the caliper area to obtain... The center of the circle is obtained by fitting a circle to each edge point using the least squares method and the RANSAC algorithm. ,radius ;

[0070] pass The wafer profile is obtained by fitting the edge points, and the roundness is calculated by the area integral between the wafer profile and the fitted circle. The integral of the product of the area between the wafer profile and the fitted circle and its corresponding angle is the deflection angle. ;

[0071] 1b) Methods for calculating image quality include:

[0072] Image preprocessing involves converting the acquired color image to a grayscale image, extracting the wafer region using the Blob algorithm, and calculating the width of the wafer region. With Gao ;

[0073] The Sobel operator is used to calculate the image gradient in the horizontal and vertical directions at each point in the wafer region. and Calculate the gradient of the composite image Calculate image sharpness ;

[0074] Calculate the signal-to-noise ratio of the entire image. ,in This is the proportionality coefficient. and These are the maximum and minimum values ​​of the local variance of all pixels in the image, respectively.

[0075] 1c) Methods for calculating pixel differences include:

[0076] Image preprocessing converts the acquired color image into a grayscale image, extracts the wafer region using the Blob algorithm, translates and aligns the centroid, and rotates and aligns the wafer Notch port.

[0077] The three channels in the wafer region are differentially processed and then combined: ,in and These are the pixel values ​​of a point in the two images, respectively. Represents the number of channels;

[0078] 2) First, calculate the image quality consistency score between the two images: Among them, weight , , , These are image quality sharpness, signal-to-noise ratio, and structural similarity, respectively.

[0079] Then calculate the physical property consistency score between the two images: Among them, weight .

[0080] Finally, the mean squared error method was used to calculate the difference between the two images. Set weights Calculate the final consistency score: ;

[0081] 3) If S q If the image quality exceeds a set threshold, the camera focal length f needs to be adjusted to improve image sharpness, and the camera level correction C needs to be adjusted to improve the signal-to-noise ratio. The camera focal length f and the camera level correction C are internal parameters of the camera.

[0082] If S p If the image size exceeds the set threshold, the imaging distance d and camera pitch angle A need to be adjusted to improve the image size and roundness.

[0083] If S q and S p If the conditions are met and the final S score is greater than the set threshold, the camera optical axis direction B needs to be adjusted to improve the consistency between units.

[0084] 4) After adjusting the above parameters, the illuminance parameter also needs to be adjusted:

[0085] Fix other hardware parameters of the machine, adjust the illumination, acquire color images of defect-free wafer samples, preprocess and divide into the smallest units, and fuse them to generate standard templates;

[0086] Fix other hardware parameters of the machine, adjust the illumination, acquire color images of the wafer to be tested, preprocess and segment into the smallest unit, and perform differential integration with the above template to obtain the defect area.

[0087] A color image of the wafer is acquired at a preset illumination level, converted into a grayscale image, and the number of pixels with a grayscale value greater than a threshold is counted. Less than the empirical threshold Increase illuminance greater than the empirical threshold Then reduce the illuminance ,in and This is an empirical coefficient;

[0088] After adjusting the illumination, acquire the wafer image again and repeat the above steps of counting grayscale pixels until... If the above standard is not reached before the maximum number of adjustments is given, the illuminance adjustment will fail and the machine will issue an alarm.

[0089] Among them, such as Figure 2 As shown, the defect detection in step two includes:

[0090] 1) Standard template creation process:

[0091] 1a) Image preprocessing process: First, the acquired color image is converted into a grayscale image, and the wafer region (ROI) is extracted from the grayscale image using the Blob algorithm.

[0092] 1b) Convert the grayscale image of the above region to the frequency domain, analyze the maximum frequency component, combine the wafer layout information (e.g., rectangular patterns of different sizes on the wafer are repeating units) to determine the size of the smallest repeating unit, use the sliding window method to identify the boundary of the repeating unit, and accurately segment the smallest unit of the wafer.

[0093] 1c) Select several repeating units from the color image segmentation of the defect-free wafer sample, and try to cover different radii and positions of the wafer;

[0094] Due to factors such as lens distortion, the size of the above units may vary and they may not be rectangular. The average width and height of the above units are taken as the size of the standard template.

[0095] 1d) By using bilinear interpolation and shape correction, the above units are unified to the size of the standard template to obtain several standard units of different sizes. The above standard units are then pixel-fused to obtain the standard template.

[0096] 2) Actual testing process:

[0097] 2a) Image preprocessing process: First, the acquired color image is converted into a grayscale image, and the wafer region (ROI) is extracted from the grayscale image using the Blob algorithm.

[0098] 2b) Convert the grayscale image of the above region to the frequency domain, analyze the maximum frequency component, combine the wafer layout information (e.g., rectangular patterns of different sizes on the wafer are repeating units) to determine the size of the smallest repeating unit, use the sliding window method to identify the boundary of the repeating unit, and accurately segment the smallest unit of the wafer.

[0099] 2c) Select several repeating units from the color image segmentation of the defect-free wafer sample, and try to cover different radii and positions of the wafer;

[0100] Due to factors such as lens distortion, the size of the above units may vary and they may not be rectangular. The average width and height of the above units are taken as the size of the standard template.

[0101] 2d) By using bilinear interpolation and shape correction, the smallest unit obtained from the segmentation of the color image of the wafer under test is unified to the size of the standard template, resulting in several standard units of different sizes;

[0102] 2e) The three channels of the standard cell are differentially compared with the standard template, and the regions with differences and areas greater than the threshold are marked. The coordinates of the above regions in the wafer image are calculated. The image and coordinates of the region are output as the defect detection result.

[0103] like Figure 3 As shown, (a) is a standard template wafer image, and (b) is the wafer image to be inspected, on which a 100-size template is printed. 100 200 200 300 300 400x400 500x500 650x650 800x800 1000x1000 1250x1250 1500x1500 and 2000x2000 The defects.

[0104] like Figure 4 As shown, (a) is a magnified view of a standard template wafer, and (b) is a magnified view of the local defect detection effect of the wafer to be inspected, according to an embodiment of the method of the present invention. As can be seen from the figures, defects greater than 200... 200 All defects were detected.

[0105] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should be considered within the scope of protection of the present invention.

Claims

1. A method for evaluating the consistency of color images of wafers and detecting defects, characterized in that, Includes the following steps: Step 1: Combining various operating conditions, use wafer color image consistency evaluation to measure and monitor the hardware of the monitoring equipment to obtain the hardware parameters related to equipment consistency that need to be adjusted. Step 2: For the batch of wafers to be tested, first create a standard template, then acquire a color image of the wafer to be tested, and perform differential integration with the standard template to obtain the defect area; Step one, color image consistency evaluation, includes: 1) Calculate the physical properties and image quality / clarity of the wafer image for both the image under test and the reference image. Pixel difference The physical properties include: center point. ,radius Roundness with deflection angle ; 2) First, calculate the image quality consistency score between the two images: Among them, weight , , , These are image quality sharpness, signal-to-noise ratio, and structural similarity, respectively. Then calculate the physical property consistency score between the two images: Among them, weight ; Finally, the mean squared error method was used to calculate the difference between the two images. Set weights Calculate the final consistency score: ; 3) If S q is greater than a set threshold, the camera focus needs to be adjusted to improve image sharpness and the camera flat-field correction needs to be improved to increase the signal-to-noise ratio. If S p If the image size exceeds the set threshold, the imaging distance and camera pitch angle need to be adjusted to improve the image size and roundness. If S q and S p If the conditions are met and the final S-score is greater than the set threshold, the camera optical axis direction needs to be adjusted to improve the consistency between units. The second defect detection step includes: 1) Fix the hardware parameters related to the consistency of the fixed machine, adjust the illumination, acquire color images of defect-free wafer samples, preprocess and divide them into the smallest units, and fuse them to generate a standard template; 2) Fix the hardware parameters related to the consistency of the fixed machine, adjust the illumination, acquire the color image of the wafer to be tested, preprocess and segment the smallest unit, and perform differential integration with the above template to obtain the defect area; The adjustment of illuminance includes: A color image of the wafer is acquired at a preset illumination level, converted into a grayscale image, and the number of pixels with a grayscale value greater than a threshold is counted. Less than the empirical threshold Increase illuminance greater than the empirical threshold Then reduce the illuminance ,in and This is an empirical coefficient; After adjusting the illumination, acquire the wafer image again, and repeat the above steps until... If the above standard is not reached before the maximum number of adjustments is given, the illuminance adjustment will fail and the machine will issue an alarm. The defect detection also includes: a) Image preprocessing process: First, the acquired color image is converted into a grayscale image. Then, the wafer region is extracted from the grayscale image as the ROI (Region of Interest) using the Blob algorithm. b) Convert the grayscale image of the above region to the frequency domain, analyze the maximum frequency component, determine the size of the smallest repeating unit by combining the wafer layout information, identify the boundary of the repeating unit by using the sliding window method, and accurately segment the smallest unit of the wafer. c) Select several repeating units from the segmented color image of the defect-free wafer sample, covering different radii and positions of the wafer; The average width and height of the above units are taken as the size of the standard template. d) By using bilinear interpolation and shape correction, the smallest unit obtained from the segmentation of the color image of the wafer under test is unified to the size of the standard template, resulting in several test units of different sizes; e) The three channels of the unit under test are differentially compared with the standard template, and the regions with differences and areas greater than the threshold are marked. The coordinates of the above regions in the wafer image are calculated. The image and coordinates of the region are output as the defect detection result.

2. The wafer color image consistency evaluation and defect detection method according to claim 1, characterized in that, Multiple operating conditions include: a) After the first device is installed and the hardware parameters are calibrated, a wafer sample image is acquired as a reference image. The newly installed machine initially acquires an image of the same wafer sample and evaluates its consistency with the reference image, thereby adjusting the hardware parameters until the score meets the threshold requirements; b) During the operation of the machine, images of the wafer sample or wafers under test acquired periodically are evaluated for consistency with other machines or historical images, thereby monitoring the machine's hardware parameters. If the threshold is exceeded, the hardware parameters are readjusted.

3. The wafer color image consistency evaluation and defect detection method according to claim 1, characterized in that, The method for calculating the physical properties of the image includes: Image preprocessing converts the acquired color image into a grayscale image, and extracts the wafer region using the Blob algorithm; Create uniformly around the wafer circumference A measuring caliper is used to perform edge detection on the caliper area to obtain... The center of the circle is obtained by fitting a circle to each edge point using the least squares method and the RANSAC algorithm. ,radius ; pass The wafer profile is obtained by fitting the edge points, and the roundness is calculated by the area integral between the wafer profile and the fitted circle. The integral of the product of the area between the wafer profile and the fitted circle and its corresponding angle is the deflection angle. .

4. The wafer color image consistency evaluation and defect detection method according to claim 1, characterized in that, The method for calculating image quality includes: Image preprocessing involves converting the acquired color image to a grayscale image, extracting the wafer region using the Blob algorithm, and calculating the width of the wafer region. With Gao ; The Sobel operator is used to calculate the image gradient in the horizontal and vertical directions at each point in the wafer region. and Calculate the gradient of the composite image Calculate image sharpness ; Calculate the signal-to-noise ratio of the entire image. ,in This is the proportionality coefficient. and These are the maximum and minimum values ​​of the local variance for all pixels in the image, respectively.

5. The wafer color image consistency evaluation and defect detection method according to claim 1, characterized in that, The method for calculating the pixel difference includes: Image preprocessing converts the acquired color image into a grayscale image, extracts the wafer region using the Blob algorithm, translates and aligns the centroid, and rotates and aligns the wafer Notch port. The three channels in the wafer region are differentially processed and then combined: ,in and These are the pixel values ​​of a point in the two images, respectively. Represents the number of channels.

6. The wafer color image consistency evaluation and defect detection method according to claim 1, characterized in that, The standard template is a series of standard cells of different sizes obtained from a pre-collected defect-free wafer through steps a)-d).

Citation Information

Patent Citations

  • Chip detection method

    CN101090083A

  • Method for detecting low-texture defects of wafer

    CN109978839A

  • Method and device for adjusting consistency of multi-camera system

    CN116668831A