Wafer defect detection method, system and storage medium

Abstract

This invention discloses a wafer defect detection method, system, and storage medium. The method includes: acquiring an image of a sawtooth region of a target to be detected in a wafer; extracting corner points from the image and calculating the angle value of each corner point; using corner points with angle values ​​within a preset range as first candidate contour corner points; obtaining the coordinate values ​​of all first candidate contour corner points; determining the tooth peak region and tooth valley region of the sawtooth region of the target to be detected based on the distribution of all first candidate contour corner points in the coordinate system; using the first candidate contour corner points located in the tooth peak region and tooth valley region as detection contour points of the sawtooth region of the target to be detected; matching the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template; and determining the sawtooth defect type based on the matching result. This invention is applicable to the detection of various sawtooth targets on wafers, especially to the detection of targets in sparse periodic roughening processes, such as wafer substrate colloids or grains.

Description

Technical Field

[0001] This application relates to the field of wafer inspection technology, and more specifically, to a wafer defect detection method, system, and storage medium. Background Technology

[0002] In the field of semiconductor packaging testing process inspection, the inspection of sawtooth marks on targets in various sparse, periodic, and rough processes is frequently involved. Due to the instability of the process of the target to be inspected, the strong randomness of its specifications and dimensions, and the inability to use a specific template for stable positioning, and the large fluctuations and randomness of various defects without a periodic distribution pattern, the available image information is limited and cannot provide sufficient effective information for target positioning. Therefore, conventional defect detection methods cannot achieve good detection results.

[0003] For example, in the wafer manufacturing process of CIS (CMOS imaging sensor) chips, the encapsulation process presents significant yield challenges because its quality directly affects subsequent bonding processes and the quality of the finished product. Therefore, inspecting the encapsulation on the glass substrate in the front-end process is crucial. However, because the encapsulation diffuses freely, the actual production results deviate significantly from the theoretical process standards, exhibiting considerable fluctuations and randomness. For defect detection in this process, traditional grayscale comparison-based die-to-goldendie and die-to-die algorithms are no longer applicable. Summary of the Invention

[0004] To address at least one defect or improvement requirement in the prior art, the present invention provides a wafer defect detection method, system, and storage medium, applicable to the detection of various sawtooth targets on wafers, particularly suitable for the detection of targets in sparse periodic rough processes, such as wafer substrate colloids or grains.

[0005] To achieve the above objectives, according to a first aspect of the present invention, a wafer defect detection method is provided, comprising:

[0006] An image of the sawtooth region of the target to be detected in the wafer is acquired, corner points are extracted from the image and the angle value of each corner point is calculated;

[0007] Corner points with angle values ​​within a preset range are selected as first candidate contour corner points. The coordinate values ​​of all first candidate contour corner points are obtained. Based on the distribution of all first candidate contour corner points in the coordinate system, the tooth peak region and tooth valley region of the serrated area of ​​the target to be detected are determined. The first candidate contour corner points located in the tooth peak region and tooth valley region are selected as the detection contour points of the serrated area of ​​the target to be detected.

[0008] Match the detection contour points of the target sawtooth area to be detected with the contour points of the template sawtooth area, and determine the sawtooth defect type of the target to be detected according to the matching result.

[0009] Further, the target to be detected is a wafer substrate colloid or a die.

[0010] Further, extracting corner points from the image and calculating the angle value of each corner point includes:

[0011] For any pixel p in the image, obtain the gray values of 16 pixels on the circumference with pixel p as the center and a radius of 3 pixels. Number these 16 pixels from 1 to 16 in clockwise or counterclockwise order. If among these 16 pixels, there are continuously n pixels whose gray values are all greater than the gray value Ip of pixel p by a preset threshold t, or continuously n pixels whose gray values are all less than the gray value Ip of pixel p by a preset threshold t, and n is greater than a preset threshold M, 1 < M < 16, then extract pixel p as a corner point;

[0012] Pre-determine the mapping relationship between the n value and the angle value, and obtain the angle value of each corner point according to the n value in the process of extracting corner points.

[0013] Further, set the preset interval according to the shape of the target to be detected. If the standard sawtooth of the target to be detected is an equilateral triangle, then set the preset interval as [60° - α, 60° + α] ∪ [300° - α, 300° + α], where α is the angle error threshold.

[0014] Further, the step of taking the first alternative contour corner points located in the tooth peak area and the tooth valley area as the contour points of the target sawtooth area to be detected includes:

[0015] Use the non-maximum suppression algorithm to process the first alternative contour corner points located in the tooth peak area and the tooth valley area, remove the non-maximum values, obtain the second alternative contour corner points, and take the second alternative contour corner points as the detection contour points of the target sawtooth area to be detected.

[0016] Further, the step of determining the tooth peak area and the tooth valley area of the target sawtooth area to be detected according to the distribution of all the first alternative contour corner points in the coordinate system includes:

[0017] Determine whether the sawtooth arrangement direction of the target to be detected is parallel to the x-axis of the coordinate system according to the distribution of all the first alternative contour corner points in the coordinate system;

[0018] If the sawtooth arrangement direction of the target to be detected is parallel to the x-axis, then determine the tooth peak area and the tooth valley area according to the y-axis coordinate values of the first alternative contour corner points;

[0019] If the sawtooth arrangement direction of the target to be detected is not parallel to the x-axis, then the angle between the sawtooth arrangement direction of the target to be detected and the x-axis is determined. The coordinate values ​​of all first candidate contour corner points are corrected according to the angle so that the corrected sawtooth arrangement direction is parallel to the x-axis. Then, the tooth peak region and tooth valley region are determined according to the corrected y-axis coordinate values ​​of the first candidate contour corner points.

[0020] Furthermore, the template jagged region image is input into the trained neural network model to generate template jagged region contour points.

[0021] Further, matching the detection contour points of the serrated region of the target to be detected with the contour points of the serrated region of the template includes:

[0022] The angle values ​​of the detection contour points in the sawtooth region of the target to be detected and the spacing between the detection contour points are statistically analyzed. The angle values ​​of the contour points in the sawtooth region of the template and the spacing between the contour points are also statistically analyzed.

[0023] The statistical information of the detection contour points of the serrated region of the target to be detected is matched with the statistical information of the contour points of the serrated region of the template.

[0024] According to a second aspect of the present invention, a wafer defect detection system is also provided, comprising:

[0025] The preprocessing module is used to acquire images of the sawtooth region of the target to be detected in the wafer, extract corner points from the images, and calculate the angle value of each corner point;

[0026] The contour point detection module is used to take corner points with angle values ​​in a preset range as first candidate contour corner points, obtain the coordinate values ​​of all first candidate contour corner points, determine the tooth peak region and tooth valley region of the serrated region of the target to be detected according to the distribution of all first candidate contour corner points in the coordinate system, and take the first candidate contour corner points located in the tooth peak region and tooth valley region as the detection contour points of the serrated region of the target to be detected.

[0027] The matching module is used to match the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template, and determine the sawtooth defect type of the target to be detected based on the matching result.

[0028] According to a third aspect of the invention, a storage medium is also provided that stores a computer program executable by a processor, which, when run on the processor, causes the processor to perform the steps of any of the methods described above.

[0029] In summary, compared with existing technologies, the technical solutions conceived in this invention achieve the following beneficial effects: Before template matching, this invention first extracts corner points and calculates the angle values ​​of each corner point, then filters corner points whose angle values ​​fall within a preset range, and then determines the peak and valley regions of the sawtooth region of the target to be detected. The detection contour points of the sawtooth region of the target to be detected are obtained from the peak and valley regions. This method is applicable to the detection of various sawtooth targets on wafers, especially for target detection in sparse, periodically rough processes, such as wafer substrate colloids or grains. Furthermore, the solution is simple and easy to operate, with high defect detection accuracy, making it the only mass-production solution for this process in the industry. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 A schematic flowchart of a wafer defect detection method provided in an embodiment of this application;

[0032] Figure 2 This is a schematic diagram illustrating the corner extraction principle provided in an embodiment of this application.

[0033] Figure 3 A schematic diagram of the detection contour points of the sawtooth region of the target to be detected provided in an embodiment of this application;

[0034] Figure 4 A schematic diagram of matching statistics provided for embodiments of this application;

[0035] Figure 5 This is a schematic diagram of the matching results provided in an embodiment of this application. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0037] In the description, claims and the above-mentioned drawings of this application, the terms "first", "second", "third", etc. are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally further includes steps or modules not listed, or optionally further includes other steps or modules inherent to these processes, methods, products or devices.

[0038] As Figure 1 shown, a wafer defect detection method according to an embodiment of the present invention includes the steps of:

[0039] S101, acquiring an image of a target serrated area to be detected in a wafer, extracting corner points from the image and calculating the angular values of each corner point.

[0040] The target to be detected in the embodiment of the present invention may be various serrated targets of a wafer, and is particularly suitable for the target detection of a sparse periodic rough process, such as wafer substrate colloid or crystal grains.

[0041] Further, for the target to be detected in a sparse periodic rough process, its image is a rough texture. In order to avoid generating too many noise feature points, only the main contour feature points are retained, and a smoothing filtering process is performed before extracting the corner points.

[0042] Since the gray-scale features of the target to be detected may be unstable and there are many interference factors, corner points are selected as feature points in the embodiment of the present invention. Corner points are the extreme points of an image, the end points of a line segment, the points with the maximum curvature of a curve, or the points with the maximum attributes in the horizontal or vertical directions, etc. These feature points are very important features of an image and play a very important role in the understanding and analysis of the image graphics. While retaining the important features of the image graphics, feature points can replace the processing of the entire image, effectively reducing the data volume of information, making its information content very high, effectively improving the calculation speed, facilitating reliable matching of the image, and making real-time processing possible.

[0043] Further, the step of extracting corner points and calculating the angular values of each corner point includes:

[0044] For any pixel p in the image, obtain the gray-scale values of r pixels on the circumference with pixel p as the center and a radius of r pixels. Number these 16 pixels in sequence from 1 to 16 in a clockwise or counterclockwise direction. If among these 16 pixels, if there are continuously n pixels whose gray-scale values are all greater than the gray-scale value Ip of pixel p by a preset threshold t, or continuously n pixels whose gray-scale values are all less than the gray-scale value Ip of pixel p by a preset threshold t, and n is greater than a preset threshold M, 1 < M < 16, then extract pixel p as a corner point; 2 个像素のグレースケール値を取得し、これら16個の画素を時計回りまたは反時計回りに1から16まで順番に番号付けし、これら16個の画素の中で、n個の画素のグレースケール値が、画素pのグレースケール値Ipよりも予め設定された閾値tだけ大きい場合、またはn個の画素のグレースケール値が、画素pのグレースケール値Ipよりも予め設定された閾値tだけ小さい場合、かつnが予め設定された閾値Mより大きく、1<M<16のとき、画素pをコーナーポイントとして抽出する。 It should be noted that the content in Japanese in the original text seems to be incorrect or an incomplete translation. I have translated it as accurately as possible based on the overall context. If there are any specific requirements or corrections regarding this part, please let me know.

[0045] Pre-determine the mapping relationship between the n value and the angle value, and obtain the angle value of each corner point according to the n value in the process of extracting the corner points.

[0046] Take r = 3 as an example for illustration.

[0047] As Figure 2 shown, for any pixel p in the image, use the 16 pixels on the circumference with a radius of 3 pixels centered on pixel p to determine whether point p is the corner point to be extracted. These 16 pixels are numbered from 1 to 16 in clockwise order. If there are continuously n pixels among these 16 pixels whose gray values are all greater than the gray value Ip of pixel p by a preset threshold t, or continuously n pixels whose gray values are all less than the gray value Ip of pixel p by a preset threshold t, and n is greater than the preset threshold M, 1 < M < 16, then extract pixel p as the corner point. Pre-determine the mapping relationship between the n value and the angle value, and obtain the angle value of each corner point according to the n value in the process of extracting the corner points. The relationship between the angle value and the n value is: angle value = 360°×(n - 1)÷16 ± β, where β is the error value. If β = 1l.25°, then n = 5, and the corresponding angle value is 78.75 - 101.25 degrees; n = 9, and the corresponding angle value is 168.75 - 191.25 degrees.

[0048] S102, take the corner points with angle values within the preset interval as the first alternative contour corner points, obtain the coordinate values of all the first alternative contour corner points, determine the tooth peak area and the tooth valley area of the sawtooth area of the target to be detected according to the distribution of all the first alternative contour corner points in the coordinate system, and take the first alternative contour corner points located in the tooth peak area and the tooth valley area as the detection contour points of the sawtooth area of the target to be detected.

[0049] Because using all the corner points to describe the contour of the target to be detected will have many redundant points, which is neither conducive to extracting features nor to reducing the processing time, so it is necessary to extract the skeleton contour points of the sawtooth area of the target to be detected.

[0050] Further, set the preset interval for screening the first alternative contour corner points according to the shape of the target to be detected. If the standard sawtooth of the target to be detected is an equilateral triangle, then set the preset interval as [60° - α, 60° + α] ∪ [300° - α, 300° + α], where α is the angle error threshold.

[0051] Taking a wafer substrate colloid as an example, the process is as follows: First, based on the colloid's manufacturing process shape, its standard serrations are abstracted as equilateral triangles, with corresponding contour point angles of 60° and 300°. Considering process instability, the angle fluctuation range is set to + / - 11.25 degrees. Second, the previously calculated corner points are filtered using the range [60°-11.25 degrees, 60°+11.25 degrees]∪[60°-11.25 degrees, 60°+11.25 degrees]. The corner points filtered within [60°-11.25 degrees, 60°+11.25 degrees] or [60°-11.25 degrees, 60°+11.25 degrees] are selected as the first candidate contour corner points, denoted as set M1. Then, obtain the coordinate values ​​of all first candidate contour corner points in set M1, determine the tooth peak region and tooth valley region of the serrated area of ​​the target to be detected based on the distribution of all first candidate contour corner points in the coordinate system, and take the first candidate contour corner points located in the tooth peak region and tooth valley region as the detection contour points of the serrated area of ​​the target to be detected.

[0052] Further, determining the peak and valley regions of the sawtooth region of the target to be detected based on the distribution of all first candidate contour corner points in the coordinate system includes: determining whether the sawtooth arrangement direction of the target to be detected is parallel to the x-axis of the coordinate system based on the distribution of all first candidate contour corner points in the coordinate system; if the sawtooth arrangement direction of the target to be detected is parallel to the x-axis, then the peak and valley regions are determined based on the y-axis coordinate values ​​of the first candidate contour corner points; if the sawtooth arrangement direction of the target to be detected is not parallel to the x-axis, then the angle between the sawtooth arrangement direction of the target to be detected and the x-axis is determined, and the coordinate values ​​of all first candidate contour corner points are corrected based on the angle so that the corrected sawtooth arrangement direction is parallel to the x-axis, and then the peak and valley regions are determined based on the corrected y-axis coordinate values ​​of the first candidate contour corner points.

[0053] Furthermore, if the direction of the sawtooth arrangement of the target to be detected is parallel to the x-axis, the tooth peak region and tooth valley region can be determined based on the y-axis coordinate value of the first candidate contour corner point in the following manner.

[0054] Method 1: Since the sawtooth arrangement of the target object is parallel to the x-axis, for an ideal standard sawtooth, the y-axis coordinates of all first candidate contour corner points can only have two values: one corresponding to the tooth peak and one corresponding to the tooth valley. However, considering error, the y-axis coordinates of all first candidate contour corner points of a standard sawtooth should be concentrated around two values: one near the y-axis coordinate corresponding to the tooth peak and one near the y-axis coordinate corresponding to the tooth valley. These two areas where the y-axis coordinates are concentrated are respectively designated as the tooth peak region and the tooth valley region. If the sawtooth of the target object has defects, the y-axis coordinates of some first candidate contour corner points may be far from these two values, not in the tooth peak region or tooth valley region. This means that the corresponding first candidate contour corner point is not a tooth peak or tooth valley and can be removed.

[0055] Method 2: Reorder all first candidate contour corner points according to the x-value of the coordinate system, and retain the elements whose positive and negative values ​​are consistent in the two comparisons by comparing the angle values ​​of each element of the sequence with the angle values ​​of the adjacent left and right elements.

[0056] Of course, other methods can also be used to determine the peak and valley regions of the sawtooth region of the target to be detected, which will not be described in detail here.

[0057] Furthermore, selecting the first candidate contour corner points located in the tooth peak region and tooth valley region as the target serrated region contour points for detection includes: processing the first candidate contour corner points located in the tooth peak region and tooth valley region using a non-maximum suppression algorithm to remove non-maximum values, obtaining a second candidate contour corner point, and using the second candidate contour corner point as the detection contour point of the target serrated region to be detected. That is, the non-maximum suppression algorithm further performs screening.

[0058] The detection contour points of the sawtooth region of the target to be detected are as follows: Figure 3 As shown, the black dots are the detection contour points of the jagged area of ​​the target to be detected.

[0059] S103, Match the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template, and determine the sawtooth defect type of the target to be detected based on the matching result.

[0060] Furthermore, the matching process includes the following steps:

[0061] (1) Calculate the angle values ​​of the detection contour points in the sawtooth region of the target to be detected and the spacing between the detection contour points. Calculate the angle values ​​of the contour points in the sawtooth region of the template and the spacing between the contour points in the sawtooth region of the template.

[0062] Since the spacing and angle of the contour points of the standard sawtooth are known, the detection contour points of the sawtooth region of the target to be detected are statistically analyzed. The angle values ​​of the detection contour points and the spacing between the detection contour points are statistically analyzed and used as features of the overall contour for the next step of analysis.

[0063] (2) Then match the statistical information of the detection contour points of the serrated area of ​​the target to be detected with the statistical information of the contour points of the serrated area of ​​the template.

[0064] Matching based on statistics from both involves comparing the statistically derived overall contour features of the target object with the known overall contour features of the template. If a tooth is missing or extra, its location can be easily determined from the positional relationships between points. For example, if a gear has an extra or missing tooth, the spacing and direction between the contour points will distinguish it from other normal contour points. Furthermore, because the calculation is based on the relative positional relationships between points, it is also resistant to tilting and noise interference.

[0065] In one embodiment, the statistical information of the detection contour points of the target sawtooth region and the statistical information of the template sawtooth region contour points are as follows: Figure 4 As shown, the matching results are as follows: Figure 5 As shown.

[0066] Furthermore, a wafer defect detection system according to an embodiment of the present invention includes:

[0067] The preprocessing module is used to acquire images of the sawtooth region of the target to be detected in the wafer, extract corner points from the images, and calculate the angle value of each corner point;

[0068] The contour point detection module is used to select corner points with angle values ​​in a preset range as first candidate contour corner points, obtain the coordinate values ​​of all first candidate contour corner points, determine the tooth peak area and tooth valley area of ​​the sawtooth region of the target to be detected based on the distribution of all first candidate contour corner points in the coordinate system, and select the first candidate contour corner points located in the tooth peak area and tooth valley area as the detection contour points of the sawtooth region of the target to be detected.

[0069] The matching module is used to match the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template, and determine the sawtooth defect type of the target to be detected based on the matching result.

[0070] This application also provides a storage medium storing a computer program executable by a processor. When the computer program runs on the processor, it causes the processor to perform the steps of any of the above-described wafer defect detection methods. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0071] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0072] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0073] In the embodiments provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual coupling or direct coupling or communication connection may be through some service interfaces; the indirect coupling or communication connection of the system or modules may be electrical or other forms.

[0074] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0075] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0076] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0077] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0078] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of embodiments of this disclosure upon considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described herein. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

[0079] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0080] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting wafer defects, characterized in that, Including: Collect an image of the target serrated area to be detected in the wafer, extract corner points from the image, and calculate the angle values of each corner point; Take the corner points with angle values within a preset interval as the first alternative contour corner points, obtain the coordinate values of all the first alternative contour corner points, determine the tooth peak area and the tooth valley area of the target serrated area to be detected according to the distribution of all the first alternative contour corner points in the coordinate system, and take the first alternative contour corner points located in the tooth peak area and the tooth valley area as the detection contour points of the target serrated area to be detected; Match the detection contour points of the target serrated area to be detected with the contour points of the template serrated area, and determine the serration defect type of the target to be detected according to the matching result; Extracting corner points from the image and calculating the angle value of each corner point includes: For any pixel p in the image, obtain the gray values of 16 pixels on the circumference with pixel p as the center and a radius of 3 pixels. Number these 16 pixels from 1 to 16 in a clockwise or counterclockwise order. If among these 16 pixels, if there are continuously n pixels whose gray values are all greater than the gray value Ip of pixel p by a preset threshold t, or continuously n pixels whose gray values are all less than the gray value Ip of pixel p by a preset threshold t, and n is greater than the preset threshold M, 1 < M < 16, then extract pixel p as a corner point; Pre-determine the mapping relationship between the n value and the angle value, and obtain the angle value of each corner point according to the n value in the process of extracting corner points.

2. The wafer defect detection method as described in claim 1, characterized in that, The target to be detected is the wafer substrate colloid or the die.

3. The wafer defect detection method as described in claim 1, characterized in that, Set the preset interval according to the shape of the target to be detected. If the standard serration of the target to be detected is an equilateral triangle, then set the preset interval as [60° - α, 60° + α] ∪ [300° - α, 300° + α], where α is the angle error threshold.

4. The wafer defect detection method as described in claim 1, characterized in that, The step of taking the first alternative contour corner points located in the tooth peak area and the tooth valley area as the detection contour points of the target serrated area includes: Use the non-maximum suppression algorithm to process the first alternative contour corner points located in the tooth peak area and the tooth valley area, remove the non-maximum values, obtain the second alternative contour corner points, and take the second alternative contour corner points as the detection contour points of the target serrated area to be detected.

5. The wafer defect detection method as described in claim 1, characterized in that, The step of determining the tooth peak area and the tooth valley area of the target serrated area to be detected according to the distribution of all the first alternative contour corner points in the coordinate system includes: Determine whether the serration arrangement direction of the target to be detected is parallel to the x-axis of the coordinate system according to the distribution of all the first alternative contour corner points in the coordinate system; If the serration arrangement direction of the target to be detected is parallel to the x-axis, then determine the tooth peak area and the tooth valley area according to the y-axis coordinate values of the first alternative contour corner points; If the serration arrangement direction of the target to be detected is not parallel to the x-axis, then determine the included angle between the serration arrangement direction of the target to be detected and the x-axis, correct the coordinate values of all the first alternative contour corner points according to the included angle so that the corrected serration arrangement direction is parallel to the x-axis, and then determine the tooth peak area and the tooth valley area according to the corrected y-axis coordinate values of the first alternative contour corner points.

6. The wafer defect detection method as described in claim 1, characterized in that, Input the template serrated area image into the trained neural network model to generate the template serrated area contour points.

7. The wafer defect detection method as described in claim 1, characterized in that, The step of matching the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template includes: The angle values ​​of the detection contour points in the sawtooth region of the target to be detected and the spacing between the detection contour points are statistically analyzed. The angle values ​​of the contour points in the sawtooth region of the template and the spacing between the contour points are also statistically analyzed. The statistical information of the detection contour points of the serrated region of the target to be detected is matched with the statistical information of the contour points of the serrated region of the template.

8. A wafer defect detection system for implementing the wafer defect detection method according to any one of claims 1 to 7, characterized in that, include: The preprocessing module is used to acquire images of the sawtooth region of the target to be detected in the wafer, extract corner points from the images, and calculate the angle value of each corner point; The contour point detection module is used to take corner points with angle values ​​in a preset range as first candidate contour corner points, obtain the coordinate values ​​of all first candidate contour corner points, determine the tooth peak region and tooth valley region of the serrated region of the target to be detected according to the distribution of all first candidate contour corner points in the coordinate system, and take the first candidate contour corner points located in the tooth peak region and tooth valley region as the detection contour points of the serrated region of the target to be detected. The matching module is used to match the detection contour points of the sawtooth region of the target to be detected with the contour points of the sawtooth region of the template, and determine the sawtooth defect type of the target to be detected based on the matching result.

9. A storage medium, characterized in that, It stores a computer program that, when run on a processor, causes the processor to perform the steps of the method according to any one of claims 1 to 7.

Images