Methods, apparatus, equipment and storage media for detecting defects in cutting surfaces

Abstract

This specification provides a method, apparatus, device, and storage medium for detecting dicing defects. The method includes: acquiring an image to be tested; the image to be tested includes a local wafer region in a wafer to be tested, the local wafer region comprising multiple grain regions and dicing regions located between adjacent grains. Identifying the dicing regions from the image to be tested, and determining the grain regions corresponding to each of the multiple grains. For any grain, determining the actual boundary and ideal boundary of the grain region corresponding to the grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in a non-destructive state. Based on the difference between the actual boundary and the ideal boundary, determining the corresponding difference region, and using the difference region as the dicing defect region corresponding to the grain.

Description

Technical Field

[0001] This specification relates to the field of wafer dicing technology, and in particular to methods, apparatus, equipment and storage media for detecting defects in dicing paths. Background Technology

[0002] During the wafer dicing process, many types of defects are generated, such as chipping at the dicing edge, particles, undicated wafers, and hidden dicing defects. These dicing defects have a significant impact on the performance and function of the die, affecting its reliability and lifespan, and also affecting subsequent processes and packaging.

[0003] Therefore, establishing a systematic cutting defect detection process can not only improve the yield and quality of the final product and ensure the long-term reliability of the product, but also provide real-time process feedback and optimization basis to pinpoint problems in the cutting process. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this specification provides methods, apparatus, equipment and storage media for detecting defects in cutting tracks.

[0005] According to a first aspect of the embodiments of this specification, a method for detecting cutting track defects is provided, the method comprising: Acquire the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, the local wafer region includes multiple grain regions and dicing regions located between adjacent grains.

[0006] The cutting kerf region is identified from the image to be tested, and the grain region corresponding to each of the multiple grains is determined.

[0007] For any grain, determine the actual boundary and ideal boundary of the grain region corresponding to the grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state.

[0008] Based on the difference between the actual boundary and the ideal boundary, the corresponding difference region is determined, and the difference region is used as the cutting defect region corresponding to the grain.

[0009] According to a second aspect of the embodiments of this specification, a cutting track defect detection device is provided, comprising: The image acquisition module is used to acquire the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, and the local wafer region includes multiple grain regions and dicing regions located between adjacent grains.

[0010] The grain region identification module is used to identify the cutting channel region from the image to be tested and determine the grain region corresponding to each of the multiple grains.

[0011] The boundary recognition module is used to determine the actual boundary and ideal boundary of the grain region corresponding to any grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state.

[0012] The defect region identification module is used to determine the corresponding difference region based on the difference between the actual boundary and the ideal boundary, and to regard the difference region as the cutting channel defect region corresponding to the grain.

[0013] According to a third aspect of the embodiments of this specification, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method as described in the first aspect.

[0014] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the method as described in the first aspect.

[0015] The technical solutions provided in the embodiments of this specification may include the following beneficial effects: In this embodiment, firstly, by acquiring images of a local wafer area, focusing on the dicing region to be detected, and without needing to capture the entire die, scanning the entire wafer is avoided, resulting in shorter processing time. Secondly, by identifying the more prominent dicing region, the location of the die region is deduced based on the spatial distribution relationship between the dicing region and the die region. Compared to directly identifying the less prominent die region, this significantly improves the accuracy and reliability of die region location. Finally, this solution can identify dicing defect regions without relying on die patterns, making it applicable to various wafer types.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification.

[0018] Figure 1 This is a flowchart illustrating a method for detecting defects in a cutting path according to an exemplary embodiment.

[0019] Figure 2This is a schematic diagram of a partial wafer region illustrated in this specification according to an exemplary embodiment.

[0020] Figure 3 This is a schematic diagram illustrating, according to an exemplary embodiment, the segmentation of an image under test into multiple connected components.

[0021] Figure 4 This is a schematic diagram of the connected region corresponding to the cutting defect area shown in this specification according to an exemplary embodiment.

[0022] Figure 5 This is a schematic diagram of a binary image corresponding to a grain region shown in this specification according to an exemplary embodiment.

[0023] Figure 6 This is a schematic diagram of a statistical curve in the X direction illustrated in this specification according to an exemplary embodiment.

[0024] Figure 7 This is a schematic diagram illustrating the actual boundary of a grain region according to an exemplary embodiment of this specification.

[0025] Figure 8 This is a schematic diagram illustrating the ideal boundary of a grain region according to an exemplary embodiment of this specification.

[0026] Figure 9 This is a schematic diagram illustrating the difference region according to an exemplary embodiment of this specification.

[0027] Figure 10 This is a schematic diagram of the minimum bounding rectangle of a grain region illustrated in this specification according to an exemplary embodiment.

[0028] Figure 11 This is a schematic diagram illustrating the distance between the cut defect region located at the corner of the grain region and the sealing line, according to an exemplary embodiment of this specification.

[0029] Figure 12 This is a schematic diagram illustrating the distance between a cut-through defect region located on the edge of a grain region and a sealing line, according to an exemplary embodiment of this specification.

[0030] Figure 13 This is a schematic diagram of the geometric region surrounding a coordinate point, illustrated in this specification according to an exemplary embodiment.

[0031] Figure 14 This is a schematic diagram of the structure of an electronic device according to an exemplary embodiment of this specification.

[0032] Figure 15 This is a block diagram illustrating a cutting track defect detection device according to an exemplary embodiment of this specification. Detailed Implementation

[0033] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this specification as detailed in the appended claims.

[0034] The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of this specification. The singular forms “a,” “the,” and “the” as used in this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0035] It should be understood that although the terms first, second, third, etc., may be used in this specification to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this specification, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0036] A wafer is the basic substrate material for semiconductor manufacturing, on which a large number of repeating integrated circuit structural units are formed. These integrated circuit structural units, which have not been physically separated after all the processes are completed, are collectively called dies. Each die is a fully functional integrated circuit die, possessing specific electrical properties and logic functions.

[0037] Dies are arranged in a regular array on the wafer, with a pre-defined non-functional area between adjacent dies, called a scribe line. The scribe line does not contain any functional circuitry; its main function is to provide operating space for mechanical cutting or laser scribing before subsequent packaging, so as to avoid damaging the functional areas of adjacent dies.

[0038] Wafer dicing is the process of dividing a whole wafer into individual dies along a dicing line. The direct result of dicing is to obtain several discrete, individually packageable bare dies. These bare dies then enter the packaging process to ultimately form a finished product that can be used in electronic systems.

[0039] During the wafer dicing process, a series of problems are often encountered, such as edge chipping, which usually occurs during the dicing process and makes the edge of the wafer irregular; particle contamination, which may be caused by tiny particles in the dicing tool or environment, resulting in an unclean wafer surface; undicated areas, which refer to the parts that should have been diced but were not fully processed, which may affect subsequent functions; and hidden dicing defects, which are usually not easy to detect, but can also affect the reliability of the die.

[0040] Therefore, it is particularly important to establish an effective mechanism for detecting cutting defects in order to detect and address these problems in a timely manner.

[0041] This disclosure provides a method for detecting defects in cutting surfaces, to complete the process of detecting defects in cutting surfaces, as detailed in the following embodiments: like Figure 1 As shown, Figure 1 This is a flowchart illustrating a method for detecting cut surface defects according to an exemplary embodiment, including steps 101-104: Step 101: Obtain the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, the local wafer region includes multiple grain regions and dicing regions located between adjacent grains.

[0042] A grain map can be generated based on the grain positioning algorithm, that is, the coordinates of all grains in the wafer under test can be obtained.

[0043] The purpose of this method is to detect dicing defects. To ensure that the complete dicing area can be captured in each image under test, the distribution of the grain and dicing area on the wafer can be determined by combining the coordinates and size of the grain, thereby planning the scanning path to capture the complete dicing area in each image under test. For example, when planning the scanning path, for each capture, the corner point of the grain can be located and positioned at the center of the field of view of the image under test. This ensures that the dicing area near the grain corner point is complete.

[0044] Local wafer regions captured in different test images can overlap to avoid missing any images. For example, for the kerf at the edge of the current test image, to avoid incomplete capture of the kerf, the kerf at the edge can be positioned at the center of the field of view of the next test image before being captured again.

[0045] Since the image to be tested is a partial wafer area captured on the wafer under test, multiple images may be captured for the wafer under test, with each image corresponding to a partial wafer area. Figure 1The method involves detecting dicing defects in a local wafer region captured by any given image, with the detection process being identical for each image. By combining the detection results from all images, a complete dicing defect detection result for the wafer under test can be obtained.

[0046] In the image under test, it's possible that the entire area of ​​a particular grain may not be captured. For example, a grain region located at the edge of the image might be only a portion of the corresponding grain. Different grain regions of the same grain can be captured in multiple images under test. Figure 2 In the image shown, the area enclosed by the red solid line is the grain region 21; the blue solid line represents the edge line 22 of the grain. Different grain regions are separated by the kerf region 23 (the area between the red solid lines in two adjacent grain regions).

[0047] exist Figure 2 In the image to be tested, there is no need to capture the entire grain or the entire wafer. Only the cut area to be detected needs to be focused, which reduces the amount of image data and makes the processing time for cut detection based on the image to be tested shorter.

[0048] Step 102: Identify the cutting kerf region from the image to be tested, and determine the grain region corresponding to each of the multiple grains.

[0049] There are several ways to identify the cutting area from the image under test: In one embodiment, the image to be tested is first preprocessed, including grayscale conversion and noise reduction, to improve the accuracy of subsequent analysis. Then, a Fast Fourier Transform (FFT) is applied to transform the image to the frequency domain, extracting its spectral features and identifying the main frequencies and directions associated with the cutting path. Next, significant amplitude peaks in the spectrum are analyzed to determine feature information such as the main period and spacing of the cutting path, constructing an ideal cutting path model. Subsequently, the frequency domain information associated with the cutting path can be transformed back to the time domain using an inverse FFT. In the time domain, the accuracy of feature extraction is verified by comparing the consistency of the cutting path model with the cutting path in the image to be tested in terms of position, angle, and geometric parameters. Finally, thresholding and edge detection are performed using the verified feature information to extract the cutting path region in the image to be tested.

[0050] In one embodiment, a brightness threshold can be determined based on the grayscale distribution of the image to be tested. Based on the brightness threshold, the image to be tested is segmented to obtain multiple connected components. From the multiple connected components, the connected components that satisfy the geometric feature conditions are selected as the cutting channel region.

[0051] Cutting channels typically exhibit relatively low grayscale values ​​in images, while grains typically exhibit relatively high grayscale values. Based on this grayscale distribution characteristic, a preliminary division of the image under test can be achieved by setting a brightness threshold, classifying the pixels in the image under test into low-grayscale pixels and high-grayscale pixels. Low-grayscale pixels can be used as candidate pixels for the cutting channel region, and high-grayscale pixels can be used as candidate pixels for the grain region.

[0052] As for how to determine the brightness threshold, the gray-level histogram distribution of the image under test can be calculated, and then the brightness threshold can be determined using Otsu's method (maximum inter-class variance method). Based on this brightness threshold, the grain region and the cut channel region can be distinguished. Of course, in addition to Otsu's method, there are other methods, such as adaptive thresholding methods and histogram valley detection methods.

[0053] Thus, by segmenting the image under test according to this brightness threshold, multiple connected components can be obtained. A connected component can be a set of pixels with the same pixel value and spatially adjacent. For example, the connected component corresponding to the cut channel region can be a connected component composed of pixels with brightness less than the brightness threshold and spatially adjacent. Conversely, the connected component corresponding to the grain region can be a connected component composed of pixels with brightness not less than the brightness threshold and spatially adjacent.

[0054] Among the multiple connected regions, a connected region composed of low grayscale pixels may be a connected region corresponding to the cutting channel region, or it may be a connected region formed by missegmentation due to factors such as local darkening of the grain.

[0055] Therefore, this scheme does not directly use the connected domains composed of pixels with brightness less than the brightness threshold as the cutting channel region. Instead, it selects the connected domains that meet the geometric feature conditions from multiple connected domains as the cutting channel region to avoid identifying the mis-segmented region as the cutting channel region, thereby improving the accuracy of determining the cutting channel region.

[0056] For example, such as Figure 3 As shown, this represents the multiple connected components obtained by segmenting the image under test. The connected components marked in gray correspond to grain regions, while those marked in red correspond to cut-line regions. However, in... Figure 3 In the segmentation results, the locally darker parts of the grain region were also mistakenly segmented as connected regions marked in red. Therefore, the connected regions marked in gray cannot be directly identified as grain regions.

[0057] from Figure 3 It can be observed that the connected domains corresponding to the cutting channel region satisfy certain geometric characteristic conditions. Therefore, as... Figure 4 As shown, the connected regions (red areas) that satisfy the geometric characteristics are selected from multiple connected regions as the cutting channel regions.

[0058] An exemplary implementation of selecting connected components that satisfy geometric feature conditions from multiple connected components as the cutting channel region: For example, actual geometric features are extracted for each connected component, including area, perimeter, aspect ratio, etc. One or more starting points (typically prominent areas in the image) are selected for region growing. During growing, neighboring pixels are continuously added to expand the connected component according to the set geometric feature conditions until the conditions are met. After region growing, all grown connected components are filtered, retaining those whose shape or area features best match the cutting path requirements; these are ultimately determined as the cutting path region.

[0059] For example, the geometric feature condition may include the length of the connected component being consistent with the height or width of the image to be measured. That is, when selecting a connected component that satisfies the geometric feature condition from multiple connected components as the cutting channel region, the connected component whose length is consistent with the height of the image to be measured, or whose length is consistent with the width of the image to be measured, may be selected as the cutting channel region.

[0060] In this embodiment, screening by geometric feature size is unaffected by surface contamination or light fluctuations, and the screening operation calculates the size comparison, resulting in higher computational efficiency.

[0061] In one embodiment, when determining the grain regions corresponding to each of the multiple grains, the complement of the cutting path region in the image to be tested can be determined. Based on the area features, multiple connected domains are selected from the connected domains of the complement as the grain regions corresponding to each of the multiple grains.

[0062] In this embodiment, since the dicing channels and the grains in a wafer are complementary, the remaining area after determining the dicing channel region is the grain region. By identifying the more prominent dicing channel region, and based on the complementary relationship between the dicing channel region and the grain region, the location of the grain region can be deduced. Compared to directly identifying the less prominent grain region, this significantly improves the accuracy and reliability of locating the grain region.

[0063] For example, from all connected components obtained by segmenting the image under test, after excluding the connected components that constitute the cut-path region, the remaining connected components can be used as the complement of the connected components corresponding to the cut-path region. This determines the expected number of grains contained in a local wafer region. It can be selected from the connected components of the complement. Each connected component is used as Each grain has its own grain region. Greater than or equal to 2.

[0064] Determining the number of grains In an exemplary implementation, the grain size can be obtained. and and the width of the image and high The maximum number of grains in the horizontal direction in the image under test is calculated as follows: and the largest number of grains in the vertical direction The expected number of grains in a local wafer region. . This indicates rounding up to the nearest integer.

[0065] For example, suppose , , , .So, ; ; .

[0066] Determining the number of grains based on image size and grain size It does not rely on image content or human intervention, and the calculation is simple and fast.

[0067] Selecting from the connected components of the complement Each connected component is used as An exemplary implementation of the individual grain regions of each grain could be to calculate the area of ​​each connected region in the complement set and select the region with the largest area from among all connected regions. Each connected component is used as Each grain has its own grain region.

[0068] By using the size of the area as the screening criterion, noise such as small connected regions (e.g., particles, contaminants) can be quickly eliminated, and only connected regions that conform to the grain size can be retained as the grain regions of the corresponding grains.

[0069] Step 103: For any grain, determine the actual boundary and ideal boundary of the grain region corresponding to the grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state.

[0070] The actual boundary can be the actual geometric boundary of the grain region extracted from the image to be tested, that is, the physical edge contour of the grain after cutting. This boundary is affected by the defects of the cutting process and presents an irregular shape.

[0071] Ideal boundaries can be generated through geometric calculations based on wafer design parameters (such as grain size). For example, if the designed grain geometry is a square, then the ideal boundaries are the four regular edges of that square.

[0072] An exemplary implementation for determining the actual boundary: The actual boundary of the grain region is determined based on the distribution differences of pixel features in the grain region.

[0073] For example, edge detection algorithms can be used to detect the actual boundaries of grain regions in an image under test. For instance, the Canny algorithm can be used to calculate the gradient of the image under test to determine the intensity and direction of the boundaries, and significant boundaries can be extracted by setting a threshold.

[0074] For example, the actual edges of grain regions can be extracted using a deep learning model for contour extraction. For instance, first, an image dataset containing grain regions is collected, and each image is labeled with its actual grain boundaries. Then, a deep learning model is trained to automatically extract the actual boundaries of grain regions from the input images.

[0075] For example, the grain region can be transformed into such Figure 5 The binary image shown represents the percentage of valid points along the X and Y directions. For example, Figure 6 For the statistical curve in the X direction, a cutoff ratio is set to represent the effective starting position of the grain region's edge. For example, when set to 0.1, the area within the first point greater than 0.1 at both ends of the statistical curve represents the actual region enclosed by the grain region's actual boundary. For instance, the edge position of each grain region can be calculated along the normal direction to find a point where the grayscale change meets the requirements. Filling the envelope around all these points yields the true region. The final result is as follows: Figure 7 The actual boundary of the grain region (represented by red serrated lines) is the envelope formed by connecting every point that meets the grayscale change requirement.

[0076] An exemplary implementation of determining the ideal boundary: The ideal boundary of the grain region corresponding to the grain is determined based on the geometric boundary characteristics of the grain region and / or the attribute information of the grain. The attribute information may include the coordinates and size of the grain.

[0077] For example, the ideal boundary of the grain region corresponding to the grain can be determined based on the grain coordinates and the grain size. For instance, the coordinates of the grain closest to the grain region can be determined, and a regular boundary can be constructed around the grain coordinates based on the grain coordinates and the grain size, and this boundary can be used as the ideal boundary of the grain region.

[0078] For example, the minimum bounding rectangle of the grain region can be determined, and the edge of the minimum bounding rectangle can be used as the ideal boundary of the grain region.

[0079] Ultimately, we can obtain the following: Figure 8The ideal boundary of the grain region shown is a smooth straight line.

[0080] Step 104: Based on the difference between the actual boundary and the ideal boundary, determine the corresponding difference region and use the difference region as the cutting defect region corresponding to the grain.

[0081] The difference region can be the part that does not coincide geometrically between the actual boundary and the ideal boundary. For example, the difference between the actual region enclosed by the actual boundary and the ideal region enclosed by the ideal boundary can be defined as the corresponding difference region. For example, such as... Figure 9 As shown, the area formed by the red line is Figure 7 The actual boundary and Figure 8 The difference region between the ideal boundaries.

[0082] In this embodiment, firstly, images are acquired from a local wafer area, focusing on the dicing region to be detected. This avoids scanning the entire wafer, thus reducing processing time. Secondly, by identifying the more prominent dicing region, the location of the grain region is deduced based on the spatial distribution relationship between the dicing region and the grain region. Compared to directly identifying less prominent grain regions, this significantly improves the accuracy and reliability of grain region location. Finally, this solution can identify dicing defect regions without relying on grain patterns, making it applicable to various wafer types.

[0083] Figure 1 The method does not limit the application of the identified cut kerf defect area; exemplary applications include the following: After step 104, the geometric features of the kerf defect region can be determined, including area and / or dimensions. Of course, the geometric features may also include the shape of the kerf defect region, etc. This embodiment does not impose any limitations on the geometric features.

[0084] After calculating the geometric characteristics of the defect area in the cutting path, its applications can include: for example, determining the severity of the defect based on its area; determining the cause of the defect based on its size and shape; for example, if the defect shape is irregular, it can be inferred that the defect is caused by bubbles, liquid dripping, or grain aggregation; if the defect shape is linear, it can be inferred that the defect is mostly related to scratches, tool errors, or object abrasion; for example, if the defect area is large, it can be inferred that the defect is caused by cutting equipment failure, substandard material quality, or uneven thickness; if the defect area is small, it can be inferred that the defect is related to minute particle contamination of the material, localized scratches, or minor wear.

[0085] After step 104, it can also be used to detect whether the grain is damaged: the location of the sealing line in the grain region can be determined, and the distance between the cut defect region and the sealing line can be determined. If the distance is less than a distance threshold, it is determined that the grain is damaged.

[0086] like Figure 2 As shown, the sealing line 22 in the grain region is illustrated. Located inside the grain, the sealing line is a protective ring added to the grain edge to protect it from environmental influences. If the defective area of ​​the cutting kerf is too close to the sealing line, it may compromise the integrity of the grain boundary, leading to sealing line failure and grain damage.

[0087] An exemplary implementation for determining the location of the sealing line in the grain region: In one embodiment, the positions of feature points in the grain region can be identified from the image to be tested, and the position of the center point of the sealing line can be determined based on the relative positional relationship between the feature points and the center point of the sealing line; the position of the sealing line in the grain region can be determined based on the position of the center point and the geometric dimensions of the grain.

[0088] For example, when creating a recipe, you can select uniquely characteristic locations at each corner of the grain as alignment feature points. In the current coordinate system, the coordinates of the center point of the sealing line are: Therefore, it can be based on feature points. and the center point coordinates of the sealing line Determine the relative positional relationship between the feature point and the center point of the sealing line. It should be noted that the above is only an example, and the feature point is not necessarily located at the corner of the grain.

[0089] During detection, feature matching can be performed on the positions of corresponding feature points. If a matching feature point can be found... Then, the coordinates of the center point of the seal line in the image coordinate system can be located based on the position of the feature points. Then, based on the grain's geometry and center point coordinates, the location of the sealing line can be determined.

[0090] In another embodiment, the minimum bounding rectangle of the grain region can be determined, and the vertex coordinates of any vertex of the minimum bounding rectangle can be determined. Based on the vertex coordinates and the size of the grain, the coordinates of the center point of the grain corresponding to the minimum bounding rectangle are determined, and the position of the sealing line is determined based on the relative positional relationship between the center point coordinates and the sealing line.

[0091] For example, for any given grain, the connected region closest to the coordinates of that grain can be considered as the grain region of that grain. Then, the minimum bounding rectangle of this connected region is determined, assuming the coordinates of its top-left corner are... The coordinates of the lower right corner are ,like Figure 10 As shown, the coordinates of the grain's center point are calculated based on the vertex coordinates of the circumscribed rectangle. For example, if the x-coordinate of the vertex... If , it means that the right half of the grain is within the image, then the x-coordinate of the center point is... If the x-coordinate of the vertex If , it means that the left half of the grain is within the image, then the x-coordinate of the center point is... Similarly, if the ordinate of the vertex... If the center point's ordinate is 0, it means the lower half of the grain is within the image. If the x-coordinate of the vertex If the upper part of the grain is within the image, then the ordinate of the center point is... .

[0092] Based on the center point coordinates of the sealing line ( , The relative positional relationship between the seal line and the sealing line can determine the position of the sealing line.

[0093] In this embodiment, it is not necessary to set feature points on the grains, and the position of the sealing line can be deduced based on the geometric relationship, which is especially suitable for scenarios where feature point matching of the grains is not possible.

[0094] In another embodiment, the positions of feature points in the grain region can be identified from the image under test, and the position of the center point of the sealing line can be determined based on the relative positional relationship between the feature points and the center point of the sealing line. The position of the sealing line in the grain region is determined based on the position of the center point of the sealing line and the geometric dimensions of the grain.

[0095] In the event of identification failure, the minimum bounding rectangle of the grain region can be determined, and the vertex coordinates of any vertex of the minimum bounding rectangle can be determined. Based on the vertex coordinates and the size of the grain, the coordinates of the center point of the grain corresponding to the minimum bounding rectangle can be determined. Based on the relative positional relationship between the center point coordinates and the sealing line, the position of the sealing line can be determined.

[0096] In this embodiment, two association identification methods are provided: feature point matching and geometric shape positioning. In the case where feature point matching cannot be used or the matching fails, geometric shape positioning can be used to supplement the determination, ensuring that the position of the sealing line can still be determined even when there are no feature points in the grain or the feature point matching fails, thus improving the robustness of the method.

[0097] An exemplary implementation of determining the distance between the defective area of ​​the cut track and the sealing line: like Figure 11As shown, if the location of the dicing defect area is at the corner of the grain region, the distance (yellow solid line) between the two sealing lines (red dashed line) where the dicing defect area intersects with the corner can be calculated, and the smallest distance is taken as the distance between the dicing defect area and the sealing line.

[0098] like Figure 12 As shown, if the location of the kerf defect area is on the edge of the grain area, the distance between the kerf defect area and the sealing line parallel to the edge (yellow solid line) can be calculated.

[0099] An exemplary implementation of determining whether the location of the kerf defect region is at a corner or edge of the grain region: The minimum distance between the point set in the dicing defect region and the diagonal of the grain can be determined. If the minimum distance is 0, the dicing defect region is located at the corner of the grain region; otherwise, the dicing defect region is located on the edge of the grain region.

[0100] In one embodiment, an exemplary implementation of calculating the distance between the defective area of ​​the cut track and the sealing line parallel to the edge is as follows: This can be done by determining the coordinates of the closest point to the sealing line in the defect area, or by determining the distance between the coordinates and the sealing line parallel to the edge.

[0101] To achieve sub-pixel level positioning, further improvements were made as follows: It can determine the coordinate point closest to the sealing line in the defect area, generate a geometric region around the coordinate point, determine the maximum value of the first derivative of the grayscale projection curve of the geometric region, determine the coordinate point of the maximum value in the geometric region, and calculate the distance between the coordinate point of the maximum value and the sealing line parallel to the edge.

[0102] For example, such as Figure 12 As shown, first obtain the coordinates of the point closest to the sealing line in the defect area of ​​the cutting track. It can generate things like Figure 13 The coordinate points are shown. The geometric region (which can be a rectangle, a square, etc.) Figure 13 (The rectangle in the middle is represented by a solid blue line). Determine the grayscale projection curve of the geometric region, then calculate the maximum value of the first derivative of the grayscale projection curve, and determine the coordinate point of the maximum value in the geometric region. Calculate the distance between the coordinate point of the maximum value and the sealing line parallel to the edge.

[0103] In this embodiment, by introducing the extreme value of the first derivative of the grayscale projection curve for localization, the measurement accuracy of the distance between the defect area and the sealing line is improved from the pixel level to the sub-pixel level, thereby improving the distance measurement accuracy.

[0104] Corresponding to the embodiments of the foregoing methods, this specification also provides embodiments of the apparatus and the terminal to which it is applied.

[0105] Figure 14 This is a schematic diagram illustrating the structure of an electronic device according to an exemplary embodiment. Figure 14 As shown, at the hardware level, the electronic device 1400 includes a processor 1402, an internal bus 1404, a network interface 1406, memory 1408, and non-volatile memory 1410, and may also include other hardware required for business operations. One or more embodiments of this specification can be implemented in software, for example, the processor 1402 reads the corresponding computer program from the non-volatile memory 1410 into the memory 1408 and then runs it. Of course, in addition to software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic module, but can also be hardware or logic devices.

[0106] Figure 15 This is a block diagram illustrating a cutting track defect detection device according to an exemplary embodiment of this specification. Figure 15 As shown, this device can be applied to, for example Figure 14 The electronic device 1400 shown implements the technical solution of this specification. The device includes: The image acquisition module 1502 is used to acquire the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, and the local wafer region includes multiple grain regions and dicing regions located between adjacent grains.

[0107] The grain region identification module 1504 is used to identify the cutting channel region from the image to be tested and determine the grain region corresponding to each of the multiple grains.

[0108] The boundary recognition module 1506 is used to determine the actual boundary and ideal boundary of the grain region corresponding to any grain for any grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state.

[0109] The defect region identification module 1508 is used to determine the corresponding difference region based on the difference between the actual boundary and the ideal boundary, and to regard the difference region as the cutting channel defect region corresponding to the grain.

[0110] Optionally, the grain region identification module 1504 is specifically used to determine a brightness threshold based on the grayscale distribution of the image under test; to segment the image under test based on the brightness threshold to obtain multiple connected regions; and to select a connected region that satisfies the geometric feature conditions from the multiple connected regions as the cutting channel region.

[0111] Optionally, the geometric feature condition includes the length of the connected region being consistent with the length or height of the image to be tested.

[0112] Optionally, the grain region identification module 1504 is specifically used to determine the complement of the cutting track region in the image to be tested; based on area features, multiple connected regions are selected from the connected regions of the complement as the grain regions corresponding to each of the multiple grains.

[0113] Optionally, the grain region identification module 1504 is specifically used to calculate the area of ​​each connected region in the complement set and determine the expected number of grains contained in the local wafer region. ; Select the largest area from each connected component Each connected component is used as Each grain has its own grain region; among which... Greater than or equal to 2.

[0114] Optionally, the device further includes a geometric feature determination module for determining the geometric features of the cut track defect region; the geometric features include area and / or size.

[0115] Optionally, the device further includes a grain quality detection module for determining the position of the sealing line in the grain region and determining the distance between the cutting defect region and the sealing line; if the distance is less than a distance threshold, it is determined that the grain is damaged.

[0116] Optionally, the grain quality detection module is specifically used to identify the positions of feature points in the grain region from the image to be tested, and determine the position of the center point of the sealing line based on the relative positional relationship between the feature points and the center point of the sealing line; determine the position of the sealing line in the grain region based on the position of the center point of the sealing line and the geometric size of the grain; or, determine the minimum bounding rectangle of the grain region, determine the vertex coordinates of any vertex of the minimum bounding rectangle; determine the center point coordinates of the grain corresponding to the minimum bounding rectangle according to the vertex coordinates and the size of the grain; and determine the position of the sealing line according to the relative positional relationship between the center point coordinates and the sealing line.

[0117] Optionally, the grain quality detection module is specifically used to calculate the distance between the two sealing lines that intersect the corner point of the grain region if the location of the dicing defect area is located at the corner point of the grain region, and take the minimum distance as the distance between the dicing defect area and the sealing line; if the location of the dicing defect area is located on the edge of the grain region, calculate the distance between the dicing defect area and the sealing line parallel to the edge.

[0118] Optionally, the grain quality detection module is specifically used to determine the coordinate point closest to the sealing line in the defect region; generate a geometric region around the coordinate point; determine the maximum value of the first derivative of the grayscale projection curve of the geometric region, and determine the coordinate point of the maximum value in the geometric region; and calculate the distance between the coordinate point of the maximum value and the sealing line parallel to the edge.

[0119] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0120] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and 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 the solution in this specification according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0121] This specification also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the aforementioned cutting surface defect detection methods provided in this application.

[0122] Specifically, computer-readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media, and memory devices, such as semiconductor memory devices (e.g., EPROM, EEPROM, and flash memory devices), magnetic disks (e.g., internal hard disks or removable disks), magneto-optical disks, and CD-ROM and DVD-ROM disks.

[0123] This specification also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of any of the aforementioned cutting surface defect detection methods.

Claims

1. A method for detecting defects in cutting surfaces, characterized in that, The method includes: Acquire the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, the local wafer region includes multiple grain regions and dicing regions located between adjacent grains; Identifying the cut-path region from the image under test and determining the grain region corresponding to each of the multiple grains includes: determining a brightness threshold based on the grayscale distribution of the image under test; segmenting the image under test based on the brightness threshold to obtain multiple connected components; selecting connected components that satisfy geometric feature conditions from the multiple connected components as cut-path regions; determining the complement of the cut-path regions in the image under test; and selecting multiple connected components from the connected components of the complement as the grain region corresponding to each of the multiple grains based on area features; wherein, the complement is the connected components remaining after excluding the connected components that are used as cut-path regions from all connected components obtained by segmenting the image under test. For any grain, determine the actual boundary and ideal boundary of the grain region corresponding to the grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state; Based on the difference between the actual boundary and the ideal boundary, the corresponding difference region is determined, and the difference region is used as the cutting defect region corresponding to the grain.

2. The method according to claim 1, characterized in that, The geometric feature conditions include the length of the connected region being consistent with the width or height of the image to be tested.

3. The method according to claim 1, characterized in that, The step of selecting multiple connected regions from the complement set as the grain regions corresponding to each of the multiple grains includes: Calculate the area of ​​each connected region in the complement set, and determine the expected number of grains to be contained in the local wafer region. ; Select the largest area from each connected component. Each connected component is used as Each grain has its own grain region; among which... Greater than or equal to 2.

4. The method according to any one of claims 1 to 3, characterized in that, After the step of identifying the difference region as the kerf defect region corresponding to the grain, the method further includes: Determine the geometric features of the defective region of the cutting track; the geometric features include area and / or dimensions; And / or, Determine the position of the sealing line in the grain region, and determine the distance between the cutting defect region and the sealing line; If the distance is less than a distance threshold, it is determined that the grain is damaged.

5. The method according to claim 4, characterized in that, Determining the position of the sealing line in the grain region includes: The positions of feature points in the grain region are identified from the image to be tested, and the position of the center point of the sealing line is determined based on the relative positional relationship between the feature points and the center point of the sealing line. The position of the sealing line in the grain region is determined based on the location of the center point of the sealing line and the geometric size of the grain; or, Determine the minimum bounding rectangle of the grain region. Determine the vertex coordinates of any vertex of the minimum bounding rectangle; Based on the vertex coordinates and the size of the grain, determine the coordinates of the center point of the grain corresponding to the minimum bounding rectangle; The position of the sealing line is determined based on the relative positional relationship between the center point coordinates and the sealing line.

6. The method according to claim 4, characterized in that, Determining the distance between the defective area of ​​the cutting channel and the sealing line includes: If the location of the cut-through defect area is at the corner of the grain region, then calculate the distance between the two sealing lines where the cut-through defect area intersects with the corner, and take the smallest distance as the distance between the cut-through defect area and the sealing line. If the location of the cut-through defect area is on the edge of the grain area, then calculate the distance between the cut-through defect area and the sealing line parallel to the edge.

7. The method according to claim 6, characterized in that, The calculation of the distance between the defective area of ​​the cutting track and the sealing line parallel to the edge includes: Determine the coordinates of the point in the defect area closest to the sealing line; Generate a geometric region surrounding the coordinate points; Determine the maximum value of the first derivative of the grayscale projection curve of the geometric region, and determine the coordinate point of the maximum value in the geometric region; Calculate the distance between the coordinate point of the maximum value and the sealing line parallel to the edge.

8. A device for detecting defects in cutting surfaces, characterized in that, The device includes: The image to be tested module is used to acquire the image to be tested; the image to be tested includes a local wafer region in the wafer to be tested, the local wafer region includes multiple grain regions and dicing regions located between adjacent grains; A grain region identification module is used to identify kerf regions from the image under test and determine the grain regions corresponding to multiple grains, including: determining a brightness threshold based on the grayscale distribution of the image under test; segmenting the image under test based on the brightness threshold to obtain multiple connected components; selecting connected components that satisfy geometric feature conditions from the multiple connected components as kerf regions; determining the complement of the kerf regions in the image under test; and filtering multiple connected components from the connected components of the complement as the grain regions corresponding to multiple grains based on area features; wherein, the complement is the remaining connected components after excluding the connected components that are used as kerf regions from all connected components obtained by segmenting the image under test; A boundary recognition module is used to determine the actual boundary and ideal boundary of the grain region corresponding to any grain; the actual boundary is the boundary of the grain region in the image, and the ideal boundary is the boundary formed by the grain region in the grain-free state. The defect region identification module is used to determine the corresponding difference region based on the difference between the actual boundary and the ideal boundary, and to regard the difference region as the cutting channel defect region corresponding to the grain.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images