A machine vision identification method, device and medium for surface defects of a silicon wafer

Abstract

The embodiment of the application discloses a machine vision identification method and device for surface defects of a silicon wafer and a medium, and belongs to the technical field of semiconductor manufacturing, and comprises the following steps: identifying a defect area to be classified based on an image of a local defect on a surface of a silicon wafer collected; determining an aspect ratio of the defect area to be classified according to a ratio of a length value and a diameter value of the defect area to be classified, wherein the length value is a maximum value of a distance between edge pixels of the defect area to be classified, and the diameter value is a maximum value of a distance between edge pixels of the defect area to be classified in a direction perpendicular to a straight line on which the length value is located; and identifying a defect type represented by the defect area to be classified according to the aspect ratio. As an outline feature of a defect area, the aspect ratio can more accurately identify and classify defects, further reduces the steps of final identification of defects, thereby realizing rapid classification of main defects on a surface of a silicon wafer and improving the monitoring level of high-end products.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor manufacturing technology, and in particular to a machine vision recognition method, apparatus and medium for silicon wafer surface defects. Background Technology

[0002] In the semiconductor field, silicon wafers are generally the raw material for integrated circuits. Various semiconductor devices can be fabricated from silicon wafers through techniques such as photolithography and ion implantation. The silicon wafer manufacturing process typically includes multiple mechanical surface treatment steps such as slicing, grinding, and polishing. These processes inevitably damage the silicon wafer surface, introducing defects such as cracks, scratches, and dislocations into the surface / subsurface of the wafer. These defects affect the performance of the final semiconductor product, and consequently, the device's lifespan and yield. Therefore, surface defects in silicon wafers have become a key factor in evaluating product quality.

[0003] With the development of advanced manufacturing processes, the feature linewidth of integrated circuits is continuously decreasing, resulting in smaller and fewer defects on the silicon wafer surface. This leads to increasingly stringent requirements for defect control, drastically increasing the difficulty of control. While SEM (Scanning Electron Microscope) measurements can provide a visual morphology of defects, their measurement speed is slow, and they are not currently used for full inspection in mass-produced products. However, future high-end products require even more stringent control, inevitably leading to an increase in the number of samples or full inspections of SEM. Therefore, classifying and statistically analyzing defects based on SEM measurement results is a necessary requirement for high-end products. However, current defect classification based on SEM results relies solely on manual visual inspection, a method far from meeting production demands. To control the large number of nanoscale defects on silicon wafers, there is an urgent need to develop an automated defect identification system to achieve rapid machine identification of major surface defect types, enabling efficient classification, quantity, and density distribution calculations. Summary of the Invention

[0004] In view of this, embodiments of the present invention aim to provide a machine vision recognition method for silicon wafer surface defects; capable of converting the acquired defect image into a grayscale matrix, identifying the defect region based on pixel grayscale values, accurately extracting and judging the aspect ratio of the defect region's shape features, thereby achieving rapid defect identification and classification.

[0005] The technical solution of this invention is implemented as follows:

[0006] In a first aspect, embodiments of the present invention provide a machine vision method for identifying defects on the surface of a silicon wafer, comprising:

[0007] Based on the images of local defects on the silicon wafer surface, the defect regions to be classified are identified.

[0008] The aspect ratio of the defect region to be classified is determined based on the ratio of its length to its diameter. The length of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length.

[0009] The defect type characterized by the defect region to be classified is identified based on the aspect ratio.

[0010] Secondly, embodiments of the present invention provide a machine vision recognition device for silicon wafer surface defects, the device comprising:

[0011] The system consists of three parts: acquisition, feature extraction, and classification / recognition.

[0012] The acquisition part is configured to identify the defect region to be classified based on the acquired image of local defects on the silicon wafer surface;

[0013] The feature extraction part is configured to determine the length-to-diameter ratio of the defect region to be classified based on the ratio of the length to the diameter of the defect region to be classified, wherein the length of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length.

[0014] The classification and identification section is configured to identify the defect type represented by the defect region to be classified based on the aspect ratio.

[0015] Thirdly, embodiments of the present invention provide a computer storage medium storing a machine vision recognition program for silicon wafer surface defects, wherein the machine vision recognition program for silicon wafer surface defects, when executed by at least one processor, implements the steps of the machine vision recognition method for silicon wafer surface defects described in the first aspect.

[0016] This invention provides a machine vision recognition method, device, and medium for silicon wafer surface defects. By using the aspect ratio as a shape feature of the defect area, defects can be more accurately identified and classified, further reducing the steps involved in the final defect identification, thereby achieving rapid classification of major defects on the silicon wafer surface and improving the monitoring level of high-end products. Attached Figure Description

[0017] Figure 1 A schematic flowchart of a machine vision method for identifying defects on the surface of a silicon wafer provided in an embodiment of the present invention;

[0018] Figure 2Examples of common silicon wafer surface defects provided in embodiments of the present invention;

[0019] Figure 3 (a) is a grayscale image of a protruding defect provided in an embodiment of the present invention;

[0020] Figure 3 (b) is a grayscale variation curve of a protruding defect provided in an embodiment of the present invention;

[0021] Figure 3 (c) is a grayscale image of a concave defect provided in an embodiment of the present invention;

[0022] Figure 3 (d) is a grayscale variation curve of a concave defect provided in an embodiment of the present invention;

[0023] Figure 4 This is a schematic diagram showing the position of linear defect pixels in a matrix according to an embodiment of the present invention;

[0024] Figure 5 A flowchart for classifying and identifying silicon wafer surface defects provided in an embodiment of the present invention;

[0025] Figure 6 A schematic diagram of a machine vision recognition device for silicon wafer surface defects provided in an embodiment of the present invention;

[0026] Figure 7 This is a schematic diagram of the hardware structure of a computing device provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0028] Preferred embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0029] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application 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 or all possible combinations of one or more of the associated listed items.

[0030] It should be understood that although the terms "first," "second," etc., may be used in this application to describe various information, such 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 application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0031] Statistical analysis reveals that the majority of small-sized defects on silicon wafer surfaces fall into several categories: fat jut defects, line jut defects, polish-induced defects (PID), scratch defects, particle defects, pit defects, residual defects, and mid-type defects. These defects each possess significantly different characteristics and originate from various process defects. Identifying the type, quantity, and density distribution of various defects on the silicon wafer can serve as a control parameter and provide feedback for process improvement.

[0032] See Figure 1 This invention provides a machine vision method for identifying defects on the surface of silicon wafers. By extracting aspect ratio features from the defects on the silicon wafer surface, the method enables rapid machine identification and efficient classification of defects. This supports in-depth analysis of defect types, quantities, and density distributions, and provides feedback for process improvement. The method includes:

[0033] S101: Identify the defect region to be classified based on the acquired images of local defects on the silicon wafer surface;

[0034] S102: Determine the length-to-diameter ratio of the defect region to be classified based on the ratio of its length to its diameter, wherein the length of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length.

[0035] S103: Identify the defect type represented by the defect region to be classified based on the aspect ratio.

[0036] It should be noted that images of local defects on the silicon wafer surface can be acquired using SEM or other acquisition methods that meet the requirements for acquiring small-sized defects. The aspect ratio, as a shape feature of the defect region, can more accurately identify and classify defects, further reducing the steps involved in the final defect identification process. For example, when identifying the shape feature of a linear defect with a certain curvature, using an aspect ratio such as that of the minimum bounding rectangle makes it easier to identify it as a non-linear defect.

[0037] for Figure 1 In some embodiments of the technical solutions shown, the step of identifying the defect region to be classified based on the image of the acquired local defects on the silicon wafer surface includes:

[0038] The acquired images of local defects on the silicon wafer surface are converted from an RGB pixel matrix to a grayscale matrix;

[0039] A first grayscale threshold and a second grayscale threshold are set based on the average grayscale value of the grayscale matrix, wherein the first grayscale threshold is less than the average grayscale value and the second grayscale threshold is greater than the average grayscale value.

[0040] In the grayscale matrix, pixels with grayscale values ​​less than the first grayscale threshold or greater than the second grayscale threshold are considered defective pixels, and the area formed by all the defective pixels is considered as a defective region to be classified.

[0041] It should be noted that in the grayscale image matrix of a defective silicon wafer, the color of the defective area is either too dark or too bright compared to the normal area, and its pixel grayscale value will be lower or higher than the grayscale value of the normal area accordingly. Therefore, a reasonable grayscale threshold is set based on the average grayscale value, and the defective area is identified by the abnormal pixel grayscale value.

[0042] Therefore, for Figure 1 In some embodiments of the technical solutions shown, identifying the defect type characterized by the defect region to be classified based on the aspect ratio includes:

[0043] If the aspect ratio of the defect region to be classified is greater than the first aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a first type of defect, which includes straight protrusion defects, scratch defects, and polishing damage defects.

[0044] The defect type represented by the defect region to be classified is determined to be a wide protrusion defect, since the aspect ratio of the defect region to be classified is between the first aspect ratio threshold and the second aspect ratio threshold. The first aspect ratio threshold is greater than the second aspect ratio threshold.

[0045] If the aspect ratio of the defect region to be classified is less than the second aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a second type of defect. The second type of defect includes pit defects, granular defects, void defects, residue defects, metal contamination defects, and crystal native defects.

[0046] In some examples of the above implementation, the first aspect ratio threshold is set in the range of [15, 20], and the second aspect ratio threshold is set in the range of [0, 3].

[0047] It should be noted that, as Figure 2 In the grayscale image of the defects shown, L represents the length of the defect region to be classified, and W represents the diameter of the defect region to be classified. Linear convex defects appear as straight convex shapes. Scratch defects generally occur during grinding or polishing when larger hard particles or fragments are mixed into the grinding or polishing powder. The mechanical polishing disc becomes locally too hard, resulting in obvious scratches, which appear as linear depressions with a certain curvature along the grinding or polishing trajectory. Polishing damage defects are defects introduced by polishing and appear as linear convex shapes with a certain curvature. The aspect ratio of these three types of defects is significantly greater than that of other types of defects. Wide convex defects are convex defects with a certain width and can be directly identified based on their aspect ratio. Granular defects are raised and generally round, possibly formed by foreign matter such as silicon slag falling onto the silicon wafer surface; pit defects are recessed pits, also relatively round in shape; residue defects have irregular morphologies, generally discontinuous protrusions, possibly caused by incomplete cleaning resulting in grinding slurry residue remaining on the silicon wafer surface; metal contamination defects are generally honeycomb-shaped pits, defects caused by metal contamination on the silicon wafer surface, forming a morphology of metal impurities accumulating; void defects are caused by SEM operations that attempt to magnify the current area and apply high voltage to find defects, ultimately resulting in a noticeable mark left on the silicon wafer surface by the high-voltage electron beam; crystal native defects are pore-like depressions formed by the aggregation of vacancies in the crystal during the silicon crystal pulling process as the temperature decreases. These pores are octahedral in shape, and their top view is rhomboid. Therefore, the classification of the above common defects based on aspect ratio provided in this embodiment of the invention can avoid identifying linear defects with a certain curvature as wide protrusion defects or second-type defects, and can directly identify wide protrusion defects using only the aspect ratio feature.

[0048] The first and second categories of defects include multiple types of defects. Further, the defect type represented by the defect region to be classified is the first category of defect. The method also includes:

[0049] Determine whether the morphological features of the defect region to be classified are convex or concave:

[0050] If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a third type of defect, including linear convex defects and polishing damage defects.

[0051] If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a scratch defect.

[0052] It should be noted that since scratch defects are concave defects while linear raised defects and polishing damage defects are raised defects, the raised or concave features of the first type of defect are extracted for further identification. Correspondingly, the first type of defect is raised, and the defect type represented by the defect area to be classified is determined to be the third type of defect, including linear raised defects and polishing damage defects. Correspondingly, the first type of defect is concave, and is identified as a scratch defect. Therefore, scratch defects can be identified as the final defect type by using two feature extraction decisions.

[0053] For the identification of raised or recessed features, the grayscale value of the defect area can be used for judgment. Preferably, in the above example, the morphological feature of the defect area to be classified is raised or recessed, including:

[0054] Based on the rows containing defective pixels in the grayscale matrix, at least one row is selected, where the leftmost defective pixel in the row is the left edge pixel, and the rightmost defective pixel in the row is the right edge pixel;

[0055] For the left edge pixel and the right edge pixel, the average gray level change rate k1 of the left edge pixel and the average gray level change rate k2 of the right edge pixel are calculated respectively according to the Sobel operator;

[0056] Calculate the grayscale change rate characteristic value deltak = k1 - λ1 × k2, where λ1 is a positive real number;

[0057] If the grayscale change rate feature value is greater than 0, then the morphological feature of the defect region to be classified is determined to be convex.

[0058] If the grayscale change rate feature value is less than 0, then the morphological feature of the defect region to be classified is determined to be concave.

[0059] In detail, Figure 3 (a) shows a grayscale image of a raised defect. Generally, the grayscale value range of a grayscale image is from 0 to 255, with 255 for white and 0 for black. Raised areas are generally lighter in color, while recessed areas are darker. Figure 3 (b) is Figure 3(a) The curve of the change of grayscale array of a row of defective pixels in the corresponding grayscale matrix shows that the grayscale value of the raised defect area is significantly higher than that of other normal areas. Figure 3 (c) shows a grayscale image of the concave defect. Figure 3 (d) is Figure 3 (c) The corresponding grayscale matrix contains a change curve of a row of grayscale values ​​for the defective region. It can be seen that the grayscale value of the concave defective region is significantly lower than that of other normal regions. Based on this phenomenon, this embodiment of the invention uses the calculation of the grayscale change rate at the edge of the defective region to identify convex or concave features. For example... Figure 3 (b) shows a grayscale curve where the grayscale change rate k1 at the left edge of the defect area is greater than 0, and the grayscale change rate k2 at the right edge of the defect area is less than 0. Therefore, the characteristic value of the grayscale change rate, deltak = k1 - λ1 × k2, is greater than 0. Figure 3 The defect shown in (a) is a protruding defect. For example, Figure 3 As shown in grayscale curve (d), the grayscale change rate k1 at the left edge of the defect area is less than 0, while the grayscale change rate k2 at the right edge of the defect area is greater than 0. Therefore, the characteristic value of the grayscale change rate is less than 0. Figure 3 (c) shows a concave defect. The gray-level change rate at the edge of the defect region can be calculated using the Sobel operator. The Sobel operator is a discrete difference operator that weights the difference between the gray-level values ​​of the four neighborhoods (up, down, left, and right) of each pixel in the image to obtain the first-order gradient of the digital image.

[0060] In the above example, further, corresponding to the defect type represented by the defect region to be classified as a third-class defect, the method further includes:

[0061] At least three sampling points are obtained by uniformly sampling along the long direction of the defect region to be classified;

[0062] Based on the row index and column index of the sampling point in the grayscale matrix, the slope of any two sampling points based on their matrix positions is calculated to obtain a slope array.

[0063] If the variance of the slope array is less than the variance threshold, then the defect type represented by the defect region to be classified is determined to be a linear protrusion defect.

[0064] If the variance of the slope array is greater than or equal to the variance threshold, then the defect type represented by the defect region to be classified is determined to be a polishing damage defect.

[0065] Preferably, the variance threshold can be set to 10. -1 .

[0066] Preferably, the number of sampling points ranges from [5, 8].

[0067] Specifically, when calculating the slope, one can iterate through and calculate the slope between any two sampling points to form a slope array, or one can calculate only the slope between adjacent sampling points to form a slope array. For example... Figure 4 In the grayscale matrix shown, a ij a st a pq This represents three sampling points on the defect region, with the slopes between any two sampling points being respectively... The variance of the slope array is D. Theoretically, when uniformly sampling along an ideal straight line, D is 0, i.e. In practical engineering applications, when D is less than the variance threshold, it is considered to have linear characteristics.

[0068] It should also be noted that metal contamination defects usually have a certain curvature, and the identification of the linear characteristics of the defects can distinguish between straight-line protrusion defects and metal contamination defects.

[0069] In some examples of the above implementation methods, the defect type represented by the defect region to be classified is a second type of defect, and the method further includes:

[0070] Calculate the radius R of the minimum circumcircle of the defect region to be classified;

[0071] The number of pixels contained in the defect region to be classified is S;

[0072] Calculate the near-circularity curve of the defect region to be classified as |π×R|. 2 -S|;

[0073] If the near-circularity ratio is less than the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fourth type of defect, including granular defects and pit defects.

[0074] If the near-circularity ratio is greater than or equal to the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fifth type of defect, including void defects, residual defects, metal contamination defects, and crystal native defects.

[0075] Preferably, the near-circularity threshold can be set to 0.1.

[0076] It should be noted that the smallest circumcircle is the circle with the smallest radius that can contain all the pixels in the defect area.

[0077] In the above example, further, corresponding to the defect type represented by the defect region to be classified as a fourth type of defect, the method further includes:

[0078] Determine whether the morphological characteristics of the defect area to be classified are convex or concave:

[0079] If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a pit defect.

[0080] If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a granular defect.

[0081] It should be noted that for granular defects and pit defects, the final defect type can be identified through three feature extraction methods: aspect ratio, near-circularity, and convexity / concavity. For void defects, residue defects, metal contamination defects, and crystal-native defects, which account for a small proportion of defects in high-end product manufacturing, they will not be further identified for now, but will be statistically analyzed as a single category, namely the fifth type of defect.

[0082] like Figure 5 As shown, this embodiment of the invention utilizes four features—the aspect ratio of the defect region, the pixel grayscale change rate feature value of the defect region, the multi-point slope variance statistics of the defect region, and the near-circularity of the defect region—for stepwise classification and identification, achieving rapid classification of major defects on the silicon wafer surface. Furthermore, the actual length, width, and height (depth) of the defect region can be calculated using the magnification ratio of the SEM image to obtain the defect size, improving the monitoring level of high-end products. In some embodiments, defects can also be classified and identified in different orders based on the above four features. Figure 5 The method steps shown are preferred embodiments that enable rapid identification of the final defect type. For example, for granular defects, pitted defects, linear protrusion defects, and polishing damage defects, identification can be completed using three features. For wide protrusion defects, identification is completed by extracting only one feature. Although the first, second, third, fourth, and fifth types of defects are not the final defect types, they are defect types with one or more common characteristics. Their statistical properties can be used to monitor and analyze defects, thereby providing feedback and guidance for defect analysis and product manufacturing processes.

[0083] Based on the same inventive concept as the aforementioned technical solution, see [link to inventive concept]. Figure 6 This illustration shows a machine vision recognition device 60 for silicon wafer surface defects according to an embodiment of the present invention. The machine vision recognition device 60 for silicon wafer surface defects includes: an acquisition part 601, a feature extraction part 602, and a classification and recognition part 603; wherein,

[0084] The acquisition section 601 is configured to identify the defect region to be classified based on the acquired image of local defects on the silicon wafer surface;

[0085] The feature extraction part 602 is configured to determine the length-to-diameter ratio of the defect region to be classified based on the ratio of the length value to the diameter value of the defect region to be classified, wherein the length value of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter value of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length value.

[0086] The classification and identification section 603 is configured to identify the defect type represented by the defect region to be classified based on the aspect ratio.

[0087] In some examples, the acquisition portion 601 is configured to identify the defect region to be classified based on the image of the acquired local defects on the silicon wafer surface, including:

[0088] The acquired images of local defects on the silicon wafer surface are converted from an RGB pixel matrix to a grayscale matrix;

[0089] A first grayscale threshold and a second grayscale threshold are set based on the average grayscale value of the grayscale matrix, wherein the first grayscale threshold is less than the average grayscale value and the second grayscale threshold is greater than the average grayscale value.

[0090] In the grayscale matrix, pixels with grayscale values ​​less than the first grayscale threshold or greater than the second grayscale threshold are considered defective pixels, and the area formed by all the defective pixels is considered as a defective region to be classified.

[0091] In some examples, the classification and identification section 603 is configured to identify the defect type characterized by the defect region to be classified based on the aspect ratio, including:

[0092] If the aspect ratio of the defect region to be classified is greater than the first aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a first type of defect, which includes straight protrusion defects, scratch defects, and polishing damage defects.

[0093] The defect type represented by the defect region to be classified is determined to be a wide protrusion defect, since the aspect ratio of the defect region to be classified is between the first aspect ratio threshold and the second aspect ratio threshold. The first aspect ratio threshold is greater than the second aspect ratio threshold.

[0094] If the aspect ratio of the defect region to be classified is less than the second aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a second type of defect. The second type of defect includes pit defects, granular defects, void defects, residue defects, metal contamination defects, and crystal native defects.

[0095] In some examples, the classification and recognition section 603 is configured such that the first aspect ratio threshold is set in the range of [15, 20] and the second aspect ratio threshold is set in the range of [0, 3].

[0096] In some examples, the classification and identification section 603 is configured to identify a first-class defect corresponding to the defect type represented by the defect region to be classified, and the method further includes:

[0097] Determine whether the morphological features of the defect region to be classified are convex or concave:

[0098] If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a third type of defect, including linear convex defects and polishing damage defects.

[0099] If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a scratch defect.

[0100] In some examples, the classification and identification section 603 is configured to identify the defect type represented by the defect region to be classified as a third-class defect, and the method further includes:

[0101] At least three sampling points are obtained by uniformly sampling along the long direction of the defect region to be classified;

[0102] Based on the row index and column index of the sampling point in the grayscale matrix, the slope of any two sampling points based on their matrix positions is calculated to obtain a slope array.

[0103] If the variance of the slope array is less than the variance threshold, then the defect type represented by the defect region to be classified is determined to be a linear protrusion defect.

[0104] If the variance of the slope array is greater than or equal to the variance threshold, then the defect type represented by the defect region to be classified is determined to be a polishing damage defect.

[0105] In some examples, the classification and recognition section 603 is configured such that the variance threshold can be set to 10. -1 .

[0106] In some examples, the classification and recognition section 603 is configured such that the number of sampling points ranges from [5, 8].

[0107] In some examples, the classification and identification section 603 is configured to identify a second type of defect corresponding to the defect type represented by the defect region to be classified, and the method further includes:

[0108] Calculate the radius R of the minimum circumcircle of the defect region to be classified;

[0109] The number of pixels contained in the defect region to be classified is S;

[0110] Calculate the near-circularity curve of the defect region to be classified as |π×R|. 2 -S|;

[0111] If the near-circularity ratio is less than the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fourth type of defect, including granular defects and pit defects.

[0112] If the near-circularity ratio is greater than or equal to the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fifth type of defect, including void defects, residual defects, metal contamination defects, and crystal native defects.

[0113] In some examples, the classification and recognition section 603 is configured such that the near-circularity threshold can be set to 0.1.

[0114] In some examples, the classification and identification section 603 is configured to identify a fourth type of defect corresponding to the defect type represented by the defect region to be classified, and the method further includes:

[0115] Determine whether the morphological characteristics of the defect area to be classified are convex or concave:

[0116] If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a pit defect.

[0117] If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a granular defect.

[0118] In some examples, the classification and recognition section 603 is configured such that the morphological features of the defect region to be classified are either convex or concave, including:

[0119] Based on the rows containing defective pixels in the grayscale matrix, at least one row is selected, where the leftmost defective pixel in the row is the left edge pixel, and the rightmost defective pixel in the row is the right edge pixel;

[0120] For the left edge pixel and the right edge pixel, the average gray level change rate k1 of the left edge pixel and the average gray level change rate k2 of the right edge pixel are calculated respectively according to the Sobel operator;

[0121] Calculate the grayscale change rate characteristic value deltak = k1 - λ1 × k2, where λ1 is a positive real number;

[0122] If the grayscale change rate feature value is greater than 0, then the morphological feature of the defect region to be classified is determined to be convex.

[0123] If the grayscale change rate feature value is less than 0, then the morphological feature of the defect region to be classified is determined to be concave.

[0124] Understandably, in this embodiment, "part" can be a part of a circuit, a part of a processor, a part of a program or software, etc., or it can be a unit, a module, or a non-modular one.

[0125] Furthermore, in this embodiment, the components can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional module.

[0126] If the integrated unit is implemented as a software functional module and not sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this embodiment, 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 storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the method described in this embodiment. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0127] Therefore, this embodiment provides a computer storage medium storing a machine vision recognition program for silicon wafer surface defects. When the machine vision recognition program for silicon wafer surface defects is executed by at least one processor, it implements the steps of the machine vision recognition method for silicon wafer surface defects described in the above technical solution.

[0128] Based on the aforementioned machine vision recognition device 60 for silicon wafer surface defects and computer storage medium, see [link to relevant documentation]. Figure 7This illustration shows the specific hardware structure of a computing device 70 capable of implementing the aforementioned machine vision recognition device 60 for silicon wafer surface defects, provided by an embodiment of the present invention. The computing device 70 can be a wireless device, mobile or cellular phone (including so-called smartphones), personal digital assistant (PDA), video game console (including video display, mobile video game device, mobile video conferencing unit), laptop computer, desktop computer, set-top box, tablet computing device, e-book reader, fixed or mobile media player, etc. The computing device 70 includes: a communication interface 701, a memory 702, and a processor 703; the various components are coupled together through a bus system 704. It is understood that the bus system 704 is used to realize the connection and communication between these components. In addition to a data bus, the bus system 704 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 5 The general designated all buses as Bus System 704. Among them,

[0129] The communication interface 701 is used for receiving and sending signals during the process of sending and receiving information with other external network elements;

[0130] The memory 702 is used to store computer programs that can run on the processor 703;

[0131] The processor 703 is used to execute the steps of the machine vision recognition method for silicon wafer surface defects described in the foregoing technical solution when running the computer program, which will not be described in detail here.

[0132] It is understood that the memory 702 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 702 of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0133] The processor 703 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 703 or by instructions in software form. The processor 703 can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory 702, and the processor 703 reads the information in memory 702 and, in conjunction with its hardware, completes the steps of the above method.

[0134] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described herein, or combinations thereof.

[0135] For software implementation, the techniques described herein can be achieved through modules (e.g., procedures, functions, etc.) that perform the functions described herein. The software code can be stored in memory and executed by a processor. The memory can be implemented within the processor or externally.

[0136] Specifically, the processor 703 is also configured to execute the steps of the machine vision recognition method for silicon wafer surface defects described in the foregoing technical solution when running the computer program, which will not be elaborated here.

[0137] It is understood that the exemplary technical solutions of the aforementioned machine vision recognition device 60 and computing device 70 for silicon wafer surface defects belong to the same concept as the aforementioned machine vision recognition method for silicon wafer surface defects. Therefore, details not described in detail above regarding the technical solutions of the machine vision recognition device 60 and computing device 70 for silicon wafer surface defects can be found in the description of the aforementioned machine vision recognition method for silicon wafer surface defects. This embodiment of the invention will not elaborate further on these details.

[0138] It should be noted that the technical solutions described in the embodiments of the present invention can be combined arbitrarily without conflict.

[0139] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A machine vision method for identifying surface defects on silicon wafers, characterized in that, include: Based on the images of local defects on the silicon wafer surface, the defect regions to be classified are identified. The aspect ratio of the defect region to be classified is determined based on the ratio of its length to its diameter. The length of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length. The defect type characterized by the defect region to be classified is identified based on the aspect ratio. The step of identifying the defect type characterized by the defect region to be classified based on the aspect ratio includes: If the aspect ratio of the defect region to be classified is greater than the first aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a first type of defect, which includes straight protrusion defects, scratch defects, and polishing damage defects. The defect type represented by the defect region to be classified is determined to be a wide protrusion defect, given that the aspect ratio of the defect region to be classified is between the first aspect ratio threshold and the second aspect ratio threshold. The first aspect ratio threshold is greater than the second aspect ratio threshold. If the aspect ratio of the defect region to be classified is less than the second aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a second type of defect. The second type of defect includes pit defects, granular defects, void defects, residue defects, metal contamination defects, and crystal native defects. The method further includes: (The first type of defect is defined as the defect region to be classified.) Determine whether the morphological features of the defect region to be classified are convex or concave: If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a third type of defect, including linear convex defects and polishing damage defects. The method further includes: (The method is defined as follows:) The defect type corresponding to the defect region to be classified is classified as a third-class defect. At least three sampling points are obtained by uniformly sampling along the long direction of the defect region to be classified; Based on the row index and column index of the sampling point in the gray matrix, the slope of any two sampling points based on their matrix positions is calculated to obtain a slope array. If the variance of the slope array is less than the variance threshold, then the defect type represented by the defect region to be classified is determined to be a linear protrusion defect. If the variance of the slope array is greater than or equal to the variance threshold, then the defect type represented by the defect region to be classified is determined to be a polishing damage defect.

2. The method according to claim 1, characterized in that, The method of identifying the defect region to be classified based on the acquired images of local defects on the silicon wafer surface includes: The acquired images of local defects on the silicon wafer surface are converted from an RGB pixel matrix to a grayscale matrix; A first grayscale threshold and a second grayscale threshold are set based on the average grayscale value of the grayscale matrix, wherein the first grayscale threshold is less than the average grayscale value and the second grayscale threshold is greater than the average grayscale value. In the grayscale matrix, pixels with grayscale values ​​less than the first grayscale threshold or greater than the second grayscale threshold are considered defective pixels, and the area formed by all the defective pixels is considered as a defective region to be classified.

3. The method according to claim 2, characterized in that, The method further includes: (The first type of defect is defined as the defect region to be classified.) Determine whether the morphological features of the defect region to be classified are convex or concave: If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a scratch defect.

4. The method according to claim 3, characterized in that, The method further includes: (The defect type corresponding to the defect region to be classified is designated as a second type of defect.) Calculate the radius R of the minimum circumcircle of the defect region to be classified; The number of pixels contained in the defect region to be classified is S; Calculate the near-circularity ratio of the defect region to be classified. ; If the near-circularity ratio is less than the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fourth type of defect, including granular defects and pit defects. If the near-circularity ratio is greater than or equal to the near-circularity ratio threshold, then the defect type represented by the defect region to be classified is determined to be a fifth type of defect, including void defects, residual defects, metal contamination defects, and crystal native defects.

5. The method according to claim 4, characterized in that, The method further includes: (The method is defined as follows:) The defect type corresponding to the defect region to be classified is classified as a fourth type of defect. Determine whether the morphological characteristics of the defect area to be classified are convex or concave: If the morphological feature of the defect region to be classified is concave, then the defect type represented by the defect region to be classified is determined to be a pit defect. If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a granular defect.

6. The method according to claim 3 or 5, characterized in that, The morphological characteristics of the defect area to be classified are determined to be either convex or concave, including: Based on the rows containing defective pixels in the grayscale matrix, at least one row is selected, where the leftmost defective pixel in the row is the left edge pixel, and the rightmost defective pixel in the row is the right edge pixel; For the left edge pixels and the right edge pixels, the average grayscale change rate k1 of the left edge pixels and the average grayscale change rate k2 of the right edge pixels are calculated respectively according to the Sobel operator; Calculate the characteristic value of grayscale change rate , where λ1 is a positive real number; If the grayscale change rate feature value is greater than 0, then the morphological feature of the defect region to be classified is determined to be convex. If the grayscale change rate feature value is less than 0, then the morphological feature of the defect region to be classified is determined to be concave.

7. A machine vision recognition device for silicon wafer surface defects, characterized in that, The device includes: an acquisition section, a feature extraction section, and a classification and recognition section; wherein... The acquisition part is configured to identify the defect region to be classified based on the acquired image of local defects on the silicon wafer surface; The feature extraction part is configured to determine the length-to-diameter ratio of the defect region to be classified based on the ratio of the length to the diameter of the defect region to be classified, wherein the length of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified, and the diameter of the defect region to be classified is the maximum distance between the edge pixels of the defect region to be classified in the direction perpendicular to the line containing the length. The classification and identification section is configured to identify the defect type represented by the defect region to be classified based on the aspect ratio. The classification and recognition section is further configured as follows: If the aspect ratio of the defect region to be classified is greater than the first aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a first type of defect, which includes straight protrusion defects, scratch defects, and polishing damage defects. The defect type represented by the defect region to be classified is determined to be a wide protrusion defect, given that the aspect ratio of the defect region to be classified is between the first aspect ratio threshold and the second aspect ratio threshold. The first aspect ratio threshold is greater than the second aspect ratio threshold. If the aspect ratio of the defect region to be classified is less than the second aspect ratio threshold, the defect type represented by the defect region to be classified is determined to be a second type of defect. The second type of defect includes pit defects, granular defects, void defects, residue defects, metal contamination defects, and crystal native defects. The defect type represented by the defect region to be classified is the first type of defect: Determine whether the morphological features of the defect region to be classified are convex or concave: If the morphological feature of the defect region to be classified is convex, then the defect type represented by the defect region to be classified is determined to be a third type of defect, including linear convex defects and polishing damage defects. The defect type represented by the defect region to be classified is the third type of defect: At least three sampling points are obtained by uniformly sampling along the long direction of the defect region to be classified; Based on the row index and column index of the sampling point in the gray matrix, the slope of any two sampling points based on their matrix positions is calculated to obtain a slope array. If the variance of the slope array is less than the variance threshold, then the defect type represented by the defect region to be classified is determined to be a linear protrusion defect. If the variance of the slope array is greater than or equal to the variance threshold, then the defect type represented by the defect region to be classified is determined to be a polishing damage defect.

8. A computer storage medium, characterized in that, The computer storage medium stores a machine vision recognition program for silicon wafer surface defects, which, when executed by at least one processor, implements the steps of the machine vision recognition method for silicon wafer surface defects according to any one of claims 1 to 6.

Images