Defect detection method, defect detection system and storage medium

Abstract

The invention provides a defect detection method, a defect detection system and a computer readable storage medium. The defect detection method comprises the following steps: acquiring an image of a sample to be detected; the sample to be detected comprises a plurality of crystal grains to be detected; extracting gray level distribution of the image, and performing clustering analysis in combination with the position of each crystal grain so as to divide each crystal grain into a plurality of detection classes; selecting crystal grains to be detected in each detection class one by one, taking at least one crystal grain closest to the crystal grains in the same detection class as a reference crystal grain, and calculating gray scale difference distribution between each crystal grain to be detected and the reference crystal grain one by one; respectively calculating the median of the gray distribution of all the crystal grains in each detection class so as to dynamically compensate a preset defect detection threshold value; and determining the pixels of which the gray difference is greater than the dynamically compensated defect detection threshold value in the gray difference distribution of the to-be-detected crystal grains as detected defects.

Description

Technical Field

[0001] The present invention relates to the technical field of semiconductor defect detection, and in particular to a defect detection method, a defect detection system, and a computer-readable storage medium. Background Art

[0002] Automatic Optical Inspection (AOI) technology can achieve fast, high-precision, non-destructive testing, and is therefore widely used in semiconductors, LEDs, pan-semiconductors, solar panels and other fields. In existing defect detection equipment, the grain to be tested is usually compared with several adjacent grains or standard grains, and the obtained difference signal is compared with the set threshold. When the difference signal is greater than the set threshold, it is considered that the grain to be tested has a defect. However, due to the influence of process fluctuations, there will be color differences in grains at different positions in the wafer. For example, there is a large difference between the grayscale values ​​of the grains at the edge of the wafer and the grayscale values ​​of the grains at the center of the wafer. This introduces additional errors when comparing grayscale differences, thereby affecting the effective detection rate of defects or causing false detections.

[0003] In order to solve the error problem caused by chromatic aberration, the existing patent CN111862076A divides the grains according to the wafer radius area to reduce the impact of chromatic aberration in grain comparison detection. However, due to the complex chromatic aberration of wafer graphics, which is not necessarily distributed according to concentric arcs, this method is not very universal in defect detection technology. In addition, the existing patent CN118505700A divides the wafer into several sub-areas, compares the grayscale differences of different wafer sub-areas, and screens out the wafers on the color difference line. The defect information of this part of the wafer can be compared and analyzed with the standard qualified wafers produced on the same production line, and the defect information of the wafers in the non-color difference line area can be obtained by judging the grayscale difference between adjacent wafers, thereby completing the defect information of the entire wafer. However, color difference distinction based only on grayscale differences between wafers is relatively rough, and cannot effectively reduce the impact of color difference on defect detection.

[0004] In order to overcome the above-mentioned defects in the prior art, there is an urgent need in the art for an improved defect detection method to reduce the influence of color difference on the semiconductor defect detection effect, thereby improving the defect detection rate. Summary of the invention

[0005] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or decisive elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0006] In order to overcome the above-mentioned defects of the prior art, the present invention provides a defect detection method, a defect detection system and a computer-readable storage medium, which can dynamically compensate for a preset defect detection threshold according to the grayscale distribution of a grain to be tested and multiple reference grains, so as to reduce the influence of chromatic aberration on the semiconductor defect detection effect, thereby improving the defect detection rate.

[0007] Specifically, the defect detection method provided according to the first aspect of the present invention includes the following steps: obtaining an image of a sample to be tested. The sample to be tested includes a plurality of grains to be tested; extracting the grayscale distribution of the image, and performing cluster analysis in combination with the positions of each of the grains to divide each of the grains into a plurality of detection classes; selecting the grains to be tested one by one in each of the detection classes, and taking at least one of the grains closest to it in the same detection class as a reference grain, so as to calculate the grayscale difference distribution between each of the grains to be tested and its reference grain one by one; respectively calculating the median of the grayscale distribution of all grains in each of the detection classes to dynamically compensate for the preset defect detection threshold; and determining the pixels in the grayscale difference distribution of each of the grains to be tested whose grayscale difference is greater than the defect detection threshold after dynamic compensation as detected defects.

[0008] Furthermore, in some embodiments of the present invention, the step of acquiring an image of the sample to be tested includes: fixing the sample to be tested on a moving stage of a defect detection system; starting an illumination light source of the defect detection system to illuminate a detection area via its optical module; moving each of the grains on the sample to be tested to the detection area in turn via the moving stage, and collecting reflected light provided by the sample to be tested via an image sensor to obtain detection images of each of the grains respectively; and splicing the detection images of each of the grains into an image of the sample to be tested.

[0009] Furthermore, in some embodiments of the present invention, the step of extracting the grayscale distribution of the image includes: performing a primary convolution process on the image to extract its first grayscale distribution; performing chromatic aberration correction on the image based on the first grayscale distribution; and performing a secondary convolution process on the image after chromatic aberration correction to extract its second grayscale distribution.

[0010] Furthermore, in some embodiments of the present invention, the step of performing the primary convolution processing and / or the secondary convolution processing includes: selecting a convolution kernel whose size is an integer multiple of the grain resolution, and performing convolution processing on the image to extract its grayscale distribution.

[0011] Furthermore, in some embodiments of the present invention, the step of performing chromatic aberration correction on the image according to the first grayscale distribution includes: obtaining the grayscale mean of standard grains; determining a correction coefficient distribution according to a ratio of the first grayscale distribution to the grayscale mean; and processing the image according to the correction coefficient distribution to perform chromatic aberration correction on it.

[0012] Furthermore, in some embodiments of the present invention, before extracting the grayscale distribution of the image, the defect detection method also includes the following steps: determining a region of interest based on the positions of the multiple grains to be inspected in the image; and applying a mask to non-interest regions in the image except the region of interest to prevent subsequent steps from extracting its grayscale distribution.

[0013] Further, in some embodiments of the present invention, the step of extracting the grayscale distribution of the image and performing cluster analysis in combination with the position of each of the grains to classify each of the grains into a plurality of detection classes includes: determining the grayscale features of each of the grains according to the grayscale distribution of the image. The grayscale features include at least one of the grayscale mean, the grayscale variance, and the grayscale distribution; obtaining the position features of each of the grains. The position features include the position coordinates of each of the grains in the sample to be tested and / or the corresponding grain group of the sample to be tested; and performing the cluster analysis according to the grayscale features and position features of each of the grains to classify each of the grains into a plurality of the detection classes.

[0014] Furthermore, in some embodiments of the present invention, the step of respectively calculating the median of the grayscale distribution of all grains in each of the detection classes to dynamically compensate for a preset defect detection threshold includes: determining a dynamic compensation coefficient based on the ratio of the grayscale difference distribution between the grain to be tested and its reference grain to the median of the grayscale distribution of all grains in the corresponding detection class; and determining the defect detection threshold after dynamic compensation based on the product of the dynamic compensation coefficient and the preset defect detection threshold.

[0015] Furthermore, in some embodiments of the present invention, the step of taking at least one grain closest to the grain in the same detection class as a reference grain to calculate the grayscale difference distribution between each grain to be tested and its reference grain one by one includes: taking the two grains closest to the grain to be tested in the same detection class as reference grains; determining a first grayscale difference distribution based on the grayscale difference distribution between the grain to be tested and the first reference grain; and determining a second grayscale difference distribution based on the grayscale difference distribution between the grain to be tested and the second reference grain.

[0016] Further, in some embodiments of the present invention, the step of determining pixels whose grayscale differences in the grayscale difference distribution of each of the tested grains are greater than the defect detection threshold after dynamic compensation as detected defects includes: comparing the first grayscale difference of each of the pixels in the first grayscale difference distribution of each of the tested grains with the defect detection threshold after dynamic compensation to determine at least one first candidate defect; comparing the second grayscale difference of each of the pixels in the second grayscale difference distribution of each of the tested grains with the defect detection threshold after dynamic compensation to determine at least one second candidate defect; and determining at least one pixel that is simultaneously determined as the first candidate defect and the second candidate defect as the detected defect.

[0017] In addition, the defect detection system provided according to the second aspect of the present invention includes a memory and a processor. The memory stores computer instructions. The processor is connected to the memory and is used to execute the computer instructions stored in the memory to implement the defect detection method provided by the first aspect of the present invention.

[0018] Furthermore, in some embodiments of the present invention, the defect detection system further comprises a motion stage, an illumination light source, an optical module and an image sensor. The motion stage is used to carry and move the sample to be tested. The illumination light source is used to provide illumination light. The optical module is used to uniformly transmit the illumination light provided by the illumination light source to the detection area of ​​the defect detection system. The image sensor is used to collect the reflected light provided by the sample to be tested in the detection area to obtain a corresponding detection image.

[0019] In addition, the computer-readable storage medium provided in the third aspect of the present invention stores computer instructions, which, when executed by a processor, implement the defect detection method provided in the first aspect of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0021] Figure 1 A schematic diagram of the architecture of a defect detection system provided according to some embodiments of the present invention is shown.

[0022] Figure 2 A schematic diagram of the architecture of a processor provided according to some embodiments of the present invention is shown.

[0023] Figure 3A schematic flow chart of the operating principle of a processor provided according to some embodiments of the present invention is shown.

[0024] Figure 4 A schematic flow chart of a defect detection method provided according to some embodiments of the present invention is shown.

[0025] Figure 5A A schematic diagram of a wafer before chromatic aberration correction according to some embodiments of the present invention is shown.

[0026] Figure 5B A schematic diagram of a wafer after chromatic aberration correction according to some embodiments of the present invention is shown.

[0027] Fig. 6A A normal grain map provided according to some embodiments of the present invention is shown.

[0028] Figure 6B A grayscale distribution diagram of normal grains provided according to some embodiments of the present invention is shown.

[0029] Figure 6C A normal grain image after chromatic aberration correction according to some embodiments of the present invention is shown.

[0030] Fig. 7A A chromatic aberration grain map provided according to some embodiments of the present invention is shown.

[0031] Figure 7B A grayscale distribution diagram of chromatic aberration grains provided according to some embodiments of the present invention is shown.

[0032] Figure 7C A chromatic aberration grain map after chromatic aberration correction according to some embodiments of the present invention is shown.

[0033] Figure 8 A schematic diagram showing an area of ​​interest and an area of ​​non-interest of a chromatic aberration grain provided according to some embodiments of the present invention is shown.

[0034] Fig. 9 A schematic diagram showing an area of ​​interest and an area of ​​non-interest of a normal grain provided according to some embodiments of the present invention is shown.

[0035] Fig.10 A schematic diagram of a mask provided according to some embodiments of the present invention is shown.

[0036] Fig.11 A schematic diagram of performing cluster analysis on grains according to some embodiments of the present invention is shown.

[0037] Fig. 12A A grayscale distribution diagram of a grain to be tested provided according to some embodiments of the present invention is shown.

[0038] Fig. 12B A grayscale distribution diagram of reference grains provided according to some embodiments of the present invention is shown.

[0039] Fig. 12C A grayscale difference distribution diagram between a test die and a standard die provided according to some embodiments of the present invention is shown.

[0040] Fig.12D An adjusted grayscale difference threshold map provided according to some embodiments of the present invention is shown. DETAILED DESCRIPTION

[0041] The following specific embodiments illustrate the implementation of the present invention, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Although the description of the present invention will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may be extended based on the claims of the present invention. In order to provide a deep understanding of the present invention, the following description will include many specific details. The present invention can also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present invention, some specific details will be omitted in the description.

[0042] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0043] In addition, the terms "upper", "lower", "left", "right", "top", "bottom", "horizontal" and "vertical" used in the following description should be understood as the directions shown in the paragraph and the related drawings. Such relative terms are only used for the convenience of description and do not mean that the device described therein must be manufactured or operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0044] It is understood that although the terms "first", "second", "third", etc. may be used herein to describe various components, regions, layers and / or parts, these components, regions, layers and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers and / or parts. Therefore, the first component, region, layer and / or part discussed below may be referred to as a second component, region, layer and / or part without departing from some embodiments of the present invention.

[0045] As mentioned above, in existing defect detection equipment, the tested grain is usually compared with several adjacent grains or standard grains, and the obtained difference signal is compared with the set threshold. When the difference signal is greater than the set threshold, it is considered that the tested grain has a defect. However, due to the influence of process fluctuations, there will be color differences in grains at different positions in the wafer. For example, there is a large difference between the grayscale value of the grain at the edge of the wafer and the grayscale value of the grain at the center of the wafer. This introduces additional errors in the grayscale difference comparison, thereby affecting the effective detection rate of defects or causing false detection.

[0046] In order to solve the error problem caused by chromatic aberration, the existing patent CN111862076A divides the grains according to the wafer radius area to reduce the impact of chromatic aberration in grain comparison detection. However, due to the complex chromatic aberration of wafer graphics, which is not necessarily distributed according to concentric arcs, this method is not very universal in defect detection technology. In addition, the existing patent CN118505700A divides the wafer into several sub-areas, compares the grayscale differences of different wafer sub-areas, and screens out the wafers on the color difference line. The defect information of this part of the wafer can be compared and analyzed with the standard qualified wafers produced on the same production line, and the defect information of the wafers in the non-color difference line area can be obtained by judging the grayscale difference between adjacent wafers, thereby completing the defect information of the entire wafer. However, color difference distinction based only on grayscale differences between wafers is relatively rough, and cannot effectively reduce the impact of color difference on defect detection.

[0047] In order to overcome the above-mentioned defects of the prior art, the present invention provides a defect detection method, a defect detection system and a computer-readable storage medium, which can dynamically compensate for a preset defect detection threshold according to the grayscale distribution of a grain to be tested and multiple reference grains, so as to reduce the influence of chromatic aberration on the semiconductor defect detection effect, thereby improving the defect detection rate.

[0048] In some non-limiting embodiments, the defect detection method provided by the first aspect of the present invention may be implemented based on the defect detection system provided by the second aspect of the present invention.

[0049] Please refer to Figure 1 . Figure 1 A schematic diagram of the architecture of a defect detection system provided according to some embodiments of the present invention is shown.

[0050] exist Figure 1In the illustrated embodiment, the defect detection system is configured with a memory and a processor 10. The memory includes but is not limited to the computer-readable storage medium provided in the third aspect of the present invention, on which computer instructions are stored. The processor 10 is connected to the memory and is configured to execute the computer instructions stored in the memory to implement the above-mentioned defect detection method provided in the first aspect of the present invention.

[0051] In addition, Figure 1 In the illustrated embodiment, the above-mentioned defect detection system provided by the second aspect of the present invention further includes a moving stage 11, an illumination light source 12, an optical module 13 and an image sensor 14. Here, the moving stage 11 is used to carry and move the sample to be tested 15. The illumination light source 12 is used to provide illumination light. The optical module 13 is used to uniformly transmit the illumination light provided by the illumination light source 12 to the detection area of ​​the defect detection system. The image sensor 14 is used to collect the reflected light provided by the sample to be tested in the detection area to obtain the corresponding detection image.

[0052] Please refer to further Figure 2 and Figure 3 . Figure 2 A schematic diagram of the architecture of a processor provided according to some embodiments of the present invention is shown. Figure 3 A schematic flow chart of the operating principle of a processor provided according to some embodiments of the present invention is shown.

[0053] exist Figure 2 and Figure 3 In the illustrated embodiment, the processor 10 includes an image preprocessing module 101, a grain clustering module 102, a grain comparison detection module 103, and a result storage module 104. Here, the image preprocessing module 101 is used to perform color difference correction on the image. The grain clustering module 102 is used to perform cluster analysis on the grayscale features and position features of each grain to divide each grain into multiple detection classes. The grain comparison detection module 103 is used to determine the grayscale difference distribution between each grain to be tested and its reference grain. The result storage module 104 is used to summarize and store the defect detection results of each grain to be tested.

[0054] The working principle of the above-mentioned defect detection system will be described below in conjunction with some examples of defect detection methods. Those skilled in the art will understand that these examples of defect detection methods are only some non-limiting implementation methods provided by the present invention, which are intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are convenient for the public to implement, rather than to limit all functions or all working modes of the defect detection system. Similarly, the defect detection system is also only a non-limiting implementation method provided by the present invention, and does not constitute a limitation on the execution subject or execution order of each step in these defect detection methods.

[0055] Please refer to the reference for details Figure 4 . Figure 4 A schematic flow chart of a defect detection method provided according to some embodiments of the present invention is shown.

[0056] like Figure 4 As shown, the defect detection system provided by the second aspect of the present invention can first obtain an image of the sample to be tested 15. Here, the sample to be tested 15 includes a plurality of grains to be detected.

[0057] Specifically, the defect detection system can first fix the sample to be tested 15 on the moving table 11. Afterwards, the defect detection system can start the illumination light source 12 to illuminate a detection area through its optical module 13. Here, the detection area can contain multiple grains at the same time. Afterwards, the defect detection system can move each grain on the sample to be tested 15 to the detection area in sequence via the moving table 11, and collect the reflected light provided by the sample to be tested 15 via the image sensor 14 (for example: a line scan camera) to obtain the detection image of each grain respectively. Afterwards, the defect detection system can splice the detection images of each grain into an image of the sample to be tested 15.

[0058] In this way, the above-mentioned defect detection system provided in the second aspect of the present invention can improve the imaging quality and detection accuracy by illuminating the detection area and using a line scan camera to scan and image each area.

[0059] Afterwards, the processor 10 of the defect detection system may extract the grayscale distribution of the image and perform cluster analysis in combination with the position of each grain to classify each grain into a plurality of detection classes.

[0060] Specifically, in the process of extracting the grayscale distribution of the image, the processor 10 may first perform a primary convolution process on the image to extract its first grayscale distribution, and perform color difference correction on the image according to the first grayscale distribution via the image preprocessing module 101. Thereafter, the processor 10 may perform a secondary convolution process on the image after the color difference correction to extract its second grayscale distribution.

[0061] Please refer to Figure 5A and Figure 5B . Figure 5A A schematic diagram of a wafer before chromatic aberration correction according to some embodiments of the present invention is shown. Figure 5B A schematic diagram of a wafer after chromatic aberration correction according to some embodiments of the present invention is shown.

[0062] Further, in Figure 5A In the illustrated embodiment, during the primary convolution processing and / or the secondary convolution processing, the processor 10 may select a convolution kernel 53 whose size is an integer multiple of the grain resolution to perform convolution processing on the image to extract its grayscale distribution.

[0063] Afterwards, in the process of performing chromatic aberration correction on the image according to the first grayscale distribution, the processor 10 may first obtain the grayscale mean value Golden_Die_gray_value of the standard grain. Here, the grayscale mean value of the standard grain may be determined according to the image of the defect-free grain (Golden Die) preset in the program of the defect detection system. Alternatively, in some embodiments where the above-mentioned defect-free grain image is not set, the median grain (Median Die) may also be calculated according to the median value of the grain grayscale as the standard grain.

[0064] Afterwards, the processor 10 may determine the correction coefficient distribution Correction_coeff according to the ratio of the first grayscale distribution to the grayscale mean:

[0065] Correction_coeff=Discolor_gray_map / Golden_Die_gray_value

[0066] Among them, Discolor_gray_map is the color difference grayscale distribution.

[0067] Afterwards, the processor 10 may process the image according to the correction coefficient distribution to perform chromatic aberration correction on it:

[0068] Image_corrected=Image / Correction_coeff

[0069] Among them, Image is the pixel value distribution of the original image, and Image_corrected is the pixel value distribution of the image after chromatic aberration correction.

[0070] like Figure 5A As shown, there is a large color difference between the normal grains 51 in the center area of ​​the wafer and the color difference grains 52 in the edge area of ​​the wafer. Figure 5B As shown, after the chromatic aberration correction is performed, the chromatic aberration between the grains 54 in various regions of the wafer is significantly reduced.

[0071] Please refer to Figure 6A to Figure 6C and Figure 7A to Figure 7C . Fig. 6A A normal grain map provided according to some embodiments of the present invention is shown. Figure 6B A grayscale distribution diagram of normal grains provided according to some embodiments of the present invention is shown. Figure 6C A normal grain image after chromatic aberration correction according to some embodiments of the present invention is shown. Fig. 7A A chromatic aberration grain map provided according to some embodiments of the present invention is shown. Figure 7B A grayscale distribution diagram of chromatic aberration grains provided according to some embodiments of the present invention is shown. Figure 7C A chromatic aberration grain map after chromatic aberration correction according to some embodiments of the present invention is shown.

[0072] like Figure 6A to Figure 6C and Figure 7A to Figure 7C As shown, in the normal grain image and the chromatic aberration grain image after chromatic aberration correction, the chromatic aberration between each grain is reduced, but the effect of reducing the chromatic aberration between each grain in the chromatic aberration grain is more obvious.

[0073] Please refer to Figure 8 to Figure 10 . Figure 8 A schematic diagram showing an area of ​​interest and an area of ​​non-interest of a chromatic aberration grain provided according to some embodiments of the present invention is shown. Fig. 9 A schematic diagram showing an area of ​​interest and an area of ​​non-interest of a normal grain provided according to some embodiments of the present invention is shown. Fig.10 A schematic diagram of a mask provided according to some embodiments of the present invention is shown.

[0074] In addition, in some preferred embodiments, before extracting the grayscale distribution of the image, the processor 10 may first determine a region of interest according to the positions of the plurality of grains to be detected in the image. Afterwards, the processor 10 may apply a mask 80 to the non-interest region in the image except the region of interest to prevent the subsequent steps from extracting its grayscale distribution.

[0075] Specifically, if Figure 8 and Fig. 9 As shown, the non-interest region 91 in the normal crystal grain 51 and the non-interest region 81 in the chromatic aberration crystal grain 52 have no chromatic aberration, and the interest region 92 in the normal crystal grain 51 and the interest region 82 in the chromatic aberration crystal grain 52 have chromatic aberration. When the defect detection system performs detection, if the non-interest regions 81, 91 and the interest regions 82, 92 are put together for chromatic aberration correction, the two regions will affect each other. Fig.10 As shown, this effect can be solved by a mask 80. Here, the white part in the mask 80 is the area of ​​interest for which the color difference is considered, and the black part is the area of ​​no interest for which the color difference is not considered. The mask 80 can be automatically generated by the data provided by the customer or the scanned image during the initial setting stage.

[0076] Please refer to Fig.11 . Fig.11 A schematic diagram of performing cluster analysis on grains according to some embodiments of the present invention is shown.

[0077] After obtaining the grayscale distribution of the image, the processor 10 can determine the grayscale feature of each grain according to the grayscale distribution of the image through the grain clustering module 102. Here, the grayscale feature includes at least one of grayscale mean, grayscale variance, and grayscale distribution.

[0078] Afterwards, the processor 10 may obtain the position feature of each grain. Here, the position feature includes the position coordinates of each grain in the sample to be tested and / or the corresponding grain group of the sample to be tested.

[0079] After that, if Fig.11 As shown, the processor 10 can perform cluster analysis based on the grayscale characteristics and position characteristics of each grain using common algorithms such as K-means, hierarchical clustering, and DBSCAN to divide each grain into multiple detection categories.

[0080] Please refer to further FIG. 12A to FIG. 12D . Fig. 12A A grayscale distribution diagram of a grain to be tested provided according to some embodiments of the present invention is shown. Fig. 12B A grayscale distribution diagram of reference grains provided according to some embodiments of the present invention is shown. Fig. 12C A grayscale difference distribution diagram between a test die and a standard die provided according to some embodiments of the present invention is shown. Fig.12D An adjusted grayscale difference threshold map provided according to some embodiments of the present invention is shown.

[0081] Afterwards, the processor 10 can select the tested grains one by one in each detection class through the grain comparison detection module 103, and use at least one grain closest to it in the same detection class as a reference grain, so as to calculate the grayscale difference distribution between each tested grain and its reference grain one by one. Here, the at least one closest grain is determined by the distance obtained by cluster analysis. The distance includes but is not limited to Euclidean distance, Manhattan distance, Chebyshev distance, and cosine similarity.

[0082] Specifically, if Fig. 12A and Fig. 12B As shown, the processor 10 can use the two crystals in the same detection category that are closest to the crystal to be detected as reference crystals. Fig. 12C As shown, the processor 10 can determine the first grayscale difference distribution according to the grayscale difference distribution between the die to be tested and the first reference die: Die_diff_discolor_gray_map=Die_test_discolor_gray_map−Die_ref_discolor_gray_map

[0083] Among them, Die_test_discolor_gray_map is the color difference grayscale distribution of the die to be tested, and Die_ref_discolor_gray_map is the color difference grayscale distribution of the reference die.

[0084] Similarly, the processor 10 may also determine the second grayscale difference distribution according to the grayscale difference distribution between the die to be tested and the second reference die.

[0085] After that, if Fig.12D As shown, the processor 10 can calculate the median of the grayscale distribution of all grains in each detection class to dynamically compensate for the preset defect detection threshold.

[0086] Specifically, the processor 10 may determine a dynamic compensation coefficient according to the ratio of the grayscale difference distribution between the tested die and the reference die to the median of the grayscale distribution of all the die in the corresponding test class:

[0087] Afterwards, the processor 10 may determine the defect detection threshold Threshold_adjust after dynamic compensation according to the product of the dynamic compensation coefficient and the preset defect detection threshold:

[0088] Threshold_adjust=Threshold*Die_diff_discolor_gray_map / Median_die_discolor_gray_map

[0089] Among them, Threshold is the defect detection threshold preset by the user.

[0090] So, like Fig.12D As shown, after the thresholds of various positions are adjusted according to the grayscale difference distribution map, the grayscale difference threshold map dynamically changes with the grayscale difference distribution map generated in real time during detection, thereby reducing the impact of color difference on semiconductor defect detection results.

[0091] Here, in the process of adjusting the defect detection threshold, the threshold is dynamically adjusted by using the inverse. In addition, in some embodiments, nonlinear operations such as exponential and logarithmic operations may also be used. At the same time, the defect detection threshold may not necessarily be relaxed but may be tightened, depending on the grayscale change compared with the color difference and the median die. In most cases, the threshold will be tightened if the color difference is dark, and the threshold will be relaxed if the color difference is bright, so that a balance can be maintained between the detection rate and the false detection rate.

[0092] Afterwards, the processor 10 may determine pixels whose grayscale differences in the grayscale difference distribution of each of the tested grains are greater than the defect detection threshold after dynamic compensation as detected defects.

[0093] Specifically, the processor 10 may compare the first grayscale difference of each pixel in the first grayscale difference distribution of each tested die with the defect detection threshold after dynamic compensation via the result storage module 104 to determine at least one first candidate defect.

[0094] Afterwards, the processor 10 may compare the second grayscale difference of each pixel in the second grayscale difference distribution of each to-be-tested die with the defect detection threshold after dynamic compensation to determine at least one second candidate defect.

[0095] Thereafter, the processor 10 may determine at least one pixel that is determined to be both the first candidate defect and the second candidate defect as a detected defect.

[0096] In summary, the above-mentioned defect detection method, defect detection system and computer-readable storage medium provided by the present invention can dynamically compensate for the preset defect detection threshold according to the grayscale distribution of the grain to be tested and multiple reference grains, so as to reduce the influence of color difference on the semiconductor defect detection effect, thereby improving the defect detection rate.

[0097] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0098] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A defect detection method, characterized in that: The following steps are involved: Acquire an image of a sample to be tested, wherein the sample to be tested includes a plurality of grains to be tested; Extracting the grayscale distribution of the image, and performing cluster analysis in combination with the position of each of the crystal grains, so as to classify each of the crystal grains into a plurality of detection classes; Selecting the crystal grains to be tested one by one in each of the detection categories, and taking at least one crystal grain closest to the crystal grain in the same detection category as a reference crystal grain, so as to calculate the grayscale difference distribution between each of the crystal grains to be tested and the reference crystal grain one by one; respectively calculating the median of the grayscale distribution of all the grains in each of the detection categories to dynamically compensate for a preset defect detection threshold; and Pixels whose grayscale differences in the grayscale difference distribution of each of the tested grains are greater than the defect detection threshold after dynamic compensation are determined as detected defects.

2. The defect detection method according to claim 1, characterized in that: The step of obtaining an image of the sample to be tested comprises: Fixing the sample to be tested on a moving table of a defect detection system; activating an illumination light source of the defect detection system to illuminate an inspection area via an optical module thereof; By means of the moving stage, the grains on the sample to be tested are moved to the detection area in sequence, and the reflected light provided by the sample to be tested is collected by means of an image sensor to obtain detection images of the grains respectively; and The detection images of the grains are stitched together to form an image of the sample to be tested.

3. The defect detection method according to claim 1, characterized in that: The step of extracting the grayscale distribution of the image comprises: Performing a primary convolution process on the image to extract a first grayscale distribution thereof; performing color difference correction on the image according to the first grayscale distribution; and A secondary convolution process is performed on the image after chromatic aberration correction to extract its second grayscale distribution.

4. The defect detection method according to claim 3, characterized in that: The steps of performing the primary convolution processing and / or the secondary convolution processing include: A convolution kernel whose size is an integer multiple of the grain resolution is selected to perform convolution processing on the image to extract its grayscale distribution.

5. The defect detection method according to claim 3, characterized in that: The step of performing chromatic aberration correction on the image according to the first grayscale distribution comprises: Obtain the grayscale mean value of the standard grains; determining a correction coefficient distribution according to a ratio of the first grayscale distribution to the grayscale mean; and The image is processed according to the correction coefficient distribution to perform chromatic aberration correction thereon.

6. The defect detection method according to claim 1, characterized in that: Before extracting the grayscale distribution of the image, the defect detection method further includes the following steps: Determining a region of interest according to positions of the plurality of grains to be inspected in the image; and A mask is applied to non-interest regions in the image except the interest region to prevent subsequent steps from extracting their grayscale distribution.

7. The defect detection method according to claim 1, characterized in that: The step of extracting the grayscale distribution of the image and performing cluster analysis in combination with the position of each of the crystal grains to classify each of the crystal grains into a plurality of detection classes includes: Determining the grayscale features of each of the grains according to the grayscale distribution of the image, wherein the grayscale features include at least one of a grayscale mean, a grayscale variance, and the grayscale distribution; Acquiring positional features of each of the crystal grains, wherein the positional features include positional coordinates of each of the crystal grains in the sample to be tested and / or a corresponding crystal grain group of the sample to be tested; and The cluster analysis is performed according to the grayscale characteristics and position characteristics of each of the crystal grains to classify each of the crystal grains into a plurality of the detection classes.

8. The defect detection method according to claim 1, characterized in that: The step of respectively calculating the median of the grayscale distribution of all grains in each of the detection categories to dynamically compensate for the preset defect detection threshold comprises: Determining a dynamic compensation coefficient according to the ratio of the grayscale difference distribution between the tested grain and the reference grain to the median of the grayscale distribution of all grains in the corresponding test class; and The defect detection threshold after dynamic compensation is determined according to the product of the dynamic compensation coefficient and the preset defect detection threshold.

9. The defect detection method according to claim 1, characterized in that: The step of taking at least one of the most similar grains in the same detection category as a reference grain to calculate the grayscale difference distribution between each of the grains to be detected and the reference grains one by one comprises: The two crystal grains closest to the crystal grain to be tested in the same detection category are used as reference crystal grains; Determining a first grayscale difference distribution according to the grayscale difference distribution between the to-be-tested crystal grain and the first reference crystal grain; and A second grayscale difference distribution is determined according to the grayscale difference distribution between the die to be tested and the second reference die.

10. The defect detection method according to claim 9, characterized in that: The step of determining pixels whose grayscale differences in the grayscale difference distribution of each of the tested grains are greater than the defect detection threshold after dynamic compensation as detected defects comprises: Comparing the first grayscale difference of each pixel in the first grayscale difference distribution of each of the tested grains with the defect detection threshold after dynamic compensation to determine at least one first candidate defect; Comparing the second grayscale difference of each pixel in the second grayscale difference distribution of each of the tested dies with the defect detection threshold after dynamic compensation to determine at least one second candidate defect; and At least one pixel determined as both the first candidate defect and the second candidate defect is determined as the detected defect.

11. A defect detection system, characterized in that: include: a memory having computer instructions stored thereon; as well as A processor is connected to the memory and is used to execute computer instructions stored in the memory to implement the defect detection method according to any one of claims 1 to 10.

12. The defect detection system according to claim 11, characterized in that: Also includes: A motion table, used to carry and move the sample to be tested; An illumination light source, used for providing illumination light; An optical module, used to uniformly transmit the illumination light provided by the illumination light source to the detection area of ​​the defect detection system; as well as The image sensor is used to collect the reflected light provided by the sample to be tested in the detection area to obtain a corresponding detection image.

13. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the defect detection method according to any one of claims 1 to 10 is implemented.

Images