Intelligent detection system and method for subfissure and fragment of silicon wafer

Abstract

The invention provides an intelligent detection system and method for hidden cracks and fragments of a silicon wafer, and the system comprises an imaging unit, an image processing unit and a control and communication unit, and is used for obtaining a surface image of the silicon wafer located on a conveying belt; the image processing unit is in communication connection with the imaging unit and is used for receiving and processing the surface image, the image processing unit comprises a processor, and the processor is configured to automatically generate an effective detection area excluding an edge area and an electrode area in the surface image based on pre-stored silicon wafer size information and electrode distribution information; the processor is used for segmenting the image in the effective detection area based on a gray threshold range determined through statistical learning, extracting a suspected defect area, performing feature extraction on the suspected defect area, and performing defect classification and judgment according to preset defect feature parameters; and the control and communication unit is connected with the image processing unit, so that efficient and accurate subfissure and fragment detection is realized on the premise of not changing the structure of a production line.

Description

TECHNICAL FIELD

[0001] The present application relates to an intelligent detection system and method for silicon wafer cracks and fragments. BACKGROUND

[0002] With the rapid development of the photovoltaic industry, the manufacturing process of solar cells has increasingly high requirements for the quality of silicon wafers. The surface integrity of silicon wafers, as the core material of the cells, directly affects the photoelectric conversion efficiency and long-term reliability of the cells. In particular, in the processes of cleaning, cutting, diffusion, deposition, lithography, and etching, cracks and fragments are prone to occur or expand. Therefore, efficient and accurate defect detection technology has become a key requirement of the industry.

[0003] Currently, the mainstream method for detecting silicon wafer cracks is based on machine vision. Cameras and light sources are arranged on both sides of the conveyor belt, and background plates or light barriers are used for image acquisition. However, this method has the following technical limitations: the installation structure is complex, the production line conveyor belt needs to be cut or a notch needs to be reserved, which requires a large amount of modification work and affects the continuous production of the production line; the detection area is determined by human experience and is easily affected by the edge area of the silicon wafer and the electrode structure, leading to false positives and false negatives; traditional image processing algorithms have poor adaptability to gray scale changes and are difficult to handle the problem of inconsistent gray scales on the surface of the cell due to process differences, resulting in insufficient detection accuracy for small size cracks, edge collapse, and fragments, and generally low detection rate.

[0004] To address the problems of complex installation structure, low detection accuracy, and poor adaptability in the prior art, there is an urgent need for a system and method that can efficiently and accurately detect cracks and fragments without modifying the production line structure. SUMMARY

[0005] The purpose of the present application is to solve the above problems in the prior art and provide an intelligent detection system and method for silicon wafer cracks and fragments that can efficiently and accurately detect cracks and fragments without modifying the production line structure. This purpose is achieved by: The application provides an intelligent detection system for silicon wafer cracks and fragments, which comprises an imaging unit, an image processing unit and a control and communication unit, the imaging unit is vertically installed above a conveying belt, used to obtain the surface image of a silicon wafer on the conveying belt, the imaging unit comprises at least one industrial camera and a light source providing illumination for the industrial camera; the image processing unit is in communication connection with the imaging unit, used to receive and process the surface image, the image processing unit comprises a processor, the processor is configured to automatically generate an effective detection area excluding the edge area and the electrode area in the surface image based on the pre-stored silicon wafer size information and electrode distribution information, the processor performs segmentation on the image in the effective detection area based on the gray threshold range determined through statistical learning, extracts a suspected defect area, performs feature extraction on the suspected defect area, and performs defect classification and judgment according to the pre-set defect feature parameters; the control and communication unit is connected with the image processing unit, used to control the image acquisition operation of the imaging unit, and communicates the detection results with an external system.

[0006] Further, the imaging unit is fixedly installed through a support, the optical axis of the industrial camera is perpendicular to the plane of the conveying belt, and the field of view of the imaging unit completely covers the passing range of the silicon wafer on the conveying belt, and the light source is a coaxial diffuse light source or a dome light source.

[0007] Further, the automatic generation of the effective detection area excluding the edge area and the electrode area in the surface image based on the pre-stored silicon wafer size information and electrode distribution information comprises: positioning the position and angle of the silicon wafer in the surface image through an edge extraction algorithm; generating a preliminary detection area by inwardly shrinking a safety distance according to the pre-stored silicon wafer size data; and generating the effective detection area by excluding the area where the electrodes are located from the preliminary detection area according to the pre-stored electrode distribution template.

[0008] Further, the safety distance is determined according to the positioning error of the silicon wafer.

[0009] Further, the gray threshold range is obtained by statistical analysis of the pixel gray values in the effective detection area of a plurality of good silicon wafer images, and the upper limit and the lower limit of the gray threshold range satisfy the following formula: wherein μ is the mean value of the pixel gray values in the effective detection area of the good silicon wafer image, σ is the standard deviation, and n is a coefficient determined according to experiments.

[0010] Further, after the feature extraction on the suspected defect area, the method further comprises: performing a morphological closing operation on the extracted suspected defect area to connect adjacent defect points to form a connected area.

[0011] Further, the defect feature parameters include one or more of area, perimeter, circumscribed rectangle length, circumscribed rectangle width, aspect ratio and circularity.

[0012] Further, the intelligent detection system for wafer cracks and fragments uses an intelligent detection method for wafer cracks and fragments, comprising the steps of: acquiring a surface image of a wafer located on a conveying belt through an imaging unit vertically installed above the conveying belt; automatically generating an effective detection area excluding edge regions and electrode regions in the surface image based on pre-stored wafer size information and electrode distribution information; segmenting the image in the effective detection area based on a gray threshold range determined through statistical learning, and extracting a suspected defect region; performing feature extraction on the suspected defect region, and performing defect classification and judgment according to pre-set defect feature parameters; and outputting a detection result.

[0013] Further, the gray threshold range is obtained by statistical analysis of pixel gray values in the effective detection area of multiple good wafer images.

[0014] Compared with the prior art, the present application has the beneficial effects of simplifying the installation structure, realizing rapid and lossless production line deployment, and being more convenient than the traditional side illumination scheme without the need for cutting and modifying the conveying belt. The effective detection area is automatically and accurately delineated through a software algorithm, eliminating the subjectivity and errors of manual ROI delineation and the interference of edge and electrode structures on the detection algorithm, and the detection accuracy is higher. The segmentation threshold is dynamically determined based on statistical learning of good wafer gray distribution, making the algorithm highly adaptable to process differences, light fluctuations and slight contamination on the wafer surface, and being able to stably and reliably detect tiny cracks and fragments, significantly improving the detection rate and detection accuracy, and meeting the stringent requirements of high-end photovoltaic manufacturing for quality control. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a flowchart of an intelligent detection method for wafer cracks and fragments; Figure 2 is a structural schematic diagram of an intelligent detection system for wafer cracks and fragments; In the figure: 1, light source, 2, industrial camera, 3, wafer. DETAILED DESCRIPTION

[0016] In order to deepen the understanding of the present application, the present application will be further described in conjunction with the embodiments and the accompanying drawings, which are only used to explain the present application and do not constitute a limitation on the protection scope of the present application. EMBODIMENT

[0017] Please refer to Figure 1 This invention proposes an intelligent detection system for microcracks and fragments on silicon wafers, comprising an imaging unit, an image processing unit, and a control and communication unit. The imaging unit is vertically mounted above a conveyor belt to acquire surface images of silicon wafers 3 located on the conveyor belt. The imaging unit includes at least one industrial camera 2 and a light source 1 providing illumination for it. The image processing unit is communicatively connected to the imaging unit and is used to receive and process the surface images. The image processing unit includes a processor configured to automatically generate an effective detection area in the surface image, excluding edge and electrode areas, based on pre-stored silicon wafer size and electrode distribution information. The processor segments the image within the effective detection area based on a grayscale threshold range determined through statistical learning, extracts suspected defect areas, performs feature extraction on the suspected defect areas, and classifies and judges defects according to preset defect feature parameters. The control and communication unit is connected to the image processing unit to control the image acquisition operation of the imaging unit and to communicate the detection results with an external system. Example

[0018] Unlike Embodiment 1, this embodiment achieves better results by specifying each step. The imaging unit is fixedly installed by a bracket, the optical axis of the industrial camera 2 is perpendicular to the plane of the conveyor belt, and the field of view of the imaging unit completely covers the passage range of the silicon wafer 3 on the conveyor belt. The light source 1 is a coaxial diffuse light source or a dome light source. Based on the pre-stored silicon wafer size information and electrode distribution information, an effective detection area excluding edge areas and electrode areas is automatically generated in the surface image. This includes: locating the position and angle of the silicon wafer 3 in the surface image using an edge extraction algorithm; generating a preliminary detection area by shrinking inward by a safe distance according to the pre-stored silicon wafer size data; subtracting the electrode area from the preliminary detection area according to the pre-stored electrode distribution template to generate an effective detection area; the safe distance is determined based on the silicon wafer positioning error; the grayscale threshold range is obtained by statistically analyzing the pixel grayscale values ​​in the effective detection area of ​​multiple images of good silicon wafers, and the upper limit of the grayscale threshold range is... and lower limit The calculation satisfies: Where μ is the mean of pixel grayscale values ​​within the effective detection area of ​​the good silicon wafer image, σ is its standard deviation, and n is a coefficient determined by experiments; after feature extraction of the suspected defect area, the method further includes: performing morphological closing operation on the extracted suspected defect area to connect adjacent defect points to form a connected region; the defect feature parameters include one or more of the following: area, perimeter, length of the circumscribed rectangle, width of the circumscribed rectangle, aspect ratio, and roundness.

[0019] Specifically, the initially extracted defect points are often discrete. To form a complete defect region, the software performs a morphological closing operation on the binary image B(x,y). First, a 3×3 pixel rectangular structuring element is used for dilation to connect adjacent defect points; then, the same structuring element is used for erosion to restore the approximate shape of the defect region. This operation can effectively connect tiny, discrete hidden cracks into lines or fill small holes inside fragmented regions. Afterward, the software uses a connected component analysis algorithm to mark all connected white regions, referred to as blobs. For each blob, a series of morphological feature parameters are calculated, including: Area A: The total number of pixels contained in the Blob; Circulating rectangle: The smallest rectangle that completely encloses the Blob, its length is denoted as A. Width is denoted as Aspect Ratio Perimeter P: The pixel perimeter of the blob outline; Circularity C is used to describe how close the shape of a blob is to a circle; for example, C=1 when it is close to a perfect circle.

[0020] The software has a built-in defect feature parameter library and decision rule library. It logically compares the feature value calculated for each blob with the preset rules to determine the defect type. For example: if A < 36, that is, the area is less than 0.09 mm², corresponding to 0.3 mm × 0.3 mm, it is judged as noise and ignored; if A >= 36 and R > 4 and A < 500, it is judged as a hidden crack, because a long and narrow shape is a typical feature of a hidden crack; if A >= 36 and C < 0.3 and A > 100, it is judged as a fragment, because the shape is irregular and the area is large; if the blob intersects with the boundary of the effective detection area, it is judged as a chipped edge or missing corner. Multiple defects may exist on a silicon wafer 3 at the same time. The software will combine the decision results of all blobs and make a final overall decision on the silicon wafer 3. The decision result includes qualified or unqualified. Example

[0021] The intelligent detection system for microcracks and fragments in silicon wafers uses an intelligent detection method for microcracks and fragments in silicon wafers, including the following steps: acquiring a surface image of a silicon wafer 3 located on the conveyor belt through an imaging unit vertically mounted directly above the conveyor belt; automatically generating an effective detection area in the surface image, excluding edge and electrode areas, based on pre-stored silicon wafer size information and electrode distribution information; segmenting the image within the effective detection area based on a grayscale threshold range determined through statistical learning, and extracting suspected defect areas; extracting features from the suspected defect areas, and classifying and judging defects according to preset defect feature parameters; outputting the detection results; the grayscale threshold range is obtained by statistically analyzing the pixel grayscale values ​​within the effective detection area in multiple images of good silicon wafers.

[0022] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent detection system for microcracks and fragments in silicon wafers, characterized in that, The system includes an imaging unit, an image processing unit, and a control and communication unit. The imaging unit is vertically mounted above a conveyor belt and is used to acquire surface images of silicon wafers located on the conveyor belt. The imaging unit includes at least one industrial camera and a light source providing illumination for it. The image processing unit is communicatively connected to the imaging unit and is used to receive and process the surface images. The image processing unit includes a processor configured to automatically generate an effective detection region in the surface image, excluding edge and electrode regions, based on pre-stored silicon wafer size and electrode distribution information. The processor segments the image within the effective detection region based on a grayscale threshold range determined through statistical learning, extracts suspected defect regions, extracts features from the suspected defect regions, and classifies and determines defects according to preset defect feature parameters. The control and communication unit is connected to the image processing unit and is used to control the image acquisition operation of the imaging unit and communicate the detection results with an external system.

2. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 1, characterized in that, The imaging unit is fixedly installed by a bracket. The optical axis of the industrial camera is perpendicular to the plane of the conveyor belt, and the field of view of the imaging unit completely covers the passage range of the silicon wafer on the conveyor belt. The light source is a coaxial diffuse light source or a dome light source.

3. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 2, characterized in that, The method of automatically generating an effective detection area that excludes edge regions and electrode regions in the surface image based on pre-stored silicon wafer size information and electrode distribution information includes: locating the position and angle of the silicon wafer in the surface image using an edge extraction algorithm; generating a preliminary detection area by shrinking inward by a safe distance according to pre-stored silicon wafer size data; and generating the effective detection area by subtracting the electrode region from the preliminary detection area according to a pre-stored electrode distribution template.

4. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 3, characterized in that, The safety distance is determined based on the silicon wafer positioning error.

5. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 4, characterized in that, The grayscale threshold range is obtained by statistical analysis of pixel grayscale values ​​within the effective detection area in multiple images of good quality silicon wafers. The upper limit of the grayscale threshold range is... and lower limit The calculation satisfies: , where μ is the mean of pixel grayscale values ​​within the effective detection area of ​​the good silicon wafer image, σ is its standard deviation, and n is a coefficient determined experimentally.

6. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 5, characterized in that, After extracting features from the suspected defect region, the method further includes performing a morphological closing operation on the extracted suspected defect region to connect adjacent defect points to form a connected region.

7. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 6, characterized in that, The defect feature parameters include one or more of the following: area, perimeter, length of circumscribed rectangle, width of circumscribed rectangle, aspect ratio, and roundness.

8. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 7, characterized in that, The intelligent detection system for silicon wafer microcracks and fragments uses an intelligent detection method for silicon wafer microcracks and fragments, including the following steps: acquiring a surface image of a silicon wafer located on the conveyor belt through an imaging unit vertically mounted directly above the conveyor belt; automatically generating an effective detection area excluding edge and electrode areas in the surface image based on pre-stored silicon wafer size information and electrode distribution information; segmenting the image within the effective detection area based on a grayscale threshold range determined through statistical learning to extract suspected defect areas; extracting features from the suspected defect areas and classifying and judging defects according to preset defect feature parameters; and outputting the detection results.

9. The intelligent detection system for microcracks and fragments in silicon wafers according to claim 8, characterized in that, The grayscale threshold range is obtained by statistical analysis of the pixel grayscale values ​​in the effective detection area of ​​multiple images of good quality silicon wafers.

Images