A semiconductor chip defect detection device and method

Abstract

The present invention provides a semiconductor chip defect detection device and method, the method comprising: using a line scanning image sensor to scan the surface of a semiconductor chip to be detected to obtain a scanned image; defining the horizontal coordinate and vertical coordinate of the semiconductor chip surface by the pixel points of the scanned image; comparing the scanned image with a standard image pre-stored in a computer, and if the scanned image is different from the pre-stored standard image, the horizontal coordinate and vertical coordinate of the semiconductor chip surface are sent to the computer and stored for subsequent detection. The device comprises: a control cabinet, a computer, and a display; the computer is connected to the line scanning image sensor through a wire. The present invention has a good promoting effect on improving the detection accuracy of semiconductor equipment. The line scanning image sensor has a fast transmission speed and is suitable for assembly line detection of semiconductor defects, which is beneficial to reducing enterprise costs and improving economic benefits; the defect position is accurately found through the coordinates of the pixel points, shortening the time of semiconductor defect detection.

Description

Technical Field

[0001] The present invention relates to the field of industrial detection technology, and in particular to a semiconductor chip defect detection device and method. Background Art

[0002] At present, semiconductor devices are made by etching and wiring on semiconductor sheets. Not only silicon chips, but also common ones include gallium arsenide (gallium arsenide is toxic, so don't be curious about decomposing some inferior circuit boards), germanium and other semiconductor materials.

[0003] Prior art 1, CN202122183419.3 A semiconductor chip defect detection device with a contact protection structure, including a fixed table, two sliding bars are fixedly installed on the top of the fixed table, one end of the two sliding bars is fixedly installed with a first electric telescopic rod, the telescopic end of the first electric telescopic rod is fixedly installed with a sliding seat, the sliding seat is slidably connected to the sliding bar, one side of the first electric telescopic rod is fixedly installed with a fixed frame, and the top of the fixed frame is fixedly installed with an appearance detection mechanism. Through the telescopic push of the second electric telescopic rod on the appearance detection mechanism, it is convenient to push the back plate on the sliding seat to move, so that the appearance high-definition camera and the semiconductor chip on the electric transmission belt are horizontally parallel, according to the size of the semiconductor chip, the rotation of the driving motor drives the sliding seat on the screw rod to slide up and down, so that the appearance high-definition camera performs appearance defect detection on the semiconductor chip, thereby improving the detection efficiency of semiconductor chips.

[0004] Prior art 2, CN201310276525.3 Integrated defect detection and positioning system and method in semiconductor chip devices, embodiments relate to systems and methods for defect detection and positioning in semiconductor chips. In one embodiment, a plurality of registers are arranged in a semiconductor chip. The specific number of registers can be changed according to the desired positioning level, and the plurality of registers are geometrically distributed so that defect detection and positioning over the entire chip area or the desired chip area such as the central active area is achieved in an embodiment. In operation, the defect detection and positioning routine can run in parallel with other normal chip functions during power-on or other stages. In an embodiment, the registers can be multifunctional, because when not used for defect detection and positioning, they can be used for other operating functions of the chip, and vice versa. The embodiment thus provides fast local defect detection.

[0005] Prior art three, CN201921685912.1, an ultrasonic-based semiconductor chip defect detection system, comprising: a transmission device, on which semiconductor chips are transported; an ultrasonic detection device including an ultrasonic probe, which is arranged directly above the transmission device and transmits ultrasonic signals to the transmission device; a main control device, which is connected to the ultrasonic detection device; an alarm device, which is connected to the main control device; when the semiconductor chip is transmitted on the transmission device, the ultrasonic detection device transmits ultrasonic waves to the semiconductor chip, and when the main control device determines that there are defects in the appearance of the semiconductor chip, the alarm device sounds an alarm. The ultrasonic-based semiconductor chip defect detection system using this structure can automatically detect semiconductor chips with high accuracy and strong reliability.

[0006] Through the above analysis, the problems and defects of the existing technology are: the existing semiconductor chip defect detection equipment has low detection efficiency and cannot realize assembly line production detection. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a semiconductor chip defect detection method, comprising the following steps:

[0008] Scanning the surface of the semiconductor chip to be inspected by a line scanning image sensor to obtain a scanned image;

[0009] The horizontal and vertical coordinates of the surface of the semiconductor chip are defined by the pixels of the scanned image;

[0010] The scanned image is compared with a standard image pre-stored in the computer. If the scanned image is different from the pre-stored standard image, the horizontal and vertical coordinates of the surface of the semiconductor chip are sent to the computer and stored for subsequent testing.

[0011] Optionally, scanning the surface of the semiconductor chip to be inspected specifically includes:

[0012] The line scan image sensor captures the image of the plastic package or pin surface of the semiconductor chip to be inspected;

[0013] The image of the semiconductor chip to be inspected is subjected to multi-threshold segmentation based on the sparrow search algorithm, and the image of the semiconductor chip to be inspected is subjected to two-dimensional threshold segmentation based on the firefly algorithm, so as to obtain a defect image of the target semiconductor chip to be inspected;

[0014] Post-process the defect image of the target semiconductor chip to be inspected after segmentation, and perform defect feature selection and defect type classification tasks;

[0015] Different classification processes are performed on the defective image of the semiconductor chip to be inspected according to the classification results of the defective features.

[0016] Optionally, the clarity of the image of the semiconductor chip plastic package or pin surface to be inspected is inspected, specifically including:

[0017] The line scanning image sensor captures the same semiconductor chip plastic package or pin surface image to be inspected at different shooting speeds to obtain a plurality of first semiconductor chip plastic package or pin surface images to be inspected;

[0018] According to the relationship between the shooting speed and the clarity of the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected, respectively setting first clarity labels for the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected;

[0019] Shooting the same semiconductor chip plastic package or pin surface to be inspected by different shooting focal lengths to obtain a plurality of second semiconductor chip plastic package or pin surface images to be inspected;

[0020] According to the relationship between the shooting focal length and the clarity of the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected, second clarity labels are respectively set for the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected;

[0021] Performing image alignment processing on a plurality of first to-be-detected semiconductor chip plastic package or pin surface images and a plurality of second to-be-detected semiconductor chip plastic package or pin surface images to detect the positions of the semiconductor chip plastic package or pin surface and key points of the semiconductor chip plastic package or pin surface;

[0022] After the alignment processing, a definition detection model of the semiconductor chip plastic package or pin surface image to be detected is trained based on a plurality of first semiconductor chip plastic package or pin surface images to be detected, a plurality of second semiconductor chip plastic package or pin surface images to be detected, and corresponding definition labels;

[0023] The clarity detection model of the semiconductor chip plastic package or pin surface image to be detected is used to perform clarity detection on the semiconductor chip plastic package or pin surface image to be detected.

[0024] Optionally, key points include the overall flatness of the semiconductor chip plastic package and the coplanarity of the pin surfaces.

[0025] Optionally, the image of the semiconductor chip plastic package or pin surface to be inspected is segmented by multi-threshold segmentation based on a sparrow search algorithm, specifically including:

[0026] Read in the color image of the semiconductor chip plastic package or pin surface to be inspected and convert it into a grayscale histogram;

[0027] Set the initialization parameters of the sparrow search algorithm;

[0028] According to the reverse learning strategy, an initialization group is generated, the position vector of the individual sparrow is initialized, and the position vector is normalized so that its components are arranged in order from small to large. The position vector is a gray value combination of the image of the plastic package or pin surface of the semiconductor chip to be detected, and its value range is [0, 255], which represents the threshold vector required for the segmentation of the image of the plastic package or pin surface of the semiconductor chip to be detected;

[0029] Calculate the fitness function value of each sparrow in the sparrow, update and save the position of the maximum function value;

[0030] Calculate the inertia weight factor f = 1-iter / max iter, where iter is the current iteration number. Use roulette wheel selection. If the selected random number is greater than f, perform Lévy flight mutation on the selected sparrow individual.

[0031] Execute the sparrow search algorithm for iterative calculation;

[0032] If the number of iterations of the algorithm reaches the preset maximum number of iterations, the optimization algorithm ends. At this time, the global optimal solution obtained is the optimal threshold vector for the image segmentation of the plastic package or pin surface of the semiconductor chip to be detected. Otherwise, the fitness function value of each sparrow in the sparrow is calculated, and the position of the maximum function value is updated and saved.

[0033] According to the position of the optimal solution after iteration, the optimal threshold vector is output, and multi-threshold segmentation is performed on the plastic package or pin surface image of the semiconductor chip to be inspected.

[0034] Optionally, two-dimensional threshold segmentation is performed on the image of the semiconductor chip plastic package or pin surface to be inspected by two-dimensional threshold segmentation based on the firefly algorithm, specifically including:

[0035] Initialize the firefly algorithm parameters: initialize the number of fireflies, initial position of fireflies, initial attraction, light intensity absorption, step factor and maximum number of iterations;

[0036] Calculate the brightness of each firefly and sort them: Calculate the two-dimensional function value corresponding to each firefly, use the two-dimensional function value as its brightness, and sort them to get the position of the firefly with the highest brightness;

[0037] Determine whether the iteration is over: Determine whether the maximum number of iterations has been reached. If so, output the position and brightness of the firefly with the highest brightness. Otherwise, calculate the brightness of each firefly and sort them.

[0038] The position of the firefly with the maximum brightness and its brightness are output, and the obtained firefly position with the maximum brightness is used as a threshold to segment the semiconductor chip plastic package or pin surface image to be detected.

[0039] Optionally, the defect image of the target semiconductor chip to be inspected obtained after segmentation is post-processed, and the tasks of defect feature selection and defect type classification are performed; there are five types of defects that are easy to occur on the plastic packaging surface of semiconductor chips: scratches, nicks, holes, defects and foreign matter; there are five main types of defects that are easy to occur on the surface of semiconductor chip pins: scratches, nicks, holes, defects and melting; a mathematical morphology algorithm process is used to extract defect areas from the semiconductor chip plastic cover image; the use of four thresholds based on a generalized inverse particle swarm and gravitational search hybrid algorithm to segment the chip pin surface defect image can obtain the best segmentation result.

[0040] Optionally, the horizontal coordinate and the vertical coordinate of the surface of the semiconductor chip are defined by the pixel points of the scanned image, specifically including:

[0041] The total horizontal pixel value and the total vertical pixel value on the scanned image are measured, and the measured physical coordinates of each characteristic pixel point on the scanned image are calculated based on the total horizontal and vertical pixel values ​​and the actual horizontal and vertical physical coordinate lengths of the semiconductor chip collected in advance. The calculation formula for the measured coordinates of the characteristic pixel points of the scanned image is as follows:

[0042] Measured X coordinate value = (actual horizontal physical coordinate length of semiconductor chip * measured X coordinate pixel value) / total horizontal pixel value of scanned image

[0043] Measured Y coordinate value = (actual longitudinal physical coordinate length of semiconductor chip * measured Y coordinate pixel value) / total longitudinal pixel value of scanned image

[0044] The measured X coordinate value and the measured Y coordinate value are the measured physical coordinates of each characteristic pixel point on the scan, and the measured X coordinate pixel value and the measured Y coordinate pixel value are the pixel coordinates of each characteristic pixel point on the scan.

[0045] The semiconductor chip defect detection device provided by the present invention comprises:

[0046] control cabinets, computers and displays;

[0047] A front panel is welded to the front of the control cabinet. The left side of the control cabinet is the outlet, and the right side is the inlet. A display, an observation window, and a computer are embedded on the front panel. The computer is connected to the line scan image sensor through a wire.

[0048] The four corners of the top of the control cabinet are equipped with lifting ears through threads, the middle part of the top of the control cabinet is embedded with three exhaust fans, and the upper four corners and the middle position of the bottom of the control cabinet are equipped with moving wheels through threads.

[0049] Optionally, the conveyor belt is installed on a bottom plate parallel to the bottom of the outlet end and the inlet end, the surface of the conveyor belt is provided with a plurality of equidistant semiconductor chip placement areas, a detection platform is provided on the semiconductor chip placement areas, a line scanning image sensor is installed above the conveyor belt, the display, camera and servo motor are connected to the computer through wires, and the detection platform is connected to the display through wires.

[0050] The present invention adopts a line scanning image sensor to scan the surface of the semiconductor chip to be detected to obtain a scanned image. The line scanning speed is fast and the accuracy is high. Especially for highly precise equipment such as semiconductors, it has a good promoting effect on improving the detection accuracy of semiconductor equipment. In addition, the line scanning image sensor has a fast transmission speed and is suitable for assembly line detection of semiconductor defects, which is beneficial to reducing enterprise costs and improving economic benefits. The horizontal and vertical coordinates of the surface of the semiconductor chip are defined by the pixel points of the scanned image. The definition method is simple. The pixel points of the scanned semiconductor image are the direct manifestation features of the image. The defect position can be accurately found through the coordinates of the pixel points, which shortens the time for semiconductor defect detection. The scanned image is compared with a standard image pre-stored in a computer. The scanned image is different from the pre-stored standard image. The scanned image is compared by a computer instead of letting the computer make an autonomous judgment. As long as there is no problem with the quality of the standard image, the accuracy of the result obtained is very high. The method is simple and easy to operate.

[0051] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0052] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0054] Figure 1 This is a flow chart of a semiconductor chip defect detection method according to an embodiment of the present invention;

[0055] Figure 2 This is a flow chart of scanning the surface of a semiconductor chip to be inspected by a line scanning image sensor in an embodiment of the present invention;

[0056] Figure 3 This is a flow chart of the clarity detection of the image of the plastic package or pin surface of the semiconductor chip to be detected in an embodiment of the present invention;

[0057] Figure 4 A schematic diagram of the structure of a semiconductor chip defect detection device according to an embodiment of the present invention;

[0058] Figure 5 A schematic diagram of a semiconductor chip defect detection device according to an embodiment of the present invention;

[0059] Figure 6 This is a physical picture of the semiconductor chip defect detection equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0061] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0062] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not have to be used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0063] Example 1

[0064] like Figure 1 As shown, an embodiment of the present invention provides a semiconductor chip defect detection method, comprising the following steps:

[0065] S100: Scanning the surface of the semiconductor chip to be inspected by using a line scanning image sensor to obtain a scanned image;

[0066] S200: defining the horizontal coordinate and the vertical coordinate of the surface of the semiconductor chip by using the pixel points of the scanned image;

[0067] S300: Compare the scanned image with a standard image pre-stored in the computer. If the scanned image is different from the pre-stored standard image, the horizontal coordinate and the vertical coordinate of the surface of the semiconductor chip are sent to the computer and stored for subsequent detection.

[0068] The working principle and beneficial effects of the above technical solution are as follows: the present invention adopts a line scanning image sensor to scan the surface of the semiconductor chip to be detected to obtain a scanned image. The line scanning speed is fast and the accuracy is high. Especially for highly precise equipment such as semiconductors, it has a good promoting effect on improving the detection accuracy of semiconductor equipment. In addition, the line scanning image sensor has a fast transmission speed and is suitable for assembly line detection of semiconductor defects, which is beneficial to reducing enterprise costs and improving economic benefits; the horizontal and vertical coordinates of the surface of the semiconductor chip are defined by the pixels of the scanned image, and the definition method is simple. The pixels of the scanned semiconductor image are the direct expression features of the image. The defect position can be accurately found through the coordinates of the pixel points, shortening the time for semiconductor defect detection; the scanned image is compared with the standard image pre-stored in the computer. The scanned image is different from the pre-stored standard image. The scanned image is compared by a computer instead of letting the computer make an autonomous judgment. As long as there is no problem with the quality of the standard image, the accuracy of the result obtained is very high. The method is simple and easy to operate.

[0069] Embodiment 2:

[0070] like Figure 2 As shown, the embodiment of the present invention provides a method of using a line scanning image sensor to scan the surface of a semiconductor chip to be inspected to obtain a scanned image, which specifically includes:

[0071] S201: a line scanning image sensor captures an image of a plastic package or a pin surface of a semiconductor chip to be inspected;

[0072] S202: performing multi-threshold segmentation on the image of the semiconductor chip package or pin surface to be inspected by multi-threshold segmentation based on the sparrow search algorithm, and performing two-dimensional threshold segmentation on the image of the semiconductor chip package or pin surface to be inspected by two-dimensional threshold segmentation based on the firefly algorithm, to obtain a defect image of the target semiconductor chip to be inspected;

[0073] S203: post-processing the defect image of the target semiconductor chip to be inspected obtained after segmentation, and performing defect feature selection and defect type classification tasks;

[0074] S204: performing different classification processes on the defect image of the semiconductor chip to be inspected according to the classification result of the defect features.

[0075] The working principle and beneficial effects of the above technical scheme are as follows: the present invention starts the line scanning image sensor to collect the image of the semiconductor chip to be detected, performs two-dimensional threshold segmentation, and obtains the defect image of the target semiconductor chip to be detected; the defect image of the target semiconductor chip to be detected obtained after segmentation is post-processed, and the defect feature selection and defect type classification tasks are performed; according to the classification results of the defect features, the defect image of the semiconductor chip to be detected is subjected to different classification processing, and in the process of online defect detection of semiconductor chips, it is convenient to classify the defective semiconductor chips. The speed and classification efficiency of the defect image processing of the target semiconductor chip to be detected are improved. After locating the defect area of ​​the defect image of the segmented target semiconductor chip to be detected, only the defect connected area with the largest area that meets the conditions in the defect image of the target semiconductor chip to be detected is extracted for mask operation to form a single defect grayscale image for feature extraction and classification, that is, the defect type of this chip is determined according to the defect type with the largest area on the current chip surface.

[0076] Embodiment 3:

[0077] like Figure 3 As shown, based on Example 2, the clarity detection of the image of the plastic package or pin surface of the semiconductor chip to be detected provided by the embodiment of the present invention specifically includes:

[0078] S301: a line scanning image sensor captures images of the same semiconductor chip plastic package or pin surface to be inspected at different shooting speeds to obtain a plurality of first semiconductor chip plastic package or pin surface images to be inspected;

[0079] S302: setting first clarity labels for the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected, respectively, according to a relationship between a shooting speed and the clarity of the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected;

[0080] S303: photographing the same semiconductor chip plastic package or pin surface to be inspected at different focal lengths to obtain a plurality of second semiconductor chip plastic package or pin surface images to be inspected;

[0081] S304: setting second clarity labels for the plurality of second semiconductor chip plastic package or pin surface images to be inspected, respectively, according to a relationship between a shooting focal length and the clarity of the plurality of second semiconductor chip plastic package or pin surface images to be inspected;

[0082] S305: performing image alignment processing on the plurality of first to-be-detected semiconductor chip plastic package or pin surface images and the plurality of second to-be-detected semiconductor chip plastic package or pin surface images, detecting the positions of the semiconductor chip plastic package or pin surface and key points of the semiconductor chip plastic package or pin surface; the key points include the overall flatness of the semiconductor chip plastic package and the coplanarity of the pin surface;

[0083] S306: training a semiconductor chip plastic package or pin surface image clarity detection model to be detected based on the plurality of first semiconductor chip plastic package or pin surface images to be detected, the plurality of second semiconductor chip plastic package or pin surface images to be detected, and the corresponding clarity labels after the alignment processing;

[0084] S307: Performing clarity detection on the image of the semiconductor chip plastic package or pin surface to be detected by using the clarity detection model of the image of the semiconductor chip plastic package or pin surface to be detected.

[0085] The working principle and beneficial effects of the above technical solution are as follows: the present invention obtains a plurality of first semiconductor chip plastic package or pin face images to be detected by different shooting speeds and sets a first clarity label for each of the plurality of first semiconductor chip plastic package or pin face images to be detected, and obtains a plurality of second semiconductor chip plastic package or pin face images to be detected by different shooting focal lengths and sets a second clarity label for each of the plurality of second semiconductor chip plastic package or pin face images to be detected, and then trains a semiconductor chip plastic package or pin face image clarity detection model to be detected based on the first semiconductor chip plastic package or pin face image to be detected, the second semiconductor chip plastic package or pin face image to be detected and the corresponding clarity label. The line scanning image sensor of the present invention will not change due to environmental factors or naked eye judgment, and each semiconductor chip plastic package or pin face image to be detected is provided with a corresponding clarity label, which can effectively improve the accuracy of the trained semiconductor chip plastic package or pin face image clarity detection model in detecting the clarity of the semiconductor chip plastic package or pin face image to be detected, thereby ensuring the accuracy of the clarity result of the semiconductor chip plastic package or pin face image to be detected.

[0086] Embodiment 4:

[0087] On the basis of Example 2, the embodiment of the present invention provides a method for performing multi-threshold segmentation on the semiconductor chip plastic package or pin surface image to be detected by multi-threshold segmentation based on the sparrow search algorithm, specifically including:

[0088] Read in the color image of the semiconductor chip plastic package or pin surface to be inspected and convert it into a grayscale histogram;

[0089] Set the initialization parameters of the sparrow search algorithm;

[0090] According to the reverse learning strategy, an initialization group is generated, the position vector of the individual sparrow is initialized, and the position vector is normalized so that its components are arranged in order from small to large. The position vector is a gray value combination of the image of the plastic package or pin surface of the semiconductor chip to be detected, and its value range is [0, 255], which represents the threshold vector required for the segmentation of the image of the plastic package or pin surface of the semiconductor chip to be detected;

[0091] Calculate the fitness function value of each sparrow in the sparrow, update and save the position of the maximum function value;

[0092] Calculate the inertia weight factor f = 1-iter / max iter, where iter is the current iteration number. Use roulette wheel selection. If the selected random number is greater than f, perform Lévy flight mutation on the selected sparrow individual.

[0093] Execute the sparrow search algorithm for iterative calculation;

[0094] If the number of iterations of the algorithm reaches the preset maximum number of iterations, the optimization algorithm ends. At this time, the global optimal solution obtained is the optimal threshold vector for the image segmentation of the plastic package or pin surface of the semiconductor chip to be detected. Otherwise, the fitness function value of each sparrow in the sparrow is calculated, and the position of the maximum function value is updated and saved.

[0095] According to the position of the optimal solution after iteration, the optimal threshold vector is output, and multi-threshold segmentation is performed on the plastic package or pin surface image of the semiconductor chip to be inspected.

[0096] The working principle and beneficial effects of the above technical scheme are as follows: the present invention performs multi-threshold segmentation on the plastic package or pin surface image of the semiconductor chip to be detected based on the multi-threshold segmentation of the sparrow search algorithm, thereby improving the real-time and accuracy of the multi-threshold segmentation of the plastic package or pin surface image of the semiconductor chip to be detected, and also improves the real-time and accuracy of the two-dimensional threshold segmentation of the plastic package or pin surface image of the semiconductor chip to be detected by the two-dimensional threshold segmentation algorithm based on the grayscale-gradient co-occurrence matrix, thereby ensuring the accuracy of the classification results of the defects of the semiconductor chip to be detected.

[0097] Embodiment 5:

[0098] On the basis of Example 2, the embodiment of the present invention provides a method for performing two-dimensional threshold segmentation on the image of the semiconductor chip plastic package or pin surface to be detected by two-dimensional threshold segmentation based on the firefly algorithm, specifically including:

[0099] Initialize the firefly algorithm parameters: initialize the number of fireflies, initial position of fireflies, initial attraction, light intensity absorption, step factor and maximum number of iterations;

[0100] Calculate the brightness of each firefly and sort them: Calculate the two-dimensional function value corresponding to each firefly, use the two-dimensional function value as its brightness, and sort them to get the position of the firefly with the highest brightness;

[0101] Determine whether the iteration is over: Determine whether the maximum number of iterations has been reached. If so, output the position and brightness of the firefly with the highest brightness. Otherwise, calculate the brightness of each firefly and sort them.

[0102] The position of the firefly with the maximum brightness and its brightness are output, and the obtained firefly position with the maximum brightness is used as a threshold to segment the semiconductor chip plastic package or pin surface image to be detected.

[0103] The working principle and beneficial effects of the above technical scheme are as follows: the present invention performs two-dimensional threshold segmentation on the image of the plastic package or pin surface of the semiconductor chip to be detected through two-dimensional threshold segmentation based on the firefly algorithm, outputs the position of the firefly with the largest brightness and its brightness, and uses the obtained firefly position with the maximum brightness as the threshold to segment the image of the plastic package or pin surface of the semiconductor chip to be detected, thereby improving the speed of two-dimensional threshold segmentation of the image of the plastic package or pin surface of the semiconductor chip to be detected, that is, promoting the detection speed of defects in the semiconductor chip to be detected, which is of great help to improving enterprise production capacity and product quality.

[0104] Embodiment 6:

[0105] On the basis of Example 2, the embodiment of the present invention provides a task of post-processing the defect image of the target semiconductor chip to be inspected after segmentation, and performing defect feature selection and defect type classification; there are five main types of defects that are prone to occur on the plastic packaging surface of semiconductor chips: scratches, scratches, holes, defects and foreign matter; there are five main types of defects that are prone to occur on the surface of semiconductor chip pins: scratches, scratches, holes, defects and melting; a mathematical morphology algorithm process is used to extract the defect area of ​​the semiconductor chip plastic cover image; four thresholds are used to segment the chip pin surface defect image based on a generalized inverse particle swarm and gravitational search hybrid algorithm to obtain the best segmentation result.

[0106] Embodiment 7:

[0107] On the basis of Example 1, the embodiment of the present invention provides a method of defining the horizontal coordinate and the vertical coordinate of the surface of the semiconductor chip by using the pixel points of the scanned image, which specifically includes:

[0108] The total horizontal pixel value and the total vertical pixel value on the scanned image are measured, and the measured physical coordinates of each characteristic pixel point on the scanned image are calculated based on the total horizontal and vertical pixel values ​​and the actual horizontal and vertical physical coordinate lengths of the semiconductor chip collected in advance. The calculation formula for the measured coordinates of the characteristic pixel points of the scanned image is as follows:

[0109] Measured X coordinate value = (actual horizontal physical coordinate length of semiconductor chip * measured X coordinate pixel value) / total horizontal pixel value of scanned image

[0110] Measured Y coordinate value = (actual longitudinal physical coordinate length of semiconductor chip * measured Y coordinate pixel value) / total longitudinal pixel value of scanned image

[0111] The measured X coordinate value and the measured Y coordinate value are the measured physical coordinates of each characteristic pixel point on the scan, and the measured X coordinate pixel value and the measured Y coordinate pixel value are the pixel coordinates of each characteristic pixel point on the scan.

[0112] The working principle and beneficial effects of the above technical solution are as follows: the present invention measures the total horizontal pixel value and the total vertical pixel value on the scanned image, and calculates the measured physical coordinates of each characteristic pixel point on the scanned image according to the total horizontal and vertical pixel values ​​and the actual horizontal and vertical physical coordinate lengths of the semiconductor chip collected in advance, the measured X coordinate value and the measured Y coordinate value are the measured physical coordinates of each characteristic pixel point on the scan, the measured X coordinate pixel value and the measured Y coordinate pixel value are the pixel coordinates of each characteristic pixel point on the scan, the obtained horizontal coordinate and vertical coordinate of the surface of the semiconductor chip are accurate, the positioning accuracy is high, and the time for the computer to quickly locate the defect position is shortened.

[0113] Embodiment 8:

[0114] like Figure 4 As shown, a front panel 2 is welded to the front end of the control cabinet 1 provided in the embodiment of the present invention, the left side of the control cabinet 1 is the outlet end, and the right side is the inlet end; a display 4, an observation window 5 and a computer 6 are embedded on the front panel 2.

[0115] The four corners of the top of the control cabinet 1 are threadedly mounted with lifting ears 7, the middle part of the top of the control cabinet 1 is embedded with three exhaust fans 8, and the upper four corners and the middle position of the bottom of the control cabinet 1 are threadedly mounted with moving wheels 9.

[0116] The working principle and beneficial effects of the above technical scheme are as follows: the semiconductor chip defect detection equipment of the present invention is installed inside the control cabinet, and the semiconductor chip to be detected enters from the inlet end and reaches the conveyor belt of the semiconductor chip defect detection equipment for defect detection; a display, an observation window and a computer are embedded on the panel, and the display is used to display the comparison results of the computer; the observation window is used to realize the whole process of detection of the semiconductor chip defect detection equipment and to monitor the detection process in real time; the computer is used to control the normal operation of the semiconductor chip defect detection equipment, such as: control of the line scanning image sensor, control of the conveyor belt, control of the display, image acquisition system (online detection of surface defects of semiconductor chips), etc.

[0117] The present invention integrates the semiconductor chip defect detection equipment inside the control cabinet, and conveys the semiconductor chips by a conveyor belt, thereby realizing assembly-line detection of the semiconductor chips and improving the detection efficiency. At the same time, the semiconductor chip defect detection equipment is used to improve the detection accuracy; a display, an observation window and a computer are installed, thereby realizing the coordinate display and behavior supervision of the semiconductor chip defect detection, and providing a hardware foundation for efficiently completing the semiconductor defect detection; lifting ears and moving wheels are installed, thereby realizing different moving modes of the control cabinet, and the movement can be realized by lifting or by moving wheels, thereby providing convenience for the enterprise-based semiconductor chip production, and the production methods and links can also be adjusted; the installation of the exhaust fan realizes the timely discharge of heat in the control cabinet, thereby preventing external environmental factors from affecting the accuracy of semiconductor chip defect detection.

[0118] Embodiment 9:

[0119] like Figure 5 As shown, based on Example 9, the semiconductor chip defect detection equipment provided in this embodiment of the present invention includes: a servo motor 10, a conveyor belt 11, a semiconductor chip placement area 12, a detection table 13, a line scan image sensor 14, a camera 15 and a servo motor 16. Figure 6 This is a physical picture of the semiconductor chip defect detection equipment.

[0120] The conveyor belt 11 is installed on a bottom plate parallel to the bottom of the outlet end and the inlet end. The surface of the conveyor belt 11 is provided with a plurality of equidistant semiconductor chip placement areas 12. A detection platform 13 is provided on the semiconductor chip placement area 12. A line scanning image sensor 14 is installed above the conveyor belt 11. The display 4, the camera 15, the servo motor 16 and the line scanning image sensor 14 are connected to the computer 6 through wires, and the detection platform 16 is connected to the display 4 through wires.

[0121] The working principle and beneficial effects of the above technical solution are: use a line scanning image sensor to scan the surface of the semiconductor chip to be inspected to obtain a scanned image; define the horizontal and vertical coordinates of the semiconductor chip surface with the pixels of the scanned image; compare the scanned image with a standard image pre-stored in a computer, and if the scanned image is different from the pre-stored standard image, send the horizontal and vertical coordinates of the semiconductor chip surface to the computer and store them for subsequent inspection.

[0122] Embodiment 10:

[0123] On the basis of Example 6, the mathematical model for selecting the characteristics of the semiconductor chip plastic packaging defects provided by the embodiment of the present invention is:

[0124]

[0125] There are five types of defects in the semiconductor chip plastic packaging defect characteristics. The number of rows of semiconductor chips is 1, 2, 3, ..., m, and the number of columns is 1, 2, 3, ..., N. F represents scratches; S represents scratches; D represents holes; O represents defects; P represents foreign matter; X represents the input matrix of the mathematical model, x1, x2...x n Represents each sample point of the input matrix, y i represents the mapping relationship of the mathematical model, A is the optimal projection matrix, d is the dimension of the low-dimensional comprehensive feature parameter; Y represents the mapping of high-dimensional semiconductor chip plastic packaging defect data to low-dimensional feature information points in the neighborhood; T represents transposition.

[0126] The working principle and beneficial effects of the above technical solution are as follows: the present invention adopts a mathematical model for selecting the characteristics of semiconductor chip plastic sealing defects, and extracts characteristics of five main defect types: scratches, scratches, holes, defects and foreign matter, thereby improving the accuracy of the selection of semiconductor chip plastic sealing defect characteristics. This mathematical model is fast and efficient, and can quickly identify the characteristics of five types of semiconductor chip plastic sealing defects and transmit them to the computer in a timely manner, so that the operator can make timely adjustments and provide suggestions for subsequent production.

[0127] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A semiconductor chip defect detection method, characterized in that: The following steps are involved: Scanning the surface of the semiconductor chip to be inspected by a line scanning image sensor to obtain a scanned image; The horizontal and vertical coordinates of the surface of the semiconductor chip are defined by the pixels of the scanned image; The scanned image is compared with a standard image pre-stored in the computer. If the scanned image is different from the pre-stored standard image, the horizontal and vertical coordinates of the surface of the semiconductor chip are sent to the computer and stored for subsequent testing; Scanning the surface of the semiconductor chip to be inspected includes: The line scan image sensor captures the image of the plastic package or pin surface of the semiconductor chip to be inspected; The image of the semiconductor chip to be inspected is subjected to multi-threshold segmentation based on the sparrow search algorithm, and the image of the semiconductor chip to be inspected is subjected to two-dimensional threshold segmentation based on the firefly algorithm, so as to obtain a defect image of the target semiconductor chip to be inspected; Post-process the defect image of the target semiconductor chip to be inspected after segmentation, and perform defect feature selection and defect type classification tasks; Performing different classification processing on the defect image of the semiconductor chip to be inspected according to the classification result of the defect feature; The clarity inspection of the semiconductor chip plastic package or pin surface image to be inspected includes: The line scanning image sensor captures the same semiconductor chip plastic package or pin surface image to be inspected at different shooting speeds to obtain a plurality of first semiconductor chip plastic package or pin surface images to be inspected; According to the relationship between the shooting speed and the clarity of the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected, respectively setting first clarity labels for the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected; Shooting the same semiconductor chip plastic package or pin surface to be inspected by different shooting focal lengths to obtain a plurality of second semiconductor chip plastic package or pin surface images to be inspected; According to the relationship between the shooting focal length and the clarity of the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected, second clarity labels are respectively set for the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected; Performing image alignment processing on a plurality of first semiconductor chip plastic package or pin surface images to be detected and a plurality of second semiconductor chip plastic package or pin surface images to be detected, and detecting the positions of the semiconductor chip plastic package or pin surface and key points of the semiconductor chip plastic package or pin surface; After the alignment processing, a semiconductor chip plastic package or pin surface image clarity detection model to be detected is trained based on a plurality of first semiconductor chip plastic package or pin surface images to be detected, a plurality of second semiconductor chip plastic package or pin surface images to be detected, and corresponding clarity labels; The clarity detection model of the semiconductor chip plastic package or pin surface image to be detected is used to perform clarity detection on the semiconductor chip plastic package or pin surface image to be detected.

2. The semiconductor chip defect detection method according to claim 1, characterized in that: The key points include the overall flatness of the semiconductor chip plastic package and the coplanarity of the pin surfaces.

3. The semiconductor chip defect detection method according to claim 1, characterized in that: The multi-threshold segmentation of the semiconductor chip plastic package or pin surface image to be inspected is performed by multi-threshold segmentation based on the sparrow search algorithm, specifically including: Read in the color image of the semiconductor chip plastic package or pin surface to be inspected and convert it into a grayscale histogram; Set the initialization parameters of the sparrow search algorithm; According to the reverse learning strategy, an initialization group is generated, the position vector of the individual sparrow is initialized, and the position vector is normalized so that its components are arranged in order from small to large. The position vector is a gray value combination of the image of the plastic package or pin surface of the semiconductor chip to be detected, and its value range is [0, 255], which represents the threshold vector required for the segmentation of the image of the plastic package or pin surface of the semiconductor chip to be detected; Calculate the fitness function value of each sparrow in the sparrow, update and save the position of the maximum function value; Calculate the inertia weight factor f = 1-iter / max iter, where iter is the current iteration number. Use roulette wheel selection. If the selected random number is greater than f, perform Lévy flight mutation on the selected sparrow individual. Execute the sparrow search algorithm for iterative calculation; If the number of iterations of the algorithm reaches the preset maximum number of iterations, the optimization algorithm ends. At this time, the global optimal solution obtained is the optimal threshold vector for the image segmentation of the plastic package or pin surface of the semiconductor chip to be detected. Otherwise, the fitness function value of each sparrow in the sparrow is calculated, and the position of the maximum function value is updated and saved. According to the position of the optimal solution after iteration, the optimal threshold vector is output, and multi-threshold segmentation is performed on the plastic package or pin surface image of the semiconductor chip to be inspected.

4. The semiconductor chip defect detection method according to claim 1, characterized in that: The two-dimensional threshold segmentation based on the firefly algorithm is used to perform two-dimensional threshold segmentation on the semiconductor chip plastic package or pin surface image to be inspected, specifically including: Initialize the firefly algorithm parameters: initialize the number of fireflies, initial position of fireflies, initial attraction, light intensity absorption, step factor and maximum number of iterations; Calculate the brightness of each firefly and sort them: Calculate the two-dimensional function value corresponding to each firefly, use the two-dimensional function value as its brightness, and sort them to get the position of the firefly with the highest brightness; Determine whether the iteration is over: Determine whether the maximum number of iterations has been reached. If so, output the position and brightness of the firefly with the highest brightness. Otherwise, calculate the brightness of each firefly and sort them. The position of the firefly with the maximum brightness and its brightness are output, and the obtained firefly position with the maximum brightness is used as a threshold to segment the semiconductor chip plastic package or pin surface image to be detected.

5. The semiconductor chip defect detection method according to claim 1, characterized in that: The defect image of the target semiconductor chip to be inspected after segmentation is post-processed, and the tasks of defect feature selection and defect type classification are performed; there are five types of defects that are prone to occur on the plastic packaging surface of semiconductor chips: scratches, scratches, holes, defects and foreign matter; there are five main types of defects that are prone to occur on the surface of semiconductor chip pins: scratches, scratches, holes, defects and melting; the mathematical morphology algorithm process is used to extract the defect area of ​​the semiconductor chip plastic cover image; the four-threshold method is used to segment the chip pin surface defect image based on the generalized inverse particle swarm and gravitational search hybrid algorithm to obtain the best segmentation result.

6. The semiconductor chip defect detection method according to claim 1, wherein: The horizontal and vertical coordinates of the surface of the semiconductor chip are defined by the pixels of the scanned image, including: The total horizontal pixel value and the total vertical pixel value on the scanned image are measured, and the measured physical coordinates of each characteristic pixel point on the scanned image are calculated based on the total horizontal and vertical pixel values ​​and the actual horizontal and vertical physical coordinate lengths of the semiconductor chip collected in advance. The calculation formula for the measured coordinates of the characteristic pixel points of the scanned image is as follows: Measured X coordinate value = (actual horizontal physical coordinate length of semiconductor chip * measured X coordinate pixel value) / total horizontal pixel value of scanned image Measured Y coordinate value = (actual longitudinal physical coordinate length of semiconductor chip * measured Y coordinate pixel value) / total longitudinal pixel value of scanned image The measured X coordinate value and the measured Y coordinate value are the measured physical coordinates of each characteristic pixel point on the scan, and the measured X coordinate pixel value and the measured Y coordinate pixel value are the pixel coordinates of each characteristic pixel point on the scan.

7. A semiconductor chip defect detection device, characterized in that: include: control cabinets, computers and displays; A front panel is welded at the front end of the control cabinet. The left side of the control cabinet is the outlet end, and the right side is the inlet end. A display, an observation window and a computer are embedded on the front panel. The computer is connected to the line scanning image sensor through a wire. The four corners of the top of the control cabinet are installed with lifting ears through threads, the middle part of the top of the control cabinet is embedded with three exhaust fans, and the upper four corners and the middle position of the bottom of the control cabinet are installed with moving wheels through threads; The surface of the semiconductor chip to be inspected is scanned by a line scanning image sensor to obtain a scanned image, and the clarity of the plastic package or pin surface image of the semiconductor chip to be inspected is inspected, specifically including: The line scanning image sensor captures the same semiconductor chip plastic package or pin surface image to be inspected at different shooting speeds to obtain a plurality of first semiconductor chip plastic package or pin surface images to be inspected; According to the relationship between the shooting speed and the clarity of the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected, respectively setting first clarity labels for the plurality of first images of the semiconductor chip plastic package or the pin surface to be inspected; Shooting the same semiconductor chip plastic package or pin surface to be inspected by different shooting focal lengths to obtain a plurality of second semiconductor chip plastic package or pin surface images to be inspected; According to the relationship between the shooting focal length and the clarity of the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected, second clarity labels are respectively set for the plurality of second images of the semiconductor chip plastic package or the pin surface to be inspected; Performing image alignment processing on a plurality of first semiconductor chip plastic package or pin surface images to be detected and a plurality of second semiconductor chip plastic package or pin surface images to be detected, and detecting the positions of the semiconductor chip plastic package or pin surface and key points of the semiconductor chip plastic package or pin surface; After the alignment processing, a semiconductor chip plastic package or pin surface image clarity detection model to be detected is trained based on a plurality of first semiconductor chip plastic package or pin surface images to be detected, a plurality of second semiconductor chip plastic package or pin surface images to be detected, and corresponding clarity labels; The clarity detection model of the semiconductor chip plastic package or pin surface image to be detected is used to perform clarity detection on the semiconductor chip plastic package or pin surface image to be detected.

8. The semiconductor chip defect detection device according to claim 7, characterized in that: The conveyor belt is installed on a bottom plate parallel to the bottom of the outlet end and the inlet end. The surface of the conveyor belt is provided with a plurality of equidistant semiconductor chip placement areas. A detection platform is provided on the semiconductor chip placement area. A line scanning image sensor is installed above the conveyor belt. The display, camera and servo motor are connected to the computer through wires, and the detection platform is connected to the display through wires.

Images