Wafer alignment method and system based on searching of right-angle side of wafer

Abstract

The invention belongs to the technical field of semiconductors, and discloses a wafer alignment method and system based on searching of a right-angle side of a wafer, and the method comprises the steps: obtaining an image: carrying out the image shooting of a wafer disposed on a scribing machine workbench through an industrial camera; preprocessing the image; constructing a template; calculating the posture of the wafer, in the extracted wafer image, detecting a straight line segment in the wafer image by using a Hough transform algorithm, and determining the X and Y coordinates and the rotation angle theta of the current wafer on the workbench so as to determine the direction and the position of the wafer; calculating an alignment error, straightening and aligning the wafer, and verifying the position; according to the scheme, alignment is carried out by directly searching and utilizing the right-angle side characteristics of the wafer, so that the image processing flow is reduced, calculation and image data processing required by multi-scale template matching and a Hough transform algorithm in an existing wafer alignment method are avoided, and the data processing time is shortened; the problem that an existing wafer alignment method is long in processing time is solved.

Description

Technical Field

[0001] The present invention belongs to the field of semiconductor technology, and specifically relates to a wafer alignment method and system based on finding the right-angle edge of the wafer. Background Art

[0002] The wafer alignment method is to use the machine vision system to pre-process the wafer image, perform image processing algorithms such as edge extraction and template matching, calculate the inclination angle between the wafer cutting path and the machine tool motion axis, and control the machine tool's rotation axis to make corresponding rotation adjustments, so that the wafer cutting path is parallel to the direction of the machine tool motion axis, thereby improving the accuracy and efficiency of wafer cutting.

[0003] Referring to the document with the existing publication (announcement) number CN118824901A, a semiconductor wafer position detection device and detection method are disclosed. The detection method calculates a first straightening angle by performing multi-scale template matching based on edge geometric features on a first low-magnification wafer image. The weighted calculation of the multi-scale features enhances the algorithm's adaptability to image changes; the wafer position is automatically preliminarily adjusted according to the first straightening angle; then a first high-magnification wafer image and a second high-magnification wafer image are obtained, and an improved Hough transform algorithm is used to calculate a second straightening angle based on the two high-magnification wafer images to ensure more precise detection of the semiconductor wafer position; and the wafer position is automatically and accurately adjusted according to the second straightening angle.

[0004] Although the above-mentioned wafer detection method introduces multi-scale feature fusion in the correlation measurement, it can capture subtle changes in the edge of the wafer more accurately by comprehensively considering the features of multiple scales. However, this method adopts a two-stage detection, that is, preliminary detection of low-magnification images is performed first, and then fine detection of high-magnification images is performed. Although this method improves the accuracy of detection, it also requires the collection and processing of more image data, which increases the data processing time. Furthermore, the above-mentioned method mainly relies on multi-scale template matching and improved Hough transform algorithm to detect the wafer position, but does not directly use the feature of the right-angle edge of the wafer. The positioning information of the right-angle edge and the edge vertex is used, which makes the data processing time longer. Summary of the invention

[0005] The purpose of this solution is to provide a wafer alignment method based on finding the right-angle edge of the wafer to solve the problem of long processing time of the existing wafer alignment method.

[0006] In order to achieve the above object, the present invention provides a wafer alignment method based on finding the right-angle edge of the wafer, comprising the following steps: Step S10: image acquisition, using an industrial camera to capture an image of the wafer placed on the workbench of the dicing machine to obtain an image of the wafer; Step S20: image preprocessing, performing grayscale transformation, filtering, threshold segmentation and edge extraction on the collected wafer image; Step S30: constructing a template, according to the known shape and size of the wafer, constructing a wafer template including the right-angle edge features of the wafer, the template is used for subsequent image comparison and error calculation; Step S40: Calculate the wafer posture. In the wafer image extracted in step S20, use the Hough transform algorithm to detect the straight line segments in the wafer image, determine the X, Y coordinates and the rotation angle θ of the current wafer on the workbench, so as to determine the direction and position of the wafer. Step S50: calculating the alignment error, comparing the wafer image determined in step S40 with the wafer template constructed in step S30, and calculating the size and direction of the alignment error of the wafer in the X and Y directions and the rotation angle θ through comparison and analysis; Step S60: the wafer is aligned and positioned. According to the wafer posture calculated in step S50 and the existing alignment error, the position and posture of the wafer on the workbench are adjusted by controlling the X, Y, and C axis movements of the dicing machine to make the wafer reach a predetermined position and posture. Step S70: Verify the position. After the wafer is aligned, use the industrial camera to capture the wafer image again, and compare the newly captured image with the wafer template for a second time. If the comparison result shows that there is an error and the error exceeds a preset threshold, fine-tune according to the feedback information until the deviation between the actual placement position of the wafer and the wafer template is less than or equal to the preset threshold. Finally, start the grinding wheel dicing machine to cut the wafer.

[0007] Furthermore, in step S20, the filtering process specifically includes the following steps: Step S21: using a sliding window of size N×N and determining it according to the noise condition of the wafer image and the desired filtering effect; Step S22: using the upper left corner of the wafer image obtained in step S10 as the starting scanning position, and determining the scanning direction and step length according to the size and dimension of the wafer image, using a row-by-row and column-by-column scanning method; Step S23: placing the sliding window at the starting scanning position of the wafer image and ensuring that the window completely covers the pixels in the area, reading the grayscale values ​​of all pixels in the sliding window and storing them in an array; Step S24: sort the grayscale values ​​of the pixels in the array in ascending order, find the middle value of the sorted array, and use it as the median value of the current scanning area, and then replace the grayscale value of the center point of the sliding window with the calculated median value; Step S25: according to the set scanning direction and step length, move the sliding window to the next scanning position until the sliding window traverses the entire wafer image.

[0008] Furthermore, in step S20, the threshold segmentation specifically includes the following steps: Step S26: traversing each pixel point in the wafer image pre-processed in step S10; Step S27: Compare the grayscale value of each pixel with a set threshold; if the grayscale value of the pixel is greater than the threshold, it is set to 255, indicating a wafer feature; if the grayscale value of the pixel is less than or equal to the threshold, it is set to 0, indicating a background area; Step S28: After traversing the entire wafer image, the original wafer binary image containing feature information is finally obtained, wherein the features on the wafer are represented by white areas and the background is represented by black areas; The edge extraction in step S20 specifically includes the following steps: Step S29: performing Gaussian smoothing on the wafer image, and using the Sobel operator to calculate the gradient amplitude and direction of each pixel in the image, and identifying the edge points of the image; Step S210: only retain the points with the largest amplitude in the gradient direction, and remove the false edges in edge detection; Step S211: set a high threshold and a low threshold, the high threshold is used to determine a strong edge, the low threshold is used to connect the edge, then use the high threshold to obtain an edge image, and use the low threshold to connect the broken parts in the edge; finally, obtain a binary image containing wafer edge information.

[0009] Furthermore, the template construction in step S30 specifically includes the following steps: Step S31: select a standard wafer of known shape and size as a reference, and ensure that the right-angle edge of the standard wafer is parallel to the movement axis direction of the dicing machine; then use an industrial camera to shoot the standard wafer, and ensure that the image is clear and contains the complete wafer shape; Step S32: converting the standard wafer image from a color image to a grayscale image, and performing filtering, threshold segmentation and edge extraction on the grayscale image; then using a corner detection algorithm and a Hough transform method to detect feature points on the edge contour of the standard wafer image, the feature points including right-angle vertices and right-angle edges, and marking and recording the detected feature points, including the coordinate position of the feature points in the image and the wafer edge information of the image; Step S32: Generate a standard template image of the wafer based on the extracted feature point information, and the template image only contains the edge contour and feature point marks of the wafer, and does not contain any background information; then save the generated template image and subsequently match it with the wafer image to be processed.

[0010] Further, in the step S40, according to the result of matching the wafer image with the template image, the centroid coordinates of the matching area and the deviation angle from the template image are extracted, and the deviation angle is regarded as the alignment angle of the wafer; if the deviation angle is positive, it means that the wafer needs to be rotated clockwise to be aligned, and if the deviation angle is negative, it means that the wafer needs to be rotated counterclockwise to be aligned; in the step S50, the following steps are specifically included: Step S51: determining an ideal position of the grinding wheel cutting focus on the wafer according to a preset cutting path and cutting road width of the dicing machine; Step S52: driving the grinding wheel to move to a preset cutting position, and photographing the actual position of the cutting focus of the grinding wheel on the wafer at this time by an industrial camera; Step S53: Compare the ideal position with the actual position, and calculate the translation error between the two in the X and Y directions and the rotation error around the Z axis; wherein the translation error is the coordinate difference between the ideal position and the actual position in the X and Y directions, and the rotation error is the angle between the actual cutting path direction and the ideal cutting path direction; Step S54: according to the calculated translation error and rotation error, drive the wafer to move and perform error compensation adjustment; for the translation error, compensate by adjusting the motion parameters of the X and Y axes; for the rotation error, compensate by adjusting the angle of the C axis; Step S55: After the error is compensated, the industrial camera is used again to photograph the actual position of the grinding wheel to verify the compensation effect; if the error between the actual position after compensation and the ideal position is within the threshold range, the cutting step is performed; if the error between the actual position after compensation and the ideal position is not within the threshold range, the above steps are repeated until the error value is within the threshold range.

[0011] The principle and effect of this solution are as follows: This solution reduces the image processing process by directly finding and using the right-angle edge features of the wafer for alignment, avoiding the calculation and image data processing required by the multi-scale template matching and Hough transform algorithms in the existing wafer alignment methods, and reducing the data processing time. At the same time, the geometric features of the right-angle edges of the wafer image are used to improve the alignment accuracy, so that the wafer can be quickly and accurately aligned to the predetermined position and posture.

[0012] A wafer alignment system based on finding the right-angle edge of a wafer, comprising: A collection unit, the collection unit is used to collect image information of the wafer to be cut; A drive unit, which is used to control the X, Y, and C axis movements of the dicing machine and adjust the position and posture of the wafer on the workbench; A cutting unit, wherein the cutting unit is used to cut the wafer; A measuring unit, the measuring unit comprising a laser transmitter, the laser transmitter being used to measure a position of a cutting focus of a cutting unit on the wafer; A data processing unit is used to receive and identify the image information and cutting focus of the wafer, then compare the image of the wafer with the wafer template, calculate the wafer posture and alignment error, control the X, Y, and C axis movement of the dicing machine, adjust the position of the wafer, and finally control the cutting unit to cut the wafer along the cutting focus.

[0013] The principle and effect of this scheme are: the image information of the wafer is obtained through the acquisition unit, and the image is recognized and processed by the data processing unit, and the right-angle edge features of the wafer are extracted to compare with the preset wafer template, so as to calculate the position and posture error of the wafer. Then, the position and posture of the wafer on the workbench are adjusted by the driving unit to ensure that the wafer reaches the predetermined alignment accuracy. Finally, the laser transmitter determines the cutting focus position and drives the cutting unit to complete the cutting.

[0014] Furthermore, the cutting unit includes a grinding wheel and a driving module for driving the grinding wheel, a sleeve is provided at the end of the grinding wheel shaft, and the cutting unit also includes a supporting platform for supporting the wafer; the laser emitter is arranged on the sleeve, and the laser emitting end of the laser emitter is inclined toward the grinding wheel tool of the grinding wheel; the measuring unit also includes a laser receiver and a controller, the controller is electrically connected to the laser receiver, the laser receiver is arranged on the supporting platform, and the laser receiver is arranged in coordination with the laser emitter.

[0015] The principle and effect of this scheme are as follows: (1) During the cutting process of the existing grinding wheel dicing machine, the grinding wheel tool and the wafer surface produce violent friction and collision, causing the blade to gradually wear out, and the debris and heat generated by the cutting may also damage the blade and make it blunt, which will cause the edge contour of the back of the wafer to be uneven when cutting the wafer, causing the cutting path on the back of the wafer to collapse, and even damage the chip. Please refer to "Research on Silicon Wafer Composite Slicing Process" (Li Yanling, Gao Aimei, Zhang Yali. Research on Silicon Wafer Composite Slicing Process [J]. Special Equipment for Electronic Industry, 2018 (268): 25-28.). Therefore, it is necessary to detect whether the grinding wheel tool has defects. (2) The grinding wheel tool in this scheme uses a diamond blade. The cutting process of the grinding wheel is all existing technology and will not be described in detail here. Before cutting the wafer, the grinding wheel tool is detected, and the laser emitter is activated to emit laser by starting the grinding wheel tool to rotate, and the laser emitting end of the laser emitter is tilted toward the grinding wheel tool of the grinding wheel, that is, the laser is tilted to irradiate the blade of the grinding wheel tool. If the tool has no defects, that is, the blade is not blunt and worn seriously, the laser will irradiate on the diamond blade and refract vertically downward, and the laser receiver will receive the laser signal, and the controller will not issue an alarm; if the tool has defects, due to the blunt blade, the laser will tilt toward the support platform and will not refract vertically downward, and the laser receiver will not receive the laser signal, and the controller will issue an alarm signal to indicate that the grinding wheel tool is damaged. (3) The laser transmitter in this scheme can locate the cutting focus of the grinding wheel tool by emitting laser, so that the grinding wheel tool cuts along the cutting focus, making the cutting path more accurate and neat. At the same time, the laser transmitter can also detect whether the tool has defects and wear by emitting laser and cooperating with the laser receiver.

[0016] Furthermore, it also includes a cleaning component, which includes a conductive rod and an electrostatic generator, the conductive rod is electrically connected to the electrostatic generator, the conductive rod is arranged on the sleeve, the supporting platform is provided with a charge plate, and the charge carried by the charge plate is opposite to that of the conductive rod; the controller is electrically connected to the electrostatic generator.

[0017] The principle and effect of this scheme are as follows: (1) When the grinding wheel tool is cutting, the workpiece is separated and grooved by the rotation of the tool itself and the action of cutting water. Although there is the effect of cutting water, dust will still float in the cutting chamber. Therefore, the dust in the cutting chamber needs to be removed. In this scheme, the electrostatic transmitter is used to give a charge to the conductive rod, and the dust generated by the tool cutting is charged by the conductive rod. When the tool is inspected before cutting, if the laser receiver receives the laser, the controller controls the electrostatic generator to energize the conductive rod, otherwise it controls the electrostatic generator not to energize the conductive rod. After the tool cutting is completed, the wafer and film on the carrier are removed. Since the charge plate and the dust have opposite charges, opposite poles attract each other, thereby attracting the dust to the charge plate. (2) This scheme determines whether to energize the conductive rod when inspecting the tool. If the conductive rod is energized, the dust generated after cutting is charged, and the dust is adsorbed by the charge plate after cutting, thereby avoiding the problem that the dust in the cutting chamber is difficult to recover and clean.

[0018] Furthermore, the sleeve includes a first sleeve and a second sleeve, the conductive rod and the laser emitter are both arranged on the outer wall of the second sleeve, the first sleeve is coaxially fixedly connected to the rotating shaft of the grinding wheel, the first sleeve is rotatably connected to the end faces of the second sleeve, a slider and a pin are slidably provided in the first sleeve, the slider and the pin are connected by a pull rope, and a socket matching the pin is provided in the second sleeve.

[0019] The principle and effect of this scheme are as follows: the first sleeve rotates at the same frequency as the grinding wheel. When testing the grinding wheel tool, the rotation speed of the grinding wheel tool is controlled to be faster, and the slider is subjected to a large centrifugal force and moves away from the rotation center axis. The slider drives the pin to move, so that the pin is away from the socket, and the second sleeve does not rotate, so that the laser emitted by the laser transmitter always tilts toward the blade of the cutting tool to detect the tool. After the cutting is completed, the cutting chamber is dusted. At this time, the rotation speed of the grinding wheel tool is reduced, and the slider is subjected to a small centrifugal force. Driven by the pull rope, the slider and the pin are reset, and the first sleeve drives the second sleeve to rotate, so that the conductive rod rotates to contact the dust in the cutting chamber and charges the dust.

[0020] Furthermore, the support platform is provided with a dust storage bin, the charge plate is arranged in the dust storage bin, a support plate is provided on the top of the support platform, and a plurality of through holes are provided on the support plate.

[0021] The principle and effect of this solution is that the wafer and the film are placed on the carrier plate during cutting. After the cutting is completed, the wafer and the film are removed, the charge plate absorbs the dust, and enters the dust storage bin through the through hole, so that the dust is collected in the dust storage bin. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1A flowchart of a wafer alignment method based on finding a right-angle edge of a wafer according to the present invention; Figure 2 It is a structural schematic diagram of a wafer alignment system based on finding the right-angle edge of the wafer according to the present invention; Figure 3 The structure of the measuring unit of the present invention is shown in FIG. Figure 1 ; Figure 4 The structure of the measuring unit of the present invention is shown in FIG. Figure 2 : Figure 5 It is a structural schematic diagram of the carrying platform of the present invention; Figure 6 It is a schematic diagram of the structure of the driving assembly of the present invention.

[0023] The figure marks in the drawings of the specification include: wafer 1, laser emitter 2, grinding wheel 3, grinding wheel tool 31, sleeve 4, first sleeve 41, second sleeve 42, slider 43, pin 44, pull rope 45, socket 46, support platform 5, dust storage bin 51, support plate 52, laser receiver 6, cleaning component 7, conductive rod 71, charge plate 72, and adhesive film 8. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the concept and technical effects of the present invention in combination with the embodiments, so as to fully understand the purpose, features and effects of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative work are all within the scope of protection of the present invention: See also Figure 1 A wafer alignment method based on finding the right-angle edge of the wafer comprises the following steps: Step S10: image acquisition, using an industrial camera to capture an image of the wafer placed on the workbench of the dicing machine to obtain an image of the wafer; Step S20: image preprocessing, performing grayscale transformation, filtering, threshold segmentation and edge extraction on the collected wafer image; Step S30: constructing a template, according to the known shape and size of the wafer, constructing a wafer template including the right-angle edge features of the wafer, the template is used for subsequent image comparison and error calculation; Step S40: Calculate the wafer posture. In the wafer image extracted in step S20, use the Hough transform algorithm to detect the straight line segments in the wafer image, determine the X, Y coordinates and the rotation angle θ of the current wafer on the workbench, so as to determine the direction and position of the wafer. Step S50: calculating the alignment error, comparing the wafer image determined in step S40 with the wafer template constructed in step S30, and calculating the size and direction of the alignment error of the wafer in the X and Y directions and the rotation angle θ through comparison and analysis; Step S60: the wafer is aligned and positioned. According to the wafer posture calculated in step S50 and the existing alignment error, the position and posture of the wafer on the workbench are adjusted by controlling the X, Y, and C axis movements of the dicing machine to make the wafer reach a predetermined position and posture. Step S70: Verify the position. After the wafer is aligned, use the industrial camera to capture the wafer image again, and compare the newly captured image with the wafer template for a second time. If the comparison result shows that there is an error and the error exceeds a preset threshold, fine-tune according to the feedback information until the deviation between the actual placement position of the wafer and the wafer template is less than or equal to the preset threshold. Finally, start the grinding wheel dicing machine to cut the wafer.

[0025] Furthermore, in step S20, the filtering process specifically includes the following steps: Step S21: using a sliding window of size N×N and determining it according to the noise condition of the wafer image and the desired filtering effect; Step S22: using the upper left corner of the wafer image obtained in step S10 as the starting scanning position, and determining the scanning direction and step length according to the size and dimension of the wafer image, using a row-by-row and column-by-column scanning method; Step S23: placing the sliding window at the starting scanning position of the wafer image and ensuring that the window completely covers the pixels in the area, reading the grayscale values ​​of all pixels in the sliding window and storing them in an array; Step S24: sort the grayscale values ​​of the pixels in the array in ascending order, find the middle value of the sorted array, and use it as the median value of the current scanning area, and then replace the grayscale value of the center point of the sliding window with the calculated median value; Step S25: according to the set scanning direction and step length, move the sliding window to the next scanning position until the sliding window traverses the entire wafer image.

[0026] Furthermore, in step S20, the threshold segmentation specifically includes the following steps: Step S26: traversing each pixel point in the wafer image pre-processed in step S10; Step S27: Compare the grayscale value of each pixel with a set threshold; if the grayscale value of the pixel is greater than the threshold, it is set to 255, indicating a wafer feature; if the grayscale value of the pixel is less than or equal to the threshold, it is set to 0, indicating a background area; Step S28: After traversing the entire wafer image, the original wafer binary image containing feature information is finally obtained, wherein the features on the wafer are represented by white areas and the background is represented by black areas; The edge extraction in step S20 specifically includes the following steps: Step S29: performing Gaussian smoothing on the wafer image, and using the Sobel operator to calculate the gradient amplitude and direction of each pixel in the image, and identifying the edge points of the image; Step S210: only retain the points with the largest amplitude in the gradient direction, and remove the false edges in edge detection; Step S211: set a high threshold and a low threshold, the high threshold is used to determine a strong edge, the low threshold is used to connect the edge, then use the high threshold to obtain an edge image, and use the low threshold to connect the broken parts in the edge; finally, obtain a binary image containing wafer edge information.

[0027] Furthermore, the template construction in step S30 specifically includes the following steps: Step S31: select a standard wafer of known shape and size as a reference, and ensure that the right-angle edge of the standard wafer is parallel to the movement axis direction of the dicing machine; then use an industrial camera to shoot the standard wafer, and ensure that the image is clear and contains the complete wafer shape; Step S32: converting the standard wafer image from a color image to a grayscale image, and performing filtering, threshold segmentation and edge extraction on the grayscale image; then using a corner detection algorithm and a Hough transform method to detect feature points on the edge contour of the standard wafer image, the feature points including right-angle vertices and right-angle edges, and marking and recording the detected feature points, including the coordinate position of the feature points in the image and the wafer edge information of the image; Step S32: Generate a standard template image of the wafer based on the extracted feature point information, and the template image only contains the edge contour and feature point marks of the wafer, and does not contain any background information; then save the generated template image and subsequently match it with the wafer image to be processed.

[0028] Further, in the step S40, according to the result of matching the wafer image with the template image, the centroid coordinates of the matching area and the deviation angle from the template image are extracted, and the deviation angle is regarded as the alignment angle of the wafer; if the deviation angle is positive, it means that the wafer needs to be rotated clockwise to be aligned, and if the deviation angle is negative, it means that the wafer needs to be rotated counterclockwise to be aligned; in the step S50, the following steps are specifically included: Step S51: determining an ideal position of the grinding wheel cutting focus on the wafer according to a preset cutting path and cutting road width of the dicing machine; Step S52: driving the grinding wheel to move to a preset cutting position, and photographing the actual position of the cutting focus of the grinding wheel on the wafer at this time by an industrial camera; Step S53: Compare the ideal position with the actual position, and calculate the translation error between the two in the X and Y directions and the rotation error around the Z axis; wherein the translation error is the coordinate difference between the ideal position and the actual position in the X and Y directions, and the rotation error is the angle between the actual cutting path direction and the ideal cutting path direction; Step S54: according to the calculated translation error and rotation error, drive the wafer to move and perform error compensation adjustment; for the translation error, compensate by adjusting the motion parameters of the X and Y axes; for the rotation error, compensate by adjusting the angle of the C axis; Step S55: After the error is compensated, the industrial camera is used again to photograph the actual position of the grinding wheel to verify the compensation effect; if the error between the actual position after compensation and the ideal position is within the threshold range, the cutting step is performed; if the error between the actual position after compensation and the ideal position is not within the threshold range, the above steps are repeated until the error value is within the threshold range.

[0029] In order to better implement the above-mentioned wafer alignment method based on finding the right-angled edge of the wafer, the present invention also provides a wafer alignment system based on finding the right-angled edge of the wafer.

[0030] See also Figure 2 , a wafer alignment system based on finding the right-angle edge of a wafer, comprising an acquisition unit, a drive unit, a cutting unit, a measuring unit and a data processing unit. Among them, the acquisition unit is used to acquire image information of the wafer 1 to be cut; the drive unit is used to control the X, Y, and C axis movement of the dicing machine, and adjust the position and posture of the wafer 1 on the workbench; the cutting unit is used to cut the wafer 1; the measuring unit includes a laser emitter 2, and the laser emitter 2 is used to measure the position of the cutting focus of the cutting unit on the wafer 1; the data processing unit is used to receive and identify the image information and cutting focus of the wafer 1, and then compare the image of the wafer 1 with the wafer template, and then calculate the wafer posture and alignment error, and control the X, Y, and C axis movement of the dicing machine, adjust the position of the wafer 1, and finally control the cutting unit to cut the wafer 1 along the cutting focus. The image information of the wafer 1 is obtained by the acquisition unit, and the image is recognized and processed by the data processing unit, and the right-angle edge features of the wafer 1 are extracted to compare with the preset wafer template, so as to calculate the position and posture error of the wafer 1. Then, the position and posture of the wafer 1 on the workbench are adjusted by the driving unit to ensure that the wafer 1 reaches a predetermined alignment accuracy. Finally, the laser transmitter 2 determines the cutting focus position and drives the cutting unit to complete the cutting.

[0031] See also Figure 2-Figure 4The cutting unit includes a grinding wheel 3 and a driving module for driving the grinding wheel 3. A sleeve 4 is provided at the end of the rotating shaft of the grinding wheel 3. The cutting unit also includes a supporting platform 5 for supporting the wafer 1. The laser emitter 2 is arranged on the sleeve 4, and the laser emitting end of the laser emitter 2 is inclined toward the grinding wheel tool 31 of the grinding wheel 3. The grinding wheel tool 31 adopts a diamond blade. The measuring unit also includes a laser receiver 6 and a controller. The controller is electrically connected to the laser receiver 6. The laser receiver 6 is arranged on the supporting platform 5. The laser receiver 6 is arranged in coordination with the laser emitter 2.

[0032] Specific working principle: before cutting the wafer 1, the grinding wheel tool 31 is detected, and the grinding wheel tool 31 is started to rotate, so that the laser emitter 2 emits laser, and the laser emitting end of the laser emitter 2 is tilted toward the grinding wheel tool 31 of the grinding wheel 3, that is, the laser is tilted to the blade of the grinding wheel tool 31. If there is no defect in the tool, that is, the blade is not blunt and worn seriously, the laser will be irradiated on the diamond blade and refracted vertically downward, and the laser receiver 6 is used to receive the laser signal, and the controller will not issue an alarm; if there is a defect in the tool, due to the blunt blade, the laser will be tilted to the support platform 5 instead of refracted vertically downward, and the laser receiver 6 will not receive the laser signal, and the controller will issue an alarm signal, indicating that the grinding wheel tool 31 is damaged.

[0033] See also Figure 2-Figure 5 , and also includes a cleaning component 7, which includes a conductive rod 71 and an electrostatic generator, the conductive rod 71 is electrically connected to the electrostatic generator, the conductive rod 71 is arranged on the sleeve 4, and the supporting platform 5 is provided with a charge plate 72, and the charge carried by the charge plate 72 is opposite to that of the conductive rod 71; the controller is electrically connected to the electrostatic generator; the supporting platform 5 is provided with a dust storage bin 51, the charge plate 72 is arranged in the dust storage bin 51, and the top of the supporting platform 5 is provided with a supporting plate 52, and the supporting plate 52 is provided with a plurality of through holes.

[0034] Specific working principle: The electrostatic transmitter is used to give electric charge to the conductive rod 71, and the dust generated by the cutting of the grinding wheel tool 31 is charged by the conductive rod 71. When the tool is inspected before cutting, if the laser receiver 6 receives the laser, the controller controls the electrostatic generator to energize the conductive rod 71, otherwise the electrostatic generator is controlled not to energize the conductive rod 71. After the tool cutting is completed, the wafer 1 and the film 8 on the carrier are removed. Since the charge plate 72 and the dust have opposite charges, opposite poles attract each other, thereby attracting the dust to the charge plate 72 and entering the dust bin 51 through the through hole, so that the dust is collected in the dust bin 51.

[0035] See also Figure 6The sleeve 4 includes a first sleeve 41 and a second sleeve 42. The conductive rod 71 and the laser emitter 6 are both arranged on the outer wall of the second sleeve 42. The first sleeve 41 is coaxially fixedly connected with the rotating shaft of the grinding wheel 3. The first sleeve 41 is rotatably connected with the end faces of the second sleeve 42. A slider 43 and a latch 44 are slidably provided in the first sleeve 41. The slider 43 and the latch 44 are connected by a pull rope 45. A socket 46 matching the latch 44 is provided in the second sleeve 42.

[0036] Specific working principle: The first sleeve 41 rotates at the same frequency as the grinding wheel 3. When the grinding wheel tool 31 is tested, the rotation speed of the grinding wheel tool 31 is controlled to be faster, and the slider 43 is subjected to a large centrifugal force and moves away from the rotation center axis. The slider 43 drives the latch 44 to move, so that the latch 44 is away from the socket 46. The second sleeve 42 does not rotate, so that the laser emitted by the laser transmitter 2 always tilts toward the blade of the cutting tool to test the tool. After the cutting is completed, the cutting chamber is dusted. At this time, the rotation speed of the grinding wheel tool 31 is reduced, and the slider 43 is subjected to a small centrifugal force. Driven by the pull rope 45, the slider 43 and the latch 44 are reset, and the first sleeve 41 drives the second sleeve 42 to rotate, so that the conductive rod 71 rotates to contact the dust in the cutting chamber and charges the dust.

[0037] The above is only an embodiment of the present invention, and the common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A wafer alignment method based on finding the right-angle edge of the wafer, characterized in that: The steps include: Step S10: image acquisition, using an industrial camera to capture an image of the wafer placed on the workbench of the dicing machine to obtain an image of the wafer; Step S20: image preprocessing, performing grayscale transformation, filtering, threshold segmentation and edge extraction on the collected wafer image; Step S30: constructing a template, according to the known shape and size of the wafer, constructing a wafer template including the right-angle edge features of the wafer, the template is used for subsequent image comparison and error calculation; Step S40: Calculate the wafer posture. In the wafer image extracted in step S20, use the Hough transform algorithm to detect the straight line segments in the wafer image, determine the X, Y coordinates and the rotation angle θ of the current wafer on the workbench, so as to determine the direction and position of the wafer. Step S50: calculating the alignment error, comparing the wafer image determined in step S40 with the wafer template constructed in step S30, and calculating the size and direction of the alignment error of the wafer in the X and Y directions and the rotation angle θ through comparison and analysis; Step S60: the wafer is aligned and positioned. According to the wafer posture calculated in step S50 and the existing alignment error, the position and posture of the wafer on the workbench are adjusted by controlling the X, Y, and C axis movements of the dicing machine to make the wafer reach a predetermined position and posture. Step S70: Verify the position. After the wafer is aligned, use the industrial camera to capture the wafer image again, and compare the newly captured image with the wafer template for a second time. If the comparison result shows that there is an error and the error exceeds a preset threshold, fine-tune according to the feedback information until the deviation between the actual placement position of the wafer and the wafer template is less than or equal to the preset threshold. Finally, start the grinding wheel dicing machine to cut the wafer.

2. The wafer alignment method based on finding the right-angle edge of the wafer according to claim 1, characterized in that: In step S20, the filtering process specifically includes the following steps: Step S21: using a sliding window of size N×N and determining it according to the noise condition of the wafer image and the desired filtering effect; Step S22: using the upper left corner of the wafer image obtained in step S10 as the starting scanning position, and determining the scanning direction and step length according to the size and dimension of the wafer image, using a row-by-row and column-by-column scanning method; Step S23: placing the sliding window at the starting scanning position of the wafer image and ensuring that the window completely covers the pixels in the area, reading the grayscale values ​​of all pixels in the sliding window and storing them in an array; Step S24: sort the grayscale values ​​of the pixels in the array in ascending order, find the middle value of the sorted array, and use it as the median value of the current scanning area, and then replace the grayscale value of the center point of the sliding window with the calculated median value; Step S25: according to the set scanning direction and step length, move the sliding window to the next scanning position until the sliding window traverses the entire wafer image.

3. The wafer alignment method based on finding the right-angle edge of the wafer according to claim 1, characterized in that: In step S20, the threshold segmentation specifically includes the following steps: Step S26: traversing each pixel point in the wafer image pre-processed in step S10; Step S27: Compare the grayscale value of each pixel with a set threshold; if the grayscale value of the pixel is greater than the threshold, it is set to 255, indicating a wafer feature; if the grayscale value of the pixel is less than or equal to the threshold, it is set to 0, indicating a background area; Step S28: After traversing the entire wafer image, the original wafer binary image containing feature information is finally obtained, wherein the features on the wafer are represented by white areas and the background is represented by black areas; The edge extraction in step S20 specifically includes the following steps: Step S29: performing Gaussian smoothing on the wafer image, and using the Sobel operator to calculate the gradient amplitude and direction of each pixel in the image, and identifying the edge points of the image; Step S210: only retain the points with the largest amplitude in the gradient direction, and remove the false edges in edge detection; Step S211: set a high threshold and a low threshold, the high threshold is used to determine a strong edge, the low threshold is used to connect the edge, then use the high threshold to obtain an edge image, and use the low threshold to connect the broken parts in the edge; finally, obtain a binary image containing wafer edge information.

4. The wafer alignment method based on finding the right-angle edge of the wafer according to claim 1, characterized in that: The template construction in step S30 specifically includes the following steps: Step S31: select a standard wafer of known shape and size as a reference, and ensure that the right-angle edge of the standard wafer is parallel to the movement axis direction of the dicing machine; then use an industrial camera to shoot the standard wafer, and ensure that the image is clear and contains the complete wafer shape; Step S32: converting the standard wafer image from a color image to a grayscale image, and performing filtering, threshold segmentation and edge extraction on the grayscale image; then using a corner detection algorithm and a Hough transform method to detect feature points on the edge contour of the standard wafer image, the feature points including right-angle vertices and right-angle edges, and marking and recording the detected feature points, including the coordinate position of the feature points in the image and the wafer edge information of the image; Step S32: Generate a standard template image of the wafer based on the extracted feature point information, and the template image only contains the edge contour and feature point marks of the wafer, and does not contain any background information; then save the generated template image and subsequently match it with the wafer image to be processed.

5. The wafer alignment method based on finding the right-angle edge of the wafer according to claim 1, characterized in that: In the step S40, according to the result of matching the wafer image with the template image, the centroid coordinates of the matching area and the deviation angle from the template image are extracted, and the deviation angle is regarded as the alignment angle of the wafer; if the deviation angle is positive, it means that the wafer needs to be rotated clockwise to be aligned, and if the deviation angle is negative, it means that the wafer needs to be rotated counterclockwise to be aligned; in the step S50, the following steps are specifically included: Step S51: determining an ideal position of the grinding wheel cutting focus on the wafer according to a preset cutting path and cutting road width of the dicing machine; Step S52: driving the grinding wheel to move to a preset cutting position, and photographing the actual position of the cutting focus of the grinding wheel on the wafer at this time by an industrial camera; Step S53: Compare the ideal position with the actual position, and calculate the translation error between the two in the X and Y directions and the rotation error around the Z axis; wherein the translation error is the coordinate difference between the ideal position and the actual position in the X and Y directions, and the rotation error is the angle between the actual cutting path direction and the ideal cutting path direction; Step S54: according to the calculated translation error and rotation error, drive the wafer to move and perform error compensation adjustment; for the translation error, compensate by adjusting the motion parameters of the X and Y axes; for the rotation error, compensate by adjusting the angle of the C axis; Step S55: After the error compensation, the actual position of the grinding wheel is photographed again using the industrial camera to verify the compensation effect; If the error between the actual position after compensation and the ideal position is within a threshold range, the cutting step is performed; If the error between the actual position after compensation and the ideal position is not within the threshold range, the above steps are repeated until the error value is within the threshold range.

6. A wafer alignment system based on finding the right-angle edge of the wafer, characterized in that: include: A collection unit, the collection unit being used to collect image information of the wafer to be cut (1); A drive unit, the drive unit being used to control the X, Y, and C axis movements of the dicing machine and to adjust the position and posture of the wafer (1) on the workbench; A cutting unit, the cutting unit being used to cut the wafer (1); A measuring unit, the measuring unit comprising a laser emitter (2), the laser emitter (2) being used to measure the position of a cutting focus of a cutting unit on the wafer (1); A data processing unit, the data processing unit is used to receive and identify image information and a cutting focus of a wafer (1), then compare the image of the wafer (1) with a wafer template, calculate the wafer posture and alignment error, control the X, Y, and C axis movement of a dicing machine, adjust the position of the wafer (1), and finally control the cutting unit to cut the wafer (1) along the cutting focus.

7. The wafer alignment system based on finding the right-angle edge of the wafer according to claim 6, characterized in that: The cutting unit comprises a grinding wheel (3) and a driving module for driving the grinding wheel (3); a sleeve (4) is provided at the end of a rotating shaft of the grinding wheel (3); the cutting unit further comprises a support platform (5) for supporting a wafer (1); the laser emitter (2) is arranged on the sleeve (4), and a laser emitting end of the laser emitter (2) is inclined toward a grinding wheel tool (31) of the grinding wheel (3); the measuring unit further comprises a laser receiver (6) and a controller, the controller being electrically connected to the laser receiver (6); the laser receiver (6) is arranged on the support platform (5), and the laser receiver (6) is arranged in coordination with the laser emitter (2).

8. The wafer alignment system based on finding the right-angle edge of the wafer according to claim 7, characterized in that: The cleaning device also comprises a cleaning component (7), wherein the cleaning component (7) comprises a conductive rod (71) and an electrostatic generator, wherein the conductive rod (71) is electrically connected to the electrostatic generator, the conductive rod (71) is arranged on the sleeve (4), the supporting platform (5) is provided with a charge plate (72), and the charge carried by the charge plate (72) is opposite to that of the conductive rod (71); and the controller is electrically connected to the electrostatic generator.

9. The wafer alignment system based on finding the right-angle edge of the wafer according to claim 8, characterized in that: The sleeve (4) comprises a first sleeve (41) and a second sleeve (42); the conductive rod (71) and the laser emitter (6) are both arranged on the outer wall of the second sleeve (42); the first sleeve (41) is coaxially fixedly connected to the rotating shaft of the grinding wheel (3); the first sleeve (41) and the end surface of the second sleeve (42) are rotatably connected; a slider (43) and a latch (44) are slidably arranged in the first sleeve (41); the slider (43) and the latch (44) are connected by a pull rope (45); and a socket (46) matching the latch (44) is arranged in the second sleeve (42).

10. The wafer alignment system based on finding the right-angle edge of the wafer according to claim 9, characterized in that: The support platform (5) is provided with a dust storage bin (51), the charge plate (72) is arranged in the dust storage bin (51), a support plate (52) is provided on the top of the support platform (5), and a plurality of through holes are provided on the support plate (52).

Images