Wafer cutting path generation and laser cutting method

Abstract

The present invention discloses a method for generating a wafer cutting path and a laser cutting method, including the following steps: obtaining cutting area information; scanning the cutting area to obtain the topography of the cutting area and generating fitting surface information of the scanned area; obtaining the current photographing height according to the fitting surface information, and performing image acquisition with a fine positioning camera based on the current photographing height; evaluating the current cutting image according to a set threshold, if the threshold is reached, analyzing and generating current first three-dimensional cutting point information of the cutting path based on the current image position and the current photographing position; if the threshold is not reached, adjusting the photographing height to obtain multiple cutting images at multiple heights, evaluating the multiple cutting images, selecting the best first image position and the corresponding adjusted photographing position to analyze and generate the current first three-dimensional cutting point information; generating multiple three-dimensional cutting point information, fitting to generate a cutting path and performing a cutting method. According to the cutting path generation method and the cutting method of the present invention, the wafer cutting accuracy is significantly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of semiconductor processing, and particularly relates to a method for generating a wafer cutting path and a laser cutting method, and more particularly, to a method for generating a wafer cutting path and a laser stealth cutting method in wafer stealth cutting. Background Art

[0002] Most of the existing laser stealth cutting devices adopt vision guidance technology for processing, that is, first collect wafer images, and then generate cutting paths according to image processing for laser cutting. The quality of the collected images affects the cutting path planning to a certain extent, which determines the quality of stealth cutting. In high-precision vision positioning scenarios, the camera field of view is very small and the focal length is also very small, resulting in the inability to clearly identify chips at a certain height. In order to obtain clear images, the existing technology uses corresponding ranging sensors to measure the topography of the wafer surface, and then adjusts the camera measurement height according to the surface undulation to collect images.

[0003] In the prior art, the topography of the wafer surface is usually measured by non-contact methods, but the topography acquisition quality in the common measurement methods also has its own limitations. For example, laser triangulation ranging is affected by the surface inclination and roughness of the measured object, the angular range of white light confocal is limited, structured light measurement is not conducive to measuring objects with complex surface structures, laser interferometry is not suitable for detecting large-size objects, nor is it suitable for measuring complex curved surfaces with large concave and convex changes. Others, such as stereo vision, have problems of complex structure and low accuracy. Nuclear magnetic resonance spectrometers cannot measure magnetic metal objects, and ultrasonic measurement technology is sensitive to temperature, etc.

[0004] The topography quality obtained by the above non-contact measurement methods, on the one hand, is limited by the measurement device itself and cannot obtain high-precision measurement results. On the other hand, since the wafer materials include silicon, germanium, silicon carbide, gallium arsenide, zinc oxide, diamond, aluminum nitride, silicon dioxide, sapphire, etc., the refractive and reflectivity of different materials are different, and the reliability of the measured topography is poor. On the third hand, affected by the entire cutting system (the cutting of the wafer includes multiple processes such as loading, high-speed movement measurement and adjustment, etc.), the high-speed movement during the cutting process or the vibration of the entire device during the image acquisition process may also cause the focus plane to deviate during the wafer image acquisition or the image impact caused by the failure to obtain the minute changes in motion in real time, and the image produces ghosting, affecting the generation and planning of the cutting path. Summary of the Invention

[0005] The present invention provides a method for generating a wafer cutting path and a laser cutting method, which can obtain a wafer contour image and accurately locate the cutting path, and further locate the cutting path based on the scanned wafer surface topography and the fine positioning method, and perform cutting according to the cutting path, which can significantly improve the wafer cutting accuracy.

[0006] The present invention discloses a method for generating a wafer cutting path, including the following steps:

[0007] Collect wafer contour image information by using a contour camera to obtain cutting area information;

[0008] Obtain the surface topography information of the cutting area and generate the fitting surface information of the scanning area;

[0009] At a cutting image acquisition position, obtain the current photographing height according to the fitting surface information, and execute current cutting image acquisition with a fine positioning camera according to the current photographing height;

[0010] Evaluate the current cutting image according to a set threshold,

[0011] If the threshold is reached, analyze and generate the current three-dimensional cutting point information of the cutting path according to the current image position and the current photographing position;

[0012] If the threshold is not reached, adjust the photographing height to obtain multiple cutting images at multiple heights, evaluate the multiple cutting images, select the best image position and the corresponding adjusted photographing position to analyze and generate the current three-dimensional cutting point information;

[0013] Update the cutting image acquisition position and repeat the above steps to correspondingly generate multiple three-dimensional cutting point information, and fit to generate a cutting path.

[0014] In one embodiment, when positioning is achieved by using a chip image, among the coordinates of the three dimensions of the three-dimensional cutting point information, the X coordinate and the Y coordinate come from the X coordinate and the Y coordinate of the chip position in the cutting image multiplied by the calibration transformation matrix between the fine positioning camera and the cutting stage, plus the correction parameter D, and the Z coordinate comes from the photographing height parameter under the best image position.

[0015] Further, the method for evaluating the current cutting image according to a set threshold is:

[0016] Locate the cutting image based on gray-scale matching, obtain the normalized cross-correlation value between the current cutting image and the template image, and compare the correlation value with the set threshold.

[0017] Further, obtaining the surface topography information of the cutting area and generating the fitting surface information of the scanning area further includes: positioning the wafer and obtaining the transformed fitting surface information after positioning.

[0018] Further, the method for positioning the wafer and obtaining the transformed fitting surface information after positioning is: the fine positioning camera collects at least one chip image at different positions, generates a fitting curve according to multiple chip positions, obtains the fine alignment angle by comparing with a template and executes angle adjustment to complete positioning, and obtains the transformed fitting surface information after positioning.

[0019] Further, the adjustment of the photographing height is adjusted in a set step distance, and at least includes multiple height adjustments upward and downward in the current height direction.

[0020] Further, the wafer is arranged on the cutting stage, and further includes a calibration step before obtaining the cutting area information by using a contour camera to collect the wafer contour image information; obtaining the positional relationship between the contour camera, the surface topography detection device, the fine positioning camera and the cutting stage through the calibration step;

[0021] Or unifying the coordinate systems of the contour camera, the surface topography detection device, the fine positioning camera and the cutting stage.

[0022] Further, the cutting image acquisition position is the chip position, and the three-dimensional cutting point information is obtained according to the distance between the cutting line and the chip requirement and the adjustment of the photographing height.

[0023] The present invention also provides a method for performing laser cutting according to the generated wafer cutting path. After fitting and generating the cutting path, it further includes performing laser cutting according to the generated cutting path;

[0024] Repeating the above cutting path generation step and laser cutting step until all cutting tasks are completed.

[0025] The present invention also provides a method for performing laser cutting according to the generated wafer cutting path,

[0026] After fitting and generating the cutting path, it further includes performing laser cutting according to the generated cutting path;

[0027] Repeating the above cutting path generation step and laser cutting step until the cutting task in the first direction is completed;

[0028] Controlling the cutting stage to rotate to rotate the wafer by 90°, and obtaining the transformed fitting surface information;

[0029] Completing the cutting task in the second direction, where the first direction and the second direction are perpendicular.

[0030] The present invention also discloses a computer device, including a memory and a processor. The memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the above method are implemented.

[0031] The present invention also discloses a computer-readable storage medium, on which a computer program is stored, and is characterized in that when the computer program is executed by a processor, the steps of the above method are implemented..

[0032] Implementing the method for generating the wafer cutting path and the laser cutting method according to the present invention can obtain the following beneficial effects:

[0033] (1) The present invention uses white light confocal to measure the surface topography of the wafer. When the fine positioning camera locates and takes pictures, the picture-taking height changes according to the topography change obtained by fitting. After the fitting analysis is completed, the picture-taking is adjusted and the taken pictures are analyzed in real time. If it is found through image analysis that the matching score of the pattern in the image is low, the picture-taking height is continuously adjusted at the corresponding position to trigger picture-taking. The picture with the highest pattern matching degree and the corresponding position information are used to analyze the scribing lanes of the stealth dicing. The positioning of each scribing lane is generated by analyzing clear pictures, which can improve the positioning accuracy.

[0034] (2) Considering that the wafer is placed on a polyester film tape during dicing, due to the stress of the polyester film tape, the wafer will expand and move during the dicing process. In the present invention, during each positioning dicing, a part of the dicing path is located first, and then the stealth dicing method is used, rather than implementing stealth dicing after locating the entire dicing path, thereby reducing the influence of expansion and movement on the positioning accuracy of the dicing path and the dicing accuracy.

[0035] (3) By using the method of the present invention, the accuracy of the distance between the final dicing line and the chip in the X and Y directions, and the distance between the dicing line and the surface of the scribing lane (stealth dicing depth) are both controlled within 1 μm. Description of the Drawings

[0036] Figure 1 It is a schematic flow chart of the wafer dicing path generation implemented according to the present invention.

[0037] Figure 2 It is a schematic overall flow chart of one implementation manner of the wafer laser dicing method implemented according to the present invention.

[0038] Figure 3 It is a schematic diagram of the wafer dicing stage involved in the wafer laser dicing method implemented according to the present invention.

[0039] Figure 4 It is a flow chart of the S10 dicing area confirmation in the wafer dicing path generation and laser dicing method implemented according to the present invention.

[0040] Figure 5 It is a corresponding flow chart of S20 in the wafer dicing path generation and laser dicing method implemented according to the present invention.

[0041] Figure 6 It is a flow chart of the single scribing lane positioning dicing in the wafer dicing path generation and laser dicing method implemented according to the present invention.

[0042] Figures 7-8 It is a schematic diagram of the wafer dicing direction and the scribing lane direction in the wafer dicing path generation and laser dicing method implemented according to the present invention.

[0043] Figure 9 Schematic diagram of obtaining the rotation center of the cutting stage in the wafer cutting path generation and laser cutting method implemented according to the present invention.

[0044] Figure 10 Schematic diagram of the calibration of the fine positioning camera and the confocal probe in the wafer cutting path generation and laser cutting method implemented according to the present invention.

[0045] Figure 11 Schematic diagram of the calibration between the fine positioning camera and the cutting stage in the wafer cutting path generation and laser cutting method implemented according to the present invention.

[0046] Figure 12 Schematic diagram of the calibration between the contour camera and the cutting stage in the wafer cutting path generation and laser cutting method implemented according to the present invention.

[0047] 11. Marble platform, 12. X-axis linear module, 13. Y-axis linear module, 14. Rotating table, 21. Confocal probe, 22. Fine positioning camera, 23. Calibration plate, 231. Mark on the calibration plate, 51. Contour camera, 52. Fine positioning camera, 53. Calibration plate. Detailed implementation manners

[0048] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0049] The present invention provides a cutting path generation method and a cutting method, which can obtain a wafer contour image and accurately locate the cutting path. Further, based on the scanned surface topography of the wafer and the fine positioning method to locate the cutting path, cutting is performed according to the cutting path, which can significantly improve the wafer cutting accuracy. The method proposed by the present invention is particularly applicable to the laser stealth cutting of miniLED chips. The following embodiments are described by taking the laser stealth cutting of miniLED chips on wafers as an example, but the present invention is not limited thereto, and it can be applied when cutting similar objects to be cut arranged in an array.

[0050] According to the technical features corresponding to the embodiments of the present invention, the wafer is usually 4 inches, 6 inches, 8 inches, 12 inches, etc. in size, and the miniLed chip is an LED chip with a size of 50-200 microns. Laser stealth cutting is performed on the wafer according to the cutting path to obtain multiple miniled chips. In one implementation, the surface topography acquisition device is a confocal probe, and it can also be other non-contact surface topography detection devices for ranging and height measurement to achieve the topography acquisition of the cutting area. The contour camera is a camera with a larger field of view, and the fine positioning camera is a camera with a smaller field of view and higher precision.

[0051] The cutting area is the position of the chip area to be cut on the wafer, that is, the area where the wafer needs to be processed invisibly. Generally, the chips to be cut are arranged in an array, and the cutting area between rows or columns can also be called a cutting lane in the following embodiments. One form is a cutting lane with grooves; the cutting path is the data information for performing specific stealth cutting operations after processing.

[0052] Figure 3 It is a schematic diagram of a wafer cutting table involved in the cutting path generation method and cutting method implemented according to the present invention. It includes a cutting stage for carrying the wafer to be cut and driving it to move in the horizontal direction. It includes a marble platform 11, which can move in three-dimensional directions on a plane through a driving mechanism. Specifically, the driving mechanism includes an X-axis linear module 12, a Y-axis linear module 13, and a DD motor 14 to achieve movement in the X direction, Y direction, and θ direction respectively, and also includes a fine positioning camera 22 arranged above the cutting carrier for image acquisition, and a confocal probe 21 also arranged above the cutting stage for topography acquisition. See Figure 12 , which is a schematic diagram of a wafer positioning table and a wafer cutting table involved in the cutting path generation method and cutting method implemented according to the present invention. The wafer positioning table is used to carry the wafer to be cut and perform contour positioning, and a contour camera 51 is arranged thereon for acquiring the wafer contour image, and then the wafer is moved to the cutting stage.

[0053] As Figure 1 shown in

[0054] S10 Use a contour camera to acquire the wafer contour image and obtain the cutting area information;

[0055] S20 Use a surface topography detection device to scan the cutting area, obtain the cutting area topography, and generate the fitting surface information of the scanned area;

[0056] S30 obtains the current photographing height according to the fitting surface information at the first cutting image acquisition position, and uses the fine positioning camera to perform the first cutting image acquisition based on the current photographing height;

[0057] Evaluate the current cutting image according to the set threshold,

[0058] If the threshold is reached, generate the current three-dimensional cutting point information of the cutting path according to the current cutting image position and the current photographing position;

[0059] If the threshold is not reached, adjust the photographing height to obtain multiple cutting images at multiple heights, evaluate the multiple cutting images, and select the best first cutting image position and the corresponding image to generate the current three-dimensional cutting point information;

[0060] Repeat this step at other cutting image acquisition positions to generate multiple cutting point information, and fit to generate the cutting path.

[0061] Among them, it also includes S40, the step of performing laser cutting according to the generated cutting path. Among them, generate a part of the cutting path and perform cutting, and repeat until cutting is completed, or, after generating all the cutting paths, perform laser cutting according to the cutting path.

[0062] As Figure 2 shown, according to an embodiment of the present invention, the wafer cutting path generation method and the cutting method include the following steps:

[0063] S10: Use a contour camera to collect the wafer contour image and obtain the cutting area information;

[0064] S20: Use a confocal probe to scan the scanning area, obtain the cutting area topography and generate the fitting surface information of the scanning area;

[0065] Among them, S20 also includes precisely aligning the wafer and obtaining the transformed fitting surface information after precise alignment;

[0066] S30: Obtain the current photographing height according to the fitting surface information at the first cutting image acquisition position, and use the fine positioning camera to perform the first cutting image acquisition based on the current photographing height;

[0067] Evaluate the current cutting image according to the set threshold,

[0068] If the threshold is reached, generate the current three-dimensional cutting point information of the cutting path according to the current cutting image position and the current photographing height;

[0069] If the threshold is not reached, adjust the photographing height to obtain multiple cutting images at multiple heights, evaluate the multiple cutting images, and select the best first cutting image position and the corresponding image to generate the current three-dimensional cutting point information;

[0070] Repeat this step at other cutting image acquisition positions to generate multiple cutting point information, and fit to generate a cutting path.

[0071] Specifically, in this embodiment, as Figure 2 shown in the cutting path generation method and the cutting method shown, in combination with S40: performing the process of laser cutting according to the generated cutting path, including first sequentially generating the cutting path in the first direction and performing cutting; then, controlling the cutting stage to rotate so that the wafer rotates 90°, obtaining the co-fitting surface information after secondary transformation, and sequentially generating and performing cutting on the cutting path in the second direction perpendicular to the first direction. In a specific embodiment, the first direction is represented by the Y direction, and the second direction is represented by the X direction.

[0072] It should be noted that in the above two-direction cutting scheme, it is also possible to complete the generation of multiple cutting paths in one direction, and then perform the cutting step, or after generating all the cutting paths in the first direction, then perform the cutting task.

[0073] Furthermore, the wafer cutting path generation method and the cutting method of the present invention further include, before S10: the step of S00 preparing materials; specifically, placing the wafer in a wafer cassette; after S40, it further includes S50: the step of receiving materials; specifically, moving the wafer back to the wafer cassette.

[0074] In order to enable the device for implementing the foregoing cutting path generation method and cutting method to operate automatically and perform laser cutting, calibration needs to be completed first. Among them, the calibration, which can be understood by those skilled in the art, is the process of unifying the coordinate systems of the cutting stage, the fine positioning camera, the confocal probe, the profile camera, and the laser processing module; or, obtaining the positional relationship of the wafer under the cutting stage, the fine positioning camera, the confocal probe, the profile camera, and the laser processing module. As a non-limiting embodiment, in the present invention, the cutting stage can move in three dimensions, and the wafer to be cut is placed on the cutting stage. In the implementation of the present invention, the calibration work to be carried out includes: obtaining the rotation center of the cutting stage, calibrating the fine positioning camera and the confocal probe, calibrating the fine positioning camera and the cutting stage, and calibrating the profile camera and the cutting stage; the present invention is not limited thereto, as long as the relevant calibration can be completed. Among them, the specific method of calibration is selected from a variety of calibration methods in the art. The execution of calibration depends on the specific situation of the device. Among them, step S10 and steps S20 to S30 can be carried out at two stations or at one station. In the embodiment of the present invention, the description is based on the two stations. When carried out at one station, the wafer positioning stage is cancelled, and the wafer is directly transferred to the cutting stage, and the profile camera is arranged above the cutting stage.

[0075] A more detailed description of the foregoing wafer cutting path generation method and cutting method is given below.

[0076] Among them, in S10, a contour camera is used to collect a wafer contour image, cutting area information is obtained, and confocal probe scanning area information is obtained based on the cutting area information. As shown in the schematic Figure 4 shown, the specific steps are as follows:

[0077] S11: Use a contour camera to collect a wafer contour image;

[0078] S12: Obtain the wafer contour area and the target cutting area, determine the centroid of the target cutting area and the circumscribed rectangle area parallel to the image coordinates;

[0079] Specifically, the wafer contour area is obtained by binarization. The largest area of the wafer contour area is selected as the target cutting area. The centroid of this area is represented by (ROICentR, ROICenterC), and the circumscribed rectangle area parallel to the image coordinates of this area is represented by Rectangle1;

[0080] S13: According to the positional relationship between the contour camera and the cutting stage obtained by calibration (the transformation matrix B in the following specific implementation manners) and the centroid (ROICenterR, ROICenterC) of the target cutting area, obtain the centroid (ROICenterX, ROICenterY) of the cutting area of the cutting stage. According to the transformation matrix B of the positional relationship between the contour camera and the cutting stage obtained by calibration and the circumscribed rectangle Rectangle1 parallel to the image coordinates, and the positional relationship (offsetX, offsetY, offsetZ) between the fine positioning camera and the confocal probe obtained by calibration, obtain the scanning area StageRectangle1 of the confocal probe; where:

[0081]

[0082] Among them, in S20, then use a confocal probe to scan the scanning area, obtain the surface topography of the cutting area and generate fitting surface information; and perform fine alignment of the wafer and obtain the transformed fitting surface information after fine alignment; as Figure 5 shown, the specific steps are as follows:

[0083] S21: The confocal probe scans the topography of the cutting area, processes the measured point cloud data, and uses the least squares method to fit to obtain the fitting surface Plane(X, Y, Z).

[0084] Specifically, in this step, according to the confocal probe scanning area StageRectangle1 and the confocal probe scanning line spacing Distance set by the recipe, the confocal probe scanning grid lines (Line1, Line2, Lime3…Line k ), where k is the number of the corresponding grid line, and each grid line must be parallel to the X-axis or Y-axis of the cutting platform. When parallel to the X-axis of the cutting platform, the mathematical expression of the kth grid line is as follows:

[0085] Y k = Y min + (k - 1) * Distance, X k ∈ [X intersect1 , X intersect2

[0086] Y min is the minimum value in the Y-axis direction of the cutting platform corresponding to the confocal probe scanning area StageRectangle1, and X intersect1 and X intersect2 are the values in the X-axis direction of the cutting platform at the two intersections of the line Y k and the area StageRectangle1.

[0087] Use the confocal probe to scan according to this grid line (Line1, Line2, Lime3…Line k ) to collect the surface topography of the cutting area.

[0088] Process the measured point cloud data, and obtain the fitted surface Plane(X, Y, Z) by least squares fitting;

[0089] In this step, the recipe mentioned above is the template information of the wafer to be processed, including wafer diameter, wafer thickness, lateral width of the cutting channel, longitudinal width of the cutting channel, length and width of the chip, template position, etc.;

[0090] In this embodiment, the scanning of the confocal probe can be at a fixed interval, or the same spacing can be used in the middle area, and the scanning interval in the edge area can be reduced.

[0091] S22 controls the cutting stage to move to the first position, so that the first chip is under the field of view of the fine positioning camera. Take a picture of the first chip area with the fine positioning camera according to the fitted surface information, and locate the first chip according to gray level matching;

[0092] In this embodiment, the first chip is the chip at the center of the wafer, which is convenient for searching and positioning. The present invention is not limited thereto, and the first chip can also be a chip at other positions.

[0093] Specifically, for the i-th time, the cutting platform moves to the first position​

[0094] (ROIStageR + i * ChipX / 3, ROIStageC, Z = Plane(ROIStageR + i * ChipX / 2, ROIStageC))

[0095] (At the initial position where the first chip is planned to be searched), the fine - positioning camera takes pictures of the chips at the corresponding positions on the wafer, and locates the first chip according to gray - scale matching, where ChipX is the width of a single chip; where i is greater than or equal to 1.

[0096] S23 evaluates the first chip image according to the set threshold, judges whether the threshold is reached. If the threshold is reached, proceed to the next step S24; if not, return to the previous step S22.

[0097] Specifically, according to the matching score, judge whether the threshold is reached, and thereby judge whether the first chip is located.

[0098] S24: Obtain the angle θ1 between the first chip and the template from the angle of the first chip calculated by positioning and the template angle recorded in the recipe; according to the template image coordinates and the position relationship (transformation matrix A) between the fine - positioning camera and the cutting stage obtained by calibration, obtain the position (ModelX1, ModelY1) of the first chip;

[0099] S25 takes the position (ModelX1, ModelY1) of the first chip as a reference, controls the cutting stage to move, and obtains the positions (ModelX2, ModelY2), (ModelX3, ModelY3) of the second chip and the third chip in the same direction. Then, use the positions of the first chip, the second chip, and the third chip to fit a straight line by the least - squares method, and obtain the angle θ2 between this and the positive direction.

[0100] In this step, the second chip and the third chip are the chips arranged adjacent to the first chip in the Y - direction for convenient search and positioning. Of course, the second chip and the third chip can also be chips adjacent or non - adjacent in the Y - direction. Specifically, during execution, each time taking (ModelX1, ModelY1) as a reference, respectively along the directions of θ1 and θ1 + 180°, move the wafer by a distance of the height of a cutting track GapY plus the height of a chip ChipY through the cutting stage, use the fine - positioning camera to take pictures, and locate the second chip (a chip above the first chip) and the third chip (a chip below the first chip) according to template matching. In the same way, obtain (ModelX2, ModelY2), (ModelX3, ModelY3);

[0101] Using the least squares method to fit a straight line based on (ModelX1, ModelY1), (ModelX2, ModelY2), and (ModelX3, ModelY3), the angle θ2 is obtained.

[0102] S26: Rotate the wafer by an angle θ2 around the rotation center (X center , Y center ) of the cutting platform by moving the cutting stage, and obtain the transformed fitting surface information.

[0103] Specifically, rotate the wafer by an angle θ2 around the center (X center , Y center ) of the cutting platform by moving the cutting stage. At this time,

[0104] The position (ModelX1, ModelY1) of the first chip

[0105] The coordinates become (ModelAfterRotatedX1, ModelAfterRotatedY1), and Plane(X, Y, Z) becomes Plane2(X, Y, Z). Among them, (ModelAfterRotatedX1, ModelAfterRotatedY1) is understood as the position of the cutting platform corresponding to the first chip at the center of the cutting area.

[0106] In the embodiments of the present invention, after the fine positioning camera acquires a clear picture of the chip, when there is no obvious angle of the chip in the image, the chip area can be selected as the matching template. The actual positioning of the cutting track needs to be achieved by positioning a single chip multiple times, and a matching algorithm based on gray values is used to position a single chip.

[0107] The specific steps of S30 and S40 are as follows: Refer to Figure 6 , which is a flowchart for positioning a cutting track in step S30 of the wafer cutting method implemented according to the embodiments of the present invention. In steps S30 and S40, according to the fitting surface data or the transformed fitting surface data, the cutting tracks in the Y direction of the wafer are confirmed one by one and invisible cutting is performed. Since the implementation methods of each cutting track in the vertical direction are similar, in this embodiment, the implementation method starts from the middle chip and searches for the last chip in one direction ( Figure 7 and Figure 8 as shown).

[0108] The judgment is based on the gray-scale matching positioning. Setting a score of 0.5 as the judgment basis can obtain a relatively accurate positioning result. After a positioning failure, 4 consecutive acquisitions will be made below and above the corresponding photographing height. The photographing range is expanded to 0.004 mm, which can avoid the situation where some points cannot be clearly clustered using the data of the fitting photographing plane Plane2, increase the data points for the cutting track fitting, and make the hidden cutting depth more stable.

[0109] S31 controls the cutting stage to move to the first position. The first chip is located under the field of view of the fine positioning camera, and the height of the cutting stage is set according to the transformed fitting surface information; use the fine positioning camera to take a picture of the first chip, and position the first chip in the image based on gray-scale matching;

[0110] Specifically, control the cutting stage to move so that the first chip is located under the field of view of the fine positioning camera, and the height of the cutting stage is set according to the transformed fitting surface.

[0111] (ModelAfterRotatedX1, ModelAfterRotatedY1 + i * j * (ChipY + GnpY)), Z = Plane2(ModelAfterRotatedX1, ModelAfterRotatedY1 + i * j * (ChipY + GapY))

[0112] At this time, the first chip is located under the field of view of the fine positioning camera. Use the fine positioning camera to take a picture of the first chip, and position the first chip in the image based on gray-scale matching (ImageCutLineR k , ImageCutLinC k );

[0113] S32: Determine whether the matching is successful. Specifically, if the score is greater than 0.5, it is determined that the matching is successful;

[0114] If the matching is successful, proceed to the next step S33;

[0115] If the matching is not successful, control the cutting stage to move up and down a certain distance. Specifically, it can move to the positions of Z - 2 * StepZ, Z - StepZ, Z + StepZ, Z + 2 * StepZ. Use the fine positioning camera to collect the image of the first chip again, and position the chip in the image based on gray-scale matching (ImageCutLineR k , ImageCutLineC k ), and determine again whether the matching is successful. If the score is greater than 0.5, it is determined that the matching is successful; if the matching is successful, proceed to the next step S33;

[0116] When the S33 multiple first chip image acquisitions meet the judgment conditions, the Z corresponding to the image with the maximum score is the optimal image height Zbest k , and its corresponding position is the position of the first chip in the image (ImageCutLineR k , ImageCutLineC k ).

[0117] According to the position of the first chip in the image, the positional relationship (matrix A) between the fine positioning camera and the cutting stage is obtained, and the planar position is obtained by adding the distance D between the cutting line and the chip requirement, and the optimal height Zbest k is used as the height position to obtain a point on the cutting path (StageCutLineX k , StageCutLineY k , Zbest k );

[0118] Specifically, from the calibrated positional relationship between the fine positioning camera and the cutting stage, first obtain how much x and y need to be moved from the target point to the camera center, and the actual position that the cutting stage needs to move to is the current photographing position plus the amount that needs to be moved.

[0119] S34: Control the cutting stage to move to other positions, other chips are under the field of view of the fine positioning camera, and the height of the cutting stage is set according to the transformed fitting surface information; use the fine positioning camera to take pictures of other chips, and based on gray scale matching, locate the positions of other chips in the image, and repeat the steps of S31 to S33 to obtain multiple points on the cutting path (StageCutLineX k , StagecutLineY k , Zbest k ), until all the chips in the Y direction corresponding to the first chip are completed;

[0120] S35: Multiple points on the cutting path (StageCutLineX k , StagecutLineY k , Zbest k ), and the offsets (deltaX, deltaY, drltaZ) between the fine positioning camera and the laser processing module are used to fit a three-dimensional straight line, which is the cutting line; specifically, in this embodiment, the least-squares method is used to fit the three-dimensional straight line.

[0121] S36: The laser processing module processes according to the path of the cutting line and performs laser cutting; specifically, in this embodiment, the cutting stage moves according to the cutting line path, which can ensure that the laser processing module processes according to the cutting line path.

[0122] S37: Position and cut the remaining Y-direction cutting lanes according to the methods of S41 to S46.

[0123] S38. Rotate the wafer on the cutting stage clockwise by 90° around the rotation center (X center , Y center ) of the cutting stage, and obtain the second transformed fitting surface information. Refer to steps S31 to S37 to complete the positioning and cutting of the X-direction cutting lanes.

[0124] Specifically, after rotating the wafer on the cutting stage clockwise by 90° around the rotation center (X center , Y center ), the center (ModelAfterRotatedX1, ModelAfterRotatedY1) of the cutting area becomes (ModelAfterRotatedX2, ModelAfterRotatedY2), and the fitting plane Plane2(X, Y, Z) of the cutting area becomes Plane3(X, Y, Z).

[0125] In addition, when specifically positioning the X-direction cutting lanes, the moving interval used for searching the chips becomes the size of the chips in the X direction plus the size of the X-direction cutting lanes, which is mathematically expressed as (ChipX + CapX).

[0126] In the above embodiments, in S30, for the first position and other positions, the sizes of the chips and cutting lanes of the known wafer can be used; preferably, start with the known sizes and then adopt the actual spacing; more preferably, adjust the actual spacing.

[0127] In the above specific embodiment of S30, when determining the cutting path and cutting according to the cutting path, the present invention preferably adopts the method of positioning one cutting path and cutting according to one cutting path. This is because the wafer is usually fixed on the polyester film tape, and due to the stress of the polyester film tape, the wafer will expand and move during the cutting process. This way reduces the influence of expansion and movement. Of course, it is also possible to determine all the cutting paths in one direction and then perform unified cutting; then determine the cutting paths in the other direction and then perform unified cutting; it is also possible to determine all the cutting paths in both directions and then perform unified cutting.

[0128] In the present invention, in both the fine alignment of the wafer in S20 and the evaluation in S30, a chip is located based on gray-scale matching to obtain the normalized cross-correlation value between the current fine alignment camera image and the template image, and the correlation value is compared with the set threshold. In the present application, the threshold is preferably set to 0.5, and it can also be other set values. Specifically, based on the ncc matching algorithm of gray-scale matching, after clearly capturing an image of a single chip using the fine alignment camera, when there is no obvious angle of the chip in the image, the chip area can be selected as the matching template. Actually, in the fine alignment of the wafer in S20 and the evaluation in S30, it is necessary to locate a single chip to achieve, and a matching algorithm based on gray-scale values is used to locate a single chip. The matching based on gray-scale values can finally return an instance with the highest score exceeding the set score.

[0129] The score is essentially the normalized cross-correlation value ncc(r, c) of the template t(r, c) and the image i(r, c):

[0130]

[0131] where n is the number of template points, R is the template area, u is the row coordinate of the template, v is the column coordinate of the template, r is the row coordinate of the image, c is the column coordinate of the image, and m t is the average gray-scale value of the template area:

[0132]

[0133] is the variance of the gray-scale values of the template area:

[0134]

[0135] m i (r, c) is to draw a template-sized area starting from the point (r, c) of the image, and the average gray-scale value of this area:

[0136]

[0137] is to draw a template-sized area starting from the point (r, c) of the image, and the variance of the gray-scale values of this area:

[0138]

[0139] In the present invention, the aforementioned calibration includes:

[0140] 1. Obtain the rotation center of the cutting stage

[0141] Such as Figure 9As shown in the figure, place the calibration plate on the wafer cutting stage. After the cutting stage is reset, rotate the angle of the DD motor to θ°, and use the fine positioning camera to observe a mark on the calibration plate. When the mark state is clear and the center position of the mark is concentric with the center of the crosshair of the camera, record the coordinates X of the cutting stage at this time. A / Y A / Z A ; After that, rotate the angle of the DD motor to (θ + 180)° or (θ - 180)°, use the fine positioning camera to observe the same mark. When the mark state is clear and the center position of the mark is concentric with the center of the crosshair of the camera, record the coordinates X of the cutting stage at this time. B / Y B / Z B ; Then, obtain the rotation center, (Xcenter, Ycenter) = 0.5 * ((X A +X B ),(Y A +Y B ))

[0142] 2. Calibration of the fine positioning camera and the confocal probe

[0143] Figure 3 、 Figure 9 and Figure 10 As shown in the figure, it includes a confocal probe 21, a fine positioning camera 22, a circular calibration plate 23, and a cross mark 231 on the calibration plate.

[0144] Place the calibration plate on the cutting stage, turn on the fine positioning camera, control the movement of the X-axis / Y-axis / Z-axis of the cutting stage to make the center of the calibration plate mark concentric with the crosshair of the camera, and record the coordinates X1 / Y1 / Z1 of the cutting stage at this time; control the movement of the X-axis / Y-axis / Z-axis of the cutting stage to make the light spot of the confocal probe align with the center of the mark pattern, and at the same time ensure that the reading of the confocal probe is in the middle area of its range, and record the coordinates X2 / Y2 / Z2 of the cutting stage at this time, and obtain the positional relationship between the fine positioning camera and the confocal probe:

[0145] (offsetX, offsetY, offsetZ) = (X1 - X2, Y1 - Y2, Z1 - Z2), that is, when the center of the fine positioning camera is aligned with a specific position of the cutting stage, by relatively moving the X-axis, Y-axis, and Z-axis of the cutting stage by offsetX, offsetY, and offsetZ, it can ensure that the confocal probe is exactly aligned with this specific position.

[0146] Here, the calibration of the fine positioning camera and the confocal probe is taken as an example. In actual implementation, the calibration of the confocal probe and the cutting stage can also be adopted to establish a connection between the fine positioning camera and the confocal probe.

[0147] 3. Calibration of the Fine Positioning Camera and the Cutting Stage

[0148] As shown in Figure 11 , the calibration board is placed on the cutting stage. Control the movement of the X-axis / Y-axis / Z-axis of the cutting stage, turn on the fine positioning camera, and make the center of the mark concentric with the cross auxiliary line of the fine positioning camera. Record the coordinates (Xencoder1, Yencoder1) of the cutting stage at this time. At the same time, use the Harris corner extraction algorithm to obtain the center position (Row1, Column1) of the mark at this time; then, control the movement of the X-axis / Y-axis / Z-axis of the cutting stage multiple times (at least 6 times) to make the mark point appear in each quadrant of the fine positioning camera's field of view, and record the corresponding coordinates (Xencoder2, Yencoder2), (Xencoder3, Yencoder3), (Xencoder4, Yencoder4),... of the cutting stage, as well as the corresponding image coordinates (Row2, Column2), (Row3, Column3), (Row4, Column4),... of the center of the mark. Assume that n compliant points are collected. The relationship between the fine positioning camera and the cutting stage for any k-th point is:

[0149]

[0150] where a, b, c, and d are constant parameters, Xencoder k and Yencoder k are the coordinates of the k-th point corresponding to the cutting platform, Xencoder1 and Yencoder1 are the coordinates of the first point corresponding to the cutting platform, Row k and Column K are the image coordinates of the k-th point, Row1 and Column1 are the image coordinates of the first point. Substitute the data of these n points into the above formula, and use the least-squares method to calculate the best transformation matrix A between the fine positioning camera and the cutting stage, that is, for any point (Rowr, Columnr) in the fine positioning camera's field of view to move to the center of the fine positioning camera's field of view, the corresponding X-axis / Y-axis needs to move relatively (a*(Rowr - Row1) + b*(Columnr - Column1), (c*(Rowr - Row1) + d*(Columnr - Column1)).

[0151] 4. Calibration of the Contour Camera and the Cutting Stage

[0152] Figure 12As shown in [figure], where there are a contour camera 51, a fine positioning camera 22, and a circular calibration plate 53. The wafer is initially located at the loading station. After taking a picture of it using the contour camera, the wafer is transferred to the cutting stage. After calibrating the field of view of the contour camera, the pixel coordinates (PixelRow, PixelColumn) of 7 or more marks evenly distributed on the circumference of the calibration plate are confirmed. Then, the cutting stage is controlled to move so that the corresponding mark is at the exact center of the field of view of the fine positioning camera, and the corresponding coordinates of the X-axis / Y-axis / Z-axis are (Encoderx, Encodery, Encoderz). In this way, a transformation matrix B between the contour camera and the cutting stage is established, and there is the following formula:

[0153]

[0154] where aa, bb, cc, dd are constant parameters, and t x and t y are the translation amounts along the X-axis and Y-axis of the cutting stage respectively.

[0155] Those skilled in the art can understand that the above calibration is a process of unifying the coordinate systems of the cutting stage, the fine positioning camera, the confocal probe, the contour camera, and the laser processing module; the present invention is not limited to this, and those skilled in the art can use the methods of the prior art for calibration according to needs. For example, the fine positioning camera, the confocal probe, the contour camera, and the laser processing module are all calibrated with the cutting stage to obtain a unified coordinate system.

[0156] The above are only embodiments of the present application and are not used to limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating a wafer cutting path, characterized in that, It includes the following steps: Collect wafer contour image information using a contour camera to obtain cutting area information; Scan the cutting area using a surface topography detection device to obtain cutting area surface topography information and generate fitting surface information of the scanned area; Obtain the current photographing height according to the fitting surface information at a cutting image acquisition position, and execute current cutting image acquisition with a fine positioning camera based on the current photographing height; Evaluate the current cutting image according to a set threshold; If the threshold is reached, analyze and generate current three-dimensional cutting point information of the cutting path based on the current image position and the current photographing position; If the threshold is not reached, adjust the photographing height to obtain multiple cutting images at multiple heights, evaluate the multiple cutting images, and select the best image position and the corresponding adjusted photographing position to analyze and generate current three-dimensional cutting point information; Update the cutting image acquisition position and repeat the above steps to correspondingly generate multiple three-dimensional cutting point information, and fit to generate a cutting path.

2. The wafer cutting path generation method according to claim 1, wherein The method for evaluating the current cutting image according to a set threshold is: Locate the cutting image based on gray-scale matching, obtain the normalized cross-correlation value between the current cutting image and the template image, and compare the correlation value with the set threshold.

3. The method for generating a wafer cutting path according to claim 1 or 2, characterized in that, Obtaining the cutting area surface topography information and generating the fitting surface information of the scanned area further includes: Locate the wafer and obtain the transformed fitting surface information after positioning.

4. The wafer dicing path generation method according to claim 3, wherein The method for locating the wafer and obtaining the transformed fitting surface information after positioning is: the fine positioning camera collects at least one chip image at different positions, generates a fitting curve according to multiple chip positions, compares with the template to obtain the fine alignment angle and performs angle adjustment to complete the positioning, and obtains the transformed fitting surface information after positioning.

5. The method for generating a wafer cutting path according to claim 1 or 2, characterized in that, The photographing height is adjusted in a set step distance, and at least includes multiple height adjustments upward and downward in the current height direction.

6. The wafer cutting path generation method according to claim 1 or 2, wherein The wafer is arranged on a cutting stage, and further includes a calibration step before collecting the wafer contour image information using the contour camera to obtain the cutting area information; through the calibration step, the position relationships of the contour camera, the surface topography detection device, the fine positioning camera and the cutting stage are obtained; Or unify the coordinate systems of the contour camera, the surface topography detection device, the fine positioning camera and the cutting stage.

7. The wafer cutting path generation method according to claim 1 or 2, wherein The cutting image acquisition position is the chip position, and the three-dimensional cutting point information is obtained according to the distance between the cutting line and the chip requirement and the adjusted photographing height.

8. A method for performing laser cutting by implementing the wafer dicing path generation method according to any one of claims 1 to 7, characterized in that, After fitting to generate the cutting path, it further includes performing laser cutting according to the generated cutting path; Repeat the above cutting path generation step and the laser cutting step until all cutting tasks are completed.

9. The method for performing laser cutting according to the wafer cutting path generation method of any one of claims 1 to 7, characterized in that: After fitting to generate the cutting path, it further includes performing laser cutting according to the generated cutting path; Repeat the cutting path generation step and the laser cutting step until the cutting task in the first direction is completed; Control the cutting stage to rotate to rotate the wafer by 90°, and obtain the transformed fitting surface information; Complete the cutting task in the second direction, and the first direction and the second direction are perpendicular.

10. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 9.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

Images