Wafer bump height and coplanarity automatic calculation method based on K-Means clustering segmentation

Abstract

The invention provides a wafer Bump height and coplanarity automatic calculation method based on K-Means clustering segmentation, and belongs to the field of semiconductor morphology detection. By combining a plane fitting method such as principal component analysis (PCA) and a K-Means clustering segmentation technology, the height and coplanarity parameters of the Bump can be efficiently and accurately extracted from three-dimensional point cloud data, and the measurement automation degree of the wafer Bump is remarkably improved. According to the method, accurate extraction of the Bump position of the wafer is achieved through two times of segmentation, and the method can adapt to Bumps of wafers with different densities, shapes and sizes in Die (the general method represents chips) and has high universality and adaptability. By accurately detecting the height of the wafer Bump and evaluating the coplanarity, defects in the manufacturing process can be found in time, and the quality control level of the wafer Bump in the semiconductor manufacturing process is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the field of computer image data processing, relates to semiconductor topography detection technology, and specifically is an automatic calculation method for wafer Bump height and coplanarity based on K-Means clustering segmentation, which is used for height measurement and coplanarity verification of wafer Bumps. Background Art

[0002] In the process of semiconductor manufacturing and packaging, the height and coplanarity of wafer Bumps (bumps) are key parameters affecting chip performance and quality. As the electrical connection points between the chip and the packaging substrate, the height and coplanarity of Bumps directly determine the reliability of soldering and the stability of electrical connection. If the Bump height is uneven or the coplanarity is poor, it may lead to problems such as poor soldering, short circuit or open circuit, thereby affecting the function and lifespan of the chip. Therefore, accurately measuring the height and coplanarity of Bumps is of great significance for ensuring chip packaging quality and improving product yield.

[0003] Currently, optical detection methods are widely used in Bump height measurement, which have the advantages of high detection efficiency, high reliability, and non-contact. The main measurement methods include spectral confocal method, white light interferometry, optical shadow method, optical triangulation method, and fringe projection profilometry, etc. Among them, the spectral confocal method has high resolution, no optical artifacts on complex measurement surfaces, and high accuracy, but the measurement speed of this method is slow and it is generally used for off-line measurement. The white light interferometry has high measurement accuracy, which can reach the nanometer level, but the measurement range is small and the measurement speed is also slow. The optical shadow method has a relatively fast measurement speed, but the measurement accuracy can only reach the micron level or above. And with the improvement of chip integration and the increase of wafer bump density, this method is easily affected by its spacing and the incident light angle, resulting in overlapping shadows and affecting the measurement. The optical triangulation method has the advantages of simple principle and fast measurement speed compared with the above methods, but the commonly used light source of this method - laser, has a relatively serious speckle problem, and this method also has the requirement that the light source line should be less than one-tenth of the size feature of the object to be measured, which greatly limits the popularization and application of this method. And microscopic fringe projection profilometry (MFPP), as a branch of microscopic three-dimensional detection technology of fringe projection profilometry, can achieve full-field measurement, with high accuracy and relatively fast speed. This technology can reconstruct the three-dimensional topography of the target object by projecting fringe patterns and capturing deformed fringe images. However, although the microscopic fringe projection technology can obtain the three-dimensional topography data of Bumps, in practical applications, how to efficiently and accurately extract the height and coplanarity parameters of Bumps from the three-dimensional point cloud data is still a technical problem. The existing methods usually rely on complex post-processing algorithms, with low calculation efficiency and being easily affected by noise, and it is difficult to meet the requirements of high-precision and high-efficiency measurement in industrial production.

[0004] After retrieval, the following publicly disclosed prior patent technologies regarding the three-dimensional topography measurement of wafer bumps are found, and these solutions are significantly different from the present invention:

[0005] Patent CN202310916032 proposes a method for three-dimensional topography measurement of wafer bumps based on white light interference. By scanning the wafer surface, spectral brightness information and height information are obtained to generate a three-dimensional point cloud topography map.

[0006] Patent CN202311165674 proposes a method for wafer defect detection based on white light interference and atomic force microscopy. Coarse positioning is performed by white light interference, and then precise measurement is carried out using the atomic force microscopy.

[0007] Patent CN202411882657 proposes a method for three-dimensional topography detection of wafers based on sparse sampling. The three-dimensional topography of the wafer is reconstructed by sparse sampling and the centroid method.

[0008] After comparison, there are significant differences between the method adopted in this patent application and the comparative documents, and the technical problems to be solved are also different. The technical differences between the comparative documents and this patent application are relatively large. The above-mentioned prior art solutions can all effectively obtain high-precision three-dimensional point clouds of wafer bumps, but they cannot efficiently and accurately extract the height and coplanarity parameters of the bumps from the three-dimensional point cloud data. The present invention proposes an automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation. By combining plane fitting methods such as principal component analysis (PCA) and K-Means clustering segmentation technology, it can efficiently and accurately extract the height and coplanarity parameters of the bumps from the three-dimensional point cloud data, solving the problem in the prior art that the height and coplanarity parameters of the bumps cannot be efficiently and accurately extracted from the three-dimensional point cloud data, and having significant technical advantages and application value. Summary of the Invention

[0009] The object of the present invention is to provide an automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation, and through the processing and analysis of three-dimensional point cloud data, realize the automatic and high-precision measurement of the height and coplanarity of wafer bumps.

[0010] The present invention solves its technical problems by adopting the following technical solutions:

[0011] An automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation, characterized in that the steps include:

[0012] S1: Fit the plane l to the obtained 3D point cloud data of wafer bumps through principal component analysis BumpPerform Z - coordinate threshold segmentation in the following way and screen the 3D point cloud data above the fitted plane;

[0013] S2: Perform clustering segmentation on the XY coordinates of the 3D point cloud data screened in S1 to obtain the centroid of the XY coordinates of each bump on the wafer;

[0014] S3: Take the centroid coordinates obtained in S2 as the center of the circle, set the radius r top and the ring radius range, and respectively obtain the XY coordinates of the top and bottom plate point clouds of the wafer bumps for calculating the height;

[0015] S4: Use the XY coordinates of the top and bottom plate point clouds obtained in S3 as indexes to extract the corresponding 3D point cloud data of the bumps, and respectively obtain the top point cloud and the point cloud distributed in the bottom plate ring;

[0016] S5: Fit a plane l bottom according to the point cloud distributed in the bottom plate ring in S4, and calculate the average distance from the top point cloud to the fitted plane l bottom as the height of the wafer bump;

[0017] S6: Calculate the maximum and minimum height differences of the distances from the top point cloud to the fitted plane in S5 as the coplanarity evaluation index of the wafer bumps;

[0018] S7: Fit a plane l top to the top point clouds of all bumps in S4, and calculate the root - mean - square error of the fitted plane l top for supplementary evaluation of the coplanarity of the wafer bumps.

[0019] Furthermore, in S1, use a three - dimensional scanning device to obtain the three - dimensional point cloud data of the measured wafer bumps, and perform plane fitting on the 3D point cloud of the wafer bumps through the principal component analysis method (PCA). The formula for the fitted plane l Bump is:

[0020] Ax + By + Cz + D = 0

[0021] where is the normal vector of the fitted plane, and D is the distance from the origin to the plane;

[0022] Compare each three - dimensional coordinate information of the 3D point cloud of the wafer bumps with the fitted plane l Bump to complete the screening. The screening condition is:

[0023]

[0024] Among them, assuming that the 3D point cloud data Data Bump of the wafer bumps is N - dimensional, (X i , Y i , Zi ) represents the three-dimensional coordinates of the 3D point cloud of the wafer bump, (X i , Y i , Z i ) ∈ Data, i = 1, 2, …, N.

[0025] Further, in S2, in order to reduce the computational complexity of the K-Means clustering algorithm, the XY coordinate information in the 3D point cloud data screened in S1 is extracted to form 2D point cloud data representing the position distribution of the wafer bumps, and the 2D point cloud of the wafer bumps is clustered and segmented. The top point cloud of each bump on the wafer is classified into a cluster, and the centroid coordinates of the corresponding cluster are obtained. Through the clustering effect formed by the segmentation, the accuracy of the centroid of the XY coordinates of the top point cloud of each bump is verified.

[0026] Further, in S3, the centroid coordinates obtained in S2 are used as the position center of each wafer bump. According to the prior information of the wafer bumps, the parameters for point cloud screening are set, specifically: setting the radius r top for obtaining the top point cloud information of the wafer bumps for subsequent height calculation and coplanarity evaluation; setting the ring radius range r small ≤ r bottom ≤ r big for obtaining the bottom point cloud information of the wafer bumps for subsequent height calculation; the point cloud information of the top and bottom of the wafer bumps obtained in step S3 is the XY coordinate information of the wafer bump point cloud.

[0027] Further, in S4, the XY coordinate information of the wafer bumps obtained according to the screening conditions in S3 is used as the index of the 3D point cloud, and the corresponding screened 3D point cloud of the wafer bumps is extracted, namely the top 3D point cloud of the wafer bump and the bottom 3D point cloud of the wafer bump.

[0028] Further, in S5, for each wafer bump at each position, the bottom point cloud with a circular ring shape distribution is plane-fitted l bottom , and the average distance between the top point cloud with a corresponding circular distribution and the fitted plane l bottom is calculated as the height value of the wafer bump;

[0029] Assume that the coordinate of the top point cloud of the circular distribution of the wafer bump is (x i , y i , y i ) ∈ data j , and the formula expression of the fitted plane l bottom is:

[0030]

[0031] where data j represents, (xi , y i , y i ) indicates that is the normal vector of the fitting plane, and D is the distance from the origin to the plane. Then the height calculation formula of the wafer bump is:

[0032]

[0033] Furthermore, in S6, the height values of the wafer bumps at each position calculated in S5 are statistically analyzed to calculate the maximum value and the minimum value of the difference Eval coplanarity , which is used to evaluate the coplanarity of the wafer bumps:

[0034]

[0035] where Max() represents the function to obtain the maximum value, and Min() represents the function to obtain the minimum value.

[0036] Furthermore, in S7, the top point clouds distributed circularly at each position screened in S4 are plane-fitted to obtain the fitting plane l top , and the root mean square error of the fitting plane l top is calculated as a supplement to the evaluation of the coplanarity of the wafer bumps.

[0037] An electronic device, the electronic device includes a memory and a processor. The memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the automatic calculation method of the height and coplanarity of the wafer bumps based on K-Means clustering segmentation described in any one of the above.

[0038] A computer-readable storage medium stores instructions, which, when executed by one or more processors, cause the electronic device to execute the automatic calculation method of the height and coplanarity of the wafer bumps based on K-Means clustering segmentation described in any one of the above.

[0039] The advantages and positive effects of the present invention are:

[0040] (1) The present invention is applicable to the height measurement and coplanarity verification of wafer bumps. The bump point cloud position recognition method proposed based on the classical machine learning algorithm K-Means clustering segmentation can efficiently classify data through unsupervised learning, thereby obtaining high-precision and highly reliable bump position information. At the same time, based on the feature distribution of the wafer bump three-dimensional point cloud and combining the point cloud processing method of PCA plane fitting and K-Means clustering segmentation, it can significantly improve the efficiency of bump position extraction and quickly complete the detection task of large-scale wafer bumps.

[0041] (2) The present invention provides a new point cloud processing method that can efficiently and accurately extract the height and coplanarity parameters of bumps from three-dimensional point cloud data, significantly improving the automation degree of measuring the key parameters of wafer bumps. The precise extraction of wafer bump positions is achieved through two segmentations, which can adapt to wafer bumps with different densities, shapes, and sizes in a die, and has strong versatility and adaptability. By precisely measuring the height and coplanarity of wafer bumps, defects in the manufacturing process can be detected in a timely manner, effectively improving the quality control level of wafer bumps in the semiconductor manufacturing process. Description of the Drawings

[0042] Figure 1 is the algorithm flowchart of the automatic calculation method for the height and coplanarity of wafer bumps in the embodiment of the present invention;

[0043] Figure 2 is the schematic diagram of the height threshold segmentation of the bump point cloud based on PCA plane fitting in the embodiment of the present invention;

[0044] Figure 3 is the effect diagram of the K-Means clustering segmentation of the XY coordinates of the point cloud data in the embodiment of the present invention;

[0045] Figure 4 is the schematic diagram of the effect of the point cloud screening of the top and bottom plates of the wafer bump in the embodiment of the present invention;

[0046] Figure 5 is the effect diagram of the distribution of the screened point cloud at the top of the bump and the fitting plane of the screened point cloud at the bottom plate in the embodiment of the present invention;

[0047] Figure 6 is the measurement result diagram of the Bruker white light interferometer scanning measurement in the embodiment of the present invention;

[0048] Figure 7 is the effect diagram of the plane fitting of the screened point cloud at the top of the bump in the embodiment of the present invention;

[0049] Figure 8It is the result diagram of the calculation of the wafer Bump heights at different heights and the coplanarity evaluation in the embodiments of the present invention. Detailed implementation manners

[0050] In order to make the objectives, technical solutions and advantages of the present application clearer and easier to understand, the present invention will be further described below in conjunction with the accompanying drawings and specific implementation manners. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] The present invention provides an automatic calculation method for wafer Bump height and coplanarity based on K-Means clustering segmentation. The overall algorithm flow is as Figure 1 shown:

[0052] Step 1: Perform Z coordinate threshold segmentation on the obtained wafer Bump three-dimensional (3D) point cloud data by fitting a plane l Bump by means of principal component analysis (PCA) and screen the 3D point cloud data above the fitting plane.

[0053] In the embodiments of the present invention, the three-dimensional topography point cloud 1 of 4 wafer Bumps is obtained through the microscopic fringe projection system structure of "monocular + projector". The wafer Bump is plane-fitted according to PCA. The plane equation of the fitting plane 2 is: 0.0078x - 0.0014y + 1.0000z + 4.6414 = 0, and the effect is as Figure 2 shown. Then, the fitting plane 2 is used as the threshold for the Z coordinate of the wafer Bump to perform segmentation, and the 3D point cloud data of the wafer Bump above the fitting plane 2 is obtained.

[0054] Step 2: Perform clustering segmentation on the XY coordinates of the 3D point cloud data screened in Step 1 to obtain the centroid and cluster of the XY coordinates of each Bump on the wafer.

[0055] In the embodiments of the present invention, the XY coordinates of the 3D point cloud data above the fitting plane 2 are extracted. The XY coordinate distributions of the 4 wafer Bumps are as Figure 3 shown. It can be seen that the point cloud distribution of each wafer Bump above the fitting plane 2 is relatively concentrated, which facilitates the high-precision implementation of the K-Means-based clustering segmentation algorithm. After simple segmentation and classification by the clustering algorithm, it can be seen that each Bump is well classified into a cluster class, and the centroid coordinates of the corresponding cluster class are obtained: the centroid 401 coordinates of Bump301 are (-0.8094, 4.2705), the centroid 402 coordinates of Bump302 are (-0.8054, 3.6210), the centroid 403 coordinates of Bump303 are (-0.1499, 4.2578), and the centroid 404 coordinates of Bump304 are (-0.1660, 3.6170).

[0056] Step 3: Taking the centroid coordinates obtained in S2 as the centers of circles, set the radius r top and the ring radius range r small ≤r bottom ≤r big , and respectively obtain the XY coordinates of the top and bottom points of the wafer Bump for calculating the height.

[0057] In the embodiment of the present invention, taking the centroid coordinates 401, centroid coordinates 402, centroid coordinates 403, and centroid coordinates 404 as the centers of circles of the screening areas of Bump301, Bump302, Bump303, and Bump304 respectively, set the radius r top = 0.01 mm, the ring radius range 0.27 mm ≤ r bottom ≤ 0.29 mm, and screen the XY coordinates of the point cloud data of the top and bottom of the wafer Bump. The screening schematic diagram is as shown in Figure 4 . 501 represents the XY coordinates of the top point cloud of Bump301 after screening, and 601 represents the XY coordinates of the bottom point cloud of Bump301 after screening; 502 represents the XY coordinates of the top point cloud of Bump302 after screening, and 602 represents the XY coordinates of the bottom point cloud of Bump302 after screening; 503 represents the XY coordinates of the top point cloud of Bump303 after screening, and 603 represents the XY coordinates of the bottom point cloud of Bump303 after screening; 504 represents the XY coordinates of the top point cloud of Bump304 after screening, and 604 represents the XY coordinates of the bottom point cloud of Bump304 after screening.

[0058] Step 4: Taking the coordinates obtained in S3 as indexes, extract the corresponding Bump 3D point cloud data.

[0059] In the embodiment of the present invention, taking the XY coordinates of the screened top and bottom point clouds as indexes, extract the 3D point cloud data at the corresponding positions of the Bump. Figure 4 The Bump 3D point cloud data corresponding to the XY coordinates 501 of the screened top point cloud in Figure 5 is 511 in Figure 4 The Bump 3D point cloud data corresponding to the XY coordinates 601 of the screened bottom point cloud in Figure 5 is 611 in Figure 4 The Bump 3D point cloud data corresponding to the XY coordinates 502 of the screened top point cloud in Figure 5 is 512 in Figure 4 The Bump 3D point cloud data corresponding to the XY coordinates 602 of the screened bottom point cloud in Figure 5 is 612 in Figure 4 The Bump 3D point cloud data corresponding to the XY coordinates 503 of the screened top point cloud in Figure 5 is 513 inFigure 4 The Bump3D point cloud data corresponding to the XY coordinates 603 of the bottom plate point cloud screened in Figure 5 is 613 in Figure 4 The Bump3D point cloud data corresponding to the XY coordinates 504 of the top point cloud screened in Figure 5 is 514 in Figure 4 The Bump3D point cloud data corresponding to the XY coordinates 604 of the bottom plate point cloud screened in Figure 5 is 614 in

[0060] Step 5: Fit the point cloud of the bottom plate ring distribution in S4 to the plane l bottom , and calculate the average distance from the top point cloud to the fitted plane l bottom as the height of the wafer Bump.

[0061] In the embodiment of the present invention, the point cloud 3D data of the screened bottom plate is plane-fitted by using the plane fitting method based on PCA

[0062] The plane equation of the fitted plane 701 is: 0.0113x - 0.0011y + 0.9999z + 4.5713 = 0,

[0063] The plane equation of the fitted plane 702 is: 0.0324x + 0.0018y + 0.9995z + 4.5798 = 0,

[0064] The plane equation of the fitted plane 703 is: 0.0085x + 0.0005y + 1.0000z + 4.5528 = 0,

[0065] The plane equation of the fitted plane 704 is: 0.0055x + 0.0007y + 1.0000z + 4.5549 = 0.

[0066] The average distance from the screened top point cloud 511 to the fitted plane 701 can be calculated through the height calculation formula of the wafer Bump The average distance from the screened top point cloud 512 to the fitted plane 702 The average distance from the screened top point cloud 513 to the fitted plane 703 The average distance from the screened top point cloud 514 to the fitted plane 704

[0067] To verify the high measurement accuracy of the wafer Bump point cloud and the effectiveness of the calculation method proposed in the present invention, the wafer Bump height value measured by the Bruker white light interferometer is used as the true value and compared with the calculated Bump height value, as Figure 6 shown. The corresponding true value 801 is 0.2870 mm, The corresponding true value 802 is 0.2870 mm, The corresponding true value 803 is 0.2841 mm, The corresponding true value 804 is 0.2874 mm. The average error is 4.6 μm, which proves the high measurement accuracy of the wafer Bump point cloud and the effectiveness of the calculation method proposed in the present invention.

[0068] Step 6: Calculate the maximum and minimum height differences in S5 as the coplanarity evaluation index of the wafer Bump.

[0069] In the embodiment of the present invention, for the coplanarity evaluation of the wafer Bump, the obtained wafer Bump height values are statistically calculated, and the difference between the maximum and minimum height values is calculated for coplanarity evaluation: 294.3 μm - 287.3 μm = 7.0 μm.

[0070] Step 7: Fit the top point cloud of all Bumps in S4 to a plane l top , and calculate the root mean square error (RMSE) of the fitted plane l top for supplementary evaluation of the coplanarity of the wafer Bump.

[0071] In the embodiment of the present invention, the top point cloud of the screened Bumps is also subjected to plane fitting, and the root mean square error of the fitted plane 9 is calculated to evaluate the coplanarity of 4 wafer Bumps, as Figure 7 shown. The equation of the fitted plane 9 is: 0.0068x + 0.0001y + 1.0000z - 4.8470 = 0, and the root mean square error is 10.2 μm.

[0072] The height calculation and coplanarity evaluation of the 3D point clouds of wafer Bumps with other 3 different heights are carried out by using the method of the present invention patent, and the processing results are as Figure 8 shown. The experimental results prove that the method of the present invention patent can efficiently and accurately extract the height and coplanarity parameters of the wafer Bump by combining the PCA plane fitting method and the Bump point cloud position recognition method based on K-Means clustering segmentation, and realizes the automatic measurement of the key parameters of the wafer Bump. The method of the present invention patent not only improves the measurement accuracy and efficiency, but also enhances the adaptability and reliability, and can effectively meet the high requirements for wafer Bump quality control in the semiconductor manufacturing process, and has significant technical advantages and application value.

[0073] This application also provides an electronic device, which may include: at least one processor and at least one memory.

[0074] Among them, the processor may include one or more processing cores. The processor uses various interfaces and circuits to connect various parts within the entire server, and by running or executing instructions, programs, code sets, or instruction sets stored in the memory, and by calling the data stored in the memory, it executes various functions of the server and processes data.

[0075] Optionally, the processor may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor may integrate one or a combination of several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor and may be implemented separately by a single chip.

[0076] Among them, the memory may include random access memory (RAM) and may also include read-only memory. Optionally, the memory includes a non-transitory computer-readable storage medium. The memory can be used to store instructions, programs, code, code sets, or instruction sets. The memory may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store the data involved in the above-mentioned method embodiments. Optionally, the memory may also be at least one storage device located far from the aforementioned processor. The memory, as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program.

[0077] This application also provides a computer-readable storage medium storing instructions. When executed by one or more processors, it causes the electronic device to execute the method as described in one or more of the above embodiments.

[0078] In addition, each functional unit in the present application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0079] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. The aforementioned memory includes various media that can store program codes, such as USB flash drives, mobile hard disks, magnetic disks, or optical discs.

[0080] The above-described embodiments only represent the specific implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. An automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation, characterized in that: steps Including: S1: Perform Z - coordinate threshold segmentation on the obtained 3D point cloud data of the wafer bumps by fitting a plane l through principal component analysis, and screen the 3D point cloud data above the fitted plane; Bump ​ S2: Cluster and segment the XY coordinates of the 3D point cloud data screened in S1 to obtain the centroid of the XY coordinates of each bump on the wafer. S3: Taking each centroid coordinate obtained in S2 as the center of a circle, set the radius r top and the range of the ring radius, respectively obtain the XY coordinates of the top and bottom plate point clouds of the wafer bumps for calculating the height; S4: Use the XY coordinates of the top and bottom plate point clouds obtained in S3 as indices to extract the corresponding 3D point cloud data of the bumps, and respectively obtain the top point cloud and the point cloud distributed in a circular ring at the bottom plate. S5: Fit the plane l based on the point cloud distributed on the bottom plate ring in S4 bottom , calculate the average distance from the top point cloud to the fitted plane l bottom as the height of the wafer bump; S6: Calculate the maximum and minimum height differences between the distances from the top point cloud in S5 to the fitting plane, and use them as the coplanarity evaluation index of the wafer bumps. S7: Fit the top point cloud of all the convex points in S4 to a plane l top , and calculate the root mean square error of the fitted plane l top for supplementary evaluation of the coplanarity of wafer bumps.

2. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, characterized in that: In the above S1, three-dimensional point cloud data of the bumps on the wafer to be measured is obtained by using a three-dimensional scanning device, and plane fitting is performed on the 3D point cloud of the wafer bumps through the principal component analysis method, and the fitting plane l Bump The formula is: Ax + By + Cz + D = 0 where, is the normal vector of the fitting plane, and D is the distance from the origin to the plane; Compare each three-dimensional coordinate information of the wafer bump 3D point cloud with the fitting plane l Bump to complete the screening. The screening conditions are as follows: Among them, assume that the 3D point cloud data Data of the wafer bump Bump is N-dimensional, and (X i , Y i , Z i ) represents the three-dimensional coordinates of the 3D point cloud of the wafer bump, (X i , Y i , Z i ) ∈ Data, where i = 1, 2, …, N.

3. The automatic calculation method for wafer bump height and coplanarity based on K-Means clustering segmentation according to claim 1, characterized in that: In S2, in order to reduce the computational complexity of the K-Means clustering algorithm, the XY coordinate information in the 3D point cloud data screened in S1 is extracted to form 2D point cloud data representing the position distribution of the wafer bumps, and the 2D point cloud of the wafer bumps is clustered and segmented. The top point cloud of each bump on the wafer is grouped into a cluster, and the centroid coordinates of the corresponding cluster are obtained. Through the clustering effect formed by the segmentation, the accuracy of the centroid of the XY coordinates of the top point cloud of each bump is verified.

4. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, characterized in that: In S3, the centroid coordinates obtained in S2 are used as the position center of each wafer bump. According to the prior information of the wafer bumps, the parameters for point cloud screening are set, specifically: setting the radius r top for obtaining the top point cloud information of the wafer bumps for subsequent height calculation and coplanarity evaluation; setting the ring radius range r small ≤ r bottom ≤ r big for obtaining the bottom point cloud information of the wafer bumps for subsequent height calculation; the point cloud information of the top and bottom of the wafer bumps obtained in step S3 is the XY coordinate information of the wafer bump point cloud.

5. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, wherein: In S4, the XY coordinate information of the wafer bumps obtained according to the screening conditions in S3 is used as the index of the 3D point cloud to extract the corresponding 3D point cloud of the screened wafer bumps, namely the 3D point cloud of the top of the wafer bumps and the 3D point cloud of the bottom plate of the wafer bumps.

6. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, wherein: In S5, for each wafer bump at a position, perform plane fitting on the floor point cloud distributed in a circular ring shape bottom , and calculate the average distance between the top point cloud distributed in a corresponding circle and the fitting plane bottom , which is used as the height value of the wafer bump; Assume that the top point cloud coordinates of the circular distribution of wafer bumps are (x i , y i , y i ) ∈ data j , and the formula for fitting the plane l bottom is expressed as: Among them, data j indicates that (x i , y i , y i ) indicates that is the normal vector of the fitting plane, D is the distance from the origin to the plane, and the height calculation formula of the wafer bump is:

7. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, characterized in that: In S6, the height values of the wafer bumps at each position calculated in S5 are statistically analyzed to calculate the maximum value of the bump height values and the minimum value of the difference Eval coplanarity , which is used to evaluate the coplanarity of the wafer bumps: Among them, Max() represents the function for obtaining the maximum value, and Min() represents the function for obtaining the minimum value.

8. The automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to claim 1, characterized in that: In S7, for the top point clouds distributed circularly at each position screened in S4, plane fitting is performed to obtain a fitting plane l top , and the root mean square error of the fitting plane l top is calculated as a supplement to the coplanarity evaluation of the wafer bumps.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by one or more processors, cause the electronic device to execute the automatic calculation method for the height and coplanarity of wafer bumps based on K-Means clustering segmentation according to any one of claims 1 to 8.

Images