Wafer surface warpage detection method and device based on automatic reference plane fitting
By acquiring 3D point cloud data through structured light and performing automatic reference plane fitting and filtering, the problem of expensive equipment and complex operation is solved, realizing low-cost, high-precision wafer warpage detection, which is suitable for small and medium-sized enterprises.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN TECH UNIV
- Filing Date
- 2026-03-16
- Publication Date
- 2026-07-24
Smart Images
Figure CN122448104A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of image processing technology, and in particular to a method and apparatus for detecting wafer surface warpage based on automatic reference plane fitting. Background Technology
[0002] Currently, wafer surface warpage inspection equipment on the market generally uses high-end industrial instruments. However, these devices are not only expensive (for example, a single inspection machine from leading machine vision companies such as Keyence and Optoelectronics can cost several million yuan), but also complex to operate, typically requiring professional engineers for maintenance and operation. This makes it difficult for small and medium-sized enterprises to afford the purchase and maintenance costs of such equipment in their production. To improve this situation, our organization has developed a low-cost, fast, and easy-to-operate wafer surface warpage inspection method and system based on structured light technology.
[0003] In this detection process, after acquiring three-dimensional point cloud data, a plane reference surface is fitted on the wafer surface by manually selecting points. However, the measurement results obtained by this method deviate significantly from the actual results. Summary of the Invention
[0004] This invention aims to at least partially solve one of the technical problems in related technologies. Therefore, one object of this invention is to provide a method and apparatus for detecting wafer surface warpage based on automatic reference plane fitting, which can detect wafer warpage using a low-cost method while ensuring detection accuracy, and is easy to operate.
[0005] The technical solution adopted in this invention is: In a first aspect, the present invention provides a method for detecting the warpage of a wafer surface, comprising: acquiring three-dimensional point cloud data of the surface of the wafer under test based on structured light; performing data preprocessing on the three-dimensional point cloud data; reconstructing the planar reference plane of the three-dimensional point cloud data using a non-collinear triangle point standard; performing data processing on the three-dimensional point cloud data using statistical filtering; and solving the surface warpage of the wafer under test based on the filtered point cloud data.
[0006] The planar reference plane for reconstructing the 3D point cloud data using a non-collinear triangle point standard includes: dividing the wafer surface formed by the 3D point cloud data into N concentric circles with equal radius differences, denoting the radius of the largest concentric circle as R, where N>=7; and within a radius of... Three points with a preset angle distribution are selected on the concentric circles; a plane fitting is performed based on these three points, and the fitted plane is used as the plane reference plane.
[0007] The method of acquiring three-dimensional point cloud data of the wafer surface based on structured light includes: acquiring a two-dimensional image of light projected onto the wafer surface, the light including four optical patterns with a phase difference of 90 degrees; and converting the two-dimensional image into three-dimensional point cloud data using a four-step phase shifting method.
[0008] The data preprocessing of the 3D point cloud data includes: using low-pass filtering to preprocess the 3D point cloud data.
[0009] The method of calculating the surface warpage of the wafer under test based on filtered point cloud data includes: aligning the Z-axis of the filtered point cloud data to the normal direction of the reference plane; obtaining the Z-axis value of the filtered point cloud data, which is the height distribution of the wafer surface under test; subtracting the minimum value from the maximum value of the Z-axis value to obtain the maximum deformation of the wafer surface in the normal direction; and dividing the maximum deformation by the reference length of the wafer under test to calculate the surface warpage of the wafer under test.
[0010] The method involves using statistical filtering to process the 3D point cloud data, including: calculating the distance distribution characteristics of points in the neighborhood of each point; and filtering outliers based on a preset standard deviation multiple.
[0011] Secondly, the present invention provides a wafer surface warpage detection device, comprising: a three-dimensional point cloud data acquisition module for acquiring three-dimensional point cloud data of the wafer surface under test based on structured light; a three-dimensional point cloud data preprocessing module for preprocessing the three-dimensional point cloud data; a planar reference plane reconstruction module for reconstructing the planar reference plane of the three-dimensional point cloud data using a non-collinear triangle point standard; a three-dimensional point cloud data filtering module for processing the three-dimensional point cloud data using statistical filtering; and a wafer surface warpage solving module for solving the surface warpage of the wafer under test based on the filtered point cloud data.
[0012] The planar reference plane reconstruction module includes a planar partitioning unit, a point selection unit, and a planar reference plane fitting unit. The planar partitioning unit divides the wafer surface formed from the 3D point cloud data into N concentric circles with equal radius differences, denoted as R for the largest concentric circle, where N>=7. The point selection unit is used to select points on the wafer surface with a radius of R. Three points with a preset angle distribution are selected on the concentric circles; the plane reference plane fitting unit is used to perform plane fitting based on the three points, and the fitted plane is used as the plane reference plane.
[0013] The 3D point cloud data filtering module includes a distance distribution calculation unit and an outlier filtering unit. The distance distribution calculation unit is used to calculate the distance distribution characteristics of points in the neighboring region for each point. The outlier filtering unit is used to filter out outliers according to a preset standard deviation multiple.
[0014] The wafer surface warpage calculation module includes a Z-axis alignment unit, a Z-axis value acquisition unit, and a wafer surface warpage calculation unit. The Z-axis alignment unit is used to align the Z-axis of the filtered point cloud data to the normal direction of the plane reference plane. The Z-axis value acquisition unit is used to acquire the Z-axis value of the filtered point cloud data, which is the height distribution of the tested wafer surface. The wafer surface warpage calculation unit is used to subtract the minimum value from the maximum value of the Z-axis values to obtain the maximum deformation of the wafer surface in the normal direction, and divide the maximum deformation by the reference length of the tested wafer to calculate the surface warpage of the tested wafer.
[0015] The beneficial effects of this invention are: This invention constructs a low-cost method for detecting wafer surface warpage. After acquiring 3D point cloud data, this method reconstructs the planar reference plane of the 3D point cloud data using a non-collinear triangle point standard, and filters the point cloud data to remove defective data, thereby ensuring detection accuracy. Furthermore, this detection method can automatically perform the detection after the initial test parameters are calibrated, making it convenient for users.
[0016] Furthermore, this invention divides the wafer surface formed by the three-dimensional point cloud data into multiple concentric circles with a radius of... Three points are selected on a concentric circle to perform plane fitting to generate a plane reference plane, and then the warpage of the wafer is calculated based on this plane reference plane. Practice has proven that this method can improve detection accuracy.
[0017] In addition, the data acquisition and data processing equipment used in the detection system of this invention are commonly used equipment on the market. Compared with high-end imported wafer detection equipment, the hardware cost is significantly reduced, and the total cost is controlled within 1 / 15 of that of imported equipment, which can meet the detection needs of small and medium-sized enterprises. Attached Figure Description
[0018] Figure 1 This is a flowchart illustrating an embodiment of the wafer surface warpage detection method based on automatic reference plane fitting of the present invention; Figure 2 yes Figure 1 A flowchart illustrating an embodiment of step S11; Figure 3 (a) is Figure 2 A schematic diagram of the structure of an embodiment of light in step S111; Figure 3 (b) is Figure 2 A schematic diagram of another embodiment of light in step S111; Figure 4 yes Figure 2 A schematic diagram of the structure of an embodiment of the two-dimensional image in step S111; Figure 5 yes Figure 2 A schematic diagram of the structure of an embodiment of the coordinate system mapping diagram in step S112; Figure 6 yes Figure 1 A schematic diagram of the structure of an embodiment of step S12; Figure 7 yes Figure 1 A flowchart illustrating an embodiment of step S13; Figure 8 yes Figure 7 A schematic diagram of the structure of one embodiment; Figure 9 yes Figure 1 A flowchart illustrating an embodiment of step S14; Figure 10 yes Figure 1 A flowchart illustrating an embodiment of step S15; Figure 11 This is a schematic diagram of an embodiment of the wafer surface warpage detection device based on automatic reference plane fitting of the present invention; Figure 12 yes Figure 11 A schematic diagram of the structure of an embodiment of the three-dimensional point cloud data acquisition module 11; Figure 13 yes Figure 11 A schematic diagram of an embodiment of the planar reference plane reconstruction module 13; Figure 14 yes Figure 11 A schematic diagram of an embodiment of the three-dimensional point cloud data filtering module 14; Figure 15 yes Figure 11 A schematic diagram of an embodiment of the wafer surface warp solving module 15; Figure 16 This is a schematic diagram of an embodiment of the wafer surface warpage detection system based on automatic reference plane fitting of the present invention. Detailed Implementation
[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.
[0020] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the wafer surface warpage detection method based on automatic reference plane fitting according to the present invention. Figure 1 As shown, the method includes the following steps: S11: Acquire three-dimensional point cloud data of the surface of the wafer under test based on structured light; Please see Figure 2 Step S11 includes the following steps: S111: Acquire a two-dimensional image of light projected onto the surface of the wafer under test, the light comprising four optical patterns with a 90-degree phase difference; Please see Figure 3 The optical pattern comprises two sets of orthogonal stripes, one horizontal and one vertical. For example... Figure 3 As shown, the horizontal stripe group includes four horizontal stripe patterns with a phase difference of 90 degrees, and the vertical stripe group includes four vertical stripe patterns with a phase difference of 90 degrees.
[0021] The light pattern is sent to a display device, which projects the light pattern onto the wafer surface. A CCD camera captures a two-dimensional image of the light projected onto the wafer surface, such as... Figure 4 As shown, the two-dimensional image is a deformed stripe. The first image is an image with a phase of 0°, the second image is an image with a phase of 90°, the third image is an image with a phase of 180°, and the fourth image is an image with a phase of 270°.
[0022] S112: The two-dimensional image is converted into three-dimensional point cloud data using a four-step phase-shifting method.
[0023] In this step, the phase information in the deformed stripe is demodulated using phase analysis software, and the three-dimensional point cloud data of the wafer surface under test is obtained according to the system structure parameters calibrated by the software.
[0024] The calculation formula for the four-step phase shift method is as follows: ; Where I(x,y) is the light intensity value captured by the camera at pixel (x,y), A is the background light intensity, i.e., the light intensity produced by ambient light and other non-light sources, and B is the modulated light intensity, representing the light intensity. This is the phase value that this study aims to measure, and it is related to the warp height of the object's surface. The phase shift quantities introduced artificially are 0, π / 2, π, and 3π / 2, respectively, in the four-step phase shift method.
[0025] The formula for the light intensity value captured at pixel (x,y) in these four images is as follows: ; ; ; ; I0, I1, I2, and I3 are formulas for the light intensity values captured at pixel (x,y) with phase differences of 0, π / 2, π, and 3π / 2.
[0026] From the above derivation, we can conclude that: .
[0027] This formula eliminates the influence of background light A and modulation amplitude B, and directly obtains the phase φ. However, this phase is a wrapped phase, which needs to be unwrapped.
[0028] In optical 3D measurement systems, phase unwrapping is a crucial step in converting a wrapped phase into a continuous absolute phase. In practical applications, high-quality phase unwrapping is essential for improving the accuracy of 3D reconstruction. This invention employs a phase unwrapping method based on complementary Gray codes to convert the wrapped phase into a continuous phase, providing a reliable guarantee for the accurate measurement of wafer warpage.
[0029] This step requires calibrating the camera and projector. It involves determining the camera's internal and external parameters to establish pixel coordinates (u, v), image coordinates (x, y), and camera coordinates (X, y). c ,Y c Z c ) and world coordinates (X) w ,Y w Z w The mapping relationship between the coordinate systems is shown in the following example: [Image of a coordinate system mapping diagram]. Figure 5 .
[0030] S12: Perform data preprocessing on the 3D point cloud data; Low-pass filtering was used to preprocess the 3D point cloud data to remove edge outliers and reduce interference.
[0031] In this step, key parameters are adjusted based on the point cloud noise characteristics: neighborhood radius (determining the scale of local features) and standard deviation coefficient (controlling the filtering intensity). The noise reduction effect is verified through real-time preview to avoid over-filtering that could lead to feature loss. Figure 6 As shown, when using this filter, you need to set the Radius filter radius (hereinafter referred to as R value). The recommended range is 0.05mm–0.15mm. A radius that is too large will cause loss of detail, while a radius that is too small will not effectively smooth noise. Optionally, you can start with an R value of 0.15mm and gradually reduce it to 0.05mm to observe the effect, and finally select the optimal R value.
[0032] S13: Reconstruct the planar reference plane of the 3D point cloud data using the standard of non-collinear triangle points; Please see Figure 7 Step S13 includes the following sub-steps: S131: Divide the wafer surface formed by the three-dimensional point cloud data into N concentric circles with equal radius differences, and denote the radius of the largest concentric circle as R, where N>=7; The height difference between each point in this 3D point cloud data is very small, therefore this 3D point cloud data can approximate a wafer surface. For example... Figure 8 As shown, the radius of the wafer surface formed by the three-dimensional point cloud data is divided into 7 equal segments, and the surface of the wafer under test is divided into 7 concentric circles.
[0033] S132: Select three points with a preset angle distribution on one of the concentric circles; Preferably, the radius of the concentric circles is [missing information]. The circle. The preset angle is such that the angle between the straight line formed by the three points and the center of the circle is 120 degrees.
[0034] like Figure 8 As shown, three points are selected on a concentric circle with a radius of 6 / 7R.
[0035] S133: Perform plane fitting based on these three points, and use the fitted plane as the plane reference plane.
[0036] Optionally, the three points can be fitted to a plane using the least squares method, and the fitted plane can be used as a reference plane.
[0037] S14: Statistical filtering is used to process the 3D point cloud data; Statistical filtering is a filtering method based on the statistical characteristics of local neighborhoods in point clouds. It is primarily used to remove noise points from point cloud data while maintaining surface smoothness and preserving the integrity of point cloud edges. The core principle is to statistically analyze the distance distribution between each point in the point cloud and the points within a defined neighborhood, specifying a threshold to effectively identify and distinguish outliers deviating from the normal distribution. The principle of statistical filtering involves calculating the average distance from each point to its n nearest neighbors and calculating its standard deviation. The average distance of the global point cloud is calculated, and a standard deviation coefficient is set. Multiplying the standard deviation of the global point cloud by this coefficient and adding it to the average distance of the global point cloud yields a distance threshold. By changing the magnitude of the standard deviation coefficient, the distance threshold can be controlled, thereby controlling the filtering strength of the statistical filter.
[0038] A statistical filtering algorithm based on the average distance between points in the k-nearest neighbor neighborhood of a sampling point can be used to remove outliers, gross errors caused by measurement mistakes, and sparse points. In point clouds, the denser the points in a given space, the greater the amount of information they contain. Since scattered noise points do not have obvious distribution characteristics and their information content is relatively small, filtering can be based on point density.
[0039] Please see Figure 9 Step S14 includes the following sub-steps: S141: Calculate the distance distribution characteristics of points in the neighboring region for each point; S142: Filter outliers based on a preset standard deviation multiple.
[0040] For example, calculate the average distance and standard deviation between neighboring points and the center point. If the average distance between a point and its neighboring points exceeds Sigma times the global average distance, it is determined to be an outlier.
[0041] In this process, the noise reduction intensity and data fidelity are balanced by adjusting the number of neighborhood points (K value) and the Sigma threshold, so as to avoid over-filtering of effective data or residual noise, and to carry out iterative optimization.
[0042] S15: Solve the surface warpage of the tested wafer based on the filtered point cloud data.
[0043] Please see Figure 10 Step S15 includes the following sub-steps: S151: Align the Z-axis of the filtered point cloud data to the normal direction of the plane reference plane; S152: Obtain the Z-axis value of the filtered point cloud data, where the Z-axis value is the height distribution of the surface of the wafer being measured. S153: Subtract the smallest value from the largest value in the Z-axis range to obtain the maximum deformation of the wafer surface in the normal direction. Divide the maximum deformation by the reference length of the wafer under test to calculate the surface warpage of the wafer under test.
[0044] The formula for calculating wafer warpage is: ,in, This is the maximum deformation of the wafer surface in the normal direction (i.e., the height difference between the highest and lowest points). This is the reference length for the wafer, usually taken as the diagonal length or diameter of the wafer.
[0045] For example, if the largest value in the Z-axis (highest point of the Z-axis) is 0.0589 mm and the smallest value in the Z-axis (lowest point of the Z-axis) is -0.0126 mm, and the reference length of the wafer being measured is 120 mm, then the wafer warpage is 0.00596%.
[0046] In this embodiment, step S11 is completed using FMS software (i.e., Flatness Measurement System software), and steps S12 to S15 are completed using CloudCompare software.
[0047] Example 2 Please see Figure 11 , Figure 11This is a schematic diagram of an embodiment of the wafer surface warpage detection device based on automatic reference plane fitting of the present invention. Figure 11 As shown, the device includes a three-dimensional point cloud data acquisition module 11, a three-dimensional point cloud data preprocessing module 12, a planar reference plane reconstruction module 13, a three-dimensional point cloud data filtering module 14, and a wafer surface warpage calculation module 15.
[0048] The 3D point cloud data acquisition module 11 is used to acquire 3D point cloud data of the surface of the wafer under test based on structured light. For example... Figure 12 As shown, the 3D point cloud data acquisition module 11 includes an image acquisition unit 111 and an image processing unit 112. The image acquisition unit 111 is used to acquire a two-dimensional image of light projected onto the surface of the wafer under test, the light including four optical patterns with a phase difference of 90 degrees. The image processing unit 112 is used to convert the two-dimensional image into 3D point cloud data using a four-step phase shifting method.
[0049] The 3D point cloud data preprocessing module 12 is used to preprocess the 3D point cloud data.
[0050] The planar reference plane reconstruction module 13 is used to reconstruct the planar reference plane of the 3D point cloud data using a non-collinear triangle point standard. For example... Figure 13 As shown, the planar reference plane reconstruction module 13 includes a planar partitioning unit 131, a point selection unit 132, and a planar reference plane fitting unit 133. The planar partitioning unit 131 is used to divide the wafer surface formed by the 3D point cloud data into N concentric circles with equal radius differences, where the radius of the largest concentric circle is denoted as R, and N>=7. The point selection unit 132 is used to select points on the wafer surface with a radius of R. Three points with a preset angle distribution are selected on the concentric circles. The plane reference plane fitting unit 133 is used to perform plane fitting based on these three points, and the fitted plane is used as the plane reference plane.
[0051] The 3D point cloud data filtering module 14 is used to process the 3D point cloud data using statistical filtering. For example... Figure 14 As shown, the 3D point cloud data filtering module 14 includes a distance distribution calculation unit 141 and an outlier filtering unit 142. The distance distribution calculation unit 141 is used to calculate the distance distribution characteristics of points in its neighboring region for each point. The outlier filtering unit 142 is used to filter outliers according to a preset standard deviation multiple.
[0052] The wafer surface warpage calculation module 15 is used to calculate the surface warpage of the tested wafer based on filtered point cloud data. For example... Figure 15As shown, the wafer surface warpage calculation module 15 includes a Z-axis alignment unit 151, a Z-axis value acquisition unit 152, and a wafer surface warpage calculation unit 153. The Z-axis alignment unit 151 aligns the Z-axis of the filtered point cloud data to the normal direction of the plane reference surface. The Z-axis value acquisition unit 152 acquires the Z-axis value of the filtered point cloud data, which represents the height distribution of the measured wafer surface. The wafer surface warpage calculation unit 153 subtracts the minimum Z-axis value from the maximum Z-axis value to obtain the maximum deformation of the wafer surface in the normal direction, and divides this maximum deformation by the reference length of the measured wafer to calculate the surface warpage of the measured wafer.
[0053] Example 3 Please see Figure 16 , Figure 16 This is a schematic diagram of an embodiment of the wafer surface warpage detection system based on automatic reference plane fitting of the present invention. Figure 16 As shown, the system includes a sample platform, projection equipment, camera, processor, and computer. In addition, the system also includes a flat beam splitter, light guide plate, three-dimensional movable platform, constant temperature and humidity system, and light shield.
[0054] The sample platform is used to place the wafer under test. This sample platform, mounted on the three-dimensional movable platform, is used to adjust the height and planar coordinate position of the wafer under test.
[0055] The projection device is used to project light onto the surface of the wafer being tested. The projection device can be a display screen or a projector. An AOC projection display can be used as the projection device.
[0056] The camera is used to capture a two-dimensional image of the light on the surface of the wafer being measured. Preferably, two cameras are used. Basler industrial cameras can be selected as such.
[0057] The computer includes a processor for performing the method as described in Embodiment 1. Preferably, the computer is a desktop computer.
[0058] A flat beam splitter is used to collimate the diverging beam projected by the projection device.
[0059] Light guide plates are used to make the brightness of the projected light uniform.
[0060] The constant temperature and humidity system is used to regulate the ambient temperature and humidity of the detection system.
[0061] The light shield is used to enclose the detection system inside the light shield to block the interference of ambient light on the detection.
[0062] Example 4 The present invention also provides a computer-readable storage medium storing computer-executable instructions for causing a computer to perform the method described in Embodiment 1.
[0063] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.
Claims
1. A method for detecting wafer surface warpage based on automatic reference plane fitting, characterized in that, include: Three-dimensional point cloud data of the surface of the wafer under test is obtained based on structured light; The three-dimensional point cloud data is preprocessed. The planar reference plane of the three-dimensional point cloud data is reconstructed using a non-collinear triangle point standard. The three-dimensional point cloud data is processed using statistical filtering. The surface warpage of the tested wafer is calculated based on the filtered point cloud data.
2. The wafer surface warpage detection method according to claim 1, characterized in that, The planar reference plane for reconstructing the 3D point cloud data using a non-collinear triangle point standard includes: The wafer surface formed by the three-dimensional point cloud data is divided into N concentric circles with equal radius differences. The radius of the largest concentric circle is denoted as R, where N>=7. At a radius of Three points with a preset angle distribution are selected on the concentric circles; Based on the three points, a plane is fitted, and the fitted plane is used as the plane reference plane.
3. The wafer surface warpage detection method according to claim 1, characterized in that, The acquisition of three-dimensional point cloud data of the surface of the wafer under test based on structured light includes: Acquire a two-dimensional image of light projected onto the surface of the wafer under test, wherein the light comprises four optical patterns with a phase difference of 90 degrees; The two-dimensional image is converted into three-dimensional point cloud data using a four-step phase-shifting method.
4. The wafer surface warpage detection method according to claim 1, characterized in that, The data preprocessing of the 3D point cloud data includes: The three-dimensional point cloud data is preprocessed using a low-pass filter.
5. The wafer surface warpage detection method according to claim 1, characterized in that, The process of determining the surface warpage of the tested wafer based on the filtered point cloud data includes: Align the Z-axis of the filtered point cloud data with the normal direction of the plane reference plane; Obtain the Z-axis value of the filtered point cloud data, where the Z-axis value is the height distribution of the surface of the measured wafer. The maximum deformation of the wafer surface in the normal direction is obtained by subtracting the minimum value from the maximum value in the Z-axis values. The surface warpage of the wafer is calculated by dividing the maximum deformation by the reference length of the wafer under test.
6. The wafer surface warpage detection method according to claim 1, characterized in that, The statistical filtering process applied to the 3D point cloud data includes: Calculate the distance distribution characteristics of points within its neighboring region for each point; Outliers are filtered based on a preset standard deviation multiple.
7. A wafer surface warpage detection device based on automatic reference plane fitting, characterized in that, include: The 3D point cloud data acquisition module is used to acquire 3D point cloud data of the surface of the wafer under test based on structured light. A 3D point cloud data preprocessing module is used to preprocess the 3D point cloud data; The planar reference plane reconstruction module is used to reconstruct the planar reference plane of the three-dimensional point cloud data using a non-collinear triangle point standard. A 3D point cloud data filtering module is used to process the 3D point cloud data using statistical filtering. The wafer surface warpage calculation module is used to calculate the surface warpage of the tested wafer based on filtered point cloud data.
8. The wafer surface warpage detection device according to claim 7, characterized in that, The planar reference plane reconstruction module includes a planar partitioning unit, a point selection unit, and a planar reference plane fitting unit. The planar partitioning unit is used to divide the wafer surface formed by the 3D point cloud data into N concentric circles with equal radius differences, and the radius of the largest concentric circle is denoted as R, where N>=7; the point selection unit is used to select points on the wafer surface with a radius of R. Three points with a preset angle distribution are selected on the concentric circles; the plane reference plane fitting unit is used to perform plane fitting based on the three points, and the fitted plane is used as the plane reference plane.
9. The wafer surface warpage detection device according to claim 7, characterized in that, The 3D point cloud data filtering module includes a distance distribution calculation unit and an outlier filtering unit. The distance distribution calculation unit is used to calculate the distance distribution characteristics of points in its neighboring region for each point. The outlier filtering unit is used to filter out outliers according to a preset standard deviation multiple.
10. The wafer surface warpage detection device according to claim 7, characterized in that, The wafer surface warpage calculation module includes a Z-axis alignment unit, a Z-axis value acquisition unit, and a wafer surface warpage calculation unit. The Z-axis alignment unit is used to align the Z-axis of the filtered point cloud data to the normal direction of the plane reference plane. The Z-axis value acquisition unit is used to acquire the Z-axis value of the filtered point cloud data, which is the height distribution of the tested wafer surface. The wafer surface warpage calculation unit is used to subtract the minimum value from the maximum value of the Z-axis values to obtain the maximum deformation of the wafer surface in the normal direction, and divide the maximum deformation by the reference length of the tested wafer to calculate the surface warpage of the tested wafer.