Wafer surface defect detection method, system and equipment and storage medium

Abstract

The invention relates to a wafer surface defect detection method, system and device and a storage medium, and relates to the technical field of semiconductors. The method comprises the following steps: acquiring point cloud data for the surface of a to-be-detected wafer, wherein the point cloud data comprises three-dimensional coordinates of all points on the surface of the to-be-detected wafer; according to the point cloud data, a first normal direction corresponding to each point is calculated, and each first normal direction represents the direction of a normal corresponding to the position where any point in the point cloud data is located; and for each first normal direction, calculating an included angle between the first normal direction and a preset contrast normal direction, and judging whether a defect exists in an area where a position corresponding to the first normal direction is located according to the included angle to obtain a surface defect detection result. The method has low requirements on ambient light, and can effectively improve the accuracy of defect detection.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a wafer surface defect detection method, system, equipment and storage medium. Background Art

[0002] Front-end measurement inspection is one of the key processes in semiconductor wafer manufacturing. Depending on the application technology, the current front-end defect detection mainly includes electron beam inspection and optical inspection. Among them, optical inspection technology is widely used due to its advantages of fast speed, non-contact, non-destructive and easy online integration. Common optical defect detection systems are divided into bright field systems and dark field systems. Bright field and dark field detection play a vital role in chip manufacturing and are an important link in improving product yield, reducing production costs and promoting process iteration. There are large differences between the two in terms of lighting methods, imaging principles, etc. Correspondingly, there are certain differences in technical difficulty. Overall, the bright field system has more stringent requirements on the physical characteristics of the illumination beam, imaging system, signal-to-noise ratio, etc. Generally speaking, the defect detection of non-patterned wafers uses a dark field system, and the defect detection of patterned wafers uses a bright field system or a combination of the two.

[0003] The surface of unpatterned wafers may sometimes be attached with particles, residues, scratches, pits, bumps, etc. These defects will affect the subsequent processing quality and the yield of the product. Therefore, defect detection of wafers is very important. However, in the existing technology, when inspecting wafers, the ambient light or reflections from other materials will affect the accuracy of the inspection results. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a wafer surface defect detection method, system, equipment and storage medium to solve the above technical problems.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A wafer surface defect detection method, comprising: obtaining point cloud data for the surface of a wafer to be detected, the point cloud data including the three-dimensional coordinates of each point on the surface of the wafer to be detected; calculating the first normal direction corresponding to each point according to the point cloud data, each first normal direction representing the direction of the normal corresponding to the position of any point in the point cloud data; for each first normal direction, calculating the angle between the first normal direction and a preset reference normal direction, and judging whether there is a defect in the area where the position corresponding to the first normal direction is located according to the angle, to obtain a surface defect detection result.

[0006] The beneficial effects of the present invention are as follows: based on the point cloud data of the wafer surface to be detected, the present invention obtains the normal direction of the wafer surface corresponding to each position of the wafer surface to be detected, and then calculates the angle between the normal direction of the wafer surface and the normal direction of the standard, and determines whether there is a defect at the position by the angle. The method has low requirements for ambient light and can effectively improve the accuracy of defect detection. In addition, the detection process of the method is simple, which can greatly improve the efficiency of wafer surface defect detection.

[0007] Based on the above technical solution, the present invention can also be improved as follows.

[0008] Furthermore, the method of calculating the first normal direction corresponding to each point based on the point cloud data includes: performing voxel filtering on the point cloud data to generate new point cloud data; and respectively calculating the direction of the normal passing through each point in the new point cloud data to obtain each of the first normal directions.

[0009] Furthermore, the point cloud data is subjected to voxel filtering processing to generate new point cloud data, including: dividing the point cloud space corresponding to the point cloud data into multiple voxels according to a preset voxel size; respectively calculating the voxel center of each voxel; and generating the new point cloud data according to each voxel center.

[0010] Further, for each point in the new point cloud data, the direction of the normal passing through the point is calculated to obtain the first normal direction, including: for each point in the new point cloud data, a plurality of neighborhood points adjacent to the point are obtained from the new point cloud data through a proximity search algorithm; for each point in the new point cloud data, a covariance matrix is ​​calculated according to the three-dimensional coordinates of each of the neighborhood points; for each point in the new point cloud data, the direction of the normal passing through the point is calculated according to the covariance matrix to obtain the first normal direction corresponding to the point.

[0011] Further, for each point in the new point cloud data, the direction of the normal passing through the point is calculated according to the covariance matrix to obtain the first normal direction corresponding to the point, including: for each point in the new point cloud data, calculating the eigenvector of the covariance matrix and the eigenvalues ​​corresponding to each of the eigenvectors; determining the eigenvalue with the smallest value according to each of the eigenvalues; determining the eigenvector corresponding to the eigenvalue with the smallest value as the normal vector; and normalizing the normal vector to obtain the first normal direction.

[0012] Further, for each point in the new point cloud data, the covariance matrix is ​​calculated according to the three-dimensional coordinates of each of the neighborhood points, including: for each point in the new point cloud data, the covariance matrix is ​​calculated by a first formula according to the three-dimensional coordinates of each of the neighborhood points, and the first formula is:

[0013]

[0014] Among them, C is the covariance matrix, k is the number of neighborhood points, and p j is the three-dimensional coordinate of the jth neighborhood point, is the average position coordinate of the neighborhood points.

[0015] Further, judging whether there is a defect in the area where the position corresponding to the first normal direction is located based on the angle includes: judging whether the angle is greater than a preset angle threshold, if so, determining that there is a defect in the area where the position corresponding to the first normal direction is located; otherwise, determining that there is no defect in the area where the position corresponding to the first normal direction is located.

[0016] In order to solve the above technical problems, the present invention also provides a wafer surface defect detection system, comprising:

[0017] A data acquisition module, used to acquire point cloud data for the surface of the wafer to be detected, wherein the point cloud data includes the three-dimensional coordinates of each point on the surface of the wafer to be detected;

[0018] A normal calculation module, used for calculating a first normal direction corresponding to each point according to the point cloud data, wherein each first normal direction represents a direction of a normal corresponding to a position of any point in the point cloud data;

[0019] The defect detection module is used to calculate the angle between the first normal direction and the preset control normal direction for each of the first normal directions, and determine whether there is a defect in the area corresponding to the position of the first normal direction according to the angle, so as to obtain a surface defect detection result.

[0020] To solve the above technical problems, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a wafer surface defect detection method as described above is implemented.

[0021] In order to solve the above technical problems, the present invention also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute a wafer surface defect detection method as described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a schematic diagram of the wafer surface;

[0023] Figure 2 Schematic diagram of particles on the wafer surface;

[0024] Figure 3 A flow chart of a wafer surface defect detection method according to the present invention;

[0025] Figure 4 A schematic diagram of an image acquisition device for a wafer surface defect detection method according to the present invention;

[0026] Figure 5 It is a schematic diagram of a wafer surface defect detection system of the present invention.

[0027] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0028] 1. Laser; 2. Sensor; 3. Rotating table. DETAILED DESCRIPTION

[0029] The principles and features of the present invention are described below. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0030] like Figure 1 and Figure 2 As shown in the figure, the surface of the unpatterned wafer sometimes has some particles, residues, scratches, pits, protrusions, etc. These defects will affect the subsequent processing quality and the product yield. Therefore, defect detection of wafers is very important. Based on this, Figure 3 As shown, this embodiment provides a wafer surface defect detection method, comprising:

[0031] S1. Acquire point cloud data for the surface of a wafer to be inspected, wherein the point cloud data includes the three-dimensional coordinates of each point on the surface of the wafer to be inspected.

[0032] The point cloud data of the wafer surface to be inspected can be obtained in the following way: an image acquisition device is used to acquire an image of the wafer surface to be inspected. The image acquisition device includes a laser 1, a sensor 2 and a rotating table 3, such as Figure 4As shown. The rotating table 3 provides a carrier for the wafer to be inspected, and the laser 1 and the sensor 2 are respectively located above the rotating table 3. When performing image acquisition, the wafer to be inspected is placed on the rotating table 3, and the rotating table 3 drives the wafer to be inspected to rotate clockwise. The laser 1 is used to emit a bright linear stripe to the surface of the wafer to be inspected. The sensor 2 is used for image acquisition, and the main imaging part is the bright stripe emitted by the laser 1. The sensor 2 can use a 2D camera, such as a CCD image collector. The reflected light or scattered light of the wafer is collected by a light collecting system composed of several lenses in the CCD image collector, and then recorded by the CCD (charge coupled device) as a detector. The CCD can detect all spectral lines within a certain wavelength range and can form an image.

[0033] When in use, the rotating table 3 drives the wafer to be inspected to rotate for one circle, and the camera collects linear stripes on the surface of the wafer to be inspected for one circle to image the surface of the wafer. The angular velocity of the rotating table 3 is consistent with the frequency of the laser line emitted by the laser 1. For example, if the frequency of the laser 1 is 360 lines / second, the rotation speed of the rotating table 3 needs to be set to 1 revolution / second.

[0034] Then the collected laser line stripe image is calibrated to determine the position of the wafer on the rotating table 3. A circular calibration plate is used to calibrate the laser 1, the camera attitude, and the rotation of the rotating table 3. Through the calibration process, a mathematical model of the spatial coordinate system is established for the entire system, so that each frame of the collected light stripe image can obtain the actual spatial coordinate value. The actual spatial coordinate system takes the upper surface of the rotating table 3 as the XOY plane, the center of the rotating table 3 as the origin, and the Z axis direction perpendicular to the upper surface of the rotating table 3 upward.

[0035] After the calibration process, the spatial coordinates of the entire wafer surface are obtained, and then a third-party interface, such as PCL's VTK visualization library, is used to make a visualization interface for the spatial coordinates. The VTK visualization library is an open source library used for point cloud visualization and supports multiple point cloud formats. For the point cloud spatial coordinates implemented by the above method, some properties of the point cloud can be set through the VTK library, such as assigning different colors to the point cloud at different point cloud heights, setting the size of the displayed points, and other operations.

[0036] S2. Calculate a first normal direction corresponding to each point according to the point cloud data, where each first normal direction represents a direction of a normal corresponding to a position of any point in the point cloud data.

[0037] Optionally, in an embodiment, calculating the first normal direction corresponding to each point based on the point cloud data includes: performing voxel filtering on the point cloud data to generate new point cloud data; respectively calculating the direction of the normal passing through each point in the new point cloud data to obtain each of the first normal directions.

[0038] Optionally, in an embodiment, the point cloud data is subjected to voxel filtering processing to generate new point cloud data, including: dividing the point cloud space corresponding to the point cloud data into multiple voxels according to a preset voxel size; respectively calculating the voxel center of each voxel; and generating the new point cloud data according to each voxel center.

[0039] Specifically, the voxel size is set according to specific usage requirements, for example, the horizontal interval, the longitudinal interval, and the vertical interval are all set to 0.01 mm. According to the preset voxel size, the point cloud space corresponding to the point cloud data is divided into multiple voxels, that is, the point cloud space corresponding to the point cloud data is divided into multiple voxels of a size equal to the preset voxel size to form a voxel network. And calculate the voxel index corresponding to each point in the point cloud data, so as to clarify the voxel where each point in the point cloud data is located, and the calculation is specifically performed according to the following formula:

[0040]

[0041] Among them, Δx is the horizontal spacing, Δy is the longitudinal spacing, Δz is the vertical spacing, and x i is the horizontal coordinate of the i-th point in the point cloud data, y i is the ordinate of the i-th point in the point cloud data, z i is the vertical coordinate of the i-th point in the point cloud data, i x is the horizontal index of the i-th point in the point cloud data, i y is the vertical index of the i-th point in the point cloud data, i z It is the vertical index of the i-th point in the point cloud data.

[0042] For each voxel, the voxel center of the voxel is calculated, specifically: sum the horizontal coordinates of all points contained in the voxel, divide the result by the number of points contained in the voxel, and obtain the horizontal coordinate of the voxel center of the voxel. Sum the vertical coordinates of all points contained in the voxel, divide the result by the number of points contained in the voxel, and obtain the vertical coordinate of the voxel center of the voxel. Sum the vertical coordinates of all points contained in the voxel, divide the result by the number of points contained in the voxel, and obtain the vertical coordinate of the voxel center of the voxel. Through the above method, the voxel center of each voxel is calculated, and all voxel centers constitute new point cloud data.

[0043] Optionally, in an embodiment, for each point in the new point cloud data, the direction of the normal passing through the point is calculated to obtain the first normal direction, including: for each point in the new point cloud data, obtaining a plurality of neighborhood points adjacent to the point from the new point cloud data through a proximity search algorithm; for each point in the new point cloud data, calculating the covariance matrix based on the three-dimensional coordinates of each of the neighborhood points; for each point in the new point cloud data, calculating the direction of the normal passing through the point based on the covariance matrix to obtain the first normal direction corresponding to the point.

[0044] K-nearest neighbor search is used to find neighboring points. Specifically, the distance between each point in the new point cloud data and the given query point is calculated respectively, and the proximity between data points is measured by distance. The most commonly used distance metric is Euclidean distance. Then, these distances are sorted in ascending order, and the first k points are selected as the neighboring points of the given query point, that is, the points closest to the given query point.

[0045] Optionally, in an embodiment, for each point in the new point cloud data, the direction of the normal passing through the point is calculated according to the covariance matrix to obtain the first normal direction corresponding to the point, including: for each point in the new point cloud data, calculating the eigenvector of the covariance matrix and the eigenvalues ​​corresponding to each of the eigenvectors; determining the eigenvalue with the smallest value according to each of the eigenvalues; determining the eigenvector corresponding to the eigenvalue with the smallest value as the normal vector; and normalizing the normal vector to obtain the first normal direction.

[0046] The normal vector is normalized, that is, the normal vector is divided by the modulus length of the normal vector to obtain a new vector, that is, the first normal direction. Through the normalization process, the lengths of the normal vectors are consistent, but the directions are different.

[0047] Optionally, in an embodiment, for each point in the new point cloud data, calculating the covariance matrix according to the three-dimensional coordinates of each of the neighborhood points includes: for each point in the new point cloud data, calculating the covariance matrix according to the three-dimensional coordinates of each of the neighborhood points by a first formula, wherein the first formula is:

[0048]

[0049] Among them, C is the covariance matrix, k is the number of neighborhood points, and p j is the three-dimensional coordinate of the jth neighborhood point, is the average position coordinate of the neighborhood points.

[0050] The calculation process of the average position coordinate of the neighborhood points is as follows: sum the horizontal coordinates of all the neighborhood points, divide the result by the number of neighborhood points k, and obtain the horizontal coordinate of the average position coordinate of the neighborhood points. Similarly, calculate the vertical coordinate and vertical coordinate of the average position coordinate of the neighborhood points respectively.

[0051] S3. For each of the first normal directions, calculate the angle between the first normal direction and a preset reference normal direction, and determine whether there is a defect in the area corresponding to the position of the first normal direction based on the angle to obtain a surface defect detection result.

[0052] The reference normal direction represents the direction of the normal corresponding to the position where there is no defect on the surface of the wafer to be inspected. In this embodiment, the reference normal direction is (0, 0, 1). The angle is calculated according to the following formula:

[0053] θ=arccos(n·n normal );

[0054] Among them, θ is the angle, n is the first normal direction, n normal is the reference normal direction.

[0055] Optionally, in an embodiment, judging whether there is a defect in the area where the position corresponding to the first normal direction is located based on the angle includes: judging whether the angle is greater than a preset angle threshold, if so, determining that there is a defect in the area where the position corresponding to the first normal direction is located; otherwise, determining that there is no defect in the area where the position corresponding to the first normal direction is located.

[0056] The angle threshold is set according to actual use requirements. The smaller the angle threshold, the smaller the fault tolerance. When the angle is greater than the preset angle threshold, it is determined that there is a defect at the position corresponding to the first normal direction corresponding to the angle, that is, there is a bulge or pit on the surface of the position.

[0057] In summary, this method is based on the point cloud data of the wafer surface to be detected, obtains the normal direction of the wafer surface corresponding to each position on the wafer surface to be detected, and then calculates the angle between the normal direction of the wafer surface and the normal direction of the standard, and judges whether there is a defect at that position by the angle. This method has low requirements for ambient light and can effectively improve the accuracy of defect detection. In addition, the detection process of this method is simple and can greatly improve the efficiency of wafer surface defect detection. This method can be applied to chip packaging, wafer production and other processes to detect potential defective products, control the yield of wafer output, and avoid unpredictable errors in subsequent testing and packaging processes.

[0058] Embodiment 2

[0059] like Figure 5As shown, this embodiment provides a wafer surface defect detection system, including:

[0060] A data acquisition module, used to acquire point cloud data for the surface of the wafer to be detected, wherein the point cloud data includes the three-dimensional coordinates of each point on the surface of the wafer to be detected;

[0061] A normal calculation module, used for calculating a first normal direction corresponding to each point according to the point cloud data, wherein each first normal direction represents a direction of a normal corresponding to a position of any point in the point cloud data;

[0062] The defect detection module is used to calculate the angle between the first normal direction and the preset control normal direction for each of the first normal directions, and determine whether there is a defect in the area corresponding to the position of the first normal direction according to the angle, so as to obtain a surface defect detection result.

[0063] Optionally, in an embodiment, the normal calculation module includes:

[0064] A preprocessing submodule, used for performing voxel filtering on the point cloud data to generate new point cloud data;

[0065] The normal calculation submodule is used to respectively calculate the direction of the normal passing through each point in the new point cloud data to obtain each first normal direction.

[0066] Optionally, in an embodiment, the preprocessing submodule includes:

[0067] A space division unit, used for dividing the point cloud space corresponding to the point cloud data into a plurality of voxels according to a preset voxel size;

[0068] A voxel center calculation unit, used for respectively calculating the voxel center of each voxel;

[0069] A data generating unit is used to generate the new point cloud data according to each of the voxel centers.

[0070] Optionally, in an embodiment, the normal calculation submodule includes:

[0071] A neighborhood point acquisition unit, configured to acquire, for each point in the new point cloud data, a plurality of neighborhood points adjacent to the point from the new point cloud data by using a proximity search algorithm;

[0072] A matrix calculation unit, used for calculating a covariance matrix for each point in the new point cloud data according to the three-dimensional coordinates of each of the neighborhood points;

[0073] The normal calculation unit is used to calculate the direction of the normal passing through each point in the new point cloud data according to the covariance matrix to obtain a first normal direction corresponding to the point.

[0074] Optionally, in an embodiment, the normal calculation unit includes:

[0075] An eigenvalue calculation subunit, used to calculate, for each point in the new point cloud data, an eigenvector of the covariance matrix and an eigenvalue corresponding to each eigenvector;

[0076] An eigenvalue determination subunit, used to determine the eigenvalue with the minimum value according to each of the eigenvalues;

[0077] A normal determination subunit, used for determining the eigenvector corresponding to the eigenvalue with the smallest value as the normal vector;

[0078] A normalization subunit is used to normalize the normal vector to obtain the first normal direction.

[0079] Optionally, in an embodiment, the matrix calculation unit includes:

[0080] The matrix calculation subunit is used to calculate the covariance matrix for each point in the new point cloud data according to the three-dimensional coordinates of each of the neighborhood points by a first formula, wherein the first formula is:

[0081]

[0082] Among them, C is the covariance matrix, k is the number of neighborhood points, and p j is the three-dimensional coordinate of the jth neighborhood point, is the average position coordinate of the neighborhood points.

[0083] Optionally, in an embodiment, the defect detection module includes:

[0084] The defect detection submodule is used to determine whether the angle is greater than a preset angle threshold. If so, it is determined that there is a defect in the area where the position corresponding to the first normal direction is located; otherwise, it is determined that there is no defect in the area where the position corresponding to the first normal direction is located.

[0085] Embodiment 3

[0086] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a wafer surface defect detection method as described in Embodiment 1 is implemented.

[0087] Embodiment 4

[0088] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute a wafer surface defect detection method as described in the first embodiment.

[0089] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0090] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A wafer surface defect detection method, characterized in that: include: Acquire point cloud data for the surface of the wafer to be inspected, wherein the point cloud data includes the three-dimensional coordinates of each point on the surface of the wafer to be inspected; Calculate, according to the point cloud data, a first normal direction corresponding to each point, wherein each of the first normal directions represents a direction of a normal corresponding to a position of any point in the point cloud data; For each of the first normal directions, the angle between the first normal direction and a preset reference normal direction is calculated, and it is determined whether there is a defect in the area corresponding to the position of the first normal direction based on the angle to obtain a surface defect detection result.

2. A wafer surface defect detection method according to claim 1, characterized in that: The step of calculating the first normal direction corresponding to each point according to the point cloud data includes: Performing voxel filtering on the point cloud data to generate new point cloud data; The directions of the normal lines passing through each point in the new point cloud data are calculated respectively to obtain the directions of the first normal lines.

3. A wafer surface defect detection method according to claim 2, characterized in that: The step of performing voxel filtering on the point cloud data to generate new point cloud data includes: Dividing the point cloud space corresponding to the point cloud data into a plurality of voxels according to a preset voxel size; Calculating the voxel center of each voxel respectively; The new point cloud data is generated according to each of the voxel centers.

4. A wafer surface defect detection method according to claim 2, characterized in that: For each point in the new point cloud data, calculating the direction of a normal line passing through the point to obtain the first normal line direction includes: For each point in the new point cloud data, a plurality of neighboring points adjacent to the point are acquired from the new point cloud data by a proximity search algorithm; For each point in the new point cloud data, calculating a covariance matrix according to the three-dimensional coordinates of each of the neighborhood points; For each point in the new point cloud data, the direction of the normal line passing through the point is calculated according to the covariance matrix to obtain the first normal line direction corresponding to the point.

5. A wafer surface defect detection method according to claim 4, characterized in that: For each point in the new point cloud data, calculating the direction of the normal line passing through the point according to the covariance matrix to obtain the first normal line direction corresponding to the point includes: For each point in the new point cloud data, calculating the eigenvector of the covariance matrix and the eigenvalues ​​corresponding to each eigenvector; According to each of the eigenvalues, determining the eigenvalue with the smallest value; Determine that the eigenvector corresponding to the eigenvalue with the smallest value is the normal vector; The normal vector is normalized to obtain the first normal direction.

6. A wafer surface defect detection method according to claim 4, characterized in that: For each point in the new point cloud data, calculating the covariance matrix according to the three-dimensional coordinates of each of the neighborhood points includes: For each point in the new point cloud data, the covariance matrix is ​​calculated by a first formula according to the three-dimensional coordinates of each of the neighborhood points. The first formula is: Among them, C is the covariance matrix, k is the number of neighborhood points, and p j is the three-dimensional coordinate of the jth neighborhood point, is the average position coordinate of the neighborhood points.

7. A wafer surface defect detection method according to claim 1, characterized in that: The determining, based on the angle, whether there is a defect in the area where the position corresponding to the first normal direction is located includes: It is determined whether the angle is greater than a preset angle threshold. If so, it is determined that a defect exists in the area where the position corresponding to the first normal direction is located; otherwise, it is determined that no defect exists in the area where the position corresponding to the first normal direction is located.

8. A wafer surface defect detection system, characterized in that: include: A data acquisition module, used to acquire point cloud data for the surface of the wafer to be detected, wherein the point cloud data includes the three-dimensional coordinates of each point on the surface of the wafer to be detected; A normal calculation module, used for calculating a first normal direction corresponding to each point according to the point cloud data, wherein each first normal direction represents a direction of a normal corresponding to a position of any point in the point cloud data; The defect detection module is used to calculate the angle between the first normal direction and the preset control normal direction for each of the first normal directions, and determine whether there is a defect in the area corresponding to the position of the first normal direction according to the angle, so as to obtain a surface defect detection result.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a wafer surface defect detection method as claimed in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to enable a computer to execute a wafer surface defect detection method as described in any one of claims 1 to 7.

Images