Semiconductor wafer surface defect detection method, device, equipment and storage medium

Abstract

The present invention provides a method, device, equipment and storage medium for detecting defects on the surface of a semiconductor wafer. The method comprises: a detection light beam emitted by an LED detection light source array is spatially filtered and then irradiated at a preset angle on the surface of the wafer to be tested. The reflected light is divided into a direct light path and a filtered light path. The direct light path is passed through a second detector to obtain an original image, and the filtered light path is passed through a 4f optical system to undergo spatial frequency filtering and then a low-pass filtered image is obtained by a third detector. A high-pass filtered image is obtained by performing a differential operation on the original image and the low-pass filtered image, and is subjected to adaptive threshold segmentation and morphological optimization processing to extract image feature data. Finally, the image feature data is input into a defect detection model to determine the defect detection results on the wafer surface. The present invention effectively reduces the interference of background noise through spatial filtering, enhances the contrast of defect edge features, and improves the detection accuracy and classification accuracy of minor defects.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a method, device, equipment and storage medium for detecting surface defects of semiconductor wafers. Background Art

[0002] Surface defect detection on semiconductor wafers is a critical process in semiconductor device manufacturing, and its detection accuracy directly impacts product yield and device reliability. With the continuous advancement of semiconductor technology, the size of wafer surface defects is shrinking, and the requirements for detection accuracy are becoming increasingly stringent. Traditional methods for detecting surface defects on semiconductor wafers rely primarily on direct optical imaging analysis, which involves directly illuminating the wafer surface with a light source to obtain defect images. Defect images obtained using this method are susceptible to interference from background noise, resulting in low contrast between the defect area and the background area in the image, making it difficult to highlight the edge features of the defect. This is particularly true for tiny defects, resulting in poor imaging quality, which seriously impacts the accuracy of subsequent defect detection and classification. Summary of the Invention

[0003] The main purpose of the present invention is to solve the technical problem that the existing semiconductor wafer surface defect detection relies on direct optical imaging analysis and is difficult to highlight the edge features of the defects;

[0004] A first aspect of the present invention provides a method for detecting surface defects of a semiconductor wafer, the method comprising:

[0005] Performing spatial filtering on the detection beam emitted by the LED detection light source array, and irradiating the detection beam after spatial filtering at a preset angle to the surface of the wafer to be tested through a preset reflector group;

[0006] Splitting the detection light path reflected from the surface of the wafer to be tested into a direct light path and a filtered light path, receiving the direct light path through a preset second detector to obtain an original image, performing spatial frequency filtering on the filtered light path through a preset 4f optical system, and then receiving the low-pass filtered image through a third detector;

[0007] performing a difference operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and performing adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data;

[0008] The image feature data is input into a preset defect detection model, and a defect detection result of the surface of the wafer to be tested is determined by the defect detection model.

[0009] Optionally, in a first implementation of the first aspect of the present invention, the performing of spatial filtering on the detection beam emitted by the LED detection light source array, and irradiating the surface of the wafer to be tested at a preset angle through a preset reflector group includes:

[0010] The preset scanning sequence and driving current control each LED in the LED detection light source array row by row to emit a detection light beam;

[0011] collimating the detection beam and performing spatial filtering on the detection beam through a preset collimating lens device and a pinhole aperture to obtain a detection beam after spatial filtering;

[0012] The reflection angle of the detection light beam after the spatial filtering process is adjusted by a plurality of preset groups of reflection mirrors, so that the detection light beam after the filtering is incident on the surface of the wafer to be measured at an angle of 45°.

[0013] Optionally, in a second implementation of the first aspect of the present invention, the collimating lens device includes a first converging lens and a second converging lens;

[0014] The method of performing collimation processing and spatial filtering processing on the detection light beam by using a preset collimating lens device and a pinhole aperture to obtain a detection light beam after spatial filtering processing includes:

[0015] Converging the detection light beam through the first converging lens to obtain a converged light beam, and spatially adjusting the convergence focus of the converged light beam so that the convergence focus is coaxial with the aperture of the pinhole diaphragm and is located on the incident surface of the pinhole diaphragm;

[0016] Performing inter-filtering processing on the detection light beam through the pinhole diaphragm to remove scattered light components in the convergent light beam to obtain a filtered divergent light beam;

[0017] The filtered divergent light beam is collimated by the second converging lens to transform the filtered divergent light beam into a parallel light beam, thereby obtaining a detection light beam after spatial filtering.

[0018] Optionally, in a third implementation of the first aspect of the present invention, the 4f optical system includes a first lens and a second lens;

[0019] The method of splitting the detection light path reflected from the surface of the wafer to be tested into a direct light path and a filtered light path, receiving the direct light path by a preset second detector to obtain an original image, performing spatial frequency filtering on the filtered light path by a preset 4f optical system, and then receiving the low-pass filtered image by a third detector comprises:

[0020] The detection light path reflected from the surface of the wafer to be tested is split by a preset spectroscope to obtain a direct light path and a filtered light path, wherein the light intensity ratio of the direct light path to the filtered light path is 1:1;

[0021] Performing photoelectric conversion processing on the direct light path through a preset second detector, converting the light intensity information of the direct light path into an electrical signal to obtain an original image;

[0022] Focusing the filtering light path through the first lens, and obtaining a Fourier spectrum containing spatial frequency information of the surface of the wafer to be measured on the back focal plane of the first lens;

[0023] performing spatial frequency filtering on the Fourier spectrum through an aperture provided on the back focal plane of the first lens, so as to retain low-frequency components in a central region of the Fourier spectrum;

[0024] The Fourier spectrum after spatial frequency filtering is imaged by the second lens, and the light intensity information after imaging is received by the third detector to obtain the low-pass filtered image.

[0025] Optionally, in a fourth implementation of the first aspect of the present invention, performing spatial frequency filtering on the Fourier spectrum by using a stop provided on the back focal plane of the first lens to retain low-frequency components in a central region of the Fourier spectrum includes:

[0026] adjusting the position of the diaphragm on the back focal plane of the first lens using a three-axis translation stage so that the aperture center of the diaphragm on the back focal plane of the first lens coincides with the optical axis of the first lens, and the diaphragm is located on the back focal plane of the first lens;

[0027] The aperture size of the diaphragm is adjusted by a preset adjustment motor so that the aperture size of the diaphragm matches the size of the bright spot at the center of the Fourier spectrum, thereby obtaining an adjusted aperture size;

[0028] The frequency of the Fourier spectrum is selected by the adjusted aperture, and only the low-frequency components in the center area of ​​the Fourier spectrum are allowed to pass through, thereby obtaining a spectrum image after spatial frequency filtering.

[0029] Optionally, in a fifth implementation of the first aspect of the present invention, performing a difference operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and performing adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data includes:

[0030] Performing pixel-wise difference calculation on the original image and the low-pass filtered image using an image processor to obtain a high-pass filtered image reflecting edge and detail information of the wafer surface;

[0031] performing multi-level threshold segmentation processing on the high-pass filtered image according to the grayscale histogram distribution of the high-pass filtered image to obtain an initial segmented image;

[0032] Performing morphological processing of opening and closing operations on the initial segmented image to eliminate noise points and burrs in the initial segmented image and obtain a segmented image with a smooth contour;

[0033] A connected domain analysis is performed on the segmented image with smooth contours using a region labeling algorithm, and characteristic parameters of each connected domain are extracted to obtain image feature data.

[0034] Optionally, in a sixth implementation of the first aspect of the present invention, inputting the image feature data into a preset defect detection model, and determining a defect detection result of the surface of the wafer to be tested by using the defect detection model includes:

[0035] Performing dimensionality reduction processing on the image feature data using a principal component analysis algorithm to obtain a principal feature vector representing the geometric form of the feature region;

[0036] Analyzing the main feature vector using a preset multidimensional anomaly detection algorithm based on Mahalanobis distance to obtain a confidence level, and determining whether a characteristic abnormal area exists on the surface of the wafer to be tested based on a preset confidence interval and the confidence level;

[0037] If there is no characteristic abnormal area, the defect detection result is set as the surface of the wafer to be tested is defect-free;

[0038] If there is a feature abnormal area, a nonlinear feature transformation is performed on the feature abnormal area through a feature mapping algorithm based on a kernel function to obtain a high-dimensional feature representation containing regional morphology and topological structure information;

[0039] The high-dimensional feature representation is analyzed by a nearest neighbor classification algorithm and a defect assessment criterion to obtain a defect type determination result and a severity determination result of the feature abnormality area, and the defect type determination result and the severity determination result are used as the defect detection result.

[0040] A second aspect of the present invention provides a semiconductor wafer surface defect detection device, the semiconductor wafer surface defect detection device comprising:

[0041] The spatial filtering module is used to perform spatial filtering on the detection beam emitted by the LED detection light source array, and illuminate the surface of the wafer to be tested at a preset angle through a preset reflector group.

[0042] A spectroscopic module is configured to split the detection light path reflected from the surface of the wafer to be tested into a direct light path and a filtered light path, receive the direct light path via a preset second detector to obtain an original image, perform spatial frequency filtering on the filtered light path via a preset 4f optical system, and receive the filtered light path via a third detector to obtain a low-pass filtered image;

[0043] an image processing module, configured to perform a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and perform adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data;

[0044] The detection module is used to input the image feature data into a preset defect detection model and determine the defect detection result of the surface of the wafer to be tested through the defect detection model.

[0045] The third aspect of the present invention provides a semiconductor wafer surface defect detection device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected through lines; the at least one processor calls the instructions in the memory so that the semiconductor wafer surface defect detection device performs the steps of the above-mentioned semiconductor wafer surface defect detection method.

[0046] A fourth aspect of the present invention provides a computer-readable storage medium having instructions stored therein, which, when executed on a computer, enables the computer to execute the steps of the above-mentioned semiconductor wafer surface defect detection method.

[0047] The above-mentioned semiconductor chip surface defect detection method, device, equipment and storage medium use an LED detection light source array to emit a detection beam, which is spatially filtered and then irradiated at a preset angle on the surface of the chip to be tested. The reflected light is divided into a direct light path and a filtered light path. The direct light path is passed through a second detector to obtain an original image, and the filtered light path is passed through a 4f optical system to obtain a low-pass filtered image through a third detector after spatial frequency filtering. A high-pass filtered image is obtained by performing a differential operation on the original image and the low-pass filtered image, and is subjected to adaptive threshold segmentation and morphological optimization processing to extract image feature data. Finally, the image feature data is input into a defect detection model to determine the defect detection results of the chip surface. The present invention effectively reduces the interference of background noise through spatial filtering, enhances the contrast of defect edge features, and improves the detection accuracy and classification accuracy of small defects.

[0048] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of a first embodiment of a method for detecting surface defects of a semiconductor wafer according to an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of an embodiment of a semiconductor wafer surface defect detection device according to an embodiment of the present invention;

[0052] Figure 3 Schematic diagram of an embodiment of a semiconductor wafer surface defect detection device in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0054] The terms "including," "having," and any variations thereof, as used in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to the process, method, product, or device.

[0055] To facilitate understanding of this embodiment, a method for detecting surface defects of a semiconductor wafer disclosed in an embodiment of the present invention is first described in detail. Figure 1 As shown, this method includes the following steps:

[0056] 101. Performing spatial filtering on the detection beam emitted by the LED detection light source array, and irradiating the detection beam after spatial filtering onto the surface of the wafer to be tested at a preset angle through a preset reflector group;

[0057] In one embodiment of the present invention, the spatial filtering processing of the detection light beam emitted by the LED detection light source array and irradiating the detection light beam after spatial filtering processing at a preset angle to the surface of the chip to be tested through a preset reflector group include: controlling each LED in the LED detection light source array row by row according to a preset scanning sequence and driving current to emit a detection light beam; collimating and spatially filtering the detection light beam through a preset collimating lens device and a pinhole aperture to obtain a detection light beam after spatial filtering processing; adjusting the reflection angle of the detection light beam after spatial filtering processing through a preset plurality of reflectors so that the filtered detection light beam is incident on the surface of the chip to be tested at an angle of 45°.

[0058] Specifically, the LED array control system consists of a main controller, a drive current control module, and a timing generation circuit. The main controller utilizes a multi-bit microprocessor with a built-in digital-to-analog converter and timer to generate precise control signals. The drive current control module includes a constant current source circuit and a current feedback regulation circuit. The constant current source circuit provides an adjustable DC drive current for each LED. The current feedback regulation circuit samples the LED's operating current in real time and generates an error signal through a comparator circuit composed of operational amplifiers. This error signal is then used to adjust the conduction state of the power device to ensure the stability of the LED's operating current. The timing generation circuit, implemented using a programmable logic device, includes a multi-stage shift register and a programmable frequency divider, enabling precise scanning timing. The scanning process proceeds according to a preset program, with each row of LEDs lighting up sequentially at a preset time interval. A protection period is set between adjacent LEDs to eliminate crosstalk. After a row is scanned, the next row is switched after a row blanking period. The shift register's clock frequency can be adjusted within a certain range to accommodate different detection speed requirements. Brightness uniformity for each LED is ensured by a photoelectric feedback circuit that compensates for temperature drift and aging effects in real time. The system also includes overvoltage protection, overcurrent protection and temperature protection circuits to ensure that the LED array operates within a safe range.

[0059] Specifically, the collimating lens can be an achromatic doublet lens structure with an appropriate focal length and numerical aperture, effectively correcting spherical and chromatic aberrations. The collimated beam has a small divergence angle. After collimation, the beam enters the 4f system. The first lens is an achromatic lens with an appropriate focal length and aperture, coated with an antireflection coating to improve transmittance. After passing through the first lens, the beam forms a Fourier transform surface on its rear focal plane, on which a precision aperture diaphragm is mounted. The aperture diaphragm is made of precision-machined metal film, achieving excellent roundness and edge smoothness. The aperture is mounted on a three-dimensional translation stage, allowing precise adjustment of the aperture's position on the optical axis. The second lens has the same specifications as the first lens, and the distance between the two lenses is controlled at twice the focal length using a precision optical mount. The entire 4f system is mounted on an optical platform with vibration reduction. The diffraction pattern at the aperture can be monitored in real time by a camera to assist in adjusting the system's optical path. The spatially filtered beam exhibits excellent spatial coherence and wavefront quality.

[0060] Specifically, the reflector assembly consists of three high-reflectivity mirrors, the base material of which is quartz glass and coated with an enhanced reflection film. The first reflector is mounted on a two-axis angle adjustment mount and can be adjusted within a certain range. The second reflector is mounted on a two-axis angle adjustment mount and also has a translation adjustment mechanism. The adjustment mechanism of the third reflector includes two-axis angle adjustment and three-axis displacement adjustment. The relative positions of the three reflectors can be calibrated using optical inspection equipment. The reason for choosing a 45° incident angle is that: first, the reflected light at this angle does not interfere with the incident light path; second, 45° incidence can produce good light-dark contrast, which is conducive to defect detection; and third, the scattered light intensity distribution at this angle is uniform, which facilitates subsequent image processing. The overall structure of the reflector assembly is made of materials with good temperature stability. The system can be equipped with an active feedback control device to compensate for angular deviations caused by mechanical vibration and thermal drift, ensuring the stability of the beam incident angle.

[0061] Furthermore, the collimating lens device includes a first converging lens and a second converging lens; the detection light beam is collimated and spatially filtered respectively by the preset collimating lens device and the pinhole aperture to obtain the detection light beam after spatial filtering, including: converging the detection light beam by the first converging lens to obtain a converged light beam, and spatially adjusting the convergence focus of the converged light beam so that the convergence focus is coaxial with the aperture hole of the pinhole aperture and is located on the incident surface of the pinhole aperture; performing inter-filtering processing on the detection light beam by the pinhole aperture to remove the scattered light components in the converged light beam to obtain a filtered divergent light beam; and collimating the filtered divergent light beam by the second converging lens to convert the filtered divergent light beam into a parallel light beam to obtain a detection light beam after spatial filtering.

[0062] Specifically, the collimating lens device is composed of two converging lenses, and the two converging lenses are selected with the same focal length to ensure the parallelism of the light beam after passing through the entire system. Both converging lenses adopt an achromatic lens structure, and the lens surface is coated with an anti-reflection film to reduce the reflection loss of light. The two converging lenses are installed on a precision adjustment seat, and the adjustment seat has axial and radial adjustment freedom to adjust the position of the lens and the alignment of the optical axis. The first converging lens is mainly used to converge the divergent detection light beam into a convergent light beam, and the second converging lens is used to re-collimate the divergent light beam after filtering through the pinhole aperture into a parallel light beam. The installation position of the two converging lenses needs to ensure that their optical axes are collinear to reduce aberrations. The material selection of the converging lens needs to take into account factors such as the transmission band, chromatic aberration correction and machining performance. The surface accuracy and surface quality of the lens directly affect the beam quality.

[0063] Specifically, after the detection light beam enters the first converging lens, the lens converges the divergent light into a convergent light beam. The focusing position of the convergent light beam is determined by the focal length of the first converging lens, and the convergent light beam forms a minimum light spot at the focus. In order for the convergent light beam to be effectively coupled with the pinhole aperture, a three-dimensional displacement adjustment device is required to accurately adjust the convergent focus of the convergent light beam. The adjustment process includes two aspects: axial position adjustment and lateral position adjustment. The axial adjustment ensures that the convergent focus is located on the incident surface of the pinhole aperture, and the lateral adjustment ensures that the convergent focus is coaxial with the aperture hole of the pinhole aperture. The position adjustment device adopts a precision displacement platform with sufficient adjustment accuracy. During adjustment, the light intensity distribution after the aperture can be monitored in real time by a photoelectric detector. When the convergent focus is coaxial with the aperture hole and is located on the incident surface, the light intensity passing through the aperture is the largest.

[0064] Specifically, after the convergent light beam is spatially adjusted, an Airy disk distribution will be formed at the pinhole diaphragm. The central bright spot of the Airy disk contains the main optical information, while the outer diffraction rings are mainly produced by the scattered light components in the light beam. The selection of the aperture size of the pinhole diaphragm requires a balance between the filtering effect and the transmitted light intensity. An aperture that is too small will result in excessive light energy loss, while an aperture that is too large will result in an insignificant filtering effect. The pinhole diaphragm spatially filters the incident convergent light beam, allowing only the light beam in the central area of ​​the Airy disk to pass through, thereby removing the scattered light components in the convergent light beam. The light beam passing through the pinhole diaphragm will form a divergent beam due to diffraction. The processing quality of the pinhole diaphragm, such as the roundness of the aperture and the smoothness of the edge, will affect the filtering effect. In order to reduce stray reflections, the surface of the pinhole diaphragm needs to be treated with an absorption coating.

[0065] Specifically, the divergent light beam after filtering by the pinhole aperture continues to be transmitted to the second converging lens. The function of the second converging lens is to convert the divergent light beam into a parallel light beam, and its focal length needs to be selected the same as that of the first converging lens to ensure that the final output is a parallel light beam. The position of the second converging lens needs to be precisely adjusted so that its front focus coincides with the position of the pinhole aperture. When the divergent light beam passes through the second converging lens, the detection light beam after spatial filtering is obtained. This light beam has good spatial coherence, and its lateral intensity distribution is more uniform, which is suitable for subsequent defect detection. The adjustment mechanism of the second converging lens needs to have angle and position adjustment functions to ensure the accuracy of the outgoing light beam. The entire optical path system needs to take necessary vibration reduction and dust prevention measures to ensure the stability of the system.

[0066] 102. Splitting the detection light path reflected from the surface of the wafer to be tested into a direct light path and a filtered light path, receiving the direct light path via a preset second detector to obtain an original image, performing spatial frequency filtering on the filtered light path via a preset 4f optical system, and then receiving the filtered light path via a third detector to obtain a low-pass filtered image;

[0067] In one embodiment of the present invention, the 4f optical system includes a first lens and a second lens; the detection light path reflected from the surface of the wafer to be tested is split into a direct light path and a filtered light path, the direct light path is received by a preset second detector to obtain an original image, the filtered light path is subjected to spatial frequency filtering by a preset 4f optical system, and then a low-pass filtered image is received by a third detector, comprising: performing a splitting process on the detection light path reflected from the surface of the wafer to be tested by a preset spectroscope to obtain a direct light path and a filtered light path, wherein the light intensity ratio between the direct light path and the filtered light path is 1:1; the filtered light path is subjected to spatial frequency filtering by a preset second detector, and then a low-pass filtered image is obtained by a third detector. The direct optical path is subjected to photoelectric conversion processing to convert the light intensity information of the direct optical path into an electrical signal to obtain an original image; the filtered optical path is focused by the first lens, and a Fourier spectrum containing spatial frequency information of the surface of the wafer to be measured is obtained on the back focal plane of the first lens; the Fourier spectrum is subjected to spatial frequency filtering processing by an aperture set on the back focal plane of the first lens to retain the low-frequency components in the central area of ​​the Fourier spectrum; the Fourier spectrum after the spatial frequency filtering processing is imaged by the second lens, and the light intensity information after imaging is received by the third detector to obtain the low-pass filtered image.

[0068] Specifically, the 4f optical system consists of two lenses with the same focal length, and the distance between the two lenses must be precisely controlled to be the sum of the two focal lengths. Both lenses utilize an achromatic lens structure, and the lens material selection requires comprehensive consideration of factors such as dispersion characteristics within the operating wavelength range, machinability, and service life. The lens surfaces are coated with a multi-layer antireflection coating to reduce reflection loss. The design of the antireflection coating must consider the incident angle range and the operating wavelength range. The two lenses are mounted on a six-degree-of-freedom precision adjustment mount, which features translation adjustment in three orthogonal directions and angular adjustment along three rotational axes for adjusting the spatial position and optical axis orientation of the lenses. The first lens primarily transforms the image information of the object under test plane into the frequency domain, forming a Fourier spectrum at its rear focal plane. The second lens transforms the frequency-filtered spectrum back into the spatial domain to produce the filtered image. The installation of the two lenses requires highly accurate collinearity of the optical axes, which requires the use of precision alignment fixtures and optical inspection equipment such as interferometers. The surface accuracy and surface quality of the lens have an important impact on the imaging quality of the system. Generally, the surface accuracy is required to be better than λ / 10 and the surface roughness is required to be better than 0.5nm.

[0069] Specifically, after the detection light path reflected from the surface of the wafer under test enters the beamsplitter, it must split the incident light into two beams of exactly equal intensity. The beamsplitter utilizes a non-polarizing beamsplitter prism structure, consisting of two right-angle prisms glued together, with the beam-splitting surfaces coated with a non-polarizing beamsplitter coating. The design of the beamsplitter coating must comprehensively consider factors such as the polarization state of the incident light, the angle of incidence, and the operating wavelength, ensuring that the intensity ratio of the two output beams remains strictly 1:1 under actual operating conditions. The beamsplitter is mounted on a precision five-degree-of-freedom adjustment mount, which includes two orthogonal tilt adjustments, one azimuth adjustment, and two lateral displacement adjustments to adjust the spatial orientation of the beamsplitter so that the two split beams propagate strictly in the predetermined directions. The direct optical path is used to acquire raw image information, while the filter optical path is used for spatial frequency domain filtering. The surface flatness of the beamsplitter directly affects wavefront quality and is generally required to be better than λ / 8. The thickness uniformity and interface flatness of the beamsplitter coating significantly influence the beam splitting effect, requiring the use of high-precision optical thin film processing. In addition, in order to reduce the influence of stray reflections, the non-working surface of the beam splitter needs to be anti-reflective treated.

[0070] Specifically, the direct light path after splitting enters the second detector, which converts the optical signal into an electrical signal to produce the original image. The second detector uses a high-sensitivity area array photodetector. Detector selection requires comprehensive consideration of parameters such as resolution, sensitivity, dynamic range, and response speed. The detector's photosensitivity surface size must match the incident beam diameter. The pixel size determines the spatial sampling interval of the image and must meet the requirements of the sampling theorem. The detector's spectral response range must fully cover the operating wavelength range of the detection light source, and the quantum efficiency must meet the detection sensitivity requirements. The analog signal output by the detector is first amplified by a preamplifier, then filtered through a bandpass filter to remove high-frequency noise. Finally, it is digitized and sampled by a high-precision analog-to-digital converter to produce the digitized original image. The detector's mounting position must be precisely controlled to ensure that the wafer surface is strictly conjugated to the detector's photosensitivity surface to ensure a clear original image. The detector's temperature control system is used to reduce dark current noise and improve the signal-to-noise ratio. Furthermore, the detector's linearity and uniformity have a significant impact on image quality, requiring calibration to obtain a response curve and perform correction.

[0071] Furthermore, the spatial frequency filtering processing is performed on the Fourier spectrum by using an aperture set on the back focal plane of the first lens to retain the low-frequency components in the central area of ​​the Fourier spectrum, including: adjusting the position of the aperture on the back focal plane of the first lens by using a three-axis translation stage so that the aperture center of the aperture on the back focal plane of the first lens coincides with the optical axis of the first lens, and the aperture is located on the back focal plane of the first lens; adjusting the aperture size of the aperture by using a preset adjustment motor so that the aperture of the aperture matches the size of the bright spot in the center of the Fourier spectrum, thereby obtaining an adjusted aperture; and performing frequency selection on the Fourier spectrum by using the adjusted aperture to allow only the low-frequency components in the central area of ​​the Fourier spectrum to pass through, thereby obtaining a spectrum image after spatial frequency filtering.

[0072] Specifically, after the filtered light path after separation enters the first lens, the light field undergoes a Fourier transform during propagation, forming a spectral distribution on the back focal plane of the first lens. This spectral distribution is actually the spatial frequency decomposition of the object space image. The center of the spectrum corresponds to the zero-frequency component, which contains the image's average brightness information. Regions farther from the center correspond to higher spatial frequency components, which contain information such as edges and details in the image. Choosing the focal length of the first lens requires a trade-off between multiple factors: a longer focal length increases the spatial spread of the spectrum and improves the frequency resolution, but also increases the system size. A shorter focal length provides a more compact system structure, but is more susceptible to aliasing. The formation of the spectrum follows the principles of Fourier optics. The complex amplitude distribution of the incident light field on the back focal plane is the Fourier transform of the object light field. The first lens must have excellent aberration correction performance, especially spherical aberration and field curvature, which must be strictly controlled as these aberrations can cause spectral distortion. The lens aperture must be large enough to receive all spatial frequency components. To achieve an ideal spectral distribution, the wavefront quality of the incident beam must also be guaranteed, which requires the incident beam to have good collimation and coherence. The installation tilt error of the first lens must be strictly controlled, as tilting will introduce additional aberrations and cause spectral distortion.

[0073] Specifically, an aperture of a specific size is placed at the spectrum-forming position on the rear focal plane of the first lens. By controlling the aperture size, the frequency components to be retained are selected. The aperture size of the aperture is directly related to the spatial frequency cutoff characteristics of the system: smaller apertures result in lower cutoff frequencies and a more blurred filtered image; larger apertures retain more high-frequency components, resulting in richer image detail. A circular aperture shape is generally chosen because it provides consistent cutoff characteristics for frequency components in all directions. Aperture quality is crucial, and the aperture edge must be smooth to avoid diffraction effects. The aperture must be precisely positioned on the spectral plane, requiring a precise three-dimensional displacement adjustment mechanism. The center of the aperture must strictly coincide with the optical axis; decentration can lead to asymmetric filtering characteristics. To reduce the influence of stray light, the aperture surface requires an absorption coating. The aperture thickness should be moderate; too thick will cause an aperture effect, while too thin will result in insufficient strength. In practical applications, multiple apertures of different diameters can be prepared, and the appropriate one can be selected based on specific needs. The diffraction effect and edge effect of the aperture will affect the filtering effect, which needs to be fully considered when designing the aperture.

[0074] Specifically, the frequency-filtered light field continues to propagate to the second lens, which performs an inverse Fourier transform on the filtered spectrum, reconstructing the frequency domain information into a spatial image. This reconstruction process is essentially an inverse transformation, requiring the parameters of the second lens to be identical to those of the first lens and the distance between them to be precisely controlled. The reconstructed image appears smoother due to the removal of high-frequency components, suppressing noise and rapidly changing edge details. The reconstructed image is received by the third detector and converted into an electrical signal. The performance requirements of the third detector are similar to those of the second detector, but given that it receives a filtered image, its dynamic range requirements can be appropriately lowered. The distance from the second lens to the third detector must be precisely adjusted to obtain a clear reconstructed image. All components of the 4f system must be fixed to a stable base to avoid mechanical vibration and thermal drift. All reflective and transmissive surfaces in the optical path require anti-reflection treatment to optimize the system's light energy utilization. The contrast and resolution of the reconstructed image are directly affected by the aperture size of the aperture, and appropriate parameters must be selected based on the detection requirements in actual applications. The system assembly and adjustment process requires the use of precision optical instruments such as interferometers to ensure precise alignment of the entire optical path.

[0075] 103. Perform a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and perform adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data;

[0076] In one embodiment of the present invention, the step of performing a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and performing adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data comprises: performing a pixel-wise differential operation on the original image and the low-pass filtered image through an image processor to obtain a high-pass filtered image reflecting the edge and detail information of the chip surface; performing multi-level threshold segmentation processing on the high-pass filtered image according to the grayscale histogram distribution of the high-pass filtered image to obtain an initial segmented image; performing morphological processing of the initial segmented image through opening and closing operations to eliminate noise points and burrs in the initial segmented image to obtain a segmented image with a smooth contour; and performing connected domain analysis on the segmented image with a smooth contour through a region labeling algorithm to extract feature parameters of each connected domain to obtain image feature data.

[0077] Specifically, the original image and the low-pass filtered image are first converted into grayscale images of the same size, ensuring a one-to-one correspondence between the pixels in the two images. The image processor then reads the grayscale values ​​of each corresponding pixel in the two images and performs a subtraction operation on these grayscale values. Since the low-pass filtered image retains low-frequency information, representing the background and slowly varying regions, while the original image contains full frequency information, the difference between the two reflects the high-frequency information, namely edges and details. This subtraction process effectively implements a high-pass filtering effect in the spatial domain. To prevent negative values ​​in the difference results, the grayscale values ​​after the difference are offset, typically by adding a fixed offset value, to ensure that all difference results fall within the valid grayscale range. Overflow processing is also performed for each pixel to ensure that the final grayscale value does not exceed the dynamic range of the image. The image processor uses a pipelined architecture to perform these operations, enabling parallel processing of multiple pixels, thereby improving processing speed. The resulting high-pass filtered image highlights edges and details on the wafer surface, which are valuable information for subsequent defect detection.

[0078] Specifically, the grayscale value distribution of all pixels in the high-pass filtered image is statistically analyzed to create a grayscale histogram. The histogram reflects the frequency of occurrence of different grayscale values ​​in the image, with peaks corresponding to the main grayscale levels. The grayscale histogram statistical process involves traversing each pixel in the image, using the pixel's grayscale value as an index and accumulating the number of occurrences of each grayscale value. After obtaining the grayscale histogram, a dynamic programming method is used to determine the optimal segmentation threshold. This method treats the histogram as a one-dimensional data sequence and calculates multiple segmentation thresholds by minimizing the between-class variance and maximizing the within-class variance. The number of thresholds can be dynamically determined based on the complexity of the image, and an iterative optimization approach is typically used to find the optimal threshold combination. After the thresholds are determined, each pixel in the high-pass filtered image is compared with these thresholds. Based on the range in which the pixel's grayscale value falls, a corresponding label is assigned, thus generating an initial segmented image. Each region in the initial segmented image represents a collection of pixels with similar grayscale characteristics, providing a preliminary indication of possible objects or defects in the image.

[0079] Specifically, morphological processing primarily uses two basic operations: opening and closing. Both operations are composed of two basic morphological operators: dilation and erosion. The opening operation involves performing erosion followed by dilation. The erosion operation uses a structuring element to scan the image. When the structuring element is completely within the target region, the location is retained; otherwise, it is removed. This eliminates noise points and artifacts smaller than the structuring element. The subsequent dilation operation uses the same structuring element. When the structuring element overlaps with the target region, the location is filled. This restores the target region reduced in size during the erosion process while maintaining edge smoothness. The closing operation involves performing dilation followed by erosion. The dilation operation can fill small holes and cracks within the target region, while the subsequent erosion operation restores the target region to its original size while maintaining the filling effect. The choice of structuring element significantly influences the effectiveness of morphological processing. A circular or square structuring element is typically used, and its size is determined based on the noise characteristics to be removed. The order of the opening and closing operations also requires careful selection. Typically, the opening operation is performed first to remove external noise, followed by the closing operation to fill internal holes. During the processing, attention should be paid to maintaining the main morphological features of the target area and avoiding excessive smoothing that may lead to loss of useful information.

[0080] Specifically, connected regions are labeled in a smooth-contour segmented image using a two-pass scanning method. The first pass scans the image from top to bottom and left to right, assigning a temporary label to each foreground pixel and recording the equivalence relationships between adjacent labels. The second pass, based on the equivalence relationships, merges labels with equivalent relationships into a single label, thereby uniquely identifying connected regions. For each identified connected region, a series of shape feature parameters are calculated, including area (number of pixels), perimeter (number of boundary pixels), circularity (the ratio of the square of the perimeter to the area to characterize the regularity of the shape), major axis length, minor axis length, and orientation angle. Furthermore, grayscale statistical features such as mean grayscale value, grayscale standard deviation, and grayscale entropy are calculated for each connected region. The calculation of these feature parameters requires traversing all pixels in each connected region, and some parameters, such as circularity, require complex mathematical operations. Together, these feature parameters form a feature vector describing the defect region. These feature vectors, as image feature data, are used for subsequent defect classification and identification. To ensure comparability of the feature parameters, some parameters are normalized to ensure they are not affected by image resolution.

[0081] 104. Input the image feature data into a preset defect detection model, and determine a defect detection result of the surface of the wafer to be tested using the defect detection model.

[0082] In one embodiment of the present invention, the inputting of the image feature data into a preset defect detection model and determining the defect detection result of the surface of the wafer to be tested by the defect detection model include: performing dimensionality reduction processing on the image feature data by a principal component analysis algorithm to obtain a principal eigenvector representing the geometric morphology of the feature area; analyzing the principal eigenvector by a preset multidimensional anomaly detection algorithm based on Mahalanobis distance to obtain a confidence level, and determining whether there is a characteristic abnormal area on the surface of the wafer to be tested based on a preset confidence interval and the confidence level; if there is no characteristic abnormal area, setting the defect detection result to the surface of the wafer to be tested as defect-free; if there is a characteristic abnormal area, performing a nonlinear feature transformation on the characteristic abnormal area by a feature mapping algorithm based on a kernel function to obtain a high-dimensional feature representation containing regional morphology and topological structure information; analyzing the high-dimensional feature representation by a nearest neighbor classification algorithm and a defect assessment criterion to obtain a defect type determination result and a severity determination result of the characteristic abnormal area, and using the defect type determination result and the severity determination result as the defect detection result.

[0083] Specifically, the original feature data must first be standardized to make features of different dimensions comparable. After standardization, the covariance matrix of the feature matrix is ​​calculated. The covariance matrix reflects the correlation between different features. The covariance matrix is ​​then subjected to eigenvalue decomposition to obtain eigenvalues ​​and corresponding eigenvectors. The eigenvalues ​​reflect the degree of data variation in the direction of the corresponding eigenvector; larger eigenvalues ​​indicate greater information in that direction. The eigenvalues ​​are sorted from largest to smallest, and the top eigenvectors with the largest contribution are selected to form a projection matrix. The original feature data is linearly transformed using this projection matrix to obtain the principal eigenvectors after dimensionality reduction. These principal eigenvectors retain the key information of the original data while significantly reducing the data dimension, simplifying subsequent processing. The selection of eigenvectors requires a comprehensive consideration of the cumulative contribution rate and the actual computational complexity. Eigenvectors with a cumulative contribution rate of at least 85% are generally selected. This dimensionality reduction process not only compresses the data but also eliminates correlations between features, making subsequent anomaly detection more reliable.

[0084] Specifically, the sample's mean vector and covariance matrix are calculated based on the principal eigenvector. These two statistics describe the distribution characteristics of normal samples. For each sample to be tested, the Mahalanobis distance is calculated between it and the mean vector. The Mahalanobis distance takes into account the scale differences and correlations across dimensions in the feature space, providing a more accurate measure of similarity between samples. The inverse of the covariance matrix is ​​used as the metric matrix in the distance calculation to eliminate the influence of correlation between features. The calculated Mahalanobis distance follows a chi-square distribution, so a threshold can be determined based on a preset confidence level. The calculated Mahalanobis distance is compared with the threshold. If the distance exceeds the threshold, the sample is considered an anomaly. The choice of confidence interval requires a balance between detection sensitivity and reliability; a confidence level of 95% or 99% is generally chosen. The results of anomaly detection directly determine the subsequent processing flow. If an anomaly is detected, further feature extraction and classification are required.

[0085] Specifically, an appropriate kernel function is selected. Common kernel functions include Gaussian kernel, polynomial kernel, etc. The selection of kernel function needs to consider the distribution characteristics and nonlinearity of the data. The samples in the original feature space are mapped to the high-dimensional feature space through the kernel function, and the samples that were originally nonlinearly separable in the high-dimensional space become linearly separable. The feature mapping process is actually performed implicitly. There is no need to explicitly calculate the coordinates in the high-dimensional space, but to directly calculate the kernel function value. The parameter selection of the kernel function has an important influence on the mapping effect. The optimal parameters can be selected through methods such as cross-validation. The high-dimensional features obtained by mapping not only contain the morphological information of the original features, but also extract deeper structural features through nonlinear transformation. This nonlinear feature transformation can better characterize the complex morphological characteristics of the defect area and provide more effective feature representation for subsequent classification. The distance measurement in the high-dimensional feature space adopts the kernel-induced measurement, which is more in line with the essential structure of the data.

[0086] Specifically, the process involves analyzing high-dimensional features using the nearest neighbor classification algorithm and defect assessment criteria. First, a k-nearest neighbor classifier is constructed in the high-dimensional feature space. The optimal value of k is determined through cross-validation. The choice of k requires a balance between the classifier's generalization ability and its ability to express local features. For each sample to be classified, its distance to all samples in the training set is calculated in kernel space. This distance calculation utilizes the properties of the kernel function and eliminates the need to explicitly calculate coordinates in the high-dimensional feature space. The k nearest neighbor samples are selected, and a weighted voting method is used to determine the class of the sample to be classified. The weights are inversely proportional to the distance between the samples, so that closer samples have greater voting weight. Furthermore, corresponding evaluation criteria are established for each defect type, including multiple evaluation metrics such as defect size, shape complexity, and edge sharpness. These evaluation criteria employ fuzzy rules to map the values ​​of each metric to a severity score. In the specific calculation, each metric is first normalized to a uniform range of [0,1]. Then, a score for each metric is calculated using a predefined membership function. Each score is weighted and averaged to arrive at the final severity determination. The weighting factors reflect the importance of different indicators. The final defect detection result includes information on both defect type and severity. This multi-dimensional assessment provides a more comprehensive picture of the wafer surface quality. The confidence level of the detection result is measured by the consistency of neighboring samples. Higher consistency indicates more reliable classification results. The entire analysis process utilizes a parallel computing architecture, capable of processing multiple samples simultaneously, improving detection efficiency.

[0087] In this embodiment, a detection beam emitted by an LED detection light source array undergoes spatial filtering and then illuminates the surface of the wafer to be tested at a preset angle. The reflected light is divided into a direct light path and a filtered light path. The direct light path passes through a second detector to obtain an original image, while the filtered light path passes through a 4f optical system to undergo spatial frequency filtering and obtains a low-pass filtered image from a third detector. A high-pass filtered image is obtained by performing a differential operation on the original image and the low-pass filtered image, and then adaptive threshold segmentation and morphological optimization are performed on the image to extract image feature data. Finally, the image feature data is input into a defect detection model to determine the defect detection results on the wafer surface. The present invention effectively reduces background noise interference through spatial filtering, enhances the contrast of defect edge features, and improves the detection accuracy and classification accuracy of minor defects.

[0088] The above describes the semiconductor wafer surface defect detection method according to the embodiment of the present invention. The following describes the semiconductor wafer surface defect detection device according to the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a semiconductor wafer surface defect detection device includes:

[0089] The spatial filtering module 201 is used to perform spatial filtering on the detection beam emitted by the LED detection light source array, and illuminate the surface of the wafer to be tested at a preset angle through a preset reflector group.

[0090] A spectrometer module 202 is configured to split the detection light path reflected from the surface of the wafer to be tested into a direct light path and a filtered light path, receive the direct light path via a preset second detector to obtain an original image, perform spatial frequency filtering on the filtered light path via a preset 4f optical system, and receive the filtered light path via a third detector to obtain a low-pass filtered image;

[0091] An image processing module 203 is configured to perform a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and perform adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data;

[0092] The detection module 204 is configured to input the image feature data into a preset defect detection model, and determine a defect detection result of the surface of the wafer to be detected by using the defect detection model.

[0093] In an embodiment of the present invention, the semiconductor wafer surface defect detection device implements the above-described semiconductor wafer surface defect detection method. The device detects a detection beam emitted by an LED detection light source array, which, after spatial filtering, illuminates the surface of the wafer to be tested at a preset angle. The reflected light is divided into a direct light path and a filtered light path. The direct light path passes through a second detector to obtain an original image, while the filtered light path passes through a 4f optical system to undergo spatial frequency filtering, and then a low-pass filtered image is obtained by a third detector. A high-pass filtered image is obtained by performing a differential operation on the original image and the low-pass filtered image, and then subjected to adaptive threshold segmentation and morphological optimization processing to extract image feature data. Finally, the image feature data is input into a defect detection model to determine the defect detection results of the wafer surface. The present invention effectively reduces the interference of background noise through spatial filtering, enhances the contrast of defect edge features, and improves the detection accuracy and classification accuracy of minor defects.

[0094] above Figure 2 The semiconductor wafer surface defect detection device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The semiconductor wafer surface defect detection device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0095] Figure 3FIG3 is a schematic diagram of the structure of a semiconductor wafer surface defect detection device provided by an embodiment of the present invention. The semiconductor wafer surface defect detection device 300 may vary significantly depending on configuration or performance. The device may include one or more central processing units (CPUs) 310 (e.g., one or more processors), memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage medium 330 may be either transient or persistent storage. The program stored in the storage medium 330 may include one or more modules (not shown), each of which may include a series of instructions for operating on the semiconductor wafer surface defect detection device 300. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, executing the series of instructions stored in the storage medium 330 on the semiconductor wafer surface defect detection device 300 to implement the steps of the semiconductor wafer surface defect detection method described above.

[0096] The semiconductor wafer surface defect detection device 300 may further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 3 The structure of the semiconductor wafer surface defect detection device shown does not constitute a limitation on the semiconductor wafer surface defect detection device provided by the present invention, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0097] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, and when the instructions are run on a computer, the computer executes the steps of the semiconductor wafer surface defect detection method.

[0098] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0099] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion 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 storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0100] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting surface defects of a semiconductor wafer, characterized in that: The semiconductor wafer surface defect detection method comprises: Performing spatial filtering on the detection beam emitted by the LED detection light source array, and irradiating the detection beam after spatial filtering at a preset angle to the surface of the wafer to be tested through a preset reflector group; The detection light path reflected from the surface of the wafer to be tested is subjected to a spectroscopic processing by a preset spectroscope to obtain a direct light path and a filtered light path, wherein the light intensity ratio between the direct light path and the filtered light path is 1:1; the direct light path is subjected to a photoelectric conversion processing by a preset second detector to convert the light intensity information of the direct light path into an electrical signal to obtain an original image; the filtered light path is subjected to a focusing processing by a first lens in a preset 4f optical system, and a Fourier spectrum containing spatial frequency information of the surface of the wafer to be tested is obtained on the back focal plane of the first lens; the position of the aperture on the back focal plane of the first lens is adjusted by a three-axis translation stage so that the aperture on the back focal plane of the first lens is The aperture center coincides with the optical axis of the first lens, and the diaphragm is located on the back focal plane of the first lens; the aperture size of the diaphragm is adjusted by a preset adjustment motor so that the aperture of the diaphragm matches the size of the bright spot at the center of the Fourier spectrum, thereby obtaining an adjusted aperture; the Fourier spectrum is frequency-selected by the adjusted diaphragm, and only low-frequency components in the central region of the Fourier spectrum are allowed to pass through, thereby obtaining a spectrum image after spatial frequency filtering; the Fourier spectrum after spatial frequency filtering is imaged by a second lens in a preset 4f optical system, and the light intensity information after imaging is received by a third detector to obtain a low-pass filtered image; performing a difference operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and performing adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data; The image feature data is input into a preset defect detection model, and a defect detection result of the surface of the wafer to be tested is determined by the defect detection model.

2. The method for detecting surface defects of a semiconductor wafer according to claim 1, wherein: The method of performing spatial filtering on the detection light beam emitted by the LED detection light source array and irradiating the detection light beam after the spatial filtering at a preset angle on the surface of the wafer to be tested through a preset reflector group includes: The preset scanning sequence and driving current control each LED in the LED detection light source array row by row to emit a detection light beam; collimating the detection beam and performing spatial filtering on the detection beam through a preset collimating lens device and a pinhole aperture to obtain a detection beam after spatial filtering; The reflection angle of the detection light beam after the spatial filtering process is adjusted by a plurality of preset groups of reflection mirrors, so that the detection light beam after the filtering is incident on the surface of the wafer to be measured at an angle of 45°.

3. The method for detecting surface defects of a semiconductor wafer according to claim 2, wherein: The collimating lens device includes a first converging lens and a second converging lens; The method of performing collimation processing and spatial filtering processing on the detection light beam by using a preset collimating lens device and a pinhole aperture to obtain a detection light beam after spatial filtering processing includes: Converging the detection light beam through the first converging lens to obtain a converged light beam, and spatially adjusting the convergence focus of the converged light beam so that the convergence focus is coaxial with the aperture of the pinhole diaphragm and is located on the incident surface of the pinhole diaphragm; Performing inter-filtering processing on the detection light beam through the pinhole diaphragm to remove scattered light components in the convergent light beam to obtain a filtered divergent light beam; The filtered divergent light beam is collimated by the second converging lens to transform the filtered divergent light beam into a parallel light beam, thereby obtaining a detection light beam after spatial filtering.

4. The method for detecting surface defects of a semiconductor wafer according to claim 1, wherein: The step of performing a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and performing adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data includes: Performing pixel-wise difference calculation on the original image and the low-pass filtered image using an image processor to obtain a high-pass filtered image reflecting edge and detail information of the wafer surface; performing multi-level threshold segmentation processing on the high-pass filtered image according to the grayscale histogram distribution of the high-pass filtered image to obtain an initial segmented image; Performing morphological processing of opening and closing operations on the initial segmented image to eliminate noise points and burrs in the initial segmented image and obtain a segmented image with a smooth contour; A connected domain analysis is performed on the segmented image with smooth contours using a region labeling algorithm, and characteristic parameters of each connected domain are extracted to obtain image feature data.

5. The method for detecting surface defects of a semiconductor wafer according to claim 1, wherein: Inputting the image feature data into a preset defect detection model and determining a defect detection result of the surface of the wafer to be tested by using the defect detection model includes: Performing dimensionality reduction processing on the image feature data using a principal component analysis algorithm to obtain a principal feature vector representing the geometric form of the feature region; Analyzing the main feature vector using a preset multidimensional anomaly detection algorithm based on Mahalanobis distance to obtain a confidence level, and determining whether a characteristic abnormal area exists on the surface of the wafer to be tested based on a preset confidence interval and the confidence level; If there is no characteristic abnormal area, the defect detection result is set as the surface of the wafer to be tested is defect-free; If there is a feature abnormal area, a nonlinear feature transformation is performed on the feature abnormal area through a feature mapping algorithm based on a kernel function to obtain a high-dimensional feature representation containing regional morphology and topological structure information; The high-dimensional feature representation is analyzed by a nearest neighbor classification algorithm and a defect assessment criterion to obtain a defect type determination result and a severity determination result of the feature abnormality area, and the defect type determination result and the severity determination result are used as the defect detection result.

6. A semiconductor wafer surface defect detection device, characterized in that: The semiconductor wafer surface defect detection device comprises: The spatial filtering module is used to perform spatial filtering on the detection beam emitted by the LED detection light source array, and illuminate the surface of the wafer to be tested at a preset angle through a preset reflector group. The spectroscopic module is used to perform spectroscopic processing on the detection light path reflected from the surface of the wafer to be tested through a preset spectroscope to obtain a direct light path and a filtered light path, wherein the light intensity ratio of the direct light path to the filtered light path is 1:1; perform photoelectric conversion processing on the direct light path through a preset second detector, convert the light intensity information of the direct light path into an electrical signal, and obtain an original image; perform focusing processing on the filtered light path through a first lens in a preset 4f optical system, and obtain a Fourier spectrum containing spatial frequency information of the surface of the wafer to be tested on the back focal plane of the first lens; adjust the position of the aperture on the back focal plane of the first lens through a three-axis translation stage so that the optical signal on the back focal plane of the first lens is not reflected; The aperture center of the diaphragm coincides with the optical axis of the first lens, and the diaphragm is located on the back focal plane of the first lens; the aperture size of the diaphragm is adjusted by a preset adjustment motor so that the aperture of the diaphragm matches the size of the bright spot at the center of the Fourier spectrum, thereby obtaining an adjusted aperture; the Fourier spectrum is frequency-selected by the adjusted diaphragm, and only low-frequency components in the central region of the Fourier spectrum are allowed to pass through, thereby obtaining a spectrum image after spatial frequency filtering; the Fourier spectrum after spatial frequency filtering is imaged by a second lens in a preset 4f optical system, and light intensity information after imaging is received by a third detector to obtain a low-pass filtered image; an image processing module, configured to perform a differential operation on the original image and the low-pass filtered image to obtain a high-pass filtered image, and perform adaptive threshold segmentation and morphological optimization processing on the high-pass filtered image to obtain image feature data; The detection module is used to input the image feature data into a preset defect detection model and determine the defect detection result of the surface of the wafer to be tested through the defect detection model.

7. A semiconductor wafer surface defect detection device, characterized in that: The semiconductor wafer surface defect detection device includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the semiconductor wafer surface defect detection device to perform the steps of the semiconductor wafer surface defect detection method according to any one of claims 1 to 5.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the steps of the semiconductor wafer surface defect detection method as described in any one of claims 1 to 5 are implemented.

Images