Wafer detection equipment and detection method

Abstract

The invention provides a wafer detection device and method, and the device comprises a wafer carrying platform which is used for placing a wafer; the detection lens group comprises a two-dimensional image acquisition device for acquiring two-dimensional image data of the wafer and a three-dimensional image acquisition device for acquiring three-dimensional structure data of the wafer; the moving device is connected with the wafer carrying table and is used for adjusting the position and the angle of the wafer; the data processing module is used for receiving the two-dimensional image data acquired by the two-dimensional image acquisition device, extracting a surface abnormal area, receiving the three-dimensional structure data acquired by the three-dimensional image acquisition device, extracting a structure abnormal area and outputting the structure abnormal area; and carrying out fusion processing on the two-dimensional image data and the three-dimensional structure data corresponding to the surface abnormal region and the structure abnormal region to realize wafer anomaly classification, and outputting an anomaly classification result and visual labeling. The wafer detection method is applied to the wafer detection equipment.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of semiconductor manufacturing equipment, and in particular to a wafer detection device and a detection method. BACKGROUND

[0002] In the field of semiconductor manufacturing, the wafer, as the core carrier for integrated circuit manufacturing, its surface quality and structural characteristics directly determine the performance and yield of the final chip. In the wafer production process, process abnormalities are key factors affecting product quality and production efficiency. The traditional detection method mainly relies on the analysis of a single data dimension, which has obvious technical limitations.

[0003] Currently, wafer detection technology mainly falls into two categories: 2D image-based surface detection and 3D structure-based topography detection. 2D detection technology obtains wafer surface images through high-resolution cameras, while 3D detection technology (such as laser triangulation, confocal microscopy, etc.) can accurately present the spatial structure and size parameters of the wafer surface, and has high measurement accuracy for three-dimensional features such as height and warping. However, the above two detection technologies are difficult to accurately classify different defects of the wafer alone. SUMMARY

[0004] The present disclosure provides a wafer detection device and a detection method.

[0005] According to one aspect of the present disclosure, a wafer detection device is provided, comprising: a wafer stage for placing a wafer; a detection lens group comprising a two-dimensional image acquisition device and a three-dimensional image acquisition device, the two-dimensional image acquisition device being configured to acquire two-dimensional image data of the wafer, and the three-dimensional image acquisition device being configured to acquire three-dimensional structure data of the wafer; a moving device connected to the wafer stage, configured to adjust the position and angle of the wafer, so that the wafer moves relative to the two-dimensional image acquisition device and the three-dimensional image acquisition device to complete scanning of the wafer; and a data processing module configured to receive the two-dimensional image data acquired by the two-dimensional image acquisition device, extract surface abnormal regions, receive the three-dimensional structure data acquired by the three-dimensional image acquisition device, extract structure abnormal regions, and fuse the two-dimensional image data corresponding to the surface abnormal regions and the three-dimensional structure data corresponding to the structure abnormal regions to realize wafer abnormality classification and output abnormality classification results and visualized annotations.

[0006] According to the technical scheme of one aspect of the present disclosure, the wafer detection device carries the wafer to be detected through a wafer carrier, and a two-dimensional image acquisition device and a three-dimensional image acquisition device of a detection lens group work cooperatively to capture two-dimensional image data related to wafer surface quality and three-dimensional structure data related to structural features respectively; a moving device adjusts the position and angle of the wafer to ensure that the two acquisition devices can scan the wafer in full area without omission; and a data processing module extracts surface abnormal areas and structural abnormal areas from the two types of data respectively, and then performs directional fusion based on wafer defect characteristics to determine the attributes of different defects and complete classification, and finally outputs the classification results and visualized labels. The technical scheme breaks through the limitation of traditional single-dimensional detection by acquiring two-dimensional image data and three-dimensional structure data, and effectively solves the problem that 2D or 3D single detection cannot accurately classify different wafer defects by fusing two-dimensional image data and three-dimensional structure data, thereby realizing comprehensive capture of wafer surface and structural features and accurate classification of defects.

[0007] According to the wafer detection device of at least one embodiment of the present disclosure, the moving device includes a Y-axis moving mechanism, an X-axis moving mechanism connected with the Y-axis moving mechanism, and a θ-axis rotating mechanism connected with the X-axis moving mechanism, and the wafer carrier is connected with the θ-axis rotating mechanism; wherein the X-axis moving mechanism is used to drive the wafer to move along the X-axis direction, the Y-axis moving mechanism is used to drive the wafer to move along the Y-axis direction, the X-axis direction is perpendicular to the Y-axis direction, and the θ-axis rotating mechanism is used to drive the wafer to rotate around a rotating axis, and the rotating axis is perpendicular to the X-axis direction and the Y-axis direction.

[0008] In the technical scheme of the present embodiment, the X-axis moving mechanism and the Y-axis moving mechanism cooperate to realize two-dimensional translation of the wafer in a horizontal plane, and the θ-axis rotating mechanism realizes rotation of the wafer around an axis perpendicular to the horizontal plane, and the three mechanisms cooperatively adjust the position and angle of the wafer, thereby being able to adjust the relative position of the wafer and the detection lens group, ensure that the detection field of view of the two-dimensional and three-dimensional acquisition devices can completely cover the wafer surface, and avoid missing detection or data misplacement caused by position deviation; at the same time, the rotating adjustment can adapt to the deviation of the placement angle of the wafer, and guarantee the consistency of detection data in different areas, thereby providing a reliable data basis for subsequent data fusion and accurate classification.

[0009] The wafer detection device according to at least one embodiment of the present disclosure, the two-dimensional image acquisition device comprises a two-dimensional detection camera, a two-dimensional detection lens group, a two-dimensional detection light source and a two-dimensional re-determination camera, the two-dimensional detection camera is optically connected with the two-dimensional detection lens group through an optical interface, the center of the photosensitive chip of the two-dimensional detection camera is collinear with the optical axis of the two-dimensional detection lens group, and the two-dimensional detection camera is used for receiving the surface optical image of the wafer transmitted through the two-dimensional detection lens group; the light source of the two-dimensional detection light source is directed to the surface of the wafer stage, and is used for providing an illumination environment for the two-dimensional detection camera; the two-dimensional re-determination camera is optically connected with the two-dimensional detection lens group, the center of the photosensitive chip of the two-dimensional re-determination camera is kept in optical path correspondence with the optical axis of the two-dimensional detection lens group after refraction through a light splitting component, and the image acquisition area of the two-dimensional re-determination camera is the same physical position as the image acquisition area of the two-dimensional detection camera.

[0010] In the technical scheme of the embodiment, the two-dimensional detection light source provides uniform illumination for the two-dimensional detection camera, and eliminates the interference of the wafer surface light shadow; the two-dimensional detection camera receives the surface optical image of the wafer through the two-dimensional detection lens group, and acquires two-dimensional image data; the two-dimensional re-determination camera shares the optical path with the two-dimensional detection lens group through a light splitting component, and performs secondary image acquisition and verification on the suspected abnormal area in the same physical position. The two-dimensional detection camera efficiently captures surface information such as surface texture and micro scratches, and the two-dimensional re-determination camera can exclude pseudo-abnormalities caused by angle shadows and slight contamination, and improve the accuracy of surface abnormal area identification.

[0011] The wafer detection device according to at least one embodiment of the present disclosure, the three-dimensional image acquisition device comprises a three-dimensional detection camera, a three-dimensional detection lens group and a three-dimensional detection light source, the three-dimensional detection camera is optically connected with the three-dimensional detection lens group through an optical interface, the center of the photosensitive chip of the three-dimensional detection camera is collinear with the optical axis of the three-dimensional detection lens group, and the three-dimensional detection camera is used for receiving the surface optical image of the wafer transmitted through the three-dimensional detection lens group; the light source of the three-dimensional detection light source is directed to the surface of the wafer stage, and is used for providing an illumination environment for the three-dimensional detection camera.

[0012] In the technical scheme of the embodiment, the three-dimensional detection light source projects specific light on the wafer surface, so that the spatial structure features (such as height, warping, protrusion and depression) of the wafer form identifiable optical signals, the three-dimensional detection camera receives the optical signals through the three-dimensional detection lens group, and converts the optical signals into three-dimensional structure data reflecting the structure parameters of the wafer. The three-dimensional image acquisition device can capture the three-dimensional topographic information of the wafer through the cooperation of the above components, make up for the lack of depth information in two-dimensional detection, provide key data support for distinguishing “only surface defects”, “only structure abnormalities” and “surface and structure composite defects”, realize more comprehensive defect attribute judgment in cooperation with two-dimensional image data, and further improve the accuracy of wafer abnormality classification.

[0013] According to the wafer detection device of at least one embodiment of the present disclosure, the included angle between the light source irradiation path of the three-dimensional detection light source and the optical axis of the three-dimensional detection lens group is 60±5°.

[0014] In the technical scheme of the present embodiment, the included angle between the irradiation path of the three-dimensional detection light source and the optical axis of the three-dimensional detection lens group is set to 60±5°. This angle can optimize the projection effect of the light source on the wafer surface, reduce the interference of structural shadows on the optical signal, and enable the three-dimensional detection camera to more clearly capture the height difference and structural details of the wafer surface.

[0015] According to another aspect of the present disclosure, a wafer detection method is provided, which is applied to the wafer detection device of any one of the above embodiments and includes the following steps: S1: uploading a wafer to be detected to a wafer stage of the wafer detection device; S2: moving the wafer stage by the moving device to achieve the alignment of the wafer with the detection lens group; S3: driving the wafer stage to move by the moving device, performing two-dimensional scanning and re-determination of the wafer by the two-dimensional image acquisition device, collecting two-dimensional image data, extracting surface abnormal regions, and simultaneously performing three-dimensional scanning of the wafer by the three-dimensional image acquisition device, collecting three-dimensional structure data, and extracting structural abnormal regions; S4: performing fusion processing on the two-dimensional image data and three-dimensional structure data corresponding to the surface abnormal regions and structural abnormal regions to achieve wafer abnormality classification; and S5: outputting the abnormality classification result and visual annotation.

[0016] According to the technical scheme of another aspect of the present disclosure, the wafer to be detected is first uploaded to the wafer stage, and the alignment of the wafer with the detection lens group is completed by the moving device. Then, the moving device drives the wafer stage to move, and the two-dimensional and three-dimensional acquisition devices are scanned synchronously to extract surface abnormal regions and structural abnormal regions, respectively. Then, the corresponding data of the two types of abnormal regions are associated by a directional fusion algorithm, and the surface and structural features are integrated to achieve defect classification. Finally, the classification result and visual annotation are output. This technical scheme breaks through the limitations of 2D or 3D single detection, effectively solves the problem of accurately classifying different wafer defects in single detection, and realizes the full-process closed-loop optimization from data acquisition to result output.

[0017] According to the wafer detection method of at least one embodiment of the present disclosure, step S2 includes: grabbing a wafer alignment point by the detection lens group, calculating the center offset and angle offset of the wafer, adjusting the angle of the wafer by the θ-axis driving mechanism, adjusting the position of the wafer by the X-axis driving mechanism and Y-axis driving mechanism, and achieving the alignment of the wafer with the detection lens group.

[0018] In the technical scheme of the embodiment, the wafer is identified by the detection lens group to identify a wafer preset alignment point (such as an edge mark point), the center offset and the angle offset of the actual position and the standard position are calculated, the angle deviation is corrected through the theta axis, and the position deviation is corrected through the X axis and the Y axis, so that the wafer center is accurately aligned with the detection center of the detection lens group. Its beneficial effects are that the detection error caused by the wafer placement deviation is eliminated, the two-dimensional and three-dimensional data are one-to-one corresponding in the spatial position, the classification confusion caused by the data misplacement is avoided, the accurate position correlation basis is provided for the subsequent fusion classification, and the accuracy of the defect classification is further improved.

[0019] According to the wafer detection method of at least one embodiment of the present disclosure, in step S3, the two-dimensional scanning and rejudgment of the wafer are performed by the two-dimensional image acquisition device, the two-dimensional image data is collected, and the surface abnormal area is extracted. The following steps are included: the moving device drives the wafer stage to move along the preset S-shaped scanning path, and dynamically calculates the number of turns of the S-shaped scanning path based on the field of view of the two-dimensional detection camera and the actual size of the wafer. The two-dimensional image acquisition device performs two-dimensional scanning on the wafer to collect the two-dimensional image data; the measured feature position of the wafer is located by matching the standard wafer sample, the two-dimensional image data of the measured feature position of the wafer is compared with the two-dimensional image data of the corresponding position of the standard wafer sample, and the measured feature position with a comparison difference greater than a preset difference threshold is determined as a surface abnormal area.

[0020] In the technical scheme of the embodiment, the S-shaped scanning path is combined with the dynamic number of turns to calculate, so as to ensure that the two-dimensional detection camera fully covers the surface of the wafer; by comparison with the standard wafer sample, the surface feature difference area is quickly screened as a suspected abnormality; and then the rejudgment camera is used for secondary verification to eliminate false abnormalities and avoid the interference of false abnormalities on the classification result, thereby assisting in accurately distinguishing different surface defects.

[0021] According to the wafer detection method of at least one embodiment of the present disclosure, in step S3, the three-dimensional scanning of the wafer is performed by the three-dimensional image acquisition device, the three-dimensional structure data is collected, and the structure abnormal area is extracted. The following steps are included: the three-dimensional scanning of the wafer is performed by the three-dimensional image acquisition device along the preset scanning path to generate a three-dimensional topographic map, the average height Z1 around the measured feature position and the vertex value Z2 of the measured feature position are calculated, the height value |H| = Z2-Z1 of the measured feature position is calculated, the height value is compared with a preset height threshold, and the measured feature position with the height value not within the height threshold is determined as the structure abnormal area.

[0022] In the technical scheme of the embodiment, the three-dimensional acquisition device synchronously scans along the same preset path as the two-dimensional scanning, ensuring that the two-dimensional and three-dimensional data of the same measured feature position correspond; the height difference is calculated by subtracting the peripheral average height from the vertex value, the structural deviation is quantified, and the structural abnormal region is screened by comparison with the preset threshold value.

[0023] According to the wafer detection method of at least one embodiment of the present disclosure, step S4 includes: finding the edge line of the surface abnormal region by a caliper operator and fitting the shape of the surface abnormal region to calculate the contour size of the surface abnormal region; and classifying wafer abnormalities in combination with the gray value, contour size and height value of the surface abnormal region, and associating the process link corresponding to the abnormalities.

[0024] In the technical scheme of the embodiment, the edge of the surface abnormal region is extracted by the caliper operator and the shape is fitted to calculate the contour size such as width, height and area; the three core parameters of gray value, contour size and height value are integrated, the classification is performed according to the wafer defect characteristic law, the limitation of single parameter judgment is broken through, and different types of defects such as scratches, protrusions and composite defects can be accurately distinguished.

[0025] According to the wafer detection method of at least one embodiment of the present disclosure, the wafer abnormality classification in step S4 includes: based on the fused data after spatial alignment, the following rules are used for abnormal type judgment: protrusion: the height abnormal value is greater than 3 μm of the average height and the area is greater than 100 μm², the gray value in the 2D image is lower than the surrounding area by more than 15%, and the protrusion abnormality is determined; recess: the height abnormal value is lower than 3 μm of the average height and the area is greater than 100 μm², the 2D image shows a dark area and the gray value is higher than the surrounding area by more than 10%, and the recess abnormality is determined; foreign matter: the gray value in the 2D image is abnormal and the width height area threshold is greater than the user-set control threshold, and the foreign matter abnormality is determined; scratch: the width w to length h ratio is less than 0.2 and the area is greater than 500 μm², the 2D image shows linear gray value change, and the scratch abnormality is determined; dwarf bump: the 2D image surface is complete and abnormal, the height value |H| is lower than 20%-40% of the standard value, and the dwarf bump abnormality is determined; wherein each type of abnormality is output to the rejudgment software for manual review by the staff.

[0026] In the technical scheme of the embodiment, the automatic classification of wafer abnormalities is realized by defining accurate multi-parameter combination judgment rules, which significantly improves the accuracy and practicality of the detection system. At the same time, the confidence threshold is set to trigger the manual review mechanism, which ensures the efficiency of automatic detection while ensuring that key defects are not missed, providing a reliable basis for rapid positioning and targeted improvement of process problems.

[0027] According to the wafer detection method of at least one embodiment of the present disclosure, step S5 comprises: the data processing module outputs the measured feature position containing the anomaly, the anomaly type, and the classification result classified based on the degree of influence of the anomaly on the wafer performance, and synchronously outputs the visualization result with two-dimensional defect labeling and three-dimensional structure anomaly labeling.

[0028] In the technical scheme of the present embodiment, the data processing module integrates the key information in the classification process to form a structured result of the anomaly position, the anomaly type, and the performance influence classification; meanwhile, the anomaly in the two-dimensional image and the three-dimensional structure data is visually labeled to intuitively present the defect features, reduce the result interpretation difficulty, make the defect information more intuitive and easy to understand, and help to quickly locate the problem and take targeted measures.

[0029] According to the wafer detection method of at least one embodiment of the present disclosure, step S5 further comprises: analyzing the association between the anomaly type and the process link, outputting the anomaly data to the re-judgment software, and after manual re-checking is completed, generating a distribution atlas with anomaly coding information, and the later production will sort the chips that pass the detection or rework the chips that fail the detection according to the distribution atlas.

[0030] In the technical scheme of the present embodiment, the association analysis between the detected anomaly type and the specific process link is performed, and the wafer distribution atlas with anomaly coding information is generated after manual re-checking, so that the subsequent production process can accurately identify and locate the problem chips. Based on the atlas, the chips that pass the detection can be directly sorted by the later process, or the chips that fail the detection can be reworked, thereby realizing efficient closed-loop linkage between the detection result and the production execution, and significantly improving the collaboration, traceability, and overall processing efficiency of the production process. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings illustrate exemplary embodiments of the present disclosure and together with the general description of the present disclosure given above, and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure. These drawings are included herewith and constitute a part of the present disclosure.

[0032] Figure 1 FIG. 1 is a structural schematic diagram of a wafer detection device according to one embodiment of the present disclosure.

[0033] Figure 2 FIG. 2 is a structural schematic diagram of the wafer detection device when the detection lens group is removed according to one embodiment of the present disclosure.

[0034] Figure 3 FIG. 3 is a structural schematic diagram of the detection lens group according to one embodiment of the present disclosure.

[0035] Figure 4is a schematic diagram of a three-dimensional light source irradiation path at an angle of 60° with a three-dimensional detection lens group optical axis according to an embodiment of the present disclosure.

[0036] Figure 5 is a step flow chart of a wafer detection method according to an embodiment of the present disclosure.

[0037] Figure 6 is a schematic diagram of a two-dimensional detection camera scanning a wafer along an S-shaped scanning path according to an embodiment of the present disclosure.

[0038] Legend of reference signs: 100, wafer stage 200, detection lens group 210, two-dimensional image acquisition device 211, two-dimensional detection camera 212, two-dimensional detection lens group 213, two-dimensional detection light source 214, two-dimensional re-determination camera 215, objective lens converter 220, three-dimensional image acquisition device 221, three-dimensional detection camera 222, three-dimensional detection lens group 223, three-dimensional detection light source 300, moving device 310, X-axis moving mechanism 320, Y-axis moving mechanism 330, θ-axis rotating mechanism 400, frame 410, gantry structure 420, Z-axis driving mechanism DETAILED DESCRIPTION

[0039] The present disclosure will be described in further detail below with reference to the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the related content, and not to limit the present disclosure. In addition, it should be noted that only parts related to the present disclosure are shown in the drawings for ease of description.

[0040] It should be noted that the embodiments in the present disclosure and the features in the embodiments can be combined with each other without conflict. The technical solutions of the present disclosure will be described in detail below with reference to the drawings and in combination with the embodiments.

[0041] Unless otherwise stated, the exemplary implementations / embodiments shown are to be understood as providing exemplary features of various details that provide ways in which the technical concepts of this disclosure can be implemented in practice. Therefore, unless otherwise stated, the features of various implementations / embodiments may be additionally combined, separated, interchanged and / or rearranged without departing from the technical concepts of this disclosure.

[0042] This disclosure relates to wafer inspection technology in the field of semiconductor manufacturing. The wafer inspection equipment and inspection methods disclosed herein are described in detail below with reference to specific implementation scenarios, so that those skilled in the art can clearly understand and implement this technical solution.

[0043] In the existing technology, in the semiconductor manufacturing field, the wafer is the core carrier of integrated circuit manufacturing. Its surface quality and structural characteristics directly determine the performance and yield of the final chip. Process abnormalities are key factors affecting product quality and production efficiency. Traditional detection methods rely on single data dimension analysis, which has obvious technical limitations. At present, wafer inspection technology is mainly divided into surface inspection based on 2D images and morphology inspection based on 3D structures. Neither of these technologies can accurately classify different defects in wafers on its own.

[0044] Figure 1 This is a schematic diagram of a wafer inspection apparatus according to one embodiment of the present disclosure. Figure 2 This is a schematic diagram of the structure of a wafer inspection apparatus according to one embodiment of the present disclosure when removing the inspection lens group.

[0045] Therefore, this disclosure provides a wafer inspection device, such as... Figure 1 and Figure 2 As shown, the wafer inspection equipment mainly consists of a wafer stage 100, an inspection lens group 200, a moving device 300, a data processing module (not shown in the figure), and a rack 400. These components work together to achieve full-dimensional inspection and accurate defect classification of the wafer. The specific structure and connection relationships are as follows: The wafer stage 100 is used to place wafers.

[0046] The inspection lens assembly 200 includes a two-dimensional image acquisition device 210 and a three-dimensional image acquisition device 220. The entire assembly is mounted on the Z-axis drive mechanism 420 of the gantry structure 410 of the rack 400 via a bracket. It can move along the Z-axis direction with the Z-axis drive mechanism 420 to adjust the relative distance with the wafer stage 100, adapting to the inspection requirements of wafers of different thicknesses. The two-dimensional image acquisition device 210 is used to acquire two-dimensional image data of the wafer, and the three-dimensional image acquisition device 220 is used to acquire three-dimensional structural data of the wafer.

[0047] The mobile device 300 is connected with the wafer carrier 100, and is used to adjust the position and angle of the wafer, so as to realize scanning of the wafer by the two-dimensional image acquisition device 210 and the three-dimensional image acquisition device 220.

[0048] The data processing module is used to receive the two-dimensional image data collected by the two-dimensional image acquisition device 210, extract the surface abnormal area, receive the three-dimensional structure data collected by the three-dimensional image acquisition device 220, extract the structure abnormal area, and perform fusion processing on the two-dimensional image data and the three-dimensional structure data corresponding to the surface abnormal area and the structure abnormal area, so as to realize wafer abnormality classification, and output the abnormality classification result and the visualized label.

[0049] The wafer detection equipment in the above technical solution stably bears the wafer to be detected through the wafer carrier 100, forms a dual-mode acquisition architecture through the two-dimensional image acquisition device 210 and the three-dimensional image acquisition device 220 of the detection lens group 200, respectively captures two-dimensional image data related to wafer surface features and three-dimensional structure data related to spatial structure, adjusts the position and angle of the wafer through the mobile device 300, ensures that the dual acquisition devices can perform full-area and dead-angle-free scanning on the wafer, extracts the surface abnormality and the structure abnormality from the two types of data through the data processing module, and then performs directional fusion based on the inherent properties of wafer defects, so as to clearly define the complete characteristics of the defects and complete classification, and finally output the classification result and the visualized label. The wafer detection equipment in the above technical solution breaks through the limitation of traditional single data dimension detection, effectively solves the problem that 2D or 3D single detection is difficult to accurately classify different wafer defects, realizes comprehensive capture of wafer surface and structure features and accurate classification of defects, and has the advantages of breaking through the limitation of traditional single data dimension detection, effectively solving the problem that 2D or 3D single detection is difficult to accurately classify different wafer defects, realizing comprehensive capture of wafer surface and structure features and accurate classification of defects, and the like.

[0050] As Figure 1 and Figure 2As shown, in one embodiment of the above mobile device 300, the mobile device 300 comprises a Y-axis moving mechanism 320, an X-axis moving mechanism 310 connected with the Y-axis moving mechanism 320, and a θ-axis rotating mechanism 330 connected with the X-axis moving mechanism 310, and the wafer carrier 100 is connected with the θ-axis rotating mechanism 330. Among them, the X-axis moving mechanism 310 is used to drive the wafer to move along the X-axis direction, the Y-axis moving mechanism 320 is used to drive the wafer to move along the Y-axis direction, and the X-axis direction is perpendicular to the Y-axis direction. The θ-axis rotating mechanism 330 is used to drive the wafer to rotate around the rotation axis, and the rotation axis is perpendicular to the X-axis direction and the Y-axis direction. In operation, the X-axis and Y-axis moving mechanism 320 cooperates to realize the translation of the wafer in the horizontal plane, and the θ-axis rotating mechanism 330 realizes the angle adjustment of the wafer around the vertical axis, and the three cooperate to complete the accurate alignment and full-area scanning coverage of the wafer and the detection lens group 200. Among them, the X-axis moving mechanism 310 and the Y-axis moving mechanism 320 can adopt a linear motor to realize high-precision translation in cooperation with a ball screw, or adopt an existing linear moving module to drive; the θ-axis rotating mechanism 330 can be driven by a servo motor, and a harmonic reducer is matched to ensure the rotation accuracy; the moving stroke of the X-axis and the Y-axis can be designed according to the size of the adapted wafer to ensure that the detection requirements of the largest size wafer can be covered.

[0051] Figure 3 is a structural schematic diagram of a detection lens group according to one embodiment of the present disclosure.

[0052] As Figure 3As shown, in one embodiment of the two-dimensional image acquisition device 210, the two-dimensional image acquisition device 210 includes a two-dimensional inspection camera 211, a two-dimensional inspection lens group 212, a two-dimensional inspection light source 213, and a two-dimensional re-judgment camera 214. The two-dimensional inspection lens group 212 serves as the core optical carrier, and is provided with a front objective lens, a lens barrel, and a rear optical interface in sequence along the optical axis. The outer wall of the lens barrel is mechanically connected to an objective lens converter 215, which can be equipped with inspection objective lenses of different magnifications to meet the inspection needs of fine defects or macroscopic areas on the wafer surface. The two-dimensional inspection camera 211 forms an optical connection with the two-dimensional inspection lens group 212 through the rear optical interface. The center of its photosensitive chip is collinear with the optical axis of the two-dimensional inspection lens group 212, enabling it to accurately receive the optical image of the wafer surface transmitted by the two-dimensional inspection lens group 212. The two-dimensional detection light source 213 is a white ring light source, which is mechanically connected to the outer periphery of the front objective lens of the two-dimensional detection lens group 212 through a light source bracket. The central hole of the ring light source is collinear with the optical axis of the detection objective lens, and the illumination direction is perpendicular to the surface of the wafer stage 100, providing uniform illumination for the two-dimensional detection camera 211 and effectively eliminating light and shadow interference on the wafer surface. The two-dimensional re-judgment camera 214 is optically connected to the lens barrel of the two-dimensional detection lens group 212 through a beam splitter assembly. The beam splitter assembly is embedded inside the lens barrel and located in the optical path between the front objective lens and the rear optical interface, which can split the incident optical image portion to the photosensitive chip of the two-dimensional re-judgment camera 214. The installation height of the two-dimensional re-judgment camera 214 is parallel to that of the two-dimensional detection camera 211. The center of its photosensitive chip is aligned with the optical axis of the two-dimensional detection lens group 212 after being refracted by the optical path of the beam splitter assembly, ensuring that the re-judgment area and the detection area of ​​the two-dimensional detection camera 211 are in the same physical location. Among them, the two-dimensional inspection camera 211 can be a high-resolution industrial camera, and the photosensitive chip can be CMOS or CCD type; the two-dimensional inspection light source 213 can be a ring LED light source, which can be adapted to the wafer surface with different reflectivity by adjusting the brightness; the beam splitting component can be a semi-transparent and semi-reflective prism, which can distribute the light path to the two-dimensional inspection camera 211 and the two-dimensional re-judgment camera 214 as needed; the resolution of the two-dimensional re-judgment camera 214 can be the same as or higher than that of the two-dimensional inspection camera 211 to ensure the clarity of the re-judged image.

[0053] like Figure 3 As shown, exemplarily, the three-dimensional image acquisition device 220 includes a three-dimensional inspection camera 221, a three-dimensional inspection lens group 222, and a three-dimensional inspection light source 223. The three-dimensional inspection camera 221 forms an optical connection with the three-dimensional inspection lens group 222 through an optical interface, and the center of its photosensitive chip is collinear with the optical axis of the three-dimensional inspection lens group 222, for receiving the optical image of the wafer surface transmitted by the three-dimensional inspection lens group 222. The illumination direction of the three-dimensional inspection light source 223 is directed towards the surface of the wafer stage 100, providing a suitable lighting environment for the three-dimensional inspection camera 221.

[0054] Figure 4is a schematic diagram of a three-dimensional light source irradiation path at an angle of 60° with a three-dimensional detection lens group optical axis according to an embodiment of the present disclosure.

[0055] As shown in Figure 4 To optimize the acquisition effect of three-dimensional structure data, the angle between the light source irradiation path of the three-dimensional detection light source 223 and the optical axis of the three-dimensional detection lens group 222 is set to 60±5°. This angle can reduce the interference of structure shadows on the optical signal, so that the three-dimensional detection camera 221 can capture the height difference and structure details of the wafer surface more clearly. The angle can be set by adjusting the angle of the mounting bracket of the three-dimensional detection light source 223. In actual application, a suitable angle can be selected within the range of 55° to 65° according to the structure characteristics of the wafer surface. For example, for a wafer with high surface flatness, a 60° angle can be used. For a wafer with complex protruding structures, the angle can be adjusted to 62° or 58° to obtain better optical signals.

[0056] Figure 5 is a step flowchart of a wafer detection method according to an embodiment of the present disclosure.

[0057] As shown in Figure 5 The present disclosure also provides a wafer detection method, which is applied to the wafer detection device of any one of the above and includes the following steps: S1: uploading a wafer to be detected to a wafer stage 100 of a wafer detection device; S2: moving the wafer stage 100 by a moving device 300 to align the wafer with a detection lens group 200; S3: moving the wafer stage 100 by the moving device 300, performing two-dimensional scanning and rejudgment on the wafer by a two-dimensional image acquisition device 210, collecting two-dimensional image data, extracting surface abnormal areas, and simultaneously performing three-dimensional scanning on the wafer by a three-dimensional image acquisition device 220, collecting three-dimensional structure data, and extracting structure abnormal areas; S4: fusing the two-dimensional image data and the three-dimensional structure data corresponding to the surface abnormal areas and the structure abnormal areas to realize wafer abnormality classification; S5: outputting the abnormality classification result and visualizing the label.

[0058] The wafer detection method of the above technical solution first uploads the wafer to be detected to the wafer loading platform 100, and completes the accurate alignment of the wafer and the detection lens group 200 through the moving device 300; then the moving device 300 drives the wafer loading platform 100 to move, and the two-dimensional and three-dimensional acquisition devices are synchronously scanned to complete the extraction of the surface abnormal area and the structure abnormal area respectively; then the corresponding data of the two types of abnormal areas are associated through the directional fusion algorithm, and the surface and structure characteristics are integrated to realize defect classification; finally, the classification results and visualized labels are output. The technical solution effectively solves the problem that separate detection cannot accurately classify different defects of the wafer, and realizes the whole-process closed-loop optimization from data acquisition to result output.

[0059] Further, step S2 includes: grabbing the wafer alignment point by the detection lens group 200, calculating the center offset and angle offset of the wafer, adjusting the angle of the wafer through the θ-axis driving mechanism, adjusting the position of the wafer through the X-axis driving mechanism and the Y-axis driving mechanism, and realizing the position alignment of the wafer and the detection lens group 200. The working principle is: using the detection lens group 200 to identify the preset alignment point of the wafer, calculating the center offset and angle offset of the actual position and the standard position, correcting the angle deviation through the θ-axis, correcting the position deviation through the X-axis and the Y-axis, and accurately aligning the center of the wafer with the detection center of the detection lens group 200. During the adjustment process, a step-by-step adjustment strategy can be adopted, i.e., first coarsely adjusting the angle and position, and then finely adjusting to ensure the alignment accuracy.

[0060] Exemplarily, step S2 includes: grabbing at least three uniformly distributed wafer alignment points on the wafer surface through the high-resolution detection lens group multiple objective lens, calculating the wafer alignment point center coordinates by using the sub-pixel level Hough transform algorithm, and fitting the actual center position of the wafer and the angle offset through the least squares method. The Hough transform principle is to map the points in the image space to the parameter space, and find the points gathered in clusters in the parameter space, which correspond to a certain shape in the image space. For circle detection, a three-dimensional parameter space (a, b, r) is constructed, where (a, b) represents the coordinates of the center, and r represents the radius of the circle. Each point (x, y) in the image space will correspond to a conical surface in the parameter space. When enough points in the parameter space intersect on a conical surface, it means that a circle is found in the image space.

[0061] Furthermore, when a wafer alignment point recognition failure is detected or an alignment error greater than 20μm is detected, the system automatically initiates a re-alignment mechanism: first, it switches to a higher magnification objective lens to re-identify the wafer alignment point; if this still fails, it triggers the stage micro-vibration function to eliminate tiny particles between the wafer and the stage through micro-vibration. The vibration lasts for 0.5-2 seconds before attempting alignment again. If alignment fails three times consecutively, the system automatically marks the wafer as "alignment abnormal" and moves it to the abnormal processing area. This scheme, as a mechanism for handling alignment failures, improves the robustness and automation of the system, preventing the entire production line from stopping due to a single wafer problem.

[0062] Figure 6 This is a schematic diagram of a two-dimensional inspection camera scanning a wafer along an S-shaped scanning path according to one embodiment of the present disclosure.

[0063] like Figure 6 As shown, in step S3, the two-dimensional image acquisition device 210 performs two-dimensional scanning and re-judgment of the wafer, collects two-dimensional image data, and extracts surface abnormal areas, including the following steps: the moving device 300 drives the wafer stage 100 to move along a preset S-shaped scanning path, and dynamically calculates the number of reversals of the S-shaped scanning path based on the field of view of the two-dimensional inspection camera 211 and the actual size of the wafer; the two-dimensional image acquisition device 210 performs two-dimensional scanning of the wafer and collects two-dimensional image data; at the same time, it matches a standard wafer sample (i.e., the Golden sample) to locate the measured feature position of the wafer, compares the two-dimensional image data of the measured feature position of the wafer with the two-dimensional image data of the corresponding position of the standard wafer sample, and determines the measured feature position with a comparison difference greater than a preset difference threshold as a suspected surface abnormal area; the two-dimensional re-judgment camera 214 re-judges the suspected surface abnormal area, and retains the real surface abnormal area after eliminating false abnormalities. The S-shaped scanning path combined with the dynamic reversal calculation ensures that the two-dimensional inspection camera 211 fully covers the wafer surface. By comparing with standard wafer samples, areas with surface feature differences are quickly screened as suspected anomalies; then, a secondary verification using a re-evaluation camera eliminates false anomalies. The number of foldbacks in the S-shaped scanning path can be dynamically calculated based on the wafer diameter and camera field of view, ensuring that adjacent scanning paths are not missed and have reasonable overlap. Standard wafer samples are defect-free, qualified wafers, and their two-dimensional image data is pre-stored in the data processing module. The preset difference threshold can be flexibly adjusted according to wafer type and detection accuracy requirements. During re-evaluation, false anomalies are eliminated by verifying the edge stability and grayscale value distribution consistency of the abnormal area.

[0064] Further, in step S3, the three-dimensional scanning of the wafer by the three-dimensional image acquisition device 220 is synchronized to collect three-dimensional structure data and extract the structure abnormal area, including the following steps: the three-dimensional image acquisition device 220 performs three-dimensional scanning on the wafer along a preset scanning path to generate a three-dimensional topographic map, calculates the average height Z1 around the measured feature position and the vertex value Z2 of the measured feature position, calculates the height value |H| = Z2-Z1 of the measured feature position, compares the height value with the preset height threshold, and determines that the measured feature position with a height value not within the height threshold is a structure abnormal area. The three-dimensional acquisition device synchronously scans along the same preset path as the two-dimensional scanning to ensure that the two-dimensional and three-dimensional data of the same measured feature position correspond; the height difference is calculated by subtracting the average height around the vertex value, the structure deviation is quantified, and the structure abnormal area is screened by comparing with the preset threshold. Among them, the average height Z1 around the measured feature position can be calculated by selecting the average height of a certain range of pixel points around the measured feature position, and this range can be set according to the size of the measured feature. The preset height threshold can be set according to the process requirements of the wafer to distinguish between normal structure deviation and abnormal structure deviation.

[0065] Exemplarily, step S4 includes: finding the edge line of the surface abnormal area by the caliper operator and fitting the shape of the surface abnormal area, and calculating the contour dimensions such as width, height, and area; the data processing module integrates the three core parameters of the gray value, the contour dimension, and the height value in the three-dimensional structure data of the surface abnormal area, and classifies the wafer abnormalities in combination with the inherent characteristics of wafer defects (such as surface scratches, which are mostly manifested as abnormal gray value, long strip-shaped appearance, and normal height value, and protruding defects, which are mostly manifested as normal gray value, regular appearance, and excessive height value); and simultaneously, according to the classification result, the corresponding process link is associated, for example, the surface scratch abnormality is associated with the raw material processing link, and the size deviation abnormality is associated with the equipment calibration link, to provide a clear direction for subsequent process optimization. Among them, the caliper operator can be an edge detection operator, and the shape fitting can be rectangular, circular, or irregular shape fitting according to the actual profile of the abnormal area. The process link association can be based on the preset corresponding relationship of the wafer production process, for example, the surface scratch is associated with the raw material processing or handling link, and the size deviation abnormality is associated with the equipment calibration link.

[0066] Further, the S4: fusing the two-dimensional image data corresponding to the surface abnormal area and the structure abnormal area and the three-dimensional structure data includes spatial alignment of the two-dimensional image data and the three-dimensional structure data, specifically including: using a calibration board to perform joint calibration of a 2D camera and a 3D detection camera, obtaining 2D actual center coordinates and 3D actual center coordinates of at least 9 feature points; calculating the affine transformation matrix of the 2D actual center coordinates and the 3D actual center coordinates to obtain the 2D / 3D position conversion model parameter initial value; removing random errors of the 2D / 3D position conversion model parameter initial value through the RANSAC algorithm, and retaining parameters with an inner point proportion greater than 80%; after calculating the 2D / 3D position conversion model parameters of all actual center coordinates, performing data regression analysis through the least square method to obtain the affine matrix of the 2D image and the 3D topographic map; after spatial alignment, the registration error of the 2D image and the 3D point cloud at the same position is less than 10μm, ensuring the accuracy of subsequent fusion processing.

[0067] As an example, the wafer abnormality classification in step S4 includes: based on the fusion data after spatial alignment, determining the abnormal type through the following rules: Bump: the height abnormal value is greater than the average height by 3μm or more, the area is greater than 100μm², the gray value in the 2D image is lower than the surrounding area by 15% or more, and the bump abnormality is determined; Depression: the height abnormal value is lower than the average height by 3μm or more, the area is greater than 100μm², the 2D image shows a dark area, and the gray value is higher than the surrounding area by 10% or more, and the depression abnormality is determined; Foreign matter: the gray value in the 2D image is abnormal, and the width, height, and area threshold value is greater than the user-set control threshold value, and the foreign matter abnormality is determined; Scratch: the ratio of width w to length h is less than 0.2, and the area is greater than 500μm², the 2D image shows linear gray value change, and the scratch abnormality is determined; Dwarf bump: the 2D image surface is complete and has no abnormality, the height value |H| is lower than 20%-40% of the standard value, and the dwarf bump abnormality is determined; Wherein, each type of abnormality is output to the review software, and the staff performs manual review.

[0068] Exemplarily, step S4 further includes the processing of excluding false abnormalities: when a suspected scratch is detected in the 2D image but the 3D data shows that the absolute value of the height difference H is less than 0.05 μm, it is determined to be a false abnormality caused by light difference; when the 3D data shows a height deviation but the corresponding area in the 2D image has a gray value change of less than 5% and no obvious edge features, it is determined to be a normal structure deviation; when the abnormal area detected in the 2D image has no corresponding structure abnormality in the 3D data, and the matching degree of the abnormal area with the known wafer surface texture feature is greater than 90%, it is determined to be a surface texture misjudgment; through the above rules, the misjudgment rate is reduced to below 0.5%, thus reducing the misjudgment rate and improving the classification accuracy.

[0069] Exemplarily, step S5 includes: the data processing module outputs the measured feature position of the abnormality, the abnormality type, and the classification result based on the performance influence degree of the abnormality on the wafer, and synchronously outputs the visualization result with two-dimensional defect labeling and three-dimensional structure abnormality labeling. The working principle is: the data processing module integrates the key information in the classification process to form a structured result of “abnormality position-abnormality type-performance influence classification”; at the same time, the abnormalities in the two-dimensional image and the three-dimensional structure data are visually labeled to intuitively present the defect features. Among them, the performance influence degree classification can be divided into different levels according to the influence of the abnormality on the chip performance, such as fatal defect, repairable defect, and slight defect. The two-dimensional defect labeling can use a frame of a specific color (such as a red frame) to identify, and the text labeling abnormality type (such as “scratch” and “missing printing”). The three-dimensional structure abnormality labeling can use a color cloud chart, different colors correspond to different height deviation ranges, such as red area representing height over standard, blue area representing height insufficient, and labeling structure abnormality type (such as “warpage” and “aperture too large”), which is convenient for intuitive identification of the distribution and severity of the structure abnormality.

[0070] Exemplarily, the classification result can be summarized by a table, which presents the “measured feature position of the abnormality, the abnormality type, and the classification result based on the performance influence degree of the abnormality on the wafer” (i.e. coded information) in the form of a table.

[0071] As an example, the anomaly classification results output in step S5 include: generating a structured data table containing anomaly ID, anomaly location coordinates, anomaly type, anomaly size parameters, height parameters, confidence level, and coding information; where the coding is divided into four levels based on the degree of impact of the anomaly on product performance: Level A represents minor anomalies with no impact on product performance, with anomaly height difference H < 0.1 μm and area < 50 μm²; Level B represents moderate anomalies that may affect product reliability, with 0.1 μm ≤ H < 0.5 μm or 50 μm² ≤ area < 500 μm²; Level C represents severe anomalies that affect product function, with 0.5 μm ≤ H < 1.0 μm or 500 μm² ≤ area < 2000 μm²; Level D represents major anomalies that must be scrapped, with H ≥ 1.0 μm or area ≥ 2000 μm²; the coding information is integrated with the MES system to automatically trigger the corresponding process handling flow.

[0072] For example, the visualization annotation in step S5 includes: marking the location of surface defects on the 2D image with red boxes, the color of which varies according to the BinCode level: green for level A, yellow for level B, orange for level C, and red for level D; displaying the height deviation on the 3D height cloud map using a heatmap, with red areas indicating height exceeding the standard (H>0) and blue areas indicating insufficient height (H<0), the color intensity being proportional to the absolute value of the deviation; generating an interactive report that allows clicking on any anomaly point to view the 2D / 3D data comparison, the basis for anomaly judgment, and suggested process improvement measures; the visualization results are stored in JSON format, containing metadata such as coordinates, type, and size, facilitating subsequent SPC statistical analysis.

[0073] As an example, step S5 also includes the correlation analysis between the anomaly type and the process: the anomaly data is output to the review software, and after manual review, a distribution map with anomaly coding information is generated. The subsequent production will use this distribution map to sort only the chips that pass the inspection or rework the chips that fail the inspection.

[0074] The working process of the wafer detection device and the detection method is as follows: first, the wafer to be detected is stably placed on the wafer table 100, ensuring that the wafer is attached to the table without deviation; then the detection lens group 200 captures the wafer alignment point, calculates the center offset and angle offset, and adjusts through the X-axis, Y-axis and theta-axis mechanisms of the moving device 300 to realize the accurate alignment of the wafer and the detection lens group 200; after alignment, the moving device 300 drives the wafer table 100 to move along the preset S-shaped scanning path, the two-dimensional image acquisition device 210 and the three-dimensional image acquisition device 220 are started simultaneously, the two-dimensional image acquisition device 210 acquires two-dimensional image data of the wafer surface, compares with the standard wafer sample to screen suspected surface abnormal areas, and then retains the real surface abnormal area after rejudgment by the two-dimensional rejudgment camera 214, the three-dimensional image acquisition device 220 synchronously generates a three-dimensional topographic map, calculates the height value of the measured feature position and compares it with the preset threshold value, and extracts the structural abnormal area; after scanning is completed, the data processing module fits the shape of the surface abnormal area and calculates the contour size by using a caliper operator, integrates the gray value, contour size and height value of the three-dimensional structure of the surface abnormal area, classifies the wafer abnormalities and correlates the corresponding process links; finally, the data processing module outputs the structured results including abnormal position, abnormal type, performance influence classification, and simultaneously outputs the visual results with two-dimensional defect labeling and three-dimensional structural abnormal labeling, completing the entire detection process.

[0075] In summary, the wafer detection device and the detection method of the present disclosure have the following advantages by fusing two-dimensional image data and three-dimensional structure data: Firstly, complementary abnormality recognition coverage is improved: 2D data captures surface defects, 3D data identifies spatial structure abnormalities, avoiding single dimension misjudgment (such as some electronic components having both surface solder scratches and pin height exceeding standards, which cannot be completely identified by single data). Secondly, misjudgment rate is reduced and classification accuracy is improved: combining 3D spatial parameters can exclude pseudo-abnormalities caused by "angle shadow" and "light difference" in 2D data; combining 2D surface information can distinguish between "normal structure deviation" and "abnormalities associated with surface defects" in 3D data (such as the same part height deviation, if accompanied by 2D surface scratches, it is determined as "process damage type abnormality", if there is no surface problem, it is determined as "size calibration type abnormality"). Finally, it adapts to customer process analysis needs: customers not only need to "identify abnormalities", but also need to "clear abnormal types and sources", and the integrated data classification can directly correlate process links (such as "surface scratches + normal size" are mostly from raw material processing links, and "size deviation + no surface defects" are mostly from equipment calibration links), providing targeted improvement for customers.

[0076] In the description of the present specification, the description of the terms "one embodiment / way", "some embodiments / ways", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment / way or example are included in at least one embodiment / way or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / way or example. Also, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments / ways or examples. Furthermore, the person skilled in the art can combine and combine the different embodiments / ways or examples described in the present specification and the features of the different embodiments / ways or examples, without contradiction.

[0077] Those skilled in the art will understand that the above embodiments are merely for the purpose of clearly illustrating the present disclosure, and are not intended to limit the scope of the present disclosure. Other changes or modifications can be made on the basis of the above disclosure, and these changes or modifications are still within the scope of the present disclosure.

Claims

1. A wafer inspection device, characterized in that, include: A wafer stage, wherein the wafer stage is used to place a wafer; The inspection lens assembly includes a two-dimensional image acquisition device and a three-dimensional image acquisition device. The two-dimensional image acquisition device is used to acquire two-dimensional image data of the wafer, and the three-dimensional image acquisition device is used to acquire three-dimensional structural data of the wafer. A moving device, connected to the wafer stage, is used to adjust the position and angle of the wafer so that the wafer moves relative to the two-dimensional image acquisition device and the three-dimensional image acquisition device to complete the scanning of the wafer. as well as The data processing module is used to receive the two-dimensional image data acquired by the two-dimensional image acquisition device, extract the surface abnormal region, receive the three-dimensional structural data acquired by the three-dimensional image acquisition device, extract the structural abnormal region, and perform fusion processing on the two-dimensional image data and three-dimensional structural data corresponding to the surface abnormal region and the structural abnormal region to realize wafer abnormality classification, and output the abnormality classification result and visual annotation.

2. The wafer inspection equipment according to claim 1, characterized in that, The moving device includes a Y-axis moving mechanism, an X-axis moving mechanism connected to the Y-axis moving mechanism, and a θ-axis rotating mechanism connected to the X-axis moving mechanism; the wafer stage is connected to the θ-axis rotating mechanism. The X-axis moving mechanism is used to drive the wafer to move along the X-axis direction, the Y-axis moving mechanism is used to drive the wafer to move along the Y-axis direction, the X-axis direction is perpendicular to the Y-axis direction, and the θ-axis rotating mechanism is used to drive the wafer to rotate around a rotation axis, the rotation axis is perpendicular to both the X-axis direction and the Y-axis direction.

3. The wafer inspection equipment according to claim 1, characterized in that, The two-dimensional image acquisition device includes a two-dimensional inspection camera, a two-dimensional inspection lens group, a two-dimensional inspection light source, and a two-dimensional re-judgment camera. The two-dimensional inspection camera is optically connected to the two-dimensional inspection lens group through an optical interface. The center of the photosensitive chip of the two-dimensional inspection camera is collinear with the optical axis of the two-dimensional inspection lens group, and is used to receive the surface optical image of the wafer transmitted by the two-dimensional inspection lens group. The illumination direction of the two-dimensional inspection light source is pointed towards the surface of the wafer stage, and is used to provide an illumination environment for the two-dimensional inspection camera. The two-dimensional re-judgment camera is optically connected to the two-dimensional inspection lens group. After the light path of its photosensitive chip is refracted by the beam splitter, it maintains the optical path correspondence with the optical axis of the two-dimensional inspection lens group. The image acquisition area of ​​the two-dimensional re-judgment camera and the image acquisition area of ​​the two-dimensional inspection camera are at the same physical location.

4. The wafer inspection equipment according to claim 1, characterized in that, The three-dimensional image acquisition device includes a three-dimensional inspection camera, a three-dimensional inspection lens group, and a three-dimensional inspection light source. The three-dimensional inspection camera is optically connected to the three-dimensional inspection lens group through an optical interface. The center of the photosensitive chip of the three-dimensional inspection camera is collinear with the optical axis of the three-dimensional inspection lens group, and is used to receive the surface optical image of the wafer transmitted by the three-dimensional inspection lens group. The illumination direction of the three-dimensional inspection light source is pointed towards the surface of the wafer stage, and is used to provide an illumination environment for the three-dimensional inspection camera.

5. The wafer inspection equipment according to claim 4, characterized in that, The angle between the illumination path of the three-dimensional detection light source and the optical axis of the three-dimensional detection mirror group is 60±5°.

6. A wafer inspection method, characterized in that, Applied to the wafer inspection equipment as described in any one of claims 1 to 5, and comprising the following steps: S1: The wafer to be inspected is uploaded to the wafer stage of the wafer inspection equipment; S2: The wafer stage is moved by the moving device to align the wafer with the detection lens group; S3: Drive the wafer stage to move through the moving device, perform two-dimensional scanning on the wafer through the two-dimensional image acquisition device, collect two-dimensional image data, extract abnormal areas on the surface, and simultaneously perform three-dimensional scanning on the wafer through the three-dimensional image acquisition device, collect three-dimensional structural data, and extract abnormal areas in the structure. S4: The two-dimensional image data and three-dimensional structural data corresponding to the surface abnormality region and the structural abnormality region are fused to achieve wafer abnormality classification; S5: Output anomaly classification results and visual annotations.

7. The wafer inspection method according to claim 6, characterized in that, Step S2 includes: using the detection lens group to grasp the wafer alignment point, calculating the wafer's center offset and angle offset, adjusting the wafer angle using the θ-axis drive mechanism, and adjusting the wafer position using the X-axis drive mechanism and the Y-axis drive mechanism to achieve alignment between the wafer and the detection lens group.

8. The wafer inspection method according to claim 6, characterized in that, In step S3, the wafer is scanned and re-judged using a two-dimensional image acquisition device to collect two-dimensional image data and extract surface abnormal areas. This includes the following steps: the moving device drives the wafer stage to move along a preset S-shaped scanning path, and dynamically calculates the number of reversals of the S-shaped scanning path based on the field of view of the two-dimensional detection camera and the actual size of the wafer; the two-dimensional image acquisition device scans the wafer in two dimensions to collect the two-dimensional image data; the measured feature position of the wafer is located by matching a standard wafer sample; the two-dimensional image data of the measured feature position of the wafer is compared with the two-dimensional image data of the corresponding position of the standard wafer sample; and the measured feature position with a comparison difference greater than a preset difference threshold is determined to be a surface abnormal area.

9. The wafer inspection method according to claim 8, characterized in that, In step S3, the wafer is simultaneously scanned in three dimensions using a three-dimensional image acquisition device to collect three-dimensional structural data and extract structurally abnormal regions. This includes the following steps: the wafer is scanned in three dimensions along the preset scanning path using the three-dimensional image acquisition device to generate a three-dimensional topography image; the average height Z1 around the measured feature location and the vertex value Z2 of the measured feature location are calculated; the height value |H| of the measured feature location is calculated as Z2 - Z1; the height value is compared with a preset height threshold; and the measured feature location whose height value is not within the height threshold is determined to be the structurally abnormal region.

10. The wafer inspection method according to claim 9, characterized in that, Step S4 includes: The edge lines of the surface anomaly region are found using a caliper operator, and the shape of the surface anomaly region is fitted. The external dimensions of the surface anomaly region are calculated. The wafer anomalies are classified by combining the gray value, external dimensions, and height value of the surface anomaly region.

11. The wafer inspection method according to claim 10, characterized in that, Step S4, which involves classifying wafer anomalies, includes determining the anomaly type based on the spatially aligned fused data using the following rules: Protrusions: If the height of the abnormal value is more than 3μm greater than the average height and the area is greater than 100μm², and the gray value in the 2D image is more than 15% lower than the surrounding area, it is judged as a protrusion anomaly. Depression: A depression is defined as an abnormal area with a height that is more than 3 μm lower than the average height and an area that is greater than 100 μm². In a 2D image, it appears as a dark area with a gray value that is more than 10% higher than the surrounding area. Foreign object: If the grayscale value of a 2D image is abnormal and the width, height and area thresholds are greater than the user-set control thresholds, it is determined to be a foreign object. Scratches: The ratio of width w to length h is less than 0.2 and the area is greater than 500μm². They appear as linear grayscale changes in 2D images and are judged as scratches. Dwarf bumps: If the surface of the 2D image is intact and without abnormalities, but the height value |H| is 20%-40% lower than the standard value, it is judged as a dwarf bump abnormality; Each type of anomaly is output to the review software for manual review by staff.

12. The wafer inspection method according to claim 10, characterized in that, Step S5 includes: The data processing module outputs the location of the abnormal measured features, the type of abnormality, and the classification results based on the degree of impact of the abnormality on wafer performance. It also outputs visualization results with two-dimensional defect annotations and three-dimensional structural abnormality annotations.

13. The wafer inspection method according to claim 11, characterized in that, Step S5 also includes: analyzing the correlation between the anomaly type and the process link, outputting the anomaly data to the review software, and generating a distribution map with anomaly coding information after manual review. The subsequent production will use this distribution map to sort only the chips that pass the inspection or rework the chips that fail the inspection.

Images