Ultrasonic displacement vision collaborative wafer detection and edging system

Abstract

The application belongs to the technical field of wafer processing, and specifically discloses an ultrasonic displacement vision collaborative wafer detection and edge grinding system. The application coordinates the grinding mechanism, the wafer carrier and the component integrating displacement and vision detection through the control module to realize collaborative detection and edge grinding. Specifically, the displacement detection module obtains the distance between the wafer and the grinding mechanism in real time, providing an accurate physical reference for compensation control; the vision detection module captures continuous images in the wafer rotation, so that the control module can extract micro-texture features to identify the offset state of the wafer. The combination of the two realizes comprehensive real-time perception of the grinding condition and overcomes the limitation of insufficient perception. Finally, the control module fuses the displacement and offset information to generate edge grinding instructions, dynamically controls the rigid body transformation of the grinding mechanism and the wafer carrier in linkage, and thus directly converts the real-time perception information into compensation actions, effectively solving the problem of uneven grinding caused by insufficient dynamic compensation capability.

Description

Technical Field

[0001] This application belongs to the field of wafer processing technology, and more specifically, relates to an ultrasonic displacement vision-assisted wafer inspection and edge grinding system. Background Technology

[0002] Silicon-based semiconductor materials are key functional materials in the microelectronics and information industry. Wafers, as high-purity, circular wafers of semiconductor material, are typically made of single-crystal silicon and serve as substrates for integrated circuits and other microelectronic devices. Before packaging, wafers require back-side thinning. During this thinning process, when grinding reaches the outer edge of the wafer, the presence of the outer arc can easily create sharp peaks, leading to cracks in the ground area. To prevent wafer breakage after thinning, it is necessary to minimize residual stress on the surface and avoid wafer warping. Pre-grinding the edges of the wafer before thinning, i.e., edge grinding, can effectively reduce crack formation during thinning, thereby improving yield, enhancing component performance, and extending lifespan.

[0003] Currently, self-rotating wafer grinding machines have become the mainstream. To improve processing efficiency, commercial silicon wafer grinding equipment often adopts a multi-spindle structure, such as installing two sets of spindles and grinding wheels on the machine. The two grinding wheels are used alternately, and while one is in use, the other can be used for correction and inspection. At the same time, the failure of a single grinding wheel will not cause the processing system to stop. However, due to insufficient real-time sensing and dynamic compensation capabilities for the working status of the grinding wheels, uneven grinding and poor wafer edge morphology quality are prone to occur. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide an ultrasonic displacement vision-assisted wafer inspection and edge grinding system, which aims to solve the problems of insufficient real-time perception and dynamic compensation capabilities due to limitations in the working state of the grinding wheel, which easily leads to uneven grinding and poor wafer edge morphology quality.

[0005] To achieve the above objectives, in a first aspect, this application provides an ultrasonic displacement-vision coordinated wafer inspection and edge grinding system, comprising: a control module, a grinding mechanism, components disposed on the grinding mechanism, and a wafer stage; the grinding mechanism, components, and wafer stage are all communicatively connected to the control module; the grinding mechanism is disposed parallel above the wafer stage so that the working area of ​​the grinding mechanism is aligned with the wafer stage; the wafer stage is used to adsorb wafers; the components include: a displacement detection module and a vision inspection module; the displacement detection module is used to detect the displacement distance from the wafer to the grinding mechanism and feed it back to the control module; The vision inspection module is used to acquire continuous surface images during wafer rotation; the control module is used to extract micro-texture features based on the continuous surface images and to acquire the wafer offset state based on the micro-texture features; the control module is also used to generate target edge grinding commands based on displacement distance and wafer offset state, and to control the grinding mechanism to drive the ultrasonic vibrating grinding wheel to rotate through the grinding mechanism spindle and perform rigid body transformation based on the y-axis or z-axis based on the target edge grinding commands, and to control the wafer stage to drive the wafer to rotate based on the wafer stage spindle and perform rigid body transformation based on the x-axis based on the target edge grinding commands, so as to achieve edge grinding.

[0006] In one embodiment, the visual inspection module includes a visual imaging unit and an illumination unit; both the visual imaging unit and the illumination unit are disposed on the grinding mechanism so that the detection area of ​​the visual imaging unit and the illumination area of ​​the illumination unit are aligned; both the visual imaging unit and the illumination unit are communicatively connected to the control module; the visual imaging unit is used to acquire continuous surface images of the detection area and transmit them to the control module; the control module is used to control the illumination unit to adjust the brightness when it is detected that the imaging quality parameters of the continuous surface images do not meet the preset values.

[0007] In one embodiment, the control module includes: a micro-texture extraction unit, a tracking and calculation unit, and a compensation control unit; the input of the micro-texture extraction unit is connected to the output of the visual detection module, and the output of the micro-texture extraction unit is connected to the input of the tracking and calculation unit; the output of the tracking and calculation unit is connected to the input of the compensation control unit; the input of the compensation control unit is also connected to the displacement detection module; the micro-texture extraction unit is used to extract micro-texture feature points based on the initial surface image and generate a current micro-texture feature set, which is then output to the tracking and calculation unit; the tracking and calculation unit is used to match the continuous surface image with the current micro-texture feature set and track the displacement of the micro-texture feature points, and obtain the center offset and rotation of the wafer based on the displacement of the micro-texture feature points; the compensation control unit is used to predict the edge grinding trajectory based on the center offset and rotation of the wafer through an edge grinding network model and generate a compensation command, and then generate a target edge grinding command based on the compensation command and the displacement distance.

[0008] In one embodiment, the control module further includes: a feedback unit; the input of the feedback unit is connected to the visual detection module, and the output of the feedback unit is connected to the compensation control unit; the feedback unit is used to generate parameter optimization instructions based on the edge grinding quality of the continuous surface image and output them to the compensation control unit; the compensation control unit is used to adjust the edge grinding network model based on the parameter optimization instructions.

[0009] In one embodiment, the component further includes: a cooling module; the cooling module is communicatively connected to the control module; the cooling module is disposed on the grinding mechanism so that the cooling area of ​​the cooling module is aligned with the contact area between the grinding wheel and the wafer; the control module is further configured to control the cooling module to cool the cooling area when the grinding wheel contacts the wafer for edge grinding.

[0010] In one embodiment, the component further includes: an air jet module; the air jet module is communicatively connected to a control module; the air jet module is disposed on the grinding mechanism to align the air jet area of ​​the air jet module with the contact area between the grinding wheel and the wafer; the control module is used to control the air jet module to spray air onto the air jet area to clean the contact area.

[0011] In one embodiment, the vision inspection module includes a CCD camera; the displacement detection module includes a displacement sensor; the components also include a cooling module and an air jet module; the cooling module includes a water spray bracket and a water spray pipe; the air jet module includes an air jet bracket and an air jet pipe; the water spray bracket and the air jet bracket are both disposed on the grinding mechanism and are used to support the water spray pipe and the air jet pipe respectively; the control module is also used to adjust the mechanical structure of the water spray bracket and the air jet bracket so that the water spray pipe and the air jet pipe are aligned with the contact area between the grinding wheel and the wafer.

[0012] In one embodiment, a plurality of fan-shaped piezoelectric ceramic groups are arranged at preset intervals along the circumferential direction on the inner side of the grinding wheel; the piezoelectric ceramic groups are excited by an alternating current signal and continuously expand and contract due to the inverse piezoelectric effect, thereby driving the grinding wheel to perform ultrasonic vibration.

[0013] In one embodiment, the main spindle of the wafer stage is an air-bearing spindle; the rigid body transformation of the wafer stage along the x-axis is achieved by a ball screw and a linear guide; a lifting structure is provided inside the wafer stage; the lifting structure is used to raise the wafer on the wafer stage after the edge grinding is completed.

[0014] Secondly, this application provides an ultrasonic displacement vision-based wafer inspection and edge grinding method, which applies the ultrasonic displacement vision-based wafer inspection and edge grinding system of the first aspect, including:

[0015] The wafer to be ground is placed on a wafer stage and fixed by vacuum adsorption via the stage's spindle. A displacement detection module detects the displacement distance between the wafer and the grinding mechanism and feeds this distance back to the control module. A vision detection module acquires continuous surface images of the wafer during rotation. The control module extracts micro-texture features from the received continuous surface images and determines the wafer offset state based on these features. The control module generates a target edge-grinding command based on the received displacement distance and wafer offset state. Based on this command, it controls the grinding mechanism to rotate the grinding wheel via the grinding mechanism's spindle and perform rigid body transformations along the y-axis or z-axis. It also controls the wafer stage to rotate the wafer along the stage's spindle and perform rigid body transformations along the x-axis, thus achieving edge grinding.

[0016] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0017] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art:

[0018] This application employs an ultrasonic displacement-vision collaborative wafer inspection and edge grinding system, consisting of a grinding mechanism uniformly coordinated by a control module, components integrating a displacement detection module and a vision inspection module, and a wafer stage. First, the displacement detection module detects the displacement distance between the wafer and the grinding mechanism in real time, providing accurate physical gap data to address the insufficient real-time sensing capability of the grinding wheel's working state—a fundamental prerequisite for precise compensation. Second, the vision inspection module acquires continuous surface images of the rotating wafer, enabling the control module to extract microscopic texture features reflecting the actual positional deviation of the wafer. This compensates for the inability to determine wafer offset solely based on displacement distance; the two work together to provide comprehensive real-time sensing of the grinding state. Based on this, the control module fuses these two complementary parameters—displacement distance and wafer offset state—to generate a target edge grinding command that simultaneously controls the grinding mechanism in the y or z axis and the wafer stage in the x axis, directly converting the sensed information into compensating actions of the actuator. This solves the problem of uneven grinding caused by insufficient dynamic compensation capability.

[0019] Ultimately, compared with existing technologies, it achieves full-process collaboration from multi-dimensional perception to closed-loop control, ensuring the stability and precision of the grinding process and realizing the technical effect of improving the quality of wafer edge morphology. Attached Figure Description

[0020] Figure 1 This is one of the structural block diagrams of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application;

[0021] Figure 2This is one of the schematic diagrams of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application;

[0022] Figure 3 This is the second structural block diagram of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application;

[0023] Figure 4 This is a second schematic diagram of a model of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application;

[0024] Figure 5 This is a schematic diagram of the workflow of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in an embodiment of this application;

[0025] Figure 6 This is a cross-sectional view of the grinding wheel of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in an embodiment of this application;

[0026] Figure 7 This is a schematic diagram of the piezoelectric ceramic of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in this application embodiment being excited by an AC signal;

[0027] Figure 8 This is a cross-sectional view of a wafer stage for an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in an embodiment of this application.

[0028] In all the accompanying drawings, the same reference numerals are used to denote the same elements or structures, wherein:

[0029] 10 is the control module; 11 is the micro-texture extraction unit; 12 is the tracking and calculation unit; 13 is the compensation control unit; 14 is the feedback unit; 20 is the grinding mechanism; 21 is the grinding wheel; 211 is the piezoelectric ceramic assembly; 30 is the component; 31 is the displacement detection module; 32 is the vision inspection module; 33 is the cooling module; 34 is the air jet module; 40 is the wafer stage; 50 is the wafer. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] In this article, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The symbol " / " in this article indicates that the related objects are in an "or" relationship; for example, A / B means A or B.

[0032] The terms "first" and "second," etc., used in the specification and claims herein are used to distinguish different objects, not to describe a specific order of objects. For example, "first response message" and "second response message," etc., are used to distinguish different response messages, not to describe a specific order of response messages.

[0033] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as superior or more advantageous than other embodiments or designs. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0034] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.

[0035] Due to limited real-time sensing capabilities of the grinding wheel's working state during the grinding process and the lack of an effective dynamic compensation mechanism, the system struggles to adjust process parameters promptly based on the actual grinding conditions. This results in uneven grinding force distribution and inconsistent material removal rates. Such uneven grinding processes easily lead to defects on the wafer surface, such as thickness variations and roughness fluctuations. Especially in structurally fragile edge regions, stress concentration and insufficient support make them more prone to morphological damage such as edge collapse, chipping, and microcracks. This severely affects the wafer's geometric accuracy and edge integrity, ultimately limiting the quality and yield of chip manufacturing.

[0036] Based on this, this application provides an ultrasonic displacement vision-assisted wafer inspection and edge grinding system. Please refer to... Figure 1 , Figure 1 This is one of the structural block diagrams of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application.

[0037] In this embodiment, the system includes: a control module 10, a grinding mechanism 20 including a grinding wheel 21, a component 30 disposed on the grinding mechanism 20, and a wafer stage 40. The grinding mechanism 20, the component 30, and the wafer stage 40 are all communicatively connected to the control module 10.

[0038] It is understandable that the communication connection between each part and the control module 10 means that the control module 10 can obtain information from other parts of the system. This information can be status information, such as whether the grinding mechanism 20 is in normal operating condition, whether the displacement detection module 31 and vision inspection module 32 in the component 30 are working properly, and whether the wafer stage 40 is firmly adsorbed onto the wafer 50. It also includes data information related to wafer inspection and edge grinding obtained by other parts during the operation, such as wafer displacement data detected by the displacement detection module 31 and wafer surface image information captured by the vision inspection module 32.

[0039] Simultaneously, the communication connection also enables the control module 10 to issue commands. The control module 10 can analyze and judge the acquired information, and then issue corresponding commands to other parts of the system. Through these commands, the control module 10 can achieve precise control over other components in the system. For example, when the vision inspection module 32 detects defects on the wafer surface, the control module 10 can issue commands to adjust the operating parameters of the grinding mechanism 20, allowing it to grind the wafer in a more suitable manner; or when the displacement detection module 31 detects a shift in the wafer position, the control module 10 can instruct the wafer stage 40 to make fine adjustments to ensure the wafer is in the correct processing position.

[0040] It should be noted that the grinding mechanism 20 should at least include a grinding mechanism spindle and a grinding wheel 21 disposed at the end of the grinding mechanism spindle, so as to realize that the grinding wheel 21 is driven to rotate by the rotation of the grinding mechanism spindle. The grinding mechanism spindle and the wafer stage spindle mentioned below are both power axes, and their positions are usually selected in various ways after design. For example, since the wafer stage 40 is generally cylindrical, the axis of symmetry of the cylinder is usually chosen for the wafer stage spindle. Its orientation is not limited in this application. It is only based on the fixed grinding mechanism spindle and wafer stage spindle that the orientation of the system is discussed.

[0041] Specifically, please refer to Figure 2 , Figure 2 This is one of the schematic diagrams of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application. Figure 2 In the middle, the grinding mechanism 20 is arranged in parallel above the wafer stage 40; the wafer stage 40 is used to adsorb the wafer 50; the component 30 includes a displacement detection module 31 and a vision inspection module 32.

[0042] It should be noted that, in terms of spatial layout, the grinding mechanism 20 is arranged parallel above the wafer stage 40, meaning that the grinding mechanism spindle of the grinding mechanism 20 is perpendicular to the wafer stage spindle, i.e., the c-axis. This reasonable positional relationship provides a basis for the grinding wheel 21 on the grinding mechanism 20 to contact the wafer 50 for edge grinding and other operations, ensuring that the working area of ​​the grinding mechanism 20 is aligned with the wafer stage. The main function of the wafer stage 40 is to hold the wafer 50, ensuring that the wafer 50 can be stably fixed during inspection and edge grinding, avoiding any impact on operational accuracy due to movement or shaking.

[0043] It should be noted that the inspection and grinding system must have space to realize rigid body transformation along the x-axis, y-axis, and z-axis. Among them, the x-axis, y-axis, and z-axis are perpendicular to each other in space. The grinding mechanism 20 can perform rigid body transformation based on any two axes, such as the y-axis or the z-axis, and the wafer stage 40 can perform rigid body transformation based on the remaining axis, such as the x-axis.

[0044] Specifically, with Figure 2 For example, the first plane formed by the x-axis and y-axis is parallel to the plane of the wafer stage 40, the y-axis is parallel to the grinding mechanism 20, the x-axis is perpendicular to the main axis of the wafer stage, and the z-axis is parallel to the main axis of the wafer stage.

[0045] It should be noted that component 30 is a functional module integrated into the edge grinding system. In this embodiment, component 30 includes two important parts: a displacement detection module 31 and a vision inspection module 32. The displacement detection module 31 may be used to detect the displacement information of the wafer or related components, while the vision inspection module 32 may be used to perform visual inspection on the wafer.

[0046] Specifically, the displacement detection module 31 is used to detect the displacement distance between the wafer 50 and the grinding mechanism 20 and feed it back to the control module 10. During the edge grinding operation, the relative position of the wafer 50 and the grinding mechanism 20 will constantly change. Even extremely small displacement deviations can have a significant impact on the final edge grinding effect, leading to problems such as inaccurate wafer edge dimensions and substandard surface quality. The displacement detection module 31 can detect the displacement distance between the two in real time with high precision and quickly and accurately feed this key data back to the control module 10.

[0047] Specifically, the vision inspection module 32 acquires continuous surface images of the wafer 50 during its rotation and feeds them back to the control module 10. During the high-speed rotation of the wafer 50 for edge grinding, its surface condition continuously changes due to the grinding action, potentially leading to various defects such as scratches, cracks, and unevenness. The vision inspection module 32 utilizes image acquisition technology to continuously capture surface images of the wafer 50 during its rotation, fully recording the detailed features of the wafer surface at different times. Subsequently, the vision inspection module 32 promptly feeds back these continuous surface images to the control module 10.

[0048] Understandably, the displacement data acquired by the displacement detection module 31, which shows the displacement between the wafer 50 and the grinding mechanism 20, can reflect the contact and pressure between the grinding wheel and the wafer in real time, allowing the system to promptly understand the physical state of the grinding wheel during operation. The image of the rotating surface of the wafer obtained by the vision inspection module can intuitively present the grinding effect of the grinding wheel and the changes on the wafer surface, assisting in judging the grinding performance of the grinding wheel. The grinding wheel parameters can be dynamically adjusted based on the displacement data, and the offset caused by defects can be accurately identified based on the image data. The combination of these two provides a key basis for the system to comprehensively and in real time perceive the state of the grinding wheel 21 and the wafer 50, and quickly make dynamic compensation decisions.

[0049] It should be noted that the control module 10 is used to extract micro-texture features based on the received continuous surface image, and to obtain the wafer 50 offset state based on the micro-texture features.

[0050] The continuous surface images include images of wafer 50 placed on wafer stage 40 when not in operation, corresponding to time 0 and images of wafer 50 continuously rotating after the grinding process begins. The micro-texture features are the micro-textures on the wafer's surface itself. Even after polishing, these inherent, randomly distributed texture features remain at the micro-level, serving as natural, unmarked feature points. Alternatively, they may be texture features generated during operation based on the grinding process.

[0051] Understandably, micro-texture features from continuous surface images include feature point extraction as well as the generation and storage of descriptors.

[0052] Specifically, in the feature point extraction stage, algorithms such as Scale-Invariant Feature Transform (SIFT), Oriented Fast and Rotated BRIEF (ORB), or improved versions of these algorithms are employed. Each of these algorithms has its advantages, enabling precise localization of unique and stable feature points in continuous surface images. For example, the SIFT algorithm performs multi-scale spatial analysis of the image, detecting feature points with good invariance to scale changes, rotation changes, and illumination changes at different scales. These feature points can stand out in complex image scenes. The ORB algorithm, combining the FAST keypoint detector and the BRIEF descriptor, features high computational speed and efficiency, quickly finding a large number of feature points in the image, making it suitable for scenarios with high real-time requirements. By processing continuous surface images using these algorithms, numerous feature points representing the microstructure information of the wafer surface can be extracted.

[0053] As we can understand, a descriptor is a quantified representation of the information surrounding a feature point. It can transform the texture, shape, and other detailed information around a feature point into a set of numerical vectors. This set of numerical vectors can uniquely and accurately describe the characteristics of the feature point. Different algorithms generate descriptors with different properties. For example, the descriptor generated by the SIFT algorithm has scale invariance and rotation invariance, and can accurately describe feature points even when the image undergoes scale changes and rotations; while the descriptor generated by the ORB algorithm focuses more on computational efficiency, and can be generated quickly while maintaining a certain level of descriptive power.

[0054] It should be noted that the wafer 50 offset state is the result of further analysis by the control module 10 based on micro-texture features. Since the wafer 50 may shift position during processing due to various factors, this shift will be reflected in changes to its surface micro-texture features. By meticulously comparing and analyzing the extracted micro-texture features, the control module 10 can perceive the changes in texture, thereby determining whether the wafer 50 has shifted, and the specific direction and degree of the shift, i.e., the offset state.

[0055] It should be noted that the control module 10 is also used to generate a target grinding command based on the received displacement distance and wafer 50 offset state, and to control the grinding mechanism 20 to drive the grinding wheel 21 to rotate and rigidly transform along the y-axis or z-axis based on the target grinding command, and to control the wafer stage 40 to drive the wafer 50 to rotate along the main axis of the wafer stage 40 and rigidly transform along the x-axis based on the target grinding command, so as to achieve grinding.

[0056] Understandably, the displacement distance data will be converted into the contact pressure state between the grinding wheel and the wafer. This conversion is based on a preset displacement-pressure mapping relationship, and the pressure is judged to be within a reasonable range by analyzing the displacement change trend. If an abnormal pressure is detected, the control module 10 will trigger a dynamic feed rate adjustment strategy. For example, when the pressure is too low, the feed rate along the z-axis will be increased to enhance the grinding force; when the pressure is too high, the feed rate will be reduced to avoid damage to the wafer surface.

[0057] Understandably, the quantitative analysis of wafer offset status is performed simultaneously. Microscopic texture features are extracted from the surface image captured by the vision inspection module 32, and the offset direction and degree are identified using a feature point matching algorithm. The control module vectorizes the pressure adjustment requirements and offset compensation requirements to generate a composite command that includes the grinding wheel rotation speed, ultrasonic vibration parameters, multi-axis rigid body transformation, and the rotation angle of the wafer stage. This fusion mechanism achieves dynamic closed-loop control during the edge grinding process by real-time sensing of the grinding wheel status and wafer position deviation. This ensures that the grinding wheel and wafer are always in optimal contact, and also improves material removal efficiency and reduces grinding wheel wear through ultrasonic vibration, effectively solving the processing error problem caused by the lag in status perception in traditional processes.

[0058] It should be noted that the target grinding command includes control of the grinding mechanism 20 and the wafer stage 40. Specifically, for the grinding mechanism 20, firstly, the grinding wheel 21 is driven to rotate at high speed by the grinding mechanism spindle to maintain the basic grinding kinetic energy; secondly, precise adjustment of spatial position and orientation is achieved through rigid body transformation along the y-axis or z-axis. Displacement along the y-axis controls the grinding wheel feed depth, directly adjusting the grinding force, while rotation θ around the y-axis... y The angle can be adjusted to change the cutting angle between the grinding wheel and the wafer contact surface, adapting to the grinding requirements of different edge profiles; displacement along the z-axis can change the perpendicular distance between the grinding wheel and the wafer, and rotation around the z-axis by θ z The angle corrects the tilt of the grinding wheel in the vertical plane, ensuring the vertical accuracy of the grinding trajectory. For the wafer stage 40, the control strategy focuses on x-axis motion and rotation compensation. Displacement along the x-axis allows for fine-tuning of the wafer's horizontal position, ensuring precise alignment of the grinding area with the grinding wheel; rotation around the x-axis by θ... x The angle can compensate for the deflection of the wafer caused by processing vibration or initial clamping error. By adjusting the wafer's own posture in real time, it maintains the optimal contact angle between the wafer and the grinding wheel. This multi-axis linkage control mechanism enables the system to adapt to complex grinding trajectories through the multi-directional movement of the grinding wheel, and to eliminate positional deviations through dynamic compensation of the wafer, ultimately achieving high-precision, low-damage edge grinding of the wafer.

[0059] In this embodiment, an ultrasonic displacement-vision collaborative wafer inspection and edge grinding system is constructed, consisting of a grinding mechanism uniformly coordinated by a control module, a component integrating a displacement detection module and a vision inspection module, and a wafer stage. The displacement detection module detects the displacement distance between the wafer and the grinding mechanism in real time, providing accurate physical gap data to address the insufficient real-time perception capability of the grinding wheel's working state—a fundamental prerequisite for precise compensation. Furthermore, the vision inspection module acquires continuous surface images of the rotating wafer, allowing the control module to extract microscopic texture features reflecting the actual positional deviation of the wafer. This compensates for the inability to determine wafer offset based solely on displacement distance; the two work together to provide comprehensive real-time perception of the grinding state. Based on this, the control module fuses the displacement distance and wafer offset state—two complementary parameters—to generate a target edge grinding command that simultaneously controls the grinding mechanism in the y or z axis and the wafer stage in the x axis, directly converting the perceived information into compensating actions of the actuator. This solves the problem of uneven grinding caused by insufficient dynamic compensation capability.

[0060] Furthermore, based on the above embodiments, a further refinement has been made, providing yet another ultrasonic displacement vision-assisted wafer inspection and edge grinding system. Please refer to... Figure 3 and Figure 4 , Figure 3 This is the second structural block diagram of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application; Figure 4 This is the second schematic diagram of an ultrasonic displacement vision-assisted wafer inspection and edge grinding system provided in the embodiments of this application.

[0061] In this embodiment, component 30 further includes: a cooling module 33; the cooling module 33 is communicatively connected to the control module 10; the cooling module 33 is disposed on the grinding mechanism 20 so that the cooling area of ​​the cooling module 33 is aligned with the contact area between the grinding wheel 21 and the wafer 50; the control module 10 is also used to control the cooling module 33 to cool the cooling area when the grinding wheel 21 contacts the wafer 50 for edge grinding.

[0062] It is understandable that when the grinding wheel 21 interacts with the wafer 50, a large amount of grinding heat will be generated at the contact point, which will affect the grinding quality of the wafer 50 and cause the grinding wheel 21 to wear. In order to ensure the normal operation of grinding, a cooling module 33 is required to cool the grinding wheel 21.

[0063] It is understandable that the cooling area refers to the working area of ​​the cooling module 33, and aligning it with the contact area between the grinding wheel 21 and the wafer 50 can better achieve direct cooling.

[0064] In this embodiment, component 30 further includes: a jet module 34; the jet module 34 is communicatively connected to the control module 10; the jet module 34 is disposed on the grinding mechanism 20 so that the jet area of ​​the jet module 34 is aligned with the contact area between the grinding wheel 21 and the wafer 50; the control module 10 is used to control the jet module 34 to jet the jet area to clean the contact area.

[0065] Understandably, considering the cooling substances that the aforementioned module may generate and the debris produced during the edge grinding process, it is necessary to eliminate these interferences to avoid inaccurate images or displacement data acquired during edge grinding obstruction or compensation. The jetting area refers to the working area of ​​the jetting module 34, and aligning it with the contact area between the grinding wheel 21 and the wafer 50 can better eliminate interference.

[0066] Specifically, the vision inspection module 32 may include a CCD camera, which needs to be a high-speed camera with a high frame rate and high resolution. The camera captures continuous surface images during the wafer rotation process by high-speed shooting. The displacement detection module 31 may include a displacement sensor; a high-precision laser displacement sensor is installed close to the wafer 50, and measures the distance by emitting a laser beam to illuminate the surface of the wafer 50 and receiving the reflected light signal. The cooling module 33 includes a water spray bracket and a water spray pipe; the air spray module 34 includes an air spray bracket and an air spray pipe; the water spray bracket and the air spray bracket are both set on the grinding mechanism 20 and are used to support the water spray pipe and the air spray pipe, respectively; the control module 10 is also used to adjust the mechanical structure of the water spray bracket and the air spray bracket so that the water spray pipe and the air spray pipe are aligned with the contact area between the grinding wheel 21 and the wafer 50.

[0067] In addition, in this embodiment, the visual inspection module 32 includes a visual imaging unit and an illumination unit. Both the visual imaging unit and the illumination unit are disposed on the grinding mechanism 20 to align the detection area of ​​the visual imaging unit and the illumination area of ​​the illumination unit; both the visual imaging unit and the illumination unit are communicatively connected to the control module 10.

[0068] It is understandable that the detection area is the working area of ​​the vision imaging unit, and the illumination area is the working area of ​​the illumination unit. Alignment between the detection and illumination areas means that the positions of the vision imaging unit and the illumination unit are synchronized, ensuring that the illumination accurately covers the detection area and avoiding image quality problems caused by positional misalignment. Furthermore, the fact that both the vision imaging unit and the illumination unit are communicatively connected to the control module 10 means that the control module can receive image data from the vision imaging unit and send commands to the illumination unit to adjust its brightness. The function of the vision imaging unit is to acquire continuous surface images of the detection area and transmit these images to the control module 10.

[0069] It should be noted that the visual imaging unit is used to acquire continuous surface images of the detection area and transmit them to the control module 10; the control module 10 is used to control the illumination module to adjust the brightness when it is detected that the imaging quality parameters of the continuous surface image do not meet the preset value.

[0070] Specifically, the visual imaging unit can be equipped with a high quantum efficiency sensor, including ultraviolet band enhancement capabilities, to enable the acquired images to include the microscopic texture of the wafer surface. The illumination unit can employ a multi-angle, multi-spectral illumination scheme to highlight the microscopic contrast of the wafer surface and suppress interference reflections.

[0071] Specifically, the system can perform real-time analysis of received images through a built-in image quality assessment algorithm. When imaging quality parameters such as insufficient brightness, low contrast, or decreased sharpness are detected and fail to meet preset standards, a brightness adjustment command is immediately sent to the illumination unit. This closed-loop control mechanism enables the system to automatically optimize illumination conditions based on the actual imaging effect. For example, it can reduce brightness to avoid overexposure when the wafer surface is highly reflective, or enhance local illumination when edge details are blurred, thereby ensuring that the vision imaging unit can always acquire high-quality surface images.

[0072] In this embodiment, the control module 10 includes a microtexture extraction unit 11, a tracking and calculation unit 12, and a compensation control unit 13. The input of the microtexture extraction unit 11 is connected to the output of the visual detection module 32, and the output of the microtexture extraction unit 11 is connected to the input of the tracking and calculation unit 12. The output of the tracking and calculation unit 12 is connected to the input of the compensation control unit 13, and the input of the compensation control unit 13 is also connected to the displacement detection module 31.

[0073] It should be noted that the micro-texture extraction unit 11 is used to extract micro-texture feature points based on the initial surface image. It is responsible for extracting feature points (such as SIFT, ORB, or their improved algorithms), generating descriptors, and storing them from the pre-sampled image, which will not be elaborated here. It also generates the current micro-texture feature set and outputs it to the tracking and solving unit 12.

[0074] It is understandable that the current micro-texture feature set refers to a dynamic data set composed of micro-texture feature points and their associated descriptors extracted from continuously acquired surface images during wafer fabrication. The core components of this set include: the spatial coordinates, scale information, principal direction angle of the feature points, and descriptor vectors generated from the surrounding regions of the feature points. As the wafer rotates and the fabrication process progresses, the visual imaging unit continuously captures new frames of images, and the feature set is updated accordingly to reflect the latest state of the wafer surface microstructure.

[0075] It should be noted that the tracking and solving unit 12 is used to match the continuous surface image with the current micro-texture feature set and track the displacement of the micro-texture feature points, and obtain the center offset and rotation of the wafer 50 based on the displacement of the micro-texture feature points. That is, it is responsible for using the feature matching algorithm to match feature points in real time, calculating the overall displacement and deformation of the feature point field, and thus solving for the center offset and rotation.

[0076] Understandably, feature matching algorithms, such as nearest neighbor search based on KD trees or optical flow optimization, are used to accurately align the current image with the current micro-texture feature set stored in the feature database. After completing feature point matching, the tracking and solving unit 12 analyzes the overall motion trend of the feature point group by calculating the spatial displacement vector field of the matched feature point pairs. For example, when the wafer is translated, the feature point field exhibits a uniform displacement distribution; when rotation occurs, the feature point field shows an arc-shaped displacement trajectory with the wafer center as the origin. By fitting these displacement data using the least squares method, the unit can accurately calculate the center offset and rotation angle of the wafer 50 in the processing plane. These parameters are updated at a millisecond-level frequency and fed back to the subsequent control loop, providing real-time basis for dynamic compensation.

[0077] It should be noted that the compensation control unit 13 is used to predict the grinding trajectory and generate compensation commands based on the center offset and rotation of the wafer 50 through the grinding network model, and then generate target grinding commands based on the compensation commands and displacement distance.

[0078] Understandably, the pre-trained grinding network model learns the displacement-trajectory mapping relationship in massive processing data through deep neural networks (such as a CNN-LSTM hybrid architecture), enabling it to predict the ideal contact trajectory between the grinding wheel and the wafer under different offset states. Subsequently, it generates compensation commands based on the predicted trajectory, including parameters such as the grinding wheel rotation speed correction value and multi-axis rigid body transformation. Then, it performs fusion calculations by combining the real-time displacement distance, and generates the target grinding command through a weighted feedback algorithm (such as a variant of PID control).

[0079] Furthermore, the control module 10 may also include: a feedback unit 14; the input of the feedback unit 14 is connected to the vision detection module 32, and the output of the feedback unit 14 is connected to the compensation control unit 13; the feedback unit 14 is used to generate parameter optimization instructions based on the grinding quality of the continuous surface image and output them to the compensation control unit 13; the compensation control unit 13 is used to adjust the grinding network model based on the parameter optimization instructions.

[0080] Specifically, the input end of the feedback unit 14 is directly connected to the vision imaging unit of the vision inspection module 32, and receives a continuous stream of surface images during the processing in real time. These images not only contain edge contour information of the wafer 50, but also capture surface quality details of the grinding area through high-resolution imaging, such as micro-features like edge burrs, waviness, and surface roughness.

[0081] It should be noted that the built-in quality assessment algorithm of the feedback unit 14 performs multi-dimensional analysis on each frame of image: first, it calculates surface roughness parameters using a texture analysis algorithm; second, it quantifies edge straightness and curvature using an edge detection algorithm; and finally, it compares the actual processing effect with the ideal model using template matching technology. When it detects that the edge grinding quality parameters do not meet the preset standards, such as edge roughness exceeding the threshold or edge straightness deviation exceeding the allowable value, the unit will generate parameter optimization instructions based on the execution effect of the previous target edge grinding instruction using a machine learning algorithm. This instruction includes suggestions for correcting the input parameters of the edge grinding network model, such as suggesting reducing the grinding wheel rotation speed to reduce the risk of surface burn, or adjusting the wafer stage rotation compensation to improve edge straightness.

[0082] Understandably, after receiving the parameter optimization command, the compensation control unit 13 inputs it into the parameter adjustment interface of the edge grinding network model. This model employs a configurable architecture, where hyperparameters such as hidden layer weights and activation function types can be dynamically adjusted via feedback commands. For example, when the feedback unit 14 indicates that periodic ripples appear on a certain edge, the compensation control unit 13 can correspondingly increase the model's suppression weight for vibration frequencies; when local over-grinding is detected, the predicted feed rate for the corresponding area is reduced. This model adjustment mechanism based on real-time quality feedback enables the system to continuously optimize its control strategy, ultimately achieving an upgrade from passive correction to proactive prevention in the processing mode, significantly improving the yield and consistency of wafer edge grinding.

[0083] Correspondingly, please refer to Figure 5 , Figure 5 This is a schematic diagram illustrating the workflow of an ultrasonic displacement vision-based collaborative wafer inspection and edge grinding system provided in an embodiment of this application. The specific workflow is as follows:

[0084] S1. Presampling and feature extraction: Before edge grinding, images of the working area are collected, and surface micro-texture features are extracted and stored.

[0085] S2. Collect continuous surface images of the working area and the displacement distance of the grinding wheel.

[0086] S3. Feature Matching and Tracking: Match the continuous table image with the pre-stored surface micro-texture features and track the displacement of the micro-texture feature points.

[0087] S4. Dynamic offset calculation: Calculate the wafer center offset and rotation, and predict the position change in the short term based on the model.

[0088] S5. Compensation Control: Generate compensation commands based on offset and rotation, and generate target edge grinding commands in conjunction with grinding wheel displacement distance.

[0089] S6. Execution Feedback: The system's six degrees of freedom are adjusted according to the target edge grinding command, and the edge grinding effect is detected. If the edge grinding effect is qualified, the process ends; otherwise, step S7 is executed.

[0090] S7. Adjust parameters: Based on the continuous surface image, the edge grinding quality is generated by optimizing the parameters based on the previous target edge grinding command to optimize the model parameters.

[0091] The specific execution details of this process can be found in the unit descriptions above, and will not be repeated here.

[0092] In one embodiment, please refer to Figure 6 Multiple fan-shaped piezoelectric ceramic groups 211 are arranged at preset intervals along the annular direction on the inner side of the grinding wheel 21. When the piezoelectric ceramic groups 211 are excited by the AC signal, they continuously expand and contract due to the inverse piezoelectric effect, which drives the grinding wheel 21 to perform ultrasonic vibration.

[0093] It is understandable that the grinding wheel 21 has fan-shaped piezoelectric ceramic assemblies 211 uniformly embedded in its interior along a circumferential direction. The piezoelectric ceramic assemblies 211 are made of high-performance PZT-8 material, which has a high electromechanical coupling coefficient and mechanical quality factor; embedding them inside the grinding wheel facilitates integration. In this embodiment, an AC signal is applied to the piezoelectric ceramic assemblies 211 via a high-frequency AC wireless power supply. For example... Figure 7 As shown, piezoelectric ceramics continuously expand and contract when excited, driving the grinding wheel to vibrate ultrasonically. The amplitude of the vibration can be controlled by adjusting the voltage amplitude and frequency of the AC signal. Ultrasonic vibration can effectively improve the efficiency of wafer edge grinding, while improving the edge morphology quality of the wafer and effectively reducing edge roughness.

[0094] In one embodiment, please refer to Figure 8 The main spindle of the wafer stage 40 is an air-bearing main spindle; the rigid body transformation of the wafer stage 40 along the x-axis is achieved by ball screws and linear guides; the wafer stage 40 is equipped with a lifting structure; the lifting structure is used to raise the wafer 50 on the wafer stage 40 after the edge grinding is completed.

[0095] Understandably, the spindle (c-axis) of the wafer stage 40 is used to adsorb and rotate the wafer 50. The spindle employs a high-precision air-bearing design to ensure the stability and accuracy of the wafer 50 during rotation. The wafer 50 is fixed to the wafer stage 40 by vacuum adsorption through air pores, with the vacuum level controlled within a suitable range to ensure that the wafer 50 does not shift during rotation. The wafer stage 40 also has a one-dimensional linear motion axis, which will not be elaborated here. Specifically, displacement and rotation in the x-axis direction can be achieved through ball screws and linear guide pairs to meet the needs of different inspection and grinding positions. After the wafer 50 is ground, a special structure between the bottom surface and the connector forms a cylinder-like mechanism to lift the lifting structure to a certain height, facilitating the removal of the ground wafer 50.

[0096] Understandably, traditional methods rely on wafer edges or specific markings for positioning. This solution requires no pre-fabrication or physical markings, utilizing the inherent micro-texture of the wafer surface. It is applicable to various wafers, including those with indistinct or damaged markings, reducing failures caused by marking recognition failures. Simultaneously, it achieves real-time, continuous, and dynamic high-precision six-degree-of-freedom tracking during the grinding process. It not only compensates for installation deviations but also counteracts errors caused by minute deformations and vibrations due to grinding forces, thermal effects, and other factors. Furthermore, by introducing machine learning algorithms to predict short-term trends in wafer position and orientation, the system can implement feedforward control, issuing compensation commands in advance to overcome system mechanical response delays.

[0097] In addition, the data from the process can be used to deeply analyze the impact of grinding process parameters on wafer behavior and quality, thereby optimizing the process formulation and achieving iterative optimization of the process.

[0098] Ultimately, compared with existing technologies, it achieves full-process collaboration from multi-dimensional perception to closed-loop control, ensuring the stability and precision of the grinding process and realizing the technical effect of improving the quality of wafer edge morphology.

[0099] Secondly, based on the ultrasonic displacement vision-coordinated wafer inspection and edge grinding system of the first aspect above, this application provides an embodiment of the ultrasonic displacement vision-coordinated wafer inspection and edge grinding method.

[0100] Specifically, the process includes: placing the wafer to be ground on a wafer stage and fixing it using vacuum adsorption via the stage's spindle; detecting the displacement distance between the wafer and the grinding mechanism using a displacement detection module and feeding this distance back to the control module; and acquiring continuous surface images of the wafer during rotation using a vision detection module. The control module extracts micro-texture features from the received continuous surface images and obtains the wafer offset state based on these features. The control module generates a target edge-grinding command based on the received displacement distance and wafer offset state, and controls the grinding mechanism to rotate the grinding wheel via the grinding mechanism's spindle and perform rigid body transformations along the y-axis or z-axis, and to control the wafer stage to rotate the wafer along the stage's spindle and perform rigid body transformations along the x-axis, thereby achieving edge grinding.

[0101] Compared with the prior art, the beneficial effects of the method provided in this application are the same as those of the system provided in the above embodiments, and will not be repeated here.

[0102] It should be understood that expressions such as “comprising” and “may include” used in this application indicate the existence of the disclosed functions, operations, or constituent elements, and do not limit one or more additional functions, operations, and constituent elements. In this application, terms such as “comprising” and / or “having” are to be interpreted as indicating a particular characteristic, number, operation, constituent element, component, or combination thereof, but not to exclude the existence or possibility of adding one or more other characteristics, numbers, operations, constituent elements, components, or combinations thereof.

[0103] Furthermore, in this application, the expression "and / or" includes any and all combinations of the associated listed words. For example, the expression "A and / or B" may include A, may include B, or may include both A and B.

[0104] In the description of the embodiments of this application, it should be noted that, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a detachable connection or a non-detachable connection; it can be a direct connection or an indirect connection through an intermediate medium. "Fixed connection" refers to a connection where the relative positional relationship remains unchanged after connection. "Rotary connection" refers to a connection where the components can rotate relative to each other after connection. "Sliding connection" refers to a connection where the components can slide relative to each other after connection. The directional terms mentioned in the embodiments of this application, such as "top," "bottom," "inner," "outer," "left," and "right," are only for reference to the directions in the accompanying drawings. Therefore, the directional terms used are for better and clearer explanation and understanding of the embodiments of this application, and are not intended to indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0105] Furthermore, the mathematical concepts mentioned in the embodiments of this application, such as symmetry, equality, parallelism, and perpendicularity, are limitations specific to the current technological level, rather than absolute and strict mathematical definitions. Slight deviations are permissible; approximations of symmetry, equality, parallelism, and perpendicularity are all acceptable. For example, "A and B are parallel" means that A and B are parallel or approximately parallel, and the angle between A and B can be between 0 and 10 degrees. "A and B are perpendicular" means that A and B are perpendicular or approximately perpendicular, and the angle between A and B can be between 80 and 100 degrees.

[0106] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An ultrasonic displacement vision-assisted wafer inspection and edge grinding system, characterized in that, include: Control module, grinding mechanism, components mounted on the grinding mechanism, and wafer stage; The grinding mechanism, the components, and the wafer stage are all communicatively connected to the control module. The grinding mechanism is arranged parallel above the wafer stage so that the working area of ​​the grinding mechanism is aligned with the wafer stage; The wafer stage is used to adsorb wafers; The components include: a displacement detection module and a vision detection module; The displacement detection module is used to detect the displacement distance from the wafer to the grinding mechanism and feed it back to the control module; the vision detection module is used to acquire continuous surface images during the wafer rotation process; The control module is used to extract micro-texture features based on the continuous surface image and to obtain the wafer offset state based on the micro-texture features; The control module is also used to generate a target grinding command based on the displacement distance and the wafer offset state, and to control the grinding mechanism to drive the ultrasonic vibrating grinding wheel to rotate through the grinding mechanism spindle and perform rigid body transformation based on the y-axis or z-axis based on the target grinding command, and to control the wafer stage to drive the wafer to rotate based on the wafer stage spindle and perform rigid body transformation based on the x-axis based on the target grinding command, so as to achieve grinding.

2. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The visual detection module includes: a visual imaging unit and an illumination unit; Both the visual imaging unit and the illumination unit are disposed on the grinding mechanism so that the detection area of ​​the visual imaging unit and the illumination area of ​​the illumination unit are aligned; both the visual imaging unit and the illumination unit are communicatively connected to the control module. The visual imaging unit is used to acquire continuous surface images of the detection area and transmit them to the control module; The control module is used to control the illumination unit to adjust the brightness when the imaging quality parameters of the continuous surface image do not meet the preset value.

3. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The control module includes: a micro-texture extraction unit, a tracking and calculation unit, and a compensation and control unit; The input terminal of the microtexture extraction unit is connected to the output terminal of the visual detection module, and the output terminal of the microtexture extraction unit is connected to the input terminal of the tracking and solving unit; the output terminal of the tracking and solving unit is connected to the input terminal of the compensation and control unit; the input terminal of the compensation and control unit is also connected to the displacement detection module. The micro-texture extraction unit is used to extract micro-texture feature points based on the initial surface image and generate the current micro-texture feature set, which is then output to the tracking and solving unit. The tracking and solving unit is used to match the continuous surface image with the current micro-texture feature set and track the displacement of the micro-texture feature points, and obtain the center offset and rotation of the wafer based on the displacement of the micro-texture feature points; The compensation control unit is used to predict the edge grinding trajectory and generate compensation instructions based on the center offset and rotation of the wafer through the edge grinding network model, and then generate target edge grinding instructions based on the compensation instructions and the displacement distance.

4. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 3, characterized in that, The control module further includes: a feedback unit; The input terminal of the feedback unit is connected to the visual detection module, and the output terminal of the feedback unit is connected to the compensation control unit. The feedback unit is used to generate parameter optimization instructions based on the previous target edge grinding instruction based on the edge grinding quality of the continuous surface image and output them to the compensation control unit. The compensation control unit is used to adjust the edge grinding network model based on the parameter optimization instructions.

5. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The component also includes: a cooling module; The cooling module is communicatively connected to the control module; the cooling module is disposed on the grinding mechanism so that the cooling area of ​​the cooling module is aligned with the contact area between the grinding wheel and the wafer; The control module is also used to control the cooling module to cool the cooling area when the grinding wheel contacts the wafer for edge grinding.

6. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The components also include: a jet module; The jet module is communicatively connected to the control module; the jet module is disposed on the grinding mechanism so that the jet area of ​​the jet module is aligned with the contact area between the grinding wheel and the wafer; The control module is used to control the jet module to spray air onto the jet area to clean the contact area.

7. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The visual detection module includes a CCD camera; the displacement detection module includes a displacement sensor; the components also include a cooling module and a jet module; the cooling module includes a water spray bracket and a water spray pipe; the jet module includes a jet bracket and a jet pipe; The water spray bracket and the air jet bracket are both disposed on the grinding mechanism and are used to support the water spray pipe and the air jet pipe, respectively. The control module is also used to adjust the mechanical structure of the water spray bracket and the air spray bracket so that the water spray pipe and the air spray pipe are aligned with the contact area between the grinding wheel and the wafer.

8. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, Multiple fan-shaped piezoelectric ceramic groups are arranged at preset intervals along the circumferential direction on the inner side of the grinding wheel; the piezoelectric ceramic groups are excited by AC signals and continuously expand and contract due to the inverse piezoelectric effect, driving the grinding wheel to perform ultrasonic vibration.

9. The ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in claim 1, characterized in that, The main spindle of the wafer stage is an air-bearing main spindle; the rigid body transformation of the wafer stage along the x-axis is achieved by ball screws and linear guides; the wafer stage is equipped with a lifting structure inside; the lifting structure is used to raise the wafer on the wafer stage after edge grinding.

10. A method for ultrasonic displacement and vision-based combined wafer inspection and edge grinding, characterized in that, The method utilizes the ultrasonic displacement vision-assisted wafer inspection and edge grinding system as described in any one of claims 1 to 9, comprising: The wafer to be repaired is placed on the wafer stage and fixed by vacuum adsorption through the spindle of the wafer stage; The displacement distance from the wafer to the grinding mechanism is detected by the displacement detection module, and continuous surface images of the wafer during the rotation process are obtained by the vision detection module. The micro-texture features are extracted from the continuous surface image by the control module, and the wafer offset state is obtained based on the micro-texture features; The control module generates a target grinding command based on the received displacement distance and wafer offset state, and controls the grinding mechanism to drive the grinding wheel to rotate through the grinding mechanism spindle and perform rigid body transformation based on the y-axis or z-axis based on the target grinding command. The wafer stage is also controlled to drive the wafer to rotate along the wafer stage spindle and perform rigid body transformation based on the x-axis based on the target grinding command, so as to achieve edge grinding.

Images