Wafer defect detection system and method based on industrial vision

Abstract

The invention discloses a wafer defect detection system and method based on industrial vision, a central control module is a system core control unit, an industrial vision detection module is connected with the central control module, the industrial vision detection module is used for executing a snakelike scanning path, a motion control module is in real-time data interaction with the central control module, and the motion control module is connected with the central control module. The motion control module is used for driving the wafer carrying platform to move, the man-machine interaction and communication module is connected with the central control module, and the man-machine interaction and communication module is used for man-machine interaction; wherein the central control module, the industrial visual detection module, the motion control module and the man-machine interaction and communication module form a master-slave control framework, and the central control module is used for receiving feedback data of each module and outputting a control instruction. In the aspect of defect identification, by arranging the multispectral light source unit, the feature extraction and classification precision of tiny defects is improved, and the identification stability in a complex defect scene is enhanced compared with a traditional machine vision algorithm.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor manufacturing technology, specifically relating to a wafer defect detection system and method based on industrial vision. Background Technology

[0002] In the current semiconductor manufacturing field, wafer defect detection technology is mainly divided into two categories: manual visual inspection and traditional automated inspection. Manual inspection relies on operators to examine the surface of each wafer one by one using an optical microscope. This not only suffers from high subjectivity and fatigue-induced missed detection rates, but also takes 15-20 minutes per wafer, making it difficult to meet the cycle time requirements of 300mm wafer mass production lines. Traditional automated inspection equipment generally has the following significant limitations: In terms of automation, most systems still require manual intervention for parameter tuning and lack the ability to intelligently adjust to process changes; in terms of control accuracy, the use of open-loop motion control architecture results in repeatability errors generally greater than ±3μm, which cannot meet the detection requirements of advanced processes for nanometer-level defects; in terms of system integration, vision acquisition, motion control, and data processing modules are mostly independent architectures connected through customized interfaces, resulting in data transmission delays and synchronization errors; in terms of programming environment, they mostly rely on equipment manufacturer-specific development platforms, which have long secondary development cycles and poor compatibility, making it difficult to seamlessly integrate with the factory's MES system.

[0003] Existing industrial vision inspection systems face three major bottlenecks in semiconductor wafer inspection scenarios: 1. Insufficient synchronous control precision: The traditional PLC + vision sensor separation architecture results in a time deviation of ±20μs or more between motion axis position feedback and image acquisition triggering, causing defect coordinates to shift during high-speed scanning; 2. Low algorithm integration: Defect recognition algorithms mostly run on independent PCs, lacking data interaction with real-time motion control, making it impossible to achieve dynamic serpentine scanning path adjustment based on defect distribution; 3. Difficulty in multi-source data interaction: Data such as wafer ID information, process parameters, and inspection results are scattered and stored in different subsystems, and data silos lead to serious delays in yield analysis.

[0004] As semiconductor process nodes advance to 3nm and below, the size of wafer surface defects has shrunk to the 0.1μm level. Traditional systems can no longer meet the inspection requirements, and hardware compatibility limitations lead to high production line upgrade costs. Therefore, this invention provides a wafer defect inspection system and method based on industrial vision to solve some of the problems mentioned in the background. Summary of the Invention

[0005] In view of the problems mentioned in the background technology above, the purpose of this invention is to provide a wafer defect detection system and method based on industrial vision. This system integrates high-precision industrial vision acquisition technology with intelligent analysis methods to achieve rapid identification and location of micro-nano-level defects on the wafer surface, and can be applied to the mid-to-back-end process quality inspection of semiconductor wafers of 12 inches and below.

[0006] To achieve the above-mentioned technical objectives, the technical solution adopted by the present invention is as follows: A wafer defect detection system based on industrial vision includes a central control module, an industrial vision inspection module, a motion control module, and a human-computer interaction and communication module. The central control module is the core control unit of the system. The industrial vision inspection module is connected to the central control module and is used to execute a serpentine scanning path. The motion control module interacts with the central control module in real time and is used to drive the wafer stage to move. The human-machine interaction and communication module is connected to the central control module and is used for human-machine interaction. The central control module, together with the industrial vision inspection module, motion control module, and human-computer interaction and communication module, forms a master-slave control architecture. The central control module is used to receive feedback data from each module and output control commands.

[0007] Further specifying, the industrial vision inspection module includes a camera, a multispectral light source unit, and an image acquisition card. The multispectral light source unit is fixedly installed inside the camera lens and is used to generate LED light sources of multiple wavelengths. The image acquisition card is integrated into the central control module and is used to transmit camera images. The camera is vertically positioned directly above the wafer and is used to capture wafer images.

[0008] Further specifying, the motion control module includes a drive unit, a fine-tuning unit, and a rotary platform unit; The drive unit is used to drive the X / Y axes, the fine-tuning unit is used to drive the Z axis, and the rotary platform unit is used to drive the motion control module to rotate as a whole.

[0009] Further, it also includes an internal local area network and a factory MES system. The human-machine interaction and communication module includes a touch screen, which is configured with dual network ports, connected to the internal local area network and the factory MES system respectively. The internal local area network communicates with the central control module, and the factory MES system is used for uploading detection data.

[0010] Further specifying, the multispectral light source unit integrates three wavelength LED light sources: a 450nm blue light source, which is used to detect wafer metal scratches; a 520nm green light source, which is used to identify wafer oxide layer defects; and an 850nm infrared light source, which is used to penetrate the wafer surface to detect subsurface damage.

[0011] A wafer defect detection method based on industrial vision, characterized by comprising the following steps: S1 System Initialization: The central control module starts a self-test program and completes the loading of parameters for each module; S2 Wafer Positioning: The drive unit of the motion control module drives the X / Y axis to move the wafer stage to the initial detection position; S3 Visual Scan: The industrial visual inspection module executes a serpentine scanning path under the synchronous control of the central control module; S4 Defect Identification: Preprocessing and feature extraction of acquired images; S5 Result Feedback: The test data is formatted and then transmitted to the human-computer interaction and communication module.

[0012] Furthermore, the serpentine scanning path generation method in S3 calculates the number of scanning rows based on the wafer diameter parameter and sets the row spacing according to the defect detection accuracy requirements.

[0013] Further specifying, the camera includes an image preprocessing unit and a feature extraction unit. The defect identification in S4 employs the image preprocessing unit and the feature extraction unit. The image preprocessing unit is used to perform Gaussian filtering and contrast enhancement, and the feature extraction unit is used to extract image features.

[0014] Furthermore, the detection data of S5 includes wafer basic information fields, defect coordinate data, defect feature parameters, and detection timestamps.

[0015] Further specifying, the basic wafer information fields include wafer ID, batch number, and process steps; the defect feature parameters include wafer type, size, and grayscale value.

[0016] The beneficial effects of this invention are: This invention achieves a significant leap in detection accuracy through multi-dimensional technological innovation. In defect identification, by setting up a multispectral light source unit, it improves the accuracy of feature extraction and classification of minute defects, and enhances the identification stability in complex defect scenarios compared to traditional machine vision algorithms. The detection efficiency of this invention is improved by optimizing the serpentine scanning path planning and motion control logic, shortening the time required for full-surface wafer inspection and meeting the inspection cycle requirements of mass production lines.

[0017] This invention significantly improves system maintenance convenience through modular design. Core functional modules can be independently disassembled and replaced, simplifying the module replacement process and reducing reliance on specialized tools. The system supports international programming standards, allowing users to expand functionality through structured text or graphical programming. Process parameter adjustments can be configured via a human-machine interface. Furthermore, the consumable replacement cycle is extended through optimized design of optical and moving parts, reducing maintenance frequency. The total lifecycle cost is improved through optimized equipment purchase cost, annual maintenance cost savings, and reduced energy consumption, combined with efficiency improvements, shortening the investment payback period. Attached Figure Description

[0018] The present invention can be further illustrated by the non-limiting embodiments given in the accompanying drawings; Figure 1 This is an overall system module connection diagram of an embodiment of a wafer defect detection system and method based on industrial vision according to the present invention; Figure 2 This invention provides an industrial vision inspection module as an embodiment of a wafer defect detection system and method based on industrial vision. Connection diagram; Figure 3 This is a connection diagram of the motion control module in an embodiment of the wafer defect detection system and method based on industrial vision of the present invention. Figure 4 This is a diagram showing the connection of the human-computer interaction and communication module in an embodiment of a wafer defect detection system and method based on industrial vision according to the present invention. Figure 5 This is a flowchart illustrating the method steps of an embodiment of a wafer defect detection system and method based on industrial vision according to the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments. The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0021] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0022] like Figure 1 As shown, the present invention provides a wafer defect detection system based on industrial vision, comprising a central control module, an industrial vision detection module, a motion control module, and a human-computer interaction and communication module. The central control module is the core control unit of the system. The industrial vision inspection module is connected to the central control module and is used to execute a serpentine scanning path. The motion control module interacts with the central control module in real time and is used to drive the wafer stage to move. The human-machine interaction and communication module is connected to the central control module and is used for human-machine interaction. The central control module, together with the industrial vision inspection module, motion control module, and human-computer interaction and communication module, forms a master-slave control architecture. The central control module is used to receive feedback data from each module and output control commands.

[0023] Specifically, the central control module adopts a modular industrial control architecture based on the CODESYS 3.5 SP16 platform. The hardware layer is configured with an Advantech IPC-610L industrial PC as the main controller, equipped with an Intel Core i7-10700TE processor (8 cores / 16 threads, 2.9GHz clock speed) and 16GB DDR4-3200 ECC memory to ensure real-time task processing capabilities.

[0024] In the practical application of this embodiment, the industrial vision inspection module includes a camera, a multispectral light source unit, and an image acquisition card. The multispectral light source unit is fixedly installed inside the camera lens and is used to generate LED light sources of multiple wavelengths. The image acquisition card is integrated into the central control module and is used to transmit camera photos. The camera is vertically positioned directly above the wafer and is used to take photos of the wafer.

[0025] Specifically, the industrial vision inspection module adopts a multi-sensor fusion architecture, with the core inspection unit configured with a composite acquisition scheme of a Basler L402k linear array camera (4096 pixel resolution, 80kHz line frequency) and a FLIR Blackfly S area array camera (24.5 million pixels, 50fps frame rate).

[0026] In the practical application of this embodiment, the motion control module includes a drive unit, a fine-tuning unit, and a rotary platform unit; The drive unit is used to drive the X / Y axes, the fine-tuning unit is used to drive the Z axis, and the rotary platform unit is used to drive the motion control module to rotate as a whole.

[0027] Specifically, the motion control module adopts a four-axis linkage architecture, with a servo driver as the drive unit and a servo motor as the fine-tuning unit. The servo driver and servo motor are responsible for driving the X, Y, and Z axes of the displacement platform as well as the wafer rotation axis (θ axis).

[0028] In practical applications of this embodiment, an internal local area network and a factory MES system are also included. The human-machine interaction and communication module includes a touch screen, which is configured with dual network ports and connected to the internal local area network and the factory MES system respectively. The internal local area network communicates with the central control module, and the factory MES system is used to detect data uploads.

[0029] In the practical application of this embodiment, the multispectral light source unit integrates three wavelength LED light sources: a 450nm blue light source, which is used to detect wafer metal scratches; a 520nm green light source, which is used to identify wafer oxide layer defects; and an 850nm infrared light source, which is used to penetrate the wafer surface to detect subsurface damage.

[0030] A wafer defect detection method based on industrial vision, characterized by comprising the following steps: S1 System Initialization: The central control module starts a self-test program and completes the loading of parameters for each module; S2 Wafer Positioning: The drive unit of the motion control module drives the X / Y axis to move the wafer stage to the initial detection position; S3 Visual Scan: The industrial visual inspection module executes a serpentine scanning path under the synchronous control of the central control module; S4 Defect Identification: Preprocessing and feature extraction of acquired images; S5 Result Feedback: The test data is formatted and then transmitted to the human-computer interaction and communication module.

[0031] In the practical application of this embodiment, the serpentine scanning path generation method in S3 calculates the number of scanning rows based on the wafer diameter parameter and sets the row spacing according to the defect detection accuracy requirements.

[0032] In the practical application of this embodiment, the camera includes an image preprocessing unit and a feature extraction unit. The defect identification in S4 uses the image preprocessing unit and the feature extraction unit. The image preprocessing unit is used to perform Gaussian filtering and contrast enhancement, and the feature extraction unit is used to extract image features.

[0033] In the practical application of this embodiment, the detection data of S5 includes wafer basic information fields, defect coordinate data, defect feature parameters, and detection timestamps.

[0034] In the practical application of this embodiment, the basic wafer information fields include wafer ID, batch number, and process steps; the defect feature parameters include wafer type, size, and grayscale value.

[0035] The working principle of this solution is as follows: After power-on, the system automatically enters the initialization process. The central control module first performs a hardware self-test: using a built-in diagnostic program to sequentially check CPU temperature, memory integrity, storage device health status, and bus communication status. After the self-test passes, the motion control module initiates the axis homing procedure: the X / Y axes search for the negative limit switch at a speed of 10mm / s, and upon triggering, perform a reverse micro-motion to the grating ruler reference point; the Z-axis performs contact-type homing, and then probes the reference plane through a force sensor (trigger threshold 500mN); the θ-axis rotates to the mechanical zero point and establishes an absolute coordinate system. The initialization phase also includes optical system calibration: automatically adjusting the lens aperture and focal length, and completing image distortion correction and pixel equivalent calibration using a standard calibration board.

[0036] Before the inspection process begins, the motion control module automatically generates a serpentine scanning path based on the wafer diameter (supporting 4-inch / 6-inch / 8-inch / 12-inch wafers): the Bresenham algorithm is used to calculate the number of scan rows, with 600 rows corresponding to a 12-inch wafer, and the row spacing is set to 50μm according to the defect detection accuracy requirements. The path data is stored in the memory buffer of the central control module and sent to the servo driver periodically via the bus.

[0037] Visual triggering employs a hybrid mode of "primarily position-triggered, secondarily time-triggered": the linear scan camera acquires data every 50μm movement along the X-axis (position trigger), while the area scan camera captures images at fixed time intervals (time trigger). Both trigger signals are generated by a hardware timer in the central control module. Defect coordinate transformation is based on the wafer coordinate system: with the wafer center as the origin, the X-axis is along the flat edge direction, and the Y-axis is perpendicular to the X-axis. Furthermore, the system's data buffer uses a dual-buffer mechanism: the current buffer receives real-time image data, while the background buffer performs preprocessing to prevent data loss.

[0038] Image data processing employs a pipeline architecture: first, preprocessing is performed (Gaussian filtering with σ=1.5 to remove noise, and the CLAHE algorithm to enhance contrast), followed by edge detection to extract defect contours. Detection results are stored in XML format, with the root node containing data such as wafer ID, batch number, and detection time, and child nodes arranged in order of defect ID, recording their X / Y coordinates, size, and grayscale features.

[0039] In summary, this invention achieves a significant leap in detection accuracy through multi-dimensional technological innovation. In defect identification, the use of a multispectral light source unit enhances the accuracy of feature extraction and classification for minute defects, improving recognition stability in complex defect scenarios compared to traditional machine vision algorithms. The invention also improves detection efficiency by optimizing the serpentine scanning path planning and motion control logic, shortening the time required for full-surface wafer inspection and meeting the inspection cycle requirements of mass production lines.

[0040] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A wafer defect detection system based on industrial vision, characterized in that: It includes a central control module, an industrial vision inspection module, a motion control module, and a human-computer interaction and communication module; The central control module is the core control unit of the system. The industrial vision inspection module is connected to the central control module and is used to execute a serpentine scanning path. The motion control module interacts with the central control module in real time and is used to drive the wafer stage to move. The human-machine interaction and communication module is connected to the central control module and is used for human-machine interaction. The central control module, together with the industrial vision inspection module, motion control module, and human-computer interaction and communication module, forms a master-slave control architecture. The central control module is used to receive feedback data from each module and output control commands.

2. The wafer defect detection system based on industrial vision according to claim 1, characterized in that: The industrial vision inspection module includes a camera, a multispectral light source unit, and an image acquisition card. The multispectral light source unit is fixedly installed inside the camera lens and is used to generate LED light sources of multiple wavelengths. The image acquisition card is integrated into the central control module and is used to transmit camera images. The camera is vertically positioned directly above the wafer and is used to capture images of the wafer.

3. The wafer defect detection system based on industrial vision according to claim 1, characterized in that: The motion control module includes a drive unit, a fine-tuning unit, and a rotary platform unit; The drive unit is used to drive the X / Y axes, the fine-tuning unit is used to drive the Z axis, and the rotary platform unit is used to drive the motion control module to rotate as a whole.

4. The wafer defect detection system based on industrial vision according to claim 1, characterized in that: It also includes an internal local area network and a factory MES system. The human-machine interaction and communication module includes a touch screen, which is equipped with dual network ports, connected to the internal local area network and the factory MES system respectively. The internal local area network communicates with the central control module, and the factory MES system is used for uploading detection data.

5. A wafer defect detection system based on industrial vision according to claim 2, characterized in that: The multispectral light source unit integrates three wavelength LED light sources: a 450nm blue light source, which is used to detect wafer metal scratches; a 520nm green light source, which is used to identify wafer oxide layer defects; and an 850nm infrared light source, which is used to penetrate the wafer surface to detect subsurface damage.

6. A wafer defect detection method based on industrial vision, characterized in that, Includes the following steps: S1 System Initialization: The central control module starts a self-test program and completes the loading of parameters for each module; S2 Wafer Positioning: The drive unit of the motion control module drives the X / Y axis to move the wafer stage to the initial detection position; S3 Visual Scan: The industrial visual inspection module executes a serpentine scanning path under the synchronous control of the central control module; S4 Defect Identification: Preprocessing and feature extraction of acquired images; S5 Result Feedback: The test data is formatted and then transmitted to the human-computer interaction and communication module.

7. The wafer defect detection method based on industrial vision according to claim 6, characterized in that: The serpentine scanning path generation method in S3 calculates the number of scanning rows based on the wafer diameter parameter and sets the row spacing according to the defect detection accuracy requirements.

8. The wafer defect detection method based on industrial vision according to claim 6, characterized in that: The camera includes an image preprocessing unit and a feature extraction unit. The defect identification in S4 uses the image preprocessing unit and the feature extraction unit. The image preprocessing unit is used to perform Gaussian filtering and contrast enhancement, and the feature extraction unit is used to extract image features.

9. The wafer defect detection method based on industrial vision according to claim 6, characterized in that: The detection data of S5 includes wafer basic information fields, defect coordinate data, defect feature parameters, and detection timestamps.

10. A wafer defect detection method based on industrial vision according to claim 9, characterized in that: The basic information fields of the wafer include wafer ID, batch number and process steps; the defect feature parameters include wafer type, size and grayscale value.

Images