Wafer cassette docking system

Abstract

This specification provides a wafer cassette docking system that, through the systematic integration of local environmental control, multimodal sensing, and active compliant adjustment functions, achieves an overall performance improvement in the wafer cassette docking process. It can form a stable positive pressure laminar flow air curtain at the initial stage of docking, effectively isolating external particles and providing a continuous ultra-clean environment for the wafer. Employing a multi-sensor fusion scheme consisting of global vision, laser ranging, and feature recognition, it can quickly and comprehensively acquire the six-degree-of-freedom pose information of the wafer cassette, providing a high-precision data foundation for subsequent fine alignment. A micro-motion adjustment platform based on piezoelectric ceramic actuators, combined with real-time closed-loop feedback control, enables nanometer-level active compliant alignment, significantly improving alignment accuracy and reliability. An electromagnetic locking mechanism ensures the stable maintenance of the final pose. The collaborative work of each module improves docking accuracy and efficiency, and also enhances the system's adaptability and stability under different operating conditions.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of equipment operation and maintenance technology, and in particular to wafer cassette docking systems. Background Technology

[0002] In semiconductor manufacturing, precise docking between the wafer cassette and the load ports of process equipment is crucial for ensuring wafer transport cleanliness and operational reliability. As semiconductor technology nodes continue to shrink, the requirements for cleanroom control and sub-millimeter alignment accuracy become increasingly stringent. Existing docking methods often rely on the repeatability of robotic arms and simple mechanical guidance, which have limitations in meeting the high standards of advanced semiconductor manufacturing. These methods typically lack active environmental isolation measures, making it difficult to effectively block external particulate contaminants; their positioning methods are mostly guided by a single sensor, unable to fully compensate for the wafer cassette's pose deviations across multiple degrees of freedom; and the adjustment process is often open-loop or semi-closed-loop control, making it difficult to achieve nanometer-level precision alignment and stable locking. These factors collectively limit the application potential of existing technologies in high-precision semiconductor manufacturing, easily leading to problems such as wafer contamination, positioning errors, and reduced equipment uptime.

[0003] Therefore, a better solution is urgently needed. Summary of the Invention

[0004] In view of this, embodiments of this specification provide a wafer cassette docking system to address the technical deficiencies existing in the prior art.

[0005] According to a first aspect of the embodiments of this specification, a wafer cassette docking system is provided, comprising:

[0006] The central coordination controller is configured to perform system self-tests and coordination control; The local environmental control module, connected to the central co-controller, includes a particulate air filter and a centrifugal fan. It is configured to supply filtered, temperature- and humidity-controlled dry air or nitrogen to the wafer box docking area and form a stable, vertically downward laminar air curtain with a pressure slightly higher than atmospheric pressure. The visual positioning module, connected to the central co-controller, includes a global wide-angle vision unit, a local laser ranging unit, and a feature point recognition unit. The global wide-angle vision unit is configured to capture images of the wafer cassette's outline and bottom guide groove and calculate macroscopic deviations in the horizontal plane and rotation angle. The local laser ranging unit is configured to measure the vertical distance between the wafer cassette's bottom surface and the load port platform reference plane, as well as the sidewall tilt angle. The feature point recognition unit is configured to image specific marks on the wafer cassette door to obtain feature point position information. The central co-controller is configured to fuse data from the global wide-angle vision unit, the local laser ranging unit, and the feature point recognition unit to generate a six-degree-of-freedom pose deviation matrix. The six-degree-of-freedom micro-motion adjustment module, connected to the central coordinating controller, includes a parallel piezoelectric ceramic drive platform and an electromagnetic locking mechanism. The piezoelectric ceramic drive platform is configured to perform nanometer-level precision micro-displacement and micro-angle adjustments based on the six-degree-of-freedom pose deviation matrix, and the electromagnetic locking mechanism is configured to lock the wafer cassette after adjustment.

[0007] In one possible implementation, the global wide-angle vision unit is configured to calculate the macroscopic deviations between the wafer box and the target location on the X, Y, and θz axes.

[0008] In one possible implementation, the local laser ranging unit is configured to measure the distance between the bottom surface of the wafer cassette and the reference surface of the load port platform on the Z-axis and the tilt angles on the θx and θy axes.

[0009] In one possible implementation, the feature point recognition unit is configured to image a QR code or crosshair target on the wafer box door.

[0010] In one possible implementation, the piezoelectric ceramic driving platform is the Stewart platform.

[0011] In one possible implementation, the electromagnetic locking mechanism is configured to generate a magnetic force to lock the wafer cell momentarily upon reaching the target position.

[0012] In one possible implementation, the local laser ranging unit and the feature point recognition unit are configured to provide real-time feedback, forming a closed-loop control with the central co-controller until the deviations of all degrees of freedom are within the tolerance range.

[0013] In one possible implementation, the local laser ranging unit is configured to calculate the height deviation. and tilt deviation , The calculation formula includes the following steps: Step 1: Calculate the average height of the three laser measurement points. :

[0014] in Let i be the height value measured by the i-th laser displacement sensor, where i = 1, 2, 3. Real-time measurement data from the local laser ranging unit; Step 2: Calculate the height deviation ΔZ:

[0015] in For the target height value, Preset values ​​from the load port platform reference plane; Step 3: Calculate the tilt deviation in the X direction :

[0016] in The mounting spacing between the first and second laser sensors in the X direction. Derived from sensor installation design values; Step 4: Calculate the tilt deviation in the Y direction :

[0017] in The mounting spacing between the first and third laser sensors in the Y direction. From sensor installation design values.

[0018] In one possible implementation, the feature point recognition unit is configured to calculate the fine translational deviations ΔX_f and ΔY_f, the calculation of which includes the following steps: Step 1: Extract the pixel coordinates of feature points from the image and convert to current world coordinates. :

[0019]

[0020] in Pixel equivalent, unit: millimeters per pixel. From camera calibration parameters, The center pixel coordinates of the image. From camera calibration parameters, Image processing results from the feature point recognition unit; Step 2: Calculate the translation deviation and :

[0021]

[0022] in The world coordinates of the feature point at the target location. From preset target coordinates.

[0023] In one possible implementation, the central coordinating controller is configured to complete a self-test procedure before the local environmental control module starts the particulate air filter and centrifugal fan.

[0024] This invention achieves a comprehensive performance improvement in the wafer cassette docking process by systematically integrating local environmental control, multimodal sensing, and active compliant adjustment functions. The system can form a stable positive-pressure laminar flow curtain at the initial docking stage, effectively isolating external particles and providing a continuous ultra-clean environment for the wafer. Employing a multi-sensor fusion scheme consisting of global vision, laser ranging, and feature recognition, it can quickly and comprehensively acquire the six-degree-of-freedom pose information of the wafer cassette, providing a high-precision data foundation for subsequent fine alignment. A micro-motion adjustment platform based on piezoelectric ceramic actuators, combined with real-time closed-loop feedback control, enables nanometer-level active compliant alignment, significantly improving alignment accuracy and reliability. An electromagnetic locking mechanism ensures the stable maintenance of the final pose. All modules work collaboratively under the unified scheduling of a central coordinating controller, not only improving docking accuracy and efficiency but also enhancing the system's adaptability and stability to different operating conditions, thereby meeting the extreme requirements of cleanliness, precision, and reliability in high-end semiconductor manufacturing. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of a wafer cassette docking system provided in one embodiment of this specification. Detailed Implementation

[0026] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0027] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0028] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0029] This specification provides a wafer cassette docking system, which will be described in detail in the following embodiments.

[0030] See Figure 1 , Figure 1 The diagram illustrates a wafer cassette docking system according to one embodiment of this specification. Specifically, it includes a central coordinating controller configured to perform system self-tests and coordinating control; a local environmental control module connected to the central coordinating controller, including a particulate air filter and a centrifugal fan, configured to supply filtered, temperature- and humidity-controlled dry air or nitrogen to the wafer cassette docking area, forming a stable, vertically downward laminar air curtain with a pressure slightly higher than atmospheric pressure; and a visual positioning module connected to the central coordinating controller, including a global wide-angle vision unit, a local laser ranging unit, and a feature point recognition unit. The global wide-angle vision unit is configured to capture images of the wafer cassette's outline and bottom guide groove and calculate the horizontal plane and rotation. For macroscopic angular deviations, the local laser ranging unit is configured to measure the vertical distance and sidewall tilt angle between the bottom surface of the wafer cassette and the reference surface of the load port platform. The feature point recognition unit is configured to image specific marks on the wafer cassette door to obtain feature point position information. The central coordinating controller is configured to fuse data from the global wide-angle vision unit, the local laser ranging unit, and the feature point recognition unit to generate a six-degree-of-freedom pose deviation matrix. The six-degree-of-freedom micro-motion adjustment module, connected to the central coordinating controller, includes a parallel piezoelectric ceramic drive platform and an electromagnetic locking mechanism. The piezoelectric ceramic drive platform is configured to perform nanometer-level precision micro-displacement and micro-angle adjustments based on the six-degree-of-freedom pose deviation matrix. The electromagnetic locking mechanism is configured to lock the wafer cassette after adjustment.

[0031] The central coordinating controller refers to the core control unit of the system, used to perform system self-tests and coordinate the operation of various modules. The local environmental control module refers to a subsystem that manages the environmental conditions of the wafer cassette docking area, capable of activating particulate air filters and centrifugal fans to supply clean air and form a laminar airflow curtain. A particulate air filter is a high-efficiency filtration device that removes ultrafine particles from the air, ensuring air cleanliness. A centrifugal fan is a device that generates airflow to propel the filtered air into a stable airflow curtain. The visual positioning module refers to a positioning system integrating multiple sensors, used to acquire the wafer cassette's pose information through visual and laser measurements. The global wide-angle vision unit refers to a wide-angle camera assembly capable of capturing the overall outline of the wafer cassette and calculating macroscopic deviations. The local laser ranging unit refers to a laser sensor array capable of accurately measuring the height and tilt angle of the wafer cassette. The feature point recognition unit refers to a high-resolution imaging unit that identifies markings on the wafer cassette doors and extracts feature point positions. The six-DOF pose deviation matrix refers to a data matrix representing the deviation of the wafer cassette in six degrees of freedom. A six-degree-of-freedom micro-motion adjustment module can refer to a precision adjustment platform capable of nanometer-level pose correction based on a deviation matrix. A parallel piezoelectric ceramic drive platform can refer to a multi-axis platform based on the piezoelectric effect, enabling high-precision micro-displacement and micro-angle adjustments. An electromagnetic locking mechanism can refer to an electromagnetically driven locking device used to fix the wafer cassette pose after adjustment.

[0032] As a specific example: After the system is powered on, the central coordinating controller completes a self-test, the local environmental control module starts, and the particulate air filter (filtration efficiency of 99.995% for 0.1μm particles) and centrifugal fan (speed of 3000 RPM) supply dry air at a temperature of 22℃ and a humidity of 45% to the wafer box to be docked, forming a laminar flow air curtain with a pressure of 101.5 kPa. The robotic arm transports the wafer cassette to the vicinity of the load port, and the vision positioning module is activated: the global wide-angle vision unit (focal length 5mm) captures the outline of the wafer cassette, calculates the X-axis deviation of -0.5mm, the Y-axis deviation of 0.3mm, and the θz-axis deviation of 0.2°; the local laser ranging unit (three laser sensors with a spacing of dx=100mm and dy=150mm) measures the Z-axis height values ​​Z1=10.1mm, Z2=10.2mm, and Z3=10.0mm, calculates the average height Zavg=10.1mm, and derives the θx tilt angle of 0.05° and the θy tilt angle of -0.03°; the feature point recognition unit (pixel equivalent s=0.01mm / pixel) images the cross-shaped target at the wafer cassette door, extracts the pixel coordinates (u=1024, v=768) and converts them into world coordinates (Xc=10.24mm, Yc=7.68mm). The central coordinating controller fuses data to generate a six-degree-of-freedom pose deviation matrix [ΔX=-0.5mm, ΔY=0.3mm, ΔZ=0.1mm, Δθx=0.05°, Δθy=-0.03°, Δθz=0.2°]. The parallel piezoelectric ceramic drive platform (1nm resolution) of the six-degree-of-freedom micro-motion adjustment module adjusts the pose within 10ms, with real-time feedback from the local laser ranging unit and feature point recognition unit, until the deviation is less than the tolerance (±0.01mm, ±0.01°). Then, the electromagnetic locking mechanism (500N magnetic force) is energized to lock the wafer cassette.

[0033] This invention achieves efficient collaboration among system modules through unified scheduling by a central coordinating controller. The positive pressure laminar flow air curtain formed by the local environmental control module effectively isolates external contaminants, improving the cleanliness of the wafer processing environment. The multi-sensor fusion scheme of the visual positioning module comprehensively acquires six-degree-of-freedom pose information, avoiding the limitations of a single sensor and improving positioning accuracy and reliability. The active compliant adjustment of the six-degree-of-freedom micro-motion adjustment module, combined with real-time closed-loop control, achieves nanometer-level precision alignment, significantly reducing docking errors. The electromagnetic locking mechanism ensures the stability of the final pose and prevents the influence of micro-vibrations. The overall system design enhances the automation and adaptability of the docking process, meets the high standards of semiconductor manufacturing, and improves production efficiency and product yield.

[0034] In one possible implementation, the global wide-angle vision unit is configured to calculate the macroscopic deviations between the wafer box and the target location on the X, Y, and θz axes.

[0035] The target position refers to the preset ideal docking pose of the wafer cassette on the load port, serving as a reference for pose adjustment. The X-axis can be an orthogonal coordinate axis defined in the horizontal plane, representing the translational dimension of the wafer cassette in the left-right direction. The Y-axis can be another orthogonal coordinate axis perpendicular to the X-axis in the horizontal plane, representing the translational dimension of the wafer cassette in the front-back direction. The θz-axis refers to the rotational degree of freedom around the vertical axis (Z-axis), representing the yaw angle of the wafer cassette in the horizontal plane. Macroscopic deviation refers to the relatively large pose error initially identified through global vision, used as the initial input for subsequent fine alignment.

[0036] As a concrete example: The global wide-angle vision unit uses an industrial camera with a focal length of 8mm, whose field of view covers the entire wafer cassette docking area. When the wafer cassette is transported by the robotic arm to a detection area approximately 50mm in front of the load port, the unit captures a global image with a resolution of 2048×1536 pixels. The image processing algorithm first identifies the rectangular outline of the wafer cassette and determines the boundary line of its bottom guide groove through edge detection. By comparing the detected outline corner coordinates (e.g., the top left corner pixel coordinates (255, 190)) with the preset target position corner coordinates (e.g., the target top left corner pixel coordinates (250, 200)), the translational deviation of the wafer cassette center in the X-axis direction is calculated to be +2.5mm, and the translational deviation in the Y-axis direction is -1.8mm. Simultaneously, by comparing the angle between the detected guide groove boundary line and the horizontal axis of the image coordinate system, the rotational deviation of the wafer cassette around the θz axis is calculated to be +0.5 degrees. These calculated macroscopic deviation data are sent to the central coordinating controller to provide initial guidance for subsequent local fine-tuning measurements and pose adjustments.

[0037] This invention utilizes a global wide-angle vision unit to rapidly acquire the overall spatial relationships of the wafer cassette, achieving initial coarse positioning during the docking process. This effectively narrows the search range for subsequent fine alignment, improving the overall system efficiency. This unit can simultaneously calculate macroscopic deviations across multiple degrees of freedom, providing the system with comprehensive initial pose information and preventing fine alignment failures or excessively long processing times due to large initial deviations. This macroscopic-to-microscopic positioning strategy optimizes the workflow and improves the robustness and response speed of the docking system.

[0038] In one possible implementation, the local laser ranging unit is configured to measure the distance between the bottom surface of the wafer cassette and the reference surface of the load port platform on the Z-axis and the tilt angles on the θx and θy axes.

[0039] In this system, the wafer cell bottom surface refers to the bottom plane of the wafer cell that contacts the load port platform, serving as a reference plane for vertical distance measurement. The load port platform reference plane refers to a precisely calibrated horizontal reference plane on the load port, serving as a spatial coordinate reference for measuring the wafer cell's pose. The Z-axis refers to a coordinate axis perpendicular to the horizontal plane in a three-dimensional coordinate system, representing the wafer cell's height dimension in the vertical direction. The θx-axis refers to the tilt angle degree of freedom around the X-axis, representing the wafer cell's pitch angle in the forward / backward direction. The θy-axis refers to the tilt angle degree of freedom around the Y-axis, representing the wafer cell's roll angle in the left / right direction. The tilt angle refers to the degree of tilt of the object's surface relative to the horizontal reference plane, used to quantify the wafer cell's angular deviation in space.

[0040] As a concrete example: the local laser ranging unit consists of three high-precision laser displacement sensors arranged in a right-angled triangle below the load port platform. After the wafer cassette is coarsely positioned, all three sensors simultaneously emit laser beams towards the bottom surface of the wafer cassette. Sensor 1 measures a distance value Z1 of 10.15 mm, sensor 2 measures Z2 of 10.25 mm, and sensor 3 measures Z3 of 10.10 mm. After receiving these data, the central coordinating controller first calculates the average height value Z_avg, which is 10.167 mm. This is the average distance on the Z-axis between the bottom surface of the wafer cassette and the platform reference plane. Subsequently, the controller calculates the tilt angles based on the installation distances between the sensors (X-direction spacing dx = 200 mm, Y-direction spacing dy = 150 mm) using trigonometric functions: the tilt angle about the θx axis is arctan((Z2-Z1) / dx) = arctan(0.10 / 200) ≈ 0.0286 degrees, representing pitch in the forward / backward direction; the tilt angle about the θy axis is arctan((Z3-Z1) / dy) = arctan(-0.05 / 150) ≈ -0.0191 degrees, representing roll in the left / right direction. These precise distance and angle data are transmitted in real-time to the central controller to update the six-degree-of-freedom pose deviation matrix.

[0041] This invention achieves high-precision non-contact measurement of the vertical distance and dual-axis tilt angle of a wafer cassette using a local laser ranging unit, providing crucial spatial attitude data for pose adjustment. This unit can monitor minute deviations between the wafer cassette and the ideal docking plane in real time, effectively ensuring the accuracy of subsequent precision alignment operations. The method of calculating the tilt angle through multi-point measurement eliminates potential local errors in single-point measurement, improving the reliability and representativeness of the measurement results. This direct spatial relationship measurement compensates for the limitations of visual positioning in vertical accuracy, complementing the vision system to jointly construct complete six-degree-of-freedom pose information, significantly improving the accuracy and stability of the entire docking system.

[0042] In one possible implementation, the feature point recognition unit is configured to image a QR code or crosshair target on the wafer box door.

[0043] Among these, a wafer enclosure door can refer to an openable and closable panel structure on a wafer enclosure, used to seal and protect the wafers inside. A QR code can refer to a geometric code distributed according to a specific pattern on a two-dimensional plane, capable of storing the wafer enclosure's identification information or location reference data. A crosshair target can refer to a calibration mark formed by the intersection of mutually perpendicular lines, providing a precise visual positioning reference point. Imaging can refer to the process of using an optical system to focus light reflected from an object onto an image sensor to form a digital image, used to acquire visual information about feature points.

[0044] As a concrete example: The feature point recognition unit employs a high-resolution industrial camera with a telecentric lens and a resolution of 5 megapixels, mounted above the load port directly opposite the wafer cassette door. After the wafer cassette is coarsely positioned, the unit images specific markings on the surface of the wafer cassette door. If a 10mm side-length QR code is attached to the wafer cassette door, the camera first captures a clear image of the QR code, reads its encoded information (e.g., batch number ABC123) using an image decoding algorithm, and simultaneously calculates the precise pixel coordinates of the QR code's center point in the image coordinate system (e.g., u=1250, v=950). If a crosshair target is used on the wafer cassette door, the camera captures an image of the crosshair and determines the precise pixel position of the intersection point using a sub-pixel edge detection algorithm (e.g., u=1248.5, v=951.2). The pixel coordinates are then converted into world coordinates (Xc, Yc) relative to the camera coordinate system based on pre-calibrated camera parameters (pixel equivalent s = 0.005 mm / pixel, image center u0 = 1280, v0 = 1024). This coordinate data is transmitted to the central co-controller as a direct input for fine translational deviations, used to correct minor deviations remaining after macroscopic positioning.

[0045] This invention achieves the highest level of positioning accuracy during the docking process by imaging and analyzing high-precision markings on wafer cell doors using a feature point recognition unit. This unit can acquire sub-pixel-level feature point position information, effectively compensating for subtle deviations remaining after macroscopic positioning and providing crucial data support for final precise alignment. Utilizing standardized feature markings such as QR codes or crosshairs improves the reliability and repeatability of the recognition process and reduces sensitivity to changes in ambient lighting. This unit works in conjunction with a global wide-angle vision and laser ranging unit, forming a progressive measurement system from macro to micro and from coarse to precise, jointly ensuring that the entire docking process ultimately achieves nanometer-level alignment requirements, significantly improving the overall performance of the system.

[0046] In one possible implementation, the piezoelectric ceramic driving platform is the Stewart platform.

[0047] The Stewart platform can refer to a parallel mechanism consisting of six independently extendable legs connecting two platforms, used to achieve precise motion control with six degrees of freedom.

[0048] As a concrete example: the Stewart platform's upper platform carries the wafer cassette via a fixture, while the lower platform is fixed to the main structure of the load port. All six legs are piezoelectric ceramic actuators, each with a travel range of 0 to 50 micrometers and a displacement resolution of 1 nanometer. When the central co-controller issues an adjustment command based on a six-degree-of-freedom pose deviation matrix, the six legs coordinate their movements according to their respective required extensions calculated by inverse kinematics algorithms. For example, to compensate for a pitch angle deviation of Δθx = 0.02 degrees, the two legs located at the front of the platform (legs 1 and 2) may need to retract by 0.5 micrometers, while the two legs at the rear (legs 5 and 6) extend by 0.5 micrometers accordingly. The side legs (legs 3 and 4) are fine-tuned to keep other degrees of freedom unchanged. All legs complete synchronous displacement with nanometer-level precision within 10 milliseconds, driving the upper platform and wafer cassette to achieve the required micrometer-level translation and milliradian-level angle adjustment. Throughout the adjustment process, the actual displacement of the outriggers is fed back by built-in strain sensors, forming a closed-loop control to ensure the accuracy of the posture adjustment.

[0049] Employing the Stewart platform as the core structure of the piezoelectric ceramic drive platform significantly enhances the performance of the micro-motion adjustment module. Its parallel mechanism boasts advantages such as high rigidity, high load capacity, and excellent motion decoupling, enabling clean and precise six-degree-of-freedom pose adjustments. The inverse kinematics-based control strategy ensures coordinated movement of all legs with rapid response, effectively eliminating the error accumulation and excessive inertia issues inherent in traditional serial mechanisms. This design guarantees that the wafer cassette can be stably and accurately guided to the target pose in an extremely short time, laying a solid foundation for reliable locking and enhancing the overall dynamic performance and stability of the system in high-precision docking tasks.

[0050] In one possible implementation, the electromagnetic locking mechanism is configured to generate a magnetic force to lock the wafer cell momentarily upon reaching the target position.

[0051] Instantaneous energization refers to the action of providing working current to the electromagnet within a very short time after receiving the locking command, enabling a rapid response in the locking operation. Magnetic force refers to the attractive or repulsive force generated by the energized electromagnet, acting on the magnetic material to achieve physical locking. Locking refers to the operation of eliminating degrees of freedom of movement and fixing relative positions through mechanical or electromagnetic means, used to maintain the stability of the adjusted posture.

[0052] As a concrete example: Once the central co-controller confirms that all elements of the six-DOF pose deviation matrix are within the preset tolerance range (e.g., translation deviation less than ±1 micrometer, angular deviation less than ±0.001 degrees), it immediately sends a TTL high-level signal to the electromagnetic locking mechanism. This signal triggers a high-current drive circuit, applying a 24V, 5A pulse current to the electromagnet coil embedded in the load port platform within less than 5 milliseconds. The energized electromagnet instantly generates a strong magnetic attraction of up to 800 Newtons, which acts directly on the low-carbon steel magnetic plate fixed to the Stewart platform. The powerful magnetic force tightly fits the upper and lower platforms, completely eliminating any tiny gaps and potential drift between them in all degrees of freedom. The locking state is confirmed by a Hall sensor integrated near the electromagnet, which detects the magnetic field strength and feeds back a "locked" signal to the central controller. Only then can the robotic arm safely detach from the wafer cassette, completing the docking process.

[0053] The introduction of the electromagnetic locking mechanism provides instantaneous and stable holding force for the wafer cassette after precision alignment, effectively preventing positional shifts caused by external vibrations or internal stress relaxation. Its rapid response ensures seamless integration of the locking action and precision alignment, significantly reducing non-productive time throughout the entire docking cycle. The non-contact magnetic locking method avoids particle contamination and impact damage that can occur with mechanical locking, particularly meeting the high cleanliness and equipment protection requirements of semiconductor manufacturing. Working in conjunction with the piezoelectric ceramic drive platform, this mechanism forms a complete "active adjustment-passive holding" work cycle, significantly improving the long-term stability and reliability of the wafer cassette during the positional holding phase, thereby ensuring the smooth execution of subsequent process steps.

[0054] In one possible implementation, the local laser ranging unit and the feature point recognition unit are configured to provide real-time feedback, forming a closed-loop control with the central co-controller until the deviations of all degrees of freedom are within the tolerance range.

[0055] Real-time feedback refers to the process by which the measurement unit immediately transmits continuously acquired data to the controller, forming an information feedback channel in the control system. Closed-loop control refers to a control method that feeds back the detected value of the system output to the input and compares it with the given value to form a deviation signal, which can automatically adjust the system behavior based on the deviation until the target is achieved. Tolerance range refers to a pre-set limit range of allowable deviation, serving as a standard for judging whether the adjustment process is complete.

[0056] As a concrete example: During the precision alignment stage, the local laser ranging unit continuously measures the height values ​​(Z1, Z2, Z3) of three points on the bottom surface of the wafer cassette at a frequency of 1 kHz, while the feature point recognition unit simultaneously captures the image of the crosshair target at a frequency of 100 Hz and calculates its center coordinates (Xc, Yc). This data is packaged and sent to the central co-controller in real time. Within each control cycle (e.g., 1 millisecond), the controller updates the six-degree-of-freedom pose deviation matrix based on the latest data. Assume that the deviation calculated in the current cycle is [ΔX=0.5μm, ΔY=0.3μm, ΔZ=0.8μm, Δθx=0.0008°, Δθy=0.0005°, Δθz=0.0002°], and the preset tolerance range is [±1μm, ±1μm, ±1μm, ±0.001°, ±0.001°, ±0.001°]. The controller determines that all deviations, while approaching, have not yet fully entered the tolerance range (e.g., the convergence threshold of ΔZ=0.8μm>0.5μm). Therefore, it generates new fine-tuning instructions and sends them to the Stewart platform. The platform then performs nanometer-level adjustments again. This measurement-calculation-adjustment-verification cycle continues until the next control cycle detects that all deviations satisfy ΔX=0.2μm, ΔY=0.1μm, ΔZ=0.4μm, Δθx=0.0004°, Δθy=0.0003°, Δθz=0.0001°, and are all within the tolerance range. Only then does the closed-loop control process end.

[0057] This invention significantly improves the accuracy, automation, and reliability of the alignment process by establishing a real-time feedback and closed-loop control mechanism. The system can dynamically respond to minute deviations generated during adjustment, achieving continuous automatic correction and effectively overcoming the shortcomings of open-loop control, which is susceptible to interference and system errors. This strategy of continuously comparing the target with the current state and adjusting in real time ensures that the alignment result ultimately stabilizes within a preset high standard tolerance range, greatly reducing the risk of docking failure. The entire control process requires no manual intervention, improving work efficiency and system intelligence, and providing a crucial guarantee for long-term stable and highly repeatable docking between the wafer cell and the load port.

[0058] In one possible implementation, the local laser ranging unit is configured to calculate the height deviation. and tilt deviation , The calculation formula includes the following steps: Step 1: Calculate the average height of the three laser measurement points. :

[0059] in Let i be the height value measured by the i-th laser displacement sensor, where i = 1, 2, 3. Real-time measurement data from the local laser ranging unit; Step 2: Calculate the height deviation ΔZ:

[0060] in For the target height value, Preset values ​​from the load port platform reference plane; Step 3: Calculate the tilt deviation in the X direction :

[0061] in The mounting spacing between the first and second laser sensors in the X direction. Derived from sensor installation design values; Step 4: Calculate the tilt deviation in the Y direction :

[0062] in The mounting spacing between the first and third laser sensors in the Y direction. From sensor installation design values.

[0063] Among them, height deviation This can refer to the vertical difference between the average height of the wafer cassette bottom surface and the target height, used to quantify the pose error in the Z-axis direction. Tilt deviation. This can refer to the pitch angle deviation caused by the rotation of the wafer cassette around the X-axis, and can represent the angular error in the forward and backward directions. Tilt deviation. This can refer to the roll angle deviation caused by the rotation of the wafer cassette around the Y-axis, representing the angular error in the left-right direction. Average height. This can refer to the arithmetic mean of measurements from three laser measuring points, used to represent the overall height level of the bottom surface of the wafer cassette. A laser displacement sensor can refer to a device that uses laser triangulation or interferometry to measure distance, accurately obtaining the relative position of an object's surface. Real-time measurement data refers to the data stream that a sensor collects and outputs in real-time during continuous operation, providing the current status information required by the control system. Target height value. This can refer to the coordinates of the pre-defined ideal mating position along the Z-axis, serving as a reference value for deviation calculation. Installation spacing. This can refer to the center-to-center distance between two laser sensors in the X-direction, used to convert height difference into angle calculations. Installation spacing. This can refer to the center distance between two laser sensors in the Y direction, to support the calculation of angular deviations in different directions.

[0064] As a concrete example: A local laser ranging unit continuously collects height data at three measuring points at a frequency of 1000 Hz. Within a certain control cycle, the measured heights are Z1 = 1002.5 μm, Z2 = 1001.8 μm, and Z3 = 1003.2 μm. The processing unit first calculates the average height. =(1002.5+1001.8+1003.2) / 3=1002.5 micrometers. Let the target height be... =1000.0 micrometers, then the height deviation ΔZ = 1002.5 - 1000.0 = +2.5 micrometers. The sensor mounting spacing is known. =50.0 mm, =60.0 mm, then calculate the tilt deviation in the X direction. =arctan((1001.8-1002.5) / 50000)=arctan(-0.000014)≈-0.0008 degrees. Calculate the tilt deviation in the Y direction. =arctan((1003.2-1002.5) / 60000)=arctan(0.0000117)≈+0.00067 degrees. These calculated deviation values, along with their sign information, are transmitted in real time to the central coordinating controller as the basis for adjusting the Stewart platform's Z-axis translation and rotation around the X and Y axes. Based on this data, the controller generates corresponding control commands to drive the piezoelectric actuators to perform compensatory displacements, such as lowering the platform by 2.5 micrometers, and fine-tuning the lengths of relevant outriggers to correct angular deviations of -0.0008 degrees and +0.00067 degrees.

[0065] This invention calculates height and tilt deviation in real time using a specific algorithm built into the local laser ranging unit, providing direct and accurate feedback to the closed-loop control system. The method of calculating the average height effectively smooths out potential random measurement errors at individual measuring points, improving the reliability of overall height judgment. By utilizing simple trigonometric functions to convert height differences into angular deviations, the quantification of spatial attitude becomes intuitive and computationally efficient, meeting the requirements of real-time control. This systematic deviation calculation process transforms raw sensor readings into parameters with clear physical meaning and control guidance value, greatly improving the accuracy and efficiency of the alignment process and ensuring that the wafer cassette can quickly and stably reach the ideal docking posture.

[0066] In one possible implementation, the feature point recognition unit is configured to calculate the fine translational deviations ΔX_f and ΔY_f, the calculation of which includes the following steps: Step 1: Extract the pixel coordinates of feature points from the image and convert to current world coordinates. :

[0067]

[0068] in Pixel equivalent, unit: millimeters per pixel. From camera calibration parameters, The center pixel coordinates of the image. From camera calibration parameters, Image processing results from the feature point recognition unit; Step 2: Calculate the translation deviation and :

[0069]

[0070] in The world coordinates of the feature point at the target location. From preset target coordinates.

[0071] Among them, fine translation deviation This can refer to the minute horizontal displacement difference between the current X-coordinate of a feature point and the target X-coordinate, used for fine-tuning the X-axis direction during the final alignment stage. Fine translation deviation. This can refer to the tiny horizontal displacement difference between the current Y-coordinate of a feature point and the target Y-coordinate, enabling fine-tuning of the Y-axis direction during the final alignment stage. Pixel coordinates This can refer to the two-dimensional position of feature points in a digital image coordinate system, expressed in pixels, to provide location information in image space. Current world coordinates It can refer to the two-dimensional position coordinates of a feature point in real physical space after coordinate transformation, used to map image positions to actual spatial scale. Pixel equivalent 's' can refer to the actual physical size scale represented by a single pixel in the image, enabling conversion from pixel units to physical length units. Image center pixel coordinates This can refer to the pixel position of the point where the camera's optical axis intersects the image plane in the image coordinate system, serving as the reference origin for image coordinate transformations. The world coordinates of the target position. These can refer to the preset two-dimensional coordinates of feature points in real physical space under ideal docking conditions, serving as the final target values ​​for fine alignment. The preset target coordinates can refer to the ideal spatial position data of feature points determined in advance through precise measurement or design, used as a standard reference for deviation calculation.

[0072] As a concrete example: The feature point recognition unit acquires an image of a crosshair target on a wafer cell door. After image processing, the pixel coordinates of the center of the crosshair are accurately extracted as (u=1250.3, v=950.7). This unit calls pre-calibrated camera parameters, where the pixel equivalent s=0.005 mm / pixel, and the image center coordinates are ( =1280, =1024).

[0073] First, perform a coordinate transformation: =0.005×(1250.3-1280)=0.005×(-29.7)=-0.1485 mm, =0.005×(950.7-1024)=0.005×(-73.3)=-0.3665 mm. The preset target world coordinates of this feature point are known to be ( =0.000 mm, =0.000 mm). Next, calculate the fine translation deviation: =-0.1485-0.000=-0.1485 mm =-0.3665-0.000=-0.3665 mm. These deviation data indicate that the wafer cassette needs to be moved 0.1485 mm to the right in the X direction and 0.3665 mm forward in the Y direction. This data is transmitted in real time to the central co-controller, which generates corresponding micro-motion commands to drive the Stewart platform to perform sub-millimeter-level precise translation compensation to eliminate the final alignment error.

[0074] This invention achieves the highest level of positioning accuracy during docking by performing precise coordinate transformation and deviation calculation through a feature point recognition unit. The method of converting image pixel coordinates to physical world coordinates allows visual measurement results to be directly used for motion control in physical space, improving the system's control accuracy and efficiency. By calculating the difference between the current coordinates and the ideal target coordinates, the translational deviation directly guides the precise alignment operation, giving the adjustment process clear directionality and target specificity. This refined deviation acquisition method, complementing macroscopic positioning and height / tilt angle measurement, constitutes a complete multi-level positioning system, ensuring that the entire docking process ultimately achieves nanometer-level ultra-high precision requirements, significantly improving the wafer transfer accuracy and reliability of semiconductor manufacturing equipment.

[0075] In one possible implementation, the central coordinating controller is configured to complete a self-test procedure before the local environmental control module starts the particulate air filter and centrifugal fan.

[0076] The self-test procedure refers to a series of diagnostic and initialization processes that the controller automatically executes after the system is powered on, used to verify the status of each hardware module and the normality of software functions.

[0077] As a concrete example: After the system is powered on, the central coordinating controller immediately initiates its built-in self-test program. This program first checks whether the controller's CPU, memory, and storage units are functioning correctly; then, it sends handshake signals to the vision positioning module and the six-degree-of-freedom micro-adjustment module via the CAN bus to confirm a smooth communication link; next, it reads the status registers of each sensor to verify that the laser displacement sensor, industrial camera, etc., are online and free of fault codes; finally, it checks whether the relay status of the electromagnetic locking mechanism is normal. The entire self-test process takes approximately 500 milliseconds. Once all self-test items pass, the controller's status indicator turns green and internally displays a "self-test complete" mark. Only then does the central coordinating controller send a start command to the local environmental control module, energizing the particulate air filter and centrifugal fan to begin operation. If any abnormality is detected during the self-test, such as a laser sensor not responding, the controller will stop the subsequent processes and display the specific fault code and handling suggestions through the human-machine interface.

[0078] By setting the self-test procedure as the first step in system startup and completing it before the environmental control module starts, this timing ensures that all critical functional units of the system are in a known good condition before being put into formal operation. This design effectively prevents equipment malfunctions or performance degradation caused by local faults, improving the reliability and safety of the entire docking system. By using the logic of starting the self-test before starting peripheral devices, the risks that may arise from starting high-power equipment (such as centrifugal fans) when the system status is unknown are avoided. At the same time, it lays a stable working foundation for subsequent accurate pose measurement and adjustment, thus ensuring the overall safety and success rate of the wafer cassette docking process.

[0079] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0080] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0081] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A wafer cassette docking system, characterized in that, include: The central coordination controller is configured to perform system self-tests and coordination control; The local environmental control module, connected to the central co-controller, includes a particulate air filter and a centrifugal fan. It is configured to supply filtered, temperature- and humidity-controlled dry air or nitrogen to the wafer cassette docking area and form a stable, vertically downward laminar air curtain with a pressure slightly higher than atmospheric pressure. The visual positioning module, connected to the central collaborative controller, includes a global wide-angle vision unit, a local laser ranging unit, and a feature point recognition unit. The global wide-angle vision unit is configured to capture images of the wafer cassette's outline and bottom guide groove and calculate macroscopic deviations in the horizontal plane and rotation angle. The local laser ranging unit is configured to measure the vertical distance between the wafer cassette's bottom surface and the load port platform reference plane, as well as the sidewall tilt angle. The feature point recognition unit is configured to image specific marks on the wafer cassette door to obtain feature point position information. The central collaborative controller is configured to fuse data from the global wide-angle vision unit, the local laser ranging unit, and the feature point recognition unit to generate a six-degree-of-freedom pose deviation matrix. The six-degree-of-freedom micro-motion adjustment module, connected to the central coordinating controller, includes a parallel piezoelectric ceramic driving platform and an electromagnetic locking mechanism. The piezoelectric ceramic driving platform is configured to perform nanometer-level precision micro-displacement and micro-angle adjustments based on the six-degree-of-freedom pose deviation matrix. The electromagnetic locking mechanism is configured to lock the wafer cassette after adjustment.

2. The wafer cassette docking system according to claim 1, characterized in that, The global wide-angle vision unit is configured to calculate the macroscopic deviation between the wafer cell and the target location on the X-axis, Y-axis, and θz-axis.

3. The wafer cassette docking system according to claim 1, characterized in that, The local laser ranging unit is configured to measure the distance between the bottom surface of the wafer cassette and the reference surface of the load port platform on the Z-axis and the tilt angle on the θx and θy axes.

4. The wafer cassette docking system according to claim 1, characterized in that, The feature point recognition unit is configured to image the QR code or cross target on the wafer box door.

5. The wafer cassette docking system according to claim 1, characterized in that, The piezoelectric ceramic drive platform is the Stewart platform.

6. The wafer cassette docking system according to claim 1, characterized in that, The electromagnetic locking mechanism is configured to generate magnetic force to lock the wafer cassette instantly upon reaching the target position.

7. The wafer cassette docking system according to claim 1, characterized in that, The local laser ranging unit and the feature point recognition unit are configured to provide real-time feedback, forming a closed-loop control with the central collaborative controller until the deviation of all degrees of freedom enters the tolerance range.

8. The wafer cassette docking system according to claim 1, characterized in that, The local laser ranging unit is configured to calculate height deviation. and tilt deviation , The calculation formula includes the following steps: Step 1: Calculate the average height of the three laser measurement points. : in The height value measured by the i-th laser displacement sensor, i=1,2,3, is... Real-time measurement data from the local laser ranging unit; Step 2: Calculate the height deviation ΔZ: in For the target height value, the Preset values ​​from the load port platform reference plane; Step 3: Calculate the tilt deviation in the X direction : in The mounting spacing between the first and second laser sensors in the X direction, the Derived from sensor installation design values; Step 4: Calculate the tilt deviation in the Y direction : in The mounting spacing between the first and third laser sensors in the Y direction, the From sensor installation design values.

9. The wafer cassette docking system according to claim 1, characterized in that, The feature point recognition unit is configured to calculate the fine translation deviations ΔX_f and ΔY_f, and the calculation formula includes the following steps: Step 1: Extract the pixel coordinates of feature points from the image and convert to current world coordinates. : in In pixel equivalent, the unit is millimeters per pixel. From camera calibration parameters, The coordinates of the center pixel of the image, From camera calibration parameters, the Image processing results from the feature point recognition unit; Step 2: Calculate the translation deviation and : in The world coordinates of the feature point at the target location, the From preset target coordinates.

10. The wafer cassette docking system according to claim 1, characterized in that, The central coordinating controller is configured to complete a self-test procedure before the local environmental control module starts the particulate air filter and centrifugal fan.

Images