Wafer pre-alignment device and pre-alignment method

Abstract

The application relates to the technical field of semiconductor manufacturing, in particular to a wafer pre-alignment device and a pre-alignment method. The device comprises a mounting platform, an XY motion platform, a rotating platform, a suction cup and an edge detection assembly. The method comprises device correction, coarse scanning and center positioning, fine scanning and gap accurate positioning and alignment execution. Through the cooperation of software and hardware, the system error is eliminated by online self-correction, and the random error and model error are inhibited by high-order fitting and feature screening algorithm, so that the repeatability positioning accuracy, gap orientation accuracy and consistency of the device are improved, and the high-precision requirement of the wafer transmission system under the advanced process can be met.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing technology, and more specifically to a wafer pre-alignment device and pre-alignment method. Background Technology

[0002] In the semiconductor integrated circuit manufacturing process, the wafer pre-alignment device, as a core component of the transmission system, primarily functions to center the wafer and orient the notch, correcting wafer center offset and notch deflection. As semiconductor process technology continues to shrink, the requirements for wafer pre-alignment accuracy are increasing.

[0003] The pre-alignment device involves multiple components such as a motion platform, a rotating shaft system, and sensors. Due to limitations in the machining accuracy of the parts and the level of manual assembly, there is often an axial deviation and distance error between the detection center of the sensor and the physical center of the rotating platform. Although the error can be reduced through high-precision mechanical calibration, this is not only costly and time-consuming, but also makes it difficult to ensure the consistency of each device.

[0004] In terms of center positioning, existing technologies mostly use conventional algorithms such as least squares circle fitting. These algorithms have limited fitting accuracy when there is a large eccentricity in the wafer or when there is noise interference in the data, making it difficult to meet the requirements of high precision. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a wafer pre-alignment device and method that can effectively compensate for hardware installation errors and possess high-precision data processing capabilities.

[0006] This invention is achieved through the following technical solution:

[0007] A pre-alignment method for a wafer pre-alignment apparatus, based on a wafer pre-alignment apparatus including an XY motion platform, a rotating platform, a chuck, and an edge detection assembly, the pre-alignment method comprising:

[0008] Device calibration: Control the movement of the XY motion platform and the rotation platform to enable the edge detection component to detect data at a specific position, calculate and compensate for the axial deviation and distance deviation between the detection center of the edge detection component and the rotation center of the rotation platform to obtain device calibration parameters; drive the XY motion platform to complete error correction;

[0009] Coarse scanning and center positioning: Control the suction cup to adsorb the wafer and rotate it one revolution, and collect the edge data of the first wafer through the edge detection component; Based on the first wafer edge data and the device correction parameters, calculate the center coordinates of the wafer and the preliminary orientation angle of the notch;

[0010] Fine scanning and precise notch positioning: The rotating platform is controlled to rotate the wafer to a preset range of the initial orientation angle, and local fine sampling is performed to collect second wafer edge data; the precise notch orientation angle is calculated based on the second wafer edge data;

[0011] Alignment execution: Based on the center coordinates of the wafer and the precise notch orientation angle, the rotating platform is controlled to rotate to achieve orientation, and the XY motion platform is controlled to translate to compensate for eccentricity, thus completing the wafer pre-alignment.

[0012] Optionally, the device calibration includes axis deviation correction and distance deviation correction; the device calibration parameters include the axis intersection deviation value between the edge detection component and the rotation center of the rotating platform and the actual distance between the detection origin of the edge detection component and the rotation center of the rotating platform.

[0013] The method for correcting axis deviation includes the following steps:

[0014] The XY motion platform is controlled to move the suction cup to the X-axis limit position, so that the detection beam of the edge detection component covers the edge of the suction cup;

[0015] The XY motion platform is controlled to move stepwise along the Y-axis, while the edge detection component collects data from multiple suction cup edge points.

[0016] An error compensation algorithm is used to fit and calculate the edge point data of the suction cup to obtain the axis intersection deviation value between the edge detection component and the rotation center of the rotating platform;

[0017] The distance deviation correction method includes the following steps:

[0018] A standard wafer of known size is placed on the suction cup, and the rotating platform is controlled to rotate the standard wafer. The edge detection component collects wafer edge data during the rotation process.

[0019] Based on the radius of the standard wafer, the preset detection length of the edge detection component, and the average value of the collected wafer edge data, the actual distance between the detection origin of the edge detection component and the rotation center of the rotating platform is calculated.

[0020] Optionally, the method for coarse scanning and center positioning includes the following steps:

[0021] The edge data of the first wafer is calculated using a one-dimensional difference algorithm to filter out the positions of the maximum and minimum difference values.

[0022] If the absolute value of the difference between the maximum difference position and the minimum difference position is less than a first preset threshold, then the average value of the angle corresponding to the maximum difference position and the angle corresponding to the minimum difference position is determined as the preliminary orientation angle of the notch; otherwise, the first wafer edge data is re-acquired.

[0023] Data within the range between the maximum and minimum difference values ​​is removed, and the remaining data is taken as valid edge data; then, the proportional difference algorithm is used to filter out noise from the valid edge data.

[0024] The optimized trigonometric series circle fitting algorithm is used to fit the processed effective edge data to calculate the center coordinates of the wafer.

[0025] Optionally, the method for fine scanning and precise notch localization includes the following steps:

[0026] The rotating platform is controlled to rotate the wafer to a preset starting position of the initial orientation angle, and rotates to a preset ending position with a preset subdivision step size. The edge detection component collects the edge data of the second wafer.

[0027] The edge data of the second wafer is processed using a differential algorithm to determine the range of the notch data;

[0028] The random sample consensus algorithm is used to iteratively filter the points within the gap data range, eliminating straight line segment data and retaining the gap arc segment data;

[0029] The least squares circle fitting algorithm is used to fit the arc segment data of the notch to obtain the coordinates of the notch fitting center.

[0030] The precise notch orientation angle of the wafer is calculated based on the center coordinates of the wafer and the notch fitting center coordinates.

[0031] Optionally, the alignment process includes the following steps:

[0032] Calculate the angle between the line connecting the notch fitting center coordinates and the wafer center coordinates and the preset coordinate axis, and determine the target rotation angle of the rotating platform based on the angle.

[0033] Control the rotating platform to rotate according to the target rotation angle to orient the wafer notch;

[0034] Using a rotation matrix algorithm, the new center coordinates of the wafer after rotation and orientation are calculated based on the target rotation angle and the wafer center coordinates before rotation.

[0035] The XY motion platform is controlled to perform reverse movement compensation based on the new center coordinates, so that the center of the wafer coincides with the rotation center of the rotating platform.

[0036] Alternatively, methods for fitting using an optimized trigonometric series circle fitting algorithm include:

[0037] Construct a trigonometric series fitting equation containing a second harmonic term: ,in, The distance value in the edge detection data. For the corresponding rotation angle, The fitting coefficients are to be determined.

[0038] The least squares method is used to solve for the fitting coefficients in the fitted equation, and the calculated coefficients are then used to solve for the fitting coefficients. The coordinates of the wafer's center are determined.

[0039] Alternatively, methods for data processing using difference algorithms include:

[0040] Calculate the difference between adjacent data points in the edge data of the second wafer;

[0041] Calculate the average of the absolute values ​​of all the differences, and set a filtering threshold based on the average value;

[0042] Traverse the difference values ​​and find two locations where the difference value is greater than the filtering threshold, and determine them as the start and end points of the gap, respectively;

[0043] The data between the starting endpoint and the ending endpoint is defined as the gap data range;

[0044] Methods for data filtering using random sample consensus algorithms include:

[0045] A subset is constructed by randomly selecting three points from the data points within the gap data range;

[0046] Calculate the temporary center and temporary radius based on the three points in the subset, and establish a temporary circle model;

[0047] Traverse other data points within the gap data range, calculate the distance from each point to the temporary circle model, and if the distance is less than a preset error threshold, mark the point as an inner point; otherwise, mark it as an outer point and count the number of inner points.

[0048] Repeat the process until the preset number of iterations is reached;

[0049] Select the set of interior points with the largest number of interior points, and determine all data points in this set as the arc segment data of the notch, and remove straight line segments that do not conform to the characteristics of the circle model.

[0050] A wafer pre-alignment apparatus for implementing the pre-alignment method as described above, comprising a mounting platform;

[0051] The XY motion platform is mounted on the mounting platform and is used to move the wafer pre-alignment device in the X and Y axis directions in the horizontal plane.

[0052] The rotating platform is mounted on the XY motion platform and is used to rotate about its vertical rotation center in the horizontal plane.

[0053] The suction cup is disposed on the rotating platform and is used to carry and adsorb the wafer;

[0054] The edge detection component is disposed on the mounting platform, and the detection end extends to the edge detection area of ​​the suction cup for detecting the edge position data of the wafer.

[0055] Optionally, the edge detection component includes:

[0056] A bracket, which is fixed to the mounting platform;

[0057] A laser micrometer is mounted on the bracket.

[0058] The detection point of the laser micrometer is located above the edge of the chuck, and is used to detect the edge morphology data of the wafer as the wafer rotates with the chuck.

[0059] Optionally, both the rotating platform and the XY motion platform have a through hole in the middle.

[0060] The bottom of the suction cup is connected to a pneumatic rotary joint, and the upper surface of the suction cup is provided with an air channel. The air channel is connected to the pneumatic rotary joint, and the pneumatic rotary joint is connected to an air pipe. The air pipe passes through the through hole of the rotating platform and the through hole of the XY motion platform in sequence and extends to the outside of the device.

[0061] Compared with the prior art, the present invention has the following features and beneficial effects:

[0062] This invention achieves multi-degree-of-freedom adjustment of the suction cup through the XY motion platform and rotation platform of the pre-alignment device, which solves the alignment accuracy bottleneck caused by errors in component processing and manual assembly in the prior art. Thus, it realizes high-precision self-testing and parameter compensation of the device without the need to introduce additional precision calibration fixtures.

[0063] This invention applies an optimized trigonometric series circle fitting algorithm to overcome the fundamental error of conventional least squares circle fitting algorithms when there is significant eccentricity or data noise in the wafer. Furthermore, it employs a secondary sampling strategy that combines coarse and fine scanning, and uses the Random Sampling Consensus Algorithm (RANSAC) to iteratively filter the gap data after fine sampling, eliminating interference from linear data.

[0064] This invention improves the repeatability, notch orientation accuracy, and equipment consistency of the wafer pre-alignment device by using online self-calibration to eliminate system errors through hardware and software collaboration, and by using high-order fitting and feature selection algorithms to suppress random and model errors. It can meet the high precision requirements of wafer transfer systems under advanced processes. Attached Figure Description

[0065] The accompanying drawings illustrate exemplary embodiments of the present invention and, together with the description thereof, serve to explain the principles of the invention. These drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, but do not constitute a limitation on the embodiments of the present invention.

[0066] Figure 1 This is a schematic diagram of the overall mechanical structure of the wafer pre-alignment device provided in an embodiment of the present invention.

[0067] Figure 2 This is a schematic diagram of the calibration model for correcting the intersection deviation of the device axes in an embodiment of the present invention.

[0068] Figure 3 This is a schematic diagram of the data acquisition results of the suction cup edge points in an embodiment of the present invention.

[0069] Figure 4 This is a schematic diagram of the simulation results of the Y-direction fitting error in an embodiment of the present invention.

[0070] Figure 5 This is a schematic diagram of the laser micrometer distance correction model in an embodiment of the present invention.

[0071] Figure 6 This is a schematic diagram illustrating the detection principle of the edge detection method in this embodiment of the invention.

[0072] Figure 7 This is a schematic diagram of the polar coordinate curve of wafer edge morphology data collected in one go in an embodiment of the present invention.

[0073] Figure 8 This is a partial structural diagram of the wafer notch in an embodiment of the present invention.

[0074] Figure 9 This is a schematic diagram of the difference calculation results of the notch morphology data in an embodiment of the present invention.

[0075] Figure 10 This is a schematic diagram of the notch arc segment data after filtering using the RANSAC algorithm in an embodiment of the present invention.

[0076] Figure 11 This is a schematic diagram of the calculation model and coordinate system for the notch rotation angle in an embodiment of the present invention.

[0077] Figure 12This is a schematic diagram of the differential calculation results of wafer edge data in an embodiment of the present invention.

[0078] Figure 13 This is a schematic diagram of the data results from fine sampling of the notch in an embodiment of the present invention.

[0079] Figure 14 This is a comparison chart of the standard deviation of the circle center fitting error between the optimized trigonometric series circle fitting method of the present invention and the existing algorithm in an embodiment of the present invention.

[0080] Figure 15 These are partial images of both sides of the calibrated wafer in an embodiment of the present invention.

[0081] Reference numerals: 1-Mounting platform; 2-XY motion platform; 3-Rotating platform; 4-Suction cup; 5-Bracket; 6-Edge detection component. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0083] It should also be noted that, for ease of description, only the parts relevant to the present invention are shown in the accompanying drawings.

[0084] Where there is no conflict, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0085] Example 1

[0086] like Figure 1 As shown, this embodiment provides a wafer pre-alignment device, which mainly consists of a fixed base (mounting platform 1), a stacked platform (XY motion platform 2 and rotation platform 3) that carries and drives the wafer movement, and an edge detection component 6 that is independent of the motion component.

[0087] The above structure will be explained in detail:

[0088] Mounting platform 1 serves as the fixed base for the entire device, providing a stable support reference.

[0089] The XY motion platform 2 is mounted on the mounting platform 1 and is used to move the wafer pre-alignment device in the X and Y directions in the horizontal plane. The XY motion platform 2 refers to a mechanical module that can move linearly in the X and Y directions in the ground plane (horizontal plane), and is usually driven by guide rails, lead screws or linear motors.

[0090] The rotating platform 3 is mounted on the XY motion platform 2 and is used to rotate in the horizontal plane around its vertical rotation center. The rotating platform 3 can be translated as a whole with the XY motion platform 2 to provide the power for rotation around its vertical central axis.

[0091] The suction cup 4 is mounted on the rotating platform 3 to support and adsorb the wafer.

[0092] The XY platform supports the stacked structure of the rotating platform 3, enabling the uppermost suction cup 4 to simultaneously perform planar movement (X / Y axes) and rotation (…). (Axis) Three degrees of freedom.

[0093] Edge detection component 6 is mounted on mounting platform 1, and its detection end extends to the edge detection area of ​​suction cup 4 to detect the edge position data of the wafer. Edge detection component 6 is independently mounted on mounting platform 1 and does not move with XY motion platform 2 or rotation platform 3. Its detection end extends to the edge detection area of ​​suction cup 4 (i.e., the trajectory area traversed by the wafer edge).

[0094] An optional embodiment is provided, in which the edge detection component 6 includes: a bracket 5 and a laser micrometer, the bracket 5 being fixed on the mounting platform 1; the laser micrometer being mounted on the bracket 5; the detection point of the laser micrometer being located above the edge of the chuck 4, and being used to detect the edge morphology data of the wafer as the wafer rotates with the chuck 4.

[0095] A laser micrometer is a high-precision, non-contact measuring instrument that uses the principle of laser beam scanning or blocking to measure the position or size of an object's edge. The detection point of the laser micrometer is precisely adjusted above the edge of the chuck 4. As the chuck 4 holds the wafer and rotates, the laser micrometer can continuously detect and record the positional changes of the wafer edge relative to the sensor, thereby acquiring edge morphology data of the wafer.

[0096] To address the air supply issue of the rotating platform under multi-degree-of-freedom motion, this embodiment employs a hollow wiring design. Both the rotating platform 3 and the XY motion platform 2 have interconnected through holes in their middle sections, forming a physical channel that runs through the entire platform.

[0097] The bottom of the suction cup 4 is connected to a pneumatic rotary joint, and the upper surface of the suction cup 4 is provided with an air passage. The air passage is connected to the pneumatic rotary joint, and the pneumatic rotary joint is connected to an air pipe. The air pipe passes through the through hole of the rotating platform 3 and the through hole of the XY motion platform 2 in sequence and extends to the outside of the device.

[0098] A pneumatic rotary joint is a connector specifically designed for transferring fluids (such as compressed air or vacuum) in rotating equipment. It allows one end of the connection to rotate with the suction cup 4, while the other end (the end connecting the air tube) remains relatively stationary or untangled. The air tube connecting the pneumatic rotary joint passes sequentially through the through holes of the rotating platform 3 and the XY motion platform 2, eventually extending to the outside of the device and connecting to the vacuum generator.

[0099] Example 2

[0100] This embodiment provides a pre-alignment method based on a wafer pre-alignment device. The method adopts the logic of "first correction, then detection, and then execution" and achieves high-precision wafer positioning and orientation through a hierarchical scanning strategy (coarse scanning and fine scanning).

[0101] Pre-alignment methods include:

[0102] S1. Device calibration: Control the movement of the XY motion platform 2 and the rotation platform 3 to enable the edge detection component 6 to detect data at a specific position, calculate and compensate for the axial deviation and distance deviation between the detection center of the edge detection component 6 and the rotation center of the rotation platform 3, and obtain device calibration parameters.

[0103] Before formally processing the wafer, the device parameters are calibrated first. The XY motion platform 2 and the rotating platform 3 are driven in a combined motion, enabling the edge detection component 6 to detect data at specific locations on the device (e.g., the edge of the suction cup 4 or the edge of a standard part). Based on the detected data, the system calculates the physical installation error between the detection center of the edge detection component 6 and the rotation center of the rotating platform 3, specifically including: axis deviation (whether the detection axis passes through the rotation center) and distance deviation (the actual distance from the detection origin to the rotation center). By calculating and compensating for these deviations, the device calibration parameters are obtained. Subsequently, the XY motion platform 2 is driven to correct the error.

[0104] S2. Coarse scanning and center positioning: Control the suction cup 4 to adsorb the wafer and rotate it one revolution, and collect the edge data of the first wafer through the edge detection component 6; Based on the edge data of the first wafer and the device correction parameters, calculate the center coordinates of the wafer and the preliminary orientation angle of the notch.

[0105] After calibration, the first stage of global data acquisition is performed. The suction cup 4 is controlled to pick up the wafer and rotate a full circle (360 degrees) with the rotating platform 3. During this process, the edge detection component 6 continuously collects the position information of the wafer edge to form the first wafer edge data. Using the first wafer edge data, combined with the device calibration parameters obtained in step S1, the current center coordinates of the wafer and the preliminary orientation angle of the wafer notch (i.e., the approximate position of the notch) are calculated.

[0106] S3. Fine scanning and precise notch positioning: Control the rotating platform 3 to rotate the wafer to a preset range of the initial orientation angle, perform local fine sampling to collect the edge data of the second wafer; calculate the precise notch orientation angle based on the edge data of the second wafer.

[0107] To improve the accuracy of notch orientation, a second stage of high-precision local data acquisition is performed. The rotating platform 3 is controlled to rotate the wafer to a preset range near the initial orientation angle calculated in step S2. Within this range, local fine sampling is performed, i.e., edge data within this region is acquired at higher resolution to obtain second wafer edge data. Based on the second wafer edge data, the precise notch orientation angle of the wafer is calculated.

[0108] S4. Alignment execution: Based on the wafer's center coordinates and precise notch orientation angle, control the rotation of the rotating platform 3 to achieve orientation, and control the translation of the XY motion platform 2 to compensate for eccentricity, thus completing the wafer pre-alignment.

[0109] Based on the precise data obtained from the above steps, the final mechanical alignment is performed. Motion control commands are generated according to the calculated wafer center coordinates and precise notch orientation angle. The rotating platform 3 is controlled to rotate by a specific angle, aligning the wafer notch to the target direction, thus achieving orientation. The XY motion platform 2 is controlled to perform translational motion to compensate for the eccentricity of the wafer center relative to the rotation center, ensuring that the wafer center coincides with the preset center, thereby completing the wafer pre-alignment.

[0110] If edge data is collected only once, the number of data points in the gap area is approximately 1 / 360th of the total edge area data. When 10,000 edge topography data points are collected, the gap area contains only about 27 points (e.g., ...). Figure 7 As shown in the figure, it is difficult to fit the gap with high precision, and the 9973 points contained in the edge part far exceed the requirements of the circle center positioning algorithm, resulting in a surge in computing power demand. In order to solve the contradiction between the supply and demand of gap points and edge points when collecting edge shape data in one go, this embodiment solves the problem by means of secondary sampling, that is, through steps S2 and S3.

[0111] During pre-alignment, morphological data is first collected from the wafer edge, with enough data points to simultaneously meet the requirements of the centering algorithm and for preliminary identification of the notch location. Subsequently, fine sampling is performed on the notch location to obtain a sufficient number of notch data points. The results of the fine notch sampling are shown below. Figure 13 As shown.

[0112] Example 3

[0113] This embodiment provides a supplementary explanation of step S1 in embodiment two.

[0114] This embodiment mainly addresses the geometric error problem that arises after hardware assembly or long-term operation of the device. The device calibration includes axis deviation correction and distance deviation correction. That is, by using an object with known geometric features (the edge of the suction cup 4 or a standard wafer), data is collected through the motion trajectory (XY translational scanning or rotational scanning), and the relative positional relationship between the edge detection component 6 and the center of the rotating platform 3 is deduced, thereby obtaining the device calibration parameters.

[0115] The device calibration parameters include the axis intersection deviation between the edge detection component 6 and the rotation center of the rotating platform 3, and the actual distance between the detection origin of the edge detection component 6 and the rotation center of the rotating platform 3.

[0116] S11. Axis deviation correction is used to correct the problem that the detection beam axis of the edge detection component 6 fails to pass through the rotation center of the rotating platform 3, including the following steps:

[0117] The XY motion platform 2 is controlled to move the suction cup 4 to the X-axis limit position, so that the detection beam of the edge detection component 6 covers the edge of the suction cup 4;

[0118] The XY motion platform 2 is controlled to move stepwise along the Y-axis (i.e., intermittent small displacement), while the edge detection component 6 collects edge point data of multiple suction cups 4.

[0119] An error compensation algorithm is used to fit and calculate the edge point data of suction cup 4. By analyzing the geometric curve features of the edge of suction cup 4, the axis intersection deviation value between the edge detection component 6 and the rotation center of the rotating platform 3 is obtained.

[0120] S12. Distance deviation correction is used to calibrate the physical distance from the sensor's detection origin to the rotation center in order to eliminate installation position errors. It includes the following steps:

[0121] A standard wafer of known size is placed on the suction cup 4, and the rotating platform 3 is controlled to rotate the standard wafer. The edge detection component 6 collects the wafer edge data during the rotation process.

[0122] Based on the radius of the standard wafer, the preset detection length of the edge detection component 6 (i.e., the inherent range of the sensor or the designed installation distance), and the average value of the collected wafer edge data, the actual distance between the detection origin of the edge detection component 6 and the rotation center of the rotating platform 3 is calculated using geometric relationships.

[0123] The wafer pre-alignment device suffers from numerous errors in component processing, assembly, and edge detection, affecting pre-alignment accuracy. Among these factors, the intersection deviation between the edge acquisition sensor's axis and the C-axis of the motion platform, as well as the distance error from the sensor to the C-axis, directly impact the accuracy of the acquired data and the fitting precision. This section provides an example for steps S11 and S12.

[0124] (1) A specific example of step S11.

[0125] To minimize the misalignment between the edge acquisition sensor's axis and the C-axis of the motion platform, wafer pre-alignment devices typically employ precise installation process control and manufacturing error control. However, in actual installation, as the number of components increases, errors gradually accumulate. To achieve higher measurement accuracy, methods such as reducing the number of stacked components, improving installation precision, or performing individual calibration on each device can be employed.

[0126] Currently, pre-alignment devices have been simplified as much as possible, but it is still difficult to guarantee installation accuracy. Furthermore, installation accuracy is closely related to the experience and skills of the installers, making it difficult to ensure consistency across all devices. Therefore, when equipment allows, performing an individual calibration on each device and calculating the deviation is an efficient and low-cost solution to ensure consistent accuracy across all devices.

[0127] During calibration, the suction cup 4 is clearly offset above the laser micrometer, and the relative position of the device's rotation center to the laser micrometer is as follows: Figure 2 As shown in the figure, The center of suction cup 4, For the axis of the laser micrometer and The intersection of the perpendicular lines to the point. Click The distance between the points is , The length to the laser micrometer is The distance from the laser micrometer to the suction cup 4 is The radius of suction cup 4 is When the motion platform moves the suction cup 4 along the positive Y-axis... At that time, the current offset distance Step-by-step acquisition of 40 edge points, such as Figure 3 As shown. Among them, This represents the actual distance reading detected by the laser micrometer. This represents the step distance that the XY motion platform moves along the Y-axis.

[0128] Using the initial offset position as Then, any position of suction cup 4 along the Y-axis can be represented as: The diameter of suction cup 4 is 90mm, and the length of the motion platform along the Y-axis is 20mm.

[0129] The error compensation algorithm used in this example is a four-parameter constraint fitting algorithm for small circular arcs, which solves the inequality constraint problem by introducing the generalized Lagrange multiplier method.

[0130] First, define a circle equation with four parameters: Transform the four-parameter model into: At this moment, the center of the circle and radius It can be represented as: .

[0131] When collected sampling points The objective function for minimizing the four-parameter circle equation is: In addition, parameters The following also needs to be followed: ,parameter These are undetermined coefficients in the algebraic circle equation, and have no direct geometric meaning.

[0132] Therefore, the fitting problem can be transformed into an optimization problem: Objective equation Rewritten in matrix form, we have ,in, ,in, The original data matrix, This is the simplified scatter matrix (real symmetric matrix).

[0133] Since matrix multiplication is independent of the order of multiplication, the above equation can be simplified to: ,in, .

[0134] The constraints of the optimization problem can be rewritten as follows: ,in, , This is the constraint matrix, a specific constant matrix used to ensure that the fitted curve is a circle.

[0135] The objective function can be defined as: And define a Lagrange function: , It is a Lagrange multiplier.

[0136] The final objective equation is obtained: This is the problem of finding the maximum and minimum of the Lagrange function, which can also be called the primal problem. Based on the duality of the Lagrange function, the dual problem of the primal problem is: .

[0137] By analyzing the dual problem, the solution to the primal problem can be obtained. Taking the derivative, we have: ,in, It is a real symmetric matrix. It is a generalized eigenvector.

[0138] Will Substitution Then we have: ,because objective function The minimum value corresponds to the smallest non-negative generalized eigenvalue. The eigenvector corresponding to the smallest non-negative generalized eigenvalue is selected as... The value of , the radius of the arc and the position of the center It can be done Obtained through calculation.

[0139] For example: Let the central angle be... If 48 edge points are collected and noise with a standard deviation of 3.5 micrometers is artificially added, then after 1000 repeated simulations, the fitting error in the Y direction is as follows: Figure 4 As shown, the standard deviation of the error is 0.0169 mm, which is much higher than the conventional installation accuracy after the platform rotation center is corrected.

[0140] (2) A specific example of step S12.

[0141] During orientation, the center of the notch needs to be connected to the center of the wafer, and the orientation angle needs to be calculated. Calculating the notch center requires accurately converting the notch data to the coordinate system of the wafer's fitting center. At this point, the distance between the laser micrometer and the center of the chuck is crucial. Typically, we use the design distance as... However, due to the influence of assembly errors and online platform calibration, the value of is subject to change. The deviation from the design distance exceeds ±0.5mm, which increases both the orientation and positioning errors.

[0142] As the wafer rotates around its center of rotation, the curve of the detection area will change depending on the position of the center. The area of ​​any detection area is: ,in, This represents the micro-displacement of the wafer during each inspection, but since the speed of the rotating platform 3 is constant, it can be approximated as a constant. The length of the wafer was detected using a laser micrometer; Let be a circle with radius equal to the length from the endpoint of the laser micrometer to the center of suction cup 4, and be a constant. The area of ​​the wafer is a constant.

[0143] Therefore, the area of ​​any detection region is a constant. If the center of the wafer coincides with the center of rotation, such as... Figure 5 As shown, then: ,at this time, Since the wafer dimensions and the laser micrometer detection distance are known, the distance from the rotation center to the right limit position of the laser micrometer can be calculated. for: ,in, This refers to the detectable length of the laser micrometer.

[0144] Taking a standard 200mm wafer as an example, it is placed on chuck 4 and rotated one revolution. The average value of the laser micrometer is calculated to be 13.256mm. The detection length of the laser micrometer is... Given a value of 28 mm, the distance from the origin of the laser micrometer to the center of rotation can be calculated to be 114.744 mm. This value will be used as a fixed value. The data is input into the calibration system to facilitate notch conversion and subsequent calculations.

[0145] The pre-alignment device uses calibrated parameters to perform repeatability positioning accuracy testing. During the experiment, the wafer is randomly placed on chuck 4 for pre-alignment, and images are acquired after pre-alignment. The experiment is repeated 20 times, and partial images of the wafer are obtained, such as... Figure 15 As shown. The left image represents a partial image of the wafer on the notch side, and the right image represents a partial image of the wafer away from the notch side. Since the offset of multiple image acquisitions is small, a fixed marker is selected in both the left and right images, and its pixel coordinates are obtained as corner points for determining the repeatability accuracy. The calculated pre-alignment accuracy of the device in the X direction is [value missing]. The pre-alignment accuracy in the Y direction is The notch orientation accuracy is .

[0146] Example 4

[0147] This embodiment further defines step S2, explaining how to quickly locate the approximate position of the notch using a one-dimensional difference algorithm, then remove the notch data, clean the remaining edge data using a proportional difference algorithm, and finally fit the cleaned data using an optimized triangular series model to obtain high-precision wafer center coordinates. The steps include:

[0148] A one-dimensional difference algorithm is used to calculate the edge data of the first wafer to identify the positions of the maximum and minimum differences. Due to the geometric abrupt change at the notch, significant extreme value jumps will appear in the difference results. The algorithm automatically selects the positions of the maximum values ​​(usually corresponding to one side of the notch edge) and the minimum values ​​(corresponding to the other side of the notch edge) in the difference data.

[0149] If the absolute value of the difference between the maximum and minimum difference positions is less than the first preset threshold (i.e., it meets the theoretical width range of the notch), then the average value of the angle corresponding to the maximum difference position and the angle corresponding to the minimum difference position is determined as the preliminary orientation angle of the notch; otherwise, the data is judged to be abnormal or unsuccessfully identified, and the first wafer edge data is re-acquired.

[0150] Data within the range between the maximum and minimum difference values ​​(i.e., data excluding the gap itself) is removed, and the remaining data is used as valid edge data. The proportional difference algorithm is then used to filter out noise from the valid edge data, i.e., to identify and filter out random noise or outliers in the data.

[0151] The optimized trigonometric series circle fitting algorithm is used to fit the processed effective edge data to calculate the center coordinates of the wafer.

[0152] Alternatively, methods for fitting using an optimized trigonometric series circle fitting algorithm include:

[0153] Construct a trigonometric series fitting equation containing a second harmonic term: ,in, The distance value in the edge detection data. For the corresponding rotation angle, The fitting coefficients are to be determined. The DC component coefficient, The first-order fundamental frequency coefficient, It is the second harmonic coefficient.

[0154] The least squares method is used to solve for the fitting coefficients in the fitted equation, and the calculated coefficients are then used to solve for the fitting coefficients. The coordinates of the wafer's center are determined.

[0155] Below is a specific example.

[0156] like Figure 6 The image shows the model during device testing. Figure 7 For the polar coordinate curves acquired by the device, any point can be represented as It exhibits a clear periodicity.

[0157] Let the center of the wafer be Point relative to the center of rotation The offset distance of the point is Then any edge point Distance to the center of rotation for: ,in, This is the eccentric phase angle.

[0158] After the wafers in the wafer cassette are placed into the pre-alignment device by the robotic arm, the center of the wafers shifts. Not exceeding 2.8mm, conventional algorithms will Transform into This transforms the circle fitting method from an algebraic fitting method into a geometric fitting method. Generally, geometric fitting methods with clear meaning are more accurate than algebraic fitting methods. However, the simplified algorithm described above discards some small quantities, making it difficult for its accuracy to surpass that of the least squares circle fitting method when no gross errors are added.

[0159] When fitting periodic functions using trigonometric series, we can approximate any given periodic function by continuously increasing the number of terms in the trigonometric series. Therefore, to improve the fitting accuracy at the wafer center and retain some discarded minute quantities, we can increase the number of terms to... Approximately To achieve a higher precision approximation, the equation is: .

[0160] Distance from the center of rotation to the laser micrometer and The difference and the distance detected by the laser micrometer Since they are approximately equal, the above equation can be expressed as: ,in, , To correct for small amounts of observation errors or noise, the center of the wafer... wafer radius .

[0161] All laser rangefinder measurements and corresponding The relationship can be represented as: ,in, , This is a matrix transpose operation, performed to make the laser measurement values ​​equal to... To achieve a high degree of fit, we must adjust the correction term. As small as possible. In fact, the correction term... It is exactly a matrix equation The error. To make the matrix equation To minimize the error, the solution vector can be obtained using the least squares approach. This minimizes the sum of squared errors on both sides of the matrix equation. Therefore, the matrix equation... The least squares solution is: .

[0162] Expand We can obtain: ,beg Compared to Taking the derivative of and setting the result to zero, we get: .

[0163] Through the above work, the optimized trigonometric series circle fitting method proposed in this invention achieves a significantly higher fitting accuracy compared to conventional algorithms. Figure 14As shown.

[0164] Example 5

[0165] This embodiment further defines and explains step S3. After determining the approximate location of the gap through coarse scanning, sufficient gap data is obtained through local high-density sampling. The gap region is extracted from the background data using a differential algorithm. A random sample consensus algorithm is introduced to intelligently remove interference from straight line segments from the gap data that mixes straight lines and arcs, retaining only the true arc segments for fitting, thereby obtaining extremely high orientation accuracy. This includes the following steps:

[0166] The control rotation platform 3 drives the wafer to rotate to a preset starting position of the initial orientation angle (e.g., initial angle -2.5°), and rotates to a preset ending position (e.g., initial angle +2.5°) with a preset subdivision step size. The edge data of the second wafer is collected by the edge detection component 6.

[0167] The edge data of the second wafer is processed using a differential algorithm to determine the range of the notch data;

[0168] The random sample consensus algorithm is used to iteratively filter the points within the gap data range, removing straight line segments and retaining the arc segments of the gap;

[0169] The least squares circle fitting algorithm is used to fit the arc segment data of the notch to obtain the coordinates of the notch fitting center.

[0170] Based on the wafer's center coordinates and the notch fitting center coordinates, combined with the previously obtained wafer center coordinates and the notch fitting center coordinates, the slope or angle of the line connecting the two points is calculated, and the precise notch orientation angle of the wafer is calculated.

[0171] Among them, the methods for data processing using the difference algorithm include:

[0172] Calculate the difference values ​​between adjacent data points in the second wafer edge data; the difference values ​​will change significantly at the positions where the notch enters and leaves.

[0173] Calculate the average of the absolute values ​​of all differences, and set a filtering threshold based on the average (e.g., 10 times the mean).

[0174] Traverse the difference values, find two locations where the difference value is greater than the filtering threshold, and determine them as the start and end points of the gap, respectively.

[0175] The data between the start and end endpoints is defined as the gap data range, and the flat arc data around the gap is removed.

[0176] Provide a specific example.

[0177] To improve fitting quality, identifying and removing outliers before fitting is essential. Typically, one-dimensional differencing algorithms are used to find outliers. When the data contains consecutive outliers, the differencing result only shows a change in amplitude when the outlier is detected; otherwise, it behaves the same as normal data, failing to effectively filter out most noise. When acquiring edge topography data, we obtain data including rotation angles. and the distance from the wafer edge to the laser range sensor Conventional difference algorithms ignore the rotation angle, leading to the aforementioned problems. This embodiment introduces the angle change during the difference calculation and designs a proportional difference algorithm to remove obvious noise data, such as... Figure 12 As shown.

[0178] The proportional difference algorithm first uses the first data Based on Calculate the difference at the next point: .

[0179] like If it is less than the threshold, then update. Calculate the difference at the next point. If If the value is greater than the threshold, remove the current traversal point from the list and do not update it. Calculate the difference for the next point directly. After traversing all points, use the points in the list directly for fitting.

[0180] Orientation refers to the process of determining the notch deflection angle using the line connecting the notch center and the wafer center. A local example of a wafer notch is shown below. Figure 8 As shown. The gap mainly consists of two straight line segments. , and circular arc segment .

[0181] To accurately filter the notched arc segments from the edge data, a two-step process is required: first, removing the flat arcs around the notch; second, filtering the notched arcs. When removing the flat arcs around the notch, the notch data is used... Calculate the difference of the notch morphology data as follows Figure 9 As shown, the difference values ​​fluctuate significantly when entering and leaving the gap. To avoid missing the gap, the mean difference value is calculated first. Then, the difference values ​​are traversed from both sides of the collected data. When the difference value is greater than a threshold, it is determined to be one end of the gap. After obtaining the two endpoints, the location of the gap can be determined. The threshold is selected based on the mean difference value and the data fluctuation range. In this example, the threshold is 10 times the mean value.

[0182] The gaps identified through differential filtering include straight segments and circular segments. The transition between straight and circular segments is smooth, making it difficult to accurately distinguish between them using only differential filtering. Therefore, this embodiment filters the data to further identify the circular segments, improving the accuracy of the gap fitting.

[0183] Methods for data filtering using the Random Sample Consensus (RANSAC) algorithm include:

[0184] A subset is constructed by randomly selecting three points from the data points within the gap data range; the three points determine a circle.

[0185] Calculate the temporary center and temporary radius based on the three points in the subset, and establish a temporary circle model;

[0186] Traverse other data points within the gap data range, calculate the distance from each point to the temporary circle model, and if the distance is less than the preset error threshold, mark the point as an inner point; otherwise, mark it as an outer point and count the number of inner points.

[0187] Repeat the process until the preset number of iterations is reached;

[0188] Select the set of interior points with the largest number of interior points, and determine all data points in this set as the arc segment data of the notched circle, and remove the straight line segment data that does not conform to the characteristics of the circle model.

[0189] Let's provide a specific example.

[0190] Although the collected gap data is in polar coordinates, the bottom of the gap is approximately circular. The RANSAC algorithm was used to analyze... The data undergoes a second round of filtering, and then the least squares circle fitting method is used to refine the filtered data. Perform a fitting operation. The specific steps of the algorithm are as follows:

[0191] (1) Randomly select from the original dataset S Each point is a subset Assuming subset All points within a point are interior points;

[0192] (2) Based on subsets Estimating parameter models;

[0193] (3) Traverse all data in S. If the data is within the given error range, mark it as an interior point; otherwise, mark it as an exterior point.

[0194] (4) All interior points form an interior point set. If the number of points in the interior point set is greater than the threshold, the parameters are re-estimated using all interior points in the interior point set, and the algorithm ends.

[0195] (5) If the number of internal points in the set is less than the threshold, repeat steps (1)-(4);

[0196] (6) After multiple iterations, select the set of interior points with the largest number of interior points, re-estimate the parameters using all the interior points in the set, and then end the algorithm.

[0197] To use subsets as much as possible When the number of interior points exceeds a threshold, we need to estimate the number of iterations. Let the proportion of interior points in the original dataset S be... In each selection of a subset of When there are n points, the probability that the selected point has at least one exterior point is 1. . The probability that the computation model samples at least one outlier in the next iteration is: The probability of sampling the correct model at this time : and obtain .

[0198] When fitting the gap, the circle equation is used as the fitting model, and the random numbers are those from the initial screening. Choose three points as a subset, the three points are , The equation of a circle can be expressed as: In the formula, Let the coordinates be the center of the circle. Using three random sampling points as the edge points of the circle, the circle parameters can be calculated as follows: ,in, , These are intermediate variables in the calculation process of defining a circle using three points. This refers to the coordinate difference in the calculation of a circle defined by three points.

[0199] After filtering the gap data, the obtained gap arc segments are as follows: Figure 10 As shown, the algorithm can quickly and effectively select the position of the notch arc, preparing for subsequent fitting.

[0200] Example 6

[0201] This embodiment further defines step S4. After the precise calculation of the wafer center and notch angle is completed, the device needs to adjust the wafer to a specified orientation (i.e., the center is aligned with the rotation center and the notch is aligned with the specified direction) through mechanical movement, including the following steps:

[0202] Calculate the angle between the line connecting the notch fitting center coordinates and the wafer center coordinates and the preset coordinate axis (e.g., the positive X-axis direction of the device), and determine the target rotation angle of the rotating platform 3 based on the angle; that is, determine the target rotation angle of the rotating platform 3 based on the difference between the actual deflection angle and the preset target alignment angle (usually 0 degrees or other specified angle).

[0203] The control platform 3 rotates according to the target rotation angle to orient the wafer notch;

[0204] Using the rotation matrix algorithm, the new center coordinates of the wafer after rotation and orientation are calculated based on the target rotation angle and the wafer center coordinates before rotation. The new center coordinates of the wafer after rotation and orientation are calculated by transforming the target rotation angle (i.e., the angle through which the rotation platform 3 rotates) and the wafer center coordinates before rotation using the rotation matrix.

[0205] The XY motion platform 2 is controlled to perform reverse movement compensation based on the new center coordinates, so that the center of the wafer coincides with the rotation center of the rotating platform 3. For example, if the new center is located at... Then the XY platform will... Directional movement. After the movement is completed, the geometric center of the wafer coincides with the rotation center of the rotating platform 3 in space, completing the entire wafer pre-alignment process.

[0206] Provide a specific example.

[0207] After obtaining the positions of the notch center and the wafer center relative to the XOY coordinate system, the wafer is rotated so that the line connecting the notch center and the wafer center is parallel to the X-axis. Then, the position of the wafer center relative to the XOY coordinate system after rotation is calculated using the rotation matrix, and it is moved in the opposite direction to complete the pre-alignment.

[0208] The quadrants where the wafer center and the notch center are located differ, leading to different methods for calculating the required rotation angle. Therefore, this embodiment establishes a coordinate system where the X-axis coincides with the line connecting the wafer center and the notch center to aid in determining the required rotation angle of the rotating platform 3. Figure 11 As shown in the figure, Center of the wafer The notch fitting center is the line connecting the wafer center and the notch fitting center with respect to the XOY coordinate system. Under different working conditions, the rotation angle of the turntable is .

[0209] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment / mode or example is included in at least one embodiment / mode or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0210] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0211] Those skilled in the art should understand that the above embodiments are merely for illustrating the present invention and are not intended to limit the scope of the invention. Those skilled in the art can make other changes or modifications based on the above invention, and these changes or modifications still fall within the scope of the present invention.

Claims

1. A pre-alignment method for a wafer pre-alignment apparatus, characterized in that, Based on a wafer pre-alignment apparatus, the wafer pre-alignment apparatus includes an XY motion platform (2), a rotation platform (3), a chuck (4), and an edge detection component (6), and the pre-alignment method includes: Device calibration: Control the movement of the XY motion platform (2) and the rotation platform (3) to make the edge detection component (6) detect data at a specific position, calculate and compensate for the axial deviation and distance deviation between the detection center of the edge detection component (6) and the rotation center of the rotation platform (3) to obtain device calibration parameters; drive the XY motion platform (2) to complete error correction; Coarse scanning and center positioning: Control the suction cup (4) to adsorb the wafer and rotate it one revolution, and collect the first wafer edge data through the edge detection component (6); Based on the first wafer edge data and the device correction parameters, calculate the center coordinates of the wafer and the preliminary orientation angle of the notch; Fine scanning and precise notch positioning: The rotating platform (3) is controlled to rotate the wafer to a preset range of the initial orientation angle, and local fine sampling is performed to collect the edge data of the second wafer; the precise notch orientation angle is calculated based on the edge data of the second wafer; Alignment execution: Based on the center coordinates of the wafer and the precise notch orientation angle, control the rotation platform (3) to rotate to achieve orientation, and control the XY motion platform (2) to translate to compensate for eccentricity, thus completing the wafer pre-alignment; The device calibration includes axis deviation calibration and distance deviation calibration; the device calibration parameters include the axis intersection deviation value between the edge detection component (6) and the rotation center of the rotating platform (3) and the actual distance between the detection origin of the edge detection component (6) and the rotation center of the rotating platform (3); The method for correcting axis deviation includes the following steps: The XY motion platform (2) is controlled to move the suction cup (4) to the X-axis limit position, so that the detection beam of the edge detection component (6) covers the edge of the suction cup (4); The XY motion platform (2) is controlled to move stepwise along the Y-axis, while the edge detection component (6) collects edge point data of multiple suction cups (4); The edge point data of the suction cup (4) is fitted and calculated using an error compensation algorithm to obtain the axis intersection deviation value between the edge detection component (6) and the rotation center of the rotating platform (3); The method for distance deviation correction includes the following steps: A standard wafer of known size is placed on the suction cup (4), and the rotating platform (3) is controlled to drive the standard wafer to rotate. The edge detection component (6) collects wafer edge data during the rotation process. Based on the radius of the standard wafer, the preset detection length of the edge detection component (6), and the average value of the collected wafer edge data, the actual distance between the detection origin of the edge detection component (6) and the rotation center of the rotating platform (3) is calculated.

2. The pre-alignment method of the wafer pre-alignment apparatus according to claim 1, characterized in that, The method for coarse scanning and center positioning includes the following steps: The edge data of the first wafer is calculated using a one-dimensional difference algorithm to filter out the positions of the maximum and minimum difference values. If the absolute value of the difference between the maximum difference position and the minimum difference position is less than a first preset threshold, then the average value of the angle corresponding to the maximum difference position and the angle corresponding to the minimum difference position is determined as the preliminary orientation angle of the notch; otherwise, the first wafer edge data is re-acquired. Data within the range between the maximum and minimum difference values ​​is removed, and the remaining data is taken as valid edge data; then, the proportional difference algorithm is used to filter out noise from the valid edge data. The optimized trigonometric series circle fitting algorithm is used to fit the processed effective edge data to calculate the center coordinates of the wafer.

3. The pre-alignment method of the wafer pre-alignment apparatus according to claim 1, characterized in that, The method for fine scanning and precise notch localization includes the following steps: The rotating platform (3) is controlled to rotate the wafer to the preset starting position of the initial orientation angle, and rotates to the preset ending position with a preset subdivision step, and the edge data of the second wafer is collected by the edge detection component (6); The edge data of the second wafer is processed using a differential algorithm to determine the range of the notch data; The random sample consensus algorithm is used to iteratively filter the points within the gap data range, eliminating straight line segment data and retaining the gap arc segment data; The least squares circle fitting algorithm is used to fit the arc segment data of the notch to obtain the coordinates of the notch fitting center. The precise notch orientation angle of the wafer is calculated based on the center coordinates of the wafer and the notch fitting center coordinates.

4. The pre-alignment method of the wafer pre-alignment apparatus according to claim 3, characterized in that, The alignment process includes the following steps: Calculate the angle between the line connecting the notch fitting center coordinates and the wafer center coordinates and the preset coordinate axis, and determine the target rotation angle of the rotating platform (3) based on the angle. Control the rotating platform (3) to rotate according to the target rotation angle to orient the wafer notch; Using a rotation matrix algorithm, the new center coordinates of the wafer after rotation and orientation are calculated based on the target rotation angle and the wafer center coordinates before rotation. The XY motion platform (2) is controlled to perform reverse movement compensation according to the new center coordinates, so that the center of the wafer coincides with the rotation center of the rotating platform (3).

5. The pre-alignment method of the wafer pre-alignment apparatus according to claim 2, characterized in that, Methods for fitting circles using optimized trigonometric series circle fitting algorithms include: Construct a trigonometric series fitting equation containing a second harmonic term: ,in, The distance value in the edge detection data. For the corresponding rotation angle, The fitting coefficients are to be determined. The least squares method is used to solve for the fitting coefficients in the fitted equation, and the calculated coefficients are then used to solve for the fitting coefficients. The coordinates of the wafer's center are determined.

6. The pre-alignment method of the wafer pre-alignment apparatus according to claim 3, characterized in that, Methods for data processing using difference algorithms include: Calculate the difference between adjacent data points in the edge data of the second wafer; Calculate the average of the absolute values ​​of all the differences, and set a filtering threshold based on the average value; Traverse the difference values ​​and find two locations where the difference value is greater than the filtering threshold, and determine them as the start and end points of the gap, respectively; The data between the starting endpoint and the ending endpoint is defined as the gap data range; Methods for data filtering using random sample consensus algorithms include: A subset is constructed by randomly selecting three points from the data points within the gap data range; Calculate the temporary center and temporary radius based on the three points in the subset, and establish a temporary circle model; Traverse other data points within the gap data range, calculate the distance from each point to the temporary circle model, and if the distance is less than a preset error threshold, mark it as an inner point; otherwise, mark it as an outer point and count the number of inner points. Repeat the process until the preset number of iterations is reached; Select the set of interior points with the largest number of interior points, and determine all data points in this set as the arc segment data of the notch, and remove straight line segments that do not conform to the characteristics of the circle model.

7. A wafer pre-alignment apparatus, characterized in that, For implementing the pre-alignment method as described in any one of claims 1-6, the apparatus includes a mounting platform (1). The XY motion platform (2) is mounted on the mounting platform (1) and is used to move the wafer pre-alignment device in the X and Y axis directions in the horizontal plane; The rotating platform (3) is mounted on the XY motion platform (2) and is used to rotate around its vertical rotation center in the horizontal plane; The suction cup (4) is disposed on the rotating platform (3) for carrying and adsorbing the wafer; The edge detection component (6) is disposed on the mounting platform (1), and the detection end extends to the edge detection area of ​​the suction cup (4) for detecting the edge position data of the wafer.

8. A wafer pre-alignment apparatus according to claim 7, characterized in that, The edge detection component (6) includes: The bracket (5) is fixed on the mounting platform (1); A laser micrometer is mounted on the bracket (5); The detection point of the laser micrometer is located above the edge of the chuck (4) and is used to detect the edge morphology data of the wafer when the wafer rotates with the chuck (4).

9. The wafer pre-alignment apparatus according to claim 7, characterized in that, Both the rotating platform (3) and the XY motion platform (2) have through holes in their middle parts; The bottom of the suction cup (4) is connected to a pneumatic rotary joint. An air passage is provided on the upper surface of the suction cup (4). The air passage is connected to the pneumatic rotary joint. The pneumatic rotary joint is connected to an air pipe. The air pipe passes through the through hole of the rotating platform (3) and the through hole of the XY motion platform (2) in sequence and extends to the outside of the device.

Images