Wafer edge finding method and device

Abstract

The invention discloses a wafer edge finding method and device, the method comprising: collecting an edge image of a wafer, the wafer being placed on a chuck table, the chuck table being movable along an X, Y and θ axis; obtaining arc feature points and notch feature points from the edge image; performing circle fitting on the arc feature points to obtain a first fitted circle; identifying and filtering out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; performing circle fitting on the remaining arc feature points again to obtain a second fitted circle; identifying and filtering out abnormal points in the remaining arc feature points based on the second fitted circle and a second preset threshold to obtain filtered arc feature points; determining the center deviation of the wafer based on the filtered arc feature points; moving the chuck table along one or more of the X, Y and θ axes to perform center compensation on the center deviation, during which the wafer and the chuck table are separated; determining the angular deviation of the wafer based on the notch feature points; rotating the chuck table around the center of the chuck table to perform rotation compensation on the angular deviation.

Description

Technical Field

[0001] The present invention relates to the field of semiconductors, and in particular to a wafer edge finding method and device. Background Art

[0002] The chuck table is an important tool used in semiconductor manufacturing, optical processing, flat panel display manufacturing, solar panel manufacturing, biomedicine and other fields. In the semiconductor field, the chuck table is used to clamp wafers, silicon wafers, films, etc. to ensure their stability and accuracy during processing. The chuck table is usually composed of a clamp, a positioner and an adjuster, and can clamp and position wafers, silicon wafers and films of various sizes, shapes and materials.

[0003] During semiconductor processing, when placing the wafer on the chuck table, problems such as inaccurate wafer positioning, center and notch direction deviation may occur. Therefore, the wafer needs to be edge-finded to adjust the wafer to the specified position and direction. Specifically, the wafer needs to be compensated for the center of the wafer so that the center of the wafer is aligned with the center of the chuck table, and the wafer needs to be rotated so that the notch on the edge of the wafer points to the specified direction.

[0004] The center compensation of the wafer involves circle fitting. The conventional circle fitting algorithm filters the collected wafer edge data (for example, the wafer edge image), and uses the least squares method to fit the center of the circle after the data processing. However, due to the acceleration and deceleration, signal fluctuations and noise during the wafer movement, the collected data may have abnormal values. At the same time, normal points may be eliminated during the filtering process, thereby increasing the fitting error. If the data is exactly on the fitted circle, the algorithm will find the correct circle. However, when the data is on a small arc, the algorithm will deviate and estimate a smaller circle than the correct circle. In addition, the least squares method is sensitive to outliers. If the filtering is not thorough, the fitting result will be biased. In addition, when the center of the wafer is compensated, the conventional compensation scheme directly moves the chuck table along the X / Y and θ axes for compensation after calculating the center deviation. This process will drive the wafer to move at the same time, which not only increases the complexity of the center compensation algorithm, but also brings certain errors to the compensation results.

[0005] In addition, when collecting images of the wafer edge, a transmissive sensor or a reflective sensor is generally used. However, a transmissive sensor cannot identify the metal-plated bond sheet, and thus cannot complete the pre-alignment compensation of the wafer. The reflective sensor has a low reflectivity for the transparent sheet and may not be able to detect it. If the sensitivity of the sensor is increased, the cost will increase. In addition, the reflective sensor cannot bear the + / -10mm eccentricity of the bond sheet, and the repeatability positioning accuracy cannot meet the requirements.

[0006] In view of this, it is desirable to provide an improved wafer edge finding method and device. Summary of the invention

[0007] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or critical elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0008] The invention provides a wafer edge finding method, comprising: collecting an edge image of a wafer, wherein the wafer is placed on a chuck table, and the chuck table can move along an X-axis, a Y-axis, and a θ-axis; obtaining arc feature points and notch feature points of the wafer edge from the edge image; performing circle fitting on the arc feature points to obtain a first fitted circle; identifying and filtering out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; performing circle fitting on the remaining arc feature points again to obtain a second fitted circle; identifying and filtering out abnormal points in the remaining arc feature points based on the second fitted circle and a second preset threshold to obtain filtered arc feature points; determining a center deviation of the wafer based on the filtered arc feature points; moving the chuck table along one or more of the X-axis, the Y-axis, and the θ-axis to perform center compensation on the center deviation, wherein the wafer and the chuck table are separated during the center compensation; determining an angular deviation of the wafer based on the notch feature points; and rotating the chuck table around the center of the chuck table to perform rotation compensation on the angular deviation.

[0009] In some embodiments, the circle fitting is performed by a Ransac algorithm, and the first preset threshold and the second preset threshold can be adjusted dynamically.

[0010] In some embodiments, determining the center deviation of the wafer based on the screened arc feature points further includes: fitting the screened arc feature points through a Pratt circle fitting algorithm to obtain the wafer center; and determining the deviation between the wafer center and the chuck table center as the center deviation.

[0011] In some embodiments, the wafer and the chuck table are separated by a Z-axis moving member, wherein the Z-axis moving member is located outside the chuck table and below the wafer and is capable of moving up and down along the Z-axis, and wherein, before the center point compensation begins, the Z-axis moving member moves upward along the Z-axis to separate the wafer and the chuck table.

[0012] In some embodiments, the Z-axis moving member includes two parallel rods respectively located on both sides of the chuck table, and the distance between the two parallel rods is greater than the diameter of the chuck table and smaller than the diameter of the wafer.

[0013] In some embodiments, after the center compensation is completed, the Z-axis moving part returns to its original position to allow the wafer to contact and be placed on the chuck stage.

[0014] In some embodiments, determining the angular deviation of the wafer based on the notch feature points further includes: obtaining a notch angle of the wafer based on the notch feature points; and determining a deviation between the notch angle and a specified angle as the angular deviation.

[0015] In some embodiments, the edge image of the wafer is obtained by an image acquisition unit, wherein the image acquisition unit acquires the image of the edge of the wafer when the chuck table is rotated.

[0016] In some embodiments, the image acquisition unit includes a camera located above the wafer and a back light source located below the wafer, and the method also includes: during the image acquisition process, detecting whether the wafer is transparent; and turning on the back light source only when it is detected that the wafer is transparent.

[0017] The present invention also provides a wafer edge-finding device, comprising: a chuck table, the chuck table is used to carry a wafer and can move along an X-axis, a Y-axis and a θ-axis; an image acquisition unit, which is configured to: acquire an edge image of the wafer; a feature point acquisition unit, which is configured to: obtain arc feature points and notch feature points of the wafer edge from the edge image; a feature point preprocessing unit, which is configured to: perform circle fitting on the arc feature points to obtain a first fitted circle; identify and filter out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; perform circle fitting again on the remaining arc feature points to obtain a second fitted circle; and Based on the second fitted circle and the second preset threshold, abnormal points in the remaining arc feature points are identified and screened out to obtain screened arc feature points; a center compensation unit is configured to: determine the center deviation of the wafer based on the screened arc feature points; move the chuck table along one or more of the X, axis, Y axis and θ axis to perform center compensation for the center deviation, wherein the wafer and the chuck table are separated during the center compensation; and a rotation compensation unit is configured to: determine the angular deviation of the wafer based on the notch feature points; and rotate the chuck table around the center of the chuck table to perform rotation compensation for the angular deviation.

[0018] In some embodiments, the image acquisition unit includes a camera located above the wafer and a back light source located below the wafer, and the image acquisition unit is further configured to: detect whether the wafer is transparent during image acquisition; and turn on the back light source only when it is detected that the wafer is transparent.

[0019] In some embodiments, the circle fitting is performed by a Ransac algorithm, and the first preset threshold and the second preset threshold can be adjusted dynamically.

[0020] In some embodiments, the center compensation unit is further configured to determine the center deviation of the wafer by: fitting the screened arc feature points through the Pratt circle fitting algorithm to obtain the wafer center; and determining the deviation between the wafer center and the chuck table center as the center deviation.

[0021] In some embodiments, the device further comprises a Z-axis moving element, which is located outside the chuck table and below the wafer and is capable of moving up and down along the Z-axis, and wherein, before the center compensation starts, the Z-axis moving element moves upward along the Z-axis to separate the wafer and the chuck table.

[0022] In some embodiments, the Z-axis moving member includes two parallel rods respectively located on both sides of the chuck table, and the distance between the two parallel rods is greater than the diameter of the chuck table and smaller than the diameter of the wafer.

[0023] In some embodiments, after the center compensation is completed, the Z-axis moving part returns to its original position to allow the wafer to contact and be placed on the chuck stage.

[0024] In some embodiments, the rotation compensation unit is further configured to determine the angular deviation of the wafer by: obtaining a notch angle of the wafer based on notch feature points; and determining a deviation between the notch angle and a specified angle as the angular deviation.

[0025] The technical solution of the present invention pre-processes the data to screen out abnormal points after collecting the wafer edge data, and then performs circle fitting on the retained normal points, thereby improving the accuracy of the fitting algorithm. When the chuck stage is moved to perform circle center compensation, the wafer and the chuck stage are separated by the Z-axis moving part, thereby reducing the complexity of circle center compensation and improving positioning accuracy. In addition, a light source is added to the back of the wafer, thereby improving the recognition efficiency of the transparent sheet, and reducing the cost relative to the reflective sensor. At the same time, by turning on the light source only when a transparent wafer is detected, energy can be saved. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The features, nature and advantages of the present invention will become more apparent when the detailed description set forth below is understood in conjunction with the accompanying drawings. In the accompanying drawings, the same reference numerals are always used for corresponding identification. It should be noted that the drawings described are only schematic and non-limiting. In the drawings, the sizes of some components may be exaggerated and are not drawn to scale for illustrative purposes.

[0027] Figure 1 A schematic diagram showing a conventional wafer edge finding process.

[0028] Figure 2 A schematic diagram of a wafer edge finding process according to an embodiment of the present invention is shown.

[0029] Figure 3 A flow chart of a wafer edge finding method according to an embodiment of the present invention is shown.

[0030] Figure 4 A schematic diagram of an image acquisition unit according to an embodiment of the present invention is shown.

[0031] Figure 5 A schematic diagram of the center deviation of a wafer according to an embodiment of the present invention is shown.

[0032] Figure 6 A schematic diagram of performing center compensation for a center deviation of a wafer according to an embodiment of the present invention is shown.

[0033] Figure 7 A schematic diagram of performing rotational compensation for angular deviation of a wafer according to an embodiment of the present invention is shown.

[0034] Figure 8 A schematic block diagram of a wafer edge-finding device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. In the following detailed description, many specific details are set forth to provide a thorough understanding of the described exemplary embodiments. However, it is obvious to those skilled in the art that the described embodiments can be practiced without some or all of these specific details. In other exemplary embodiments, well-known structures are not described in detail to avoid unnecessarily obscuring the concepts of the present disclosure. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. At the same time, in the absence of conflict, the various aspects described in the embodiments can be combined arbitrarily.

[0036] Figure 1 A schematic diagram showing a conventional wafer edge finding process.

[0037] As shown in the figure, in a conventional wafer edge finding process, after the wafer is placed on the chuck stage, the chuck stage first moves from the wafer loading position to the image acquisition position.

[0038] Next, the arc image of the wafer edge is collected at the image collection position, and a circle fitting is performed to calculate the center deviation. For example, the chuck table can be rotated to drive the wafer to rotate together, and the edge image of the wafer can be collected during the wafer rotation process.

[0039] Due to the acceleration and deceleration, signal fluctuations and noise during wafer movement, the collected data may have outliers. Conventional circle fitting algorithms filter the collected image data, and normal points will be eliminated during the filtering process, thereby increasing the fitting error. After data processing, the least squares method is used to fit the center of the circle. If the data is exactly on the circle, the algorithm will find the correct circle. However, when the image data is on a small arc, the algorithm will deviate and estimate a smaller circle. At the same time, the least squares method is sensitive to outliers. If the filtering is not thorough, the fitting result will be biased.

[0040] After calculating the center deviation, center compensation can be performed for the center deviation. Conventional center compensation solutions directly move the chuck table to compensate the X / Y and θ axes after calculating the center deviation. However, the movement of the chuck table will cause the wafer to move at the same time, which not only increases the complexity of the center compensation algorithm, but also brings certain errors to the compensation results.

[0041] After the circle center compensation is performed, the edge notch image can be collected to obtain the deviation of the notch angle relative to the set angle, and rotation compensation can be performed for the angle deviation.

[0042] Finally, after completing the center compensation and rotation compensation, the chuck table moves to the unloading position, whereby the wafer can be transferred from the chuck table to another location.

[0043] In the conventional image acquisition process, transmission sensors and reflection sensors are usually used. Transmission sensors cannot identify metal-plated bond pads, so they cannot complete wafer pre-alignment. The working principle of reflection sensors is to irradiate the light source onto the detection object and receive the reflected light from the detection object for detection. However, the reflectivity of transparent wafers is low and may not be detected by reflection sensors. If the sensitivity of the sensor is increased, the cost will increase. In addition, reflection sensors cannot bear the eccentricity of the bond pad + / - 10mm, and the repeatability positioning accuracy cannot meet the requirements.

[0044] It can be seen that the circle center compensation algorithm used in the existing wafer edge finding process is relatively complex and the circle fitting error is relatively large. Transmissive sensors cannot identify and detect metal-plated bonded wafers, and ordinary reflective sensors cannot detect transparent wafers with low reflectivity. High-sensitivity reflective sensors are expensive and costly.

[0045] Figure 2 A schematic diagram of a wafer edge finding process according to an embodiment of the present invention is shown.

[0046] Figure 2 The wafer edge finding process and Figure 1 The main differences of the wafer edge finding process are: adding a light source on the back of the wafer, and when collecting edge images, the light source is turned on only when the wafer is detected to be transparent, thereby saving energy; after collecting the edge image, firstly, two circle fittings are performed to screen and eliminate abnormal points, and then the circle center deviation is calculated by fitting; in addition, before starting the circle center compensation, the Z-axis moving part outside the chuck table is moved to separate the chuck table from the wafer, such as Figure 2 Shown in the shaded box.

[0047] The image acquisition scheme adopted in the wafer edge finding process of the present invention can effectively increase the recognition efficiency of the transparent sheet and reduce the cost compared with the reflective sensor. In addition, before fitting and calculating the center deviation, the abnormal points can be more accurately eliminated and the normal points can be retained to the greatest extent through two circle fittings, thereby improving the accuracy of circle fitting. At the same time, during the center compensation, the wafer and the chuck stage are separated, so that the chuck stage moves independently during the center compensation process, reducing the compensation error and reducing the complexity of the center compensation.

[0048] Figure 3 A flowchart of a wafer edge finding method 300 according to an embodiment of the present invention is shown.

[0049] like Figure 3 As shown, method 300 starts at step 305. At step 305, an edge image of a wafer is acquired, wherein the wafer is placed on a chuck stage, and the chuck stage can move along the X-axis, the Y-axis, and the θ-axis. Movement along the θ-axis means rotation on the XY plane around the Z-axis direction. In other words, the chuck stage can move along the X-axis and the Y-axis and rotate around the Z-axis.

[0050] Before starting to perform each step of method 300, the chuck stage may be first moved from the wafer loading position to the image acquisition position. For example, in the wafer loading position, the wafer may be placed on the chuck stage. At this time, the wafer and the chuck stage are roughly aligned. Subsequently, the chuck stage is moved to the image acquisition position, thereby moving the wafer to the image acquisition position together.

[0051] When the chuck stage is in the image acquisition position, step 305 may be performed to acquire an image. For example, the image acquisition unit may acquire an image of the edge of the wafer while the chuck stage is rotating.

[0052] The image acquisition unit may be configured to acquire an image of a wafer located at an image acquisition position. For example, the image acquisition unit may include a camera located above an edge of the wafer, and when the wafer is located at the image acquisition position, the edge of the wafer is within the field of view of the camera. Thus, when the chuck stage rotates at the image acquisition position, the image acquisition unit may acquire an edge image of the wafer. After the chuck stage rotates one circle, the image acquisition unit may acquire an edge image of the entire wafer.

[0053] In a preferred embodiment, in addition to the camera located above the wafer, the image acquisition unit may also include a back light source located below the wafer.

[0054] As an example, combining Figure 4 A schematic diagram of an image acquisition unit is shown.

[0055] For ease of explanation, Figure 4 The chuck table 401 and the image acquisition unit of the wafer edge finding device of the present invention are shown. Figure 4 As shown, the camera 402 of the image acquisition unit is located above the wafer, and the back light source 403 is located below the wafer, and the positions of the camera and the back light source are aligned. In a preferred embodiment, the camera is a high-resolution camera so that a clear wafer image can be captured.

[0056] For non-transparent wafers, the edge of the wafer can be identified from the image captured by a high-resolution camera (e.g., using various edge detection algorithms in the field of image processing). However, for transparent wafers, it may be difficult to identify the edge of the wafer from the image captured by the camera because the edge is not obvious. By way of example and not limitation, Figure 4 FIG. 4 shows a transparent wafer (transparent sheet 404 ) placed on a chuck table.

[0057] Thus, during the image acquisition process performed by the image acquisition unit, it is possible to detect whether the wafer is transparent, and only turn on the back light source when the wafer is detected to be transparent, and turn off the back light source when the wafer is detected to be non-transparent, thereby improving the recognition efficiency of the transparent sheet and saving energy. In some implementations, it is possible to detect whether the wafer is transparent based on image processing of the captured image, based on light transmittance, based on a visual detection method, and the like, and then control the turning on / off of the back light source based on the detection result. By turning on the back light source only in the case of a transparent wafer, energy can be saved. At the same time, setting a back light source reduces costs relative to a highly sensitive reflective sensor.

[0058] Return to Figure 3 In step 310, arc feature points and notch feature points of the wafer edge are obtained from the edge image.

[0059] After acquiring the edge image of the wafer, the arc portion and the notch (V-groove or flat) portion of the edge can be identified. Thus, multiple sampling points in the arc portion can be selected to obtain arc feature points, and multiple sampling points in the notch portion can be selected to obtain notch feature points. In various embodiments of the present invention, the feature points can be represented by the coordinates of the selected sampling points.

[0060] When the accuracy requirement is high, a larger number of sampling points can be selected; when the accuracy requirement is not so high, a relatively smaller number of sampling points can be selected. In actual implementation, those skilled in the art can select an appropriate number of sampling points as needed.

[0061] Although it is described above that the captured edge image includes both the arc portion and the notch portion and the arc feature points and the notch feature points are obtained therefrom, this is only exemplary and not restrictive. In a specific implementation, those skilled in the art may also perform image acquisition on the arc portion and the notch portion of the wafer edge separately to obtain the arc image and the notch image, and then obtain the arc feature points from the arc image and obtain the notch feature points from the notch image. For the image acquisition of the arc portion, image acquisition may be performed on arc portions of different curvatures.

[0062] In step 315 , abnormal points in the arc feature points are screened out to obtain screened arc feature points.

[0063] In the present invention, the abnormal points in the arc feature points can be screened out by two rounds of circle fitting. Specifically, step 315 can further include: performing circle fitting on the arc feature points to obtain a first fitted circle; identifying and screening out the abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; performing circle fitting on the remaining arc feature points again to obtain a second fitted circle; identifying and screening out the abnormal points in the remaining arc feature points based on the second fitted circle and a second preset threshold.

[0064] In some embodiments, circle fitting may be performed by a Ransac algorithm.

[0065] In the field of computer vision and machine learning, it is often necessary to process data containing noise and outliers. In these cases, traditional optimization methods such as least squares often fail to obtain satisfactory results. The Ransac (Random Sample Consensus) algorithm can effectively estimate the parameters of the model from a large amount of data through random sampling and model verification, even if the data contains a lot of noise and outliers.

[0066] The basic idea of ​​the Ransac algorithm is to randomly extract a subset of data from the data, use this subset to estimate the parameters of the model, and then verify whether other data points also conform to this model. Repeat this process many times, and the model generated each time is either discarded because there are too few data points, or selected because it is better than the existing model. The general process of the Ransac algorithm is as follows: randomly extract a subset of data, which is called the inliers; use this set of inliers to estimate the parameters of the model; use the estimated model to test other data points; if a point is suitable for this model, it is considered to be an inlier, and if enough points are classified as inliers, then the model is considered to be reasonable; use all assumed inliers to re-estimate the model; evaluate the quality of the model by estimating the error rate of the inliers and the model. Repeat the above steps many times and finally select the optimal model. The Ransac algorithm is well known in the fields of computer vision and machine learning, so I will not go into details here.

[0067] In various embodiments of the present invention, the Ransac algorithm may be used to perform two circle fittings, and based on the results of the two circle fittings and corresponding preset thresholds, abnormal points in the arc feature points may be identified and screened out.

[0068] For example, after performing the first circle fitting on the arc feature points, a first fitted circle can be obtained. Subsequently, the threshold between each arc feature point and the first fitted circle can be determined and the first preset threshold can be set as the first radial distance threshold. Thus, the arc feature points whose radial distance from the first fitted circle is less than or equal to the first preset threshold can be regarded as normal points, and the arc feature points whose radial distance from the first fitted circle is greater than the first preset threshold can be regarded as abnormal points. In this way, abnormal points can be identified and screened out.

[0069] After the above processing, the remaining arc feature points can be fitted again to obtain a second fitted circle. Since the remaining arc feature points are obtained by filtering out a part of abnormal points from all arc feature points, the second fitted circle obtained based on the remaining arc feature points is closer to the real circle where the wafer is located than the first fitted circle, thereby further reducing the fitting error.

[0070] After obtaining the second fitted circle, the radial distance between each remaining arc feature point and the second fitted circle can be determined, and the second preset threshold is set as the second radial distance threshold. Thus, the arc feature points whose radial distances from the second fitted circle are less than or equal to the second preset threshold among the remaining arc feature points can be regarded as normal points, and the arc feature points whose radial distances from the second fitted circle are greater than the second preset threshold can be regarded as abnormal points. In this way, abnormal points among the remaining arc feature points can be identified and screened out.

[0071] In some implementations, the first preset threshold and the second preset threshold may be the same. In other implementations, the first preset threshold and the second preset threshold may be different.

[0072] In a preferred embodiment, the first preset threshold and the second preset threshold can be adjusted dynamically. Taking the first preset threshold as an example, the first preset threshold can be dynamically adjusted within a certain specific range. For example, the first preset threshold can be initially set to a certain initial value (the initial value is within the specific range), and then the first preset threshold can be dynamically adjusted based on a preset rule.

[0073] For example, the preset rule may be that the number or proportion of abnormal points screened out reaches a certain number. In such cases, the first preset threshold value may be dynamically adjusted based on the number or proportion of abnormal points actually screened out. When the first preset threshold value is an initial value (e.g., 0.5 mm), if the number or proportion of abnormal points screened out does not reach the required or desired number or proportion, the first preset threshold value may be adjusted within a specific range (e.g., 0.1 to 1 mm) so that under the adjusted first preset threshold value, the number or proportion of abnormal points screened out reaches the required or desired number or proportion. When adjusting the first preset threshold value within a specific range, the adjustment may be made randomly, in a certain step size, or in other ways.

[0074] In some other examples, the preset rule may be that the degree of deviation between the arc feature points remaining after the outliers are screened out and the fitted circle does not exceed the expected maximum degree of deviation. In such cases, the degree of deviation between the arc feature points remaining after the outliers are screened out and the first fitted circle based on the initial value may be calculated when the first preset threshold is the initial value. For example, if the degree of deviation is too large (exceeding the expected maximum degree of deviation), the first preset threshold may be adjusted within a specific range so that under the adjusted first preset threshold, the degree of deviation between the arc feature points remaining after the outliers are screened out and the first fitted circle does not exceed the expected maximum degree of deviation.

[0075] It should be noted that the above preset rules are only exemplary. In actual implementation, those skilled in the art may also dynamically adjust the first preset threshold and the second preset threshold based on other preset rules.

[0076] In step 320 , the center deviation of the wafer is determined based on the screened arc feature points.

[0077] In some embodiments, determining the center deviation of the wafer based on the screened arc feature points further includes: fitting the screened arc feature points through a Pratt circle fitting algorithm to obtain the wafer center; and determining the deviation between the wafer center and the chuck table center as the center deviation.

[0078] After filtering out the abnormal points in the arc feature points, the Pratt circle fitting algorithm can be used to fit the center of the filtered arc feature points to obtain the center of the wafer. By filtering out the abnormal points twice in the previous step 315, the filtered arc feature points can be made closer to the real circle where the wafer is located, and the center of the circle obtained using these arc feature points is also closer to the real center of the wafer.

[0079] Compared with the traditional least squares algorithm, the advantage of the Pratt circle fitting algorithm is that it eliminates the influence of the radius, so that the algorithm can run correctly even when the curvature radius is very large and the noise is large.

[0080] In an ideal situation, the center of the wafer coincides with the center of the chuck table, and there is no center deviation. However, in practice, there is always a certain deviation between the two. Therefore, the deviation between the center of the wafer and the center of the chuck table can be used as the center deviation of the wafer.

[0081] As an example, combining Figure 5 A schematic diagram of the center deviation of the wafer is shown.

[0082] Assume that the fitted circle is C1, the center of the wafer is P1, the circumference of the chuck table is C2, and the center of the chuck table is P2.

[0083] As an example, an XY coordinate system can be established with P1 as the origin. Thus, the coordinates of the wafer center P1 are (0, 0), and the coordinates of the chuck table center P2 are (x, y), the distance between P2 and P1 is d, and the angle of P2 relative to the X axis is a.

[0084] Therefore, the center deviation of the wafer can be expressed by coordinates (x, y), or by distance d and angle a.

[0085] Return to Figure 3 In step 325, the chuck stage is moved along one or more of the X-axis, the Y-axis, and the θ-axis to perform center compensation for the center deviation, wherein the wafer and the chuck stage are separated during the center compensation.

[0086] After the circle center deviation is determined, the circle center deviation can be compensated. Specifically, the chuck stage can be moved along one or more of the X-axis, the Y-axis and the θ-axis to perform circle center compensation.

[0087] Combination Figure 5 The center deviation shown can move the chuck table -x along the X-axis and -y along the Y-axis to align the chuck table center P2 with the wafer center P1. In an alternative solution, the chuck table can be moved by angle a along the θ-axis (i.e., rotated by angle a around the Z-axis) with P1 as the center, and then moved -d along the X-axis to align the chuck table center P2 with the wafer center P1.

[0088] By filtering out abnormal points twice, the center obtained by using the filtered arc feature points is closer to the true center of the wafer, so that the center deviation determined in step 320 is more accurate. Therefore, the effect of center compensation for the center deviation is also better.

[0089] As an example, combining Figure 6 A schematic diagram of performing center compensation on a center deviation of a wafer is shown.

[0090] like Figure 6 As shown, the chuck table 601 is capable of moving along one or more of an X-axis, a Y-axis, and a θ-axis.

[0091] Figure 6 Also shown in the figure are Z-axis moving parts 605 located on both sides of the chuck stage and below the wafer (including the bonding substrate 602, the silicon wafer 603, and the metal layer 604 plated on the bonding substrate). The Z-axis moving parts can move up and down along the Z-axis.

[0092] exist Figure 6 In the embodiment of the present invention, the Z-axis moving member comprises two parallel rods respectively located at both sides of the chuck table, and the distance between the two parallel rods is greater than the diameter of the chuck table and less than the diameter of the wafer. In order to adapt to the chuck table and wafer of different sizes, the distance between the parallel rods of the Z-axis moving member can be set to be adjustable, and the distance between the parallel rods is adjusted according to the actual size of the chuck table and the wafer. For example, the size of the chuck table and the wafer can be detected, and the distance between the parallel rods is adjusted so that the distance is greater than the diameter of the chuck table and less than the diameter of the wafer. In this way, when the Z-axis moving member moves upward, the wafer can be driven to move upward together and not contact the chuck table, thereby separating the wafer from the chuck table. In certain embodiments, the Z-axis moving member can be moved upward by a preset distance.

[0093] In a preferred embodiment, an attachment member (not shown) may be provided on the Z-axis moving member to enable the wafer to be attached to the Z-axis moving member when the Z-axis moving member contacts the wafer. In this way, the wafer can be prevented from sliding and the stability of the wafer position can be ensured.

[0094] After the wafer and the chuck stage are separated, the chuck stage can be moved along one or more of the X-axis, the Y-axis, and the θ-axis to perform center compensation. In this way, the chuck stage can be moved independently of the wafer, reducing the center compensation error and the complexity of the center compensation algorithm.

[0095] After center compensation, the center of the wafer is aligned with the center of the chuck table.

[0096] After the circle center compensation is completed, the Z-axis moving part returns to its original position so that the wafer contacts the chuck stage and is placed on the chuck stage.

[0097] It should be noted that although Figure 6 The Z-axis moving member is shown to be two parallel rods, but this is only exemplary and non-restrictive. In a specific implementation, those skilled in the art can adopt different ways to realize the Z-axis moving member as needed.

[0098] Back to Figure 3 After the circle center compensation is completed, in step 330, the angle deviation of the wafer is determined based on the notch feature points.

[0099] In some embodiments, determining the angular deviation of the wafer based on the notch feature points further includes: obtaining a notch angle of the wafer based on the notch feature points; and determining a deviation between the notch angle and a specified angle as the angular deviation.

[0100] At step 335, the chuck stage is rotated about the center of the chuck stage to perform rotational compensation for the angular deviation.

[0101] As an example, combining Figure 7 A schematic diagram showing rotational compensation for angular deviation of a wafer is shown.

[0102] After the center compensation, the wafer center P1 coincides with the chuck center P2. Figure 7 as shown in .

[0103] For ease of explanation, Figure 7 The notch-shaped notch (N) at the edge of the wafer is shown in FIG. Assuming that the angle between the center of the notch N and the center of the wafer P1 relative to the reference direction (e.g., the X-axis direction) is b1 (notch angle), and the specified angle is b2, the angular deviation Δb of the wafer is b2-b1.

[0104] As shown in the figure, the chuck stage rotates around the center P1 (P2) in the XY plane by an angle Δb, and the direction of rotation is shown by the arrow in the figure. By rotating, the angle of the wafer notch can be changed from b1 to b2, so that the wafer notch is located in the specified direction. In this way, the angle deviation can be rotationally compensated.

[0105] By executing the above steps of method 300, the wafer can be compensated for the center and rotation, the center deviation can be eliminated to align the center, and the wafer notch can be oriented in a specified direction, thereby accurately positioning the wafer.

[0106] Figure 8 A schematic block diagram of a wafer edge-finding device 800 according to an embodiment of the present invention is shown.

[0107] See also Figure 8 , showing that the wafer edge-finding device 800 includes a chuck table, an image acquisition unit, a feature point acquisition unit, a feature point preprocessing unit, a center compensation unit, a rotation compensation unit and a Z-axis moving part.

[0108] The chuck stage is used to mount the wafer and can move along the X-axis, Y-axis, and θ-axis. Figure 3 and Figure 6 An example of a chuck station is shown in FIG.

[0109] The image acquisition unit may be configured to acquire an edge image of the wafer.

[0110] In some embodiments, the image acquisition unit may include a camera located above the wafer and a back light source located below the wafer, and the image acquisition unit is also configured to: during the image acquisition process, detect whether the wafer is transparent, and turn on the back light source only when it is detected that the wafer is transparent. Figure 4 An example of an image acquisition unit is shown in FIG.

[0111] The feature point acquisition unit may be configured to: obtain arc feature points and notch feature points of the wafer edge from the edge image.

[0112] The feature point preprocessing unit can be configured to: perform circle fitting on the arc feature points to obtain a first fitted circle; identify and filter out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; perform circle fitting on the remaining arc feature points again to obtain a second fitted circle; and identify and filter out abnormal points in the remaining arc feature points based on the second fitted circle and a second preset threshold to obtain filtered arc feature points.

[0113] The center compensation unit can be configured to: determine the center deviation of the wafer based on the screened arc feature points; and move the chuck table along one or more of the X-axis, the Y-axis and the θ-axis to perform center compensation for the center deviation, wherein the wafer and the chuck table are separated during the center compensation.

[0114] The rotation compensation unit may be configured to: determine an angle deviation of the wafer based on the notch feature point; and rotate the chuck table around a center of the chuck table to rotationally compensate for the angle deviation.

[0115] In practice, the feature point acquisition unit, the feature point preprocessing unit, the center compensation unit and the rotation compensation unit can be implemented separately or integrated. For example, the feature point acquisition unit, the feature point preprocessing unit, the center compensation unit and the rotation compensation unit can be integrated in a processor.

[0116] The Z-axis moving part is located outside the chuck table and below the wafer and can move up and down along the Z-axis. When the Z-axis moving part moves upward along the Z-axis, the wafer and the chuck table can be separated. When the Z-axis moving part returns to its original position, the wafer can be contacted and placed on the chuck table.

[0117] In some embodiments, the Z-axis moving member includes two parallel rods respectively located on both sides of the chuck table, and a distance between the two parallel rods is greater than a diameter of the chuck table and smaller than a diameter of the wafer. Figure 6 An example of a Z-axis moving part is shown in FIG.

[0118] Figure 8 The interaction between the various units / components is also shown. For example, the image acquisition unit performs image acquisition on the chuck table (and the wafer placed on the chuck table), and provides the acquired edge image to the feature point acquisition unit. The feature point acquisition unit provides the acquired arc feature points and gap feature points to the feature point preprocessing unit. The feature point preprocessing unit provides the screened arc feature points and gap feature points to the center compensation unit and the rotation compensation unit, respectively. The center compensation unit and the rotation compensation unit cause the chuck table to move to perform center compensation and rotation compensation, respectively. Before the center compensation unit performs center compensation, the Z-axis moving part can be instructed to move upward to separate the chuck table and the wafer.

[0119] Although Figure 8 Specific units / components of the wafer edge-finding device 800 are shown, but it should be understood that these units / components are only exemplary and not restrictive. In different implementations, one or more of these units / components can be combined, split, removed, or additional units / components can be added. For example, in some implementations, the center compensation unit and the rotation compensation unit can be combined into a single unit. In some implementations, the wafer edge-finding device 800 may also include additional units / components. For example, in some embodiments, the wafer edge-finding device may include a drive structure (not shown in the figure) to drive the movement of the chuck table and / or the Z-axis moving part. In some other embodiments, the chuck table and / or the Z-axis moving part may also have their own drive structure. In some implementations, the wafer edge-finding device may also include a control unit (not shown in the figure) to control the operation of the drive structure of the chuck table and / or the Z-axis moving part.

[0120] The technical solution of the present invention has the following advantages:

[0121] (1) For data preprocessing, the Ransac algorithm is used to dynamically adjust the threshold adaptively and perform two-time circle fitting to screen and eliminate abnormal points, retain normal points to the greatest extent, and improve the accuracy of circle fitting; then the Pratt circle fitting algorithm is used to calculate the center of the circle to improve the stability of circle fitting;

[0122] (2) By adding an external Z-axis moving part to the chuck table, the complexity of the center compensation algorithm is reduced, a high-precision positioning compensation algorithm is implemented, and pre-alignment compensation of the wafer is completed;

[0123] (3) By using a high-resolution camera, adding a light source on the back of the wafer, and cooperating with a high-precision recognition algorithm and positioning compensation algorithm, pre-alignment compensation for various types of wafers such as metal layer bonding wafers and transparent wafers can be completed, reducing costs. At the same time, the light source is only turned on when a transparent wafer is detected, saving energy.

[0124] The detailed description described above in conjunction with the accompanying drawings describes examples and does not represent all examples that can be implemented or fall within the scope of the claims. The terms "example" and "exemplary" when used in this specification mean "used as an example, instance or illustration" and do not mean "better or better than other examples".

[0125] Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, use of these phrases may refer to more than just one embodiment. Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0126] The previous description is provided to enable any person skilled in the art to practice the various aspects described herein. Various modifications to these aspects will be readily apparent to those skilled in the art, and the universal principles defined herein may be applied to other aspects. Therefore, the claims are not intended to be limited to the aspects shown herein, but should be granted the full scope consistent with the claims in language, wherein the singular reference to the elements is not intended to mean "there is and only one", but "one or more", unless otherwise specifically stated. Unless otherwise specifically stated, the term "some" refers to one or more. The elements of the various aspects described throughout the present invention are all structurally and functionally equivalent schemes currently or hereafter known to those of ordinary skill in the art and are expressly incorporated herein by reference, and are intended to be covered by the claims.

[0127] It should also be noted that these embodiments may be described as a process depicted as a flow chart, flow diagram, structure diagram, or block diagram. Although the flow chart may describe the operations as sequential processes, many of these operations can be performed in parallel or concurrently. In addition, the order of these operations can be rearranged.

[0128] Although various embodiments have been illustrated and described, it should be understood that the embodiments are not limited to the precise configuration and components described above. Various modifications, substitutions and improvements obvious to those skilled in the art may be made in the arrangement, operation and details of the devices disclosed herein without departing from the scope of the claims.

Claims

1. A wafer edge finding method, comprising: Acquiring an edge image of a wafer, wherein the wafer is placed on a chuck stage, and the chuck stage is movable along an X-axis, a Y-axis, and a θ-axis; Obtain arc feature points and notch feature points of the wafer edge from the edge image; Performing circle fitting on the arc feature points to obtain a first fitted circle; Identifying and filtering out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; Performing circle fitting again on the remaining arc feature points to obtain a second fitted circle, wherein the second fitted circle is closer to the real circle where the wafer is located than the first fitted circle; Based on the second fitted circle and the second preset threshold, identifying and filtering out abnormal points in the remaining arc feature points to obtain filtered arc feature points, wherein the first preset threshold and the second preset threshold can be dynamically adjusted within a specific range based on a preset rule; Determining the center deviation of the wafer based on the screened arc feature points; Moving the chuck stage along one or more of an X-axis, a Y-axis, and a θ-axis to perform center compensation for the center deviation, wherein the wafer and the chuck stage are separated during the center compensation; Determining an angle deviation of the wafer based on the notch feature points; and Rotating the chuck table around the center of the chuck table to perform rotational compensation for the angular deviation, The edge image of the wafer is obtained by an image acquisition unit, wherein the image acquisition unit acquires an image of the edge of the wafer when the chuck table is rotated, and wherein the image acquisition unit includes a camera located above the wafer and a back light source located below the wafer, and the method further includes: During the image acquisition process, detecting whether the wafer is transparent; and The back light source is turned on only when it is detected that the wafer is transparent.

2. The method according to claim 1, wherein: The circle fitting is performed by the Ransac algorithm.

3. The method according to claim 1, wherein: Determining the center deviation of the wafer based on the screened arc feature points further includes: Fitting the selected arc feature points by a Pratt circle fitting algorithm to obtain the center of the wafer; and The deviation between the center of the wafer and the center of the chuck table is determined as the center deviation.

4. The method according to claim 1, wherein: The wafer and the chuck table are separated by a Z-axis moving part, wherein the Z-axis moving part is located outside the chuck table and below the wafer and is capable of moving up and down along the Z-axis, and wherein, before the center of circle compensation begins, the Z-axis moving part moves upward along the Z-axis to separate the wafer and the chuck table.

5. The method according to claim 4, wherein: The Z-axis moving part includes two parallel rods respectively located on both sides of the chuck table, and the distance between the two parallel rods is greater than the diameter of the chuck table and smaller than the diameter of the wafer.

6. The method according to claim 4, wherein: After the circle center compensation is completed, the Z-axis moving part returns to its original position so that the wafer contacts and is placed on the chuck table.

7. The method according to claim 1, wherein: Determining the angular deviation of the wafer based on the notch feature point further includes: Obtaining a notch angle of the wafer based on the notch feature points; and The deviation between the notch angle and a specified angle is determined as the angle deviation.

8. A wafer edge finding device, comprising: A chuck stage, which is used to place the wafer and can move along an X-axis, a Y-axis, and a θ-axis; An image acquisition unit, configured to: acquire an edge image of the wafer; A feature point acquisition unit, configured to: obtain arc feature points and notch feature points of the wafer edge from the edge image; A feature point preprocessing unit is configured to: Performing circle fitting on the arc feature points to obtain a first fitted circle; Identifying and filtering out abnormal points in the arc feature points based on the first fitted circle and a first preset threshold; Performing circle fitting again on the remaining arc feature points to obtain a second fitted circle, wherein the second fitted circle is closer to the real circle where the wafer is located than the first fitted circle; as well as Based on the second fitted circle and the second preset threshold, identifying and filtering out abnormal points in the remaining arc feature points to obtain filtered arc feature points, wherein the first preset threshold and the second preset threshold can be dynamically adjusted within a specific range based on a preset rule; A center compensation unit, which is configured to: Determining the center deviation of the wafer based on the screened arc feature points; and Moving the chuck stage along one or more of the X, axis, the Y axis, and the θ axis to perform center compensation for the center deviation, wherein the wafer and the chuck stage are separated during the center compensation; and A rotation compensation unit configured to: Determining an angle deviation of the wafer based on the notch feature points; and Rotating the chuck table around the center of the chuck table to perform rotational compensation for the angular deviation, The image acquisition unit includes a camera located above the wafer and a back light source located below the wafer, and the image acquisition unit is further configured as follows: During the image acquisition process, detecting whether the wafer is transparent; and The back light source is turned on only when it is detected that the wafer is transparent.

9. The device according to claim 8, wherein: The circle fitting is performed by the Ransac algorithm.

10. The device according to claim 8, wherein: The center compensation unit is further configured to determine the center deviation of the wafer by the following operations: Fitting the selected arc feature points by a Pratt circle fitting algorithm to obtain the center of the wafer; and The deviation between the center of the wafer and the center of the chuck table is determined as the center deviation.

11. The device according to claim 8, further comprising a Z-axis moving member, the Z-axis moving member being located outside the chuck table and below the wafer and being movable up and down along the Z-axis, and wherein, Before the center compensation starts, the Z-axis moving member moves upward along the Z-axis to separate the wafer from the chuck stage.

12. The device according to claim 11, wherein The Z-axis moving part includes two parallel rods respectively located on both sides of the chuck table, and the distance between the two parallel rods is greater than the diameter of the chuck table and smaller than the diameter of the wafer.

13. The device according to claim 11, wherein: After the circle center compensation is completed, the Z-axis moving part returns to its original position so that the wafer contacts and is placed on the chuck table.

14. The device according to claim 8, wherein: The rotation compensation unit is further configured to determine the angular deviation of the wafer by: Obtaining a notch angle of the wafer based on the notch feature points; and The deviation between the notch angle and a specified angle is determined as the angle deviation.

Images