Wafer alignment method, wafer bonding method, and wafer bonding apparatus

Abstract

The application provides a wafer alignment method, a wafer bonding method and a wafer bonding device. The wafer alignment method comprises: for each wafer of two wafers to be aligned, acquiring at least two arc edge images collected for the arc edge of the wafer; determining the center position of the center of the corresponding wafer according to at least two first fitting tangent lines corresponding to the arc edge contained in each of the at least two arc edge images; acquiring two positioning feature images collected one-to-one for the positioning features of the two wafers; determining the feature point position of the geometric feature point of the positioning feature of the corresponding wafer based on each positioning feature image of the two positioning feature images; and controlling the relative motion of the two wafers according to the center position of each of the two wafers and the feature point position of each of the two wafers to align the two wafers. The stability of the center position of the wafer determined by this scheme is high, and the alignment accuracy can be greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor manufacturing, and particularly relates to a wafer alignment method, a wafer bonding method, a wafer bonding apparatus, a computer-readable storage medium, and a computer program product. Background Technology

[0002] In semiconductor manufacturing, the wafer bonding process includes alignment and bonding. Alignment requirements typically include: the core positions of the two wafers coincide, and the orientation of the positioning features of the two wafers (such as notches, edges, markings, etc. formed by wafer processing) is consistent.

[0003] In related technologies, edge feature points at specific locations on the arc edge of the wafer are often used as one of the alignment reference points, or the wafer center position is used as one of the alignment reference points. However, the position of edge feature points is easily affected by noise or local defects (such as wear, stains, reflections, etc.), resulting in poor positioning stability. The wafer center position is usually determined by the "three-point circle determination" method, which is also easily affected by image noise or local defects, causing the extracted arc edge point positions to be offset. If any of the extracted arc edge points has a positional deviation, it will reduce the positioning accuracy of the center position, resulting in low fault tolerance and affecting alignment accuracy and subsequent bonding processes. Summary of the Invention

[0004] This invention provides a wafer alignment method, a wafer bonding method, a wafer bonding apparatus, a computer-readable storage medium, and a computer program product. This approach determines the center position of the wafer by constructing geometric constraints through at least two fitted tangents of the local edge of the arc edge. It has better robustness to image noise and local defects, and the determined center position of the wafer has high stability, which can greatly improve alignment accuracy.

[0005] According to one aspect of the present invention, a wafer alignment method is provided, the method comprising: for each of two wafers to be aligned, acquiring at least two arc edge images of the arc edges of the wafer, wherein the arc edges contained in each of the at least two arc edge images are at least partially non-overlapping on the wafer; determining the center position of the center of the corresponding wafer based on at least two first fitted tangents corresponding one-to-one with the arc edges contained in each of the at least two arc edge images; acquiring two positioning feature images corresponding one-to-one with the positioning features of the two wafers; determining the feature point position of the geometric feature point of the positioning feature of the corresponding wafer based on each of the two positioning feature images; and controlling the relative movement of the two wafers to align the two wafers based on the center position of the two wafers and the feature point position of the two wafers.

[0006] Optionally, the number of arc edge images for each of the two wafers is three, and the number of first fitted tangents is three. The center position of the wafer is determined based on at least two first fitted tangents that correspond one-to-one with the arc edges contained in each of the at least two arc edge images. This includes: for each of the two wafers, the center position of the wafer is determined by using the least squares method based on a preset wafer reference radius and the linear equations of the three first fitted tangents of the wafer.

[0007] Optionally, determining the center position of the wafer's center based on at least two first fitted tangent lines corresponding one-to-one with the arc edges contained in each of the at least two arc edge images includes: extracting multiple edge sampling points on the arc edges contained in each of the at least two arc edge images; performing linear fitting on the multiple edge sampling points contained in each of the at least two arc edge images to obtain at least two first fitted tangent lines corresponding one-to-one with the at least two arc edge images; and determining the center position of the wafer based on the at least two first fitted tangent lines.

[0008] Optionally, extracting multiple edge sampling points on the arc edges contained in each of the at least two arc edge images includes: for each arc edge image of the at least two arc edge images, determining multiple caliper detection positions of the first measuring caliper on the arc edge contained in the arc edge image, and generating multiple scanning regions based on the multiple caliper detection positions; and performing a first sub-pixel edge detection operation along the normal of the first measuring caliper in the multiple scanning regions according to a preset sampling point extraction rule, so as to determine multiple edge sampling points on the arc edges contained in the arc edge image.

[0009] Optionally, the sampling point extraction rules include edge transition polarity, edge amplitude threshold, and edge selection strategy. Edge transition polarity is used to indicate the brightness change direction that the edge sampling point needs to conform to. Edge amplitude threshold is used to indicate the minimum grayscale jump intensity that the edge sampling point needs to conform to. Edge selection strategy is used to indicate the selection strategy of choosing a unique edge sampling point from at least two edge sampling points when the number of edge sampling points identified in any scanning area is at least two.

[0010] Optionally, a straight line fitting operation is performed on the multiple edge sampling points contained in each of the at least two arc edge images to obtain at least two first fitted tangent lines corresponding one-to-one with the at least two arc edge images. This includes: performing a first straight line fitting operation on the multiple edge sampling points contained in each of the at least two arc edge images to obtain initial fitted straight lines for each of the at least two arc edge images; for each arc edge image of the at least two arc edge images, determining a first distance between the multiple edge sampling points of the arc edge image and the initial fitted straight line of the arc edge image; removing edge sampling points whose first distance is greater than a first preset distance, and performing a second straight line fitting operation on the edge sampling points retained after removal to obtain the first fitted tangent line corresponding to the arc edge image.

[0011] Optionally, the positioning features include cuts formed after wafer processing to determine the orientation of the wafer. Determining the feature point positions of geometric feature points of the corresponding wafer's positioning features based on each of the two positioning feature images includes: for each of the two positioning feature images, determining two sets of contour sampling points located at specific positions on the cut contour of that positioning feature image; each set of contour sampling points includes multiple contour sampling points, and the two sets of contour sampling points are located on both sides of the cut vertex on the cut contour; performing linear fitting on the two sets of contour sampling points to obtain two second fitted tangent lines of the cut contour; determining the feature point positions of the geometric feature points of the cut according to the two second fitted tangent lines and a pre-configured second preset distance between the geometric feature points and the two second fitted tangent lines; wherein, the two second fitted tangent lines of one of the two positioning feature images correspond one-to-one with the two second fitted tangent lines of the other positioning feature image, and the distance between a geometric feature point of one of the two positioning feature images and any second fitted tangent line of its corresponding positioning feature image is equal to the distance between a geometric feature point of the other positioning feature image and its corresponding second fitted tangent line.

[0012] Optionally, determining two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: performing a threshold segmentation operation on the positioning feature image to obtain the cut region of the positioning feature image, the cut region being the image region where the cut is located; determining the minimum bounding rectangle of the cut region, the minimum bounding rectangle having a first direction axis extending along the cut depth direction of the cut region and a second direction axis extending along the cut width direction of the cut region; determining the positions of two second measuring calipers based on the center of the minimum bounding rectangle, the axial direction of either the first direction axis or the second direction axis, wherein the scanning areas of each of the two measuring calipers can respectively cover a portion of the cut contour, the two second measuring calipers are respectively located on both sides of the center of the rectangle, and the angles between the two second measuring calipers and the direction axis are corresponding preset angles; and performing a second sub-pixel edge detection operation along the normal of each of the two second measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

[0013] Optionally, determining two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: performing template matching between a preset cut contour template and at least a portion of the image region of the positioning feature image to determine the cut contour of the positioning feature image, wherein the cut contour template can indicate the shape of the cut contour, and at least a portion of the image region includes the image region where the cut is located; determining the positions of two third measuring calipers of the positioning feature image according to the relative positional relationship between the pre-configured measuring calipers and the cut contour template; and performing a third sub-pixel edge detection operation along the normal of each of the two third measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

[0014] Optionally, the feature point position of the geometric feature point of the cut is determined based on the two second fitted tangents and the second preset distance between the pre-configured geometric feature point and the two second fitted tangents, including: determining the angle bisector of the target angle formed by the two sub-tangents of each of the two second fitted tangents, wherein the target angle is the angle formed by the two sub-tangents on the same side of the intersection of the two second fitted tangents; and determining the position of the point on the angle bisector located on the side away from the corresponding wafer and at a distance of the second preset distance from the two second fitted tangents as the feature point position of the geometric feature point.

[0015] Optionally, the location of a point on the angle bisector that is a second preset distance from the two second fitted tangents is determined as the feature point location of the geometric feature point. This includes: calculating the distance between the geometric feature point on the angle bisector and the vertex using trigonometric functions based on the half angle value of the target angle and the second preset distance, so as to determine the feature point location of the geometric feature point.

[0016] Optionally, a straight line fitting operation is performed on the two sets of contour sampling points to obtain two second fitted tangent lines of the cut contour, including: performing a third straight line fitting operation on the two sets of contour sampling points respectively to obtain the initial fitted straight lines of the two sets of contour sampling points; for each set of contour sampling points in the two sets of contour sampling points, determining the second distance between each contour sampling point in the set of contour sampling points and the initial fitted straight line of the set of contour sampling points; removing contour sampling points whose second distance is greater than a third preset distance, and performing a fourth straight line fitting operation on the contour sampling points retained after removal to obtain the second fitted tangent line corresponding to the set of contour sampling points.

[0017] Optionally, the relative movement of the two wafers is controlled based on their respective center positions and feature point positions to align them. This includes: determining the azimuth angle difference between the two wafers based on their respective center positions and feature point positions; controlling at least one of the wafers to rotate based on the azimuth angle difference until the azimuth angle difference meets a preset azimuth angle difference requirement; and controlling at least one of the wafers to move based on the positional deviation between the center positions and / or the positional deviation between the feature point positions after rotation until the positional deviation between the center positions and / or the positional deviation between the feature point positions meets a preset positional deviation requirement.

[0018] Optionally, the wafer alignment method is applied to a wafer bonding apparatus, which includes: a first stage for holding one of two wafers during wafer bonding; a second stage for holding the other of the two wafers during wafer bonding; at least two first image acquisition devices for acquiring at least two arc edge images of each of the two wafers; and a second image acquisition device for acquiring a positioning feature image of each of the two wafers. The wafer alignment method includes: acquiring arc edge images of each of the two wafers by the first image acquisition devices; determining the center position of the corresponding wafer's center based on at least two first fitted tangents corresponding one-to-one with the arc edges contained in each of the at least two arc edge images; acquiring positioning feature images of each of the two wafers by the second image acquisition devices; determining the feature point position of the geometric feature point of the positioning feature of the corresponding wafer based on each of the two positioning feature images; and driving the first stage and the second stage to move relative to each other based on the center position and the feature point position of each of the two wafers to align the two wafers.

[0019] Optionally, before acquiring at least two arc edge images for the arc edge of the wafer, the method further includes: acquiring, for each of the at least two first image acquisition devices and the second image acquisition device, a target image acquired by the image acquisition device for a corresponding target placed on a calibration piece on a second stage, wherein at least two first targets on the calibration piece on the second stage are respectively located within the field of view of the at least two first image acquisition devices, and a second target on the calibration piece on the second stage is located within the field of view of the second image acquisition device; controlling the second stage to move according to a preset 3×3 grid, and during the control of the second stage to move, acquiring, for each of the at least two first image acquisition devices and the second image acquisition device, nine target images acquired by the image acquisition device for a target within the corresponding field of view, wherein the physical positions of the targets contained in each of the nine target images are different; constructing a two-dimensional affine transformation matrix based on the image positions of the targets contained in each of the nine target images and the physical positions of the targets contained in each of the nine target images, the two-dimensional affine transformation matrix being used to transform the image into a target image. The image position in the image coordinate system used by the acquisition device is converted into the physical position in the planar physical coordinate system used by the second platform; the center position of the corresponding wafer is determined according to at least two first fitted tangents corresponding one-to-one with the arc edges contained in each of the at least two arc edge images, including: using a two-dimensional affine transformation matrix to transform the tangent image positions of the at least two first fitted tangents in the image coordinate system of their respective arc edge images into tangent physical positions in the planar physical coordinate system; the center position of the corresponding wafer is determined according to the tangent physical positions of the at least two first fitted tangents in the planar physical coordinate system; the feature point positions of the geometric feature points of the positioning features of the corresponding wafer are determined based on each of the two positioning feature images, including: determining the feature point positions of the geometric feature points of the two wafers in the image coordinate system used by their respective positioning feature images based on the two positioning feature images respectively; the feature point positions of the geometric feature points in the image coordinate system used by their respective positioning feature images are converted into feature point positions in the planar physical coordinate system using a two-dimensional affine transformation matrix.

[0020] According to another aspect of the present invention, a wafer bonding method is also provided, comprising: aligning two wafers according to the wafer alignment method described above; and bonding the aligned two wafers together.

[0021] According to another aspect of the present invention, a wafer bonding apparatus is also provided for bonding two wafers, the wafers having arc edges and positioning features. The apparatus includes: a first stage for holding one of the two wafers during wafer bonding; a second stage for holding the other of the two wafers during wafer bonding; at least two first image acquisition devices for acquiring at least two arc edge images of each of the two wafers, wherein the arc edges contained in each of the at least two arc edge images are at least partially non-overlapping on the wafer; a second image acquisition device for acquiring a positioning feature image of each of the two wafers; a processor; and a memory storing computer program instructions, wherein the processor reads and executes the computer program instructions to implement the wafer alignment method described above, and / or the wafer bonding method described above.

[0022] According to another aspect of the present invention, a computer-readable storage medium is also provided, on which computer program instructions are stored, which, when executed by a processor, implement the wafer alignment method described above, and / or the wafer bonding method described above.

[0023] According to another aspect of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the above-described wafer alignment method and / or the above-described wafer bonding method.

[0024] The above-mentioned technical solution determines the center position of the wafer's center by using the first fitted tangent line based on at least two partially non-overlapping arc edge images on the wafer. This effectively overcomes the problem of insufficient stability in three-point centering and improves the accuracy of wafer centering. Furthermore, this solution obtains the center position of the wafer and the feature point positions of the geometric feature points of the positioning features to control the relative movement of the two wafers to achieve alignment. This helps to achieve both center alignment and azimuth alignment of the wafers, thus taking into account the alignment requirements of both translation and rotation. The overall alignment accuracy of the two wafers is high. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 This is a schematic flowchart of a wafer alignment method according to an embodiment of the present invention;

[0027] Figure 2This is a schematic diagram illustrating the determination of the center position using three first fitted tangent lines and a wafer reference radius according to an embodiment of the present invention;

[0028] Figure 3 This is a schematic diagram of a first measuring caliper according to an embodiment of the present invention;

[0029] Figure 4 This is a schematic diagram of removing discrete edge sampling points according to an embodiment of the present invention;

[0030] Figure 5 This is a schematic diagram illustrating the principle of determining geometric feature points according to an embodiment of the present invention;

[0031] Figure 6 This is a schematic diagram of the minimum circumscribed rectangle of the cut area and the corresponding second measuring caliper according to an embodiment of the present invention;

[0032] Figure 7 This is a schematic diagram showing the relative positions of a measuring caliper and a cut contour template according to an embodiment of the present invention;

[0033] Figure 8 This is a schematic diagram illustrating the determination of the azimuth angle of a wafer according to an embodiment of the present invention;

[0034] Figure 9 This is a schematic diagram of the layout of the download station and the image acquisition device according to an embodiment of the present invention;

[0035] Figure 10 This is a schematic diagram of the target arrangement of a calibration plate according to an embodiment of the present invention;

[0036] Figure 11 This is a schematic diagram of a nine-point calibration according to an embodiment of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0038] To at least partially solve the aforementioned technical problems, embodiments of the present invention provide a wafer alignment method, a wafer bonding method, a wafer bonding apparatus, a computer-readable storage medium, and a computer program product. This approach determines the wafer's center position by constructing geometric constraints using at least two fitted tangents to the local edge of the arc edge. It exhibits better robustness to image noise and local defects, and the determined wafer center position is highly stable, significantly improving alignment accuracy.

[0039] According to one aspect of the present invention, a wafer alignment method is provided. See also... Figure 1 The diagram shown is a schematic flowchart of a wafer alignment method according to an embodiment of the present invention. The method includes steps S110 to S150.

[0040] In step S110, for each of the two wafers to be aligned, at least two arc edge images are acquired for the arc edge of the wafer, and the arc edges contained in each of the at least two arc edge images do not overlap at least partially on the wafer.

[0041] For example, the image acquisition device used to acquire images of the arc edge is typically a high-precision small-field-of-view camera, whose field of view usually only includes one arc segment of the arc edge of the wafer (i.e., the arc edge at a local location). In embodiments of the present invention, at least two arc edge images can be acquired by different image acquisition devices, and each image acquisition device can acquire images for different local locations of the arc edge of each wafer. More specifically, "the arc edges contained in at least two arc edge images do not overlap at least partially on the wafer" means that the arc edges contained in at least two arc edge images correspond to different arc segment regions of the wafer edge; there may be overlap between different arc segment regions, that is, different arc segment regions may not overlap completely or may partially overlap.

[0042] In step S120, the center position of the corresponding wafer is determined based on at least two first fitted tangents that correspond one-to-one with the arc edges contained in each of the at least two arc edge images.

[0043] For example, for each arc edge image, edge extraction can be performed on the arc edge image. The extracted arc edge can have a corresponding fitted tangent line, denoted as the first fitted tangent line. In a specific embodiment, the first fitted tangent line of the arc edge is the line that passes through the most edge sampling points on the arc edge among all candidate lines. The edge sampling points can be all points on the arc edge, or they can be some points on the arc edge. Preferably, the edge sampling points are points with sub-pixel precision on the arc edge of the arc edge image. It should be noted that for any edge sampling point on the arc edge, if the distance between a certain line and the sampling point is less than or equal to a preset distance threshold, then the line can be considered to pass through the edge sampling point. For example, the first fitted tangent line can be obtained by fitting a line to the edge sampling points on the arc edge. It should be noted that, due to the small field of view of the image acquisition device used to acquire the arc edge image, the arc edge in the arc edge image is a small segment of the entire arc edge of the wafer. Since the arc edge image uses pixels as the smallest unit, the arc edge in the arc edge image is not an ideal smooth arc. Consequently, it is difficult to directly locate the center position of the wafer based on the arc in the arc edge image.

[0044] For example, each arc edge image may contain arc edges that have corresponding first fitted tangents. In some embodiments, the number of arc edge images may be two, and correspondingly, the number of first fitted tangents may also be two. The nominal radius (i.e., the nominal radius) of the wafer to be aligned is a known value. Given the positions of the two first fitted tangents and the nominal radius of the wafer, the center position of the wafer's center can be calculated. It can be understood that, given two straight lines and the radius of the circle, four center positions can be calculated. In this embodiment of the invention, the center position located inside the wafer can be selected from the four calculated center positions as the final center position. In other embodiments, the number of arc edge images may be three, and correspondingly, the number of first fitted tangents may also be three. In this case, the center position of the wafer can be directly calculated using the positions of the three first fitted tangents. By calculating the point equidistant from the three first fitted tangents, this point can be determined as the center position of the wafer. Alternatively, the positions of the three first fitted tangents, combined with the nominal radius of the wafer, can be used to solve for the center position of the wafer using the least squares method. This approach will be described in detail below and will not be repeated here. When the number of arc edge images is greater than or equal to 3, that is, when the number of first fitted tangents is greater than or equal to 3, the embodiment of the present invention preferably uses the least squares method to solve for the center position of the wafer.

[0045] In step S130, two positioning feature images are acquired, corresponding to the positioning features of the two wafers.

[0046] For example, wafer positioning features may include notches (also known as "grooves"), diced edges, and markings formed after wafer processing. Positioning features are used to determine the wafer's orientation, i.e., its azimuth. Similarly, the image acquisition device used to capture images of positioning features is typically a high-precision, small-field-of-view camera, whose field of view may include at least the wafer's positioning features.

[0047] In step S140, the feature point positions of the geometric feature points of the corresponding wafer's positioning features are determined based on each of the two positioning feature images.

[0048] For example, geometric feature points are points that can be uniquely determined for a positioning feature. Taking a notch as an example, the geometric feature point of the notch could be, for example, the notch vertex (i.e., the point on the notch profile closest to the center of the wafer), the center of the circle passing through the notch vertex obtained by contour transformation of the notch, or the center of the largest inscribed circle of the notch, etc. Taking a cut edge as an example, the geometric feature point of the cut edge could be, for example, the endpoint of the cut edge, the midpoint of the cut edge, etc. It can be understood that the two wafers to be aligned have the same type of positioning feature, and the geometric feature points of their respective positioning features are of the same type.

[0049] In step S150, the two wafers are controlled to move relative to each other based on the center position of each wafer and the feature point position of each wafer, so as to align the two wafers.

[0050] For example, during wafer alignment, the wafer can be controlled to move only within a plane, including rotation and translation. In some embodiments, only one of the two wafers can be controlled to move; specifically, that wafer can be controlled to translate along a first direction and a second direction in a horizontal plane, and to rotate within the horizontal plane, to achieve alignment of the two wafers. In other embodiments, one of the two wafers can be controlled to translate along a first direction and a second direction in a horizontal plane, while the other wafer can be controlled to rotate within the horizontal plane. Preferably, the first direction and the second direction are perpendicular to each other.

[0051] It should be noted that the image acquisition devices in the embodiments of the present invention (including the first image acquisition device and the second image acquisition device hereinafter referred to as the first image acquisition device and the second image acquisition device) can all have their intrinsic and extrinsic parameters calibrated in advance, or they can all have their corresponding two-dimensional affine transformation matrices established in advance. In other words, the image coordinates of each pixel of the image acquired by the image acquisition device in the image coordinate system are transformed to the physical coordinates in the preset physical coordinate system. In some embodiments, the position of the first fitted tangent, the center position of the wafer, and the feature point position of the geometric feature point involved in the above steps S110 to S140 can all be pixel positions (i.e., image coordinates) in the image coordinate system of the respective image. More specifically, when the intrinsic and extrinsic parameters of the image acquisition device have been calibrated in advance, the image positions of the first fitted tangents in different image spaces (i.e., different image coordinate systems) can be transformed to the same image space, and the center position of the wafer in the image space can be solved using the first fitted tangent in the same image space. In step S150, the center position and feature point position in the image coordinate system can be transformed to the physical coordinate system first, and then alignment is performed according to the center position and feature point position in the physical coordinate system. In other embodiments, the image position of the first fitted tangent in the image can be converted to the physical position in the physical coordinate system. Accordingly, the center position obtained in step S120 is also the position in the physical coordinate system.

[0052] For example, the positional deviation between the centers of the two wafers can be obtained based on their respective center positions; the positional deviation between the geometric feature points of the two wafers can be obtained based on their feature point positions. It is understood that, at least based on the center positional deviation and the feature point positional deviation, the relative movement of the two wafers is controlled to ensure that the positional deviations of the centers and the geometric feature points of the aligned wafers meet the alignment requirements. Preferably, for each wafer, the azimuth angle can be determined based on its center position and feature point positions, and the angular difference between the azimuth angles of the two wafers can characterize their relative tilt angle. Therefore, the relative movement of the two wafers can be controlled based on their relative tilt angle, center positional deviation, and feature point positional deviation to ensure that the relative tilt angle, center positional deviation, and geometric feature point positional deviation of the aligned wafers meet the alignment requirements.

[0053] The above-mentioned technical solution determines the center position of the wafer's center by using the first fitted tangent line based on at least two partially non-overlapping arc edge images on the wafer. This effectively overcomes the problem of insufficient stability in three-point centering and improves the accuracy of wafer centering. Furthermore, this solution obtains the center position of the wafer and the feature point positions of the geometric feature points of the positioning features to control the relative movement of the two wafers to achieve alignment. This helps to achieve both center alignment and azimuth alignment of the wafers, thus taking into account the alignment requirements of both translation and rotation. The overall alignment accuracy of the two wafers is high.

[0054] Optionally, the number of arc edge images for each of the two wafers is three, and the number of first fitted tangents is three. The center position of the wafer is determined based on at least two first fitted tangents that correspond one-to-one with the arc edges contained in each of the at least two arc edge images. This includes: for each of the two wafers, the center position of the wafer is determined by using the least squares method based on a preset wafer reference radius and the linear equations of the three first fitted tangents of the wafer.

[0055] For example, the wafer reference radius is the nominal radius of the wafer mentioned above. Preferably, this embodiment of the invention employs the least squares method to solve for the center position based on the wafer reference radius and the positions of the three first fitted tangent lines. The position of the first fitted tangent lines can be represented by a linear equation. Specifically, the unit normal linear equation of the first fitted tangent lines can be expressed as:

[0056] a i X+b i Y+c i =0

[0057] Among them, ||(a i , b i )||=1,a i b i and c i Both methods can be solved using the coordinates of any two points on the fitted tangent line, where i represents the i-th first fitted tangent line, and the normal vector points inwards on the wafer. The coordinates of the points on the fitted tangent line can be image coordinates in the image coordinate system. In this case, the solved linear equation can represent the image position of the fitted tangent line in image space. Alternatively, the image coordinates of the points on the fitted tangent line in the image coordinate system can be converted to physical coordinates in the physical coordinate system. In this case, the solved linear equation can represent the physical position of the fitted tangent line in physical space. In one specific embodiment, the linear equation of the first fitted tangent line can represent its physical position on the plane containing the wafer surface in physical space.

[0058] The wafer reference radius is denoted as R. nomThe position of the center of the wafer in the plane containing the wafer surface in physical space is denoted as (X). c , Y c For the i-th first fitted tangent line, the following equation can be constructed:

[0059] s i (a) i X c +b i Y c +c i =R nom ,

[0060] Among them, s i is a sign factor, taking the value 1 or -1. If the center of the circle is located on the side of the positive direction of the normal vector of the i-th first fitted tangent line, the value is 1; if the center of the circle is located on the side of the negative direction of the normal vector of the i-th first fitted tangent line, the value is -1. The three first fitted tangent lines can construct three equations, where X... c and Y c Given two unknowns, the least squares method can be used to solve for (X). c , Y c Please see ). Figure 2 As shown, it is a schematic diagram of determining the center position using three first fitting tangents and the wafer reference radius according to an embodiment of the present invention. Figure 2 The three first-fit tangent lines shown are denoted as L1, L2, and L3, with L2 and L3 being parallel. It can be understood that, due to the sign factor, even though L2 and L3 are parallel, three equations can still be constructed.

[0061] In one specific embodiment, the overdetermined system of equations (three equations, two unknowns) constructed for the three first fitted tangent lines can be represented in matrix form as follows:

[0062] Ap = b, p = (Xc, Yc) T

[0063] Matrix A can be represented as follows:

[0064]

[0065] Matrix b can be represented as follows:

[0066]

[0067] Solve p=(A) using the least squares closed-form method. T A) -1 A T b, then the position of the center of the circle can be obtained, or p can be solved using numerical equivalent methods such as QR / SVD.

[0068] It should be noted that, theoretically, Rnom can be treated as an unknown in the solution (three equations can solve for three unknowns). However, due to noise in the arc edge image, the arc edge obtained by edge extraction from the arc edge image may have deviations. Furthermore, the extracted arc edge is affected by pixel discretization, which leads to a deviation between the first fitted tangent determined by the image and the true tangent of the (actual) arc edge of the wafer. This causes the obtained wafer radius to drift, resulting in poor stability of the center position. The embodiments of this invention use a wafer reference radius for constraint, resulting in better stability of the solved center position.

[0069] In some embodiments, self-test gating logic can be configured. The self-test gating logic is represented by the following formula:

[0070]

[0071] In the formula, MaxErr represents the maximum absolute deviation between the distance between the center position of the wafer and the three first fitted tangents and the wafer reference radius. When MaxErr is greater than or equal to the corresponding preset deviation threshold, the wafer is rejected (the obtained center position is determined to be invalid) or a prompt message is output.

[0072] In the above technical solution, the wafer radius is not calculated by back-calculating through the three first fitting tangents, which can effectively avoid the drift of the wafer radius solution caused by image noise. Furthermore, since both wafers use the same wafer reference radius, the positioning error introduced by the wafer reference radius deviation can be partially canceled out during differential alignment calculation. This helps to avoid further amplification of alignment calculation error due to the difference in wafer reference radii between the two wafers.

[0073] Optionally, determining the center position of the wafer's center based on at least two first fitted tangent lines corresponding one-to-one with the arc edges contained in each of the at least two arc edge images includes: extracting multiple edge sampling points on the arc edges contained in each of the at least two arc edge images; performing linear fitting on the multiple edge sampling points contained in each of the at least two arc edge images to obtain at least two first fitted tangent lines corresponding one-to-one with the at least two arc edge images; and determining the center position of the wafer based on the at least two first fitted tangent lines.

[0074] For example, edge sampling points can be obtained by performing sub-pixel edge extraction, contour point acquisition, threshold boundary sampling, edge tracking, or other equivalent sub-pixel level feature point acquisition methods on the arc edges contained in the arc edge image. In this embodiment of the invention, a preferred method is to use measuring calipers arranged along the arc edge to perform sub-pixel edge measurement, which will be described in detail below. For any arc edge image, a first fitted tangent line can be obtained by performing straight line fitting on multiple edge sampling points of the arc edge image. This embodiment of the invention does not specifically limit the fitting method for straight line fitting. For example, the straight line fitting method can be a common least squares straight line fitting method, an orthogonal least squares (total least squares) straight line fitting method, a weighted least squares straight line fitting method, a robust weighted straight line fitting method (such as Tukey / Huber weight functions), a RANSAC straight line estimation method, a Hough transform straight line detection method, and other equivalent methods for obtaining straight lines from point sets. By performing straight line fitting on the edge sampling points to obtain the fitted tangent line, the influence of image noise on the arc edge can be effectively reduced, and a relatively stable first fitted tangent line can be obtained.

[0075] Optionally, extracting multiple edge sampling points on the arc edges contained in each of the at least two arc edge images includes: for each arc edge image of the at least two arc edge images, determining multiple caliper detection positions of the first measuring caliper on the arc edge contained in the arc edge image, and generating multiple scanning regions based on the multiple caliper detection positions; and performing a first sub-pixel edge detection operation along the normal of the first measuring caliper in the multiple scanning regions according to a preset sampling point extraction rule, so as to determine multiple edge sampling points on the arc edges contained in the arc edge image.

[0076] For example, for any image with an arc edge, a first measuring caliper can be positioned along the arc edge in the image. The first measuring caliper has multiple detection positions, which can be understood as reference detection points where the first measuring caliper is positioned on the arc edge. For each caliper detection position, a rectangular region of interest (ROI) can be generated based on that caliper detection position; this rectangular ROI is also the scanned area. See [link to relevant documentation]. Figure 3 As shown, this is a schematic diagram of a first measuring caliper according to an embodiment of the present invention. Each solid dot can represent a caliper detection position, and a scanning area can be generated between every two caliper detection positions. Figure 3 There are 15 caliper detection positions in total, generating 14 scanning areas.

[0077] In one specific embodiment, the scanning area is a local search window centered on the caliper detection position. Exemplarily, the "normal direction of the first measuring caliper" is the direction pointing towards the center of the wafer (i.e., the radial direction of the arc edge). The first sub-pixel edge detection operation can employ any sub-pixel edge algorithm (e.g., grayscale centroid method, polynomial interpolation method, Zernike moment, etc.), as long as it can solve for the sub-pixel level edge coordinates along the normal direction of the first measuring caliper.

[0078] The above technical solution uses a first measuring caliper to delineate the scanning area and performs sub-pixel edge detection along the normal of the caliper to extract edge sampling points. This can effectively reduce the search range of edge detection, reduce image noise interference, and improve the reliability of wafer center positioning by obtaining sub-pixel precision edge sampling points.

[0079] Optionally, the sampling point extraction rules include edge transition polarity, edge amplitude threshold, and edge selection strategy. Edge transition polarity is used to indicate the brightness change direction that the edge sampling point needs to conform to. Edge amplitude threshold is used to indicate the minimum grayscale jump intensity that the edge sampling point needs to conform to. Edge selection strategy is used to indicate the selection strategy of choosing a unique edge sampling point from at least two edge sampling points when the number of edge sampling points identified in any scanning area is at least two.

[0080] For example, edge transition polarity is used to define the direction of grayscale brightness change corresponding to the edge sampling point, and only positions that satisfy this direction of brightness change are judged as candidate edge sampling points; edge amplitude threshold is used to define the lower limit of the grayscale jump intensity required for a valid edge, and positions with grayscale change amplitude below the threshold are eliminated; edge selection strategy is used to handle scenarios where multiple candidate edge sampling points are detected in a single scanning area. When at least two candidate edge sampling points are identified in the same scanning area, a unique edge sampling point can be determined according to this strategy as the final output edge sampling point of the scanning area. The edge selection strategy can be, for example, selecting the first / last edge sampling point as the final output edge sampling point of the corresponding scanning area, or selecting the edge sampling point with the largest grayscale change amplitude as the final output edge sampling point of the corresponding scanning area, etc. The embodiments of the present invention do not make specific limitations on the edge sampling point selection strategy.

[0081] The above technical solution, by configuring edge transition polarity, edge amplitude threshold and edge selection strategy, can effectively filter image noise and false edges, and stably output a unique and highly reliable edge sampling point in each scanning area, which is beneficial to ensuring the accuracy of subsequent arc edge fitting tangent and wafer center solution.

[0082] Optionally, a straight line fitting operation is performed on the multiple edge sampling points contained in each of the at least two arc edge images to obtain at least two first fitted tangent lines corresponding one-to-one with the at least two arc edge images. This includes: performing a first straight line fitting operation on the multiple edge sampling points contained in each of the at least two arc edge images to obtain initial fitted straight lines for each of the at least two arc edge images; for each arc edge image of the at least two arc edge images, determining a first distance between the multiple edge sampling points of the arc edge image and the initial fitted straight line of the arc edge image; removing edge sampling points whose first distance is greater than a first preset distance, and performing a second straight line fitting operation on the edge sampling points retained after removal to obtain the first fitted tangent line corresponding to the arc edge image.

[0083] For example, when performing straight-line fitting on multiple edge sampling points of any circular arc edge image, a first straight-line fitting operation can be performed first to obtain an initial fitted straight line. The fitting method for the first straight-line fitting operation can be referred to the relevant description in the foregoing embodiments, and will not be repeated here. After obtaining the initial fitted straight line, a first distance between each edge sampling point on the circular arc edge image and the initial fitted straight line can be calculated, and edge sampling points whose first distance is greater than a first preset distance can be removed. A second straight-line fitting operation is then performed on the remaining edge sampling points after removal to obtain the final first fitted tangent line. It is understood that the fitting methods used in the first and second straight-line fitting operations can be the same or different. Please refer to [link to relevant documentation]. Figure 4 The diagram illustrates the process of removing discrete edge sampling points according to an embodiment of the present invention. The resulting edge sampling points may include outliers at the ends of arc edges, dirt points, endpoints with discrete features, etc. These points usually deviate significantly from the initial fitted line. Therefore, these points are removed by distance constraints, and a second line fitting is performed on the remaining edge sampling points after removal.

[0084] In this embodiment of the invention, the number of edge sampling points retained after removal can optionally be detected. Specifically, the second straight-line fitting operation can be performed when the number of retained edge sampling points is greater than or equal to a preset minimum number of points, or when the ratio of the number of retained edge sampling points to the number of edge sampling points before removal is greater than or equal to a preset ratio. After obtaining the final first fitted tangent line, the root mean square residual (RMS residual) from the retained edge sampling points to the first fitted tangent line can also be calculated. If the RMS residual is greater than a preset residual threshold, the first fitted tangent line can be determined to be invalid.

[0085] The above technical solution can effectively suppress the interference of discrete noise points on the circular arc edge by performing a second straight line fitting on the edge sampling points, reduce the influence of abnormal edge sampling points on the fitted tangent, and obtain a more accurate first fitted tangent.

[0086] Optionally, the positioning features include cuts formed after wafer processing to determine the orientation of the wafer. Determining the feature point positions of geometric feature points of the corresponding wafer's positioning features based on each of the two positioning feature images includes: for each of the two positioning feature images, determining two sets of contour sampling points located at specific positions on the cut contour of that positioning feature image; each set of contour sampling points includes multiple contour sampling points, and the two sets of contour sampling points are located on both sides of the cut vertex on the cut contour; performing linear fitting on the two sets of contour sampling points to obtain two second fitted tangent lines of the cut contour; determining the feature point positions of the geometric feature points of the cut according to the two second fitted tangent lines and a pre-configured second preset distance between the geometric feature points and the two second fitted tangent lines; wherein, the two second fitted tangent lines of one of the two positioning feature images correspond one-to-one with the two second fitted tangent lines of the other positioning feature image, and the distance between a geometric feature point of one of the two positioning feature images and any second fitted tangent line of its corresponding positioning feature image is equal to the distance between a geometric feature point of the other positioning feature image and its corresponding second fitted tangent line.

[0087] For example, the shape of the cut can be U-shaped. It is understood that algorithms such as edge detection, template matching, or image segmentation can be used to identify and determine the cut contour. A specific position on the cut contour is a relative selection area of ​​the selected contour sampling points relative to the cut contour. It is understood that the relative selection areas corresponding to two positioning feature images are basically consistent in relative position to their respective cut contours. For example, the relative selection areas of the two sets of contour sampling points in each positioning feature image are located on both sides of the cut vertex of the cut contour in that positioning feature image. Correspondingly, the two sets of contour sampling points are located on both sides of the corresponding cut vertex. This embodiment of the invention does not specifically limit the method of determining the specific position; the specific position of each set of contour sampling points only needs to be uniquely identifiable. For example, contour sampling points can be obtained by performing sub-pixel edge extraction, contour point acquisition, threshold boundary sampling, edge tracking, or other equivalent sub-pixel level feature point acquisition methods on specific positions of the cut contour contained in the positioning feature image. This embodiment of the invention preferably uses a method of performing sub-pixel edge measurement at a specific position using calipers; this scheme will be described in detail below.

[0088] For example, performing straight-line fitting on each contour sampling point of each group of contour sampling points can generate corresponding fitted tangents, denoted as the second fitted tangents. For any localization feature image, after determining the two second fitted tangents of the localization feature image, points whose distances from the two second fitted tangents are respectively second preset distances can be determined, and these points can be used as geometric feature points. It should be noted that the second preset distance d1 between the geometric feature point and one of the second fitted tangents and the second preset distance d2 between the geometric feature point and the other second fitted tangent can be the same or different. When d1 and d2 are different, the determined geometric feature points need to satisfy the following conditions: the two second fitted tangents of the two localization feature images with a distance of d1 between them and the corresponding geometric feature points have a corresponding relationship, and the two second fitted tangents of the two localization feature images with a distance of d2 between them and the corresponding geometric feature points have a corresponding relationship. Constrained by the feeding mechanism, the azimuth angles of the two wafers before alignment are usually small, and correspondingly, the orientation of the cuts of the two wafers is small. Therefore, the correspondence between the second fitted tangents of the two positioning feature images can be determined. For example, it can be determined by the sign of the tangent slope in the image coordinate system of their respective images. Second fitted tangents with the same sign have a corresponding relationship.

[0089] It is understood that by using the distance constraints with the two second fitted tangents, two points can be determined: one located closer to the wafer center and the other farther away. In this embodiment of the invention, the point farther away from the wafer center is preferably used as the geometric feature point of the corresponding positioning feature image. In some embodiments, the position of the second fitted tangent in the image space of the corresponding image can be first converted to physical space to obtain the position of the geometric feature point in physical space. In other embodiments, the position of the geometric feature point in the image space can be solved using the position of the second fitted tangent in the image space of the corresponding image, and then the image position of the geometric feature point can be converted to its physical position in physical space. Please refer to [link to relevant documentation]. Figure 5 As shown, it is a schematic diagram illustrating the principle of determining geometric feature points according to an embodiment of the present invention. Figure 5 The notch shown is a U-shaped notch. In this embodiment, the determined geometric feature points are all equidistant from the two second fitted tangents, each at a distance of R. Figure 5 The center position shown is the location of the geometric feature point in this embodiment.

[0090] The above technical solution obtains the second fitted tangent by collecting contour sampling points on both sides of the cut vertex and fitting them, and then solves the geometric feature points of the cut according to the second preset distance. This can effectively overcome the problems of imaging distortion at the cut vertex or blurred cut contour edges, and obtain the geometric feature points of the cut relatively stably.

[0091] Optionally, determining two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: performing a threshold segmentation operation on the positioning feature image to obtain the cut region of the positioning feature image, the cut region being the image region where the cut is located; determining the minimum bounding rectangle of the cut region, the minimum bounding rectangle having a first direction axis extending along the cut depth direction of the cut region and a second direction axis extending along the cut width direction of the cut region; determining the positions of two second measuring calipers based on the center of the minimum bounding rectangle, the axial direction of either the first direction axis or the second direction axis, wherein the scanning areas of each of the two measuring calipers can respectively cover a portion of the cut contour, the two second measuring calipers are respectively located on both sides of the center of the rectangle, and the angles between the two second measuring calipers and the direction axis are corresponding preset angles; and performing a second sub-pixel edge detection operation along the normal of each of the two second measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

[0092] In some embodiments, the cut region of the localization feature image can be obtained by performing a threshold segmentation operation on the localization feature image. Preferably, an adaptive threshold segmentation algorithm can be used for the threshold segmentation operation. Based on the cut region, the minimum bounding rectangle of the cut region can be determined. The minimum bounding rectangle has a major axis and a minor axis. In some embodiments, the extension direction of the major axis of the minimum bounding rectangle is substantially consistent with the cut width direction, and the extension direction of the minor axis of the minimum bounding rectangle is substantially consistent with the cut depth direction. In this case, the first direction axis is the minor axis of the minimum bounding rectangle, and the second direction axis is the major axis of the minimum bounding rectangle. In other embodiments, the extension direction of the major axis of the minimum bounding rectangle is substantially consistent with the cut depth direction, and the extension direction of the minor axis of the minimum bounding rectangle is substantially consistent with the cut width direction. In this case, the first direction axis is the major axis of the minimum bounding rectangle, and the second direction axis is the minor axis of the minimum bounding rectangle.

[0093] For example, the center of the smallest bounding rectangle of the cut area is used as the position reference datum, and the axial direction of either the first or second direction axis is selected as the angle reference datum. Under the aforementioned datum, two second measuring calipers are arranged on both sides of the center of the smallest bounding rectangle. The two second measuring calipers are at corresponding preset angles to the selected direction axis, so that the scanning areas of each of the two second measuring calipers can respectively cover a portion of the cut contour. For a description of the second measuring calipers and their corresponding scanning areas, please refer to the relevant explanation of the first measuring calipers in the aforementioned embodiments; it will not be repeated here. The process of performing the second sub-pixel edge detection operation along the normal of the second measuring calipers can refer to the aforementioned process of performing the first sub-pixel edge detection operation along the normal of the first measuring calipers; it will not be repeated here. Please refer to... Figure 6As shown, it is a schematic diagram of the minimum circumscribed rectangle of the cut area and the corresponding second measuring caliper according to an embodiment of the present invention. Figure 6 The gray-filled area is the cut area. The first direction axis of the smallest bounding rectangle is the minor axis, and the second direction axis is the major axis. The second measuring calipers are located on both sides of the vertex of the cut contour.

[0094] The above technical solution utilizes the center of the smallest bounding rectangle and the axial direction of the direction axis to arrange the second measuring caliper, and performs sub-pixel edge detection along the normal of the caliper. This can effectively enhance the adaptability of the second measuring caliper to cuts at different image positions. Even when there is a positional angular deviation between the cuts contained in the two positioning feature images, it can still stably obtain two sets of contour sampling points of the cut, which is beneficial to improving the accuracy and robustness of solving the geometric feature points of the cut.

[0095] Optionally, determining two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: performing template matching between a preset cut contour template and at least a portion of the image region of the positioning feature image to determine the cut contour of the positioning feature image, wherein the cut contour template can indicate the shape of the cut contour, and at least a portion of the image region includes the image region where the cut is located; determining the positions of two third measuring calipers of the positioning feature image according to the relative positional relationship between the pre-configured measuring calipers and the cut contour template; and performing a third sub-pixel edge detection operation along the normal of each of the two third measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

[0096] In other embodiments, a pre-configured cut contour template can be used for template matching with at least a portion of the image region of the positioning feature image. Specifically, the cut contour template can be used to perform template matching with the entire positioning feature image to obtain the image position of the cut contour in the positioning feature image. Alternatively, a region of interest (ROI) can be extracted from the positioning feature image first; the ROI may be, for example, a rectangular image region including the cut region. Template matching using the cut contour template with the ROI of the positioning feature image can obtain the image position of the cut contour in the positioning feature image. Optionally, a matching angle range (e.g., ±15°) can be set before template matching. For example, the relative position of a calibration measuring caliper and the cut contour template can be pre-configured; please refer to [link to relevant documentation]. Figure 7 As shown, it is a schematic diagram of the relative positions of a measuring caliper and a cut contour template according to an embodiment of the present invention. Figure 7The standard cut contour of the cut contour template and the relative position of the measuring calipers with respect to the standard cut contour are shown. After obtaining the position and pose of the cut in the positioning feature image through template matching, the measuring calipers pre-configured on the template can be mapped to the positioning feature image to obtain two measuring calipers in the positioning feature image, i.e., two third measuring calipers. The description of the third measuring calipers can refer to the relevant explanation of the first measuring calipers in the aforementioned embodiments, and will not be repeated here. The process of performing the second sub-pixel edge detection operation along the normal of the third measuring calipers can refer to the process of performing the first sub-pixel edge detection operation along the normal of the first measuring calipers, and will not be repeated here. This embodiment of the invention can also set a matching score gate; if the matching score of the matched cut contour is less than a preset minimum matching score, the matched cut contour can be determined to be invalid.

[0097] The above technical solution obtains contour sampling points by template matching and combining the template with a pre-calibrated third measuring caliper. This enables the measuring caliper to adaptively map to the positioning feature image following the template posture. Even when there is a positional angle deviation between the cuts contained in the two positioning feature images, it is still possible to stably extract two sets of contour sampling points at specific positions on the cut contour.

[0098] Optionally, the feature point position of the geometric feature point of the cut is determined based on the two second fitted tangents and the second preset distance between the pre-configured geometric feature point and the two second fitted tangents, including: determining the angle bisector of the target angle formed by the two sub-tangents of each of the two second fitted tangents, wherein the target angle is the angle formed by the two sub-tangents on the same side of the intersection of the two second fitted tangents; and determining the position of the point on the angle bisector located on the side away from the corresponding wafer and at a distance of the second preset distance from the two second fitted tangents as the feature point position of the geometric feature point.

[0099] For example, the intersection of two second-fitted tangent lines can form four included angles, and each second-fitted tangent line is divided into two sub-tangent lines by the intersection point. Using the intersection point of the two second-fitted tangent lines as the dividing point, a side closer to the wafer and a side farther from the wafer can be distinguished. Two sub-tangent lines located on the same side belong to different second-fitted tangent lines; the included angle formed by the two sub-tangent lines closer to the wafer and the included angle formed by the two sub-tangent lines farther from the wafer are vertical angles, and their angle bisectors are the same straight line. Therefore, any of these included angles can be used as target included angles. On the angle bisector of the target included angle, a point on the side farther from the wafer, at a distance of a second preset distance from the second-fitted tangent line, can be determined; this point is the geometric feature point. This scheme can quickly and accurately determine the geometric feature points of the localization feature image.

[0100] Optionally, the location of a point on the angle bisector that is a second preset distance from the two second fitted tangents is determined as the feature point location of the geometric feature point. This includes: calculating the distance between the geometric feature point on the angle bisector and the vertex using trigonometric functions based on the half angle value of the target angle and the second preset distance, so as to determine the feature point location of the geometric feature point.

[0101] Specifically, the intersection of the two second fitted tangent lines is denoted as (r v ,c v Let the second preset distance be denoted as R. px The unit direction vectors pointing from the intersection of the two second-fitted tangents away from the wafer are summed and normalized to obtain the unit direction of the interior angle bisector (AxisRow, AxisCol); the half-angle HalfAngle is obtained from the target angle between the two second-fitted tangents. The distance between the geometric feature point to be solved and the intersection of the two second-fitted tangents is denoted as CenterOffset, and its calculation formula is as follows: CenterOffset = R px / sin(HalfAngle). Correspondingly, the two candidate geometric feature points are (r v ±CenterOffset,c v (±CenterOffset). The feature point position is selected from the side of the wafer furthest from the wafer. This scheme has a relatively simple calculation logic and strong interpretability.

[0102] Optionally, a straight line fitting operation is performed on the two sets of contour sampling points to obtain two second fitted tangent lines of the cut contour, including: performing a third straight line fitting operation on the two sets of contour sampling points respectively to obtain the initial fitted straight lines of the two sets of contour sampling points; for each set of contour sampling points in the two sets of contour sampling points, determining the second distance between each contour sampling point in the set of contour sampling points and the initial fitted straight line of the set of contour sampling points; removing contour sampling points whose second distance is greater than a third preset distance, and performing a fourth straight line fitting operation on the contour sampling points retained after removal to obtain the second fitted tangent line corresponding to the set of contour sampling points.

[0103] It is understandable that a secondary straight-line fitting can also be performed on each set of contour sampling points. The specific process can be referred to the relevant description of the secondary straight-line fitting of edge sampling points in the previous embodiment, which will not be repeated here. By performing a secondary straight-line fitting on the contour sampling points, the interference of discrete noise points of the cut contour can be effectively suppressed, the influence of abnormal contour sampling points on the fitted tangent can be reduced, and a more accurate second fitted tangent can be obtained.

[0104] Optionally, the relative movement of the two wafers is controlled based on their respective center positions and feature point positions to align them. This includes: determining the azimuth angle difference between the two wafers based on their respective center positions and feature point positions; controlling at least one of the wafers to rotate based on the azimuth angle difference until the azimuth angle difference meets a preset azimuth angle difference requirement; and controlling at least one of the wafers to move based on the positional deviation between the center positions and / or the positional deviation between the feature point positions after rotation until the positional deviation between the center positions and / or the positional deviation between the feature point positions meets a preset positional deviation requirement.

[0105] For example, the azimuth angle of any wafer can be represented by the angle between the line connecting its center position and the feature point position and the reference axis of the plane containing the wafer surface in physical space. Please refer to [link to relevant documentation]. Figure 8 The diagram shown is a schematic representation of determining the azimuth angle of a wafer according to an embodiment of the present invention. Figure 8In the illustrated embodiment, P1 represents the center of the wafer, with a center position of (x1, y1), and P2 represents the geometric feature point of the wafer's notch, with a feature point position of (x2, y2). The reference axis in this embodiment is horizontal, and correspondingly, the azimuth angle of the wafer is θ. Using trigonometric functions, θ can be calculated based on the center position (x1, y1) and the feature point position (x2, y2). In a specific embodiment, θ = atan2(y2-y1, x2-x1). The relative rotation of the two wafers can be controlled using the azimuth angle difference. Preferably, one wafer is rotated based on the azimuth difference until the azimuth angle difference between the two wafers meets a preset azimuth angle difference requirement. This preset azimuth angle difference requirement may include, for example, that the azimuth angle difference between the two wafers is less than or equal to a preset azimuth angle difference threshold. For example, after controlling the rotation of the two wafers until the azimuth angle difference meets the preset azimuth angle difference requirement, the center positions and feature point positions of the two wafers can be re-acquired. In some embodiments, the relative translation of two wafers can be controlled based on the positional deviation of their center positions. In other embodiments, the relative translation can be controlled based on the positional deviation of their feature point positions. In still other embodiments, the relative translation can be controlled based on the average of the center position deviation and the feature point position deviation. Preferably, the relative translation of the two wafers can be achieved by controlling the movement of one of the wafers. After the relative translation, the positional deviation between the center positions and / or the positional deviation between the feature point positions of the two wafers must meet a preset positional deviation requirement. This preset positional deviation requirement may include, for example, that the distance between the centers of the two wafers is less than or equal to a preset center distance threshold, and / or that the distance between the geometric feature points of the positioning features of the two wafers is less than or equal to a preset feature point distance threshold.

[0106] The above technical solution adopts a rotation-then-translation alignment method, which can decouple the rotation and translation movements, and help avoid the disturbance of the center position caused by the rotation. By combining the position deviation of the center position and / or feature points as alignment criteria, the alignment stability can be improved, the angle and position deviation can be reduced, and the wafer alignment accuracy can be improved.

[0107] Optionally, the wafer alignment method is applied to a wafer bonding apparatus, which includes: a first stage for holding one of two wafers during wafer bonding; a second stage for holding the other of the two wafers during wafer bonding; at least two first image acquisition devices for acquiring at least two arc edge images of each of the two wafers; and a second image acquisition device for acquiring a positioning feature image of each of the two wafers. The wafer alignment method includes: acquiring arc edge images of each of the two wafers by the first image acquisition devices; determining the center position of the corresponding wafer's center based on at least two first fitted tangents corresponding one-to-one with the arc edges contained in each of the at least two arc edge images; acquiring positioning feature images of each of the two wafers by the second image acquisition devices; determining the feature point position of the geometric feature point of the positioning feature of the corresponding wafer based on each of the two positioning feature images; and driving the first stage and the second stage to move relative to each other based on the center position and the feature point position of each of the two wafers to align the two wafers.

[0108] For example, the first stage can be an upper stage, and the second stage can be a lower stage. One of the two wafers is the upper wafer, and the other is the lower wafer. After the upper wafer is placed on the lower stage, each image acquisition device can acquire images of the upper wafer. Then, the upper stage can pick up the upper wafer by adsorption / clamping. The upper stage holds the upper wafer, and at this time, the projection position of the upper wafer onto the plane of the lower stage is consistent with the placement position of the upper wafer on the lower stage during image acquisition. Then, the lower wafer can be placed on the lower stage, and each image acquisition device can acquire images of the lower wafer. Using the wafer alignment method of the aforementioned embodiment, the alignment of the two wafers can be achieved by driving the relative movement of the two stages. It can be understood that the upper stage has a motion dimension of moving along the vertical axis (called the z-axis). In addition, there are three motion dimensions: a motion dimension of moving along two axes (denoted as the x-axis and y-axis) in the horizontal plane, and a motion dimension of rotating around the z-axis, which are achieved by the cooperation of the two stages. For example, the download station can handle only the three motion dimensions, or the download station can handle some of the motion dimensions, while the upload station handles the other motion dimensions.

[0109] Please see Figure 9The diagram shows a layout of a download station and an image acquisition device according to an embodiment of the present invention. The central circular area represents the wafer, which is supported by a wafer chuck. A notch for orientation positioning is provided at the bottom of the wafer, and the area where the notch is located is denoted as the Notch area. In this embodiment, three first image acquisition devices and one second image acquisition device are configured, each being a high-precision, small-field-of-view visible light camera. The three first image acquisition devices are located above, to the left, and to the right of the download station, respectively, while the second image acquisition device is located below the download station. It should be noted that "above," "below," "left," and "right" here refer to the horizontal viewpoint, not the vertical height direction in space. The top camera faces the top of the wafer, with its field of view covering the upper arc edge (i.e., the upper edge arc segment) of the wafer; the bottom camera faces the bottom of the wafer, with its field of view covering the Notch area at the bottom of the wafer, used for observing and locating the notch; the left camera faces the left side of the wafer, with its field of view covering the left arc edge of the wafer; and the right camera faces the right side of the wafer, with its field of view covering the right arc edge of the wafer.

[0110] The above-mentioned technical solution determines the center position of the wafer's center by using the first fitted tangent line based on at least two partially non-overlapping arc edge images on the wafer. This effectively overcomes the problem of insufficient stability in three-point centering and improves the accuracy of wafer centering. Furthermore, this solution obtains the center position of the wafer and the feature point positions of the geometric feature points of the positioning features to control the relative movement of the two wafers to achieve alignment. This helps to achieve both center alignment and azimuth alignment of the wafers, thus taking into account the alignment requirements of both translation and rotation. The overall alignment accuracy of the two wafers is high.

[0111] Optionally, before acquiring at least two arc edge images for the arc edge of the wafer, the method further includes: acquiring, for each of the at least two first image acquisition devices and the second image acquisition device, a target image acquired by the image acquisition device for a corresponding target placed on a calibration piece on a second stage, wherein at least two first targets on the calibration piece on the second stage are respectively located within the field of view of the at least two first image acquisition devices, and a second target on the calibration piece on the second stage is located within the field of view of the second image acquisition device; controlling the second stage to move according to a preset 3×3 grid, and during the control of the second stage to move, acquiring, for each of the at least two first image acquisition devices and the second image acquisition device, nine target images acquired by the image acquisition device for a target within the corresponding field of view, wherein the physical positions of the targets contained in each of the nine target images are different; constructing a two-dimensional affine transformation matrix based on the image positions of the targets contained in each of the nine target images and the physical positions of the targets contained in each of the nine target images, the two-dimensional affine transformation matrix being used to transform the image into a target image. The image position in the image coordinate system used by the acquisition device is converted into the physical position in the planar physical coordinate system used by the second platform; the center position of the corresponding wafer is determined according to at least two first fitted tangents corresponding one-to-one with the arc edges contained in each of the at least two arc edge images, including: using a two-dimensional affine transformation matrix to transform the tangent image positions of the at least two first fitted tangents in the image coordinate system of their respective arc edge images into tangent physical positions in the planar physical coordinate system; the center position of the corresponding wafer is determined according to the tangent physical positions of the at least two first fitted tangents in the planar physical coordinate system; the feature point positions of the geometric feature points of the positioning features of the corresponding wafer are determined based on each of the two positioning feature images, including: determining the feature point positions of the geometric feature points of the two wafers in the image coordinate system used by their respective positioning feature images based on the two positioning feature images respectively; the feature point positions of the geometric feature points in the image coordinate system used by their respective positioning feature images are converted into feature point positions in the planar physical coordinate system using a two-dimensional affine transformation matrix.

[0112] For example, a two-dimensional affine transformation matrix for each image acquisition device can be constructed using a nine-point calibration method. The calibration plate can be a dedicated calibration disc or calibration plate. At least two first targets for calibrating at least two first image acquisition devices and a second target for calibrating a second image acquisition device can be pre-fabricated on the calibration plate. See also... Figure 10 As shown, it is a schematic diagram of the target arrangement of a calibration plate according to an embodiment of the present invention. Figure 10 The target arrangement of the calibration plate shown is similar to... Figure 9 The layout of each image acquisition device shown is compatible. Figure 10 The targets include those used for calibration Figure 9 The upper target of the camera above is used for calibration. Figure 9 The lower target of the camera is used for calibration. Figure 9 The left target of the left camera, and the target used for calibration. Figure 9 The right target of the right-side camera. In this embodiment, the first target includes an upper target, a left target, and a right target, and the second target is a lower target. Before calibration, the calibration sheet can be adsorbed or clamped onto the Chuck, so that it moves synchronously with the second stage. Within their respective calibration strokes, each target can fall into the field of view of the corresponding camera and be clearly imaged.

[0113] Please see Figure 11 The diagram shown is a schematic representation of a nine-point calibration according to an embodiment of the present invention. Figure 9 and Figure 10 Taking the upper camera and upper target as an example in the corresponding embodiment, the second stage can be driven to move according to a preset 3×3 grid. Correspondingly, the calibration plate can move with the second stage, so that the upper target appears at nine different pixel positions (i.e., image positions) within the field of view of the upper camera. For the k-th grid point of the 3×3 grid, k=1,2,…,9: when the second stage stops at the k-th grid point, the machine tool encoder is read to obtain the stage coordinates (T). x,k ,T y,k The platform coordinates can be directly used as the world coordinates of the upper target. Using the upper camera, images of the upper target on the calibration plate can be acquired, obtaining the target image corresponding to the k-th grid point. By identifying the upper target in the target image, the pixel coordinates of the upper target's center can be obtained ((Row...)). k Col k It can be understood that each camera can acquire nine target images, and correspondingly, the pixel coordinates of the centers of the nine targets can be obtained. Based on the one-to-one mapping relationship between the pixel coordinates of the centers of the nine targets and the world coordinates of the nine targets, a two-dimensional affine transformation matrix can be solved to establish the transformation relationship from the image coordinate system used by the camera to the planar physical coordinate system used by the second platform. It can be understood that using this two-dimensional affine transformation matrix, the three first fitted tangent lines can be transformed from image space to physical space, and correspondingly, the position of the wafer's center in the planar physical coordinate system can be obtained. Similarly, using this two-dimensional affine transformation matrix, the feature point positions of geometric feature points can be transformed from image positions to physical positions.

[0114] The above technical solution controls the movement of the second stage to drive the target to move according to a 3×3 grid, thereby constructing a two-dimensional affine transformation matrix. This can uniformly transform the pixel positions in the image coordinate system used by each image acquisition device to the planar physical coordinate system of the second stage, thereby unifying the coordinate reference of each image acquisition device and improving the wafer alignment accuracy.

[0115] According to another aspect of the present invention, a wafer bonding method is also provided, comprising: aligning two wafers according to the wafer alignment method described above; and bonding the aligned two wafers together.

[0116] According to another aspect of the present invention, a wafer bonding apparatus is also provided for bonding two wafers, the wafers having arc edges and positioning features. The apparatus includes: a first stage for holding one of the two wafers during wafer bonding; a second stage for holding the other of the two wafers during wafer bonding; at least two first image acquisition devices for acquiring at least two arc edge images of each of the two wafers, wherein the arc edges contained in each of the at least two arc edge images are at least partially non-overlapping on the wafer; a second image acquisition device for acquiring a positioning feature image of each of the two wafers; a processor; and a memory storing computer program instructions, wherein the processor reads and executes the computer program instructions to implement the wafer alignment method described above, and / or the wafer bonding method described above.

[0117] The logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to perform the above-described wafer alignment method and / or the above-described wafer bonding method.

[0119] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the above-described wafer alignment method and / or the above-described wafer bonding method.

[0120] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0121] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A wafer alignment method, characterized in that, The method includes: For each of the two wafers to be aligned, acquire at least two arc edge images for the arc edge of the wafer, wherein the arc edges contained in each of the at least two arc edge images do not overlap at least partially on the wafer; The center position of the corresponding wafer is determined based on at least two first fitted tangent lines that correspond one-to-one with the arc edges contained in each of the at least two arc edge images. Acquire two positioning feature images that correspond one-to-one with the positioning features of the two wafers; Based on each of the two positioning feature images, the feature point position of the geometric feature point of the positioning feature of the corresponding wafer is determined; Based on the center positions and feature point positions of the two wafers, the relative movement of the two wafers is controlled to align the two wafers.

2. The method according to claim 1, characterized in that, The number of arc edge images for each of the two wafers is three, and the number of first fitted tangents is three. The step of determining the center position of the wafer's center based on at least two first fitted tangents that correspond one-to-one with the arc edges contained in each of the at least two arc edge images includes: For each of the two wafers, the center position of the wafer is determined by the least squares method based on the preset wafer reference radius and the linear equations of the three first fitted tangents of the wafer.

3. The method according to claim 1, characterized in that, Determining the center position of the wafer based on at least two first fitted tangent lines corresponding one-to-one with the arc edges contained in each of the at least two arc edge images includes: Extract multiple edge sampling points on the arc edges contained in each of the at least two arc edge images; Straight line fitting is performed on the plurality of edge sampling points contained in each of the at least two circular arc edge images to obtain at least two first fitting tangent lines that correspond one-to-one with the at least two circular arc edge images; The center position of the wafer is determined based on the at least two first fitted tangents.

4. The method according to claim 3, characterized in that, The step of extracting multiple edge sampling points on the arc edges contained in each of the at least two arc edge images includes: For each of the at least two arc edge images, multiple caliper detection positions of the first measuring caliper are determined on the arc edge contained in the arc edge image, and multiple scanning areas are generated based on the multiple caliper detection positions. According to the preset sampling point extraction rules, a first sub-pixel edge detection operation is performed along the normal of the first measuring caliper in the multiple scanning areas to determine the multiple edge sampling points on the arc edge contained in the arc edge image.

5. The method according to claim 4, characterized in that, The sampling point extraction rules include edge transition polarity, edge amplitude threshold, and edge selection strategy. The edge transition polarity is used to indicate the brightness change direction that the edge sampling point needs to conform to. The edge amplitude threshold is used to indicate the minimum grayscale jump intensity that the edge sampling point needs to conform to. The edge selection strategy is used to indicate the selection strategy of choosing a unique edge sampling point from at least two edge sampling points when the number of edge sampling points identified in any scanning area is at least two.

6. The method according to claim 3, characterized in that, The step of performing linear fitting on the plurality of edge sampling points contained in each of the at least two circular arc edge images to obtain at least two first fitting tangent lines corresponding one-to-one with the at least two circular arc edge images includes: A first straight line fitting operation is performed on the plurality of edge sampling points contained in each of the at least two circular arc edge images to obtain the initial fitted straight line for each of the at least two circular arc edge images; For each of the at least two circular arc edge images, a first distance is determined between the plurality of edge sampling points of the circular arc edge image and the initial fitted straight line of the circular arc edge image; Edge sampling points whose first distance is greater than the first preset distance are removed, and a second straight line fitting operation is performed on the remaining edge sampling points after removal to obtain the first fitted tangent line corresponding to the arc edge image.

7. The method according to claim 1, characterized in that, The positioning feature includes a notch formed after the wafer is processed to determine the orientation of the wafer. The step of determining the feature point position of the geometric feature point of the positioning feature corresponding to the wafer based on each of the two positioning feature images includes: For each of the two positioning feature images, two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image are determined. Each set of contour sampling points includes multiple contour sampling points, and the two sets of contour sampling points are located on both sides of the cut vertex on the cut contour. By performing linear fitting on the two sets of contour sampling points, two second fitted tangent lines of the cut contour are obtained; The feature point position of the geometric feature point of the cut is determined based on the two second fitting tangents and the second preset distance between the pre-configured geometric feature point and the two second fitting tangents; Wherein, the two second fitted tangents of one of the two positioning feature images correspond one-to-one with the two second fitted tangents of the other positioning feature image, and the distance between the geometric feature point of one of the two positioning feature images and any second fitted tangent of its corresponding positioning feature image is equal to the distance between the geometric feature point of the other positioning feature image and the corresponding second fitted tangent of its corresponding positioning feature image.

8. The method according to claim 7, characterized in that, The determination of two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: A threshold segmentation operation is performed on the localization feature image to obtain the cut region of the localization feature image, wherein the cut region is the image region where the cut is located; Determine the minimum bounding rectangle of the cut area, the minimum bounding rectangle having a first direction axis extending along the cut depth direction of the cut area and a second direction axis extending along the cut width direction of the cut area; The positions of the two second measuring calipers are determined based on the center of the minimum bounding rectangle, the axis of the first direction axis, and the axis of the second direction axis. The scanning areas of the two measuring calipers can cover part of the edge sampling points on the cut contour. The two second measuring calipers are located on both sides of the center of the rectangle. The angles between the two second measuring calipers and the direction axis are the corresponding preset angles. A second subpixel edge detection operation is performed along the normal of each of the two second measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

9. The method according to claim 7, characterized in that, The determination of two sets of contour sampling points located at specific positions on the cut contour of the positioning feature image includes: The cut contour of the positioning feature image is determined by template matching with at least a portion of the image region using a preset cut contour template. The cut contour template can indicate the shape of the cut contour, and the at least a portion of the image region includes the image region where the cut is located. Based on the relative positional relationship between the pre-configured measuring calipers and the cut contour template, the positions of the two third measuring calipers of the positioning feature image are determined; A third subpixel edge detection operation is performed along the normal of each of the two third measuring calipers to determine the two sets of contour sampling points of the positioning feature image.

10. The method according to claim 7, characterized in that, The step of determining the feature point position of the geometric feature point of the cut based on the two second fitted tangents and the second preset distance between the pre-configured geometric feature point and the two second fitted tangents includes: Determine the angle bisector of the target angle formed by the two sub-tangents of each of the two second fitted tangents, wherein the target angle is the angle formed by the two sub-tangents located on the same side of the intersection point formed by the two second fitted tangents; The position of the point on the angle bisector that is located on the side away from the corresponding wafer and whose distance from the two second fitted tangents is the second preset distance is determined as the feature point position of the geometric feature point.

11. The method according to claim 10, characterized in that, The step of determining the position of the point on the angle bisector that is at a distance of the second preset distance from the two second fitted tangents as the geometric feature point includes: Based on the half-angle value of the target included angle and the second preset distance, the distance between the geometric feature point located on the angle bisector and the vertex is calculated using trigonometric functions to determine the feature point position of the geometric feature point.

12. The method according to claim 7, characterized in that, The step of performing linear fitting on the two sets of contour sampling points to obtain two second fitted tangent lines for the cut contour includes: A third line fitting operation is performed on the two sets of contour sampling points respectively to obtain the initial fitted lines for each of the two sets of contour sampling points; For each of the two sets of contour sampling points, determine the second distance between each contour sampling point in the set and the initial fitted line of the set of contour sampling points. Contour sampling points whose second distance is greater than the third preset distance are removed, and a fourth straight line fitting operation is performed on the remaining contour sampling points after removal to obtain the second fitted tangent line corresponding to the set of contour sampling points.

13. The method according to any one of claims 1-12, characterized in that, The step of controlling the relative movement of the two wafers based on their respective center positions and feature point positions to align them includes: The azimuth angle difference between the two wafers is determined based on the center positions of the two wafers and the feature point positions of the two wafers. Control at least one of the two wafers to rotate according to the angle difference until the angle difference between the two wafers meets the preset angle difference requirement; Based on the positional deviation between the center positions and / or the positional deviation between the feature points of the two wafers after rotation, at least one of the two wafers is controlled to move until the positional deviation between the center positions and / or the positional deviation between the feature points of the two wafers meets the preset positional deviation requirements.

14. The method according to any one of claims 1-12, characterized in that, The wafer alignment method is applied to a wafer bonding apparatus, the wafer bonding apparatus comprising: A first stage is used to hold one of the two wafers during wafer bonding; A second stage is used to hold one of the two wafers during wafer bonding; At least two first image acquisition devices are used to acquire images of the at least two arc edges of each of the two wafers; The second image acquisition device is used to acquire the positioning feature image of each of the two wafers. The wafer alignment method includes: The first image acquisition device acquires images of the arc edges of each of the two wafers; The center position of the corresponding wafer is determined based on at least two first fitted tangent lines that correspond one-to-one with the arc edges contained in each of the at least two arc edge images. The second image acquisition device acquires the positioning feature images of each of the two wafers; Based on each of the two positioning feature images, the feature point position of the geometric feature point of the positioning feature of the corresponding wafer is determined; Based on the center positions and feature point positions of the two wafers, the first stage and the second stage are driven to move relative to each other to align the two wafers.

15. The method according to claim 14, characterized in that, Before acquiring at least two arc edge images for the arc edge of the wafer, the method further includes: For each of the at least two first image acquisition devices and the second image acquisition device, a target image is acquired by the image acquisition device for the corresponding target placed on the calibration plate on the second platform. At least two first targets on the calibration plate on the second platform are respectively located within the field of view of the at least two first image acquisition devices, and a second target on the calibration plate on the second platform is located within the field of view of the second image acquisition device. The second platform is controlled to move according to a preset 3×3 grid. During the movement of the second platform, for each of the at least two first image acquisition devices and the second image acquisition device, nine target images are acquired by the image acquisition device for the target in the corresponding field of view. The physical positions of the targets contained in each of the nine target images are different. Based on the image positions of the targets contained in each of the nine target images and the physical positions of the targets contained in each of the nine target images, a two-dimensional affine transformation matrix is ​​constructed. The two-dimensional affine transformation matrix is ​​used to convert the image positions in the image coordinate system adopted by the image acquisition device into the physical positions in the planar physical coordinate system adopted by the second platform. The step of determining the center position of the corresponding wafer's center based on at least two first fitted tangent lines that correspond one-to-one with the arc edges contained in each of the at least two arc edge images includes: Using the two-dimensional affine transformation matrix, the tangent image positions of the at least two first fitted tangents in the image coordinate system of their respective circular arc edge images are transformed into the tangent physical positions in the planar physical coordinate system. Based on the physical positions of the tangents of the at least two first fitted tangents, the center position of the corresponding wafer in the plane physical coordinate system is determined. The step of determining the feature point position of the geometric feature point of the corresponding wafer based on each of the two positioning feature images includes: Based on the two positioning feature images, the positions of the geometric feature points of each of the two wafers in the image coordinate system adopted by the respective positioning feature images are determined; Using the two-dimensional affine transformation matrix, the position of the geometric feature point in the image coordinate system of the corresponding positioning feature image is converted into the position of the feature point in the physical coordinate system of the plane.

16. A wafer bonding method, characterized in that, include: The wafer alignment method according to any one of claims 1-15 aligns the two wafers; The two aligned wafers are then bonded together.

17. A wafer bonding apparatus, characterized in that, For bonding two wafers, the wafers having arc edges and positioning features, the device includes: A first stage is used to hold one of the two wafers during wafer bonding; A second stage is used to hold one of the two wafers during wafer bonding; At least two first image acquisition devices are used to acquire at least two arc edge images of each of the two wafers, wherein the arc edges contained in each of the at least two arc edge images do not overlap at least partially on the wafer; The second image acquisition device is used to acquire the positioning feature image of each of the two wafers; A processor and a memory storing computer program instructions, wherein the processor reads and executes the computer program instructions to implement the wafer alignment method as claimed in any one of claims 1-15, and / or the wafer bonding method as claimed in claim 16.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the wafer alignment method as described in any one of claims 1-15, and / or the wafer bonding method as described in claim 16.

19. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the wafer alignment method as described in any one of claims 1-15, and / or the wafer bonding method as described in claim 16.

Images