Wafer image compensation method and wafer defect detection system

Abstract

This application discloses a wafer image compensation method and a wafer defect detection system, relating to wafer defect detection technology. The method includes: determining the object-side pixel size when an image sensor acquires an image and determining the scanning speed of the stage; determining a reference die image; obtaining the image position deviation corresponding to the real-time die image based on the reference die image and the real-time die image; determining the position deviation sequence in the current scanning path based on the image position deviation and the object-side pixel size; compensating the next scanning path based on the position deviation sequence in the current scanning path to obtain a compensated scanning path, until all multiple scanning paths of the die row to be tested in the wafer under test have been scanned, thus obtaining a wafer compensated image corresponding to the die row to be tested. This application can reduce the position deviation in subsequent image acquisition processes, ultimately reducing the amount of compensation required for image registration interpolation, reducing noise caused by image interpolation, and improving the sensitivity of wafer defect detection.

Description

Technical Field

[0001] This application relates to the field of wafer defect detection technology, and in particular to a wafer image compensation method and a wafer defect detection system. Background Technology

[0002] Bright-field defect detection equipment uses a high-precision motion control system to sequentially transport the pattern on the wafer into the field of view of a high-resolution microscopic imaging system. The high-resolution microscopic imaging system images the pattern on the wafer onto the camera's photosensitive surface. The camera records and quantifies the light intensity distribution falling on the photosensitive surface, ultimately obtaining a digital image corresponding to the pattern. By comparing the digital images formed by patterns at the same location within different grains, abnormal grayscale changes are used to determine which locations have defects.

[0003] During this process, due to the mechanical vibrations of the stage, objective lens, mirrors in the optomechanical structure, and camera, the actual acquired pattern deviates from the desired pattern's position on the wafer. Therefore, the multiple digital images being compared do not strictly originate from the same location within different grains. In this case, even if the pattern itself is defect-free, grayscale changes will occur during comparison. If these grayscale changes are close to those caused by the target defect, the defect will be masked by noise and cannot be detected. To avoid this problem, existing technologies typically perform image registration on the digital images first, then image interpolation, aligning the images before comparison, thereby minimizing the impact of positional deviations.

[0004] However, during digital image registration and interpolation, the image interpolation function cannot reflect the true light intensity distribution, resulting in a difference between the image obtained by interpolation based on the registration position and the image actually acquired by the camera at that position. This difference acts as noise, reducing the sensitivity of defect detection. Moreover, the greater the difference in pixel position during interpolation, the greater the adverse impact on the sensitivity of wafer defect detection. Summary of the Invention

[0005] This application provides a wafer image compensation method and a wafer defect detection system, which can improve the sensitivity of wafer defect detection.

[0006] In a first aspect, this application provides a wafer image compensation method, comprising:

[0007] For a wafer under test placed on a motion stage, the object-side pixel size when the image sensor acquires an image is determined based on the grain distance between adjacent dies in the same grain row of the wafer under test, wherein the grain distance between adjacent dies is an integer multiple of the object-side pixel size; the scanning speed of the motion stage is determined based on the object-side pixel size; the wafer under test includes multiple grain rows, and each grain row includes multiple scanning paths;

[0008] A reference grain row is determined from the plurality of said grain rows, and a reference grain image is determined based on the plurality of reference grains contained in the reference grain row;

[0009] Select the current scanning path, control the motion stage to move according to the scanning speed, and control the image sensor to acquire images of each die according to the object pixel size, so as to obtain the real-time die image corresponding to each die in sequence;

[0010] Based on the reference grain image, image registration is performed on each of the real-time grain images to obtain the image position deviation corresponding to each real-time grain image; the position deviation sequence of the motion stage in the current scanning path is obtained according to the image position deviation corresponding to each real-time grain image and the object pixel size;

[0011] For the next scanning path, the next scanning path is compensated according to the position deviation sequence in the current scanning path to obtain a compensated scanning path; the compensated scanning path is determined as the current scanning path, and the operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size is repeated to obtain the real-time die image corresponding to each die in sequence until all multiple scanning paths of the die row to be tested in the wafer under test are scanned, and the wafer compensation image corresponding to the die row to be tested is obtained.

[0012] Secondly, this application provides a circular defect detection system, the device comprising:

[0013] The defect detection system includes a motion controller, an illumination system, an imaging system, an image sensor, and an image processing system; the image processing system is connected to the motion controller, the illumination system, the imaging system, and the image sensor, respectively.

[0014] The motion controller is used to control the movement of the motion stage; the illumination system is used to output a light beam so that the light beam is focused onto the surface of the wafer under test through the objective lens; the imaging system is used to receive the reflected light from the wafer under test returned through the objective lens, and the imaging system is also used to adjust the imaging magnification to adjust the object-side pixel size when the image sensor acquires the wafer image of the wafer under test; the image sensor is used to acquire images of each die and transmit the acquired wafer images to the image processing system; the image processing system implements the wafer image compensation method as described in any embodiment of the present invention, obtains a compensated image of the wafer under test, and performs defect detection on the wafer under test based on the compensated image.

[0015] The wafer image compensation scheme and circular defect detection system provided in this application predetermine the object-space pixel size when the image sensor acquires images, and the scanning speed of the stage is determined based on the object-space pixel size before the image sensor acquires images based on the object-space pixel size. Furthermore, the distance between adjacent dies is limited to an integer multiple of the object-space pixel size. This ensures that when acquiring images based on the currently set object-space pixel size, the pixel positions on the camera are strictly consistent when the image sensor acquires images of the same position on different dies, thereby avoiding pixel position offsets. To reduce noise introduced by image registration and interpolation, the detection sensitivity is improved. Furthermore, during the process of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size, the next scanning path can be modified based on the positional deviation sequence obtained from the current scanning path. This reduces positional deviations in subsequent image acquisition caused by process errors of devices in the detection system, thermal expansion and contraction of the wafer due to temperature changes, and process deviations in the position of the dies on the wafer. It also reduces errors caused by process or temperature and noise introduced by image interpolation, improves the quality of acquired images, and thus increases the sensitivity of wafer defect detection.

[0016] The beneficial effects described in the second aspect of this application can be referred to in the analysis of the beneficial effects of the first aspect, and will not be repeated here.

[0017] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent from the following description.

[0018] It is understood that before using the technical solutions disclosed in the various embodiments of this application, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this application in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic flowchart of a wafer image compensation method provided in an embodiment of this application;

[0021] Figure 2 This is a schematic diagram of the hardware architecture of the wafer defect detection system provided in this application;

[0022] Figure 3 This is a schematic diagram of a motion table scanning path provided in an embodiment of this application;

[0023] Figure 4 This is provided by the embodiments of this application. Figure 3 A magnified view of a row of grains in the middle row;

[0024] Figure 5 This is an illustrative diagram illustrating the impact of image interpolation on defect detection provided in an embodiment of this application;

[0025] Figure 6 This is a schematic diagram of an image scanning strategy provided in an embodiment of this application;

[0026] Figure 7 This is a schematic diagram illustrating the defined pixel position provided in an embodiment of this application;

[0027] Figure 8 This is a schematic diagram of determining a reference grain row provided in an embodiment of this application;

[0028] Figure 9 This is a schematic flowchart illustrating the process of determining image position deviation provided in an embodiment of this application;

[0029] Figure 10 This is a schematic diagram of a reference grain image block provided in an embodiment of this application;

[0030] Figure 11 This is a schematic diagram of a real-time grain image block provided in an embodiment of this application;

[0031] Figure 12 This is a flowchart illustrating the process of determining a position deviation sequence provided in an embodiment of this application;

[0032] Figure 13 This is another flowchart illustrating the determination of the position deviation sequence provided in an embodiment of this application;

[0033] Figure 14 This is another hardware architecture diagram of the wafer defect detection system provided in the embodiments of this application. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present application, the technical solutions of the present application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0035] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present application, not the entire structure.

[0037] Figure 1 This is a schematic flowchart of a wafer image compensation method provided in an embodiment of this application. This embodiment can be applied to the case of wafer defect detection based on the wafer image compensation method provided in this embodiment. The method can be executed by an image processing system in a wafer defect detection system. The image processing system can be implemented in hardware and / or software and integrated into the wafer defect detection system that executes this method.

[0038] Before introducing the wafer image compensation scheme provided in this embodiment, please refer to Figure 2. Figure 2 This is a schematic diagram of the hardware architecture of the wafer defect detection system provided in this application.

[0039] Generally, based on Figure 2The provided wafer defect detection system places the wafer under test above a motion stage during defect detection. The motion stage can be adjusted in both the X and Y directions. A motion stage controller moves the wafer to bring patterns at different coordinate positions below the objective lens. The illumination system outputs a beam that passes sequentially through a reflector, beam splitter, and objective lens to converge onto the wafer pattern at the current coordinate position. The reflected light is then used by an imaging system to acquire an image of the wafer. Finally, the wafer image is transmitted to an image processing system where image registration and interpolation are performed. Defect detection is then performed based on the processed differential image.

[0040] Further, please refer to Figure 3 , Figure 3 This is a schematic diagram of a motion table scanning path provided in an embodiment of this application. During defect detection, the movement trajectory of the motion table will... Figure 3 The dashed line representing the scanning path serves as the movement path, and the arrow direction indicates the movement direction. As the stage moves along each scanning path, the stage's Y-axis position remains unchanged during the scanning process. After the current scanning path is completed, the stage's Y-axis position moves downwards, while the X-axis scanning direction changes. During each scanning path, the patterns of the chips along the same path are sequentially recorded by the image sensor to form a digital image.

[0041] For further details, please refer to [link / reference]. Figure 4 , Figure 4 This is provided by the embodiments of this application. Figure 3 A magnified view of a section of a row of grains. Figure 4 China and Israel Figure 3 The example uses the first grain row as an example. The first grain row represents the first row of grains among the multiple rows of grains that make up a wafer. The first grain row includes three grains: grain 1, grain 2, and grain 3. Grains 1, 2, and 3 within the first grain row contain multiple scan paths, as shown in the example with four scan paths: scan path 1, scan path 2, scan path 3, and scan path 4. During the actual scanning process, the stage scans each scan path sequentially until a grain row is scanned, thus obtaining the scanned image corresponding to that grain row.

[0042] Specifically, please refer to Figure 5 , Figure 5 This is an illustrative diagram illustrating the impact of image interpolation provided in this application on defect detection. Generally, the system selects images of the same location on multiple grains, registers them, and determines a reference image by taking the median. Figure 5Taking the "H" pattern as an example, in reference image 301, each square 304 represents a pixel, and the color depth represents the grayscale value of that pixel, with white being 0 grayscale and pure black being 255 grayscale. Candidate image 302 is a pattern with a defect, and its pixel position is exactly the same as the reference image during image acquisition. For both the reference and candidate images, a difference image 303 is determined. In difference image 303, except for the defect location which has a grayscale difference of 255, the other locations are 0 grayscale. Therefore, based on this image, we can easily determine whether a defect exists and pinpoint its precise location.

[0043] But Figure 3 In the scan path shown, the X-stage continuously scans one row of grains. The pixel position of candidate image 302 on the grain in the X direction is actually arbitrary. The acquired candidate... Figure X Taking a pixel position offset of 0.5 pixels as an example (i.e., candidate image 305); in this case, candidate image 305 generally needs to be registered and interpolated to obtain registered image 306, and then compared with reference image 301 to obtain difference image 307. However, through registration and interpolation, the obtained registered image 306 cannot completely restore the state of candidate image 302, and difference image 307 also has large grayscale differences in non-defect positions. These noises interfere with the judgment of real defects.

[0044] In actual defect detection, if the defect detection system structure is absolutely stable, images can be directly compared without image registration interpolation. However, during actual detection, mechanical vibrations of the equipment (e.g., bending of the guide rails moving in different directions), temperature drift (thermal expansion and contraction of the guide rails and wafers due to temperature changes), and product process errors (e.g., process errors in the position of the die on the wafer) can interfere with the actual position of the captured pattern on the wafer, causing a deviation between the pattern position acquired by the equipment and the theoretically acquired position at the current moment. To reduce this error, a higher precision motion stage and a more stable machine environment can generally be developed. However, as the detection size continues to shrink, the hardware development cost increases dramatically and the development difficulty becomes extremely high.

[0045] In view of this, this embodiment provides a wafer image compensation method for a defect detection system to avoid noise caused by image registration interpolation due to pixel position offset, thereby improving detection sensitivity.

[0046] For details, please refer to Figure 1 The wafer image compensation method in this embodiment includes, but is not limited to, the following steps:

[0047] S110. For the wafer to be tested placed on the motion stage, determine the object-side pixel size when the image sensor acquires the image based on the grain distance between adjacent grains in the same grain row of the wafer to be tested.

[0048] Please continue to refer to Figure 4 ,exist Figure 4 In the example, the positions of grain 1, grain 2, and grain 3 represent the same grain row. Grain 1 and grain 2, or grain 2 and grain 3, represent adjacent grains. The grain distance between adjacent grains can be understood as the period (DiePitchX) of adjacent grains in the x-direction, and the current grain distance can be a pre-stored parameter of the wafer under test. For details, please refer to... Figure 6 , Figure 6 This is a schematic diagram of an image scanning strategy provided in an embodiment of this application. The two "H"s in the diagram can represent images of the pattern in grain 1 and the pattern in grain 2 obtained during scanning the first grain row, respectively. Without considering wafer deformation, the grain distance can be determined from the images of the two grains acquired by the image sensor, such as... Figure 6 As shown, the distance between grain 1 and grain 2, or the distance between the two "H"s, is DiePitchX.

[0049] Specifically, in this embodiment, the object-side pixel size is determined when the image sensor acquires an image. This can be achieved by setting the object-side pixel size based on the grain distance between adjacent dies, ensuring that the grain distance between adjacent dies is an integer multiple of the object-side pixel size; then, the scanning speed of the motion stage is determined based on the object-side pixel size. It should be noted that when a pattern of a certain size on the object surface is imaged onto the image sensor and just covers one pixel of the image sensor, this size on the object surface can be referred to as the object-side pixel size.

[0050] For example, in this application, when a pattern of a certain size in the die is imaged onto the image sensor and just covers one pixel of the image sensor, this size in the die is called the object-side pixel size, which is usually denoted as PixelSize. Alternatively, a pattern of a certain size on the object surface can be imaged onto the image sensor via an imaging system included in the detection system, which is not limited here.

[0051] In this embodiment, the object-side pixel size (PixelSize) of the image sensor during image acquisition is changed by adjusting the magnification of the variable magnification module in the imaging system. The object-side pixel size acquired by the image sensor includes the pixel size in the X direction and the pixel size in the Y direction, and the pixel size in the X direction is the same as the pixel size in the Y direction. To ensure consistency, the object-side pixel size mentioned throughout this document can be the pixel size in the X direction, denoted as PixelSizeX, such that the modified object-side pixel size satisfies DiePitchX / PixelSizeX (the distance between adjacent dies divided by the object-side pixel size in the X direction) as an integer. The purpose of this is that, theoretically, when the image sensor is based on... Figure 3When the scanning path shown acquires images along the X direction, because images of the same position within different grains are spaced an integer number of pixels apart in the X direction, the X pixel positions of the two images are the same; and because the Y position remains unchanged, the Y pixel positions are also the same. Therefore, image registration interpolation of the candidate images is unnecessary for direct comparison. According to the method provided in this embodiment, "setting the object-side pixel size based on the grain distance between adjacent grains so that the grain distance between adjacent grains is an integer multiple of the object-side pixel size," controlling the image sensor to acquire images can avoid noise caused by image registration interpolation and improve detection sensitivity.

[0052] In a preferred embodiment, the wafer image compensation scheme provided in this embodiment determines the scanning speed of the motion stage based on the object pixel size, including: determining the scanning speed of the motion stage based on the object pixel size and the line frequency of the image sensor.

[0053] When modifying the object-side pixel size during image acquisition by the image sensor, the scanning speed of the stage must be modified accordingly. This is because the scanning speed (V) and object-side pixel size are linearly related during image acquisition. Specifically, this relationship can be expressed as V = PixelSizeX * LR, where LR represents the line frequency of the image sensor, and PixelSizeX is the object-side pixel size in the X direction. Since the line frequency of the image sensor is fixed, the scanning speed of the stage can be determined based on the modified object-side pixel size, thus controlling the stage's scanning speed and the object-side pixel size during image acquisition to be modified synchronously.

[0054] Please refer to Figure 6 and Figure 7 , Figure 7 This is a schematic diagram illustrating the defined pixel positions provided in an embodiment of this application. For example... Figure 6 As shown, the distance between the image of the pattern set on the die and the starting position of the die is DistX. In the finally acquired image, the position of pixel X is ( Figure 7 As shown, it is maintained as: DistX%PixelSizeX (the remainder of DistX and the object-side pixel size in the X direction). In this way, the images from the same location on the die used for comparison are kept consistent in the acquisition pixel position, thus avoiding noise caused by image registration interpolation due to pixel position offset.

[0055] In the above embodiments, the image sensor can be a delay integration camera (TDI).

[0056] In this embodiment, the wafer under test includes multiple grain rows, and each grain row includes multiple scan paths. Specifically, this can be explained as being based on... Figure 3 The provided schematic diagram of the wafer under test can include five rows of dies. The first row includes three dies, the second row includes five dies, the third row includes seven dies, the fourth row includes five dies, and the fifth row includes three dies, and so on. Within each row, multiple scanning paths can be formed by different dies. During wafer inspection, the image sensor is controlled to acquire images based on a determined object-side pixel size, and the motion stage is controlled to move along the scanning path at a determined scanning speed, performing scans line by line to execute the wafer image compensation scheme provided in this embodiment during the scanning process.

[0057] S120. Determine a reference grain row from multiple grain rows, and determine a reference grain image based on the multiple reference grains contained in the reference grain row.

[0058] The purpose of determining the reference grain image based on multiple reference grains contained in the reference grain row is to provide a unified standard for determining the image position deviation corresponding to the real-time grain image in subsequent steps. This facilitates compensation for the subsequent scanning path based on the image position deviation. By scanning the wafer under test according to the compensated scanning path, the positional accuracy of image acquisition and the detection sensitivity can be improved.

[0059] In this embodiment, the reference grain row can be determined by selecting the grain row located at the center of the entire grain distribution area as the reference grain row; or by observing the shape, size and other morphological characteristics of the grains in each grain row and selecting the row with the most regular and consistent grain morphology as the reference grain row; the specific method of determining the reference grain row is not limited here.

[0060] In a preferred embodiment, the wafer image compensation scheme provided in this embodiment, which determines the reference grain row and the reference grain image based on the multiple reference grains contained in the reference grain row, can be implemented in the following way:

[0061] Obtain the grain identifier corresponding to each grain, and determine the row of the grain whose grain identifier is a preset identifier as the reference grain row; obtain the image corresponding to each grain in the reference grain row, process the image corresponding to each grain to obtain the grain grayscale image of each grain; determine the reference grain image based on the grayscale value in the grayscale image of each grain.

[0062] In this embodiment, the grain identifier can be represented in the form of coordinates (x, y) to indicate the location of each grain. Figure 3In the example, taking the three grains contained in the first grain row as an example, the grain identifiers corresponding to each grain can be (1,1), (1,2), and (1,3) respectively. Among them, the horizontal coordinate in each grain identifier can be used to represent the row identifier, so that the row containing the grain with the preset identifier in the grain identifier can be determined as the reference grain row.

[0063] Optionally, the preset identifier in this embodiment can be the row identifier where the wafer under test contains the most grains. Specifically, please refer to... Figure 8 , Figure 8 This is a schematic diagram illustrating the determination of a reference grain row according to an embodiment of this application. In the example of Figure 3, after determining the grain identifier for each grain, the grain row containing the most grains can be determined based on the vertical coordinate corresponding to each grain identifier. For example, analysis of the identifiers corresponding to each grain reveals that the Nth grain row contains M grains, where each grain identifier in the Nth row is (N,1), (N,2), ..., (N,M), and M is the maximum value of the vertical coordinates corresponding to all grain identifiers. Therefore, the row containing the grain with the horizontal coordinate N can be determined as the reference grain row. In this embodiment, the reason for determining the row identifier containing the most grains using a preset identifier is that the longest row can guarantee the inclusion of more grains, resulting in a higher fault tolerance. Furthermore, when the wafer is placed on the motion stage, the deformation of the grains in the middle row is smaller, making the reference grain image obtained using the grains in the middle row more accurate. The preset identifier is not limited to the row identifier where the most grains are located. It can also be the grain identifier corresponding to multiple grains in the middle region of the wafer under test. The specific method of selecting reference grains based on the preset identifier is not limited here.

[0064] For each grain in the reference grain row, a digital image of each reference grain is recorded. Then, the images of different reference grains are aligned through image registration and image interpolation. Finally, the median gray value of the corresponding position of different reference grains is calculated to obtain the reference grain image, and the reference grain image is saved in the image processing system.

[0065] S130. Select the current scanning path, control the motion stage to move according to the scanning speed, and control the image sensor to acquire images of each die according to the object pixel size, so as to obtain the real-time die image corresponding to each die in sequence.

[0066] Before selecting the current scan path, an initial scan grain row is first determined so that each scan path can be sequentially scanned within the selected initial scan grain row. The initial scan grain row can be selected based on... Figure 3The example scan planning path selects the topmost first die row as the initial scan die row, and controls the stage to scan sequentially from top to bottom along the scan path contained in the first die row; alternatively, the initial scan die row can also be selected in the planning... Figure 3 When scanning the entire wafer as shown, determine the intermediate grain rows (e.g.) Figure 3 The third grain row in the process is used as the initial scan grain row. After the scan path contained in the middle grain row is completed, the next grain row is scanned upwards until the path in the upper half of the wafer is completed. Then, the process returns to the middle and scans the path contained in the lower half of the wafer from top to bottom. The specific method for determining the initial scan grain row is not limited here.

[0067] For any given row of grains in the current scanning path, the motion stage is controlled to move according to the scanning speed determined in step S110. Simultaneously, the image sensor is controlled to acquire images of each grain in the current scanning path based on the object-side pixel size. Thus, after scanning a grain, the image processing system can obtain a real-time grain image. This can be achieved by the image processing system obtaining the real-time grain image according to the image pixel rows. In other words, when the image sensor acquires images of each grain, it sends the acquired images to the image processing system according to the preset pixel rows of the acquired images. For example, it can send the images line by line or multiple lines. In this way, the image processing system obtains the pixel rows in real time, ultimately forming a real-time grain image. Furthermore, all grains included in the current scanning path are scanned, or the real-time grain image corresponding to each grain is obtained sequentially.

[0068] S140. Based on the reference grain image, perform image registration on each real-time grain image to obtain the image position deviation corresponding to each real-time grain image; obtain the position deviation sequence of the motion stage in the current scanning path according to the image position deviation corresponding to each real-time grain image and the object pixel size.

[0069] In an image processing system, after receiving each real-time grain image, an image registration operation can be performed on the current real-time grain image based on a pre-stored reference grain image.

[0070] For a preferred implementation, please refer to [link / reference]. Figure 9 , Figure 9 This is a flowchart illustrating the process of determining image position deviation according to an embodiment of this application. The implementation method provided by this application for "performing image registration for each real-time grain image based on a reference grain image to obtain the image position deviation corresponding to each real-time grain image" includes, but is not limited to, the following steps S141 to S143:

[0071] S141. Determine the preset image segmentation period, and divide the real-time grain image and the reference grain image into the same number of real-time grain image blocks and reference grain image blocks according to the preset image segmentation period.

[0072] Please refer to the following in the current step. Figure 10 and Figure 11 ,in, Figure 10 This is a schematic diagram of a reference grain image block provided in an embodiment of this application; Figure 11 This is a schematic diagram of a real-time grain image block provided in an embodiment of this application. In the image processing system, the reference grain image is pre-divided into N reference grain image blocks based on a preset image segmentation period, such as... Figure 10 In the context of Frame1, Frame2, Frame3, ..., FrameN, during the current scanning path, as the image sensor sequentially acquires images of each die based on the object-side pixel size, a real-time die image block is obtained when an image corresponding to a preset image segmentation period is acquired. After scanning one die, multiple real-time die image blocks can be obtained, such as... Figure 11 The number and size of the real-time grain image blocks contained in the real-time grain image are the same as the number and size of the reference grain image blocks contained in the reference grain image.

[0073] The above-mentioned preset image segmentation period can be 6 pixels, or it can be 512 pixels or 1024 pixels, etc. The specific selection of the preset image segmentation period is not limited here.

[0074] S142. For the current reference die image block and the current real-time die image block located at the same position, perform image registration on the current real-time die image block according to the current reference die image block to obtain the pixel position deviation of the current real-time die image block.

[0075] During real-time processing by the image processor, image registration is performed on the real-time transmitted die image blocks. For example, after obtaining the first real-time die image block (Frame1) in the real-time die image corresponding to die 1 at the current moment, the first real-time die image block is registered with the first pre-divided reference die image block in the reference die image to obtain the pixel position deviation of the current real-time die image block. In this embodiment, the obtained pixel deviation may include pixel deviation in the X direction and pixel deviation in the Y direction.

[0076] S143. After obtaining the corresponding pixel position deviation for each real-time grain image block in the real-time grain image, obtain the image position deviation corresponding to the real-time grain image.

[0077] For a real-time grain image corresponding to a grain, after obtaining the corresponding pixel position deviation for each real-time grain image block, the pixel position deviation corresponding to each real-time grain image block is used as the image position deviation of the real-time grain image.

[0078] This can be understood as the image position deviation corresponding to a real-time grain image being a set of pixel position deviations obtained for each real-time grain image block. Specifically, the pixel position deviations corresponding to each real-time grain image block in the X direction can be represented as Δx1, Δx2, ..., Δx N The image position deviation corresponding to the real-time grain image is represented as D. i Let i represent the real-time grain image corresponding to any grain in the current scan path, then D i ={△x1, △x2, ..., △x N}

[0079] Based on the image position deviation scheme for determining the real-time grain image provided in steps S141 to S143 above, the method of analyzing and registering each image block reduces the amount of data and computational complexity of each processing step. The pixel position deviation is calculated for each image block, and finally integrated to obtain the position deviation of the entire real-time grain image, which can accurately reflect the change in spatial position of the real-time grain image relative to the reference grain image.

[0080] Example 1, please refer to Figure 12 , Figure 12 This is a flowchart illustrating the determination of a position deviation sequence provided in an embodiment of this application. The implementation method provided in this application for "obtaining the position deviation sequence of the motion stage in the current scanning path based on the image position deviation and object pixel size corresponding to each real-time grain image" includes, but is not limited to, the following steps S144 to S146:

[0081] S144. For the current image position deviation corresponding to the current real-time die image, obtain the motion stage position deviation corresponding to the current real-time die image based on the pixel position deviation and object pixel size corresponding to each real-time die image block in the current image position deviation.

[0082] One real-time grain image corresponds to one image position deviation, which includes the pixel position deviation corresponding to each real-time grain image block. Further, the motion stage position deviation corresponding to one real-time grain image is calculated by multiplying the pixel position deviation corresponding to each real-time grain image block in the current image position deviation by the pixel size (the pixel deviation in the X direction is multiplied by the pixel size in the X direction, and the pixel deviation in the Y direction is multiplied by the pixel size in the Y direction). The current motion stage position deviation is the position deviation in the motion stage coordinate system. That is, the motion stage position deviation includes the motion stage position deviation in the first direction and the motion stage position deviation in the second direction.

[0083] S145. Obtain the initial deviation sequence of the motion stage in the current scanning path based on the motion stage position deviation corresponding to each real-time grain image.

[0084] After the current scan path is completed, the initial deviation sequence of the motion stage in the current scan path can be obtained by using the position deviation of the motion stage corresponding to the multiple real-time grain images contained in the current scan path.

[0085] The initial deviation sequence can be represented as DX0(x i ),DY0(x i ), representing the deviation of the position of the motion stage when the i-th real-time grain image block is acquired from the position of the motion stage when the i-th reference grain image block is acquired. Where x i This indicates the position of the motion stage in the X direction when the i-th reference grain image block within the current scan path is acquired.

[0086] The purpose of obtaining the initial deviation sequence in the current scanning path in this embodiment is to compensate for the next scanning path in the subsequent step S160 using the initial deviation sequence in the previous scanning path, so as to control the motion stage to scan according to the compensated scanning path, thereby reducing the image position deviation corresponding to each real-time grain image in the next scanning path and improving the defect detection sensitivity.

[0087] S146. For the initial deviation sequence, interpolation calculation is performed according to the scanning coordinates corresponding to the movement time of the motion stage to obtain the position deviation sequence of the motion stage in the current scanning path. The set of scanning coordinates corresponding to different movement times constitutes the current scanning path.

[0088] The purpose of interpolating the initial deviation sequence in this embodiment is that the deviation of each real-time grain image block obtained at the moment cannot represent the deviation corresponding to each row of pixels in the acquired real-time grain image block. After interpolation calculation, the deviation corresponding to all rows of pixels is characterized, and the position deviation sequence obtained in this way improves the accuracy of the compensation result for the next scanning path.

[0089] As can be seen from the solution provided in steps S144 to S146 above, the method for image compensation of multiple dies included in the scanning path in the current embodiment is to compensate the next scanning path according to the positional deviation sequence in the current scanning path. This allows for sequential scanning of each die in the next scanning path based on the compensated scanning path, reducing the positional deviation of the real-time die image block relative to the reference die image block in the next row. This helps reduce the impact of temperature drift of the equipment and wafer, as well as process errors in die position, on the image acquisition position, reducing the search range for image registration interpolation during defect detection, while also reducing noise caused by image interpolation and improving detection sensitivity.

[0090] Example 2: In addition to the method provided in Example 1, this example also provides another parallel implementation scheme for determining the position deviation sequence. Specifically, please refer to... Figure 13 , Figure 13 This is another flowchart illustrating the determination of the positional deviation sequence provided in this application embodiment. This application embodiment provides an implementation method for "performing image registration for each real-time grain image based on a reference grain image to obtain the image positional deviation corresponding to each real-time grain image," including, but not limited to, the following steps S1401 to S1407:

[0091] S1401. For the current real-time grain image block corresponding to the current real-time grain image, obtain the coordinate position deviation corresponding to the current real-time grain image block based on the pixel position deviation and object pixel size of the current real-time grain image block.

[0092] In the current embodiment, when the image processing system acquires a real-time grain image block based on a preset image segmentation cycle, it obtains the coordinate position deviation of the current real-time grain image block by multiplying the pixel position deviation of the current real-time grain image block with the object pixel size.

[0093] The pixel position deviation corresponding to the current real-time die image block includes the pixel deviation in the X direction and the pixel deviation in the Y direction; the object-side pixel size includes the pixel size in the X direction and the pixel size in the Y direction; the coordinate position deviation corresponding to the current real-time die image block can be determined by multiplying the pixel deviation in the X direction by the pixel size in the X direction and multiplying the pixel deviation in the Y direction by the pixel size in the Y direction, thereby obtaining the coordinate position deviation corresponding to the current real-time die image block, including the coordinate position deviation dX_1 in the X direction and the coordinate position deviation dY_1 in the Y direction. It should be noted that the coordinate position deviation in this embodiment can characterize the position deviation of the motion stage.

[0094] S1402. Obtain historical position deviation. The historical position deviation is the sum of the coordinate position deviations of the motion stage arranged sequentially before the current real-time die image block when acquiring each real-time die image block.

[0095] Before acquiring the current real-time die image block i, historical time has been used to acquire i-1 real-time die image blocks, and each of the i-1 real-time die image blocks corresponds to a coordinate position deviation. Therefore, by summing the coordinate position deviations corresponding to the i-1 real-time die image blocks, the historical position deviation can be obtained.

[0096] In a special case, when the current real-time grain image block i=1, step S1402 does not need to be executed.

[0097] S1403. Determine the actual position deviation of the motion stage when acquiring the current real-time grain image block based on the sum of the coordinate position deviation corresponding to the current real-time grain image block and the historical position deviation.

[0098] The actual position deviation of the current real-time die image block can be obtained by summing the coordinate position deviation of the current real-time die image block and the historical position deviation. The current actual position deviation is used to characterize the offset between the scan coordinates planned for the current real-time die image block and the actual scan coordinates in the scan path. The scan coordinates planned for the real-time die image block can be a part of the current scan path, that is, it can be the pre-planned current scan path, which is equivalent to pre-planning the scan coordinates of the real-time die image block.

[0099] In another special case, when the current real-time grain image block i = 1, the coordinate position deviation of the current real-time grain image block is the corresponding actual position deviation.

[0100] S1404. Update the scan coordinates of the current real-time grain image block according to the actual position deviation to obtain the updated scan coordinates.

[0101] The current real-time die image block has corresponding scan coordinates, which are part of the current scan path. These scan coordinates are the ones initially planned when the current scan path is planned, and are determined during the initial scan path planning. After obtaining the actual position deviation of the current real-time die image block, it indicates that the actual scan coordinates when scanning the current real-time die image block deviate from the initially planned scan coordinates. Therefore, the initially planned scan coordinates need to be updated according to the actual position deviation to obtain updated scan coordinates. Thus, scanning of the current real-time die image block is achieved based on the updated scan coordinates. It should be noted that the scan coordinates can include scan coordinates in a first direction and scan coordinates in a second direction. The first direction and the second direction are two different directions of motion of the stage, and the first direction and the second direction can intersect.

[0102] For example, taking the scan coordinates in the first scan path as an example, before updating the scan coordinates of the current real-time die image block, the scan coordinates of the motion stage in the X and Y directions of the current real-time die image block can be expressed by the following formula:

[0103]

[0104] After obtaining the actual position deviation at time t0, the updated scan coordinates of the current real-time grain image block can be expressed by the following formula:

[0105]

[0106] In the above, This represents the position coordinates of the motion stage in the X direction when the initial image position of the first real-time grain image on the first scan path is acquired in the image sensor. This represents the Y-axis coordinate of the motion stage when scanning the initial image position of the first real-time die image acquired by the image sensor along the first scanning path. The initial image position described above can be the position of the motion stage when the image sensor acquires the first row of pixels of the first real-time die image. Moreover, the image sensor acquires the first real-time die image from the first real-time die image block. This initial image position can be understood as the position of the motion stage when the image sensor acquires the first row of pixels of the first real-time die image block. dY1 represents the actual position deviation of the motion table in the first direction at time t0. 1 This represents the actual position deviation of the motion table in the second direction at time t0.

[0107] As can be seen from the solution provided in the current steps, the method of image compensation for multiple dies contained in a scanning path in the current embodiment is as follows: during the scanning of a scanning path, for each real-time die image block pixel position deviation acquired, the coordinate position deviation is determined in real time based on the pixel position deviation, and then the real-time position deviation is determined based on the coordinate position deviation and the historical position deviation. Thus, the scanning path of the motion stage scanning the current die image block is updated based on the real-time position deviation, so that when acquiring images of a row of dies, the effects of motion stage drift or even jitter can be eliminated in real time.

[0108] S1405. After all real-time die image blocks in the current scanning path have been scanned, the updated scanning path of the current scanning path is obtained according to the updated scanning coordinates corresponding to each real-time die image block.

[0109] The updated scan path for the current scan path is obtained based on the updated scan coordinates corresponding to each real-time die image block. In other words, the set of updated scan coordinates corresponding to each real-time die image block constitutes the updated scan path. In this embodiment, the purpose of determining the updated scan path for the current scan path is to compensate for the next scan path based on the current updated scan path, thereby pre-adjusting the motion stage scan path to achieve accurate scanning.

[0110] In one embodiment, the updated scan path is obtained using the following formula:

[0111]

[0112] in, This constitutes the update scan path. This indicates the j-th update scan path in the first direction. This indicates the j-th update scan path in the second direction; This indicates the position of the motion stage in the first direction when the image sensor acquires the initial image position of the first real-time die image block on the j-th scan path; This represents the position of the motion stage in the second direction when the image sensor acquires the initial image position of the first real-time grain image block on the j-scan paths; DX j (x i ) refers to the actual position deviation corresponding to the i-th real-time grain image block in the first direction on the j-th scan path, DY j (x i ) refers to the actual position deviation of the i-th real-time grain image block in the second direction on the j-th scan path.

[0113] The first direction mentioned above is the X direction, the second direction is the Y direction, and the initial image position is the same as the concept mentioned above, so it will not be repeated here.

[0114] S1406. After scanning all the test dies based on the updated scanning path, the initial deviation sequence of the motion stage in the current scanning path is obtained according to the actual position deviation corresponding to each real-time die image block.

[0115] After all the test grains have been scanned, the initial deviation sequence obtained can be represented as DX0(x i ),DY0(x i ), representing the deviation of the position of the motion stage when the i-th real-time grain image block is acquired from the position of the motion stage when the i-th reference grain image block is acquired. Where x i This indicates the position of the motion stage in the X direction when the i-th reference grain image block within the current scan path is acquired.

[0116] S1407. For the initial deviation sequence, interpolation calculation is performed according to the scanning coordinates corresponding to the movement time of the motion stage to obtain the position deviation sequence of the motion stage in the current scanning path. The set of scanning coordinates corresponding to different movement times constitutes the current scanning path.

[0117] It should be noted that the technical solutions provided in Embodiment 1 and Embodiment 2 are two parallel solutions for "determining the position deviation sequence of the motion stage in the current scanning path". In specific implementation, it is not limited to using the implementation method of either embodiment.

[0118] In a preferred embodiment, based on the above step S130, this embodiment provides two approaches to determining the initial scan grain row. Therefore, based on the different methods of determining the initial scan grain row, after completing step S140 to obtain the position deviation sequence of the stage in the current scan path, this embodiment needs to perform different operations accordingly. Specifically, the corresponding operation steps for different scanning methods are as follows:

[0119] A) Obtain the scanning mode of the motion stage for the wafer under test.

[0120] In this embodiment, the scanning methods include a first method and a second method. The first method involves the motion stage performing a sequential scan based on each row of grains in the wafer under test; the second method involves the motion stage initiating the scan from the first row of grains containing the largest number of grains in the wafer under test.

[0121] B) When the scanning mode is the first mode, determine whether the current scanning path is the first scanning path in each row of grains to be tested.

[0122] by Figure 3Taking the structure diagram of the wafer under test as an example, the reason for determining whether the current scanning path is the first scanning path in each die row under test in the current step is that when the current die row under test is the first die row, when scanning the first scanning path of the first die row, the position deviation sequence of the motion stage in the first scanning path can only be obtained after scanning the first scanning path, and the scanning image obtained by the first scanning path is not compensated. Therefore, to improve the detection sensitivity of the scanned image corresponding to the first scan path, when the current scan path is the first scan path in the first row of the test grains, it is necessary to repeat the scan of the first scan path once based on the current position deviation sequence. After all scanned paths in the first row of the test grains have been scanned, when scanning the second row of the test grains, since the number of grains in the first row is less than the number of grains in the second row, the position deviation sequence corresponding to the last scan path in the first row is not meaningful for image compensation of the first scan path in the second row. Therefore, it is necessary to re-scan the first scan path in the second row to obtain the position deviation sequence, and then scan the first scan path in the second row again based on the position deviation sequence to compensate for the first scan path in the second row, and so on. For details, please refer to [reference needed]. Figure 11 When the current scan path is detected as the first scan path in each row of the grains to be tested, i.e., scan path 1, two scans are required.

[0123] If the current scan path is the first scan path in each row of the die to be tested, then execute step b1); if the current scan path is not the first scan path in each row of the die to be tested, then execute step b2).

[0124] b1) Compensate the first scan path in each row of the die to be tested according to the position deviation sequence to obtain the first updated scan path corresponding to the first scan path in each row of the die to be tested. In each row of the die to be tested, the first updated scan path is determined as the current scan path. Repeatedly execute the operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object-side pixel size, and sequentially obtain the real-time die image corresponding to each die. Then, perform image registration on each real-time die image based on the reference die image to obtain the image position deviation corresponding to each real-time die image. Based on the image position deviation corresponding to each real-time die image and the object-side pixel size, obtain the position deviation sequence of the motion stage in the current scan path. Continue until the first scan path in the row of the die to be tested is scanned. At this time, when the first scan path is scanned for the second time, the position deviation sequence of the motion stage when scanning the first scan path for the second time is obtained to compensate the next scan path. Continue until multiple scan paths in each row of the die to be tested are scanned to obtain the wafer compensation image corresponding to each row of the die to be tested.

[0125] b2) Perform the operation of compensating the next scan path according to the position deviation sequence to obtain the compensated scan path for the next scan path.

[0126] C) When the scanning mode is the first mode, determine whether the current scanning path is the first scanning path of the first grain row in the wafer under test.

[0127] Since the first scan path is the row containing the most grains selected from the wafer under test, from... Figure 3 As can be seen from the example diagram of the wafer under test, the positional deviation sequence of the stage in the first scan path can only be obtained after scanning the first scan path of the first grain row, and no compensation is performed for the scan image obtained in the first scan path. Therefore, in order to improve the detection sensitivity of the scan image corresponding to the first scan path of the first grain row, when the current scan path is the first scan path in the first grain row under test, it is necessary to repeat the scan of the first scan path again based on the current positional deviation sequence.

[0128] If the current scan path is the first scan path of the first grain row in the wafer under test, then execute step c1); if the current scan path is not the first scan path of the first grain row in the wafer under test, then execute step c2).

[0129] c1) Compensate the first scan path of the first die row according to the position deviation sequence to obtain the first updated scan path, and determine the first updated scan path as the current scan path. Repeat the operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object-side pixel size, and sequentially obtain the real-time die image corresponding to each die. Based on the reference die image, perform image registration on each real-time die image to obtain the image position deviation corresponding to each real-time die image. Based on the image position deviation corresponding to each real-time die image and the object-side pixel size, obtain the position deviation sequence of the motion stage in the current scan path. Continue the operation until the first scan path in the first die row is scanned. When the first scan path is scanned for the second time in the first die row, the final position deviation sequence of the motion stage when scanning the first scan path for the second time is obtained. Compensate the next scan path until multiple scan paths in the first die row are scanned. Obtain the wafer compensation image corresponding to each die row to be tested.

[0130] c2) Perform the operation of compensating the next scan path according to the position deviation sequence to obtain the compensated scan path for the next scan path.

[0131] As can be seen from the comparison of the two scanning methods above, the solution provided in this embodiment preferentially adopts the second scanning method. The reason is that after scanning the first scan path of the first grain row twice, since the first grain row contains the most grains, when scanning upwards to the next row of grains until the path scanning in the upper half of the wafer is completed, and then returning to the middle to scan the scan path contained in the lower half of the wafer from top to bottom, the number of grains in each grain row to be tested decreases row by row; that is, the number of grains in the previous row is always more than the number in the next row. Therefore, the first scan path corresponding to the next grain row to be tested can be compensated based on the positional deviation sequence corresponding to the last scan path of the previous grain row, thereby avoiding the need to repeatedly scan the first scan path for each grain row to be tested in the first method, thus improving scanning efficiency.

[0132] S150. Determine whether all the multiple scan paths of the test grain row in the test wafer have been scanned.

[0133] If not all scans are completed, proceed to step S160.

[0134] If all scans have been completed, then proceed to step S170.

[0135] S160. For the next scanning path, compensate the next scanning path according to the position deviation sequence in the current scanning path to obtain the compensated scanning path, and determine the compensated scanning path as the current scanning path.

[0136] After determining the compensation scan path as the current scan path in the current step, repeat steps S130 to S150. That is, control the motion stage to move according to the scanning speed, and control the image sensor to acquire images of each die according to the object-side pixel size, sequentially obtaining the real-time die image corresponding to each die, and continuing to perform image registration on each real-time die image based on the reference die image to obtain the image position deviation corresponding to each real-time die image; the operation of obtaining the position deviation sequence of the motion stage in the current scan path according to the image position deviation corresponding to each real-time die image and the object-side pixel size, until all multiple scan paths of the die row to be tested in the wafer under test have been scanned.

[0137] Specifically, regarding the method of "compensating the next scanning path based on the position deviation sequence in the current scanning path to obtain a compensated scanning path," step S140 provides two implementation methods for "determining the position deviation sequence of the motion stage in the current scanning path." Consequently, in this step, there are also two implementation methods for obtaining the compensated scanning path, which are explained in detail below:

[0138] In one implementation, when determining the position deviation sequence of the motion stage in the current scanning path based on the scheme provided in Embodiment 1 in step S140 above, the method for obtaining the compensated scanning path for the next scanning path can be obtained by the following formula:

[0139]

[0140] in, This indicates the position of the motion stage in the first direction when the image sensor acquires the initial image position of the first real-time grain image on the j-th scanning path; V represents the position of the motion stage in the second direction when the image sensor acquires the initial image position of the first real-time grain image along the j-th scanning path;

[0141]

[0142] The sum of the path's positional deviation sequence in the second direction; and Construct a compensation scan path, This represents the j-th compensated scan path in the first direction. This represents the j-th compensated scan path in the second direction, and m represents the change in the number of scan paths.

[0143] The initial image position mentioned above refers to the position of the motion stage when the first row of pixels of the first image is acquired. For a detailed explanation, please refer to the above, and it will not be repeated here.

[0144] It should be noted that when determining the compensation scan path based on the above formula, when m=1, the first scan path needs to be scanned twice. That is, the first scan is used to determine the position deviation sequence of the first scan path; the second scan is used to compensate the first scan path based on the current position deviation sequence, so as to make the obtained scan image more accurate and reduce the noise caused by image interpolation. It should be noted that in the second scan, the position deviation sequence is the position deviation sequence obtained in the first scan of the first scan path.

[0145] By using the solution provided in this embodiment, the influence of temperature drift of equipment and wafer and process errors of die position on image acquisition position can be reduced, the search range of image registration interpolation can be reduced, and the noise caused by image interpolation can be reduced, thereby improving detection sensitivity.

[0146] Another implementation method, when determining the position deviation sequence of the motion stage in the current scanning path based on the scheme provided in Embodiment 2 in step S140 above, the method for obtaining the compensated scanning path for the next scanning path can be obtained by the following formula:

[0147] The compensation scan path is obtained using the following formula:

[0148]

[0149]

[0150] This refers to the j-th update scan path in the first direction. This refers to the j-th update scan path in the second direction; and Construct a compensation scan path, This represents the j-th compensated scan path in the first direction. This represents the j-th compensated scan path in the second direction; m represents the change in the number of scan paths.

[0151] By using the solution provided in this embodiment, compared with the above embodiment, the solution provided in this embodiment, during the scanning of a scanning path, immediately performs image registration with the reference grain image block after obtaining each real-time grain image block, and corrects the scanning path of the motion stage to scan the next real-time grain image block based on the registration result. By determining the actual position deviation in real time, the bandwidth and accuracy of position compensation are improved, further reducing the noise caused by image interpolation during defect detection and improving the detection sensitivity of the equipment.

[0152] S170. Until all multiple scanning paths of the test grain row in the wafer under test have been scanned, the wafer compensation image corresponding to the test grain row is obtained.

[0153] The wafer image compensation method provided in this embodiment predetermines the object-space pixel size when the image sensor acquires images, and determines the scanning speed of the stage based on the object-space pixel size before controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die based on the object-space pixel size. It also limits the die distance between adjacent dies to an integer multiple of the object-space pixel size. This ensures that when acquiring images based on the currently set object-space pixel size, the pixel positions on the camera are strictly consistent when the image sensor acquires images of the same position on different dies. This avoids issues caused by pixel position differences. The noise introduced by image registration interpolation due to offset is reduced, thus improving detection sensitivity. Furthermore, during the process of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size, the next scanning path can be modified based on the positional deviation sequence obtained from the current scanning path. This reduces the positional deviation in subsequent image acquisition caused by process errors of devices in the detection system, thermal expansion and contraction of the wafer due to temperature changes, and process deviations in the position of the dies on the wafer. It also reduces errors caused by process or temperature and noise introduced by image interpolation, thereby improving detection sensitivity.

[0154] Please refer to Figure 14 This embodiment, based on the above-described wafer image compensation method, also provides a wafer defect detection system. Specifically, the defect detection system includes a motion controller 10, an illumination system 20, an imaging system 30, an image sensor 31, and an image processing system 40. The image processing system 40 is connected to the motion controller 10, the illumination system 20, the imaging system 30, and the image sensor 31. The motion controller 10 controls the movement of the motion stage. The illumination system 20 outputs a light beam so that the beam is focused onto the surface of the wafer under test through the objective lens. The imaging system 30 receives the reflected light from the wafer under test returned through the objective lens and adjusts the imaging magnification to adjust the object-side pixel size when the image sensor 31 acquires the wafer image of the wafer under test. The image sensor 31 acquires images of each die and transmits the acquired wafer images to the image processing system 40. The image processing system 40 executes the wafer image compensation method of any embodiment of this application to obtain a compensated image of the wafer under test and performs defect detection on the wafer under test based on the compensated image.

[0155] This can be further explained as being based on Figure 14When performing defect detection on a wafer under test, the provided wafer defect detection system places the wafer under test above a motion stage. The motion stage can adjust the position of the wafer under test from at least the x and y directions. The motion stage controller controls the motion stage to move the pattern at different X and Y positions on the wafer under test below the objective lens. The light beam output by the illumination system is sequentially converged onto the wafer pattern at the current X and Y position through a reflector, a beam splitter, and an objective lens. The reflected light is then passed through the objective lens and a zoom module. The zoom module adjusts the imaging magnification so that the image sensor acquires the object-side pixel size of the wafer image based on the adjusted imaging magnification. The wafer image is then transmitted to the image processing system 40, where the wafer image compensation method provided in this embodiment is executed to perform defect detection.

[0156] The wafer defect detection system provided in this embodiment of the invention can execute the wafer image compensation method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0157] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A wafer image compensation method, characterized in that, include: For a wafer under test placed on a motion stage, the object-side pixel size when the image sensor acquires an image is determined based on the grain distance between adjacent dies in the same grain row of the wafer under test, wherein the grain distance between adjacent dies is an integer multiple of the object-side pixel size; the scanning speed of the motion stage is determined based on the object-side pixel size; the wafer under test includes multiple grain rows, and each grain row includes multiple scanning paths; A reference grain row is determined from the plurality of said grain rows, and a reference grain image is determined based on the plurality of reference grains contained in the reference grain row; Select the current scanning path, control the motion stage to move according to the scanning speed, and control the image sensor to acquire images of each die according to the object pixel size, so as to obtain the real-time die image corresponding to each die in sequence; Based on the reference grain image, image registration is performed on each of the real-time grain images to obtain the image position deviation corresponding to each real-time grain image; the position deviation sequence of the motion stage in the current scanning path is obtained according to the image position deviation corresponding to each real-time grain image and the object pixel size; For the next scanning path, the next scanning path is compensated according to the position deviation sequence in the current scanning path to obtain a compensated scanning path; the compensated scanning path is determined as the current scanning path, and the operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size is repeated to obtain the real-time die image corresponding to each die in sequence until all multiple scanning paths of the die row to be tested in the wafer under test are scanned, and the wafer compensation image corresponding to the die row to be tested is obtained.

2. The wafer image compensation method according to claim 1, characterized in that, The step of performing image registration on each of the real-time grain images based on the reference grain image to obtain the image position deviation corresponding to each real-time grain image includes: A preset image segmentation period is determined, and the real-time grain image and the reference grain image are divided into equal numbers of real-time grain image blocks and reference grain image blocks according to the preset image segmentation period; For a current reference die image block and a current real-time die image block located at the same position, image registration is performed on the current real-time die image block based on the current reference die image block to obtain the pixel position deviation of the current real-time die image block; After obtaining the corresponding pixel position deviation for each real-time grain image block in the real-time grain image, the image position deviation corresponding to the real-time grain image is obtained.

3. The wafer image compensation method according to claim 2, characterized in that, The step of obtaining the position deviation sequence of the motion stage in the current scanning path based on each image position deviation and the object pixel size includes: For the current image position deviation corresponding to the current real-time die image, the motion stage position deviation corresponding to the current real-time die image is obtained according to the pixel position deviation corresponding to each real-time die image block and the object pixel size. The initial deviation sequence of the motion stage in the current scanning path is obtained based on the motion stage position deviation corresponding to each of the real-time grain images; For the initial deviation sequence, interpolation calculation is performed according to the scanning coordinates corresponding to the movement time of the motion stage to obtain the position deviation sequence of the motion stage in the current scanning path. The set of scanning coordinates corresponding to different movement times constitutes the current scanning path.

4. The wafer image compensation method according to claim 3, characterized in that, The compensated scanning path is obtained by compensating the next scanning path based on the positional deviation sequence in the current scanning path, using the following formula: in, This indicates the position of the motion stage in the first direction when the image sensor acquires the initial image position of the first real-time grain image on the j-th scanning path; V represents the position of the motion stage in the second direction when the image sensor acquires the initial image position of the first real-time grain image along the j-th scanning path; This represents the j-th compensated scan path in the first direction. This represents the j-th compensated scan path in the second direction, and m represents the change in the number of scan paths.

5. The wafer image compensation method according to claim 3, characterized in that, After obtaining the position deviation sequence of the motion stage in the current scanning path, the method further includes: The scanning mode of the motion stage on the wafer under test is obtained. The scanning mode includes a first mode and a second mode. The first mode is that the motion stage performs sequential scanning based on each grain row in the wafer under test. The second mode is that the motion stage starts scanning from the first grain row with the largest number of grains in the wafer under test. Wherein, when the scanning method is the first method, the method further includes: Determine whether the current scan path is the first scan path in each row of grains to be tested; If so, the first scan path in each of the rows of the die to be tested is compensated according to the position deviation sequence to obtain the first updated scan path corresponding to the first scan path in each row of the die to be tested. The first updated scan path is determined as the current scan path in each row of the die to be tested. The operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size is repeated to obtain the real-time die image corresponding to each die in sequence until all the multiple scan paths in each row of the die to be tested are scanned, and the wafer compensation image corresponding to each row of the die to be tested is obtained. If not, then perform the operation of compensating the next scanning path according to the position deviation sequence to obtain a compensated scanning path for the next scanning path; Wherein, when the scanning method is the second method, the method further includes: Determine whether the current scan path is the first scan path of the first grain row in the wafer under test; If so, the first scan path of the first die row is compensated according to the position deviation sequence to obtain the first updated scan path, and the first updated scan path is determined as the current scan path. The operation of controlling the motion stage to move according to the scanning speed and controlling the image sensor to acquire images of each die according to the object pixel size is repeated to obtain the real-time die image corresponding to each die in sequence until all the multiple scan paths in the first die row are scanned, and the wafer compensation image corresponding to the first die row is obtained. If not, then perform the operation of compensating the next scan path according to the position deviation sequence to obtain a compensated scan path for the next scan path.

6. The wafer image compensation method according to claim 2, characterized in that, The step of obtaining the position deviation sequence of the motion stage in the current scanning path based on each image position deviation and the object pixel size includes: For the current real-time grain image block corresponding to the current real-time grain image, the coordinate position deviation corresponding to the current real-time grain image block is obtained according to the pixel position deviation corresponding to the current real-time grain image block and the object-side pixel size; The historical position deviation is obtained, which is the sum of the coordinate position deviations of the motion stage arranged sequentially before the current real-time grain image block when acquiring each real-time grain image block; The actual position deviation of the motion stage when acquiring the current real-time grain image block is determined based on the sum of the coordinate position deviation corresponding to the current real-time grain image block and the historical position deviation. The scan coordinates of the current real-time grain image block are updated based on the actual position deviation to obtain the updated scan coordinates; After all real-time die image blocks in the current scanning path have been scanned, the updated scanning path of the current scanning path is obtained according to the updated scanning coordinates corresponding to each real-time die image block. After scanning each of the test dies based on the updated scanning path, the initial deviation sequence of the motion stage in the current scanning path is obtained according to the actual position deviation corresponding to each real-time die image block. For the initial deviation sequence, interpolation calculation is performed according to the scanning coordinates corresponding to the movement time of the motion stage to obtain the position deviation sequence of the motion stage in the current scanning path. The set of scanning coordinates corresponding to different movement times constitutes the current scanning path.

7. The wafer image compensation method according to claim 6, characterized in that, The update scan path is obtained using the following formula: in, This constitutes the update scan path. This indicates the j-th update scan path in the first direction. This indicates the j-th update scan path in the second direction; This indicates the position of the motion stage in the first direction when the image sensor acquires the initial image position of the first real-time die image block on the j-th scan path; This represents the position of the motion stage in the second direction when the image sensor acquires the initial image position of the first real-time grain image block on the j-scan paths; DX j (x i ) refers to the actual position deviation corresponding to the i-th real-time grain image block in the first direction on the j-th scan path, DY j (x i ) refers to the actual position deviation of the i-th real-time grain image block in the second direction on the j-th scan path.

8. The wafer image compensation method according to claim 6, characterized in that, The step of compensating the next scanning path based on the position deviation sequence in the current scanning path to obtain a compensated scanning path includes: The next scan path is compensated based on the updated scan path and the position deviation sequence to obtain the compensated scan path; The compensation scan path is obtained using the following formula: This refers to the j-th update scan path in the first direction. This refers to the j-th update scan path in the second direction; and Construct a compensation scan path, This represents the j-th compensated scan path in the first direction. This represents the j-th compensated scan path in the second direction; m represents the change in the number of scan paths.

9. The wafer image compensation method according to claim 1, characterized in that, Determining the scanning speed of the motion stage based on the object pixel size includes: The scanning speed of the motion stage is determined based on the object pixel size and the line frequency of the image sensor.

10. The real-time wafer image compensation method according to claim 1, characterized in that, The step of determining a reference grain row from a plurality of said grain rows, and determining a reference grain image based on a plurality of reference grains contained in the reference grain row, includes: Obtain the grain identifier corresponding to each grain, and determine the row where the grain identifier is a preset identifier as the reference grain row; Obtain the image corresponding to each grain in the reference grain row, process the image corresponding to each grain, and obtain the grain grayscale image of each grain; The reference grain image is determined based on the grayscale value in the grayscale image of each grain.

11. A wafer defect detection system, characterized in that, The defect detection system includes a motion controller, an illumination system, an imaging system, an image sensor, and an image processing system; the image processing system is connected to the motion controller, the illumination system, the imaging system, and the image sensor, respectively. The motion controller is used to control the movement of the motion stage; the illumination system is used to output a light beam so that the light beam is focused onto the surface of the wafer under test through the objective lens; the imaging system is used to receive the reflected light from the wafer under test returned through the objective lens, and the imaging system is also used to adjust the imaging magnification to adjust the object-side pixel size when the image sensor acquires the wafer image of the wafer under test; the image sensor is used to acquire an image of each die and transmit the acquired wafer image to the image processing system; the image processing system executes the wafer image compensation method according to any one of claims 1-10 to obtain a compensated image of the wafer under test, and performs defect detection on the wafer under test based on the compensated image.

Images