A grain shift detection method and related products

Abstract

The application discloses a wafer grain offset detection method and related products, and the method comprises the following steps: determining the theoretical parameters of a target wafer grain based on an initial wafer map of a target wafer; performing dynamic correction processing on the initial wafer map to obtain an actual wafer map of the target wafer; determining the actual parameters and field parameters of the target wafer grain based on the actual wafer map; the actual parameters at least include actual center point coordinates and target edge profile coordinates; the field parameters at least include field edge profile coordinates of each field wafer grain adjacent to the target wafer grain; and obtaining the offset detection result of the target wafer grain based on the theoretical parameters, the actual parameters and the field parameters of the target wafer grain. The application can comprehensively evaluate the wafer cutting quality, correct the global coordinate system deviation, and finely control the local wafer cutting risk, thereby meeting the strict requirements of modern semiconductor manufacturing on high-precision and high-reliability detection.

Description

Technical Field

[0001] This application relates to the field of semiconductor technology, and in particular to a method for detecting grain offset and related products. Background Technology

[0002] Currently, the industry's conventional inspection method typically involves pre-planning a fixed die map in the inspection recipe based on the standard wafer layout, and recording the theoretical standard coordinates of each die. During actual inspection, the equipment identifies the actual position of each die by acquiring wafer images, compares it with the preset theoretical coordinates in the die map, and calculates the absolute offset distance of each die. Based on this value, it determines whether there is a cutting misalignment defect. However, this comparison method based on a fixed distribution map has significant drawbacks in actual mass production and is difficult to meet the requirements of high-precision on-site judgment.

[0003] On the one hand, a fixed die map cannot effectively adapt to the actual deformation of the wafer, resulting in large detection errors. During processes such as lamination, expansion, and mechanical dicing, wafers are prone to overall displacement, local warping, or colloidal stretching deformation, causing an overall deviation between the actual die arrangement and the preset theoretical die map. If only theoretical fixed coordinates are used to calculate single-point offsets, it will not only fail to accurately reflect the physical state after dicing, but will also easily lead to misjudgments or missed judgments.

[0004] On the other hand, existing solutions only use the absolute offset of a single die as the criterion, ignoring the relative positional relationship between dies and failing to calculate the actual spacing between adjacent dies. This often leads to two extreme situations on-site: first, although the offset value of a single die is within the acceptable range, the spacing between adjacent dies is too small, posing a risk of damage to subsequent processing; second, although the offset value of a single point is large, the actual die spacing still meets the process safety requirements, yet it is still judged as scrapped without cause. This one-sided judgment logic obviously cannot fully guarantee product quality and yield. Summary of the Invention

[0005] To address the aforementioned issues, this application provides a method for detecting grain displacement and related products.

[0006] The embodiments of this application disclose the following technical solutions: A first aspect of this application provides a method for detecting grain offset, the method comprising: Based on the initial wafer map of the target wafer, the theoretical parameters of the target grain are determined; the initial wafer map includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain; the theoretical parameters of the target grain include at least the coordinates of the theoretical center point; The initial wafer image is dynamically corrected to obtain the actual wafer image of the target wafer; Based on the actual wafer image, the actual parameters and neighborhood parameters of the target grain are determined; the actual parameters include at least the actual center point coordinates and the target edge contour coordinates; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain. Based on the theoretical parameters, actual parameters, and neighborhood parameters of the target grain, the offset detection results of the target grain are obtained.

[0007] A second aspect of this application provides a grain offset detection device, the grain offset detection device comprising: The parameter determination module is used to determine the theoretical parameters of the target grain based on the initial wafer map of the target wafer; the initial wafer map includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain; the theoretical parameters of the target grain include at least the coordinates of the theoretical center point; The correction module is used to dynamically correct the initial wafer image to obtain the actual wafer image of the target wafer. The parameter determination module is further configured to determine the actual parameters and neighborhood parameters of the target grain based on the actual wafer image; the actual parameters include at least the actual center point coordinates and the target edge contour coordinates; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain; The detection module is used to obtain the offset detection result of the target grain based on the theoretical parameters, actual parameters and neighborhood parameters of the target grain.

[0008] A third aspect of this application provides a computer program product, including a computer program that, when executed by a processor, implements the grain offset detection method provided in the first aspect.

[0009] Compared with the prior art, this application has the following advantages: This application includes determining the theoretical parameters of a target die based on an initial wafer map of a target wafer; the initial wafer map includes multiple dies with fixed distribution positions, and any die can be configured as the target die; the theoretical parameters of the target die include at least the coordinates of a theoretical center point; the initial wafer map is dynamically corrected to obtain an actual wafer map of the target wafer; based on the actual wafer map, the actual parameters and neighborhood parameters of the target die are determined; the actual parameters include at least the coordinates of an actual center point and the coordinates of a target edge contour; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood die adjacent to the target die; based on the theoretical parameters, actual parameters, and neighborhood parameters of the target die, the offset detection result of the target die is obtained.

[0010] This application obtains the actual wafer image of the target wafer by dynamically correcting the initial wafer image. This allows for real-time tracking and fitting of deformations such as overall displacement, local warpage, or colloidal stretching that occur during wafer lamination, expansion, and dicing. By combining theoretical parameters with the dynamically corrected actual distribution map, this application more realistically reflects the physical state of the wafer after dicing, effectively solving the detection error problem caused by the deviation between fixed coordinates and actual positions, and significantly improving the robustness and accuracy of the detection.

[0011] This application introduces "neighborhood parameters" during the inspection process, specifically including "the boundary edge contour coordinates of each neighboring die adjacent to the target die," and combines this with the target die's "actual center point coordinates and target edge contour coordinates" for comprehensive judgment. This application not only focuses on whether a single die deviates from its theoretical position, but also crucially calculates the actual physical distance between the target die and adjacent dies, avoiding subsequent processing damage or unwarranted scrap caused by biased judgment logic in existing technologies, effectively improving the final product yield. By integrating the target die's theoretical parameters, its own actual parameters (center point and contour), and surrounding neighborhood parameters (neighborhood contours), this application can comprehensively evaluate the cutting quality, correcting global coordinate system deviations and finely controlling local cutting risks, meeting the stringent requirements of modern semiconductor manufacturing for high-precision and high-reliability inspection. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application 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 only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a grain offset detection method provided in an embodiment of this application; Figure 2 An initial wafer pattern provided for an embodiment of this application; Figure 3 This is an actual wafer diagram of the target wafer provided in the embodiments of this application; Figure 4 This is a structural diagram of a grain offset detection device provided in an embodiment of this application. Detailed Implementation

[0014] As described earlier, existing conventional inspection methods rely on a pre-set fixed die map, judging dicing misalignment defects by comparing the absolute offset of the actual position of a single die with its theoretical coordinates. However, this method has two major drawbacks in actual mass production: First, the fixed die map cannot adapt to the overall displacement or local warping physical deformation of the wafer during processes such as lamination, expansion, and dicing, easily leading to inspection errors and misjudgments or missed judgments; Second, it only uses the absolute offset of a single die as the judgment basis, ignoring the relative position and actual spacing between adjacent dies. This can easily lead to one-sided judgments such as "single-point offset is acceptable but the adjacent spacing is too small (risk of damage)" or "single-point offset is too large but the spacing is safe (unjustifiably scrapped)," failing to comprehensively guarantee product quality and yield.

[0015] In view of the above problems, this application provides a method for generating a grain offset detection product and related products. The method includes: determining the theoretical parameters of a target grain based on an initial wafer image of a target wafer; the initial wafer image includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain; the theoretical parameters of the target wafer include at least the coordinates of a theoretical center point; performing dynamic correction processing on the initial wafer image to obtain an actual wafer image of the target wafer; determining the actual parameters and neighborhood parameters of the target grain based on the actual wafer image; the actual parameters include at least the coordinates of an actual center point and the coordinates of a target edge contour; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain; and obtaining the offset detection result of the target grain based on the theoretical parameters, actual parameters, and neighborhood parameters of the target grain.

[0016] This application dynamically corrects the initial wafer image, enabling real-time tracking and fitting of various deformations of the wafer during processing. This more accurately reflects the physical state of the wafer after dicing, effectively eliminating detection errors caused by fixed coordinates and significantly improving the robustness and accuracy of the inspection. Simultaneously, this application introduces "territory parameters," such as the contour coordinates of the edge of adjacent dies, and makes a comprehensive judgment based on the actual center point and contour coordinates of the target die. This scheme not only corrects global coordinate deviations but also accurately calculates the actual physical spacing between dies, avoiding subsequent processing damage or unwarranted scrap due to partial judgments. By comprehensively evaluating dicing quality, this application achieves refined control of local dicing risks, meeting the stringent requirements of modern semiconductor manufacturing for high-precision and high-reliability inspection.

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0018] Figure 1 A flowchart of a grain offset detection method provided in an embodiment of this application is shown below. Figure 1 As shown, a grain offset detection method includes: S101: Determine the theoretical parameters of the target grain based on the initial wafer diagram of the target wafer.

[0019] This application does not strictly limit the specific generation method and representation of the initial wafer map (which can also be understood as a die map), but it must ensure that the initial wafer map can accurately represent the arrangement of multiple dies with fixed distribution positions on the target wafer, and the target die to be detected must be included among the multiple dies. Furthermore, the theoretical parameters of the target die include at least the theoretical center point coordinates of the target die, as an absolute reference for subsequent alignment and offset calculations.

[0020] For example, the specific implementation process for determining the theoretical parameters of the target grain may include the following steps: Obtain the basic data of the target wafer.

[0021] This basic data serves as the basis for constructing an idealized distribution model, and includes at least the width and spacing parameters of the scribe lines, the length and width dimensions of a single die, and the effective area diameter of the wafer, among other specifications.

[0022] Based on the basic data, an initial wafer diagram is planned and constructed.

[0023] Specifically, during the equipment's recipe setup and debugging phase, the system accurately inputs core parameters such as the standard wafer dicing spacing, individual die length and width dimensions, and overall wafer layout specifications. Through the equipment's built-in array planning algorithm, it automatically calculates and constructs an initial fixed die distribution map, such as... Figure 2 As shown, this initial wafer diagram is essentially an idealized, physically unbiased baseline digital model that fully stores the core theoretical parameters of all grains within the entire wafer. Specifically, it includes: the theoretical center point coordinates of each grain, the theoretical edge coordinates of its four sides, the global row and column arrangement topology, the standard spacing between adjacent grains in the horizontal and vertical directions, and the standard dicing width, among other complete baseline data.

[0024] Select any grain in the initial wafer diagram as the target grain and determine the theoretical parameters of the target grain.

[0025] In the actual testing process, when the device moves to the target location and locks onto a specific die (e.g., the die located in row X, column Y) as the target die, the theoretical center point coordinates corresponding to the target die are directly retrieved from the initial wafer map in memory. These theoretical center point coordinates will serve as an absolute reference frame and will be compared with the physical center point actually acquired and identified by the camera in subsequent steps to accurately calculate the actual translational coordinate offset value of the target die.

[0026] S102: Perform dynamic correction processing on the initial wafer image to obtain the actual wafer image of the target wafer.

[0027] For example, the specific implementation process of revising the initial wafer pattern may include the following steps: Images of the target wafer are acquired using a preset scanning path, resulting in multiple scanned images. In actual inspection processes, full-area image acquisition of the entire wafer can be achieved using a high-resolution industrial camera and a precision optical imaging system. For example, images of the target wafer are acquired using a preset scanning path (such as an S-shaped or serpentine scanning path), and the acquired images constitute multiple scanned images. This path planning effectively avoids problems such as missed scans, duplicate scans, and edge imaging distortion, ensuring the integrity and continuity of image acquisition.

[0028] Based on multiple scanned images and an initial wafer image, the coordinate offset value of each die is obtained. In one exemplary embodiment, the multiple scanned images are stitched together to obtain an initial global image. Since the actual acquired images are easily affected by ambient light, wafer surface reflection, or sensor noise, feature extraction processing is required on the initial global image to obtain an enhanced global image. The specific processing flow is as follows: the initial global image is preprocessed to obtain a preprocessed global image; the preprocessing includes one or more of grayscale conversion, filtering and noise reduction, and contrast enhancement to highlight die features and suppress background noise; the preprocessed global image is processed using an edge detection algorithm to extract die edges, resulting in an extracted global image; the extracted global image is then subjected to angle fitting processing to obtain an enhanced global image, thereby eliminating coordinate deviations caused by overall wafer tilt or local deformation.

[0029] After image fitting is completed, a grayscale template matching algorithm is used to process the enhanced global image and identify the actual physical coordinates of each grain in the image.

[0030] Based on the actual physical coordinates and theoretical physical coordinates of each die, the coordinate offset value of each die is obtained. For example, the difference between the actual physical coordinates and the theoretical physical coordinates of each die is obtained, that is, the coordinate offset value of a single die (including the translation of the X-axis and Y-axis, as well as possible rotation angle deviations). The coordinate offset value constitutes a mathematical model of wafer surface deformation or cutting displacement.

[0031] As an optional embodiment, the initial wafer image is dynamically corrected based on the coordinate offset value to obtain the actual wafer image of the target wafer. By calculating the average coordinate offset of all grains or fitting a global trend, the overall translation and rotation deviations of the wafer are determined, and the initial wafer image is subjected to global coordinate system rotation correction and translation compensation. For nonlinear deformations such as warping and colloidal stretching caused by local wafer bonding, film expansion, or stress release, the initial wafer image is adaptively elastically deformed based on the actual offset vector of the grains in each region.

[0032] The calculated offset compensation is then applied inversely to the initial diemap. This involves adding (or subtracting) the corresponding dynamic coordinate offset value to the theoretical standard coordinates of each die in the initial diemap, thereby generating an actual diemap reflecting the current physical arrangement of the wafer. Figure 3 As shown, the actual wafer diagram displays the true distribution of each die, making it convenient for users to view and record data. For example, the global row and column arrangement topology can accurately identify each die, and even use it as the target die for subsequent processing. The actual physical parameters of each die can also be recorded through charts, background data, and other methods.

[0033] After the above dynamic correction, the original fixed theoretical initial wafer image is updated to an actual wafer image that conforms to the actual deformation, completely eliminating the detection error caused by the overall displacement or local deformation of the wafer, and avoiding the misjudgment and scrapping of qualified products with "safe spacing but absolute coordinate offset".

[0034] S103: Based on the actual wafer diagram, determine the actual parameters and neighborhood parameters of the target die.

[0035] The embodiments of this application do not strictly limit the specific methods and extraction algorithms for determining actual parameters and neighborhood parameters, but it must be ensured that the actual parameters obtained include at least the actual center point coordinates and target edge contour coordinates of the target grain; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain (i.e., the surrounding adjacent grains) adjacent to the target grain.

[0036] For example, based on the row and column topology of the actual wafer diagram, multiple neighboring dies are identified. Since the actual wafer diagram not only shows the true distribution of each die on the diced wafer but also records the true physical parameters of each die, such as the center coordinates and edge contour point coordinates, processing the actual wafer diagram is sufficient to identify the actual center coordinates and edge contour coordinates of the target die, as well as the edge contour coordinates of each neighboring die, thus obtaining the actual parameters and neighborhood parameters of the target die. Specifically, since the true physical parameters of each die are already known during the generation of the actual wafer diagram, the target die can be found based on the row and column topology of the actual wafer diagram, and then the corresponding true physical parameters of the target die, i.e., the actual parameters and neighborhood parameters of the target die, can be directly obtained.

[0037] Of course, algorithms can also be used to obtain the actual parameters and neighborhood parameters of the target grain. For example, a grayscale template matching algorithm combined with a sub-pixel edge extraction algorithm can be used to perform high-precision processing on the actual wafer image. This algorithm accurately identifies the coordinates of the four corners and complete edge contours of each independent grain by matching a preset grain template, and calculates the actual physical coordinates of the target grain in the current wafer image (i.e., the actual center point coordinates and the target edge contour coordinates), as well as the neighborhood edge contour coordinates of each adjacent grain. Through the above identification process, the system generates a dataset of all actual grain coordinates detected in this study, providing a reliable measured data foundation for subsequent calculations of the absolute translational coordinate offset, rotation angle, and evaluation of cutting quality.

[0038] S104: Based on the theoretical parameters, actual parameters, and neighborhood parameters of the target grain, the offset detection results of the target grain are obtained.

[0039] The core of this step lies in constructing a multi-dimensional comprehensive judgment model that takes into account both absolute position and relative safety distance. For example, the specific implementation process for determining the target grain offset detection result may include the following steps: Based on the theoretical center point coordinates in the theoretical parameters and the actual center point coordinates in the actual parameters, the center offset value (i.e., absolute translational deviation) of the target grain is obtained; based on the target edge contour coordinates in the actual parameters and the theoretical edge contour coordinates in the theoretical parameters, the angular offset value of the target grain is obtained; based on the target edge contour coordinates in the actual parameters and the neighborhood edge contour coordinates in the neighborhood parameters, the edge distance data of the target grain is obtained. Using the aforementioned actual wafer map, all adjacent grains of the target grain can be traversed to accurately calculate the actual edge spacing and center distance between adjacent grains. Specifically, it not only calculates the theoretical center offset distance between the current target grain and the theoretical center offset distance before the map update, but more importantly, it calculates the actual edge distance between the current target grain and its surrounding adjacent grains (such as 8-neighborhood grains).

[0040] Based on the center offset value, angle offset value and edge distance data of the target grain, the offset detection result of the target grain is obtained.

[0041] In one exemplary implementation, a dual-condition joint judgment mechanism is employed: if the center offset value, angle offset value, and edge distance data of the target grain meet the alignment conditions, then the qualified cutting position is taken as the offset detection result of the target grain; the alignment conditions specifically include: the center offset value is less than a preset maximum center offset threshold, the angle offset value is less than a preset maximum angle offset threshold, and the edge distance data (i.e., the actual kerf width) is greater than a preset minimum spacing threshold. If the center offset value, angle offset value, and edge distance data of the target grain do not meet the alignment conditions, then the unqualified cutting position is taken as the offset detection result of the target grain. For example, in a real-world scenario, if the center offset value and angle offset value of the target grain (such as die04, die05, or die08 shown in 3) may meet the alignment conditions, but its edge distance with the adjacent grain is too close and does not meet the alignment conditions, then the cutting position is determined to be unqualified.

[0042] In practice, if the offset detection results indicate that the cutting position of the target grain is unqualified, the target grain is marked with anomalies and classified into defects, generating a global distribution map of the anomaly points. The system can clearly distinguish different defect types such as excessive center offset, abnormal edge spacing, and multi-angle offset, and generate a global distribution map of the anomaly points. This not only provides a reliable basis for accurate rejection in the back-end processes but also provides intuitive data support for anomaly investigation and parameter optimization in the front-end cutting process.

[0043] Traditional die cutting inspection relies solely on a single center distance to determine offset, which has limitations such as limited detection dimensions, missed detection of minor offsets, and misjudgment of edge anomalies. To address this pain point, this application innovatively constructs a dual-index detection system based on dynamically corrected real wafer images, consisting of "center offset distance + 8-neighborhood edge spacing," to comprehensively determine the compliance of die cutting offsets and significantly improve the accuracy of defect identification.

[0044] The core detection parameters of this system include two key indicators: first, accurately quantifying the overall displacement deviation of the die, i.e., calculating the offset distance between the current actual center point of a single die and the theoretical center point of the initial wafer pattern; second, comprehensively detecting local anomalies such as proximity, offset, and tilt of the die, i.e., accurately measuring the actual edge spacing between a single die and all its eight neighboring dies. During the acceptance judgment stage, a dual-threshold verification mechanism is adopted, with a maximum center offset threshold, a maximum angle offset threshold, and a minimum safe edge spacing threshold preset in advance according to wafer process standards. Only when the center offset distance of a single die does not exceed the maximum center offset threshold, the angle offset value is less than the preset maximum angle offset threshold, and the edge spacing of all its eight neighboring dies is greater than the minimum safe spacing threshold, can the die cutting position be judged to be compliant and without offset defects; if any one of these indicators exceeds the standard, it is judged as an abnormal cutting offset.

[0045] Through the aforementioned precise screening and interception mechanism, this application can preemptively remove substandard dies, completely preventing abnormal products from flowing into downstream bonding, die bonding, and packaging processes. This effectively reduces material waste and processing losses in subsequent stages, significantly improving finished product yield. Simultaneously, the corrected actual wafer image, full wafer offset data, and abnormal point distribution data will be stored and retained. This data can be used for quality review and batch defect statistics of the current dicing process; furthermore, it can serve as a true alignment benchmark for downstream die bonding and die bonding processes, significantly reducing secondary positioning costs and positioning errors in subsequent processes.

[0046] Overall, this application effectively solves the problems of misjudgment and missed judgment caused by wafer deformation. By combining the actual die offset and the relative spacing between adjacent dies for dual comprehensive evaluation, it successfully avoids the misjudgment and scrapping caused by "large single-point offset but compliant layout spacing". It closely matches the judgment requirements of actual production on site, effectively controlling the product scrap rate while significantly saving production material and labor manufacturing costs. This application abandons the traditional detection mode of a single benchmark with a fixed die distribution map. Through a closed-loop detection process of initial benchmark construction, full-area image acquisition and actual coordinate correction, multi-dimensional offset parameter calculation, and anomaly screening, it achieves high-precision, full-area, and intelligent detection of the cutting offset of each die on the wafer. It can effectively identify quality defects such as die offset, spacing abnormality, and layout deformation during the wafer cutting process. It is suitable for conventional wafers and wafer detection scenarios with slight process deformation, greatly improving the yield and detection accuracy of the wafer cutting process.

[0047] Figure 4 A structural diagram of a grain offset detection device provided in an embodiment of this application is shown below. Figure 4 As shown, based on the grain offset detection method provided in the preceding embodiments, this application also provides a grain offset detection device, including: The parameter determination module is used to determine the theoretical parameters of the target grain based on the initial wafer map of the target wafer. The initial wafer map includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain. The theoretical parameters of the target grain include at least the coordinates of the theoretical center point.

[0048] The correction module is used to dynamically correct the initial wafer image to obtain the actual wafer image of the target wafer.

[0049] The parameter determination module is also used to determine the actual parameters and neighborhood parameters of the target grain based on the actual wafer image; the actual parameters include at least the actual center point coordinates and the target edge contour coordinates; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain.

[0050] The detection module is used to obtain the offset detection results of the target grain based on the theoretical parameters, actual parameters and neighborhood parameters of the target grain.

[0051] As an optional embodiment, the parameter determination module is specifically used for: Obtain the basic data of the target wafer; the basic data includes dicing spacing and die size parameters; plan and construct the initial wafer map based on the basic data.

[0052] Select any grain in the initial wafer diagram as the target grain and determine the theoretical parameters of the target grain.

[0053] As an optional embodiment, the correction module specifically includes: The scanning unit is used to acquire images of the target wafer using a preset scanning path, thereby obtaining multiple scanned images.

[0054] The coordinate offset calculation unit is used to obtain the coordinate offset value of each grain based on multiple scanned images and the initial wafer image.

[0055] The correction unit is used to dynamically correct the initial wafer image based on the coordinate offset value to obtain the actual wafer image of the target wafer.

[0056] As an optional embodiment, the theoretical parameters also include theoretical physical coordinates and a coordinate offset calculation unit, specifically including: The stitching subunit is used to stitch together multiple scanned images to obtain an initial global image; The fitting sub-unit is used to perform feature extraction processing on the initial global image to obtain an enhanced global image.

[0057] The identification sub-unit is used to process the enhanced global image using a grayscale template matching algorithm to identify the actual physical coordinates of each grain.

[0058] The computational sub-unit is used to obtain the coordinate offset value of each grain based on its actual physical coordinates and theoretical physical coordinates.

[0059] As an optional embodiment, the fitting subunit is specifically used for: The initial global image is preprocessed to obtain the preprocessed global image; the preprocessing includes one or more of grayscale conversion, filtering and noise reduction, and contrast enhancement.

[0060] The preprocessed global image is processed by edge detection algorithm to extract grain edges, resulting in the extracted global image.

[0061] Angle fitting is performed on the extracted global image to obtain an enhanced global image.

[0062] As an optional embodiment, the detection module specifically includes: The center offset calculation unit is used to obtain the center offset value of the target grain based on the theoretical center point coordinates in the theoretical parameters and the actual center point coordinates in the actual parameters.

[0063] The angle offset calculation unit is used to obtain the angle offset value of the target grain based on the target edge contour coordinates in the actual parameters and the theoretical edge contour coordinates in the theoretical parameters.

[0064] Edge units are used to obtain the edge distance data of the target grain based on the target edge contour coordinates in the actual parameters and the neighborhood edge contour coordinates in the neighborhood parameters.

[0065] The detection unit is used to obtain the offset detection result of the target grain based on the center offset value, angle offset value and edge distance data of the target grain.

[0066] As an optional embodiment, the detection unit is specifically used for: If the center offset, angle offset, and edge distance data of the target grain meet the alignment conditions, the qualified cutting position is taken as the offset detection result of the target grain. The alignment conditions include the center offset value being less than the preset maximum center offset threshold, the angle offset value being less than the preset maximum angle offset threshold, and the edge distance data being greater than the preset minimum spacing threshold.

[0067] If the center offset, angle offset, and edge distance data of the target grain do not meet the alignment conditions, the unqualified cutting position will be taken as the offset detection result of the target grain.

[0068] This application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a grain offset detection method.

[0069] This application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a grain offset detection method.

[0070] This application provides a computer program product, including a computer program that, when executed by a processor, implements a grain offset detection method.

[0071] It should be noted that the various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for the device and equipment embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the description of the method embodiments. The device and equipment embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components indicated 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 the solution in this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0072] The above description is merely one specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for detecting grain displacement, characterized in that, The method includes: Based on the initial wafer map of the target wafer, the theoretical parameters of the target grain are determined; the initial wafer map includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain; the theoretical parameters of the target grain include at least the coordinates of the theoretical center point; The initial wafer image is dynamically corrected to obtain the actual wafer image of the target wafer; Based on the actual wafer image, the actual parameters and neighborhood parameters of the target grain are determined; the actual parameters include at least the actual center point coordinates and the target edge contour coordinates; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain. Based on the theoretical parameters, actual parameters, and neighborhood parameters of the target grain, the offset detection results of the target grain are obtained.

2. The grain offset detection method according to claim 1, characterized in that, The determination of the theoretical parameters of the target grain based on the initial wafer map of the target wafer specifically includes: Acquire the basic data of the target wafer; the basic data includes dicing spacing and grain size parameters; Based on the aforementioned basic data, an initial wafer diagram is planned and constructed. Select any grain in the initial wafer diagram as the target grain, and determine the theoretical parameters of the target grain.

3. The grain offset detection method according to claim 1, characterized in that, The step of dynamically correcting the initial wafer image to obtain the actual wafer image of the target wafer specifically includes: The target wafer is image acquired using a preset scanning path to obtain multiple scanned images; Based on the multiple scanned images and the initial wafer image, the coordinate offset value of each grain is obtained; The initial wafer image is dynamically corrected based on the coordinate offset value to obtain the actual wafer image of the target wafer.

4. The grain offset detection method according to claim 3, characterized in that, The theoretical parameters also include theoretical physical coordinates. The coordinate offset value for each grain, obtained based on the multiple scanned images and the initial wafer image, specifically includes: The multiple scanned images are stitched together to obtain an initial global image; The initial global image is subjected to feature extraction processing to obtain an enhanced global image; The enhanced global image is processed using a grayscale template matching algorithm to identify the actual physical coordinates of each grain; Based on the actual physical coordinates and theoretical physical coordinates of each grain, the coordinate offset value of each grain is obtained.

5. The grain offset detection method according to claim 4, characterized in that, The step of performing feature extraction processing on the initial global image to obtain the enhanced global image specifically includes: The initial global image is preprocessed to obtain a preprocessed global image; the preprocessing includes one or more of grayscale conversion, filtering and noise reduction, and contrast enhancement. The preprocessed global image is subjected to grain edge extraction processing using an edge detection algorithm to obtain the extracted global image. Angle fitting processing is performed on the extracted global image to obtain an enhanced global image.

6. The grain offset detection method according to claim 1, characterized in that, The process of obtaining the offset detection result of the target grain based on its theoretical parameters, actual parameters, and neighborhood parameters specifically includes: Based on the theoretical center point coordinates in the theoretical parameters and the actual center point coordinates in the actual parameters, the center offset value of the target grain is obtained; The angular offset value of the target grain is obtained based on the target edge contour coordinates in the actual parameters and the theoretical edge contour coordinates in the theoretical parameters. Based on the target edge contour coordinates in the actual parameters and the domain edge contour coordinates in the domain parameters, the edge distance data of the target grains is obtained; Based on the center offset value, angle offset value, and edge distance data of the target grain, the offset detection result of the target grain is obtained.

7. The grain offset detection method according to claim 6, characterized in that, The process of obtaining the offset detection result of the target grain based on the center offset value, angle offset value, and edge distance data of the target grain specifically includes: If the center offset value, angle offset value, and edge distance data of the target grain meet the alignment conditions, then the qualified cutting position is taken as the offset detection result of the target grain; the alignment conditions include the center offset value being less than the preset maximum center offset threshold, the angle offset value being less than the preset maximum angle offset threshold, and the edge distance data being greater than the preset minimum spacing threshold. If the center offset value, angle offset value, and edge distance data of the target grain do not meet the alignment conditions, then the unqualified cutting position is taken as the offset detection result of the target grain.

8. The grain offset detection method according to claim 7, characterized in that, The method further includes: If the offset detection result indicates that the cutting position of the target grain is unqualified, the target grain is marked as abnormal, and a global distribution map of abnormal points is generated.

9. A grain offset detection device, characterized in that, The grain offset detection device includes: The parameter determination module is used to determine the theoretical parameters of the target grain based on the initial wafer map of the target wafer; the initial wafer map includes multiple grains with fixed distribution positions, and any grain can be configured as the target grain; the theoretical parameters of the target grain include at least the coordinates of the theoretical center point; The correction module is used to dynamically correct the initial wafer image to obtain the actual wafer image of the target wafer. The parameter determination module is further configured to determine the actual parameters and neighborhood parameters of the target grain based on the actual wafer image; the actual parameters include at least the actual center point coordinates and the target edge contour coordinates; the neighborhood parameters include at least the neighborhood edge contour coordinates of each neighborhood grain adjacent to the target grain; The detection module is used to obtain the offset detection result of the target grain based on the theoretical parameters, actual parameters and neighborhood parameters of the target grain.

10. The grain offset detection device according to claim 9, characterized in that, The correction module specifically includes: The scanning unit is used to acquire images of the target wafer using a preset scanning path to obtain multiple scanned images; The coordinate offset calculation unit is used to obtain the coordinate offset value of each grain based on the multiple scanned images and the initial wafer image; The correction unit is used to dynamically correct the initial wafer image based on the coordinate offset value to obtain the actual wafer image of the target wafer.

11. The grain offset detection device according to claim 10, characterized in that, The coordinate offset calculation unit specifically includes: The stitching subunit is used to stitch together the multiple scanned images to obtain an initial global image; A fitting subunit is used to perform feature extraction processing on the initial global image to obtain an enhanced global image; The identification subunit is used to process the enhanced global image using a grayscale template matching algorithm to identify the actual physical coordinates of each grain; The calculation subunit is used to obtain the coordinate offset value of each grain based on the actual physical coordinates and theoretical physical coordinates of each grain.

12. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the grain offset detection method according to any one of claims 1-8.

Images