A mask detection program creation method and related device

Abstract

The application discloses a mask detection program creation method and related device, which can be used in the field of defect detection. In the method, first, the received detection trigger instruction is analyzed, and the basic information of the mask to be detected is retrieved and extracted from the mask information database. Then, based on the basic information, the historical detection data is feature extracted and pattern recognized through a preset learning model to generate a first data set. Next, based on the design parameters of the mask to be detected, a second data set, a focusing parameter set, a positioning point and a light adjusting parameter are generated. Finally, the first data set, the second data set, the focusing parameter set, the positioning point and the light adjusting parameter are combined to obtain a detection program. Thus, the creation process of the entire detection program does not need to rely on manual adjustment of parameters, greatly reduces the human participation link, saves a large amount of machine time, avoids the tedious operation and subjective deviation of manual input of parameters, and improves the parameter consistency of the mask detection program.

Description

Technical Field

[0001] This application relates to the field of defect detection technology, and in particular to a method for creating a mask plate detection program and related apparatus. Background Technology

[0002] As the semiconductor industry rapidly develops towards higher precision and higher integration, the pattern accuracy of the photomask, as the core template of the photolithography process, directly determines the chip yield. Therefore, more stringent requirements are placed on the defect detection accuracy and efficiency of the photomask.

[0003] Currently, the creation of mask inspection recipes mainly relies on manual configuration. Operators need to use a camera to collect optical images on the inspection machine or manually delineate the inspection area (ROI), set the defect judgment threshold, and adjust the optical parameters and focusing strategy according to the mask design documents (such as GDSII / OASIS). The whole process is time-consuming and depends on experience accumulation. The parameters of the inspection program are inconsistent, and different operators have different subjective settings for defect judgment thresholds and inspection priorities, which can easily lead to missed or false defects.

[0004] Therefore, how to reduce machine time usage and improve the parameter consistency of the mask detection program has become a problem that needs to be solved. Summary of the Invention

[0005] To address the aforementioned issues, this application provides a method and related apparatus for creating a mask detection program, which can reduce machine time usage and improve the parameter consistency of the mask detection program.

[0006] The embodiments of this application disclose the following technical solutions: In a first aspect, embodiments of this application provide a method for creating a mask detection program, the method comprising: The received detection trigger command is parsed, and the basic information of the mask to be detected is retrieved and extracted from the mask information database. The basic information includes at least the design parameters of the mask to be detected and historical detection data of masks of the same type as the mask to be detected. Based on the aforementioned basic information, a first dataset is generated by performing feature extraction and pattern recognition on the historical detection data using a preset learning model. The first dataset includes global baseline resolution, detection area priority, and defect determination confidence threshold. Based on the design parameters of the mask to be detected, a second dataset, a focusing parameter set, alignment points, and dimming parameters are generated; the second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. The detection procedure is obtained by combining the first dataset, the second dataset, the focusing parameter set, the alignment point, and the dimming parameter.

[0007] Optionally, the design parameters of the mask to be tested are obtained by parsing the design file of the mask to be tested. The generation of a second dataset, a focusing parameter set, alignment points, and dimming parameters based on the obtained design parameters of the mask to be tested includes: Based on the obtained design parameters of the mask to be tested, the coordinates of the detection area of ​​the mask to be tested are extracted, and a second dataset is generated based on the coordinates of the detection area and the design parameters of the mask to be tested. Based on the obtained design parameters of the mask to be tested, the focusing parameters are configured to generate a focusing parameter set; Based on the obtained design parameters of the mask to be tested, the alignment marks in the layout file of the mask to be tested are extracted to determine the alignment points; Based on the design parameters of the mask to be tested, the layer attributes of the mask to be tested are extracted. According to the mapping relationship between the layer attributes and the material, and the target value of the optical parameters defined based on the optical properties of the material, the dimming parameters are generated.

[0008] Optionally, the step of extracting the detection region coordinates of the mask to be detected based on the obtained design parameters of the mask to be detected, and generating a second dataset based on the detection region coordinates and the design parameters of the mask to be detected, includes: Based on the obtained design parameters of the mask to be tested, the regional features of the mask to be tested are determined by a preset regional feature recognition algorithm. If the regional features indicate that the mask to be detected contains a repeating region, then the layout data in the design file of the mask to be detected is scanned layer by layer to extract the starting coordinates, row number, column number and cell spacing of the repeating units, so as to obtain the detection region coordinates of the mask to be detected. If the region feature indicates that the mask to be detected does not contain a repeating region, then the minimum bounding rectangle of the effective pattern in the mask to be detected is extracted to obtain the detection region coordinates of the mask to be detected. A second dataset is generated based on the coordinates of the detection area and the design parameters of the mask to be detected.

[0009] Optionally, the step of extracting the minimum bounding rectangle of the effective pattern in the mask to be detected to obtain the coordinates of the detection area of ​​the mask to be detected includes: Extract the minimum bounding rectangle of the valid pattern in the mask to be detected; The boundary of the minimum bounding rectangle is expanded based on the preset edge redundancy to obtain the detection area coordinates of the mask to be detected.

[0010] Optionally, generating the second dataset based on the coordinates of the detection area and the design parameters of the mask to be detected includes: The coordinates of the detection area are verified to obtain valid coordinates of non-overlapping areas with boundaries within the effective map area; Based on the standard coordinate protocol corresponding to the detection equipment, the effective coordinates are converted into standard coordinates; A second dataset is generated based on the standard coordinates and the design parameters of the mask to be tested.

[0011] Optionally, the step of configuring focusing parameters based on the acquired design parameters of the mask to be detected, and generating a focusing parameter set, includes: Based on the design parameters of the mask to be detected, feature information of each scanning area is extracted; Based on the feature information and the focusing parameter database in the historical detection data, the initial focusing interval is predicted by a deep learning-based aggregation prediction model, generating a focusing parameter set that includes axial focusing position, lateral scanning interval, and focusing stability judgment threshold.

[0012] Optionally, the step of configuring focusing parameters based on the acquired design parameters of the mask to be detected, and generating a focusing parameter set, includes: The area to be detected of the mask to be tested is divided into segmented regions arranged in an array; By designing a file parser, the material parameters and thickness of each segmented region can be identified; Based on the material parameters and thickness of each segmented region, an initial focusing surface is formed through surface fitting. The data of the initial focus surface is synchronized to the automatic focus configuration file to generate a focus parameter set.

[0013] Optionally, the design parameters of the mask to be tested are obtained by parsing the design file of the mask to be tested. The generation of a second dataset, a focusing parameter set, alignment points, and dimming parameters based on the obtained design parameters of the mask to be tested includes: Based on the obtained design parameters of the mask to be tested, the coordinates of the detection area of ​​the mask to be tested are extracted and the coordinates are verified to generate a second dataset; Based on the obtained design parameters of the mask to be tested, the focusing parameters are configured to generate a focusing parameter set; Based on the design parameters of the mask to be tested, a target alignment template matching the mask to be tested is selected from the preset alignment template library, and the alignment marks in the target alignment template are extracted to determine the alignment point. Based on the design parameters of the mask to be tested, the layer attributes of the mask to be tested are extracted. According to the mapping relationship between the layer attributes and the material, and the target value of the optical parameters defined based on the optical properties of the material, the dimming parameters are generated.

[0014] Optionally, the design parameters of the mask to be detected include the user-inputted detection area image, autofocus parameters, alignment template image, and auto-optical parameters. The generation of a second dataset, a set of focus parameters, alignment points, and dimming parameters based on the acquired design parameters of the mask to be detected includes: Based on the image of the detection area and the historical detection data, the priority of the scanning area, resolution requirements, detection mode and algorithm type are recommended to generate a second dataset. Based on the autofocus parameters and the historical detection data, the axial focus position, lateral scanning interval and focus stability judgment threshold are recommended, and a focus parameter set is generated. Based on the alignment template image and the historical detection data, the number and distribution information of alignment sites are generated to determine the alignment sites; Based on the automatic optical parameters and the mapping relationship between material parameters and optical parameters, dimming parameters corresponding to each material region are generated.

[0015] Optionally, based on the autofocus parameters and the historical detection data, the step of recommending axial focus positions, lateral scanning intervals, and focus stability judgment thresholds to generate a focus parameter set includes: Mark the material parameters of each segmented region in the mask to be tested; Based on the autofocus parameters and the material parameters, an initial focus surface is generated; Based on the autofocus parameters corresponding to the initial focusing surface and the historical detection data, the axial focusing position, lateral scanning interval and focusing stability judgment threshold are recommended to generate a focusing parameter set.

[0016] Optionally, the design parameters of the mask to be tested include the detection area selection data, focus point, alignment point, and material area point selected by the user through the plugin. The generation of a second dataset, a focus parameter set, an alignment template, and lighting parameters based on the acquired design parameters of the mask to be tested includes: Based on the detection area selection data and the historical detection data, the priority of the scanning area, resolution requirements, detection mode and algorithm type are recommended to generate a second dataset. Based on the focal points and the material parameters and thickness of each focal point, and combined with the focal data with the same material parameter type in the historical detection data, a focal parameter set is generated; Based on the alignment site and the historical detection data, the alignment site is determined; Based on the location of the material region and the mapping relationship between material parameters and optical parameters, dimming parameters corresponding to each material region are generated.

[0017] Secondly, embodiments of this application provide a mask detection program creation apparatus, the apparatus comprising: The extraction module is used to parse the received detection trigger command, retrieve and extract the basic information of the mask to be detected from the mask information database; the basic information includes at least the design parameters of the mask to be detected and historical detection data of masks of the same type as the mask to be detected; The first data generation module is used to perform feature extraction and pattern recognition on the historical detection data based on the basic information and through a preset learning model to generate a first dataset; the first dataset includes global baseline resolution, detection area priority, and defect judgment confidence threshold. The second data generation module is used to generate a second dataset, a focusing parameter set, alignment points, and dimming parameters based on the design parameters of the mask to be detected. The second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. The combination module is used to combine the first dataset, the second dataset, the focusing parameter set, the alignment point, and the dimming parameters to obtain the detection program.

[0018] Thirdly, embodiments of this application provide a mask detection program creation device, the device comprising: a memory and a processor; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute, according to the program code, the steps of the mask detection program creation method described in any embodiment of the first aspect.

[0019] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program. When the computer program is run on a mask detection program creation device, the mask detection program creation device performs the steps of the mask detection program creation method described in any embodiment of the first aspect.

[0020] Compared with the prior art, this application has the following advantages: This application provides a method for creating a mask detection program. First, a received detection trigger command is parsed, and basic information about the mask to be detected is retrieved and extracted from a mask information database. This basic information includes at least the design parameters of the mask to be detected and historical detection data of masks of the same type as the mask to be detected. Then, based on the mask type, design pattern complexity, material transmittance and reflectivity, and historical defect distribution characteristics in the basic information, feature extraction and pattern recognition are performed on the historical detection data using a preset learning model to generate a first dataset. This first dataset includes a global baseline resolution, detection area priority, and defect judgment confidence threshold. Next, based on the obtained design parameters of the mask to be detected, a second dataset, a focusing parameter set, alignment points, and dimming parameters are generated. This second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. Finally, the first dataset, the second dataset, the focusing parameter set, the alignment points, and the dimming parameters are combined to obtain the detection program.

[0021] Therefore, the entire detection program creation process does not rely on manual parameter adjustment based on human experience, significantly reducing manual intervention, saving considerable machine time, reducing reliance on operator expertise, avoiding tedious manual parameter input and subjective bias, improving parameter consistency of the mask detection program, significantly shortening the detection program creation time, reducing offline waiting costs, and improving the overall efficiency of the detection process. Furthermore, by generating core parameters required for creating the detection program based on historical detection data of similar masks and a pre-set learning model, the consistency of key configurations such as defect judgment confidence thresholds and detection area priorities is ensured, eliminating detection bias caused by differences in operator experience and reducing the risk of missed or false detections. Based on the design parameters of the mask to be inspected, a second dataset, focusing parameter set, alignment point, and lighting parameters are generated, allowing the detection program to better adapt to the detection needs of different masks, improving the detection stability of minute defects and enhancing the program's adaptability. Attached Figure Description

[0022] 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.

[0023] Figure 1 A flowchart illustrating a mask detection program creation method provided in this application embodiment; Figure 2A flowchart of a parameter generation method provided in an embodiment of this application; Figure 3 A flowchart of another parameter generation method provided in this application embodiment; Figure 4 A flowchart illustrating another parameter generation method provided in this application embodiment; Figure 5 A schematic diagram of a mask detection program creation device provided in an embodiment of this application; Figure 6 This is a structural diagram of a fault diagnosis device for a fuel evaporation emission control system provided in an embodiment of this application. Detailed Implementation

[0024] The mask detection program creation method and related apparatus provided in this application can be used in the field of defect detection. The above is only an example and does not limit the application field of the mask detection program creation method and related apparatus provided in this application.

[0025] The terms "first," "second," "third," and "fourth," etc., used in this application specification, claims, and drawings are used to distinguish different objects, not to limit a specific order.

[0026] In the embodiments of this application, the terms "as an example" or "for example" are used to indicate that they are examples, illustrations, or explanations. Any embodiment or design that is described as "as an example" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Specifically, the use of terms such as "as an example" or "for example" is intended to present the relevant concepts in a specific manner.

[0027] The terminology used in the implementation section of this application is for the purpose of explaining specific embodiments of this application only, and is not intended to limit this application.

[0028] 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.

[0029] See Figure 1 The figure is a flowchart of a mask detection program creation method provided in an embodiment of this application. The method includes: S101: Parse the received detection trigger command, retrieve and extract the basic information of the mask to be detected from the mask information database, as well as the historical detection data of masks of the same type as the mask to be detected.

[0030] As an example, the detection trigger command may include the identification of the mask to be detected, such as an ID code; when the same mask ID exists in multiple design iteration versions, the detection trigger command may also include the design version identifier of the mask to be detected, such as a version number, to further clarify the specific version of the mask to be detected and avoid calling incorrect historical detection data or design parameters.

[0031] As an example, after retrieving and extracting the basic information of the mask to be inspected from the mask information database, the basic information can be standardized according to predefined inspection rules, such as defect classification rules, process parameter thresholds, and data verification algorithms. This basic information is then converted into a general information data structure to facilitate subsequent use in driving the created mask inspection program. The basic information of the mask to be inspected includes, but is not limited to: mask ID, design version, material properties, mask type, design pattern complexity, material transmittance and reflectivity, design parameters of the mask to be inspected, historical defect distribution characteristics, and historical inspection data of masks of the same type as the mask to be inspected.

[0032] Specifically, data cleaning can be performed first to filter out outliers, null values ​​or format errors in material properties from historical test data, and then the basic information extracted can be verified to be consistent with the mask to be tested indicated by the test trigger command through a data verification algorithm.

[0033] Then, key features such as high-frequency defect characteristics (such as the proportion of defects like pinholes and scratches), typical defect size distribution, and defect detection rate under different detection parameters are extracted from historical detection data. Detection-related features such as light transmittance range and surface flatness are extracted from material properties.

[0034] Subsequently, the unstructured basic information that has undergone data cleaning and feature extraction is converted into a structured format.

[0035] Finally, the processed basic information is integrated into a general information data structure, which strictly matches the structural specifications of the detection system. This includes: equipment hardware parameters determined based on material properties and historical detection data, such as camera resolution, light source wavelength, and lens magnification; an algorithm version selected based on the algorithm performance in historical detection data and adapted to the current detection; defect classification criteria determined based on predefined defect classification rules; defect judgment thresholds set by combining process parameter thresholds and historical detection data; and a data transmission protocol that specifies the data interaction format and rules during the detection process.

[0036] Using a general information data structure as the execution framework for the inspection system, it can directly drive optical inspection equipment to start the pre-created mask inspection program: the hardware runs according to the equipment's hardware parameters, the algorithm executes according to the specified algorithm version, and the results are output according to the defect classification criteria and defect judgment thresholds, with data interaction conducted through data transmission protocols. The entire process requires no manual intervention, achieving automation and standardization of the inspection workflow.

[0037] S102: Based on basic information, feature extraction and pattern recognition are performed on historical detection data through a preset learning model to generate the first dataset.

[0038] The first dataset includes global baseline resolution, detection area priority, and defect determination confidence threshold.

[0039] Specifically, based on the mask type, design pattern complexity, material light transmittance and reflectivity, and historical defect distribution characteristics in the basic information, a preset learning model, such as a machine learning model based on random forest threshold optimization algorithm or a deep learning model, can be used to extract features and recognize patterns from historical detection data containing a large number of defect samples, including grayscale values, edge gradients, and signal-to-noise ratio data. For different defect types such as pinholes, scratches, and particles, the optimal detection algorithm parameters for each defect type can be determined through model iteration optimization. The optimal detection algorithm parameters include grayscale threshold range, edge detection sensitivity coefficient, and noise filtering parameters.

[0040] Furthermore, the detection algorithm parameters can be dynamically adjusted by combining the design parameters of the mask to be inspected, such as linewidth accuracy requirements and pattern density. Finally, based on data such as linewidth accuracy requirements and minimum defect size, and combined with the correlation between resolution and detection rate in historical detection data, a global baseline resolution is determined. Based on historical defect distribution heatmaps in the historical detection data, detection area priorities are set, such as giving higher priority to high-density defect areas than low-density defect areas. The detection algorithm parameters are converted into quantified confidence scores; for example, a pinhole defect is considered a valid defect if it meets the following conditions: grayscale ≤ 110, edge coefficient 0.8, and filtering parameters 2×2. Thus, a first dataset (Sensitivity dataset) is generated, including the global baseline resolution, detection area priorities, and defect judgment confidence thresholds.

[0041] The first dataset serves as the core parameter configuration file for the optical inspection equipment to perform adaptive scanning, thereby improving the detection rate of minute defects and reducing the false detection rate. This approach avoids the iterative process of manually setting thresholds, performing lengthy inspection scans, and readjusting threshold settings based on scan results, thus improving the efficiency of mask inspection program creation.

[0042] S103: Based on the obtained design parameters of the mask to be detected, generate a second dataset, a focusing parameter set, an alignment template, and dimming parameters.

[0043] The second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type.

[0044] Specifically, in this embodiment, the design parameters of the mask to be tested can be obtained by parsing the design file of the mask to be tested. The flowchart of the parameter generation method for the second dataset, focusing parameter set, alignment template, and dimming parameters is shown below. Figure 2 As shown, the method includes: S21: Based on the obtained design parameters of the mask to be tested, extract the coordinates of the detection area of ​​the mask to be tested, and generate a second dataset based on the coordinates of the detection area and the design parameters of the mask to be tested.

[0045] Specifically, based on the design parameters of the mask to be detected, such as the array distribution representation, the repetition period of the repeating region (Die), and the layer-level structure, the regional features of the mask to be detected can be determined by a preset regional feature recognition algorithm to determine whether the mask to be detected contains repeating regions.

[0046] If the regional features indicate that the mask to be detected contains repeating regions, a design file parser such as GDSII / OASIS can be called to scan the layout data in the design file of the mask to be detected layer by layer, extract the starting coordinates (X0, Y0), number of rows (M), number of columns (N), and cell spacing (ΔX, ΔY) of the repeating cells, and output the array data containing M×N rectangular regions after coordinate transformation, thus obtaining the detection area coordinates of the mask to be detected; where each rectangular region is defined as [X0+i×ΔX, Y0+j×ΔY, Width, Height], where i=0..M-1, j=0..N-1.

[0047] If the region features indicate that the mask to be detected does not contain repeating regions, the minimum bounding rectangle of the effective pattern in the mask to be detected can be extracted by the design file parser. Combined with the user-preset edge redundancy, the boundary is extended to obtain the detection region coordinates of the mask to be detected. Among them, the single detection region rectangle is defined as [MinX, MinY, MaxX, MaxY].

[0048] At the boundary between the effective pattern and the non-patterned area of ​​the mask, there is often a transition effect in optical imaging. By expanding the boundary of the minimum bounding rectangle with the user-preset edge redundancy, the boundary of the resulting detection area extends beyond the edge of the actual pattern. This ensures the detection accuracy of the effective pattern area while avoiding misjudgments caused by the boundary transition effect. It ensures that all effective patterns on the mask to be detected can be completely scanned, which helps to improve the accuracy of the mask detection results.

[0049] As an example, after obtaining the coordinates of the detection area, the coordinates of the detection area can be verified first to obtain valid coordinates that ensure no overlapping areas and that the boundaries are within the effective map area; then, using the standard coordinate protocol supported by the detection device, the valid coordinates can be converted into standard coordinates that the detection device can directly recognize.

[0050] Finally, based on the layout structure priority, line width accuracy requirements, layer material properties, mapping relationship between regional features and algorithm type, and detection area coordinates (or standard coordinates) in the design file of the mask to be detected, a second dataset (ROI dataset) can be generated, which includes scanning area priority, resolution requirements, detection mode, and detection algorithm type.

[0051] The priority of the scanning area can be automatically determined based on the functional partitions in the design parameters. For example, within the detection area, the priority of the core circuit area is higher than that of the peripheral marking area. This is used to plan the detection path, guide the scanning system to scan different areas in priority order, improve detection efficiency and accuracy, and avoid detection omissions and false detections.

[0052] Based on the line width accuracy and pattern density within the detection area extracted from the design documents, resolution requirements can be configured. For example, the finer the line width accuracy, the higher the resolution requirement; the higher the pattern density, the higher the resolution requirement. This allows for the rational allocation of detection resources, enabling accurate detection in core areas and efficient detection in non-critical areas, thereby balancing accuracy and efficiency.

[0053] The material properties and thickness of the detection area can be extracted from the basic information of the mask. Combined with the mapping relationship between materials and detection modes in the process specification library, the detection mode can be configured. For example, the transmission mode can be configured first for transparent materials, and the reflection mode can be configured first for opaque metal layers.

[0054] The detection algorithm type can be determined based on the regional features of the mask to be detected. For example, for a mask to be detected with overlapping regions, an algorithm such as the Die to Die comparison algorithm, which can quickly locate differences by utilizing cell consistency, can be automatically matched. For a mask to be detected without overlapping regions, an algorithm such as the Cell to Cell comparison algorithm can be automatically matched.

[0055] Using the second dataset as input to the detection path planning module guides the detection device to achieve dynamic field of view adjustment. Compared with manually drawing ROIs (Regions of Interest), it improves drawing accuracy and efficiency, and avoids false detections caused by ROIs going out of bounds at the boundary between patterned and non-patterned areas due to low accuracy of manual drawing, as well as missed detections caused by ROI shrinkage.

[0056] S22: Configure focusing parameters based on the obtained design parameters of the mask to be tested, and generate a focusing parameter set.

[0057] One implementation method involves first extracting feature information for each scanning area based on the design parameters of the mask to be tested. This feature information can include pattern density distribution, material reflectivity, and surface smoothness data. Then, combining this feature information with a focusing parameter database from historical testing data, an initial focusing interval is predicted using a deep learning-based aggregation prediction model. This narrows down the focusing interval, and leveraging the consistency of repeating units within repeating regions, batch mapping of focusing parameters is performed within these regions. A high-precision focusing mode is enabled for the core pattern area (e.g., configuring the focus step size to 0.1 μm), while a fast focusing mode is used for the peripheral marking areas (e.g., configuring the focus step size to 0.5 μm). This ultimately generates a focusing parameter set containing axial focusing position, lateral scanning interval, and focusing stability thresholds. The focusing parameter database includes parameters such as focal length and illumination conditions under successful focusing conditions for masks of the same type as the mask to be tested.

[0058] As another implementation method, the area to be detected of the mask to be detected can be divided into arrayed segmented regions first; then, by designing a file parser, the material parameters and thickness of each segmented region can be identified; then, based on the material parameters and thickness of each segmented region, an initial focusing surface can be formed by surface fitting; finally, the data of the initial focusing surface can be synchronized to the automatic focusing configuration file to generate a focusing parameter set.

[0059] In the process of manually configuring automatic focusing parameters, it is impossible to identify the initial focusing surface of the scanning path in advance. It relies entirely on the focusing system calculation during detection, which requires a large stroke and cannot meet the needs of high-precision focusing scenarios. However, the automatic focusing parameter configuration process provided in this application preprocesses the initial focusing surface for early compensation, reduces the stroke of the focusing probe during detection, and can achieve a higher precision automatic focusing effect. The focusing parameters learned by the machine are closer to the actual hardware conditions of the mask detection equipment, which greatly improves the defect detection rate.

[0060] S23: Based on the obtained design parameters of the mask to be tested, extract the alignment marks in the layout of the mask to be tested and determine the alignment points.

[0061] As an example, a three-level linkage mechanism of design file parsing, specification mapping, and feature matching can be used to achieve fully automated extraction of alignment points. Specifically, for masks to be detected with explicit alignment marks, the explicit alignment marks (Marks), such as cross-shaped, T-shaped, or box-shaped alignment marks, can be extracted from the mask layout file by calling the design file parser to determine the alignment points. For masks to be detected without explicit alignment marks, array features such as the angular coordinates of repeating regions can be extracted through array periodic analysis, and alignment point constraint rules such as quantity, distribution spacing, and edge distance in the Specification document can be parsed to construct a candidate set of alignment points. Then, based on the candidate set of alignment points, the regions of candidate alignment points are extracted from the layout file, optical simulation and rendering are performed to generate an alignment template image, and the alignment marks in the alignment template image are extracted to determine the alignment points.

[0062] Furthermore, for masks to be detected without a layout file, multi-scale template matching can be introduced: that is, constructing a matching template library corresponding to different pixel sizes and different lighting conditions to cope with pattern changes caused by different mask types and different detection conditions, as well as pattern size changes caused by mask manufacturing deviations. Then, a target matching template that matches the mask to be detected is selected from the matching template library, and the matching marks in the target matching template are extracted to determine the matching point.

[0063] Therefore, by replacing reliance on manual experience with data-driven design from the source, the accuracy and consistency of alignment point settings are improved, eliminating discrepancies in alignment point settings caused by operator subjective judgment, and resulting in better repeatability of alignment point settings. This application also supports cluster analysis of alignment point data from historical detection data, continuously optimizing the feature weights of the recognition algorithm, such as adding frequently occurring non-standard alignment markers to the template library, further enhancing adaptability to complex masks.

[0064] S24: Based on the obtained design parameters of the mask to be tested, extract the layer attributes of the mask to be tested, and generate dimming parameters according to the mapping relationship between the layer attributes and the material and the preset optical parameters defined based on the optical properties of the material.

[0065] Specifically, layer attributes (Layer / Data Type) can be extracted using a design file parser, and a mapping relationship between layer attributes and materials can be established using the Fab process specification library. In regions with different material types, material area points with regional representativeness, optical stability, and imaging contrast are selected. Preset target values ​​of optical parameters defined based on the material's optical properties are used as input parameters for automatic optical parameter adjustment to generate dimming parameters.

[0066] Multiple material area points can be selected for each material for uniformity correction.

[0067] During the testing process, the program automatically marks the material areas with deviations greater than the alarm threshold and triggers a retest, generating a light intensity distribution heat map and a pass rate report, which are saved to the machine learning sample library. The optical parameter target values ​​are dynamically corrected when the dimming parameters are automatically set next time.

[0068] Manually selecting dimming points is difficult to find suitable locations, while automatic dimming point setting by the program is faster and more targeted. It can also be linked with the detection process, outputting deviation curves and automatically calibrating based on the deviation curves, efficiently generating dimming parameters that include dimming points and optical parameters.

[0069] S104: Combine the first dataset, the second dataset, the focusing parameter set, the alignment template, and the dimming parameters to obtain the detection program.

[0070] Therefore, in this embodiment, the entire detection program creation process does not rely on manual parameter adjustment based on human experience, significantly reducing manual intervention, saving considerable machine time, reducing reliance on operator expertise, avoiding tedious manual parameter input and subjective bias, improving parameter consistency of the mask detection program, significantly shortening the detection program creation time, reducing machine offline waiting costs, and improving the overall efficiency of the detection process. Furthermore, generating the core parameters required for creating the detection program based on historical detection data of similar masks and a preset learning model ensures consistency in key configurations such as defect judgment confidence thresholds and detection area priorities, eliminating detection bias caused by differences in operator experience and reducing the risk of missed or false defects. Based on the design parameters of the mask to be detected, generating a second dataset, focusing parameter set, alignment points, and lighting parameters allows the detection program to better adapt to the detection needs of different masks, improving the detection stability of minute defects and enhancing the program's adaptability.

[0071] In other embodiments provided in this application, the design parameters of the mask to be detected include a user-inputted detection area image, autofocus parameters, alignment template image, and autooptic parameters.

[0072] As an example, in cases where the mask design file is missing or permissions are insufficient, an offline tool can be provided. Users can directly input or import detection area images, autofocus parameters, alignment template images, and automatic optical parameters through the offline tool. The offline tool combines the user-input design parameters of the mask to be detected, historical detection data, and machine learning database to automatically generate detection program files.

[0073] Specifically, the flowchart for generating the parameters of the second dataset, the focus parameter set, the alignment template, and the dimming parameters is as follows: Figure 3 As shown, the method includes: S31: Based on the detection area image and historical detection data, recommend scanning area priority, resolution requirements, detection mode and algorithm type, and generate a second dataset.

[0074] The offline tool can optimize the creation of detection areas through a combination of interactive operation and intelligent assistance. It provides manual marking functions such as canvas drawing and image import, as well as a built-in standard array layout template library for quick selection. At the same time, it combines historical databases to recommend complete array schemes for the detection area images input by users and obtain the coordinates of the detection areas.

[0075] The offline tool supports two parameterized configuration modes: repeating regions (automatically generating M×N regions by inputting the starting coordinates, number of rows and columns, and spacing) and single regions (inputting coordinates and customizing edge redundancy). It can automatically verify the compliance of overlapping regions and boundaries, highlight anomalies, and provide one-click correction to obtain more accurate detection region coordinates.

[0076] Finally, based on historical cases of detection area coordinates, we recommend scanning area priorities, resolution requirements, detection modes, and algorithm types to generate a second dataset that accurately matches the detection requirements.

[0077] S32: Generate a set of focusing parameters based on the automatic focusing parameters, the initial focusing surface generated according to the material parameters of each segmented region in the mask to be detected, and historical detection data.

[0078] The offline tool offers multi-mode focusing parameter configuration. For example, in quick mode, users can select the mask type to retrieve the best historical focusing parameters for the same type. Advanced mode supports manual adjustment of the core and peripheral region focusing step sizes to obtain automatic focusing parameters. Furthermore, the area to be inspected on the mask can be divided into arrayed segments. After the user marks the material parameters for each segment, the system can generate an initial focusing surface based on the material thickness database and provide a 3D preview. It also supports importing and reusing historical focusing curves as automatic focusing parameters. The material parameters include the physical or chemical properties exhibited by the material under specific processes.

[0079] Based on autofocus parameters and historical inspection data, the offline tool can automatically recommend axial focusing positions, lateral scanning intervals, and focusing stability judgment thresholds, generating a set of focusing parameters to help users find a balance between inspection speed and accuracy, reduce the travel of the focusing probe during inspection, and improve focusing accuracy and defect detection rate.

[0080] S33: Based on the alignment template image and historical detection data, generate information on the number and distribution of alignment points to determine the alignment points.

[0081] The offline tool has a built-in library of standard alignment marker templates such as cross and T-shapes, supporting direct selection, creation via camera capture, or importing and reusing historical templates. It automatically generates alignment template variations under different resolutions and lighting conditions to improve matching robustness and provides a preview of matching accuracy. Then, based on historical detection data, it recommends the optimal number and distribution of alignment points, or suggests optimal supplementary points after the user manually marks alignment points to confirm their location. Furthermore, the offline tool can be configured with an alignment point accuracy evaluation function to quantify the reliability of each point, replacing manual online point selection, solving accuracy, repeatability, and consistency issues, and eliminating differences in the detection process caused by operator subjective judgment.

[0082] S34: Based on the automatic optical parameters and the mapping relationship between material parameters and optical parameters, generate the dimming parameters corresponding to each material region.

[0083] The offline tool can establish a mapping relationship between material parameters and optical parameters. After the user marks the material of a region, the tool can automatically match and recommend optical parameters. It also supports custom material tags and parameters and saves them to the user library. Based on the material region, the tool can automatically recommend representative dimming points. Users can also manually add or delete dimming points. The offline tool updates the light intensity distribution prediction in real time and provides an interface for fine-tuning optical parameters and effect previews, so as to generate dimming parameters corresponding to each material region based on the dimming points and optical parameters.

[0084] After the inspection, the offline tool can also generate a light intensity distribution report and record the history of the user's adjustment of optical parameters to form a personalized optimization model. This not only solves the problem of manually selecting appropriate dimming points on complex patterns, but also continuously improves the adaptability of optical parameters through closed-loop optimization, ensuring the defect detection capability.

[0085] Therefore, in this embodiment of the application, without relying on mask design documents, a unified input window is used, and a templated and configurable approach can save time in creating detection programs and reduce the possibility of errors.

[0086] In some other embodiments provided in this application, the design parameters of the mask to be tested include the detection area selection data, focus point, alignment point, and material area point selected by the user through the plugin.

[0087] As an example, when the mask design file is complete but access is lacking, a plug-in program can be written. When the user is creating a JDV (job deck view, overall inspection workflow view), it interfaces with the corresponding design file viewer and EDA software. While previewing the design file, the user can use the plug-in program to select focal points for autofocus and label the material type for each focal point, select material area points for automatic optical parameter adjustment, select the inspection area to be inspected, and select alignment points. Finally, after collecting all the information, the plug-in program converts it into a data structure adapted to the inspection program file, providing inspection area selection data, focal points, alignment points, and material area points to support the automated creation of the inspection program.

[0088] Specifically, the flowchart for generating the parameters of the second dataset, the focus parameter set, the alignment template, and the dimming parameters is as follows: Figure 4 As shown, the method includes: S41: Based on the detection area selection data and historical detection data, recommend scanning area priority, resolution requirements, detection mode and algorithm type, and generate a second dataset.

[0089] As an example, the plugin is deeply integrated with EDA software or JDV views, allowing users to directly select the detection area when previewing design files. It supports various selection tools such as polygons, rectangles, and circles, and also provides intelligent assistance functions: automatically generating suggestions for repeating areas by recognizing array features in the design file, and automatically filling in the number of rows, columns, and spacing parameters after user confirmation; supporting batch selection and editing, allowing translation, scaling, and copying of selected areas; the plugin can check in real time whether the boundaries of the detection area exceed the effective map range, highlighting overlapping areas and providing merge and split options to obtain the coordinates of the detection area; after the user completes the selection, the plugin automatically associates historical detection data, recommends the corresponding scanning detection priority, resolution requirements, detection mode, and algorithm type for this type of detection area, and generates a second dataset that conforms to the detection device coordinate protocol after one-click confirmation.

[0090] S42: Based on the focal point and the material parameters and thickness of each focal point, and combined with the focal data with the same material parameter type in the historical test data, generate a focal parameter set.

[0091] The plugin provides a focus point marking tool in the JDV view. Users can directly select key areas as focus points in the design file preview interface, such as the core pattern area and the outer marking area, and label the material type for each focus point. The plugin can automatically record the coordinates of the focus points, calculate the material height difference by combining it with the layer thickness information in the design file, generate an initial height matrix, and form focus point marking data. For array areas, it supports single-point marking and batch application functions. After the user marks the focus point of a repeating unit, it automatically maps to all repeating units. A focus parameter preset panel is provided, which allows configuration of focal length steps for the core area and the outer area respectively. The plugin combines the focus data with the same material parameter type in historical detection data to recommend axial focus positions and stability thresholds, generating a focus parameter set. Optionally, the final generated focus parameter set can be previewed in real time in the 3D view to facilitate fine-tuning and optimization by the user.

[0092] S43: Determine the alignment site based on the alignment site and historical detection data.

[0093] The plugin provides a quick alignment point marking function in the design file preview interface, automatically identifying and highlighting explicit alignment points in the view. Users can directly select or manually select implicit feature points. After selection, the plugin automatically extracts the marked area image, generates a multi-scale template with alignment point variations of different scaling ratios and rotation angles, and simultaneously parses the coordinate information in the design file as a reference value. It supports batch verification of alignment points, simulating the matching effect under offset conditions in the preview interface and displaying the score. After user confirmation, the plugin associates and stores the alignment template image with the coordinate data, and compares it with the historical alignment point database in historical detection data to recommend the optimal alignment point distribution scheme, such as a five-point layout with four corners and a center, to determine the alignment points and ensure that the number and spacing of alignment points meet the specifications. The alignment template generated based on the determined alignment points can be directly used for matching calibration during detection.

[0094] S44: Based on the location of material regions and the mapping relationship between material parameters and optical parameters, generate the dimming parameters corresponding to each material region.

[0095] The plugin supports quick filtering of regions by layer and / or material in the JDV view. Users can mark representative points (material region points) for different materials. The plugin automatically links to the Fab process specification library and matches the initial values ​​of the corresponding optical parameters for each material based on the mapping relationship between material parameters and optical parameters. It provides a parameter preview function, simulating the imaging effect under different light intensities and wavelengths in real time in the view. It supports point uniformity verification, automatically calculating the distribution density of marked material region points and prompting whether to add edge region points. After user confirmation, the plugin can bind the coordinates of the material region points to the optical parameters, generating a configuration set containing target values ​​and deviation thresholds, and synchronizing it to the machine learning library to generate the corresponding dimming parameters for each material region. If parameter deviations occur during detection, the plugin can backtrack and mark the corresponding material region points in the JDV view, facilitating recalibration in the design file later, forming a closed loop of marking-detection-optimization.

[0096] Therefore, through deep integration between the plugin and design file viewer (such as EDA software or JDV view), the original offline manual input is transformed into a visual interaction based on design source data. This avoids the permission restrictions of directly parsing design files, preserves the accuracy of design information, improves the efficiency of creating the detection program, and ensures the consistency of parameter configuration through historical detection data and intelligent recommendation function.

[0097] See Figure 5 The figure is a schematic diagram of a mask detection program creation device provided in an embodiment of this application. The device includes: The extraction module 501 is used to parse the received detection trigger command, retrieve and extract the basic information of the mask to be detected from the mask information database; the basic information includes at least the design parameters of the mask to be detected and the historical detection data of masks of the same type as the mask to be detected; The first data generation module 502 is used to extract features and recognize patterns from historical detection data based on basic information and through a preset learning model to generate a first dataset; the first dataset includes global baseline resolution, detection area priority, and defect judgment confidence threshold. The second data generation module 503 is used to generate a second dataset, a focusing parameter set, alignment points, and dimming parameters based on the design parameters of the acquired mask to be detected. The second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. The combination module 504 is used to combine the first dataset, the second dataset, the focus parameter set, the alignment point, and the dimming parameters to obtain the detection program.

[0098] Therefore, in this embodiment, the entire detection program creation process does not rely on manual parameter adjustment based on human experience, significantly reducing manual intervention, lowering reliance on operator skills, avoiding tedious manual parameter input and subjective bias, improving parameter consistency of the mask detection program, significantly shortening the detection program creation time, reducing offline waiting costs, and improving the overall efficiency of the detection process. Furthermore, generating the core parameters required for creating the detection program based on historical detection data of similar masks and a preset learning model ensures consistency in key configurations such as defect judgment confidence thresholds and detection area priorities, eliminating detection bias caused by differences in operator experience and reducing the risk of missed or false detections. Based on the design parameters of the mask to be detected, generating a second dataset, focusing parameter set, alignment points, and lighting parameters allows the detection program to better adapt to the detection needs of different masks, improving the detection stability of minute defects and enhancing the program's adaptability.

[0099] Optionally, the second data generation module 503 includes a second dataset generation unit, a focus parameter set generation unit, a determination unit, and a dimming parameter generation unit. The second dataset generation unit is used to extract the detection area coordinates of the mask to be tested based on the acquired design parameters of the mask to be tested, and generate a second dataset based on the detection area coordinates and the design parameters of the mask to be tested. The focus parameter set generation unit is used to configure focus parameters based on the acquired design parameters of the mask to be tested, and generate a focus parameter set. The determination unit is used to extract alignment marks from the layout file of the mask to be tested based on the acquired design parameters of the mask to be tested, and determine alignment points. The dimming parameter generation unit is used to extract the layer attributes of the mask to be tested based on the acquired design parameters of the mask to be tested, and generate dimming parameters according to the mapping relationship between layer attributes and materials and the target values ​​of optical parameters defined based on the optical properties of the materials.

[0100] Optionally, the second dataset generation unit is specifically used for: determining the regional features of the mask to be detected based on the obtained design parameters of the mask to be detected using a preset regional feature recognition algorithm; if the regional features indicate that the mask to be detected contains repeating regions, then scanning the layout data in the design file of the mask to be detected layer by layer, extracting the starting coordinates, row number, column number, and cell spacing of the repeating units to obtain the detection region coordinates of the mask to be detected; if the regional features indicate that the mask to be detected does not contain repeating regions, then extracting the minimum bounding rectangle of the effective pattern in the mask to be detected to obtain the detection region coordinates of the mask to be detected; and generating the second dataset based on the detection region coordinates and the design parameters of the mask to be detected.

[0101] Optionally, the focusing parameter set generation unit is specifically used to: extract feature information of each scanning area based on the design parameters of the obtained mask to be detected; and predict the initial focusing interval based on the feature information and the focusing parameter database in the historical detection data through a deep learning-based aggregation prediction model, thereby generating a focusing parameter set that includes axial focusing position, lateral scanning interval and focusing stability judgment threshold.

[0102] Optionally, the focusing parameter set generation unit is specifically used for: dividing the area to be detected of the mask to be detected into arrayed segmented regions; identifying the material parameters and thickness of each segmented region through a design file parser; forming an initial focusing surface based on the material parameters and thickness of each segmented region through surface fitting; and synchronizing the data of the initial focusing surface to the automatic focusing configuration file to generate a focusing parameter set.

[0103] Optionally, the second data generation module 503 includes a second dataset generation unit, a focus parameter set generation unit, a determination unit, and a dimming parameter generation unit. The second dataset generation unit is used to extract the detection area coordinates of the mask to be tested based on the acquired design parameters of the mask to be tested and perform coordinate verification to generate a second dataset. The focus parameter set generation unit is used to configure focus parameters based on the acquired design parameters of the mask to be tested to generate a focus parameter set. The determination unit is used to select a target alignment template matching the mask to be tested from a preset alignment template library based on the acquired design parameters of the mask to be tested, and extract alignment marks from the target alignment template to determine alignment points. The dimming parameter generation unit is used to extract the layer attributes of the mask to be tested based on the acquired design parameters of the mask to be tested, and generate dimming parameters according to the mapping relationship between the layer attributes and the material, and the target values ​​of optical parameters defined based on the optical properties of the material.

[0104] Optionally, the design parameters of the mask to be tested include the user-inputted detection area image, autofocus parameters, alignment template image, and automatic optical parameters. The second data generation module 503 is specifically used to: recommend scanning area priority, resolution requirements, detection mode, and algorithm type based on the detection area image and historical detection data, and generate a second dataset; recommend axial focus position, lateral scanning interval, and focus stability judgment threshold based on the autofocus parameters and historical detection data, and generate a focus parameter set; generate the number and distribution information of alignment points based on the alignment template image and historical detection data to determine the alignment points; and generate the corresponding dimming parameters for each material region based on the automatic optical parameters and the mapping relationship between material parameters and optical parameters.

[0105] Optionally, the design parameters of the mask to be tested include the detection area selection data, focus points, alignment points, and material area points selected by the user through the plugin. The second data generation module 503 is specifically used for: recommending scanning area priority, resolution requirements, detection modes, and algorithm types based on the detection area selection data and historical detection data, and generating a second dataset; generating a focus parameter set based on the focus points and the material parameters and thickness of each focus point, combined with focus data with the same material parameter type in the historical detection data; determining alignment points based on alignment points and historical detection data; and generating the corresponding dimming parameters for each material area based on the material area points and the mapping relationship between material parameters and optical parameters.

[0106] See Figure 6 The figure is a structural diagram of a mask detection program creation device provided in an embodiment of this application. The device includes a memory 601 and a processor 602.

[0107] Memory 601: Used to store program code and transfer program code to the processor.

[0108] Processor 602: Used to execute the steps of the mask detection program creation method described above according to the instructions in the program code.

[0109] In addition, this application also provides a computer-readable storage medium storing computer instructions, which, when executed on a mask detection program creation device, perform the steps of the above-described mask detection program creation method.

[0110] 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 storage medium 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 storage medium 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 this embodiment solution according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0111] 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 creating a mask detection program, characterized in that, The method includes: The received detection trigger command is parsed, and the basic information of the mask to be detected is retrieved and extracted from the mask information database. The basic information includes at least the design parameters of the mask to be detected and historical detection data of masks of the same type as the mask to be detected. Based on the aforementioned basic information, a first dataset is generated by performing feature extraction and pattern recognition on the historical detection data using a preset learning model. The first dataset includes global baseline resolution, detection area priority, and defect determination confidence threshold. Based on the design parameters of the mask to be detected, a second dataset, a focusing parameter set, alignment points, and dimming parameters are generated; the second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. The detection procedure is obtained by combining the first dataset, the second dataset, the focusing parameter set, the alignment point, and the dimming parameter.

2. The method according to claim 1, characterized in that, The design parameters of the mask to be tested are obtained by parsing the design file of the mask to be tested. Based on the obtained design parameters of the mask to be tested, a second dataset, a focusing parameter set, alignment points, and dimming parameters are generated, including: Based on the obtained design parameters of the mask to be tested, the coordinates of the detection area of ​​the mask to be tested are extracted, and a second dataset is generated based on the coordinates of the detection area and the design parameters of the mask to be tested. Based on the obtained design parameters of the mask to be tested, the focusing parameters are configured to generate a focusing parameter set; Based on the obtained design parameters of the mask to be tested, the alignment marks in the layout file of the mask to be tested are extracted to determine the alignment points; Based on the design parameters of the mask to be tested, the layer attributes of the mask to be tested are extracted. According to the mapping relationship between the layer attributes and the material, and the target value of the optical parameters defined based on the optical properties of the material, the dimming parameters are generated.

3. The method according to claim 2, characterized in that, The process involves extracting the detection region coordinates of the mask under test based on the acquired design parameters of the mask, and generating a second dataset based on the detection region coordinates and the design parameters of the mask under test, including: Based on the obtained design parameters of the mask to be tested, the regional features of the mask to be tested are determined by a preset regional feature recognition algorithm. If the regional features indicate that the mask to be detected contains a repeating region, then the layout data in the design file of the mask to be detected is scanned layer by layer to extract the starting coordinates, row number, column number and cell spacing of the repeating units, so as to obtain the detection region coordinates of the mask to be detected. If the region feature indicates that the mask to be detected does not contain a repeating region, then the minimum bounding rectangle of the effective pattern in the mask to be detected is extracted to obtain the detection region coordinates of the mask to be detected. A second dataset is generated based on the coordinates of the detection area and the design parameters of the mask to be detected.

4. The method according to claim 3, characterized in that, The step of extracting the minimum bounding rectangle of the effective pattern in the mask to be detected, and obtaining the coordinates of the detection area of ​​the mask to be detected, includes: Extract the minimum bounding rectangle of the valid pattern in the mask to be detected; The boundary of the minimum bounding rectangle is expanded based on the preset edge redundancy to obtain the detection area coordinates of the mask to be detected.

5. The method according to claim 3, characterized in that, The generation of the second dataset based on the detection region coordinates and the design parameters of the mask to be detected includes: The coordinates of the detection area are verified to obtain valid coordinates of non-overlapping areas with boundaries within the effective map area; Based on the standard coordinate protocol corresponding to the detection equipment, the effective coordinates are converted into standard coordinates; A second dataset is generated based on the standard coordinates and the design parameters of the mask to be tested.

6. The method according to claim 2, characterized in that, The focusing parameter configuration is performed based on the obtained design parameters of the mask to be detected, generating a focusing parameter set, including: Based on the design parameters of the mask to be detected, feature information of each scanning area is extracted; Based on the feature information and the focusing parameter database in the historical detection data, the initial focusing interval is predicted by a deep learning-based aggregation prediction model, generating a focusing parameter set that includes axial focusing position, lateral scanning interval, and focusing stability judgment threshold.

7. The method according to claim 2, characterized in that, The focusing parameter configuration is performed based on the obtained design parameters of the mask to be detected, generating a focusing parameter set, including: The area to be detected of the mask to be tested is divided into segmented regions arranged in an array; By designing a file parser, the material parameters and thickness of each segmented region can be identified; Based on the material parameters and thickness of each segmented region, an initial focusing surface is formed through surface fitting. The data of the initial focus surface is synchronized to the automatic focus configuration file to generate a focus parameter set.

8. The method according to claim 1, characterized in that, The design parameters of the mask to be tested are obtained by parsing the design file of the mask to be tested. Based on the obtained design parameters of the mask to be tested, a second dataset, a focusing parameter set, alignment points, and dimming parameters are generated, including: Based on the obtained design parameters of the mask to be tested, the coordinates of the detection area of ​​the mask to be tested are extracted and the coordinates are verified to generate a second dataset; Based on the obtained design parameters of the mask to be tested, the focusing parameters are configured to generate a focusing parameter set; Based on the design parameters of the mask to be tested, a target alignment template matching the mask to be tested is selected from the preset alignment template library, and the alignment marks in the target alignment template are extracted to determine the alignment point. Based on the design parameters of the mask to be tested, the layer attributes of the mask to be tested are extracted. According to the mapping relationship between the layer attributes and the material, and the target value of the optical parameters defined based on the optical properties of the material, the dimming parameters are generated.

9. The method according to claim 1, characterized in that, The design parameters of the mask to be tested include the user-input detection area image, autofocus parameters, alignment template image, and automatic optical parameters. Based on the acquired design parameters of the mask to be tested, a second dataset, a set of focus parameters, alignment points, and dimming parameters are generated, including: Based on the image of the detection area and the historical detection data, the priority of the scanning area, resolution requirements, detection mode and algorithm type are recommended to generate a second dataset. Based on the autofocus parameters and the historical detection data, the axial focus position, lateral scanning interval and focus stability judgment threshold are recommended, and a focus parameter set is generated. Based on the alignment template image and the historical detection data, the number and distribution information of alignment sites are generated to determine the alignment sites; Based on the automatic optical parameters and the mapping relationship between material parameters and optical parameters, dimming parameters corresponding to each material region are generated.

10. The method according to claim 9, characterized in that, Based on the autofocus parameters and the historical detection data, the system recommends axial focus positions, lateral scanning intervals, and focus stability thresholds, generating a focus parameter set, including: Mark the material parameters of each segmented region in the mask to be tested; Based on the autofocus parameters and the material parameters, an initial focus surface is generated; Based on the autofocus parameters corresponding to the initial focusing surface and the historical detection data, the axial focusing position, lateral scanning interval and focusing stability judgment threshold are recommended to generate a focusing parameter set.

11. The method according to claim 1, characterized in that, The design parameters of the mask to be tested include the detection area selection data, focus point, alignment point, and material area point selected by the user through the plugin. Based on the acquired design parameters of the mask to be tested, a second dataset, a focus parameter set, an alignment template, and lighting parameters are generated, including: Based on the detection area selection data and the historical detection data, the priority of the scanning area, resolution requirements, detection mode and algorithm type are recommended to generate a second dataset. Based on the focal points and the material parameters and thickness of each focal point, and combined with the focal data with the same material parameter type in the historical detection data, a focal parameter set is generated; Based on the alignment site and the historical detection data, the alignment site is determined; Based on the location of the material region and the mapping relationship between material parameters and optical parameters, dimming parameters corresponding to each material region are generated.

12. A mask plate detection program creation device, characterized in that, The device includes: The extraction module is used to parse the received detection trigger command, retrieve and extract the basic information of the mask to be detected from the mask information database; the basic information includes at least the design parameters of the mask to be detected and historical detection data of masks of the same type as the mask to be detected; The first data generation module is used to perform feature extraction and pattern recognition on the historical detection data based on the basic information and through a preset learning model to generate a first dataset; the first dataset includes global baseline resolution, detection area priority, and defect judgment confidence threshold. The second data generation module is used to generate a second dataset, a focusing parameter set, alignment points, and dimming parameters based on the design parameters of the mask to be detected. The second dataset includes scanning area priority, resolution requirements, detection mode, and detection algorithm type. The combination module is used to combine the first dataset, the second dataset, the focusing parameter set, the alignment point, and the dimming parameters to obtain the detection program.

13. A mask plate detection program creation device, characterized in that, The device includes: a memory and a processor; The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the steps of the mask detection program creation method according to any one of claims 1-11, based on the program code.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is run on the mask detection program creation device, the mask detection program creation device performs the steps of the mask detection program creation method as described in any one of claims 1-11.

Images