Mask pattern processing methods, mask pattern processing systems, and masks.

Abstract

This application provides a method, system, and mask for processing mask patterns. The method is applied to a mask formed by splicing at least two sub-masks, the mask including a splicing region formed based on the overlapping areas of adjacent sub-masks. The method includes: acquiring a mask pattern on the mask and splitting it into a splicing pattern and a non-splicing pattern; predicting the photolithography behavior of the mask pattern using a full-pattern optical proximity effect model of the corresponding mask pattern to obtain a predicted photolithography pattern; when a defect is detected in the predicted splicing photolithography pattern within the predicted photolithography pattern, adjusting the critical dimensions at the position corresponding to the defect in the splicing pattern according to the defect detection result and a critical dimension adjustment rule of the splicing pattern; and outputting the processed target mask pattern. This application improves the efficiency of defect processing for mask patterns located within the splicing region before mask fabrication.

Description

Technical Field

[0001] The embodiments in this application relate to the field of semiconductor technology, specifically to a mask pattern processing method, a mask pattern processing system, and a mask. Background Technology

[0002] A Complementary Metal Oxide Semiconductor Image Sensor (CIS) is an integrated circuit used to convert light signals into digital signals to create images. Due to its advantages such as low power consumption, high performance, and ease of integration, it has been widely used in many fields, including consumer electronics, security monitoring, driver assistance systems, and medical devices. In current CIS manufacturing processes, the photomask is crucial for transferring the designed circuit pattern to the wafer. The transmittance, phase, and other parameters of the photomask directly affect the lithographic resolution and the fidelity of the lithographic pattern.

[0003] Limited by the optical system setup in lithography equipment and the processing precision of mask materials, the physical size of the mask has an upper limit, which in turn restricts the effective photosensitive area size of a single CIS chip. To develop CIS products with large size and high resolution, existing technologies typically utilize a stitching process to expand the physical size of the mask, forming a stitched mask, thereby increasing the effective photosensitive area of ​​the CIS chip to meet the manufacturing requirements of full-frame or large-size sensors.

[0004] However, in the existing process of manufacturing splicing masks, before the mask tapeout, defects in the mask pattern located in the splicing area are handled by the joint design and verification (JDV) method, which has low defect handling efficiency. Summary of the Invention

[0005] In view of this, several embodiments of this application provide a method for processing mask patterns, a system for processing mask patterns, and a mask, so as to improve the efficiency of defect processing of mask patterns located in the splicing area before mask fabrication.

[0006] In one aspect, an embodiment of this application provides a method for processing a mask pattern, the method being applied to a mask formed by splicing at least two sub-masks; wherein any of the sub-masks includes an overlapping region; a splicing region of the mask is formed based on the overlapping regions of adjacent sub-masks; the method for processing the mask pattern includes: acquiring a mask pattern on the mask; splitting the mask pattern into a splicing pattern and a non-splicing pattern; wherein the splicing pattern is a mask pattern on the sub-mask that falls within the splicing region; the non-splicing pattern is a mask pattern on the sub-mask that falls outside the splicing region; the splicing pattern The corresponding transmittance is different from that of the non-stitched pattern; using the full-pattern optical proximity effect model corresponding to the mask pattern, the photolithography behavior of the mask pattern is predicted to obtain a predicted photolithography pattern; wherein, the predicted photolithography pattern includes a predicted stitched photolithography pattern corresponding to the stitched pattern; when a defect is detected in the predicted stitched photolithography pattern, the key dimensions of the position corresponding to the defect in the predicted stitched photolithography pattern are adjusted according to the defect detection result of the predicted stitched photolithography pattern and the key dimension adjustment rule of the stitched pattern; the processed target mask pattern is output.

[0007] Optionally, the predicted lithography pattern further includes a predicted non-stitched lithography pattern corresponding to the non-stitched pattern; the mask pattern processing method further includes: when a defect is detected in the predicted non-stitched lithography pattern, optical proximity effect correction processing is performed on the position in the non-stitched pattern corresponding to the defect in the predicted non-stitched lithography pattern according to the defect detection result of the predicted non-stitched lithography pattern.

[0008] Optionally, the mask pattern includes multiple mask patterns; the key dimension composite adjustment rule of the splicing pattern includes a baseline adjustment rule and a pre-trained linear regression fitting model; when a defect is detected in the predicted splicing lithography pattern, the step of adjusting the key dimensions of the position corresponding to the defect in the splicing pattern according to the defect detection result of the predicted splicing lithography pattern and the key dimension composite adjustment rule of the splicing pattern includes: dividing the mask pattern into multiple pattern types according to the distribution density of the mask patterns in the mask pattern; determining the defect mask pattern corresponding to the defect in the predicted splicing lithography pattern according to the defect detection result of the predicted splicing lithography pattern; and obtaining the target image corresponding to the defect mask pattern. The method includes: a case type, a target key dimension, and the target non-stitched key dimension and predicted key dimension corresponding to the target pattern type and the target key dimension; wherein, the predicted key dimension is predicted based on the target key dimension using the full-pattern optical proximity effect model; according to the target non-stitched key dimension and the benchmark adjustment rule, the target key dimension is adjusted by a benchmark to obtain the adjustment benchmark of the target key dimension; based on the target pattern type, the target key dimension and the predicted key dimension, the target key dimension is adjusted by an offset using the pre-trained linear regression fitting model to obtain the adjustment offset of the target key dimension; the adjustment benchmark and the adjustment offset are summed to obtain the adjusted key dimension corresponding to the target key dimension.

[0009] Optionally, the linear regression fitting model includes multiple sub-fitting models; wherein each sub-fitting model has a corresponding pattern type and target key dimension value range; the step of adjusting the target key dimension by offsetting it using the pre-trained linear regression fitting model based on the target pattern type, the target key dimension, and the predicted key dimension to obtain the adjusted offset of the target key dimension includes: inputting the target key dimension and the predicted key dimension into a specified sub-fitting model, and using the output of the specified sub-fitting model as the iterative offset of the target key dimension; wherein the specified sub-fitting model is determined based on the target key dimension value range into which the target pattern type and the target key dimension fall; updating the adjustment benchmark using the sum of the adjustment benchmark and the iterative offset of the target key dimension; making a prediction using the full-pattern optical proximity effect model based on the updated adjustment benchmark, and updating the predicted key dimension with the output of the full-pattern optical proximity effect model; performing a difference operation on the target key dimension and the updated predicted key dimension to obtain the prediction error; repeating the above steps if the prediction error falls outside the specified error range until the prediction error falls within the specified error range.

[0010] Optionally, before the step of adjusting the offset of the target key size using the pre-trained linear regression fitting model based on the target pattern type, the target key size, and the predicted key size to obtain the adjusted offset of the target key size, the mask pattern processing method further includes: obtaining a training key size for training the linear regression fitting model; wherein, the training key size includes a training target key size corresponding to the plurality of pattern types, a training predicted key size corresponding to the training target key size, and a training offset corresponding to the training target key size; the training predicted key size utilizes the full-pattern optical proximity effect model. The training target key dimensions are predicted; the following sub-steps are repeated until a linear regression fitting model is determined for adjusting the offset of the target key dimensions corresponding to the plurality of pattern types; the sub-steps include: analyzing the training target key dimensions corresponding to any pattern type using the K-means clustering analysis algorithm, dividing the training target key dimensions corresponding to any pattern type into multiple value intervals; performing linear regression fitting on all training target key dimensions falling within any of the value intervals, the training predicted key dimensions corresponding to all training target key dimensions, and the training offset, to determine the linear regression fitting model corresponding to any of the value intervals.

[0011] Optionally, before the step of predicting the photolithography behavior of the mask pattern using a full-pattern optical proximity effect model corresponding to the mask pattern to obtain a predicted photolithography pattern, the mask pattern processing method further includes: exposing a test semiconductor structure based on a test mask to obtain a test photolithography structure; wherein, the test mask is formed by splicing at least two sub-test masks; the test mask pattern on the test mask includes a test splicing pattern and a test non-splicing pattern; the test photolithography structure includes a splicing photolithography structure corresponding to the test splicing pattern and a non-splicing photolithography structure corresponding to the test non-splicing pattern; both the splicing photolithography structure and the non-splicing photolithography structure are composed of multiple pattern structures; for the... The key dimensions of the spliced ​​lithography structure and the non-spliced ​​lithography structure are measured to obtain key dimension measurement data of multiple pattern structures in the spliced ​​lithography structure and the non-spliced ​​lithography structure. Using the key dimensions falling within a specified numerical range as a filtering condition, the pattern structures constituting the spliced ​​lithography structure and the non-spliced ​​lithography structure are filtered to determine the target spliced ​​pattern structure and the target non-spliced ​​pattern structure. Based on the target spliced ​​pattern structure and the target non-spliced ​​pattern structure, target spliced ​​pattern data and target non-spliced ​​pattern data are obtained. Based on the target spliced ​​pattern data and the target non-spliced ​​pattern data, the full-pattern optical proximity effect model is established.

[0012] Optionally, the step of establishing the full-pattern optical proximity effect model based on the target stitched graphic data and the target unstitched graphic data includes: generating a simulated mask pattern based on the target stitched graphic data, the target unstitched graphic data, and simulated graphic data; wherein the transmittance corresponding to the target stitched graphic data and the transmittance corresponding to the simulated graphic data are the same; the transmittance corresponding to the target stitched graphic data is different from the transmittance corresponding to the target unstitched graphic data; inputting the simulated mask pattern into a pre-trained lithography process simulation model, and using the output of the lithography process simulation model as the full-pattern optical proximity effect model.

[0013] Optionally, the overlapping region has an overlapping boundary located inside the sub-mask; the splicing region is formed by the overlapping regions of adjacent sub-masks by coinciding the overlapping boundaries of adjacent sub-masks.

[0014] Optionally, the predicted lithography pattern is composed of multiple predicted lithography patterns; the defects existing in the predicted spliced ​​lithography pattern include at least one of the following: bridging defects formed by the interconnection of at least two pre-set unconnected predicted lithography patterns; narrow defects formed by local size inhomogeneity of a single predicted lithography pattern.

[0015] In another aspect, one embodiment of this application provides a mask pattern processing system for processing mask patterns on a mask formed by stitching together at least two sub-masks; wherein any of the sub-masks includes an overlapping region; a stitching region of the mask is formed based on the overlapping regions of adjacent sub-masks; the mask pattern processing system includes: a mask pattern acquisition module for acquiring mask patterns on the mask; a mask pattern splitting module for splitting the mask pattern into a stitched pattern and a non-stitched pattern; wherein the stitched pattern is a mask pattern on the sub-mask that falls within the stitching region; the non-stitched pattern is a mask pattern on the sub-mask that falls outside the stitching region; the stitched pattern corresponds to... The transmittance is different from that of the non-stitched pattern; the lithography behavior prediction module is used to predict the lithography behavior of the mask pattern using the full-pattern optical proximity effect model corresponding to the mask pattern, to obtain a predicted lithography pattern; wherein, the predicted lithography pattern includes a predicted stitched lithography pattern corresponding to the stitched pattern; the stitched pattern key size adjustment module is used to adjust the key size of the position corresponding to the defect in the predicted stitched lithography pattern according to the defect detection result of the predicted stitched lithography pattern and the key size composite adjustment rule of the stitched pattern when a defect is detected in the predicted stitched lithography pattern; the mask pattern output module is used to output the processed target mask pattern.

[0016] In another aspect, one embodiment of this application provides a mask formed by splicing together at least two sub-masks; the mask pattern on the mask is a target mask pattern output after processing by the mask pattern processing method described in the above embodiments, or the mask pattern on the mask is a target mask pattern output by the mask pattern processing system described in the above embodiments.

[0017] In several embodiments provided in this application, a mask pattern on a mask formed by splicing at least two sub-masks is obtained, and then split into spliced ​​patterns and non-spliced ​​patterns with different transmittances. The lithography behavior of the mask pattern is predicted using a full-pattern optical proximity effect model of the corresponding mask pattern to obtain a predicted lithography pattern including a predicted spliced ​​lithography pattern. If a defect is detected in the predicted spliced ​​lithography pattern, the key dimensions at corresponding positions in the spliced ​​pattern are adjusted according to the defect detection result of the predicted spliced ​​lithography pattern and the key dimension adjustment rule of the spliced ​​pattern, and the processed target mask pattern is output. Unexpected effects achieved include: when the mask image... After the case is split into spliced ​​patterns and non-spliced ​​patterns with different transmittance, the overall photolithography behavior of the mask pattern is predicted by using the full-pattern optical proximity effect model corresponding to the complete mask pattern. This makes the predicted spliced ​​photolithography pattern closer to the spliced ​​photolithography pattern obtained by the actual photolithography process. On this basis, the key dimensions of the spliced ​​pattern located in the splicing area are adjusted according to the defect detection results of the predicted spliced ​​photolithography pattern and the key dimension adjustment rules of the spliced ​​pattern. This realizes the replacement of the manual defect handling process that relies on human experience with the automated defect handling process for spliced ​​patterns, thereby improving the efficiency of defect handling for spliced ​​patterns before mask fabrication. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in describing the embodiments 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.

[0019] Figure 1 This is a schematic diagram illustrating the process of forming a mask by splicing together multiple sub-masks, which is provided for related technologies.

[0020] Figure 2 A schematic diagram illustrating the mask type for changing the portion of two sub-masks located within the splicing area, provided for related technologies.

[0021] Figure 3 A schematic diagram of a splicing mask including splicing and non-splicing areas, provided for related technologies.

[0022] Figure 4 A schematic diagram of a stitched mask displayed using a visual interface of mask design software, provided for related technologies.

[0023] Figure 5 A flowchart illustrating the mask pattern processing method provided for related technologies.

[0024] Figure 6 This is a schematic flowchart illustrating a mask pattern processing method provided in one embodiment of this application.

[0025] Figure 7 This is a schematic diagram illustrating the process of establishing a full-pattern optical proximity effect model according to one embodiment of this application.

[0026] Figure 8 This is a schematic diagram of a test mask provided in one embodiment of this application.

[0027] Figure 9 This is a schematic diagram of a first simulated submask pattern provided in one embodiment of this application.

[0028] Figure 10 This is a schematic diagram of a second simulated submask pattern provided in one embodiment of this application.

[0029] Figure 11 This is a schematic diagram of a simulated mask pattern provided in one embodiment of this application.

[0030] Figure 12 This is a schematic diagram showing the defect detection result of a bridging defect provided in one embodiment of this application.

[0031] Figure 13 This is a schematic flowchart illustrating the process of adjusting the key dimensions of a position in a spliced ​​pattern that corresponds to a predicted defect in the spliced ​​lithography pattern, according to one embodiment of this application.

[0032] Figure 14 This is a flowchart illustrating a method for processing a mask pattern according to another embodiment of this application.

[0033] Figure 15 This is a schematic diagram of a predicted lithographic pattern for a dense pattern type mask pattern in related technologies.

[0034] Figure 16 This is a schematic diagram of a predicted lithographic pattern for a dense pattern type mask pattern provided in an embodiment of this application.

[0035] Figure 17 This is a schematic diagram of the predicted photolithography pattern for sparse pattern type mask patterns in related technologies.

[0036] Figure 18This is a schematic diagram of a predicted lithographic pattern for a sparse pattern type mask pattern provided in an embodiment of this application.

[0037] Figure 19 This is a schematic diagram comparing the predicted critical dimensions of the predicted stitching lithography pattern with the measured critical dimensions of the actual lithography pattern, as provided in the embodiments of this application.

[0038] Figure 20 This is a schematic diagram comparing the predicted lithographic pattern and the actual lithographic pattern provided in the embodiments of this application. Detailed Implementation

[0039] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0040] The accompanying drawings provided in this application are only schematic illustrations of the basic concept of this application. The drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the shape, quantity and proportion of each component may be changed, and the layout of the components may also be more complex.

[0041] In the description of the embodiments of this application, it should be understood that the terms "upper," "lower," "left," "right," "vertical," "horizontal," "inner," "outer," "center," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" and "second" may explicitly or implicitly include one or more of the stated features.

[0042] Please see Figure 1 In related technologies, the physical size limit of a single photomask is typically 26mm × 33mm. To overcome the limitation of the physical size of a single mask on the effective photosensitive area size of a CIS chip, a splicing process is usually used to obtain a spliced ​​mask with a physical size exceeding 26mm × 33mm. Subsequently, a double exposure process is used to transfer the mask pattern on the spliced ​​mask to the wafer, thereby achieving continuous patterning of a larger-sized CIS chip.

[0043] Please see Figures 2 to 4 The specific process for manufacturing splicing masks using splicing technology is generally as follows.

[0044] First, based on mask pattern segmentation rules, segmentation regions and segmentation boundaries within larger mask patterns that exceed the upper limit of the physical size of a single photomask are determined, and the larger mask pattern is segmented into at least two sub-mask patterns. Specifically, mask pattern segmentation rules may include: defining regions where the mask pattern complexity is less than a specified threshold, and / or where the critical size of the mask pattern is greater than a specified value, and / or where the patterns in the mask pattern are arranged in a specified direction as segmentation regions.

[0045] Subsequently, submasks are fabricated for each sub-mask pattern, resulting in at least two submasks. Specifically, each submask has an overlapping region for subsequent formation of the stitching area, and this overlapping region has an overlapping boundary within the submask. To accommodate the transfer of complex mask patterns and improve lithography accuracy, in related technologies, the submask type is typically a phase-shift mask (PSM) with a transmittance of 0.06.

[0046] Next, at least two sub-masks are arranged according to the mask pattern splicing rules corresponding to the mask pattern splitting rules, and the relative positions of adjacent sub-masks are adjusted so that the overlapping boundaries of the overlapping areas of adjacent sub-masks overlap, thereby forming a splicing area based on the overlapping areas of adjacent sub-masks, and the non-overlapping areas on the sub-masks are taken as non-splicing areas, resulting in a splicing mask containing splicing and non-splicing areas.

[0047] As can be seen from the above process of forming a spliced ​​mask, if the mask type of some sub-masks located within the splicing area remains a phase-shifting mask, the transmittance of these sub-masks within the splicing area will undergo linear superposition due to the fluctuation of light. That is, the transmittance of some sub-masks within the splicing area will change to 0.12, making the lithography behavior of these sub-masks in subsequent photolithography processes unpredictable. At this point, due to the change in transmittance, even if lithography parameters such as exposure dose and critical dimensions remain unchanged, performing the photolithography process based on the spliced ​​mask may lead to a decrease in lithography accuracy or even lithography defects.

[0048] To improve the ability to predict the lithographic behavior of mask patterns on spliced ​​masks, researchers use mask design software during the mask design stage to change the mask type of some sub-masks located in the splicing area from phase-shift masks to binary masks with zero transmittance, thereby reducing the complexity of predicting the lithographic behavior of mask patterns on spliced ​​masks.

[0049] Please see Figure 5To improve the production yield of CIS chips based on splicing masks, in related technologies, the mask pattern on the splicing mask is inspected and processed through the following steps S110 to S150.

[0050] S110: Obtain the mask pattern on the stitching mask.

[0051] First, the design data of the mask pattern on the splicing mask can be imported into the mask pattern adjustment software, so that the mask pattern adjustment software can extract the mask pattern on the splicing mask according to the design data of the mask pattern.

[0052] S120: Perform optical proximity correction on the mask pattern.

[0053] Secondly, optical proximity correction (OPC) can be performed on the obtained mask pattern using mask pattern adjustment software to compensate for the influence of light fluctuations on smaller-sized patterns in the mask pattern, optimize the pattern resolution and pattern fidelity during the pattern transfer process, and expand the photolithography process window.

[0054] S130: Perform lithography rule checks on the corrected mask pattern.

[0055] Subsequently, the corrected mask pattern can be subjected to a lithography rule check (LRC) using mask pattern adjustment software to identify any defects that may exist in the corrected mask pattern.

[0056] S140: Adjust for defects in the detected mask pattern.

[0057] If defects are found in the corrected mask pattern, the defects in the mask pattern can be adjusted by increasing or decreasing the key size of the mask pattern at the defect location, or by adjusting the orientation of the mask pattern. This allows for stable imaging within the depth of focus and exposure dose tolerance range of the photolithography process when the photolithography process is performed based on the adjusted mask pattern.

[0058] S150: Output the adjusted mask pattern.

[0059] Researchers performed a photolithography process based on the adjusted mask pattern and inspected the patterned structure formed by the photolithography. They found that there were still many defects in the patterned structure formed by the photolithography, which made it difficult to meet the production yield requirements for manufacturing CIS chips.

[0060] To improve the production yield of CIS chips, researchers further investigated the reasons why there are still many defects in the patterned structure formed by photolithography.

[0061] In related technologies, the mask types of the portion of the spliced ​​mask located in the splicing region differ from those of the portion located in the non-splicing region. That is, the spliced ​​mask has two mask types. Due to effects such as diffraction crosstalk, standing wave effects, and the amplification effect of process errors, the mask pattern adjustment software cannot predict the lithographic behavior of mask patterns with two mask types before the spliced ​​mask is fabricated. This results in uncertainty in the position and shape of the patterned structure corresponding to the mask pattern on the portion of the mask located in the splicing region on the wafer. To reduce the risk of defects in the patterned structure, before the spliced ​​mask is fabricated, defects in the mask pattern on the portion of the mask located in the splicing region are typically detected and corrected using historical experience and a manual joint design and verification method. This defect detection and correction method is not only inefficient but also prone to missed detections (Miss Operation, MO) due to human error.

[0062] Furthermore, due to insufficient defect handling of the mask pattern before fabrication of the stitched mask, the presence of defects in the patterned structure formed by photolithography needs to be determined through the actual photolithography process. Therefore, the development process for stitched masks includes: designing the mask pattern for the stitched mask; fabricating the stitched mask based on the designed mask pattern; performing the photolithography process based on the stitched mask to form a patterned structure; inspecting the patterned structure; adjusting the mask pattern of the stitched mask according to the detected defects in the patterned structure; and fabricating the stitched mask based on the adjusted mask pattern. As can be seen from the above development process, the current development cycle for stitched masks is relatively long and the development cost is high.

[0063] Therefore, it is necessary to provide a method for processing mask patterns that can predict the photolithographic behavior of mask patterns on a spliced ​​mask containing two types of mask before the spliced ​​mask fabrication, perform defect detection on the mask patterns on a portion of the mask located in the splicing area based on the prediction results, and adjust the mask patterns when defects are detected, thereby improving the efficiency of defect processing on the mask patterns on a portion of the mask located in the splicing area.

[0064] Please see Figure 6 One embodiment of this application provides a method for processing a mask pattern. This method can be applied to a mask formed by stitching together at least two sub-masks. Specifically, any sub-mask may include an overlapping area. A stitching area of ​​the mask can be formed based on the overlapping areas of adjacent sub-masks. The method for processing the mask pattern may include steps S210, S230, S250, S270, and S290.

[0065] S210: Obtain the mask pattern on the mask.

[0066] In this embodiment, the design data of the mask pattern on the mask can be imported into the mask pattern adjustment software so that the mask pattern adjustment software can extract the mask pattern according to the design data of the mask pattern.

[0067] S230: Decompose the mask pattern into a spliced ​​pattern and a non-spliced ​​pattern.

[0068] In this embodiment, the mask pattern can be split according to the splicing area. Specifically, the splicing pattern can be a mask pattern on a sub-mask that falls within the splicing area. The non-splicing pattern can be a mask pattern on a sub-mask that falls outside the splicing area. The transmittance corresponding to the splicing pattern is different from the transmittance corresponding to the non-splicing pattern. For example, the transmittance corresponding to the splicing pattern can be 0, while the transmittance corresponding to the non-splicing pattern can be 0.06.

[0069] Please continue reading. Figure 3 To improve the efficiency of defect detection and processing in spliced ​​patterns, in this embodiment, the splicing area can be defined based on the overlapping area of ​​sub-masks. Specifically, the overlapping area of ​​any sub-mask can be defined by the sub-mask boundary and the overlapping boundary located inside the sub-mask. Sub-masks can be arranged according to the patterns in the corresponding mask patterns, and the position of one of two adjacent sub-masks can be fixed. The other sub-mask can be moved until the overlapping boundaries of the two sub-masks coincide. The area formed by the overlapping areas of the two sub-masks is then used as the splicing area.

[0070] To improve the ability to predict the photolithography behavior of mask patterns, in some embodiments, before performing step S250, the mask pattern processing method may further include step S240: establishing a full-pattern optical proximity effect model of the corresponding mask pattern.

[0071] Since the transmittance of spliced ​​and non-spliced ​​patterns is different, in order to predict the overall lithography behavior of the mask pattern, we can first establish a full-pattern optical proximity effect model corresponding to all mask patterns, and then use the full-pattern optical proximity effect model to predict the overall lithography behavior of the mask pattern.

[0072] Please see Figure 7 In this embodiment, establishing a full-pattern optical proximity effect model corresponding to the mask pattern may include sub-steps S241, S242, S243, S244, and S245.

[0073] S241: Expose the test semiconductor structure based on the test mask to obtain the test photolithography structure.

[0074] To obtain data for building a full-pattern optical proximity effect model, a photolithography process can be performed on a test mask formed by a splicing process to form a test photolithography structure on the test semiconductor structure, and then data from the test photolithography structure can be collected.

[0075] In order to acquire data for predicting the lithographic behavior of the mask pattern on the mask to be processed based on the actual lithographic behavior of the mask pattern on the test mask, in this embodiment, the test mask can be similar in structure to the mask structure to be processed. Specifically, the test mask can be formed by splicing at least two sub-test masks. Therefore, similar to the mask pattern on the mask, the test mask pattern on the test mask can also include a test spliced ​​pattern and a test non-spliced ​​pattern.

[0076] In this embodiment, the test lithography structure can be a patterned structure formed by transferring the mask pattern of a test mask onto a test semiconductor structure using a photolithography process. Specifically, the test lithography structure can include a spliced ​​lithography structure corresponding to the test spliced ​​pattern and a non-spliced ​​lithography structure corresponding to the test non-spliced ​​pattern. Both the spliced ​​and non-spliced ​​lithography structures can be composed of multiple patterned structures.

[0077] S242: Measure the key dimensions of the spliced ​​lithography structure and the non-spliced ​​lithography structure to obtain the key dimension measurement data of multiple pattern structures in the spliced ​​lithography structure and the key dimension measurement data of multiple pattern structures in the non-spliced ​​lithography structure.

[0078] In this embodiment, a key dimension measurement tool for patterned structures, such as a scanning electron microscope, can be used to measure the key dimensions of multiple patterned structures constituting a spliced ​​lithography structure and the spacing between adjacent patterned structures, thereby obtaining key dimension measurement data for each patterned structure.

[0079] S243: Using the critical dimensions falling within a specified numerical range as a filtering condition, filter the graphic structures that constitute the spliced ​​lithography structure and the non-spliced ​​lithography structure to determine the target spliced ​​graphic structure and the target non-spliced ​​graphic structure.

[0080] To ensure that the patterned structure obtained by photolithography based on the processed mask pattern meets the precision requirements of the photolithography process, the full-pattern optical proximity effect model needs to be established based on data that meets these precision requirements. Therefore, target stitched and target non-stitched patterned structures with key dimensions meeting the precision requirements can be selected from multiple patterned structures constituting the stitched and non-stitched photolithography structures. Data can then be collected based on these target stitched and target non-stitched patterned structures.

[0081] In this embodiment, the specified numerical range can be the critical dimension error range allowed by the photolithography process. Specifically, the specified numerical range can be determined based on the preset critical dimension and the process window of the photolithography process. For example, if the preset critical dimension is 120nm and the process window is ±30nm, then the specified numerical range can be 90nm~150nm.

[0082] In this embodiment, the target stitched pattern structure can be a pattern structure whose critical dimensions fall within a specified numerical range among multiple pattern structures constituting the stitched lithography structure. The target non-stitched pattern structure can be a pattern structure whose critical dimensions fall within a specified numerical range among multiple pattern structures constituting the non-stitched lithography structure.

[0083] S244: Based on the target stitched graphic structure and the target non-stitched graphic structure, obtain the target stitched graphic data and the target non-stitched graphic data.

[0084] In this embodiment, image data related to the target stitched graphic structure and the target non-stitched graphic structure can be imported into mask pattern adjustment software. The mask pattern adjustment software can then be used to extract the target stitched graphic data and the target non-stitched graphic data from the image data. For example, when using a scanning electron microscope to obtain key dimension measurement data, the scanning electron microscope images of the target stitched graphic structure and the target non-stitched graphic structure can be imported into the mask pattern adjustment software to obtain the target stitched graphic data and the target non-stitched graphic data.

[0085] S245: Establish a full-pattern optical proximity effect model based on the target stitched graphic data and the target non-stitched graphic data.

[0086] To improve the predictive performance of the full-pattern optical proximity effect model for actual lithography behavior, in this embodiment, a simulated mask pattern is first generated based on the target stitched pattern data, the target non-stitched pattern data, and the simulated pattern data. This simulated mask pattern is then input into a pre-trained lithography process simulation model, and the output of the simulation model is used as the full-pattern optical proximity effect model. Specifically, the transmittance corresponding to the target stitched pattern data is the same as the transmittance corresponding to the simulated pattern data. The transmittance corresponding to the target stitched pattern data is different from the transmittance corresponding to the target non-stitched pattern data. In particular, taking the test mask formed by stitching together a first test sub-mask and a second test sub-mask as an example, the process of establishing a full-pattern optical proximity effect model based on the target stitched pattern data and the target non-stitched pattern data is briefly described.

[0087] Please see Figure 8 The test mask pattern on the test mask can include a first non-stitched test pattern, a stitched test pattern, and a second non-stitched test pattern. The transmittance of both the first and second non-stitched test patterns can be 0.06, while the transmittance of the stitched test pattern can be 0. Furthermore, the test mask pattern also includes a background layer pattern, which can be a fully transparent pattern, meaning its transmittance can be 1. Target stitched graphic data, first target non-stitched graphic data, and second target non-stitched graphic data can be obtained from the test mask pattern.

[0088] Please see Figure 9 A first simulated sub-mask pattern can be generated based on the target stitched graphic data, the first target non-stitched graphic data, and the simulated graphic data. The transmittance of the target stitched graphic data and the simulated graphic data can both be 0, while the transmittance of the first target non-stitched graphic data can be 0.06. In the first simulated sub-mask pattern, the first target non-stitched graphic corresponding to the first target non-stitched graphic data and the simulated graphic corresponding to the simulated graphic data are located on either side of the target stitched graphic corresponding to the target stitched graphic data.

[0089] Please see Figure 10 A second simulated sub-mask pattern can be generated based on the target stitched graphic data, the second target unstitched graphic data, and the simulated graphic data. The transmittance of different parts and the relative positional relationship of the corresponding graphics in the second simulated sub-mask pattern are similar to those in the first simulated sub-mask pattern, and will not be repeated here.

[0090] Please see Figure 11 Using the target stitching pattern as the stitching reference, the first and second simulated sub-mask patterns can be stitched together to obtain the simulated mask pattern. Subsequently, the simulated mask pattern can be input into a pre-trained lithography process simulation model, which will then simulate the double exposure process. The output of the lithography process simulation model can be used as the full-pattern optical proximity effect model.

[0091] S250: Using the full-pattern optical proximity effect model of the corresponding mask pattern, the lithography behavior of the mask pattern is predicted to obtain the predicted lithography pattern.

[0092] To achieve overall defect detection and processing of mask patterns, the process of performing photolithography based on mask patterns can be simulated using a full-pattern optical proximity effect model.

[0093] In this embodiment, the predicted lithography pattern can be the predicted position and outline of the patterned structure formed by transferring the mask pattern to the semiconductor structure on the semiconductor structure. Specifically, the predicted lithography pattern can include a predicted splicing lithography pattern corresponding to the splicing pattern.

[0094] S270: When a defect is detected in the predicted lithographic pattern, the critical dimension of the position in the lithographic pattern corresponding to the defect in the predicted lithographic pattern is adjusted according to the defect detection result of the predicted lithographic pattern and the critical dimension adjustment rule of the lithographic pattern.

[0095] In this embodiment, the mask pattern may include multiple mask graphics. This application does not impose specific limitations on the shape and number of mask graphics.

[0096] Since a mask pattern can include multiple mask patterns, a predicted lithography pattern can be composed of multiple predicted lithography patterns corresponding to the multiple mask patterns. To improve the production yield of CIS chips manufactured based on the processed mask pattern, in this embodiment, defects existing in the predicted stitched lithography pattern can include at least one of the following: a bridging defect formed by the interconnection of at least two pre-defined unconnected predicted lithography patterns; or a pinch defect formed by local dimensional inhomogeneity of a single predicted lithography pattern. Specifically, a bridging defect may cause a short circuit, and a pinch defect may cause a broken wire.

[0097] In this embodiment, the predicted lithography pattern can be detected based on the key dimensions of the predicted lithography pattern in the predicted lithography pattern, and the probability of defects appearing at different positions in the predicted lithography pattern can be determined. If the probability value corresponding to any position is greater than the probability threshold, it is determined that there is a defect in the predicted lithography pattern, and the position corresponding to this probability is the defect position.

[0098] Please see Figure 12 Taking bridging defects as an example, in the predicted lithography pattern, the predicted minimum spacing of two adjacent predicted lithography patterns with a design spacing of 100nm was detected to be 89nm. In this case, even if these two predicted lithography patterns are not connected, the probability of a short circuit due to the interconnection of the corresponding patterned structures after the actual lithography process is relatively high. Therefore, it can be determined that a bridging defect exists in the predicted lithography pattern, and the location of these two predicted lithography patterns can be taken as the defect location.

[0099] To enhance the ability to adjust the critical dimensions of the spliced ​​pattern, in this embodiment, the composite adjustment rule for the critical dimensions of the spliced ​​pattern may include a baseline adjustment rule and a pre-trained linear regression fitting model. Specifically, the baseline adjustment rule can be used to determine the adjustment baseline amount for the critical dimensions of the spliced ​​pattern based on process experience or experimental data. The pre-trained linear regression fitting model can be used to determine the adjustment offset of the critical dimensions of the spliced ​​pattern based on the pattern type and corresponding target critical dimensions of the mask pattern with defects in the spliced ​​pattern.

[0100] Please see Figure 13 To improve the accuracy of adjusting the critical dimensions of the splicing pattern, in this embodiment, adjusting the critical dimensions of the positions in the splicing pattern corresponding to the predicted defects in the splicing lithography pattern may include sub-steps S271, S272, S273, S274, S275, and S276.

[0101] S271: Based on the density of the mask pattern distribution in the mask pattern, the mask pattern is divided into multiple pattern types.

[0102] Since the density of the mask pattern distribution may affect the lithography behavior, in this embodiment, the pattern type can include dense pattern type, semi-sparse (Semi-Isolation, Semi-ISO) pattern type and sparse (Isolation, ISO) pattern type, depending on the gradual decrease in the density of the mask pattern distribution.

[0103] S272: Based on the defect detection results of the predicted splicing lithography pattern, determine the defect mask pattern in the splicing pattern that corresponds to the defect in the predicted splicing lithography pattern.

[0104] In this embodiment, the defect detection results of the predicted stitched lithography pattern can be used to indicate the location of defects in the predicted stitched lithography pattern and the predicted lithography pattern of the defect location. Since the predicted stitched lithography pattern is obtained by predicting the patterned structure formed by lithography based on the stitched pattern, there is a correspondence between the predicted stitched lithography pattern and the stitched pattern. Therefore, by combining the defect detection results of the predicted stitched lithography pattern and the correspondence between the predicted stitched lithography pattern and the stitched pattern, the defect mask pattern corresponding to the predicted lithography pattern of the predicted stitched lithography pattern containing defects in the predicted stitched lithography pattern can be determined.

[0105] S273: Obtain the target pattern type and target key dimension corresponding to the defect mask pattern, as well as the target non-stitching key dimension and predicted key dimension corresponding to the target pattern type and target key dimension.

[0106] In this embodiment, the target pattern type corresponding to the defect mask pattern can be determined based on the density of mask pattern distribution near the location of the defect mask pattern.

[0107] In this embodiment, the target critical dimension can be the critical dimension of the patterned structure expected to be obtained by exposure using Automated Defect Inspection (ADI). Specifically, the target critical dimension corresponding to the defect mask pattern can be determined based on the design data of the mask pattern.

[0108] In this embodiment, the target non-stitching critical dimension can be the target critical dimension corresponding to the defect mask pattern assuming the defect mask pattern is located outside the stitching area. Specifically, based on process experience and experimental data, a correspondence between the target pattern type, target critical dimension, and target non-stitching critical dimension is preset. Furthermore, when the target critical dimension is the same but the target pattern type is different, the target non-stitching critical dimension corresponding to the target critical dimension is different. Taking a target critical dimension of 120nm as an example, when the target pattern type is a dense pattern type, the corresponding target non-stitching critical dimension is 121nm; when the target pattern type is a sparse pattern type, the corresponding target non-stitching critical dimension is 136.5nm. Therefore, the target non-stitching critical dimension corresponding to the target critical dimension of the defect mask pattern can be determined based on the target pattern type, the target critical dimension, and the correspondence between the two and the target non-stitching critical dimension.

[0109] In this embodiment, the critical dimension can be predicted based on the target critical dimension using a full-pattern optical proximity effect model. The specific prediction process is similar to the process of obtaining the predicted lithographic pattern using a full-pattern optical proximity effect model, and will not be described in detail here.

[0110] S274: Based on the target non-splitting critical dimension and the reference quantity adjustment rules, adjust the target critical dimension according to the reference quantity to obtain the adjustment reference quantity of the target critical dimension.

[0111] In this embodiment, the reference adjustment rule can be to set the adjustment reference value (Stitching Pattern Basic Mask Critical Dimension, SBMC) of the target critical dimension as the sum of the target non-stitching pattern basic mask critical dimension and the process empirical value. For example, if the process empirical value can be +15nm per side, then 30nm can be added to the target non-stitching pattern basic mask critical dimension to obtain the adjustment reference value of the target critical dimension. Taking a target critical dimension of 120nm and a target pattern type of dense pattern as an example, the target non-stitching pattern critical dimension corresponding to the target critical dimension is 121nm, then the adjustment reference value of the target critical dimension is 151nm.

[0112] S275: Based on the target pattern type, target key size, and predicted key size, the target key size is adjusted by offset using a pre-trained linear regression fitting model to obtain the adjusted offset of the target key size.

[0113] In this embodiment, a linear regression fitting model can be used to represent the linear relationship between the Stitching Pattern Mask Critical Dimension Bias (SMCB), the Target Critical Dimension (T), and the prediction error (E). The prediction error can be the difference between the Target Critical Dimension and the prediction error. Specifically, the linear regression fitting model can include multiple sub-fitting models. Each sub-fitting model has a corresponding pattern type and a range of Target Critical Dimension values.

[0114] In this embodiment, adjusting the offset of the target key dimensions using a linear regression fitting model may include sub-steps S2751, S2752, S2753, and S2754.

[0115] S2751: Input the target key dimension and the predicted key dimension into the specified subfit model, and use the output of the specified subfit model as the iterative offset of the target key dimension.

[0116] In this embodiment, the specified subfit model can be determined based on the target pattern type and the target critical size value range into which the target critical size falls. Taking a target critical size of 120nm and a target pattern type of dense pattern as an example, the specified subfit model is SMCB=0.1211*T-0.2343*E-10.6399.

[0117] S2752: Update the adjustment reference value using the sum of the adjustment reference value and the iterative offset of the target critical dimension.

[0118] In this embodiment, after obtaining the iterative offset of the target key dimension, the adjustment reference amount and the iterative offset can be summed, and the summation result can be assigned to the adjustment reference amount to update the adjustment reference amount.

[0119] S2753: The critical dimensions are updated using the output of the full-pattern optical proximity model based on the updated adjusted baseline.

[0120] In this embodiment, after the adjustment reference value is updated, the updated adjustment reference value can be input into the full-pattern optical proximity effect model, the full-pattern optical proximity effect model is used for prediction, and the output of the full-pattern optical proximity effect model is assigned to the predicted key dimension to update the predicted key dimension.

[0121] S2754: Calculate the difference between the target critical dimension and the updated predicted critical dimension to obtain the prediction error.

[0122] In this embodiment, if the prediction error falls outside the specified error range, steps S2751 to S2754 are repeated until the prediction error falls within the specified error range. In this embodiment, the specified error range can be determined in conjunction with the process window of the photolithography process. For example, the specified error range can be ±3nm.

[0123] Since the target critical dimension is the expected critical dimension of the graphical structure corresponding to the defect mask pattern, in this embodiment, at least one iteration is required for adjusting the target critical dimension offset.

[0124] S276: Sum the adjustment reference amount and the adjustment offset amount to obtain the adjustment critical dimension (Stitching Pattern Final Mask Critical Dimension, SFMC) corresponding to the target critical dimension.

[0125] Please refer to Formula 1. In this embodiment, the adjustment reference amount of the target critical dimension is updated in each iteration. Therefore, the adjustment critical dimension includes the iterative offset of the target critical dimension obtained from multiple iterations.

[0126] Formula 1

[0127] in, This indicates the adjustment critical dimension corresponding to the target critical dimension. Indicates the adjustment reference amount for the target critical dimension. This indicates the adjustment offset for the target's critical dimensions.

[0128] To provide a basis for calculating the adjustment offset, in some embodiments, a linear regression fitting model can be trained before performing sub-step S275. Specifically, training key dimensions for training the linear regression fitting model can be obtained first, and then the following sub-steps S275a and S275b can be repeated until a linear regression fitting model for adjusting the offset of the target key dimensions corresponding to multiple pattern types is determined.

[0129] In this embodiment, the training key dimension may include a training target key dimension corresponding to multiple pattern types, a training predicted key dimension corresponding to the training target key dimension, and a training offset corresponding to the training target key dimension. The training predicted key dimension can be predicted based on the training target key dimension using a full-pattern optical proximity effect model.

[0130] In this embodiment, sub-step S275a can be: using the K-means clustering analysis algorithm to analyze the key dimensions of the training target corresponding to any pattern type, and dividing the key dimensions of the training target corresponding to any pattern type into multiple value intervals.

[0131] Since the goal of the K-means clustering algorithm is to divide the data into K clusters, making data points within the same cluster as similar as possible and data points in different clusters as different as possible, please refer to Equations 2 to 4. For the training target key size corresponding to any pattern type, we can first calculate the Euclidean distance between each training target key size and the center of each data cluster according to Equation 2, and then calculate the sum of squared errors within each data cluster according to Equation 3. The iteration objective is to minimize the sum of squared errors within all data clusters. According to Equation 4, the center of each data cluster is updated to the mean of the training target key size within the data cluster after each iteration, until the change in the cluster center obtained from two iterations is less than a specified threshold, at which point the iteration stops. For example, the specified threshold can be 0.001. After obtaining the analysis results of the K-means clustering algorithm, the training target key size corresponding to any pattern type can be divided into multiple value intervals according to the data patterns contained in the analysis results.

[0132] Formula 2

[0133] in, Indicates the first Key dimensions of each training objective. Indicates the first The center of each data cluster Indicates the first The key dimensions of the training target and the first The Euclidean distance between the centers of the data clusters.

[0134] Formula 3

[0135] in, This represents the sum of squared errors within the cluster. Indicates the number of data clusters. Indicates the first The set of key dimensions of the training target in each data cluster.

[0136] Formula 4

[0137] In this embodiment, sub-step S275b can be: performing linear regression fitting on all training target key dimensions that fall within any value interval, the training prediction key dimensions corresponding to all training target key dimensions, and the training offset to determine the linear regression fitting model corresponding to any value interval.

[0138] Please refer to Formulas 5 and 6. After dividing the value range of the key dimensions of the training target for any pattern type, linear regression can be performed on all data within any value range according to Formula 5, and the linear regression fitting results within that value range can be evaluated according to Formula 6.

[0139] Formula 5

[0140] in, The training offset represents the key dimensions of the training target. This represents the difference between the target key size and the corresponding predicted key size. Indicates the critical dimensions of the training target. , , All of these represent constant coefficients.

[0141] Formula 6

[0142] in, Indicates the goodness of fit. Represents the sum of squared residuals. This represents the total sum of squares.

[0143] Please refer to Table 1. In Table 1, Min represents the minimum allowable value for the critical dimension of the training target, and Max represents the maximum allowable value for the critical dimension. Min and Max can be determined according to the lithography process requirements. In this embodiment, Min is 70nm and Max is 1000nm. For simplicity, in Table 1, , , Let SMCB, E, and T represent these values ​​respectively. According to the data in Table 1, the minimum goodness of fit for the multiple target key dimension ranges is 0.715, indicating that the linear regression fit is relatively good. When the target key dimension range is from 600 to Max, since the training target key dimensions for different pattern types do not vary, the sub-fit model degenerates into a constant.

[0144] Table 1 Training results of the linear regression fitting model

[0145]

[0146] S290: Output the processed target mask pattern.

[0147] After completing the critical dimension adjustment processing of the positions corresponding to the defects in the splicing pattern and the predicted splicing lithography pattern, the processed target mask pattern can be exported using mask pattern adjustment software to realize mask fabrication based on the target mask pattern.

[0148] As described above, the entire process of processing the mask pattern can be completed using mask pattern adjustment software. Based on this, by setting task instructions for the mask pattern adjustment software, the software can be used to achieve automated defect detection and processing of spliced ​​patterns in the mask pattern, thereby replacing the manual processing process in related technologies and improving the efficiency of defect processing of spliced ​​patterns.

[0149] Please see Figure 14 Another embodiment of this application provides a method for processing a mask pattern. This method can be applied to a mask formed by stitching together at least two sub-masks. Specifically, any sub-mask may include an overlapping area. A stitching area of ​​the mask can be formed based on the overlapping areas of adjacent sub-masks. The method for processing the mask pattern may include steps S210, S230, S250, S270a, and S290.

[0150] In this embodiment, steps S210, S230, S250 and S290 are similar to those in the above embodiments, and will not be described again here.

[0151] In this embodiment, the predicted lithography pattern may further include a predicted non-stitched lithography pattern corresponding to the non-stitched pattern.

[0152] In this embodiment, step S270a can be: when a defect is detected in the predicted non-stitched lithography pattern, optical proximity effect correction processing is performed on the position in the non-stitched pattern corresponding to the defect in the predicted non-stitched lithography pattern according to the defect detection result of the predicted non-stitched lithography pattern.

[0153] In this embodiment, a preset non-stitched optical proximity effect model corresponding to the non-stitched pattern can be used to detect defects in the predicted non-stitched lithography pattern. The defect detection method for the predicted non-stitched lithography pattern is similar to the defect detection method for the predicted stitched lithography pattern described in the above embodiment, and will not be repeated here.

[0154] In this embodiment, optical proximity effect correction processing is performed on the positions in the non-stitched pattern corresponding to the defects in the predicted non-stitched lithography pattern. The non-stitched optical proximity effect model can be used to correct the key dimensions, shape, orientation, etc. of the defect mask pattern in the non-stitched pattern.

[0155] Please refer to Table 2. Taking a target critical size of 120nm and target pattern types of dense and sparse patterns as examples, the mask patterns on the masks formed by the splicing process are processed using the mask pattern processing methods provided by related technologies and the mask pattern processing methods provided in the embodiments of this application, respectively. The mask pattern processing methods provided by related technologies do not process the spliced ​​pattern, while the mask pattern processing methods provided in the embodiments of this application do process the spliced ​​pattern. The error represents the difference between the predicted critical size of the lithographic pattern and the target critical size.

[0156] Table 2. Processing results of the two mask pattern processing methods

[0157]

[0158] As can be seen from the error data in Table 2, for both dense and sparse pattern types, the errors obtained by using the mask pattern processing method provided in this application embodiment are smaller than the errors obtained by using the mask pattern processing method provided by related technologies.

[0159] Please see Figures 15 to 18 . Figure 15 and Figure 17 The red lines in the diagram represent the predicted lithographic pattern outline obtained by predicting the lithographic behavior of the mask pattern using the mask pattern processing method provided by the relevant technology. Figure 16 and Figure 18 The green lines in the diagram represent the predicted lithographic pattern outline obtained by predicting the lithographic behavior of the mask pattern using the mask pattern processing method provided in this application embodiment. (Comparison) Figure 15 and Figure 16 ,contrast Figure 17 and Figure 18 It is known that, compared with the predicted lithographic pattern outline obtained by the mask pattern processing method provided by related technologies, the predicted lithographic pattern outline obtained by the mask pattern processing method provided in this application is closer to the mask pattern outline.

[0160] Based on the above Table 2 and Figures 15 to 18 As can be seen from the information presented, after processing the mask pattern using the mask pattern processing method provided in the embodiments of this application, the pattern transfer accuracy of performing photolithography based on the processed target mask pattern is relatively high.

[0161] Please see Figure 19After processing the mask pattern using the mask pattern processing method provided in this application embodiment, a photolithography process is performed based on the processed target mask pattern to obtain the actual photolithography pattern. The key dimensions of the actual photolithography pattern are measured using a photolithography pattern key dimension measurement tool. Seven positions (P1, P2, P3, P4, P5, P6, and P7) are selected in the actual photolithography pattern, and the key dimensions of these positions are extracted. The extracted key dimensions are then compared with the predicted key dimensions obtained using the mask pattern processing method provided in this application embodiment and displayed in the same two-dimensional coordinate system. The error represents the difference between the measured key dimensions of the actual photolithography pattern and the predicted key dimensions. Figure 19 It can be seen that the measured critical dimension of each position of the actual photolithography pattern is very close to the predicted critical dimension, and the error at all positions falls within the allowable error range.

[0162] Please see Figure 20 . Figure 20 The comparison shows the actual and predicted lithographic patterns at one of the seven locations mentioned above. Figure 20 It can be seen that the outline of the predicted lithographic pattern basically coincides with the outline of the actual lithographic pattern.

[0163] comprehensive Figure 19 and Figure 20 As can be seen from the information presented, the mask pattern processing method provided in the embodiments of this application has a better predictive effect on the photolithography behavior of the mask pattern.

[0164] Another embodiment of this application provides a mask pattern processing system. This mask pattern processing system can be used to process mask patterns on a mask formed by stitching together at least two sub-masks. Each sub-mask may include an overlapping region. A stitching region of the mask can be formed based on the overlapping regions of adjacent sub-masks. The mask pattern processing system may include a mask pattern acquisition module, a mask pattern splitting module, a lithography behavior prediction module, a stitching pattern key size adjustment module, and a mask pattern output module. Specifically, the functions of each part are described below.

[0165] The mask pattern acquisition module can be used to acquire the mask pattern on the mask.

[0166] The mask pattern splitting module can be used to split a mask pattern into a stitched pattern and a non-stitched pattern. Specifically, a stitched pattern can be a mask pattern on a sub-mask that falls within the stitching area. A non-stitched pattern can be a mask pattern on a sub-mask that falls outside the stitching area.

[0167] The lithography behavior prediction module can be used to predict the lithography behavior of a mask pattern using a full-pattern optical proximity effect model, thereby obtaining a predicted lithography pattern. Specifically, the predicted lithography pattern can include a predicted splicing lithography pattern corresponding to the splicing pattern.

[0168] The critical dimension adjustment module for the splicing pattern can be used to adjust the critical dimensions of the position corresponding to the defect in the predicted splicing lithography pattern when a defect is detected in the predicted splicing lithography pattern, based on the defect detection result of the predicted splicing lithography pattern and the critical dimension adjustment rules of the splicing pattern.

[0169] The mask pattern output module can be used to output the processed target mask pattern.

[0170] In another embodiment of this application, a mask is provided, which can be formed by splicing together at least two sub-masks. The mask pattern on the mask can be a target mask pattern output after processing by the mask pattern processing method described in the above embodiments, or the mask pattern on the mask can be a target mask pattern output by the mask pattern processing system described in the above embodiments.

[0171] For the mask pattern processing system and other technical effects of the mask described in the above embodiments, please refer to other embodiments of this application for comparison and explanation, and they will not be repeated here.

[0172] In several embodiments provided in this application, a mask pattern on a mask formed by splicing at least two sub-masks is obtained, and then split into spliced ​​patterns and non-spliced ​​patterns with corresponding different transmittances. The lithographic behavior of the mask pattern is predicted using a full-pattern optical proximity effect model of the corresponding mask pattern to obtain a predicted lithographic pattern including a predicted spliced ​​lithographic pattern. If a defect is detected in the predicted spliced ​​lithographic pattern, the critical dimensions at corresponding positions in the spliced ​​pattern are adjusted according to the defect detection result of the predicted spliced ​​lithographic pattern and the critical dimension adjustment rule of the spliced ​​pattern, and the processed target mask pattern is output. Unexpected effects achieved include: after splitting the mask pattern into corresponding... After combining spliced ​​and non-spliced ​​patterns with different transmittance, the overall photolithography behavior of the mask pattern is predicted using the full-pattern optical proximity effect model corresponding to the complete mask pattern. This improves the prediction effect of the photolithography behavior of the completed mask pattern. Based on this, the key dimensions of the spliced ​​pattern located in the splicing area are adjusted according to the defect detection results of the predicted spliced ​​lithography pattern and the key dimension adjustment rules of the spliced ​​pattern. This realizes the replacement of the manual defect handling process relying on human experience with an automated defect handling process for spliced ​​patterns, improves the efficiency of defect handling for spliced ​​patterns before mask fabrication, and improves the pattern transfer accuracy of performing photolithography based on the processed target mask pattern.

[0173] It is understood that the specific examples in this application are only intended to help those skilled in the art better understand the embodiments of this application, and are not intended to limit the scope of this application.

[0174] It is understood that in the various embodiments of this application, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not limit the implementation process of the embodiments of this application in any way.

[0175] It is understood that the various embodiments described in this application can be implemented individually or in combination, and the embodiments of this application are not limited in this respect.

[0176] Unless otherwise stated, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in the embodiments of this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0177] In the several embodiments provided in this application, it should be understood that the disclosed mask pattern processing system and mask can be implemented in other ways. For example, the embodiments of the mask pattern processing system and mask described above are merely illustrative.

[0178] The above description is merely a 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 scope of the technology 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 of processing a mask pattern, characterized by, The mask pattern processing method is applied to a mask formed by splicing at least two sub-masks; any of the sub-masks comprises an overlapping area; a splicing area of the mask is formed based on the overlapping areas of adjacent sub-masks; the mask pattern processing method comprises: acquiring a mask pattern on the mask; splitting the mask pattern into a splicing pattern and a non-splicing pattern; the splicing pattern is a mask pattern on the sub-masks falling within the splicing area; the non-splicing pattern is a mask pattern on the sub-masks falling outside the splicing area; the splicing pattern corresponds to a different transmittance from the non-splicing pattern; using a full-pattern optical proximity effect model corresponding to the mask pattern to predict a lithography behavior of the mask pattern, to obtain a predicted lithography pattern; the predicted lithography pattern comprises a predicted splicing lithography pattern corresponding to the splicing pattern; in a case where it is detected that there is a defect in the predicted splicing lithography pattern, adjusting a critical dimension of a position in the splicing pattern corresponding to the defect in the predicted splicing lithography pattern according to a defect detection result of the predicted splicing lithography pattern and a critical dimension composite adjustment rule of the splicing pattern; outputting a processed target mask pattern.

2. The method of claim 1, wherein The predicted lithography pattern further comprises a predicted non-splicing lithography pattern corresponding to the non-splicing pattern; the mask pattern processing method further comprises: in a case where it is detected that there is a defect in the predicted non-splicing lithography pattern, performing optical proximity effect correction processing on a position in the non-splicing pattern corresponding to the defect in the predicted non-splicing lithography pattern according to a defect detection result of the predicted non-splicing lithography pattern.

3. The method of claim 1, wherein The mask pattern comprises a plurality of mask patterns; the critical dimension composite adjustment rule of the splicing pattern comprises a reference amount adjustment rule and a pre-trained linear regression fitting model; in a case where it is detected that there is a defect in the predicted splicing lithography pattern, the step of adjusting a critical dimension of a position in the splicing pattern corresponding to the defect in the predicted splicing lithography pattern according to a defect detection result of the predicted splicing lithography pattern and a critical dimension composite adjustment rule of the splicing pattern comprises: dividing the mask pattern into a plurality of pattern types according to a distribution density of mask patterns in the mask pattern; determining a defect mask pattern in the splicing pattern corresponding to the defect in the predicted splicing lithography pattern according to a defect detection result of the predicted splicing lithography pattern; acquiring a target pattern type, a target critical dimension corresponding to the defect mask pattern, and a target non-splicing critical dimension and a predicted critical dimension corresponding to the target pattern type and the target critical dimension; the predicted critical dimension is obtained by predicting the target critical dimension based on the full-pattern optical proximity effect model; adjusting the target critical dimension by a reference amount according to the target non-splicing critical dimension and the reference amount adjustment rule, to obtain an adjustment reference amount of the target critical dimension; adjusting the target critical dimension by using the pre-trained linear regression fitting model based on the target pattern type, the target critical dimension and the predicted critical dimension, to obtain an adjustment offset of the target critical dimension; performing summation operation on the adjustment reference and the adjustment offset, to obtain an adjusted critical dimension corresponding to the target critical dimension.

4. The method of claim 3, wherein The linear regression fitting model comprises a plurality of sub-fitting models; wherein any of the sub-fitting models has a corresponding pattern type and a target critical dimension value interval; the step of adjusting the target critical dimension by using the pre-trained linear regression fitting model based on the target pattern type, the target critical dimension and the predicted critical dimension, to obtain an adjustment offset of the target critical dimension, comprises: inputting the target critical dimension and the predicted critical dimension into a specified sub-fitting model, and taking the output of the specified sub-fitting model as an iterative offset of the target critical dimension; wherein the specified sub-fitting model is determined based on the target pattern type and the target critical dimension falling into a target critical dimension value interval; updating the adjustment reference by using the sum of the adjustment reference and the iterative offset of the target critical dimension; updating the predicted critical dimension by using the full-pattern optical proximity effect model based on the updated adjustment reference, and updating the predicted critical dimension by using the output of the full-pattern optical proximity effect model; performing difference operation on the target critical dimension and the updated predicted critical dimension, to obtain a prediction error; repeating the above steps until the prediction error falls within a specified error range, in the case that the prediction error falls outside the specified error range.

5. The method of claim 3, wherein Before the step of adjusting the target critical dimension by using the pre-trained linear regression fitting model based on the target pattern type, the target critical dimension and the predicted critical dimension, to obtain an adjustment offset of the target critical dimension, the mask pattern processing method further comprises: obtaining training critical dimensions for training the linear regression fitting model; wherein the training critical dimensions comprise training target critical dimensions corresponding to the plurality of pattern types, training predicted critical dimensions corresponding to the training target critical dimensions, and training offsets corresponding to the training target critical dimensions; the training predicted critical dimensions are obtained by using the full-pattern optical proximity effect model based on the training target critical dimensions; repeating the following sub-steps until a linear regression fitting model for adjusting the offset of the target critical dimension corresponding to the plurality of pattern types is determined; the sub-steps comprise: analyzing the training target critical dimensions corresponding to any pattern type by using a K-means clustering analysis algorithm, and dividing the training target critical dimensions corresponding to any pattern type into a plurality of value intervals; performing linear regression fitting on all training target critical dimensions falling into any of the value intervals, training predicted critical dimensions corresponding to the training target critical dimensions, and training offsets corresponding to the training target critical dimensions, to determine a linear regression fitting model corresponding to any of the value intervals.

6. The method of claim 1, wherein Before the step of predicting a lithography behavior of the mask pattern by using a full pattern optical proximity effect model corresponding to the mask pattern, the processing method of the mask pattern further comprises: exposing a test semiconductor structure based on a test mask to obtain a test lithography structure; wherein the test mask is formed by splicing at least two sub-test masks; a test mask pattern on the test mask comprises a test splicing pattern and a test non-splicing pattern; the test lithography structure comprises a splicing lithography structure corresponding to the test splicing pattern and a non-splicing lithography structure corresponding to the test non-splicing pattern; the splicing lithography structure and the non-splicing lithography structure are both composed of a plurality of pattern structures; measuring critical dimensions of the splicing lithography structure and the non-splicing lithography structure to obtain critical dimension measurement data of the plurality of pattern structures in the splicing lithography structure and critical dimension measurement data of the plurality of pattern structures in the non-splicing lithography structure; screening the pattern structures constituting the splicing lithography structure and the non-splicing lithography structure by taking critical dimensions falling within a specified numerical range as a screening condition to determine target splicing pattern structures and target non-splicing pattern structures; obtaining target splicing pattern data and target non-splicing pattern data according to the target splicing pattern structures and the target non-splicing pattern structures; establishing the full pattern optical proximity effect model according to the target splicing pattern data and the target non-splicing pattern data.

7. The method of claim 6, wherein The step of establishing the full pattern optical proximity effect model according to the target splicing pattern data and the target non-splicing pattern data comprises: generating a simulation mask pattern according to the target splicing pattern data, the target non-splicing pattern data and simulation pattern data; wherein a transmittance corresponding to the target splicing pattern data is the same as a transmittance corresponding to the simulation pattern data; the transmittance corresponding to the target splicing pattern data is different from a transmittance corresponding to the target non-splicing pattern data; inputting the simulation mask pattern to a pre-trained lithography process simulation model and taking an output of the lithography process simulation model as the full pattern optical proximity effect model.

8. The method of claim 1, wherein The overlapping region has an overlapping boundary located inside the sub-mask; the splicing region is formed by overlapping the overlapping boundaries of adjacent sub-masks and is jointly constituted by the overlapping regions of adjacent sub-masks.

9. The method of claim 1, wherein The predicted lithography pattern is composed of a plurality of predicted lithography patterns; defects existing in the predicted splicing lithography pattern include at least one of the following: a bridging defect formed by connecting at least two preset unconnected predicted lithography patterns to each other; a narrow defect formed by local size unevenness of a single predicted lithography pattern.

10. A system for processing a mask pattern, the system comprising: The mask pattern processing system is used for processing a mask pattern on a mask formed by splicing at least two sub-masks; wherein any sub-mask comprises an overlapping region; a splicing region of the mask is formed based on overlapping regions of adjacent sub-masks; the mask pattern processing system comprises: a mask pattern acquisition module configured to acquire the mask pattern on the mask; a mask pattern splitting module, configured to split the mask pattern into a stitching pattern and a non-stitching pattern; wherein the stitching pattern is a mask pattern on the sub-mask falling within the stitching area; the non-stitching pattern is a mask pattern on the sub-mask falling outside the stitching area; the stitching pattern corresponds to a different transmittance from the non-stitching pattern; a lithography behavior prediction module, configured to predict a lithography behavior of the mask pattern by using a full-pattern optical proximity effect model corresponding to the mask pattern, to obtain a predicted lithography pattern; wherein the predicted lithography pattern includes a predicted stitching lithography pattern corresponding to the stitching pattern; a stitching pattern critical dimension adjustment module, configured to, in a case where it is detected that there is a defect in the predicted stitching lithography pattern, adjust a critical dimension of a position corresponding to the defect in the predicted stitching lithography pattern in the stitching pattern according to a defect detection result of the predicted stitching lithography pattern and a critical dimension composite adjustment rule of the stitching pattern; a mask pattern output module, configured to output a target mask pattern after processing.

11. A mask, characterized in that The mask is formed by splicing at least two sub-masks; the mask pattern on the mask is a target mask pattern output after processing by the mask pattern processing method in any one of claims 1 to 9, or the mask pattern on the mask is a target mask pattern output by the mask pattern processing system in claim 10.

Images