Photolithography hotspot detection method and device, electronic equipment and readable storage medium

Abstract

The application provides a photolithography hotspot detection method and device, electronic equipment and readable storage medium. The method comprises the following steps: obtaining a mask pattern of photolithography, and processing the mask pattern through N communication channels to obtain N eigenimages; inputting the N eigenimages into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure graph; extracting a contour graph from the predicted photoresist structure graph using a specified photoresist thickness, and comparing the contour graph corresponding to the specified photoresist thickness with a contour graph at the bottom of the predicted photoresist structure graph to find a photolithography hotspot of the photoresist. In this way, the calculation time can be saved to realize photolithography hotspot detection related to the photoresist structure of the whole chip; further, the accuracy of finding the photolithography hotspot can be improved.

Description

Technical Field

[0001] This application relates to the field of integrated circuit manufacturing technology, and in particular to a method, apparatus, electronic device and readable storage medium for detecting photolithographic hotspots. Background Technology

[0002] In advanced lithography processes, the health of the 3D structure of the photoresist after lithography encompasses both the accuracy of the photoresist edge positions and the photoresist thickness. Currently, verification and inspection of photomask data after Optical Proximity Correction (OPC) only involves the accuracy of the photoresist edge positions after full-chip lithography; the photoresist thickness or 3D structure cannot yet be inspected across the entire chip. However, the 3D structure of the photoresist plays a crucial role in verifying the data quality after OPC technology.

[0003] In the prior art, the three-dimensional light intensity distribution function in the photoresist can be approximated by calculating the light intensity distribution at several different depths in the photoresist. Furthermore, the three-dimensional light intensity distribution function is input into the detection model for prediction, the prediction result is obtained, and the prediction result is verified to determine the lithographic hotspots related to the 3D structure of the photoresist.

[0004] However, the above methods require a large amount of computing resources and a long computation time, and the model prediction accuracy is low, the processing speed is slow, and the accuracy is not high. Summary of the Invention

[0005] This application provides a method, apparatus, electronic device, and readable storage medium for detecting photolithographic hotspots, which can save computing time, reduce resource waste, improve prediction accuracy and processing speed, and further improve the accuracy of finding photolithographic hotspots related to the 3D structure of photoresist.

[0006] In a first aspect, this application provides a method for detecting photolithographic hotspots, the method comprising:

[0007] The photolithographic mask pattern is obtained, and the mask pattern is processed through N communication channels to obtain N intrinsic images;

[0008] The N intrinsic images are input into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure diagram.

[0009] Using a specified photoresist thickness, a contour map is extracted from the predicted photoresist structure map, and the contour map is compared with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithographic hot spots.

[0010] Optionally, the mask pattern is processed through N communication channels to obtain N intrinsic images, including:

[0011] Extract the imaging conditions of the mask pattern, and determine the feature function set of the mask pattern based on the imaging conditions;

[0012] Based on the imaging equation, the feature function set is decomposed into feature function information of N communication channels;

[0013] Obtain the mask transfer function of the mask pattern, and calculate the approximate information of the mask pattern based on the mask transfer function and N feature function information respectively to obtain N intrinsic images.

[0014] Optionally, the intrinsic image is determined by the following formula:

[0015]

[0016] Among them, S i (x, y; z = z0) represents the i-th intrinsic image. Let M(x, y) represent the characteristic function information of the i-th communication channel at any depth z0, and M(x, y) represent the mask transfer function of the mask pattern. This represents the convolution operation.

[0017] Optionally, the method further includes:

[0018] Obtain a dataset, which includes a training subset, a validation subset, and a test subset; wherein each subset of the dataset includes: multiple intrinsic images processed by the communication channel and their corresponding photoresist structure diagrams;

[0019] Based on the dataset, train the machine learning model;

[0020] Accordingly, the N intrinsic images are input into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure map, including:

[0021] The N intrinsic images are input into a machine learning model trained on the dataset for prediction, and a predicted photoresist structure diagram synthesized from the N intrinsic images is predicted.

[0022] Optionally, the method further includes: verifying the degree of matching between the predicted photoresist structure diagram and the photoresist structure diagram obtained by experiment or rigorous simulation;

[0023] Accordingly, the matching degree is determined by the following formula:

[0024]

[0025] Where NCC represents the matching degree, and G(i,j) represents the photoresist thickness value in the i-th row and j-th column of the predicted photoresist structure diagram. The mean value of the photoresist thickness in the predicted photoresist structure diagram is represented by g(i,j), where g(i,j) represents the photoresist thickness value in the i-th row and j-th column of the photoresist structure diagram obtained experimentally or through rigorous simulation. The mean value of the photoresist thickness in the photoresist structure diagram obtained by experiment or rigorous simulation is represented by m and n, where m and n represent m rows and n columns of the photoresist structure diagram.

[0026] Optionally, comparing the contour map with the contour map at the bottom of the predicted photoresist structure map to find photolithographic hotspots in the photoresist includes:

[0027] Calculate the contour variation degree of the bottom contour map of the contour map and the bottom contour map of the predicted photoresist structure map respectively;

[0028] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map is less than a preset threshold, then there are no photolithographic hot spots in the detected area.

[0029] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map exceeds a preset threshold at a certain position, then a photolithography hotspot exists at this position, and a prompt message is sent to remind the user to find the photolithography hotspot of the photolithography mask data based on the position.

[0030] Optionally, the method further includes:

[0031] If a photolithographic hotspot is detected in the photolithographic mask data, the optical proximity correction (OPC) formula is adjusted to repair the detected hotspot. The OPC formula is a series of preset software calculation operation instructions designed to convert the chip design pattern into a mask pattern. The resulting photolithographic mask data is used to determine the exposure pattern of the photoresist.

[0032] Secondly, this application also provides a photolithography hotspot detection device, the device comprising:

[0033] The acquisition module is used to acquire the photolithographic mask pattern and process the mask pattern through N communication channels to obtain N intrinsic images;

[0034] The prediction module is used to input the N intrinsic images into a pre-trained machine learning model for prediction, and obtain a predicted photoresist structure diagram.

[0035] The search module extracts a contour map from the predicted photoresist structure map using a specified photoresist thickness, and compares the contour map with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithography hotspots.

[0036] Thirdly, this application also provides an electronic device, including: a processor, a memory, and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor, and the computer program includes instructions for performing the lithographic hotspot detection method as described in any of the first aspects.

[0037] Fourthly, this application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the lithographic hotspot detection method as described in any of the first aspects.

[0038] Fifthly, this application also provides a computer program product, including program code, which, when a computer runs the computer program, performs the method as described in any of the first aspects.

[0039] In summary, this application provides a method, apparatus, electronic device, and readable storage medium for detecting photolithographic hotspots. It acquires a photolithographic mask pattern, processes the mask pattern information through a series of independent communication channels to obtain the desired intrinsic image, where each channel forms one intrinsic image. Furthermore, the obtained intrinsic image is input into a pre-trained machine learning model for prediction, resulting in a predicted photoresist structure map, which is a three-dimensional thickness map of the photoresist. Further, a contour map at a specified photoresist thickness can be extracted from the predicted photoresist structure map, and the contour map is compared with the contour map at the bottom of the predicted photoresist structure map to locate photolithographic hotspots. This saves computation time, reduces resource waste, improves prediction accuracy and processing speed, and further improves the accuracy of locating photoresist thickness-related hotspots. Attached Figure Description

[0040] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0041] Figure 1 A photolithography process flow diagram provided for an embodiment of this application;

[0042] Figure 2 This application provides a schematic diagram of a structure in which photoresist has photolithographic hotspots after photolithography.

[0043] Figure 3 This is a schematic diagram of an application scenario provided by an embodiment of this application;

[0044] Figure 4 A schematic flowchart of a photolithographic hotspot detection method provided in an embodiment of this application;

[0045] Figure 5 This is a schematic diagram of the structure of a communication channel processing mask pattern provided in an embodiment of this application;

[0046] Figure 6 A schematic diagram of a structure for prediction using a deep convolutional neural network model is provided in an embodiment of this application;

[0047] Figure 7 This is a schematic diagram illustrating the results of model prediction using different model inputs, as provided in an embodiment of this application.

[0048] Figure 8A A schematic diagram comparing the prediction of a DCNN model with the actual 3D structure of photoresist after photolithography, provided for an embodiment of this application;

[0049] Figure 8B A schematic diagram comparing another DCNN model prediction with the actual 3D structure of photoresist after lithography, provided for an embodiment of this application;

[0050] Figure 9 This is a schematic diagram of the structure of a photolithography hotspot detection device provided in an embodiment of this application;

[0051] Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0052] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0053] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0054] To facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and purpose. For example, "first device" and "second device" are merely used to distinguish different devices and do not limit their order of execution. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that "first" and "second" do not necessarily imply that they are different.

[0055] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0056] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0057] The embodiments of this application will now be described in conjunction with the accompanying drawings. To provide a clearer description of this application, the relevant process flow and problems encountered will be briefly described below. Figure 1 A photolithography process flow diagram is provided for an embodiment of this application, such as... Figure 1As shown, firstly, photoresist is coated onto the silicon wafer to form a uniform, highly adhesive, and defect-free photoresist film. Next, the spin-coated photoresist film is baked at a high temperature to evaporate as much solvent as possible. Then, a photomask, created using photomask data corrected by Optical Proximity Correction (OPC) technology, is used to expose the photoresist, i.e., to irradiate it. At this point, a photo-induced acid reaction occurs. Next, the exposed wafer is moved to a hot plate for high-temperature baking, triggering an acid-catalyzed chemical reaction. The difference in solubility between the exposed and unexposed areas of the photoresist creates a difference in solubility. Finally, the wafer, after the acid-catalyzed chemical reaction, is immersed in a developing solution. During development, the exposed areas of the positive photoresist or the unexposed areas of the negative photoresist dissolve, thus presenting a three-dimensional pattern. This process is called development. Ultimately, a 3D structure of the photoresist after photolithography is formed (only a portion is shown in the figure). After the 3D structure of the photoresist is formed by photolithography, the next step is to etch the wafer with the 3D structure covered by the photoresist. The material in the areas not protected by the photoresist is etched. To ensure the health of the etched pattern, the quality of the 3D structure of the photoresist needs to be predicted.

[0058] Because the top loss of negative tone development (NTD) photoresist is more severe than that of positive tone development (PTD) photoresist, and in low K1 (K1 is called the process factor or resolution constant, low K1 means K1 < 0.4) lithography processes, some patterned areas have low imaging contrast, which may lead to severe top loss of the photoresist. For these patterned areas, even if the bottom edge of the photoresist is correctly positioned, there is a high risk that the pattern will fail after etching. This is because severe top loss of the photoresist results in insufficient photoresist thickness for subsequent etching processes, usually leading to patterning failure after etching and the appearance of lithographic hotspots. Specifically… Figure 2 This application provides a schematic diagram of a structure where photoresist has photolithographic hotspots after photolithography, as shown in the embodiments of this application. Figure 2 As shown in the figure, Figure A shows the three-dimensional structure after photoresist etching, and Figure B is a magnified three-dimensional structure of the photoresist after photolithography, which is the circle in Figure A on the right. In Figure B, area 201 is the area where photolithography hotspots exist. Photolithography hotspots are the locations where problems and defects exist.

[0059] Post-Optical Photolithography (OPC) technology can lead to lithographic hotspots related to the photoresist thickness, causing delays in technology development or product delivery. To prevent lithographic hotspots related to the 3D structure of the photoresist after OPC, it is necessary to predict the 3D structure of the photoresist corresponding to the post-OPC photomask data to ensure safe production of the entire chip.

[0060] For example, Figure 3 This is a schematic diagram illustrating an application scenario provided by an embodiment of this application. The photolithographic hotspot detection method provided in this application can be applied to, for example... Figure 3 The application scenario shown includes: a first mask pattern 301, a second mask pattern 302, and a third mask pattern 303; an image imaging device 304; a server 305; a terminal device 306; and a user 307. The first mask pattern 301 is a head-to-head regular pattern, the second mask pattern 302 is a head-to-line regular pattern, and the third mask pattern 303 is a random 2D pattern. In this scenario, the image imaging device 304 can send the acquired first mask pattern 301, second mask pattern 302, or third mask pattern 303 to the server 307. 05. Further processing: The server 305 can receive the first mask pattern 301, the second mask pattern 302, or the third mask pattern 303 and perform a series of prediction processes to obtain prediction results. The prediction results are then compared with the actual photoresist 3D model to find the photolithographic hotspots in the photoresist 3D structure. Furthermore, the server 305 can also send the photoresist 3D structure with the found photolithographic hotspots to the terminal device 306 with a display screen. The terminal device 306 receives the photoresist 3D structure and displays it on the screen for the user 307 to view.

[0061] It is understood that server 305 can be a server integrated with image imaging device 304, and the two are an integrated device that can acquire and process images of the 3D structure of photoresist. Alternatively, they can be separate devices, each performing different functions. Image imaging device 304 only acquires images and sends them to server 305 for a series of image processing steps. This application does not specifically limit this aspect.

[0062] It should be noted that the terminal device 306 may be a large screen (or smart screen), mobile phone, tablet computer, smartwatch, smart bracelet, smart glasses or other terminal devices with a display screen, etc., and this application embodiment does not specifically limit it.

[0063] In existing technologies, the three-dimensional light intensity distribution function in photoresist can be approximated by calculating the light intensity distribution at several different depths in the photoresist. Furthermore, the three-dimensional light intensity distribution function is input into the detection model for prediction, the prediction result is obtained, and the prediction result is verified to determine the photolithography hotspot.

[0064] However, the above methods require a large amount of computing resources and a long computation time, and the model prediction accuracy is low, the processing speed is slow, and the accuracy is not high.

[0065] Therefore, this application provides a method for detecting photolithographic hotspots. This method acquires a photolithographic mask pattern and processes the acquired mask pattern through a series of independent communication channels to obtain the desired intrinsic image, where each channel forms an intrinsic image. Further, the obtained intrinsic image is input into a pre-trained machine learning model for prediction, resulting in a predicted photoresist structure map. This machine learning model has a fixed mapping function for the three-dimensional light intensity distribution function in the photoresist and the corresponding three-dimensional structure of the photoresist after photolithography. Further, a contour map is extracted from the predicted photoresist structure map using a specified photoresist thickness, and the contour map corresponding to the specified photoresist thickness is compared with the contour map at the bottom of the predicted photoresist structure map to locate photolithographic hotspots in the photoresist.

[0066] For example, Figure 4 This is a flowchart illustrating a photolithographic hotspot detection method provided in an embodiment of this application, as shown below. Figure 4 As shown, the method in this application embodiment includes:

[0067] S401. Obtain the photolithographic mask pattern and process the mask pattern through N communication channels to obtain N intrinsic images.

[0068] In this step, the mask pattern refers to the pattern used in photolithography on the mask. The mask can refer to a structure on a thin film, plastic or glass substrate material where various functional patterns are fabricated and precisely positioned for selective exposure of photoresist coatings.

[0069] The mask pattern can be graphically split according to the needs of the process flow, and designed into two or more masks to represent patterns at different locations.

[0070] In this embodiment, the intrinsic image refers to the intrinsic image of multiple independent imaging channels at a single depth Z of the photoresist, used to characterize the photomask pattern information and imaging formation feature information (i.e., imaging conditions). The imaging formation feature information may include exposure wavelength, numerical aperture, illumination source, photoresist stack, etc., and each communication channel can form an intrinsic image. Different intrinsic images are used to represent the images corresponding to different independent imaging channels under different imaging conditions.

[0071] The communication channel refers to the process of decomposing the light intensity distribution function under partially coherent illumination conditions at the photoresist depth Z0 into a series of light intensity distribution functions under coherent imaging systems, based on imaging theory. This decomposition is used to transform the mask pattern into an intrinsic image through a series of independent transformations. The processing of this communication channel can be understood as an optimal and most efficient spatial image information encoding scheme. This scheme has been experimentally proven to have optimal decomposition in terms of computational efficiency. The sum of the coherent imaging systems constitutes a complete light intensity distribution image of the mask pattern. Specifically, the formula for the sum of the coherent imaging systems I(x, y; z = z0) can be expressed as follows:

[0072]

[0073] Where (x, y; z = z0) represents an arbitrary (x, y) plane of depth z0, α i (z = z0) represents the weight of the intrinsic image formed after passing through the i-th communication channel at any depth z0. The characteristic function information of the i-th communication channel at any depth z0 is represented by M(x, y), where M(x, y) represents the mask transfer function of the mask pattern. This represents the convolution operation.

[0074] It should be noted that α i The weights can be set according to the actual decomposition situation or determined according to the decomposition equation. This application embodiment does not make specific limitations. The mask transfer function M(x, y) of the mask pattern is determined by the optical properties of the mask pattern and the photolithographic mask. Each mask pattern has a corresponding mask function. This application embodiment does not limit the specific mask function.

[0075] For example, in Figure 3 In the application scenario, the image imaging device 304 can acquire the first mask pattern 301 and send the first mask pattern 301 to the server 305. Correspondingly, the server 305 obtains the first mask pattern 301 and processes the first mask pattern 301 through 5 communication channels to obtain 5 intrinsic images.

[0076] It is understood that the number of communication channels is determined by the number of coherent imaging systems decomposed from the imaging equation, and this application does not specifically limit this.

[0077] S402. Input the N intrinsic images into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure diagram.

[0078] In this embodiment, the machine learning model is a deep learning model built upon a deep convolutional neural network (DCNN). DCNN is a type of feedforward neural network that includes convolutional computation and has a deep structure; it is one of the representative algorithms of deep learning. DCNN possesses representation learning capabilities, enabling translation-invariant classification of input information according to its hierarchical structure. The shared parameters of the convolutional kernels within its hidden layers and the sparsity of inter-layer connections allow DCNNs to learn gridded features, such as pixels in an image, with relatively low computational cost, achieving stable results and requiring no additional feature engineering of the data.

[0079] Understandably, deep convolutional neural networks differ from feedforward neural networks. Deep convolutional neural networks perform better in image processing. Specifically, deep convolutional neural networks can use spatial transformations to define axes and image boundaries before prediction, thereby helping to correct imbalances caused by scaling and rotation, as well as noise in attention mechanisms. In other words, deep convolutional neural networks can denoise images and solve the problem of severe image degradation under low-light conditions.

[0080] In this step, a deep learning model can refer to an architecture based on a deep learning algorithm that can be used to iteratively train training data. Examples include the U-net model and the GAN model.

[0081] It should be noted that in the embodiments of this application, the feature image is used as the input of the DCNN model. The DCNN model can learn a nonlinear mapping function from a set of feature images to a three-dimensional thickness map of the photoresist through training. That is, the input is a two-dimensional intrinsic image, and the output is a three-dimensional photoresist structure map, that is, a thickness map of the photoresist.

[0082] For example, in Figure 3 In application scenarios, taking the first mask pattern 301 acquired by the image imaging device 304 as an example, the server 305 inputs the five intrinsic images obtained by processing the first mask pattern 301 through five communication channels into the pre-trained DCNN model for prediction, and can obtain the predicted 3D structure map of the photoresist.

[0083] S403. Extract a contour map from the predicted photoresist structure map using a specified photoresist thickness, and compare the contour map with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithography hotspots.

[0084] In this step, after obtaining the predicted photoresist structure diagram, the specified photoresist thickness is input into the server to obtain a planar contour diagram of the photoresist at any thickness (i.e., depth). By comparing the planar contour diagram of the photoresist at any thickness with the planar contour diagram at the bottom of the predicted photoresist structure diagram, the photolithography hotspots can be found by comparing the differences between the two.

[0085] For example, in Figure 3 In the application scenario, after the server 305 inputs five intrinsic images into the pre-trained DCNN model to obtain the predicted 3D structure map of the photoresist, the server 305 can further extract the contour map of a certain photoresist thickness in the 3D structure map, such as the contour map at a depth of z = 500nm. Further, the server can obtain the contour map of the bottom of the 3D structure map of the photoresist, and compare the contour map at a depth of z = 500nm with the bottom contour map to find the photoresist lithography hot spots. Preferably, the photoresist lithography hot spots can be found by calculating the contour change degree of the two.

[0086] It is understood that the specified photoresist thickness can be a preset thickness of the system, which can be used to search for photolithography hotspots each time, or it can be a thickness that is manually entered. This application does not specifically limit this.

[0087] Therefore, the lithography hotspot detection method proposed in this application uses the intrinsic images of multiple independent imaging channels at a single depth Z of the photoresist as the input of the model, which reduces the waste of computing resources and computing time, and improves processing efficiency. The model used is a pre-trained machine learning model with high prediction accuracy, which greatly improves the accuracy of finding lithography hotspots.

[0088] Optionally, the mask pattern is processed through N communication channels to obtain N intrinsic images, including:

[0089] Extract the imaging conditions of the mask pattern, and determine the feature function set of the mask pattern based on the imaging conditions;

[0090] Based on the imaging equation, the feature function set is decomposed into feature function information of N communication channels;

[0091] Obtain the mask transfer function of the mask pattern, and calculate the approximate information of the mask pattern based on the mask transfer function and N feature function information respectively to obtain N intrinsic images.

[0092] In this embodiment, imaging conditions can refer to the conditional information required to convert the mask pattern into a 3D structure diagram, such as exposure wavelength, numerical aperture, illumination source, photoresist stack, photoresist thickness, etc. The feature function set can refer to the feature function set of the transmission cross coefficient matrix, which is composed of various factors in the imaging conditions. The transmission cross coefficient matrix contains information about the complete imaging conditions, while the feature function information refers to the partial feature data in the transmission cross coefficient matrix corresponding to each communication channel.

[0093] It should be noted that the data in the feature function information is determined by the decomposition of the imaging equation, and this application embodiment does not make specific limitations. However, the input of the machine learning model must include information about the mask pattern itself and the lithographic imaging conditions.

[0094] It is understandable that the mask transfer function is determined by the mask pattern. After obtaining the photolithographic mask pattern, the mask transfer function corresponding to the mask pattern is already known. The mask transfer function can be a mask transfer function pre-stored in the system corresponding to each mask pattern, or it can be a mask transfer function manually input corresponding to the mask pattern.

[0095] In this step, approximate information can refer to all the information that contains information about the mask pattern and imaging conditions. This approximate information contains the information necessary for the DCNN structure to learn and construct the light intensity distribution function in the photoresist. In other words, the approximate information represents the information that the intrinsic image possesses.

[0096] For example, Figure 5 This is a schematic diagram of the structure of a communication channel processing mask pattern provided in an embodiment of this application, as shown below. Figure 5 As shown, the imaging conditions of the mask pattern in the image are extracted. Based on these imaging conditions, the feature function set of the mask pattern is determined. Furthermore, based on the imaging equation, this feature function set is decomposed into five communication channels, which are respectively... Furthermore, the mask pattern is processed... The processing yields S1-S5 intrinsic images. Specifically, these images are processed through a communication channel. The processing can be described as follows: obtain the mask transfer function M of the mask pattern in the figure, and calculate the approximate information of the mask pattern based on the mask transfer function M and the characteristic function information of the five communication channels to obtain five intrinsic images.

[0097] Therefore, processing the mask pattern through a communication channel to obtain the model input not only reduces the amount of computation, but also ensures that the result of the communication channel processing is the optimal decomposition, thus improving accuracy.

[0098] Optionally, the intrinsic image is determined by the following formula:

[0099]

[0100] Among them, S i (x, y; z = z0) represents the i-th intrinsic image. The characteristic function information of the i-th communication channel at any depth z0 is represented by M(x, y), where M(x, y) represents the mask transfer function of the mask pattern. This represents the convolution operation.

[0101] For example, taking the second communication channel at a depth z=1 as an example, the above formula can be used to calculate the second intrinsic image. Where S2(x, y; z = 1) is the eigenimage of the plane with depth z = 1, and M(x, y) is a known mask transfer function. This refers to the characteristic function information of the second communication channel with depth z = 1.

[0102] Therefore, using the above formula to calculate the intrinsic image greatly improves computational efficiency and simplicity.

[0103] Optionally, the method further includes:

[0104] Obtain a dataset, which includes a training subset, a validation subset, and a test subset; wherein each subset of the dataset includes: multiple intrinsic images processed by the communication channel and their corresponding photoresist structure diagrams;

[0105] Based on the dataset, train the machine learning model;

[0106] Accordingly, the N intrinsic images are input into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure map, including:

[0107] The N intrinsic images are input into a machine learning model trained on the dataset for prediction, and a predicted photoresist structure diagram synthesized from the N intrinsic images is predicted.

[0108] In this step, a manually provided dataset can be obtained, which includes a training subset, a validation subset, and a test subset in a ratio of 6:2:2. Each subset may include multiple intrinsic images processed by the communication channel and their corresponding photoresist structure diagrams. The intrinsic images are two-dimensional feature images, and the photoresist structure diagrams are photoresist thickness diagrams characterizing the three-dimensional structure of the photoresist.

[0109] It should be noted that the ratio of the training subset, the validation subset, and the test subset can also be 6:1:1. This application does not specifically limit this, but the ratio of the validation subset to the test subset should not be greater than that of the training subset. The ratio of the validation subset to the test subset can be the same or different.

[0110] Understandably, acquiring the dataset and training the machine learning model only needs to be done once, and then the trained machine learning model can be directly used to predict N intrinsic images.

[0111] For example, the server can train the machine learning model in advance, i.e., acquire a dataset that may include a training subset, a validation subset, and a test subset in a ratio of 6:2:2. Each subset may include multiple two-dimensional feature images processed by the communication channel and their corresponding three-dimensional structure maps of photoresist. Further, the DCNN model is trained based on the dataset.

[0112] Correspondingly, Figure 6 This application provides a schematic diagram of a structure for prediction using a deep convolutional neural network model, as shown in the embodiments of this application. Figure 6 As shown, the photolithographic mask pattern is processed through a communication channel to obtain 5 intrinsic images. These 5 intrinsic images are then input into a DCNN model trained on a dataset for prediction, which produces a predicted 3D structure diagram of the photoresist synthesized from the 5 intrinsic images.

[0113] Therefore, using a pre-trained machine learning model to predict intrinsic images can improve the prediction speed and accuracy.

[0114] Optionally, the method further includes: verifying the degree of matching between the predicted photoresist structure diagram and the photoresist structure diagram obtained by experiment or rigorous simulation;

[0115] Accordingly, the matching degree is determined by the following formula:

[0116]

[0117] Where NCC represents the matching degree, and G(i,j) represents the photoresist thickness value in the i-th row and j-th column of the predicted photoresist structure diagram. The mean value of the photoresist thickness in the predicted photoresist structure diagram is represented by g(i,j), where g(i,j) represents the photoresist thickness value in the i-th row and j-th column of the photoresist structure diagram obtained experimentally or through rigorous simulation. The mean value of the photoresist thickness in the photoresist structure diagram obtained by experiment or rigorous simulation is represented by m and n, where m and n represent m rows and n columns of the photoresist structure diagram.

[0118] In this embodiment, the matching degree can refer to the similarity between the predicted photoresist structure diagram and the actual photoresist structure diagram. It can be obtained by calculating the normalized cross correlation (NCC). NCC is used to describe the correlation between two targets, that is, it can be used to characterize the similarity between targets. Generally, NCC is also used for image matching, that is, searching for the region with the highest NCC in an image and matching it with a small known region, and then matching it with the whole image.

[0119] In this step, the photoresist thickness value can refer to the depth value of each point in each plane of the photoresist structure diagram.

[0120] For example, by setting the calculation of the photoresist structure diagram to 2 rows and 2 columns, the photoresist thickness values ​​at each point in the following plane can be extracted: G(1,1)=150nm, G(1,2)=150nm, G(2,1)=200nm, G(2,2)=300nm; g(1,1)=100nm, g(1,2)=150nm, g(2,1)=200nm, g(2,2)=150nm; Based on the above data, NCC = 1 - 0.2887 = 0.7113 can be calculated.

[0121] It is understandable that the number of m rows and n columns in the photoresist structure diagram is not limited to the 2 rows and 2 columns mentioned above; the number should be quite large. The above is just an example.

[0122] It should be noted that NCC=1 means a perfect match between the image predicted by the DCNN model and the real image. The larger the NCC value, the higher the degree of matching between the model's predicted image and the real image.

[0123] Therefore, by verifying the matching degree between the predicted photoresist structure diagram and the actual photoresist structure diagram using the above formula, the accuracy of the predicted photoresist structure diagram obtained in this application is confirmed.

[0124] To demonstrate that the intrinsic images used as input to the machine learning model in this application have good prediction accuracy, experiments were conducted to verify this. For example, Figure 7 This is a schematic diagram illustrating the results of model prediction using different model inputs, as provided in an embodiment of this application. Figure 7As shown, three models were constructed using three different types of model inputs. The input of Model 1 is the light intensity distribution function of a single Z-plane in the photoresist; the input of Model 2 is the light intensity distribution function of 5 different Z-planes in the photoresist; and the input of Model 3 is 5 intrinsic images on a single Z-plane in the photoresist. By calculating their NCC using the matching degree calculation formula, and their histograms as shown in Figure 7, the conclusions are as follows: For the input of the light intensity distribution function of a single Z-plane in the photoresist, P(NCC >= 0.94) = 87.4%, with a minimum NCC of 0.69; for the input of the light intensity distribution functions of 5 different Z-planes in the photoresist, P(NCC >= 0.94) = 94.4%, with a minimum NCC of 0.86; for the input of 5 intrinsic images on a single Z-plane in the photoresist, P(NCC >= 0.94) = 99.4%, with a minimum NCC of 0.88. Therefore, the scheme proposed in this application, which uses communication channel processing to obtain intrinsic images, performs best in terms of model prediction accuracy.

[0125] The histogram is used to represent the trend of NCC between the predicted photoresist structure diagram and the actual photoresist structure diagram. The horizontal axis represents NCC, and the vertical axis represents the probability of NCC in each interval.

[0126] It should be noted that Model 3 uses a single Graphics Processing Unit (GPU), and the prediction time for each image (area = 1800nm ​​x 1800nm) is 7.14ms. Compared to Model 1's 6.92ms and Model 2's 8.22ms, this speed is fast enough. The Central Processing Unit (CPU) of Model 3 has an execution time of 0.856s, which is also fast enough compared to Model 1's 0.84036s and Model 2's 0.894s. Therefore, Model 3 can be used to realize the detection of 3D structure of photoresist after full-chip photolithography.

[0127] For example, Figure 8A This application provides a schematic diagram comparing the prediction of a DCNN model with the actual 3D structure of the photoresist after photolithography, as shown in the embodiments of this application. Figure 8A As shown, the top image is a magnified view of the actual 3D structure of the photoresist after photolithography, and the bottom image is a corresponding magnified view of the 3D structure of the photoresist predicted by the DCNN model. Figure 8B This is a schematic diagram comparing the prediction of another DCNN model with the actual 3D structure of the photoresist after photolithography, as provided in an embodiment of this application; Figure 8BAs shown, the upper part is a magnified view of the actual 3D structure of the photoresist after lithography, and the lower part is a corresponding magnified view of the 3D structure of the photoresist predicted by the DCNN model. The circled areas represent the locations of lithography hotspots. Through comparison, it was found that the two 3D structure images of the photoresist are highly similar. The difference between the actual 3D structure of the photoresist after lithography and the 3D structure predicted by the DCNN model is not significant. Therefore, using the DCNN model to predict the 3D structure of the photoresist has high accuracy.

[0128] Optionally, comparing the contour map with the contour map at the bottom of the predicted photoresist structure map to find photolithographic hotspots in the photoresist includes:

[0129] Calculate the contour variation degree of the bottom contour map of the contour map and the bottom contour map of the predicted photoresist structure map respectively;

[0130] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map is less than a preset threshold, then there are no photolithographic hot spots in the detected area.

[0131] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map exceeds a preset threshold at a certain position, then a photolithography hotspot exists at that position; and a prompt message is sent to remind the user to find the photolithography hotspot in the photolithography mask data based on the position.

[0132] In this embodiment, the preset threshold can refer to the maximum allowable change in the contour of a certain thickness of photoresist and the contour at the bottom of the photoresist structure diagram that the system sets to determine whether there are photolithographic hot spots in the image. The preset threshold can be determined by experience or experiment. As an example, it can be assumed to be 0.94. This embodiment does not limit the specific value of the preset threshold.

[0133] It should be noted that the method for calculating the contour variation degree in this application embodiment is not specifically limited, and can refer to the calculation method in the prior art.

[0134] In this step, the prompt message is used to indicate the presence of lithographic hotspots in the 3D structure diagram of the photoresist. The method of sending the prompt message is not specifically limited in this embodiment. It can be to display a message prompt box on the display screen of the server, displaying "There is a lithographic hotspot at position 1 in the photoresist structure diagram," or to directly display an image of the photoresist structure diagram on the display screen, marking the lithographic hotspots in the photoresist structure diagram with red boxes to prompt the user to promptly locate and correct the lithographic hotspots in the photoresist based on the prompt message. Alternatively, the server can send a message prompt to the user terminal device, stating "There is a lithographic hotspot at position 1 in the photoresist structure diagram," or send image information to the display screen of the user terminal device, further displaying an image of the photoresist structure diagram on the display screen of the user terminal device, marking the lithographic hotspots in the photoresist structure diagram with red boxes to prompt the user to promptly locate and correct the lithographic hotspots in the photoresist based on the prompt message.

[0135] For example, if the server calculates that the contour variation of all positions in the predicted photoresist structure diagram is 0.80, which is less than the preset threshold of 0.94, it means that there are no photoresist hot spots. If the server calculates that the contour variation of the predicted photoresist structure diagram at a certain position is 0.99, which is greater than the preset threshold of 0.94, a message prompt box is displayed on the display screen carried by the server. The message prompt box displays "There is a photoresist hot spot at position 1 in the photoresist structure diagram", which is used to remind the user to find the photoresist hot spot based on position 1.

[0136] It should be noted that this application can also determine whether there is a difference between the contour map corresponding to the specified photoresist thickness and the contour map at the bottom of the predicted photoresist structure map by calculating the similarity or other characteristic factors, and further find photolithography hot spots. The embodiments of this application do not specifically limit the method of comparing the contour map corresponding to the specified photoresist thickness and the contour map at the bottom of the predicted photoresist structure map.

[0137] Therefore, by calculating the contour variation between the contour map of any thickness of the predicted photoresist structure and the contour map of the bottom of the predicted photoresist structure, it is possible to determine whether there are photolithography hot spots in the photoresist. Furthermore, the location of the photolithography hot spots can be found and displayed to the user, which greatly improves the processing speed, makes it easier for users to understand the situation at any time, and correct the photolithography hot spots.

[0138] Optionally, the method further includes:

[0139] If a photolithographic hotspot is detected in the photolithographic mask data, the optical proximity correction (OPC) formula is adjusted to repair the detected hotspot. The OPC formula is a series of preset software calculation operation instructions designed to convert the chip design pattern into a mask pattern. The resulting photolithographic mask data is used to determine the exposure pattern of the photoresist.

[0140] In this step, the OPC formula is a series of pre-defined software calculation instructions designed to convert chip design patterns into mask patterns. The resulting photolithography mask data allows the server to eliminate photolithography hotspots that are related to photoresist thickness, based on the finely tuned OPC formula.

[0141] Among them, photolithography mask data can refer to the design pattern data carried on the mask. Based on this data, light can be passed through the mask to project the design pattern onto the photoresist.

[0142] For example, if the server determines that there are lithographic hot spots in a certain photomask data, it can adjust the OPC formula based on this photomask data. Furthermore, the adjusted OPC formula can be used to generate the required photoresist structure pattern to achieve the purpose of repairing the detected lithographic hot spots.

[0143] Therefore, this application can also correct the photomask data corresponding to photoresist with photolithographic hotspots, reduce resource waste, and improve the success rate of photolithography masks.

[0144] In the foregoing embodiments, the lithographic hotspot detection method provided by the embodiments of this application has been described. To implement the functions of the methods provided by the embodiments of this application, the electronic device serving as the execution subject may include hardware structures and / or software modules, implementing the above functions in the form of hardware structures, software modules, or a combination of hardware structures and software modules. Whether a particular function is executed in the form of hardware structures, software modules, or a combination of hardware structures and software modules depends on the specific application and design constraints of the technical solution.

[0145] For example, Figure 9 This is a schematic diagram of the structure of a photolithography hotspot detection device provided in an embodiment of this application, as shown below. Figure 9 As shown, the device includes: an acquisition module 910, a prediction module 920, and a search module 930; wherein, the acquisition module 910 is used to acquire the photolithographic mask pattern and process the mask pattern through N communication channels to obtain N intrinsic images;

[0146] The prediction module 920 is used to input the N intrinsic images into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure diagram.

[0147] The lookup module 930 is used to extract a contour map from the predicted photoresist structure map using a specified photoresist thickness, and compare the contour map with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithography hot spots.

[0148] Optionally, module 910 is used for:

[0149] Extract the imaging conditions of the mask pattern, and determine the feature function set of the mask pattern based on the imaging conditions;

[0150] Based on the imaging equation, the feature function set is decomposed into feature function information of N communication channels;

[0151] Obtain the mask transfer function of the mask pattern, and calculate the approximate information of the mask pattern based on the mask transfer function and N feature function information to obtain N intrinsic images.

[0152] Optionally, the intrinsic image is determined by the following formula:

[0153]

[0154] Among them, S i (x, y; z = z0) represents the i-th intrinsic image. The characteristic function information of the i-th communication channel at any depth z0 is represented by M(x, y), where M(x, y) represents the mask transfer function of the mask pattern. This represents the convolution operation.

[0155] Optionally, the device further includes a training module, the training module being used for:

[0156] Obtain a dataset, which includes a training subset, a validation subset, and a test subset; wherein each subset of the dataset includes: multiple intrinsic images processed by the communication channel and their corresponding photoresist structure diagrams;

[0157] Based on the dataset, train the machine learning model;

[0158] Correspondingly, the prediction module 920 is specifically used for:

[0159] The N intrinsic images are input into a machine learning model trained on the dataset for prediction, and a predicted photoresist structure diagram synthesized from the N intrinsic images is predicted.

[0160] Optionally, the device further includes a verification module, which is used to verify the degree of matching between the predicted photoresist structure pattern and the photoresist structure pattern obtained by experiment or rigorous simulation;

[0161] Specifically, the matching degree is determined by the following formula:

[0162]

[0163] Where NCC represents the matching degree, and G(i,j) represents the photoresist thickness value in the i-th row and j-th column of the predicted photoresist structure diagram. The mean value of the photoresist thickness in the predicted photoresist structure diagram is represented by g(i,j), where g(i,j) represents the photoresist thickness value in the i-th row and j-th column of the photoresist structure diagram obtained experimentally or through rigorous simulation. The mean value of the photoresist thickness in the photoresist structure diagram obtained by experiment or rigorous simulation is represented by m and n, where m and n represent m rows and n columns of the photoresist structure diagram.

[0164] Optionally, the lookup module 930 is specifically used for:

[0165] Calculate the contour variation degree of the bottom contour map of the contour map and the bottom contour map of the predicted photoresist structure map respectively;

[0166] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map is less than a preset threshold, then there are no photolithographic hot spots in the detected area.

[0167] If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map exceeds a preset threshold at a certain position, then a photolithography hotspot exists at that position; and a prompt message is sent to remind the user to find the photolithography hotspot in the photolithography mask data based on the position.

[0168] Optionally, the device further includes a correction module, the correction module being configured to:

[0169] If a photolithographic hotspot is detected in the photolithographic mask data, the optical proximity correction (OPC) formula is adjusted to repair the detected hotspot. The OPC formula is a series of preset software calculation operation instructions designed to convert the chip design pattern into a mask pattern. The resulting photolithographic mask data is used to determine the exposure pattern of the photoresist.

[0170] The specific implementation principle and effects of the photolithography hotspot detection device provided in this application embodiment can be found in the relevant descriptions and effects of the above embodiments, and will not be elaborated further here.

[0171] This application also provides a schematic diagram of the structure of an electronic device. Figure 10 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 10As shown, the electronic device may include: a processor 1002 and a memory 1001 communicatively connected to the processor; the memory 1001 stores a computer program; the processor 1002 executes the computer program stored in the memory 1001, causing the processor 1002 to perform the method described in any of the above embodiments.

[0172] The memory 1001 and the processor 1002 can be connected via a bus 1003.

[0173] This application also provides a computer-readable storage medium storing computer program execution instructions, which, when executed by a processor, are used to implement the methods described in any of the foregoing embodiments of this application.

[0174] This application also provides a chip for executing instructions, which is used to perform the methods described in any of the foregoing embodiments of this application as executed by an electronic device.

[0175] This application also provides a computer program product, which includes a computer program that, when executed by a processor, can implement the methods described in any of the foregoing embodiments executed by an electronic device as described in any of the foregoing embodiments of this application.

[0176] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0177] The modules described as separate components may or may not be physically separate. The components shown as modules 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 implement the solution of this embodiment according to actual needs.

[0178] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit composed of the above modules can be implemented in hardware or in the form of hardware plus software functional units.

[0179] The integrated modules implemented as software functional modules described above can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application.

[0180] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0181] The memory may include high-speed random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0182] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0183] The aforementioned storage medium can be implemented from any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium accessible to general-purpose or special-purpose computers.

[0184] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor. Both the processor and the storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and storage medium can exist as discrete components in an electronic device or host device.

[0185] The above description is merely a specific implementation of the embodiments of this application, but the protection scope of the embodiments of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the embodiments of this application should be covered within the protection scope of the embodiments of this application. Therefore, the protection scope of the embodiments of this application should be determined by the protection scope of the claims.

Claims

1. A method for detecting photolithographic hotspots, characterized in that, The method includes: A photolithographic mask pattern is obtained, and the imaging conditions of the mask pattern are extracted. Based on the imaging conditions, the feature function set of the mask pattern is determined. Based on the imaging equation, the feature function set is decomposed into the feature function information of N communication channels. The mask transfer function of the mask pattern is obtained, and the approximate information of the mask pattern is calculated based on the mask transfer function and the N feature function information respectively to obtain N intrinsic images. The N intrinsic images are input into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure diagram. Using a specified photoresist thickness, a contour map is extracted from the predicted photoresist structure map, and the contour map is compared with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithographic hot spots.

2. The method according to claim 1, characterized in that, The intrinsic image is determined by the following formula: in, Represents the i-th intrinsic image. Represents any depth The characteristic function information of the i-th communication channel The mask transfer function representing the mask pattern. This represents the convolution operation.

3. The method according to claim 1, characterized in that, Also includes: Obtain a dataset, which includes a training subset, a validation subset, and a test subset; wherein each subset of the dataset includes: multiple intrinsic images processed by the communication channel and their corresponding photoresist structure diagrams; Based on the dataset, train the machine learning model; Accordingly, the N intrinsic images are input into a pre-trained machine learning model for prediction to obtain a predicted photoresist structure map, including: The N intrinsic images are input into a machine learning model trained on the dataset for prediction, and a predicted photoresist structure diagram synthesized from the N intrinsic images is predicted.

4. The method according to claim 1, characterized in that, The method further includes: verifying the matching degree between the predicted photoresist structure diagram and the photoresist structure diagram obtained by experiment or rigorous simulation; Accordingly, the matching degree is determined by the following formula: in, Represents the degree of matching. The value representing the photoresist thickness in the i-th row and j-th column of the predicted photoresist structure diagram. The mean value of the photoresist thickness represents the predicted photoresist structure. The value in the i-th row and j-th column of the photoresist structure diagram obtained from experiments or rigorous simulations represents the photoresist thickness value. The mean value of the photoresist thickness in the photoresist structure diagram obtained by experiment or rigorous simulation is represented by m and n, where m and n represent m rows and n columns of the photoresist structure diagram.

5. The method according to claim 1, characterized in that, The process of comparing the contour map with the bottom contour map of the predicted photoresist structure to locate photolithographic hotspots in the photoresist includes: Calculate the contour variation degree of the bottom contour map of the contour map and the bottom contour map of the predicted photoresist structure map respectively; If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map is less than a preset threshold, then there are no photolithographic hot spots in the detected area. If the contour variation between the contour map and the bottom contour map of the predicted photoresist structure map exceeds a preset threshold at a certain position, then a photolithography hotspot exists at that position; and a prompt message is sent to remind the user to find the photolithography hotspot in the photolithography mask data based on the position.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: If a photolithographic hotspot is detected in the photolithographic mask data, the optical proximity correction (OPC) formula is adjusted to repair the detected hotspot. The OPC formula is a series of preset software calculation operation instructions designed to convert the chip design pattern into a mask pattern. The resulting photolithographic mask data is used to determine the exposure pattern of the photoresist.

7. A photolithography hotspot detection device, characterized in that, The device includes: An acquisition module is used to acquire a photolithographic mask pattern and extract the imaging conditions of the mask pattern, determine the feature function set of the mask pattern based on the imaging conditions, decompose the feature function set into feature function information of N communication channels based on the imaging equation, acquire the mask transfer function of the mask pattern, and calculate the approximate information of the mask pattern based on the mask transfer function and the N feature function information to obtain N intrinsic images. The prediction module is used to input the N intrinsic images into a pre-trained machine learning model for prediction, and obtain a predicted photoresist structure diagram. The lookup module is used to extract a contour map from the predicted photoresist structure map using a specified photoresist thickness, and compare the contour map with the contour map at the bottom of the predicted photoresist structure map to find photoresist lithography hotspots.

8. An electronic device, characterized in that, include: A processor, a memory, and a computer program; wherein the computer program is stored in the memory and configured to be executed by the processor, the computer program including instructions for performing the lithographic hotspot detection method as claimed in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the lithographic hotspot detection method as described in any one of claims 1-6.

10. A computer program product, characterized in that, Includes program code that, when the computer runs the computer program, performs the method as described in any one of claims 1-6.

Images