A photolithography hotspot detection method, device, storage medium and equipment

Abstract

A photolithography hot spot detection method, device, storage medium and equipment applied to the field of semiconductor technology. The method comprises the following steps: first, obtaining a photolithography layout to be detected, and extracting a typical pattern feature representing pattern information of the photolithography layout; then inputting the typical pattern feature into a pre-constructed photolithography hot spot detection model (including a convolution layer, a full connection layer and an output layer) to obtain a photolithography hot spot in the photolithography layout. It can be seen that, since the method is to input the typical pattern feature representing the pattern information of the photolithography layout to be detected into the pre-constructed photolithography hot spot detection model for detection, the detection process of the model fully considers the typical pattern feature (i.e. the geometric feature of the layout of the wiring in the photolithography layout) contained in the photolithography layout, overcomes the problem that the existing detection based on the features such as the layout density is easy to lose other pattern features of the photolithography layout, resulting in an inaccurate detection result, and thus the accuracy of the detection result can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor technology, and in particular to a lithography hotspot detection method and device, a storage medium and equipment. BACKGROUND

[0002] With the continuous reduction of semiconductor process size, the special layout pattern of the key layer in the process manufacturing is very sensitive to the change of the lithography process window or the optical diffraction, although various optical resolution enhancement techniques are introduced to correct the original layout to obtain more accurate imaging quality, but the effectiveness of these techniques still strongly depends on the compatibility of the layout pattern to the lithography process. Therefore, in the design stage, the hotspots in the layout pattern which are difficult to implement by the process and are easy to introduce circuit failure are detected and corrected in advance, which has become an effective means to ensure the consistency of design and manufacturing in advanced processes.

[0003] At present, there are usually two methods for lithography hotspot (HS) detection: one is to use machine learning to identify the lithography hotspot, which usually cannot directly use the layout as the input of the model, and needs to extract the feature quantity of the layout, the commonly used feature quantities include pattern density, critical dimension area, etc., and then based on these feature quantities, image classification and recognition are performed, but since the feature quantity extraction is easy to lose other key features of the pattern, and the detection accuracy and recall rate of the machine learning model are highly dependent on the feature quantity extraction result of the layout, therefore, the false positive rate of the model is high, and the recall rate and accuracy are relatively low; and the other commonly used detection method is to use deep learning for detection, which can improve the detection accuracy without feature quantity extraction, but this method simplifies the hotspot recognition to simple image recognition when identifying the lithography hotspot, and does not consider the characteristics of the lithography process, and in the model training process, the overall translation algorithm is used for data enhancement, which will cause a small amount of blank area in the sample image, since only the hotspot data is usually enhanced, the increased blank area is inconsistent with the actual layout pattern, which will cause the model to read in false features when reading the pattern features, resulting in inaccurate detection results of the model, and the false positive rate is high and the recall rate is low. SUMMARY

[0004] The embodiments of the present application provide a lithography hotspot detection method, device, storage medium and equipment, which can overcome the shortcomings of the existing lithography hotspot detection method, make the detection process fully consider the typical pattern features of the lithography layout, and further improve the accuracy of the detection result.

[0005] In a first aspect, the present application provides a photolithography hotspot detection method, which comprises: first acquiring a photolithography layout to be detected; then extracting a typical pattern feature representing the pattern information of the photolithography layout to be detected (i.e. the geometric features of the layout of the traces in the photolithography layout, such as the diagonal spacing between the ends of the traces, etc.) from the photolithography layout to be detected; and then inputting the typical pattern feature into a pre-constructed photolithography hotspot detection model to obtain the photolithography hotspot in the photolithography layout to be detected, wherein the photolithography hotspot detection model comprises a convolution layer, a fully connected layer and an output layer.

[0006] Compared with the prior art, in the embodiments of the present application, the extracted typical pattern feature representing the pattern information of the photolithography layout to be detected is input into the pre-constructed photolithography hotspot detection model comprising the convolution layer, the fully connected layer and the output layer for detection, so that the detection process of the model fully considers the typical pattern features (i.e. the geometric features of the layout of the traces in the photolithography layout) contained in the photolithography layout, thereby overcoming the problem that the detection result is not accurate enough due to the loss of other pattern features of the photolithography layout when the existing detection is based on the features such as the layout density, and further effectively improving the accuracy of the detection result.

[0007] In a possible implementation, the typical pattern feature comprises one or more of the following features in the photolithography layout: the area of the minimum spacing region between lines; the area of the minimum spacing region between a line end and a line; the area of the minimum spacing region between a line end and a line end; the diagonal spacing between line ends; the number of T-shaped trace anchor points; and the number of U-shaped trace anchor points.

[0008] In a possible implementation, the pre-constructed photolithography hotspot detection model comprises N different detection models, where N is a positive integer greater than or equal to 2; and inputting the typical pattern feature into the pre-constructed photolithography hotspot detection model to obtain the photolithography hotspot in the photolithography layout comprises: inputting the typical pattern feature into the N different detection models respectively to obtain N detection results; and inputting the N detection results into a preset fully connected layer to determine the photolithography hotspot in the photolithography layout. In this way, a detection result with higher accuracy and recall rate can be obtained.

[0009] In a possible implementation, the photolithography hotspot detection model is constructed in the following manner: acquiring a sample photolithography layout; and training a pre-constructed initial photolithography hotspot detection model using the sample photolithography layout to obtain the photolithography hotspot detection model.

[0010] In one possible implementation, a pre-built initial lithographic hotspot detection model is trained using a sample lithographic pattern to obtain a lithographic hotspot detection model. This includes: performing data augmentation on the sample lithographic pattern to obtain an enhanced sample lithographic pattern; extracting typical pattern features from the enhanced sample lithographic pattern; inputting the enhanced sample lithographic pattern into the pre-built initial lithographic hotspot detection model, and inputting the typical pattern features of the enhanced sample lithographic pattern into the initial lithographic hotspot detection model for training, thereby generating a lithographic hotspot detection model. This further improves the training accuracy and detection accuracy of the model.

[0011] In one possible implementation, the method further includes: acquiring a verification lithography pattern; extracting typical pattern features from the verification lithography pattern; inputting the typical pattern features of the verification lithography pattern into a lithography hotspot detection model to obtain the detection result of the verification lithography pattern; when the lithography hotspot detection result of the verification lithography pattern is inconsistent with the lithography hotspot marking result corresponding to the verification lithography pattern, the verification lithography pattern is used again as a sample lithography pattern to update the parameters of the lithography hotspot detection model. In this way, the verification lithography pattern can be used to effectively verify the lithography hotspot detection model and to adjust and update the model in a timely manner, thereby helping to improve the detection accuracy and precision of the detection model.

[0012] In one possible implementation, the initial lithographic hotspot detection model is a convolutional neural network, which includes M convolutional layers, fully connected layers, and an output layer; M is a positive integer greater than or equal to 2.

[0013] Secondly, this application also provides a lithography hotspot detection device, which includes: a first acquisition unit for acquiring a lithography pattern to be detected; a first extraction unit for extracting typical pattern features from the lithography pattern; and a first obtaining unit for inputting the typical pattern features into a pre-constructed lithography hotspot detection model to obtain lithography hotspots in the lithography pattern; wherein the lithography hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer.

[0014] In one possible implementation, typical pattern features include one or more of the following features in the photolithographic layout: area of ​​the minimum spacing region between lines; area of ​​the minimum spacing region between line ends; area of ​​the minimum spacing region between line ends; diagonal spacing between line ends; number of T-shaped trace anchor points; number of U-shaped trace anchor points.

[0015] In one possible implementation, the pre-constructed lithography hotspot detection model includes N different detection models, where N is a positive integer greater than or equal to 2; the first obtaining unit includes: an obtaining subunit, used to input typical pattern features into the N different detection models respectively, and predict N detection results; and a determining subunit, used to input the N detection results into a preset fully connected layer, and determine the lithography hotspots in the lithography pattern.

[0016] In one possible implementation, the device further includes: a second acquisition unit for acquiring a sample photolithography pattern; and a training unit for training a pre-constructed initial photolithography hotspot detection model using the sample photolithography pattern to obtain a photolithography hotspot detection model.

[0017] In one possible implementation, the training unit includes: an enhancement subunit for performing data augmentation processing on the sample lithography pattern to obtain an enhanced sample lithography pattern; an extraction subunit for extracting typical pattern features of the enhanced sample lithography pattern from the enhanced sample lithography pattern; and a training subunit for inputting the enhanced sample lithography pattern into a pre-built initial lithography hotspot detection model and inputting the typical pattern features of the enhanced sample lithography pattern into the initial lithography hotspot detection model for training to generate a lithography hotspot detection model.

[0018] In one possible implementation, the device further includes: a third acquisition unit for acquiring a verification lithography pattern; a second extraction unit for extracting typical pattern features of the verification lithography pattern from the verification lithography pattern; a second obtaining unit for inputting the typical pattern features of the verification lithography pattern into a lithography hotspot detection model to obtain the detection result of the verification lithography pattern; and an update unit for updating the parameters of the lithography hotspot detection model by reusing the verification lithography pattern as a sample lithography pattern when the lithography hotspot detection result of the verification lithography pattern is inconsistent with the lithography hotspot marking result corresponding to the verification lithography pattern.

[0019] In one possible implementation, the initial lithographic hotspot detection model is a convolutional neural network, which includes M convolutional layers, fully connected layers, and an output layer; M is a positive integer greater than or equal to 2.

[0020] Thirdly, this application also provides a photolithography hotspot detection device, which includes: a memory and a processor;

[0021] The memory is used to store instructions; the processor is used to execute the instructions in the memory, performing the methods described in the first aspect above and any of its possible implementations.

[0022] Fourthly, this application also provides a computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the methods described in the first aspect and any possible implementation thereof.

[0023] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0024] In this embodiment of the application, when performing lithographic hotspot detection, the lithographic layout to be detected is first acquired. Then, typical pattern features characterizing its pattern information are extracted from the lithographic layout. Next, these typical pattern features are input into a pre-constructed lithographic hotspot detection model to obtain the lithographic hotspots in the lithographic layout. The lithographic hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. It can be seen that because this embodiment of the application inputs the extracted typical pattern features characterizing the pattern information of the lithographic layout to be detected into a pre-constructed lithographic hotspot detection model including a convolutional layer, a fully connected layer, and an output layer, the detection process of the model fully considers the typical pattern features contained in the lithographic layout (i.e., the geometric features of the trace layout in the lithographic layout, such as the diagonal spacing between trace ends), thus overcoming the problem that existing detection methods based on features such as layout density easily lose other pattern features of the lithographic layout, resulting in inaccurate detection results. Therefore, the accuracy of the detection results can be effectively improved. Attached Figure Description

[0025] Figure 1 A schematic diagram of the main framework of artificial intelligence provided in the embodiments of this application;

[0026] Figure 2 This is a schematic diagram illustrating an application scenario of an embodiment of this application;

[0027] Figure 3 A flowchart of a photolithographic hotspot detection method provided in an embodiment of this application;

[0028] Figure 4 A schematic diagram of the pattern features of the photolithographic pattern provided in the embodiments of this application;

[0029] Figure 5 One of the schematic diagrams provided in this application illustrates inputting pattern features into a fully connected layer of a pre-built lithographic hotspot detection model to detect hotspots in a lithographic pattern;

[0030] Figure 6 This is a second schematic diagram illustrating how pattern features are input into a pre-built fully connected layer of a lithographic hotspot detection model to detect hotspots in a lithographic pattern, as provided in an embodiment of this application.

[0031] Figure 7A schematic diagram illustrating data enhancement processing of a sample photolithography pattern provided in an embodiment of this application;

[0032] Figure 8 This is a schematic diagram illustrating the input of the enhanced sample lithography pattern into the initial lithography hotspot detection model, as provided in an embodiment of this application.

[0033] Figure 9 A structural block diagram of a photolithography hotspot detection device provided in an embodiment of this application;

[0034] Figure 10 This is a schematic diagram of the structure of a photolithography hotspot detection device provided in an embodiment of this application. Detailed Implementation

[0035] This application provides a method, apparatus, storage medium, and device for detecting photolithographic hotspots, which fully considers the pattern characteristics of the photolithographic pattern during the detection process, thereby further improving the accuracy of the detection results.

[0036] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0037] First, the overall workflow of the artificial intelligence system is described; please refer to [link / reference]. Figure 1 , Figure 1 The diagram illustrates a structural framework for artificial intelligence (AI). The framework is further elaborated below along two dimensions: the "Intelligent Information Chain" (horizontal axis) and the "IT Value Chain" (vertical axis). The "Intelligent Information Chain" reflects a series of processes from data acquisition to processing. For example, it could be the general process of intelligent information perception, intelligent information representation and formation, intelligent reasoning, intelligent decision-making, and intelligent execution and output. In this process, data undergoes a condensation process of "data—information—knowledge—wisdom." The "IT Value Chain" reflects the value that AI brings to the information technology industry, from the underlying infrastructure of human intelligence and information (provided and processed through technological means) to the industrial ecosystem of the system.

[0038] (1) Infrastructure

[0039] Infrastructure provides computing power to support artificial intelligence systems, enabling communication with the external world and providing support through a basic platform. This communication occurs through sensors; computing power is provided by intelligent chips (hardware acceleration chips such as CPUs, NPUs, GPUs, ASICs, and FPGAs); and the basic platform includes distributed computing frameworks and related platform guarantees and support, which may include cloud storage and computing, interconnected networks, etc. For example, sensors communicate with the outside world to acquire data, and this data is provided to intelligent chips in the distributed computing system provided by the basic platform for computation.

[0040] (2) Data

[0041] The data at the next layer of infrastructure is used to represent the data sources in the field of artificial intelligence. The data involves graphics, images, voice, text, and IoT data from traditional devices, including business data from existing systems and sensor data such as force, displacement, liquid level, temperature, and humidity.

[0042] (3) Data processing

[0043] Data processing typically includes methods such as data training, machine learning, deep learning, search, reasoning, and decision-making.

[0044] Among them, machine learning and deep learning can perform intelligent information modeling, extraction, preprocessing, and training of data by symbolizing and formalizing it.

[0045] Reasoning refers to the process in which, in a computer or intelligent system, the machine thinks and solves problems by simulating human intelligent reasoning, based on reasoning control strategies and using formalized information. Typical functions include search and matching.

[0046] Decision-making refers to the process of making decisions based on intelligent information after reasoning, and it typically provides functions such as classification, sorting, and prediction.

[0047] (4) General ability

[0048] After the data processing mentioned above, the results of the data processing can be used to form some general capabilities, such as algorithms or a general system, for example, translation, text analysis, computer vision processing, speech recognition, image recognition, etc.

[0049] The embodiments of this application can be applied to the field of information technology. The following description will take the anomaly detection of storage devices as an example. The embodiments of this invention can also be applied to other devices, such as servers, networks and other devices. The embodiments of this invention are not limited to these.

[0050] The process for detecting photolithographic hotspots in computer equipment is as follows:

[0051] The lithographic hotspot detection method provided in this application can be applied to the lithographic hotspot detection process in computer equipment. See also Figure 2 , Figure 2 This is a schematic diagram illustrating an application scenario of an embodiment of this application, such as... Figure 2 As shown, the computer device 201 includes an AI system for detecting photolithographic hotspots, used to acquire a photolithographic pattern to be detected. This pattern may include a photolithographic design used in semiconductor manufacturing to etch a film layer of a specific shape. The photolithographic pattern is highly sensitive to the effects of the photolithography process window or optical diffraction, thus significantly impacting the imaging quality of the photolithography. The photolithography process window, also known as the photolithography process tolerance, specifically refers to the range of exposure dose and defocus amount that ensures the mask pattern can be correctly copied onto the silicon wafer. Different photolithography patterns may have different photolithography process windows. After extracting the typical pattern features from the obtained lithography pattern to be inspected, these features can be input into the fully connected layer of the AI ​​detection system model to further detect the typical pattern features. The output results determine the lithography hotspot detection results in the lithography pattern to be inspected. This ensures that the entire detection process fully considers the typical pattern features of the lithography pattern, overcoming the problem of inaccurate detection results caused by the loss of other pattern features when detecting based on features such as pattern density. This effectively improves the accuracy of the detection results. Thus, after the lithography pattern to be inspected is designed, before etching it into a film layer of a specific shape in semiconductor manufacturing, the AI ​​system in computer device 201, which has the function of detecting lithography hotspots, can be used to perform lithography hotspot detection on the lithography pattern to be inspected using the aforementioned detection method. Then, based on the detection results, the detected lithography hotspots can be corrected in advance to avoid problems and significant economic losses when using the lithography pattern to form a film layer of a specific shape later.

[0052] As an example, computer device 201 can be any device capable of analyzing photolithographic patterns and detecting photolithographic hotspots, including but not limited to: smartphones, feature phones, tablets, laptops, desktop computers, minicomputers, mid-range computers, mainframe computers, etc. It should be understood that the embodiments of this application can also be applied to other scenarios requiring photolithographic hotspot detection; other application scenarios will not be listed here.

[0053] Based on the above application scenarios, this application provides a method for detecting photolithographic hotspots, which will be described below.

[0054] S301: Obtain the lithographic pattern to be inspected.

[0055] It should be noted that as semiconductor process dimensions continue to shrink—that is, as the number of circuit elements per unit area increases and the size of these miniaturized elements increases—the patterns of circuit elements fabricated on the substrate become smaller and closer together. This reduction in feature size increases the difficulty of fabricating the desired layout on the substrate. This is partly because light diffraction during photolithography creates defects, preventing the desired image from being accurately projected onto the substrate, thus resulting in flaws in the final device structure.

[0056] Therefore, to avoid such defects, after the photolithography pattern is designed, before etching it into a film layer of a specific shape in semiconductor manufacturing, it is necessary to accurately detect hot spots in the photolithography pattern that are difficult to process and prone to introducing circuit failures. Then, based on the detection results, the detected photolithography hot spots are corrected in advance to avoid problems and significant economic losses when using the photolithography pattern to etch a film layer of a specific shape later, thereby ensuring design and manufacturing consistency. Therefore, in this embodiment of the application, after obtaining the photolithography pattern to be detected, the photolithography hot spots in the photolithography pattern can be detected through subsequent steps S302-S303.

[0057] S302: Extract typical pattern features from the photolithography pattern.

[0058] In this embodiment, after obtaining the photolithography pattern to be detected in step S301, feature extraction methods can be used to extract its features, and typical pattern features that can characterize its pattern information can be extracted for subsequent step S303. The typical pattern features of the photolithography pattern refer to the geometric features of the trace layout in the photolithography pattern, such as the diagonal spacing between the ends of the traces.

[0059] In one optional implementation of this embodiment, the extracted typical pattern features may include one or more of the following features in the photolithography pattern:

[0060] Area of ​​the minimum spacing between lines; area of ​​the minimum spacing between line ends; area of ​​the minimum spacing between line ends; diagonal spacing between line ends; number of T-shaped wiring anchor points; number of U-shaped wiring anchor points.

[0061] Specifically, such as Figure 4 As shown, where, Figure 4 In diagram 'a', the shaded area between the two rectangles represents the area of ​​the minimum line-to-line spacing region in the lithographic print. Figure 4In b, the shaded area between the two rectangles represents the area of ​​the minimum line-end to line spacing region in the photolithography pattern. Figure 4 In c, the shaded area between the two rectangles represents the area of ​​the minimum line-end to line-end spacing region in the photolithography pattern. Figure 4 The arrow pointing between the two rectangles in 'd' indicates the line-end to line-end corner spacing in the photolithography pattern. Figure 4 The two bold black dots in 'e' represent the number of T-shape anchor points in the photolithography layout. Figure 4 The four bold black dots in f represent the number of U-shaped anchor points in the photolithography pattern.

[0062] S303: Input typical pattern features into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout. The lithography hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer.

[0063] In this embodiment, after extracting the typical pattern features of the lithography pattern to be detected in step S302, it can be input into a pre-constructed lithography hotspot detection model that includes convolutional layers, fully connected layers, and an output layer. The model is then merged with each array vector array obtained in the fully connected layer after being processed by multiple convolutional layers in the layout array corresponding to the lithography pattern. The merged array is then used as the output result of the fully connected layer. Based on this output result, the detection result of the lithography hotspot in the lithography pattern (such as the probability that the detection position is a lithography hotspot) can be output through the output layer of the model.

[0064] For example: Figure 5 As shown, to improve the detection accuracy of the model, after extracting the typical pattern features of the lithographic layout to be detected, preprocessing operations (such as normalization) can be performed on these typical pattern feature data according to the actual process nodes and layers of the layout, to obtain the preprocessed typical pattern feature vector (vector dimension greater than or equal to 1 and less than or equal to 6, e.g., ...). Figure 5 The values ​​shown are between 1 and 6 (e.g., 4×1 dimension), and then they are passed through multiple convolutional layers in the layout array corresponding to the photolithography pattern (e.g., ...). Figure 5 Each array vector (with dimensions such as the 3 convolutional layers in the middle) is obtained after processing. Figure 5The 1024×1 dimension vectors are input together into the fully connected layer to form a new feature vector (vector dimension greater than or equal to 1025 and less than or equal to 1030, such as...). Figure 5 The array is defined as 1025 to 1030 (e.g., 1028×1 dimension), and the output layer outputs whether the corresponding layout location is a lithographic hotspot. For example, a probability value P can be output as the detection result. The value of P ranges from [0,1], where 1 indicates that the layout location is a hotspot, 0 indicates that the layout location is not a hotspot, 0.9 indicates that the layout location is 90% likely to be a hotspot, and so on. A lithographic hotspot refers to a specific layout pattern that, after the lithographic layout design is completed, is difficult to implement in terms of process, easily leading to circuit failure.

[0065] In one optional implementation of this embodiment, to improve the detection accuracy and recall of the model, N different lithographic hotspot detection models can be pre-trained by setting different hyperparameters, where N is a positive integer greater than or equal to 2, and recall refers to the proportion of all "lithographic patterns with accurately detected lithographic hotspots" to all "lithographic patterns that should have been detected with lithographic hotspots". Furthermore, different models correspond to different attributes of the detection results, such as accuracy, false positive rate, and recall. Some models have high accuracy and a high false positive rate, while others have low accuracy and a low false positive rate, and still others have low accuracy but a high F-score (representing the harmonic value of accuracy and recall). Therefore, to obtain detection results with higher accuracy and recall, fully connected layers can be used to further process the detection results of different models to obtain the final detection result. Specifically, for example... Figure 6 As shown, the execution process of step S303 may include the following steps A1-A2:

[0066] Step A1: Input the typical pattern features into N different detection models to predict N detection results.

[0067] In this implementation, after extracting the typical pattern features of the lithography pattern to be detected in step S302, these typical pattern features can be input into N pre-constructed lithography hotspot detection models with different attributes to obtain N detection results. Among them, some models have the highest accuracy, some have the best recall, and some have the highest F-score, etc.

[0068] For example: Figure 6As shown, N is set to 3, and the detection results of the three pre-constructed lithography hotspot detection models have the best recall, best F-score, and best precision, respectively. After inputting the typical pattern features of the lithography pattern to be detected into these three lithography hotspot detection models with different attributes, the three detection results obtained are P1, P2, and P3, respectively. It can be seen that P1 corresponds to the best recall, P2 corresponds to the best F-score, and P3 corresponds to the best precision.

[0069] Step A2: Input the N detection results into the preset fully connected layer to determine the lithographic hotspots in the lithography pattern.

[0070] In this implementation, after obtaining three detection results through step A1, these N detection results can be further input into a pre-set fully connected layer for comprehensive processing, and the processing results can be used to determine whether the layout position corresponding to each array in the layout array corresponding to the photolithography layout is a hot spot.

[0071] For example: Based on the above examples, such as Figure 6 As shown, after obtaining three detection results P1, P2, and P3, these three detection results can be input into a pre-set fully connected layer for comprehensive processing. Then, based on the processing results, it can be determined whether the corresponding layout position is a photolithography hot spot.

[0072] It should be noted that one optional implementation is to pre-train a large amount of training pattern data and obtain various detection models with different attributes by setting different hyperparameters (e.g., some models have high accuracy and high false positive rate, while others have low accuracy and low false positive rate). Then, after obtaining the detection results of these different models, the output results of these models can be retrained using pre-defined fully connected layers to obtain a final model with high accuracy and recall. Subsequently, this single final model can be used to perform hotspot detection on the lithography pattern. The specific implementation process can be found in the above description and will not be repeated here.

[0073] In summary, the lithography hotspot detection method provided in this embodiment first acquires the lithography pattern to be detected. Then, it extracts typical pattern features representing the pattern information from the lithography pattern to be detected. Next, it inputs these typical pattern features into a pre-constructed lithography hotspot detection model to obtain the lithography hotspots in the lithography pattern to be detected. The lithography hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. It can be seen that because this embodiment inputs the extracted typical pattern features representing the pattern information of the lithography pattern to be detected into a pre-constructed lithography hotspot detection model including a convolutional layer, a fully connected layer, and an output layer, the detection process fully considers the typical pattern features contained in the lithography pattern (i.e., the geometric features of the trace layout in the lithography pattern, such as the diagonal spacing between trace ends) during detection. This overcomes the problem that existing detection methods based on features such as layout density easily lose other pattern features of the lithography pattern, resulting in inaccurate detection results. Therefore, it can effectively improve the accuracy of the detection results.

[0074] Next, this embodiment will describe the construction process of the photolithography hotspot detection model, which may specifically include the following steps B1-B2:

[0075] Step B1: Obtain the sample photolithography pattern.

[0076] In this embodiment, in order to build the lithography hotspot detection model, a lot of preparatory work is required. First, a large amount of sample lithography pattern data containing hotspots and non-hotspots needs to be collected. For example, 100 lithography patterns can be collected in advance, and each collected lithography pattern can be used as a sample lithography pattern to train the lithography hotspot detection model.

[0077] It should be noted that, in order to reduce the machine resource consumption and running time of computer equipment, after obtaining the sample photolithography layout, the image size of these sample photolithography layout data can first be normalized. Each sample photolithography layout is evenly divided into multiple small windows, and the size of each small window should be larger than the influence radius of the optical wavelength of the current photolithography process to ensure coverage of the entire photolithography influence range. The sample photolithography layout is then converted into an array according to a certain pixel ratio. This allows the scaling ratio to effectively reduce the machine resource consumption and running time of computer equipment without affecting the layout accuracy.

[0078] Step B2: Using the sample lithography pattern, train the pre-constructed initial lithography hotspot detection model to obtain the lithography hotspot detection model.

[0079] In this embodiment, an initial lithographic hotspot detection model can be pre-constructed and its parameters initialized. One optional implementation is that the initial lithographic hotspot detection model can be a convolutional neural network (CNN), comprising M convolutional layers, fully connected layers, and an output layer, where M is a positive integer greater than or equal to 2. The initial lithographic hotspot detection model is used to obtain the lithographic hotspot detection model based on typical pattern features in the sample lithographic pattern and by performing pattern density encoding and other processing on the sample lithographic pattern. The specific implementation process may include the following steps B21-B23:

[0080] Step B21: Perform data augmentation processing on the sample photolithography pattern to obtain the enhanced sample photolithography pattern.

[0081] In practical applications, the hot data in a lithography pattern is often far less than the non-hot data. Therefore, if the sample lithography pattern is directly used as the training dataset after obtaining it, the extreme imbalance between the distribution of hot and non-hot data will lead to insufficient accuracy of the training model. Therefore, it is necessary to perform data augmentation on the obtained sample lithography pattern to obtain an augmented sample lithography pattern, which expands the amount of hot data in the sample lithography pattern to increase the generalization ability of the model and improve the accuracy of the model.

[0082] Specifically, this embodiment expands the data volume of the sample photolithography pattern by translating, flipping, and rotating upwards, downwards, leftwards, and rightwards in steps of a certain size. For example, as... Figure 7 As shown, translation, vertical flip, horizontal flip, and 180-degree rotation can be used respectively (90-degree / 180-degree / 270-degree rotation can be used for layers that support 2D direction) to obtain a larger number of enhanced sample photolithography patterns for subsequent step B22.

[0083] Step B22: Extract typical pattern features from the enhanced sample lithography pattern.

[0084] After obtaining the enhanced sample lithography pattern through step B21, typical pattern features characterizing the pattern information of each enhanced sample lithography pattern can be extracted from each enhanced sample lithography pattern. Specifically, a method similar to extracting typical pattern features of the lithography pattern from the lithography pattern to be detected in step S302 above can be used. The lithography pattern to be detected is replaced with the enhanced sample lithography pattern, and typical pattern features characterizing the pattern information of each enhanced sample lithography pattern can be extracted from each enhanced sample lithography pattern (i.e., the geometric features of the trace layout in the enhanced sample lithography pattern, including one or more of the following features in the enhanced sample lithography pattern: area of ​​the minimum spacing region between lines; area of ​​the minimum spacing region between line ends; area of ​​the minimum spacing region between line ends; diagonal spacing between line ends; number of T-shaped trace anchor points; number of U-shaped trace anchor points). For relevant details, please refer to the description of step S302 above, which will not be repeated here.

[0085] Step B23: Input the enhanced sample lithography pattern into the pre-built initial lithography hotspot detection model, and input the typical pattern features of the enhanced sample lithography pattern into the fully connected layer of the initial lithography hotspot detection model for training to generate the lithography hotspot detection model.

[0086] After obtaining the enhanced sample lithography pattern through step B21, the corresponding encoding arrays of the enhanced sample lithography pattern can be input one by one into the pre-constructed initial lithography hotspot detection model, such as... Figure 8 As shown, the initial lithographic hotspot detection model includes multiple convolutional layers (such as...). Figure 8 The model consists of three layers: a fully connected layer, a production layer, and an output layer. This embodiment uses a linear correction unit as the activation function for the convolutional layer. In practical applications, other functions (such as the hyperbolic tangent function) can be used for training, depending on the specific circumstances. To reduce overfitting and improve model accuracy, regularization, batch normalization, and other algorithms can also be added to the model.

[0087] It should be noted that, in order to preserve the complete information of the lithographic pattern as much as possible, the initial lithographic hotspot detection model constructed in this embodiment does not use pooling layers to reduce feature dimensions. In practical applications, pooling layers can be appropriately added to compress data and parameters to improve the training efficiency of the model.

[0088] Furthermore, this embodiment uses a sigmoid function as the output layer activation function for classification (distinguishing between hotspots and non-hotspots). In practical applications, a normalized exponential function (softmax) can also be used as the classification output layer function. Simultaneously, loss functions such as mean squared error and cross-entropy, along with gradient descent or adaptive learning rate algorithms and their variations, can be used as optimizers for model training. In this case, focal loss is used as the loss function, and different class weights are set to focus training on map data containing hotspots and imbalanced hotspot data, thereby reducing the impact of dataset imbalance. Thus, during training, by setting different hyperparameters, at least one well-trained initial lithographic hotspot detection model can be obtained, or N different well-trained initial lithographic hotspot detection models can be obtained, where N is a positive integer greater than or equal to 2. Different models correspond to different attributes of the detection results, such as accuracy, false positive rate, and recall. Some models have high accuracy and a high false positive rate, while others have low accuracy and a low false positive rate, and still others have low accuracy but a high F-score, etc.

[0089] After extracting the pattern features of the enhanced sample lithography pattern in step B22, the typical pattern features of the sample lithography pattern and each encoded array vector obtained after processing by the multi-layer convolutional layers of the initial lithography hotspot detection model in the corresponding pattern array of the sample lithography pattern can be jointly input into the trained initial lithography hotspot detection model for training. The model outputs a probability value (e.g., a probability value in the range [0,1]) indicating whether the pattern location corresponding to the array is a hotspot. Then, this probability value can be compared with the corresponding real detection result (e.g., 1 indicates that the pattern location is a hotspot, and 0 indicates that the pattern location is not a hotspot), and the model parameters can be updated according to the difference between the two until a preset condition is met, such as a very small change in the difference. Then, the update of the model parameters is stopped, the training of the lithography hotspot detection model is completed, and a trained lithography hotspot detection model is generated.

[0090] It should be noted that, as an alternative implementation, typical pattern features of typical lithography patterns (i.e., at least one of the following: minimum spacing area between lines; minimum spacing area between line ends; minimum spacing area between line ends; diagonal spacing between line ends; number of T-shaped trace anchor points; number of U-shaped trace anchor points) can be extracted and input into the initial lithography hotspot detection model. Each encoded array vector obtained in the fully connected layer of the layout array corresponding to the sample lithography pattern, after being processed by multiple convolutional layers of the initial lithography hotspot detection model, is then used for training to generate a trained lithography hotspot detection model. The specific implementation process will not be elaborated here.

[0091] Through the above embodiments, a lithographic hotspot detection model can be trained and generated using a sample lithographic pattern. Furthermore, the generated lithographic hotspot detection model can be validated using a validation lithographic pattern. The specific validation process may include the following steps C1-C4:

[0092] Step C1: Obtain the verification photolithography pattern.

[0093] In this embodiment, in order to verify the lithography hotspot detection model, it is first necessary to obtain a large amount of verification lithography layout data. The verification lithography layout refers to the lithography layout that can be used to verify the lithography hotspot detection model. After obtaining the verification lithography layout, the subsequent step C2 can be executed.

[0094] Step C2: Extract typical pattern features from the verification lithography pattern.

[0095] After obtaining the verification lithography layout in step C1, it cannot be directly used to verify the lithography hotspot detection model. Instead, it is necessary to first extract the typical pattern features that characterize the pattern information of the verification lithography layout (i.e., the geometric features of the trace layout in the verification lithography layout, including at least one of the following: the area of ​​the minimum spacing region between lines; the area of ​​the minimum spacing region between line ends; the area of ​​the minimum spacing region between line ends; the diagonal spacing between line ends; the number of T-shaped trace anchor points; and the number of U-shaped trace anchor points). Then, the extracted typical pattern features of the verification lithography layout can be used to verify the lithography hotspot detection model.

[0096] Step C3: Input the typical pattern features of the verification lithography pattern into the lithography hotspot detection model to obtain the detection results of the verification lithography pattern.

[0097] After extracting the typical pattern features of the verification lithography pattern in step C2, the typical pattern features of the verification lithography pattern can be further input into the lithography hotspot detection model to obtain the detection results of the verification lithography pattern, and then the subsequent step C4 can be executed.

[0098] Step C4: When the lithographic hotspot detection results of the verification lithography pattern are inconsistent with the lithographic hotspot marking results corresponding to the verification lithography pattern, the verification lithography pattern is used again as the sample lithography pattern to update the parameters of the lithography hotspot detection model.

[0099] After obtaining the detection results of the verification lithography pattern through step C3, if the detection results of the lithography hotspots of the verification lithography pattern are inconsistent with the lithography hotspot marking results (i.e., the actual detection results) corresponding to the verification lithography pattern, the verification lithography pattern can be used again as the sample lithography pattern to update the parameters of the lithography hotspot detection model.

[0100] Through the above embodiments, the lithography hotspot detection model can be effectively verified using the verification lithography pattern. When the lithography hotspot detection results of the verification lithography pattern are inconsistent with the actual lithography hotspot detection results corresponding to the verification lithography pattern, the lithography hotspot detection model can be adjusted and updated in a timely manner, thereby helping to improve the detection accuracy and precision of the detection model.

[0101] In summary, the lithography hotspot detection model trained using this embodiment can quickly and accurately detect the hotspot locations of the lithography pattern using typical pattern features of the pattern to be detected, effectively improving the efficiency and accuracy of hotspot detection of the lithography pattern to be detected.

[0102] To facilitate better implementation of the above-described solutions in the embodiments of this application, related apparatus for implementing the above solutions is also provided below. Please refer to... Figure 9 As shown in the figure, this application embodiment provides a photolithography hotspot detection device 900. The device 900 may include: a first acquisition unit 901, a first extraction unit 902, and a first obtaining unit 902. The first acquisition unit 901 is used to support the device 900 in performing... Figure 3 S301 in the illustrated embodiment. The first extraction unit 902 is used to support the device 900 in performing... Figure 3 S302 in the illustrated embodiment. The first obtaining unit 903 is used to support the device 900 in performing... Figure 3 S303 in the illustrated embodiment. Specifically,

[0103] The first acquisition unit 901 is used to acquire the photolithography pattern to be detected;

[0104] The first extraction unit 902 is used to extract typical pattern features of the photolithography pattern from the photolithography pattern;

[0105] The first acquisition unit 903 is used to input typical pattern features into a pre-constructed lithography hotspot detection model to obtain lithography hotspots in the lithography pattern; wherein, the lithography hotspot detection model includes a convolutional layer, a fully connected layer and an output layer.

[0106] In one implementation of this embodiment, typical pattern features include one or more of the following features in the photolithographic pattern:

[0107] The area of ​​the region with the minimum spacing between lines;

[0108] The area of ​​the region with the minimum spacing between the ends of the lines;

[0109] The area of ​​the region with the minimum distance between the ends of a line;

[0110] The diagonal distance between the ends of the lines;

[0111] Number of T-shaped cable anchor points;

[0112] Number of U-shaped cable anchor points.

[0113] In one implementation of this embodiment, the pre-constructed lithography hotspot detection model includes N different detection models, where N is a positive integer greater than or equal to 2; the first obtaining unit 903 includes:

[0114] A sub-unit is obtained, which is used to input typical pattern features into N different detection models to predict N detection results;

[0115] The sub-unit is determined to input N detection results into a preset fully connected layer to determine the lithographic hotspots in the lithography pattern.

[0116] In one implementation of this embodiment, the device further includes:

[0117] The second acquisition unit is used to acquire the sample photolithography pattern;

[0118] The training unit is used to train the pre-built initial lithographic hotspot detection model using sample lithographic patterns to obtain the lithographic hotspot detection model.

[0119] In one implementation of this embodiment, the training unit includes:

[0120] The enhancement subunit is used to perform data enhancement processing on the sample photolithography pattern to obtain the enhanced sample photolithography pattern.

[0121] Extraction sub-units are used to extract typical pattern features from the enhanced sample lithography pattern;

[0122] The training subunit is used to input the enhanced sample lithography pattern into the pre-built initial lithography hotspot detection model, and to input the typical pattern features of the enhanced sample lithography pattern into the initial lithography hotspot detection model for training, thereby generating the lithography hotspot detection model.

[0123] In one implementation of this embodiment, the device further includes:

[0124] The third acquisition unit is used to acquire and verify the photolithography pattern;

[0125] The second extraction unit is used to extract typical pattern features from the verification photolithography pattern.

[0126] The second acquisition unit is used to input the typical pattern features of the verification lithography pattern into the lithography hotspot detection model to obtain the detection result of the verification lithography pattern.

[0127] The update unit is used to update the parameters of the lithography hotspot detection model when the lithography hotspot detection result of the verification lithography pattern is inconsistent with the lithography hotspot marking result corresponding to the verification lithography pattern.

[0128] In one implementation of this embodiment, the initial lithographic hotspot detection model is a convolutional neural network, wherein the convolutional neural network includes M convolutional layers, fully connected layers, and an output layer; M is a positive integer greater than or equal to 2.

[0129] In summary, the lithography hotspot detection device provided in this embodiment first acquires the lithography pattern to be detected during lithography hotspot detection. Then, it extracts typical pattern features representing the pattern information from the lithography pattern to be detected. Next, it inputs these typical pattern features into a pre-constructed lithography hotspot detection model to obtain the lithography hotspots in the lithography pattern to be detected. The lithography hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. It can be seen that because this embodiment inputs the extracted typical pattern features representing the pattern information of the lithography pattern to be detected into a pre-constructed lithography hotspot detection model including a convolutional layer, a fully connected layer, and an output layer for detection, the detection process of the model fully considers the typical pattern features contained in the lithography pattern (i.e., the geometric features of the trace layout in the lithography pattern, such as the diagonal spacing between trace ends), thereby overcoming the problem that existing detection methods based on features such as layout density easily lose other pattern features of the lithography pattern, resulting in inaccurate detection results. This effectively improves the accuracy of the detection results.

[0130] See Figure 10 This application provides a photolithography hotspot detection device 1000, which includes a memory 1001, a processor 1002, and a communication interface 1003.

[0131] Memory 1001 is used to store instructions;

[0132] Processor 1002 is used to execute instructions stored in memory 1001, performing the aforementioned operations applied to... Figure 3 The lithographic hotspot detection method in the illustrated embodiment;

[0133] Communication interface 1003 is used for communication.

[0134] The memory 1001, processor 1002, and communication interface 1003 are interconnected via bus 1004. Bus 1004 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 10 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0135] In a specific embodiment, the processor 1002, when performing lithography hotspot detection, first acquires the lithography pattern to be detected, then extracts typical pattern features representing its pattern information from the lithography pattern to be detected, and then inputs these typical pattern features into a pre-constructed lithography hotspot detection model to obtain the lithography hotspots in the lithography pattern to be detected. The lithography hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. For detailed processing procedures of the processor 1002, please refer to the above. Figure 3 The detailed descriptions of S301, S302, and S303 in the illustrated embodiment will not be repeated here.

[0136] The aforementioned memory 1001 may be random-access memory (RAM), flash memory, read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), register, hard disk, portable hard disk, CD-ROM, or any other form of storage medium known to those skilled in the art.

[0137] The processor 1002 described above may be, for example, a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of embodiments of this application. The processor may also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0138] The aforementioned communication interface 1003 may be, for example, an interface card, and may be an Ethernet interface or an asynchronous transfer mode (ATM) interface.

[0139] This application also provides a computer-readable storage medium including instructions that, when run on a computer, cause the computer to perform the above-described photolithography hotspot detection method.

[0140] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0141] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0142] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical module division, and in actual implementation, there may be other division methods. For example, multiple units 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 an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0143] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be obtained according to actual needs to achieve the purpose of this embodiment.

[0144] Furthermore, the module units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software module unit.

[0145] If the integrated unit is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0146] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of a computer program from one place to another. Storage media can be any available medium accessible to a general-purpose or special-purpose computer.

[0147] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention.

[0148] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0149] Furthermore, the embodiments of this application can also be applied to other future communication technologies. The network architecture and service scenarios described in this application are for the purpose of more clearly illustrating the technical solutions of this application, and do not constitute a limitation on the technical solutions provided in this application. As those skilled in the art will understand, with the evolution of network architecture and the emergence of new service scenarios, the technical solutions provided in this application are also applicable to similar technical problems.

Claims

1. A method for detecting photolithographic hotspots, characterized in that, The method includes: Obtain the lithographic pattern to be inspected; From the photolithography pattern, typical pattern features of the photolithography pattern are extracted using an image feature extraction method. The typical pattern features of the photolithography pattern are the geometric features of the trace layout in the photolithography pattern. The typical pattern features are input into a pre-constructed lithographic hotspot detection model to obtain the lithographic hotspots in the lithographic pattern; the lithographic hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. The step of inputting the typical pattern features into a pre-built lithographic hotspot detection model to obtain the lithographic hotspots in the lithographic pattern includes: The typical pattern features and array vector are input into the fully connected layer to obtain the output of the fully connected layer. The array vector is obtained by processing the layout array corresponding to the photolithography pattern through the convolutional layer. The output of the fully connected layer is input to the output layer to obtain the lithographic hotspots in the lithography pattern.

2. The method according to claim 1, characterized in that, The typical pattern features include one or more of the following features in the photolithography pattern: The area of ​​the region with the minimum spacing between lines; The area of ​​the region with the minimum spacing between the ends of the lines; The area of ​​the region with the minimum distance between the ends of a line; The diagonal distance between the ends of the lines; Number of T-shaped cable anchor points; Number of U-shaped cable anchor points.

3. The method according to claim 1 or 2, characterized in that, The pre-built lithography hotspot detection model includes N different detection models, where N is a positive integer greater than or equal to 2; The step of inputting the typical pattern features into a pre-built lithographic hotspot detection model to obtain the lithographic hotspots in the lithographic pattern includes: The typical pattern features are input into the N different detection models respectively, and N detection results are predicted; The N detection results are input into a preset fully connected layer to determine the lithographic hotspots in the lithography pattern.

4. The method according to claim 1, characterized in that, The lithographic hotspot detection model is constructed as follows: Obtain the sample photolithography pattern; Using the sample photolithography pattern, the pre-constructed initial photolithography hotspot detection model is trained to obtain the photolithography hotspot detection model.

5. The method according to claim 4, characterized in that, The step of training a pre-constructed initial lithographic hotspot detection model using the sample lithographic pattern to obtain the lithographic hotspot detection model includes: The sample photolithography pattern is subjected to data augmentation processing to obtain an enhanced sample photolithography pattern; From the enhanced sample lithography pattern, extract the typical pattern features of the enhanced sample lithography pattern; The enhanced sample lithography pattern is input into a pre-constructed initial lithography hotspot detection model, and the typical pattern features of the enhanced sample lithography pattern are input into the initial lithography hotspot detection model for training, thereby generating the lithography hotspot detection model.

6. The method according to claim 4, characterized in that, The method further includes: Obtain the verification photolithography pattern; Extract typical pattern features from the verification photolithography pattern; The typical pattern features of the verification lithography pattern are input into the lithography hotspot detection model to obtain the detection result of the verification lithography pattern; When the lithographic hotspot detection result of the verification lithographic pattern is inconsistent with the lithographic hotspot marking result corresponding to the verification lithographic pattern, the verification lithographic pattern is used again as the sample lithographic pattern, and the parameters of the lithographic hotspot detection model are updated.

7. The method according to claim 4, characterized in that, The initial lithographic hotspot detection model is a convolutional neural network, which includes M convolutional layers, fully connected layers, and an output layer; where M is a positive integer greater than or equal to 2.

8. A photolithography hotspot detection device, characterized in that, The device includes: The first acquisition unit is used to acquire the photolithography pattern to be inspected; The first extraction unit is used to extract typical pattern features of the photolithography pattern from the photolithography pattern using an image feature extraction method. The typical pattern features of the photolithography pattern are the geometric features of the trace layout in the photolithography pattern. The first obtaining unit is used to input the typical pattern features into a pre-constructed lithographic hotspot detection model to obtain the lithographic hotspots in the lithographic pattern; the lithographic hotspot detection model includes a convolutional layer, a fully connected layer, and an output layer. The first obtaining unit is specifically used to input the typical pattern features and array vector into the fully connected layer to obtain the output result of the fully connected layer. The array vector is obtained by processing the layout array corresponding to the photolithography pattern through the convolutional layer. The output result of the fully connected layer is input into the output layer to obtain the photolithography hotspots in the photolithography pattern.

9. The apparatus according to claim 8, characterized in that, The typical pattern features include one or more of the following features in the photolithography pattern: The area of ​​the region with the minimum spacing between lines; The area of ​​the region with the minimum spacing between the ends of the lines; The area of ​​the region with the minimum distance between the ends of a line; The diagonal distance between the ends of the lines; Number of T-shaped cable anchor points; Number of U-shaped cable anchor points.

10. The apparatus according to claim 8 or 9, characterized in that, The pre-built lithography hotspot detection model includes N different detection models, where N is a positive integer greater than or equal to 2; The first obtaining unit includes: A subunit is obtained, which is used to input the typical pattern features into the N different detection models respectively, and predict N detection results; A sub-unit is defined to input the N detection results into a preset fully connected layer to determine the lithographic hotspots in the lithography pattern.

11. The apparatus according to claim 8, characterized in that, The device further includes: The second acquisition unit is used to acquire the sample photolithography pattern; The training unit is used to train the pre-constructed initial lithographic hotspot detection model using the sample lithographic pattern to obtain the lithographic hotspot detection model.

12. The apparatus according to claim 11, characterized in that, The training unit includes: An enhancement subunit is used to perform data enhancement processing on the sample photolithography pattern to obtain an enhanced sample photolithography pattern. An extraction subunit is used to extract typical pattern features of the enhanced sample photolithography pattern from the enhanced sample photolithography pattern; The training subunit is used to input the enhanced sample lithography pattern into a pre-constructed initial lithography hotspot detection model, and to input the typical pattern features of the enhanced sample lithography pattern into the initial lithography hotspot detection model for training, thereby generating the lithography hotspot detection model.

13. The apparatus according to claim 11, characterized in that, The device further includes: The third acquisition unit is used to acquire and verify the photolithography pattern; The second extraction unit is used to extract typical pattern features from the verification photolithography pattern. The second obtaining unit is used to input the typical pattern features of the verification lithography pattern into the lithography hotspot detection model to obtain the detection result of the verification lithography pattern; The update unit is used to update the parameters of the lithography hotspot detection model when the lithography hotspot detection result of the verification lithography pattern is inconsistent with the lithography hotspot marking result corresponding to the verification lithography pattern.

14. The apparatus according to claim 11, characterized in that, The initial lithographic hotspot detection model is a convolutional neural network, which includes M convolutional layers, fully connected layers, and an output layer; where M is a positive integer greater than or equal to 2.

15. A photolithography hotspot detection device, characterized in that, The device includes a memory and a processor; The memory is used to store instructions; The processor is configured to execute the instructions in the memory to perform the method according to any one of claims 1-7.

16. A computer-readable storage medium comprising instructions, which, when executed on a computer, cause the computer to perform the method described in any one of claims 1-7.

Images