WO2022082692A1

WO2022082692A1 - Lithography hotspot detection method and apparatus, and storage medium and device

Info

Publication number: WO2022082692A1
Application number: PCT/CN2020/123036
Authority: WO
Inventors: 王佩瑶; 张锐; 陈成
Original assignee: 华为技术有限公司
Priority date: 2020-10-23
Filing date: 2020-10-23
Publication date: 2022-04-28
Also published as: CN116324788A

Abstract

A lithography hotspot detection method and apparatus, and a storage medium and a device, which are applied to the technical field of semiconductors. The method comprises: first, acquiring a lithography layout to be subjected to detection, and extracting a typical pattern feature that represents pattern information thereof; and then inputting the typical pattern feature into a preconstructed lithography hotspot detection model (comprising a convolutional layer, a fully connected layer and an output layer), so as to obtain a lithography hotspot in the lithography layout. It can be seen that the method involves inputting a typical pattern feature that represents pattern information of a lithography layout to be subjected to detection into a preconstructed lithography hotspot detection model for detection, such that the typical pattern feature (i.e. a geometric feature of a wiring arrangement in the lithography layout) contained in the lithography layout is fully taken into consideration during a detection process of the model, so as to overcome the existing problem of an inaccurate detection result caused by other pattern features of the lithography layout being easily lost when detection is performed on the basis of a feature such as the layout density, thereby effectively improving the accuracy of a detection result.

Description

A lithography hot spot detection method, device, storage medium and equipment

technical field

The present application relates to the field of semiconductor technology, and in particular, to a method, device, storage medium and device for detecting hot spots in lithography.

Background technique

As the size of semiconductor processes continues to shrink, the special layout patterns of key layers in process manufacturing are very sensitive to the effects of lithography process window changes or optical diffraction, although various optical resolution enhancement techniques have been introduced to correct the original layout to obtain better Accurate imaging quality, but the effectiveness of these techniques still strongly depends on the compatibility of the layout pattern with the lithography process. Therefore, in the design stage, it has become an effective means to ensure the consistency of design and manufacture in the advanced process to detect hotspots in the layout pattern that are difficult to implement and easy to introduce circuit failures, and to correct them in advance.

At present, there are usually two methods for lithography hotspot (HS) detection: one is to use machine learning to identify lithography hotspots. This method usually cannot directly use the layout as the input of the model, and requires the feature quantity of the layout. Extraction, commonly used feature quantities include graphic density, key size area, etc., and then perform image classification and recognition based on these feature quantities. The recall rate is highly dependent on the feature extraction results of the layout. Therefore, the false positive rate of the model is relatively high, and the recall rate and accuracy rate are relatively low. Another common detection method is to use deep learning for detection. Although this method can improve the detection accuracy without the need for feature extraction, this method only simplifies the hot spot identification into simple image recognition when performing lithography hot spot recognition, and does not consider the lithography process. In addition, during the model training process, the overall translation algorithm is used for data enhancement, which will lead to a small amount of blank areas in the sample images. Since only the hotspot data is usually enhanced, the increased blank areas are inconsistent with the actual layout pattern. As a result, the model will read pseudo features when reading graphic features, resulting in inaccurate detection results of the model, and there are disadvantages of high false positive rate and low recall rate.

SUMMARY OF THE INVENTION

The embodiments of the present application provide a lithography hot spot detection method, device, storage medium and equipment, which help to overcome the shortcomings of the existing lithography hot spot detection method, so that the detection process fully considers the typical pattern characteristics of the lithography layout, and further Improve the accuracy of detection results.

In a first aspect, the present application provides a method for detecting hot spots in lithography, the method comprising: first acquiring a lithography layout to be detected, and then extracting typical pattern features representing pattern information from the lithography layout to be detected (that is, the geometric features of the trace layout in the lithography layout, such as the diagonal spacing between the ends of the traces, etc.), then, the typical pattern features are input into the pre-built lithography hot spot detection model to obtain the lithography to be detected. Lithography hotspots in the layout, where the lithography hotspot detection model includes a convolutional layer, a fully connected layer and an output layer.

Compared with the traditional technology, because the embodiment of the present application inputs the extracted typical pattern features representing the lithography layout pattern information to be detected into the pre-built lithography hotspot including the convolution layer, the fully connected layer and the output layer. The detection model is used for detection, so that the detection process of the model fully considers the typical pattern features contained in the lithography layout (that is, the geometric features of the trace layout in the lithography layout), thereby overcoming the existing detection based on layout density and other characteristics. Loss of other pattern features of the lithography layout leads to the problem of inaccurate detection results, which can effectively improve the accuracy of the detection results.

In a possible implementation, the typical pattern features include one or more of the following features in the lithography layout: the area of the minimum spacing area between lines; the area of the minimum spacing area between the end of the line and the line; the end of the line and the line The area of the minimum spacing area at the end; the diagonal spacing between the end of the line and the end of the line; the number of T-shaped trace anchor points; the number of U-shaped trace anchor points.

In a possible implementation manner, the pre-built lithography hot spot detection model includes N different detection models, where N is a positive integer greater than or equal to 2; then the typical pattern features are input into the pre-built lithography hot spot detection model. model to obtain lithography hotspots in the lithography layout, including: inputting the typical pattern features into N different detection models, and predicting N detection results; inputting the N detection results to the preset fully connected layer to determine Lithographic hotspots in the lithography layout. In this way, detection results with higher accuracy and recall rate can be obtained.

In a possible implementation, a lithography hotspot detection model is constructed in the following manner: obtaining a sample lithography layout; using the sample lithography layout to train a pre-built initial lithography hotspot detection model to obtain a lithography hotspot detection model .

In a possible implementation manner, the sample lithography layout is used to train a pre-built initial lithography hotspot detection model to obtain a lithography hotspot detection model, including: performing data enhancement processing on the sample lithography layout to obtain an enhanced lithography hotspot detection model. Sample lithography layout; from the enhanced sample lithography layout, extract the typical pattern features of the enhanced sample lithography layout; input the enhanced sample lithography layout into the pre-built initial lithography hot spot detection model, and use the The typical pattern features of the enhanced sample lithography layout are input to the initial lithography hot spot detection model for training, and the lithography hot spot detection model is generated. Thus, the training accuracy of the model can be further improved and the detection accuracy of the model can be improved.

In a possible implementation manner, the method further includes: acquiring a verification lithography layout; extracting typical pattern features of the verification lithography layout from the verification lithography layout; inputting the typical pattern features of the verification lithography layout into a lithography hotspot The detection model is used to obtain the detection results of the verification lithography layout; when the lithography hot spot detection results of the verification lithography layout are inconsistent with the lithography hot spot marking results corresponding to the verification lithography layout, the verification lithography layout is re-used as the sample lithography layout. Parameter update for the lithography hotspot detection model. In this way, the lithography hotspot detection model can be effectively verified by using the verification lithography layout, and the lithography hotspot detection model can be adjusted and updated in time, thereby helping to improve the detection precision and accuracy of the detection model.

In a possible implementation manner, the initial lithography hotspot detection model is a convolutional neural network, wherein the convolutional neural network includes M layers of convolutional layers, a fully connected layer and an output layer; M is a positive integer greater than or equal to 2.

In a second aspect, the present application also provides a lithography hot spot detection device, the device includes: a first acquisition unit for acquiring a lithography layout to be detected; a first extraction unit for extracting from the lithography layout Typical pattern features of the lithography layout; the first obtaining unit is used to input the typical pattern features into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout; wherein, the lithography hotspot detection model includes convolution layer, fully connected layer and output layer.

In a possible implementation manner, 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: an obtaining subunit, which is used to The pattern features are respectively input to N different detection models, and N detection results are predicted to be obtained; the determination subunit is used to input the N detection results to the preset fully connected layer to determine the lithography hotspots in the lithography layout.

In a possible implementation manner, the device further includes: a second acquisition unit for acquiring a sample lithography layout; a training unit for training a pre-built initial lithography hot spot detection model by using the sample lithography layout, Obtain the lithography hot spot detection model.

In a possible implementation manner, the training unit includes: an enhancement subunit, which is used to perform data enhancement processing on the sample lithography layout to obtain an enhanced sample lithography layout; an extraction subunit is used to extract the sample lithography from the enhanced sample lithography. In the layout, the typical pattern features of the enhanced sample lithography layout are extracted; the training subunit is used to input the enhanced sample lithography layout into the pre-built initial lithography hot spot detection model, and the enhanced sample lithography The typical pattern features of the layout are input to the initial lithography hotspot detection model for training, and the lithography hotspot detection model is generated.

In a possible implementation manner, the device further includes: a third acquisition unit for acquiring a verification lithography layout; a second extraction unit for extracting typical pattern features of the verification lithography layout from the verification lithography layout; The second obtaining unit is used to input the typical pattern features of the verification lithography layout into the lithography hot spot detection model, and obtain the detection result of the verification lithography layout; the updating unit is used for when the lithography hot spot detection result of the verification lithography layout matches the When the lithography hot spot marking results corresponding to the verification lithography layout are inconsistent, the verification lithography layout is re-used as the sample lithography layout, and the parameters of the lithography hot spot detection model are updated.

In a third aspect, the present application also provides a lithography hot spot detection device, the device comprising: a memory and a processor;

The memory is used to store the instructions; the processor is used to execute the instructions in the memory, and execute the method in the first aspect and any possible implementation manners thereof.

In a fourth aspect, the present application further provides a computer-readable storage medium, including instructions, which, when executed on a computer, cause the computer to execute the method in the first aspect and any possible implementations thereof.

As can be seen from the above technical solutions, the embodiments of the present application have the following advantages:

When performing lithography hot spot detection in the embodiments of the present application, the lithography layout to be detected is first obtained, and then, typical pattern features representing pattern information thereof are extracted from the lithography layout to be detected, and then the typical pattern features are extracted. Input to a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout to be detected, wherein the lithography hotspot detection model includes a convolution layer, a fully connected layer and an output layer. It can be seen that in the embodiments of the present application, the typical pattern features that characterize the lithography layout pattern information to be detected are input into the pre-built lithography hotspot detection model including 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 contained in the lithography layout (that is, the geometric features of the trace layout in the lithography layout, such as the diagonal spacing between trace ends, etc.), thus overcoming the existing When detecting features such as layout density, it is easy to lose other pattern features of the lithography layout, resulting in an inaccurate detection result, which can effectively improve the accuracy of the detection result.

Description of drawings

1 is a schematic structural diagram of an artificial intelligence main frame provided by an embodiment of the present application;

FIG. 2 is a schematic diagram of an application scenario of an embodiment of the present application;

3 is a flowchart of a method for detecting hot spots in lithography provided by an embodiment of the present application;

4 is a schematic diagram of a pattern feature of a lithography layout provided by an embodiment of the present application;

5 is one of the schematic diagrams of inputting pattern features into a fully connected layer of a pre-built lithography hotspot detection model to detect and obtain hotspots in a lithography layout provided by an embodiment of the present application;

6 is the second schematic diagram of inputting pattern features into a fully connected layer of a pre-built lithography hot spot detection model to detect and obtain hot spots in a lithography layout provided by an embodiment of the present application;

7 is a schematic diagram of performing data enhancement processing on a sample lithography layout according to an embodiment of the present application;

8 is a schematic diagram of inputting an enhanced sample lithography layout to an initial lithography hot spot detection model provided by an embodiment of the present application;

9 is a structural block diagram of a lithography hot spot detection device provided by an embodiment of the present application;

FIG. 10 is a schematic structural diagram of a lithography hot spot detection device according to an embodiment of the present application.

Detailed ways

The embodiments of the present application provide a lithography hot spot detection method, device, storage medium and equipment, so that the detection process fully considers the pattern features of the lithography layout, and further improves the accuracy of the detection result.

The embodiments of the present application will be described below with reference to the accompanying drawings. Those of ordinary skill in the art know that with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

First, the overall workflow of the artificial intelligence system will be described. Please refer to Figure 1. Figure 1 shows a schematic structural diagram of the main frame of artificial intelligence. The above-mentioned artificial intelligence theme framework is explained in two dimensions (vertical axis). Among them, the "intelligent information chain" reflects a series of processes from data acquisition to processing. For example, it can be the general process of intelligent information perception, intelligent information representation and formation, intelligent reasoning, intelligent decision-making, intelligent execution and output. In this process, data has gone through the process of "data-information-knowledge-wisdom". The "IT value chain" reflects the value brought by artificial intelligence to the information technology industry from the underlying infrastructure of human intelligence, information (providing and processing technology implementation) to the industrial ecological process of the system.

(1) Infrastructure

The infrastructure provides computing power support for artificial intelligence systems, realizes communication with the outside world, and supports through the basic platform. Communication with the outside world through sensors; computing power is provided by smart chips (hardware acceleration chips such as CPU, NPU, GPU, ASIC, FPGA); the basic platform includes distributed computing framework and network-related platform guarantee and support, which can include cloud storage and computing, interconnection networks, etc. For example, sensors communicate with external parties to obtain data, and these data are provided to the intelligent chips in the distributed computing system provided by the basic platform for calculation.

(2) Data

The data on the upper layer of the infrastructure is used to represent the data sources in the field of artificial intelligence. The data involves graphics, images, voice, and text, as well as IoT data from traditional devices, including business data from existing systems and sensory data such as force, displacement, liquid level, temperature, and humidity.

(3) Data processing

Data processing usually includes data training, machine learning, deep learning, search, reasoning, decision-making, etc.

Among them, machine learning and deep learning can perform symbolic and formalized intelligent information modeling, extraction, preprocessing, training, etc. on data.

Reasoning refers to the process of simulating human's intelligent reasoning method in a computer or intelligent system, using formalized information to carry out machine thinking and solving problems according to the reasoning control strategy, and the typical function is search and matching.

Decision-making refers to the process of making decisions after intelligent information is reasoned, usually providing functions such as classification, sorting, and prediction.

(4) General ability

After the above-mentioned data processing, some general capabilities can be formed based on the results of data processing, such as algorithms or a general system, such as translation, text analysis, computer vision processing, speech recognition, image identification, etc.

The embodiments of the present application may be applied in the field of information technology. The following will take anomaly detection on a storage device as an example for description. The embodiments of the present invention may also be applied to other devices, such as servers, networks, and other devices. Not limited.

The process of lithography hot spot detection applied in computer equipment is as follows:

The lithography hot spot detection method provided by the embodiments of the present application can be applied to the lithography hot spot detection process in computer equipment. Referring to FIG. 2, FIG. 2 is a schematic diagram of an application scenario of an embodiment of the present application. As shown in FIG. 2, the computer device 201 is provided with an AI system that implements a function of lithography hot spot detection, so as to obtain a lithography layout to be detected, wherein the to-be-detected lithography layout is The detected lithography layout may include a lithography pattern, which is used to etch a film layer of a specific shape in the semiconductor process manufacturing, and the to-be-detected lithography layout is more sensitive to the influence of the lithography process window or optical diffraction, so, The lithography layout to be detected has a great influence on the imaging quality of the lithography. Among them, the lithography process window is also called the lithography process tolerance, which specifically refers to the exposure dose and defocus amount range to ensure that the mask pattern can be correctly copied to the silicon wafer. Engraved craft window. After feature extraction is performed on the obtained lithography layout to be detected to obtain its corresponding typical pattern features, it can be input into the fully connected layer in the AI detection system model to continue to detect the obtained typical pattern features, so that according to the output The results determine the detection results of lithography hot spots in the lithography layout to be detected, so that the entire detection process fully considers the typical pattern characteristics of the lithography layout, and overcomes the fact that other patterns of the lithography layout are easily lost when detecting based on characteristics such as layout density. The problem of inaccurate detection results is caused by the characteristics, which can effectively improve the accuracy of the detection results. In this way, after the lithography layout design to be detected is completed, before using it to etch a film layer of a specific shape in the semiconductor process manufacturing, the AI system in the computer equipment 201 with the function of realizing lithography hot spot detection can be used first, Perform the lithography hot spot detection on the lithography layout to be detected by performing the aforementioned detection method, and then correct the detected lithography hot spots in advance according to the detection results, so as to avoid the subsequent use of the lithography layout to perform lithography etching to form specific lithography hot spots. There is a problem with the shape of the film layer, resulting in large economic losses.

Wherein, as an example, the computer device 201 can be any device capable of analyzing the lithography layout and detecting lithography hot spots, including but not limited to: smart phones, non-smart phones, tablet computers, laptop personal computers, Desktop personal computers, minicomputers, medium computers, mainframe computers, etc. It should be understood that the embodiments of the present application may also be applied to other scenarios where lithography hot spot detection is required, and other application scenarios will not be listed one by one here.

Based on the above application scenarios, an embodiment of the present application provides a method for detecting hot spots in lithography, and the method is introduced below.

S301: Obtain a lithography layout to be detected.

It should be noted that, as the size of the semiconductor process continues to shrink, that is, the number of circuit elements per unit area and the size of the miniature circuit elements continue to increase, the circuit element patterns produced on the substrate are getting smaller and smaller and closer to each other. . The reduction in the feature size of circuit elements increases the difficulty in fabricating the desired layout pattern on the substrate. This is partly due to the phenomenon of diffraction of light causing defects during the lithographic manufacturing process, so that the desired image is not accurately imaged on the substrate, thereby creating imperfections in the final device structure.

Therefore, in order to avoid such defects, after the lithography layout design is completed, before using it to etch a film layer of a specific shape in the semiconductor process manufacturing, it is necessary to firstly analyze the lithography layout process that is difficult to achieve and easy to introduce. The hot spot of circuit failure is accurately detected, and then the detected lithography hot spot is corrected in advance according to the detection result, so as to avoid the subsequent use of the lithography layout to etch to form a specific shape of the film. economic loss, so as to ensure the consistency of design and manufacture. Therefore, after the lithography layout to be detected is acquired in the embodiment of the present application, the lithography hot spots in the lithography layout can be detected through subsequent steps S302-S303.

S302: From the lithography layout, extract typical pattern features of the lithography layout.

In this embodiment, after the lithography layout to be detected is acquired in step S301, the image feature extraction method can be used to extract features, and typical pattern features that can characterize the pattern information can be extracted from them to perform subsequent steps. S303. Among them, the typical pattern features of the lithography layout refer to the geometric features of the trace layout in the photolithography layout, such as the diagonal spacing between trace ends.

In an optional implementation manner of this embodiment, the extracted typical pattern features may include one or more of the following features in the lithography layout:

The area of the minimum distance between the line and the line; the area of the minimum distance between the end of the line and the line; the area of the minimum distance between the end of the line and the end of the line; the diagonal distance between the end of the line and the end of the line; the number of T-shaped trace anchor points; the U-shaped The number of trace anchor points.

Specifically, as shown in Figure 4, the shaded part between the two rectangular bars in Figure 4a represents the area of the minimum spacing area (line to line minimum spacing area) between lines in the lithography layout. The shaded part between the two rectangular bars in 4b represents the line-end to line minimum spacing area between the line end and the line in the lithography layout. The area between the two rectangular bars in Figure 4c is The shaded part represents the area of the line-end to line-end minimum spacing area in the lithography layout, and the part indicated by the arrow between the two rectangular bars in Figure 4d represents the lithography layout The line-end to line-end corner spacing (line-end to line-end corner spacing), the two bold black dots in Figure 4e represent the number of T-shape traces in the lithography layout. anchor point), the four bold black dots in Figure 4f represent the number of u-shape anchor points in the lithography layout.

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

In this embodiment, after the typical pattern feature of the lithography layout to be detected is extracted through step S302, it can be input into a pre-built lithography hotspot detection model including a convolution layer, a fully connected layer and an output layer, And merge each array vector array obtained in the fully connected layer after processing by the multi-layer convolution layer in the layout array corresponding to the lithography layout, and then use the combined array as the output result of the fully connected layer, and then you can According to the output result, the detection result of the lithography hot spot in the lithography layout is output through the output layer of the model (such as the probability that the detection position is the lithography hot spot).

Example: As shown in Figure 5, in order to improve the detection accuracy of the model, after the typical pattern features of the lithography layout to be detected are extracted, the typical pattern feature data can be analyzed according to the actual process nodes and levels of the layout. Perform preprocessing operations (such as normalization operations, etc.) to obtain a typical pattern feature vector after preprocessing (the dimension of the vector is greater than or equal to 1 and less than or equal to 6, as shown in Figure 5, between 1 and 6, for example, it can be 4 ×1 dimension), and then process each array vector (the dimension of the vector as shown in Fig. 5) obtained by processing the multi-layer convolution layer (3-layer convolution layer in Fig. 5) in the layout array corresponding to the lithography layout. 1024×1 dimension) are jointly input to the fully connected layer to form a new feature vector (the dimension of the vector is greater than or equal to 1025 and less than or equal to 1030, as shown in Figure 5, between 1025 and 1030, for example, it can be 1028×1 dimension) , and output through the model output layer whether the layout position corresponding to the array is a lithography hotspot. For example, a probability value P can be output as the detection result. And the value range of P is [0,1], where 1 indicates that the layout position is a hot spot, 0 indicates that the layout position is not a hot spot, 0.9 indicates that the layout position is 90% likely to be a hot spot, and so on. Among them, the lithography hotspot refers to a specific layout pattern that is difficult to implement in the layout pattern after the lithography layout design is completed, and it is easy to introduce circuit failure.

In an optional implementation manner of this embodiment, in order to improve the detection accuracy and recall rate of the model, N different lithography hotspot detection models can be obtained by pre-training by setting different hyperparameters, where N is A positive integer greater than or equal to 2, the recall rate refers to the proportion of all "lithography layouts with lithography hotspots detected accurately" to all "lithography layouts that should be detected with lithography hotspots". And different models have different attributes such as the accuracy rate, false positive rate, and recall rate of the detection results. Some models have high accuracy and high false positive rates, and some models have low accuracy and low false positive rates. The accuracy of the model is low, but the F score (representing the harmonic value of accuracy and recall) is high. Therefore, in order to obtain detection results with higher accuracy and recall, the fully connected layer can be used to detect different models. The results are processed again to obtain the final detection result. Specifically, as shown in FIG. 6 , the execution process of this step S303 may include the following steps A1-A2:

Step A1: Input the typical pattern features into N different detection models respectively, and predict to obtain N detection results.

In this implementation manner, after the typical pattern features of the lithography layout to be detected are extracted in step S302, the typical pattern features can be respectively input into pre-built lithography hot spot detection models with N different attributes to obtain N detection models. result. Among them, the detection results corresponding to some models have the highest accuracy rate, the detection results corresponding to some models have the best recall rate, and the detection results corresponding to some models have the highest F score.

For example: As shown in Figure 6, N is set to 3, and the attributes of the detection results corresponding to the three pre-built lithography hot spot detection models are the best recall rate, the best F score, and the best precision. After inputting the typical pattern features of the lithography layout to be detected into the lithography hot spot detection models with different attributes, the three detection results obtained are P ₁ , P ₂ , and P ₃ respectively. It can be seen that P ₁ corresponds to The recall rate of P 2 is the best, the F score corresponding to P ₂ is the best, and the precision corresponding to P ₃ is the best.

Step A2: Input the N detection results to the preset fully connected layer, and determine the lithography hot spots in the lithography layout.

In this implementation manner, after three detection results are obtained through step A1, the N detection results can be further input into a preset fully connected layer for comprehensive processing, and a layout corresponding to the lithography layout is determined according to the processing results Whether the layout position corresponding to each array in the array is a hot spot.

Example: Based on the above example, as shown in Figure 6, after three detection results P ₁ , P ₂ , and P ₃ are obtained, these three detection results can be input to the preset fully connected layer for comprehensive processing, and then Whether the corresponding layout position is a lithography hot spot can be determined according to the processing result.

It should be noted that an optional implementation method is that a large amount of training layout data can be used in advance, and a variety of detection models with different attributes can be obtained by training by setting different hyperparameters (such as models with high accuracy and false positive rate). Also higher, some models have low accuracy and low false positive rate). Then, after obtaining the respective output detection results through these different models, the output results of these models can be retrained by using the preset fully connected layer to obtain the final model with high accuracy and recall rate. Only this final model is used to perform hot spot detection on the lithography layout. For the specific implementation process, please refer to the above introduction, which will not be repeated here.

To sum up, in a method for detecting hot spots in lithography provided by this embodiment, when detecting hot spots in lithography, first obtain the lithography layout to be detected, and then extract pattern information representing the lithography layout from the lithography layout to be detected The typical pattern features of the layer and output layer. It can be seen that in the embodiments of the present application, the typical pattern features that characterize the lithography layout pattern information to be detected are input into the pre-built lithography hotspot detection model including the convolution layer, the fully connected layer and the output layer for detection. , so that the model detection process fully considers the typical pattern features contained in the lithography layout (that is, the geometric characteristics of the trace layout in the lithography layout, such as the diagonal spacing between trace ends, etc.), thus overcoming the existing layout-based layout. When detecting features such as density, it is easy to lose other pattern features of the lithography layout, resulting in an inaccurate detection result, which can effectively improve the accuracy of the detection result.

Next, this embodiment will introduce the construction process of the lithography hot spot detection model, which may specifically include the following steps B1-B2:

Step B1: Obtain a sample lithography layout.

In this embodiment, in order to build a lithography hot spot detection model, a lot of preparation work needs to be done in advance. First, a large number of sample lithography layout data including hot spots and non-hot spots need to be collected. For example, 100 lithography layouts can be collected in advance, Each collected lithography layout is used as a sample lithography layout to train a lithography hot spot detection model.

It should be noted that, in order to reduce the machine resource occupation and running time of computer equipment, after obtaining the sample lithography layout, the image size of the sample lithography layout data can be normalized first, and each sample lithography can be processed by normalizing the image size. The 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 lithography process to ensure that all lithography influence ranges are covered, and the sample lithography layout is converted into an array according to a certain pixel ratio, so that This scaling ratio can effectively reduce machine resource occupation and running time of computer equipment without affecting layout accuracy.

Step B2: Using the sample lithography layout, the pre-built initial lithography hot spot detection model is trained to obtain a lithography hot spot detection model.

In this embodiment, an initial lithography hotspot detection model may be pre-built and model parameters are initialized. An optional implementation manner is that the initial lithography hotspot detection model may be a convolutional neural network (CNN) ), and the convolutional neural network includes M layers of convolution layers, fully connected layers and output layers, where M is a positive integer greater than or equal to 2. The initial lithography hotspot detection model is used to obtain the lithography hotspot detection model according to the typical pattern features in the sample lithography layout and the pattern density coding on the sample lithography layout. The specific implementation process may include the following steps B21-B23:

Step B21: Perform data enhancement processing on the sample lithography layout to obtain an enhanced sample lithography layout.

In practical applications, the hotspot data of the lithography layout is often far less than the non-hotspot data. Therefore, after obtaining the sample lithography layout, if it is directly used as the training data set for training, it will be due to the hotspot data in the data. The extremely unbalanced distribution of data and non-hotspot data leads to insufficient training model accuracy. Therefore, it is necessary to perform data enhancement processing on the obtained sample lithography layout to obtain an enhanced sample lithography layout to expand the hotspots of the sample lithography layout. The amount of data can increase the generalization ability of the model and improve the accuracy of the model.

Specifically, in this embodiment, the data volume of the sample lithography layout is expanded by means of up, down, left, right translation, flip, and rotation according to a certain step size. For example, as shown in FIG. Translate, flip vertically, flip horizontally and rotate 180 degrees (for layers that support 2D orientation, you can use 90 degrees/180 degrees/270 degrees of rotation) to obtain a larger number of enhanced sample lithography layouts, using to perform the subsequent step B22.

Step B22: From the enhanced sample lithography layout, extract typical pattern features of the enhanced sample lithography layout.

After the enhanced sample lithography layout is obtained through step B21, typical pattern features representing the pattern information of the enhanced sample lithography layout can be extracted from each enhanced sample lithography layout. In the above-mentioned step S302, a method similar to the typical pattern features of the lithography layout to be detected is extracted from the lithography layout to be detected, and the lithography layout to be detected is replaced with an enhanced sample lithography layout. The typical pattern features that characterize the pattern information of each enhanced sample lithography layout (that is, the geometric features of the trace layout in the enhanced sample lithography layout, including the following one in the enhanced sample lithography layout) are extracted from the lithography. Item or more features: area of minimum spacing area between line and line; area area of minimum spacing area between line end and line; area area of minimum spacing area between line end and line end; diagonal spacing between line end and line end; T-shaped routing The number of anchor points; the number of anchor points of the U-shaped line), please refer to the introduction of the above step S302 for related details, which will not be repeated here.

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

After obtaining the enhanced sample lithography layout through step B21, the encoding arrays corresponding to the enhanced sample lithography layout can be further input into the pre-built initial lithography hot spot detection model, as shown in FIG. The engraved hot spot detection model consists of multiple convolutional layers (3 layers in Figure 8), fully connected layers and output layers. In this embodiment, the linear correction unit is used as the activation function of the convolution layer. In practical applications, other functions (such as the hyperbolic tangent function) can also be used for training and learning according to the actual situation. Algorithms such as regularization and batch normalization are added to the model.

It should be noted that, in order to preserve the complete information of the lithography layout pattern as much as possible, the initial lithography hot spot detection model constructed in this embodiment does not use a pooling layer with reduced feature dimensions. In practical applications, it can also be appropriate according to the actual situation. A pooling layer is added to compress data and parameters to improve the training efficiency of the model.

In addition, the embodiment of the present application adopts the sigmoid function as the output layer activation function for classification (distinguishes hot spots and non-hot spots), and the normalized exponential function (softmax) can also be used as the classification output layer function in practical applications. At the same time, loss functions such as mean square error, cross entropy and gradient descent or adaptive learning rate algorithm and its deformation can also be used as optimizers for model training. In this case, focal-loss is used as the loss function, and different categories are set. Weight, centralized training contains hotspots and hotspot data imbalanced layout data to reduce the impact of imbalanced datasets. Therefore, during the training process, by setting different hyperparameters, at least one trained initial lithography hotspot detection model can be obtained by training, or N different trained initial lithography hotspot detection models can be obtained, where N is greater than or A positive integer equal to 2, and different models have different attributes such as the accuracy rate, false positive rate, and recall rate of the detection results. Some models have high accuracy and high false positive rates, and some models have low accuracy. The false positive rate is low, and some models have low accuracy, but high F scores.

After the pattern features of the enhanced sample lithography layout are extracted through step B22, the typical pattern features of the sample lithography layout can be further compared with the multi-layer volume of the initial lithography hot spot detection model in the layout array corresponding to the sample lithography layout. Each coding array vector obtained after the multi-layer processing is jointly input to the trained initial lithography hotspot detection model for training, and the model outputs a probability value of whether the layout position corresponding to the array is a hotspot (such as the value range). is a probability value in [0,1]). Then, the probability value can be compared with the corresponding real detection results (for example, 1 indicates that the layout position is a hot spot, and 0 indicates that the layout position is not a hot spot), and the model parameters are updated according to the difference between the two until the preset is satisfied. For example, if the variation of the difference is small, 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.

It should be noted that, in an optional implementation manner, some typical pattern features of typical lithographic layout patterns can also be extracted (ie, the area of the minimum spacing area between lines; the area of the minimum spacing area between the end of the line and the line) ; area area of the minimum distance between the end of the line and the end of the line; the diagonal distance between the end of the line and the end of the line; the number of T-shaped trace anchor points; at least one of the number of U-shaped trace anchor points), input to the initial lithography hot spot detection model , each encoding array vector obtained in the fully connected layer after processing by the multi-layer convolution layer of the initial lithography hotspot detection model in the layout array corresponding to the sample lithography layout is jointly trained to generate a trained light Engraved hot spot detection model. The specific implementation process is not repeated here.

Through the above embodiment, a lithography hotspot detection model can be generated by using the sample lithography layout training, and further, the generated lithography hotspot detection model can be verified by using the verification lithography layout. The specific verification process may include the following steps C1-C4:

Step C1: Obtain a verification lithography layout.

In this embodiment, in order to verify the lithography hotspot detection model, a large amount of verification lithography layout data needs to be obtained first, wherein the verification lithography layout refers to a lithography layout that can be used to verify the lithography hotspot detection model , after the verification lithography layout is obtained, the subsequent step C2 can be continued.

Step C2: From the verification lithography layout, extract typical pattern features of the verification lithography layout.

After the verification lithography layout is obtained through step C1, it cannot be directly used to verify the lithography hot spot detection model, but it is necessary to first extract the typical pattern features that characterize the verification lithography layout pattern information (ie, verify the trace layout in the lithography layout). The geometric features of the lithography, including verifying the area of the minimum space between lines and lines in the lithography layout; the area of the minimum space between the end of the line and the line; the area of the minimum space between the end of the line and the end of the line; the diagonal distance between the end of the line and the end of the line; T The number of anchor points of the U-shaped trace; at least one of the number of anchor points of the U-shaped trace), and then the extracted typical pattern features of the verification lithography layout can be used to verify the obtained lithography hot spot detection model.

Step C3: Input the typical pattern features of the verification lithography layout into the lithography hot spot detection model, and obtain the detection result of the verification lithography layout.

After the typical pattern features of the verification lithography layout are extracted through step C2, further, the typical pattern features of the verification lithography layout can be input into the lithography hot spot detection model to obtain the detection results of the verification lithography layout, and then the subsequent steps can be continued. C4.

Step C4: When the lithography hot spot detection result of the verification lithography layout is inconsistent with the lithography hot spot marking result corresponding to the verification lithography layout, the verification lithography layout is re-used as the sample lithography layout, and the parameters of the lithography hot spot detection model are updated. .

After obtaining the detection result of the verification lithography layout through step C3, if the lithography hot spot detection result of the verification lithography layout is inconsistent with the lithography hot spot marking result (ie, the real detection result) corresponding to the verification lithography layout, the verification photo The lithography map is re-used as the sample lithography layout, and the parameters of the lithography hot spot detection model are updated.

Through the above embodiment, the lithography hotspot detection model can be effectively verified by using the verification lithography layout. When the lithography hotspot detection result of the verification lithography layout is inconsistent with the actual detection result of the lithography hotspot corresponding to the verification lithography layout, the lithography hotspot detection result can be timely verified. Adjust and update the lithography hotspot detection model, thereby helping to improve the detection accuracy and accuracy of the detection model.

In summary, using the lithography hot spot detection model trained in this embodiment, the typical pattern features of the lithography layout to be detected can be used to quickly and accurately detect the hot spot position of the lithography layout, which effectively improves the light intensity to be detected. Efficiency and accuracy of hot spot detection in lithography.

In order to better implement the above solutions of the embodiments of the present application, related devices for implementing the above solutions are also provided below. Referring to FIG. 9 , an embodiment of the present application provides a lithography hot spot detection apparatus 900 . The apparatus 900 may include: a first obtaining unit 901 , a first extracting unit 902 and a first obtaining unit 902 . The first obtaining unit 901 is configured to support the apparatus 900 to perform S301 in the embodiment shown in FIG. 3 . The first extraction unit 902 is configured to support the apparatus 900 to perform S302 in the embodiment shown in FIG. 3 . The first obtaining unit 903 is configured to support the apparatus 900 to perform S303 in the embodiment shown in FIG. 3 . specific,

The first obtaining unit 901 is used to obtain the lithography layout to be detected;

a first extraction unit 902, configured to extract typical pattern features of the lithography layout from the lithography layout;

The first obtaining unit 903 is used for inputting typical pattern features into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout; wherein, the lithography hotspot detection model includes a convolutional layer, a fully connected layer and an output layer.

In an implementation of this embodiment, the typical pattern features include one or more of the following features in the lithography layout:

The area of the minimum spacing area between lines;

The area of the minimum spacing area between the end of the line and the line;

The area of the minimum distance between the end of the line and the end of the line;

The diagonal distance between the end of the line and the end of the line;

The number of anchor points of the T-shaped trace;

The number of U-shaped trace anchor points.

In an implementation of this embodiment, 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 903 includes:

Obtaining subunits for inputting typical pattern features into N different detection models respectively, and predicting N detection results;

The determining subunit is used for inputting the N detection results to the preset fully connected layer, and determining the lithography hot spot in the lithography layout.

In an implementation manner of this embodiment, the device further includes:

a second acquisition unit, configured to acquire a sample lithography layout;

The training unit is used for using the sample lithography layout to train the pre-built initial lithography hot spot detection model to obtain the lithography hot spot detection model.

In an implementation of this embodiment, the training unit includes:

The enhancer unit is used to perform data enhancement processing on the sample lithography layout to obtain the enhanced sample lithography layout;

an extraction subunit for extracting typical pattern features of the enhanced sample lithography layout from the enhanced sample lithography layout;

The training subunit is used to input the enhanced sample lithography layout into the pre-built initial lithography hotspot detection model, and input the typical pattern features of the enhanced sample lithography layout into the initial lithography hotspot detection model for training, Generate a lithographic hotspot detection model.

In an implementation manner of this embodiment, the device further includes:

The third acquisition unit is used to acquire the verification lithography layout;

The second extraction unit is used for extracting typical pattern features of the verification lithography layout from the verification lithography layout;

The second obtaining unit is used to input the typical pattern features of the verification lithography layout into the lithography hot spot detection model to obtain the detection result of the verification lithography layout;

The update unit is used to re-use the verification lithography layout as the sample lithography layout when the lithography hot spot detection result of the verification lithography layout is inconsistent with the lithography hot spot mark result corresponding to the verification lithography layout, and perform the lithography hot spot detection model. Parameter update.

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

To sum up, a lithography hot spot detection device provided in this embodiment, when performing lithography hot spot detection, firstly obtains the lithography layout to be detected, and then extracts pattern information representing the lithography layout from the lithography layout to be detected The typical pattern features of the layer and output layer. It can be seen that in the embodiments of the present application, the typical pattern features that characterize the lithography layout pattern information to be detected are input into the pre-built lithography hotspot detection model including 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 contained in the lithography layout (that is, the geometric features of the trace layout in the lithography layout, such as the diagonal spacing between trace ends, etc.), thus overcoming the existing When detecting features such as layout density, it is easy to lose other pattern features of the lithography layout, resulting in an inaccurate detection result, which can effectively improve the accuracy of the detection result.

Referring to FIG. 10, an embodiment of the present application provides a lithography hot spot detection device 1000, the device includes a memory 1001, a processor 1002, and a communication interface 1003,

memory 1001 for storing instructions;

The processor 1002 is configured to execute the instructions in the memory 1001, and execute the above-mentioned lithography hot spot detection method applied to the embodiment shown in FIG. 3;

The communication interface 1003 is used for communication.

The memory 1001, the processor 1002 and the communication interface 1003 are connected to each other through a bus 1004; the bus 1004 may be a peripheral component interconnect (PCI for short) bus or an extended industry standard architecture (EISA for short) bus Wait. The bus can be divided into address bus, data bus, control bus and so on. For ease of presentation, only one thick line is used in FIG. 10, but it does not mean that there is only one bus or one type of bus.

In a specific embodiment, the processor 1002 is configured to first obtain the lithography layout to be detected when performing lithography hot spot detection, and then extract typical pattern features representing pattern information from the lithography layout to be detected, and then , and input the typical pattern features into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout to be detected, wherein the lithography hotspot detection model includes a convolutional layer, a fully connected layer and an output layer. For the detailed processing process of the processor 1002, please refer to the detailed descriptions of S301, S302, and S303 in the embodiment shown in FIG. 3 above, which will not be repeated here.

The above-mentioned memory 1001 can be random-access memory (random-access memory, RAM), flash memory (flash), read only memory (read only memory, ROM), erasable programmable read only memory (erasable programmable read only memory, EPROM) ), electrically erasable programmable read only memory (electrically erasable programmable read only memory, EEPROM), register (register), hard disk, removable hard disk, CD-ROM or any other form of storage medium known to those skilled in the art.

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

The above-mentioned communication interface 1003 may be, for example, an interface card or the like, and may be an Ethernet (ethernet) interface or an asynchronous transfer mode (Asynchronous transfer mode, ATM) interface.

Embodiments of the present application also provide a computer-readable storage medium, including instructions, which, when executed on a computer, cause the computer to execute the above-mentioned method for detecting hot spots in lithography.

The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that data so used can be interchanged under appropriate circumstances so that the embodiments described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having", and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those expressly listed Rather, those steps or units may include other steps or units not expressly listed or inherent to these processes, methods, products or devices.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above may refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical module division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be acquired according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each module unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of software module units.

The integrated unit, if implemented in the form of a software module unit and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of software products in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, and the computer software products are stored in a storage medium , including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .

Those skilled in the art should appreciate that, in one or more of the above examples, the functions described in the present invention may be implemented in hardware, software, firmware, or any combination thereof. When implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage medium can be any available medium that can be accessed by a general purpose or special purpose computer.

The specific embodiments described above further describe the objectives, technical solutions and beneficial effects of the present invention in further detail, and it should be understood that the above descriptions are only specific embodiments of the present invention.

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

In addition, the embodiments of the present application may also be applicable to other future-oriented communication technologies. The network architecture and service scenarios described in this application are for the purpose of illustrating the technical solutions of this application more clearly, and do not constitute a limitation on the technical solutions provided in this application. appears, the technical solutions provided in this application are also applicable to similar technical problems.

Claims

A lithography hot spot detection method, characterized in that the method comprises:

Obtain the lithography layout to be detected;

from the lithography layout, extracting typical pattern features of the lithography layout;

The typical pattern features are input into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout; the lithography hotspot detection model includes a convolution layer, a fully connected layer and an output layer.
The method of claim 1, wherein the typical pattern features include one or more of the following features in the lithography layout:

The area of the minimum spacing area between lines;

The area of the minimum spacing area between the end of the line and the line;

The area of the minimum distance between the end of the line and the end of the line;

The diagonal distance between the end of the line and the end of the line;

The number of anchor points of the T-shaped trace;

The number of U-shaped trace anchor points.
The method according to claim 1 or 2, wherein the pre-built lithography hot spot detection model includes N different detection models, and N is a positive integer greater than or equal to 2;

Inputting the typical pattern features into a pre-built lithography hotspot detection model to obtain lithography hotspots in the lithography layout, including:

Inputting the typical pattern features into the N different detection models respectively, and predicting to obtain N detection results;

The N detection results are input to a preset fully connected layer, and the lithography hotspots in the lithography layout are determined.
The method according to claim 1, wherein the lithography hot spot detection model is constructed in the following manner:

Obtain sample lithography layout;

Using the sample lithography layout, the pre-built initial lithography hot spot detection model is trained to obtain the lithography hot spot detection model.
The method according to claim 4, wherein, using the sample lithography layout to train a pre-built initial lithography hotspot detection model to obtain the lithography hotspot detection model, comprising:

performing data enhancement processing on the sample lithography layout to obtain an enhanced sample lithography layout;

extracting typical pattern features of the enhanced sample lithography layout from the enhanced sample lithography layout;

inputting the enhanced sample lithography layout into a pre-built initial lithography hot spot detection model, and inputting the typical pattern features of the enhanced sample lithography layout into the initial lithography hot spot detection model for training, The lithography hotspot detection model is generated.
The method according to claim 4, wherein the method further comprises:

Obtain the verification lithography layout;

from the verification lithography layout, extracting typical pattern features of the verification lithography layout;

Inputting the typical pattern features of the verification lithography layout into the lithography hot spot detection model to obtain a detection result of the verification lithography layout;

When the lithography hot spot detection result of the verification lithography layout is inconsistent with the lithography hot spot marking result corresponding to the verification lithography layout, the verification lithography layout is re-used as the sample lithography layout, and The parameters of the hot spot detection model are updated.
The method according to claim 4, wherein the initial lithography hotspot detection model is a convolutional neural network, and the convolutional neural network comprises M layers of convolution layers, fully connected layers and an output layer; the M layers is a positive integer greater than or equal to 2.
A lithography hot spot detection device, characterized in that the device comprises:

a first acquisition unit, used for acquiring the lithography layout to be detected;

a first extraction unit for extracting typical pattern features of the lithography layout from the lithography layout;

The first obtaining unit is used to input the 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 convolution layer, a fully connected layer and output layer.
The apparatus of claim 8, wherein the typical pattern features include one or more of the following features in the lithography layout:

The area of the minimum spacing area between lines;

The area of the minimum spacing area between the end of the line and the line;

The area of the minimum distance between the end of the line and the end of the line;

The diagonal distance between the end of the line and the end of the line;

The number of anchor points of the T-shaped trace;

The number of U-shaped trace anchor points.
The device according to claim 8 or 9, wherein the pre-built lithography hotspot detection model includes N different detection models, and N is a positive integer greater than or equal to 2;

The first obtaining unit includes:

obtaining subunits for inputting the typical pattern features into the N different detection models respectively, and predicting to obtain N detection results;

A determination subunit, configured to input the N detection results to a preset fully connected layer, and determine a lithography hotspot in the lithography layout.
The apparatus according to claim 8, wherein the apparatus further comprises:

a second acquisition unit, configured to acquire a sample lithography layout;

The training unit is configured to use the sample lithography layout to train a pre-built initial lithography hotspot detection model to obtain the lithography hotspot detection model.
The apparatus according to claim 11, wherein the training unit comprises:

an enhancer unit for performing data enhancement processing on the sample lithography layout to obtain an enhanced sample lithography layout;

an extraction subunit for extracting typical pattern features of the enhanced sample lithography layout from the enhanced sample lithography layout;

A training subunit for inputting the enhanced sample lithography layout into a pre-built initial lithography hotspot detection model, and inputting the typical pattern features of the enhanced sample lithography layout into the initial lithography The hot spot detection model is trained to generate the lithography hot spot detection model.
The apparatus of claim 11, wherein the apparatus further comprises:

The third acquisition unit is used to acquire the verification lithography layout;

a second extraction unit, configured to extract typical pattern features of the verification lithography layout from the verification lithography layout;

a second obtaining unit, configured to input the typical pattern features of the verification lithography layout into the lithography hot spot detection model, and obtain a detection result of the verification lithography layout;

an update unit, configured to re-use the verification lithography layout as the sample lithography layout when the lithography hot spot detection result of the verification lithography layout is inconsistent with the lithography hot spot marking result corresponding to the verification lithography layout , and update the parameters of the lithography hot spot detection model.
The device according to claim 11, wherein the initial lithography hotspot detection model is a convolutional neural network, and the convolutional neural network comprises M layers of convolutional layers, a fully connected layer and an output layer; the M is a positive integer greater than or equal to 2.
A lithography hot spot detection device, characterized in that the device includes a memory and a processor;

the memory for storing instructions;

The processor is configured to execute the instructions in the memory, and execute the method of any one of claims 1-7.
A computer-readable storage medium comprising instructions which, when run on a computer, cause the computer to perform the method of any one of claims 1-7 above.