A hotspot detection method and apparatus

Abstract

The application relates to the field of hotspot recognition, in particular to a hotspot detection method and device, which can solve the problem of poor accuracy when performing hotspot detection through photolithography simulation. The hotspot detection method comprises the following steps: storing a plurality of first images comprising circuit layouts into an image library, wherein the circuit layouts contain hotspots; determining a matching image in the image library which matches a to-be-detected image, the first image comprises the matching image, and the to-be-detected image comprises a wafer which needs to be subjected to hotspot detection; acquiring a feature vector of the matching image; inputting the feature vector into a preset detection model; and determining the type and position of the hotspots in the wafer based on the output of the detection model.

Description

Technical Field

[0001] This application relates to the field of hotspot identification, and more specifically, to a hotspot detection method and apparatus. Background Technology

[0002] A crucial step in chip manufacturing is photolithography, where the wafer is lithographically etched using a photolithography machine. This process involves removing specific areas from the wafer's surface according to a pre-designed circuit layout, creating a pattern on the wafer that matches the circuit layout. The lithographically etched wafer then undergoes further processing to form the final chip. Currently, with increasing demands for chip performance and integration, the wavelengths used in photolithography are constantly decreasing. This results in the Optical Proximity Effect (OPE), which distorts the wafer's surface pattern, leading to defects such as bridging and breaking. These defects are known as lithography hotspots, or simply hotspots. The presence of hotspots causes chip errors and affects normal chip operation. Therefore, it is necessary to detect hotspots after photolithography to address problematic wafers promptly.

[0003] Current hotspot detection is generally based on photolithography simulation. Specifically, a series of initial process parameters are first set to simulate the photolithography process of the wafer. During this process, the initial process parameters are fitted multiple times, and the process parameters are continuously corrected based on the fitting results until the final simulation result is obtained. The simulation result is a virtual pattern of the wafer obtained after simulating the photolithography process. The simulation result is compared with the real pattern on the surface of the wafer obtained by photolithography based on the same circuit layout to determine the difference points between the two. The location of the difference point on the wafer is the location of the hotspot.

[0004] However, the method of hot spot detection through photolithography simulation requires multiple fitting and correction of a large number of process parameters to obtain simulation results. The various parameters in this process are usually interrelated, so the process of continuous fitting and correction is prone to errors, resulting in inaccurate simulation results, which in turn leads to poor accuracy of the hot spot location in the final determined wafer. Summary of the Invention

[0005] To address the issue of poor accuracy in hotspot detection using photolithography simulation, this application provides a hotspot detection method and apparatus.

[0006] The first aspect of this application provides a hotspot detection method, the method comprising:

[0007] Multiple first images, including circuit layouts, are stored in an image library. The circuit layouts contain hotspots, and the hotspots in at least two of the first images are different.

[0008] Determine a matching image in the image library that matches the image to be tested, wherein the first image includes the matching image, and the image to be tested includes a wafer for which hotspot detection is required;

[0009] Obtain the feature vector of the matched image;

[0010] The feature vector is input into a preset detection model, and the type and location of hot spots in the wafer are determined based on the output of the detection model.

[0011] In some possible implementations, the detection model includes at least one classifier, and after storing multiple first images including circuit layouts to an image library, the method further includes:

[0012] The first image is divided into at least one image set, and one image set corresponds to one classifier;

[0013] The step of inputting the feature vector into a preset detection model includes:

[0014] The feature vector is input into the classifier corresponding to the image set containing the matched image.

[0015] In some possible implementations, dividing the first image into at least one image set includes:

[0016] The first image is divided according to the type of the hotspot and / or based on the area surrounding the hotspot in the circuit layout.

[0017] In some possible implementations, before segmenting the first image according to the type of hotspot, the method further includes:

[0018] Based on the difference between the area of ​​the circuit layout and the actual area of ​​the circuit layout in the first image, the hotspot type in the first image is determined;

[0019] Wherein, if the difference between the area of ​​the circuit layout and the actual area is a first preset value, and the area of ​​the circuit layout is smaller than the actual area, the hot spot type is a wire break type;

[0020] If the difference between the area of ​​the circuit layout and the actual area is a second preset value, and the area of ​​the circuit layout is greater than the actual area, the hotspot type is a wire bridging type.

[0021] If the difference between the area of ​​the circuit layout and the actual area is a third preset value, and the hot spot type is a contact hole type, wherein the third preset value is less than the second preset value and the third preset value is less than the first preset value.

[0022] In some possible implementations, dividing the first image based on the region surrounding the hotspot in the circuit layout includes:

[0023] The first region is determined by cropping the area around the hot spot in the first image;

[0024] The first images that are identical in the first region are identified as images of the same type.

[0025] In some possible implementations, the feature vector includes at least two of the following parameters:

[0026] The shortest distance from the center of the hotspot to the edge of the matching image, the shortest straight distance from the center of the hotspot to the edge of the circuit layout, the longest distance from the center of the hotspot to the edge of the circuit layout, the shortest distance from the edge of the matching image to the edge of the circuit layout, the longest distance from the edge of the matching image to the edge of the circuit layout, and the shortest distance from the center of the hotspot to the four corners of the circuit layout.

[0027] In some possible implementations, obtaining the matching image from the image library that matches the image to be tested includes:

[0028] Determine the matching score between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested;

[0029] If the matching score is greater than the first threshold, the first image is determined to be the matching image.

[0030] In some possible implementations, the method further includes:

[0031] If there is no matching image in the image library that matches the image to be tested, the feature vector of the image to be tested is input into a preset detection model to obtain the detection image output by the detection model;

[0032] The detected image is saved to the image library.

[0033] In some possible implementations, the training process of the detection model includes:

[0034] Obtain multiple sample images and the feature vectors corresponding to each sample image, wherein the sample images contain the real hotspot locations and real hotspot types;

[0035] The feature vectors of each sample image are sequentially input into the detection model to be trained, and the detection model to be trained predicts the sample hotspot location and sample hotspot type in each sample image;

[0036] Based on the actual hotspot location, the sample hotspot location, the actual hotspot type, and the sample hotspot type, the detection model to be trained is iteratively trained, and after training, the detection model is obtained.

[0037] A second aspect of this application provides a hotspot detection device, the device comprising:

[0038] An image storage module is used to store multiple first images, including circuit layouts, into an image library, wherein the circuit layouts contain hotspots;

[0039] The image matching module is used to determine the matching image in the image library that matches the image to be tested. The first image includes the matching image, and the image to be tested includes a wafer that needs to be detected for hotspots.

[0040] A feature vector acquisition module is used to acquire the feature vector of the matching image;

[0041] Detection module: used to input the feature vector into a preset detection model, and determine the type and location of hot spots in the wafer based on the output of the detection model.

[0042] A third aspect of this application provides a hotspot detection system, the system comprising:

[0043] A memory and a processor, the memory storing a computer program, the processor retrieving and executing the computer program from the memory to implement the steps of the method described in any one of the first aspects above.

[0044] A fourth aspect of this application provides a computer-readable storage medium storing at least one computer program that, when executed by a processor, implements the steps of the method described in any one of the first aspects.

[0045] The technical solution provided in this application can achieve at least the following beneficial effects:

[0046] By storing multiple first images in an image library, it is easy to find matching images that match the image to be tested in the image library. The feature vector of the matching image is input into a preset detection model. The preset detection model detects the feature vector and determines the hot spots in the detection image. In the whole process, the feature vector of the matching image can be directly input into the preset detection model for detection. The preset detection model does not need to extract features from each image to be tested, which reduces the amount of data processing and makes the final detection result more accurate. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of the structure of a computer device shown in an exemplary embodiment of this application;

[0049] Figure 2 This is a schematic flowchart illustrating a hotspot detection method according to an exemplary embodiment of this application;

[0050] Figure 3 This is a schematic diagram illustrating the specific process of step 220 in a hotspot detection method according to an exemplary embodiment of this application;

[0051] Figure 4 This is a schematic diagram illustrating a hotspot according to an exemplary embodiment of this application;

[0052] Figure 5 This is a schematic diagram illustrating a hotspot according to an exemplary embodiment of this application;

[0053] Figure 6 This is a schematic diagram illustrating a hotspot according to an exemplary embodiment of this application;

[0054] Figure 7 This is a flowchart illustrating the first image segmentation step in a hotspot detection method according to an exemplary embodiment of this application;

[0055] Figure 8A This is a schematic diagram of a first image including a first region, as illustrated in an exemplary embodiment of this application;

[0056] Figure 8B This is a schematic diagram of a first image including a first region, as illustrated in an exemplary embodiment of this application;

[0057] Figure 8C This is a schematic diagram of a first image including a first region, as illustrated in an exemplary embodiment of this application;

[0058] Figure 8D This is a schematic diagram of a first image including a first region, as illustrated in an exemplary embodiment of this application;

[0059] Figure 9 This is a schematic diagram of parameters in a feature vector shown in an exemplary embodiment of this application;

[0060] Figure 10This is a schematic flowchart illustrating the training process of a detection model in a hotspot detection method according to an exemplary embodiment of this application;

[0061] Figure 11 This is a schematic diagram illustrating the entire process of a hotspot detection method according to an exemplary embodiment of this application;

[0062] Figure 12 This is a schematic diagram of the structure of a hotspot detection device shown in an exemplary embodiment of this application. Detailed Implementation

[0063] To make the objectives, implementation methods and advantages of this application clearer, the exemplary implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings of the exemplary embodiments of this application. Obviously, the exemplary embodiments described are only some embodiments of this application, and not all embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0064] It should be noted that the brief descriptions of terms in this application are only for the convenience of understanding the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise stated, these terms should be understood in their ordinary and common meaning.

[0065] The terms "first," "second," "third," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar or related objects or entities, and do not necessarily imply a specific order or sequence, unless otherwise specified. It should be understood that such terms are interchangeable where appropriate.

[0066] The terms “comprising” and “having”, and any variations thereof, are intended to cover but not exclude inclusion, for example, a product or device that includes a range of components is not necessarily limited to all of the components that are clearly listed, but may include other components that are not clearly listed or that are inherent to such product or device.

[0067] To facilitate a clear description of the technical solutions of the embodiments of this application, some concepts involved in this application will be explained first below.

[0068] (1) Wafer

[0069] A wafer is a silicon wafer used to fabricate silicon semiconductor circuits; its raw material is silicon. High-purity polycrystalline silicon is dissolved, doped with silicon crystal seeds, and then slowly pulled out to form a cylindrical single-crystal silicon wafer. After grinding, polishing, and slicing, the silicon crystal ingot is formed into a silicon wafer, or wafer. The wafer processing generally involves photolithography.

[0070] (2) Photolithography

[0071] Photolithography is a core step in integrated circuit manufacturing. It involves a series of processes to remove specific portions of the silicon wafer surface, transferring the designed circuit pattern to the various layers of material on the wafer, thus forming a stacked structure including devices. A typical photolithography system consists of four basic elements: a light source system, a photomask, a projection system, and a wafer. The photomask is the pattern template used in photolithography, carrying the designed circuit layout pattern; that is, the photomask solidifies the data of each layer of the circuit layout. During the photolithography process, the light waves emitted by the light source system pass through the transparent areas of the photomask, carrying the modulation information of the circuit layout. This creates areas of different exposure intensities on the wafer surface covered with photoresist. At this point, a photolithographic image pattern is formed on the wafer surface, and this image pattern corresponds to the circuit layout pattern.

[0072] It should be noted that, considering process errors, the consistency of the patterns mentioned here can be understood as the overlap between the photolithographic pattern and the circuit layout pattern being within a preset error range, not that they are required to be completely identical.

[0073] (3) Optical Proximity Effect (OPE)

[0074] With the development of semiconductor technology, the feature size of semiconductor devices is getting smaller and smaller. In the photolithography process, when the feature size of semiconductor devices is close to or even smaller than the wavelength of light used in the photolithography process, due to the diffraction and interference of light, there is a certain distortion between the photolithography pattern obtained on the actual wafer and the pattern on the mask. For example, the photolithography pattern on the wafer surface may have distortions such as shortened line ends, uneven line width, or rounded corners. This phenomenon is called the Optical Proximity Effect (OPE).

[0075] (4) Hot Topics

[0076] The area on a wafer that causes pattern distortion in photolithography due to the optical proximity effect and thus results in defects is called a photolithography hotspot, or simply a hotspot. Hotspots have a significant negative impact on the manufacturing yield of subsequent integrated circuits.

[0077] (5) Neural Network (ANN)

[0078] Neural networks are a subset of machine learning and are the core of deep learning algorithms. Their name and structure are inspired by the human brain, mimicking the way biological neurons transmit signals to each other.

[0079] A neural network consists of layers of nodes, each containing an input layer, one or more hidden layers, and an output layer. Each node, also called an artificial neuron, is connected to another node and has associated weights and a threshold. If the output of any single node exceeds a specified threshold, that node is activated and sends data to the next layer of the network. Otherwise, data is not passed to the next layer.

[0080] Next, the application scenarios and implementation environment of the embodiments of this application will be introduced.

[0081] During the transfer of circuit layout patterns onto a wafer, the optical proximity effect causes significant distortion in the lithographic pattern. To mitigate this, phase-shifting masks or optical proximity correction are commonly used to optimize the pattern structure. However, hot spots on the wafer remain unavoidable in subwavelength lithography, severely impacting the yield of subsequent integrated circuits. Therefore, hot spot detection is essential. Hot spot detection can be defined as accurately locating hot spots within an acceptable timeframe.

[0082] Currently, the main method for hotspot detection is based on photolithography simulation. Specifically, a series of initial process parameters are set for a photolithography model identical to the wafer to be lithographicated. These initial parameters are specific photolithography process parameters, such as exposure intensity and depth of focus. Then, the photolithography model is simulated, allowing it to mimic the photolithography process of the wafer under the same conditions. During this simulation, the initial process parameters are fitted multiple times, and the parameters are continuously adjusted based on the fitting results until the final simulation result is obtained. The simulation result is a virtual pattern on the wafer obtained after simulating the photolithography process. By comparing the simulation result with the actual pattern on the surface of the wafer obtained by photolithography based on the same circuit layout, the differences between the two are determined. The locations of these differences on the actual wafer are the locations of the hotspots.

[0083] In the process of determining the distinguishing points, hot spots that may have problems in the actual wafer can be screened out based on parameters such as the error in the edge position between the virtual pattern and the real pattern.

[0084] However, in the above process, the method of hot spot detection through photolithography simulation requires multiple fitting and correction of a large number of process parameters to obtain simulation results. The various parameters in this process are usually interrelated, so the process of continuous fitting and correction is prone to errors, resulting in inaccurate simulation results, and consequently, poor accuracy of the hot spot positions in the final determined wafer.

[0085] Therefore, to address the issue of poor accuracy in detecting hotspot locations, this application provides a hotspot detection method. This method involves defining an image library and conveniently obtaining matching images from the image library. The feature vectors of the matching images are then input into a preset detection model to obtain the hotspot type and location output by the preset detection model. In this process, by inputting the feature vectors of the matching images into the preset detection model for detection, the preset detection model does not need to extract features from each image, reducing the amount of data processing and resulting in a more accurate final detection result.

[0086] The hotspot detection method provided in this application can utilize computer equipment to store a trained detection model, and by running the detection model, determine the type and location of hotspots in the wafer based on the feature vector of the matched image.

[0087] In some embodiments, the structure of the computer device is as follows: Figure 1 As shown, the computer device 100 includes at least one processor 110, a memory 120, a communication bus 130, and at least one communication interface 140.

[0088] The processor 110 may be a general-purpose CPU, a network processor (NP), a microprocessor, or one or more integrated circuits for implementing the scheme of this application, such as an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof.

[0089] The aforementioned PLD can be a Complex Programmable Logic Device (CPLD), a Field-Programmable Gate Array (FPGA), a Generic Array Logic (GAL), or any combination thereof.

[0090] The computer device 100 may include multiple processors 110, and each processor 110 may include one or more CPUs. Each of these processors 110 may be a single-core processor or a multi-core processor.

[0091] It should be noted that the processor 110 here may refer to one or more devices, circuits and / or processing cores used to process data (such as computer program instructions).

[0092] The memory 120 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions; it may also be a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions; it may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0093] It should be noted that the memory 120 can exist independently and be connected to the processor 110 via the communication bus 130. Of course, the memory 120 can also be integrated with the processor 110.

[0094] The communication bus 130 is used to transfer information between components (such as between the processor and memory). The communication bus 130 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 1 The diagram uses only one communication bus, but this does not mean that there is only one bus or one type of bus.

[0095] The communication interface 140 is used for the computer device 100 to communicate with other devices or communication networks. The communication interface 140 includes a wired communication interface or a wireless communication interface. The wired communication interface can be, for example, an Ethernet interface. The Ethernet interface can be an optical interface, an electrical interface, or a combination thereof. The wireless communication interface can be a Wireless Local Area Network (WLAN) interface, a cellular network communication interface, or a combination thereof.

[0096] In some embodiments, the computer device 100 may further include output devices and input devices. Figure 1(Not shown in the image). The output device communicates with the processor 110 to display information in various ways; for example, the output device can be a liquid crystal display (LCD), a light-emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector. The input device communicates with the processor 110 to receive user input in various ways; for example, the input device can be a mouse, keyboard, touchscreen device, or sensor device.

[0097] In some embodiments, memory 120 may be used to store computer programs that execute the solutions of this application, and processor 110 may execute the computer programs stored in memory 120. For example, the computer device 100 may use processor 110 to call and execute the computer programs stored in memory 120 to implement the steps of the optical proximity correction method provided in the embodiments of this application.

[0098] It should be understood that the hotspot detection method provided in this application can also be applied to a hotspot detection device. This hotspot detection and correction device can be implemented as part or all of the processor through software, hardware, or a combination of software and hardware, and can be integrated into any computer device.

[0099] Next, the technical solutions of this application and how they solve the aforementioned technical problems will be described in detail through embodiments and in conjunction with the accompanying drawings. The embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application.

[0100] See Figure 2 This application provides a hotspot detection method, which can be applied to the above-mentioned... Figure 1 The computer device 100 shown is used as an example. The method may include the following steps:

[0101] Step 210: Store multiple first images, including circuit layouts, into an image library. The circuit layouts contain hotspots, and the hotspots in at least two of the first images are different.

[0102] In this process, conventional image acquisition equipment can be used to acquire images of multiple circuit layouts containing hotspots, thereby obtaining multiple first images.

[0103] The circuit layout included in the first image of this application can be a complete circuit layout of a single-layer process of an integrated circuit device, or it can be a circuit layout of each unit circuit in the circuit layout. The circuit layout in each first image can be the same or different, and the embodiments of this application do not limit this.

[0104] It should be noted that the hotspots in at least the two first images are of different types or in different locations, or both types and locations of the hotspots may be different.

[0105] In some possible implementations, the same image preprocessing can be performed on each first image, such as image noise reduction, to make the first image clearer.

[0106] Step 220: Determine the matching image in the image library that matches the image to be tested. The first image includes the matching image, and the image to be tested includes the wafer for which hotspot detection needs to be performed.

[0107] The image to be tested can also be acquired using conventional image acquisition equipment by photographing the wafer to be inspected for hotspots, and the same image preprocessing operations as the first image can be performed on the image to be tested. In determining the matching image, a region recognition algorithm in the image to be tested and a matching image search algorithm in the image library can be used, resulting in a highly accurate acquisition process for the matching image.

[0108] Specifically, in some possible implementations, such as Figure 3 As shown, the steps for determining the matching image specifically include:

[0109] Step 310: Determine the matching score between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested.

[0110] Step 320: If the matching score is greater than the first threshold, the first image is determined to be the matching image.

[0111] Specifically, by acquiring the features of the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested, and then calculating the matching score of the two features, the higher the matching score, the more consistent the features of the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested are, and the higher the similarity between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested.

[0112] Based on this, when the matching score between a first image in the image library and the image to be tested is greater than a first threshold, the first image can be identified as the matching image. The first threshold can be set according to actual needs, and its empirical value is generally between 0.96 and 1, with a preferred value of 0.98.

[0113] When the image library stores a large number of first images, such as 50 first images, some first images may have the same hotspot type and position. Therefore, it is tedious and time-consuming to calculate the matching score of each image with the image to be tested in turn. Therefore, in some possible implementations, in order to quickly complete the process of determining the matching image, the first images can be classified and then step 220 can be executed.

[0114] Specifically, after storing multiple first images, including circuit layouts, in an image library, the first images can be divided into at least one image set, that is, the multiple first images can be divided into several types according to specific classification criteria. When determining a matching image later, the matching image can be quickly identified from the determined types.

[0115] The first image can be divided according to the type of hotspot and / or based on the area surrounding the hotspot in the circuit layout.

[0116] When dividing the first image according to the type of hotspot, the following steps need to be performed before dividing the first image:

[0117] The hotspot type in the first image is determined based on the difference between the area of ​​the circuit layout and the actual area of ​​the circuit layout in the first image.

[0118] Specifically, if the difference between the circuit layout area and the actual area is a second preset value, and the circuit layout area is larger than the actual area, the hotspot type is a wire-bridged type. A wire-bridged hotspot is as follows: Figure 4 As shown in a and b, line bridging hotspots refer to situations where two photoresist lines, which should be separated, are not completely separated. This can also be described as the photoresist lines being "connected by thin threads." Such hotspots have a significant impact, completely altering the electrical properties, and in most cases, they result in short circuits.

[0119] If the difference between the circuit layout area and the actual area is a first preset value, and the circuit layout area is smaller than the actual area, the hotspot type is a wire-pinch type. A hotspot of the wire-pinch type is as follows: Figure 5 As shown in c and d, the hot spot of the line clamp type refers to the situation where the photoresist line is broken. The impact of this hot spot is mostly circuit breakage.

[0120] If the difference between the circuit layout area and the actual area is a third preset value, the hotspot type is a contact hole type, wherein the third preset value is less than the second preset value and less than the first preset value, and the hotspot of the contact hole type is as follows. Figure 6 As shown in 'e', ​​this type of hot spot is caused by incomplete contact holes and is very likely to cause a short circuit.

[0121] It should be noted that the values ​​of the first, second, and third preset values ​​are generally selected based on the actual scenario. As an example, the first preset value is 0.05, the second preset value is 0.06, and the third preset value is 0.03.

[0122] In some possible implementations, when dividing the first image based on the area surrounding a hot spot in the circuit layout, such as... Figure 7 As shown, perform the following steps:

[0123] Step 710: Determine the first region by cropping the area around the hot spot in the first image.

[0124] Step 720: Determine that the first images with the same first region belong to the same type of image set.

[0125] It should be noted that although the graphical structures of hot spots differ, the first image can still be divided by classifying the topological characteristics of the circuit layout. In the circuit layout, the topology describes the relationship between the hot spot and its surrounding area, without concern for the details of the entire circuit layout or the proportional relationships between different areas.

[0126] As an example, such as Figures 8A-8D As shown, Figures 8A-8D All are first images. Figures 8A-8D The region M in the diagram represents the topological structure containing the hotspot and its surrounding area; M is the first region. The first image can be classified based on the shape of this first region, such as by classifying the first region M. Figures 8A-8D Divided into two different groups, among which, Figure 8A and Figure 8B The first region M in the two is the same, therefore 8A and Figure 8B The images were identified as being of the same type; Figure 8C and Figure 8D The first region M in the two is the same, therefore 8C and Figure 8D The images were identified as being of the same type.

[0127] Understandably, when the first image is divided according to the type of hotspot and the area around the hotspot in the circuit layout, it is possible to more accurately divide the first image into different image sets, thereby further improving the convenience of the matching image process for obtaining the image to be tested in the later stage.

[0128] Step 230: Obtain the feature vector of the matching image.

[0129] Each matching image has its own attributes, and different attributes are represented by different attribute values. Multiple attribute values ​​combined can be represented by a vector, which is called a feature vector. In other words, a feature vector is a set of attributes. The feature vector of each matching image is attached to a matching image and is used to represent the matching image.

[0130] To accurately represent the matched images, feature vectors are often multi-dimensional vectors, such as two-dimensional, four-dimensional, or five-dimensional vectors, i.e. Figure 9 As shown, Figure 9 In the diagram, label A indicates the location of a hotspot in the matching image B. The feature vector may include at least two of the following parameters: the shortest distance from the center of the hotspot to the edge of the matching image (e.g., ...). Figure 9 In d1), the shortest straight-line distance from the center of the hot spot to the edge of the circuit layout (e.g., d1) Figure 9 d2), the longest distance from the center of the hot spot to the edge of the circuit layout (e.g., d2), Figure 9 In d3), the shortest distance from the edge of the matching image to the edge of the circuit layout (e.g., d3) Figure 9 In d4), the longest distance from the edge of the matching image to the edge of the circuit layout (e.g., d4) Figure 9 (d5) and the center of the hotspot to the circuit board Figure 4 The shortest distance between the corners (e.g.) Figure 9 (d6 in the middle).

[0131] As an example, when the parameters of the feature vector include all the above parameters, the feature vector is a six-dimensional parameter, which can be specifically represented as P = (d1, d2, d3, d4, d5, d6). In this case, the feature vector P can represent the entire matching image.

[0132] After determining the feature vector, a pre-trained detection model can be used to predict the image to be tested corresponding to the feature vector, so as to obtain the hotspot type and hotspot location in the image to be tested, i.e., the following step 240 is performed:

[0133] Step 240: Input the feature vector into the preset detection model, and determine the type and location of hot spots in the wafer based on the output of the detection model.

[0134] The preset detection model is trained using sample images of known hotspots; specifically, for example... Figure 10 As shown, the training process of the preset detection model includes:

[0135] Step 1010: Input the feature vectors of each sample image into the detection model to be trained in sequence, and predict the sample hotspot location and sample hotspot type in each sample image through the detection model to be trained.

[0136] The size information of the sample image may be the same as or different from that of the first image; this application embodiment does not impose any restrictions on this.

[0137] Step 1020: Based on the real hotspot location, sample hotspot location, real hotspot type, and sample hotspot type, iteratively train the detection model to be trained. After training, the detection model is obtained.

[0138] The detection model to be trained can be an Artificial Neural Network (ANN) detection model. An ANN is a complex network structure formed by a large number of interconnected processing units (neurons). It is an abstraction, simplification, and simulation of the structure and operation mechanism of the human brain, using mathematical models to simulate neuronal activity. It is an information processing system built upon imitating the structure and function of the brain's neural network. Artificial neural networks achieve the purpose of processing information and simulating the relationship between input and output by repeatedly learning and training on known information and gradually adjusting the connection weights of neurons.

[0139] In addition, the number of iterations can be set according to actual needs, such as 10 iterations, 13 iterations or 15 iterations. This application embodiment does not limit the specific type of detection model to be trained or the number of iterations.

[0140] In the above process, such as Figure 11 As shown, the initial detection model is trained using a rich dataset of known hotspot sample images. During training, the model extracts rules from the training dataset to obtain the target results (i.e., sample hotspot locations and sample hotspot types). Then, the parameters of the model are adjusted based on the differences between the actual hotspot locations and sample hotspot locations, and the differences between the actual hotspot types and sample hotspot types. Through repeated training, the errors between the actual hotspot locations and sample hotspot locations, and between the actual hotspot types and sample hotspot types, are all less than the preset deviation thresholds. At this point, training can be terminated, and the trained detection model, i.e., the preset detection model, is obtained. The preset detection model can accurately predict the hotspot type and location in the test image based on the input feature vector.

[0141] It should be noted that when the first image is divided into at least one image set, the preset detection model includes at least one classifier, and one classifier corresponds to one image set. When training the detection model to be trained, the feature vectors of the sample images are input into the detection model containing the corresponding classifiers. When inputting the feature vectors of the matching images into the preset detection model, the feature vectors are input into the classifier corresponding to the image set containing the matching image. At this point, different types of sample images are treated as separate classification problems by their respective classifiers. Multiple classifiers in the preset detection model are combined into a combined classifier, and the result of the combined classifier represents the result of the entire preset detection model.

[0142] Understandably, when the preset detection model needs to identify new hotspots, adjustments can be made simply by adding or updating the classifier in the preset detection model, making the whole process simple and convenient.

[0143] In some possible implementations, since the circuit layout patterns in the first image are mostly patterns of various shapes, and hotspots only occupy a small portion of the pattern in the first image, in order to further improve the accuracy of the preset detection model, a certain amount of non-hotspot data can be input into the detection model to be trained during training. Specifically, the same number of second images as the first images are input into the detection model to be trained. The second images are images that do not contain hotspots, but contain circuit layouts. The circuit layout patterns in the first image and the circuit layout patterns in the second image can be the same or different. At this time, the ratio of hotspots to non-hotspots in the training dataset is relatively balanced, and the detection model to be trained can fully learn and identify hotspots and non-hotspots (i.e., background), improving the accuracy of the detection model during the training phase. This allows the preset detection model to accurately detect the type and location of hotspots in the image under test.

[0144] like Figure 11 As shown, based on the above technical solution, this application divides multiple first images into different types of image sets and stores them in an image library. This facilitates the rapid retrieval of matching images that correspond to the image to be tested within the image library. The feature vectors of the matching images are input into a preset detection model. The preset detection model detects the feature vectors and determines the hotspot locations in the detection image. Throughout the process, the feature vectors of the matching images can be directly input into the preset detection model for detection. The preset detection model does not need to extract features from each image to be tested, reducing the amount of data processing and resulting in more accurate detection results. Compared with the method of detecting hotspots based on photolithography simulation, the hotspot detection scheme of this application, while ensuring high accuracy, has low computational complexity and consumes less time.

[0145] Since the number of first images in the image library is limited, and the types of first images are also limited, there are cases where no matching image exists in the image library. Therefore, in some possible implementations, such as... Figure 11 As shown, if no matching image exists in the image library, the feature vector of the image to be tested is input into a preset detection model to obtain the detection image output by the detection model; the detection image is then saved to the image library. During this process, the detection model can identify the type and location of hotspots in the image to be tested based on the feature vector, and store the image to be tested as the first image in the image library to increase the comprehensiveness of the image library, thereby facilitating subsequent hotspot detection.

[0146] In this embodiment, by combining the process of determining a matching image with a preset detection model, when a matching image exists, the feature vector of the matching image is input into the preset detection model. In this case, the preset detection model no longer needs to perform complex feature extraction on the image under test, greatly simplifying the computational processing of the matching image. When no matching image exists, the preset detection model can effectively determine the hotspot location and type in the image under test based on the feature vector of the image under test. That is, the solution in this application has good predictive ability even when no matching image exists, and has a wide range of application scenarios.

[0147] Based on the aforementioned hotspot detection method and employing the same technical concept, this application also provides a hotspot detection device for implementing the aforementioned hotspot detection method. The solution provided by this device is similar to the implementation scheme described in the above method embodiments.

[0148] In one exemplary embodiment, such as Figure 12 As shown, the hotspot detection device in this application includes:

[0149] The image storage module is used to store multiple first images, including circuit layouts, into an image library. The circuit layouts contain hotspots.

[0150] The image matching module is used to determine the matching image in the image library that matches the image to be tested. The first image includes the matching image, and the image to be tested includes the wafer that needs to be detected for hotspots.

[0151] The feature vector acquisition module is used to acquire the feature vectors of the matching image;

[0152] Detection module: Used to input feature vectors into a preset detection model, and determine the type and location of hot spots in the wafer based on the output of the detection model.

[0153] It is understood that the feature vector includes at least two of the following parameters: the shortest distance from the center of the hotspot to the edge of the matching image, the shortest straight-line distance from the center of the hotspot to the edge of the circuit layout, the longest distance from the center of the hotspot to the edge of the circuit layout, the shortest distance from the edge of the matching image to the edge of the circuit layout, the longest distance from the edge of the matching image to the edge of the circuit layout, and the shortest distance from the center of the hotspot to the four corners of the circuit layout.

[0154] It should be noted that the training process of the preset detection model includes:

[0155] Obtain multiple sample images and the feature vectors corresponding to each sample image. The sample images contain the real hotspot locations and real hotspot types.

[0156] The feature vectors of each sample image are sequentially input into the detection model to be trained, and the detection model to be trained predicts the sample hotspot location and sample hotspot type in each sample image;

[0157] Based on the real hotspot location, sample hotspot location, real hotspot type, and sample hotspot type, the detection model to be trained is iteratively trained, and after training, the preset detection model is obtained.

[0158] In some possible implementations, the preset detection model includes at least one classifier. After storing multiple first images, including circuit layouts, in an image library, the device further includes an image segmentation module, which is used for:

[0159] The first image is divided into at least one image set, and each image set corresponds to a classifier. At this time, the feature vector is input into the preset detection model. The detection module includes a detection unit, which is used to input the feature vector into the classifier corresponding to the image set where the matching image is located.

[0160] In some possible implementations, when dividing the first image into at least one image set, the partitioning module includes a partitioning unit, which is used for:

[0161] The first image is divided based on the type of hotspot and / or based on the area surrounding the hotspot in the circuit layout.

[0162] Specifically, before dividing the first image according to the type of hotspot, the dividing unit is used to determine the type of hotspot in the first image based on the difference between the area of ​​the circuit layout and the actual area of ​​the circuit layout in the first image;

[0163] If the difference between the area of ​​the circuit layout and the actual area is a first preset value, and the area of ​​the circuit layout is smaller than the actual area, the hot spot type is the wire clamp type.

[0164] If the difference between the area of ​​the circuit layout and the actual area is the second preset value, and the area of ​​the circuit layout is greater than the actual area, the hotspot type is the wire bridging type.

[0165] If the difference between the area of ​​the circuit layout and the actual area is a third preset value, the hot spot type is a contact hole type, wherein the third preset value is less than the second preset value and the third preset value is less than the first preset value.

[0166] Based on the area around the hot spot in the circuit layout, the first image is divided. The division unit is used to determine the first region by cropping the area around the hot spot in the first image. First images with the same first region are determined to be the same type of image set.

[0167] The image matching module retrieves matching images from the image library that match the image to be tested. The image matching module includes an image matching unit, which is used for:

[0168] Determine the matching score between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested; if the matching score is greater than a first threshold, determine the first image as the matching image.

[0169] In some possible implementations, if there is no matching image in the image library that matches the image to be tested, the detection module includes a direct detection unit. The direct detection unit is used to input the feature vector of the image to be tested into a preset detection model, obtain the detection image output by the detection model, and save the detection image to the image library.

[0170] It should be noted that the specific limitations on hotspot detection devices can be found in the limitations on hotspot detection methods mentioned above, and will not be repeated here.

[0171] It should be noted that each module in the aforementioned hotspot detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0172] Furthermore, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above hotspot detection method.

[0173] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0174] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the invention patent. It should be noted that any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without departing from the concept of this application should be included within the protection scope of the embodiments of this application.

Claims

1. A hotspot detection method, characterized in that, include: Multiple first images, including circuit layouts, are stored in an image library. The circuit layouts contain hotspots, and the hotspots in at least two of the first images are different. Determine a matching image in the image library that matches the image to be tested, wherein the first image includes the matching image, and the image to be tested includes a wafer for which hotspot detection is required; The step of determining the matching image in the image library that matches the image to be tested includes: Determine the matching score between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested; If the matching score is greater than the first threshold, the first image is determined to be the matching image; Obtain the feature vector of the matched image; The feature vector is input into a preset detection model, and the type and location of hot spots in the wafer are determined based on the output of the detection model. The detection model includes at least one classifier, and after storing multiple first images including circuit layouts into an image library, the method further includes: The first image is divided into at least one image set, and one image set corresponds to one classifier; The step of inputting the feature vector into a preset detection model includes: The feature vector is input into the classifier corresponding to the image set containing the matched image; The step of dividing the first image into at least one image set includes: The first image is divided according to the type of the hotspot and / or based on the area surrounding the hotspot in the circuit layout.

2. The hotspot detection method according to claim 1, characterized in that, Before segmenting the first image according to the type of hotspot, the method further includes: Based on the difference between the area of ​​the circuit layout and the actual area of ​​the circuit layout in the first image, the hotspot type in the first image is determined; Wherein, if the difference between the area of ​​the circuit layout and the actual area is a first preset value, and the area of ​​the circuit layout is smaller than the actual area, the hot spot type is a wire break type; If the difference between the area of ​​the circuit layout and the actual area is a second preset value, and the area of ​​the circuit layout is greater than the actual area, the hotspot type is a wire bridging type. If the difference between the area of ​​the circuit layout and the actual area is a third preset value, and the hot spot type is a contact hole type, wherein the third preset value is less than the second preset value and the third preset value is less than the first preset value.

3. The hotspot detection method according to claim 1, characterized in that, The step of dividing the first image based on the area surrounding the hotspot in the circuit layout includes: The first region is determined by cropping the area around the hot spot in the first image; The first images that are identical in the first region are identified as images of the same type.

4. The hotspot detection method according to claim 1, characterized in that, The feature vector includes at least two of the following parameters: The shortest distance from the center of the hotspot to the edge of the matching image, the shortest straight distance from the center of the hotspot to the edge of the circuit layout, the longest distance from the center of the hotspot to the edge of the circuit layout, the shortest distance from the edge of the matching image to the edge of the circuit layout, the longest distance from the edge of the matching image to the edge of the circuit layout, and the shortest distance from the center of the hotspot to the four corners of the circuit layout.

5. The hotspot detection method according to claim 1, characterized in that, Also includes: If there is no matching image in the image library that matches the image to be tested, the feature vector of the image to be tested is input into a preset detection model to obtain the detection image output by the detection model; The detected image is saved to the image library.

6. The hotspot detection method according to any one of claims 1-5, characterized in that, The training process of the detection model includes: Obtain multiple sample images and the feature vectors corresponding to each sample image, wherein the sample images contain the real hotspot locations and real hotspot types; The feature vectors of each sample image are sequentially input into the detection model to be trained, and the detection model to be trained predicts the sample hotspot location and sample hotspot type in each sample image; Based on the actual hotspot location, the sample hotspot location, the actual hotspot type, and the sample hotspot type, the detection model to be trained is iteratively trained, and after training, the detection model is obtained.

7. A hotspot detection device, characterized in that, include: An image storage module is used to store multiple first images, including circuit layouts, into an image library, wherein the circuit layouts contain hotspots; The image matching module is used to determine the matching image in the image library that matches the image to be tested. The first image includes the matching image, and the image to be tested includes a wafer that needs to be detected for hotspots. The step of determining the matching image in the image library that matches the image to be tested includes: Determine the matching score between the circuit layout pattern in the first image and the circuit layout pattern in the image to be tested; If the matching score is greater than the first threshold, the first image is determined to be the matching image; A feature vector acquisition module is used to acquire the feature vector of the matching image; Detection module: used to input the feature vector into a preset detection model, and determine the type and location of hot spots in the wafer based on the output of the detection model; The detection model includes at least one classifier, and after storing multiple first images including circuit layouts into the image library, it further includes: The first image is divided into at least one image set, and one image set corresponds to one classifier; The step of inputting the feature vector into a preset detection model includes: The feature vector is input into the classifier corresponding to the image set containing the matched image; The step of dividing the first image into at least one image set includes: The first image is divided according to the type of the hotspot and / or based on the area surrounding the hotspot in the circuit layout.

Images