Method and device for generating SEM image

Abstract

The invention discloses a method and a device for generating an SEM (scanning electron microscope) image, which fully utilize structured information of simulation data and combine with an image processing technology to obtain an SEM image which takes photoetching simulation as a core and is based on layout design data, ensure that the simulation image has high authenticity and flexible adjustability, save time and labor and improve the efficiency of generating the SEM image. And the requirements of high efficiency and automation are met.

Description

Technical Field

[0001] The present application relates to but is not limited to semiconductor integrated circuit technology, and in particular to a method and device for generating a SEM image based on a simulated lithography process. Background Art

[0002] As chip manufacturing processes continue to shrink, the optical proximity effect (OPE) has become one of the main challenges in semiconductor manufacturing. OPE can cause deformation and distortion of patterns during the photolithography process, which directly affects the accuracy of parameters such as the critical dimension (CD) on the wafer. Whether in the R&D stage of new processes or in verifying manufacturing processes, it is particularly important to quickly and accurately obtain key indicators after photolithography exposure.

[0003] Critical dimension scanning electron microscope (CD-SEM) is an important tool for measuring CD values ​​in semiconductor manufacturing. Modern CD-SEM equipment combines high-resolution imaging technology and intelligent image comparison algorithms. Through the automated analysis of SEM images, it has achieved a transition from traditional manual measurement to semi-automated or even fully automated. This automation capability significantly saves labor costs while improving detection efficiency and accuracy. Although modern CD-SEM machines have semi-automatic or automated capabilities, the reliance on a large amount of real wafer sampling data for model training and parameter optimization has led to high data acquisition costs; moreover, CD-SEM equipment needs to focus and adjust imaging parameters in specific areas, and cannot be previewed in real time like ordinary cameras. It needs to rely on imaging results for repeated adjustments, which is time-consuming and labor-intensive. In addition, the automated algorithm has limited adaptability to new processes or abnormal defects. When faced with different process conditions, it is often necessary to readjust or retrain the model. In addition, automation based on traditional image comparison methods cannot make fine adjustments to local brightness, noise or specific defects, which reduces the model's generalization ability. Summary of the invention

[0004] The present application provides a method and device for generating SEM images, which saves time and effort and can meet the needs of efficient automation.

[0005] An embodiment of the present invention provides a method for generating a SEM image, comprising: Inputting the design layout and the lithography simulation model into the lithography simulation engine to generate a first simulation result; Extracting contour edge information and generating white edges according to the first simulation result, and acquiring a simulation image with white edges according to the generated white edges and the simulation image in the first simulation result; The masks of the inner and outer regions of the defined pattern are determined according to the contour information in the first simulation result, and the brightness is adjusted on the simulation image with white edges according to the inner and outer regions to obtain a SEM image.

[0006] In an exemplary embodiment, the lithography simulation model is input into the lithography simulation engine via a parameter configuration file, a script interface, a model library, or an experimental data calibration result; The design layout is used as a geometric input and loaded into the lithography simulation engine through a layout analysis tool, and drives the optical simulation engine together with the lithography simulation model to generate the first simulation result including a simulation image Image and contour Contour information.

[0007] In an exemplary embodiment, the step of acquiring a simulated image with a white edge includes: Extracting the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge; Convolve the captured contour edge information with the preset convolution kernel to generate a white edge; The generated white edge is superimposed on the simulated image in the first simulation result to form the simulated image with the white edge.

[0008] In an exemplary embodiment, extracting the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge includes: A first-order derivative is performed on the simulation image in the first simulation result to extract the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge.

[0009] In an exemplary embodiment, the convolution kernel includes: a Gaussian kernel or a difference kernel.

[0010] In an exemplary embodiment, the step of superimposing the generated white edge on the simulated image in the first simulation result further includes: controlling the intensity of the white edge by adjusting a coefficient.

[0011] In an exemplary embodiment, the step of determining a mask of inner and outer regions of the defined pattern according to the contour information in the first simulation result includes: The contour information in the first simulation result is used to divide the inner area and the outer area of ​​the pattern, and generate two binary masks; the masks include: an inner mask and an outer mask.

[0012] In an exemplary embodiment, the method further includes: superimposing noise on the SEM image to simulate the noise generated during a real SEM imaging process.

[0013] An embodiment of the present application further provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute any of the above methods for generating a SEM image.

[0014] An embodiment of the present application further provides a computer device, including a memory and a processor, wherein the memory stores the following instructions executable by the processor: for executing the steps of any of the above-mentioned methods for generating a SEM image.

[0015] The embodiment of the present application also provides a device for generating a SEM image, comprising: a generating module, a first processing module, and a second processing module; wherein: A generation module, used for inputting the design layout and the lithography simulation model into the lithography simulation engine to generate a first simulation result; A first processing module, used for extracting contour edge information and generating white edges according to the first simulation result, and obtaining a simulation image with white edges according to the generated white edges and the simulation image in the first simulation result; The second processing module is used to determine the mask of the inner and outer areas of the defined pattern according to the contour information in the first simulation result, and to adjust the brightness on the simulation image with white edges according to the inner and outer areas to obtain a SEM image.

[0016] In an exemplary embodiment, the second processing module is further used to: perform noise superposition on the SEM image to simulate the noise generated in a real SEM imaging process.

[0017] The method for generating SEM images provided in the embodiment of the present application utilizes the structured information of the simulation data, and combines the image processing technology to obtain a SEM image based on the layout design data with lithography simulation as the core, ensuring that the simulation image has both high authenticity and flexible adjustability. The method for generating SEM images provided in the embodiment of the present application saves time and effort and meets the needs of efficient automation. The method for generating SEM images provided in the embodiment of the present application fully considers the possible adjustment needs, so that the SEM image can be close to the real image taken using the SEM machine, providing a strong guarantee for the automated shooting of the SEM.

[0018] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide further understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application and do not constitute a limitation on the technical solution of the present application.

[0020] Figure 1 A schematic diagram of a process for generating a SEM image in an embodiment of the present application; Figure 2 Schematic diagram of the composition structure of the device for generating SEM images in the embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present application more clear, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other arbitrarily without conflict.

[0022] In a typical configuration of the present application, a computing device includes one or more processors (CPU), an input / output interface, a network interface, and a memory.

[0023] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0024] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0025] The steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. Also, although a logical sequence is shown in the flowchart, in some cases, the steps shown or described can be performed in a sequence different from that shown here.

[0026] When using traditional manual methods to obtain SEM images, the operation process is relatively complicated. The SEM machine needs to set the focus and other parameters in non-core areas, and these steps often require repeated manual adjustments. A significant difference between SEM and ordinary "cameras" is that SEM does not have a viewfinder to preview the image, and can only judge whether the set parameters are reasonable based on the results after imaging. Therefore, this method is not only time-consuming and labor-intensive, but also difficult to meet the needs of efficient automation, especially in large-scale wafer inspection scenarios.

[0027] If artificial intelligence (AI) or neural network methods are used to solve these problems, the quality and quantity of real data are required to be high. AI-generated SEM images usually require a large amount of labeled data for training, but even so, there are some practical challenges. This is because AI methods have weak direct control over the generation process. For example, it is difficult to flexibly adjust details such as brightness and noise in local areas. Moreover, the generated SEM images may lack realism, and the degree of freedom of the generated results is relatively low, which cannot be well adapted to certain specific usage scenarios such as those that require higher authenticity and detail fidelity.

[0028] Although some current solutions can generate SEM images, they are difficult to meet the needs of complex scenes due to their limited degrees of freedom and insufficient image quality. Therefore, the present application provides a method for generating SEM images, aiming to find a balance between the authenticity and adaptability of the generated images, saving time and effort, and meeting the needs of efficient automation to promote further development in this field.

[0029] Figure 1 Schematic diagram of the process of generating SEM images in the embodiment of the present application, such as Figure 1 As shown, it may include: Step 100: Input the design layout and the lithography simulation model into a lithography simulation engine to generate a first simulation result.

[0030] In an exemplary embodiment, the lithography simulation model is established based on the actual lithography machine parameters and physical models, which usually include light source parameters (such as wavelength, numerical aperture), objective lens characteristics and photoresist response characteristics. The lithography simulation model can be obtained through experimental measurement, optimization fitting or from an existing process database, and calibrated with experimental data to ensure that the simulation results are consistent with the actual lithography effect. In one embodiment, the input of the lithography simulation model is not provided in the form of a single data, but consists of multiple process parameters and physical models, which are passed to the lithography simulation engine in the form of a configuration file, a parameter interface or a model library.

[0031] In an exemplary embodiment, the design layout defines the two-dimensional geometric structure of the chip (such as polygonal graphics, line width, etc.), and is usually in a standard format, such as a GDSII file or an OASIS file.

[0032] In an exemplary embodiment, a lithography simulation engine is a core computing module used to simulate the interaction between light and materials during semiconductor lithography. The lithography simulation engine is based on physical optics, diffraction theory, photoresist chemical reaction models, etc., and can predict and analyze the pattern formation process on the wafer after exposure and development. By inputting the design layout and lithography simulation model, the lithography simulation engine can output key results such as exposure images, contour boundaries, critical dimensions (CD), etc., to help engineers evaluate the performance of lithography processes without the need for actual wafers. Commonly used lithography simulation software and platforms may include but are not limited to: PROLITH (ASML / Cymer), Panoramic Technology, Sentaurus Lithography (Synopsys), Dr.LiTHO (Fraunhofer), etc.

[0033] In one embodiment, the lithography simulation model may include multiple sub-models such as an optical model, a chemical reaction model, and a process deviation model. These models are input to the lithography simulation engine in the form of parameter configuration files, script interfaces, model libraries, or experimental data calibration results. In one embodiment, the design layout is used as a geometric input and loaded into the lithography simulation engine through a layout parsing tool, and the optical simulation engine is driven together with the lithography simulation model to ultimately generate output results such as simulated images and contour information. The modular input method in the embodiment of the present application ensures the flexibility and scalability of the lithography simulation engine and adapts to different process nodes and application scenarios.

[0034] Through step 100, the design layout and the lithography simulation model are taken as inputs and input into the lithography simulation engine. The lithography simulation engine simulates the lithography exposure process based on optical propagation theory (such as optical transfer function OTF, diffraction model, etc.), generates a light intensity distribution map for simulating exposure, an Image reflecting the grayscale characteristics of the actual lithography pattern, and a Contour of the boundary extracted from the lithography image, which is used to define the position and shape of the CD.

[0035] Step 100 is adapted to high-precision SEM image generation by cleverly using the design layout and lithography simulation model as input information of the lithography simulation engine, so that the lithography simulation engine can generate high-quality SEM images without real wafer data, thereby reducing data dependency costs; moreover, by adding real defects, noise, brightness and other parameter adjustments during the simulation stage, the image authenticity and flexibility are improved; thereby providing controllable, high-fidelity simulation data for AI model training, thereby promoting detection automation.

[0036] Step 101: extracting contour edge information and generating white edges according to the first simulation result, and acquiring a simulated image with white edges according to the generated white edges and the simulated image in the first simulation result.

[0037] In an exemplary embodiment, step 101 may include: Extracting the gray gradient change of the simulation image in the first simulation result to capture the information of the contour edge; Convolve the captured contour edge information with the preset convolution kernel to generate a white edge; The generated white edge is superimposed on the simulation image in the first simulation result to form a simulation image with a white edge.

[0038] In one embodiment, extracting the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge may include: A first-order derivative is taken on the simulation image in the first simulation result, and a grayscale gradient change of the simulation image in the first simulation result is extracted to capture information on the contour edge.

[0039] In one embodiment, convolving the captured contour edge information with a preset convolution kernel to generate a white edge may include: According to the preset convolution kernel, the convolution kernel is convolved with the captured contour edge information to control the width of the white edge. The white edge includes the inner white edge and the outer white edge. The inner white edge is used to adjust the direction of the convolution kernel so that the white edge expands into the inside of the pattern, and the outer white edge is used to adjust the direction of the convolution kernel so that the white edge expands out of the pattern.

[0040] In one embodiment, a convolution kernel is a mathematical tool for processing image edges, which can enhance or smooth edge areas. Commonly used convolution kernels may include: Gaussian kernel, which is used to blur and smooth edges and generate soft white edges, and its parameters (such as standard deviation σ) can control the width and blur of the white edges. Difference kernel, which is used to emphasize the gradient changes of edges and generate clearer and sharper white edges. In the embodiment of the present application, by adjusting the size and weight distribution of the convolution kernel, the width, brightness gradient and blur of the white edges can be controlled.

[0041] In the embodiment of the present application, the generated white edge is superimposed on the simulated image in the first simulation result, so that the simulated image forms a realistic edge characteristic. In one embodiment, during the superposition, the strength of the white edge can be controlled by adjusting the coefficient. The embodiment of the present application combines the physical model (Contour) and the image processing technology (convolution), and provides flexible parameter adjustment capabilities while ensuring realism.

[0042] In the embodiment of the present application, the Contour in the first simulation result is used to locate the boundary and generate the mask of the inner and outer areas. Through the Contour, we can know where the boundary is, but we cannot know how the brightness on both sides of the boundary changes. However, through the first-order derivative result of step 101, the grayscale gradient is captured, the grayscale change at the edge is extracted, and a real brightness gradient effect is generated, thereby generating a realistic blur or transition effect in the edge area. Finally, by combining the two, the accuracy of the image structure is guaranteed, and the realism of the edge area is enhanced.

[0043] Step 102: determining masks of inner and outer regions of the defined pattern according to the contour information in the first simulation result, and adjusting the brightness on the simulation image with white edges according to the inner and outer regions to obtain a SEM image.

[0044] In an exemplary embodiment, determining a mask of inner and outer regions of the defined pattern according to the contour information in the first simulation result may include: Use the Contour in the first simulation result to divide the inner area and the outer area of ​​the pattern, and generate two binary masks, namely the inner Mask, which represents the inner area of ​​the pattern (for example, the value is 1, and the other areas are 0), and the outer Mask, which represents the background area (that is, the outside of the Contour, for example, the value is 1, and the inside is 0).

[0045] In one embodiment, a contour extraction algorithm (such as findContours of OpenCV) may be used to determine the pattern boundary, and then a polygon fill algorithm (Polygon Fill) or a scan line algorithm is used to distinguish the inner and outer areas of the boundary to generate a binary mask.

[0046] In an exemplary embodiment, different brightness gain coefficients can be used to adjust the brightness on the simulated image with white edges according to the masks of the inner and outer areas. In one embodiment, by adjusting the internal brightness of the inner area, the brightness can be increased or decreased to simulate the contrast effect of the material, and by adjusting the external brightness of the outer area, the background brightness can be appropriately adjusted to enhance the contrast with the internal pattern.

[0047] Furthermore, in an exemplary embodiment, step 102 may also include: Noise overlay was performed on the SEM images.

[0048] In one embodiment, the noise source may be real noise data input, such as noise data from an actual machine, and these real noise data are directly used to analyze the noise distribution characteristics (such as Gaussian noise and Poisson noise). In one embodiment, the noise source may be a preset noise pattern, such as applying a common noise model (such as Gaussian noise, salt and pepper noise, etc.) for simulation.

[0049] Through noise superposition, the noise generated in the real SEM imaging process is simulated, such as electron beam noise and random noise caused by equipment vibration.

[0050] In one embodiment, superimposing noise on the simulated image after brightness adjustment may include superimposing noise of different intensities on the inner region and the outer region respectively, so as to simulate the noise difference between the pattern region and the background region in the real SEM image.

[0051] In the embodiment of the present application, SEM images are generated based on layout design data with lithography simulation as the core. On the one hand, the layout of each node (i.e., process node) can be adapted. Since the simulation method has good scalability, it can handle various layout designs from mature processes to advanced processes. On the other hand, the field of view (FOV) can be adjusted, that is, the field of view of simulation or imaging can be flexibly adjusted to meet simulation requirements of different accuracy and range. This flexibility improves the versatility and adaptability of the simulation tool, and is suitable for different process nodes and application scenarios.

[0052] The method for generating SEM images provided in the embodiment of the present application utilizes the structured information of the simulation data, and at the same time combines the image processing technology to obtain a SEM image based on the layout design data with lithography simulation as the core, ensuring that the simulation image has both high authenticity and flexible adjustability. The method for generating SEM images provided in the embodiment of the present application saves time and effort and meets the needs of efficient automation. The method for generating SEM images provided in the embodiment of the present application fully considers the possible adjustment needs, so that the SEM image can be close to the real image taken using the SEM machine, providing a strong guarantee for the automated shooting of the SEM.

[0053] The embodiment of the present application does not require data collection in advance, and generates SEM images based on the simulated lithography process, which can ensure authenticity without the need for data collection in advance, and because the simulation model can be calibrated for the data, and real noise, real defects, etc. can be added during the simulation process, so that authenticity is guaranteed. The embodiment of the present application also opens many configurable parameters, which can fine-tune the brightness, edge bandwidth, etc.

[0054] The present application also provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute any of the above methods for generating a SEM image.

[0055] The present application further provides a computer device, comprising a memory and a processor, wherein the memory stores the following instructions executable by the processor: used to execute the steps of any of the above-mentioned methods for generating a SEM image.

[0056] Figure 2 Schematic diagram of the composition structure of the device for generating SEM images in the embodiment of the present application, such as Figure 2 As shown, it may include: a generating module, a first processing module, and a second processing module; wherein, A generation module, used for inputting the design layout and the lithography simulation model into the lithography simulation engine to generate a first simulation result; A first processing module, used for extracting contour edge information and generating white edges according to the first simulation result, and obtaining a simulation image with white edges according to the generated white edges and the simulation image in the first simulation result; The second processing module is used to determine the mask of the inner and outer areas of the defined pattern according to the contour information in the first simulation result, and to adjust the brightness on the simulation image with white edges according to the inner and outer areas to obtain a SEM image.

[0057] In an exemplary embodiment, the second processing module is further used to: perform noise superposition on the SEM image to simulate the noise generated in a real SEM imaging process.

[0058] The device for generating SEM images provided in the embodiment of the present application utilizes the structured information of the simulation data, and combines the image processing technology to obtain a SEM image based on the layout design data with lithography simulation as the core, ensuring that the simulation image has both high authenticity and flexible adjustability. The method for generating SEM images provided in the embodiment of the present application saves time and effort and meets the needs of efficient automation. The method for generating SEM images provided in the embodiment of the present application fully considers the possible adjustment needs, so that the SEM image can be close to the real image taken using the SEM machine, providing a strong guarantee for the automated shooting of the SEM.

[0059] Although the embodiments disclosed in this application are as above, the contents described are only embodiments adopted to facilitate understanding of this application and are not intended to limit this application. Any technician in the field to which this application belongs can make any modifications and changes in the form and details of implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be based on the scope defined in the attached claims.

Claims

1. A method for generating a SEM image, characterized in that: include: Inputting the design layout and the lithography simulation model into the lithography simulation engine to generate a first simulation result; Extracting contour edge information and generating white edges according to the first simulation result, and acquiring a simulation image with white edges according to the generated white edges and the simulation image in the first simulation result; The masks of the inner and outer regions of the defined pattern are determined according to the contour information in the first simulation result, and the brightness is adjusted on the simulation image with white edges according to the inner and outer regions to obtain a SEM image.

2. The method according to claim 1, wherein: The lithography simulation model is input into the lithography simulation engine through a parameter configuration file, a script interface, a model library or an experimental data calibration result; The design layout is used as a geometric input and loaded into the lithography simulation engine through a layout analysis tool, and drives the optical simulation engine together with the lithography simulation model to generate the first simulation result including a simulation image Image and contour Contour information.

3. The method according to claim 1, wherein: The step of obtaining a simulated image with a white edge comprises: Extracting the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge; Convolve the captured contour edge information with the preset convolution kernel to generate a white edge; The generated white edge is superimposed on the simulated image in the first simulation result to form the simulated image with the white edge.

4. The method according to claim 3, wherein: The step of extracting the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge includes: A first-order derivative is performed on the simulation image in the first simulation result to extract the grayscale gradient change of the simulation image in the first simulation result to capture the information of the contour edge.

5. The method according to claim 3, wherein: The convolution kernel includes: a Gaussian kernel or a difference kernel.

6. The method according to claim 3, wherein: The step of superimposing the generated white edge on the simulated image in the first simulation result further includes: controlling the intensity of the white edge by adjusting a coefficient.

7. The method according to claim 1, wherein: The step of determining the mask of the inner and outer areas of the defined pattern according to the contour information in the first simulation result includes: The contour information in the first simulation result is used to divide the inner area and the outer area of ​​the pattern, and generate two binary masks; the masks include: an inner mask and an outer mask.

8. The method according to claim 1 or 7, wherein: Also includes: Noise is superimposed on the SEM image to simulate the noise generated during the actual SEM imaging process.

9. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the method for generating a SEM image according to any one of claims 1 to 8.

10. A computer device comprising a memory and a processor, wherein: The memory stores the following instructions executable by the processor: used to execute the steps of the method for generating a SEM image as described in any one of claims 1 to 8.

11. A device for generating a SEM image, characterized in that: include: A generating module, a first processing module, and a second processing module; wherein, A generation module, used for inputting the design layout and the lithography simulation model into the lithography simulation engine to generate a first simulation result; A first processing module, used for extracting contour edge information and generating white edges according to the first simulation result, and obtaining a simulation image with white edges according to the generated white edges and the simulation image in the first simulation result; The second processing module is used to determine the mask of the inner and outer areas of the defined pattern according to the contour information in the first simulation result, and to adjust the brightness on the simulation image with white edges according to the inner and outer areas to obtain a SEM image.

12. The device according to claim 11, wherein the second processing module is further used to: perform noise superposition on the SEM image to simulate the noise generated in a real SEM imaging process.

Images