A method, device, equipment and medium for optimizing a layout lithography process window

Abstract

The application provides a layout photolithography process window optimization method, device, equipment and medium, can divide the original layout to obtain a plurality of first edition picture segments; the first edition picture segment is optimized by using a process window optimization algorithm to obtain a second edition picture segment; the first focal depth corresponding to the first edition picture segment is less than the second focal depth corresponding to the second edition picture segment; the first focal depth and the second focal depth are obtained by inputting the first edition picture segment and the second edition picture segment into a convolution network model for prediction respectively. By using the process window optimization algorithm, the focal depth of the first edition picture segment can be optimized, that is, the process window is expanded, and the optimization is performed before the mask design stage, which can maximize the process window. The present application combines the process window optimization algorithm and the convolution network model to maximize the process window and improve the optimization accuracy.

Description

Technical Field

[0001] This application relates to the field of semiconductor technology, and in particular to a method, apparatus, equipment and medium for optimizing a layout photolithography process window. Background Technology

[0002] The manufacturing process of integrated circuits heavily relies on the photolithographic conformance of the layout. The simplified design flow of digital circuits includes logic design, layout design, and mask design. Various resolution enhancement techniques, such as optical nearest neighbor correction, sub-resolution assisted feature insertion, and reverse lithography, are employed to optimize photolithographic printability during the mask design stage—that is, to optimize the photolithographic process window. However, these techniques are performed during the mask design stage, when the layout and wiring are already determined, and the optimization effect is not significant. Therefore, the scope for enhancing photolithographic printability is extremely limited, and optimizing the photolithographic process window has certain limitations. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, apparatus, equipment, and medium for optimizing the lithography process window, which can maximize the process window and improve optimization accuracy. The specific solution is as follows:

[0004] On the one hand, this application provides a method for optimizing the layout photolithography process window, including:

[0005] The original map was divided into multiple first-version image segments;

[0006] The first image segment is optimized using a process window optimization algorithm to obtain a second image segment; the first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment; the first depth of focus and the second depth of focus are obtained by inputting the first image segment and the second image segment into a convolutional network model for prediction.

[0007] Specifically, the first version of the image segment is optimized using a process window optimization algorithm to obtain the second version of the image segment, which includes:

[0008] When the graphic information of the first image segment is lower than the preset graphic information, the first image segment is optimized by using a brute-force algorithm to obtain the second image segment.

[0009] Specifically, the method further includes:

[0010] When the graphic information of the first image segment is not lower than the preset graphic information, the first image segment is optimized using a gradient optimization algorithm to obtain the second image segment.

[0011] Specifically, the training process of the convolutional network model includes:

[0012] Retrieve multiple third-edition image segments;

[0013] Optical simulation is performed on multiple third-version image segments to obtain multiple third focal depths, each of which corresponds to each third-version image segment;

[0014] The preset model is trained using multiple third-version image segments and multiple third-depth-of-focus images, and the convolutional network model is obtained after training is completed.

[0015] In another aspect, embodiments of this application also provide an apparatus for optimizing the layout photolithography process window, including:

[0016] The division unit is used to divide the original map into multiple first-version image segments;

[0017] The first optimization unit is used to optimize the first image segment using a process window optimization algorithm to obtain a second image segment; the first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment; the first depth of focus and the second depth of focus are obtained by inputting the first image segment and the second image segment into a convolutional network model for prediction.

[0018] Specifically, the first optimization unit is used for:

[0019] When the graphic information of the first image segment is lower than the preset graphic information, the first image segment is optimized by using a brute-force algorithm to obtain the second image segment.

[0020] Specifically, the device further includes:

[0021] The second optimization unit is used to optimize the process window of the first image segment using a gradient optimization algorithm when the graphic information of the first image segment is not lower than the preset graphic information, so as to obtain the second image segment.

[0022] Specifically, the training process of the convolutional network model includes:

[0023] Retrieve multiple third-edition image segments;

[0024] Optical simulation is performed on multiple third-version image segments to obtain multiple third focal depths, each of which corresponds to each third-version image segment;

[0025] The preset model is trained using multiple third-version image segments and multiple third-depth-of-focus images, and the convolutional network model is obtained after training is completed.

[0026] In another aspect, embodiments of this application provide a computer device, the computer device including a processor and a memory:

[0027] The memory is used to store program code and transmit the program code to the processor;

[0028] The processor is used to execute the methods described above according to the instructions in the program code.

[0029] In another aspect, embodiments of this application provide a computer-readable storage medium for storing a computer program for performing the methods described above.

[0030] This application provides a method, apparatus, device, and medium for optimizing the lithography process window of a layout. The method involves dividing the original layout into multiple first image segments; optimizing the process window of each first image segment using a process window optimization algorithm to obtain a second image segment; the first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment; the first and second depths of focus are obtained by inputting the first and second image segments into a convolutional network model for prediction. By utilizing the process window optimization algorithm, the depth of focus of the first image segment can be optimized, i.e., the process window can be enlarged. Optimization before the mask design stage can maximize the process window size. This solution combines the process window optimization algorithm and the convolutional network model to maximize the process window size and improve optimization accuracy. Attached Figure Description

[0031] 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.

[0032] Figure 1 This illustration shows a flowchart of a method for optimizing a layout photolithography process window according to an embodiment of this application;

[0033] Figure 2 This illustration shows a flowchart of another method for optimizing a layout photolithography process window provided in an embodiment of this application;

[0034] Figure 3 A schematic diagram of a photolithography process window provided in an embodiment of this application is shown;

[0035] Figure 4 A schematic diagram of a gradient optimization algorithm provided in an embodiment of this application is shown;

[0036] Figure 5 A structural block diagram of an optimization apparatus for a layout photolithography process window provided in an embodiment of this application;

[0037] Figure 6 This is a structural diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0038] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0039] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0040] For ease of understanding, the following detailed description, in conjunction with the accompanying drawings, provides an embodiment of the optimization method, apparatus, equipment, and medium for a layout photolithography process window provided in this application.

[0041] refer to Figure 1 The diagram shown is a flowchart illustrating a method for optimizing a layout photolithography process window according to an embodiment of this application. The method may include the following steps.

[0042] S101, the original map is divided into multiple first-version image segments.

[0043] In this embodiment, the original layout is the layout to be optimized. The original layout can be divided and cut to obtain multiple image clips, denoted as the first image clip. That is, the first image clip is a smaller image clip cut from the original layout. Subsequently, the photolithography process window can be optimized for each first image clip. Compared to directly optimizing the larger original layout, optimizing each image clip allows for process window optimization of each region, resulting in more optimization iterations, greater detail, and better optimization effect of the photolithography process window.

[0044] Specifically, refer to Figure 2 As shown, this is another method for optimizing the layout photolithography process window provided in this application embodiment, which can segment the original layout to obtain a first image segment (initial clip).

[0045] S102, the process window of the first image segment is optimized using the process window optimization algorithm to obtain the second image segment; the first depth of focus of the first image segment is less than the second depth of focus of the second image segment; the first depth of focus and the second depth of focus are obtained by inputting the first image segment and the second image segment into the convolutional network model for prediction.

[0046] Specifically, for each first-version image segment, a process window optimization algorithm can be used to optimize the process window, resulting in an optimized image segment, denoted as the second-version image segment. The depth of focus (DOF) corresponding to the image segment can be used to evaluate the lithography process window; a larger DOF indicates a larger lithography process window. The DOF of the first-version image segment is denoted as the first DOF, and the DOF of the second-version image segment is denoted as the second DOF, where the second DOF is greater than the first DOF. A brief explanation of the process window follows.

[0047] In projection lithography systems, for a given feature size, the range of deviations from the optimal focal plane that allows for ensuring line quality is defined as the depth of focus. For lithography processes, a greater depth of focus is more favorable for the exposure of the lithographic pattern. The relationship between depth of focus, exposure wavelength, and the numerical aperture of the projection lens is as follows:

[0048]

[0049] In the formula, k2 is the depth-of-focus process factor. As the position of the focal plane changes, the quality of the exposure lines also changes. Thus, there must exist a position with the best exposure line quality, which is called the optimal focal plane. The value by which the actual imaging plane deviates from the optimal focal plane is called the defocusing amount.

[0050] In other words, depth of focus is the range within which the imaging position is allowed to deviate from the optimal focal plane position, while defocus is the range within which the actual imaging position deviates from the optimal focal plane position.

[0051] The purpose of photolithography is to copy the pattern on the photomask into a photoresist of a certain thickness. In order to obtain lines with steep sidewalls, the imaging intensity should be as consistent as possible within the thickness range of the photoresist. This requires that the depth of focus of projection lithography be greater than the thickness of the photoresist.

[0052] Relating to the formula for lithographic resolution, the expression for depth of focus can be rewritten as:

[0053]

[0054] The feature size CD of a photolithographic pattern fluctuates significantly with variations in exposure dose and defocusing amount. In actual photolithography processes, the electrical performance of integrated circuit chips allows for a certain degree of error in the feature size of the photolithographic pattern; this permissible error range is typically ±10% CD. Based on this standard, points meeting the conditions can be plotted on the Poisson curve of the photolithographic image, and connecting these points yields the following result: Figure 3 The process window shown.

[0055] Figure 3 The portion of the curve shown that allows for the largest possible rectangular or elliptical enclosed area is called the process window. Elliptical process windows are generally used in photolithography, where only the largest possible ellipse can be obtained within the closed area. It should be noted that the rectangles or circles shown in the figure are drawn within a closed area formed by two curves, representing the exposure dose at different focal plane positions, satisfying both CD(1+10%) and CD(1-10%) conditions.

[0056] In actual photolithography processes, the standard for measuring the process window is generally the depth of focus (i.e., the length of the Y-axis of a fixed ellipse, measuring the length of the X-axis of the ellipse) when the exposure dose varies by 5%. The larger this value is, the larger the process window is. Therefore, the depth of focus can be used to measure the size of the process window.

[0057] In this embodiment, a Convolutional Neural Network (CNN) model can be used to predict the depth of focus. A first image segment is input into the CNN model to obtain a first depth of focus, and a second image segment is input into the CNN model to obtain a second depth of focus. Specifically, during the optimization process, the first image segment and the first depth of focus can be input into a process window optimization algorithm for optimization. After obtaining a new image segment, the CNN model is used to predict the depth of focus value until the optimal image segment is found, thus obtaining the second image segment with the second depth of focus.

[0058] In this way, by using the process window optimization algorithm, the depth of focus of the first image segment can be optimized, that is, the process window can be expanded. Moreover, the optimization process is carried out before the mask design stage, which can maximize the process window. In addition, this solution combines the process window optimization algorithm and the convolutional network model, which can maximize the expansion of the process window, have high optimization efficiency, and improve optimization accuracy.

[0059] In the embodiments of this application, the convolutional network model can be trained before using it to predict the depth of focus, thereby improving the model's prediction accuracy.

[0060] Specifically, multiple third-version image segments can be obtained through clips collection. These third-version image segments are segmented from the unoptimized version image segments. Optical simulation can be performed on these multiple third-version image segments to obtain multiple third depths of focus (DFCs). Each DFC corresponds to a specific third-version image segment, thus obtaining a large number of third-version image segments and their corresponding DFC values, increasing the size of the training set. Then, the preset model is trained using these multiple third-version image segments and DFCs. After iteration stops, the training is complete, resulting in a convolutional network model. It is understandable that after model training is complete, the first-version image segments requiring optimization can be directly input into the convolutional network model and the process window optimization algorithm, without needing to be input into the optical simulation model.

[0061] To further improve optimization efficiency and accuracy, different process window optimization algorithms can be selected based on the graphic information in the first version of the image segment, i.e., algorithm selection.

[0062] In one possible implementation, when the graphic information of the first image segment is lower than the preset graphic information, the process window of the first image segment is optimized using a brute-force algorithm to obtain a second image segment (clip after LRE). By splicing multiple second image segments, an optimized layout (layout after LRE) can be obtained.

[0063] Specifically, the graphic information of the first image segment includes the size, shape, and quantity of the graphics in the image segment. For example, the graphic information could be the total number of graphics in the first image segment, or the total area of ​​the graphics, etc., without specific limitations. The preset graphic information can also be a preset total number of graphics and a preset total area of ​​graphics. In other words, the graphic information of the first image segment and the preset graphic information are information of the same dimension, such as the total number of graphics.

[0064] When the graphic information of the first image segment is lower than the preset graphic information, the process window optimization algorithm can be an exposure cracking algorithm, which uses a brute-force cracking algorithm for optimization. The brute-force cracking algorithm can try to explore all optimization schemes and perform various optimizations on the first image segment, such as graphic stretching and shifting, to obtain a large number of new image segments. From these, the image segment with the largest photolithography process window can be determined, thus obtaining the second image segment.

[0065] Thus, when the graphic information in the first image segment is relatively simple, using a brute-force algorithm for optimization can improve the optimization effect and accurately determine the second image segment with the largest process window.

[0066] In another possible implementation, when the graphic information of the first image segment is not lower than the preset graphic information, the graphic information is relatively complex. Optimizing it using a brute-force algorithm would be very time-consuming. In this case, the first image segment can be optimized using the gradient optimization algorithm (GREA) to obtain the second image segment.

[0067] Specifically, refer to Figure 4 The diagram shown is a schematic of a gradient optimization algorithm provided in an embodiment of this application. When using the gradient optimization algorithm for optimization, X can represent the first image segment, X' can represent the optimized image segment (i.e., the temporary optimal solution), and X” represents the second image segment (i.e., the optimal solution), which is the optimized image segment with the largest photolithography process window.

[0068] In the initial stage (1-4), X, X', and X” can be initialized. The optimization scheme S is assigned to X and X'. ΔF represents the DOF increment, which is the difference between the focal depth of the first image segment and the focal depth of the second image segment. It can be initially set to 1, and the key optimal solution X” is cleared.

[0069] If X is not equal to X`, proceed to steps 6-12. If the DOF increment is greater than 0, assign X` to X, and then continue executing the algorithm. The function G(x) i ) can be represented as x along the direction that increases DOF. i The graphic in the layout, such as a small line segment, can be represented by a function L. This function can perform operations on the graphic in the layout segment, such as extending, translating, or scaling, to obtain the optimized layout segment X', which is the temporary optimal solution.

[0070] Step 13 indicates that a local brute-force search is performed near the temporary optimal solution, rather than a global brute-force search. The local brute-force search will limit the search range. If the temporary optimal solution is also the optimal solution in its neighborhood, the algorithm ends and this solution is the optimal solution. Otherwise, the optimal solution in the neighborhood is used as the starting point and the optimization continues until the optimal solution is found.

[0071] Therefore, it can be seen that by using gradient-based optimization algorithms, we can find the second image segment with the largest lithography process window while reducing the amount of computation and improving efficiency.

[0072] This application provides a method for optimizing the lithography process window of a layout. The original layout is divided into multiple first image segments. A process window optimization algorithm is used to optimize the process window of each first image segment to obtain a second image segment. The first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment. The first and second depths of focus are predicted by inputting the first and second image segments into a convolutional network model, respectively. By utilizing the process window optimization algorithm, the depth of focus of the first image segment can be optimized, i.e., the process window can be expanded. Optimization is performed before the mask design stage, maximizing the process window size. This solution combines the process window optimization algorithm and the convolutional network model to maximize the process window size and improve optimization accuracy.

[0073] Based on the above optimization method for the layout photolithography process window, this application embodiment also provides an optimization apparatus for the layout photolithography process window, referencing... Figure 5 The diagram shown is a structural block diagram of an optimization apparatus for a layout photolithography process window provided in an embodiment of this application. The apparatus may include:

[0074] Division unit 201 is used to divide the original map into multiple first-version image segments;

[0075] The first optimization unit 202 is used to optimize the first version of the image segment using a process window optimization algorithm to obtain a second version of the image segment; the first depth of focus corresponding to the first version of the image segment is less than the second depth of focus corresponding to the second version of the image segment; the first depth of focus and the second depth of focus are obtained by inputting the first version of the image segment and the second version of the image segment into the convolutional network model for prediction.

[0076] Specifically, the first optimization unit is used for:

[0077] When the graphic information of the first image segment is lower than the preset graphic information, the first image segment is optimized by using a brute-force algorithm to obtain the second image segment.

[0078] Specifically, the device further includes:

[0079] The second optimization unit is used to optimize the process window of the first image segment using a gradient optimization algorithm when the graphic information of the first image segment is not lower than the preset graphic information, so as to obtain the second image segment.

[0080] Specifically, the training process of the convolutional network model includes:

[0081] Retrieve multiple third-edition image segments;

[0082] Optical simulation is performed on multiple third-version image segments to obtain multiple third focal depths, each of which corresponds to each third-version image segment;

[0083] The preset model is trained using multiple third-version image segments and multiple third-depth-of-focus images, and the convolutional network model is obtained after training is completed.

[0084] This application provides an optimization device for the lithography process window of a layout. A partitioning unit divides the original layout into multiple first image segments. A first optimization unit optimizes the process window of the first image segments using a process window optimization algorithm to obtain second image segments. The first depth of focus corresponding to the first image segments is less than the second depth of focus corresponding to the second image segments. The first and second depths of focus are predicted by inputting the first and second image segments into a convolutional network model, respectively. By utilizing the process window optimization algorithm, the depth of focus of the first image segments can be optimized, i.e., the process window can be expanded. Optimization before the mask design stage can maximize the process window size. This solution combines the process window optimization algorithm and the convolutional network model to maximize the process window size and improve optimization accuracy.

[0085] In another aspect, embodiments of this application provide a computer device, with reference to Figure 6 The diagram shown is a structural diagram of a computer device provided in an embodiment of this application. The computer device includes a processor 310 and a memory 320.

[0086] The memory 320 is used to store program code and transmit the program code to the processor 310;

[0087] The processor 310 is used to execute the method provided in the above embodiments according to the instructions in the program code.

[0088] The computer device may include a terminal device or a server, and the aforementioned apparatus may be configured in the computer device.

[0089] In another aspect, embodiments of this application also provide a storage medium for storing a computer program for executing the methods provided in the above embodiments.

[0090] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by program instructions in hardware. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium can be at least one of the following media: read-only memory (ROM), RAM, magnetic disk, or optical disk, etc., and other media capable of storing program code.

[0091] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, the apparatus embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0092] The above description is merely a preferred embodiment of this application. Although this application has disclosed preferred embodiments above, it is not intended to limit this application. Any person skilled in the art can make many possible variations and modifications to the technical solutions of this application using the methods and techniques disclosed above, or modify them into equivalent embodiments with equivalent changes, without departing from the scope of the technical solutions of this application. Therefore, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of this application without departing from the content of the technical solutions of this application shall still fall within the protection scope of the technical solutions of this application.

Claims

1. A method for optimizing a layout photolithography process window, characterized in that, include: The original map was divided into multiple first-version image segments; The first image segment is optimized using a process window optimization algorithm to obtain a second image segment; the first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment; the first depth of focus and the second depth of focus are obtained by inputting the first image segment and the second image segment into a convolutional network model for prediction; The first version of the image segment is optimized using a process window optimization algorithm to obtain the second version of the image segment, which includes: When the graphic information of the first image segment is lower than the preset graphic information, the first image segment is optimized by a brute-force algorithm to obtain the second image segment. When the graphic information of the first image segment is not lower than the preset graphic information, the first image segment is optimized using a gradient optimization algorithm to obtain the second image segment.

2. The method according to claim 1, characterized in that, The training process of the convolutional network model includes: Retrieve multiple third-edition image segments; Optical simulation is performed on multiple third-version image segments to obtain multiple third focal depths, each of which corresponds to each third-version image segment; The preset model is trained using multiple third-version image segments and multiple third-depth-of-focus images, and the convolutional network model is obtained after training is completed.

3. An optimization device for a layout photolithography process window, characterized in that, include: The division unit is used to divide the original map into multiple first-version image segments; The first optimization unit is used to optimize the first image segment using a process window optimization algorithm to obtain a second image segment; the first depth of focus corresponding to the first image segment is less than the second depth of focus corresponding to the second image segment; the first depth of focus and the second depth of focus are obtained by inputting the first image segment and the second image segment into a convolutional network model for prediction; The first optimization unit is used for: When the graphic information of the first image segment is lower than the preset graphic information, the first image segment is optimized by a brute-force algorithm to obtain the second image segment. The second optimization unit is used to optimize the process window of the first image segment using a gradient optimization algorithm when the graphic information of the first image segment is not lower than the preset graphic information, so as to obtain the second image segment.

4. The apparatus according to claim 3, characterized in that, The training process of the convolutional network model includes: Retrieve multiple third-edition image segments; Optical simulation is performed on multiple third-version image segments to obtain multiple third focal depths, each of which corresponds to each third-version image segment; The preset model is trained using multiple third-version image segments and multiple third-depth-of-focus images, and the convolutional network model is obtained after training is completed.

5. A computer device, characterized in that, The computer device includes a processor and memory: The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the method described in any one of claims 1-2 according to the instructions in the program code.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method according to any one of claims 1-2.

Images