Inversion lithography method and system based on adaptive threshold regularization

Abstract

This application discloses an inversion lithography method and system based on adaptive threshold regularization, which relates to the fields of inversion lithography and optical proximity correction in computational lithography. The method includes obtaining an initial mask image; optimizing the initial mask image using a CTM algorithm to obtain a grayscale value mask containing SRAFs; performing regional division to obtain a grayscale value mask after regional division; and using an adaptive threshold regularization algorithm to dynamically adjust the thresholds of each region in the grayscale value mask after regional division to obtain a binary mask containing SRAFs. The dynamic threshold adjustment optimization is to adjust the thresholds corresponding to each region in the grayscale value mask after regional division by adaptively adjusting parameter a to determine the optimal threshold for each region. Parameter a is a parameter in the regularization function in the adaptive threshold regularization algorithm. This application can obtain a binary mask that meets industrial manufacturing needs and improves the accuracy of mask imaging.

Description

Technical Field

[0001] The present application relates to the technical fields of inversion lithography and optical proximity correction in computational lithography, and in particular to an inversion lithography method and system based on adaptive threshold regularization. Background Art

[0002] With the rapid advancement of semiconductor manufacturing technology, the design of integrated circuits has become increasingly complex, and the amount of mask layout data has surged to the billions. Against this backdrop, mask binarization plays a particularly significant role. It ensures that the pattern can be transferred from the mask to the silicon wafer with extremely high precision during the photolithography process, a step that is crucial for accurately replicating the microscopic features in integrated circuits. Furthermore, the use of electron beam lithography machines to manufacture masks necessitates optical proximity correction (OPC) to output a binary mask and prepare mask data. Therefore, mask binarization is not only key to improving production efficiency, ensuring product quality, and guaranteeing manufacturability, but also a significant driver of technological innovation in the semiconductor industry.

[0003] The mask binarization process involves converting the grayscale mask into a binary representation consisting of only two states: fully transparent and fully opaque. This conversion is crucial to meeting industrial production standards. However, conventional mask binarization techniques rely on a single threshold method, resulting in low mask manufacturing precision and inaccurate mask imaging. Summary of the Invention

[0004] The purpose of this application is to provide an inversion lithography method and system based on adaptive threshold regularization, which can improve the accuracy of mask imaging.

[0005] To achieve the above objectives, this application provides the following solutions.

[0006] In a first aspect, the present application provides an inversion lithography method based on adaptive threshold regularization, and the inversion lithography method based on adaptive threshold regularization includes the following steps.

[0007] An initial mask image is obtained; the initial mask image is a mask image obtained after initializing a target mask, and the target mask is a mask to be processed.

[0008] The initial mask image is optimized using a CTM algorithm to obtain a grayscale value mask containing SRAFs.

[0009] According to the grayscale value of each region in the grayscale value mask containing SRAFs, the grayscale value mask containing SRAFs is divided into regions to obtain a grayscale value mask after region division.

[0010] An adaptive threshold regularization algorithm is used to dynamically adjust and optimize the threshold values ​​of each region in the grayscale value mask after the region division to obtain a binary mask containing SRAFs; the dynamic threshold adjustment optimization refers to adjusting the threshold values ​​corresponding to each region in the grayscale value mask after the region division by adaptively adjusting parameter a, thereby determining the optimal threshold value for each region; the parameter a refers to a parameter in the regularization function in the adaptive threshold regularization algorithm.

[0011] In a second aspect, the present application provides an inversion lithography system based on adaptive threshold regularization, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the inversion lithography method based on adaptive threshold regularization described in the first aspect.

[0012] According to the specific embodiments provided in this application, this application discloses the following technical effects.

[0013] The present application provides an inversion lithography method and system based on adaptive threshold regularization. The initial mask image is optimized using the CTM algorithm to obtain a grayscale value mask containing SRAFs. This optimization process based on the CTM algorithm can preliminarily ensure the accuracy of mask imaging. By dividing the grayscale value mask containing SRAFs into regions, it is convenient to adjust the threshold of each region in the grayscale value mask containing SRAFs. By introducing the parameter a in the regularization function, the threshold corresponding to each region in the grayscale value mask after region division is adjusted by adaptively adjusting the parameter a, thereby determining the optimal threshold for each region, and finally obtaining a more accurate and reliable binary mask containing SRAFs. This method can improve the details of each region, thereby reflecting the characteristics of the overall mask and improving the accuracy of mask imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 A schematic flow chart of an inversion lithography method based on adaptive threshold regularization provided in one embodiment of the present application.

[0016] Figure 2 A schematic diagram of an inversion lithography method based on adaptive threshold regularization provided in one embodiment of the present application.

[0017] Figure 3 This is a graph showing how the regularization function value of the adjustment parameter a varies with mask values ​​in the range of 0 to 0.5 according to an embodiment of the present application.

[0018] Figure 4 This is a graph showing how the regularization function value of the adjustment parameter a varies with mask values ​​in the range of 0.5 to 1, according to an embodiment of the present application.

[0019] Figure 5 This is a loss comparison curve diagram of the imaging process of the adaptive threshold method, CTM algorithm and single threshold method provided in one embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0021] like Figure 1 As shown, this embodiment proposes an inversion lithography method based on adaptive threshold regularization, and the inversion lithography method based on adaptive threshold regularization specifically includes the following steps.

[0022] Step S1: Obtain an initial mask image. The initial mask image is a mask image obtained after initializing a target mask, and the target mask is a mask to be processed.

[0023] Step S2: Optimize the initial mask image using a CTM (Continuous Transmission Mask) algorithm to obtain a grayscale mask containing SRAFs (Sub-Resolution Assist Features).

[0024] Step S3: Divide the grayscale value mask containing SRAFs into regions according to the distribution of grayscale values ​​of each region in the grayscale value mask containing SRAFs to obtain a grayscale value mask after region division.

[0025] In this embodiment, the grayscale value mask after region division includes a main feature region and SRAFs regions of different orders. The main feature region refers to the main graphic region of the grayscale value mask obtained by the CTM algorithm. Based on this main graphic region, regions of different grayscales are sequentially obtained based on distance and grayscale distribution by expanding outward from the main graphic region. These regions are called SRAFs regions of different orders. The main graphic region is a pattern similar to the target mask portion. Then, based on distance and grayscale value, the regions expanding outward from the main graphic region are divided into first-order SRAFs regions, second-order SRAFs regions, and third-order SRAFs regions.

[0026] Step S4: Using an adaptive threshold regularization algorithm, dynamically adjust and optimize the thresholds of each region in the grayscale value mask after region division to obtain a binary mask containing SRAFs. Dynamic threshold adjustment and optimization refers to adaptively adjusting parameter a to adjust the thresholds corresponding to each region in the grayscale value mask after region division, thereby determining the optimal threshold for each region. Parameter a refers to a parameter in the regularization function of the adaptive threshold regularization algorithm.

[0027] In this embodiment, step S1 of acquiring an initial mask image specifically includes the following steps.

[0028] Step S11: Read the pattern of the target mask.

[0029] Step S12: Perform initialization processing according to the target mask pattern to obtain an initial mask image.

[0030] In this embodiment, step S2 uses the CTM algorithm to optimize the initial mask image to obtain a grayscale value mask containing SRAFs, which specifically includes the following steps.

[0031] The CTM algorithm is used to optimize the pixels of the initial mask image, optimizing the pixels of 0 and 1 to pixels between 0 and 1, thereby obtaining a grayscale value mask containing SRAFs. The pixels of the initial mask image are 0 and 1, and the pixels of the grayscale value mask containing SRAFs are between 0 and 1.

[0032] In this embodiment, step S4 adopts an adaptive threshold regularization algorithm to perform dynamic threshold adjustment optimization on each region in the grayscale value mask after region division to obtain a binary mask containing SRAFs, which specifically includes the following steps.

[0033] Step S41: Adopting an adaptive threshold regularization algorithm, the regularization function of the adaptive threshold regularization algorithm is added as a regularization term to the objective function of the CTM algorithm, which includes a loss function and a regularization term. With minimizing this objective function as the optimization goal, the grayscale value mask after region division is optimized. Before the optimization begins, an initial parameter a is set for each region in the grayscale value mask after region division.

[0034] Step S42: Adopting the gradient descent method, by adaptively adjusting the initial parameter a corresponding to each region, the threshold corresponding to each region is iteratively optimized and adjusted, so as to determine the optimal threshold corresponding to each region, and at this time, a binary mask containing SRAFs is obtained.

[0035] In this embodiment, the regularization function is added to the objective function in the form of a regularization term. Therefore, the regularization function is actually a regularization term. In the regularization function, the following relationship exists between the regularization function value and the mask value of the regularization function.

[0036] (1) When the mask value is 0 and 1, the regularization function value is the smallest and both are 0.

[0037] (2) When the mask value is any value between 0 and 1, the regularization function value has only one controllable maximum value. By adjusting the size of parameter a, the maximum value of the regularization function value can be adjusted so that the regularization function reaches the maximum value under different mask values.

[0038] The principle of the inversion lithography method based on adaptive threshold regularization in this embodiment is as follows.

[0039] Based on the target mask image, the initial mask image is first optimized using the CTM algorithm. During this process, the pattern error is set as the objective function to ensure high imaging accuracy. After completing this step, a grayscale value mask containing SRAFs is obtained. Secondly, based on the grayscale value distribution of the grayscale value mask containing SRAFs, the grayscale value mask containing SRAFs is dynamically divided into primary feature regions and SRAF regions of different orders, laying the foundation for subsequent refined optimization. A regularization function is then applied, integrated into the objective function, which is key to achieving adaptive threshold adjustment and ensuring optimal binarization in each region. Finally, the primary feature regions and SRAF regions of different orders are jointly optimized. This process adaptively adjusts the optimal threshold for each region, and through iterative optimization, gradually approaches the global optimal solution, ultimately achieving accurate binarization of the mask image. This significantly improves the accuracy of mask imaging and effectively addresses the low accuracy of traditional methods.

[0040] This embodiment introduces a regularization term that can adaptively adjust the threshold, which not only optimizes the binarization problem in mask design and manufacturing, but also significantly improves the accuracy of mask imaging. By dynamically adjusting the threshold to adapt to the diversity and subtle differences of lithography masks, it overcomes the limitations of traditional mask binarization technology when processing complex patterns.

[0041] In order to make the technical solution of this embodiment clearer, the specific implementation steps of the technical solution of this embodiment are described in detail below in the form of examples.

[0042] Step 101: Read the target mask pattern and initialize to obtain an initial mask image.

[0043] Step 102: Optimize the initial mask image using the CTM algorithm. To ensure the accuracy of mask imaging, edge placement error (EPE) and pattern error (PE) are used as metrics to evaluate mask imaging accuracy. This embodiment defines pattern error as the loss function in the optimization process. In practical applications, edge placement error or other metrics can also be defined as the loss function. This objective function is then used as the objective function in subsequent optimization steps. This objective function is designed to minimize mask imaging errors and ensure high fidelity of the mask image. The formula is as follows.

[0044]

[0045] in, To calculate the loss, and Respectively represent the number of rows and columns of the target graph, and are row and column indices, respectively. Represents the target graphic, Indicates the wafer image, represents the initial mask image, Represents the imaging of the mask after each optimization.

[0046] In this embodiment, after the optimization is completed, the CTM algorithm will generate a pixel map with intensity values ​​between 0 and 1, that is, a grayscale value mask containing SRAFs.

[0047] Step 103: A grayscale value mask containing SRAFs can be obtained through step 102. According to the grayscale value distribution, the grayscale value mask containing SRAFs is dynamically divided into a main feature region and SRAFs regions of different orders.

[0048] Step 104: Through step 103, the main feature regions of the grayscale value mask and the SRAFs regions of different orders can be obtained. To dynamically adjust the optimal threshold for each region, this embodiment designs a regularization function. The regularization function reaches its minimum value, both 0, when the mask value is 0 and 1. There is only one controllable maximum point in the mask value range from 0 to 1. This embodiment defines this parameter as a. By adjusting parameter a, this maximum point can be flexibly adjusted. Therefore, using this regularization function, the threshold can be adaptively adjusted. Figure 3 The graph of the regularization function value of the adjustment parameter a changes with the mask value in the range of 0 to 0.5. Figure 4 It is a graph showing the change of the regularization function value of the adjustment parameter a with the mask value in the range of 0.5 to 1, where Figure 3 and Figure 4 The six curves in represent the specific changes of the regularization function by adjusting the value of parameter a. Figure 3 and Figure 4 It can be seen that the regularization function reaches its minimum value when the mask value is 0 and 1, both of which are 0, and there is only one controllable maximum point in the interval of mask value from 0 to 1, and this maximum point can be adjusted by adjusting the parameter a.

[0049] Step 105: Add the regularization function in step 104 as a regularization term to the objective function. The objective function is essentially a cost function, and the goal is to minimize the cost function. At the same time, in step 105, it is necessary to select a suitable weight for the regularization term so that it can ensure the accuracy of mask imaging and the obtained mask is in a binary form.

[0050] Step 106: Jointly optimize the main feature regions and SRAFs of different orders from step 103. Each region is assigned an initial parameter a. Gradient descent is used to adaptively adjust the parameter a for each region. A determination is then made as to whether the cost function is less than a set value. If so, adjustment and optimization are stopped, resulting in an optimal threshold for each region. Otherwise, parameter a is iteratively adjusted until the cost function is less than the set value, indicating that all regions have their optimal thresholds, thus obtaining a binary mask containing SRAFs.

[0051] Step 107: Verify the effectiveness of the technical solution of this embodiment by comparing it with the traditional method. The method of this embodiment is compared with the mask optimized by only using the CTM algorithm and the single threshold algorithm in the optimization process. Figure 5The following graph compares the loss of the imaging process using the adaptive threshold regularization method, the CTM algorithm, and the single threshold algorithm in this embodiment. It can be seen that while the mask optimized by the CTM algorithm has a smaller loss, with a final loss of 66.2696, the optimized mask is a grayscale value mask, not a binary mask, which does not meet actual production requirements. While the single threshold method achieves mask binarization, its loss is excessive, reaching a final loss of 130.3158, which cannot guarantee mask manufacturing accuracy and results in low mask imaging precision. However, using the method of this embodiment, the resulting mask not only meets the binary mask requirements of production and manufacturing, but also ensures accurate mask imaging.

[0052] This embodiment can adaptively adjust the threshold of each region, achieving accurate matching of the optimal threshold for each region. Compared with the traditional inversion lithography method, this embodiment not only optimizes the binarization effect of the local region, but also helps to discover the global optimal solution, thereby significantly improving the accuracy of mask imaging. This adaptive threshold adjustment mechanism provides strong support for the accurate binarization of the mask and meets the strict requirements of high-end semiconductor manufacturing. Moreover, by dividing the grayscale value mask containing SRAFs into multiple regions and applying an adaptive threshold regularization adjustment strategy to each region, not only the flexibility of the design is ensured, but also the process of finding the global optimal solution is effectively promoted. Therefore, the accuracy of mask binarization can be significantly improved, the accuracy of mask imaging can be guaranteed, and a more accurate and efficient mask production technology can be provided for the semiconductor manufacturing field.

[0053] This embodiment adopts an adaptive threshold regularization adjustment strategy to ensure the degree of simulation and mask manufacturability during the mask optimization process. First, the CTM algorithm is used to optimize the initial mask image to generate a grayscale value mask containing SRAFs. The generated grayscale value mask containing SRAFs is then dynamically divided into main feature regions and SRAFs regions of different orders according to the distribution of its grayscale values. Next, the grayscale value mask after regional division is used as input and optimized. At the same time, the adaptive threshold regularization adjustment strategy is integrated into the mask optimization process. During the optimization process, dynamic threshold adjustment is performed on the main feature regions and SRAFs regions of the grayscale value mask after regional division. The optimization result selects the most appropriate threshold for each region and ensures the accuracy of mask imaging. Finally, a binary mask containing SRAFs is obtained. Among them, this embodiment designs a regularization function specifically for optimizing the mask binarization process. This regularization function reaches its minimum value, 0, when the mask values ​​are 0 and 1, and has a single, controllable maximum point within the mask value range from 0 to 1. This maximum point can be flexibly adjusted by adjusting the parameter a. This regularization function enables precise control of the threshold. This design allows different regions of the mask image to automatically adapt to the optimal threshold, effectively ensuring the manufacturability of the binary mask and the accurate representation of SRAFs.

[0054] In an exemplary embodiment, an inversion lithography system based on adaptive threshold regularization is also provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned inversion lithography method based on adaptive threshold regularization.

[0055] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0056] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. An inversion lithography method based on adaptive threshold regularization, characterized in that: The inversion lithography method based on adaptive threshold regularization includes: Acquire an initial mask image; the initial mask image is a mask image obtained after initializing a target mask, and the target mask is a mask to be processed; The initial mask image is optimized using a CTM algorithm to obtain a grayscale value mask containing SRAFs; Dividing the grayscale value mask containing SRAFs into regions according to the grayscale value of each region in the grayscale value mask containing SRAFs to obtain a grayscale value mask after region division; Adaptive threshold regularization algorithm is used to dynamically adjust and optimize the thresholds of the respective regions in the grayscale value mask after the region division to obtain a binary mask containing SRAFs; the dynamic threshold adjustment optimization refers to adjusting the thresholds corresponding to the respective regions in the grayscale value mask after the region division by adaptively adjusting parameter a, thereby determining the optimal threshold for each region; parameter a refers to a parameter in the regularization function in the adaptive threshold regularization algorithm; Adopting an adaptive threshold regularization algorithm, dynamically adjusting and optimizing the threshold values ​​of each region in the grayscale value mask after the region division, and obtaining a binary mask containing SRAFs, specifically including: Adopting the adaptive threshold regularization algorithm, adding the regularization function of the adaptive threshold regularization algorithm to the objective function of the CTM algorithm in the form of a regularization term, optimizing the grayscale value mask after the region division with minimization of the objective function as the optimization goal, and setting an initial parameter a for each region in the grayscale value mask after the region division during the optimization process; The gradient descent method is used to adaptively adjust the initial parameter a corresponding to each region to determine the optimal threshold corresponding to each region and obtain a binary mask containing SRAFs; The regularization function reaches its minimum value when the mask value is 0 and 1, both being 0, and there is only one controllable maximum point in the interval of the mask value from 0 to 1, and this maximum point is adjusted by adjusting the parameter a.

2. The inversion lithography method based on adaptive threshold regularization according to claim 1, characterized in that: Obtain the initial mask image, including: Read the pattern of the target mask; Initialization processing is performed according to the graphic of the target mask to obtain an initial mask image.

3. The inversion lithography method based on adaptive threshold regularization according to claim 1, characterized in that: The initial mask image is optimized using the CTM algorithm to obtain a grayscale value mask containing SRAFs, specifically including: The pixels of the initial mask image are optimized using a CTM algorithm, and pixels of 0 and 1 are optimized to pixels between 0 and 1, thereby obtaining the grayscale value mask containing SRAFs; the pixels of the initial mask image are 0 and 1, and the pixels of the grayscale value mask containing SRAFs are between 0 and 1.

4. The inversion lithography method based on adaptive threshold regularization according to claim 1, characterized in that: The loss function of the CTM algorithm is the pattern error.

5. The inversion lithography method based on adaptive threshold regularization according to claim 4, characterized in that: The expression of the loss function is: Among them, F loss To calculate the loss, M and N represent the number of rows and columns of the target graph, i and j are the row index and column index, respectively. Represents the target graphic, z i，j Represents the wafer image, z * represents the initial mask image, and z represents the image of the mask after each optimization.

6. An inversion lithography system based on adaptive threshold regularization, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the inversion lithography method based on adaptive threshold regularization according to any one of claims 1 to 5.

Images