Extreme ultraviolet lithography mask optimization method based on localized guidance of implicit field evolution

Abstract

The present application relates to the technical field of semiconductor manufacturing, and particularly relates to an extreme ultraviolet lithography mask optimization method, comprising the following steps: S1: performing lithography process simulation on the extreme ultraviolet lithography mask, obtaining a simulated physical response, and calculating an error field between the simulated physical response and a target physical response; S2: converting the error field into an exposure requirement field based on locality; S3: iteratively updating an implicit field with the exposure requirement field as a driving item; and S4: reconstructing the implicit field into an optimized mask pattern. The present application can realize extreme ultraviolet lithography mask optimization with strict physical simulation driving, low evaluation frequency and high optimization quality. Meanwhile, the method can be used for multi-layer film defect compensation, and supports two kinds of mask representation forms, i.e. curvilinear and pixelated.

Description

Technical Field

[0001] This invention relates to the field of semiconductor manufacturing technology, and in particular to an extreme ultraviolet lithography mask optimization method. Background Technology

[0002] Photolithography is a manufacturing technology used in semiconductor production to transfer patterns in integrated circuits. As the size of integrated circuits continues to shrink, the fidelity of images printed on wafers is facing increasingly severe challenges. Mask optimization, by actively controlling the pattern morphology, to tap the imaging potential of photolithography systems, has been widely researched and applied in deep ultraviolet lithography.

[0003] However, compared to deep ultraviolet (DUV) lithography, extreme ultraviolet (EUV) lithography involves more complex physical mechanisms. Significant mask thickness effects, angular dependence of reflectivity, and defects in multilayer films combine to make the near-field distribution of the mask difficult to accurately simulate using thin masks, typically requiring rigorous electromagnetic field solutions based on Maxwell's equations. Furthermore, the lower photon count and the complexity of chemically amplified photoresists further enhance random fluctuations in EUV lithography, making the development process difficult to accurately describe using simplified models. Therefore, to accurately simulate the complex imaging and development behavior in EUV lithography, rigorous physical modeling of the lithography process is necessary.

[0004] Prior technique 1 (ZHENG S, YU B, WONG M. OpenILT: An open source inverselithography technique framework. proceedings of the 2023 IEEE 15th International Conference on ASIC (ASICON) [C].IEEE,2023: 1–4.) employs a gradient optimization method based on automatic differentiation. By establishing a differentiable lithography model, it iteratively optimizes the mask pattern using a gradient descent algorithm. However, the differentiable model in this method uses a simplified description of the lithography process, which is difficult to fully characterize the complex physical effects in extreme ultraviolet lithography, thus limiting the accuracy of the optimization results. Prior technique 2 (HUANG W, LI Y, LI Z, et al. High-fidelitycurvilinear mask optical proximity correction using tangent angle-arc lengthcurve [J]. Optics Express, 2025, 33(11): 23796–23808.) transforms the mask optimization problem into a parameter search problem by constructing a parameterized mask representation and uses a heuristic algorithm to find a solution that satisfies the optimization objective. However, the inherent high-dimensional search characteristic of the mask optimization problem makes heuristic optimization require maintaining a large population size, making it difficult to overcome the efficiency bottleneck. Prior technique 3 (HUANG W, LI Y, YANG H, et al. Fast and high-fidelitycurvilinear optical proximity correction using boundary iterative optimization [J]. Applied Optics, 2025, 65(2): 374–381.) optimizes the mask profile by driving the movement of parametric curve control points based on the local centroid deviation between the printed pattern and the mask pattern, thus avoiding the heuristic search process. However, its optimization quality still has considerable room for improvement. Summary of the Invention

[0005] This invention addresses the limitations of existing extreme ultraviolet (EUV) lithography mask optimization techniques in overcoming the constraints of rigorous simulation, search efficiency, and optimization quality. It proposes an EUV lithography mask optimization method based on locality-guided implicit field evolution, achieving mask optimization that balances rigorous physical simulation-driven optimization, low evaluation times, and high quality.

[0006] The extreme ultraviolet lithography mask optimization method provided by this invention specifically includes the following steps: S1: Simulate the lithography process of the extreme ultraviolet lithography mask, obtain the simulated physical response, and calculate the error field between it and the target physical response. The error field can characterize the difference between the target physical response and the actual physical response generated by the current mask, providing error-driven information for subsequent mask optimization.

[0007] Preferably, the simulated physical response includes the photoresist contour pattern after exposure and development on the wafer side, the spatial image threshold pattern on the wafer side, and the near-field electric field intensity distribution on the mask side; Preferably, the error field is represented as: (1); in, Represents the error field. Represents the distribution of the target physical response. Represents the simulated physical response distribution. Represents spatial coordinates.

[0008] Preferably, the photoresist outline pattern after exposure and development is used as the simulated physical response, and step S1 specifically includes the following steps: S111: Construct extreme ultraviolet lithography simulation models, including mask near-field electromagnetic field simulation models, projection imaging models, and photoresist exposure and development models; S112: Obtain the photoresist outline pattern after exposure and development through simulation, and calculate the error field between it and the target pattern.

[0009] Preferably, using a spatial image threshold pattern as the simulated physical response specifically includes the following steps: S121: Construct an extreme ultraviolet lithography simulation model, including a mask near-field electromagnetic field simulation model and a projection imaging model; S122: Obtain the spatial image intensity distribution on the wafer side through simulation, generate a spatial image threshold pattern according to a preset threshold, and calculate the error field between the pattern and the target pattern.

[0010] Preferably, the near-field electric field intensity distribution of the mask is used as the simulated physical response, specifically including the following steps: S131: Construct a near-field electromagnetic field simulation model for the mask; S132: Obtain the near-field electric field intensity distribution of the mask through simulation, and calculate the error field between it and the target electric field intensity distribution.

[0011] S2: Based on locality, the error field is converted into an exposure demand field on the mask side. Locality refers to the significant spatial local response characteristics of the photolithography process. That is, the exposure result at any location on the wafer is mainly determined by the corresponding mask pattern and its neighboring area, while the influence of distant patterns is relatively limited. This characteristic is related to factors such as the attenuation of the point spread function response with increasing spatial distance in projection optics systems, and the limited range of influence of the photoresist reaction process. This feedback relationship can be abstracted into two basic characteristics: the range of influence and the spatial attenuation characteristics.

[0012] Construct localized convolution kernels, and determine their form based on the influence range of the photolithography process and its spatial attenuation characteristics: (2); in, Represents a local convolution kernel. This is a normalization coefficient used to maintain the consistency of the total feedback. The kernel size is used to describe the range of influence. This represents the number of standard deviation intervals covered by the convolution kernel, used to characterize the attenuation of feedback intensity with spatial distance; S22: Convolve the error field using a localized convolution kernel to obtain the mask-side exposure requirement field: (3); in, This indicates a demand for exposure. Represents the error field. Represents a local convolution kernel. Represents spatial coordinates.

[0013] S3: The implicit field is iteratively updated using the exposure demand field as the driving factor; compared with the explicit boundary representation, the implicit field avoids complex boundary topology changes and curve parameterization problems, and can naturally support morphological evolution such as boundary expansion, contraction, and structural splitting and merging, thus improving the freedom of mask design. Preferably, the iterative update relationship of the implicit field is expressed as: (4); in, and Let these represent the implicit fields corresponding to the t-th and t+1-th iterations, respectively. This represents the exposure demand field corresponding to the t-th iteration. The iteration step size, Represents spatial coordinates; It is worth noting that, since the implicit field uses cumulative updates, its boundary evolution depends on the feedback consistency across multiple iterations. Only when a specific region consistently exhibits underexposure will the boundary evolution be affected. When the field value in a given region decreases, the corresponding field value gradually increases, thus forming a reflection zone; conversely, when the field value decreases, an absorption zone is formed. This cumulative update mechanism helps smooth out perturbations in a single iteration and improves the evolutionary stability of the algorithm.

[0014] S4: Reconstruct the implicit field into an optimized mask pattern.

[0015] Preferably, the mask pattern reconstruction method includes pixel-based reconstruction or curve-based reconstruction.

[0016] Preferably, the pixelated mask pattern reconstruction method specifically includes the following steps: S411: Reconstruct the mask pattern based on the updated implicit field distribution, converting the implicit field into a pixelated mask form. The reconstruction relationship is expressed as follows: (5); in, and These represent the binary mask patterns corresponding to the t-th and t+1-th iterations, respectively. This represents the updated implicit field. Represents spatial coordinates; S412: Calculate the optimization evaluation index based on the pixelated mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; if the termination condition is not met, return to steps S1 to S4 to perform iterative optimization until the termination condition is met.

[0017] Preferably, the method for reconstructing the curved mask pattern specifically includes the following steps: S421: Reconstruct the mask pattern based on the updated implicit field distribution, use the zero contour lines of the implicit field to characterize the boundary of the mask pattern, and fit the boundary with a parametric curve to obtain a curved mask pattern. S422: Calculate the optimization evaluation index based on the curved mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; if the termination condition is not met, return to steps S1 to S4 to perform iterative optimization until the termination condition is met.

[0018] Compared with the prior art, the present invention can achieve the following beneficial effects: This invention employs a locality-guided optimization method that does not rely on gradient information, overcoming the dependence of traditional gradient optimization methods on model differentiability. This method can be combined with rigorously physically solved lithography simulation models, eliminating the need to adjust the strategy when replacing different physical models or introducing more complex physical effects. Simultaneously, it avoids the extensive simulation evaluation requirements of heuristic search in mask optimization, improving optimization efficiency and making it suitable for high-degree-of-freedom mask optimization problems. Compared to traditional iterative optimization methods, implicit field evolution overcomes the limitations of boundary representation methods on topological changes, expanding the design freedom of mask pattern optimization and enabling this invention to obtain high-quality optimization results. Furthermore, it supports both curve-based and pixel-based mask representations, accommodating the needs of data compression and discrete simulation computation. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of error field calculation provided according to an embodiment of the present invention; Figure 3 This is a schematic diagram of an iterative update process provided according to an embodiment of the present invention. Detailed Implementation

[0020] In the following description, embodiments of the invention will be described with reference to the accompanying drawings. In the description below, the same modules are denoted by the same reference numerals. Where the same reference numerals are used, their names and functions are also the same. Therefore, their detailed description will not be repeated.

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not constitute a limitation thereof.

[0022] The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0023] Figure 1 The flowchart of the extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution provided in the embodiments of the present invention is shown.

[0024] S1: Simulate the photolithography process of the extreme ultraviolet lithography mask, obtain the simulated physical response, and calculate the error field between it and the target physical response; The simulated physical response includes the photoresist contour pattern after exposure and development on the wafer side, the spatial image threshold pattern on the wafer side, and the near-field electric field intensity distribution on the mask side. The error field is represented as: (1); in, Represents the error field. Represents the distribution of the target physical response. Represents the simulated physical response distribution. Representing spatial coordinates (e.g.) Figure 2 (As shown).

[0025] Using the exposed and developed photoresist outline pattern as the simulated physical response, step S1 specifically includes the following steps: S111: Construct extreme ultraviolet lithography simulation models, including a mask near-field electromagnetic field simulation model based on waveguide method or finite-difference time-domain method, a projection imaging model based on Abbe imaging theory, and a photoresist exposure and development model based on Dill-Mack model. S112: Obtain the photoresist outline pattern after exposure and development through simulation, and calculate the error field between it and the target pattern.

[0026] Using spatial image threshold patterns as the simulated physical response includes the following steps: S121: Construct extreme ultraviolet lithography simulation models, including mask near-field electromagnetic field simulation models based on waveguide method or finite-difference time-domain method and projection imaging models based on Abbe imaging theory. S122: Obtain the spatial image intensity distribution on the wafer side through simulation, generate a spatial image threshold pattern according to a preset threshold, and calculate the error field between the pattern and the target pattern.

[0027] Using the near-field electric field intensity distribution of the mask as the simulated physical response, the specific steps include the following: S131: Construct a near-field electromagnetic field simulation model of a mask based on the waveguide method or the finite-difference time-domain method; S132: Obtain the near-field electric field intensity distribution of the mask through simulation, and calculate the error field between it and the target electric field intensity distribution.

[0028] S2: Based on locality, the error field is converted into the exposure demand field on the mask side; S21: Construct localized convolution kernels, determining their form based on the influence range and spatial attenuation characteristics of the photolithography process: (2); in, Represents a local convolution kernel. This is a normalization coefficient used to maintain the consistency of the total feedback. The kernel size is used to describe the range of influence. This represents the number of standard deviation intervals covered by the convolution kernel, used to characterize the attenuation of feedback intensity with spatial distance; S22: Convolve the error field using a localized convolution kernel to obtain the mask-side exposure requirement field: (3); in, This indicates a demand for exposure. Represents the error field. Represents a local convolution kernel. Represents spatial coordinates.

[0029] S3: Iteratively update the implicit field using the exposure demand field as the driving factor; In this embodiment, before the iterative update begins, the implicit field is... Perform initialization, with the initialization form being zero field; The iterative update relationship of the implicit field is represented as: (4); in, and Let these represent the implicit fields corresponding to the t-th and t+1-th iterations, respectively. This represents the exposure demand field corresponding to the t-th iteration. The iteration step size, Represents spatial coordinates.

[0030] The implicit field is reconstructed into an optimized mask pattern.

[0031] Mask pattern reconstruction methods include pixel-based reconstruction or curve-based reconstruction.

[0032] The pixelated mask pattern reconstruction method specifically includes the following steps: In this embodiment, before the iterative update begins, the pixelated mask pattern is... Perform initialization, using the target pattern as the initialization format; S411: Reconstruct the mask pattern based on the updated implicit field distribution, converting the implicit field into a pixelated mask form. The reconstruction relationship is expressed as follows: (5); in, and These represent the binary mask patterns corresponding to the t-th and t+1-th iterations, respectively. This represents the updated implicit field. Represents spatial coordinates; S412: Calculate the optimization evaluation index based on the pixelated mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; if the termination condition is not met, return to steps S1 to S4 for iterative optimization until the termination condition is met (e.g., ...). Figure 3 (As shown).

[0033] The method for reconstructing the curved mask pattern specifically includes the following steps: In this embodiment, before the iterative update begins, the curved mask pattern is initialized, and the initialization form is the target pattern; S421: Reconstruct the mask pattern based on the updated implicit field distribution, use the zero contour lines of the implicit field to characterize the boundary of the mask pattern, and fit the boundary with a parametric curve to obtain a curved mask pattern. S422: Calculate the optimization evaluation index based on the curved mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; if the termination condition is not met, return to steps S1 to S4 for iterative optimization until the termination condition is met (e.g., ...). Figure 3 (As shown).

[0034] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0035] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for optimizing extreme ultraviolet lithography masks based on locality-guided implicit field evolution, characterized in that, Specifically, the steps include the following: S1: Simulate the photolithography process of the extreme ultraviolet lithography mask, obtain the simulated physical response, and calculate the error field between it and the target physical response; S2: Based on locality, the error field is converted into the exposure demand field on the mask side; S3: Iteratively update the implicit field using the exposure demand field as the driving factor; S4: Reconstruct the implicit field into an optimized mask pattern.

2. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 1, characterized in that, The simulated physical response in step S1 includes the photoresist contour pattern after exposure and development on the wafer side, the spatial image threshold pattern on the wafer side, and the near-field electric field intensity distribution on the mask side. The error field is represented as: （1）； in, Represents the error field. Represents the distribution of the target physical response. Represents the simulated physical response distribution. Represents spatial coordinates.

3. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 2, characterized in that, In step S1, the photoresist outline pattern after exposure and development is used as the simulated physical response, specifically including the following steps: S111: Construct extreme ultraviolet lithography simulation models, including mask near-field electromagnetic field simulation models, projection imaging models, and photoresist exposure and development models; S112: Obtain the photoresist outline pattern after exposure and development through simulation, and calculate the error field between it and the target pattern.

4. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 2, characterized in that, In step S1, using the spatial image threshold pattern as the simulated physical response specifically includes the following steps: S121: Construct an extreme ultraviolet lithography simulation model, including a mask near-field electromagnetic field simulation model and a projection imaging model; S122: Obtain the spatial image intensity distribution on the wafer side through simulation, generate a spatial image threshold pattern according to a preset threshold, and calculate the error field between the pattern and the target pattern.

5. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 2, characterized in that, In step S1, the near-field electric field intensity distribution of the mask is used as the simulated physical response, which specifically includes the following steps: S131: Construct a near-field electromagnetic field simulation model for the mask; S132: Obtain the near-field electric field intensity distribution of the mask through simulation, and calculate the error field between it and the target electric field intensity distribution.

6. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 1, characterized in that, Step S2 specifically includes the following steps: S21: Construct localized convolution kernels, determining their form based on the influence range and spatial attenuation characteristics of the photolithography process: （2）； in, Represents a local convolution kernel. This is a normalization coefficient used to maintain the consistency of the total feedback. The kernel size is used to describe the range of influence. This represents the number of standard deviation intervals covered by the convolution kernel, used to characterize the attenuation of feedback intensity with spatial distance; S22: Convolve the error field using a localized convolution kernel to obtain the mask-side exposure requirement field: （3）； in, This indicates a demand for exposure. Represents the error field. Represents a local convolution kernel. Represents spatial coordinates.

7. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 1, characterized in that, In step S3, the iterative update relationship of the implicit field is expressed as follows: （4）； in, and Let these represent the implicit fields corresponding to the t-th and t+1-th iterations, respectively. This represents the exposure demand field corresponding to the t-th iteration. The iteration step size, Represents spatial coordinates.

8. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 1, characterized in that, In step S4, the mask pattern reconstruction method includes pixel-based reconstruction or curve-based reconstruction.

9. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 8, characterized in that, The pixelated mask pattern reconstruction method specifically includes the following steps: S411: Reconstruct the mask pattern based on the updated implicit field distribution, converting the implicit field into a pixelated mask form. The reconstruction relationship is expressed as follows: （5）； in, and These represent the binary mask patterns corresponding to the t-th and t+1-th iterations, respectively. This represents the updated implicit field. Represents spatial coordinates; S412: Calculate the optimization evaluation index based on the pixelated mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; if the termination condition is not met, return to steps S1 to S4 to perform iterative optimization until the termination condition is met.

10. The extreme ultraviolet lithography mask optimization method based on locality-guided implicit field evolution according to claim 8, characterized in that, The method for reconstructing the curved mask pattern specifically includes the following steps: S421: Reconstruct the mask pattern based on the updated implicit field distribution, use the zero contour lines of the implicit field to characterize the boundary of the mask pattern, and fit the boundary with a parametric curve to obtain a curved mask pattern. S422: Calculate the optimization evaluation index based on the curved mask pattern, and determine whether the current mask pattern meets the preset optimization termination condition; If the termination condition is not met, return to steps S1 to S4 for iterative optimization until the termination condition is met.

Images