High-fidelity digital photoetching method based on adaptive curvature segmentation and region fitting

Abstract

The invention relates to a high-fidelity digital photoetching method based on self-adaptive curvature segmentation and region fitting, which comprises the following steps: acquiring a to-be-processed exposure image, performing curvature analysis and self-adaptive segmentation on the to-be-processed exposure image, and acquiring a plurality of sub-images; fitting the sub-images according to the curvature features of the sub-images, wherein fitting comprises linear fitting, polynomial fitting and local fitting; and obtaining a displacement path of the three-dimensional nano mobile platform based on a fitting result, calculating a local expansion effect of the sub-image through the displacement path, performing preprocessing, and further performing exposure on the preprocessed sub-image to obtain an exposure image. According to the method, the problems of edge sawteeth and line width errors in DMD photoetching are effectively solved through self-adaptive curvature segmentation and a region specificity fitting strategy, the exposure quality and size precision of complex patterns are remarkably improved, and the method is suitable for the fields of high-resolution DMD photoetching, multi-layer patterning, free morphology microstructure manufacturing and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital micro-mirror device (DMD) imaging exposure, and particularly to a high-fidelity digital lithography method based on adaptive curvature segmentation and region fitting. BACKGROUND

[0002] Digital micro-mirror device (DMD) technology has become a cornerstone of digital lithography technology in the fields of semiconductor manufacturing, biomedical engineering and additive manufacturing due to its cost-effectiveness, high throughput and system compactness. This technology can replace traditional photomasks to achieve rapid patterning of printed circuit boards, tissue scaffolds and micro-optical elements. However, the inherent discrete nature of the micro-mirror array can cause pixelization effects at the edges of the pattern, resulting in obvious jagged lines (>1 μm) and line width deviations. These defects seriously affect the dimensional accuracy in sub-micron applications, hindering the application of DMD in high-resolution lithography.

[0003] To solve this problem, global researchers have proposed various improvement schemes. For example, the existing technical solution uses a wobble technique to compensate for the jaggedness by controlling the sub-pixel level relative displacement between the DMD micro-mirror and the stage, but this method requires a high-precision displacement stage, significantly increasing the system complexity. The existing technical solution also proposes an optical proximity correction (OPC) technique based on a genetic algorithm to pre-compensate for line width errors in mask design, but its iterative optimization requires a large number of calculation cycles, limiting real-time exposure applications. Gray-scale lithography smooths the edges by adjusting the exposure energy gradient, but requires a 12-bit high-precision light intensity modulation module, resulting in a significant increase in equipment cost. In addition, optimization strategies based on pixel overlap (such as pulse scanning exposure, intensity modulation OPC) can fill in the geometric discontinuities through sub-pixel displacement, reducing the jagged error to 45-55% of the initial value. Since 2016, existing edge jaggedness compensation strategies (such as wobble technology, pulse scanning, cooperative modulation) have reduced the jagged error by an average of about 50%. The scheme for optimizing edge roughness through a large number of sub-patterns (such as time-space modulation technology) requires more than 400 continuous overlapping sub-patterns to improve edge smoothness. The previously proposed dynamic blur exposure technology based on curvature effectively alleviates edge jaggedness through curve fitting combined with a three-dimensional nanometer-level positioning platform. Although this method is effective for simple geometric structures, its dependence on fixed thresholds and global fitting limits its adaptability to non-linear curved surface patterns. In view of the limitations of existing methods. SUMMARY

[0004] In order to solve the problems existing in the prior art, the purpose of the present application is to provide a high-fidelity digital lithography method based on adaptive curvature segmentation and regional fitting, which can dynamically divide the edge region according to the local curvature through an adaptive threshold segmentation technology, and realize flexible regulation of the number of sub-regions through parameter adjustment, so as to balance efficiency and accuracy, and can also flexibly adopt optimal fitting strategies such as least square method or spline method to process different sub-regions, while effectively preserving linear and nonlinear boundary characteristics, significantly reducing fitting distortion. The present application realizes high-quality optimization of complex pattern edges, effectively solves the problems of edge sawtooth and line width error in DMD lithography.

[0005] In order to achieve the above purpose, the present application provides the following scheme:

[0006] The high-fidelity digital lithography method based on adaptive curvature segmentation and regional fitting comprises:

[0007] Obtaining a to-be-processed exposure image, performing curvature analysis and adaptive segmentation on the to-be-processed exposure image, and obtaining a plurality of sub-images;

[0008] According to the curvature characteristics of the sub-images, fitting the sub-images, the fitting comprising linear fitting, polynomial fitting and local fitting;

[0009] Based on the fitting result, obtaining a displacement path of a three-dimensional nanometer moving platform, calculating the local swelling effect of the sub-images through the displacement path, and preprocessing, further exposing the preprocessed sub-images to obtain an exposure image.

[0010] Optionally, the curvature analysis on the to-be-processed exposure image comprises:

[0011] Calculating the first-order gradient and the second-order partial derivative of the to-be-processed exposure image, and according to the first-order gradient and the second-order partial derivative, obtaining the curvature value of each pixel point in the to-be-processed exposure image:

[0012]

[0013] wherein, is the curvature value of each pixel point, G xx , G yy , G xy is the second-order partial derivative, G x , G y is the first-order gradient.

[0014] Optionally, the adaptive segmentation on the to-be-processed exposure image comprises:

[0015]

[0016] wherein, a self-adaptive threshold, a global median of curvatures of the whole image, median(k i−ω:i+ω represents a local curvature median with a current point as a center and a window width as ω, as a local reference basis, mad(k i−ω:i+ω represents a median absolute deviation of curvatures in the window, and β is a weight parameter.

[0017] Optionally, the median absolute deviation of curvatures in the window comprises:

[0018]

[0019] wherein, is a curvature value of a jth point in the window, is a median of all curvatures in the window.

[0020] Optionally, fitting the sub-image according to the curvature feature of the sub-image comprises:

[0021] when an absolute value of a curvature of a pixel point in the sub-image is less than a first threshold, a region corresponding to the pixel point is regarded as a low curvature region, and the low curvature region is linearly fitted;

[0022] when the absolute value of the curvature of the pixel point in the sub-image is greater than the first threshold and less than a second threshold, the region corresponding to the pixel point is regarded as a medium-high curvature region, and the medium-high curvature region is polynomially fitted;

[0023] when the absolute value of the curvature of the pixel point in the sub-image is greater than the second threshold, the region corresponding to the pixel point is regarded as an extremely high curvature region, and the extremely high curvature region is locally fitted after being subdivided.

[0024] Optionally, the displacement path of the three-dimensional nanometer moving platform comprises:

[0025] When the absolute value of the fitting result is greater than a third threshold value, then the x-axis of the three-dimensional nanometer moving platform is moved by a target pixel distance, and the y-axis is moved by Ly=Lx*k; when the absolute value of the fitting result is less than the third threshold value, then the y-axis of the three-dimensional nanometer moving platform is moved by the target pixel distance, and the x-axis is moved by Lx=Ly / k; when the absolute value of the fitting result is equal to the third threshold value, then the x-axis and the y-axis of the three-dimensional nanometer moving platform are both moved by the target pixel distance; wherein Ly is the displacement of the y-axis, Lx is the displacement of the x-axis, k is a proportional coefficient also indicating curvature, used to correlate the displacement relationship between the x-axis and the y-axis, so that Lx and Ly are linked according to the rule of Ly=Lx*k or Lx=Ly / k, Lx is the displacement of the x-axis, Ly is the displacement of the y-axis, and k is a proportional coefficient, used to correlate the displacement relationship between the x-axis and the y-axis, so that Lx and Ly are linked according to the rule of Ly=Lx*k or Lx=Ly / k.

[0026] Optionally, the preprocessing of the sub-image comprises:

[0027] The edge optimization processing of the sub-image comprises edge pixel elimination and energy distribution adjustment.

[0028] The sub-image after the edge optimization processing is finely adjusted by using the local swelling effect.

[0029] To achieve the above object, the application further provides a high-fidelity digital lithography system based on adaptive curvature segmentation and region fitting, comprising:

[0030] An adaptive segmentation module is configured to acquire a to-be-processed exposure image, perform curvature analysis and adaptive segmentation on the to-be-processed exposure image, and acquire a plurality of sub-images.

[0031] An image fitting module is configured to fit the sub-images according to the curvature features of the sub-images, wherein the fitting comprises linear fitting, polynomial fitting and local fitting.

[0032] An image exposure module is configured to acquire a displacement path of a three-dimensional nanometer moving platform based on the fitting result, calculate a local swelling effect of the sub-image through the displacement path, perform preprocessing, further expose the sub-image after the preprocessing, and acquire an exposure image.

[0033] To achieve the above object, the application further provides a medium storing a computer program, wherein the program is executed to perform the method.

[0034] The application has the following beneficial effects:

[0035] The application overcomes the limitations of traditional fixed threshold segmentation by adaptive curvature segmentation and region-specific fitting strategy, dynamically adjusts the segmentation threshold and fitting method, significantly improves the edge smoothness of the pattern by optimizing the motion trajectory of the three-dimensional nano platform and the dynamic blur exposure parameters, and ensures the reliability and practicability of the method by using the hopkins model for simulation verification.

[0036] The application reduces the edge sawtooth from 1.32 mu m to 0.15 mu m without changing the structure of the optical system, reduces the line width error rate by 45.16%, greatly improves the exposure quality and size precision of complex patterns, and provides an effective technical solution for high-precision micro-nano manufacturing. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0038] Figure 1 The flow chart of the high-fidelity digital lithography method based on adaptive curvature segmentation and region fitting of the embodiments of the present application;

[0039] Figure 2 The schematic diagram of the DMD digital lithography system structure of the embodiments of the present application;

[0040] Figure 3 The three-dimensional nano platform and DMD collaborative work schematic diagram of the embodiments of the present application; (a) is the target segment to be exposed with a slope of 1 / 2, (b) is the initial exposure position, (c) is the relative position between the DMD exposure mode and the diagonal line after moving the stage position, (d) is the return to the initial position after exposure, wherein the dark block represents the region optimized by the dynamic blur effect;

[0041] Figure 4 The image segmentation example diagram of the embodiments of the present application; (a) is a complex pattern to be exposed, (b) is a sub-mode 1, (c) is a sub-mode 2, (d) is a sub-mode 3, and (e) is a sub-mode 4;

[0042] Figure 5 The mask optimization comparison diagram under different fitting strategies of the embodiments of the present application; (a) is an original mask image without processing, (b) is a mask image with sub-pixel overlap optimized by using the conventional DBE method, and (c) is an image with sub-pixel overlap optimized by using the improved DBE method;

[0043] Figure 6Figure 1 is a comparison chart of non-linear pattern exposure results of an embodiment of the present application; (a) is an exposure pattern without processing, (b) is an exposure pattern processed using the original fitting strategy, and (c) is an exposure pattern processed using the improved fitting strategy;

[0044] Figure 7 Figure 2 is a comparison chart of Hopkins exposure simulation results of an embodiment of the present application; (a) is an initial mask image, (b) is a Hopkins simulation of the initial mask, (c) is an initial exposure simulation result, (d) is a mask image of the original optimization method, (e) is a Hopkins simulation of the mask of the original optimization method, (f) is an exposure simulation result of the original optimization method, (g) is a mask image after improved optimization, (h) is a Hopkins simulation of the improved optimization mask, and (i) is an exposure simulation result after improved optimization;

[0045] Figure 8 Figure 3 is a comparison chart of line width measurement of an embodiment of the present application; (a) is a line width before optimization, (b) is a line width after optimization using the traditional method, and (c) is a line width after optimization using the improved method;

[0046] Figure 9 Figure 4 is a comparison chart of experimental exposure results of an embodiment of the present application; (a) is an original exposure image, (b) is an exposure image after optimization, (c) is an exposure image after improved optimization, (d) is an edge of the original exposure image, (e) is an edge of the optimized exposure image, and (f) is an edge of the exposure image after improved optimization;

[0047] Figure 10 Figure 5 is a schematic diagram of experimental verification results of various closed patterns of an embodiment of the present application; (a) is an original Q-type closed pattern, (b) is an optimized Q-type closed pattern, (c) is an improved and optimized Q-type closed pattern, (d) is an original droplet-type closed pattern, (e) is an optimized droplet-type closed pattern, (f) is an improved and optimized droplet-type closed pattern, (g) is an original crescent-type closed pattern, (h) is an optimized crescent-type closed pattern, and (i) is an improved and optimized crescent-type closed pattern. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0049] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0050] As Figure 1 shown, the embodiment discloses a high-fidelity digital lithography method based on adaptive curvature segmentation and regional fitting, comprising: acquiring a to-be-processed exposure image, performing curvature analysis and adaptive segmentation on the to-be-processed exposure image, and acquiring a plurality of sub-images; fitting the sub-images according to the curvature characteristics of the sub-images, the fitting including linear fitting, polynomial fitting, and local fitting; based on the fitting result, acquiring a displacement path of a three-dimensional nanometer moving platform, calculating the local swelling effect of the sub-images through the displacement path, and performing preprocessing, and further exposing the preprocessed sub-images to acquire an exposure image.

[0051] Specifically, the embodiment discloses a high-fidelity digital lithography method based on adaptive curvature segmentation and regional fitting, comprising: selecting a to-be-processed exposure image; performing curvature analysis and adaptive segmentation on the to-be-processed exposure image to acquire a plurality of sub-images; selecting a fitting strategy according to the curvature characteristics of the sub-images and calculating corresponding parameters; designing a displacement path based on the fitting parameters; calculating the local swelling of the sub-images based on the displacement path and preprocessing the sub-images; exposing the preprocessed sub-images to acquire a complete exposure image.

[0052] Further, selecting the to-be-processed exposure image comprises: adjusting the DMD lithography system to a working state, so that the ultraviolet light source can pass through the engineering scattering sheet, the lens group, the DMD chip, the reflecting mirror, and the projection objective lens in the DMD lithography system in sequence, expose the target image on the photosensitive material on the three-dimensional nanometer moving platform, and obtain the to-be-processed exposure image.

[0053] As Figure 2 shown, after the ultraviolet light source (LUMINUS UV-LED, CBT-39-UV-C32-FB400-22) is turned on, it is projected onto the DMD through the lens group, and the resolution is 1920x1080 (the pixel size is 7.56 μm). The typical peak wavelength of the light source is 405 nm, but due to the large beam divergence angle, the light intensity is in a Gaussian distribution, and a large light energy loss will occur during propagation. In order to improve the utilization rate of the light source, the DMD lithography system adopts a uniform light module composed of an engineering scattering body (ED1-C20-MD, Thorlabs) and a lens group (1 plano-convex lens and 3 double-convex lenses placed in sequence). The DMD controller uses Texas Instruments DLP 6500 model. The photosensitive material uses W200 water-washed resin (eSUN), and precise positioning is achieved through a three-dimensional nanometer positioning platform (Coremorrow XYZ100, E70.D3S closed-loop piezoelectric controller).

[0054] In the aspect of substrate processing, clean cover glass is placed on the photosensitive resin, ensuring that no bubbles are generated at the contact position. The embodiment ensures that the ultraviolet light source can pass through the engineering scattering sheet, lens group, DMD chip, mirror, and projection objective lens to accurately expose the target image on the photosensitive material on the three-dimensional nanometer moving platform by precisely adjusting the DMD photolithography system.

[0055] As an optional embodiment, the three-dimensional nanometer moving platform adopts a piezoelectric nanometer ceramic moving platform, which has nanometer-level positioning accuracy and fast response characteristics.

[0056] Further, the curvature analysis on the to-be-processed exposure image includes: calculating a first-order gradient and a second-order partial derivative of the to-be-processed exposure image, and obtaining a curvature value of each pixel point in the to-be-processed exposure image according to the first-order gradient and the second-order partial derivative.

[0057] Specifically, the segmentation of the to-be-processed exposure image to obtain a plurality of sub-images includes: processing the to-be-processed exposure image by an adaptive curvature segmentation algorithm to obtain sub-images with different curvature characteristics. The process first calculates the first-order gradient G x and G y of the image by a Sobel operator, and then calculates the second-order partial derivatives G xx , G yy and G xy based on the Hessian matrix, and finally calculates the curvature value k of each pixel point by the following formula:

[0058]

[0059] The segmentation and processing of the sub-image are realized by an adaptive threshold segmentation algorithm:

[0060]

[0061] wherein median(k i−ω:i+ω ) represents the local curvature median with the current point as the center and the window width as ω, which is taken as a local reference benchmark, and C is the global median of the entire image curvature, and the calculation formula is as follows:

[0062]

[0063] wherein median(k global ) represents the global curvature median, k global represents a set of global curvatures, and C is taken as a global offset to ensure that the segmentation strategy can respond to the overall characteristics of the image.

[0064] mad(k i−ω:i+ω ) represents the median absolute deviation of the curvature in the above window, and the calculation formula is as follows: , to quantify the fluctuation intensity of local curvature.

[0065] α and β are adjustable weight parameters to cooperatively control the contribution of global offset and local fluctuation to the threshold value. By adjusting α and β, the balance between the fineness and efficiency of segmentation can be achieved, and the number of generated sub-images can be controlled.

[0066] The determination of the detection window width ω needs to follow the Nyquist sampling theorem to ensure the accurate capture of edge curvature features. Its value is determined by the following formula:

[0067]

[0068] where f max is the highest spatial frequency of edge curvature. In the experimental system of the embodiment, the resolution of a DMD single pixel on the actual substrate is 1.89 μm. Considering factors such as pixel discreteness, processing efficiency, and noise interference, the preferred window width is set to ω = 5 after calculation and rounding.

[0069] The value of the weight parameter β is related to the statistical properties of curvature fluctuation. For approximately normally distributed curvature features, the value of β can be determined by reference to the ratio of the curvature standard deviation σ k to the median absolute deviation mad(k). The value is determined by the following formula:

[0070]

[0071] As a preferred embodiment, β = 1.5 is taken in the embodiment to achieve a good balance between the sensitivity and stability of the dynamic threshold.

[0072] The weight parameter α is usually valued between 0.3 and 0.7. Adjusting α can intuitively control the number of sub-patterns generated by image segmentation. In the embodiment, α = 0.5 is selected after simulation analysis for fair performance comparison under the same number of sub-images.

[0073] Further, fitting the sub-image according to the curvature features of the sub-image comprises: when the absolute value of the curvature of a pixel point in the sub-image is less than a first threshold value, the region corresponding to the pixel point is regarded as a low-curvature region, and the low-curvature region is linearly fitted; when the absolute value of the curvature of a pixel point in the sub-image is greater than the first threshold value and less than a second threshold value, the region corresponding to the pixel point is regarded as a medium-high curvature region, and the medium-high curvature region is polynomially fitted; when the absolute value of the curvature of a pixel point in the sub-image is greater than the second threshold value, the region corresponding to the pixel point is regarded as an extremely high curvature region, and the extremely high curvature region is locally fitted after being subdivided.

[0074] Specifically, different fitting strategies are adopted based on the curvature characteristics of the sub-images. The curvature characteristics are determined by calculating the statistical distribution (such as median and standard deviation) of the curvature values ​​of all pixels within the sub-image. For low curvature regions (|k| < 0.05), the least squares method is used for linear fitting, and the fitting function is:

[0075]

[0076] Where x is the x-coordinate of the pixel, y is the y-coordinate of the pixel, b is the intercept, and the slope a is calculated using the following formula:

[0077]

[0078] The change in the vertical axis, This represents the change in the horizontal axis.

[0079] Parameters a and b are theoretical values, but their calculation formulas are difficult to determine in practical engineering applications. and Therefore, the estimated value is calculated using the residual sum formula below through an algorithm. , To replace the actual a and b. This illustrates the difference between local conventions and general expressions:

[0080]

[0081]

[0082] in, To fit the estimated slope of the straight line, To fit the estimated intercept of the line, Let x be the x-coordinate of the i-th pixel in the sub-image. Let be the ordinate of the i-th pixel in the sub-image.

[0083] For the medium-to-high curvature region (0.05≤|k|<0.2), a quadratic polynomial fitting is used:

[0084]

[0085] in, y is the x-coordinate of the pixel, c, d, and e are the quadratic polynomial fitting coefficients.

[0086] The coefficients c, d, and e are obtained by solving the following system of linear equations:

[0087]

[0088] After determining the quadratic polynomial fitting function, the average slope a2 is calculated:

[0089]

[0090] where x end is the right endpoint of the interval (i.e., the endpoint of the range of values of the independent variable x when calculating the average slope), and x start is the left endpoint of the interval (i.e., the starting point of the range of values of the independent variable x when calculating the average slope).

[0091] For regions with extremely high curvature (∣k∣≥0.2), further subdivision is performed, and a local fitting strategy is used to avoid distortion in the fitting.

[0092] Further, the displacement path of the three-dimensional nanometer moving platform includes: when the absolute value of the fitting result is greater than a third threshold, moving the three-dimensional nanometer moving platform by a target pixel distance on the x-axis and Ly=Lx⋅k on the y-axis; when the absolute value of the fitting result is less than the third threshold, moving the three-dimensional nanometer moving platform by a target pixel distance on the y-axis and Lx=Ly / k on the x-axis; and when the absolute value of the fitting result is equal to the third threshold, moving the three-dimensional nanometer moving platform by a target pixel distance on the x-axis and y-axis.

[0093] Specifically, the displacement path is designed based on the fitting parameters, including:

[0094] According to the slope value a2 obtained by fitting, the moving path of the three-dimensional nanometer moving platform is designed. When ∣k∣>1, the x-axis moves by a distance of 0.5 pixels (0.945 μm), and the y-axis moves by Ly=Lx⋅k; when ∣k∣<1, the y-axis moves by a distance of 0.5 pixels, and the x-axis moves by Lx=Ly / k; and when ∣k∣=1, the x-axis and y-axis both move by a distance of 0.5 pixels, as shown in Figure 3 (a)-(d).

[0095] Further, the preprocessing of the sub-image includes edge optimization processing, which includes edge pixel elimination and energy distribution adjustment; and fine adjustment of the sub-image after edge optimization processing using local inflation effect.

[0096] Specifically, the parameters of the dynamic blur are determined based on the displacement path, and a corresponding motion blur kernel is generated. The motion blur kernel is a two-dimensional convolution template, whose shape is defined by the vector direction of the displacement and the scalar length in the image pixel scale, which is used to predict and compensate for the light intensity diffusion effect (i.e., local inflation) caused by platform motion in the physical exposure process in the digital domain. The blur kernel matrix is constructed as follows:

[0097]

[0098] where (ci ,c j ) is the nuclear coordinate, length is the blur length, and sum is the normalization coefficient.

[0099] The preprocessing of the sub-image includes edge optimization processing of the sub-image using a Python image processing algorithm, mainly including edge pixel elimination and energy distribution adjustment. The local expansion effect is calculated through a motion blur kernel function, and then refined adjustment is performed using an image processing algorithm to obtain the preprocessed sub-image, as shown in Figure 4 (a)-(e).

[0100] The exposure of the preprocessed sub-image includes:

[0101] Further, a Hopkins exposure model is used to simulate and verify the preprocessed sub-image. The model is based on frequency domain analysis and can accurately simulate the imaging process of partially coherent light. The core is to multiply the frequency spectrum O(f x ,f y ) of the mask and the transfer function H(f x ,f y ) of the optical system in the frequency domain, and then obtain the spatial light intensity distribution of the image plane through inverse Fourier transform, and the calculation formula is as follows:

[0102]

[0103] where O(f x ,f y ) is the mask frequency spectrum, H(f x ,f y ) is the optical transfer function, represents the two-dimensional inverse Fourier transform.

[0104] Further, the exposure process further includes: sorting the sub-images according to the curvature characteristics, and sequentially performing exposure. For sub-images with the same curvature, a parallel exposure strategy is used to improve efficiency. During the exposure process, the motion trajectory of the three-dimensional nanometer moving platform and the pattern switching of the DMD are accurately controlled to realize high-quality pattern splicing, as shown in Figure 5 (a)-(c).

[0105] The specific exposure implementation process is as follows:

[0106] A Python processing program is created through a dynamic blur kernel function, the preprocessed sub-image is imported into the system, and the corresponding displacement parameters are set to obtain the exposure energy distribution graph optimized through the dynamic blur effect, as shown in Figure 6 (a)-(c).

[0107] The energy simulation results are imported into the Hopkins Python exposure model to obtain the optimized exposure simulation results. Compared with the simulation results before optimization, as shown in Figure 7 (a)-(i), it can be observed that the edge quality is improved.

[0108] In the actual exposure process, the immersion exposure method is adopted, and the photosensitive resin and the substrate are placed on a three-dimensional moving platform. Each sub-image is imported into the DMD controller, and the corresponding exposure time parameter is set. At the same time, the displacement data corresponding to each sub-image is imported into the moving platform control system to ensure the strict synchronization of the image sequence and the displacement data.

[0109] When the ultraviolet light source is activated, the light beam irradiates on the photosensitive resin surface, and the platform starts to move according to the preset trajectory, and the sub-images are exposed in turn. Under the illumination intensity of 8.11 mW / cm², the exposure time of each sub-image is controlled within 6-7 seconds. After each sub-image is exposed, the DMD and the moving platform will be temporarily paused, and the exposure process will continue after the next sub-image and displacement data are ready.

[0110] After all the sub-images are exposed, the complete exposure pattern is obtained through the image stitching algorithm. The optimized exposure experimental results are compared with those before optimization, as shown in Figure 8 (a)-(c) and Figure 9 (a)-(c), it can be observed that the edge sawtooth is reduced and the line width precision is improved. To further demonstrate the efficiency and applicability of the present application, the experimental verification results of a variety of closed patterns are supplemented, as shown in Figure 10 (a)-(i).

[0111] This embodiment realizes high-quality optimization of complex pattern edges through adaptive curvature segmentation and region-specific fitting strategies. By using the high-precision positioning capability of the three-dimensional nanometer moving platform and the spatial light modulation characteristics of the DMD, combined with the motion blur effect and the optimization algorithm, the edge quality and size precision of the exposure pattern are significantly improved. This method is not only suitable for conventional microstructure processing, but also can be used for multi-layer patterning and complex free-form manufacturing, and has a wide application prospect.

[0112] The embodiment also provides a high-fidelity digital lithography system based on adaptive curvature segmentation and region fitting, comprising: an adaptive segmentation module, configured to acquire an exposure image to be processed, perform curvature analysis and adaptive segmentation on the exposure image to be processed, and acquire a plurality of sub-images; an image fitting module, configured to fit the sub-images according to curvature characteristics of the sub-images, the fitting comprising linear fitting, polynomial fitting and local fitting; and an image exposure module, configured to acquire a displacement path of a three-dimensional nanometer moving platform based on the fitting result, calculate a local swelling effect of the sub-images through the displacement path, perform preprocessing, further expose the preprocessed sub-images, and acquire an exposure image.

[0113] The embodiment also provides a medium storing a computer program, the program being executed to perform the method.

[0114] The above-described embodiments are merely used to describe the preferred modes of the present application, and are not used to limit the scope of the present application, and various modifications and improvements made by those skilled in the art to the technical solutions of the present application without departing from the design spirit of the present application shall fall within the protection scope of the present application defined by the claims.

Claims

1. A high fidelity digital lithography method based on adaptive curvature segmentation and area fitting, characterized in that, The method comprises the following steps: obtaining a to-be-processed exposure image, performing curvature analysis and adaptive segmentation on the to-be-processed exposure image, and obtaining a plurality of sub-images; fitting the sub-images according to the curvature characteristics of the sub-images, wherein the fitting comprises linear fitting, polynomial fitting and local fitting; based on the fitting result, obtaining a displacement path of a three-dimensional nanometer moving platform, calculating the local swelling effect of the sub-image through the displacement path, and performing preprocessing, and further exposing the preprocessed sub-image to obtain an exposure image.

2. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 1, wherein, The curvature analysis on the to-be-processed exposure image comprises: calculating the first-order gradient and the second-order partial derivative of the to-be-processed exposure image, and obtaining the curvature value of each pixel point in the to-be-processed exposure image according to the first-order gradient and the second-order partial derivative: wherein, is the curvature value of each pixel point, G xx , G yy , G xy is the second-order partial derivative, G x , G y is the first-order gradient.

3. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 1, wherein, The adaptive segmentation of the to-be-processed exposure image comprises: wherein, is an adaptive threshold, is a global median of the curvature of the entire image, median(k i−ω:i+ω ) represents a local median of the curvature centered at the current point with a window width of is used as a local reference, and mad(k i−ω:i+ω ) represents the median absolute deviation of the curvature within the window, and β is a weight parameter.

4. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 3, wherein, obtaining the median absolute deviation of the curvature in the window comprises: wherein, is the curvature value of the jth point of the window, is the median of all curvatures in the window range.

5. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 1, wherein, The fitting of the sub-image according to the curvature characteristics of the sub-image comprises: when the absolute value of the curvature of a pixel point in the sub-image is less than a first threshold value, the region corresponding to the pixel point is regarded as a low-curvature region, and linear fitting is performed on the low-curvature region; when the absolute value of the curvature of a pixel point in the sub-image is greater than the first threshold value and less than a second threshold value, the region corresponding to the pixel point is regarded as a medium-high curvature region, and polynomial fitting is performed on the medium-high curvature region; when the absolute value of the curvature of a pixel point in the sub-image is greater than the second threshold value, the region corresponding to the pixel point is regarded as an extremely high curvature region, and local fitting is performed after subdivision on the extremely high curvature region.

6. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 1, wherein, The displacement path of the three-dimensional nanometer moving platform comprises: when the absolute value of the fitting result is greater than a third threshold value, the three-dimensional nanometer moving platform is moved by a target pixel distance on the x-axis and Ly=Lx·k on the y-axis; when the absolute value of the fitting result is less than the third threshold value, the three-dimensional nanometer moving platform is moved by the target pixel distance on the y-axis and Lx=Ly / k on the x-axis; when the absolute value of the fitting result is equal to the third threshold value, the three-dimensional nanometer moving platform is moved by the target pixel distance on the x-axis and the y-axis; wherein Ly is the displacement amount of the y-axis, Lx is the displacement amount of the x-axis, k is the proportion coefficient, which is also the curvature, and is used to associate the displacement relationship between the x-axis and the y-axis, so that Lx and Ly are linked according to the rule of Ly=Lx·k or Lx=Ly / k.

7. The adaptive curvature segmentation and region fitting based high fidelity digital lithography method of claim 1, wherein, The preprocessing of the sub-image comprises: edge optimization processing is performed on the sub-image, and the edge optimization processing comprises edge pixel elimination and energy distribution adjustment; the local swelling effect is used to perform fine adjustment on the sub-image after the edge optimization processing.

8. The high fidelity digital lithography system based on adaptive curvature segmentation and area fitting of any of claims 1-7, wherein, The method comprises the following steps: an adaptive segmentation module is used to obtain a to-be-processed exposure image, perform curvature analysis and adaptive segmentation on the to-be-processed exposure image, and obtain a plurality of sub-images; an image fitting module, configured to fit the sub-image according to a curvature feature of the sub-image, the fitting including linear fitting, polynomial fitting and local fitting; an image exposure module, configured to obtain a displacement path of the three-dimensional nanometer moving platform based on the fitting result, calculate a local swelling effect of the sub-image through the displacement path, and perform preprocessing, and further expose the preprocessed sub-image to obtain an exposed image.

9. A medium storing a computer program, characterized by The program is executed to implement the method of any one of claims 1-7.

Images