Imaging system and method for image acquisition

Abstract

The application discloses an imaging system and a method for image acquisition, and relates to the technical field of image acquisition. The disclosed imaging system comprises: an image acquisition module configured to control an electron beam to scan a sample according to a first angle in a preset angle sequence, so as to generate a first scanning image; an image processing module configured to acquire a plurality of scanning images corresponding to a plurality of angles in the preset angle sequence; fuse the plurality of scanning images, so as to obtain a composite image; perform morphing processing on the composite image, so as to obtain an output image; and calculate a correction coefficient based on the sizes of the composite image and the output image, so as to correct measurement data of the output image. The technical scheme of the application can utilize the imaging system to acquire an output image of a sample to be imaged, which effectively suppresses the charging effect artifact, improves the compatibility of the output image with software, and the accuracy of measurement data obtained based on the output image.

Description

Technical Field

[0001] This application generally relates to the field of image acquisition technology. More specifically, this application relates to an imaging system and a method for image acquisition. Background Technology

[0002] Scanning electron microscopy (SEM) uses a focused electron beam to scan a sample and generates an image by detecting the excitation signal. In semiconductor manufacturing and inspection, SEM is widely used for measuring critical dimensions (CDs) and observing structures. However, when imaging semiconductor materials with SEM, a charging effect is prone to occur. This charging effect degrades image quality, causing image distortion, unclear structural boundaries, and affecting the accuracy of subsequent image analysis and CD measurements. To address the charging effect problem, HOLON has proposed a rotating scanning scheme: using an electron beam to acquire mask images from different scanning directions, performing pixel-to-pixel scanning of the electron beam to acquire pixel signals at the same location on the mask and generate multiple images, and then synthesizing a high-precision mask image from these multiple images.

[0003] Some factory automation software (such as CD-SEM automated measurement machines) mandates that input images be of standard size to match recipe settings, and does not support input images of non-standard sizes. When the size of a SEM scan image, or a composite image obtained by fusing SEM scan images, differs from the standard size, deformation processing is required. This deformation processing alters the pixel size in the image (i.e., the actual physical length of the sample corresponding to a single pixel). Performing critical dimension measurements or structural feature observations based on the deformed image can lead to erroneous results. However, current rotational scanning solutions lack a dynamic correction mechanism for pixel size changes. After images undergo rotation, resampling, stretching, and other deformation processing, without precise physical dimension traceability, systematic deviations in measurement results will occur, failing to meet the stringent requirements for measurement repeatability in advanced processes.

[0004] In view of this, this application proposes an imaging system and a method for image acquisition, so as to acquire images of the sample to be imaged that effectively suppress charging effect artifacts, improve the compatibility of the images with software, and improve the accuracy of the measurement data acquired based on the images. Summary of the Invention

[0005] In order to at least solve one or more of the technical problems mentioned above, this application proposes an imaging system and a method for image acquisition in several aspects.

[0006] In a first aspect, this application provides an imaging system, comprising: an image acquisition module configured to: control an electron beam to scan a sample according to a first angle in a preset angle sequence, so as to generate a first scan image; and an image processing module configured to: acquire multiple scan images corresponding to multiple angles in the preset angle sequence; fuse the multiple scan images to obtain a composite image; perform deformation processing on the composite image to obtain an output image; and calculate a correction coefficient based on the dimensions of the composite image and the output image to correct measurement data of the output image.

[0007] In a second aspect, this application provides a method for image acquisition based on an imaging system as described in the first aspect.

[0008] This application controls the electron beam to scan and image a sample based on multiple angles in a preset angle sequence. After obtaining a scanned image based on each angle in the preset angle sequence, multiple scanned images are fused to obtain a composite image. Since the composite image incorporates scanned images from multiple angles, it helps suppress charging artifacts in the scanned images at each angle. Furthermore, this application performs deformation processing on the composite image to obtain an output image, ensuring that the output image conforms to the format requirements of the software processing. Additionally, this application calculates correction coefficients based on the dimensions of the composite image and the output image, and uses these correction coefficients to correct the measurement data of the output image. This helps ensure the measurement accuracy of the output image under any operating condition, ensuring that the physical accuracy of CD measurements is not affected by the image processing flow. Attached Figure Description

[0009] The above and other objects, features, and advantages of exemplary embodiments of this application will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the drawings, several embodiments of this application are illustrated by way of example and not limitation, and the same or corresponding reference numerals denote the same or corresponding parts, wherein:

[0010] Figure 1 An exemplary structural diagram of a SEM in some embodiments of this application is shown.

[0011] Figure 2 Exemplary schematic diagrams of SEM imaging in some embodiments of this application are shown.

[0012] Figure 3 An exemplary structural schematic diagram of an imaging system in some embodiments of this application is shown.

[0013] Figure 4 An exemplary structural diagram of an image processing module in some embodiments of this application is shown.

[0014] Figure 5a An exemplary schematic diagram of the electron beam scanning angle in some embodiments of this application is shown.

[0015] Figure 5b An exemplary schematic diagram is shown in some embodiments of this application, illustrating the determination of the scanning area based on a first angle and a first scanning size.

[0016] Figure 5c An exemplary schematic diagram of the target observation field of view outside the scanning area in some embodiments of this application is shown.

[0017] Figure 6 I0, I are shown in some embodiments of this application. 180 and I e .

[0018] Figure 7 I0, I are shown in some embodiments of this application. 90 I 180 I 270 I e and I f . Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] It should be understood that the terms "comprising" and "including" used in the specification and claims of this application indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application. As used in this specification and claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this specification and claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations.

[0022] As used in this specification and claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0023] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise expressly specified. "Several" means one or more, unless otherwise expressly specified.

[0024] The specific embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0025] SEM uses a focused electron beam to scan a sample and generates an image by detecting the excitation signal.

[0026] Figure 1 Exemplary structural diagrams of the SEM100 in some embodiments of this application are shown. For example... Figure 1 As shown, in some embodiments, the SEM 100 includes: an electron gun 101, an electron beam 102, a condenser lens 103, an objective aperture 104, a deflection coil 105, an objective lens 106, a sample chamber 107, a sample 108, a stage 109, an electron detector 110, and a display screen 111. The electron gun 101 can be implemented as a thermionic electron gun or a field emission electron gun to generate the electron beam 102. The condenser lens 103 focuses and shapes the electron beam 102. The objective aperture 104 limits the aperture angle of the electron beam 102, filtering out large-angle stray electrons. The deflection coil 105 controls the electron beam 102 to scan the sample 108 based on a preset scanning mode to facilitate image construction. The objective lens 106 focuses the electron beam 102 onto the sample 108. The sample chamber 107 includes the sample 108 and the stage 109. The sample 108 can be fixed on the stage 109, and the stage 109 can move horizontally, vertically, and perpendicularly to the plane of the stage. Furthermore, the stage 109 can also rotate and tilt the sample 108.

[0027] continue Figure 1Once the electron beam 102 reaches the sample 108 and interacts with it, it generates various detectable excitation signals, such as secondary electrons, backscattered electrons, and X-rays. The electron detector 110 may include a secondary electron detector, a backscattered electron detector, and an X-ray detector to convert secondary electrons, backscattered electrons, and X-rays into electrical signals. The display screen 111 converts the electrical signals output by the electron detector 110 into grayscale values, thereby generating a SEM image.

[0028] Figure 2 Exemplary schematic diagrams of SEM imaging in some embodiments of this application are shown. As mentioned above, the deflection coil 105 controls the electron beam 102 to scan the sample 108 based on a preset scanning mode to facilitate image construction. In some embodiments, a target observation field of view is defined on the sample 108, and the deflection coil controls the electron beam to scan the target observation field of view, thereby generating a corresponding SEM image. In these embodiments, the electron beam scans the target observation field of view along two mutually perpendicular directions; specifically, the electron beam first completes a scan along the fast scanning direction, then changes to a new line along the slow scanning direction, and then repeats the scan in the fast scanning direction, ultimately making the scanned area cover the entire target observation field of view. Further, the scanning process of the electron beam on the target observation field of view consists of a series of dwell times. At each dwell time, the electron beam forms a corresponding electron beam spot on the target observation field of view and excites a corresponding signal. Based on the excitation signal corresponding to each electron beam spot, the gray value corresponding to a pixel in the SEM image can be obtained.

[0029] In semiconductor manufacturing and inspection processes, SEM is widely used for CD measurement and structural observation. For example, in chip manufacturing, the photolithography process copies the circuit pattern on the mask onto the wafer through light projection; any pattern deviation or defect in the mask will be copied to every chip in batches, thus affecting chip yield; therefore, it is essential to accurately measure the critical dimensions of the mask (including linewidth, aperture, edge roughness, and area). Since the mask is at the nanoscale, SEM is needed to scan the mask surface with a focused electron beam in order to generate high-resolution images for measuring critical dimensions.

[0030] When imaging semiconductor materials using SEM, the charging effect is prone to occur. This effect leads to localized electric field distortion, which in turn affects the electron beam scanning trajectory and secondary electron release, ultimately producing various structural artifacts in the image, such as edge blackening, contrast inhomogeneity, structural blurring, and local distortion. These artifacts degrade image quality, causing image distortion, unclear structural boundaries, and affecting the accuracy of subsequent image analysis and CD measurements, thus becoming a significant factor limiting the improvement of SEM detection accuracy.

[0031] The industry has attempted to remove charging artifacts from SEM images using artificial intelligence algorithms. Recent deep learning models, such as those based on autoencoders or adversarial networks, are essentially probabilistic generative models. In semiconductor scenarios lacking absolute truth supervision, these models are prone to incorrectly smoothing out critical line edge roughness or "filling in" non-existent microstructures while removing artifacts. For CD measurements, which strive for absolute physical accuracy, such nondeterministic pixel manipulation is unacceptable. Furthermore, these methods heavily rely on synthetic data for training, resulting in weak model generalization ability and difficulty in perfectly replicating the complex shot noise and material contrast in real production lines, thus lacking the versatility required for industrial environments.

[0032] Holon's patent application, US10553391B2, discloses a rotational scanning scheme. Specifically, Holon proposes using an electron beam to acquire mask images from different scanning directions, performing pixel-to-pixel scanning on the electron beam to acquire pixel signals at the same location on the mask and generate multiple images, and then synthesizing a high-precision mask image based on the generated multiple images.

[0033] Some factory automation software mandates that input images be of standard size to match recipe settings, and does not support input images of non-standard sizes. When the size of a SEM scan image or a composite image obtained by fusing SEM scan images differs from the standard size, deformation processing is required. This deformation processing alters the pixel dimensions in the image, leading to erroneous results when performing critical dimension measurements or structural feature observations based on the deformed image. However, current rotational scanning schemes lack dynamic correction mechanisms for pixel dimension changes. After images undergo deformation processing such as rotation, resampling, and stretching, without precise physical dimension tracing, systematic deviations in measurement results will occur, failing to meet the stringent requirements for measurement repeatability in advanced processes. Therefore, this application proposes an imaging system and a method for image acquisition, enabling the acquisition of output images from the sample to be imaged that effectively suppress charging effect artifacts, improving the software compatibility of the output image and the accuracy of measurement data acquired based on the output image.

[0034] Figure 3 An exemplary structural schematic diagram of an imaging system 300 according to some embodiments of this application is shown. For example... Figure 3 As shown, in some embodiments, the imaging system 300 includes: an image acquisition module 310 configured to: control an electron beam to scan a sample according to a first angle in a preset angle sequence, so as to generate a first scan image; and an image processing module 320 configured to: acquire multiple scan images corresponding to multiple angles in a preset angle sequence; fuse the multiple scan images to obtain a composite image; perform deformation processing on the composite image to obtain an output image; and calculate correction coefficients based on the dimensions of the composite image and the output image to correct the measurement data of the output image.

[0035] In some embodiments, the imaging system 300 is used to acquire images of a sample to facilitate analysis and measurement based on the acquired images. For example, the imaging system 300 can be used to acquire images of semiconductor materials, such as photomasks, involved in a semiconductor manufacturing process to facilitate critical dimension measurements and structural observation based on the acquired images.

[0036] In some embodiments, the image acquisition module 310 includes a SEM, for example... Figure 1 The SEM100 is shown.

[0037] Figure 4 An exemplary structural diagram of the image processing module 320 in some embodiments of this application is shown. For example... Figure 4As shown, in some embodiments, the image processing module 320 includes a processor 321, main memory 322, auxiliary memory 323, network interface device 324, input / output device 326, battery 327, and bus 328. The processor 321 may include a central processing unit (CPU) or a graphics processing unit (GPU) for running code embedded in hardware or flexibly loaded software code. The main memory 322 may be volatile or non-volatile memory for providing the processor 321 with the instructions and data required for computation in real time. The auxiliary memory 323 may be volatile or non-volatile memory for long-term, large-capacity data storage. The network interface device 324 may include a wireless adapter for providing connectivity to a network 325, which may include a wide area network (WAN), local area network (LAN), wireless LAN, wireless personal area network (WPAN), wireless wide area network (WAN), or other networks. The input / output device 326 may include a keyboard, mouse, touchpad, touchscreen input device, and display. The battery 327 provides power to the various components of the image processing module 320. Bus 328 is used to transmit communication data between various hardware components. In these embodiments, image processing module 320 acquires multiple scan images from image acquisition module 310 via input / output device 326 or network interface device 324, stores the acquired multiple scan images in main memory 322 or loads them into main memory 322 via auxiliary memory 323, so that processor 321 can acquire multiple scan images from main memory 322 and process the multiple scan images.

[0038] Figure 5a Exemplary schematic diagrams of electron beam scanning angles in some embodiments of this application are shown. For example... Figure 5a As shown, in some embodiments, a horizontal X-axis and a vertical Y-axis are defined on the stage plane, wherein the positive direction of the X-axis is horizontal to the right and the positive direction of the Y-axis is vertical to the down. The electron beam can scan the sample based on a preset scanning mode under the control of the deflection coil. The scanning mode includes a fast scanning direction and a slow scanning direction that are perpendicular to each other, wherein the angle α of the fast scanning direction rotating counterclockwise relative to the positive X-axis direction is the electron beam scanning angle.

[0039] In some embodiments, the preset angle sequence includes multiple preset electron beam scanning angles, and the first angle can be any preset electron beam scanning angle in the preset angle sequence. In these embodiments, the image acquisition module 310 traverses all electron beam scanning angles in the preset angle sequence; for the currently traversed first angle: the electron beam is controlled to scan the sample at the first angle to generate a first scan image. Thus, after traversing all electron beam scanning angles in the preset angle sequence, a scan image corresponding to each electron beam scanning angle can be obtained.

[0040] In some embodiments, fusing multiple scanned images to obtain a composite image includes: performing pixel-by-pixel fusion on the multiple scanned images to obtain a composite image. In some embodiments, a pixel in the scanned image corresponds to an imaging point on the actual sample surface, and the maximum, minimum, or weighted average pixel value of the pixels corresponding to the same imaging point in the multiple scanned images is taken to obtain a pixel in the composite image; after completing the traversal processing of all pixels in the scanned images, a complete composite image can be obtained.

[0041] In some embodiments, when the synthesized image is processed into an output image by deformation operations such as stretching, resampling or scaling, a correction coefficient q is calculated based on the width W1 of the synthesized image and the width W2 of the output image. For example, q = W1 / W2.

[0042] In some embodiments, the measurement data M1 of the output image is corrected according to the correction coefficient q to obtain the corrected result M2, for example, M2 = M1 × q.

[0043] In some embodiments, the pixel size of the output image is corrected according to q. For example, P corr =P origin ×q, where P origin P corr These represent the pixel size of the output image before and after correction, respectively, with units of nm / pixel. In these embodiments, the actual length is calculated based on the corrected pixel size and the pixel length data for the output image.

[0044] In some embodiments, if only the composite image has been cropped, the correction factor can be set to 1.

[0045] This application controls the electron beam to scan and image a sample based on multiple angles in a preset angle sequence. After obtaining a scanned image based on each angle in the preset angle sequence, multiple scanned images are fused to obtain a composite image. Since the composite image incorporates scanned images from multiple angles, it helps suppress charging artifacts in the scanned images at each angle. Furthermore, this application performs deformation processing on the composite image to obtain an output image, ensuring that the output image conforms to the format requirements of the software processing. In addition, this application calculates correction coefficients based on the dimensions of the composite image and the output image, and corrects the measurement data of the output image based on these correction coefficients. This helps ensure the measurement accuracy of the output image under any operating condition, ensuring that the physical accuracy of CD measurements is not affected by the image processing flow.

[0046] In some embodiments, the imaging system 300 allows the user to configure a preset angle sequence and a total number of image acquisition frames. The image acquisition module 310 evenly distributes the total number of image acquisition frames to each angle in the preset angle sequence, so that each angle is assigned a corresponding number of acquisition frames (NF). For the first angle in the angle sequence... The image acquisition module 310 controls the electron beam to... The image acquisition module 310 or the image processing module 320 acquires NF images of the same angle for the sample at the scanning angle. Then, the image acquisition module 310 or the image processing module 320 performs pixel-by-pixel fusion on the NF images of the same angle. For example, the maximum value, minimum value or weighted average value of the pixel values ​​of the pixels corresponding to the same imaging point in the NF images of the same angle is taken to obtain the first scan image.

[0047] Rotational scanning schemes are prone to field-of-view loss, such as the scanned area deviating from the effective observation area or the four corners of the target observation field of view being missing. Furthermore, rotational scanning schemes also suffer from redundant scanning, with scanning invalid areas reducing image acquisition efficiency and increasing the power consumption of the imaging system. Therefore, in some embodiments, the image acquisition module 310 is further configured to: calculate a first scan size for a first angle, such that the scanned area determined by the first angle and the first scan size circumferentially borders the target observation field of view; and control the electron beam to scan the sample based on the first angle and the first scan size to generate a first scan image.

[0048] Figure 5b This document illustrates exemplary schematic diagrams of determining the scanning area based on a first angle and a first scanning size in some embodiments of this application. For example... Figure 5b As shown, in some embodiments, the first scan size includes the scan width (W). base ) and scan height (H) base The scanning area determined by the first angle and the first scanning size is a rectangular area with a width of W. baseHeight is H base The center is located at the origin O of the stage, and the shape of the rectangle is rotated counterclockwise by a first angle relative to the stage.

[0049] In some embodiments, the scanning width in the first scanning dimension is calculated for the first angle using the following formula (1), and the scanning height in the first scanning dimension is calculated for the first angle using the following formula (2).

[0050] Formula (1): .

[0051] Formula (2): .

[0052] In the two formulas above, Indicates the width of the target observation field of view. Indicates the height of the target observation field of view. Indicates the first angle.

[0053] Figure 5c This illustration shows an exemplary schematic diagram of the target observation field of view circumscribing the scanning area in some embodiments of this application. For example... Figure 5c As shown, the target observation field of view can be set by the user of the imaging system according to the observation needs; it can be a part of the sample surface or the entire sample surface. A target observation field of view circumscribed outside the scanning area means that all vertices of the target observation field of view lie on the outline of the scanning area, and the target observation field of view is located inside the scanning area.

[0054] It is understood that in the process of acquiring scanned images based on each angle in the preset angle sequence, this application calculates the corresponding scan size for each preset angle, thereby making the target observation field of view circumscribed by the scanned area determined by the preset angle and its corresponding scan size. The beneficial effects of the adaptive circumscribed scan area strategy formed by this include: ensuring that the scanned images acquired based on each angle in the preset angle sequence accurately correspond to the observation field of view of interest to the user, improving the suppression effect of the synthetic image on charging effect artifacts; avoiding scanning unnecessary redundant areas during image acquisition, and maximizing acquisition efficiency and data transmission bandwidth.

[0055] The image acquisition module 310 has certain mechanical errors. In particular, this application requires continuous control of the electron beam to scan the sample at multiple angles. The mechanical errors of the deflection coil will cause the electron beam to be unable to scan accurately at the preset angle, or the scanning area to be unable to accurately connect to the target observation field of view. In order to prevent the mechanical errors of the image acquisition module 310 from affecting the scanning area, and to avoid the problem that the synthesized image may not be able to effectively suppress charging effect artifacts due to the influence of the scanning area, in some embodiments, the image acquisition module 320 is further configured to: calculate the first scanning size for the first angle according to the preset error margin. In these embodiments, the preset error margin includes the bilateral error margin in the fast scanning direction and the bilateral error margin in the slow scanning direction, and the scanning width in the first scanning size is calculated for the first angle using the following formula (3), and the scanning height in the first scanning size is calculated for the first angle using the following formula (4).

[0056] Formula (3): .

[0057] Formula (4): .

[0058] In the two formulas above, This represents the unilateral error margin in the fast / slow scan direction.

[0059] It is understandable that by adding a preset error margin during the calculation of the first scanning size based on the first angle, even if the image acquisition module 310 has mechanical errors, it can ensure that the scanning area accurately connects to the target observation field of view.

[0060] It is understood that there is a relative rotation angle between the multiple scanned images corresponding to multiple angles in the preset angle sequence; in other words, due to the different scanning angles, the multiple scanned images are not aligned. In some embodiments, the multiple scanned images are aligned to facilitate subsequent pixel-by-pixel fusion of the multiple scanned images to obtain a composite image; furthermore, in these embodiments, the suppression effect of charging effect artifacts on the composite image is improved by increasing the alignment between the multiple scanned images.

[0061] In some embodiments, the image processing module 320 is further configured to rotate the first scanned image counterclockwise by a first angle to obtain a rotated image. It is understood that the scanned image obtained with a scanned angle of θ is rotated clockwise by θ degrees compared to the scanned image obtained with a scanned angle of 0°, and counterclockwise rotation helps improve the alignment between multiple scanned images.

[0062] In some embodiments, the image processing module 320 is further configured to perform noise reduction processing and / or grayscale histogram equalization processing on the first scanned image before rotating it. In these embodiments, noise reduction processing and / or grayscale histogram equalization processing are performed on multiple scanned images respectively to eliminate overall brightness fluctuations caused by differences in scanning time.

[0063] According to the translation theorem of Fourier transform, if the image It is an image Translation What was obtained, that is , and Fourier transform and satisfy: In other words, , The amplitudes in the frequency domain are exactly the same, but the phases differ, and Compared to Translation in the spatial domain is equivalent to linear phase shift in the frequency domain. Based on this, to further align the scanned images, in some embodiments, a preset angle sequence includes 0°, and the scanned image corresponding to 0° is selected as the reference scanned image. In these embodiments, the phase difference information between the rotated image and the reference scanned image is extracted, and a first translation vector of the rotated image relative to the reference image is calculated based on the phase difference information. The first translated image is obtained by translating the rotated image according to the first translation vector. It is understood that any angle in the preset angle sequence other than 0° has a corresponding rotated image. By converting each rotated image into a first translated image using the above method, multiple first translated images and the 0° scanned image have a high degree of alignment.

[0064] In some embodiments, the image processing module 320 is further configured to: acquire phase difference information between the rotated image and the reference scan image in the frequency domain; calculate a first translation vector of the rotated image relative to the reference scan image based on the phase difference information; and translate the rotated image based on the first translation vector to obtain a first translated image.

[0065] In some embodiments, a reference scan image and rotated image Perform two-dimensional fast Fourier transforms separately to obtain and Where (x, y) represents pixel coordinates and (u, v) represents frequency coordinates. Then, the normalized cross-power spectrum R(u, v) is calculated according to the following formula (5): Formula (5): .

[0066] In formula (5), yes The conjugate of complex numbers, This represents the modulo operation. The normalized cross-power spectrum R(u, v) calculated using frequency domain data eliminates the grayscale and contrast differences between the reference scanned image and the rotated image, while preserving the phase difference information between the two images. Furthermore, an inverse Fourier transform is performed on the cross-power spectrum to obtain the Dirac impulse function r(x, y) in the spatial domain. The peak coordinates of r(x, y) are then... As the first translation vector of the rotated image relative to the reference scanned image, the first translation vector can be understood as: the integer pixel displacement of the rotated image relative to the reference scanned image. Finally, the rotated image is translated... To obtain the first translated image.

[0067] The Dirac pulse function r(x, y) can be understood as the correspondence between discrete coordinate points and the degree of overlap between two images. The peak coordinates extracted from the Dirac pulse function are integer pixels; however, the actual displacement may be at the non-integer pixel level. In view of this, in some embodiments, the image processing module 320 is further configured to: fit the correspondence between the translation vector and the degree of overlap between the rotated image and the reference scan image within the neighborhood of the first translation vector; extract the peak coordinates (which may be at the sub-pixel level) corresponding to the highest degree of overlap in the correspondence to obtain a second translation vector; and translate the rotated image according to the second translation vector to obtain a second translated image.

[0068] In some embodiments, the correspondence between the degree of overlap and the translation vector is set as a quadratic surface model. Peak coordinates extracted from the Dirac impulse function By least squares in 3 3. Solve for parameters a, b, c, d, e, and f within the neighborhood to fit the correspondence between the translation vector and the degree of overlap between the rotated image and the reference scan image. Let The derivative is 0, thus the peak coordinates can be solved. and will This serves as the second translation vector. Further, the rotated image is translated according to the second translation vector. To obtain the second translated image.

[0069] Existing traditional image fusion algorithms mostly employ "pixel-level averaging" or "linear superposition." For charging effect artifacts that manifest as "single-sided shadows" or "asymmetric tails," these methods can only reduce the grayscale contrast of the artifacts (lighten them), but cannot completely remove them from the image. The inventors of this application discovered in practice that in SEM images, the gradient values ​​of true structural edges are high, while the gradient values ​​of charging effect artifacts are low and divergent. Based on this characteristic, the gradient magnitude is used as a criterion for pixel-level sharpness. Multiple second translation images are fused into a composite image based on the gradient magnitude, achieving image reconstruction that automatically preserves sharp edges and suppresses blurry artifacts. In some embodiments, the image processing module 320 is further configured to: acquire multiple second translation images corresponding to multiple angles in a preset angle sequence; acquire the gradient magnitudes of the multiple second translation images; and fuse the multiple second translation images based on the gradient magnitudes to obtain a composite image.

[0070] In some embodiments, the image processing module 320 is further configured to: acquire the horizontal gradient and the vertical gradient of the second translated image; and acquire the gradient magnitude based on the horizontal gradient and the vertical gradient.

[0071] Let there be multiple second translation images, including P1~P2. n , will P i Input the Sobel operator and calculate P respectively. i Horizontal gradient and vertical gradient and according to To get P i gradient magnitude G i In these embodiments, G1~G n Gaussian filtering is performed to suppress shot noise by utilizing the spatial continuity of the structure, generating smooth gradient magnitudes SG1~SG2. n Furthermore, for SG1~SG n Normalization is performed to obtain the weights AG1~AG n And AG1~AG n The sum is 1. Then, based on AG1~AG... n For P1~P n Element-wise weighted fusion is performed to obtain a synthetic image (I). e ),Right now .

[0072] This application also discloses a method for image acquisition, which is based on the imaging system in any of the preceding embodiments.

[0073] The technical solution proposed in this application will be described below through two embodiments.

[0074] Example 1 Implementation Step 1: Set the acquisition parameters.

[0075] In this embodiment, the target structure to be imaged is a linewidth structure with obvious directional artifacts. The preset angle sequence is (0°, 180°), where 0° is the reference phase and 180° is the inverted phase, to counteract the directional charging effect artifacts. In the target image size, and All are 512 pixels. The total number of image frames acquired is 8, thus acquiring 4 frames of images at each preset angle.

[0076] To optimize acquisition efficiency while ensuring field-of-view integrity, an adaptive circumscribed rectangle strategy is employed to dynamically calculate the base image size based on a preset angle θ. In this embodiment, a reserved margin for single-sided error is set. Pixels. According to formulas (3) and (4), when θ = 0°, .

[0077] .

[0078] Similarly, it can also be calculated when θ = 180°. Pixels. Therefore, when acquiring images at two preset angles, the image acquisition module sets the scanning resolution to 522×522 pixels.

[0079] Step 2: Image acquisition process.

[0080] The image acquisition module controls the electron beam to scan the sample surface. First, it scans at 0°, acquiring four frames at the same angle, then averages them pixel-by-pixel to obtain the first scan image I0 (522×522). Then, it scans at 180°, acquiring four frames at the same angle, then averages them pixel-by-pixel to obtain the first scan image I0. 180 (522×522). Because I 180 The actual structure in the image is inverted compared to I0, and because the scanning direction is opposite, the direction of the charging effect artifacts is also opposite to that of I0.

[0081] Step 3: Image processing and fusion.

[0082] Angle normalization: The image processing module obtains I0 and I from the image acquisition module. 180 Identify the angle markers of the two images and use a pixel matrix to group the I... 180 Rotate 180° without damage to obtain It can be assumed that: The spatial orientation is consistent with I0, but there may still be slight translation.

[0083] Precise registration: Using I0 as a reference, calculate using a frequency domain phase correlation algorithm. The integer-level pixel displacement (Δx0, Δy0) is used to estimate the sub-pixel-level peak position (Δx1, Δy1) within a 3×3 neighborhood of (Δx0, Δy0) using quadratic surface fitting. Translate (-Δx1, -Δy1) to obtain , Nanoscale alignment with I0 was achieved at a scale of 522×522.

[0084] Structure-aware weighted fusion: The image processing module calculates separately Gradient magnitude G of I0 180 G0, for G 180 Normalize G0 to obtain the weight AG 180 AG0, after weighted fusion, yields the synthetic image I. e =AG 180 +AG0I0.

[0085] Effective area output: After fusion, the image processing module outputs the fused image I from the 522×522 pixel fusion region. e The target region of 512×512 pixels is cropped out as the output image I. o .

[0086] Figure 6 I0 and I are shown in this embodiment. 180 and I e ; where subgraph (a) corresponds to I0, and subgraph (b) corresponds to I 180 Subgraph (c) corresponds to I e .according to Figure 6 It is understandable that the fused image I e Compared to the original image: the left and right brightness are more symmetrical, edge artifacts (such as black trails) are significantly reduced, and the sharpness of structural edges is significantly improved. This is due to a mechanical drift margin of ±5 pixels allowed during acquisition. Even if translation occurs during registration, the central area of ​​the 512×512 pixel image remains within the effective imaging range, ensuring that the output image has no black borders and a complete field of view. The entire processing flow involves only pixel cropping without any geometric scaling, so the output image strictly maintains the original pixel size at the time of acquisition, requiring no additional size correction and ensuring the absolute physical fidelity of subsequent linewidth measurement data.

[0087] Example 2 When observing the morphology of the bottom structure of a borehole, such structures often have a high aspect ratio, and scanning from a single angle frequently results in blurred edges or asymmetrical charging tails, severely affecting the accuracy of aperture measurement. This embodiment employs four-directional scan fusion to improve structural integrity and imaging uniformity, and focuses on demonstrating how the system ensures measurement accuracy through a physical size traceability mechanism under "limited field-of-view acquisition" conditions (i.e., the adaptive circumscribed rectangle strategy in Embodiment 1 cannot be used). Unlike the adaptive circumscribed rectangle strategy in Embodiment 1, which can actively expand the acquisition range, this embodiment simulates scenarios limited by the maximum scanning field of view of the hardware or requiring adherence to a fixed scanning magnification.

[0088] Implementation Step 1: Set the acquisition parameters.

[0089] In this embodiment, the target image size and All are 1024 pixels. The preset angle sequence is (0°, 90°, 180°, 270°), and the total number of image acquisition frames is 16 frames.

[0090] Step 2: Image acquisition process.

[0091] The image acquisition module controls the electron beam to scan the sample surface. Four frames of 1024×1024 pixel images are acquired at each preset angle. The first scan image I0 is obtained by averaging the pixels of each preset angle's corresponding scan image. 90 I 180 I 270 And I0, I 90 I 180 I 270 The pixel resolution is 1024×1024.

[0092] Step 3: Image processing and fusion.

[0093] Angle normalization: The image processing module uses matrix transformation to normalize I... 90 I 180 I 270 Rotate each image counterclockwise to 0° to obtain the rotated image SI. 90 SI 180 SI 270 .

[0094] Sub-pixel registration: Calculate SI with I0 as a reference. 90 SI 180 SI 270 The subpixel displacement, based on the subpixel displacement for SI 90 SI 180 SI 270 Perform a translation to obtain MI 90MI 180 MI 270 .

[0095] Analysis of the black border phenomenon: Due to the lack of an adaptive circumscribed rectangle strategy and the presence of minor mechanical repetitive positioning errors (e.g., ±20 pixels) in the SEM sample stage, I0 and MI... 90 MI 180 MI 270 They cannot perfectly overlap in space. Specifically, assume MI 180 MI is offset 20 pixels to the right relative to I0; after the algorithm forces them to align, MI 180 A 20-pixel blank area, the so-called "black border," will appear on the left edge. (This refers to the I0 and MI...) 90 MI 180 MI 270 After overlay, only the central common intersection area is valid data, and the size of this area must be smaller than the original 1024×1024.

[0096] Step 4: Multi-angle image fusion and cropping.

[0097] A structure-aware weighted fusion algorithm similar to that used in Example 1 is employed to fuse I0 and MI. 90 MI 180 MI 270 Gradient weights are used to eliminate the charging tail at the bottom edge of the hole to obtain the synthetic image I. e To remove the black edges caused by registration, calculate I0 and MI. 90 MI 180 MI 270 The common intersection of the values ​​is used to crop the largest inscribed rectangle, thus obtaining the output image I. f Assuming that the effective area is reduced due to mechanical offset, the cropped output image I... f The size has been changed to 980×980 pixels.

[0098] Figure 7 I0 and I are shown in this embodiment. 90 I 180 I 270 I e and I f ; where subgraph (a) corresponds to I0, and subgraph (b) corresponds to I 90 Subgraph (c) corresponds to I 180 Subgraph (d) corresponds to I 270 Subgraph (e) corresponds to I e Subgraph (f) corresponds to I f The arrows in sub-figures (a~d) indicate the black trailing effect in the image.

[0099] Step 5: Image resampling and physical size correction.

[0100] Because factory automation software mandates that input images maintain a standard 1024×1024 pixels, and the underlying hardware does not support non-standard output sizes (such as 980×980 pixels), the cropped output image must be... f Restored to standard size. The image processing module uses a bicubic interpolation algorithm to resize the 980×980 I... f Stretch and enlarge to 1024×1024 pixels to obtain image I. g .

[0101] Understandably, the stretching operation reduces the actual physical length represented by a single pixel. To ensure the accuracy of subsequent aperture measurements, the image processing module triggers the physical size tracing module to calculate the corrected pixel size P based on the following formula (6). corr .

[0102] Formula (6): .

[0103] In formula (6) above, For the stretched image I g In this embodiment, the pixel size is... Correct pixel size P corr =1.1487nm / pixel. The image processing module will... corr It is written into the image metadata and used in subsequent CD measurement algorithms.

[0104] The specific functions implemented by the image acquisition method provided in the embodiments of this specification can be explained in comparison with the foregoing embodiments in this specification, and can achieve the technical effects of the foregoing embodiments, so they will not be repeated here.

[0105] While numerous embodiments of this application have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will arise for those skilled in the art without departing from the spirit and intent of this application. It should be understood that various alternatives to the embodiments of this application described herein may be employed in the practice of this application. The appended claims are intended to define the scope of protection of this application and therefore cover equivalents or alternatives within the scope of these claims.

Claims

1. An imaging system, characterized in that, include: The image acquisition module is configured as follows: According to the first angle in the preset angle sequence, the electron beam is controlled to scan the sample in order to generate the first scan image; The image processing module is configured as follows: Acquire multiple scan images corresponding to multiple angles in a preset angle sequence; The multiple scanned images are fused to obtain a composite image; The synthesized image is deformed to obtain the output image; as well as Correction coefficients are calculated based on the dimensions of the synthesized image and the output image to correct the measurement data of the output image.

2. The imaging system according to claim 1, characterized in that, The image acquisition module is further configured as follows: Calculate the first scanning size for the first angle, so that the target observation field is circumscribed in the scanning area determined by the first angle and the first scanning size; Based on the first angle and the first scanning size, the electron beam is controlled to scan the sample in order to generate a first scan image.

3. The imaging system according to claim 2, characterized in that, The image acquisition module is further configured to calculate the first scanning size for the first angle based on a preset error margin.

4. The imaging system according to claim 2, characterized in that, The image processing module is further configured to rotate the first scanned image counterclockwise by a first angle to obtain a rotated image.

5. The imaging system according to claim 4, characterized in that, The image processing module is further configured to perform noise reduction processing and / or grayscale histogram equalization processing on the first scanned image before rotating it.

6. The imaging system according to claim 4, characterized in that, The image processing module is further configured as follows: Obtain the phase difference information between the rotated image and the reference scan image in the frequency domain; The first translation vector of the rotated image relative to the reference scan image is calculated based on the phase difference information; The rotated image is translated according to the first translation vector to obtain the first translated image.

7. The imaging system according to claim 6, characterized in that, The image processing module is further configured as follows: Within the neighborhood of the first translation vector, the correspondence between the translation vector and the degree of overlap between the rotated image and the reference scan image is fitted; Extract the peak coordinates corresponding to the highest degree of overlap in the correspondence to obtain the second translation vector; The rotated image is translated according to the second translation vector to obtain the second translated image.

8. The imaging system according to claim 7, characterized in that, The image processing module is further configured as follows: Acquire multiple second translation images corresponding to multiple angles in a preset angle sequence; Obtain the gradient magnitude of the plurality of second translation images; The multiple second translation images are fused based on the gradient magnitude to obtain the composite image.

9. The imaging system according to claim 8, characterized in that, The image processing module is further configured as follows: Obtain the horizontal and vertical gradients of the second translated image; The gradient magnitude is obtained based on the horizontal gradient and the vertical gradient.

10. A method for image acquisition, characterized in that, Image acquisition is performed using the imaging system described in any one of claims 1-9.

Images