Calculation microscopic imaging method for nanoscale microstructure defect recognition

Abstract

The invention discloses a calculation microscopic imaging method for nanoscale microstructure defect recognition, and belongs to the technical field of microscopic imaging, and the specific operation is as follows: step 1, a scanning system controls a light beam to collect a low-resolution image matrix at a fixed interval, and N is the number of scanning frames; step 2, iterating and reversely deducing a point spread function from the low-resolution image matrix; step 3, a low-resolution image matrix and a point spread function are used to calculate and restore an image; 4, calculating a local frequency amplitude entropy for the recovered image; and 5, judging whether the local frequency amplitude entropy is greater than a threshold value, if so, marking the region as a suspected defect region, and if not, marking the region as a normal region. According to the method, a real structure image is restored through an airspace regular smoothing and path compensation mechanism, high-fidelity nano defect identification can be realized under dye-free and non-vacuum conditions, and a nano-scale defect target is extracted.

Description

Technical Field

[0001] This invention relates to the field of microscopic imaging technology, and more specifically to a computational microscopic imaging method for identifying defects in nanoscale microstructures. Background Technology

[0002] With the increasing demands for precision in material structures in fields such as semiconductor manufacturing, nanomaterials science, and biomedical engineering, defect detection at the microscale has become a crucial link in ensuring fundamental processes and quality control. In the nanoscale manufacturing environment, defects such as linewidth deformation in wafer fabrication, microcracks in photolithography masks, voids in metal interconnect layers, and residual particles often directly determine the yield of finished products and device performance. However, traditional optical microscopy imaging systems are limited by the Abbe diffraction limit, and their resolution is constrained by both the wavelength λ and the numerical aperture NA.

[0003] Based on this, the present invention designs a computational microscopy imaging method for identifying defects in nanoscale microstructures to solve the above problems. Summary of the Invention

[0004] To address the aforementioned shortcomings of existing technologies, this invention provides a computational microscopy imaging method for identifying defects in nanoscale microstructures.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A computational microscopy imaging method for identifying defects in nanoscale microstructures is described below:

[0007] Step 1: The scanning system controls the light beam to acquire a low-resolution image matrix at fixed intervals. N is the number of scan frames;

[0008] Step 2: From the low-resolution image matrix Iterative back-calculation of point spread function ;

[0009] Step 3: Low-resolution image matrix and point spread function Calculate and restore the image ;

[0010] Step 4: For the restored image Calculate local frequency amplitude entropy ;

[0011] Step 5: Determine the local frequency amplitude entropy Is it greater than the threshold? If the judgment is yes, mark it as a suspected defect area; if the judgment is no, mark it as a normal area.

[0012] Furthermore, the fixed spacing is 1µm.

[0013] Furthermore, the spatial resolution of low-resolution images is 1µm, while the spatial resolution of high-resolution images is 50-60nm.

[0014] Furthermore, the specific steps for step two are as follows:

[0015] Step 21: Low-resolution image matrix low-resolution images in An initial guess image of the high-resolution image is obtained by upsampling using bilinear interpolation. ;

[0016] Step 22: From low-resolution image Statistical fuzzy marginal gradient, determining the estimated ;

[0017] Step 23: Based on estimation Solving for the optimal estimate of a high-resolution image ;

[0018] Step 24: Optimal estimation of high-resolution images Solve ;

[0019] Step 25: Repeat steps 23 and 24. Matching low-resolution image matrix after downsampling All low-resolution images At the same time, Apply spatial smoothness regularization constraints and output the point spread function. .

[0020] Furthermore, based on estimation Solving for the optimal estimate of a high-resolution image The optimization objective is:

[0021]

[0022] This indicates a downsampling operation. Frequency domain and spatial domain canonical intensity coefficients.

[0023] Furthermore, the spatial smoothness regularization constraint is as follows:

[0024]

[0025] The weight of the regularization term. Used to ensure spatial smoothness.

[0026] Furthermore, image restoration The calculation is as follows:

[0027] .

[0028] Furthermore, regarding image recovery Calculate local frequency amplitude entropy The calculation method is as follows:

[0029]

[0030]

[0031] Represents a two-dimensional Fourier transform. This represents the image spectrum.

[0032] Beneficial effects: This invention restores the real structural image through spatial regularization smoothing and path compensation mechanism, and can achieve high-fidelity nanodefect identification and extract nanoscale defect targets under dye-free and non-vacuum conditions. It can be applied to semiconductor production lines, materials science laboratories and high-precision manufacturing fields, and is especially suitable for flexible, automated and intelligent inspection processes. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0034] Figure 1 This is a flowchart of a computational microscopy imaging method for identifying defects in nanoscale microstructures according to the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention 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 the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0036] The present invention will be further described below with reference to embodiments.

[0037] Example 1: Please refer to Figure 1 A computational microscopy imaging method for identifying defects in nanoscale microstructures is described below:

[0038] Step 1: The scanning system controls the light beam to acquire a low-resolution image matrix at fixed intervals. N is the number of scan frames;

[0039] Step 2: From the low-resolution image matrix Iterative back-calculation of point spread function ;

[0040] Step 3: Low-resolution image matrix and point spread function Calculate and restore the image ;

[0041] Step 4: For the restored image Calculate local frequency amplitude entropy ;

[0042] Step 5: Determine the local frequency amplitude entropy Is it greater than the threshold? If the judgment is yes, mark it as a suspected defect area; if the judgment is no, mark it as a normal area.

[0043] The fixed spacing is 1µm.

[0044] The spatial resolution of low-resolution images is 1µm, while the spatial resolution of high-resolution images is 50-60nm.

[0045] Step two involves the following steps:

[0046] Step 21: Low-resolution image matrix low-resolution images in An initial guess image of the high-resolution image is obtained by upsampling using bilinear interpolation. ;

[0047] Step 22: From low-resolution image Statistical fuzzy marginal gradient, determining the estimated ;

[0048] Step 23: Based on estimation Solving for the optimal estimate of a high-resolution image ;

[0049] Step 24: Optimal estimation of high-resolution images Solve ;

[0050] Step 25: Repeat steps 23 and 24. Matching low-resolution image matrix after downsampling All low-resolution images At the same time, Apply spatial smoothness regularization constraints and output the point spread function. .

[0051] Based on estimation Solving for the optimal estimate of a high-resolution image The optimization objective is:

[0052]

[0053] This indicates a downsampling operation. These are the frequency and spatial canonical intensity coefficients.

[0054] The spatial smoothness regularization constraints are as follows:

[0055]

[0056] The weight of the regularization term. Used to ensure spatial smoothness.

[0057] Restore image The calculation is as follows:

[0058] .

[0059] For image recovery Calculate local frequency amplitude entropy The calculation method is as follows:

[0060]

[0061]

[0062] Represents a two-dimensional Fourier transform. This represents the image spectrum.

[0063] By restoring real structural images through spatial regularization smoothing and path compensation mechanisms, high-fidelity nanodefect identification can be achieved under dye-free and non-vacuum conditions, extracting nanoscale defect targets. It can be applied to semiconductor production lines, materials science laboratories, and high-precision manufacturing fields, and is especially suitable for flexible, automated, and intelligent inspection processes.

[0064] Semiconductor lithography mask inspection: Traditional EUV mask inspection instruments rely on extreme light sources and vacuum environments, which are costly and difficult to deploy flexibly. This system, based on a confocal microscopy platform, can directly perform high-throughput inspection of defects such as microcracks, edge collapse, and inconsistent pattern transfer on the mask surface, providing an efficient and safe inspection method for the front-end of chip manufacturing.

[0065] Microelectronic Packaging Defect Analysis: In Fan-out packaging and SiP structures, this system can identify problems such as bare die cracks, TSV via blockage, and uneven metal filling through 3D stack reconstruction, which plays a key role in ensuring the integrity of electrical interconnects.

[0066] MEMS and Micro / Nano Structure Manufacturing Inspection: Micromirror arrays, microcavity structures, etc. are extremely sensitive to size and stress response. This system can capture structural displacement and strain cracks below 50nm, assisting in the reliability verification of structural components.

[0067] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for computational microscopy for nanoscale microstructure defect identification, the method comprising: The procedure is as follows: ​ Step 1 : The scanning system controls the light beam to capture a matrix of low resolution images at a fixed pitch N is the number of scanning frames; Step 2: Iteratively back-propagating the point spread function from the low resolution image matrix ;​ Step 3: Low resolution image matrix and point spread function computing a restored image ; Step 4: For the restored image Computing local frequency amplitude entropy ; Step 5: judging local frequency amplitude entropy whether it is greater than a threshold value If the judgment is yes, mark as a suspected defect region, and if the judgment is no, mark as a normal region.

2. The method of claim 1, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. The fixed interval is 1 pm.

3. The method of claim 2, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. The spatial resolution of the low-resolution image is 1 pm and the spatial resolution of the high-resolution image is 50-60 nm.

4. The method of claim 3, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. The procedure of step two is as follows: Step 21: Low resolution image matrix Low resolution image in Upsample by bilinear interpolation method to get initial guess image of high resolution image ; Step 22: From the low resolution image statistical blur edge gradients, determine an estimate ; Step 23: estimating based on solving for an optimal estimate of a high resolution image ; Step 24: Optimal estimation of high resolution image solving ; Step 25: repeat iteration steps 23 and 24 until Down-sampled matching low resolution image matrix All low resolution images in the middle At the same time, the point spread function is output Applying spatial smoothness regularization constraint .

5. The method of claim 4, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. based on the estimate solving for an optimal estimate of a high resolution image the optimization objective is ； denotes a down-sampling operation, are the frequency and spatial regularized intensity coefficients.

6. The method of claim 5, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. The spatial smoothness regularisation constraint is as follows: The spatial smoothness regularisation constraint is as follows: ； weight of the regularizer, for ensuring spatial smoothness.

7. The method of claim 6, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. Recovering an image The calculation is as follows: 。 8. The method of claim 7, wherein the method is a method of computational microscopy for nanoscale microstructure defect identification. For restoring images Computing local frequency amplitude entropy The computation method is as follows: ； ； denotes a two-dimensional Fourier transform, is the image spectrum.

Images