Wafer alignment mark-oriented focusing evaluation method, system and equipment and medium

By combining the Brenner and Roberts functions for focus evaluation, and through variance calculation and adaptive sampling, the problems of poor noise immunity and stability in the prior art are solved, and efficient and accurate focus evaluation is achieved, which is suitable for multi-directional edge changes. Wafer alignment mark autofocus.

CN119027371AActive Publication Date: 2024-11-26NANJING UNIV OF AERONAUTICS & ASTRONAUTICS +1

Patent Information

Application Number
CN202410990631.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-23
Publication Date
2024-11-26
Estimated Expiration
2044-07-23

AI Technical Summary

Technical Problem

In the prior art, the focus evaluation function has poor noise resistance and poor stability, complex calculations and poor real-time performance, making it difficult to apply to a variety of measurement objects, especially in the case of variation in the edge direction of the image feature of the wafer alignment mark. Down.

Method used

Combining the Brenner function and Roberts function for edge extraction and focus evaluation, and improving noise immunity and stability through variance calculation, adaptive sampling is used to reduce the computational complexity and achieve efficient calculation.

Benefits of technology

High-precision focus evaluation is achieved, the problems of poor noise resistance and stability of the traditional focus evaluation function are overcome, robustness and real-time are improved, and it is suitable for automatic focus of wafer alignment marks with multi-directional edge changes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027371A_ABST
    Figure CN119027371A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of optical metrology, and discloses a wafer alignment mark-oriented focusing evaluation method, system and device and a medium, and the method comprises the steps: obtaining a sequence image set of a wafer alignment mark in a defocusing-focusing-defocusing process; performing edge extraction on each sequence image in the sequence image set; based on a Brenner function and a Roberts function, performing focusing evaluation in different directions of each sequence image after edge extraction to obtain an initial focusing evaluation function value corresponding to each sequence image; performing variance calculation on the initial focusing evaluation function value corresponding to each sequence image to obtain a final focusing evaluation function value; and judging the focusing state of the wafer alignment mark based on the final focusing evaluation function value. According to the technical scheme, the wafer alignment mark can be automatically focused quickly and accurately.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of optical metrology, and in particular relates to a focusing evaluation method, system, equipment and medium for wafer alignment marks. Background Art

[0002] In the integrated circuit industry, the measurement of micro-nano structures runs through the entire manufacturing process, including key links such as circuit design, wafer preparation, silicon wafer manufacturing, packaging and testing. Although scanning electron microscopes, transmission electron microscopes, and atomic force microscopes can all achieve nanoscale dimensional measurement, their disadvantages are significant: slow measurement speed, high cost, and complex equipment operation. Therefore, optical measurement technology with the advantages of fast measurement speed, non-contact, non-destructive, and easy online integration is widely used in chip manufacturing process control and yield management. In the field of optical measurement technology, autofocus technology, as a key image acquisition and processing method, is a core technology in this field. In response to the requirements of wafer alignment mark quantity detection, autofocus technology can provide high-precision, high-speed and reliable focus positioning methods to obtain clear images and accurate measurement results. The core of this technology mainly includes the selection of the focus window, the selection of the focus evaluation function, and the feedback control of the stepper motor by the search algorithm. Among them, the focus evaluation function is the most important of the three, and its performance can intuitively reflect the efficiency and accuracy of the autofocus.

[0003] At present, focusing evaluation functions are mainly divided into spatial domain, frequency domain and information entropy types: spatial domain functions mainly include Roberts function, Brenner function, Tenengrad function, etc. This type of function describes the focus of the image by the change of image grayscale gradient. It is simple to calculate and has high stability. It is a type of focusing evaluation function that is currently widely used, but the noise resistance of this type of function is weak; frequency domain functions mainly include Fourier transform function and discrete cosine transform function, etc. This type of function has high sensitivity, but is sensitive to noise. The calculation amount is too large, resulting in poor real-time performance and difficult to apply to actual situations; information entropy functions describe the richness of information. The focus of the image is proportional to the change of the corresponding grayscale value interval, that is, the information entropy value of the entire area. This type of function has good unbiasedness, but the calculation process is complicated and easily affected by light, and has low sensitivity and noise resistance.

[0004] In the prior art, the SMD-Roberts evaluation function proposed by combining SML with Roberts function has higher focus sensitivity. This method can improve the accuracy and stability of focus evaluation value calculation, but the time cost required for multiple focus evaluation operations is relatively large. The improved dual-threshold Tenengrad focus evaluation function suitable for tool sequence images, although this improved method makes full use of the high-frequency information of the image, the final focus evaluation result and measurement accuracy depend on the selection of appropriate thresholds, and it is difficult to apply to various measurement objects.

[0005] In the prior art, an improved Laplace focusing evaluation function is used to evaluate the focus of the imaging. This method can improve the sensitivity of the evaluation algorithm to the x and y directions, but has weak anti-noise capability and low stability.

[0006] The prior art also uses an improved gradient-weighted high-precision focusing evaluation function for part images for evaluation, which obtains the gradient value of edge pixels based on a 4-directional Sobel operator to improve the accuracy of the gradient. Then, the pixel gradient weighting coefficient is obtained based on the difference in grayscale distribution between the edge pixel and its 8 neighboring pixels, thereby enhancing the sensitivity and noise resistance of the gradient weighting algorithm. However, it also leads to more complex calculations and poor real-time performance. Summary of the invention

[0007] The purpose of the present invention is to provide a focus evaluation method, system, device and medium for wafer alignment marks to solve the problems existing in the above-mentioned prior art.

[0008] To achieve the above object, the present invention provides a focus evaluation method for wafer alignment marks, comprising:

[0009] Acquire a sequence image set of the wafer alignment mark in the defocus-focus-defocus process;

[0010] Performing edge extraction on each sequence image in the sequence image set;

[0011] Based on the Brenner function and the Roberts function, focusing evaluation is performed in different directions of each sequence image after edge extraction, so as to obtain an initial focusing evaluation function value corresponding to each sequence image;

[0012] Performing variance calculation on the initial focus evaluation function values ​​corresponding to each of the sequence images to obtain a final focus evaluation function value;

[0013] The focus state of the wafer alignment mark is determined based on the final focus evaluation function value.

[0014] Optionally, edge extraction is performed on each sequence image in the sequence image set, specifically including:

[0015] Determine the initial sampling interval, and calculate the standard deviation of the grayscale value of each of the sequence images;

[0016] Calculating the grayscale value difference between each pixel to be marked and adjacent pixels in each of the sequence images, wherein the adjacent pixels include pixels to the right and below the current pixel to be marked;

[0017] The gray value standard deviation is compared with the gray value difference of adjacent pixels, and based on the comparison result, it is determined whether the current pixel to be marked is an edge pixel.

[0018] Optionally, comparing the gray value standard deviation with the gray value difference of adjacent pixels specifically includes:

[0019] If the grayscale value differences of adjacent pixels are all smaller than the grayscale value standard deviation, it is determined that the current pixel to be marked is not an edge pixel;

[0020] If there is a grayscale value difference between the grayscale value differences of adjacent pixels that is greater than the grayscale value standard deviation, then the current pixel to be marked is determined to be an edge pixel;

[0021] After determining that the current pixel to be marked is an edge pixel, the current pixel to be marked is marked and the initial sampling interval is reduced.

[0022] Optionally, focusing evaluation is performed in different directions of each sequence image after edge extraction based on the Brenner function and the Roberts function, specifically including:

[0023] The Brenner function is used to perform evaluation calculation in the 0° and 90° directions of the sequence images, and the pixel grayscale difference calculation in the 90° direction is added to obtain a first focus evaluation function value; the 3×3 Roberts operator is used to perform grayscale gradient calculation in the 45° and 135° directions of the sequence images to obtain a second focus evaluation function value, and the first focus evaluation function value and the second focus evaluation function value are multiplied to obtain an initial focus evaluation function value.

[0024] Optionally, variance calculation is performed on the initial focus evaluation function values ​​corresponding to each of the sequence images, and the specific calculation formula is:

[0025]

[0026] Where N is the number of images in the sequence, U Bre2d_Rob2d is the initial average evaluation function value of the sequence image set, F Bre2d_Rob2d_Var is the final focusing evaluation function value.

[0027] A focus evaluation system for wafer alignment marks, comprising:

[0028] An image acquisition module, used for acquiring a sequence image set of the wafer alignment mark in the process of defocusing-focusing-defocusing;

[0029] A focusing evaluation module is used to perform edge extraction on each sequence image in the sequence image set; perform focus evaluation in different directions of each sequence image after edge extraction based on a Brenner function and a Roberts function to obtain an initial focus evaluation function value corresponding to each sequence image; perform variance calculation on the initial focus evaluation function value corresponding to each sequence image to obtain a final focus evaluation function value; and judge the focus state of the wafer alignment mark based on the final focus evaluation function value.

[0030] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a focus evaluation method for wafer alignment marks.

[0031] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the focus evaluation method for wafer alignment marks is implemented.

[0032] The technical effects of the present invention are:

[0033] 1. Aiming at the application background of the diversity of edge directions of wafer alignment mark image features, the present invention combines the advantages of Brenner function and Roberts function, and can efficiently extract edge gradient information in multiple directions of the mark image, thereby achieving high-precision focusing.

[0034] 2. The present invention integrates the traditional focusing evaluation function and adaptive sampling. Although the computational complexity is improved, the amount of calculation is greatly reduced through adaptive sampling, thereby achieving efficient calculation. It not only overcomes the problems of poor noise resistance and stability of the traditional focusing evaluation function, but also solves the problems of complex calculation and poor real-time performance of the fused focusing function.

[0035] 3. The present invention improves the problem of poor noise resistance of spatial domain focusing function by finally calculating the variance of the evaluation function, and has better robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0038] Figure 1 Schematic diagram of a wafer alignment mark sequence image set in an embodiment of the present invention;

[0039] Figure 2 is a specific implementation flow chart of the focusing evaluation method in an embodiment of the present invention;

[0040] Figure 3 This is a comparison diagram of focus evaluation function curves obtained by evaluating and calculating 11 alignment mark defocus-focus-defocus sequence images using the six focus evaluation functions in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0042] It should be understood that the terms described in the present invention are only for describing special embodiments and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that each intermediate value between the upper and lower limits of the scope is also specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the described range is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded in the scope.

[0043] Unless otherwise indicated, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art generally understood by the present invention. Although the present invention describes only preferred methods, any method similar or equivalent to that described herein may also be used in the implementation or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods related to the documents. In the event of a conflict with any incorporated document, the content of this specification shall prevail.

[0044] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention description without departing from the scope or spirit of the present invention. Other embodiments derived from the present invention description will be apparent to those skilled in the art. The present application description and examples are exemplary only.

[0045] The words “include,” “including,” “have,” “contain,” etc. used in this article are open-ended terms, meaning including but not limited to.

[0046] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0047] Embodiment 1

[0048] like Figure 1 As shown in the figure, this embodiment provides a focus evaluation method for wafer alignment marks, including: obtaining a set of sequential images of the wafer alignment marks in the defocus-focus-defocus process; performing edge extraction on each sequential image in the sequential image set; performing focus evaluation in different directions of each sequential image after edge extraction based on the Brenner function and the Roberts function to obtain an initial focus evaluation function value corresponding to each sequential image; performing variance calculation on the initial focus evaluation function value corresponding to each sequential image to obtain a final focus evaluation function value; and judging the focus state of the wafer alignment mark based on the final focus evaluation function value.

[0049] The acquired wafer alignment mark defocus-focus-defocus sequence images are subjected to pixel adaptive sampling and mark feature edge pixel extraction; then, an improved function combining the Brenner function and the Roberts function is used to evaluate and calculate the sampled pixels to obtain the initial focus evaluation function value of each image; in order to reduce the sensitivity of the improved function to noise, the initial focus evaluation function value of the image set is subjected to variance calculation to obtain the final focus evaluation function value; according to the focus evaluation function curve of the image, the focal length can be adjusted in real time to obtain a precisely focused image, thereby realizing high-precision detection of wafer alignment marks. The focus evaluation function constructed in this embodiment has strong noise resistance, good stability and timeliness, and is suitable for automatic focusing of wafer alignment marks with diverse changes in edge directions.

[0050] 1. This embodiment is aimed at the application background of the diversity of edge directions of wafer alignment mark image features, combines the advantages of Brenner function and Roberts function, and can efficiently extract edge gradient information in multiple directions of the mark image, thereby achieving high-precision focusing.

[0051] 2. This embodiment integrates the traditional focusing evaluation function and adaptive sampling. Although the computational complexity is increased, the amount of calculation is greatly reduced through adaptive sampling, thereby achieving efficient calculation. It not only overcomes the problems of poor noise resistance and stability of the traditional focusing evaluation function, but also solves the problems of complex calculation and poor real-time performance of the fused focusing function.

[0052] 3. This embodiment improves the problem of poor noise resistance of the spatial domain focusing function by finally calculating the variance of the evaluation function, and has better robustness.

[0053] This embodiment not only meets the unimodal and unbiased properties of the focusing curve, but also has good stability and noise resistance, and also has an adaptive focusing evaluation function with high computational efficiency, thereby achieving fast and accurate automatic focusing of the wafer alignment mark.

[0054] This embodiment takes the poor noise resistance and stability of traditional focus evaluation functions and the large amount of calculation and poor real-time performance of fusion focus evaluation functions as the research basis, and proposes an adaptive focus evaluation method for wafer alignment marks in view of the application background of the diverse changes in the edge directions of wafer alignment marks.

[0055] In order to achieve this purpose, the technical solution adopted in this embodiment is:

[0056] An adaptive focus evaluation method for wafer alignment marks is disclosed. Firstly, the acquired defocus-focus-defocus sequence images of the wafer alignment marks are adaptively sampled for the pixel points of the mark images and their characteristic edge pixel points are extracted. Then, the improved function Brenner2d-Roberts2d is used to evaluate and calculate the sampled pixel points to obtain the initial focus evaluation function value of each image. In order to reduce the sensitivity of the improved function to noise and enhance its stability and noise resistance, the variance of the initial evaluation function value is calculated to obtain the final focus evaluation function Brenner2d-Roberts2d-Var.

[0057] The specific steps include:

[0058] (1) Adaptive sampling of marked image pixels:

[0059] According to the size of the wafer alignment mark defocus-focus-defocus sequence image acquired by the camera, different sampling intervals S1 are determined:

[0060] S1=max(X,Y) / n

[0061] Wherein, X and Y are the height and width of each image; n is the number of sampling times, which is determined according to the calculation efficiency of the evaluation function and the size of the wafer alignment mark.

[0062] (2) Extraction of edge pixels of marked image features:

[0063] In order to reduce the amount of calculation, the simplest grayscale difference is used to judge the edge pixels of the alignment mark image feature. First, the grayscale value standard deviation σ of each image is calculated as the basis for judging whether it is an edge pixel:

[0064]

[0065] Where f(x,y) is the grayscale value of the pixel (x,y); XY is the image pixel size; u is the average pixel value of the entire image.

[0066] Then calculate the grayscale difference between the pixel point (x, y) and its right pixel point (x+1, y) and the pixel point below (x, y+1), and judge it with the grayscale value standard deviation σ of the image to determine whether it is an edge pixel point, and adjust the adaptive sampling spacing S in time:

[0067] R=|f(x,y)-f(x+1,y)

[0068] D=|f(x,y)-f(x,y+1)

[0069] Where R is the absolute value of the grayscale difference between the pixel (x, y) and the pixel to its right (x+1, y); D is the absolute value of the grayscale difference between the pixel (x, y) and the pixel to its right (x, y+1).

[0070]

[0071] By judging the grayscale difference R, D and the standard deviation σ, if R and D are both smaller than the standard deviation σ, it means that the pixel is not the mark edge, and continue to sample at the large spacing S1. If R or D is larger than the standard deviation σ, it means that the pixel is the mark edge, and the sampling spacing is reduced to S2. S2 is determined based on the calculation efficiency of the evaluation function and the size of the wafer alignment mark.

[0072] (3) Improved focusing evaluation function:

[0073] Considering that the wafer alignment mark has multiple characteristic edge directions, this embodiment intends to perform evaluation function calculations on the four directions (0°, 45°, 90°, and 135°) of the mark image. First, the Brenner function is selected in the 0° and 90° directions. This function is one of the spatial domain focusing evaluation functions with the lowest computational complexity, but it only calculates the square of the pixel grayscale difference of two units in the 0° direction. Therefore, without increasing the amount of calculation, the calculation of the pixel grayscale difference of two units in the 90° direction is added:

[0074]

[0075] Then select the Roberts function in the 45° and 135° directions. This function calculates the grayscale gradient value in the diagonal direction. However, since the Brenner function operator is a 3×3 matrix, in order to unify the grayscale gradient calculation dimension, the Roberts operator is adjusted from 2×2 to 3×3 matrix:

[0076]

[0077] Finally, multiply the two functions to obtain the initial focusing evaluation function that satisfies the four directions:

[0078]

[0079] (4) Variance calculation

[0080] Since the number of calculation directions of the evaluation function has increased to 4, although it solves the azimuth limitation problem of the traditional spatial focus evaluation function in the gray gradient calculation to a certain extent, and is better adapted to the evaluation calculation of the wafer alignment mark image, making the calculation result more accurate and the focusing sensitivity improved, it also increases the noise sensitivity of the evaluation function and greatly weakens the anti-noise performance. Therefore, the variance calculation is added on the basis of the above improved function. The variance function can enhance the stability of the evaluation function and ensure that the evaluation function is not easily disturbed by noise:

[0081]

[0082] Where N is the number of defocus-focus-defocus sequence image sets, u Bre2d_Rob2d is the initial average evaluation function value of the out-of-focus sequence image set, F Bre2d_Rob2d_Var is the final focusing evaluation function.

[0083] Finally, based on the focus evaluation function value of the defocus-focus-defocus sequence image, the focus evaluation function curve is fitted to achieve precise focus positioning and obtain a clear focused image, thereby realizing high-precision detection of wafer alignment marks.

[0084] The specific application object of this embodiment is a wafer bonding alignment mark automatic focusing system. The system first captures a sequence of images of alignment mark defocus-focus-defocus, such as Figure 1 After the image set is acquired, the adaptive focus evaluation method for wafer alignment marks proposed in this embodiment is used for calculation. The process is as follows: Figure 2 shown.

[0085] The detailed calculation process is as follows:

[0086] (1) Determination of pixel adaptive sampling interval S1 and sampling times n:

[0087] The size of the sequential images of the wafer alignment mark defocus-focus-defocus acquired by the camera is 1280×1024; n is the number of sampling times, which is determined to be 60 according to the calculation efficiency of the evaluation function and the size of the wafer alignment mark; the sampling interval S1 is rounded to obtain:

[0088] S1=max(X,Y) / n

[0089] =max(1280,1024) / 60

[0090] ≈21

[0091] (2) Determination of the threshold σ for extracting the characteristic edge pixel points of the wafer alignment mark:

[0092] First, calculate the standard deviation σ of the gray value of each image as the basis for judging whether it is an edge pixel:

[0093]

[0094] Where f(x,y) is the grayscale value of the pixel (x,y); XY is the image pixel size; u is the average pixel value of the entire image.

[0095] Then calculate the grayscale difference between the pixel point (x, y) and its right pixel point (x+1, y) and the pixel point below (x, y+1), and judge it with the grayscale value standard deviation σ of the image to determine whether it is an edge pixel point, and adjust the adaptive sampling spacing in time:

[0096] R=|f(x,y)-f(x+1,y)

[0097] D=|f(x,y)-f(x,y+1)

[0098] Where R is the absolute value of the grayscale difference between the pixel (x, y) and the pixel to its right (x+1, y); D is the absolute value of the grayscale difference between the pixel (x, y) and the pixel to its right (x, y+1).

[0099]

[0100] By judging the gray value difference R, D and the standard deviation σ, if R and D are both smaller than the standard deviation σ, it means that the pixel is not the mark edge, and continue to sample at the large spacing S1. If R or D is larger than the standard deviation σ, it means that the pixel is the mark edge, and the sampling spacing is reduced to S2. S2 is determined to be 1 based on the calculation efficiency of the evaluation function and the size of the alignment mark structure.

[0101] (3) Use the multi-directional gradient focusing evaluation function to perform convolution calculation on the above sampling pixels:

[0102] First, use the Brenner operator to calculate the pixel grayscale difference of two units in the 0° and 90° directions:

[0103]

[0104] Then, the improved Roberts operator is used to calculate the pixel grayscale difference of two diagonal units in the 45° and 135° directions:

[0105]

[0106] Finally, multiply the two functions to obtain the initial focusing evaluation function that satisfies the four directions:

[0107]

[0108] (4) Perform variance calculation on the focus evaluation function of the labeled sequence image set:

[0109]

[0110] Where N is the number of alignment mark defocus-focus-defocus sequence image sets, which is 11, u Bre2d_Rob2d is the initial average evaluation function value of the out-of-focus sequence image set, F Bre2d_Rob2d_Var is the final focusing evaluation function.

[0111] Finally, based on the focus evaluation function value of the alignment mark defocus-focus-defocus sequence image, the focus evaluation function curve is fitted to achieve precise focus positioning and obtain a clear focused image, thereby realizing high-precision detection of the wafer alignment mark.

[0112] In order to verify the performance improvement of the method of this embodiment, performance comparison experiments and analyses were conducted on this embodiment and traditional focusing evaluation functions: Brenner, Roberts, Tenengrad functions, DFT two-dimensional discrete Fourier transform image focusing evaluation function, and Bre4d_var evaluation function proposed by Dong Zhengqiong et al. in 2023.

[0113] The above focus evaluation function is used to calculate the defocus-focus-defocus sequence images of the same wafer alignment mark, and the obtained focus evaluation values ​​are normalized to obtain a comparison chart of the focus evaluation function curve, as shown in FIG. Figure 3 In order to more objectively compare and analyze the performance differences of different focusing evaluation functions, the steep area width W, clarity ratio R, steepness Stp and sensitivity f are selected. sen Performance Indicators The above focusing evaluation functions are analyzed and compared, as shown in Table 1. The meanings of relevant indicators are as follows.

[0114] Steep zone width W: The steep zone width W represents the width of the region with a larger curve change rate on the evaluation function curve. It depends on the specific shape and change rate of the evaluation function curve and is a very important parameter. It can be used as a reference indicator for selecting the maximum step length of the focusing process.

[0115] Clarity ratio R: The clarity ratio R represents the ability of the focus evaluation function to distinguish images with different defocus levels. The larger the focus ratio, the greater the difference in focus values ​​between the clear image and the blurred image, and the easier it is to distinguish the focus function:

[0116]

[0117] where f max is the maximum value of the focusing function, f min is the minimum value of the focusing function.

[0118] Steepness Stp: Steepness Stp indicates the steepness of the focus evaluation function curve at the peak. The greater the steepness, the stronger the function's ability to resolve images with different defocus levels.

[0119] Sensitivity f sen : Sensitivity f sen Indicates the intensity of changes near the maximum value of the focus evaluation function. The higher the sensitivity, the more drastic the function changes, and the better the image focusing effect. Conversely, the function changes more slowly:

[0120]

[0121] where f max is the maximum value of the focusing function, f(z max +θ) is the function value when the horizontal axis changes by θ.

[0122] Table 1 Summary of performance evaluation indicators of focusing evaluation function

[0123]

[0124]

[0125] As shown in Table 1, the steep zone width W of the improved focus evaluation function of this embodiment is the smallest compared with other evaluation functions; the clarity ratio R is several orders of magnitude higher than other evaluation functions; the steepness Stp is 7 times the minimum value of other evaluation functions; the sensitivity f sen It is 14 times the minimum value of other evaluation functions.

[0126] In summary, this embodiment is an automatic focusing evaluation algorithm with strong anti-noise performance, good stability and timeliness, and suitable for wafer alignment mark images with diverse changes in edge directions.

[0127] A focus evaluation system for wafer alignment marks, comprising:

[0128] An image acquisition module, used for acquiring a sequence image set of the wafer alignment mark in the process of defocusing-focusing-defocusing;

[0129] A focusing evaluation module is used to perform edge extraction on each sequence image in the sequence image set; perform focus evaluation in different directions of each sequence image after edge extraction based on a Brenner function and a Roberts function to obtain an initial focus evaluation function value corresponding to each sequence image; perform variance calculation on the initial focus evaluation function value corresponding to each sequence image to obtain a final focus evaluation function value; and judge the focus state of the wafer alignment mark based on the final focus evaluation function value.

[0130] An electronic device comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a focus evaluation method for wafer alignment marks.

[0131] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the focus evaluation method for wafer alignment marks is implemented.

[0132] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A focus evaluation method for wafer alignment marks, characterized in that: include: Acquire a sequence image set of the wafer alignment mark in the defocus-focus-defocus process; Performing edge extraction on each sequence image in the sequence image set; Based on the Brenner function and the Roberts function, focusing evaluation is performed in different directions of each sequence image after edge extraction, so as to obtain an initial focusing evaluation function value corresponding to each sequence image; Performing variance calculation on the initial focus evaluation function values ​​corresponding to each of the sequence images to obtain a final focus evaluation function value; The focus state of the wafer alignment mark is determined based on the final focus evaluation function value.

2. A focus evaluation method for wafer alignment marks according to claim 1, characterized in that: Performing edge extraction on each sequence image in the sequence image set specifically includes: Determine the initial sampling interval, and calculate the standard deviation of the grayscale value of each of the sequence images; Calculating the grayscale value difference between each pixel to be marked and adjacent pixels in each of the sequence images, wherein the adjacent pixels include pixels to the right and below the current pixel to be marked; The gray value standard deviation is compared with the gray value difference of adjacent pixels, and based on the comparison result, it is determined whether the current pixel to be marked is an edge pixel.

3. The focus evaluation method for wafer alignment marks according to claim 12, characterized in that: Comparing the gray value standard deviation with the gray value difference of adjacent pixels specifically includes: If the grayscale value differences of adjacent pixels are all smaller than the grayscale value standard deviation, it is determined that the current pixel to be marked is not an edge pixel; If there is a grayscale value difference between the grayscale value differences of adjacent pixels that is greater than the grayscale value standard deviation, then the current pixel to be marked is determined to be an edge pixel; After determining that the current pixel to be marked is an edge pixel, the current pixel to be marked is marked and the initial sampling interval is reduced.

4. The focus evaluation method for wafer alignment marks according to claim 1, characterized in that: Based on the Brenner function and the Roberts function, focus evaluation is performed in different directions of each sequence image after edge extraction, specifically including: The Brenner function is used to perform evaluation calculation in the 0° and 90° directions of the sequence images, and the pixel grayscale difference calculation in the 90° direction is added to obtain a first focus evaluation function value; the 3×3 Roberts operator is used to perform grayscale gradient calculation in the 45° and 135° directions of the sequence images to obtain a second focus evaluation function value, and the first focus evaluation function value and the second focus evaluation function value are multiplied to obtain an initial focus evaluation function value.

5. The focus evaluation method for wafer alignment marks according to claim 1, characterized in that: The variance calculation is performed on the initial focus evaluation function values ​​corresponding to each of the sequence images. The specific calculation formula is: Where N is the number of images in the sequence, U Bre2d_Rob2d is the initial average evaluation function value of the sequence image set, F Bre2d_Rob2d_Var is the final focusing evaluation function value.

6. A focus evaluation system for wafer alignment marks, characterized in that: include: An image acquisition module, used for acquiring a sequence image set of the wafer alignment mark in the process of defocusing-focusing-defocusing; A focus evaluation module is used to perform edge extraction on each sequence image in the sequence image set; perform focus evaluation in different directions of each sequence image after edge extraction based on a Brenner function and a Roberts function to obtain an initial focus evaluation function value corresponding to each sequence image; The variance of the initial focus evaluation function values ​​corresponding to each of the sequence images is calculated to obtain a final focus evaluation function value; and the focus state of the wafer alignment mark is judged based on the final focus evaluation function value.

7. An electronic device, characterized in that: It comprises a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a focus evaluation method for wafer alignment marks according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements a focus evaluation method for wafer alignment marks as described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Self-adaptive automatic focusing method based on gray gradient value

    CN109669264A

  • Microscope fine focusing method based on random forest

    CN112433339A

  • Microscope automatic focusing method and device based on image definition evaluation

    CN113109936A

  • Image focusing accuracy evaluation method and system

    CN114972084A

  • Part image high-precision focusing method and device based on improved gradient weighting

    CN117252915A

Cited By

  • Wafer defect detection method and system and electronic equipment

    CN121033033A