Wafer alignment mark-oriented variable step curve fitting focusing search method

Abstract

The invention belongs to the technical field of optical metrology, and discloses a wafer alignment mark-oriented variable step curve fitting focusing search method, which comprises the following steps of: after acquiring an initial focusing range of an imaging system, performing traversal search at a larger first sampling step to obtain a first wafer mark image set; therefore, the first focusing position is judged and the focusing range is shrunk. And then, traversing again in the contraction range by adopting a smaller second sampling step length, and determining an image acquisition range required by curve fitting. And a third wafer mark image set is collected within the range, and a fitting curve is obtained through a curve fitting method, so that the optimal focusing point position is accurately determined. And finally, the focal length of the imaging system is adjusted according to the position, and high-precision focusing of the wafer mark is realized. According to the method, the focus range is quickly shrunk by adopting large-step traversal search, the influence of a local peak value is avoided by improving the search precision, curve fitting is carried out in a final range, and the precision and efficiency of a focus search algorithm are ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of optical metrology, and particularly relates to a variable-step curve fitting focusing search method for wafer alignment marks. Background Art

[0002] In the process of semiconductor manufacturing, the precise detection of wafer marks runs through all aspects of the process manufacturing. This is not only the basis for achieving precise positioning of process steps but also a key link to ensure that the equipment can complete operations in specific areas of the wafer. As the core technology for precise detection of wafer marks, the autofocus technology in optical detection methods is the key to optical detection imaging. Through autofocus, it is possible to ensure that the camera obtains a clear mark image, thereby achieving precise identification of wafer marks, reducing positioning errors, and significantly improving the yield of chip products. In this technology, the focusing search algorithm adjusts the imaging object distance through a control mechanism to achieve precise focusing. This algorithm is the key technical bridge for analyzing the focusing evaluation results and converting them into physical focusing actions. However, with the evolution of chip technology, traditional focusing search algorithms are prone to problems of insufficient accuracy when facing micro-nano structure marks, such as blurred edge features or improper selection of the focusing window.

[0003] Typical focusing search algorithms include the traversal search method, the Fibonacci search method, the curve fitting search algorithm, and the hill climbing search algorithm, etc. The traversal search method is accurate but has a long search time and a large amount of calculation, making it difficult to meet the real-time requirements. The Fibonacci search method has wide applicability but is prone to focusing failure due to the influence of local peaks. The curve fitting search algorithm determines the optimal focusing position by fitting the focusing evaluation function curve, and the focusing accuracy depends on the accuracy of the function expression. The hill climbing search algorithm uses the unimodality of the focusing evaluation function for search but is also easily interfered by local peaks.

[0004] To solve these problems, in the prior art, there are algorithms that combine an adaptive step size and a two-step search method, adjusting the search strategy according to the size of the focusing evaluation function value and the curve shape to improve the speed and accuracy, but they are still restricted by the motor stepping accuracy and have a low focusing efficiency. Another method is based on the threshold method and local maximum values, ignoring local extrema caused by interference factors by setting thresholds to improve the focusing success rate, but the threshold selection is greatly affected by the environment and has poor universality. There is also an infrared autofocus fast search method that uses a variable-step hill climbing method, combines the method of taking the median of the image sharpness obtained multiple times to ensure the evaluation accuracy, and uses the hill climbing method with momentum and acceleration in the coarse focusing stage to reduce the number of steps, but it is still difficult to completely solve the problem of falling into local optimal solutions. Summary of the Invention

[0005] The object of the present invention is to provide a variable-step curve fitting focusing search method for wafer alignment marks to solve the problems existing in the above prior art.

[0006] To achieve the above object, the present invention provides a variable-step curve fitting focusing search method for wafer alignment marks, including:

[0007] Obtain the initial focusing range in the imaging system;

[0008] Traverse and search the initial focusing range based on the first sampling step to obtain the first wafer mark image set; judge the first focusing position based on the first wafer mark image set, and shrink the focusing range based on the first focusing position;

[0009] Traverse and search in the shrunk focusing range based on the second sampling step to obtain the second wafer mark image set, and determine the curve fitting image acquisition range based on the second wafer mark image set; wherein, the second sampling step is smaller than the first sampling step;

[0010] Collect a number of wafer mark images based on the curve fitting image acquisition range to obtain the third wafer mark image set; perform fitting calculation on the third wafer mark image set based on the curve fitting method to obtain a fitting curve;

[0011] Determine the best focus point position based on the fitting curve, and adjust the focal length of the imaging system according to the best focus point position to achieve high-precision focusing on the wafer mark.

[0012] Optionally, the judging of the first focusing position based on the first wafer mark image set specifically includes:

[0013] Drive the motor to traverse the initial focusing range based on the first sampling step, and collect wafer mark images at each position to obtain the first wafer mark image set;

[0014] Evaluate and calculate the images in the first wafer mark image set based on the focusing evaluation function, judge the first focusing position according to the evaluation results of the images in the first wafer mark image set, and shrink the focusing range with the first focusing position as the center.

[0015] Optionally, the judging process of the first focusing position specifically includes:

[0016] Judge the evaluation results of the images in the first wafer mark image set. If the evaluation results continuously increase or decrease, the initial focusing range does not contain the best imaging position, and the overall focusing range is displaced by a preset length in the direction of the increasing evaluation results and traversed again; if the evaluation results do not show a continuous increasing or decreasing trend, the image position corresponding to the maximum evaluation result is the first focusing position.

[0017] Optionally, determining the curve fitting image acquisition range based on the second set of wafer marking images specifically includes:

[0018] Within the shrunk focusing range, traverse and acquire at a second sampling step size to obtain the second set of wafer marking images;

[0019] Evaluate and calculate the images in the second set of wafer marking images based on a focus evaluation function, and take the image position corresponding to the maximum evaluation result as the second focusing position; take the second focusing position as the midpoint to determine the curve fitting image acquisition range.

[0020] Optionally, the process of obtaining the third set of wafer marking images specifically includes:

[0021] Within the curve fitting image acquisition range, determine a third sampling step size based on the preset number of optimal sampling points for curve fitting, and equally spacedly acquire a number of wafer marking images based on the third sampling step size to obtain the third set of wafer marking images.

[0022] Optionally, the process of obtaining the fitting curve specifically includes:

[0023] Evaluate and calculate the images in the third set of wafer marking images based on a focus evaluation function to generate a fitting data set;

[0024] Perform fitting calculation on the fitting data set based on a curve fitting method to obtain a fitting curve.

[0025] Optionally, the process of obtaining the optimal focus point position specifically includes:

[0026] Calculate the extreme point of the fitting curve, which is the optimal focus point position.

[0027] The technical effects of the present invention are as follows:

[0028] 1. In the initial search process of the present invention, a large step size is used for traversal search. According to the principle that the imaging object distance varies greatly so that the gray level gradient of the marking edge changes greatly and the clarity of the image is more obvious, the focus search range can be quickly shrunk, avoiding the interference of environmental factors such as noise in the imaging system.

[0029] 2. By improving the traversal search accuracy and acquiring images with a smaller step size, the present invention can effectively avoid the influence of local peaks of the evaluation function, and at the same time more accurately determine the range where the peak is located, providing a good data basis for subsequent curve fitting.

[0030] 3. By performing curve fitting within the final range obtained from the traversal search, the present invention can fit an accurate evaluation function curve without a large number of image acquisitions, determine the optimal imaging position, and ensure the search accuracy and efficiency of the focus search algorithm. Brief Description of the Drawings

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0032] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application. In the drawings:

[0033] Figure 1 is a specific implementation flowchart of the variable step - length curve fitting focusing search algorithm for wafer - facing alignment marks in the embodiments of the present invention;

[0034] Figure 2 is a schematic diagram of the focusing experiment data of the search algorithm in the embodiments of the present invention; wherein, Figure 2 in (a) is a schematic diagram of the focusing experiment data during the first - stage traversal search in the embodiments of the present invention; Figure 2 in (b) is a schematic diagram of the focusing experiment data during the second - stage traversal search in the embodiments of the present invention; Figure 2 in (c) is a schematic diagram of the focusing experiment data during the curve fitting process in the embodiments of the present invention;

[0035] Figure 3 is a comparison diagram of the focusing experiment effects in the embodiments of the present invention. Detailed Description of the Embodiments

[0036] Now, various exemplary embodiments of the present invention will be described in detail. This detailed description should not be considered as a limitation of the present invention, but rather as a more detailed description of certain aspects, characteristics, and implementation schemes of the present invention.

[0037] It should be understood that the terms described in the present invention are only for describing specific embodiments and are not used to limit the present invention. Additionally, for the numerical ranges in the present invention, it should be understood that each intermediate value between the upper and lower limits of the range is also specifically disclosed. Each intermediate value within any stated value or stated range, as well as each smaller range between any other stated value or intermediate value within the stated range, is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded from the range.

[0038] Without departing from the scope or spirit of the present invention, various modifications and variations to the specific embodiments of the description of the present invention will be apparent to those skilled in the art. Other embodiments obtained from the description of the present invention will be apparent to those skilled in the art. The description and examples of this application are merely exemplary.

[0039] Regarding the use of "comprising", "including", "having", "containing", etc. in this article, they are all open-ended terms, meaning including but not limited to.

[0040] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will detail this application with reference to the drawings and in combination with the embodiments.

[0041] As Figure 1 - Figure 3 shown, in this embodiment, a variable-step curve fitting focusing search method for wafer alignment marks is provided, including: obtaining an initial focusing range in the imaging system; traversing and searching the initial focusing range based on a first sampling step to obtain a first set of wafer mark images; determining a first focusing position based on the first set of wafer mark images, and shrinking the focusing range based on the first focusing position; traversing and searching within the shrunk focusing range based on a second sampling step to obtain a second set of wafer mark images, and determining a curve fitting image acquisition range based on the second set of wafer mark images; wherein, the second sampling step is less than the first sampling step; collecting a plurality of wafer mark images based on the curve fitting image acquisition range to obtain a third set of wafer mark images; performing fitting calculation on the third set of wafer mark images based on a curve fitting method to obtain a fitting curve; determining an optimal focus point position based on the fitting curve, and adjusting the focal length of the imaging system according to the optimal focus point position to achieve high-precision focusing on the wafer marks.

[0042] This embodiment relates to a variable-step curve fitting focusing search algorithm for wafer alignment marks. The search range is converged by a traversal search algorithm with continuously improving accuracy, and the focus position is obtained by curve fitting. The specific steps are as follows: First, in the initial traversal search process, the imaging object distance varies greatly, so that the gray gradient of the mark edge changes greatly, and the clarity of the image is more obvious. A large-step traversal sampling is used to quickly contract the focus search range and avoid the interference of environmental factors such as noise in the imaging system. Then, after the accuracy of the traversal search is improved, by collecting images with a smaller step size, the influence of local peaks of the evaluation function is effectively avoided, and the range where the peak is located is accurately positioned, providing a good data basis for subsequent curve fitting. Finally, curve fitting is performed within the final range obtained by the traversal search to find the extreme point of the fitting curve, which is the best imaging position, avoiding the redundant image acquisition process in the traditional method. Through this staged traversal fitting search method, not only can a large amount of sampling work be reduced, the algorithm time consumption be reduced, but also the influence of local peaks of the evaluation function can be effectively removed, and the robustness of the search algorithm can be enhanced.

[0043] Based on the problems of low search efficiency, poor anti-noise performance, and insufficient adaptability to complex scenarios of traditional focusing search algorithms, this embodiment combines the traversal search algorithm and the curve fitting algorithm in the application background of high-precision focusing requirements for wafer marks, and proposes a variable-step curve fitting focusing search algorithm for wafer alignment marks.

[0044] To achieve this goal, the technical solutions adopted in this embodiment specifically include the following steps:

[0045] The traversal search in the first stage:

[0046] (1) Determine the initial sampling step size L of the motor 1 , L 1 =Nd 1 , d 1 is the minimum moving step size of the displacement platform in the imaging system, and the value of N needs to be determined according to the relevant parameters of the actual focusing imaging system;

[0047] (2) Drive the motor to traverse the initial focusing range A with a length of a at the step size L 1 , collect the wafer mark images, generate the image set P 0 , and use the focusing evaluation function to evaluate and calculate the images in the image set P 1 ; 1 (3) If the data obtained from the image evaluation calculation in the image set P

[0048] is continuously increasing or decreasing, it means that the initial focusing range A 1 is correct, and the focus is within the range A 0If the best imaging position is not included, shift the overall focusing range by a in the direction of increasing evaluation value, and return to step (1) to traverse again. Otherwise, skip this step and continue to step (4);

[0049] (4) According to the data obtained from the evaluation calculation, find the focusing position x corresponding to the image with the maximum evaluation value 1 , with point x 1 as the midpoint, shrink the search range to A 1 [x 1 -L 1 , x 1 +L 1 ;

[0050] Traversal search in the second stage:

[0051] (5) Improve the traversal sampling accuracy of the motor, and the sampling step size is L 2 , L 2 = L 1 / 2;

[0052] (6) Within the search range A 1 , perform a more accurate traversal acquisition to obtain the image set P 2 , and use the focusing evaluation function to evaluate and calculate the images in the image set P 2 again;

[0053] (7) Find the focusing position x corresponding to the image with the maximum evaluation value 2 , with point x 2 as the midpoint, determine the curve fitting image acquisition range A 2 [x 2 -L 2 , x 1 +L 2 ;

[0054] Since the traversal search method is relatively simple and efficient, and it meets the requirements of the timeliness of autofocus, the number and accuracy of traversal searches can be increased according to the actual situation.

[0055] Curve fitting stage:

[0056] (8) Within the fitting range A 2 , according to the number of best sampling points X for curve fitting, determine the sampling step size L 3 , and equally spaced collect X marked images to obtain the image set P 3 ;

[0057] (9) Use the focusing evaluation function to evaluate and calculate the images in the image set P 3 to generate the fitting data set D; use the curve fitting method to perform fitting calculations on the fitting data set D to obtain the fitting curve F;

[0058] (10) Find the extreme point S of the fitting curve F, which is the optimal focus position.

[0059] In the initial search process of this embodiment, a large step size is used for traversal search. According to the principle that the imaging object distance difference is large so that the gray level gradient change at the edge of the mark is large and the image clarity is more obvious, the focus search range can be quickly shrunk, avoiding the interference of environmental factors such as noise in the imaging system.

[0060] By improving the traversal search accuracy and collecting images with a smaller step size in this embodiment, the influence of local peaks of the evaluation function can be effectively avoided, and at the same time, the range where the peak is located can be more accurately determined, providing a good data basis for subsequent curve fitting.

[0061] By performing curve fitting within the final range obtained from the traversal search in this embodiment, an accurate evaluation function curve can be fitted without a large number of image acquisitions, and the optimal imaging position can be determined, ensuring the search accuracy and efficiency of the focusing search algorithm.

[0062] The specific application object of this embodiment - the automatic focusing system for wafer marking. Through this system, a wafer marking focusing experiment is carried out, and calculations are performed according to the variable step size curve fitting focusing search algorithm for wafer alignment marks proposed in this embodiment. The process is as Figure 1 shown.

[0063] The detailed calculation process is as follows:

[0064] In the experimental preparation stage, relevant parameters such as the search step size in the variable step size curve fitting search algorithm are determined through the parameters of the built imaging system. From the technical parameters of the infrared camera, it can be known that the pixel size p of the camera is 5μm, and the limit resolution distance e of the camera sensor can be calculated by the following formula:

[0065] e = 2P = 2×5 = 10μm

[0066] The depth of field range of the system can be calculated through the depth of field calculation formula in the microscopic imaging system. The refractive index n of the medium between the objective lens and the object takes a value of 1 in the formula:

[0067]

[0068] According to the wavelength range of the halogen light source selected in this embodiment and the response band of the infrared camera, it can be known that the wavelength λ ranges from 400 to 1700nm, and the depth of field range of the system is 24.2 to 30.6μm.

[0069] Combined with the positioning accuracy of the displacement platform, the two traversal sampling step sizes L 1 、L 2They are determined to be 40μm and 20μm respectively. From the experiment on the number of image sampling for fitting, it can be seen that the optimal sampling times for curve fitting is 10 times. Therefore, the sampling step L 3 is 4μm.

[0070] After the experimental parameters are set, adjust the marking structure into the camera's field of view and ensure that the best imaging position is at the middle position of the search range. On this basis, conduct the wafer marking focusing experiment under different defocus states. For example Figure 2 Shown is a schematic diagram of the change in EVG marking focusing experiment data with a search range of 1mm and a defocus distance of 200μm.

[0071] During the first-stage traversal search process, a total of 26 marked images are collected within the initial focusing range A 0 . The change in the sharpness evaluation value is as shown in Figure 2 (a). Among them, the 18th marked image has the largest sharpness evaluation value, and the corresponding position x 1 is 680μm, and the search range shrinks to A 1 [640, 720]μm;

[0072] During the second-stage traversal search process, a total of 5 marked images are collected. Among them, the 4th image has the largest sharpness evaluation value as shown in Figure 2 (b). The corresponding position x 2 is 700μm, and the search range shrinks again. Finally, the fitting sampling range is determined to be A 2 [680, 720]μm;

[0073] Within the fitting sampling range, a total of 10 marked images are sampled. By using the curve fitting method of this embodiment to fit the sharpness evaluation values corresponding to the 10 marked images, the peak value of the fitting curve is finally obtained as 701.8μm, as shown in Figure 2 (c). This point is the best focusing position of the system. As shown in Figure 3 is the comparison image of the initial defocus state and the marked image after focusing.

[0074] As can be seen from the above experimental results, the autofocus algorithm shows good focusing effect after being integrated into the hardware platform. To further verify the performance of the overall system in terms of focusing accuracy and stability, a large number of focusing experiments with different defocus degrees were carried out. To ensure the accuracy and comparability of the experimental results, multiple different types of marks were selected on the same wafer sample for focusing experiments to comprehensively evaluate the performance of the system under various conditions. First, the cross mark with the simplest structure was used to determine the ideal imaging position. After 30 groups of focusing experiments were carried out on it, the average focal position was obtained as the best imaging position of the current system, and the imaging object distance was adjusted to make the system in a quasi-focused state. Secondly, on the basis of the best imaging position, focusing experiments were carried out on single-layer Bar, EVG sub, EVG mother marks, double-layer EVG and Bar-in-Bar marks at different defocus distances with the defocus distance varying from 0.1 to 1 mm and the interval being 0.1 mm. To avoid the influence of hardware system errors on the experimental results, 10 experiments were carried out for each defocus distance, and the maximum deviation, minimum deviation and average deviation of the focal position in each group of experiments were respectively counted. Tables 1 and 2 are the focusing experimental data of single-layer and double-layer marks respectively. The plus or minus sign of the distance in the table indicates whether the defocus position is above or below the best imaging position.

[0075] Table 1 Focusing Experimental Data of Single-Layer Marks

[0076]

[0077] Table 2 Focusing Experimental Data of Double-Layer Marks

[0078]

[0079] Since the depth of field range is 24.2 - 30.6 μm, according to the principle of optical imaging, the focusing results with the focal position deviation greater than the minimum depth of field of 24.2 μm are determined as focusing failures. Among the above 1000 focusing experiments, there are 8 focusing failures in total, and the focusing success rate is 99.2%, which fully verifies the high stability of the improved autofocus algorithm in the infrared imaging system of wafer marks in this embodiment. In addition, it can be seen from the experimental data that the average deviation of the double-layer marks at different defocus distances increases slightly compared with that of the single-layer marks, which indicates that when the double-layer wafer is focused and imaged, it is indeed affected by the focusing evaluation of the actual imaging object distance difference between the mother and son marks. As the defocus distance increases, the interference of environmental factors such as external light gradually increases, and the focal position deviation will show a gradually increasing trend. Nevertheless, whether it is a single-layer or double-layer wafer mark, its average deviation is less than the minimum depth of field, and the overall average deviation of 1000 focusing experiments is 9.9 μm, which is less than half of the minimum depth of field of the imaging system. This shows that the autofocus search algorithm proposed in this embodiment can achieve accurate focusing and imaging of wafer marks, and has strong practicability and reliability.

[0080] As described above, it is only the preferred specific implementation manner of the present application. However, the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed in the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A variable step length curve fitting focus search method for wafer alignment marks, characterized in that: include: Acquiring an initial focus range in an imaging system; Performing a traversal search on the initial focus range based on a first sampling step length to obtain a first wafer mark image set; Determining a first focus position based on the first wafer mark image set, and shrinking a focus range based on the first focus position; Performing a traversal search within the shrunken focus range based on a second sampling step size to obtain a second wafer mark image set, and determining a curve fitting image acquisition range based on the second wafer mark image set; wherein the second sampling step size is smaller than the first sampling step size; Collecting a number of wafer mark images based on the curve fitting image acquisition range to obtain a third wafer mark image set; performing fitting calculation on the third wafer mark image set based on a curve fitting method to obtain a fitting curve; The best focus point position is determined based on the fitting curve, and the focal length of the imaging system is adjusted according to the best focus point position to achieve high-precision focusing on the wafer mark.

2. The variable step length curve fitting focusing search method for wafer alignment marks according to claim 1, characterized in that: The determining a first focus position based on the first wafer mark image set specifically includes: Based on the first sampling step length, the motor is driven to traverse the initial focus range, and a wafer mark image is collected at each position to obtain a first wafer mark image set; An evaluation calculation is performed on the images in the first wafer mark image set based on a focus evaluation function, a first focus position is determined according to the evaluation results of each image in the first wafer mark image set, and a focus range is narrowed with the first focus position as the center.

3. The variable step length curve fitting focusing search method for wafer alignment marks according to claim 2, characterized in that: The first focus position determination process specifically includes: The evaluation results of each image in the first wafer marking image set are judged. If the evaluation results are continuously increasing or decreasing, the optimal imaging position is not included in the initial focus range, and the overall focus range is shifted by a preset length in the direction of increasing evaluation results and re-traversed; if the evaluation results do not show a continuous increasing or decreasing trend, the image position corresponding to the maximum evaluation result is the first focus position.

4. The variable step length curve fitting focus search method for wafer alignment marks according to claim 1, characterized in that: The determining of the curve fitting image acquisition range based on the second wafer mark image set specifically includes: Within the shrunken focus range, performing traversal acquisition with a second sampling step length to obtain a second wafer mark image set; Based on the focus evaluation function, the images in the second wafer mark image set are evaluated and calculated, and the image position corresponding to the maximum evaluation result is used as the second focus position; and the curve fitting image acquisition range is determined with the second focus position as the midpoint.

5. The variable step length curve fitting focus search method for wafer alignment marks according to claim 1, characterized in that: The process of acquiring the third wafer mark image set specifically includes: Within the curve fitting image acquisition range, a third sampling step is determined based on a preset number of curve fitting optimal sampling points, and a plurality of wafer mark images are evenly spaced based on the third sampling step to obtain a third wafer mark image set.

6. The variable step length curve fitting focus search method for wafer alignment marks according to claim 1, characterized in that: The process of obtaining the fitting curve specifically includes: Performing evaluation calculation on the images in the third wafer mark image set based on the focus evaluation function to generate a fitting data set; The fitting data set is fitted based on the curve fitting method to obtain the fitting curve.

7. The variable step length curve fitting focus search method for wafer alignment marks according to claim 1, characterized in that: The process of obtaining the best focus point position specifically includes: The extreme point of the fitting curve is calculated to be the optimal focus point position.

Images