Overlay mark asymmetry degree metering method, device and system and storage medium

Abstract

The invention provides an overlay mark asymmetry degree metering method, device and system and a storage medium, and the method comprises the steps: collecting an image of a to-be-measured target structure based on an optical imaging system, and obtaining a discrete sampling signal; determining an initial center position of the overlay mark through preset algorithm processing, establishing an analysis function, and performing Taylor expansion on the analysis function to obtain each analysis item; the analysis item is used for reflecting asymmetry of the overlay mark in the to-be-measured target structure; and counting the global item and the local item, and calculating the geometric offset in the overlay mark asymmetry according to the error offset corresponding to each analysis item. According to the method, the asymmetry of the overlay mark is accurately quantified on global and local levels by utilizing Taylor expansion, a nondestructive measurement means is realized, the fine asymmetry of a target structure is quantified, the recognition precision of the asymmetry is improved, measurement errors caused by the asymmetry can be eliminated, and the measurement accuracy of the overlay mark is improved.

Description

Technical Field

[0001] The present application relates to the field of semiconductor manufacturing technology, and in particular to a method, device, system and storage medium for measuring the degree of asymmetry of overlay marks. Background Art

[0002] In the semiconductor manufacturing process, as the process nodes continue to shrink, the reduction in device feature size and the complexity of multi-layer structures have put forward higher requirements for the alignment accuracy of the production line. Accurate alignment between layers is an important link to ensure process consistency, production yield and capacity. Due to factors such as process fluctuations and changes in photoresist thickness, slight asymmetry or local damage to the target structure often occurs. These asymmetries bring significant measurement errors during optical overlay measurement (OVL). This error directly affects the accuracy and reliability of OVL measurement, which in turn leads to a decrease in yield and serious damage to performance during chip production.

[0003] Traditional methods such as slice measurement or scanning electron microscope (SEM) measurement are lossy in identifying asymmetry, which makes it inefficient in actual production and difficult to meet the needs of efficient and high-precision measurement.

[0004] Therefore, a new scheme for measuring the degree of asymmetry of overlay marks is needed. Summary of the invention

[0005] In view of this, embodiments of the present disclosure provide a method, device, system, and storage medium for measuring the degree of asymmetry of an overlay mark.

[0006] The embodiments of this specification provide the following technical solutions:

[0007] The embodiment of this specification provides a method for measuring the degree of asymmetry of an overlay mark, including:

[0008] Based on the optical imaging system, an image of the target structure to be measured is collected to obtain a discrete sampling signal;

[0009] The discrete sampling signal is processed by a preset algorithm to determine the preliminary center position of the overlay mark;

[0010] Establishing an analysis function according to the preliminary center position, and performing Taylor expansion on the analysis function to obtain various analysis items; the analysis items include global items of the target structure to be measured and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured;

[0011] The geometric offset in the overlay mark asymmetry is calculated according to the error offsets corresponding to the global term and the local term respectively.

[0012] The embodiment of the present specification provides a device for measuring the degree of asymmetry of an overlay mark, comprising:

[0013] A sampling module, used for acquiring an image of the target structure to be measured based on an optical imaging system to obtain a discrete sampling signal;

[0014] An acquisition module, used to process the discrete sampling signal through a preset algorithm to determine the preliminary center position of the overlay mark;

[0015] An analysis module is used to establish an analysis function according to the preliminary center position, and to perform Taylor expansion on the analysis function to obtain various analysis items; the analysis items include global items of the target structure to be measured and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured;

[0016] A calculation module is used to calculate the geometric offset in the overlay mark asymmetry according to the error offsets respectively corresponding to the global term and the local term.

[0017] Compared with the prior art, the at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects:

[0018] The embodiments of this specification combine traditional algorithms with Taylor expansion to accurately quantify the asymmetry of overlay marks at global and local levels. The global asymmetry is captured by low-order derivatives, and the local asymmetry is characterized by high-order derivatives, thereby realizing a non-destructive measurement method. Compared with traditional algorithms, it reduces the measurement error caused by ignoring details and quantifies the subtle asymmetry of the target structure. It not only improves the recognition accuracy of asymmetric shapes, but can also be used to eliminate the measurement error introduced by asymmetry, thereby improving the accuracy of overlay mark measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, 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 application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 These are schematic diagrams of four common target asymmetric structures;

[0021] Figure 2 is a schematic diagram of a marking signal in an embodiment of the present application;

[0022] Figure 3 is a flow chart of a method for calculating the degree of asymmetry of an overlay mark according to an embodiment of the present application;

[0023] Figure 4 It is a schematic diagram of the calculation principle of the asymmetry degree of an overlay mark according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0025] The following describes the implementation methods of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the following embodiments and the features in the embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0026] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspect described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0027] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0028] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the examples can be practiced without these specific details.

[0029] In the semiconductor production and processing process, each layer of manufacturing processes such as lithography, chemical mechanical polishing (CMP), and shallow trench isolation (STI) needs to be aligned with high precision based on the previous layer to ensure that the final device performance meets the design specifications.

[0030] Due to factors such as process fluctuations and photoresist thickness variations, small asymmetries or local damage to the target structure are often caused. These asymmetries can cause significant measurement errors during optical overlay measurement (OVL).

[0031] In the actual production process, non-ideal factors of various processes will introduce asymmetry in the physical morphology, which increases the difficulty of the alignment process. For example:

[0032] Chemical mechanical polishing (CMP) may result in non-uniform material removal, which may cause surface unevenness.

[0033] Asymmetric etching or filling in shallow trench isolation (STI) processes leads to differences in local geometry.

[0034] Uneven photoresist coating will affect the edge quality and morphology of the pattern, further exacerbating the asymmetry of the shape.

[0035] The glue wall angle in the thick glue process is difficult to control, resulting in different side wall angles.

[0036] Especially when the photoresist thickness reaches a certain level (such as 4μm), the asymmetry error may even exceed 50nm. This error amplification directly affects the accuracy and reliability of OVL measurement, which in turn leads to a decrease in yield and serious damage to performance during chip production.

[0037] These physical morphology asymmetries introduced by the process are particularly prominent in the optical imaging and measurement process. Due to the discrete sampling characteristics of optical imaging, key details of asymmetric morphology may be ignored or lost during the sampling process, thus appearing as errors in the final measurement results. For example, when the target morphology loses important edge or contour information in optical imaging, the measurement system's recognition and positioning of the target will be significantly affected.

[0038] Traditional optical measurement methods, such as edge detection technology based on threshold method or center positioning method based on cross-correlation algorithm, usually assume that the measured target has a high degree of symmetry or a regular shape. However, when faced with asymmetric target structures introduced in actual production, the accuracy and reliability of these methods will be greatly reduced:

[0039] The threshold method determines the target edge position by setting a fixed brightness or grayscale threshold, but when the edge is asymmetric, the threshold selection often cannot accurately reflect the actual geometric center.

[0040] The cross-correlation algorithm locates the target center by finding the best matching area in the image, but the morphological deviation of the asymmetric structure will lead to error accumulation and affect the final alignment accuracy.

[0041] In this case, the optical measurement system is more sensitive to the shape of the target, especially when facing asymmetric structures. Traditional measurement methods will cause measurement deviations due to the lack of key information or misjudgment, significantly reducing the accuracy and precision of the measurement. This error accumulation is particularly prominent in multi-layer alignment, which ultimately affects the process yield and the electrical performance of the chip.

[0042] Therefore, there is an urgent need for a method that can accurately quantify the asymmetry of the target structure based on the optical imaging process without relying on traditional destructive testing methods (such as slicing or SEM, the abbreviation of Scanning Electron Microscope). This method can help Fab process engineers to timely identify and evaluate process-induced asymmetry problems during the production process, quickly adjust process parameters or eliminate error data, thereby effectively reducing yield loss.

[0043] Based on this, the embodiment of this specification proposes a new scheme for measuring the degree of asymmetry of the overlay mark, which uses the idea of ​​Taylor expansion to accurately capture the local and global asymmetry of the target structure by analyzing the high-order expansion of the OVL measurement signal at a specific point. Through detailed analysis of derivatives of various orders, the microscopic and macroscopic structural features described by derivatives of different orders can quantitatively reflect the subtle asymmetry of the target structure. This method not only improves the recognition accuracy of asymmetry, but also effectively reduces the measurement error introduced by asymmetry, especially the error amplification phenomenon that occurs when the adhesive layer is thicker, and its effectiveness is verified by SEM-CD, aiming to solve the error problem in the existing OVL measurement process, thereby improving the measurement accuracy and yield in the semiconductor manufacturing process.

[0044] The following are explanations of professional terms:

[0045] Overlay error (OVL) describes the deviation between the current layer lithography and the previous layer lithography in the lithography process, thereby achieving monitoring and providing data to correct the process.

[0046] Overlay marks (OVL-Mark) are graphic marks used to measure overlay errors. These marks are distributed in a certain number in the lithography unit. In terms of measurement type, they are generally divided into imaging measurement (IBO, Image-based Overlay) and non-imaging diffraction measurement (DBO, diffraction based overlay). The overlay mark graphics will also vary due to processing errors, such as Figure 1 Parts A, B, C, and D respectively illustrate an overlay mark pattern.

[0047] Mark-Signal is a marker signal used to measure the OVL error. It is a spatial signal obtained by CIS and is divided into two directions: x (horizontal direction) and y (vertical direction). Figure 2 .

[0048] CIS (CMOS-Image-Sensor) is a sensor used to receive light signals marked by test samples.

[0049] The technical solutions provided by various embodiments of the present application are described below in conjunction with the accompanying drawings.

[0050] In the existing optical measurement process, the threshold method and the cross-correlation algorithm are two commonly used edge detection and center positioning technologies. However, as the precision requirements of semiconductor manufacturing processes continue to increase, these traditional algorithms have exposed a series of limitations when dealing with asymmetric target structures, resulting in error accumulation in the measurement results, which ultimately affects the measurement accuracy and process alignment accuracy.

[0051] Among them, the limitations of the threshold method

[0052] The threshold method sets a fixed brightness or grayscale threshold and regards the area exceeding the threshold as the edge of the target. Under ideal conditions, this method can effectively determine the edge position for symmetrical and clearly defined targets. However, when the target structure is asymmetrical, the reliability of the threshold method is greatly reduced. The main reasons are as follows:

[0053] Errors due to edge asymmetry: In an asymmetric structure, the two edges of the target may have different brightness or grayscale gradients. The set fixed threshold cannot accurately capture the position of both edges at the same time. For example, one edge of the target structure may become blurred or deformed due to photoresist thickness or process errors, while the other edge is still relatively clear. In this case, the fixed threshold will capture different positions on both sides, resulting in a shift in the calculated geometric center, which in turn introduces measurement errors.

[0054] Sensitivity of threshold selection: The threshold setting is very sensitive, especially when dealing with complex, asymmetric or disturbed images. A slightly improper threshold selection may lead to significant deviations in edge positions. Since the edge characteristics of asymmetric structures behave differently in different locations, a single threshold setting cannot adapt to the differences in all areas, further exacerbating the accumulation of errors.

[0055] Therefore, the threshold method is difficult to accurately reflect the actual geometric center when dealing with complex and asymmetric structures, especially when the target edge is unclear or there is shape variation, which can easily cause significant measurement errors. This makes its measurement results unreliable under high-precision requirements and cannot meet the strict requirements for alignment accuracy in the semiconductor manufacturing process.

[0056] Among them, the limitations of the cross-correlation algorithm

[0057] The cross-correlation algorithm determines the optimal alignment position of the target by calculating the degree of match between two images or signals. In optical measurement, it is often used to find the best matching area and locate the geometric center of the target by cross-correlating the acquired image with the reference image. However, for asymmetric structures, the cross-correlation algorithm is prone to error accumulation in the following cases:

[0058] Matching errors caused by morphological deviations: The core assumption of the cross-correlation algorithm is that the target morphology should remain consistent or symmetrical between different positions. However, when the target morphology is asymmetric, such as when there is a significant geometric deviation or local deformation between one side of the structure and the other, cross-correlation matching becomes unreliable. Since cross-correlation is essentially a match based on the overall shape, asymmetric morphology will cause the best matching position to deviate from the actual center position, ultimately leading to measurement errors.

[0059] Amplification effect of local asymmetry: In asymmetric structures, even small shape deviations may cause error accumulation in the cross-correlation algorithm. For example, if a part of the target has an edge deformation or irregularity due to process error, this local asymmetry will be amplified in the cross-correlation operation, causing the center position of the measurement to deviate from the actual center. As the degree of asymmetry increases, errors will continue to accumulate, affecting the reliability and stability of the measurement.

[0060] Multiple solutions caused by asymmetry: The cross-correlation algorithm may cause multiple solutions in the case of asymmetric structures, that is, due to the diversity of asymmetric shapes, the algorithm may find high matches in multiple locations. This makes it difficult for the system to determine the only center position, thereby increasing the uncertainty of the measurement.

[0061] Therefore, the cross-correlation algorithm is difficult to provide accurate target center positioning when dealing with asymmetric structures, especially for complex structures or structures with slight local differences. The measurement results are easily affected by asymmetric morphological deviations, resulting in error accumulation and ultimately a significant decrease in measurement accuracy.

[0062] The existing threshold method and cross-correlation algorithm both have significant limitations when facing asymmetric target structures. The threshold method cannot adapt to edge asymmetry due to the selection of a fixed threshold, resulting in geometric center positioning deviation; the cross-correlation algorithm is difficult to accurately locate the actual center of the asymmetric structure due to the amplification effect of morphological deviation and the problem of multiple solutions. These deficiencies have led to a significant reduction in the accuracy of the measurement results of the existing technology when dealing with asymmetric targets, which in turn affects the alignment accuracy in the production process and cannot meet the needs of modern semiconductor manufacturing for high-precision measurement.

[0063] like Figure 3 and Figure 4 As shown, the embodiment of this specification provides a method for measuring the degree of asymmetry of the overlay mark, including steps S301 to S304. In step S301, an image of the target structure to be measured is collected based on an optical imaging system to obtain a discrete sampling signal; in step S302, the discrete sampling signal is processed by a preset algorithm to determine the preliminary center position of the overlay mark; in step S303, an analysis function is established according to the preliminary center position, and the analysis function is Taylor expanded to obtain various analysis items, the analysis items include global items of the target structure to be measured, and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured; in step S304, the geometric offset in the asymmetry of the overlay mark is calculated according to the error offsets corresponding to the global items and the local items respectively.

[0064] Specifically, step S301, based on the optical imaging system, an image of the target structure to be measured is collected to obtain a discrete sampling signal, wherein there is an overlay mark in the target structure to be measured.

[0065] Discrete sampling principle in the imaging process. In the optical measurement process, the formation of the image is based on the sampling process of the optical reflection signal and the imaging system. The light source is irradiated on the target structure to be measured, and the reflection signal is generated through the reflection and scattering of light. The optical imaging system captures these reflection signals through optical lenses and sensors and forms discrete pixel images. The formula of this process is as follows:

[0066]

[0067] Among them, I(x,y) is the discrete optical sampling signal acquired by the imaging system, A(x′,y′) is the actual physical morphology of the target structure, and PSF(x,y) is the point spread function (Point Spread Function) in the optical imaging system. This formula describes the optical response of each target point in the system.

[0068] In the actual imaging process, discrete sampling is performed based on the pixel resolution of the sensor, and the pixels in the image are the sampling results of the target physical structure. Since the sampling points are discrete, discontinuous sampling will lose some detailed information of the target structure, especially when the target structure is asymmetric or has slight damage, the sampling process cannot fully capture these asymmetries, thus affecting the accuracy of subsequent measurements.

[0069] In addition, the optical reflection coupling function in the optical system further affects the performance of the imaging signal. The reflection, interference and scattering of light on complex structures will produce different optical responses due to surface irregularities or asymmetries. The coupling function describes the combined effect of this light reflection, and its response to complex and asymmetric target structures will lead to deviations in the optical imaging results.

[0070] Step S302: Process the discrete sampling signal by a preset algorithm to determine the preliminary center position of the overlay mark, wherein the center of the overlay mark is preliminarily determined by using the preset algorithm.

[0071] In some embodiments, the preset algorithms include but are not limited to traditional algorithms such as threshold method, cross-correlation algorithm, edge detection algorithm, morphological algorithm, NCC matching algorithm and peak detection algorithm.

[0072] In the embodiments of this specification, the cross-correlation algorithm is used as an example to explain in detail. The cross-correlation algorithm compares the matching degree of two signals or images, such as comparing the matching degree of discrete sampling signals with reference signals, to find the most similar parts. In optical measurement, cross-correlation is used to locate the center of the target, and the center of the overlay mark can be preliminarily determined. Specifically, the discrete sampling signal and the reference signal are cross-correlated to determine the preliminary center position of the overlay mark. The reference signal is used to compare and evaluate the similarity with the discrete sampling signal, and is a known signal.

[0073] However, for asymmetric structures, cross-correlation only considers global similarity and ignores subtle local changes in the signal, which can easily lead to error accumulation and make it difficult to accurately reflect the geometric center of the asymmetric target.

[0074] Therefore, in order to improve the defects of traditional algorithms and realize accurate identification and quantification of asymmetric structures, the Taylor expansion method is introduced on the basis of the traditional algorithm to capture the local and global characteristics of asymmetric targets through derivatives of different orders.

[0075] Step S303: establish an analysis function according to the preliminary center position, and perform Taylor expansion on the analysis function to obtain various analysis items. The analysis items include global items of the target structure to be measured and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured.

[0076] Specifically, the calculation formula 2 of the analysis function obtained by combining the above cross-correlation function is:

[0077] R a,b (τ)=∫f a (x)f b (x+τ)dx (Formula 2)

[0078] Among them, f a (x) and f b (x) represents the sampling signal and the reference signal, and τ represents the offset.

[0079] Although the above formula can calculate the similarity of two signals at different offsets, it cannot characterize the local details and asymmetry of the signals, and then conduct analysis in combination with Taylor.

[0080] The above cross-correlation function (i.e., analysis function) f R(τ) Taylor expansion analysis is performed to accurately describe the local area characteristics of the signal.

[0081] Taylor expansion formula 3 is as follows:

[0082]

[0083] Near a certain point x0 of the target structure, we perform Taylor expansion on the optical signal I(x) (assuming that only one-dimensional signals are expanded here, and the two-dimensional case is similar), and expand it into the sum of derivatives of various orders as shown in Formula 4:

[0084]

[0085] Combined with the expression of optical imaging signals, this formula indicates that the signal of the imaging system near x0 can be described by first-order, second-order, third-order derivatives, and each derivative represents local changes at different levels.

[0086] Global asymmetry: The low-order terms of the Taylor expansion, especially the first and second derivatives, can describe the overall change trend of the signal. If two signals show global asymmetry (for example, different overall shapes or slopes), the first and second derivatives will show significant differences. These differences can help identify the global asymmetry of the target structure.

[0087] Local asymmetry: Higher-order derivatives are particularly sensitive to local characteristics of the signal. For example, third-order or fourth-order derivatives can reveal subtle fluctuations and local morphological changes in the signal. For asymmetric structures, local morphological differences may not be obvious globally, but will be more prominent in higher-order derivatives. Therefore, the higher-order terms of the Taylor expansion can capture local differences in asymmetric structures, reflecting small asymmetries in the target structure that are not easy to detect.

[0088] In addition, Taylor expansion is performed on traditional algorithms to achieve improvements in different directions to accurately quantify asymmetry. For example, the threshold method usually determines the edge position based on brightness or grayscale thresholds, but this method is sensitive to asymmetry, especially when the brightness gradients of the left and right sides or the upper and lower edges of the target are inconsistent, it is difficult to obtain accurate edge positioning. By introducing Taylor expansion, the accuracy of the threshold method can be improved. Another example is that edge detection algorithms (such as Canny edge detection, Sobel operator, Laplace operator, etc.) are widely used to identify edge features in images. However, these algorithms often lose details or misjudge edge positions when faced with asymmetric structures or discontinuous edges. The introduction of Taylor expansion can improve the accuracy of edge detection. Another example is that morphological operations (such as corrosion, expansion, opening operation, and closing operation) are usually used to process the morphological features of binary images, such as filling holes or separating connected target objects. However, morphological operations are more sensitive to the asymmetry of the target structure. By introducing Taylor expansion, the processing effect of the morphological algorithm can be improved. The high-order terms of Taylor expansion can help identify smaller and finer local features in the image. In morphological processing, these detail information can be retained to avoid local asymmetric features being erased or filled due to morphological operations, thereby improving the adaptability of morphological processing to asymmetric structures.

[0089] Another example is the NCC (Normalized Cross-Correlation) matching algorithm, which is used to find the position that best matches the template in the image, and is usually used to locate a specific shape or target. However, when the target structure is asymmetric, the accuracy of template matching will decrease. After the introduction of Taylor expansion, the local feature differences between the template and the actual target can be described by derivatives of various orders, thereby enhancing the sensitivity of matching. Another example is in peak detection or extreme value detection applications, where the extreme points of the signal are used to identify certain features. However, for asymmetric signals or complex backgrounds, the peak position may shift. By introducing Taylor expansion, the accuracy of peak detection can be improved, and Taylor expansion can be performed on the signal near the peak, and the peak point can be accurately located by derivative analysis. If the first-order derivative is close to zero but the second-order derivative is large, the shape and position of the peak can be analyzed by high-order derivatives to adapt to the characteristics of asymmetric signals. The derivatives of various orders of Taylor expansion can help distinguish signal peaks from noise backgrounds. For example, when the high-order derivative terms of the signal change significantly, it can be determined that the signal is a true peak rather than a noise fluctuation.

[0090] The embodiments of this specification can accurately quantify asymmetry at the global and local levels by combining a preset algorithm and Taylor expansion. Global asymmetry is captured by low-order derivatives, while local asymmetry is characterized by high-order derivatives. This method reduces the measurement error caused by ignoring details in traditional algorithms, quantifies the subtle asymmetry of the target structure, and not only improves the recognition accuracy of asymmetric shapes, but can also be used to eliminate measurement errors introduced by asymmetry and improve the accuracy of overlay mark measurement.

[0091] Step S304: Calculate the geometric offset in the overlay mark asymmetry according to the error offsets corresponding to the global term and the local term respectively.

[0092] Since each derivative represents a local change at a different level, it can be calculated by error offset.

[0093] Assume that the geometric center of the signal is x0, but due to asymmetry, the center actually measured is offset by an amount Δx. According to the high-order derivatives of Taylor expansion, the error offset is expressed as Formula 5:

[0094]

[0095] Where n represents the order of Taylor expansion, represents the function value of the nth derivative at x0. This formula actually shows that the signal deviation caused by asymmetry has a cumulative effect under different derivatives. The signal characteristics corresponding to each derivative can reveal different levels of geometric deviation.

[0096] The embodiments of this specification can effectively characterize the global and local asymmetry of the target structure by combining the Taylor expansion of the optical imaging signal with the analysis of derivatives of different orders. For example, the measurement of the asymmetry degree of overlay measurement caused by surface unevenness in the chemical mechanical polishing process can be eliminated by analyzing the global term, and the measurement of the asymmetry degree of overlay measurement caused by uneven distribution of photoresist in the photolithography process can be eliminated by analyzing the global term, and the measurement of the asymmetry degree of overlay measurement caused by asymmetric etching or filling in the shallow trench isolation process can be eliminated by analyzing the local term.

[0097] In some embodiments, Taylor expansion is performed on the analysis function to obtain various analysis items, including: Taylor expansion is performed on the optical signal at a preset range position of the target structure to be measured, and the sum of each order derivative and the analysis item represented by each order derivative are obtained by expansion; wherein the global item includes the first-order derivative and the second-order derivative, the first-order derivative reflects the global change trend of the overlay mark, and the second-order derivative reflects the local convexity or surface change of the overlay mark; the local item includes the third-order derivative and the fourth-order derivative, and the third-order derivative or the fourth-order derivative reflects the local fluctuation and morphological change of the overlay mark.

[0098] Specifically, near a certain point x0 of the target structure, we perform Taylor expansion on the optical signal I(x) (assuming that only one-dimensional signal is expanded here, the two-dimensional case is similar and will not be repeated here), expand it into the sum of derivatives of various orders, and obtain the analysis terms represented by each order derivative.

[0099] Combined with the Taylor expansion of the above embodiment, the signal of the imaging system near x0 can be described by first-order, second-order, third-order derivatives, etc., and each derivative represents a local change at a different level:

[0100] First Derivative Indicates the slope change of the signal, reflecting the trend of global changes.

[0101] Second Derivative Represents the curvature of the signal, revealing local convexity or surface changes.

[0102] Third level and higher Derivatives describe more subtle local changes and are more sensitive to asymmetries. For example, third- and fourth-order derivatives can reveal subtle fluctuations and local morphological changes in the signal.

[0103] In some embodiments, the error offset formula is

[0104] Where n represents the order of Taylor expansion, represents the function value of the nth-order derivative at x0. This formula shows that the signal deviation caused by asymmetry has a cumulative effect under different orders of derivatives; among them, the signal characteristics corresponding to each order of derivative represent different levels of geometric deviation.

[0105] In combination with the above embodiment, the traditional algorithm and Taylor expansion, for the error caused by the asymmetric structure, the specific geometric offset can be quantified by the offset of the derivative peak. Assuming that the geometric center of the signal is x0, but due to the asymmetry, the center actually measured is offset by an amount Δx, according to the high-order derivative of Taylor expansion, the error offset can be expressed as:

[0106] Where n represents the order of Taylor expansion, represents the function value of the nth derivative at x0. This formula actually shows that the signal deviation caused by asymmetry has a cumulative effect under different derivatives. That is, the signal characteristics corresponding to each derivative can be calculated by calculating the above error offset formula to represent different levels of geometric deviation.

[0107] In some embodiments, when the first-order derivative When a certain position reaches the zero point, the position should be the "center point" of the signal. Then, according to the above error offset formula, the "center point" is used as the origin to analyze the different positions of the two x, and the positions of the two x are symmetrical with the "center point". The geometric deviations corresponding to each order derivative corresponding to the different positions of the two x can be analyzed. Then, the asymmetric manifestation of the overlay mark can be obtained by statistically analyzing the geometric deviations of multiple different x positions within a certain range.

[0108] In some embodiments, the method further includes constructing an error distribution model by statistically analyzing the peak distribution of different-order derivatives in each analysis item, wherein the error distribution model quantifies the different levels of geometric deviation represented by each-order derivative.

[0109] Specifically, in order to quantify the asymmetry and its impact on the measurement accuracy, we can analyze the peak positions of different derivatives. Assuming that the derivative of the signal shows a certain deviation on the asymmetric structure, we can get the error distribution by the following steps:

[0110] First-order derivative peak: When the first-order derivative When the zero point is reached at a certain position, that position should be the "center point" of the signal. If there is global asymmetry in the structure, the center point will be offset from the true geometric center of the target. This offset can reflect the error caused by global asymmetry.

[0111] Second-order derivative peak: The peak of the second-order derivative corresponds to the change in curvature of the signal. If the two sides of the target structure are asymmetrical, the extreme points of the second-order derivative will shift. By analyzing the extreme value distribution of the second-order derivative, the error caused by local asymmetry can be further quantified.

[0112] Third-order and higher derivative peaks: Third-order and higher derivatives are particularly sensitive to small changes and details in the signal, especially in complex asymmetric structures, where these higher-order derivatives can show local errors more significantly. The peak shift of higher-order derivatives can reflect more subtle asymmetry characteristics and their impact on measurement accuracy.

[0113] By statistically analyzing the peak distribution of different-order derivatives, we can construct an error distribution model that can quantitatively describe the impact of asymmetric structures in imaging signals and evaluate the global and local error sources based on the contribution of each order derivative.

[0114] In some embodiments, the position point corresponding to when the first-order derivative in the global term is 0 is used as the target center position.

[0115] Assuming that the derivative of the signal shows a certain deviation on the asymmetric structure, the first-order derivative peak: when the first-order derivative When the zero point is reached at a certain position, that position should be the "center point" of the signal. If there is global asymmetry in the structure, the center point will be offset from the true geometric center of the target. This offset can reflect the error caused by global asymmetry.

[0116] In some embodiments, the geometric offset in the overlay mark asymmetry is used as a process reference to adjust the process or correct the measurement results.

[0117] In combination with the above embodiments, the signal characteristics corresponding to each order derivative can represent different levels of geometric deviations, and then the geometric deviations are output to provide asymmetry results for reference by process engineers to adjust the process or correct the measurement results.

[0118] In some embodiments, the semiconductor manufacturing process includes photolithography, chemical mechanical polishing, and trench isolation.

[0119] In combination with the above-mentioned embodiments, the Taylor expansion formula of the optical imaging signal is combined with different order derivatives and their peak analysis, and the present invention can effectively characterize the global and local asymmetry of the target structure. The signal characteristics corresponding to each order derivative can reveal different levels of geometric deviations, and finally quantify the impact of these asymmetries on the measurement accuracy through the error distribution model. This method greatly improves the sensitivity to asymmetric structures and can provide more accurate measurement results. Particularly suitable manufacturing processes include but are not limited to photolithography, chemical mechanical polishing (CMP), shallow trench isolation (STI), etc.

[0120] In addition, this method also provides a solid foundation for further optimizing measurement accuracy, implementing error compensation and improving asymmetry, ensuring the continuity of the production process and improving product quality.

[0121] The same or similar parts between the various embodiments in this specification can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the product embodiment described later, since it corresponds to the method, the description is relatively simple, and the relevant parts can be referred to the partial description of the system embodiment.

[0122] The above is only a 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 shall be based on the protection scope of the claims.

Claims

1. A method for measuring the degree of asymmetry of an overlay mark, characterized in that: include: Based on the optical imaging system, an image of the target structure to be measured is collected to obtain a discrete sampling signal; The discrete sampling signal is processed by a preset algorithm to determine the preliminary center position of the overlay mark; Establishing an analysis function according to the preliminary center position, and performing Taylor expansion on the analysis function to obtain various analysis items; the analysis items include global items of the target structure to be measured and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured; The geometric offset in the overlay mark asymmetry is calculated according to the error offsets corresponding to the global term and the local term respectively.

2. The method for measuring the degree of asymmetry of overlay marks according to claim 1, characterized in that: Taylor expansion is performed on the analysis function to obtain various analysis items, including: Performing Taylor expansion on the optical signal at a preset range position of the target structure to be measured, and obtaining the sum of each order derivative and the analysis item represented by each order derivative; The global term includes the first-order derivative and the second-order derivative. The first-order derivative reflects the global change trend of the overlay mark, and the second-order derivative reflects the local concavity or surface change of the overlay mark. The local terms include third-order derivatives and fourth-order derivatives, and the third-order derivatives or fourth-order derivatives reflect the local fluctuations and morphological changes of the overlay mark.

3. The method for measuring the degree of asymmetry of overlay marks according to claim 1, characterized in that: The error offset formula is: Where n represents the order of Taylor expansion, represents the function value of the nth-order derivative at x0. This formula shows that the signal deviation caused by asymmetry has a cumulative effect at different orders of derivatives; Among them, the signal features corresponding to each order derivative represent different levels of geometric deviations.

4. The method for measuring the degree of asymmetry of overlay marks according to claim 1, characterized in that: The position point corresponding to the first-order derivative of the global term being 0 is taken as the target center position.

5. The method for measuring the degree of asymmetry of overlay marks according to claim 1, characterized in that: The method also includes constructing an error distribution model by statistically analyzing the peak distribution of different-order derivatives in each analysis item, wherein the error distribution model is used to quantify geometric deviations of different levels represented by each-order derivative.

6. The method for measuring the asymmetry degree of overlay marks according to any one of claims 1 to 5, characterized in that: The geometric offset in the overlay mark asymmetry is used as a process reference to adjust the process or correct the measurement results.

7. The method for measuring the degree of asymmetry of overlay marks according to claim 6, characterized in that: Semiconductor manufacturing processes include photolithography, chemical mechanical polishing, and shallow trench isolation.

8. A device for measuring the degree of asymmetry of overlay marks, characterized in that: include: A sampling module, used for acquiring an image of the target structure to be measured based on an optical imaging system to obtain a discrete sampling signal; An acquisition module, used to process the discrete sampling signal through a preset algorithm to determine the preliminary center position of the overlay mark; An analysis module is used to establish an analysis function according to the preliminary center position, and to perform Taylor expansion on the analysis function to obtain various analysis items; the analysis items include global items of the target structure to be measured and local items of the target structure to be measured, and the analysis items are used to reflect the asymmetry of the overlay mark in the target structure to be measured; A calculation module is used to calculate the geometric offset in the overlay mark asymmetry according to the error offsets respectively corresponding to the global term and the local term.

9. A system for measuring the degree of asymmetry of overlay marks, characterized in that: include: A memory, a processor and a computer program, wherein the computer program is stored in the memory, and the processor runs the computer program to execute the overlay mark asymmetry degree measurement method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, it is used to implement the method for measuring the degree of asymmetry of an overlay mark as claimed in any one of claims 1 to 7.

Images