Method for monitoring performance of a photographing device, method for monitoring integrated semiconductor patterning and measurement processes and system therefor

Abstract

A method for monitoring the performance of a shooting device according to an embodiment comprises: a step of collecting an image to be analyzed by shooting an object to be analyzed using a shooting device; a step of deriving a pixel histogram of the image to be analyzed and calculating a DW value (Dark or White value) of the image to be analyzed using the pixel histogram; and a step of evaluating whether an abnormality has occurred in the shooting device by comparing the DW value of the image to be analyzed with reference information.

Description

Technology Field

[0001] The present invention relates to a method for monitoring the performance of an imaging device, a method for monitoring by integrating semiconductor patterning and measurement processes, and a system for the same. Background Technology

[0002] Recently, as the integration density of semiconductor devices increases, the critical dimension (CD) of semiconductor patterns is becoming smaller. Generally, to manufacture highly integrated semiconductor devices, securing a patterning process capable of accurately forming fine semiconductor patterns at desired locations and creating patterns with high critical dimension uniformity is a crucial factor. Accordingly, various methods are being researched and developed to improve pattern overlay and CD uniformity.

[0003] The semiconductor photolithography process generally utilized to manufacture the above-mentioned semiconductor device includes a patterning process in which laser light is irradiated onto a mask, the laser light passing through the mask is collected, and the mask pattern is transferred onto a wafer to form a pattern on the wafer. In the patterning process, it is important to form a deep, clear, and noise-free pattern by ensuring that the focal point of the laser light, collected through a lens, precisely aligns with the surface of the wafer. In other words, determining the precise focal point of the laser light acts as a critical factor in improving the reliability of fine pattern formation in the patterning process. In particular, when transferring multiple semiconductor patterns with various critical dimensions and shapes onto a wafer in combination through the patterning process, if the process is not performed at the optimal laser focal point that takes all of these factors into account, it frequently occurs that some of the multiple patterns are not transferred in the correct position and shape.

[0004] Meanwhile, in the semiconductor industry, after performing a patterning process on a wafer as described above, a measurement process is performed to verify whether the target pattern has been correctly formed on the wafer.

[0005] The above measurement process is performed by using a photographing device to capture a pattern formed on a wafer to collect a pattern image, and by verifying the pattern in the collected pattern image to check for the occurrence of defects in the pattern. In such a measurement process, since the quality of the pattern image is affected by the focus of the photographing device, the pattern image is collected while the photographing device maintains an optimal focus.

[0006] As mentioned above, since focus acts as a major factor in determining process quality in semiconductor patterning and measurement processes, a method capable of accurately detecting the optimal focus is required to maintain the performance of semiconductor exposure and imaging devices.

[0007] However, since there is currently no method to accurately express focus as objective numerical values ​​when performing processes using semiconductor lithography and imaging devices, there is a problem in maintaining the performance of semiconductor patterning and measurement processes, so research is needed on methods to compensate for this. Prior art literature

[65535] Korean Registered Patent No. 10-1369670 (Publication Date: Dec. 11, 2013) Korean Published Patent No. 10-2016-0034131 (Publication Date: March 29, 2016) The problem to be solved

[0008] According to one embodiment, the invention aims to provide technical content regarding a method for monitoring the performance of a shooting device that can accurately determine whether an abnormality has occurred due to focus fluctuations of the shooting device.

[0009] According to one embodiment, the present invention aims to provide technical content regarding an integrated monitoring method for semiconductor patterning and measurement processes and a system for such process, which can comprehensively monitor whether an abnormality occurs in a semiconductor exposure device and an imaging device to maintain the performance of the semiconductor patterning process and the semiconductor measurement process. means of solving the problem

[0010] A method for monitoring the performance of a photographing device according to an embodiment may include: a step of collecting an image of an object to be analyzed by photographing it using a photographing device; a step of deriving a pixel histogram of the image to be analyzed and calculating a DW value (Dark or White value) of the image to be analyzed using the pixel histogram; and a step of evaluating whether an abnormality has occurred in the photographing device by comparing the DW value of the image to be analyzed with reference information. An integrated monitoring method for semiconductor patterning and measurement processes according to an embodiment may include: a step of forming a pattern on a wafer by performing an exposure process using a semiconductor exposure device; a step of collecting a pattern image of an analysis target pattern using a photographing device; a step of deriving a pixel histogram of the pattern image and calculating a DW value (Dark or White value) of the pattern image using the pixel histogram; and a step of evaluating whether an abnormality has occurred in the semiconductor exposure device and the photographing device by comparing at least one of the pixel histogram and the DW value of the pattern image with reference information.

[0011] An integrated monitoring system for semiconductor patterning and measurement processes according to an embodiment may include: a pattern forming unit that performs an exposure process to form a pattern on a wafer; an image collecting unit that collects a pattern image of a pattern to be analyzed formed on the wafer; and an abnormality detection unit that evaluates whether an abnormality has occurred in the semiconductor exposure device and the imaging device by comparing at least one of the pixel histogram and DW value of the pattern image with reference information. Effects of the invention

[0012] The performance monitoring method of a shooting device according to an embodiment can evaluate whether an abnormality has occurred in the shooting device by calculating a pixel histogram and a DW value from an image to be analyzed collected using the shooting device, clearly confirm the level of abnormality in the shooting device numerically, and track and manage the performance of the shooting device by recognizing changes in focus of the shooting device using the said numerical value.

[0013] The integrated monitoring method for semiconductor patterning and measurement processes according to the embodiment calculates a pixel histogram and DW value from an image to be analyzed to recognize changes in focus between a patterning device and a shooting device, thereby enabling simultaneous management of the semiconductor patterning process and the measurement process for inspecting the semiconductor device, and allows for the identification of productivity and trends by numerically representing the level of abnormal occurrence, and enables the exclusion of arbitrary evaluations by managers, so that the same process management standards can be applied to multiple devices.

[0014] In addition, the integrated monitoring method for semiconductor patterning and measurement processes according to the embodiment enables the analysis of causal factors using data mining techniques when an anomaly occurs, and can prevent problems that may occur in the patterning process in advance through learning, thereby enabling the realization of effects such as reduced management costs, improved work efficiency, and rapid response to problems. Brief explanation of the drawing

[0015] FIG. 1 is a process diagram showing a method for monitoring the performance of a shooting device according to an embodiment. FIG. 2 shows (a) an image of a pattern formed on a wafer at an optimal focus position and (b) an image of a pattern formed on a wafer at a defocus position, collected using a CD-SEM, which is a photographic device according to one embodiment. FIG. 3 is a pixel histogram calculated from (a) an image to be analyzed collected by capturing a pattern formed on a wafer at an optimal focus position and (b) an image to be analyzed collected by capturing a pattern formed on a wafer at a defocus position using a CD-SEM, which is a shooting device according to one embodiment. FIG. 4 is a graph in which a plurality of images to be analyzed are collected using two shooting devices according to one embodiment, DW values ​​are calculated from the collected images to be analyzed, and the calculated DW values ​​are displayed by shooting process execution date. FIG. 5 is a schematic diagram showing a semiconductor exposure apparatus according to one embodiment. Figure 6a is a schematic diagram showing the state where the laser light and the lens are in optimal focus (f=0). FIG. 6b is a schematic diagram showing the state where the laser light and the lens are in negative defocus (f<0). FIG. 6c is a schematic diagram showing the state where the laser light and the lens are in positive defocus (f>0). FIG. 7a is a CD-SEM pattern image of a pattern formed by performing a patterning process while the light source of a semiconductor exposure device is in the optimal focus position according to one embodiment. FIG. 7b is a CD-SEM pattern image of a pattern formed by performing a patterning process while the light source of a semiconductor exposure device is in a defocus position according to one embodiment. FIG. 8 is a process diagram showing an integrated monitoring method for semiconductor patterning and measurement processes according to an embodiment. FIG. 9 is an image showing an example of a pixel histogram for each focus position calculated from the collected pattern image, wherein a pattern having the same shape and critical dimension scale is formed at a total of 11 focus positions (focus1 to focus11) according to one embodiment, and then the pattern image is collected. Figure 10 is an image showing the DW curve and approximate correction curve derived from the pixel histogram for each focus position of the pattern generated at 11 different focus positions as in Figure 9. FIG. 11a is a state diagram showing the state in which a window is created on a pattern image of a pattern to be analyzed according to one embodiment. FIG. 11b is a state diagram showing the state of comparing similarity by creating a window in a pattern image of a pattern to be analyzed using the SSIM technique according to one embodiment. FIG. 12 is the result of predicting the time of an anomaly during the process of forming a pattern on a wafer by repeatedly performing an exposure process using two different semiconductor exposure equipment according to one embodiment. FIG. 13 is the result of predicting the time of an abnormality occurring during the measurement process by repeatedly performing the shooting process using two different shooting devices according to one embodiment. FIG. 14 is an example of a pattern image collected from a pattern including five sub-patterns with different shapes and critical dimensions according to one embodiment. FIG. 15 is a focus graph calculated by forming five sub-patterns with different shapes and critical dimensions at a total of 11 focus positions according to one embodiment, and then deriving a histogram scatter function for each of the five sub-patterns. Figure 16 is an approximate correction graph of the focus-by-focus graph of Figure 14. Figure 17 is the result of calculating the derivative at an arbitrary focus position (focus4) of the approximate correction graph of Figure 16. FIG. 18 is a focus graph for the entire pattern set including five sub-patterns calculated by weighted averaging the derivatives of FIG. 16, and an approximate correction graph thereof. FIG. 19 is a configuration diagram showing an integrated monitoring system for semiconductor patterning and measurement processes according to an embodiment. Specific details for implementing the invention

[0016] Generally, imaging devices such as cameras, image sensors, and electron microscopes perform the imaging process by adjusting the focus to collect a clear image of the subject. In this imaging process, if shooting is performed in a defocused state due to an unsuitable focal length of the imaging device, the sharpness of the captured image decreases depending on the degree of defocus. Accordingly, in measurement methods that collect images to analyze a subject, it is a crucial factor to check the degree of defocus of the imaging device and adjust it to maintain optimal focus in order to collect a clear image.

[0017] In particular, for measurement equipment operated on semiconductor production lines, the accuracy, repeatability, and reproducibility of the data measured by the equipment are considered critical factors in quality control due to the precision of the semiconductor process. Therefore, to ensure accuracy, repeatability, and reproducibility, it is essential to acquire stable and clear images of the target for analysis and to employ appropriate image processing techniques.

[0018] In addition, to improve the productivity and yield of semiconductor devices, the acquisition and measurement of target images must be performed in a short time.

[0019] The performance monitoring method of a shooting device according to an embodiment aims to provide a performance monitoring method of a shooting device capable of accurately determining whether an abnormality occurs due to a change in focus of the shooting device.

[0020] Hereinafter, a method for monitoring the performance of a shooting device according to an embodiment will be described in detail.

[0021] FIG. 1 is a process diagram showing a method for monitoring the performance of a shooting device according to an embodiment.

[0022] Referring to FIG. 1, a method for monitoring the performance of a shooting device according to an embodiment may include the steps of collecting an image to be analyzed (S110), deriving a pixel histogram of the image to be analyzed (S130), calculating a DW value of the image to be analyzed (S150), and evaluating whether an abnormality has occurred in the shooting device (S170).

[0023] First, in the step of collecting the image to be analyzed (S110), the image to be analyzed is collected by photographing the object to be analyzed using a shooting device.

[0024] The above-mentioned imaging device may be a conventional imaging device of various forms used to capture an object and collect an image, such as a camera, an image sensor, or an electron microscope. In particular, the above-mentioned imaging device may be used to capture a semiconductor pattern and collect a pattern image. The image to be analyzed may be a pattern image collected by capturing a pattern formed on a wafer.

[0025] For example, the above imaging device may be a scanning electron microscope (SEM).

[0026] Based on the advantages of having very low resolution (10 to 0.5 nm) and the ability to achieve deep penetration depths (0.1 to 1 mm), the above-mentioned SEM is widely used as a measurement device in the semiconductor business sector where high integration and high aspect ratio processing are performed. The above-mentioned SEM can be described as equipment specialized for monitoring semiconductor manufacturing processes and inspecting defects in semiconductor chips. The above-mentioned SEM can collect a three-dimensional image of the object to be analyzed by scanning accelerated electrons at high speed onto the object to be analyzed, detecting and amplifying secondary electrons generated from uneven areas. In particular, the above-mentioned imaging device may be a CD-SEM (Critical Dimension Scanning Electron Microscope), and CD-SEM is mainly utilized to analyze fine patterns formed on the wafer surface.

[0027] The above CD-SEM can adjust the image resolution of the target image according to the size of the electron beam reaching the target. The CD-SEM can increase image resolution by reducing the size of the electron beam reaching the target. Therefore, in order to obtain a clear image in the CD-SEM, it is necessary to minimize the size of the electron beam irradiated onto the target to maintain appropriate image resolution.

[0028] Conventionally, to control the image resolution of a CD-SEM—that is, the focus of the CD-SEM—methods such as extracting edge values ​​from the image under analysis or correcting astigmatism in the frequency domain have been utilized. However, the method of extracting edge values ​​requires determining the components and directions of the edges, and since the operation is inaccurate in areas where edges are not identified, it is only possible to operate in areas with a narrow focus. Additionally, the method of correcting astigmatism in the frequency domain has limitations in resolution control because it is difficult to predict the accurate direction.

[0029] FIG. 2 shows (a) an image of a pattern formed on a wafer at an optimal focus position and (b) an image of a pattern formed on a wafer at a defocus position, collected using a CD-SEM, which is a photographic device according to one embodiment.

[0030] Referring to FIG. 2(a), the image to be analyzed collected by the imaging device in an optimal focus state exhibits the characteristic that the pattern is clearly displayed. On the other hand, referring to FIG. 2(b), the image to be analyzed collected in an inappropriate focus position, i.e., in a defocus state, exhibits the characteristic that the pattern formed on the wafer is blurry, there is a lot of noise in the image, and the entire image to be analyzed is either very bright or, conversely, very dark compared to the image collected in the optimal focus state. Accordingly, in the image to be analyzed collected in the defocus position, it is difficult to clearly determine whether a defect has occurred because the image is blurry.

[0031] Therefore, the resolution of the image subject to analysis may vary depending on the focus position of the imaging device. In monitoring the performance of the imaging device, it is important to accurately evaluate whether an abnormality has occurred due to changes in the focus of the imaging device. To this end, in the method for monitoring the performance of an imaging device according to the embodiment, the performance of the imaging device can be monitored by collecting an image subject to analysis using the imaging device and calculating the DW value of the collected image subject to analysis. The DW value and performance monitoring will be explained in more detail below.

[0032] The above-mentioned object of analysis may be used to generate an image of the object of analysis using a commercial imaging device, and to check for the occurrence of abnormalities, such as defects, through the generated image of the object of analysis.

[0033] Specifically, the object of analysis may be a semiconductor. In particular, the object of analysis may include a semiconductor pattern formed through a semiconductor photolithography process.

[0034] Next, in the step of deriving a pixel histogram of the image to be analyzed (S1130), the pixel histogram of the image to be analyzed is derived using the image to be analyzed.

[0035] Referring again to FIG. 2, when an image to be analyzed is collected at a suitable focus position using the above-described imaging device, a clear image to be analyzed can be collected as shown in FIG. 2(a). On the other hand, when an image to be analyzed is collected at an unsuitable defocus position, a relatively blurry, low-resolution image to be analyzed is collected as shown in FIG. 2(b). The pixel brightness values ​​of such images to be analyzed appear differently.

[0036] In this step, a pixel histogram of the image to be analyzed is derived. The pixel histogram can be generated by subdividing the image to be analyzed into pixel units and converting the brightness values ​​of the subdivided pixels into a statistical histogram.

[0037] Specifically, first, the collected image to be analyzed is divided into pixel units. The pixel is the smallest unit into which the image is divided. Next, a pixel histogram of the image to be analyzed is derived based on the brightness values ​​of the divided pixels. Each divided pixel has a specific brightness value, and the brightness value may have a value between 0 and 255.

[0038] FIG. 3 is a pixel histogram calculated from (a) an image to be analyzed collected by capturing a pattern formed on a wafer at an optimal focus position and (b) an image to be analyzed collected by capturing a pattern formed on a wafer at a defocus position using a CD-SEM, which is a shooting device according to one embodiment.

[0039] Referring to FIG. 3, a pixel histogram for a single image to be analyzed can be calculated using the brightness values ​​of pixels displayed in the image to be analyzed and the number of pixels for each brightness value. In the pixel histogram calculated from the image to be analyzed, the x-axis represents the brightness value of each pixel included in the image to be analyzed, and the y-axis represents the number of pixels included in the image to be analyzed. The pixel histogram can be generated by counting the number of pixels having a specific brightness value in the image to be analyzed, and from the pixel histogram, the number of pixels for each brightness value can be verified and the pixel dispersion can be verified. For example, an image to be analyzed collected at a specific focus position of a shooting device can be divided into 512 pixels horizontally and vertically, respectively. Then, by generating a pixel histogram for the collected image to be analyzed, it is possible to verify how pixels of a specific value are distributed in the image to be analyzed.

[0040] Accordingly, pixel histograms can be generated from high-resolution images and low-resolution images, respectively, and the generated pixel histograms can exhibit different pixel histogram dispersion functions.

[0041] The pixel histogram described above can represent the statistical characteristics of the object of analysis identified in the image of analysis. That is, a pixel histogram for a single image of analysis can be derived using the brightness values ​​of the pixels displayed in the image of analysis and the number of pixels corresponding to each brightness value. Furthermore, a single DW value representing the characteristics of the image of analysis can be calculated using the pixel histogram.

[0042] For reference, the image and pixel histogram to be analyzed shown in FIGS. 2 and FIGS. 3 are merely examples, and various shapes can be represented without being limited to the image and pixel histogram.

[0043] Next, in the step (S150) of calculating the DW value (Dark or White value) of the image to be analyzed using the pixel histogram, the DW value can be calculated using the pixel histogram generated for each image to be analyzed. The DW value refers to the standard deviation (SD) of the histogram of the image to be analyzed. The term DW value is not a term currently in use, but a term coined to explain the performance monitoring method of a shooting device according to one embodiment.

[0044] The above DW value refers to the Standard Deviation (SD) of a pixel histogram when the pattern image is subdivided into pixel units, the number of pixels having a specific brightness value is calculated, and the brightness at a pixel location is represented as a statistical pixel histogram. The above DW value can be generated individually for each image under analysis.

[0045] Since the pixel histograms above were calculated from images under analysis collected at different focus positions, the images under analysis each have different pixel histogram dispersion functions and DW values. The DW value may refer to the average brightness value that explains the variation in the images under analysis.

[0046] The resolution of the image under analysis is a critical factor in determining the quality of the measurement under analysis. An image under analysis collected at the optimal focus position exhibits a sharp image, whereas an image under analysis collected at the defocus position exhibits a blurry image.

[0047] Accordingly, pixel histograms generated from low-resolution target images collected at non-optimal focus locations tend to have lower heights and greater dispersion compared to pixel histograms generated from high-resolution target images. The aforementioned dispersion refers to the degree of distribution of pixel brightness values.

[0048] Next, in the step (S170) of evaluating whether an abnormality has occurred in the imaging device, the DW value of the image to be analyzed can be compared with reference information to evaluate whether an abnormality has occurred in the imaging device.

[0049] The above reference information may include a reference DW value derived using a pixel histogram calculated from an image to be analyzed collected at an optimal focus position.

[0050] In this step, the occurrence of an abnormality in the imaging device can be evaluated by comparing the DW value of the image to be analyzed with the reference DW value of the reference information. If the deviation resulting from the comparison between the DW value of the image to be analyzed and the reference DW value exceeds an allowable threshold, it can be determined that an abnormality has occurred in the performance of the imaging device. If the deviation resulting from the comparison between the DW value and the reference DW value is within the allowable threshold, it can be determined that the imaging device is maintaining appropriate performance.

[0051] The above reference information may include a reference pixel histogram, a reference DW value, an upper DW value, and a lower DW value. The reference DW value may be calculated using an image to be analyzed collected at the optimal focus position of the imaging device. The reference information collects images to be analyzed at multiple focus positions and selects the image with the highest resolution among the collected images to be analyzed as the reference image. Then, the reference DW value can be calculated by calculating a pixel histogram using the selected reference image and deriving a DW value from the calculated pixel histogram. The upper DW value and the lower DW value may each be calculated from an image to be analyzed that has an acceptable DW value deviation among the images to be analyzed collected using the imaging device.

[0052] FIG. 4 is a graph in which a plurality of images to be analyzed are collected using two shooting devices according to one embodiment, DW values ​​are calculated from the collected images to be analyzed, and the calculated DW values ​​are displayed by shooting process execution date.

[0053] Referring to FIG. 4, in this step, the DW value calculated from the image to be analyzed is compared with reference information, and if the DW value of the image to be analyzed is located between the upper DW value and the lower DW value, it can be determined that the shooting device is operating normally and performing the shooting process. On the other hand, if the DW value calculated from the image to be analyzed is compared with reference information and the DW value of the image to be analyzed exceeds the upper DW value or falls short of the lower DW value, it can be determined that the shooting device is operating abnormally due to a malfunction.

[0054] The above reference information refers to information collected at the optimal focus position, and the generation of the above reference information will be explained in more detail below.

[0055] The method for monitoring the performance of a shooting device according to the embodiment described above can monitor the performance of the shooting device by a simple method of deriving a pixel histogram from an analysis target image collected by shooting an analysis target using the shooting device, calculating a DW value from the derived pixel histogram, and comparing the calculated DW value with reference information. In addition, the focus position of the shooting device can be verified as an accurate numerical value, the DW value, allowing for intuitive verification of the performance of the shooting device.

[0056] In the method for monitoring the performance of a shooting device according to an embodiment, the reference information of the shooting device is obtained by capturing a plurality of analysis targets while adjusting the focus position of the shooting device and collecting analysis target images for each focus position. A pixel histogram and a DW value are derived for each analysis target image for each focus position, and a DW curve is generated. An inflection point is identified in the generated DW curve to select a reference analysis target image, and a DW value and a pixel histogram are extracted from the selected reference analysis target image to generate reference information.

[0057] However, the reference information of the above-mentioned imaging device is not necessarily limited to the description above, and can also be calculated using the same method as the method for collecting reference information of the semiconductor exposure device in the integrated monitoring method to be described later.

[0058] Meanwhile, the integrated monitoring method for semiconductor patterning and measurement processes according to the embodiments will be examined in detail below.

[0059] First, we will explain in detail the semiconductor patterning process, which involves forming patterns on a wafer using a semiconductor lithography device.

[0060] The above semiconductor patterning process refers to a process of forming a pattern on a wafer. The above semiconductor patterning process can be performed using a semiconductor exposure device.

[0061] Referring to FIG. 5, the semiconductor exposure device (210) may have a structure comprising a light source (211) that generates light for performing a patterning process, a lens (212, 212') that controls the focus of the light, a mask (213) on which a transfer pattern is formed, and a stage (214) that forms a structure on which the wafer (W) is placed and can move. A semiconductor exposure device having such a structure can form a pattern on a wafer (W) by irradiating the light source (211) with a pattern formed on the mask (213) and transferring it to the wafer (W). The semiconductor exposure device controls the focus by adjusting the focal distance between the light source (211) and the lens (212, 212'). In a semiconductor patterning process performed by irradiating light as described above, the focus of the light source (211) is important, and if the light source (211) performs the semiconductor patterning process at an optimal focus position, a fine-sized pattern can be accurately transferred onto the wafer (W).

[0062] Specifically, referring to FIG. 6a, at the optimal focus position, the distance between the laser light, which is the light source (211), and the lens is suitable, so the focus is placed on the surface of the wafer (W). Referring to FIG. 6b, at the negative defocus position, the distance between the laser light and the lens is unsuitable, so the focus is placed on the inside or bottom of the wafer (W). Referring to FIG. 6c, at the positive defocus position, the distance between the laser light and the lens is unsuitable, so the focus is placed on the top of the wafer (W). Depending on the focus position of the light source (211), the clarity and shape of the pattern transferred to the wafer (W) through the patterning process may differ.

[0063] Referring to FIG. 7a, a pattern image collected by forming a pattern at an optimal focus position can be seen to show a deep, clear, and low-noise pattern on the wafer (W).

[0064] Specifically, when a semiconductor pattern is formed by performing an exposure process at an optimal focus position, light and energy are concentrated at that optimal focus position. Consequently, the pattern image collected at that focus position can be observed to have deep, sharp characteristics with low noise in the surroundings.

[0065] In contrast, referring to Fig. 7b, in the pattern image collected by forming a pattern at the defocus position, a pattern with unclear lines and shapes and a lot of noise can be observed.

[0066] Specifically, when a semiconductor pattern is formed by performing an exposure process at a defocus position unsuitable for the semiconductor photolithography process, light and energy are dispersed. Consequently, in pattern images collected at the defocus position, the pattern is not accurately transferred, resulting in damaged or indistinct sub-patterns, noise, and a blurry pattern.

[0067] As described above, depending on the focus position of the light source (211) for pattern formation, the clarity, depth, noise, and shape (whether damage has occurred) of the pattern confirmed in the pattern image are different.

[0068] Therefore, by collecting information regarding the image level for each pattern image, the process suitability level or the optimal focus position of the light source can be identified. However, as shown in FIGS. 7a and 7b, it is difficult to visually confirm very accurately how the difference between the pattern images of the patterns generated at the optimal focus position and the defocus position appears.

[0069] In addition, there is a disadvantage in that, previously, there was no method to convert pattern images collected from semiconductor patterns into objective numerical values, making it difficult to accurately verify the performance of the device and process numerically after performing semiconductor photolithography and measurement processes.

[0070] For example, conventionally, a method is used to evaluate the suitability of a semiconductor lithography process by measuring the length of a specific part, or the critical dimension (CD), in a pattern image collected after performing the semiconductor lithography process, and comparing the measured critical dimensions.

[0071] However, the method of measuring critical dimensions has a problem in that the focus position of the semiconductor exposure device is inaccurate because the critical dimension is measured identically even when the focus position of the light source changes. In addition, since the measurement quantity relies on sampling inspection at the level of 5% of the total semiconductor production volume, it is difficult to identify trends in product production, and there is a disadvantage that it is impossible to predict the cause factors or process in the event of a problem.

[0072] Furthermore, in semiconductor measurement processes, the process is performed under the assumption that the conditions of imaging devices, such as CD-SEMs, are always in an optimal state. Although multiple imaging devices may be utilized in the aforementioned measurement process, there is a problem in that it is difficult to guarantee process defects and uniformity because pattern images are collected by arbitrarily deriving optimal conditions without considering differences in device standards or process conditions. Additionally, if a malfunction occurs in the imaging device used in the semiconductor lithography process, it is difficult to identify the problem in a short period of time, resulting in wasted costs and time.

[0073] In the integrated monitoring method for semiconductor patterning and measurement processes according to the embodiment, a method is provided that can calculate the optimal focus position of each semiconductor exposure device and imaging device by utilizing a pixel histogram, DW value, and DW curve, which are statistical methods, and can accurately monitor whether an abnormality occurs in said devices.

[0074] FIG. 8 is a process diagram showing an integrated monitoring method for semiconductor patterning and measurement processes according to an embodiment.

[0075] Referring to FIG. 8, the integrated monitoring method for semiconductor patterning and measurement processes according to an embodiment may include the steps of: forming a pattern on a wafer (S210); collecting a pattern image (S230); calculating a DW value of the pattern image (S250); and evaluating whether an abnormality has occurred in the semiconductor exposure device and the imaging device (S270).

[0076] First, in the step of forming a pattern on a wafer (S210), a photolithography process is performed using a semiconductor photolithography device to form a pattern on the wafer.

[0077] In this step, light is generated using a semiconductor exposure device, and a pattern can be formed on a wafer by performing a photolithography process using the generated light.

[0078] The above semiconductor exposure device can be implemented using various conventional types of semiconductor exposure devices that are utilized to form a semiconductor pattern on a wafer by performing an exposure process using a light source.

[0079] In this step, a pattern including multiple unit patterns of the same size can be formed on the wafer. In addition, in this step, a pattern including multiple unit patterns of different sizes and shapes can be formed on the wafer.

[0080] The light source (211) may be a laser light source. The laser light source (211) may include at least one of a KrF excimer laser light source, an ArF excimer laser light source, an ArFi laser light source, and an EUV laser light source.

[0081] The above laser light may be formed using various conventional forms of light sources used to generate light through light amplification by stimulated emission of radiation, regardless of wavelength and type. For example, representative examples of the above laser light sources include a KrF excimer laser light source (248 nm), an ArF excimer laser light source (193 nm), an ArFi laser light source (38 nm), and an EUV laser light source (13.5 nm).

[0082] Next, in the step of collecting pattern images (S230), a pattern image of a pattern to be analyzed among the patterns formed on the wafer (W) using the semiconductor exposure device can be collected.

[0083] In this step, pattern images of the pattern can be collected using various conventional imaging devices used to capture patterns in the photolithography process for manufacturing semiconductor devices. According to one embodiment, the imaging device may be a CD-SEM, and CD-SEM pattern images can be collected. In this step, multiple pattern images may be collected for a single pattern to be analyzed.

[0084] In this step, a pattern image of the pattern to be analyzed can be collected using a capturing device. At this time, the capturing device adjusts the focus to generate a high-resolution pattern image that allows the pattern to be analyzed to be clearly identified. If the pattern image is collected in a defocused state, where the focus of the capturing device is in an inappropriate position, the pattern to be analyzed is displayed blurry in the pattern image, resulting in a low-resolution pattern image and making accurate semiconductor measurement impossible.

[0085] Next, in the step of calculating the DW value of the pattern image (S250), a pixel histogram of the pattern image is derived, and a DW value for each pattern image can be calculated using the pixel histogram.

[0086] In this step, the DW value of the pattern image is obtained by dividing the pattern image into pixel units. Then, a pixel histogram is generated based on the brightness values ​​of the divided pixels, and the DW value can be calculated using the generated pixel histogram. The calculation of the DW value can be performed using the same method as the method for generating the DW value from the analysis target image described above, and a detailed explanation thereof will be omitted.

[0087] In the step (S270) of evaluating whether an abnormality has occurred in the semiconductor exposure device and the imaging device, the abnormality of the semiconductor exposure device and the imaging device can be evaluated by comparing either the pattern image or the DW value with reference information.

[0088] Reference information is important for evaluating whether an anomaly has occurred at this stage, so we will first explain the reference information in detail.

[0089] The above reference information may include information regarding a reference pattern image of a pattern formed by performing an exposure process with the semiconductor exposure device in optimal focus, and extracting a reference pixel histogram, a reference DW value, an upper DW value, and a lower DW value from the selected reference pattern image. The upper DW value and the lower DW value may be set by applying an acceptable range based on the reference DW value.

[0090] The above reference information may include a reference pixel histogram and a reference DW value calculated from the above reference pattern image. The above reference information may include an upper DW value and a lower DW value calculated based on the above reference DW value. The upper DW value and the lower DW value may selectively adjust the setting range as needed.

[0091] The above reference information can be calculated in the following way.

[0092] Specifically, the reference information can be calculated by a method comprising the steps of: performing a semiconductor exposure process at multiple focus positions to form a pattern on a wafer for each focus position; collecting pattern images for each pattern formed for each focus position; deriving a pixel histogram for each focus position using the pattern images for each focus position; calculating a DW value using the pixel histogram for each focus position; generating a DW curve using the DW value for each focus position; selecting an optimal focus position using the DW curve; and generating reference information including the selected optimal focus position information.

[0093] More specifically, first, a semiconductor exposure process is performed at multiple focus positions using the semiconductor exposure device to form patterns on a wafer, thereby preparing multiple samples, each having a pattern formed at a specific focus position. Next, the samples with patterns formed at each focus position, i.e., the objects to be analyzed, are photographed using an imaging device to collect pattern images at each focus position.

[0094] FIG. 9 is an image showing an example of a pixel histogram for each focus position calculated from the collected pattern image, wherein a pattern having the same shape and critical dimension scale is formed at a total of 11 focus positions (focus1 to focus11) according to one embodiment, and then the pattern image is collected.

[0095] Referring to Fig. 9, the collected pattern images by focus position are divided into pixel units to generate a pixel histogram by focus position.

[0096] Then, the DW value for each focus position is calculated from the generated pixel histogram for each focus position.

[0097] Next, the calculated DW values ​​for each focus position are converted into a quadratic function to generate a DW curve. Using the generated DW curve, the optimal focus position is determined, and the pattern image of the pattern generated at the optimal focus position can be set as reference information.

[0098] In other words, the semiconductor lithography device can adjust the focus position differently, and to calculate reference information, the semiconductor lithography process is performed for each focus position to form a pattern on the wafer. Then, pattern images for each focus position regarding the formed pattern are collected, and pixel histograms are generated using the collected pattern images for each focus position. Since the pixel distribution differs among the generated pixel histograms for each focus position, pixel histograms with different shapes may all be produced.

[0099] Figure 10 is an image showing the DW curve and approximate correction curve derived from the pixel histogram for each focus position of the pattern generated at 11 different focus positions as in Figure 9.

[0100] Referring to FIG. 10, the DW value for each focus position is calculated using the pixel histogram for each focus position. Since pixel histograms for each focus position were derived for a total of 11 focus positions to generate reference information, there are a total of 11 DW values ​​for each focus position, and the DW curve can be calculated by converting the DW values ​​for each focus position into a quadratic function.

[0101] The above DW curve can represent the level of overall physical interaction according to the change in the focus position of the light source equipped in the semiconductor exposure device. The inflection point of the above DW curve can be determined as the optimal focus position that can transfer the pattern most accurately and clearly to the target position in the semiconductor exposure process. In FIG. 10, the inflection point is identified as the sixth focus position, so the sixth focus position can be determined as the optimal focus position of the light source equipped in the semiconductor exposure device.

[0102] As described above, a pattern image selected as the optimal focus position can be extracted and set as reference information. The reference information may include the DW value of the pattern image as a reference DW value. Additionally, the reference information may generate an upper and lower DW value by arbitrarily specifying an acceptable range based on the reference DW value. The upper and lower DW values ​​can be derived by selecting a pattern image having an acceptable deviation from the pattern image for each focus position.

[0103] At this time, the focus position can be adjusted differently by considering the distance between the light source and the lens, and the focus position is not limited to 11, but can selectively divide the focus position as needed to collect pattern images for each focus position. By adjusting the focus position as described above to calculate a pixel histogram, the DW value, which is the dispersion function of the pattern image, can be derived for each.

[0104] Meanwhile, the integrated monitoring method according to the embodiment can simultaneously manage the semiconductor patterning process and the semiconductor measurement process. That is, it can simultaneously monitor whether an abnormality occurs in the semiconductor exposure device for forming a pattern on a wafer and the measurement device for evaluating whether a defect occurs in the formed semiconductor pattern.

[0105] To this end, in this step, the semiconductor exposure device and the imaging device can be monitored integrally through a method of evaluating whether an abnormality has occurred in the semiconductor exposure device, and then evaluating whether an abnormality has occurred in the imaging device if it is confirmed that no abnormality has occurred in the semiconductor exposure device.

[0106] First, in this step, the occurrence of abnormalities in the semiconductor exposure device is evaluated using the following method.

[0107] In this step, a similarity analysis is performed by comparing the reference pixel histogram extracted from the reference pattern image selected as the optimal focus position with the pixel histogram extracted from the pattern image of the pattern to be analyzed. If, as a result of the similarity analysis, the pixel histogram of the collected pattern image is determined to be similar to the reference pixel histogram, it is determined that there is no abnormality in the performance of the semiconductor exposure device.

[0108] Referring again to FIG. 9, the pixel histogram extracted from the pattern image may exhibit different shapes depending on the focus position, and if the sixth focus (focus 6) is determined to be the optimal focus position, the pattern image of the pattern generated at other focuses exhibits a pixel histogram different from the pattern image of the pattern generated at the sixth focus. In particular, when the distance deviation increases significantly at the sixth focus position, it becomes a defocus position, and when a pixel histogram is calculated from the pattern image of the pattern generated at the defocus position, the pixel histogram at the defocus position is significantly different in shape from the reference pixel histogram, and the center of the pixel histogram is positioned at a location that is farther away from the center of the reference pixel histogram.

[0109] Using this principle, in this step, if it is confirmed that the similarity between the reference pixel histogram and the pixel histogram extracted from the collected pattern image falls below a specific value, it is determined that an abnormality has occurred in the performance of the semiconductor exposure device.

[0110] Referring again to FIGS. 7a and 7b, the reference pixel histogram calculated from the reference pattern image (Fig. 7a) of the pattern formed by the semiconductor exposure device in the optimal focus position and the reference histogram calculated from the pattern image (Fig. 7b) of the pattern formed in the defocus position have different shapes, so when comparing similarities, the similarity deviation may be large, and in such cases, it can be determined that there is an abnormality in the performance of the semiconductor exposure device, that is, that it is in a defocus state. On the other hand, if the reference pixel histogram and the pixel histogram of the collected pattern image have an acceptable similarity deviation, it can be determined that there is no abnormality in the performance of the semiconductor exposure device.

[0111] The above similarity analysis utilizes a vector similarity-based method to quantify pattern images into a single vector, and can detect the level of anomaly occurrence based on the similarity of the vector's magnitude and direction.

[0112] The above vector similarity-based method is a method for comparing the similarity of the above pixel histogram. Representative examples of the above vector similarity-based method include cross-correlation distance, chi-square distance, intersection distance, Bhattacharyya distance, and cosine distance. The pixel histogram generated from the above pattern image is a visualization of a frequency distribution table. In the above pixel histogram, the horizontal axis represents the class, and the vertical axis represents the frequency.

[0113] In this step, the pixel histogram of the pattern image of the pattern to be analyzed is compared with the pixel histogram of the reference pattern image or the pattern image before anomaly occurrence (pre), and a similarity vector for the pattern image of the pattern to be analyzed can be calculated.

[0114] More specifically, the cross-correlation distance can measure how similar the distributions of two histograms are through the inner product. The cross-correlation distance has a value of 1 when the two histograms match perfectly, -1 when they do not match perfectly, and 0 when there is no correlation. In other words, the closer the cross-correlation distance is to 1, the more similar the two pixel histograms are, and the closer it is to -1, the more different the distributions of the two pixel histograms are.

[0115] The chi-square distance is a method for measuring the probability that a distribution of a new pixel histogram will appear relative to a reference pixel histogram among two pixel histograms. The chi-square distance has a value from 0 to ∞, and the closer the two distributions are to 0, the closer the value is to 0. In other words, the chi-square distance indicates the probability that a pixel histogram vector calculated based on a reference pixel histogram vector among two pixel histograms will appear. Therefore, the closer the chi-square distance is to 0, the more similar the two pixel histograms are, and the further the value is from 0, the more different the distributions of the two pixel histograms are.

[0116] The above intersection distance is a method for calculating the area where two pixel histograms overlap. The above intersection distance has a value from 0 to 1, and the closer the distribution of the two pixel histograms is to 1, the closer the value is to 1. That is, the closer the above intersection distance is to 1, the more similar the two pixel histograms are, and the closer the value is to 0, the more different the distributions of the two pixel histograms are.

[0117] The Bhattachaya distance mentioned above is a method for measuring the distance similarity of the probability distributions of two pixel histogram vectors. It calculates the distribution similarity of the pixel histogram vectors by dividing the two pixel histogram vectors by the total size of the histograms so that the sum becomes a probability distribution of 1. It is calculated using a formula that adds up the sizes of the overlapping areas of the bins of the two histogram vectors and then subtracts 1 from the sum. A value closer to 1 indicates a smaller overlapping area, while a value closer to 0 indicates a larger overlapping area, meaning that the data of the two pixel histogram vectors are similar.

[0118] The above cosine distance is a method for detecting similarity by dividing the inner product of two pixel histogram vectors by the product of the sizes of the two pixel histograms. The above cosine distance has a range value from -1 to 1; the closer the two pixel histogram vectors are to each other, the closer the value is to 1; if they are completely mismatched, the value is -1; and if there is no correlation, the value is 0. In other words, the closer the above cosine distance is to 1, the more similar the two pixel histograms are, and the closer it is to -1, the more different the distributions of the two pixel histograms are.

[0119] In this step, a vector similarity-based method is utilized to quantify the pattern image into a single vector. Specifically, by employing the aforementioned vector similarity-based method, the vector similarity between the calculated pixel histogram and a reference pixel histogram vector is determined, thereby identifying outliers and assessing the level of anomaly occurrence. Furthermore, this step compares the magnitude of the vector with the cosine angle. If it is confirmed that the magnitude and cosine angle of the vector are completely identical, the images are determined to be the same. This vector similarity-based method allows for the quantification and comparison of similarity between a reference pixel histogram and an analysis target pixel histogram in a short period of time, even with limited resources.

[0120] Accordingly, in this step, the level of abnormality occurrence can be derived through the similarity comparison as described above, and through this, it becomes possible to verify whether there is an abnormality in the semiconductor lithography device.

[0121] Furthermore, when utilizing the aforementioned DW curve during the actual semiconductor lithography process, it is possible to observe the extent to which the focus of the light source for each pattern formation deviates from the reference optimal focus position. That is, by performing the semiconductor lithography process to form a pattern and collecting pattern images, calculating a DW value from the collected pattern images, and comparing the DW value with the inflection point of the DW curve, it is possible to verify the extent to which the patterning process performed using the semiconductor lithography device deviates from the reference optimal focus position. Accordingly, utilizing the aforementioned DW curve allows for tracking the focus position of the light source during the lithography process using the semiconductor lithography device.

[0122] For reference, in the semiconductor photolithography process, while it is important for the pattern formed on the wafer to be formed in an accurate shape, concentrating light at a single focal point to enhance energy transfer and ensure the pattern appears deep and sharp is also a critical factor in determining quality.

[0123] Specifically, when a pattern is formed at the optimal focus position, the pattern image at the optimal focus position generates patterns accurately and clearly only at the intended locations. When a pattern is formed at a non-optimal focus, the energy of the laser light is dispersed, resulting in shallow and blurry patterns being generated over a wide area outside the intended locations. Consequently, in the pattern image collected from a pattern formed at a non-optimal focus position, the pixel distribution at the intended locations appears relatively lower compared to the pixel histogram generated from the pattern image at the optimal focus position, while the pixel distribution at locations where a pattern should not be formed tends to increase relatively. As the defocus becomes more severe while moving further away from the optimal focus position, the molecular structure collapse effect decreases, leading to a tendency for the number of dark pixels to decrease. Accordingly, from the pattern image where defocus becomes increasingly severe from the optimal focus position, a pixel histogram is derived that moves from left to right, gradually decreasing in height or curvature depth and increasing in dispersion.

[0124] Accordingly, the DW curve generated from pattern images collected from patterns formed at non-optimal focus locations tends to exhibit low height and large dispersion. The aforementioned dispersion refers to the degree of distribution of pixel brightness values.

[0125] Meanwhile, in this step, if it is confirmed that there is no abnormality in the performance of the semiconductor lithography device based on the similarity analysis results, an analysis is performed to determine whether there is an abnormality in the imaging device of the image to be analyzed.

[0126] In this step, whether an abnormality has occurred in the performance of the imaging device can be evaluated by comparing the DW value calculated from the collected pattern image with reference information.

[0127] Specifically, the calculation of the DW value of a pattern image collected from a pattern formed using the aforementioned semiconductor exposure device may, from the perspective of the imaging device, mean the focus of the imaging device for verifying the patterning result in the measurement process after performing the semiconductor exposure process.

[0128] Referring again to FIG. 2b, if the image to be analyzed, i.e., the pattern image, is collected in a defocused state where the focus of the imaging device is not appropriate, a blurry, low-resolution image to be analyzed is collected from the object to be analyzed. In particular, in a defocused state where the focus of the imaging device does not exactly match the phase of the object to be analyzed, the entire image to be analyzed is blurry and noisy, and the entire image to be analyzed becomes very bright or, conversely, very dark. When a pixel histogram is calculated from the image to be analyzed and a DW value is derived, the DW value of the image to be analyzed collected in a defocused state is different from the DW value of the image to be analyzed collected in an optimal focus state.

[0129] Conversely, referring to FIG. 2a, when a pattern image is collected with the focus of the imaging device set correctly, a clear, high-resolution image of the target for analysis is collected.

[0130] Referring again to FIG. 4, if the DW value calculated from the image to be analyzed is within the allowable deviation from the reference DW value, the DW value calculated from the image to be analyzed may be located between the upper DW value and the lower DW value.

[0131] Conversely, if the DW value calculated from the image under analysis exceeds the reference DW value and the allowable deviation, that is, if the DW value calculated from the image under analysis exceeds the upper limit DW value or falls short of the lower limit DW value, it may result in a outcome where the calculated DW value exceeds the upper limit DW value or falls short of the lower limit DW value.

[0132] Therefore, in this step, the calculated DW value is compared with the reference DW value, the upper DW value, and the lower DW value. If it is confirmed that the calculated DW value lies between the upper and lower DW values, it is determined to be within the range of variation within the group, and the imaging device is determined to be operating normally. In other words, while the focus of the imaging device varies within a certain range from optimal conditions, only the focus of the exposure device for forming the semiconductor pattern—or other auxiliary devices, imaging conditions, or the image to be analyzed being a semiconductor pattern—varies within a normal range. Conversely, if the DW value calculated from the image to be analyzed exceeds the allowable deviation from the reference DW value—that is, if the DW value calculated from the image to be analyzed deviates from the upper or lower DW value—it is confirmed to be between the range of variation, and it can be determined that an abnormality has occurred in the imaging device. In other words, because the focus of the imaging device is out of optimal condition, even if the focus of other auxiliary devices or exposure devices other than the imaging device actually fluctuates within a normal range and there are no abnormalities, the result of the DW value is recorded at a position significantly outside the optimal focus range of the existing imaging device.

[0133] When the DW value calculated as above indicates a level of variation between groups, the content of the image to be analyzed is noisy and difficult to accurately identify, and the entire image appears very bright or dark. If such a pattern of variation between groups is confirmed in the image to be analyzed, it can be determined that the imaging device is in a defocused state. The defocus of the imaging device can occur due to various causes in addition to the distance between the lens and the object to be analyzed.

[0134] For reference, the above within-group variation and between-group variation are each factors representing the degree of variation in process capability analysis.

[0135] For example, regarding the semiconductor lithography process in which semiconductor patterning is performed using an exposure device and semiconductor pattern images are collected and measured using a imaging device, the aforementioned within-group and between-group variations are explained in detail. The within-group variation refers to fluctuations caused by routine factors affecting the focal position of the exposure equipment, and is mostly due to causes that affect the performance of the patterning process. These within-group variations can be caused by numerous factors, such as the condition of the wafer stage, lens aberrations, vibration of the exposure device, and the height of the wafer surface. More specifically, since the imaging focus of the imaging device is optimized during the imaging process and the focus fluctuation of the patterning device is managed using the optimized imaging device, it can be considered that there is no influence from the imaging process itself. On the other hand, the aforementioned between-group variation refers to fluctuations caused by the process conditions of the imaging device and performance abnormalities of the imaging device itself. As this is a variation caused by abnormal causes, it significantly affects the average value of the process. When the above between-group variation occurs, the imaging device is repaired or the process conditions of the imaging process are corrected.

[0136] If, when collecting images to be analyzed using a shooting device, the focus fluctuation of the shooting device is not significant—that is, if the shooting process is performed at an appropriate focus—the DW values ​​of multiple images to be analyzed collected by the said shooting device will exhibit a dispersion within the range of a reference lower limit and a reference upper limit, where the deviation from the average is not large. On the other hand, if the DW value calculated after collecting images to be analyzed using the said shooting device exceeds either the reference lower limit or the reference upper limit, it can be determined that a malfunction has occurred in the shooting device because the shooting process was not performed at an appropriate focus.

[0137] Specifically, if the DW value collected from the image under analysis falls within the range of the reference lower and upper limits, it exhibits within-group variation behavior, and in such cases, it can be determined that the imaging device maintains an appropriate focus. Furthermore, in the case of a process where semiconductor patterning is performed using an exposure device and semiconductor pattern images are collected and measured using an imaging device, it can likewise be determined that the focus of both the imaging device and the exposure equipment is also at an appropriate level. On the other hand, if the DW value collected from the image under analysis deviates from the reference lower and upper limits, it exhibits between-group variation behavior, and in such cases, it can be determined that the imaging device is in an outlier state where it is defocused.

[0138] Meanwhile, in this step, if it is determined that an abnormality has occurred in the semiconductor exposure device, a step of deriving the location and level of the abnormality in the semiconductor exposure device may be further included. However, it is difficult to determine the location of the abnormality in a two-dimensional pattern image using only the vector similarity-based method.

[0139] Accordingly, in this step, the structural similarity between the image vector of the transformed pattern and the vector obtained by transforming the reference pattern image is evaluated. Then, the location of anomalies in the pattern to be analyzed can be detected.

[0140] The above anomaly location can be derived from a two-dimensional pattern image by utilizing the SSIM technique, which divides the pattern image to be analyzed into local ranges and allows for a detailed comparison of local similarity with a reference pattern image.

[0141] The above SSIM technique can evaluate the similarity between a reference pattern image and a pattern image to be analyzed by considering three factors such as the structure, brightness, and contrast of the pattern image. By mimicking the way human systems perceive images, the SSIM technique can evaluate the similarity of pattern images by comparing their structural characteristics rather than simply the difference in pixel values. The SSIM technique evaluates whether the texture, structure, and brightness of the pattern images are similar, and because it focuses on determining how similar the pattern images appear rather than on simple color accuracy, it can calculate the location of anomalies by precisely evaluating the qualitative aspects of the image.

[0142] FIG. 11a is a state diagram showing the state in which a window is created on a pattern image of a pattern to be analyzed according to one embodiment. FIG. 11b is a state diagram showing the state in which a window is created on a pattern image of a pattern to be analyzed using the SSIM technique to compare similarity according to one embodiment.

[0143] First, in this step, a window is created on the pattern image of the pattern to be analyzed as shown in FIG. 11a. Then, as shown in FIG. 11b, the similarity between the reference pattern image and the pattern image of the pattern to be analyzed is evaluated while moving the created window on the pattern image. FIG. 11a and FIG. 11b respectively show that a window is created only on the pattern image of the pattern to be analyzed, but in the SSIM technique, the window is also formed on the reference pattern image at the same location as the pattern image of the pattern to be analyzed, so that the areas where the window is created can be compared.

[0144] In this step, the texture, structural brightness, etc., of the pattern image of the pattern to be analyzed is evaluated using the method described above to determine whether they are similar to the reference pattern image. Accordingly, the location of anomalies in the pattern image of the pattern to be analyzed can be derived.

[0145] In the integrated monitoring method according to the embodiment, the level of abnormal occurrence is derived as described above, and the location of the abnormal occurrence is derived from the pattern image of the pattern to be analyzed, thereby providing a reliable analysis result to the manager regarding the focus abnormality of the light source equipped in the semiconductor exposure device in the semiconductor patterning process.

[0146] In addition, the level of abnormal occurrence in the semiconductor exposure device can also be detected by the following method.

[0147] Again, referring to FIG. 7a, in the case of the optimal pattern image for the pattern formed at the optimal focus position, the pattern is formed at the most accurate position, and the clearest and deepest pattern can be confirmed in the pattern image.

[0148] Referring to Fig. 7b, in the case of a pattern image formed at a defocus position that is excessively separated from the optimal focus position, the pattern is lost or the pattern is observed to have low clarity and shallow depth.

[0149] The characteristics of the pattern image described above may show a tendency for the defocus position to increase in proportion to the distance separated from the optimal focus position, and thus the level of anomaly occurrence can be calculated using these characteristics, and the level of anomaly occurrence can be calculated in the following way.

[0150] First, the DW value is calculated from the pattern image of the pattern under analysis. Then, the DW deviation is calculated by comparing it with the reference DW value used to derive reference information, thereby determining the level of anomaly occurrence.

[0151] If the pattern image of a pattern formed at an optimal focus position separated by a certain distance from the optimal focus position represents a DW value located between the optimal DW value and the upper DW value, or if the pattern image of a pattern formed at the optimal focus position represents a DW value located between the optimal DW value and the lower DW value, it can be confirmed through the pattern image that a pattern having a suitable shape and critical dimension scale has been formed, although the clarity and depth have decreased relatively compared to the optimal pattern image.

[0152] However, when a pattern is formed at a defocus position, a DW value exceeding the upper limit DW value or falling short of the lower limit DW value is calculated in the pattern image. By comparing the calculated DW value of the pattern image with the reference DW value, the DW deviation between them can be confirmed, and it can be determined that the level of anomaly occurrence increases as the DW deviation increases.

[0153] In this step, only the evaluation of the level of anomaly occurrence is described when it is determined that an anomaly has occurred, but the method is not limited thereto. That is, the above evaluation of the level of anomaly occurrence can be performed even when the DW value differs from the upper DW value and the reference DW value, or from the lower DW value and the reference DW value.

[0154] More specifically, in this step, if the DW value calculated from the pattern image is confirmed to be between the upper DW value and the reference DW value, or between the lower DW value and the reference DW value, it can be confirmed that the degree of defocus is relatively low depending on the deviation. Conversely, if the DW value calculated from the pattern image is detected in an area outside the upper or lower DW value, it can be confirmed that the degree of defocus is high, making it possible to detect the level of anomaly occurrence.

[0155] In addition, during the process of repeatedly forming multiple exposure processes and collecting pattern images in this step, if the calculated DW value is found to exceed an upper DW value or fall below a lower DW value, the corresponding pattern image can be checked to determine the point at which an anomaly occurs.

[0156] This step may further include a step of predicting the time of occurrence of an abnormality in the semiconductor lithography device.

[0157] FIG. 12 is the result of predicting the time of an anomaly during the process of forming a pattern on a wafer by repeatedly performing an exposure process using two different semiconductor exposure equipment according to one embodiment.

[0158] Referring to Fig. 12, it can be seen that when a photolithography process is repeatedly performed using a specific semiconductor photolithography device, the DW value fluctuates according to the date of the photolithography process. The fluctuation of the DW value may exhibit a certain trend. By utilizing such a trend, the time of an abnormality in the semiconductor photolithography device can be predicted.

[0159] For example, semiconductor lithography device 1 may have a DW value between an upper DW value and a lower DW value. It can be confirmed that the semiconductor lithography device 1 has a trend in which the DW value gradually decreases during the process of repeating the lithography process. In the case of lithography device 1 having such a trend, if the slope (Arrow 1) of the average DW value for each lithography process execution date is calculated, it is possible to predict the level of change reaching the lower DW value and the time of anomaly occurrence.

[0160] In addition, in the case of semiconductor lithography device 2, all DW values ​​exist within the upper and lower DW values, but it can be confirmed that the DW values ​​have a decreasing trend during the process of repeating the lithography process. In the case of lithography device 2 having such a trend, the slope (Arrow 2) of the average DW value for each day of the lithography process execution is greater than the slope (Arrow 1) of lithography device 1, and when the lithography process is repeated for the same amount of time, the point at which the lower DW value is reached may be faster compared to lithography device 1.

[0161] If a machine learning model is constructed based on the distribution of DW values ​​as described above, the machine learning model can predict the timing of an anomaly in a specific semiconductor lithography device by comparing the differences in performance levels of each lithography device using multiple semiconductor lithography devices or by checking the trend of DW values ​​of the semiconductor lithography device.

[0162] In this step, if it is determined that an abnormality has occurred in the semiconductor lithography device, a step of deriving the major abnormality factors that cause the abnormality in the semiconductor lithography device may be further included.

[0163] Specifically, in the photolithography process using the semiconductor photolithography device described above, multiple abnormality factors may be involved in the occurrence of abnormalities. In this case, the major abnormality factor refers to the abnormality factor that contributes most significantly among the multiple abnormality factors affecting the occurrence of abnormalities in the photolithography process.

[0164] In this step, a machine learning model can be generated to detect the cause of an anomaly in the semiconductor exposure device based on the setting conditions of the semiconductor exposure device for performing the exposure process and the DW value. The machine learning model can predict not only the cause of the anomaly but also the time of the anomaly, the level of the anomaly, etc., using the same method.

[0165] The machine learning model described above can predict the time and level of anomaly occurrence based on process and equipment conditions for performing the exposure process, and the direction, trend, and level of change of the DW value. The time of anomaly occurrence is calculated by deriving the trend direction and trend angle of the DW value from the change of the DW value over time, and the level of anomaly occurrence can be calculated using the trend angle.

[0166] The above machine learning model can identify anomaly occurrence factors related to the exposure equipment and the exposure process that are expected to have a significant impact on the change in the DW value by considering the correlation with the trend direction and trend angle of the DW value, and can provide results regarding the level of improvement of the changed DW value by controlling the identified anomaly occurrence factors through simulation. The above machine learning model will be explained in more detail below.

[0167] The above-mentioned abnormal occurrence factors may include exposure process conditions, operating conditions of the exposure device, etc. For example, the above-mentioned abnormal occurrence factors may include the focus position of the light source, the energy level of the light source, damage to the light source and lens installed in the semiconductor exposure device, the angle of incidence of the light source and the lens, contamination of the light source and the lens, the result of a wafer processing process performed prior to the exposure process, vibration of the exposure device, and conditions of the exposure process. The above-mentioned abnormal occurrence may be caused by any one of the above-mentioned abnormal occurrence factors, or it may be caused by multiple of the above-mentioned abnormal occurrence factors.

[0168] In this step, major anomaly factors that significantly influence anomaly occurrence can be derived from the pattern images of the pattern to be analyzed using a machine learning model. The major anomaly factors can be derived in the following manner.

[0169] In this step, major anomaly occurrence factors can be derived by a method comprising the step of extracting anomaly features from a pattern image of a pattern to be analyzed and the step of deriving major anomaly occurrence factors from the extracted anomaly features using a machine learning model.

[0170] First, in the step of extracting outlier features, outlier features are extracted from the pattern image determined to have an anomaly. The outlier features can be extracted using a feature extraction filter. The feature extraction filter can be configured to extract features related to pixel values ​​(color), the amount of light energy (illumination), noise, and the shape of the pattern from the pattern image under analysis through comparison with a reference pattern image.

[0171] Specifically, the feature extraction filter can extract pixel values, illuminance, and pattern shapes of the analysis target area from the analysis target pattern image. At this time, the feature extraction filter is used to better extract the features of anomalies that occur according to the type of anomaly occurrence. The feature extraction filter can compare the reference pattern image with the analysis target pattern image and extract differences in pixel values, illuminance, and pattern shapes as anomaly features. At this time, the types of anomalies may differ depending on the type of anomaly occurrence factor.

[0172] The above feature extraction filter extracts abnormal features such as pixel values, illuminance, and pattern shape by comparing the reference pattern image and the pattern image to be analyzed.

[0173] Next, in the step of deriving major anomaly occurrence factors, major anomaly occurrence factors can be derived from the above anomaly features extracted using a machine learning model.

[0174] The contribution of each factor to anomaly occurrence in the above machine learning model is based on the level to which the variation of each of the multiple anomaly occurrence factors (X-axis values) influences the change in the result factor, the DW value (Y-axis).

[0175] Specifically, if any of the aforementioned anomaly factors changes, the Y-axis value (DW) is affected and changes; the level of change in the DW value caused by the change in the anomaly factors can be referred to as correlation. If an anomaly factor with high correlation changes by a large amount, the DW value will change over a wide range. Conversely, in the case of an anomaly factor with low correlation, even if it changes by a large amount, the DW value will change within a small range. Based on the above logic, the machine learning model can present anomaly causes listed in priority from the most probable causes of anomaly to detect the true cause of anomaly that significantly changes the DW value to a level where an anomaly is determined to have occurred in the pattern image under analysis.

[0176] The machine learning model described above is not derived solely by utilizing the correlation between a single anomaly factor and a DW value. Instead, the model must consider noise-related issues, such as multicollinearity, where multiple anomaly factors interact with each other due to high correlations. The model can identify key anomaly factors by considering their priorities using mining techniques. By learning from big data, the model derives the contribution of each factor to the anomaly factors and can provide reliable results regarding the causes of anomalies based on feature importance.

[0177] After the aforementioned machine learning model is constructed, a separate component is established to collect pattern images of patterns determined to have anomalies and to gather information on the type of anomaly from the collected pattern images. Then, multiple anomaly factors associated with the aforementioned type of anomaly are learned using the machine learning model. Furthermore, the machine learning model learns from big data to derive the contribution of each factor of the anomaly factors. The machine learning model can provide reliable results regarding the cause of the anomaly by evaluating the analysis algorithm based on the feature importance of each factor. For analysis targets determined by the machine learning model to have anomalies, not only are the time of occurrence, level of occurrence, and major causative factors presented according to feature importance, but pattern images are also collected to analyze and learn the exact location and characteristics of the anomaly. Local areas are compared with existing normal pattern images, and the coordinates and level of the anomaly in those areas are learned and stored separately from the machine learning model. Alternatively, the performance of the model can be enhanced so that the machine learning model can prioritize the analysis of the relevant area first when examining new pattern images for anomalies in the future. The machine learning model described above acquires data consisting of all measurable process and equipment parameters in the photolithography process. Next, the machine learning model enhances the robustness and accuracy of the model by excluding from model construction any anomaly factors that are clearly unaffected by changes in the dependent variable, the DW value (Y-axis value). The machine learning model repeats this exclusion process to train a dataset composed of the finally selected anomaly factors and the DW values ​​of the pattern images. When predicting anomalies, the machine learning model utilizes an analysis algorithm to select the model with the best consistency (correlation with data, explanatory power).The above machine learning model refers to the process of building a model by applying machine learning methods to data corresponding to 70 to 80% of the total dataset. By repeating the process of evaluating and rebuilding the model's performance by inputting the test dataset corresponding to the remaining 20 to 30% into the ideal learning training model built through learning as described above, the performance of the machine learning model can be improved.

[0178] At this time, various methods such as Mean Squared Error (MSE) and Root Mean Squared Error (RMSE) can be used in combination to evaluate the performance of the model, and the evaluation metric with the best performance among them can be selected and continuously applied or changed. Through such a robust and highly reliable machine learning model, it is possible to manage and predict the timing and level of anomalies in DW values, and to reliably track the causes of these anomalies.

[0179] In addition, the machine learning model may utilize an analysis algorithm that includes at least one of Linear Regression, Lasso, Ridge, Decision Tree, Random Forest, GBM (Gradient Boost), LGBM (Light GBM), and XGB (XG Boost).

[0180] In addition, once sufficient data loading and model reliability are verified, it is possible to scale up to large-scale models such as artificial neural networks.

[0181] The performance of the machine learning model described above can be tested by comparing the predicted DW values ​​(Y-axis values) with the actual DW values ​​(Y-axis values) within the pattern image dataset. Various variables existing within the formulas of each learning model are referred to as hyperparameters; by tuning these hyperparameters, a machine learning model with an optimal function capable of effectively explaining the dataset can be derived. In this step, the derivation of such a learning model merely explains the method and principles of constructing the machine learning model, and the types, principles, and formulas of each learning model may vary. Accordingly, characteristics such as the learning model construction process, analysis performance based on the amount of data, and analysis time also vary. The learning model can be utilized selectively as needed and may be configured to continuously modify the model by reflecting external conditions.

[0182] In addition, the machine learning model may be configured to derive major anomaly occurrence factors using the pattern image, and then generate simulation information capable of predicting the level of change of major anomaly occurrence factors to reduce anomaly occurrences and the patterning results for each level of change of major anomaly occurrence factors, and provide this information to the manager.

[0183] In addition, the above machine learning model can also predict the time of an anomaly in a specific semiconductor lithography equipment during the process of performing a lithography process using multiple semiconductor lithography equipment.

[0184] In addition, the machine learning model may be configured to provide a simulation function that detects the major anomaly occurrence factors and suggests the level of improvement required to control process variables of the semiconductor lithography process associated with the major anomaly occurrence factors, or to directly perform automatic control. In this case, the major anomaly occurrence factors may be presented according to priority among multiple anomaly occurrence factors, considering the correlation with the impact on the occurrence of anomalies.

[0185] In addition, the step of controlling process variables of the semiconductor lithography device may further include a step of generating and providing patterning prediction information based on a DW value derived when a pattern is formed on the wafer by the semiconductor lithography device by controlling process variables of the semiconductor lithography device associated with the major abnormal occurrence factor.

[0186] In addition, this step may further include a step of generating a control signal to control process variables of an exposure process using the semiconductor exposure device when it is determined that an abnormality has occurred in the semiconductor exposure device.

[0187] In addition, the step of determining whether an anomaly has occurred in the above-mentioned analysis target pattern may further include the step of performing history management of DW values ​​and anomaly occurrence factors for each exposure process and exposure device, comparing and analyzing the difference between the anomaly occurrence factors and the level of anomaly occurrence, and outputting and providing an improvement method that can improve the occurrence of anomalies.

[0188] In addition, this step may be configured to further include a step of evaluating whether an abnormality has occurred in the imaging device, and then calculating the level of abnormality and the time of abnormality occurrence of the imaging device.

[0189] FIG. 13 is the result of predicting the time of an abnormality occurring during the measurement process by repeatedly performing the shooting process using two different shooting devices according to one embodiment.

[0190] Referring to Fig. 13, it can be seen that when a shooting process is repeatedly performed using a shooting device, the DW value fluctuates according to the date of the shooting process. The fluctuation of the DW value may exhibit a certain trend. By utilizing such a trend, the time of occurrence of an abnormality in the shooting device can be predicted.

[0191] Since the level of abnormal occurrence in the above-mentioned imaging device can be evaluated using the same method as the aforementioned method for evaluating the level of abnormal occurrence in the semiconductor exposure device by verifying the deviation between the reference DW value and the calculated DW value, a detailed explanation thereof will be omitted.

[0192] In addition, in this step, the time of an anomaly in the imaging device is extracted by comparing the DW value with reference information, and it is possible to evaluate whether a significant change in the focus of the imaging device occurred in the period close to the time of the anomaly. That is, candidate causes of problems, such as repairs or restarts, that affect the focus of the imaging device in the period adjacent to the time of the anomaly can be provided according to priority. Furthermore, the priority can be derived by calculating the correlation and contribution of causal factors affecting the focus change in the imaging device and the semiconductor lithography device, respectively, using a machine learning method.

[0193] In addition, by checking the distribution of DW values ​​at the point where the imaging device maintains optimal focus and the focus change over time, a focus change trend can be presented, and correction information necessary to adjust the focus position of the imaging device can be provided.

[0194] Furthermore, the distribution of DW values ​​at the point where the imaging device maintains optimal focus and the change in focus over time can be identified, and a focus prediction model can be generated by applying machine learning to this. The focus prediction model can predict in advance when significant changes in DW values ​​occur, thereby preventing process accidents caused by the imaging device and providing appropriate maintenance timing.

[0195] Information related to the occurrence of an abnormality in the above-mentioned imaging device can be calculated using the same method as the machine learning model described above.

[0196] Meanwhile, in this step, the Cpk (capability of process Katayori) index, which is a capability of process index for indicating the process capability of a semiconductor lithography device and an imaging device, can be calculated using a reference pixel histogram and a calculated pixel histogram. The Cpk index is a representative process capability index that simultaneously considers the dispersion of the data and the level of deviation from the center. In this step, the Cpk index is presented merely as an example of a process capability index; the system may also be configured to calculate various conventional indicators capable of indicating the process capability of the device in addition to the Cpk index.

[0197] Specifically, the aforementioned capability of process index refers to an indicator used to measure process capability and the level required to improve the process. This capability is an indicator that evaluates the future performance ability of a semiconductor manufacturing process. This performance ability is determined by how uniformly the process can be executed in relation to the requirements. The capability of process index is a numerical representation of this capability. The Cpk index numerically represents the ability of the process to perform relative to specification limits.

[0198] The above Cpk index can be calculated regardless of whether an abnormality occurs in the semiconductor lithography device and the imaging device.

[0199] The above Cpk index can be calculated using the following Equations 1 and 2.

[0200] [Equation 1]

[0201] CP = Allowable pixel histogram spread / Calculated pixel histogram spread

[0202] Cp = (Upper DW value - Lower DW value) / (6 * sigma)

[0203] [Equation 2]

[0204] Cpk = (1-k) × CP

[0205] k = abs(((UCL + LCL) / 2 - DW_mean) / (UCL - LCL) / 2)

[0206] However, in Equation 1, Sigma represents the value derived from the Standard Deviation of the DW Value distribution. Also, in Equation 2 above, UCL represents the upper limit DW value, LCL represents the lower limit DW value, and DW_mean represents the average of all DW values.

[0207] In Equation 1 above, CP (Capability of Process) refers to process capability. The above-mentioned allowable DW value refers to a preset deviation range from the reference DW value. That is, it refers to the upper limit SW value Su and the lower limit DW value Sl. The above-mentioned allowable DW value may refer to the DW value calculated from the pattern image of the pattern generated at the optimal focus position and the allowable range of focus positions. The above-mentioned allowable DW value can be selectively changed as needed and in accordance with changes in various conditions, such as process, equipment conditions, and pattern characteristics.

[0208] When the above CP is 1 or greater (CP=1, CP>1), it means that the overall DW values ​​exist stably within UCL and LCL. As CP becomes greater than 1, there are more DW values ​​that gradually shift toward UCL and LCL. In other words, while the focus of the imaging equipment is maintained normally at an optimal level, the focus of the exposure equipment is also fluctuating within a normal range.

[0209] However, 0 <CP<1인 경우 UCL, LCL을 넘는 DW값이 많아지므로, 반도체 노광 장치 혹은 촬영 장치 중 적어도 어느 하나에 이상이 발생한 것을 의미할 수 있다. 상기 Cp가 1.33 이상일 때 공정능력을 만족한다고 판단할 수 있다.

[0210] In the above Equation 2, k represents the bias factor constant.

[0211] For reference, the difference is that while the above Cp refers only to the distribution of DW values, the above Cpk is a more practical index as it can consider the extent to which the mean of the DW value distribution is skewed in a specific direction.

[0212] The above Cp and Cpk values ​​can have values ​​of 0.33 (1σ), 0.67 (2σ), 1 (3σ), 1.33 (4σ), 1.67 (5σ), and 2 (6σ). When the value is 1, the fluctuation of the DW value is at an average level, and when it is 1.33 or higher, the fluctuation range of the DW value is small, indicating a good level of DW value management. Specifically, it means that the center of the distribution of DW values ​​including the new DW value is skewed to the left or right from the center of the distribution of existing DW values ​​that do not include the new DW value. When Cpk > 1.33, it means that the center of the calculated DW value distribution is less skewed compared to the center of the acceptable DW value distribution and shows a tendency to be clustered at the center of the acceptable DW value distribution. In the case where Cpk > 1.67 above, the center of the calculated DW value distribution tends to cluster around the center of the reference DW value distribution, indicating that the semiconductor lithography and imaging devices are performing the process at the optimal focus. In other words, it means that the process capability is sufficient. The above 0 <Cpk<1.33인 경우 DW값 분포의 중심이 허용 가능한 DW값 분포의 중심에서 멀어져 치우침이 상대적으로 커 반도체 노광 장치 및 촬영 장치가 디포커스인 상태에서 공정을 수행할 가능성이 있음을 의미한다. 보다 구체적으로, DW값이 상기 UCL(상한 DW값)과 상기 LCL(하한 DW값) 안에 분포하는 경우에는 여전히 촬영 장치의 포커스가 최적이라 해도, 노광 장치의 특정 문제로 인해 패턴의 변화가 기존보다 적거나 커지는 등 포커스 변동이 일어났을 수도 있음을 시사하는 것일 수 있다. 반대로, DW값이 상기 UCL(상한 DW값)과 상기 LCL(하한 DW값) 안에 분포하지 않는 경우 촬영 장비의 포커스가 서서히 변동하고 있을 가능성이 있음을 시사할 수도 있다.

[0213] Accordingly, in this step, the distribution of DW values ​​calculated from the collected pattern images can be compared with reference information to generate and display a Cpk index, which is a process capability index, for the semiconductor lithography device and the imaging device. Furthermore, the integrated monitoring method for the semiconductor patterning and measurement process according to the embodiment can be configured to generate an alert signal and transmit it to an administrator when the Cpk index exceeds a preset management range.

[0214] In addition, the integrated monitoring method for semiconductor patterning and measurement processes according to the embodiment can identify the cause of an anomaly and provide it to the manager when the Cpk index deviates from a preset management range or when an outlier in the DW value occurs.

[0215] Meanwhile, the foregoing description assumes a case where a pattern including a single sub-pattern is formed on a wafer using a semiconductor lithography device. However, in a semiconductor patterning process, a pattern including multiple sub-patterns may also be formed on a wafer. The sub-pattern refers to a small unit pattern that forms the single pattern. That is, the pattern may include at least one sub-pattern.

[0216] Specifically, in the step of forming the above pattern, a pattern set including multiple sub-patterns with different shapes and critical dimensions may be formed on a wafer through a single semiconductor photolithography process.

[0217] FIG. 14 is an example of a pattern image collected from a pattern containing five sub-patterns with different shapes and critical dimensions according to one embodiment. FIG. 15 is a focus graph calculated by forming five sub-patterns with different shapes and critical dimensions at a total of 11 focus positions according to one embodiment, and then deriving a histogram dispersion function for each of the five sub-patterns. FIG. 16 is an approximate correction graph obtained by approximate correction of the focus-specific graph of FIG. 14. FIG. 17 is the result of calculating the derivative at an arbitrary focus position (focus4) of the approximate correction graph of FIG. 16. FIG. 18 is a focus graph for the entire pattern set containing five sub-patterns and an approximate correction graph thereof, calculated by weighted averaging the derivatives of FIG. 16.

[0218] Referring to FIGS. 14 to 18, when first to fifth sub-patterns with different shapes and critical dimensions are formed at a total of 11 focus positions, a pixel histogram for each of the first to fifth sub-patterns can be derived as a single diagram.

[0219] In addition, a focus graph for one sub-pattern is first generated using the DW values ​​per sub-pattern image for the first to fifth sub-patterns, i.e., the pixel histogram scatter function. Then, this generation of focus graphs is repeatedly performed for the remaining sub-patterns to generate multiple focus graphs for multiple sub-patterns and visualize them as a single diagram.

[0220] In addition, approximation processing is performed on each of the multiple focus graphs to generate multiple approximation processing graphs for the first to fifth sub-patterns, respectively.

[0221] Next, derivatives representing the slope of the pattern-specific function at each focus position (x-value) are derived from the focus positions and DW values ​​of all approximated pattern images. Through this process, it becomes possible to verify the level of change in the individual pixel histogram scatter value (y-axis) according to changes in a specific focus position (x-value).

[0222] Next, using the derivatives of the first to fifth sub-patterns derived therefrom, an overall focus graph for multiple sub-patterns can be generated by considering the pattern-specific weights of all sub-patterns. Then, a single overall weighted average graph for all multiple sub-patterns can be generated by weighting the overall focus graphs. Subsequently, an optimal focus position considering all multiple sub-patterns can be calculated using the overall approximate weighted average graph. The optimal focus position considering all multiple sub-patterns may be an inflection point of the overall approximate weighted average graph.

[0223] That is, in this step, the slopes of the focus and DW values ​​of all sub-patterns are calculated for each focus position (x-value), and a weighted average is calculated to represent a single pattern containing multiple sub-patterns. The weighted average value allows the pattern image with the greatest change to be considered as a priority when the focus changes. Additionally, the weighted average value can indicate the total level of change of all sub-patterns according to the change in focus position.

[0224] Furthermore, for each focus position (x-value), a weighted average value for all of the first to fifth sub-patterns is calculated, and this is represented as a DW value (y-axis) based on the focus (x-axis) and the weighted average value of all sub-patterns, thereby allowing the inflection point of this function to be calculated. Through this method, the optimal focus position of the laser light considering all multiple sub-patterns can be derived.

[0225] And, as shown in FIG. 18, the weighted average final focus and DW value can be derived by using the derivatives of all derived sub-patterns and considering the pattern-specific weights of all sub-patterns. By identifying the inflection point based on the above weighted average result, the optimal focus position for all sub-patterns can be calculated when multiple sub-patterns are formed in a single photolithography process.

[0226] In addition, in this stage, by utilizing the method described above, a unified mathematical / scientific logic can be established to objectively analyze the pattern-specific focus influence, which refers to the level of change in the physical interaction of the pattern due to changes in focus, regardless of the shape of the pattern.

[0227] Furthermore, the optimal focus of the laser light source is formed at a very fine size ranging from several nanometers (nm) to several micrometers (µm). The laser light source exhibits sensitive characteristics, such as being easily defocused if vibrations occur in the installed equipment. Accordingly, since it is difficult to transfer a pattern from a single position, which is the optimal focus of the light source, in all processes, an error range for focus position changes can be calculated to minimize external influences on pattern transfer. Additionally, a technology is required to monitor the focus position to maintain the said error range. Accordingly, the integrated monitoring method according to the embodiment can calculate an error range for focus changes that minimizes external influences on pattern transfer by utilizing approximate correction results. The said error range can be calculated by verifying the change in pattern shape due to focus changes by checking the approximate correction results, and determining the error range when the level of change in the DW value (y-axis) maintains a preset level.

[0228] The level of change in the above DW value (y-axis) can be calculated as an error range of greater than 0 to less than or equal to 10%, or other values ​​other than the above value can be selected to calculate the error range.

[0229] In addition, in this step, the level of change in the defocused pattern image compared to the most ideal pattern image formed at the optimal focus can be quantitatively verified through the DW value, indicating how much it has deviated from the optimal focus position.

[0230] The deviation between the optimal focus position and the defocus position as described above, i.e., the quantitative value, can be calculated using the DW value for each pattern image. The above DW value can be calculated using the number of pixels for each pixel brightness value.

[0231] Accordingly, the difference between the pixel brightness value and the number of pixels is calculated between the pattern image of the optimal focus position and the pattern image of the worst focus position. The calculated difference is then divided by the number of focus positions to determine the unit deviation value. Subsequently, the pixel brightness value and the number of pixels at the pattern image of the optimal focus position are compared with those at an unknown focus position, and the unit deviation value is applied to quantitatively identify the defocused position. Alternatively, methods other than those described may also be utilized.

[0232] Furthermore, in the current semiconductor industry, there are no metrics or methods to verify the level of focus achieved during the photolithography process used to produce a specific product. However, by utilizing the integrated monitoring method according to the embodiment, the focus level of the light source used during the production of a specific product can be accurately determined from the pattern image. In other words, when light is irradiated with the optimal focus properly set, the exact location where the light source is concentrated can be identified, allowing fine patterns to be transferred accurately and clearly to the target location.

[0233] For reference, the multiple sub-pattern images shown in FIG. 14 are merely examples for representing CD-SEM images of patterns, and the results of FIG. 15 to FIG. 18 may not have been produced using the multiple sub-pattern images.

[0234] Meanwhile, FIG. 19 is a configuration diagram showing an integrated monitoring system for semiconductor patterning and measurement processes according to an embodiment.

[0235] Referring to FIG. 19, the integrated monitoring system for semiconductor patterning and measurement processes according to the embodiment may include a pattern forming unit (210), an image collection unit (220), an abnormality detection unit (230), a cause detection unit (240), and a history management unit (250).

[0236] The above pattern forming unit (210) performs a semiconductor photolithography process to form a pattern on a wafer.

[0237] The pattern forming unit (210) can be implemented using various types of conventional semiconductor exposure devices used to form a pattern through a semiconductor exposure process. In particular, the pattern forming unit (210) may include a semiconductor exposure device that includes a laser light source and is used to form a pattern on a wafer through a photolithography process.

[0238] For example, the pattern forming unit (210) may use a semiconductor exposure device having a structure including a laser light source (211) for generating laser light, a lens (212, 212') for adjusting the focus of the laser light, a mask (213) on which a transfer pattern is formed, and a stage (214) that forms a structure on which the wafer (W) is placed and can move.

[0239] The laser light source (211) may include at least one of a KrF excimer laser light source, an ArF excimer laser light source, an ArFi laser light source, and an EUV laser light source.

[0240] In the pattern forming unit (210) above, at least one lens (212, 212') may be installed as needed. Although the drawing shows an example of a semiconductor exposure device having a structure with two lenses installed, it is not limited to such a structure.

[0241] Additionally, the pattern forming unit (210) can form a structure that selectively adjusts the focus position. The pattern forming unit (210) can form a pattern by changing the focus position, and can form a pattern on the wafer according to the focus position.

[0242] Additionally, the pattern forming unit (210) may include a plurality of semiconductor exposure devices. The plurality of semiconductor exposure devices may differ from one another in terms of manufacturers, patterning conditions, etc. Each of the plurality of semiconductor exposure devices may independently perform an exposure process to simultaneously form a plurality of patterns having the same characteristics.

[0243] The above image collection unit (220) has the role of collecting an image of an object to be analyzed by photographing the object to be analyzed, or collecting a pattern image of an object to be analyzed pattern.

[0244] The above image collection unit (220) can be implemented using various types of conventional shooting devices used to collect pattern images by photographing a pattern formed on a wafer.

[0245] For example, the image collection unit (220) may include a CD-SEM (Critical Dimension Scanning Electron Microscope) device and may collect CD-SEM pattern images.

[0246] The image collection unit (220) can collect pattern images for a pattern formed by the pattern forming unit (210). When at least one of the plurality of patterns formed by the pattern forming unit (210) is designated as a pattern to be analyzed, the image collection unit (220) can collect pattern images for the pattern to be analyzed and transmit the collected pattern images to the anomaly detection unit (230).

[0247] The above anomaly detection unit (230) receives a pattern image collected by the image collection unit (220). The above anomaly detection unit (230) subdivides the pattern image collected by the image collection unit (220) into pixel units. The above anomaly detection unit (230) calculates a pixel histogram indicating the degree of dispersion of pixel values ​​from the collected pattern image. Then, the above anomaly detection unit (230) derives a pixel histogram of the pattern image that statistically indicates the brightness at each pixel location and calculates the DW value of the collected pattern image. The above anomaly detection unit (230) can determine whether an outlier has occurred in the focus of the semiconductor exposure device by comparing the calculated DW value of the pattern image with reference information.

[0248] The above reference information may each include a reference pattern image for a pattern formed at an optimal focus position, a reference pixel histogram of the reference pattern image, a reference DW value, and an upper DW value and a lower DW value set based on the reference DW value.

[0249] The above abnormality detection unit (230) can determine whether an abnormality has occurred in the semiconductor exposure device by performing a similarity analysis by comparing a pixel histogram generated from a pattern image of a pattern to be analyzed with reference information.

[0250] The above anomaly detection unit (230) can perform similarity analysis through vector similarity-based methods such as cross-correlation distance, chi-square distance, intersection distance, Bhattacharyya distance, and cosine distance.

[0251] If the above abnormality detection unit (230) confirms that no abnormality has occurred in the semiconductor exposure device, it can evaluate whether an abnormality has occurred in the imaging device.

[0252] The above abnormality detection unit (230) can determine whether an abnormality has occurred in the shooting device by comparing the calculated DW value generated from the pattern image of the pattern to be analyzed with reference information.

[0253] Additionally, the above abnormality detection unit (230) compares the DW value calculated from the pattern image of the pattern to be analyzed with the upper DW value and the lower DW value. The above abnormality detection unit (230) can determine that an abnormality has occurred in the imaging device by determining that the level of variation between groups is such that if the DW value of the pattern image to be analyzed exceeds the upper DW value or if the DW value of the pattern image to be analyzed falls short of the lower DW value, an abnormality has occurred in the imaging device.

[0254] Alternatively, the above abnormality detection unit (230) may determine that there is no abnormality in the imaging device by determining that the DW value of the pattern image to be analyzed is located between the upper DW value and the lower DW value, and that there is no abnormality in the imaging device.

[0255] The above anomaly detection unit (230) can generate reference information and store the generated reference information. The above anomaly detection unit (230) can select an upper DW value and a lower DW value, respectively, based on the reference DW value. The above anomaly detection unit (230) can select an upper DW value and a lower DW value, respectively, according to a preset standard. The above anomaly detection unit (230) can change and set the upper DW value and the lower DW value in various ways as needed.

[0256] The above anomaly detection unit (230) can calculate the reference information in the following manner. First, a pattern is repeatedly formed on the wafer using laser light at multiple focus positions using the semiconductor exposure equipment. Then, pattern images for each focus position are collected for the pattern formed at the multiple focus positions. Next, pixel histograms of the pattern images for each focus position are derived. Next, the extracted pattern images are selected as reference pattern images, and reference DW values ​​of the reference pattern images can be derived. Then, an upper DW value and a lower DW value can be set using the reference DW values ​​according to a preset standard.

[0257] At this time, the anomaly detection unit (230) calculates DW values ​​for each focus position and calculates a DW curve using the multiple calculated DW values. Then, the anomaly detection unit (230) can select the inflection point of the DW curve as the optimal focus position. Additionally, after deriving the optimal focus position, the anomaly detection unit (230) can select an upper DW value and a lower DW value based on the optimal focus position according to a preset standard.

[0258] Additionally, the anomaly detection unit (230) can derive the level of anomaly occurrence, the location of the anomaly occurrence, and the time of the anomaly occurrence, respectively, of the pattern forming unit (210). The anomaly detection unit (230) can derive the level of anomaly occurrence and the time of the anomaly occurrence, respectively, of the image collection unit (220).

[0259] The above-mentioned anomaly detection unit (230) can detect the level of anomaly occurrence in the pattern forming unit (210) and the image collecting unit (220) by considering the characteristic that the pattern images collected for each focus position yield different DW values. The above-mentioned anomaly detection unit (230) can derive the location of anomaly occurrence in the pattern forming unit (210) by applying the vector similarity-based method and the SSIM technique. In addition, the time of anomaly occurrence in the pattern forming unit (210) and the image collecting unit (220) can be derived by using the results of DW value fluctuations by exposure process execution date and the results of DW value fluctuations by shooting process execution date.

[0260] In addition, the above-mentioned abnormality detection unit (230) calculates a Cpk (capability of process Katayori) index, which is a capability of process index for indicating the process capability of a semiconductor lithography device and an imaging device, using a reference DW value distribution and a calculated DW value distribution, and transmits the calculated Cpk index so that an administrator can numerically verify the process capability of the semiconductor lithography device and an imaging device.

[0261] Meanwhile, the semiconductor exposure system (200) according to the embodiment may further include a cause detection unit (240) and a history management unit (250).

[0262] The above cause detection unit (240) serves to derive an abnormality factor that is identified as affecting the occurrence of an abnormality in the pattern forming unit (210) and the image collection unit (220).

[0263] The cause detection unit (240) can derive a correlation between a plurality of abnormal occurrence factors and the level of variation of the DW value, derive a major abnormal occurrence factor among the plurality of abnormal occurrence factors that has a major influence on the variation of the DW value, and generate a control signal that controls a process variable associated with the major abnormal occurrence factor and provide it to the pattern forming unit (210).

[0264] Specifically, the cause detection unit (240) can construct a machine learning model based on the DW value and multiple various process and equipment-related variables that may affect the change in the DW value, and the machine learning model can determine whether the DW value is at a normal level or an abnormal level when the process and equipment-related variables are combined at a certain level. Subsequently, after patterning is performed on a new wafer and the DW value is extracted therefrom, if the level exceeds the upper and lower limits of the management limits of the DW value level recognized by the existing machine learning model, it is equipped with a function to notify the manager. In addition, notification information can be generated by comparing the pattern image of the wafer with the pattern image database of other wafers held by the existing machine learning model to determine which part of the wafer and at what level of abnormality has occurred.

[0265] Furthermore, if the combination of multiple diverse process and equipment variables for such a new wafer and the resulting DW value are normal, the machine learning model incorporates this data into the existing data held by the model and recognizes it, thereby strengthening the function of determining normal data.

[0266] Similarly, if the combination of multiple various process and equipment variables for a new wafer and the resulting DW value are abnormal, the machine learning model adds this data to the data held by the existing model; however, if the combination of multiple various process and equipment variables and the resulting DW value for another wafer in the future are similar to the case of the aforementioned abnormal wafer, the abnormal data detection function is enhanced so that the trained machine learning model can recognize and report this as abnormal data.

[0267] In this process, it should be noted that machine learning models for normal data and anomalous data do not exist separately, but rather a single machine learning model is constructed by mixing normal and anomalous data.

[0268] The machine learning model described above may be configured to receive an abnormal pattern image for a pattern determined to have an abnormality and to repeatedly perform machine learning in advance on the type of abnormality. The machine learning model may apply a machine learning algorithm to learn the correlation between the abnormality factor and the level of variation of the DW value, and derive the major abnormality factors according to their feature importance. The cause detection unit (240) provides the machine learning model generated in the above manner and the feature importance of the major factors contributing to the abnormality to the abnormality analysis unit (230) according to priority. Furthermore, the abnormality analysis unit (230) provides a function to predict, based on the data within the machine learning model, which factor among the factors with high contribution to the abnormality must be improved to a certain level for the DW value to return to a normal level, and what level the DW value will be at that time.

[0269] The above cause detection unit (240) can be implemented by applying analysis algorithms such as Linear Regression, Lasso, Ridge, Decision Tree, Random Forest, GBM (Gradient Boost), LGBM (Light GBM), XGB (XG Boost), etc.

[0270] The above history management unit (250) performs the role of managing the history of DW values ​​and abnormal occurrence factors for each exposure equipment included in the photolithography process and the pattern forming unit. The DW values ​​calculated from the pattern image can objectively describe the patterning characteristics without limitation on the type and form of the exposure process and exposure equipment.

[0271] The history management unit (250) can collect and store history information such as the process of the photolithography process, equipment conditions and equipment conditions of the exposure equipment included in the pattern forming unit, the pattern image, the pixel histogram derived from the pattern image, the DW value derived from the pattern image, the DW curve generated from the DW value, the level of abnormal occurrence, the location of abnormal occurrence, the time of abnormal occurrence, and the factor of abnormal occurrence. The history management unit (250) can provide the stored history information to the cause detection unit (240).

[0272] In addition, the history management unit (250) may provide the history information stored so that the cause detection unit (240) can provide simulation results that present the effect of improving patterning performance.

[0273] The integrated monitoring system for semiconductor patterning and measurement processes according to the embodiment can be utilized to manage a plurality of semiconductor exposure devices and imaging devices.

[0274] Specifically, the semiconductor industry utilizes multiple semiconductor lithography and imaging devices for the mass production of semiconductor devices. In particular, semiconductor lithography devices transfer ultrafine patterns in the nanometer range, while imaging devices collect pattern images to measure these ultrafine patterns. Due to the characteristics of ultrafine semiconductor processes, even minute differences in operation or errors among the multiple devices can result in significant discrepancies.

[0275] The integrated monitoring system for semiconductor patterning and measurement processes according to the embodiment utilizes data mining techniques to set priorities for each device regarding key factors and abnormality levels that significantly affect changes in DW values ​​in light sources and semiconductor exposure devices, and can provide information that allows for optimal maintenance and management of process conditions for each device by considering these priorities. Accordingly, it can guide all semiconductor devices to exhibit uniform and excellent performance during the semiconductor patterning and measurement processes.

[0276] For example, semiconductor lithography device A may exhibit a large variation in DW values ​​because the height of the stage where the wafer is placed is lower compared to other semiconductor lithography devices. Semiconductor lithography device B may exhibit a large variation in DW values ​​because the lens curvature is lower compared to other semiconductor lithography devices. Semiconductor lithography device C may exhibit a large variation in DW values ​​due to minute height differences caused by chuck tolerances or differences in wear rates.

[0277] The integrated monitoring system for semiconductor patterning and measurement processes according to the embodiment may provide a function to continuously manage the timing, level, and trend of changes by recommending, as a priority, the height of the wafer stage under the device conditions of semiconductor exposure device A, the curvature of the lens under the device conditions of semiconductor exposure device B, and the height difference between chucks under the device conditions of semiconductor exposure device C.

[0278] In addition, the integrated monitoring system for semiconductor patterning and measurement processes according to the embodiment can display the process conditions of the semiconductor exposure device and the imaging device as scientific numerical values. Accordingly, when performing the exposure and imaging processes, objectively identical process control standards can be applied, and even minute differences in conditions can be displayed as accurate numerical values, allowing all equipment to be calibrated to the same conditions even when multiple devices are utilized.

[0279] The performance evaluation method of the imaging device according to the embodiment described above, the integrated monitoring method of the semiconductor patterning and measurement process, and the system technology for the same are each performed in the order of collecting images to be analyzed, calculating pixel histograms and DW values, deriving DW curves, determining the optimal focus position of the semiconductor exposure device, evaluating whether an abnormality has occurred in the semiconductor imaging device, machine learning, evaluating trends in the semiconductor exposure process, and deriving the cause of the abnormality, thereby enabling the simultaneous integrated management of the performance of the imaging device and the semiconductor exposure device.

[0280] In this specification, the DW value refers to the Standard Deviation (SD) of a pixel histogram when a pattern image is subdivided into pixel units, the number of pixels having a specific brightness value is calculated, and the brightness at a pixel location is represented as a statistical pixel histogram. The DW value may be generated individually for each image subject to analysis.

[0281] In this specification, a DW curve refers to a DW value collected for each focus position while varying the focus position of a light source provided in a semiconductor lithography device, and a multiple collected DW values ​​for each focus position converted into a quadratic function. The DW curve can indicate the overall level of physical interaction according to changes in the focus position of the light source. The DW curve provides information that allows observation of how much the focus of the light source has changed from reference information (i.e., optimal focus position) when forming a semiconductor pattern through an actual semiconductor lithography process. That is, the focus position of the light source can be tracked in the semiconductor lithography process by utilizing the DW curve.

[0282] In this specification, the optimal focus position may refer to an inflection point, which is a position on the DW curve where the pattern can be formed most accurately and clearly on the wafer.

[0283] As described above, the detailed description of the preferred embodiments of the present invention disclosed is provided to enable those skilled in the art to implement and practice the present invention. Although the present invention has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the present invention without departing from the scope of the invention. For example, those skilled in the art may utilize each configuration described in the embodiments described above in combination with one another. Accordingly, the present invention is not intended to be limited to the embodiments shown herein, but to be given the broadest scope consistent with the principles and novel features disclosed herein.

[0284] The present invention may be embodied in other specific forms without departing from the spirit and essential features of the invention. Accordingly, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention. The invention is not intended to be limited to the embodiments shown herein, but to be given the broadest possible scope consistent with the principles and novel features disclosed herein. Furthermore, embodiments may be constructed by combining claims that are not explicitly related in the claims, or by including them as new claims through amendments made after filing. Explanation of the symbols

[0285] 200: Integrated Monitoring System 210: Pattern forming section 220: Image Collection Department 230: Anomaly detection unit 240: Cause detection unit 250: History Management Department

Claims

Claim 1 A method for monitoring the performance of a photographing device, comprising: a step of collecting an image of an object to be analyzed by photographing it using a photographing device; a step of deriving a pixel histogram of the image to be analyzed; a step of calculating a DW value (Dark or White value) of the image to be analyzed using the pixel histogram; and a step of evaluating whether an abnormality has occurred due to a change in focus of the photographing device by comparing the DW value of the image to be analyzed with reference information of a reference image to be analyzed; wherein the reference image to be analyzed is selected by collecting an image to be analyzed according to a change in focus of the photographing device, calculating a DW value from each image to be analyzed according to the focus, generating a DW curve using a plurality of calculated DW values, and selecting an image to be analyzed having a DW value located at an inflection point of the generated DW curve. Claim 2 In claim 1, the above imaging device is a CD-SEM (critical dimension scanning electron microscope) device, and the method for monitoring the performance of the imaging device. Claim 3 In claim 1, the analysis target is a method for monitoring the performance of an imaging device comprising a semiconductor pattern formed through a semiconductor photolithography process. Claim 4 A method for monitoring the performance of a shooting device according to claim 1, wherein the step of calculating the DW value of the image to be analyzed comprises dividing the image to be analyzed into pixel units, classifying the divided pixels by brightness value to calculate the number of pixels for each brightness value, calculating the pixel histogram using the calculated number of pixels for each brightness value, and then calculating the DW value of the image to be analyzed from the pixel histogram. Claim 5 A method for monitoring the performance of a shooting device according to claim 1, wherein the step of evaluating whether an abnormality has occurred in the shooting device comprises comparing the DW value of the image to be analyzed with the upper DW value and lower DW value of the reference information, and determining that the shooting device is operating normally by determining the level of variation within the group when it is confirmed that the DW value of the image to be analyzed is between the upper DW value and lower DW value. Claim 6 A method for monitoring the performance of a shooting device according to claim 1, wherein the step of evaluating whether an abnormality has occurred in the shooting device comprises comparing the DW value of the image to be analyzed with the upper DW value and lower DW value of the reference information, and determining that an abnormality has occurred in the operation of the shooting device by determining the level of variation between groups when it is confirmed that the DW value of the image to be analyzed exceeds the upper DW value or falls short of the lower DW value. Claim 7 A method for integrated monitoring of a semiconductor patterning and measurement process, comprising: a step of forming a pattern on a wafer by performing an exposure process using a semiconductor exposure device; a step of collecting a pattern image of a pattern to be analyzed using a capturing device; a step of deriving a pixel histogram of the pattern image and calculating a DW value (Dark or White value) of the pattern image using the pixel histogram; and a step of evaluating whether an abnormality has occurred due to a focus change of the semiconductor exposure device and the capturing device by comparing at least one of the pixel histogram and the DW value of the pattern image with reference information, wherein the reference information comprises: collecting a pattern image according to the focus change of the capturing device, calculating a DW value from each of the collected pattern images according to the focus change, generating a DW curve from a plurality of calculated DW values, selecting a pattern image having a DW value located at an inflection point of the generated DW curve as a reference pattern image, and including the DW value of the selected reference pattern image. Claim 8 In claim 7, the semiconductor exposure device comprises a laser light source, and the imaging device is a CD-SEM device, in an integrated monitoring method for semiconductor patterning and measurement processes. Claim 9 In claim 7, the above reference information comprises a method comprising: a step of performing an exposure process with the semiconductor exposure device at multiple focus positions to form a pattern on the wafer for each focus position; a step of collecting pattern images for each focus position for the patterns formed at each of the multiple focus positions; a step of deriving a pixel histogram for each focus position using the pattern images for each focus position; a step of deriving a DW value for each focus position from the pixel histogram for each focus position; a step of generating a DW curve using the DW value for each focus position; and a step of determining an optimal focus position of the semiconductor exposure device using the DW curve, and extracting a pattern image, a pixel histogram, and a DW value corresponding to the optimal focus position and selecting them as reference information of the semiconductor exposure device. Claim 10 In claim 7, the step of evaluating whether an abnormality has occurred comprises performing a similarity analysis between a pixel histogram calculated from the pattern image and a reference pixel histogram included in the reference information, and evaluating that an abnormality has occurred in the semiconductor exposure device if the result of the similarity analysis falls short of a preset similarity value. Claim 11 In claim 10, the above similarity analysis is performed using a vector similarity-based method comprising at least one of cross-correlation distance, chi-square distance, intersection distance, Bhattacharyya distance, and cosine distance, in an integrated monitoring method for semiconductor patterning and measurement processes. Claim 12 In claim 7, the step of evaluating whether an abnormality has occurred comprises comparing a DW value calculated from the pattern image with an upper DW value and a lower DW value included in the reference information, and evaluating that an abnormality has occurred in the imaging device if it is confirmed that the DW value calculated from the pattern image exceeds the upper DW value or falls short of the lower DW value. Claim 13 In claim 7, the step of evaluating whether an abnormality has occurred further comprises, when it is evaluated that an abnormality has occurred in at least one of the semiconductor exposure device and the imaging device, a step of determining the level of abnormality occurrence in the semiconductor exposure device and the imaging device by comparing the DW value calculated from the pattern image with the reference DW value included in the reference information. Integrated monitoring method for semiconductor patterning and measurement processes. Claim 14 In claim 7, the step of evaluating whether an abnormality has occurred further comprises the step of determining the location of the abnormality in the pattern to be analyzed by evaluating the structural similarity index measure (SSIM) between the pattern image collected from the pattern to be analyzed and the reference pattern image included in the reference information when it is evaluated that an abnormality has occurred in the semiconductor exposure device. Claim 15 In claim 7, the step of evaluating whether an abnormality has occurred further comprises the step of deriving the cause of the abnormality in the semiconductor exposure device and the imaging device when it is evaluated that an abnormality has occurred in either the semiconductor exposure device or the imaging device. Claim 16 In claim 7, the step of evaluating whether an abnormality occurs further comprises the step of predicting the time of occurrence of an abnormality in the semiconductor exposure device and the imaging device using a DW value calculated from a pattern image collected from the pattern to be analyzed. This describes an integrated monitoring method for semiconductor patterning and measurement processes. Claim 17 In claim 7, the step of evaluating whether an abnormality has occurred further comprises the step of calculating a Cpk (capability of process Katayori) index to indicate the process capability of the semiconductor exposure device and the imaging device using a pixel histogram calculated from a pattern image collected from the analysis target pattern and a reference pixel histogram included in the reference information. This describes an integrated monitoring method for semiconductor patterning and measurement processes. Claim 18 A semiconductor lithography system comprising: a pattern forming unit including a semiconductor lithography device that performs a lithography process to form a pattern on a wafer; an image collecting unit including a shooting device that collects a pattern image of a pattern to be analyzed formed on the wafer; and an anomaly detection unit that evaluates whether an anomaly occurs due to a change in focus of the semiconductor lithography device and the shooting device by comparing at least one of a pixel histogram and a DW value of the pattern image with reference information, wherein the reference information includes reference information of the shooting device that collects a pattern image according to a change in focus of the shooting device, calculates a DW value from each of the collected pattern images according to the change in focus, generates a DW curve with a plurality of calculated DW values, selects a pattern image having a DW value located at an inflection point of the generated DW curve as a reference pattern image, and includes a DW value of the selected reference pattern image. Claim 19 A semiconductor lithography system according to claim 18, further comprising a cause detection unit for deriving an abnormality occurrence factor predicted to affect the occurrence of an abnormality in the semiconductor lithography device and the imaging device. Claim 20 A semiconductor lithography system according to claim 19, further comprising a history management unit that collects and stores history information of the semiconductor lithography device and the imaging device, wherein the history information is provided to the cause detection unit, and wherein the stored history information is provided to the cause detection unit.

Images