Photo response non-uniformity correction during semiconductor inspection
Patent Information
- Application Number
- TW113124349
- Authority / Receiving Office
- TW · TW
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2023-06-30
- Filing Date
- 2024-06-28
- Publication Date
- 2026-11-11
- Estimated Expiration
- 2044-06-27
Smart Images

Figure TWG2TB001920289_001 
Figure TWG2TB001920289_002 
Figure TWG2TB001920289_003
Abstract
Description
Technical Field
[0001] Cross - Reference to Related Applications
[0002] This application claims priority to the filing date of U.S. Patent Application No. 18 / 217,359, filed on June 30, 2023, and titled "PHOTO RESPONSE NON - UNIFORMITY CORRECTION DURING SEMICONDUCTOR INSPECTION", the entire content of which is incorporated herein by reference.
[0003] This disclosure is generally related to semiconductor devices, and particularly to techniques for inspecting semiconductor devices. Background Art
[0004] Semiconductor devices are essential components of modern electronic and computing devices. A semiconductor device is an electronic component that utilizes the electrical properties of semiconductor materials. The conductivity of semiconductor materials can be manipulated by introducing an electric or magnetic field. Improvements in manufacturing processes have led to exponential improvements in the size, speed, and cost of semiconductor devices.
[0005] In a typical semiconductor manufacturing process, bare substrates (such as substrates, panels, and wafers) are processed using lithography techniques to build circuitry thereon. These substrates with circuitry are typically divided into smaller pieces called dies. These dies form the basis of common electronic devices. Summary of the Invention
[0006] In semiconductor manufacturing processes, image sensors assist in inspecting semiconductor substrates (including semiconductor wafers). However, due to the non - uniformity of cells between image sensors (which is referred to as "photo response non - uniformity (PRNU)"), when illuminated with a uniform light source, each image sensor may generate different voltage levels, resulting in inaccurate pixel - level brightness, which is referred to as "photo response non - uniformity" (or "illumination flat field"). Photo response non - uniformity (PRNU) can be corrected using a system calibration, which calibrates a substrate (including a wafer) under a given illumination condition, objective lens, and camera. Subsequently, the PRNU correction is applied to all images acquired using that given setting to produce a uniform image. This disclosure describes a technique for correcting PRNU at runtime, even when the substrate has a profile (e.g., tilt or warp) relative to the normal axis of the optical system.
[0007] In some aspects, the present disclosure relates to a method of using a semiconductor substrate inspection system to determine an outline of a semiconductor substrate, comprising: generating an image of the semiconductor substrate; determining a grayscale value of each pixel of the image; and using a relationship between the grayscale value of the image and an incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate to determine the outline of the semiconductor substrate. Brief Description of the Drawings
[0008] In the drawings which are not necessarily drawn to scale, like reference numerals may describe similar components in different views. Similar reference numerals with different letter suffixes may represent different instances of similar components. The drawings generally illustrate, by way of example and not limitation, various embodiments discussed in this document. [FIG. 1] illustrates an example portion of a manufacturing system 100 that can be used to perform one or more of the techniques shown and described herein. [FIG. 2] depicts an example of a composite image of the substrate of FIG. 1. [FIG. 3] is a side view of the substrate of FIG. 1. [FIG. 4] is a graphical illustration of an example of an outline diagram of a substrate. [FIG. 5] is a conceptual diagram illustrating an example of an image of a substrate obtained using an inspection camera. [FIG. 6] depicts a simulation of a scatter plot of the incident angles of a substrate. [FIG. 7] graphically depicts the polar angle θ. [FIG. 8] is a conceptual diagram illustrating an example of a grouping of polar angles. [FIG. 9] is a conceptual diagram depicting an example of a group of images associated with a corresponding incident angle range of a camera of a semiconductor substrate inspection system relative to the semiconductor substrate. [FIG. 10] depicts an example graph depicting the relationship between the incident angle and the grayscale value of a specific pixel. [FIG. 11] depicts a flowchart of an example of a method of training a calibration model. [FIG. 12] is a graphical illustration of an example of a training outline diagram of a substrate. [FIG. 13] depicts a simulation of a scatter plot of the incident angles of the combination of the substrate and the gasket of FIG. 12. [FIG. 14] depicts a flowchart of an example of a method of calibrating a semiconductor substrate inspection system in accordance with various techniques of the present disclosure. [FIG. 15] depicts a flowchart of another example of a method of calibrating a semiconductor substrate inspection system in accordance with various techniques of the present disclosure. [FIG. 16] is a flowchart depicting an example of a method for determining an outline of a semiconductor substrate using a semiconductor substrate inspection system. Embodiments
[0009] In semiconductor manufacturing processes, image sensors assist in inspecting semiconductor substrates (including semiconductor wafers). However, due to the non-uniformity of the units between image sensors, when illuminated with a uniform light source, each image sensor may generate different voltage levels, resulting in inaccurate pixel-level brightness, which is referred to as "photo-response non-uniformity" (or illuminated flat field).
[0010] Photo-response non-uniformity (PRNU) can be corrected using a system calibration, which calibrates a substrate (including wafers) under a given illumination condition, objective lens, and camera. Subsequently, the PRNU correction is applied to all images acquired using the given settings to generate a uniform image.
[0011] The inventors have recognized that different responses may occur when an outline (e.g., tilt or warp) in the substrate involves the normal axis of the optical system. Optically, this is caused by the movement of the pupil with different angular inputs, but in practice, it manifests as different PRNUs that vary with the angle of a substrate (relative to the optical system).
[0012] To make matters more complex, the angle continuously changes when inspecting the substrate. Imagine looking down at a mountain. Since the surface area is rasterized, the angle of the mountain surface in each image is different. The inventors have recognized that due to the outlines (e.g., tilt) in the semiconductor substrate, a semiconductor substrate inspection system needs to be corrected.
[0013] This disclosure describes a technique for dynamically correcting PRNU at runtime, which is based on a) a PRNU model of how imaging responds to tilts in a customer substrate; and b) the tilts measured by using a point sensor that maps the outline (e.g., once per recipe, per lot, or per substrate). This outline map establishes the angle of the substrate relative to the optical system.
[0014] FIG. 1 illustrates an example portion of a manufacturing system 100 that can be used to perform one or more of the techniques shown and described herein. The manufacturing system 100 can be used to manufacture and inspect a substrate 102 (such as a semiconductor device).
[0015] The manufacturing system 100 may include one or more tools 110 (also referred to as "machines") to perform a processing step on the substrate 102 (e.g., a semiconductor wafer) for manufacturing. In some examples, the tool 110 may include a chemical vapor deposition (CVD) machine for depositing one or more layers on the substrate 102. In some examples, the tool 110 may include: a polishing tool for performing CMP; an illumination light source for emitting a radiation beam (e.g., an electromagnetic wave) that is projected onto the substrate 102 for lithography; an electroplating tool for electroplating the substrate; and / or other suitable tools for semiconductor manufacturing.
[0016] The manufacturing system 100 may include a metrology instrument 112 (also referred to as an "inspection instrument") to measure various features on the substrate 102 or various properties of the substrate 102. For example, the various features and properties may include film thickness measurement, critical-dimension (CD) measurement of the features formed on the substrate (in the x-dimension, y-dimension, and / or z-dimension), pitch of line-space features formed on the substrate, overlay offset from one layer to another on the substrate, or several other measurements or properties known to those of ordinary skill in the art. The inspection instrument can be used to inspect the compliance of the features formed or otherwise found on the substrate. For example, the inspection can be directed to a substrate on which integrated circuit dies have been manufactured, the location of the dies, the location of unwanted particulate matter or other unwanted or unplanned features, and so on.
[0017] The tool 110 and the metrology instrument 112 may be coupled to a control system 120. The control system 120 may include a central processing unit (CPU) 122, a graphics processing unit (GPU) 124, a field-programmable gate array (FPGA) 131 (or other suitable accelerators such as a data processing unit (DPU), RNN, artificial neuron network (ANN), and the like), a memory 126, a display 128, an input device 130, and a communication interface 132 (e.g., a high performance network (HPC)). The control system 120 may also include a front-end circuit system such as a transmission signal chain, a reception signal chain, a switching circuit system, digital and analog circuit systems, etc. The transmission signal chain can provide control signals for the tool 110. The reception signal chain can receive process parameter measurements from the metrology instrument 112.
[0018] The front-end circuit system can be coupled to and controlled by one or more processor circuits, such as CPU 122, GPU 124, and FPGA 131. CPU 122 can be configured as one or more multi-core processors. GPU 124 and FPGA 131 can be used to accelerate data processing and the performance of machine learning networks as described herein. The techniques shown and described herein can be executed by CPU 122 working in conjunction with GPU 124 for faster processing.
[0019] CPU 122 and GPU 124 can be coupled to memory 126, such as to execute instructions that cause control system 120 to perform one or more of manufacturing control, processing, or inspection-related data storage, or otherwise execute instructions of the techniques shown and described herein. Control system 120 can be communicatively coupled to other parts of system 100, such as using a wired or wireless communication interface 132.
[0020] For example, the performance of one or more of the techniques shown and described herein can be completed on control system 120, or using other processing facilities or storage facilities, such as using a computing facility 140 (e.g., a general-purpose computing device such as a server, cloud processing system, data warehouse, laptop, tablet, smartphone, desktop computer, or the like). For example, if execution on control system 120 would be undesirably slow or beyond the capabilities of control system 120, then task processing can be performed remotely (e.g., on another system), such as in response to a request from control system 120. Similarly, the storage of inspection data or intermediate data can be completed using a remote facility communicatively coupled to control system 120. Control system 120 can also include a display 128, such as for presenting configuration information or results; and an input device 130, such as including one or more of a keyboard, trackball, function keys or soft keys, mouse interface, touch screen, stylus, or the like, for receiving operator commands, configuration information, or responses to queries.
[0021] FIG. 2 depicts an example of a composite image 200 of the substrate 102 of FIG. 1. During inspection, the metrology instrument 112 of FIG. 1 rotates the substrate 102, and an inspection camera (shown in FIG. 3) captures images of the substrate 102 along a first circular path within the field of view of the inspection camera. Subsequently, a robotic mechanism can direct the inspection camera to move radially, for example, inwardly toward the center of the substrate 102, such that the inspection camera can capture images of the substrate 102 along a second circular path. Finally, the inspection camera captures an image of the entire substrate 102, and the metrology instrument 112 can use the captured images to generate the composite image 200 shown in FIG. 2.
[0022] FIG. 3 is a side view of the substrate 102 of FIG. 1. FIG. 3 depicts portions of the metrology instrument 112 of FIG. 1, including an inspection camera 300 and a sensor 302. In a non-limiting example, each image captured by the inspection camera 300 can have two million pixels. The inspection camera 300 has a field of view 304. Although shown positioned below the substrate 102, the inspection camera 300 and the sensor 302 can be positioned above the substrate 102. In other examples, the metrology instrument 112 is a dual-sided inspection instrument, and thus, an inspection camera 300 and a sensor 302 can be positioned on both sides of the substrate 102. The term "substrate" refers to semiconductor substrates, panels, and wafers, as these techniques can be applied to all specific examples of such semiconductor structures.
[0023] The magnification of the metrology instrument 112 determines how many images the inspection camera 300 will capture. In some examples, the inspection camera 300 can capture 700 to 800 images, such that the control system 120 can use these images to generate the composite image 200 of FIG. 2 for display on the display 128.
[0024] The control system 120 can control the sensor 302 (e.g., a point sensor) to measure a Z distance to the substrate 102. Based on the Z distance determined using the sensor 302, the control system 120 can generate a contour (or topography) map of the substrate 102, for example, for display on the display 128. The contour map depicts the tilt of the substrate 102 measured by the sensor 302 at various points. Like a topography map, the contour map can show variations in the height and the steepness of the slopes of the substrate 102 due to warping of the substrate 102. An example of a contour map is shown in FIG. 4.
[0025] FIG. 4 is a graphical illustration of an example of an outline view 400 of a substrate. The outline view 400 of FIG. 4 is for a wafer having a diameter of 300 millimeters (mm) and a thickness of 780 micrometers (μm). The outline view 400 is a 2D view that depicts the tilt of the wafer in 3D space.
[0026] The control system 120 knows the position on the outline view of each image captured using the inspection camera 300. The control system 120 can determine the normal vector of a plane and determine the angle relative to the inspection camera 300 (which is the incident angle). Using the various techniques of the present disclosure, the control system 120 can determine how light reflects away and determine what type of shadow is produced for each given incident angle.
[0027] Using the various techniques of the present disclosure and as detailed below, the control system 120 can generate a plurality of image groups of a semiconductor substrate (such as the substrate 102 in FIG. 1), where each group in the plurality of image groups is associated with a corresponding incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate. For example, each group in the plurality of image groups can be associated with a corresponding incident angle range of the inspection camera 300 (FIG. 3) in the metrology instrument 112 (FIG. 1). Subsequently, the control system 120 can determine a correction model for each pixel of an image and use the correction model to generate a corrected image of the semiconductor substrate, thereby reducing or eliminating PRNU.
[0028] FIG. 5 is a conceptual diagram illustrating an example of an image of a substrate obtained using an inspection camera. An inspection camera (such as the inspection camera 300 in FIG. 3) can acquire images 500A to 500D along a path 502 (for example, a circular path when the metrology instrument 112 in FIG. 1 rotates the substrate 102).
[0029] The control system 120 can determine a normal vector 504 of a plane 506 to simulate the positioning of the inspection camera 300. Subsequently, for each of the images 500A to 500D, the control system 120 can determine an angle relative to the normal vector 504 (which is an incident angle relative to the inspection camera 300).
[0030] In the conceptual diagram shown in FIG. 5, the control system 120 determines that the images 500A and 500B are associated with a vector 508 relative to the plane 506, and the vector 508 has a first incident angle relative to the normal vector 504. The control system 120 determines that the images 500C and 500D are associated with a vector 510 relative to the plane 506, and the vector 510 has a second incident angle relative to the normal vector 504, where the second incident angle is greater than the first incident angle associated with the images 500A and 500B.
[0031] FIG. 6 depicts a simulation of a scatter plot of the incident angles of a substrate. The y-axis of the scatter plot 600 depicts the incident angle, and the x-axis depicts the polar coordinate angle θ in degrees (0 to 360 degrees), which corresponds to an outline of a substrate (such as the outline 400 corresponding to FIG. 4).
[0032] The control system 120 can group the images associated with a corresponding range of incident angles. For example, referring to FIG. 5, the control system 120 can group the images 500A and 500B together in a first group and group the images 500C and 500D together in a second group. In a non-limiting example, the 360-degree range of the x-axis can be conceptually divided into four quadrants, such as those shown in FIG. 8. In this example, the control system 120 can group the images 500A and 500B together in one quadrant, and the control system 120 can group the images 500C and 500D together in another quadrant.
[0033] FIG. 7 graphically depicts the polar coordinate angle θ. The polar coordinate angle θ is the angle between the incident angle and the vector normal to the substrate 102.
[0034] FIG. 8 is a conceptual diagram depicting an example of a grouping of polar coordinate angles. The field of view can be divided into multiple regions. In the non-limiting example shown in FIG. 8, the field of view (e.g., the polar coordinate angle grouping 800) includes four regions, such as quadrants 0 to 3. If the incident angle vector falls within the region (e.g., quadrant), then all the images falling within that region are grouped together. These regions (e.g., quadrants) serve as a rough estimate of the incident angle, thereby simplifying the calculations performed by the control system 120. For example, due to the grouping, the control system 120 can later determine a third-order polynomial fit versus a fourth-order polynomial fit (or higher order).
[0035] FIG. 9 is a conceptual diagram depicting an example of a group of images associated with a corresponding range of incident angles of a camera of a semiconductor substrate inspection system relative to the semiconductor substrate. A group of images may include one or more images. For example, group 900 in FIG. 9 includes five images, including image 500A and image 500B of FIG. 5.
[0036] Each image has a plurality of pixels, such as two million pixels. A pixel 902 of image 500A is shown in FIG. 9. Each pixel in the image has a grayscale value. Using the techniques of the present disclosure, the control system 120 determines the relationship between an incident angle and the grayscale value for each pixel, which is shown in a graph in FIG. 10. The top image (e.g., image 500A) in group 900 has an incident angle, and the grayscale value of pixel 902 in image 500A at that incident angle is X. The next image (e.g., image 500B) in group 900 has a different incident angle (but is still in the same quadrant), and the grayscale value of pixel 902 in image 500B at that incident angle is Y, where the grayscale value Y may be different from the grayscale value X. The other images in group 900 will have corresponding grayscale values of pixel 902 in a similar manner.
[0037] For each pixel 902 of an image in group 900, the control system 120 can determine a calibration model. For example, the control system 120 can apply linear regression techniques to determine a calibration model. Linear regression generates the coefficients of a polynomial representing a calibration model. For example, for each pixel 902 of an image in group 900, the control system 120 performs a third-order polynomial regression fit. Subsequently, the obtained coefficients can be used to calculate an expected grayscale value for the given incident angle.
[0038] The inventors have recognized that in some instances, it may be desirable for the control system 120 to perform linear regression techniques in image form rather than converting the images into a corresponding numerical form. Performing polynomial linear regression on each pixel (e.g., two million pixels) can be computationally expensive. Therefore, to utilize the parallel capabilities of the CPU / GPU, the data can be kept in image form, and linear regression can be performed on all two million pixels simultaneously. This procedure can be regarded as multiplying and adding the images, rather than single scalar values.
[0039] For example, the control system 120 can multiply two images and retain their product in another image, which can simplify the calculation and thus optimize the process. Thus, linear regression can generate an A coefficient image, a B coefficient image, a C coefficient image, etc. for a polynomial, and these images can be used to determine a new image based on the corrected model input of a given input angle.
[0040] In this way, the control system 120 generates a model for each pixel in the field of view for the specific incident angle as to what the gray scale value of each pixel should be. During operation, the control system 120 can view the incident angle associated with an image of a substrate and apply the correction model to the image to generate a corrected image of the substrate.
[0041] In some examples, the control system 120 can filter (or limit) each group in a plurality of image groups to determine a central tendency gray scale value for each pixel of the image. Filtering helps to distinguish artifacts of the substrate from artifacts of the inspection system. For example, the control system 120 can group multiple images of a substrate based on the incident angle and then apply a filter (e.g., a median filter) to determine a nominal gray scale value (e.g., central tendency gray scale value) for each pixel of the image. The control system 120 can subtract the nominal gray scale value of the pixel from the gray scale value obtained using the correction model to generate a difference relative to the model. The control system 120 can subtract the nominal gray scale value of the pixel from each image in FIG. 9 to determine whether a specific pixel value should be added to the model. As an example, if a pixel in image 500A is abnormally different from "nominal", the control system 120 can reject the pixel in image 500A from participating in the linear regression fit. In some examples, a given pixel can be considered abnormal using a statistical threshold value such as two σ. Not all pixels of all images in group 900 will be used to generate the correction model. In this way, the control system 120 can filter the content of the image before allowing each image to be incorporated into the model.
[0042] FIG. 10 depicts an example of a graph 1000 that depicts the relationship between the incident angle and the gray scale value for a single pixel of a quadrant or group. The x-axis represents the incident angle and the y-axis represents the gray scale value. As shown in FIG. 10, the gray scale value changes with the change of the incident angle.
[0043] FIG. 11 depicts a flowchart of an example of a method 1100 for training a calibration model. Referring back to FIG. 6, the maximum incident angle shown is approximately 0.18. However, the incident angles of other substrates may be higher than 0.18. The inventors have recognized that it may be desirable to induce all incident angles up to the maximum incident angle that the control system 120 will encounter. In this way, the control system 120 can train a calibration model.
[0044] For example, a special tool can be used to induce all incident angles up to the maximum incident angle that the control system 120 will encounter. As a non-limiting example, a spacer can be positioned to induce the desired range of incident angles.
[0045] In block 1102, the method 1100 includes training a calibration model by generating a training profile of a semiconductor substrate. An example of a training profile is shown in FIG. 12.
[0046] In block 1104, the method 1100 includes: using the training profile to generate a plurality of incident angles, where the plurality of incident angles includes more incident angles than the plurality of incident angles determined from the substrate under inspection. An example of a scatter plot is shown in FIG. 13, which depicts incident angles up to approximately 0.3.
[0047] A system (such as one or a part of the manufacturing system 100 of FIG. 1) can perform the techniques shown in FIG. 11.
[0048] FIG. 12 is a graphical illustration of an example of a training profile 1200 of a substrate. The training profile 1200 of FIG. 12 includes a wafer with a diameter of 300 millimeters (mm) and a thickness of 780 micrometers (μm), which is used in combination with a special tool (e.g., a spacer) to induce a plurality of incident angles.
[0049] FIG. 13 is a simulation of a scatter plot depicting the incident angles of the combination of the substrate of FIG. 12 and the spacer. The y-axis of the scatter plot 1300 depicts the incident angles, and the x-axis depicts the polar coordinate angle θ in degrees (0 to 360 degrees), which corresponds to a profile (such as the profile 1200 corresponding to FIG. 12). As shown in FIG. 13, incident angles from 0 to approximately 0.3 are induced.
[0050] FIG. 14 depicts a flowchart of an example of a method 1400 for calibrating a semiconductor substrate inspection system according to various techniques of the present disclosure. A system (such as one or a part of the manufacturing system 100 of FIG. 1) can perform the techniques shown in FIG. 14.
[0051] In block 1402, method 1400 includes determining a relationship between a grayscale value of each pixel of an image of a semiconductor substrate and an incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate. In block 1404, method 1400 includes determining a correction model for each pixel of an image based on the relationship. In block 1406, method 1400 includes applying the correction model to the image. In block 1408, method 1400 includes generating a corrected image of the semiconductor substrate.
[0052] FIG. 15 is a flowchart depicting another example of a method 1500 for calibrating a semiconductor substrate inspection system in accordance with various techniques of the present disclosure. A system (such as one or a portion of the manufacturing system 100 of FIG. 1) may perform the techniques shown in FIG. 15.
[0053] In block 1502, method 1500 includes generating a plurality of image groups of a semiconductor substrate, wherein each group of the plurality of image groups is associated with a corresponding incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate. In block 1504, method 1500 includes determining a correction model for each pixel of an image. In block 1506, method 1500 includes using the correction model to generate a corrected image of the semiconductor substrate.
[0054] In addition to the above techniques, the inventors have further recognized that it is desirable to use the grayscale values of the pixels in the corresponding images of a substrate to determine the profile of the substrate. The control system 120 uses the above techniques to determine a calibration model, and thus the control system 120 knows the incident angle and how the incident angle affects the grayscale value of a pixel of an image. The control system 120 can obtain an uncorrected image, such as the image 500A of FIG. 5, and for each pixel, determine an incident angle from the grayscale value of the corresponding pixel.
[0055] Using the relationship between the grayscale values of the image and the incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate, the control system 120 can determine the profile of the substrate for the image. In other words, the control system 120 can determine how the surface of the substrate is inclined for the image. For example, for each pixel of an image, the control system 120 can use the grayscale value of the pixel and apply the correction model to back-calculate to determine the incident angle of the image, whereby it can be used to determine the profile, such as the inclination (of the substrate), for the image.
[0056] By using these techniques, the images obtained by an inspection camera (such as inspection camera 300 in FIG. 3) can be used to determine the shape (e.g., tilt), rather than using a point sensor (such as sensor 302 in FIG. 3). Different from point sensors, since an inspection camera has a field of view, using the images obtained by an inspection camera can be much faster than using the data obtained by using a point sensor. A flowchart of an example of the method is shown in FIG. 16.
[0057] FIG. 16 depicts a flowchart of an example of a method 1600 for using a semiconductor substrate inspection system to determine the shape of a semiconductor substrate. A system (such as one or part of manufacturing system 100 in FIG. 1) can perform the techniques shown in FIG. 16.
[0058] In block 1602, method 1600 includes generating an image of the semiconductor substrate. For example, an inspection camera (such as inspection camera 300 in FIG. 3) can generate an image (such as image 500A in FIG. 5).
[0059] In block 1604, method 1600 can include determining a grayscale value of each pixel of the image. For example, control system 120 can determine a grayscale value of a pixel 902 of image 500A in FIG. 9.
[0060] In block 1606, method 1600 can include using the relationship between the grayscale values in the image and a range of incident angles of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate to determine the shape (e.g., tilt) of the semiconductor substrate. For example, control system 120 can determine the shape (e.g., tilt) of each of the generated images 500A to 500D, etc., and use those shapes that are available to determine the shape (e.g., tilt) of substrate 102. Various considerations
[0061] Each of the non-limiting aspects or examples described herein can stand alone, or can be combined with one or more of other examples in various arrangements or combinations.
[0062] The above embodiments include the accompanying drawings that form part of the embodiments. The drawings show, by way of illustration, specific embodiments in which the present invention may be practiced. These embodiments are also referred to herein as "examples". Such examples may include elements in addition to those shown or described. However, the inventors also contemplate examples in which only those elements shown or described are provided. In addition, the inventors also contemplate examples using any combination or arrangement (or one or more aspects) of those elements shown or described, whether with respect to the specific examples (or one or more aspects) shown or described herein, or with respect to other examples (or one or more aspects).
[0063] In the event of inconsistent usage between this document and any document incorporated herein by reference, the usage of this document shall prevail.
[0064] In this document, the term "a / an" is used (as is common in patent documents) to include one, or more than one, distinct from any other case or usage of "at least one" or "one or more". In this document, the term "or" is used to indicate non-exclusivity, such that "A or B" includes "A but not B", "B but not A", and "A and B", unless otherwise indicated. In this document, the terms "including" and "in which" are used as the ordinary English equivalents of the respective terms "comprising" and "wherein". Further, in the following aspects, the terms "including" and "comprising" are open-ended, i.e., a system, apparatus, article, composition, formulation, or process that includes elements in addition to those listed after this term in the aspect is still considered to fall within the scope of that aspect. In addition, in the following aspects, the terms "first", "second", "third", etc. are used only as labels and are not intended to impose a numerical requirement on their objects.
[0065] The method examples described herein can be implemented, at least in part, by a machine or computer. Some examples can include a computer-readable medium or machine-readable medium encoded with instructions that can be operative to configure an electronic device to perform the methods described as examples above. Implementations of such methods can include code, such as microcode, assembly language code, high-level language code, or the like. Such code can include computer-readable instructions for performing various methods. The code can form part of a computer program product. Further, in one example, the code can be tangibly stored on one or more volatile non-transitory, or non-volatile tangible computer-readable media, such as during execution or at other times. Examples of such tangible computer-readable media can include, but are not limited to, hard disks, removable magnetic disks, removable optical disks (e.g., optical disks and digital video disks), magnetic tape cartridges, memory cards or sticks, random access memory (RAM), read only memory (ROM), and the like.
[0066] The foregoing description is intended to be illustrative and not restrictive. For example, the above examples (or one or more aspects thereof) can be used in combination with each other. Other embodiments can be used, such as by one of ordinary skill in the art reviewing the foregoing description. The abstract is provided to comply with 37 C.F.R. § 1.72(b) to allow the reader to quickly ascertain the nature of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the aspects. Also, in the foregoing embodiments, various features can be combined together to simplify the disclosure. This should not be construed as intending that non-claimed disclosed features are essential to any aspect. Rather, the subject matter of the invention can lie in less than all of the features of a particular disclosed embodiment. Accordingly, the following aspects are hereby incorporated into the embodiments as examples or embodiments, where each aspect is independently a separate embodiment, and it is contemplated that such embodiments can be combined with each other in various combinations or permutations. The scope of the invention should be determined with reference to the appended aspects, along with the full scope of equivalents to which such aspects are entitled.
[0067] 100: Manufacturing system 102: Substrate 110: Tool 112: Metering instrument 120: Control system 122: Computer processing unit 124: Graphics processing unit 126: Memory 128: Display 130: Input device 131: Field Programmable Gate Array 132: Communication Interface 140: Computing Facility 200: Composite Image 300: Inspection Camera 302: Sensor 304: Field of View 400: Outline 502: Path 500A: Image 500B: Image 500C: Image 500D: Image 504: Normal Vector 506: Plane 508: Vector 510: Vector 600: Scatter Plot 800: Polar Coordinate Angle Grouping 900: Image Group 902: Pixel 1000: Chart 1200: Training Outline 1300: Scatter Plot
Claims
1. A method for calibrating a semiconductor substrate inspection system, comprising: generating a plurality of image groups of a semiconductor substrate, wherein each of the plurality of image groups is associated with a corresponding incident angle range of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate; determining a calibration model for each pixel of the image, wherein the calibration model is configured to correct photoresponse inhomogeneities of an image sensor based on a measurement of the three-dimensional shape of the semiconductor substrate; and using the calibration model to generate a calibrated image of the semiconductor substrate.
2. The method of claim 1, wherein determining the correction model for each pixel of the image includes: For each pixel, determine the relationship between the grayscale value of that pixel and the corresponding incident angle.
3. The method of claim 2, comprising: using the relationship between the grayscale values of the pixels in the image and the corresponding incident angle to determine the three-dimensional shape of the semiconductor substrate.
4. The method of claim 1, comprising: filtering each group in the plurality of image groups to determine the central tendency grayscale value for each pixel of the image.
5. The method of claim 1, comprising: generating a shape drawing of the semiconductor substrate; and using the shape drawing to generate a plurality of incident angles.
6. The method of claim 5, wherein the plurality of incident angles is a first plurality of incident angles, and wherein the correction model is trained by: generating a training profile of the semiconductor substrate; and using the training profile to generate a second plurality of incident angles, wherein the second plurality of incident angles includes more incident angles than the first plurality of incident angles.
7. The method of claim 1, wherein determining the correction model for each pixel of the image includes applying linear regression.
8. The method of claim 7, wherein applying the linear regression includes applying the linear regression to the images in image form.
9. A semiconductor substrate inspection system, comprising: a camera configured to generate a plurality of image groups of a semiconductor substrate, wherein each of the plurality of image groups is associated with a corresponding angle range of incidence of the camera of the semiconductor substrate inspection system relative to the semiconductor substrate; and a processor configured to: determine a correction model for each pixel of the image, wherein the correction model is configured to correct for light response inhomogeneities of an image sensor based on a measurement of the three-dimensional shape of the semiconductor substrate; and use the correction model to generate a corrected image of the semiconductor substrate.
10. The semiconductor substrate inspection system of claim 9, wherein the processor configured to determine the correction model for each pixel of the image is configured to: determine, for each pixel, the relationship between the grayscale value of the pixel and the corresponding incident angle.
11. The semiconductor substrate inspection system of claim 10, wherein the processor is further configured to: determine the three-dimensional shape of the semiconductor substrate by using the relationship between the grayscale values of the pixels in the image and the corresponding incident angle.
12. The semiconductor substrate inspection system of claim 9, wherein the processor is further configured to: filter each group in the plurality of image groups to determine the central tendency grayscale value for each pixel of the image.
13. The semiconductor substrate inspection system of claim 9, wherein the processor is further configured to: generate a shape drawing of the semiconductor substrate; and use the shape drawing to generate a plurality of incident angles.
14. The semiconductor substrate inspection system of claim 13, wherein the plurality of incident angles is a first plurality of incident angles, and wherein the calibration model is trained by: generating a training profile of the semiconductor substrate; and using the training profile to generate a second plurality of incident angles, wherein the second plurality of incident angles includes more incident angles than the first plurality of incident angles.
15. The semiconductor substrate inspection system of claim 9, wherein the processor configured to determine the correction model for each pixel of the image is configured to apply linear regression.
16. The semiconductor substrate inspection system of claim 15, wherein the processor configured to apply the linear regression is configured to apply the linear regression to the images in image form.
17. A semiconductor substrate inspection system comprising: a component for generating a plurality of image groups of a semiconductor substrate, wherein each of the plurality of image groups is associated with a corresponding angle range of incidence of a camera of the semiconductor substrate inspection system relative to the semiconductor substrate; a component for determining a correction model for each pixel of the image, wherein the correction model is configured to correct for light response inhomogeneities of an image sensor based on a measurement of the three-dimensional shape of the semiconductor substrate; and a component for using the correction model to generate a corrected image of the semiconductor substrate.
18. The semiconductor substrate inspection system of claim 17, wherein the component for determining the correction model for each pixel of the image includes: A component is used to determine the relationship between the grayscale value of a pixel and the corresponding incident angle for each pixel.
19. The semiconductor substrate inspection system of claim 18, comprising: a component for determining the three-dimensional shape of the semiconductor substrate by using the relationship between the grayscale value of a pixel in an image and the corresponding incident angle.
20. The semiconductor substrate inspection system of claim 17, comprising: a component for filtering each group in the plurality of image groups to determine the central tendency grayscale value for each pixel of the image.
Citation Information
Patent Citations
Mura correction system
TW202025132A
Contour based defect detection
US20180293721A1
Training methods for machine learning assisted optical proximity error correction
US20200356011A1
