JP2024087006A - Method and apparatus for judging defect quality - Google Patents

Abstract

To correctly determine the quality of a defect that has occurred by correctly measuring the occurrence position in the height direction of a three-dimensional structure and determining the relative height with a pattern of a background.SOLUTION: A method for judging defect quality includes: acquiring plural images with a predetermined step in a height direction by optical imaging means (22) to an inspection subject (10) which includes multilayer transparent thin films (1, 2, 3, 4, 5, 6); calculating sharpness of partial images from luminance differences to adjacent pixels against each pixel of the plural images; calculating height information of the partial images from an image number which a calculating result of the sharpness at a same pixel position is maximum over all images of the plural images; obtaining three-dimensional information of all the images from calculating the height position; and judging the defect quality of the inspection subject based on the three-dimensional information.SELECTED DRAWING: Figure 4B

Description

The present invention relates to a defect quality determination method and device that uses an optical imaging means to measure height information of defects that occur during the manufacture of inspection objects such as semiconductor wafers and thin-film transistor display devices that use multi-layer thin films, and identifies the height information of the defect-occurring portion and the layer in which the defect occurred to determine the quality of the defect.

In the manufacture of semiconductors, thin-film transistor displays, and similar devices that use multiple thin films, minute patterns are formed using photolithography. During these manufacturing processes, defects such as pattern anomalies and pinholes can occur due to a variety of factors, resulting in reduced yields. In order to eliminate factors that reduce yields and increase production efficiency, work is being carried out to manage these manufacturing processes, inspect the defects that occur, and identify the causes.

In such manufacturing processes, in processes in which multiple layers are formed continuously, there are cases in which it is not possible to inspect for pattern abnormalities, pinholes, and similar defects during the process of forming each layer. Therefore, the only way is to inspect for defects after the formation of the final layer and identify the process that caused the defect from the height position information of the defect. For example, in the sealing process of flexible organic EL (Electro-Luminescence), a technology is used to prevent oxygen and moisture in the atmosphere from entering the device by laminating inorganic thin films such as silicon nitride and organic thin films such as polyimide in multiple layers. However, the presence of tiny pinholes that occur in each layer is fatal to the life of the device. In particular, it is necessary to accurately measure the positions of pinholes that occur in different layers and are close to each other to judge the quality of the product. However, these layers must be formed in a vacuum or nitrogen atmosphere in a short time, and it is not possible to stop the process and inspect the target object midway.

Figure 1 shows the cross-sectional structure of a typical flexible organic EL display device, where the organic EL emits light by an electric circuit pattern 1. The electric circuit pattern 1 is formed on a substrate consisting of a first substrate 5 and a second substrate 6, and is sealed by a transparent film consisting of a first sealing layer (inorganic film) 2, a second sealing layer (organic film) 3, and a third sealing layer (inorganic film) 4. Usually, the first sealing layer 2 and the third sealing layer 4 are silicon nitride films, which are inorganic materials, and are formed by a chemical vapor deposition (CVD) process. The second sealing layer 3 is made of polyimide, which is an organic material, and is formed, for example, by using an inkjet printing device. The first substrate 5 is a transparent substrate. For example, a resin substrate such as polyethylene terephthalate (PET) or polycarbonate (PC) is used as the first substrate 5.

2 shows a state where two pinholes 7A and 7B are formed in the sealing layer 4 during the process of forming a sealing layer (e.g., a transparent film for sealing made of an organic or inorganic material). The pinholes 7A and 7B are present in the silicon nitride film of the third sealing layer 4. Even if oxygen (O ₂ ) or water (H ₂ O) in the air permeates the polyimide film, the oxygen (O ₂ ) or water (H ₂ O) in the air is blocked by the first sealing layer 2 (silicon nitride film). The EL elements directly below the pinholes 7A and 7B are not immediately destroyed, and the pinholes 7A and 7B are not a fatal defect for the organic EL display device.

On the other hand, FIG. 3 similarly shows a state in which defective pinholes 8A and 8B are present in adjacent (vertical) locations of the silicon nitride films of the first sealing layer 2 and the third sealing layer 4 during the process of forming the sealing layer of a flexible organic EL display device. In this case, oxygen and water that have entered through pinhole 8A in sealing layer 4, which is in contact with air, permeate the polyimide film (sealing layer 3) over time. Eventually, the oxygen and water reach pinhole 8B in sealing layer 2, destroying the EL element directly below pinhole 8B. The destruction of the EL element in this way, shortening the lifespan of the organic EL display device, is fatal to the display device.

As explained using the example of an organic electroluminescence display, it is necessary to determine whether defects (e.g., pinholes and foreign matter) on the sealing layer of a display device are fatal to the display device as shown in FIG. 3. However, since the thickness of the sealing layer is about a few μm, in order to identify the transparent sealing film in which the defect has occurred, it is necessary to measure the height of the defect with a resolution on the order of submicrons. Known technologies that meet such precise measurement accuracy include, for example, a distance measuring device using a laser triangulation method, a white light interferometer (JP Patent Publication No. 2013-19767 (Patent Document 1)), a confocal microscope (JP Patent Publication No. 2012-237647 (Patent Document 2)), and similar devices.

Japanese Patent Publication No. 2014-148735 (Patent Document 3) discloses a multifocal confocal Raman spectroscopic microscope, in which Raman scattered light from a sample is observed using a laser observation optical system. Furthermore, Japanese Patent Publication No. 2012-220490 (Patent Document 4) discloses a technique for creating a three-dimensional profile map by obtaining multiple images of a three-dimensional sample according to the focal length using a camera device such as an optical microscope, combining these images to evaluate the degree of focus at which the light intensity and brightness contrast are maximized, and generating a three-dimensional profile map (height map) of the sample. Furthermore, Japanese Patent Publication No. 2005-172805 (Patent Document 5) discloses a technique for moving a Z revolver at a moving pitch ΔZ in a scanning confocal microscope, and generating brightness and height information (three-dimensional information) about the sample based on the maximum intensity point of the confocal image acquired for each Z relative position.

JP 2013-19767 A JP 2012-237647 A JP 2014-148735 A JP 2012-220490 A JP 2005-172805 A Japanese Patent Application Laid-Open No. 11-337313

Height measurement technologies such as distance measuring devices using laser triangulation, white light interferometers, and confocal microscopes can measure the height of flat pattern defects of 10 μm or more. However, they cannot accurately measure the height information of defects of a few μm in size (e.g., pinholes and foreign objects) or foreign objects that are not flat. In particular, they cannot measure defects of 1 μm or less.

Measurement methods that rely on interference between light reflected by an object and a reference light, or detection methods that use a confocal optical system to focus the light reflected by an object, are not capable of detecting defects that are 1 μm or smaller in size. If the size of the object is 1 μm or smaller, or if the object has an uneven surface that scatters the illumination light used for observation rather than reflecting it specularly, the reflected light itself cannot be observed and defects cannot be detected. Furthermore, conventional height measurement devices cannot detect defects that are 1 μm or smaller.

The laser observation optical system disclosed in Patent Document 3 does not have the functionality to inspect defects in objects such as semiconductor wafers and thin-film transistors, and in particular the sealing layers (sealing transparent films) of flexible organic EL display devices, and to calculate height information of the defects.

Furthermore, Depth From Defocus (DFD) or Depth From Focus (DFF) is known, which acquires images while changing the focal position and extracts height information of an object based on the amount of change in brightness in areas where the brightness changes sharply (JP Patent Publication No. 11-337313 (Patent Document 6)). However, in conventional DFD processing as shown in Patent Document 6, due to errors caused by misalignment between images and pixel noise in images, height information cannot be accurately measured for objects with an imaging pixel size of about one pixel, such as pattern defects with a size of 1 μm or less or minute defects on a layer (pinholes and foreign objects), due to errors caused by image misalignment and pixel noise. The absolute position in the height direction in measuring height information using DFD or DFF depends on the accuracy of the machine connecting the substrate and the microscope.

Currently, in order to increase manufacturing efficiency, G6 size (1500 x 1850 mm) glass substrates are used in display devices that use EL elements. And because the stage on which the substrate is loaded (mounted) inside the inspection device becomes larger to accommodate the G6 size, it is not realistic to adjust the absolute height position of the substrate surface to within 1 μm. For this reason, it is very difficult to measure the height information (absolute height position) of defects that occur in patterns formed on glass substrates in 1 μm units.

The techniques disclosed in Patent Documents 4 and 5 generate a three-dimensional profile map (height map) of a sample, but do not have the functionality to calculate height information of defects in transparent sealing films of flexible organic electroluminescence display devices and the like using a resolution of μm or less.

The present invention was made based on the above-mentioned circumstances, and the object of the present invention is to provide a defect quality determination method and device that can accurately determine the quality of a defect by accurately measuring the position in the height direction of a three-dimensional structure where the defect occurs, even for defects such as pattern defects less than 1 μm in size or minute pinholes in a film (layer), and determining the height relative to the background pattern.

The above object of the present invention is achieved by including the following steps. That is, the object of the present invention is achieved by including: acquiring a plurality of images of an object to be inspected, which is used for a multi-layer transparent thin film, by an optical imaging means, at a predetermined interval in the height direction; calculating the sharpness of a partial image from the brightness difference between adjacent pixels for each pixel of the plurality of images; calculating height information of the partial image from the image number which has the maximum sharpness calculation result at the same pixel position among all the images of the plurality of images; obtaining three-dimensional information of the entire image from the calculation of the height information; and judging the defect quality of the object to be inspected based on the three-dimensional information.

The above object of the present invention is efficiently achieved by further including the following steps. That is, by further including the steps of detecting a pattern defect in the image with the maximum sharpness, extracting an image with the maximum density of partial images with high sharpness from the multiple images, setting the image as a reference position 1 in the height direction of a three-dimensional pattern structure, and measuring the height of the pattern defect in the three-dimensional pattern structure generated from the relationship between the height information of the pattern defect and the reference position 1, or by further including the steps of detecting a pattern defect in the image with the maximum sharpness, extracting an image in the multiple images in which an interference image of interference fringes generated at the edge of the transparent thin film is the sharpest, setting the image as a reference position 2 in the height direction of a three-dimensional pattern structure, and measuring the height of the pattern defect in the three-dimensional pattern structure generated from the relationship between the height information of the pattern defect and the reference position 2, or by further including the steps of repairing the pattern defect using the height information of the pattern defect.

The above object of the present invention is achieved by providing the following configuration. That is, the object of the present invention is achieved by providing an imaging means that acquires multiple image data of an object having a multi-layer transparent thin film by using an optical imaging means that can move up and down at a predetermined interval, and assigns image numbers to the image data; an extraction unit that extracts features of the image data; an evaluation value calculation unit that calculates an evaluation value based on the features; an evaluation value comparison unit that compares the evaluation value with a previous evaluation value that matches the positional relationship with the evaluation value in the image data and generates a comparison result; an evaluation value storage unit that stores the evaluation value based on the comparison result; an image number storage unit that stores the image number based on the comparison result; a three-dimensional information extraction unit that extracts three-dimensional information of the object based on the image number stored in the image number storage unit; a three-dimensional information extraction unit that extracts height information of defects present in the object based on the three-dimensional information; and a pass/fail judgment unit that judges the pass/fail of the object based on the difference in height information between the defects when there are multiple defects.

The above object of the present invention is efficiently achieved by providing the following configuration. That is, the three-dimensional information extraction unit extracts the three-dimensional information based on the image number with the highest evaluation value, or the evaluation value is calculated based on the luminance difference between the pixel of interest and the adjacent pixel adjacent to the pixel of interest, or the three-dimensional information extraction unit determines the standard of the height information based on the evaluation value of the image data in which the electrode pattern of the inspection object and the interference fringes of the sealing layer of the inspection object are photographed, or the defect is a pattern defect, a pinhole, or a foreign object, or the evaluation value is a sharpness calculated based on the difference in luminance value of the pixel of interest relative to the luminance value of the adjacent pixel, or the inspection object is an organic EL display device, or the inspection object is a flexible organic EL display device formed on a flexible substrate, or the object further includes at least one function of repairing the defect based on the height information calculated by the three-dimensional information extraction unit, or further includes a function of selecting the function according to the height information.

According to the present invention, the height of minute patterns and foreign objects is observed not only by the reflected light from the patterns themselves, but also by observing the reduction in observation light passing through the edge of the pattern or near the foreign object. This makes it possible to accurately measure the height information of pinholes and foreign objects with a diameter of about 1 μm, which was impossible to measure with conventional technology.

In the present invention, even if horizontal positional errors occur between the acquired images due to vibrations of the device itself or the floor in the process of acquiring and processing multiple images, the height information of any minute defects that occur can be measured accurately and without error by taking into account the horizontal amplitude between the images. When comparing the image sharpness of pattern edges or foreign objects between images, the sharpness values between adjacent pixel positions of the pixel of interest are compared, and the image numbers stored around them are replaced with the image number with the highest sharpness. In this way, the vibration resistance characteristics of the entire device can be improved, and the above process contributes greatly to reducing the cost of inspection devices.

In addition, according to the present invention, in the manufacture of thin-film, multi-layer devices, it is possible to identify layers in which pinholes with diameters of approximately 1 μm occur, which were impossible to measure using conventional techniques, and therefore it is possible to select the most suitable repair method for the material of the film that forms that layer.

FIG. 1 is a cross-sectional view showing an example of the structure of a typical flexible organic EL display device. FIG. 11 is a cross-sectional view showing a state in which a pinhole is formed in the same layer in the sealing layer of the flexible organic EL display device. 1 is a cross-sectional view showing a state in which pinholes are formed in a first sealing layer and a third sealing layer of a flexible organic EL display device. 1 is a block diagram showing a portion of a configuration example of a defect pass/fail determination device according to the present invention; FIG. 4B is a block diagram showing a part of an example of the configuration of a defect quality determination device according to the present invention, and is connected to FIG. 4A to show the entire block diagram. 4 is a flowchart showing an operation example of the present invention. FIG. 11 is a perspective view showing a state in which two minute foreign bodies are generated in one pixel of a flexible organic EL device. FIG. 2 is a cross-sectional view showing the shape of a light-emitting portion of a flexible organic EL device. FIG. 2 is a plan view showing a state in which the flexible organic EL substrate is observed from a direction perpendicular to the flexible organic EL substrate. 10 is a flowchart showing a detailed example of an imaging operation. 11 is a flowchart showing a portion of a flowchart illustrating details of an example of an operation for extracting a pattern edge of an image. FIG. 10B is a diagram showing a part of a flowchart showing details of an example of an operation for extracting a pattern edge of an image, and is linked to FIG. 10A to show the entire flowchart. 13 is a flowchart showing a portion of the flowchart illustrating details of an example of an operation for extracting a minute foreign matter image from an image. FIG. 11B is a diagram showing a part of a flowchart showing details of an example of an operation for extracting a minute foreign substance image from an image, and is linked to FIG. 11A to show the entire flowchart. 11 is a flowchart showing a portion of a flowchart illustrating details of an example of an operation for extracting three-dimensional information from an image. 12B is a flowchart showing a portion of a flowchart illustrating an example of an operation for extracting three-dimensional information from an image in detail, and is linked to FIG. 12A to show the entire flowchart. FIG. 11 is an exploded perspective view showing the first, tenth, thirtieth and fortieth images out of 40 captured images arranged in the height direction. FIG. 13 is a diagram showing array data of luminance values (gray values) of a first image. FIG. 13 is a diagram showing array data of luminance values (gray values) of the tenth image. FIG. 13 is a diagram showing array data of luminance values (gray values) of the 30th image. FIG. 13 is a diagram showing array data of luminance values (gray values) of the 40th image. FIG. 13 is a diagram showing edge evaluation values E(i, j) of a first image. FIG. 13 is a diagram showing the edge evaluation value E(i, j) of the tenth image. FIG. 13 is a diagram showing the edge evaluation value E(i, j) of the 30th image. FIG. 13 is a diagram showing the edge evaluation value E(i, j) of the 40th image. FIG. 13 is a diagram showing a state in which an edge evaluation storage value EM(i,j) is updated using an edge evaluation value E(i,j) of a first image. FIG. 13 is a diagram showing a state in which the edge evaluation storage value EM(i, j) is updated using the edge evaluation value E(i, j) of the tenth image. FIG. 13 is a diagram showing a state in which the edge evaluation storage value EM(i, j) is updated using the edge evaluation value E(i, j) of the 30th image. FIG. 13 is a diagram showing a state in which the edge evaluation storage value EM(i, j) is updated using the edge evaluation value E(i, j) of the 40th image. FIG. 13 is a diagram showing a state in which the foreign matter evaluation storage value FM(i,j) is updated using the foreign matter evaluation value F(i,j) of the 30th image. FIG. 13 is a diagram showing a state in which the foreign matter evaluation storage value FM(i,j) is updated using the foreign matter evaluation value F(i,j) of the 40th image. FIG. 13 is a diagram showing a state in which the image number in the edge image number storage unit is updated after the evaluation value update for the first image is completed. FIG. 13 is a diagram showing how the image number in the edge image number storage unit is updated after the evaluation value update for the tenth image is completed. FIG. 13 is a diagram showing how the image number in the edge image number storage unit is updated after the evaluation value update for the 30th image is completed. FIG. 13 is a diagram showing how the image number in the edge image number storage unit is updated after the evaluation value update for the 40th image is completed. FIG. 13 is a diagram showing how the image number in the foreign object image number storage unit is updated after the evaluation value update for the 30th image is completed. FIG. 13 is a diagram showing how the image number in the foreign object image number storage unit is updated after the evaluation value update for the 40th image is completed. FIG. 13 is a diagram showing three-dimensional information of a sample (flexible organic EL device) by a contour line display. FIG. 2 is a schematic side view of the inspection device. 36 is a schematic plot of vibrations measured with the apparatus of FIG. 35 in accordance with an embodiment of the present invention; 37 is a plot that illustrates a schematic of focus quality measurements made by the apparatus of FIG. 36 in accordance with an embodiment of the present invention; 1 is a flow chart that illustrates a schematic diagram of a method for mapping features in a sample, according to an embodiment of the present invention;

In the present invention, in the manufacture of semiconductors, display devices, and the like using multi-layer transparent thin films, the height information of minute pattern defects that occur during film formation is measured by an optical inspection means. The height information of the defect occurrence site and the type of film in which the defect occurs are identified, and the quality of the defect is judged. More specifically, the microscope imaging device has a mechanism for mechanically scanning the focal position in the height direction, and while scanning in the height direction, it continuously captures and stores multiple images, and calculates the contrast difference between adjacent pixels of the image information as a numerical evaluation value. By judging the magnitude of the obtained evaluation value between the images for each pixel, the number of the sharpest image of the pattern edge image is selected, and the image number is converted to the vertical height position where the image was obtained and measured as the vertical height of the corresponding image part. The position that serves as the reference for the height is obtained from the density of the maximum contrast evaluation value or the interference fringe image generated at the edge of the transparent film. The evaluation value is also calculated by a similar method for images of defect points extracted as minute image points such as pattern abnormalities, pinholes, and foreign objects, and the height of the defect point is measured based on the relative positional relationship with the reference height. The layer where the defect occurred is identified based on the vertical height at which the defect occurred, and the quality of the defect is determined.

An embodiment of the present invention will be described below with reference to the drawings.

First, an example of the configuration of an embodiment of the present invention will be described with reference to Figures 4A and 4B.

In this embodiment, the flexible organic EL display device 10 is the object to be inspected, and the display device 10 is mounted on a predetermined stage (not shown) and installed below the microscope 20. An objective lens 21 is attached to the object to be inspected side of the lens barrel of the microscope 20, and an image capturing camera 21 is attached to the opposite side. The sequence control unit 30 controls the height direction drive motor 23 and the image acquisition unit 31. The height direction drive motor 23 is connected to the microscope 20 via a rack and pinion or the like, and the microscope 20 moves up and down when the motor 23 is driven by the sequence control unit 30. The camera 22 continuously captures the flexible organic EL display device 10, and the image acquisition unit 31 acquires image data from the camera 22 in response to an instruction from the sequence control unit 30. The image memory 32 stores the images sent from the image acquisition unit 31. The sequence control unit 30 controls the height of the microscope 20 in a predetermined increment (up and down amount) via the height direction drive motor 23, adjusts the focal position of the objective lens 21, and can capture the display device 10 to be inspected. The predetermined increment is the resolution in the height direction, so the smaller the predetermined increment, the more images can be acquired for each height direction within the measurement range. Conversely, the larger the predetermined increment, the fewer images can be acquired for each height direction within the measurement range. By adjusting the predetermined increment, the resolution of the image in the height direction can be adjusted. In this way, when the microscope 20 moves in the height of the measurement range, all images corresponding to the predetermined increment are stored in the image memory 32. When the optical system of the microscope 20 is configured as an infinity correction optical system, the motor 23 may drive only the objective lens 21 up and down instead of driving the microscope 20 up and down.

The data stored in the image memory 32 is processed by a judgment processing section, which will be described below. The judgment processing section is composed of an edge processing section 40 which extracts pattern edges, a foreign matter processing section 50 which extracts and processes minute foreign matters, and a pass/fail judgment section 60 which judges the pass/fail quality of defects based on the three-dimensional information ED from the edge processing section 40 and the three-dimensional foreign matter information FM from the foreign matter processing section 50.

The edge processing unit 40 is composed of a pattern edge extraction unit 41 that extracts pattern edges, an edge evaluation value calculation unit 42 that calculates edge evaluation values, an edge evaluation value comparison unit 43 that compares edge evaluation values, an edge evaluation value memory unit 44 that stores edge evaluation values, an edge image number memory unit 45 that stores edge image numbers (including symbols and the like), and an edge 3D information extraction unit 46 that extracts edge 3D information ED based on the information in the edge evaluation value memory unit 44 and the edge image number memory unit 45. The foreign matter processing unit 50 is made up of a minute foreign matter extraction unit 51, a foreign matter evaluation value calculation unit 52 that calculates a foreign matter evaluation value, a foreign matter evaluation value comparison unit 53 that compares foreign matter evaluation values, a foreign matter evaluation value storage unit 54 that stores foreign matter evaluation values, a foreign matter image number storage unit 55 that stores foreign matter image numbers (including symbols and the like), and a foreign matter 3D information extraction unit 56 that extracts foreign matter 3D information FM based on the information in the foreign matter evaluation value storage unit 54 and the foreign matter image number storage unit 55.

In such a configuration, an example of the operation is shown in the flowchart in FIG. 5. First, the microscope 20 is driven by the sequence control unit 30 to capture an image of the flexible organic EL display device 10, which is the object to be inspected (step S100). Next, the edge processing unit 40 extracts the pattern edge of the image (step S200), and the foreign matter processing unit 50 extracts minute foreign matters from the image (step S300). The order of the edge extraction process and the foreign matter extraction process may be changed. The edge three-dimensional information extraction unit 46 in the edge processing unit 40 and the foreign matter three-dimensional information extraction unit 56 in the foreign matter processing unit 50 perform the extraction process of three-dimensional information (step S400). The three-dimensional information ED from the edge three-dimensional information extraction unit 46 and the foreign matter three-dimensional information FM from the foreign matter three-dimensional information extraction unit 56 are input to the pass/fail judgment unit 60, which judges the pass/fail of the defect (step S500).

First, we will explain how to extract edge pixels of patterns, defects, etc. from images acquired within the measurement range.

In principle, an evaluation value for a pixel of interest is calculated based on the difference in luminance between the pixel of interest and the adjacent pixels adjacent to the pixel of interest in the image information. The evaluation value is then compared with a predetermined reference threshold value to determine the degree of defocus for the partial image around the pixel of interest. Based on the result of the defocus degree determination, the device determines whether the pixel of interest is an edge pixel of a pattern, defect, or the like. If the partial image is not defocused, the partial image has a high degree of sharpness. If the partial image is defocused, the partial image has a low degree of sharpness. As will be described later, a similar method is used to determine the degree of defocus for pixels that may contain foreign matter, even when extracting minute foreign matter pixels.

Examples of patterns and defects that exist in an image as image information and are subject to extraction include the electrode pattern of the flexible organic EL display device 10, organic films, inorganic films and similar patterns, and pinholes and similar defects. In order to evaluate a pixel of interest, a variable representing the pixel is defined as follows. That is, when the position of each pixel that constitutes an image is horizontal position i and vertical position j in the image, the gray value (brightness value) of any pixel in the image is represented by G(i,j).

Next, a method of calculating the edge evaluation value E(i,j) using a function will be described. The functions used are a function MAX(X,Y) which compares X and Y and outputs the larger value, and a function ABS(X) which outputs the absolute value of X. Using these functions, the edge evaluation value E(i,j) can be calculated. Then, the edge evaluation value E(i,j) can be compared with an edge threshold value to extract edge pixels (edge portion images) of a pattern. If the calculated edge evaluation value E(i,j) is greater than the edge threshold value, the pixel having G(i,j) is considered to be an edge pixel. If the calculated edge evaluation value E(i,j) is not greater than the edge threshold value, it is considered not to be an edge pixel. More specifically, the edge evaluation value E(i,j) is given as in Equation 1. Then, whether or not a pixel having G(i,j) is an edge pixel is determined using Equation 2.
[Equation 1]
Edge evaluation value E(i,j)=MAX(MAX(ABS(G(i,j)-G(i+1,j)), ABS(G(i,j)-G(i,j+1))),
MAX(ABS(G(i,j)-G(i-1,j)), ABS(G(i,j)-G(i,j-1)))
[Equation 2]
Edge evaluation value E(i, j)>edge threshold

The process of the edge processing unit 40 for extracting edge pixels of patterns, defects, etc. is performed as follows.

First, the pattern edge extraction unit 41 acquires from the image memory 32 the gray value G(i,j) of an arbitrary pixel in the image, the gray values G(i-1,j) and G(i+1,j) of pixels adjacent to the arbitrary pixel in the horizontal direction, and the gray values G(i,j-1) and G(i,j+1) of pixels adjacent to the arbitrary pixel in the vertical direction. Then, the edge evaluation value calculation unit 42 calculates the edge evaluation value E(i,j) using the calculation method shown in Equation 1, and compares the edge evaluation value E(i,j) with the edge threshold value. If the edge evaluation value E(i,j) is greater than the edge threshold value as a result of the comparison, the pixel having G(i,j) is considered to be an edge pixel. If the edge evaluation value E(i,j) is not greater than the edge threshold value, the pixel having G(i,j) is not considered to be an edge pixel.

The edge evaluation value comparison unit 43 compares the next acquired edge evaluation value E(i,j) with the edge evaluation storage value EM(i,j) stored in the edge evaluation value storage unit 44 at the corresponding position (horizontal position i, vertical position j). The edge evaluation storage value EM(i,j) is the previous edge evaluation value E(i,j) acquired before the edge evaluation value E(i,j) is acquired subsequently. If the comparison result shows that the edge evaluation value E(i,j) is greater than the edge evaluation storage value EM(i,j), the edge evaluation value storage unit 44 rewrites the edge evaluation storage value EM(i,j) to the edge evaluation value E(i,j). When rewriting is performed, the edge image number storage unit 45 updates the edge image number EN(i,j), which is the corresponding position element, to the image number currently being processed, and associates the edge evaluation value E(i,j) with the image number. The number does not have to be a number, but may be a symbol that can be distinguished from others. In this way, all pixels of the image number currently being processed are sequentially judged as to whether they are edge pixels. Depending on the judgment, the edge evaluation value EM(i,j) and edge image number EN(i,j) are rewritten to the edge evaluation value E(i,j) and image number N currently being processed, respectively.

When the above process is completed for all pixels in the image with image number N currently being processed, a similar process is performed on the image with the next image number. Then, when the process is completed for all images with image numbers, the edge 3D information extraction unit 46 generates height information for edge pixels of patterns, defects, etc., based on the image numbers EN(i,j) stored in the edge image number storage unit 45.

Next, we will explain the foreign matter processing unit 50, which extracts images of minute foreign matters and extracts three-dimensional information about the minute foreign matters. The process of extracting images of minute foreign matters may be performed after the process of generating pattern edge height information, may be performed before the process, or may be performed simultaneously in parallel.

As with the process of extracting edge pixels of patterns and defects, if we assume that the position of each pixel that composes an image is horizontal position i and vertical position j in the image, the gray value (brightness value) of any pixel in the image is expressed as G(i,j). The function used in the method of calculating the foreign matter evaluation value F(i,j) is the function MIN(X,Y), which compares X and Y and gives the smaller value. Using this function, the foreign matter evaluation value F(i,j) can be calculated.

By comparing the foreign matter evaluation value F(i,j) with the micro foreign matter threshold, micro foreign matter pixels (micro foreign matter partial images) of the pattern can be extracted. If the calculated foreign matter evaluation value F(i,j) is greater than the micro foreign matter threshold, the pixel having G(i,j) is deemed to be a micro foreign matter pixel, and if the calculated foreign matter evaluation value F(i,j) is not greater than the micro foreign matter threshold, the pixel having G(i,j) is deemed not to be a micro foreign matter pixel. More specifically, the foreign matter evaluation value F(i,j) is given by Equation 3. Then, whether or not the pixel having G(i,j) is a micro foreign matter pixel is determined using Equation 4.
[Equation 3]
Foreign matter evaluation value F(i, j)
=MIN(G(i-1,j)+G(i+1,j)-G(i,j)*2,G(i,j-1)+
G(i,j+1)-G(i,j)*2)
[Equation 4]
Foreign matter evaluation value F(i, j)>micro-foreign matter threshold

By using Equation 4, it is possible to extract pixels that are lower in luminance than the surrounding pixels in both the horizontal and vertical directions on the image, that is, dark spots with a size of about one pixel.

The process of the foreign matter processing unit 50 to extract minute foreign matter pixels is performed as follows:

First, the minute foreign matter extraction unit 51 obtains from the image memory 32 the gray value G(i,j) of an arbitrary pixel in the image, the gray values G(i-1,j) and G(i+1,j) of arbitrary horizontally adjacent pixels, and the gray values G(i,j-1) and G(i,j+1) of arbitrary vertically adjacent pixels. The foreign matter evaluation value calculation unit 52 then calculates the foreign matter evaluation value F(i,j) using the calculation method shown in Equation 3, and compares the foreign matter evaluation value F(i,j) with the minute foreign matter threshold value. If the comparison result shows that the minute foreign matter evaluation value F(i,j) is greater than the foreign matter threshold value, the pixel having G(i,j) is deemed to be a minute foreign matter pixel, and if the minute foreign matter evaluation value F(i,j) is not greater than the foreign matter threshold value, the pixel having G(i,j) is not deemed to be a minute foreign matter pixel.

The foreign matter evaluation value comparison unit 53 compares the next acquired foreign matter evaluation value F(i,j) with the foreign matter evaluation memory value FM(i,j) of the corresponding position (horizontal position i, vertical position j) stored in the foreign matter evaluation value memory unit 54. The foreign matter evaluation memory value FM(i,j) is the foreign matter evaluation value F(i,j) acquired before the next acquisition of the foreign matter evaluation value F(i,j). If the comparison result shows that the foreign matter evaluation value F(i,j) is greater than the foreign matter evaluation memory value FM(i,j), the foreign matter evaluation value memory unit 54 rewrites the foreign matter evaluation memory value FM(i,j) to the next acquired foreign matter evaluation value F(i,j). When the foreign substance evaluation storage value FM(i,j) is rewritten, the foreign substance image number storage unit 55 updates the foreign substance image number FN(i,j), which is the element at the corresponding position, to the image number N currently being processed, and associates the foreign substance evaluation value F(i,j) with the image number N.

All pixels of the image number N currently being processed are sequentially judged as to whether they are foreign object pixels. Depending on the judgment, the foreign object evaluation memory value FM(i,j) and foreign object image number FN(i,j) are rewritten to the edge evaluation value E(i,j) and image number N currently being processed, respectively. Then, when processing of all pixels in the image currently being processed with image number N has been completed, the same processing is performed on the image with the next image number (N+1).

When processing for all images with image numbers is completed, the foreign object 3D information extraction unit 56 generates height information for the micro foreign object based on the foreign object image number FN(i, j) stored in the foreign object image number storage unit 55.

The pass/fail judgment unit 60 judges the pass/fail quality of the sample flexible organic EL display device 10 based on the relative height between the edge 3D information ED (height information of the edge pixels of the pattern) from the edge 3D information extraction unit 46 and the foreign matter 3D information FM (height information of minute foreign matters) from the foreign matter 3D information extraction unit 56, and outputs the judgment result.

First, as an example of an inspection target (sample), FIG. 6 shows two minute foreign bodies 101 and 102 occurring in one pixel of a flexible organic EL display device 10. As described later, the first minute foreign body 101 is located on the plane 10-30 on which the 30th image is focused, and the second minute foreign body 102 is located on the plane 10-40 on which the 40th image is focused. The reference plane for height is the electrode pattern 103. In this embodiment, since images are acquired every 0.1 μm, the minute foreign body 101 is detected to be located 2.0 μm above the pattern, and the minute foreign body 102 is detected to be located 3.0 μm above the pattern.

Next, the height standard of the three-dimensional information (structure) will be described. As a structure in which the pattern edge of the flexible organic EL circuit is difficult to detect, the shape of the light-emitting layer 107 may be a rectangle (window) formed by the organic film 105 covering the cathode electrode 104, as shown in FIG. 7. In the following description, the light-emitting part is referred to as a window. The illumination light is reflected by the interface between the organic film 105 and the transparent film 106 that form the window. The illumination light using the coaxial epi-illumination forms a ring-shaped interference pattern around the window. The interference pattern can be observed using image processing. The innermost interference pattern 107A is the end (edge) of the window 107 of the organic film 105. As a result, the edge evaluation value E(i, j) detected in the area with this edge is suitable as the height standard in the three-dimensional information (structure). That is, the pattern edges of the circuit of the flexible organic EL substrate are difficult to detect due to its structure, but the end (edge) of the window 107 of the organic film 105 is suitable as a reference for the height of the three-dimensional information (structure). Furthermore, when the flexible organic EL substrate is observed from the vertical direction of the flexible organic EL substrate (for example, using a microscope), the cathode electrode 104 acts like a mirror, and strong ring-shaped interference fringes are observed around the window due to the illumination light reflected in the opposite direction at the interface between the organic film 105 that forms the window and the transparent film 106. This is shown in FIG. 8. FIG. 8 shows an edge image 104A of the cathode electrode 104, an edge image 107A of the organic EL light-emitting layer 107, and an interference image 107B that occurs at the end of the organic film 105 that forms the window.

The details of each operation described above will be explained with reference to the flowchart.

A detailed example of the operation of the imaging (step S100 in FIG. 5) will be described with reference to the flowchart in FIG. 9. First, the sequence control unit 30 initializes the image number (step S101), and adjusts the height of the microscope 20 to the measurement start position by driving the height direction drive motor 23 (step S102). In this state, the objective lens 10 and the camera 22 are in a focused position relationship, and the image of the display device 10 is captured via the image reading unit 31 (step S103), and the image data is stored in the image memory 32 (step S104). Next, the microscope 20 is moved in the height direction by a predetermined interval (step S105), and it is determined whether the height of the microscope 20 is within the measurement range (step S106). If the height of the microscope 20 is within the measurement range, the image of the display device 10 is repeatedly captured, and the image data is stored sequentially. If the height of the microscope 20 is outside the measurement range, the image number N currently being processed is stored in the image memory 32 as the maximum image number Nmax (step S107), and imaging is completed.

Next, the details of the pattern edge extraction operation (step S200 in FIG. 5) of the display device 10 will be described with reference to the flowcharts of FIGS. 10A and 10B. First, values indicating the image number N, the height of the lens barrel of the microscope 20, the horizontal position i and the vertical position j, and the edge pixel count EC(N) are initialized (step S201). Next, the edge evaluation value calculation unit 42 acquires the gray values G(i,j), G(i+1,j), G(i,j+1), G(i-1,j), and G(i,j-1) of the image number N currently being processed from the image memory 16 (step S202), and calculates the edge evaluation value E(i,j) from the gray values based on Equation 1 (step S203). The edge evaluation value comparison unit 43 compares the edge evaluation value E(i,j) with the edge threshold value (step S204). If the edge evaluation value E(i,j) is greater than the edge threshold, the edge evaluation value comparison unit 43 rewrites the edge evaluation storage value EM(i,j) in the edge evaluation value storage unit 44 to the calculated edge evaluation value E(i,j) (step S205), and rewrites the edge image number EN(i,j) in the edge image number storage unit 45 to the image number N currently being processed (step S206). Furthermore, the edge pixel number EC(N), which indicates the number of pixels in the image number N that are determined to be edge pixels, is counted up ("+1") (step S207), and the horizontal position i of the pixel is counted up ("+1") (step S208). Also, in the above step S204, if the edge evaluation value E(i,j) is equal to or less than the edge threshold, only the horizontal position i of the pixel is increased by one ("+1") (step S208).

Then, the horizontal position i of the pixel is compared with the maximum horizontal position imax (the horizontal position is the position of the edge of the image) (step S209), and if the horizontal position i of the pixel is equal to or greater than the maximum horizontal position imax, the vertical position j is counted up ("+1") (step S210), and the horizontal position i is initialized (step S210). If the horizontal position i of the pixel is less than the maximum horizontal position imax, the process returns to step S202, and the gray values G(i,j), G(i+1,j), G(i,j+1), G(i-1,j), and G(i,j-1) of the pixel are reacquired, and the process of extracting edge pixels of the pattern is performed.

Next, the vertical position j of the pixel is compared with the maximum vertical position jmax (step S211), and if the vertical position j of the pixel is equal to or greater than the maximum vertical position jmax, the image number N is counted up ("+1") and the vertical position j is initialized (step S212). If the vertical position j of the pixel is less than the maximum vertical position jmax, the process returns to step S202, and the gray values G(i,j), G(i+1,j), G(i,j+1), G(i-1,j), and G(i,j-1) of the pixels are similarly obtained, and the process of extracting edge pixels of the pattern is performed.

Then, the image number N is compared with the maximum image number Nmax (step S213), and if the image number N is equal to or greater than the maximum image number Nmax, the process of extracting edge pixels of the pattern is terminated. If the image number N is less than the maximum image number Nmax, the process returns to step S202, and the process of extracting edge pixels of the pattern is performed.

Next, the details of the display device 10's micro-foreign substance image extraction operation (step S300 in FIG. 5) will be described with reference to the flowcharts of FIGS. 11A and 11B. First, values indicating the image number N, the microscope height, the horizontal position i and vertical position j, and the number of foreign substance pixels FC(N) are initialized (step S301). Next, the foreign substance evaluation value calculation unit 52 obtains the gray values G(i-1,j), G(i,j), G(i+1,j), G(i,j-1), and G(i,j+1) of the image (step S302), and then calculates the foreign substance evaluation value F(i,j) according to Equation 3 (step S303). Then, the foreign substance evaluation value comparison unit 53 compares the foreign substance evaluation value F(i,j) with the micro-foreign substance threshold value (step S304). If the foreign substance evaluation value F(i,j) is greater than the micro foreign substance threshold, the foreign substance value comparison unit 53 rewrites the foreign substance evaluation storage value FM(i,j) in the foreign substance evaluation value storage unit 54 to the micro foreign substance evaluation value F(i,j) (step S305), and the foreign substance image number FN(i,j) in the foreign substance image number storage unit 55 is rewritten to the image number N currently being processed (step S306). The foreign substance pixel count FC(N), which indicates the number of pixels of image number N that have been determined to be foreign substances, is counted up ("+1") (step S307), and the horizontal position i of the pixel is counted up ("+1") (step S308). If the foreign substance evaluation value F(i,j) is equal to or less than the micro foreign substance threshold, only the horizontal position i of the pixel is incremented by one ("+1") (step S308).

Next, the horizontal position i of the pixel is compared with the maximum horizontal position imax (the horizontal position is the position of the edge of the image) (step S309), and if the horizontal position i of the pixel is equal to or greater than the maximum horizontal position imax, the vertical position j of the pixel is counted up ("+1") and the horizontal position i is initialized (step S310). If the horizontal position i of the pixel is less than the maximum horizontal position imax, the process returns to the above step S302, and the gray values G(i-1,j), G(i,j), G(i+1,j), G(i,j-1), and G(i,j+1) of the pixel are reacquired (step S302), and the extraction process of the minute foreign object image is performed. The vertical position j of the pixel is compared with the maximum vertical position jmax (step S311), and if the vertical position j of the pixel is equal to or greater than the maximum vertical position jmax, the image number N is counted up ("+1") and the vertical position j is initialized (step S312). If the vertical position j of the pixel is less than the maximum vertical position jmax, the process returns to step S302, where the pixel gray values G(i-1,j), G(i,j), G(i+1,j), G(i,j-1), and G(i,j+1) are similarly obtained, and a process of extracting a minute foreign object image is performed. The image number is compared with the maximum image number (step S313), and if the image number N is equal to or greater than the maximum image number Nmax, the process of extracting a minute foreign object image is terminated. If the image number N is less than the maximum image number Nmax, the process returns to step S302, and a process of extracting a minute foreign object image is performed.

Next, the details of the 3D information extraction operation (step S400 in FIG. 5) in the edge 3D information extraction unit 46 and the foreign object 3D information extraction unit 56 will be described with reference to the flowcharts in FIGS. 12A and 12B. Here, an example will be described in which edge processing is performed first, and then foreign object processing is performed, but the order can be changed, and both processes can also be performed simultaneously in parallel.

First, the image number N with the largest number of edge pixels is detected. First, the image number N and the maximum edge pixel number ECmax are initialized (step S401), and the edge pixel number EC(N) is obtained (step S402). Next, it is determined whether the edge pixel number EC(N) is greater than the maximum edge pixel number ECmax (step S403). If the edge pixel number EC(N) is greater than the maximum edge pixel number ECmax, the device replaces the image number ECNmax with the image number N (step S404), and the image number N is counted up ("+1") (step S405). If the edge pixel number EC(N) is equal to or less than the maximum edge pixel number ECmax, only the image number N is counted up ("+1") (step S405). Then, it is determined whether the image number N is equal to or greater than the maximum image number Nmax (step S406). If the image number N is less than the maximum image number Nmax, the process returns to step S403 and the above process is repeated. In this way, the device detects the image number with the largest number of edge pixels and detects the image number on which the electrode pattern of the display device 10 is in focus.

In the next step, the image number N with the largest number of foreign matter pixels is detected. First, the image number N, the first maximum foreign matter pixel number FCN1max, and the second maximum foreign matter pixel number FCN2max are initialized (step S407), and the foreign matter pixel number FC(N) is obtained (step S408). Next, the device determines whether the foreign matter pixel number FC(N) is greater than the first maximum foreign matter pixel number FCN1max (step S409). If the foreign matter pixel number FC(N) is greater than the first maximum foreign matter pixel number FCN1max, the device rewrites the value of the first foreign matter pixel number image number FCN1max to the value of the second foreign matter pixel number image number FCN2max, and rewrites the value of the first foreign matter pixel number image number to the image number N (step S410). The image number N is counted up ("+1") (step S411). Also, if it is determined that the foreign substance pixel count FC(N) is equal to or less than the first maximum foreign substance pixel count FCN1max, the image number N is increased by one ("+1") (step S411). After that, the device determines whether the image number N is equal to or greater than the maximum image number Nmax (step S412). If the image number N is less than the maximum image number Nmax, the process returns to step S409, and the process is repeated. In this way, the image number with the maximum foreign substance pixel count, i.e., the first foreign substance pixel count image number FCN1max, is detected, and the image number with the maximum foreign substance rewritten pixel count or the image number with the second highest number, i.e., the second foreign substance pixel count image number FCN2max, is detected. The image number on which the foreign substance is in focus is detected.

Finally, a first difference between the first foreign object pixel count image number FCN1max and the maximum edge pixel count image number ECNmax, and a second difference between the second foreign object pixel count image number FCN2max and the maximum edge pixel count image number ECNmax are calculated. Then, foreign object height information is extracted based on the first difference and the second difference (step S413).

Next, we will explain step by step how the process of extracting three-dimensional information (height information) of the sample edge and minute foreign objects based on an image actually captured of a flexible organic EL sample (one pixel).

First, the height of the microscope is changed at equal intervals (predetermined intervals, for example, 0.1 μm) in the height direction, and 40 images of the flexible organic EL (one pixel size) to be inspected are acquired. Then, the first, tenth, thirtieth, and fortieth images of the 40 acquired images are arranged in the height direction as shown in FIG. 13. As shown in FIG. 13, the first image 10-1 is an image acquired from a height of 1 μm below the electrode pattern 103. From a height of 1 μm below the electrode pattern 103, images are acquired every 0.1 μm upward. Image 10-10, in which the electrode pattern 103 is completely in focus, is acquired in the tenth image. Image 10-30, in which the first foreign object 101 is in focus, is acquired in the 30th image. Image 10-40, in which the second foreign object 102 is in focus, is acquired in the 40th image. A focused image is characterized by a large difference in brightness between the pixel of interest and the adjacent pixels, and high image sharpness. For example, interference images of samples, film defects, and the like in a focused image are present as thin lines or dots with a large brightness difference (brightness difference) compared to the surroundings, that is, as a (non-defocused) partial image with high sharpness.

Here, flexible organic EL (one pixel size) is imaged in an area with 20 pixels arranged vertically and 20 pixels arranged horizontally, and the state in which it is converted into luminance value array data is shown in Figs. 14 to 17. Figs. 14 to 17 respectively show array data of luminance values (gray values) of the first image 36, array data of luminance values (gray values) of the tenth image 37, array data of luminance values (gray values) of the 30th image 38, and array data of luminance values (gray values) of the 40th image 39. In Figs. 14 to 17, the horizontal position corresponds to horizontal position i, and the vertical position corresponds to vertical position j. The positional relationship between the image position and the array data can also be applied to the luminance value array data described later.

Next, the edge evaluation value E(i,j) of each image is calculated using Equation 1. The distribution of edge evaluation values E(i,j) in an area where 20 pixels are arranged vertically and 20 pixels are arranged horizontally on the flexible organic EL (one pixel size) of the inspection object is shown in Figures 18 to 21. Figures 18 to 21 respectively show the edge evaluation value E(i,j) of the first image 36, the edge evaluation value E(i,j) of the tenth image 37, the edge evaluation value E(i,j) of the 30th image 38, and the edge evaluation value E(i,j) of the 40th image 39. Similarly, the foreign matter evaluation value F(i,j) of each image can be calculated using Equation 3.

After the edge evaluation values E(i,j) corresponding to all image numbers (N=1 to 40) from the first to the fortieth sheets are calculated, the edge evaluation storage value EM(i,j) is updated in the evaluation value storage unit 44 as shown in Figs. 22 to 25. Figs. 22 to 25 respectively show how the edge evaluation storage value EM(i,j) is updated using the edge evaluation value E(i,j) of the first image 36, how the edge evaluation storage value EM(i,j) is updated using the edge evaluation value E(i,j) of the tenth image 37, how the edge evaluation storage value EM(i,j) is updated using the edge evaluation value E(i,j) and foreign matter evaluation value F(i,j) of the 30th image 38, and finally how the edge evaluation storage value EM(i,j) is updated using the edge evaluation value E(i,j) and foreign matter evaluation value F(i,j) of the 40th image 39.

After the edge evaluation storage value EM(i,j) is updated using the edge evaluation value E(i,j) in this manner (hereinafter referred to as "evaluation value update"), the image number N is updated in the edge image number storage unit 45 as shown in Figures 28 to 31.

In the process of detecting foreign matter, FIG. 26 shows how the foreign matter evaluation memory value FM(i,j) is updated using the foreign matter evaluation value F(i,j) of the 30th image. FIG. 27 shows how the foreign matter evaluation memory value FM(i,j) is updated using the foreign matter evaluation value F(i,j) of the 40th image. The foreign matter evaluation values F(i,j) of the 1st and 10th images are below the foreign matter threshold value, so are set to 0 (zero).

Next, in the process of detecting edges, Figures 28 to 31 show how the image number N in the image number storage unit 45 is updated after the evaluation value update for the first image is completed, how the image number N is updated in the image number storage unit 45 after the evaluation value update for the 10th image is completed, how the image number N is updated in the image number storage unit 45 after the evaluation value update for the 30th image is completed, and how the image number N is updated in the image number storage unit 45 after the evaluation value update for the 40th image is completed.

Next, in the process of detecting foreign objects, FIG. 32 shows how the image number in the foreign object image number storage unit 55 is updated after the evaluation value update for the 30th image is completed. FIG. 33 shows how the image number in the foreign object image number storage unit 55 is updated after the evaluation value update for the 40th image is completed. The foreign object 3D information extraction unit 56 determines that the pixel displaying the foreign object is present at position (5, 14) in image number 30 (30th image) and position (15, 4) in image number 40 (40th image) as shown in FIG. 33. Note that the foreign object evaluation value F(i, j) for the first and tenth images is set to 0, which is less than the foreign object threshold, so the evaluation value update is not performed in the first and tenth images, and no pixel displaying a foreign object is detected.

When the process of updating the image numbers in which edges and foreign objects are detected is thus completed, the device finally creates a contour graph for the inspection object based on the image number array data shown in FIG. 31. Then, as shown in FIG. 34, three-dimensional information (height information) of the sample can be analyzed from the contour graph. Based on FIG. 34, the three-dimensional information for the inspection object is determined in detail.

The arrangement of edge evaluation values E(i,j) of the 10th image shown in FIG. 19 can be said to be the densest, so the 10th image can be determined to be an image focused on the thin film transistor (TFT) circuit portion at the bottom of the object, which is the reference for the object's three-dimensional information (structure). Therefore, the height of the 10th image is set as the reference height (0 μm). The image number corresponding to the reference height of the three-dimensional information is 10. In other words, the height of the 10th image can be set as the reference.

Since the image number of the image on which the first minute foreign matter 101 is focused is 30, the device can determine that the first minute foreign matter 101 is present at a height of 2.0 μm above the reference. Since the image number of the image on which the second minute foreign matter 102 is focused is 40, the device can determine that the second minute foreign matter 102 is present at a height of 3.0 μm above the reference. The device can then determine that the first minute foreign matter 101 is present 2 μm above the electrode pattern 103. Furthermore, the device can determine that the second minute foreign matter 102 is present 3 μm above the first foreign matter, that is, 5 μm above the electrode pattern 103. The height of the peak in the contour graph shown in FIG. 34 indicates three-dimensional information about the minute foreign matter in the inspection object.

When a defect such as a minute foreign object or a pinhole is detected, the pass/fail judgment unit 60 analyzes the three-dimensional information of the defect. From the analysis result, the device judges whether the defect exists on the same sealing layer. If the defect exists on the same sealing layer, the organic EL display device, which is the object, is judged as a good product. Then, the device judges whether multiple defects exist at different heights (thickness direction). If multiple defects exist at different heights (thickness direction), the object is judged as a defective product. As described above, oxygen and water that have entered through defects on the organic film permeate the organic film over time. If the oxygen and water reach the defects below the organic film, the measurement object (e.g., the EL display device) directly below the defects is destroyed. As a result, the life of the measurement object is shortened.

Furthermore, a means for repairing defects may be added to the defect quality determination device of the present invention. For example, in the manufacture of thin-film multi-layer devices such as organic EL display devices, the defect quality determination device of the present invention can identify a layer having a defect (e.g., a pinhole or foreign matter) with a diameter of 1 μm or less, but the conventional technology cannot determine it. Therefore, by using the defect quality determination device of the present invention, it is possible to select an optimal repair means according to the material of the sealing film forming the layer in which the defect exists. If the defect is a foreign matter in the organic film, the foreign matter can be removed by using a laser, and then the film can be repaired. Furthermore, by selecting the wavelength [nm] and energy density [J/cm ² ] of the laser light, optimal repair can be performed. If it is not possible to use a laser light, a repair method in which the foreign matter is pushed downward can be selected. If the defect is a pinhole, the following method can be adopted. This method is a method in which a small amount of film material is applied to the defective pinhole using a cartridge equipped with a fine tip processing tube (microdispenser), and then cured by baking or ultraviolet irradiation.

In this embodiment, one image number with the maximum number of edge pixels of the pattern is detected, and also the image number with the maximum number of foreign matter pixels and the image number with the second largest number of foreign matter pixels (including the case where the number is the same as the maximum number of foreign matter pixels) are detected, however, this is not limited to the above embodiment. For example, appropriate changes can be made depending on the organic film, inorganic film, and electrode pattern that make up the sample, or depending on the size, density, location of occurrence, and the like of the foreign matter or defect.

The edge evaluation value E(i,j) and the foreign matter evaluation value F(i,j) can be stored in the edge evaluation value memory value EM(i,j). Which of the edge evaluation value E(i,j) and the foreign matter evaluation value F(i,j) is to be prioritized may be determined based on, for example, the size of the edge evaluation value E(i,j) and the foreign matter evaluation value F(i,j) and the sharpness of the partial image.

In the defect quality determination method and device according to the present invention, horizontal positional errors occur between the images acquired during the imaging process due to vibration of the device itself or the floor. By detecting the image number with the highest sharpness, it is possible to accurately measure the height information at which minute defects occur without being affected by horizontal positional errors. This also improves the vibration resistance characteristics of the defect quality determination device, and the above process contributes greatly to reducing the cost of the device.

In the above embodiment, an example was shown in which height measurement of edge pixels of a pattern and height measurement of a minute foreign particle were performed independently, but when the edge evaluation value of a pattern and the minute foreign particle evaluation value are converted to values of the same order of magnitude, it is possible to treat the processing from the edge evaluation value comparison unit 43 to the edge 3D information extraction unit 46 and the processing from the foreign particle evaluation value comparison unit 53 to the foreign particle 3D information extraction unit 56 as a common processing. By sharing the processing units in this way, it is possible to simplify the defect pass/fail determination device of the present invention. As such examples, the configuration of the defect quality determination device of the present invention can be simplified by configuring it as follows: a feature extraction section that integrates the pattern edge extraction section 41 and the minute foreign matter extraction section 51; an evaluation value calculation section that integrates the edge evaluation value calculation section 42 and the foreign matter evaluation value calculation section 52; an evaluation value comparison section that integrates the edge evaluation value comparison section 43 and the foreign matter evaluation value comparison section 53; an evaluation value storage section that integrates the edge evaluation value storage section 44 and the foreign matter evaluation value storage section 54; an image number storage section that integrates the edge image number storage section 45 and the foreign matter image number storage section 55; and a three-dimensional information extraction section that integrates the edge three-dimensional information extraction section 46 and the foreign matter three-dimensional information extraction section 56. The edge processing section 40, the foreign matter processing section 50, and the quality determination section 60 can be executed by software processing, except for the storage section.

Alternative Embodiments of the Invention
35 is a schematic side view of an inspection apparatus 500 for mapping features in a sample 502, according to one embodiment of the present invention. The apparatus 500 operates based on similar principles to the previous embodiments, with additions and modifications as described below. As described in previous embodiments, e.g., in Figures 1 and 3, the sample 502 typically includes multiple thin film layers, including a transparent layer, that are overlaid on the surface of the sample.

The apparatus 500 includes a video camera 506 that captures an electronic image of the sample 502 through a lens 508, typically a microscope lens with high magnification, high numerical aperture, and shallow depth of focus. An illumination source 504 illuminates the sample 502 while the camera 506 captures the image. In this embodiment, the illumination source 504 emits monochromatic light, i.e. light having a bandwidth (full width at half maximum) of 40 nm or less. Such enhanced monochromatic illumination is advantageous for eliminating the effects of chromatic aberration in the image captured by the camera 506. It is also advantageous for the illumination source 504 to illuminate the sample 506 in a dark field mode to enhance the contrast of image features. However, alternatively, the illumination source 504 may emit white light or other broadband light, or may provide bright field illumination.

A motor 510 scans the front focal plane of the camera 506 in a direction perpendicular to the surface of the sample 502. The scan can be continuous or stepwise. In the illustrated embodiment, the motor 510 translates the camera 506 and the lens 508 up and down. Alternatively, or in addition, the motor can shift the vertical position of the sample 502 or adjust the focus setting of the lens 508 to scan the focal plane. Over the course of the scan, the camera 506 captures a series of images of the thin film layer on the sample 502, each at a different focal depth within the sample. As a result, features located at different depths within the sample are in focus in a series of different images, where the sharpest focus occurs when the front focal plane of the camera coincides with the location of the feature. For features that span a range in the depth dimension (i.e., the dimension perpendicular to the surface of the sample 502), the top of the feature will be in sharp focus in one image and the bottom will be in sharp focus in the other image.

The processor 512 processes the sequence of images captured by the camera 506 over the course of the motor 510 to identify features of interest in the images. Such features may include, for example, defects in the thin film layer, as described above. The processor 512 generally comprises a general-purpose computer processor, has suitable interfaces for receiving electronic images from the camera 506 and signals from other components of the device 500, and is programmed with software to perform the functions described herein. Alternatively or additionally, at least some of the functions of the processor 512 may be implemented in programmable or hardwired logic. Having identified a feature of interest, the processor 512 calculates the optimum optical focal depth of that feature in the sequence of images and then estimates the location of that feature, particularly the location of the depth (vertical) dimension, within the thin film layer. To this end, as described in detail above, the processor 512 calculates a measure of the sharpness of the edges of the feature in the images and finds the depth that maximizes the sharpness.

In this embodiment, the device 500 includes a range finder including a laser 514 and a detector 516 to measure the distance between the camera 506 and the sample 502. The illustrated range finder operates by sensing the shift in the position of the laser spot reflected from the sample 502 onto the detector 516 as the distance between the camera and the sample changes. Alternatively, other types of range finders known in the art can be used, such as ultrasonic or interferometric range finders. The processor 512 applies the distance measured by the range finder in estimating the position of the feature of interest, and more specifically corrects for variations in the position of the front focal plane of the camera 506 within a thin film layer on the sample 502 that may occur, for example, due to vibration of the sample. The processor 512 can detect this vibration based on the periodic change in the distance measured by the range finder over time, and can then correct the depth measurements in the captured images to compensate for this vibration, thus estimating the position of the feature of the sample 502 with greater accuracy.

36 is a schematic plot of vibrations measured in an apparatus 500 according to an embodiment of the invention. A number of data points 520 in the plot indicate the relative height (in microns) of the camera 506 above the sample 502 relative to the baseline height as a function of time (in seconds) in the absence of vibration. Each data point 520 corresponds to a reading made by the range finder detector 516. The processor 512 fits a periodic function to the data points 520 to generate a curve 522, which gives the estimated amplitude of vibration at any time. The camera 506 captures images at times indicated by a number of marks 524 on the curve 522. At each such time, the processor 512 reads the values of the curve 522 to provide a height correction, which it adds (or subtracts) to the nominal depth provided by the scan of the motor 510 to calculate the corrected depth of focus. Thus, the processor 512 can compensate for the vibrations of the sample 502 and can more accurately estimate the location of features appearing in the image.

37 is a schematic plot showing measurements of focus quality made by the apparatus 500 according to an embodiment of the present invention. The data points 530 correspond to the calculated focus scores for a given feature, plotted as a function of the focal depth of the camera 506 in the thin film layer on the sample 502. The nominal focal depth may be corrected for sample vibration as explained above. The focus score measures the sharpness of the edges of the feature of interest, for example based on the derived image. The Z positions of the data points 530 have been corrected for the measured vibration and may therefore be unevenly spaced in the plot. The focus score is inverse parabolic as a function of the depth of the front focal plane of the camera 506. The processor 512 therefore fits a suitable curve 532 to the data points 530 and finds the peak of the curve 532, which indicates the depth of the feature in the sample 502.

Figure 38 is a flow chart that illustrates, in a schematic manner, a method for mapping features in a sample according to an embodiment of the present invention. The method is described below for convenience and clarity with reference to features of apparatus 500 (see Figure 35). Alternatively, the method may be performed using the apparatus of the previous embodiment, or using any other suitable inspection system, mutatis mutandis, as will be apparent to those skilled in the art after reading this specification.

In a distance measurement step 540, the processor 512 measures the distance of the camera 506 and lens 508 from the sample 502 using a range finder, such as a laser 514 or sensor 516. Generally, in a scanning step 542, the apparatus 500 is configured so that this distance remains approximately constant (other than slight movements due to vibration) while the motor 510 scans the depth of the front focal plane of the camera through the thin film layer on the sample 502. Alternatively, in this step, the range finder may measure the shift caused by the operation of the motor 510. As the motor 510 scans the focal point of the camera in the depth dimension, the processor 512 acquires an image of the sample 502 from the camera 506.

Based on the range finder measurements made in step 540, the processor 512 reconstructs the vibration pattern of the sample 502, for example, in a vibration reconstruction step 544, as shown in FIG. 36. The processor 512 can then correct the nominal focus depth of the image acquired in step 542, in a depth correction step 546, to compensate for errors caused by vibration. The processor 512 identifies one or more features of interest in the image, e.g., potential defects, and calculates focus scores for these features as a function of corrected depth, in a focus scoring step 548. For each such feature, the processor 512 fits a curve to the calculated focus scores and then finds the three-dimensional coordinates of the feature, in a position calculation step 550.

It will be understood that the above embodiments are cited by way of example, and that the present invention is not limited to what has been specified and described above. Rather, the scope of the present invention includes combinations and subcombinations of the various features described above, as well as variations and modifications thereof not disclosed in the prior art that would occur to one of ordinary skill in the art upon reading the above description.

REFERENCE SIGNS LIST 1 Electric circuit pattern 2 First sealing layer 3 Second sealing layer 4 Third sealing layer 5 First substrate 6 Second substrate 10 Flexible organic EL display device 20 Microscope (lens barrel)
21 Objective lens 22 Camera 23 Height direction drive motor 30 Sequence control unit 31 Image acquisition unit 32 Image memory 40 Edge processing unit 41 Pattern edge extraction unit 42 Edge evaluation value calculation unit 43 Edge evaluation value comparison unit 44 Edge evaluation value storage unit 45 Edge image number storage unit 46 Edge three-dimensional information extraction unit 50 Foreign matter processing unit 51 Micro foreign matter extraction unit 52 Foreign matter value calculation unit 53 Foreign matter value comparison unit 54 Foreign matter value storage unit 55 Foreign matter image number storage unit 56 Foreign matter three-dimensional information extraction unit 60 Good/bad judgment unit 101 First micro foreign matter 102 Second micro foreign matter 103 Electrode pattern 104 Cathode electrode 105 Organic film 106 Transparent film 107 Light-emitting layer 500 Inspection device 502 Sample 504 Illumination light source 506 Camera 508 Microscope lens 510 Focus adjustment motor 512 Processor 514 Distance Meter Laser 516 Distance Meter Detector

Claims (25)

A step of acquiring a plurality of images of an object to be inspected, the image having a predetermined interval in a height direction, by an optical imaging means, for the object to be inspected, the image having a multi-layer transparent thin film;
a step of calculating a sharpness of a partial image from a luminance difference between each pixel of the plurality of images and adjacent pixels, the partial image being composed of each pixel of the plurality of images, pixels adjacent to each pixel in the horizontal direction, and pixels adjacent to each pixel in the vertical direction;
calculating height information of the partial image from an image number having a maximum calculation result of the sharpness at a single pixel position among all of the plurality of images;
A step of obtaining three-dimensional information of the entire image from the calculation of the height information, the three-dimensional information including at least one of three-dimensional information of an edge of the entire image and three-dimensional information of a foreign object of the entire image; and
determining whether the inspection object is defective or not based on the edge three-dimensional information and the foreign matter three-dimensional information;
the object to be inspected is an organic EL device, an edge of an electrode pattern of the organic EL device is used as a reference for the height of the three-dimensional information, and when the edge of the electrode pattern is difficult to detect, an edge of a light-emitting portion of the organic EL device is used as a reference for the height of the three-dimensional information;
Method for determining whether defects are acceptable or not. detecting a pattern defect in the image having the greatest sharpness;
extracting an image having a maximum density of the partial images with high sharpness from the plurality of images;
A step of setting the image as a reference position 1 in the height direction of a three-dimensional pattern structure; and
2. The defect quality determination method according to claim 1, further comprising the step of measuring a height of the generated pattern defect in the three-dimensional pattern structure from a relationship between the height information of the pattern defect and the reference position 1. detecting pattern defects in the image with the greatest sharpness;
extracting an image from the plurality of images in which an interference image of interference fringes occurring at an edge of the transparent thin film is sharpest;
A step of setting the image as a reference position 2 in the height direction of a three-dimensional pattern structure; and
2. The defect quality determination method according to claim 1, further comprising the step of measuring a height of the generated pattern defect in the three-dimensional pattern structure from a relationship between height information of the pattern defect and the reference position (2). The defect pass/fail determination method according to claim 2 or 3, further comprising a step of repairing the pattern defect using the height information of the pattern defect. an imaging means for acquiring a plurality of image data of an inspection object having a multi-layer transparent thin film by using an optical imaging means that can be moved up and down at a predetermined interval, and assigning image numbers to the image data;
An extraction unit that extracts features of the image data;
an evaluation value calculation unit that calculates an evaluation value based on the features, the evaluation value representing a luminance difference between a pixel of interest and an adjacent pixel adjacent to the pixel of interest;
an evaluation value comparison unit that compares the evaluation value with a previous evaluation value that matches a positional relationship in the image data and generates a comparison result;
an evaluation value storage unit that stores the evaluation value based on the comparison result;
an image number storage unit that stores the image number based on the comparison result;
an edge three-dimensional information extraction unit that extracts edge three-dimensional information of the inspection object based on the image number stored in the image number storage unit;
a foreign matter three-dimensional information extraction unit that extracts height information of a defect present in the inspection object based on the edge three-dimensional information; and
a quality determination unit that determines quality of the inspection object based on a difference in height information between the defects when the plurality of defects are present;
the object to be inspected is an organic EL device, and an edge of an electrode pattern of the organic EL device is used as a reference for the height of the three-dimensional information, and when the edge of the electrode pattern is difficult to detect, an edge of a light-emitting portion of the organic EL device is used as the reference for the height of the three-dimensional information. The defect quality determination device according to claim 5, wherein the edge 3D information extraction unit extracts the edge 3D information based on the image number with the highest evaluation value. The defect quality determination device according to claim 5 or 6, wherein the evaluation value is calculated based on the luminance difference between a pixel of interest and an adjacent pixel adjacent to the pixel of interest. The defect quality determination device according to any one of claims 5 to 7, wherein the foreign object 3D information extraction unit determines the standard for the height information based on the evaluation value of the image data in which the electrode pattern of the inspection object and the interference fringes of the sealing layer of the inspection object are captured. The defect quality determination device according to any one of claims 5 to 8, wherein the defect is a pattern defect, a pinhole, or a foreign object. The defect quality determination device according to any one of claims 7 to 9, wherein the evaluation value is a sharpness calculated based on the difference in luminance value of the pixel of interest relative to the luminance values of the adjacent pixels. The defect quality determination device according to any one of claims 5 to 10, wherein the object to be inspected is an organic electroluminescence display device. The defect quality determination device according to any one of claims 5 to 10, wherein the object to be inspected is a flexible organic EL display device formed on a flexible substrate. The defect quality determination device according to any one of claims 5 to 12, further comprising at least one function for repairing the defect based on the height information calculated by the foreign object 3D information extraction unit. The defect quality determination device according to claim 13, further comprising a function for selecting the function according to the height information. a camera configured to capture an image of a sample including a plurality of thin film layers overlaid on a surface thereof;
a motor coupled to scan a front focal plane of the camera in a direction perpendicular to a surface of the sample, thereby capturing a series of images of a thin film layer at different focal depths within the sample; and
a processor configured to process images in the series of images to identify features of interest in the images, calculate an optimum depth of focus for the features of interest in the series of images, and estimate a location of the features of interest;
Equipped with
the processor is further configured such that calculating the optimum focal depth for the feature of interest includes calculating a sharpness of an edge of the feature of interest and finding the depth that maximizes the sharpness;
the sample is an organic EL device, an edge of an electrode pattern of the organic EL device is used as a reference for the height of the three-dimensional information, and when the edge of the electrode pattern is difficult to detect, an edge of a light-emitting portion of the organic EL device is used as a reference for the height of the feature of interest;
Inspection equipment. The apparatus of claim 15, comprising an illumination source configured to illuminate the sample with monochromatic light while the camera captures the image. The apparatus of claim 16, wherein the illumination source is configured to illuminate the sample in a dark field mode. The apparatus of claim 15, further comprising a range finder configured to measure a distance between the camera and the sample, and the processor configured to use the measured distance in estimating a position of the feature of interest. The apparatus of claim 18, wherein the processor is configured to detect vibration of the sample relative to the camera based on periodic changes in the measured distance over time and to correct the focal depth of the captured image to compensate for the detected vibration. capturing a series of images of a sample, the sample including a plurality of thin film layers overlaid on a surface of the sample, at different respective focal depths within the sample;
Identifying features of interest within the image;
calculating an optimal depth of focus of the feature of interest in the series of images; and estimating a location of the feature of interest within the thin film layer based on the optimal depth of focus.
Including,
Calculating the optimum depth of focus includes calculating the sharpness of edges of the feature of interest and finding the depth that maximizes the sharpness;
the sample is an organic EL device, an edge of an electrode pattern of the organic EL device is used as a reference for the height of the three-dimensional information, and when the edge of the electrode pattern is difficult to detect, an edge of a light-emitting portion of the organic EL device is used as a reference for the height of the feature of interest;
Inspection method. 21. The method of claim 20, wherein capturing the sequence of images includes illuminating the sample with monochromatic light while the images are being captured. 22. The method of claim 21, wherein illuminating the sample includes directing light at the sample in a dark field mode. 21. The method of claim 20, wherein capturing the series of images includes scanning a front focal plane of a camera in a direction perpendicular to a surface of the sample, whereby the camera captures the series of images of the thin film layer at the different focal depths. 24. The method of claim 23, wherein calculating the optimal depth of focus includes measuring a distance between the camera and the sample and applying the measured distance in estimating the position of the feature of interest. 25. The method of claim 24, wherein applying the measured distance includes detecting vibration of the sample relative to the camera based on periodic changes in the measured distance over time, and correcting the focal depth of the captured image to compensate for the detected vibration.