Pattern inspection apparatus, and pattern inspection method

Abstract

To inspect a sample using a reference image generated with image generation parameters optimized for an inspection target pattern.SOLUTION: A pattern inspection apparatus comprises: a filter coefficient calculation part that uses a plurality of pattern types to calculate a plurality of parameters for generating a reference image; an optical image acquisition mechanism 150 that acquires a representative optical image; an expansion circuit 111 that generates a representative expanded image by expanding the image using design data of a pattern formed in the representative optical image; a reference circuit 112 that generates a representative reference image by filtering the representative expanded image; a similarity parameter calculation part that calculates, for each case, a similarity parameter indicating the similarity between the representative reference image and the representative optical image; a similarity determination part that determines parameters having a predetermined rank of high similarity; and a comparison circuit 108 that compares an inspection reference image with an inspection optical image, the comparison circuit 108 comparing the inspection reference image generated using the parameters having the predetermined rank of high similarity with the inspection optical image.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] TECHNICAL FIELD One aspect of the present invention relates to a pattern inspection apparatus and a pattern inspection method, for example, to an apparatus and a method for inspecting pattern defects on a sample used in semiconductor manufacturing. [Background technology]

[0002] In recent years, with the increasing integration and capacity of large-scale integrated circuits (LSI), the circuit line width required for semiconductor elements has become increasingly narrow. These semiconductor elements are manufactured by forming circuits by exposing and transferring the pattern onto a wafer using a reduced projection exposure device called a stepper, using an original pattern (also called a mask or reticle, hereinafter collectively referred to as a mask) on which the circuit pattern is formed.

[0003] Improving yield is essential for the manufacture of LSIs, which incur huge manufacturing costs. However, as exemplified by 1-gigabit-class DRAMs (random access memories), the patterns that make up LSIs are approaching the submicron to nanometer order. One of the major factors reducing yields is pattern defects in the masks used to expose and transfer ultra-fine patterns onto semiconductor wafers using photolithography technology. In recent years, as the dimensions of LSI patterns formed on semiconductor wafers have become increasingly miniaturized, the dimensions that must be detected as pattern defects have also become extremely small. This has led to a need for higher-precision pattern inspection systems to inspect the transfer masks used in LSI manufacturing for defects.

[0004] One known inspection method involves comparing optical images of a pattern formed on a sample, such as a lithography mask, captured at a predetermined magnification using a magnifying optical system with design data. For example, "die-to-database" inspection involves converting the pattern design data (design pattern data) into an input format for the lithography device used to write the pattern on the mask. This data is then input into an inspection device, where a design image (reference image) is generated based on the design image. This image is then compared with an optical image of the pattern, which serves as measurement data. In this type of inspection method, the sample is placed on a stage, and the stage moves, scanning the sample with a beam of light. The sample is illuminated by a light source and illumination optical system. The light transmitted through or reflected from the sample is focused on a sensor via the optical system. The image captured by the sensor is then sent to a comparison circuit as measurement data. After aligning the images, the comparison circuit compares the measurement data with the reference data using an appropriate algorithm. If the images do not match, it is determined that a pattern defect exists.

[0005] Conventionally, reference images, which serve as reference data to be compared with measurement images serving as measurement data, have been created using parameters common to each position within a mask. These parameters are created to achieve a high overall similarity for multiple pattern images within the mask. However, such common parameters, for example, tend to be ambivalent between resembling fine patterns and resembling large patterns, resulting in a low similarity between the created reference image and the measurement image. This has led to the problem of generating many false defects. On the other hand, loosening the decision threshold to suppress false defects has led to the problem of missing defects. Therefore, more effective reference image generation parameters are needed.

[0006] Here, a set of multiple reference image generation parameters obtained from multiple sample images is calculated. Then, the parameter that has the highest degree of match between the sample reference image and the sample optical image using each of the multiple generation parameters is designated as the common parameter. A reference image for each region in the mask is created using this common parameter, and the reference image and measurement image are compared for each region. Then, for regions where defects exist, a reference image is created using the remaining parameters in the set of multiple generation parameters, and is compared again with the measurement image. Then, the comparison result that results in the smallest number of defects is adopted. This technique is disclosed (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-222627 Summary of the Invention [Problem to be solved by the invention]

[0008] Therefore, one aspect of the present invention provides an inspection apparatus and method capable of performing inspection using a reference image created with effective reference image generation parameters suited to the pattern to be inspected. [Means for solving the problem]

[0009] A pattern inspection apparatus according to one aspect of the present invention comprises: a parameter calculation unit that calculates a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; an optical image acquisition mechanism for acquiring a plurality of representative optical images of a plurality of representative regions of the pattern-formed sample to be inspected; an exfoliated image creating unit that creates a representative exfoliated image for each of the representative regions by exposing the representative region using design data of a pattern to be formed in the representative optical image of the representative region; a reference image creation unit that creates a representative reference image by filtering a representative exfoliated image of a representative region using each of the plurality of parameters; a similarity calculation unit that calculates, for each of the plurality of parameters, a similarity parameter that indicates a similarity between a representative reference image created using the parameter and a representative optical image of the representative region; a similarity determination unit that determines a parameter having a predetermined ranking of high similarity from among the plurality of parameters; a comparison unit that compares a reference image for inspection with an optical image for inspection; Equipped with the reference image creation unit creates a reference image for inspection using the parameters in a predetermined order with the highest similarity; The optical image acquisition mechanism acquires an optical image for inspection, the comparison unit compares the optical image for inspection with a reference image for inspection created using parameters in a predetermined order with high similarity; It is characterized by:

[0010] The reference image creation unit creates a plurality of reference images for inspection using a plurality of parameters in a predetermined order of high similarity, the comparison unit compares the optical image for inspection with a plurality of reference images for inspection created using a plurality of parameters in a predetermined order of high similarity, It is preferable to further include a counting unit that counts the comparison results for each of the compared reference images and outputs the results.

[0011] Furthermore, the parameters in a predetermined order with high similarity are preferably the top one parameter or the top two parameters.

[0012] Another aspect of the pattern inspection apparatus of the present invention is a parameter calculation unit that calculates a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; an optical image acquisition mechanism for acquiring an optical image of a predetermined area of ​​the pattern-formed sample to be inspected; an unfolded image creating unit that creates an unfolded image by unfolding an image using design data of a pattern to be formed on an optical image of a predetermined area; a reference image generating unit that generates a plurality of reference images by filtering the unfolded image using different ones of the plurality of parameters; a comparison unit that compares each of the optical image with a plurality of reference images; a counting unit that counts the comparison results for each of the compared reference images and outputs the results; The present invention is characterized by the following features.

[0013] Preferably, the plurality of parameters include a parameter estimated from a fine pattern and a parameter estimated from a large pattern.

[0014] Preferably, the counting unit outputs, as a defect, a portion that is commonly determined to be a defect from the comparison results for each reference image.

[0015] Alternatively, it is preferable that the counting unit outputs, as a defect, a portion that is determined to be a defect in any of the comparison results for each reference image.

[0016] A pattern inspection method according to one aspect of the present invention includes: calculating a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; acquiring a plurality of representative optical images of a plurality of representative regions of a patterned sample to be inspected; creating a representative expanded image for each of the representative regions by using design data of a pattern to be formed in the representative optical image of the representative region; a step of generating a representative reference image by filtering the representative exfoliated image of the representative region using each of the plurality of parameters; calculating, for each of the plurality of parameters, a similarity parameter indicating a similarity between a representative reference image created using the parameter and a representative optical image of the representative region; determining a parameter having a predetermined rank with a high degree of similarity from among the plurality of parameters; creating a reference image for inspection using parameters ranked in a predetermined order with high similarity; acquiring an optical image for inspection; comparing the optical image for inspection with a reference image for inspection created using parameters in a predetermined order with a high degree of similarity; The present invention is characterized by the following features.

[0017] A pattern inspection method according to another aspect of the present invention includes: calculating a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; acquiring an optical image of a predetermined area of ​​the patterned sample to be inspected; creating an expanded image by expanding the image using design data of a pattern to be formed in the optical image of a predetermined area; creating a plurality of reference images by filtering the unfolded image using different ones of the plurality of parameters; comparing the optical image to a plurality of reference images, respectively; a step of aggregating the comparison results for each of the compared reference images and outputting the results; The present invention is characterized by the following features. [Effects of the Invention]

[0018] According to one aspect of the present invention, inspection can be performed using a reference image created with effective reference image generation parameters suited to the pattern to be inspected, thereby suppressing or reducing false defects and / or missed defect detection. [Brief explanation of the drawings]

[0019] [Figure 1] FIG. 1 is a configuration diagram showing a configuration of a pattern inspection device according to a first embodiment. [Figure 2] FIG. 2 is a conceptual diagram for explaining an inspection area in the first embodiment. [Figure 3] FIG. 10 is a diagram showing an example of an image for estimating filter coefficients in a comparative example of the first embodiment. [Figure 4] FIG. 2 is a block diagram showing an example of the internal configuration of a filter function calculation circuit according to the first embodiment. [Figure 5] FIG. 3 is a flowchart showing an example of main steps of the inspection method according to the first embodiment. [Figure 6] FIG. 2 is a diagram showing an example of a representative frame according to the first embodiment. [Figure 7] FIG. 2 is a diagram showing an example of a fine pattern image according to the first embodiment. [Figure 8] FIG. 2 is a diagram showing an example of a large pattern image according to the first embodiment. [Figure 9] 4A to 4C are diagrams showing examples of graphic patterns before and after resizing in the first embodiment. [Figure 10] 5A to 5C are diagrams showing examples of figure patterns before and after rounding processing in the first embodiment. [Figure 11] FIG. 4 is a diagram for explaining an example of the relationship between a filter function, an unfolded image, and a measurement image in the first embodiment. [Figure 12] FIG. 3 is a diagram illustrating an example of a method for calculating coefficients of a filter function according to the first embodiment. [Figure 13] 3 is a diagram illustrating an example of an internal configuration of a comparison circuit according to the first embodiment. FIG. [Figure 14] FIG. 10 is a configuration diagram showing the configuration of a pattern inspection device according to a second embodiment. [Figure 15] FIG. 10 is a flowchart showing an example of main steps of an inspection method according to the second embodiment. [Figure 16] FIG. 10 is a diagram showing an example of a counting method in the second embodiment. [Figure 17] FIG. 10 is a diagram showing another example of a counting method in the second embodiment. [Figure 18] FIG. 11 is a block diagram showing an example of the internal configuration of a filter function calculation circuit according to a third embodiment. [Figure 19]FIG. 11 is a flowchart showing an example of main steps of an inspection method according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0020] Embodiment 1 Fig. 1 is a configuration diagram showing the configuration of a pattern inspection apparatus according to embodiment 1. In Fig. 1, an inspection apparatus 100 that inspects a sample, for example, a pattern formed on a mask, for defects includes an optical image acquisition mechanism 150 and a control system circuit 160 (control unit).

[0021] The optical image acquisition mechanism 150 includes a light source 103, an illumination optical system 170, a movably arranged XYθ table 102, a magnifying optical system 104, an imaging sensor 105 (an example of a sensor), a sensor circuit 106, a stripe pattern memory 123, a laser measurement system 122, and an autoloader 130. A sample 101 is placed on the XYθ table 102. The sample 101 includes, for example, a photomask for exposure that transfers a pattern onto a wafer. The photomask has a pattern formed thereon that is made up of a plurality of geometric patterns to be inspected. The sample 101 is placed on the XYθ table 102 with the pattern-formed surface facing downward, for example.

[0022] A line sensor or a two-dimensional sensor is used as the image sensor 105. For example, a TDI (time delay integration) sensor is preferably used. A TDI sensor has multiple photosensor elements arranged two-dimensionally. A predetermined image accumulation time is set for each photosensor element when capturing an image. In a TDI sensor, the outputs of multiple photosensor elements arranged in the scanning direction are integrated and output. Multiple photosensor elements arranged in the scanning direction capture the same pixel at different times in accordance with the movement of the XYθ table 102. When a line sensor is used, multiple photosensor elements are arranged in a direction perpendicular to the scanning direction.

[0023] In the control system circuit 160, a control calculator 110 serving as a computer is connected via a bus 120 to a position circuit 107, a comparison circuit 108, an expansion circuit 111, a reference circuit 112, an autoloader control circuit 113, a table control circuit 114, a filter function calculation circuit 140, a magnetic disk drive 109, a flexible disk drive (FD) 115, a memory 116, a CRT 117, a pattern monitor 118, and a printer 119. The sensor circuit 106 is connected to a stripe pattern memory 123, which is connected to the comparison circuit 108. The XYθ table 102 is driven by an X-axis motor, a Y-axis motor, and a θ-axis motor. The XYθ table 102 is an example of a stage.

[0024] Note that a series of "circuits" such as the position circuit 107, the comparison circuit 108, the expansion circuit 111, the reference circuit 112, the autoloader control circuit 113, the table control circuit 114, and the filter function calculation circuit 140 include a processing circuit. Such processing circuits include an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. Each circuit may be configured using the same processing circuit (one processing circuit), or different processing circuits (separate processing circuits). For example, a series of "circuits" such as the position circuit 107, the comparison circuit 108, the expansion circuit 111, the reference circuit 112, the autoloader control circuit 113, the table control circuit 114, and the filter function calculation circuit 140 may be configured and executed by the control computer 110. Input data or calculation results required for the position circuit 107, comparison circuit 108, expansion circuit 111, reference circuit 112, autoloader control circuit 113, table control circuit 114, and filter function calculation circuit 140 are stored in a memory (not shown) within each circuit or in memory 116. Input data or calculation results required for the control computer 110 are stored in a memory (not shown) within the control computer 110 or in memory 116. The program for executing the processor etc. may be recorded on a recording medium such as the magnetic disk device 109, FD 115, or ROM (read only memory).

[0025] In the inspection apparatus 100, a high-magnification inspection optical system is configured with a light source 103, an XYθ table 102, an illumination optical system 170, a magnifying optical system 104, an imaging sensor 105, and a sensor circuit 106. The XYθ table 102 is driven by a table control circuit 114 under the control of a control computer 110. It is movable by a drive system such as a three-axis (XY-θ) motor that drives in the X, Y, and θ directions. These X, Y, and θ motors can be, for example, step motors. The XYθ table 102 can be moved horizontally and in rotational directions by motors on each of the X, Y, and θ axes. The movement position of the sample 101 placed on the XYθ table 102 is measured by a laser measurement system 122 and supplied to a position circuit 107.

[0026] Drawing data (design data) that is the basis for forming a pattern on the sample 101 to be inspected is input from outside the inspection device 100 and stored in the magnetic disk device 109. The drawing data defines a plurality of figure patterns, and each figure pattern is usually composed of a combination of a plurality of element figures. However, a figure pattern composed of a single figure may also be present. On the sample 101, corresponding patterns are formed based on each of the figure patterns defined in the drawing data.

[0027] 1 shows components necessary for explaining the first embodiment. It goes without saying that the inspection device 100 may include other components that are normally required.

[0028] FIG. 2 is a conceptual diagram illustrating the inspection area in the first embodiment. As shown in FIG. 2, the inspection area 10 (the entire inspection area) of the sample 101 is virtually divided, for example, in the Y direction, into a plurality of rectangular inspection stripes 20 each having a scan width W of the image sensor 105. The inspection device 100 then acquires an image (stripe area image) for each inspection stripe 20. For each inspection stripe 20, a laser beam (inspection light) is used to capture an image of a graphic pattern arranged within the inspection stripe 20 in the longitudinal direction (X direction) of the stripe area. Note that, to prevent missing images, the inspection stripes 20 are preferably set so that adjacent inspection stripes 20 overlap with each other by a predetermined margin width.

[0029] The movement of the XYθ table 102 causes the image sensor 105 to continuously move relatively in the X direction, thereby acquiring optical images. The image sensor 105 continuously captures optical images with a scan width W as shown in FIG. 2. In the first embodiment, after capturing an optical image of one inspection stripe 20, the image sensor 105 moves in the Y direction to the position of the next inspection stripe 20, and then moves in the reverse direction while similarly capturing optical images with the scan width W continuously. In other words, imaging is repeated in the forward (FWD)-backward (BWD) directions, which are opposite directions on the outward and return paths.

[0030] Furthermore, in actual inspection, the stripe region image of each inspection stripe 20 is divided into a plurality of rectangular frame region 30 images (frame images 31), as shown in FIG. 2. Then, inspection is performed for each frame image 31 of the frame region 30. For example, the frame region 30 is divided into a size of 512 × 512 pixels. Therefore, a reference image to be compared with the frame images 31 of the frame region 30 is also created for each frame region 30.

[0031] Here, the imaging direction is not limited to repeated forward (FWD)-backward (BWD). Imaging may be performed from one direction. For example, FWD-FWD may be repeated. Alternatively, BWD-BWD may be repeated.

[0032] FIG. 3 is a diagram showing an example of an image for estimating filter coefficients in a comparative example of the first embodiment. As shown in FIG. 3, sub-images of different pattern types A, B, C, and D are combined to form a single image. For example, sub-measurement images of different pattern types A, B, C, and D are captured from the inspection area 10 of the sample 101 and combined to create a single measurement image. Then, an unfolded image is created by combining sub-unfolded images obtained by unfolding the design pattern data of the areas corresponding to each sub-measurement image. Then, coefficients of a filter function for matching this unfolded image to the measurement image as closely as possible are calculated, and the obtained coefficients are used as common parameters for generating a reference image.

[0033] The common parameters obtained in this way, for example, are in a negotiable relationship between the effect of making the image resemble a fine pattern and the effect of making the image resemble a large pattern, resulting in a low similarity between the created reference image and the measurement image. This results in the generation of many false defects. On the other hand, if the judgment threshold is relaxed to suppress false defects, it will result in defects not being detected.

[0034] Therefore, in the first embodiment, a filter coefficient is calculated for each pattern type. Then, a plurality of reference images are created using the filter coefficient for each pattern type. Then, an optimum reference image is selected from the plurality of reference images and used. This will be explained in detail below.

[0035] Fig. 4 is a block diagram showing an example of the internal configuration of the filter function calculation circuit in Embodiment 1. In Fig. 4, filter function calculation circuit 140 includes storage devices 70, 71, 75, and 78 such as magnetic disk devices, a frame selection unit 72, a resizing / rounding processing unit 73, a frame image creation unit 74, a filter coefficient calculation unit 76, a similarity parameter calculation unit 80, a similarity determination unit 82, and a determination unit 84. A series of "units" such as the frame selection unit 72, resizing / rounding processing unit 73, frame image creation unit 74, filter coefficient calculation unit 76, similarity parameter calculation unit 80, similarity determination unit 82, and determination unit 84 each have a processing circuit. Such processing circuits include an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. Furthermore, each "unit" may use a common processing circuit (the same processing circuit), or may use different processing circuits (separate processing circuits). Input data or calculation results required for the frame selection unit 72, resizing / rounding processing unit 73, frame image creation unit 74, filter coefficient calculation unit 76, similarity parameter calculation unit 80, similarity determination unit 82, and determination unit 84 are stored each time in a memory (not shown) within the filter function calculation circuit 140 or in the memory 116.

[0036] Fig. 5 is a flowchart showing an example of main steps of the inspection method according to Embodiment 1. In Fig. 5, a series of steps are performed: a pattern type frame selection step (S102), a stripe image acquisition step (S104), a representative frame image creation step (S106), a representative exfoliated image creation step (S110), a filter coefficient calculation step (S130), a representative reference image creation step (S140), a similarity parameter calculation step (S150), a similarity determination step (S152), a filter function determination step (S154), a stripe image acquisition step (S202), a frame image creation step (S204), an exfoliated image creation step (S212), a reference image creation step (S214), and a comparison step (S230).

[0037] In the pattern type frame selection step (S102), the frame selection unit 72 selects a plurality of representative frames of different pattern types from the inspection area 10. As shown in FIG. 2, the stripe area of ​​each inspection stripe 20 is divided into, for example, a plurality of frame areas 30, and inspection is performed for each frame area 30. In other words, the inspection area 10 of the sample 101 is virtually divided into a plurality of frame areas 30, and inspection is performed for each frame area 30. Here, a plurality of frame areas (representative frames) of different pattern types are selected from the inspection area 10 of the sample 101.

[0038] Fig. 6 is a diagram showing an example of a representative frame in the first embodiment. In the example of Fig. 6, a plurality of alignment marks 11 are formed on the sample 101 around an inspection area 10 on which a pattern is formed. A plurality of representative frames 13a-c are selected from the inspection area 10. In practice, a frame area 30 on which a plurality of pattern types are formed is extracted and selected from the design image data that is the source of the pattern formed on the sample 101. The plurality of pattern types includes at least a fine pattern and a large pattern.

[0039] FIG. 7 is a diagram showing an example of a fine pattern image in the first embodiment. In the example of FIG. 7, examples of the fine patterns shown include a line and space pattern (vertical L / S pattern) extending in the y direction, a line and space pattern (horizontal L / S pattern) extending in the x direction, a plurality of hole patterns (black hole patterns) arranged in an array, and a plurality of hole patterns (white hole patterns) in which the black hole patterns are inverted. However, the fine patterns are not limited to these. Other fine patterns may also be used. Note that a white pattern refers to a pattern with a high gradation value in the image, and a black pattern refers to a pattern with a low gradation value in the image. For example, in an image defined with a resolution of 256 gradations, a pattern with approximately 200 gradations may be used as a white pattern, and a pattern with approximately 20 gradations may be used as a black pattern.

[0040] Fig. 8 is a diagram showing an example of a large pattern image in the first embodiment. In the example of Fig. 8, the large patterns shown include, for example, a cross pattern (black cross pattern) formed large in the image, a cross pattern (white cross pattern) obtained by inverting the black cross pattern, a rectangular pattern (black rectangular pattern) formed large in the image, a rectangular pattern (white rectangular pattern) obtained by inverting the black rectangular pattern, an L-shaped pattern (black L-shaped pattern) formed large in the image, and an L-shaped pattern (white L-shaped pattern) obtained by inverting the black L-shaped pattern. However, the large patterns are not limited to these. Other large patterns may also be used.

[0041] Not all of these fine patterns and large patterns are necessarily formed in the sample 101 to be inspected. Therefore, a frame region in which at least one of the fine patterns is formed and a frame region in which at least one of the large patterns is formed may be selected as representative frames.

[0042] Other pattern types may include multiple basic geometric patterns in which the orientation of the basic geometric figure is reversed, such as a pattern in which half of the image is black and the other half is white. If these multiple basic geometric patterns do not exist in the sample 101, they may be prepared as learning templates. The multiple basic geometric patterns may be omitted when calculating the filter coefficients.

[0043] The optical image acquisition mechanism 150 acquires a plurality of representative frame images (representative optical images) of a plurality of representative frames 13 (representative regions) of the pattern-formed inspection sample 101. This will be explained in detail below.

[0044] In the stripe image acquisition step (S104), the optical image acquisition mechanism 150 acquires, for each representative frame, an optical image of the inspection stripe 20 including the representative frame 13 of the sample 101. Specifically, the operation is as follows.

[0045] First, the XYθ table 102 is moved to a position where an inspection stripe 20 including a target representative frame 13 among a plurality of representative frames of different pattern types can be imaged. A laser beam (e.g., DUV light) having a wavelength in the ultraviolet range or shorter, which serves as inspection light, is irradiated onto the pattern formed on the sample 101 from an appropriate light source 103 via an illumination optical system 170. The light transmitted through the sample 101 passes through a magnifying optical system 104 and is focused as an optical image on an imaging sensor 105 (an example of a sensor), where it is incident.

[0046] The pattern image formed on the image sensor 105 is photoelectrically converted by each light-receiving element of the image sensor 105, and then A / D (analog-to-digital) converted by the sensor circuit 106. Then, pixel data of the inspection stripe 20 to be measured is stored in the stripe pattern memory 123. Thereafter, the stripe area image is sent to the filter function calculation circuit 140 together with data indicating the position of the sample 101 on the XYθ table 102 output from the position circuit 107. The measurement data (pixel data) is, for example, 8-bit unsigned data, and represents the brightness gradation (light amount) of each pixel. The stripe area image output to the filter function calculation circuit 140 is stored in the storage device 71.

[0047] In the representative frame image creation step (S106), the frame image creation unit 74 creates a frame image (representative frame image) of the target representative frame 13 for each representative frame. Specifically, the stripe region image is divided into a predetermined size in the x and y directions so as to extract the frame image (representative frame image) of the target representative frame 13 from the stripe region image (optical image) of the inspection stripe 20 including the representative frame 13. For example, the stripe region image is divided into frame images of 512 x 512 pixels. Data of the divided representative frame images is output to and stored in the storage device 75.

[0048] As a result, a plurality of representative frame images of a plurality of representative frames with different pattern types are acquired and stored in the storage device 75.

[0049] In the representative exfoliated image creation step (S110), the unfolding circuit 111 (an example of an unfolded image creation unit) creates a representative exfoliated image for each of the plurality of representative frames 13 by unfolding the image using design data of the pattern to be formed in the representative frame image of that representative frame 13. Specifically, for each of the plurality of representative frames 13 having different pattern types, the design data is read from the magnetic disk device 109 via the control computer 110, and each graphic pattern in the area of ​​the target representative frame 13 defined in the read design data is converted into binary or multi-value image data (image unfolding) to create an unfolded image (representative exfoliated image).

[0050] Here, the figures defined in the design pattern data are, for example, rectangles or triangles as basic figures, and graphic data (vector data) is stored that defines the shape, size, position, etc. of each pattern figure using information such as the coordinates (x, y) at the reference position of the figure, the length of the sides, and a figure code that serves as an identifier to distinguish the type of figure, such as a rectangle or triangle.

[0051] When the information on the design pattern that becomes such graphic data is input to the expansion circuit 111, it is expanded into data for each graphic, and the graphic code and graphic dimensions that indicate the graphic shape of the graphic data are interpreted. Then, binary or multi-valued design image data is expanded as patterns to be arranged in squares with a grid of a predetermined quantized dimension as a unit, and output. In other words, the design data is read, the inspection area is virtually divided into squares with a predetermined dimension as a unit, and the occupancy rate of the graphic in the design pattern is calculated for each square, and n-bit occupancy data is output. For example, it is preferable to set one square as one pixel. Then, 1 / 2 is assigned to one pixel. 8 If a pixel has a resolution of (=1 / 256), a small area of ​​1 / 256 is allocated to the area of ​​the figure placed within the pixel, and the occupancy rate within the pixel is calculated. Then, a representative exfoliated image of 8-bit occupancy data is created for each pixel. The representative exfoliated image data is output to the filter function calculation circuit 140 and stored in the storage device 70.

[0052] As a result, a plurality of representative exfoliated images of a plurality of representative frames with different pattern types are stored in the storage device 70.

[0053] Here, for the representative developed image and representative optical image for each pattern type, it is also preferable to combine multiple pattern images of the same pattern type to create one representative developed image and one representative optical image. For example, for a fine pattern, two or more images of the four pattern images shown in Figure 7 may be combined to create one image. Similarly, for a large pattern, two or more images of the six pattern images shown in Figure 8 may be combined to create one image.

[0054] In the filter coefficient calculation step (S130), first, the resizing / rounding processor 73 performs resizing and / or rounding on patterns in a plurality of representative exfoliated images of different pattern types.

[0055] FIG. 9 shows an example of a graphic pattern before and after resizing in the first embodiment. The example in FIG. 9 shows a graphic pattern 14 in which the width of the graphic pattern 12 in the representative expanded image is resized by a resize amount A'. The resize amount is set in advance. For example, the graphic pattern 14 is resized in 0.5 pixel increments within a range of -5 pixels to +5 pixels. A negative value indicates a resize that decreases the width. A positive value indicates a resize that increases the width. For example, 0.5 pixels indicates a resize amount that is half the pixel size. The example in FIG. 9 shows a case in which the width of the graphic pattern 12 is resized to decrease. While the example in FIG. 9 shows an example in which the width is resized in the x direction, the width may be resized in the y direction instead of the x direction. Alternatively, the width may be resized in both the x direction and the y direction.

[0056] FIG. 10 shows an example of a graphic pattern before and after rounding in the first embodiment. The example in FIG. 10 shows a graphic pattern 16 in which the corners of a resized graphic pattern 14 are rounded by a rounding amount B'. When the rounding amount ranges from 0 to +2 pixels, for example, the rounding amount is rounded in increments of 0.2 pixels. For example, 0.2 pixels indicates a rounding amount that is 1 / 5 of the pixel size. The rounding amount indicates the radius of curvature. In the example in FIG. 10, the upper left corner is shown, but other corners are also rounded in the same manner.

[0057] A plurality of combinations of these resizing amounts and rounding amounts are set, and for each combination, resizing and rounding are performed on a plurality of representative exfoliated images of different pattern types. As a result, a plurality of representative exfoliated images of different pattern types are created, each corresponding to the number of combinations of resizing amounts and rounding amounts.

[0058] Although it is desirable to perform resizing and / or rounding, this is not a limitation, and resizing and / or rounding may be omitted.

[0059] Next, the filter coefficient calculation unit 76 (parameter calculation unit) calculates a plurality of generation parameters for creating reference images using a plurality of different pattern types formed in a plurality of representative frames (representative regions) of the inspected sample 101. Specifically, it operates as follows. The filter coefficient calculation unit 76 calculates the coefficients of a filter function for creating a representative reference image for each pattern type. Here, the filter coefficient calculation unit 76 calculates the coefficients of a filter function for creating a representative reference image for each combination of resizing amount and rounding amount and for each pattern type, for example.

[0060] FIG. 11 is a diagram for explaining an example of the relationship between the filter function, the unfolded image, and the measurement image according to the first embodiment. FIG. 12 is a diagram illustrating an example of a technique for calculating coefficients of a filter function in the first embodiment. For example, as shown in FIG. 12, an unknown coefficient matrix a(i,j) (an example of a filter function) consisting of k×k elements, which is fewer than the number of pixels in the frame area, is calculated using the following technique. For example, for a representative frame image consisting of 512×512 pixels, a 15×15 coefficient matrix a(i,j) is calculated. The sum of the products of k×k pixels and the coefficient matrix a(i,j) centered on a pixel of interest d(i,i) in the expanded image is divided by the number of pixels N (=k×k) to calculate the coefficient matrix a(i,j) that most closely matches the pixel of interest r(i,i) of the frame image of the representative frame corresponding to the pixel of interest d(i,i). The relational expression (1) is shown below.

[0061]

number

[0062] As shown in FIG. 12, while the pixel of interest is moved within the target representative frame, relational expression (1) is calculated each time. Then, the coefficient matrix a(i,j) that best satisfies relational expression (1) defined using the unknown coefficient matrix a(i,j) obtained for each pixel in the target representative frame is found. The number of elements k×k of the coefficient matrix a(i,j) can be set appropriately. If it is too small, accuracy will deteriorate, and if it is too large, the calculation time will be long. Also, when the pixel of interest moves within the representative frame, if it is close to the edge, there may not be enough surrounding pixels on the edge side. In such cases, calculation can be done using the surrounding pixels and the number of pixels N from which values ​​can be obtained.

[0063] As a result of the above, the coefficient matrix a(i,j) of the filter function is obtained for each pattern type. Here, for example, the coefficient matrix a(i,j) of the filter function is obtained for each combination of the resizing amount and the rounding amount and for each pattern type.

[0064] As described above, the coefficient matrix a(i,j) obtained for each pattern type is an example of a reference image generation parameter. Here, for example, a set of the resizing amount and rounding amount and the coefficient matrix a(i,j) obtained for each combination of the resizing amount and rounding amount and for each pattern type is an example of a reference image generation parameter.

[0065] As a result of the above, n reference image generation parameters (n is an integer equal to or greater than 2) are obtained. Here, the multiple reference image generation parameters include, for example, parameters estimated from fine patterns and parameters estimated from large patterns.

[0066] In the representative reference image creation step (S140), the reference circuit 112 (reference image creation unit) creates a representative reference image by filtering a representative exfoliated image of the representative frame 13 that was the basis for calculating the reference image generation parameters, using the reference image generation parameters for each of a plurality of reference image generation parameters. Specifically, the operation is as follows: The reference circuit 112 creates a representative reference image for each pattern type using the generated filter function. Here, for example, a representative reference image is created for each combination of resizing amount and rounding amount and for each pattern type. Specifically, the operation is as follows: For each combination of resizing amount and rounding amount and for each pattern type, the reference circuit 112 creates reference image data for each pixel in the exfoliated image that has been processed using the combination of resizing amount and rounding amount of the target representative frame, by dividing the sum of the products of k×k pixels with the pixel d(i,i) at the center and a coefficient matrix a(i,j) that serves as a filter function by the number of pixels N (=k×k).

[0067] For each combination of resizing amount and rounding amount and for each pattern type, a reference image is created using the reference image generation parameters obtained. The obtained multiple reference image data are sent to the filter function calculation circuit 140 and stored in the storage device 78.

[0068] In the similarity parameter calculation step (S150), the similarity parameter calculation unit 80 (similarity calculation unit) calculates, for each of the plurality of reference image generation parameters, a similarity parameter indicating the similarity between a representative reference image created using the reference image generation parameter and a representative optical image of the representative frame. In other words, the similarity parameter calculation unit 80 (similarity calculation unit) calculates, for each of the plurality of reference image generation parameters, a similarity parameter indicating the similarity between a representative reference image created using the reference image generation parameter and a representative frame image of the representative frame 13 from which the reference image generation parameter was calculated. Specifically, the operation is as follows: The similarity parameter calculation unit 80 calculates, for each pattern type, a similarity parameter indicating the similarity between the representative reference image and the representative frame image. Here, the similarity parameter calculation unit 80 calculates, for example, a similarity parameter indicating the similarity between the representative reference image and the representative frame image for each combination of resizing amount and rounding amount and for each pattern type. As the similarity parameter, it is preferable to use, for example, the sum of squares of the gray level difference as shown in the following equation (2) using pixel f(i,j) of the representative reference image and pixel r(i,j) of the representative frame image.

[0069]

number

[0070] In the similarity determination step (S152), the similarity determination unit 82 determines the reference image generation parameter with the highest similarity among the multiple reference image generation parameters, which is ranked in a predetermined order. Here, the top reference image generation parameter in the predetermined order with the highest similarity is determined. Specifically, the similarity determination unit 82 determines the top reference image generation parameter with the highest similarity between the representative reference image created using the reference image generation parameters and the representative frame image, from the multiple reference image generation parameters created for each pattern type. Here, the similarity determination unit 82 determines the top reference image generation parameter with the highest similarity between the representative reference image created using the reference image generation parameters and the representative frame image, from the multiple reference image generation parameters created for each combination of resizing amount and rounding amount and for each pattern type, for example.

[0071] In the filter function determination step (S154), the determination unit 84 determines the determined top reference image generation parameter as a common parameter. The determined common parameter is set in the reference circuit 112. Then, the inspection process for the entire sample 101 is started.

[0072] The optical image acquisition mechanism acquires an inspection frame image (inspection optical image) of the sample 101. This will be specifically described below. In the stripe image acquisition step (S202), the optical image acquisition mechanism 150 acquires an optical image of the pattern-formed sample 101 to be inspected. The method for acquiring the stripe image is the same as that described above. Here, stripe images are acquired in order as shown in FIG. 2. Then, pixel data is stored in the stripe pattern memory 123 for each inspection stripe 20. Thereafter, the stripe region image is sent to the comparison circuit 108 together with data indicating the position of the photomask 101 on the XYθ table 102 output from the position circuit 107. The measurement data (pixel data) is, for example, 8-bit unsigned data, and represents the brightness gradation (light amount) of each pixel.

[0073] FIG. 13 is a diagram showing an example of the internal configuration of a comparison circuit according to the first embodiment. In FIG. 13, comparison circuit 108 includes storage devices 50, 52, and 56, such as a magnetic disk device, a frame image creation unit 54, an alignment unit 57, and a comparison unit 58. A series of "units" such as frame image creation unit 54, alignment unit 57, and comparison unit 58 each have a processing circuit. Such processing circuits include an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. Each "unit" may share a common processing circuit (the same processing circuit), or may use different processing circuits (separate processing circuits). Input data or calculation results required for frame image creation unit 54, alignment unit 57, and comparison unit 58 are stored in a memory (not shown) within comparison circuit 108 or memory 116 each time.

[0074] The stripe region image outputted into the comparison circuit 108 is stored in the storage device 50 .

[0075] In the frame image creation step (S204), the frame image creation unit 54 generates a plurality of frame images 31 by dividing the stripe region image (optical image) at a predetermined width. Specifically, as shown in FIG. 2, the stripe region image is divided into frame images of a plurality of rectangular frame regions 30. For example, the size is divided into frames of 512 x 512 pixels. Data for each frame region 30 is stored in the storage device 56.

[0076] In this way, the optical image acquisition mechanism 150 acquires a frame image 31 (measurement image, optical image) for inspection of the frame region 30 (predetermined region) of the sample 101 to be inspected on which a pattern has been formed.

[0077] In the exfoliated image creation step (S212), the exfoliated image creation circuit 111 (exfoliated image creation unit) creates an exfoliated image for inspection by image development using design data that is the basis for a pattern to be formed in the optical image for inspection. In other words, the exfoliated image creation circuit 111 (exfoliated image creation unit) creates an exfoliated image for inspection by image development using design data that is the basis for a pattern to be formed in the optical image of a predetermined area. Specifically, the exfoliated image creation circuit 111 reads the design data from the magnetic disk device 109 through the control computer 110, and converts each graphic pattern of each frame area 30 defined in the read design data into binary or multi-value image data. Then, a design image of 8-bit occupancy data for each pixel is created. The design image data (image data) is output to the reference circuit 112.

[0078] In the reference image creation step (S214), the reference circuit 112 creates a reference image for inspection using parameters with a predetermined ranking that have the highest similarity. Here, the reference circuit 112 creates a reference image for inspection by filtering the unfolded image for inspection using the top reference image creation parameter. Specifically, the reference circuit 112 resizes and rounds the exfoliated image (exfoliated image for inspection) according to the set resizing and rounding amounts, and then performs filtering using the set coefficient matrix a(i,j) (an example of a filter function) to create a reference image. The created reference image is output to the comparison circuit 108, and the reference image output to the comparison circuit 108 is stored in the storage device 52.

[0079] In the comparison step (S230), the comparison circuit 108 (an example of a comparison unit) compares the reference image for inspection with the optical image for inspection. Here, the comparison circuit 108 compares the reference image for inspection created using parameters with a predetermined ranking that have the highest similarity with the optical image for inspection. In other words, the comparison circuit 108 (an example of a comparison unit) compares the frame image for inspection (optical image) with the reference image (reference image for inspection) created using the top reference image generation parameter. Specifically, the operation is as follows. First, the alignment unit 57 reads the frame image (optical image) to be compared from the storage device 56, and similarly reads the reference image to be compared from the storage device 52. Then, alignment is performed using a predetermined algorithm. For example, alignment is performed using the least squares method. The comparison unit 58 then compares the two for each pixel in accordance with predetermined judgment conditions to determine whether or not there is a defect, such as a shape defect. The judgment conditions may, for example, be a predetermined algorithm for comparing the two for each pixel to determine whether or not there is a defect. The comparison result is then output. The comparison result may be output to, for example, the magnetic disk device 109, the flexible disk device (FD) 116, the CRT 117, or the pattern monitor 118, or may be output from the printer 119.

[0080] As described above, according to the first embodiment, the generation parameters with the highest similarity are used, and therefore inspection can be performed using a reference image created with effective reference image generation parameters suited to the pattern to-be-inspected, thereby suppressing or reducing false defects and / or missed defect detection.

[0081] Embodiment 2 In the first embodiment, a case has been described in which the comparison result with a reference image created using the generation parameter with the highest similarity among n (n≧2) reference image generation parameters is used as is, but this is not limited to this. In the second embodiment, a configuration will be described in which a plurality of comparison results are used. The contents below are the same as those in the first embodiment except for points that will be particularly described.

[0082] Fig. 14 is a configuration diagram showing the configuration of a pattern inspection device according to embodiment 2. Fig. 14 is the same as Fig. 1 except that one comparison circuit 108 is replaced by two comparison circuits 108a and 108b, and that a summing circuit 142 is added. The internal configurations of the comparison circuits 108a and 108b are the same as those in Fig. 13.

[0083] Fig. 15 is a flowchart showing an example of the main steps of the inspection method according to embodiment 2. Fig. 15 is the same as Fig. 5 except that a reference image creation step (S224), a comparison step (S232), and a counting step (S240) are added.

[0084] The contents of each of the pattern type frame selection process (S102), stripe image acquisition process (S104), representative frame image creation process (S106), representative exfoliated image creation process (S110), filter coefficient calculation process (S130), representative reference image creation process (S140), and similarity parameter calculation process (S150) are the same as those in embodiment 1.

[0085] In the second embodiment, n (n≧3) reference image generation parameters are obtained. Here, the multiple reference image generation parameters include, for example, parameters estimated from fine patterns and parameters estimated from large patterns.

[0086] In the similarity determination step (S152), the similarity determination unit 82 determines, from among the multiple reference image generation parameters, multiple parameters in a predetermined order that have the highest similarity. Here, the similarity determination unit 82 determines the top two parameters with the highest similarity from among the multiple reference image generation parameters. Specifically, from among the multiple reference image generation parameters created for each pattern type, the similarity determination unit 82 determines the top two reference image generation parameters that have the highest similarity between the representative reference image created using the reference image generation parameters and the representative frame image. Here, the similarity determination unit 82 determines, for example, from among the multiple reference image generation parameters created for each combination of resizing amount and rounding amount and for each pattern type, the top two reference image generation parameters that have the highest similarity between the representative reference image created using the reference image generation parameters and the representative frame image.

[0087] In the filter function determination step (S154), the determination unit 84 determines the determined top two reference image generating parameters as two common parameters. The determined two common parameters are set in the reference circuit 112.

[0088] In the above example, the top two reference image generation parameters are determined as common parameters, but the top three or more reference image generation parameters may be determined as common parameters. Then, the inspection process for the entire sample 101 is started.

[0089] The contents of each of the stripe image acquisition step (S202), frame image creation step (S204), and unfolded image creation step (S212) are the same as those in the first embodiment. Therefore, the optical image acquisition mechanism 150 acquires a frame image 31 (optical image, optical image for inspection) for each frame area 30 (predetermined area) of the pattern-formed sample to be inspected 101. The unfolding circuit 111 creates an unfolded image for inspection (unfolded image for inspection) for each frame area 30, which is unfolded using design data that is the basis of the pattern to be formed in the frame image of the frame area 30 (predetermined area).

[0090] In the reference image creation step (S214) and the reference image creation step (S224), the reference circuit 112 (reference image creation unit) creates multiple reference images by filtering the unfolded image using different ones of multiple reference image generation parameters. In the second embodiment, the reference circuit 112 (reference image creation unit) creates multiple reference images for inspection using multiple parameters with a predetermined ranking of high similarity. Here, two reference images are created from the same unfolded image using the top two reference image generation parameters with high similarity. Specifically, the operation is as follows.

[0091] In the reference image creation step (S214), the reference circuit 112 (reference image creation unit) creates a first reference image for inspection (first inspection reference image) by filtering the unfolded image for inspection using one of the set top two reference image creation parameters. The reference circuit 112 resizes and rounds the unfolded image (unfolded image for inspection) using the resizing amount and rounding amount of one of the top two reference image creation parameters, and then creates a reference image by filtering using a coefficient matrix a(i,j) (an example of a filter function) of one of the top two reference image creation parameters. The created reference image is output to the comparison circuit 108a, and the reference image output to the comparison circuit 108a is stored in the storage device 52.

[0092] In the reference image creation step (S224), the reference circuit 112 (reference image creation unit) creates a second reference image for inspection (second inspection reference image) that has been subjected to inspection exfoliated image filtering using the other of the two highest-ranked reference image creation parameters that have been set. The reference circuit 112 resizes and rounds the exfoliated image (exfoliated image for inspection) using the resizing and rounding amounts of the other of the two highest-ranked reference image creation parameters, and then filters the exfoliated image using the coefficient matrix a(i,j) (an example of a filter function) of the other of the two highest-ranked reference image creation parameters to create a reference image. The created reference image is output to the comparison circuit 108b, and the reference image output to the comparison circuit 108b is stored in the storage device 52.

[0093] In the comparison step (S230) and the comparison step (S232), the comparison circuits 108a and 108b compare a plurality of reference images for inspection (reference images for inspection) with the frame image for inspection (optical image, optical image for inspection), respectively. In other words, the comparison circuits 108a and 108b compare a plurality of reference images for inspection created using a plurality of parameters in a predetermined order (here, the top two) with high similarity with the optical image for inspection, respectively. Specifically, the operation is as follows.

[0094] In the comparison step (S230), the comparison circuit 108a (an example of a comparison unit) compares an inspection frame image (optical image) with a reference image (inspection reference image) created using one of the top two reference image generation parameters. Specifically, the operation is as follows. First, the alignment unit 57 in the comparison circuit 108a reads the frame image (optical image) to be compared from the storage device 56, and similarly reads the reference image to be compared from the storage device 52. Then, alignment is performed using a predetermined algorithm. For example, alignment is performed using the least squares method. The comparison unit 58 in the comparison circuit 108a then compares the two for each pixel in accordance with predetermined criteria to determine whether or not there is a defect, such as a shape defect. The criteria may, for example, be a predetermined algorithm for comparing the two for each pixel to determine whether or not there is a defect. The comparison result is then output. The comparison result is then output to the summing circuit 142.

[0095] In the comparison step (S232), the comparison circuit 108b (an example of a comparison unit) compares the frame image (optical image) for inspection with a reference image (reference image for inspection) created using the other of the top two reference image generation parameters. Specifically, the operation is as follows. First, the alignment unit 57 in the comparison circuit 108b reads the frame image (optical image) to be compared from the storage device 56, and similarly reads the reference image to be compared from the storage device 52. Then, alignment is performed using a predetermined algorithm. For example, alignment is performed using the least squares method. The comparison unit 58 in the comparison circuit 108b then compares the two for each pixel in accordance with predetermined criteria to determine whether or not there is a defect, such as a shape defect. The criteria may, for example, be a predetermined algorithm for comparing the two for each pixel to determine whether or not there is a defect. The comparison result is then output. The comparison result is then output to the summing circuit 142.

[0096] In the above example, the comparison step (S230) and the comparison step (S232) are performed by two comparison circuits 108a and 108b, but this is not limiting and the steps may be performed by the same comparison circuit. In this case, the order of the steps may be any.

[0097] In the counting step (S240), the counting circuit 142 (counting unit) counts the comparison results for each of the compared reference images and outputs the results. As a counting method, a false defect reduction priority mode and a sensitivity priority mode can be used.

[0098] Fig. 16 is a diagram showing an example of a method of counting in the second embodiment. In Fig. 16, for each frame region 30, a frame image (sensor image) is compared with a reference image (reference image A) created using one of the top two reference image generation parameters in terms of similarity and a reference image (reference image B) created using the other of the top two reference image generation parameters in terms of similarity. Thus, a defect map of the comparison result A with reference image A and a defect map of the comparison result B with reference image B are obtained.

[0099] In the example of FIG. 16, the summing circuit 142 outputs, as a defect, a location that is commonly determined to be a defect in comparison results A and B for each reference image (false defect reduction priority mode). The example of FIG. 16 shows a case where a defect in one location in the center of the image is determined to be a defect in comparison result A and also in comparison result B. Therefore, this common defect is output as the summation result. Alternatively, the inspection results may be summed as follows. Therefore, false defects can be suppressed or reduced.

[0100] FIG. 17 is a diagram showing another example of the counting method in the second embodiment. In the example of FIG. 17, the counting circuit 142 outputs, as defects, any location determined to be defective in either comparison result A or B for each reference image (sensitivity priority mode). The example of FIG. 17 shows multiple defect locations determined to be defective in comparison result A and multiple defect locations determined to be defective in comparison result B. All locations determined to be defective in either of the two are counted as defects. This counting result is output. Therefore, it is possible to suppress or reduce missed defect detection.

[0101] The results of the counting may be output to, for example, the magnetic disk device 109, the flexible disk device (FD) 116, the CRT 117, the pattern monitor 118, or the printer 119.

[0102] As described above, according to the second embodiment, a comparison is made between a plurality of reference images using a plurality of generation parameters with high similarity, and therefore inspection can be performed using a reference image created using effective reference image generation parameters that match the pattern to be inspected. This makes it possible to suppress or reduce false defects and / or defect detection omissions. In other words, it is possible to achieve both high-sensitivity inspection and false defect suppression.

[0103] Embodiment 3 In the second embodiment, a configuration was described in which the top two reference image generation parameters with the highest similarity are selected from n (n≧3) reference image generation parameters, two reference images are created using different reference image generation parameters, and both are compared with the frame image. In the third embodiment, a configuration will be described in which, for example, n=2 reference image generation parameters are created and similarity determination is omitted. The contents below are the same as those in the second embodiment except for points that will be particularly described.

[0104] The configuration of the pattern inspection device in the third embodiment is the same as that shown in FIG.

[0105] 18 is a block diagram showing an example of the internal configuration of a filter function calculation circuit according to embodiment 3. It is the same as FIG. 4 except that the similarity parameter calculation unit 80, the similarity determination unit 82, and the determination unit 84 are omitted.

[0106] A series of "units" such as the frame selection unit 72, resizing / rounding processing unit 73, frame image creation unit 74, and filter coefficient calculation unit 76 have processing circuits. Such processing circuits include electric circuits, computers, processors, circuit boards, quantum circuits, semiconductor devices, etc. Furthermore, each "unit" may use a common processing circuit (the same processing circuit), or may use different processing circuits (separate processing circuits). Input data or calculation results required for the frame selection unit 72, resizing / rounding processing unit 73, frame image creation unit 74, and filter coefficient calculation unit 76 are stored each time in a memory (not shown) within the filter function calculation circuit 140 or in memory 116.

[0107] Fig. 19 is a flowchart showing an example of the main steps of the inspection method according to embodiment 3. Fig. 19 is the same as Fig. 15 except that the representative reference image creation step (S140), the similarity parameter calculation step (S150), the similarity determination step (S152), and the filter function determination step (S154) are omitted.

[0108] In the pattern type frame selection step (S102) in the third embodiment, the frame selection unit 72 selects a plurality of representative frames 13 of different pattern types from the inspection area 10. Examples of different pattern types include fine patterns and large patterns. Here, a representative frame in which a fine pattern is arranged and a representative frame in which a large pattern is arranged are selected from the inspection area 10 of the sample 101.

[0109] The optical image acquisition mechanism 150 acquires a plurality of representative frame images (image representative optical images) of a plurality of representative frames 13 (representative regions). Here, two representative frame images (image representative optical images) of two representative frames 13 (representative regions) are acquired. The stripe image acquisition step (S104) and the representative frame image creation step (S106) are the same as those in the second embodiment (first embodiment).

[0110] As a result, two representative frame images (representative optical images) of two representative frames 13 (representative regions) with different pattern types are acquired and stored in the storage device 75.

[0111] The contents of the representative exfoliated picture creating step (S110) are the same as those in the second embodiment (first embodiment).

[0112] As a result, two representative exfoliated images of two representative frames with different pattern types are stored in the storage device 70.

[0113] In the filter coefficient calculation step (S130), the filter coefficient calculation unit 76 (parameter calculation unit) calculates a plurality of generation parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative frames (representative regions) of the sample 101 to be inspected. Here, the filter coefficient calculation unit 76 (parameter calculation unit) calculates two generation parameters for creating a reference image using a plurality of different pattern types formed in two representative frames (representative regions) of the sample 101 to be inspected. In the third embodiment, two reference image generation parameters for different pattern types are calculated, so resizing and / or rounding are omitted. Alternatively, one combination of resizing amount and rounding amount may be set for each pattern type.

[0114] Here, for the representative developed image and representative optical image for each pattern type, it is also preferable to combine multiple pattern images of the same pattern type to create one representative developed image and one representative optical image. For example, for a fine pattern, two or more images of the four pattern images shown in Figure 7 may be combined to create one image. Similarly, for a large pattern, two or more images of the six pattern images shown in Figure 8 may be combined to create one image.

[0115] As a result of the above, a coefficient matrix a(i,j) of the filter function is obtained for each pattern type. Here, for example, for two pattern types, a coefficient matrix a(i,j) of the filter function is obtained for each pattern type. Here, the multiple reference image generation parameters include, for example, parameters estimated from fine patterns and parameters estimated from large patterns.

[0116] As a result of the above, n (n=2) reference image generation parameters are obtained.

[0117] The contents of each of the stripe image acquisition step (S202), frame image creation step (S204), exfoliated image creation step (S212), reference image creation step (S214), comparison step (S230), reference image creation step (S224), comparison step (S232), and tallying step (S240) are the same as those in the second embodiment. The reference circuit 112 creates multiple reference images by filtering the exfoliated image using different ones of multiple reference image generation parameters. Here, two reference images are created from the same exfoliated image using two reference image generation parameters.

[0118] As described above, according to the third embodiment, since a comparison is made between a plurality of reference images using a plurality of generation parameters for different pattern types without performing a similarity determination, inspection can be performed using a reference image created using effective reference image generation parameters that match the pattern to be inspected. This makes it possible to suppress or reduce false defects and / or missed defect detection. In other words, it is possible to achieve both high-sensitivity inspection and false defect suppression.

[0119] The above describes the embodiments with reference to specific examples. However, the present invention is not limited to these specific examples. For example, in the embodiments, a transmitted illumination optical system using transmitted light is shown as the illumination optical system 170, but the present invention is not limited to this. For example, a reflective illumination optical system using reflected light may also be used. Alternatively, a transmitted illumination optical system and a reflective illumination optical system may be combined to use transmitted light and reflected light simultaneously.

[0120] The above-described filter functions and coefficients of the filter functions are merely examples, and the present invention is not limited to these. Other filter functions and coefficients of the filter functions may also be used.

[0121] Furthermore, although descriptions of the device configuration, control method, and other parts not directly necessary for the explanation of the present invention have been omitted, the required device configuration and control method can be appropriately selected and used. For example, although the description of the control unit configuration that controls the inspection device 100 has been omitted, it goes without saying that the required control unit configuration can be appropriately selected and used.

[0122] In addition, all pattern inspection devices and pattern inspection methods that include the elements of the present invention and that can be appropriately modified by those skilled in the art are included in the scope of the present invention. [Explanation of symbols]

[0123] 10 Inspection Area 12, 14, 16 geometric patterns 20 Inspection Stripes 30 Frame Area 31 frame images 32 representative frames 54 Frame image creation section 57 Alignment section 58 Comparison Section 61,63,65 Storage device 70,71,75,78 Storage device 72 Frame selection section 73 Resize / Rounding Processor 74 Frame image creation section 76 Filter coefficient calculation section 80 Similarity parameter calculation unit 82 Similarity determination section 84 Decision Section 100 Inspection equipment 101 Sample 102 XYθ table 103 Light source 104 Magnifying Optical System 105 Photodiode array 106 Sensor Circuit 107 Position circuit 108 Comparison circuit 109 Magnetic disk unit 110 Control computer 111 Deployment Circuit 112 Reference circuit 113 Autoloader control circuit 114 Table control circuit 115FD 116 memory 117 CRT 118 Pattern Monitor 119 Printer 120 Bus 122 Laser length measurement system 123 Stripe Pattern Memory 140 Filter function calculation circuit 142 Totaling Circuit 150 Optical image acquisition unit 160 Control Circuits 170 Illumination optical system

Claims

1. a parameter calculation unit that calculates a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; an optical image acquisition mechanism for acquiring a plurality of representative optical images of the plurality of representative regions of the pattern-formed sample to be inspected; an exfoliated image creating unit that creates a representative exfoliated image for each of the representative regions by exposing the representative region using design data of a pattern to be formed in the representative optical image of the representative region; a reference image creation unit that creates a representative reference image by filtering a representative exfoliated image of the representative region using each of the plurality of parameters; a similarity calculation unit that calculates, for each of the plurality of parameters, a similarity parameter that indicates a similarity between a representative reference image created using the parameter and a representative optical image of the representative region; a similarity determination unit that determines a parameter having a predetermined rank of high similarity among the plurality of parameters; a comparison unit that compares a reference image for inspection with an optical image for inspection; Equipped with the reference image creation unit creates the reference image for inspection using parameters in a predetermined order with high similarity; the optical image acquisition mechanism acquires the optical image for inspection; the comparison unit compares the reference image for inspection created using parameters in a predetermined order with a high degree of similarity with the optical image for inspection; A pattern inspection device characterized by:

2. the reference image creation unit creates a plurality of reference images for inspection using a plurality of parameters in a predetermined order of high similarity; the comparison unit compares the plurality of reference images for inspection created using a plurality of parameters in a predetermined order of high similarity with the optical image for inspection, 2. The pattern inspection apparatus according to claim 1, further comprising a counting unit that counts the comparison results for each of the compared reference images and outputs the results.

3. 2. The pattern inspection apparatus according to claim 1, wherein the parameters in the predetermined order having the highest similarity are the top one parameter or the top two parameters.

4. a parameter calculation unit that calculates a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; an optical image acquisition mechanism for acquiring an optical image of a predetermined area of ​​the pattern-formed sample to be inspected; an unfolded image creating unit that creates an unfolded image by unfolding the image using design data of a pattern to be formed in the optical image of the predetermined area; a reference image generating unit that generates a plurality of reference images by filtering the unfolded image using different ones of the plurality of parameters; a comparison unit that compares the optical image with each of the plurality of reference images; a counting unit that counts the comparison results for each of the compared reference images and outputs the results; A pattern inspection device comprising:

5. 5. The pattern inspection apparatus according to claim 1, wherein the plurality of parameters include a parameter estimated from a fine pattern and a parameter estimated from a large pattern.

6. 5. The pattern inspection apparatus according to claim 2, wherein the counting unit outputs, as defects, portions that are commonly determined to be defects from the comparison results for each reference image.

7. 5. The pattern inspection apparatus according to claim 2, wherein the counting unit outputs, as a defect, a portion determined to be a defect in any of the comparison results for each reference image.

8. calculating a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; acquiring a plurality of representative optical images of the plurality of representative regions of the patterned sample to be inspected; creating a representative expanded image for each of the plurality of representative regions by image expansion using design data of a pattern to be formed in the representative optical image of the representative region; creating a representative reference image by filtering a representative exfoliated image of the representative region using each of the plurality of parameters; calculating, for each of the plurality of parameters, a similarity parameter indicating a similarity between a representative reference image created using the parameter and a representative optical image of the representative region; determining a parameter having a predetermined rank that has the highest degree of similarity among the plurality of parameters; creating a reference image for inspection using the parameters with the highest similarity in a predetermined order; acquiring an optical image for inspection; comparing the reference image for inspection created using parameters in a predetermined order with the highest similarity with the optical image for inspection; A pattern inspection method comprising:

9. calculating a plurality of parameters for creating a reference image using a plurality of different pattern types formed in a plurality of representative regions of the sample to be inspected; acquiring an optical image of a predetermined area of ​​the patterned sample to be inspected; creating an expanded image by expanding the image using design data of a pattern to be formed in the optical image of the predetermined area; generating a plurality of reference images by filtering the unfolded image using different ones of the plurality of parameters; comparing the optical image with each of the plurality of reference images; a step of aggregating the comparison results for each of the compared reference images and outputting the results; A pattern inspection method comprising:

Images