Inspection device and inspection method

Abstract

To reduce the detection of a pseudo defect caused by a positional deviation between an inspection image and a reference image.SOLUTION: An inspection device 1 includes: an imaging mechanism 10; an image acquisition circuit 214 which extracts a first contour line; an expansion circuit 211; a classification circuit 212 which classifies pattern edges; a reference parameter estimation circuit 216 which estimates a reference parameter 222 class-by-class; and a reference image generation circuit 213 which executes resize processing and corner rounding processing based on the reference parameter to extract a second contour line from a pattern after the resize processing and the corner rounding processing; a comparison circuit which calculates a positional deviation amount between the first contour line and the second contour line pixel-by-pixel to calculate an error amount of the positional deviation amount class-by-class; and a correction circuit 217 which corrects the positional deviation amount using the class-by-class error amount.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an inspection apparatus and an inspection method for inspecting defects in a pattern formed on a sample. [Background technology]

[0002] In the manufacturing process of semiconductor devices, a circuit pattern is transferred onto a semiconductor substrate by reduction exposure using an exposure apparatus (also called a "stepper" or "scanner"). In the exposure apparatus, a mask (also called a "reticle") on which an original pattern (hereinafter simply referred to as "pattern") is formed is used to transfer the circuit pattern onto a semiconductor substrate (hereinafter also referred to as a "wafer").

[0003] For example, cutting-edge devices require the formation of circuit patterns with line widths of several nm. As circuit patterns become finer, the master patterns on masks also become finer. In addition, as circuit patterns become finer, super-resolution technologies such as OPC (Optical Proximity Correction) and SRAF (Sub Resolution Assist Feature) are being introduced, and master patterns become more complex. For this reason, mask inspection equipment is required to have high defect detection performance that can handle fine and complex master patterns.

[0004] There are two types of defect inspection methods: the D-DB (Die to Database) method, which compares an inspection image based on an image of a mask with a reference image based on design data, and the DD (Die to Die) method, which compares images of multiple areas consisting of the same pattern formed on a mask.

[0005] The pattern formed on the mask may have a dimensional shift from the design data due to the pattern formation (drawing) process. In the case of the D-DB method, when a dimensional shift occurs, the edge position of the pattern in the inspection image does not match the edge position of the reference image. When a positional shift occurs between the reference image and the inspection image, the misaligned area may be detected as a false defect.

[0006] For example, Patent Document 1 discloses a technique for detecting defects based on the amount of positional deviation between the contour of a pattern in an inspection image and the contour of a reference image.

[0007] In order to deal with the dimensional shift, when creating a reference image, corrections such as resizing to move the edge positions of the pattern and corner rounding to round the corners of the pattern are performed. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] JP 2009-44070 A Summary of the Invention [Problem to be solved by the invention]

[0009] The amount of misalignment between the inspection image and the reference image differs depending on the edge position of the pattern (e.g., corner portion and straight line portion, etc.). In other words, the degree of match between the inspection image and the reference image differs depending on the edge position of the pattern. For this reason, it is difficult to inspect the entire pattern using the same standard (threshold value).

[0010] The present invention has been made in consideration of these points. That is, the present invention is capable of performing class classification based on the edge shape of a pattern in an inspection device, correcting the amount of misalignment between an inspection image and a reference image based on the class classification, and inspecting the entire pattern using the same standard. As a result, an object of the present invention is to provide an inspection device and an inspection method that can reduce the detection of false defects caused by misalignment. [Means for solving the problem]

[0011] According to a first aspect of the present invention, an inspection apparatus includes an imaging mechanism for a sample, an image acquisition circuit for extracting a first contour from image data of the sample imaged by the imaging mechanism, an unfolding circuit for generating an unfolded image from design data, a class classification circuit for classifying edges of a pattern in the unfolded image, a reference parameter estimation circuit for estimating reference parameters including a resizing amount and a corner rounding amount for each classified class, a reference image generation circuit for performing resizing processing and corner rounding processing of the pattern based on the reference parameters and extracting a second contour from the pattern after the resizing processing and the corner rounding processing, a comparison circuit for comparing the first contour and the second contour to calculate an amount of positional deviation between the first contour and the second contour for each pixel and to calculate an amount of error in the positional deviation for each class, and a correction circuit for correcting the amount of positional deviation using the amount of error of the class corresponding to each pixel.

[0012] According to a second aspect of the present invention, an inspection method includes the steps of extracting a first contour from image data of a sample, generating an unfolded image from design data, classifying edges of a pattern from the unfolded image, estimating reference parameters including a resizing amount and a corner rounding amount for each classified class, performing resizing processing and corner rounding processing of the pattern based on the reference parameters, extracting a second contour from the pattern after resizing processing and corner rounding processing, comparing the first contour and the second contour to calculate an amount of misalignment between the first contour and the second contour for each pixel, calculating an amount of error in the amount of misalignment for each class, and correcting the amount of misalignment using the amount of error for each class corresponding to each pixel. Effect of the Invention

[0013] According to the inspection apparatus and inspection method of the present invention, when comparing an inspection image with a reference image, the patterns can be classified and the amount of misalignment can be corrected, thereby reducing the detection of false defects caused by misalignment. [Brief description of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram showing an overall configuration of an inspection device according to an embodiment. [Diagram 2] FIG. 2 is a flowchart of a defect inspection process in an inspection device according to an embodiment. [Diagram 3] FIG. 3 is a flowchart of a reference parameter estimation process in an inspection apparatus according to an embodiment. [Figure 4] FIG. 4 is a diagram showing an example of class classification in the inspection device according to an embodiment. [Diagram 5] FIG. 5 is a diagram showing an example of a reference parameter map in an inspection apparatus according to an embodiment. [Figure 6] FIG. 6 is a diagram showing an example of a relationship between a figure pattern and a reference contour in an inspection apparatus according to an embodiment. [Figure 7] FIG. 7 is a diagram showing an example of a relationship between an actual contour line and a reference contour line in an inspection apparatus according to an embodiment. [Figure 8] FIG. 8 is a diagram illustrating an example of the error amount of a class in an inspection device according to an embodiment. [Figure 9] FIG. 9 is a diagram illustrating an example of a correction process in the inspection device according to an embodiment. [Figure 10] FIG. 10 is a diagram showing an example of a resizing process in the inspection device according to an embodiment. [Figure 11] FIG. 11 is a diagram showing an example of extraction of contour points of an unfolded image in an inspection device according to an embodiment. [Figure 12] FIG. 12 is a diagram showing an example of an isotropic filter used in corner rounding processing in an inspection apparatus according to an embodiment. [Figure 13] FIG. 13 is a diagram illustrating an example of an isotropic filter in an inspection apparatus according to an embodiment. [Figure 14] FIG. 14 is a diagram showing an example of a broken line function in the inspection device according to an embodiment. [Figure 15]FIG. 15 is a diagram showing an example of pixel values ​​of each pixel before and after corner rounding processing in one class area in an inspection device according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] Hereinafter, the embodiments will be described with reference to the drawings. The embodiments exemplify devices and methods for embodying the technical ideas of the invention. The drawings are schematic or conceptual, and the dimensions and ratios of each drawing are not necessarily the same as those in reality. The technical ideas of the present invention are not specified by the shapes, structures, arrangements, etc. of the components.

[0016] In the following, an inspection device that captures an electron beam image (hereinafter also referred to as an "SEM image" or "actual image") of a pattern to be measured using a scanning electron microscope (hereinafter referred to as "SEM (Scanning Electron Microscope)") will be described as an inspection device for a sample. The inspection device may capture an optical image of a pattern using an optical microscope, or may capture an optical image of light reflected or transmitted through the sample using a light receiving element. In addition, although the present embodiment describes a case where the sample to be inspected is a mask, the sample may be any sample with a pattern on its surface, such as a wafer or a substrate used in a liquid crystal display device or the like.

[0017] 1. Overall configuration of the inspection device First, an example of the overall configuration of an inspection device will be described with reference to Fig. 1. Fig. 1 is a diagram showing the overall configuration of an inspection device 1.

[0018] As shown in FIG. 1, the inspection device 1 includes an imaging mechanism 10 and a control mechanism 20.

[0019] The imaging mechanism 10 includes a sample chamber 11 and a lens barrel 12. The lens barrel 12 is installed above the sample chamber 11. For example, the lens barrel 12 has a cylindrical shape extending perpendicularly to the sample chamber 11. The sample chamber 11 and the lens barrel 12 have openings on their mutually contacting surfaces. The space formed by the sample chamber 11 and the lens barrel 12 is maintained in a vacuum (reduced pressure) state using a turbo molecular pump or the like.

[0020] In the sample chamber 11, a stage 13, a stage driving mechanism 14, and a detector 15 are provided.

[0021] A sample (mask) 300 is placed on the stage 13. The stage 13 is movable in an X direction parallel to the surface of the stage 13 and in a Y direction parallel to the surface of the stage 13 and intersecting the X direction. The stage 13 may also be movable in a Z direction perpendicular to the surface of the stage 13, or may be rotatable around a rotation axis on an XY plane with the Z direction as the rotation axis.

[0022] The stage driving mechanism 14 has a driving mechanism for moving the stage 13 in the X direction and the Y direction. Note that the stage driving mechanism 14 may have, for example, a mechanism for moving the stage 13 in the Z direction, or a mechanism for rotating the stage 13 around a rotation axis in the Z direction on the XY plane.

[0023] The detector 15 detects secondary electrons or reflected electrons emitted from the sample, etc. The detector 15 transmits a signal of the detected secondary electrons or reflected electrons, etc., that is, data of the SEM image, to the image acquisition circuit 214.

[0024] An electron gun 16 and an electron optical system 17, which are components of the SEM, are provided within the column 12.

[0025] The electron gun 16 is installed so as to emit an electron beam toward the sample chamber 11 .

[0026] The electron optical system 17 focuses the electron beam emitted from the electron gun 16 on a predetermined position on the sample 300 and irradiates the sample. For example, the electron optical system 17 includes a plurality of focusing lenses 101 and 102, a plurality of scanning coils 103 and 104, and an objective lens 105. The electron beam emitted from the electron gun 16 is accelerated and then focused as an electron spot on the surface of the sample 300 placed on the stage 13 by the focusing lenses 101 and 102 and the objective lens 105. The scanning coils 103 and 104 control the position of the electron spot on the sample 300.

[0027] The control mechanism 20 includes a control circuit 21 , a memory device 22 , a display device 23 , an input device 24 , and a communication device 25 .

[0028] The control circuit 21 controls the entire inspection device 1. More specifically, the control circuit 21 controls the imaging mechanism 10 to acquire an SEM image. The control circuit 21 also controls the control mechanism 20 to compare the generated reference image with the inspection image and detect defects. That is, the control circuit 21 is a processor for performing defect inspection. For example, the control circuit 21 includes a CPU (Central Processing Unit), a RAM (Random Access Memory), and a ROM (Read Only Memory), which are not shown. For example, the CPU loads a program stored in the ROM or the storage device 22 as a non-transitory storage medium into the RAM. Then, the control circuit 21 interprets and executes the program loaded into the RAM by the CPU to control the inspection device 1. The control circuit 21 may be, for example, a CPU device such as a microprocessor, or a computer device such as a personal computer. In addition, the control circuit 21 may include a dedicated circuit (dedicated processor) in which at least some of the functions are performed by other integrated circuits such as an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Alley (FPGA), or a Graphics Processing Unit (GPU).

[0029] The control circuit 21 includes an unfolding circuit 211, a classifying circuit 212, a reference image generating circuit 213, an image acquiring circuit 214, and a comparing circuit 215. These may be configured by a program executed by an integrated circuit such as a CPU, an ASIC, an FPGA, or a GPU, may be configured by hardware or firmware included in the integrated circuits, or may be configured by individual circuits controlled by the integrated circuits. In the following, a case will be described in which the control circuit 21 realizes the functions of the unfolding circuit 211, the classifying circuit 212, the reference image generating circuit 213, the image acquiring circuit 214, and the comparing circuit 215 by a program executed by the control circuit 21.

[0030] The expansion circuit 211 expands, for example, the design data 221 held in the storage device 22 into data for each figure, and interprets the figure code and figure dimensions indicating the figure shape of the figure data. The expansion circuit 211 then expands the design data 221 into a binary or multi-value (for example, 8-bit) image (hereinafter also referred to as a "CAD image" or "expanded image") as a pattern arranged in a square with a grid of a predetermined quantization dimension as a unit. The expansion circuit 211 calculates the occupancy rate of the figure for each pixel of the expanded image. The figure occupancy rate in each pixel calculated in this way is the pixel value. For example, when the expanded image is expressed by 8-bit gradation data, the pixel value of each pixel is expressed by a gradation value of 0 to 255. When the pixel value is 0, the figure occupancy rate is 0%, and when the pixel value is 255, the figure occupancy rate is 100%.

[0031] The classification circuit 212 classifies the edges of each pattern of the expanded image into classes based on the shape of the edges. For example, the classification circuit 212 classifies the edges of the patterns into three classes: straight line parts, 90-degree corner parts, and 270-degree corner parts. The number and types of classes can be set arbitrarily.

[0032] The reference image generating circuit 213 generates a reference image based on the classified exfoliated image. More specifically, the reference image generating circuit 213 executes resizing and corner rounding of the exfoliated image based on the reference parameters 222. The reference parameters 222 include a resizing amount and a corner rounding amount. The resizing amount is used in the resizing process. The corner rounding amount is used in the corner rounding process. The reference parameters 222 are set for each class. The reference image generating circuit 213 extracts a contour line (hereinafter, referred to as a "reference contour line") from the exfoliated image after the resizing process and the corner rounding process to generate a reference image (contour image). The contour line is a line connecting the contour points of each pixel. The extraction of the contour points will be described later. The reference image generating circuit 213 transmits the generated reference image to the comparison circuit 215 and the storage device 22.

[0033] The reference image generating circuit 213 includes a reference parameter estimation circuit 216. The reference parameter estimation circuit 216 estimates the reference parameters 222 of each class. More specifically, a plurality of estimation regions used for estimating the reference parameters 222 are set in the sample 300. The estimation regions may be set by a user (operator) of the inspection device 1, or may be set by the control circuit 21 based on the design data 221. The reference parameter estimation circuit 216 estimates the reference parameters 222, i.e., the resizing amount and the corner rounding amount, based on the class classification of the edge of the pattern in the estimation region. Then, the reference parameter estimation circuit 216 generates a reference parameter map based on the estimation result. Note that, when the reference parameter estimation circuit 216 has previously performed an inspection of a sample identical or similar to the sample 300 to be inspected this time, the reference parameter estimation circuit 216 may use the reference parameters 222 of the past inspection stored in the storage device 22. For example, the same or similar sample is a sample having the same mask name, mask series, or pattern type and minimum dimension as the sample 300 to be inspected this time. The decision as to whether or not to use the past reference parameters 222 may be made by an operator or by the reference parameter estimation circuit 216 .

[0034] The image acquisition circuitry 214 acquires SEM image data from the detector 15 of the imaging mechanism 10. The image acquisition circuitry 214 extracts a contour line (hereinafter, referred to as an "actual image contour line") from the SEM image (actual image) to generate an inspection image (contour image).

[0035] The comparison circuit 215 uses an appropriate algorithm to compare the inspection image with the reference image. More specifically, the comparison circuit 215 aligns the inspection image with the reference image. Based on the alignment result, the comparison circuit 215 measures the amount of distortion of the inspection image in the surface of the sample 300 and calculates a distortion coefficient. Based on the alignment result and the amount of distortion, the comparison circuit 215 calculates the amount of positional deviation of each pixel. The amount of positional deviation indicates the distance (the magnitude of the relative vector) between the reference contour line and the actual contour line. If the amount of positional deviation corrected by the correction circuit 217 exceeds a preset threshold, the comparison circuit 215 determines that there is a defect at the corresponding coordinate position of the sample 300. For example, the comparison circuit uses a common threshold to determine the amount of corrected positional deviation of each pixel.

[0036] In the process of estimating the reference parameters 222, the comparison circuit 215 calculates the amount of error in the amount of positional deviation for each class (hereinafter referred to as the "error amount of the class"). In other words, the comparison circuit 215 calculates the error in the distance (magnitude of the relative vector) between the reference contour and the actual image contour of each pixel including an edge for each area classified into classes (hereinafter referred to as the "class area"). The comparison circuit 215 calculates, for example, the root mean square error (RMSE) of the amount of positional deviation of pixels including an edge in the class area as the error amount of the class. Note that other error calculation methods may be used to calculate the amount of error for each class.

[0037] The comparison circuit 215 includes a correction circuit 217. The correction circuit 217 corrects the positional deviation amount for each pixel using the error amount of the class corresponding to the pixel. More specifically, in the estimation process of the reference parameters, the correction circuit 217 sets a correction coefficient for each class based on the error amount (e.g., RMSE) of the class. The correction coefficient may be common to all classes. In order to correct the positional deviation amount, the correction circuit 217 sets a correction function for each class having the error amount and the correction coefficient of the class as variables. In the process of comparing the inspection image with the reference image, the correction circuit 217 corrects the positional deviation amount of each pixel using the correction function of the class to which the pixel belongs. For example, the positional deviation amount after correction is smaller than the positional deviation amount before correction.

[0038] The storage device 22 stores data and programs related to defect inspection. For example, the storage device 22 stores design data 221, reference parameters 222, parameters 223 of inspection conditions, and inspection data 224. The design data 221 is design data of the sample 300. The reference parameters 222 include resizing amounts and corner rounding amounts of each class. The parameters 223 of inspection conditions include, for example, the imaging conditions of the imaging mechanism 10, information on class classification, error amounts of classes, correction coefficients, and defect detection conditions (algorithms). The inspection data 224 includes, for example, information on the amount of positional deviation between an inspection image and a reference image, image data (unfolded image, reference image, SEM image, and inspection image), and data on detected defects (coordinates, size, etc.). The storage device 22 also stores a defect inspection program 225 as a non-transient storage medium. The defect inspection program 225 is a program for causing the control circuit 21 to execute a defect inspection.

[0039] The storage device 22 may include various storage devices such as a magnetic disk storage device (HDD: Hard Disk Drive) or a solid state drive (SSD) as an external storage. Furthermore, the storage device 22 may include a drive for reading a program stored in a non-transitory storage medium such as a CD (Compact Disc) or a DVD (Digital Versatile Disc).

[0040] The display device 23 includes, for example, a display screen (for example, an LCD (Liquid Crystal Display) or an EL (Electroluminescence) display, etc.) Under the control of the control circuit 21, the display device 23 displays, for example, a defect detection result.

[0041] The input device 24 is an input device such as a keyboard, a mouse, a touch panel, or a button switch.

[0042] The communication device 25 is a device for connecting to a network to transmit and receive data to and from an external device. Various communication standards can be used for communication. For example, the communication device 25 receives design data from an external device and transmits defect inspection results and the like to the external device.

[0043] 2. Overall flow of defect inspection process Next, an example of the overall flow of the defect inspection process will be described with reference to Fig. 2. Fig. 2 is a flowchart of the defect inspection process.

[0044] As shown in FIG. 2, the inspection process roughly includes an inspection image acquisition process (step S1), a reference image generation process (step S2), and a comparison process (step S3).

[0045] 2.1 Inspection image acquisition process First, an example of the inspection image acquisition process in step S1 will be described.

[0046] The image acquisition circuit 214 acquires an SEM image of the sample 300 from the imaging mechanism 10 (step S11).

[0047] Next, the image acquisition circuitry 214 executes a filter process to remove noise from the acquired SEM image (step S12).

[0048] Next, the image acquisition circuitry 214 extracts the actual contour line of each pattern from the filtered SEM image (step S13). That is, the image acquisition circuitry 214 generates an inspection image (contour image).

[0049] The image acquisition circuitry 214 transmits the generated inspection image to the comparison circuitry 215 and the storage device 22 .

[0050] 2.2 Reference image acquisition process Next, an example of the reference image acquisition step will be described.

[0051] First, the inspection device 1 acquires the design data 221 via the communication device 25 (step S21). The acquired design data 221 is stored in the storage device 22, for example.

[0052] Next, the unfolding circuit 211 reads out the design data 221 stored in the storage device 22. Then, the unfolding circuit 211 executes an unfolding process to unfold (convert) the design data 221 into, for example, 8-bit image data (unfolded image) (step S22). The unfolding circuit 211 transmits the unfolded image to the classification circuit 212 and the storage device 22.

[0053] Next, the classification circuit 212 executes classification of the edges of the pattern of the developed image (step S23).

[0054] Next, the reference image generating circuit 213 executes resizing and corner rounding based on the reference parameters 222 of each class (step S24).

[0055] Next, the reference image generating circuit 213 extracts a reference contour line of the pattern from the expanded image that has been resized and rounded (step S25). That is, the reference image generating circuit 213 generates a reference image (contour image). The reference image generating circuit 213 transmits the generated reference image to the comparison circuit 215 and the storage device 22.

[0056] 2.3 Comparison process Next, an example of the comparison process will be described. First, the comparison circuit 215 performs alignment using the inspection image and the reference image (step S31). That is, the comparison circuit 215 aligns the pattern in the inspection image with the pattern in the reference image. At this time, the comparison circuit 215 obtains a relative vector between the position of the actual image contour line and the position of the reference contour line. Then, the comparison circuit 215 calculates the average value of the relative vectors in each pattern as the amount of alignment shift of the inspection image relative to the reference image.

[0057] Next, the comparison circuit 215 measures the amount of distortion of the inspection image and calculates a distortion coefficient (step S32). For example, due to stage movement accuracy or distortion of the sample 300, a positional deviation may occur between the coordinate information based on the design data 221 and the coordinates of the pattern calculated from the captured image. The comparison circuit 215 measures the amount of distortion of the inspection image from, for example, the distribution of local alignment shift amounts within the surface of the sample 300, and calculates the distortion coefficient.

[0058] Next, the comparison circuit 215 calculates the amount of positional deviation between the actual image contour and the reference contour for each pixel based on the relative vector and the distortion coefficient (step S33).

[0059] Next, the correction circuit 217 corrects the positional deviation amount of the pixel using the error amount of the class corresponding to each pixel (step S34). In other words, the correction circuit 217 corrects the positional deviation amount of the pixel based on the correction function of the class corresponding to each pixel.

[0060] Next, the comparison circuit 215 detects a defect based on the corrected amount of misalignment (step S35). If the corrected amount of misalignment exceeds a threshold, the contour position is determined to be defective. After storing the result of the defect inspection in the storage device 22, the control circuit 21 may display the result on the display device 23, or may output the result to an external device (e.g., a review device, etc.) via the communication device 25.

[0061] 3. Reference parameter estimation process Next, an example of the estimation process of the reference parameters 222 will be described with reference to Fig. 3. Fig. 3 is a flowchart of the estimation process of the reference parameters 222.

[0062] The estimation process of the reference parameters 222 may be performed before the inspection process described in Fig. 2 as a pre-inspection operation of the sample 300, or may be performed during the reference image generation process in Fig. 2. When the estimation process is performed before the inspection process, it is preferable to use a predetermined adjustment pattern or the like. Below, a case where the estimation process of the reference parameters 222 is performed without using the past reference parameters 222 will be described.

[0063] 3, steps S101 to S104 are executed as the process of estimating the reference parameters 222. Each step will be described in detail.

[0064] [Step S101] First, the unfolding circuit 211 generates an unfolded image of the estimated area based on the design data 221. Next, the classification circuit 212 executes classification of the edges of the pattern (graphic pattern) in the estimated area. An example of classification in the estimated area is shown in FIG. 4. In the example of FIG. 4, the edges of the graphic pattern are classified into three class areas A to C based on their shapes. Class area A includes a 90-degree corner portion of the graphic pattern. Class area B includes a straight line portion of the edge of the graphic pattern. Class area C includes a 270-degree corner portion of the graphic pattern. The size (number of pixels) of each class area is arbitrary.

[0065] [Step S102] Next, the reference parameter estimation circuit 216 estimates optimal reference parameters for each class, and generates a reference parameter map for the estimation area. An example of the reference parameter map is shown in FIG. 5. As shown in FIG. 5, the reference parameter estimation circuit 216 generates a reference parameter map indicating class areas and corresponding reference parameters 222. For example, in class area A, 250 is set as the resize amount, and 2.2 is set as the corner rounding amount. In class area B, 250 is set as the resize amount, and 0 is set as the corner rounding amount. In class area C, 250 is set as the resize amount, and 1.8 is set as the corner rounding amount.

[0066] The reference parameters can be estimated using an estimation method such as the least squares method or the conjugate gradient method, which minimizes an objective function (error function).

[0067] [Step S103] Next, the reference image generating circuit 213 executes resizing and corner rounding of the pattern of the estimated area based on the reference parameter map. Then, the reference image generating circuit 213 extracts a reference contour from the expanded image that has been resized and corner rounded. An example of the relationship between the graphic pattern and the reference contour is shown in FIG. 6. As shown in FIG. 6, the resizing and corner rounding are executed based on the reference parameters set (estimated) for each class. Then, the reference contour is extracted from the expanded image that has been resized and corner rounded. In the example of FIG. 6, the reference contour is contracted with respect to the graphic pattern, but is not limited to this. The reference contour may be expanded with respect to the graphic pattern.

[0068] [Step S104] Next, the image acquisition circuit 214 extracts the real image contour of the estimated area in the same procedure as the inspection image acquisition process in Fig. 2. Note that the extraction of the real image contour may be performed before step S101 or step S103. The method of extracting the contour position may be a conventional method. For example, a differential filter process is performed in the X direction and the Y direction using a Sobel filter or the like, and the tilt direction of the real image is obtained from each differential value. Then, a gradation value profile of the tilt direction of the real image is obtained, and the peak position is extracted as the contour position on the contour (real image contour).

[0069] [Step S105] Next, the comparison circuit 215 compares the actual image contour with the reference contour for each pixel. An example of the relationship between the actual image contour and the reference contour is shown in Fig. 7. As shown in Fig. 7, the comparison circuit 215 compares the actual image contour of the estimated area with the reference contour, and calculates the amount of positional deviation (the magnitude of the relative vector) between the actual image contour and the reference contour for each pixel.

[0070] [Step S106] Next, the comparison circuit 215 calculates the error amount of the class. The process of calculating the error amount of the class may be performed before the inspection process described in FIG. 2 as a pre-inspection operation of the sample 300, or may be performed during the comparison process in FIG. 2. When the process is performed before the inspection process, it is preferable to use a predetermined adjustment pattern or the like. An example of the error amount of the class is shown in FIG. 8. As shown in FIG. 8, the comparison circuit 215 calculates the RMSE of the positional deviation amount for each pixel in the class area for each class, and sets this as the error amount of the class. For example, the error amount of the class area A is 0.21. The error amount of the class area B is 0.12. The error amount of the class area C is 0.30. The comparison circuit 215 stores the error amount of each class in the storage device 22.

[0071] When the reference parameter estimation and the class error amount calculation are performed before the inspection process, it is preferable to save the final result of the objective function (error function) optimized during the estimation as the error amount, since this reduces the overall processing amount.

[0072] [Step S107] Next, the correction circuit 217 calculates a correction coefficient for each class based on the error amount of the class. The correction coefficient may be set by an operator. Then, the correction circuit 217 sets a correction function for each class based on the error amount and the correction coefficient of the class.

[0073] 4. Correction process Next, an example of the correction of the positional deviation amount by the correction circuit 217 described in step S34 of Fig. 2 will be described with reference to Fig. 9. Fig. 9 is a diagram showing an example of the correction process.

[0074] As shown in FIG. 9, the positional deviation amounts before correction for class areas A, B, and C are denoted as L1, L2, and L3, respectively. The positional deviation amounts after correction for class areas A, B, and C are denoted as L1', L2', and L3', respectively. The correction coefficients for class areas A, B, and C are k 1 , k 2 , and k 3 For example, when the error amounts of class areas A, B, and C are 0.21, 0.12, and 0.30, respectively, the positional deviation amount after correction of each pixel is represented by the following correction function.

[0075] (Class area A): L1' = max(0, L1-0.21k 1 ) (Class region B): L2' = max(0, L2-0.12k 2 ) (Class region C): L3' = max(0, L3-0.30k 3 ) For example, in any pixel corresponding to class area A, the position shift amount is L1=1 and the correction coefficient is k 1 = 1, the positional deviation amount corrected using the correction function is L1' = 0.79. Also, for example, if the positional deviation amount of class area A is L1 = 0.1, then 0.1 - 0.21 × 1 = -0.11, and so the corrected positional deviation amount is L1' = 0.

[0076] The comparator circuit 215 compares the corrected amount of positional deviation with a threshold value to determine whether or not there is a defect.

[0077] For example, if the threshold for determining defects is 0.5, then this pixel is determined to be defective because 0.79 is greater than 0.5. Similar processing is performed for the other pixels.

[0078] 5.Resizing Next, an example of resizing will be described with reference to FIG. 10. FIG. 10 is a diagram showing an example of resizing. In FIG. 10, the pattern is a visualization of a pattern (figure) on the design data 221, and each pixel of the expanded image has a gradation value corresponding to the occupancy rate of the figure. Therefore, the sum of the gradation values ​​of the expanded image is proportional to the area of ​​the figure in that region. In the example of FIG. 10, the resizing is performed in the direction to expand the figure, but when contracting the figure, it is preferable to invert the gradation of the figure before resizing, and then invert it again after the processing. In the example of FIG. 10, a case where the resizing is performed using a Sobel filter will be described. The filter used in the resizing is not limited to the Sobel filter.

[0079] As shown in FIG. 10, the reference image generating circuit 213 executes resizing processing for each pixel based on the resizing amount of the reference parameter 222 described with reference to FIG. 5. For example, the reference image generating circuit 213 repeats resizing processing in units of one pixel. The resizing processing in units of one pixel is a resizing processing in which the upper limit of one resizing processing is set to the resizing amount of 100, for example, when the resizing amount for moving the edge position by one pixel is defined as 100. For example, as shown in the reference parameter map of FIG. 5, when the resizing amount of the reference parameter 222 is 250, three resizing processings are executed. In a 3×3 expanded image centered on a pixel of interest including a pattern edge, a Sobel filter is used to obtain the normal direction of the contour at the pixel of interest. More specifically, the reference image generating circuit 213 executes convolution calculations with the kernels of the Sobel filter in the X direction and the Y direction, respectively, on the expanded image. Then, the reference image generating circuit 213 calculates the normal angle θ of the contour vector obtained by combining the calculated value in the X direction and the calculated value in the Y direction at the pixel of interest. For example, if the value in the X direction after the Sobel filter process is Fx and the value in the Y direction is Fy, then it can be expressed by the formula θ=atan(Fy / Fx).

[0080] Next, the reference image generation circuit 213 forms a half-plane whose total pixel values ​​(tone values) of the 3×3 expanded image (i.e., the total area of ​​3×3 pixels) is the same and whose boundary line is perpendicular to the contour vector found by the Sobel filter.

[0081] Next, the reference image generating circuit 213 moves the half plane in the direction of the resize amount at the pixel of interest by a resize amount of up to one pixel (in this example, the resize amount is 100 or less). That is, the reference image generating circuit 213 corrects the pixel value of the pixel of interest according to the resize amount, and executes resize processing in units of one pixel. The resize amount has two components in the X direction and the Y direction. The resize amount in the figure is expressed as a vector (Rx·cosθ, Ry·sinθ), where Rx is the resize amount in the X direction and Ry is the resize amount in the Y direction.

[0082] The reference image generating circuit 213 repeats the expansion process, the resizing process, and the subtraction process of the resizing amount until the resizing amount of each pixel becomes equal to or less than a preset threshold value (for example, 0). The expansion process is a process of adding a resizing amount to pixels with a resizing amount of 0 among pixels adjacent to a pixel with a resizing amount of 1 or more set, and increasing (expanding) the pixels to which a resizing amount has been added. The subtraction process of the resizing amount is a process of subtracting the resizing amount of one pixel from the resizing amount after the resizing process of one pixel.

[0083] 6. Extraction of contour points Next, an example of extraction of contour points in an exfoliated image will be described with reference to FIG. Fig. 11 is a diagram showing an example of extraction of contour points in an exfoliated image. In Fig. 11, the pattern is a visualization of a pattern (figure) on the design data 221, and each pixel in the exfoliated image has a gradation value corresponding to the occupancy rate of the figure. Therefore, the sum of the gradation values ​​of the exfoliated image is proportional to the area of ​​the figure in that region.

[0084] As shown in Fig. 11, a contour vector of a pixel of interest is obtained using a Sobel filter in the same manner as in Fig. 10, and a half-plane corresponding to the contour vector is formed. A contour point is then set at the midpoint in the X and Y directions on the boundary line of the half-plane. The contour position is set in sub-pixel units within each contour pixel.

[0085] 7. Corner rounding Next, an example of corner rounding processing will be described with reference to Fig. 12 to Fig. 14. Fig. 12 is a diagram showing an example of an isotropic filter used in the corner rounding processing. Fig. 13 is a diagram showing an example of an isotropic filter. Fig. 14 is a diagram showing an example of a broken line function.

[0086] For example, the reference image generation circuit 213 performs corner rounding processing for each pixel by performing filter processing using an isotropic filter and calculating a broken line function using the result of the filter processing, based on the corner rounding amount of the reference parameter 222 set for each class region described with reference to FIG. 5. For example, by applying a broken line function to the result of the filter processing using an isotropic filter, corner rounding processing (correction of pixel values) of the target pixel is performed.

[0087] As shown in FIG. 12, for example, when the corner rounding amount of the class region corresponding to the pixel of interest is R, filter processing (convolution operation) using an isotropic filter with a radius r = R is performed on the corner rounding amount R. More specifically, for example, when 1.5 < R ≤ 2.5, the size of the isotropic filter f is represented by a 5×5 pixel kernel centered on the pixel of interest. Here, assuming the coordinates of the pixel of interest are (0, 0), the coordinates of each pixel are represented by (x, y) = (-2, 2) to (2, -2) from the lower left to the upper right of the paper surface.

[0088] The reference image generation circuit 213 first performs the operation of Equation (1) to perform a convolution operation between the isotropic filter f and the pixel of interest. Here, I represents the pixel value of the pixel of interest. I' represents the calculated value.

[0089]

Equation

[0090] Next, the reference image generation circuit 213 performs the operations of Equation (2) and Equation (3) to calculate the coefficients k 1 and k 2 . The coefficient k 1 represents the total value (area) of the isotropic filter f with an X coordinate of 0. The coefficient k 2 represents the total value (area) of the isotropic filter f with an X coordinate of 1 or more and an upper limit of N (N is the maximum value of the X coordinate of the isotropic filter).

[0091]

Equation

[0092]

number

[0093] Next, the reference image generating circuit 213 performs the operation of the broken line function L of the formula (4) to calculate the pixel value of the target pixel after the corner rounding process. max indicates the maximum pixel value. For example, when pixel values ​​are expressed as 0 to 255, I max =255.

[0094]

number

[0095] As shown in FIG. 13, for example, a case will be described where the corner rounding amount of class area A shown in the reference parameter map of FIG. 5 is R=2.2. In a 5×5 pixel isotropic filter f, for example, the values ​​of coordinates (-2,2), (2,2), (-2,-2), and (2,-2) are set to 0.0068. The values ​​of coordinates (-1,2), (1,2), (-2,1), (2,1), (-2,-1), (2,-1), (-1,-2), and (1,-2) are set to 0.4326. The values ​​of coordinates (0,2), (-2,0), (2,0), and (0,-2) are set to 0.6812. And the values ​​of other coordinates are set to 1. In this case, k1=4.3624 and k2=5.4252. Therefore, I max The broken line function L when L=255 is expressed by the graph shown in FIG.

[0096] A specific example of corner rounding using the 5×5 pixel isotropic filter described in Fig. 13 will be described with reference to Fig. 15. Fig. 15 is a diagram showing an example of pixel values ​​of each pixel before and after corner rounding in one class area. In the example of Fig. 15, the same values ​​continue in the left and right and above and below the pixels outside the class area.

[0097] As shown in FIG. 15, for example, if a class region is 7×7 pixels with the coordinates (4,4) of the corner detection pixel as the center pixel, the pixel value of 165 at the coordinates (4,4) becomes 0 due to corner rounding. More specifically, when the coordinates of the pixel of interest are (4,4), the result of the convolution operation using the isotropic filter shown in equation (1) is I'=1335. This is because I max ×K 2 Since I' is smaller than I'=1383, the pixel value at coordinate (4,4) is 0. Also, when the coordinates of the pixel of interest are (4,5), the result of the convolution operation using the isotropic filter is I'=1906. This is because I max ×K 2 = 1383 or greater, I max (K 1 +K 2 ) = 2496. Therefore, the pixel value 196 at coordinates (4,5) becomes 120. Similar processing is performed for the other coordinates. As a result, the pixel value 196 at coordinates (4,6) becomes 187. The pixel value 216 at coordinates (5,4) becomes 126. The pixel value 216 at coordinates (6,4) becomes 202. Note that in the example of FIG. 15, the pixel values ​​of the other pixels in the class area do not change.

[0098] 8. Effects of this embodiment In defect inspection, when comparing an inspection image (real image contour line) with a reference image (reference contour line), a positional deviation occurs between the real image contour line and the reference contour line due to a dimensional shift of the pattern. The amount of positional deviation differs depending on the shape of the edge of the pattern. Therefore, if the positional deviation amount of each pixel including the edge is judged using the same threshold value, the detection frequency of false defects will differ depending on the shape of the edge.

[0099] In contrast, with the configuration according to this embodiment, the inspection device can classify the edges of the pattern based on their shapes when generating a reference image, and set different resizing amounts and corner rounding amounts for each class. The inspection device can calculate an error amount for each class based on the positional deviation amount between the actual image contour and the reference contour calculated for each pixel. The inspection device can correct the positional deviation amount of each pixel based on the error amount of the corresponding class. The inspection device can then compare the corrected positional deviation amount with a threshold value to determine a defect. This allows the corrected positional deviation amount of each pixel to be determined using the same threshold value. Therefore, the extraction of false defects by defect inspection can be reduced, and the reliability of defect inspection can be improved.

[0100] 9. Modifications, etc. In the above embodiment, a case where a reference image is generated in an inspection device has been described, but the method for generating a reference image is not limited to the inspection device, and may be applied to other devices that generate a reference image based on data, such as a measurement device.

[0101] In the above-described embodiment, machine learning or the like may be used for classifying the patterns. That is, AI (artificial intelligence) may learn the patterns and set the classes.

[0102] The present invention is not limited to the above-mentioned embodiment, and can be modified in various ways without departing from the gist of the present invention. The embodiments may be combined as appropriate, and in that case, the combined effect can be obtained. Furthermore, the above-mentioned embodiment includes various inventions, and various inventions can be extracted by combinations selected from the multiple components disclosed. For example, if the problem can be solved and the effect can be obtained even if some components are deleted from all the components shown in the embodiment, the configuration from which the components are deleted can be extracted as an invention. [Explanation of symbols]

[0103] 1...inspection apparatus, 10...imaging mechanism, 11...sample chamber, 12...optical tube, 13...stage, 14...stage driving mechanism, 15...detector, 16...electron gun, 17...electron optical system, 20...control mechanism, 21...control circuit, 22...storage device, 23...display device, 24...input device, 25...communication device, 101, 102...focusing lens, 103, 104...scanning coil, 105...objective lens, 211...development circuit, 212...classification circuit, 213...reference image generation circuit, 214...image acquisition circuit, 215...comparison circuit, 216...reference parameter estimation circuit, 217...correction circuit, 221...design data, 222...reference parameters, 223...inspection condition parameters, 224...inspection data, 225...defect inspection program, 300...sample

Claims

1. A sample imaging mechanism; an image acquisition circuit that extracts a first contour line from image data of the sample captured by the imaging mechanism; an unfolded circuit for generating an unfolded image from design data; a classification circuit for classifying edges of a pattern in the exfoliated image; a reference parameter estimation circuit for estimating reference parameters including a resize amount and a corner rounding amount for each of the classified classes; a reference image generating circuit for resizing the pattern and rounding the corners based on the reference parameters, and extracting a second contour line from the pattern after the resizing and the rounding; a comparison circuit that compares the first contour line with the second contour line to calculate a positional deviation amount between the first contour line and the second contour line for each pixel, and calculates an error amount of the positional deviation amount for each class; a correction circuit that corrects the positional deviation amount by using the error amount of the class corresponding to each pixel; An inspection device comprising:

2. the correction circuit sets a function having the amount of error as a variable for each of the classes.

2. The inspection device according to claim 1.

3. the classification circuit sets a class region based on the shape of an edge of the pattern; 3. The inspection device according to claim 1 or 2.

4. The comparison circuit determines that a defect exists when the corrected positional deviation amount exceeds a threshold value.

4. The inspection device according to claim 1.

5. the reference parameter estimation circuit generates a map based on the classification and the reference parameters.

5. An inspection device according to claim 1.

6. Extracting a first contour line from image data obtained by imaging the sample; generating an expanded image from the design data; classifying edges of a pattern from the developed image; estimating reference parameters for each classified class, including resizing and corner rounding; resizing and rounding the corners of the pattern based on the reference parameters; extracting a second contour line from the pattern after the resizing and the corner rounding; comparing the first contour line with the second contour line to calculate a positional deviation between the first contour line and the second contour line for each pixel; calculating an error amount of the positional deviation amount for each of the classes; correcting the positional deviation amount by using the error amount of the class corresponding to each of the pixels; An inspection method comprising:

7. setting a function having the error amount as a variable for each of the classes, The inspection method according to claim 6.

8. The step of classifying the patterns includes a step of setting a class region based on a shape of an edge of the pattern. The inspection method according to claim 6 or 7.

9. The method further includes a step of determining that there is a defect when the corrected positional deviation amount exceeds a threshold value. The inspection method according to any one of claims 6 to 8.

10. The step of estimating the reference parameters includes generating a map based on the classification and the reference parameters. The inspection method according to any one of claims 6 to 9.

Images