Pattern inspection device and pattern inspection method

Abstract

A pattern inspection device according to one embodiment of the present invention is characterized by being provided with: an optical image acquisition means for acquiring an optical image of a sample to be inspected on which a pattern is formed; a parameter range setting circuit that sets a range of reference image generation parameters on the basis of the degree of coincidence with past inspection condition parameters; a filter function generation circuit that generates a plurality of filter function candidates using values within the set range of the reference image generation parameters; a determination circuit that determines a filter function for creating a reference image from among the generated candidates of the plurality of filter functions; a reference image creation circuit that creates a reference image using the determined filter function; and a comparison circuit that compares the optical image with the reference image.

Description

TECHNICAL FIELD

[0001] This application is based on a priority application JP 2023-095655 (application number) filed in Japan on June 9, 2023. All the contents described in JP 2023-095655 are incorporated into this application by reference.

[0002] One embodiment of the present application relates to a pattern inspection device and a pattern inspection method. For example, it relates to a pattern inspection technology that inspects a pattern defect of an object that is a sample used in semiconductor manufacturing, and a method that inspects a defect of a mask used when a semiconductor element is manufactured. BACKGROUND

[0003] In recent years, along with high integration and large capacity of large scale integrated circuits (LSIs), circuit line widths required for semiconductor elements have become narrower. These semiconductor elements use a rough pattern (also referred to as a mask or an intermediate mask, hereinafter referred to as a mask) on which a circuit pattern is formed, and form a circuit by exposing and transferring the pattern on a wafer using a reduction projection exposure device called a so-called stepper, thereby being manufactured.

[0004] Furthermore, for manufacturing of an LSI which is expensive, improvement of yield is essential. However, as represented by a DRAM (Dynamic Random Access Memory) of 1 gigabit level, a pattern constituting an LSI has changed from submicron to nanometer level. As one of the main factors that reduce the yield, a pattern defect of a mask used when a super-fine pattern is exposed and transferred on a semiconductor wafer using a photolithography technology can be listed. In recent years, along with miniaturization of an LSI pattern formed on a semiconductor wafer, a size that must be detected as a pattern defect has become extremely small. Therefore, a pattern inspection device that inspects a defect of a transfer mask used for LSI manufacturing needs to be highly precise.

[0005] As a method of inspection, there is known a method in which an optical image obtained by photographing a pattern formed on a sample such as a photomask using a magnifying optical system at a prescribed magnification is compared with design data, whereby inspection is performed. For example, there is a "die to database inspection" in which CAD data after pattern design is converted into a device input format used when a pattern is drawn on a mask by a drawing device, the converted drawing data (design pattern data) is input to an inspection device, and a design image (reference image) is generated on the basis thereof, and the design image is compared with an optical image which is measurement data obtained by photographing a pattern. In the inspection method of the inspection device, a sample is placed on a stage, and inspection is performed by moving the stage and scanning a light beam on the sample. A light beam is irradiated on the sample by a light source and an illumination optical system. Light which has passed through the sample or has been reflected by the sample is imaged on a sensor via an optical system. An image photographed by the sensor is sent to a comparison circuit as measurement data. In the comparison circuit, after the images are aligned with each other, the measurement data and the reference data are compared in accordance with an appropriate algorithm, and in the case of disagreement, it is determined that there is a pattern defect.

[0006] Pixel data of an optical image photographed from a sample is in a state in which a filter acts in accordance with resolution characteristics and the like of an optical system used for photographing, in other words, in a continuously changing analog state, and thus an image intensity (gradation value) is different from a design image which is a digital value. Therefore, a filter process is performed on the design image, and a comparison process is performed on the basis of making a reference image close to a measurement image.

[0007] In recent demands for improvement of inspection accuracy along with miniaturization of a pattern, in order to achieve improvement of inspection accuracy, it is necessary to improve accuracy of the filter process. Therefore, a plurality of filter functions are generated using a wide range of reference image generation parameters. Then, a case in which a filter function capable of approximating a measurement image is obtained from many filter functions is performed. In order to generate such a plurality of filter functions, the amount of processing becomes enormous, and the time taken until an optimal filter function is obtained also becomes long.

[0008] Here, in the case of inspection similar to past inspection, a large number of filter functions are also generated, and thus there is a problem in that the amount of processing becomes enormous, the processing time until an optimal filter function is obtained becomes long, and a long time is required until actual inspection starts. Therefore, it is being studied that, in the case of being the same as or similar to a past inspection condition, a filter function used in the past is used as it is for this time's inspection (for example, refer to Patent Literature 1).

[0009] However, it is known that, as long as the inspection conditions are not completely identical, a case where even if a filter function used for a similar past inspection is used for this time's inspection, desired accuracy cannot be obtained occurs. On the other hand, in a case where a larger number of filter functions are generated as with the past, the processing time becomes longer, and a problem that a long time is required until the actual inspection starts has not been solved.

[0010] Prior Art Documents Patent Documents Patent Document 1: Japanese Patent Application Laid-Open No. 2022-182497 SUMMARY

[0011] PROBLEMS TO BE SOLVED BY THE INVENTION Therefore, one embodiment of the present application provides an inspection device and method capable of shortening the processing time until a filter function suitable for this time's inspection is obtained.

[0012] MEANS FOR SOLVING THE PROBLEMS A pattern inspection device of one embodiment of the present application includes: an optical image acquisition mechanism that acquires an optical image of a sample to be inspected on which a pattern is formed; a parameter range setting circuit that sets a range of a reference image generation parameter in accordance with a degree of identity with a past inspection condition parameter; a filter function generation circuit that generates a plurality of candidates of filter functions using values within the set range of the reference image generation parameter; a decision circuit that decides a filter function used for making a reference image from the generated plurality of candidates of filter functions; a reference image making circuit that makes a reference image using the decided filter function; and a comparison circuit that compares the optical image and the reference image.

[0013] A pattern inspection method of one embodiment of the present application includes: acquiring an optical image of a sample to be inspected on which a pattern is formed, setting a range of a reference image generation parameter in accordance with a degree of identity with a past inspection condition parameter, generating a plurality of candidates of filter functions using values within the set range of the reference image generation parameter, deciding a filter function used for making a reference image from the generated plurality of candidates of filter functions, making a reference image using the decided filter function, comparing the optical image and the reference image, and outputting a result.

[0014] Inventive Effects According to one embodiment of the present application, the amount of processing until a filter function suitable for the present inspection is obtained can be reduced. Furthermore, the processing time until a filter function suitable for the present inspection is obtained can be shortened. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a configuration view showing the configuration of the pattern inspection apparatus in Embodiment 1.

[0016] Figure 2 is a conceptual view for explaining the inspection region in Embodiment 1.

[0017] Figure 3 is a block diagram showing an example of the internal configuration of the filter function operation circuit in Embodiment 1.

[0018] Figure 4 is a flowchart showing the main part of the procedure of the pattern inspection method in Embodiment 1.

[0019] Figure 5 is a view showing an example of the inspection condition parameters and the proportion (weighting) of each parameter in Embodiment 1.

[0020] Figure 6 shows an example of the relationship between the degree of agreement and the proportion of the reduced range in Embodiment 1.

[0021] Figure 7 is a view showing an example of the relationship between the reference image generation parameters, the filter function, the score, and the order in the past inspection in Embodiment 1.

[0022] Figure 8 is a view showing an example of the reference image generation parameter range and the set results in Embodiment 1.

[0023] Figure 9 is a view showing an example of the internal procedure of the filter function generation procedure in Embodiment 1.

[0024] Figure 10 is a view showing an example of the graphic pattern before and after the size adjustment in Embodiment 1.

[0025] Figure 11 is a view showing an example of the graphic pattern before and after the rounding process in Embodiment 1.

[0026] Figure 12 is a view showing an example of the relationship between the filter function, the design image, and the measurement image in Embodiment 1.

[0027] Figure 13 is a view showing an example of the method of operating the coefficients of the filter function in Embodiment 1.

[0028] Figure 14 Fig. 1 is a diagram showing an example of the internal configuration of the comparison circuit in Embodiment 1.

[0029] Figure 15 Fig. 2 is a diagram showing an example of the reference image generation parameter range and the result of the setting in the modification example of Embodiment 1. DETAILED DESCRIPTION

[0030] Embodiment 1 Figure 1 Fig. 1 is a configuration diagram showing the configuration of the pattern inspection apparatus in Embodiment 1. In Embodiment 1, an inspection apparatus 100 that inspects defects of a pattern formed on a sample, such as a mask, is provided with an optical image acquisition mechanism 150 and a control system circuit 160 (control section). Figure 1

[0031] The optical image acquisition mechanism 150 has a light source 103, an illumination optical system 170, a movably arranged XYθ stage 102, a magnification optical system 104, and a shooting sensor 105 (an example of a sensor), a sensor circuit 106, a strip pattern memory 123, and a laser length measuring system 122. The XYθ stage 102 is arranged with a sample 101. As the sample 101, for example, a photomask that transfers a pattern to a wafer is included. In addition, a pattern composed of a plurality of graphic patterns that are the inspection targets is formed on the photomask. The sample 101 is arranged on the XYθ stage 102 with the pattern formation surface facing the lower side, for example.

[0032] A line sensor or a two-dimensional sensor is used as the shooting sensor 105. For example, it is preferable to use a TDI (Time Delay Integration) sensor. The TDI sensor has a plurality of photoelectric sensing elements arranged in two dimensions. Each photoelectric sensing element is set to a prescribed image accumulation time when shooting an image. In the TDI sensor, the outputs of the plurality of photoelectric sensing elements arranged in the scanning direction are integrated and output. The plurality of photoelectric sensing elements arranged in the scanning direction shoot the same pixel while shifting in time according to the movement of the XYθ stage 102. In the case of using a line sensor, the plurality of photoelectric sensing elements are arranged in a direction orthogonal to the scanning direction.

[0033] ​In the control system circuit 160, the control computer 110, which is a computer, is connected to the position circuit 107, the comparison circuit 108, the unwinding circuit 111, the reference circuit 112, the auto loader control circuit 113, the stage control circuit 114, the filter function operation circuit 140, the magnetic disk device 109, the memory 116, the floppy disk device (FD) 115, the CRT 117, the pattern monitor 118, and the printer 119 via the bus 120. In addition, the sensor circuit 106 is connected to the strip pattern memory 123, which is connected to the comparison circuit 108. In addition, the XYθ stage 102 is driven by an X-axis motor, a Y-axis motor, and a θ-axis motor. The XYθ stage 102 is an example of a stage.

[0034] A series of circuits including the position circuit 107, the comparison circuit 108, the unwinding circuit 111, the reference circuit 112, the auto loader control circuit 113, the stage control circuit 114, and the filter function operation circuit 140 includes a processing circuit. The respective processing circuits include a circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device, or the like. In the case where the same processing circuit (one processing circuit) is used, a different processing circuit (a respective processing circuit) can also be used. For example, the series of "circuit" of the position circuit 107, the comparison circuit 108, the unwinding circuit 111, the reference circuit 112, the auto loader control circuit 113, the stage control circuit 114, and the filter function operation circuit 140 can also be constituted by and executed by the control computer 110. Input data required for the position circuit 107, the comparison circuit 108, the unwinding circuit 111, the reference circuit 112, the auto loader control circuit 113, the stage control circuit 114, and the filter function operation circuit 140 or a result obtained by operation is stored in a memory not shown in the respective circuits or the memory 116 each time. Input data required for the control computer 110 or a result obtained by operation is stored in a memory not shown in the control computer 110 or the memory 116 each time. A program of a processing unit or the like can be recorded in the magnetic disk device 109, the FD 115, or a recording medium such as a ROM (Read Only Memory).

[0035] In the inspection apparatus 100, a high-magnification inspection optical system is constituted by the light source 103, the XYθ stage 102, the illumination optical system 170, the magnification optical system 104, the imaging sensor 105, and the sensor circuit 106. In addition, the XYθ stage 102 is driven by the stage control circuit 114 under the control of the control computer 110. It is possible to move by a driving system such as a three-axis (X-Y-θ) motor driven in the X direction, the Y direction, and the θ direction. These X motor, Y motor, and θ motor can use, for example, a stepping motor. The XYθ stage 102 can move in the horizontal direction and the rotational direction by the motors of the respective axes of the XYθ. The sample 101 is transported onto the XYθ stage 102 from an automatic loader 130 controlled by an automatic loader control circuit 113. Then, the moving position of the sample 101 disposed on the XYθ stage 102 is measured by a laser length measuring system 122 and supplied to the position circuit 107.

[0036] Design data (design data) on which a pattern of the sample 101 to be inspected is formed is input from the outside of the inspection apparatus 100 and stored in the disk device 109. In the design data, a plurality of graphic patterns are defined, and each graphic pattern is generally constituted by a combination of a plurality of element graphics. In addition, it can also be a graphic pattern constituted by one graphic. On the sample 101 to be inspected, a corresponding pattern is formed based on each graphic pattern defined by the relevant design data.

[0037] Here, in Figure 1 , the components required for the embodiment 1 are described. For the inspection apparatus 100, of course, other components required can also be generally included.

[0038] Figure 2 is a conceptual diagram for explaining the inspection region in the embodiment 1. As Figure 2 indicated, the inspection region 10 (entire inspection region) of the sample 101 is virtually divided into a plurality of inspection strips 20 in the shape of a long strip of the scan width W of the imaging sensor 105, for example, in the Y direction. Also, in the inspection apparatus 100, an image (strip region image) is acquired for each inspection strip 20. For each inspection strip 20, an image of a graphic pattern disposed in the inspection strip 20 in the long side direction (X direction) of the strip region is imaged using a laser (inspection light). In addition, in order to prevent omission of the image, it is preferable that the plurality of inspection strips 20 are set so that the adjacent inspection strips 20 overlap each other by a prescribed margin width.

[0039] The imaging sensor 105 acquires an optical image while continuously moving in the X direction by the movement of the XYθ stage 102. In the imaging sensor 105, an image of the entire inspection region 10 is continuously imaged while moving in the X direction. Figure 2optical image of the scanning width W. In Embodiment 1, after the optical image in one inspection strip 20 is photographed, the position of the next inspection strip 20 is moved in the Y direction, and the optical image of the scanning width W is continuously photographed again in the same manner while moving in the opposite direction. That is, the photographing is repeated in the forward (FWD) - backward (BWD) direction toward the opposite direction in the forward and backward directions.

[0040] In addition, at the time of actual inspection, as shown in Figure 2 the strip area image of each inspection strip 20 is divided into the images (frame images 31) of a plurality of frame areas 30 in a rectangular shape. Further, the inspection is performed for each frame image 31 of each frame area 30. For example, the division is performed in a size of 512 x 512 pixels. Therefore, the reference image compared with the frame image 31 of the frame area 30 is also created for each frame area 30.

[0041] Here, the direction of photographing is not limited to the repetition of the forward (FWD) - backward (BWD). Photographing can be performed from one direction. For example, it can be the repetition of FWD-FWD. Or, it can be the repetition of BWD-BWD.

[0042] Here, the pixel data of the optical image photographed from the sample 101 is in a state where a filter is applied in accordance with the resolution characteristics and the like of the optical system used for the photographing, in other words, in a continuously changing analog state, and therefore the image intensity (gradation value) is different from the design image of the digital value described later. Therefore, the filter processing is performed on the design image, and the comparison processing is performed on the basis close to the measurement data. In Embodiment 1, before the inspection processing of the sample 101 is performed, the filter function used for the filter processing is first calculated.

[0043] Figure 3 is a block diagram showing an example of the internal configuration of the filter function calculation circuit in Embodiment 1. In Figure 3 the filter function calculation circuit 140, a storage device 61, 70, 71, 73, 75 such as a magnetic disk, a coincidence degree calculation section 60, a parameter range setting section 62, a filter function generation section 64, a frame image creation section 72, a gray scale difference calculation section 80, a determination section 82, a decision section 84, and an evaluation value calculation section 86 are arranged. In addition, in the filter function generation section 64, a size adjustment processing section 66, a rounding processing section 67, and a filter coefficient calculation section 68 are arranged.

[0044] The series of "sections" such as the consistency degree calculation section 60, the parameter range setting section 62, the filter function generation section 64 (the size adjustment processing section 66, the rounding processing section 67, and the filter coefficient calculation section 68), the frame image production section 72, the gray scale difference calculation section 80, the determination section 82, the decision section 84, and the evaluation value calculation section 86 have processing circuits. The respective processing circuits include circuits, computers, processors, circuit boards, quantum circuits, or semiconductor devices, and the like. In addition, the respective "sections" can use a common processing circuit (the same processing circuit). Alternatively, different processing circuits (respective processing circuits) can be used. The input data or the operation results required for the consistency degree calculation section 60, the parameter range setting section 62, the filter function generation section 64 (the size adjustment processing section 66, the rounding processing section 67, and the filter coefficient calculation section 68), the frame image production section 72, the gray scale difference calculation section 80, the determination section 82, the decision section 84, and the evaluation value calculation section 86 are stored in a memory not shown in the filter function operation circuit 140 or the memory 116 each time.

[0045] Figure 4 is a flowchart showing the main part processes of the pattern inspection method in Embodiment 1. In Figure 4 , the pattern inspection method in Embodiment 1 implements the series of processes of the consistency degree calculation process (S102), the parameter range setting process (S104), the representative frame selection process (S110), the strip image acquisition process (S114), the representative frame image production process (S116), the representative design image production process (S118), the filter function operation process (S130), the strip image acquisition process (S202), the frame image production process (S204), the design image production process (S212), the reference image production process (S214), and the comparison process (S230).

[0046] As the consistency degree calculation process (S102), the consistency degree calculation section 60 inputs the inspection condition parameters of the inspection, and calculates the consistency degrees with the inspection condition parameters implemented in the past.

[0047] Figure 5 is a diagram showing an example of the inspection condition parameters and the weight (weighting) of each parameter in Embodiment 1. In Figure 5 , as the inspection condition parameters in Embodiment 1, for example, the parameters of the pattern conditions and the parameters of the photographing conditions can be given. As the parameters of the pattern conditions, for example, the series name of the mask, the layer name of the mask, the layout name of the mask, the film type, and the like can be given. As the parameters of the photographing conditions, for example, the stage speed, the inspection sensitivity (S), and the like can be given.

[0048] In Figure 5In the example of FIG. 9, the weight of the mask series name is 1, the weight of the mask layer name is 2, the weight of the mask layout name is 5, the weight of the film type is 1, the weight of the stage speed is 1, and the weight of the inspection sensitivity (threshold) is 2. The values of the weights can be set based on empirical values or the like.

[0049] The parameters of the pattern condition are defined as additional information, for example, of the drawing data. Alternatively, they can be obtained by capturing an unillustrated ID image formed outside the inspection region 10 of the sample 101. In the additional information or the ID image, for example, the identification number of the mask of the sample 101, the mask series name, the mask layer name, the mask layout name, the film type, and the like are defined. As the mask series name, for example, a name corresponding to the generation of the manufactured mask is used, which indicates the name of a front-end mask or a mask on which a pattern of a rough pattern size is formed. As the mask layer name, for example, the name of the level in the multilayer wiring layer of the semiconductor device is used. As the mask layout name, for example, the name of the circuit of the semiconductor device is used. As the film type, for example, the name indicating the type of the film used in the pattern formation of the mask of the sample 101 is used.

[0050] The coincidence degree calculating section 60 calculates the coincidence degree using at least one of the parameters of the pattern condition and the parameters of the capturing condition. Here, a case where both are used for the calculation is described. The inspection condition parameters of the past inspection are accumulated in the storage device 61 for each sample of the past inspection. The coincidence degree is defined by the following relation (1) using the weights of the respective parameters of the inspection condition parameters.

[0051] (1) Coincidence degree = sum of weights of coincident parameters / sum of weights of all parameters In addition, in a case where the data of a plurality of samples of the past inspection are accumulated in the storage device 61, it is preferable to use the sample that gives the highest coincidence degree as the sample of the past inspection. Alternatively, the samples of a number or a ratio that are set in advance in order of the coincidence degree from the sample that gives the highest coincidence degree can be used as the targets.

[0052] As part of the parameter range setting process (S104), the parameter range setting unit 62, in order to determine the reference image generation parameters used in reference image generation, changes and sets the range of the reference image generation parameters for the object based on the consistency with past inspection condition parameters. Examples of reference image generation parameters include at least one of resizing amount and chamfer amount. The range of the object's reference image generation parameters is changed from a preset default range. Furthermore, the range of the object's reference image generation parameters is set to a range narrower than the preset default range. Specifically, the parameter range setting unit 62 inputs the inspection condition parameters for the inspection, and based on the consistency with previously implemented inspection condition parameters, changes and sets at least one of the resizing amount and chamfer amount to multiple values ​​used in reference image generation of the reference image generation parameters within a range narrower than the preset default range of the reference image generation parameters.

[0053] Figure 6 This is an example illustrating the relationship between the degree of consistency and the proportion of the narrowed scope in Implementation Method 1. Figure 6 In the examples, when the consistency is 0 (0%) or higher and less than 0.2 (20%), 100% of the default reference image generation parameter range is used. When the consistency is 0.2 (20%) or higher and less than 0.4 (40%), 80% of the default reference image generation parameter range is used. When the consistency is 0.4 (80%) or higher and less than 0.6 (60%), 60% of the default reference image generation parameter range is used. When the consistency is 0.6 (80%) or higher and less than 0.8 (80%), 40% of the default reference image generation parameter range is used. When the consistency is 0.8 (80%) or higher and less than 1 (100%), 20% of the default reference image generation parameter range is used. Alternatively, when the consistency is 1 (100%), 20% of the default reference image generation parameter range can be used, or the consistent past reference image generation parameters can be used as is. Or, a preset percentage within the default reference image generation parameter range of 0 to 20% can be used.

[0054] Alternatively, a threshold Tth can be preset for the consistency, and the parameter range setting unit 62 sets a default range for the reference image generation parameters when the calculated consistency is less than the threshold Tth. In other words, even when the consistency is less than the threshold Tth, it is preferable to use 100% of the default range of reference image generation parameters. Figure 6 In the example, for instance, if the consistency is less than 0.6, it means using 100% of the default reference image generation parameter range.

[0055] The default settings for the scale used in the reference image generation parameter range are as follows.

[0056] Figure 7 is an example of a relationship of a reference image generation parameter, a filter function, a score, and an order in the past examination in Embodiment 1. In Figure 7 In the example of FIG. 17, as the reference image generation parameter, a size adjustment amount and a rounding amount are cited. In Figure 7 In the example of FIG. 17, a case where a range of Al to An is set as a range of the size adjustment amount as a default is shown. Also, as a range of the rounding amount as a default, a case where a range of Bl to Bm is set is shown. The range can be determined by an empirical rule or the like. In calculating the filter function for the reference image generation, a plurality of size adjustment amounts Al, A2, A3,..., An separated by a predetermined interval in the range of the size adjustment amount is used. Similarly, a plurality of rounding amounts Bl, B2, B3,..., Bm separated by a predetermined interval in the range of the rounding amount is used.

[0057] Then, for all combinations obtained by combining the plurality of size adjustment amounts Al, A2, A3,..., An and the plurality of rounding amounts Bl, B2, B3,..., Bm one by one, filter functions al, a2, a3,... are generated.

[0058] At this time, for each combination, an evaluation value of the generated filter function is calculated as a value of the score. As the evaluation value, for example, a sum of squares of differences (sum of difference squares) of gray scales of each pixel of the reference image made using the generated filter function and the corresponding measurement image (optical image), or a square root of the sum of difference squares is preferably used. Alternatively, as the evaluation value, for example, a maximum gray scale difference of the gray scale difference of each pixel of the reference image made using the generated filter function and the corresponding measurement image (optical image) is preferably used. Figure 7 In the example of FIG. 17, a case where the square root of the sum of difference squares is used is shown.

[0059] Then, for each combination, an order from the smallest to the largest of the scores is recorded.

[0060] For each examination in the past, a list of the relationship of the reference image generation parameter, the filter function, the score, and the order is made, and is accumulated in the storage device 61 as past data in association with the test sample of the subject.

[0061] In the past inspection, the best filter function was selected and decided from among all the combinations of filter functions in the default range each time. Thus, in the past inspection, the default reference image generation parameter range was used. Therefore, a plurality of combinations of filter functions needed to be generated, and the operation for generating the filter function for reference image generation required a large amount of processing and time. Therefore, in Embodiment 1, instead of generating all the combinations of filter functions, the range of the reference image generation parameter or the number of used values is changed or reduced. Specifically, the range of the parameter value used is changed for at least one of the default range of the size adjustment amount and the default range of the rounding amount. Specifically, the following operation is performed.

[0062] The parameter range setting section 62 sets the combinations of the size adjustment amount and the rounding amount in the proportion corresponding to the degree of coincidence from the combination side in which the order is the upper order, based on the data of the past inspection. For example, if the degree of coincidence is 0.9, the combinations of the size adjustment amount and the rounding amount in the range in which the order is reduced to 20% of the upper order are extracted, and the extracted combinations of the size adjustment amount and the rounding amount are set.

[0063] Figure 8 is a diagram showing an example of the result of the setting of the reference image generation parameter range in Embodiment 1. In Figure 8 In the example of FIG. 10, the vertical axis represents the rounding amount, and the horizontal axis represents the size adjustment amount. There are n values of the size adjustment amount in the range of Al to An as the default range of the size adjustment amount. There are m values of the rounding amount in the range of Bl to Bm as the default range of the rounding amount. The default range can be set by experience. Thus, there are (m x n) combinations of the size adjustment amount and the rounding amount. In Embodiment 1, of the (m x n) combinations, the combinations corresponding to N% of the degree of coincidence are extracted in order from the side of the smaller score, and thus the number of combinations used for generating the filter function can be reduced to be smaller than the number of combinations of the default range. In Figure 8 In the example of FIG. 10, an example of the plurality of combinations after the reduction is shown.

[0064] As the representative frame selection process (S110), at the time of the inspection of the test object 101, a representative frame used for calculating the filter function is selected. As shown in FIG. 11, the representative frame is selected from among the plurality of frames of the test object 101. The representative frame is selected by the operator or the like. Figure 2As shown, the strip region of each inspection strip 20 is divided into a plurality of frame regions 30, for example, and inspection is performed for each frame region 30. In other words, the inspection region 10 of the sample 101 is virtually divided into a plurality of frame regions 30, and inspection is performed for each frame region 30. Here, a frame region (representative frame 32) for generating a filter function is selected from the inspection region 10 of the sample 101. The position of the selected frame region (representative frame 32) can be arbitrarily set by the user in advance. Then, the position of the selected frame region can be set in the control computer 110. Alternatively, a prescribed selection condition can be set in advance, and the control computer 110 can select the position of the frame region.

[0065] As the strip image acquisition process (S114), the optical image acquisition mechanism 150 acquires an optical image of the inspection strip 20 including the representative frame 32 of the photomask that is the sample 101. Specifically, the following operations are performed.

[0066] First, the XYθ stage 102 is moved to a position at which the inspection strip 20 including the representative frame 32 can be imaged. Laser light (e.g., DUV light) of a wavelength in the ultraviolet region or below that becomes inspection light is irradiated from the appropriate light source 103 via the illumination optical system 170 to the pattern formed on the sample 101. The light that has passed through the sample 101 is imaged as an optical image on the imaging sensor 105 (an example of a sensor) via the magnification optical system 104.

[0067] The image of the pattern imaged on the imaging sensor 105 is photoelectrically converted by each light-receiving element of the imaging sensor 105, and further A / D (analog / digital) converted by the sensor circuit 106. Furthermore, the pixel data of the inspection strip 20 that is the measurement target is stored in the strip pattern memory 123. Then, the strip region image is sent to the filter function operation circuit 140 together with data indicating the position of the sample 101 on the XYθ stage 102 that is output from the position circuit 107. The measurement data (pixel data) is, for example, 8-bit unsigned data indicating the level (light amount) of brightness of each pixel. The strip region image output to the filter function operation circuit 140 is stored in the storage device 71.

[0068] As the representative frame image production process (S116), the frame image production section 72 produces a frame image (representative frame image) of the representative frame 32. Specifically, the strip region image (optical image) of the inspection strip 20 including the representative frame 32 is divided into frames of a prescribed size in the x direction so as to cut out a frame image (representative frame image) of the representative frame 32. For example, the strip region image is divided into 512 x 512 pixel frame images. The data of the divided representative frame image is output to the storage device 73 and stored.

[0069] As a representative design image creation process (S118), an expansion circuit 111 (one example of a design image creation section) creates a design image (representative design image) by performing image expansion based on design pattern data that is the basis of pattern formation of the inspection sample 101. Specifically, the design data is read out from the disk device 109 by the control computer 110, and each graphic pattern of the area of the representative frame 32 defined in the read-out design data is converted into 2-value or multi-value image data (image expansion) to create a design image (representative design image).

[0070] Here, the graphic defined in the design pattern data is, for example, a basic graphic of a rectangle or a triangle, and stores information such as a graphic code as an identifier that distinguishes the graphic type such as a rectangle or a triangle, and coordinates (x, y) at a reference position of the graphic and the length of the side, for example, to define the shape, size, position, and the like of each pattern graphic (vector data).

[0071] If the information of the design pattern as the corresponding graphic data is input to the expansion circuit 111, the data of each graphic is expanded, and the graphic code that represents the graphic shape of the graphic data, the graphic size, and the like are interpreted. Further, as a pattern arranged in a grid in units of a grid of a prescribed quantization size, the design image data of 2-value or multi-value is expanded and output. In other words, the design data is read in, and for each grid into which the inspection area is virtually divided in units of a grid of a prescribed size, the occupancy rate of the graphic in the design pattern is calculated, and n-bit occupancy rate data is output. For example, it is preferable that one grid be set to one pixel. Further, if one pixel has a resolution of 1 / 256 (=1 / 256), the area of the graphic arranged in the pixel is allocated a small area of 1 / 256 to calculate the occupancy rate in the pixel. Then, the design image of 8-bit occupancy rate data is created for each pixel. The data of the design image is output to the filter function operation circuit 140 and stored in the storage device 70. 8

[0072] As a filter function operation process (S130), the filter function operation circuit 140 operates the filter function used to create the reference image. Specifically, it operates as follows.

[0073] Figure 9 is a diagram that represents one example of an internal process of the filter function generation process in Embodiment 1. In Figure 9 ​In the present embodiment, the filter function operation step (S130) implements, as its internal steps, the filter function generation step (S131), the gradation difference calculation step (S138), the determination step (S140), the default setting step (S142), the evaluation value calculation step (S144), and the filter function decision step (S146). The temporary filter function generation step (S131) implements, as its internal steps, the size adjustment step (S132), the rounding processing step (S134), and the filter coefficient calculation step (S136).

[0074] As the filter function generation step (S131), the filter function generation section 64 generates a filter function for creating a reference image using a value within a range of the set reference image generation parameter. Specifically, the filter function generation section 64 generates candidates of a plurality of filter functions for creating a reference image using each of a plurality of values within a range of the changed or reduced reference image generation parameter. Specifically, a candidate of a filter function is generated for each of a plurality of combinations in which the size adjustment amount and the rounding amount are reduced. The following will be described in further detail.

[0075] First, as the size adjustment step (S132), the size adjustment processing section 66 performs size adjustment on the graphic pattern within the design image of the representative frame 32 using each of a plurality of values within a range of the changed or reduced reference image generation parameter. Specifically, size adjustment is performed on the graphic pattern within the design image of the representative frame 32 for each of a plurality of combinations in which the size adjustment amount and the rounding amount are reduced.

[0076] Figure 10 is a drawing showing an example of a graphic pattern before and after size adjustment in Embodiment 1. In Figure 10 In the example of FIG. 13, a graphic pattern 14 after size adjustment of the width dimension of the graphic pattern 12 within the design image by the size adjustment amount A’ is shown. In a case where the default range of the size adjustment amount is, for example, a range from -5 pixels to +5 pixels, a range of, for example, seven values (-1 pixel, -0.5 pixel, 0, +0.5 pixel, +1 pixel, +1.5 pixel, +2 pixel) separated by 0.5 pixels from -1 pixel to +2 pixel is changed or reduced, and the reduced range is set. A negative value indicates that size adjustment is performed in a direction in which the width dimension is reduced. A positive value indicates that size adjustment is performed in a direction in which the width dimension is increased. Further, for example, 0.5 pixel indicates a size adjustment amount of 1 / 2 of a pixel dimension. In Figure 10 In the example of FIG. 13, a case where size adjustment is performed in a direction in which the width dimension of the graphic pattern 12 is reduced is shown. In Figure 10In the example of FIG. 12, an example of resizing the width dimension in the x direction is shown, but it can also be a case of resizing the width dimension in the y direction instead of the width dimension in the x direction. Alternatively, both the width dimension in the x direction and the width dimension in the y direction can be resized.

[0077] Next, as the rounding process (S134), the rounding processing section 67 performs rounding processing on the corner portions of the graphic pattern within the design image of the representative frame 32 using each of the plurality of values within the range of the reference image generation parameters after the change setting or the reduction. Specifically, the rounding processing is performed on the corner portions of the graphic pattern within the design image of the representative frame 32 for each of the plurality of combinations of the size adjustment amount and the rounding amount that are reduced in size. The rounding processing can be performed on the graphic pattern after the size adjustment.

[0078] Figure 11 FIG. 11 is a diagram for illustrating an example of the graphic pattern before and after the rounding processing in Embodiment 1. In FIG. 11, the example of FIG. 10 is shown as an example of the graphic pattern before the rounding processing. Figure 11 In the example of FIG. 12, an example of the corner (corner portion) of the graphic pattern 14 after the size adjustment is rounded with the rounding amount B' of the graphic pattern 16. In a case where the default range of the rounding amount is, for example, a range from 0 to +2 pixels, for example, it is assumed that the default range of the rounding amount is changed or reduced to five values (0.2 pixels, 0.4 pixels, 0.6 pixels, 0.8 pixels, and 1 pixel) spaced apart by 0.2 pixels in a range from 0.2 pixels to 1 pixel. In addition, for example, 0.2 pixels indicates a rounding amount of a size of 1 / 5 of the pixel size. The rounding amount indicates the radius of curvature. In the example of FIG. 12, the upper left corner is shown, but the other corners are also rounded in the same manner. Figure 11 In the example of FIG. 12, the upper left corner is shown, but the other corners are also rounded in the same manner.

[0079] As the filter coefficient calculation process (S136), the filter coefficient calculation section 68 calculates the coefficients of the filter function using the design image on which the size adjustment and / or the rounding processing is performed using each of the plurality of values within the range of the reference image generation parameters after the change setting or the reduction. Specifically, the coefficients of the filter function are calculated using the design image on which the size adjustment and / or the rounding processing is performed for each of the plurality of combinations of the size adjustment amount and the rounding amount that are reduced in size.

[0080] Figure 12 FIG. 10 is a diagram for illustrating an example of the relationship between the filter function, the design image, and the measurement image in Embodiment 1.

[0081] Figure 13 FIG. 13 is a diagram for illustrating an example of a method of calculating the coefficients of the filter function in Embodiment 1. For example, as shown in FIG. 13, the filter coefficient calculation section 68 calculates the coefficients of the filter function using the design image on which the size adjustment and / or the rounding processing is performed for each of the plurality of combinations of the size adjustment amount and the rounding amount that are reduced in size. Figure 12As shown, the unknown coefficient matrix a(i,j) consisting of k×k fewer elements than the number of pixels in the frame region is obtained by the following method (an example of a filtering function). For example, for an image with a frame region consisting of 512×512 pixels, a 15×15 coefficient matrix a(i,j) is obtained. Centered on the focus pixel d(i,i) of the representative design image that has undergone resizing and / or rounding, the value obtained by dividing the sum of the products of k×k pixels and the coefficient matrix a(i,j) by the number of pixels N (=k×k) is closer to the coefficient matrix a(i,j) of the focus pixel r(i,i) of the representative frame image corresponding to the focus pixel d(i,i). This relationship (2) is shown below.

[0082] like Figure 13 As shown, while moving the pixel of interest within the representative frame 32, the relation (2) is calculated each time. Furthermore, the coefficient matrix a(i,j) that best satisfies the relation (2) defined using the unknown coefficient matrix a(i,j) is obtained for all pixels within the representative frame 32. The number of elements k×k in the coefficient matrix a(i,j) can be appropriately set. If it is too small, the accuracy deteriorates; if it is too large, the computation time increases. Additionally, when the pixel of interest moves within the representative frame 32, if it is near the end, sometimes the necessary number of surrounding pixels on the end side may not exist. However, in this case, it is sufficient to perform the calculation using the surrounding pixels with the obtained value and the number of pixels N.

[0083] This yields the coefficient matrix a(i,j) of the filter function for each of several combinations of reduced sizing and rounding amounts. These coefficient matrices a(i,j) are candidate filter functions for creating the reference image. The actual filter function needs to be selected from these coefficient matrices a(i,j).

[0084] Furthermore, when the consistency is less than the threshold Tth, the parameter range setting unit 62 sometimes sets a default range for the image generation parameters. In this case, the filter function generation unit 64 uses each of the multiple image generation parameters within the default range of the image generation parameters to generate candidate filter functions for creating the image. This way, even with low consistency, the range of image generation parameters that can be used is not narrowed.

[0085] As the gray scale difference calculation process (S138), first, the reference circuit 112 makes candidates of a plurality of reference images using candidates of a plurality of filter functions generated in the range of the reference image generation parameters after the change in setting or the reduction. Specifically, the reference circuit 112 acts as follows. The reference circuit 112 makes a candidate of a reference image corresponding to the representative frame image using a candidate of a filter function for each of a plurality of combinations of the size adjustment amount and the round amount after the reduction, for each of the combinations. The reference circuit 112 first performs the size adjustment and the round processing of the combination on the pattern within the design image of the representative frame region 32, and then, for each pixel of the design image, makes reference image data obtained by dividing, by the number N (= k x k) of pixels, the sum of the product of the k x k pixels and the coefficient matrix a (i, j) that becomes the filter function for the combination, with the pixel d (i, i) at the center. The obtained reference image data is sent to the filter function operation circuit 140 and stored in the storage device 75.

[0086] Next, the gray scale difference calculation section 80 calculates the gray scale difference Δ obtained by subtracting the pixel value of the measurement image from the pixel value of the reference image, for each of a plurality of combinations of the size adjustment amount and the round amount after the reduction. The gray scale difference is calculated for each pixel within the image.

[0087] As the determination process (S140), the determination section 82 determines whether there is a candidate of a reference image for which the maximum gray scale difference Δmax of the reference image and the representative frame image (optical image) is equal to or less than the threshold value ΔTh, using the maximum gray scale difference Δmax of each of the candidates of a plurality of reference images. The maximum gray scale difference Δmax indicates the maximum value among a plurality of gray scale differences Δ obtained by all the pixels within the image.

[0088] In a case where there is no candidate of a reference image for which the maximum gray scale difference Δmax is equal to or less than the threshold value ΔTh among the candidates of a plurality of reference images, the default setting process (S142) is entered. In a case where there is a candidate of a reference image for which the maximum gray scale difference Δmax is equal to or less than the threshold value ΔTh, the evaluation value calculation process (S144) is entered.

[0089] As the default setting process (S142), the parameter range setting section 62 re-sets the default range of the reference image generation parameters in a case where the maximum gray scale difference Δmax exceeds the threshold value ΔTh among all the candidates of a plurality of reference images. In other words, among a plurality of combinations of the size adjustment amount and the round amount after the reduction, the range is too narrow, and thus the setting of the default range is returned. Then, the filter function generation process (S131) is returned to, and the filter function generation section 64 generates a plurality of filter functions for making a candidate of a reference image again using each of a plurality of reference image generation parameters within the default range of the reference image generation parameters. Then, the gray scale difference calculation process (S138) and the determination process (S140) are performed.

[0090] As the evaluation value calculation step (S144), the evaluation value calculation section 86 calculates the evaluation value of the filter function that has made the candidate of the reference image in which the maximum gray scale difference Δmax is below the threshold value ΔTh. As the evaluation value, it is preferable to use the already calculated gray scale difference Δ, such as the sum of squares of differences (sum of squares of differences) or the square root of the sum of squares of differences within the image. In the case of using the maximum gray scale difference as the evaluation value, since it has already been calculated, this step is omitted. The calculated evaluation value is recorded as the value of the above-mentioned score together with the reference image generation parameter and the filter function. It is used as one of the past data at the time of inspection of the future sample.

[0091] As the filter function decision step (S146), the decision section 84 decides the true filter function used to make the reference image among the candidates of the plurality of filter functions generated. For example, if the candidate of the reference image in which the maximum gray scale difference Δmax is below the threshold value ΔTh is one, the filter function used in the making of the reference image is decided as the true filter function. For example, in the case where there are two or more candidates of the reference image in which the maximum gray scale difference Δmax is below the threshold value ΔTh, it is preferable to decide the filter function used in the making of the reference image of the side (the side of the higher order of the score) having the smaller evaluation value (such as the square root of the sum of squares of differences) obtained in the evaluation value calculation step (S144) as the true filter function.

[0092] As described above, by changing the setting according to the degree of agreement or narrowing the range of the reference image generation parameter, it is possible to reduce the amount of processing generated in the generation of the filter function, and it is possible to shorten the time. The decided filter function (coefficient matrix) is set in the reference circuit. Similarly, the amount of size adjustment and the amount of rounding at that time are set in the reference circuit. Then, the inspection processing of the entire sample 101 is started.

[0093] Figure 14 is a diagram showing an example of the internal configuration of the comparison circuit in Embodiment 1. In Figure 14 In the comparison circuit 108, the storage devices 50, 52, 56 such as a magnetic disk device, the frame image making section 54, the alignment section 57, and the comparison section 58 are arranged. The series of “~ sections” of the frame image making section 54, the alignment section 57, and the comparison section 58 have processing circuits. In the corresponding processing circuits, circuits, computers, processors, circuit boards, quantum circuits, or semiconductor devices, or the like are included. In addition, each of the “~ sections” can use a common processing circuit (the same processing circuit). Or a different processing circuit (each processing circuit) can be used. The input data required for the frame image making section 54, the alignment section 57, and the comparison section 58 or the results obtained by the operation are stored in a not-shown memory or the memory 115 within the comparison circuit 108 each time.

[0094] As part of the strip image acquisition process (S202) (also known as the scanning process or optical image acquisition process), the optical image acquisition mechanism 150 acquires an optical image of the photomask that forms the sample 101. The method for acquiring the strip image is the same as described above. However, here, as... Figure 2 As shown, strip images are acquired sequentially. Furthermore, pixel data is stored in the strip pattern memory 123 for each inspected strip 20. Then, the strip region image, along with data representing the position of the photomask 101 on the XYθ stage 102 output from the position circuit 107, is sent to the comparison circuit 108. The measurement data (pixel data) is, for example, 8 bits of unsigned data, representing the brightness level (light intensity) of each pixel. The strip region image output to the comparison circuit 108 is stored in the storage device 50.

[0095] As part of the frame image creation process (S204), within the comparison circuit 108, the frame image creation unit 54 generates multiple frame images 31 that divide the striped region image (optical image) by a predetermined width. Specifically, as... Figure 2 As shown, the striped region image is divided into multiple rectangular frame regions 30. For example, the segments are divided into 512×512 pixel sizes. The data of each frame region 30 is stored in the storage device 56.

[0096] As part of the design image creation process (S212), the unfolding circuit 111 (design image creation unit) creates a design image by unfolding the design pattern data that forms the basis of the pattern formation of the sample 101 to be inspected. Specifically, the unfolding circuit 111 reads the design data from the disk device 109 via the control computer 110, and transforms each graphic pattern of each frame region 30 defined in the read design data into two-valued or multi-valued image data. Then, an 8-bit occupancy data design image is created for each pixel. The design image data (image data) is output to the reference circuit 112.

[0097] As part of the reference image creation process (S214), the reference circuit 112 (reference image creation unit) uses a determined actual filtering function to create a reference image for comparison with an optical image. Specifically, after adjusting and rounding the design image according to the set size adjustment amount and rounding amount, the reference circuit 112 performs filtering using a set coefficient matrix a(i,j) (an example of a filtering 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. Through the above, image (reference image) data for comparison is generated for inspection.

[0098] In addition, the coefficient matrix a(i, j) (an example of a filter function) calculated using the representative design image representing the frame is used for filter processing of all design images (a plurality of design images) in which the frame region 30 within the inspection region 10 differs from each other. In other words, the plurality of design images are subjected to filter processing using the same filter function calculated respectively.

[0099] The comparison process (S230) and the comparison circuit 108 (an example of a comparison section) compare the captured optical image and the reference image and output the result. Specifically, the following operation is performed. First, the alignment section 57 reads out the frame image (optical image) to be compared from the storage device 56, and similarly, the reference image to be compared is read out from the storage device 52. Then, alignment is performed with a predetermined algorithm. For example, alignment is performed using the least square method. Further, the comparison section 58 compares the two for each pixel according to a predetermined determination condition, for example, determines the presence or absence of a shape defect as a defect. As the determination condition, for example, the two are compared for each pixel according to a predetermined algorithm to determine the presence or absence of a defect. Then, the comparison result is output. The comparison result is output, for example, to the disk device 109, the floppy disk device (FD) 115, the CRT 117, the pattern monitor 118, or can be output from the printer 119.

[0100] Figure 15 is a diagram showing an example of a range of reference image generation parameters and a result of the setting in the modification example of Embodiment 1. In Figure 15 In the example of Figure 8 In the example of Figure 15 As shown in Figure 15 In the example of

[0101] As described above, according to Embodiment 1, it is possible to reduce the amount of processing until a filter function suitable for the present inspection is obtained. Further, it is possible to shorten the processing time until a filter function suitable for the present inspection is obtained.

[0102] The above describes the embodiments with reference to specific examples. However, the present application is not limited to these specific examples. For example, in the embodiments, as the illumination optical system 170, a transmissive illumination optical system using transmitted light is shown, but is not limited thereto. For example, it can also be a reflective illumination optical system using reflected light. Or, it can also be a combination of a transmissive illumination optical system and a reflective illumination optical system, and use of both transmitted light and reflected light.

[0103] In addition, the filter function and the coefficient of the filter function described above are one example, and are not limited thereto. It can also be a case where another filter function and the coefficient of the filter function are used.

[0104] In addition, parts and the like for which description of the present application is not directly necessary, such as device configuration and control method, are omitted from the description, but necessary device configuration and control method can be appropriately selected and used. For example, for the control section configuration for controlling the inspection device 100, description is omitted, but of course, necessary control section configuration can be appropriately selected and used.

[0105] Further, all the pattern inspection devices and pattern inspection methods which are designed to be changed by those skilled in the art with the elements of the present application included therein are within the scope of the present application.

[0106] Industrial Applicability One embodiment of the present application relates to a pattern inspection method and a pattern inspection device. For example, it relates to a pattern inspection technique for inspecting a pattern defect of an object which is a sample used in semiconductor manufacturing, and can be used in a method for inspecting a defect of a mask used when a semiconductor element is manufactured.

[0107] Explanation of Reference Numerals 10 Inspection region 20 Inspection strip 30 Frame region 31 Frame image 32 Representative frame 54 Frame image creation section 57 Alignment section 58 Comparison section 60 Degree of agreement calculation section 61, 70, 71, 73, 75 Storage device 62 Parameter range setting section 64 Filter function generation section 66 Size adjustment processing section 67 Round processing section 68 Filter coefficient calculation section 72 Frame image creation section 80 Gray scale difference calculation section 82 Determination section 84 decision section 86 evaluation value calculation section 100 inspection apparatus 101 sample 102 XYθ stage 103 light source 104 magnifying optical system 105 photodiode array 106 sensor circuit 107 position circuit 108 comparison circuit 109 disk device 110 control computer 111 expansion circuit 112 reference circuit 113 auto loader control circuit 114 stage control circuit 115 FD 116 memory 117 CRT 118 pattern monitor 119 printer 120 bus 122 laser length measuring system 123 strip pattern memory 130 auto loader 140 coefficient operation circuit 150 optical image acquisition section 160 control system circuit 170 illumination optical system

Claims

1. A pattern inspection device, characterized in that, have: An optical image acquisition mechanism that acquires an optical image of a patterned specimen under inspection; The parameter range setting circuit sets the range of parameters for generating reference images based on the consistency with past inspection condition parameters. A filter function generation circuit generates multiple candidate filter functions using values ​​within the set range of the reference image generation parameters; A decision circuit that selects a filter function from the generated candidate filter functions to create a reference image; A reference image fabrication circuit uses the determined filtering function to fabricate a reference image; and A comparison circuit that compares the optical image with the reference image.

2. The pattern inspection device according to claim 1, characterized in that, The reference image generation parameters are at least one of size adjustment amount and chamfer amount.

3. The pattern inspection device according to claim 1, characterized in that, The range of the reference image generation parameters is set to a smaller range than the preset default range.

4. The pattern inspection device according to claim 1, characterized in that, When the consistency is less than a threshold, the parameter range setting circuit sets a preset default range for the reference image generation parameters. The filter function generation circuit uses each of the plurality of reference image generation parameters within the default range of the reference image generation parameters to generate candidate filter functions for creating the reference image.

5. The pattern inspection device according to claim 1, characterized in that, It also includes a consistency calculation circuit that uses at least one of the parameters of the pattern conditions and the parameters of the shooting conditions to calculate the consistency.

6. The pattern inspection device according to claim 1, characterized in that, It also includes a determination circuit that uses candidate filters generated within the set range of the reference image generation parameters, and uses the maximum grayscale difference between each of the candidate reference images generated by the reference image production unit and the optical image to determine whether there is a candidate reference image whose maximum grayscale difference is below a threshold. If the maximum grayscale difference among all the candidate reference images exceeds a threshold, the parameter range setting circuit resets the preset default range of the reference image generation parameters. The filter function generation circuit uses each of the plurality of reference image generation parameters within the default range of the reference image generation parameters to generate candidate filter functions for creating the reference image.

7. A pattern inspection method, characterized in that, Obtain an optical image of the patterned specimen under inspection. Based on the consistency with past inspection condition parameters, set the range of parameters for generating reference images. Using the values ​​within the set range of the reference image generation parameters, generate multiple candidate filter functions. The filter function to be used to create the reference image is determined from the candidate filter functions generated. Use the determined filtering function to create a reference image. The optical image is compared with the reference image, and the result is output.

8. The pattern inspection method according to claim 7, characterized in that, The reference image generation parameters are at least one of size adjustment amount and chamfer amount.

9. The pattern inspection method according to claim 7, characterized in that, The range of the reference image generation parameters for the object is set to a smaller range than the preset default range.

10. The pattern inspection method according to claim 7, characterized in that, When the consistency is less than a threshold, the parameter range setting circuit sets a preset default range for the reference image generation parameters. Candidates for the plurality of filter functions used to create the reference image are generated using each of the plurality of reference image generation parameters within the default range of the reference image generation parameters.

11. The pattern inspection method according to claim 7, characterized in that, The degree of consistency is calculated using at least one of the parameters of the pattern conditions and the parameters of the shooting conditions.

12. The pattern inspection method according to claim 7, characterized in that, Using the candidate filter functions generated within the set range of the reference image generation parameters, and using the maximum grayscale difference between each of the candidate reference images generated by the reference image production unit and the optical image, it is determined whether there are candidate reference images whose maximum grayscale difference is below a threshold. Among all the candidate reference images, if the maximum grayscale difference exceeds a threshold, the preset default range of the reference image generation parameters is reset. Using each of the multiple reference image generation parameters within the default range of the reference image generation parameters, candidates for the multiple filtering functions used to create the reference image are generated again.

Images