JP7547138B2 - Electron beam inspection device and electron beam inspection method - Google Patents

Description

One aspect of the present invention relates to an electron beam inspection device and an electron beam inspection method. For example, the present invention relates to an inspection device and method that irradiates a substrate with multiple electron beams and uses a secondary electron image of the emitted pattern to perform inspection.

In recent years, with the increasing integration and capacity of large-scale integrated circuits (LSIs), the circuit line width required for semiconductor elements has become narrower and narrower. In addition, improving the yield is essential for the manufacture of LSIs, which require a large manufacturing cost. However, the patterns that make up LSIs are approaching the order of 10 nanometers or less, and the dimensions that must be detected as pattern defects are also extremely small. Therefore, there is a need to improve the accuracy of pattern inspection devices that inspect defects in ultra-fine patterns transferred onto semiconductor wafers. Another major factor that reduces the yield is pattern defects in masks used when exposing and transferring ultra-fine patterns onto semiconductor wafers using photolithography technology. Therefore, there is a need to improve the accuracy of pattern inspection devices that inspect defects in transfer masks used in LSI manufacturing.

A known defect inspection method is to use ultraviolet light or an electron beam to capture a measurement image of a pattern formed on a substrate, such as a semiconductor wafer or a lithography mask, and compare this measurement image with design data or a measurement image of the same pattern on the substrate.

In particular, in die-database inspection, which compares a measurement image with a reference image based on design pattern data, a reference image that resembles the measurement image is created by filtering a design image created from design pattern data. Conventionally, inspection devices that capture images using ultraviolet light have used a method of calculating the filter model and coefficients for creating a reference image using machine learning or the like before starting inspection (see, for example, Patent Document 1).

On the other hand, in die-database inspection, which compares a SEM (Scanning Electron Microscope) image obtained by scanning a sample with an electron beam with a reference image, the image quality of the SEM image obtained by scanning a sample with an electron beam is expected to change little by little over time from the time the scanning operation started. As a result, die-database inspection has the problem that the difference between the SEM image and the reference image tends to increase over time, resulting in frequent false defects.

JP 2016-133613 A

One aspect of the present invention provides an apparatus and method capable of creating a reference image to track changes in image quality of a measurement image obtained by scanning a substrate with an electron beam.

An electron beam inspection apparatus according to one aspect of the present invention comprises:
a reference image creation unit that creates a reference image corresponding to the target electron optical image by filtering a design image corresponding to the target electron optical image for each of the plurality of electron optical images using a predetermined image processing filter;
a comparison unit for comparing the electro-optical image of the object with a reference image corresponding to the electro-optical image of the object;
a database storing information on a plurality of first image processing filters that can be searched for using feature information indicating features;
a search unit that searches for a plurality of second image processing filters having predetermined characteristics based on characteristics of an image processing filter currently selected as a predetermined image processing filter from among the plurality of first image processing filters using a database;
a selection unit that selects, from among a plurality of second image processing filters, a second image processing filter that produces a reference image that matches the electro-optical image better than the reference image produced using the image processing filter currently selected for comparison with the electro-optical image;
The present invention is characterized by comprising:

In addition, it is preferable to use, as a characteristic of each first image processing filter, the area of the reference pattern after filtering when filtering an image of the reference pattern with the first image processing filter.

In addition, it is preferable to use, as a characteristic of each first image processing filter, the distance between the mid-tone position of the reference pattern after filtering when the image of the reference pattern is filtered with the first image processing filter and a reference position outside the pattern.

In addition, it is preferable to use, as a characteristic of each first image processing filter, the maximum and minimum gradation values of the reference pattern after filtering when filtering an image of the reference pattern with the first image processing filter.

In addition, it is preferable to use the rank of the micro-figures in the design image used to obtain the first image processing filter and the number of micro-figures in the rank as the features of each first image processing filter.

The selection unit selects, from among the second image processing filters, two or more second image processing filters that generate a reference image having a high degree of matching;
It is preferable that the reference image creation unit uses the two or more selected second image processing filters to perform filter processing on a design image among the remaining electron-optical images, which corresponds to the target electron-optical image, to create two or more reference image candidates corresponding to the target electron-optical image, and creates a reference image by combining the two or more reference image candidates.

In addition, it is preferable that the reference image creation unit uses the selected second image processing filter to perform filter processing on a design image corresponding to the target electron-optical image among the remaining electron-optical images of the multiple electron-optical images for which comparison processing has not yet been performed, thereby creating a reference image corresponding to the target electron-optical image.

An electron beam inspection method according to one aspect of the present invention includes:
creating a reference image corresponding to the target electron-optical image by filtering a design image corresponding to the target electron-optical image for each of the plurality of electron-optical images using a predetermined image processing filter;
comparing the electro-optical image of the object with a reference image corresponding to the electro-optical image of the object and outputting the result;
a step of searching for a plurality of second image processing filters having predetermined characteristics based on the characteristics of an image processing filter currently selected as a predetermined image processing filter from among the plurality of first image processing filters, using a database storing information on a plurality of first image processing filters that can be searched using characteristic information indicating the characteristics;
selecting a second image processing filter from among the plurality of second image processing filters that produces a reference image that matches more closely with a remaining electro-optical image of the plurality of electro-optical images than a reference image produced using the currently selected image processing filter;
The present invention is characterized by comprising:

In addition, it is preferable to use, as a characteristic of each first image processing filter, the area of the reference pattern after filtering when filtering an image of the reference pattern with the first image processing filter.

In addition, it is preferable to use, as a characteristic of each first image processing filter, the distance between the mid-tone position of the reference pattern after filtering when the image of the reference pattern is filtered with the first image processing filter and a reference position outside the pattern.

In addition, it is preferable to use, as a characteristic of each first image processing filter, the maximum and minimum gradation values of the reference pattern after filtering when filtering an image of the reference pattern with the first image processing filter.

According to one aspect of the present invention, a reference image can be created to track changes in image quality of a measurement image obtained by scanning a substrate with an electron beam.

1 is a configuration diagram showing an example of a configuration of an inspection device according to a first embodiment; 1 is a conceptual diagram showing a configuration of a shaping aperture array substrate in embodiment 1. 1 is a diagram showing an example of a plurality of chip regions formed on a semiconductor substrate in a first embodiment; 1 is a diagram for explaining a multi-beam scanning operation in embodiment 1. FIG. FIG. 2 is a flowchart showing main steps of the inspection method according to the first embodiment. 4 is a diagram for explaining an example of a method for calculating a filter function in the first embodiment. FIG. 2 is a block diagram showing an example of an internal configuration of a filter updating circuit according to the first embodiment. FIG. 4 is a diagram showing an example of a reference design image in the first embodiment. FIG. 4 is a diagram showing an example of a reference image after filtering in the first embodiment. 5A to 5C are diagrams showing an example of features of a reference image in the first embodiment. 13A and 13B are diagrams showing another example of features of the reference image in the first embodiment. FIG. 2 is a diagram showing an example of a hierarchy table in the first embodiment. 2 is a configuration diagram showing an example of a configuration within a comparison circuit according to the first embodiment; FIG. FIG. 11 is a diagram for explaining similar filter functions in the first embodiment. FIG. 2 is a diagram for explaining a search method in the first embodiment. 13A to 13C are diagrams showing an example of a plurality of reference image candidates and a synthetic reference image in embodiment 2. FIG. 13 is a block diagram showing an example of an internal configuration of a filter updating circuit according to a third embodiment. 13 is a diagram for explaining a micro graphic rank in the third embodiment. FIG. FIG. 13 is a diagram showing an example of a hierarchy table in the third embodiment.

In the following embodiments, an electron beam inspection device will be described as an example of an inspection device. In addition, in the embodiments, an inspection device that acquires an image using a multi-beam of multiple electron beams will be described, but the present invention is not limited to this. An inspection device that acquires an image using a single beam of one electron beam may also be used.

Embodiment 1.
FIG. 1 is a configuration diagram showing an example of the configuration of an inspection apparatus in the first embodiment. In FIG. 1, an inspection apparatus 100 for inspecting a pattern formed on a substrate is an example of a multi-electron beam inspection apparatus. The inspection apparatus 100 includes an image acquisition mechanism 150 (secondary electron image acquisition mechanism) and a control circuit 160. The image acquisition mechanism 150 includes an electron beam column 102 (electron lens barrel) and an inspection chamber 103. In the electron beam column 102, an electron gun 201, an electromagnetic lens 202, a shaping aperture array substrate 203, an electromagnetic lens 205, a collective blanking deflector 212, a limiting aperture substrate 213, an electromagnetic lens 206, an electromagnetic lens 207 (objective lens), a main deflector 208, a sub-deflector 209, a beam separator 214, a deflector 218, an electromagnetic lens 224, an electromagnetic lens 226, and a multi-detector 222 are arranged. 1, the electron gun 201, the electromagnetic lens 202, the shaping aperture array substrate 203, the electromagnetic lens 205, the collective blanking deflector 212, the limiting aperture substrate 213, the electromagnetic lens 206, the electromagnetic lens 207 (objective lens), the main deflector 208, and the sub-deflector 209 constitute a primary electron optical system that irradiates the substrate 101 with the multiple primary electron beams 20. The beam separator 214, the deflector 218, the electromagnetic lens 224, and the electromagnetic lens 226 constitute a secondary electron optical system that irradiates the substrate 222 with the multiple secondary electron beams 300.

In the inspection chamber 103, a stage 105 that can move in the XY direction is placed. On the stage 105, a substrate 101 (sample) to be inspected is placed. Here, the substrate 101 is supported by, for example, three points. The substrate 101 includes an exposure mask substrate and a semiconductor substrate such as a silicon wafer. When the substrate 101 is a semiconductor substrate, a plurality of chip patterns (wafer dies) are formed on the semiconductor substrate. When the substrate 101 is an exposure mask substrate, a chip pattern is formed on the exposure mask substrate. The chip pattern is composed of a plurality of figure patterns. The chip pattern formed on the exposure mask substrate is exposed and transferred onto the semiconductor substrate a plurality of times, so that a plurality of chip patterns (wafer dies) are formed on the semiconductor substrate. The following mainly describes the case where the substrate 101 is a semiconductor substrate. The substrate 101 is placed on the stage 105, for example, with the pattern forming surface facing upward. Also, a mirror 216 is placed on the stage 105 to reflect the laser light for laser length measurement irradiated from a laser length measurement system 122 placed outside the inspection chamber 103. The multi-detector 222 is connected to the detection circuit 106 outside the electron beam column 102.

In the control system circuit 160, a control computer 110 that controls the entire inspection device 100 is connected to a position circuit 107, a comparison circuit 108, an unfolding circuit 111, a reference image creation circuit 112, a stage control circuit 114, a lens control circuit 124, a blanking control circuit 126, a deflection control circuit 128, a filter update circuit 132, a filter calculation circuit 134, a database 130, a storage device 109 such as a magnetic disk device, a monitor 117, and a memory 118 via a bus 120. The deflection control circuit 128 is also connected to DAC (digital-to-analog conversion) amplifiers 144, 146, and 148. The DAC amplifier 146 is connected to a main deflector 208, and the DAC amplifier 144 is connected to a sub-deflector 209. The DAC amplifier 148 is connected to a deflector 218.

The detection circuit 106 is also connected to the chip pattern memory 123. The chip pattern memory 123 is connected to the comparison circuit 108 and the filter calculation circuit 134. The stage 105 is also driven by the drive mechanism 142 under the control of the stage control circuit 114. The drive mechanism 142 is configured with a drive system such as a three-axis (X-Y-θ) motor that drives in the X, Y, and θ directions in the stage coordinate system, and the stage 105 is movable. The stage 105 can be moved in the horizontal and rotational directions by the motors of the X, Y, and θ axes. The moving position of the stage 105 is measured by the laser measurement system 122 and supplied to the position circuit 107. The laser measurement system 122 measures the position of the stage 105 by receiving the reflected light from the mirror 216 based on the principle of laser interference. For example, the X, Y, and θ directions of the stage coordinate system are set with respect to a plane perpendicular to the optical axis (electron orbit central axis) of the multi-primary electron beam.

The electromagnetic lens 202, the electromagnetic lens 205, the electromagnetic lens 206, the electromagnetic lens 207 (objective lens), the electromagnetic lens 224, the electromagnetic lens 226, and the beam separator 214 are controlled by the lens control circuit 124. The collective blanking deflector 212 is composed of two or more electrodes, and is controlled by the blanking control circuit 126 via a DAC amplifier (not shown) for each electrode. The sub-deflector 209 is composed of four or more electrodes, and is controlled by the deflection control circuit 128 via a DAC amplifier 144 for each electrode. The main deflector 208 is composed of four or more electrodes, and is controlled by the deflection control circuit 128 via a DAC amplifier 146 for each electrode. The deflector 218 is composed of four or more electrodes, and is controlled by the deflection control circuit 128 via a DAC amplifier 148 for each electrode.

The electron gun 201 is connected to a high-voltage power supply circuit (not shown), and an acceleration voltage is applied from the high-voltage power supply circuit between a filament (cathode) (not shown) and an extraction electrode (anode) inside the electron gun 201. In addition, a voltage is applied to another extraction electrode (Wehnelt) and the cathode is heated to a predetermined temperature, causing the group of electrons emitted from the cathode to be accelerated and emitted as an electron beam 200.

Here, FIG. 1 shows the configuration necessary for explaining the first embodiment. The inspection device 100 may also be provided with other configurations that are normally required.

FIG. 2 is a conceptual diagram showing the configuration of the shaping aperture array substrate in the first embodiment. In FIG. 2, the shaping aperture array substrate 203 has two-dimensional holes (apertures) 22 of m ₁ columns (x direction) by n ₁ rows (y direction) (one of m ₁ and n ₁ is an integer of 2 or more, and the other is an integer of 1 or more) formed at a predetermined arrangement pitch in the x and y directions. The example of FIG. 2 shows a case where 23×23 holes (apertures) 22 are formed. Ideally, each hole 22 is formed as a rectangle of the same size and shape. Alternatively, ideally, they may be circles of the same outer diameter. When a part of the electron beam 200 passes through each of these multiple holes 22, m ₁ ×n ₁ (=N) multi-primary electron beams 20 are formed.

Next, we will explain the operation of the image acquisition mechanism 150 in the inspection device 100.

The electron beam 200 emitted from the electron gun 201 (emission source) is refracted by the electromagnetic lens 202 and illuminates the entire shaping aperture array substrate 203. As shown in FIG. 2, a plurality of holes 22 (openings) are formed in the shaping aperture array substrate 203, and the electron beam 200 illuminates the area including all of the plurality of holes 22. Each portion of the electron beam 200 irradiated at the position of the plurality of holes 22 passes through each of the plurality of holes 22 in the shaping aperture array substrate 203, thereby forming a multi-primary electron beam 20.

The multi-primary electron beam 20 thus formed is refracted by the electromagnetic lens 205 and the electromagnetic lens 206, respectively, and while repeating intermediate images and crossovers, it passes through a beam separator 214 arranged at the crossover position (intermediate image position of each beam) of the multi-primary electron beam 20 and proceeds to the electromagnetic lens 207 (objective lens). Then, the electromagnetic lens 207 focuses the multi-primary electron beam 20 on the substrate 101. The multi-primary electron beam 20 focused on the substrate 101 (sample) surface by the objective lens 207 is deflected collectively by the main deflector 208 and the sub-deflector 209, and each beam is irradiated at its respective irradiation position on the substrate 101. Note that when the entire multi-primary electron beam 20 is deflected collectively by the collective blanking deflector 212, it is shifted from the center hole of the limiting aperture substrate 213 and is shielded by the limiting aperture substrate 213. On the other hand, the multi-primary electron beams 20 that are not deflected by the collective blanking deflector 212 pass through a hole in the center of the limiting aperture substrate 213 as shown in FIG. 1. Blanking control is performed by turning the collective blanking deflector 212 ON/OFF, and the beam ON/OFF is controlled collectively. In this way, the limiting aperture substrate 213 shields the multi-primary electron beams 20 that are deflected by the collective blanking deflector 212 to be in the beam OFF state. Then, the multi-primary electron beams 20 for inspection (for image acquisition) are formed by a group of beams that pass through the limiting aperture substrate 213 and are formed from when the beam is turned ON until it is turned OFF.

When the multi-primary electron beams 20 are irradiated onto a desired position on the substrate 101, a bundle of secondary electrons (multi-secondary electron beams 300) including reflected electrons corresponding to each beam of the multi-primary electron beams 20 is emitted from the substrate 101 as a result of the irradiation of the multi-primary electron beams 20.

The multiple secondary electron beams 300 emitted from the substrate 101 pass through the electromagnetic lens 207 and proceed to the beam separator 214.

Here, the beam separator 214 generates an electric field and a magnetic field in perpendicular directions on a plane perpendicular to the direction in which the central beam of the multi-primary electron beam 20 travels (the central axis of the electron orbit). The electric field exerts a force in the same direction regardless of the direction of electron travel. In contrast, the magnetic field exerts a force according to Fleming's left-hand rule. Therefore, the direction of the force acting on the electrons can be changed depending on the direction in which the electrons enter. For the multi-primary electron beams 20 entering the beam separator 214 from above, the force due to the electric field and the force due to the magnetic field cancel each other out, and the multi-primary electron beams 20 travel straight downward. In contrast, for the multi-secondary electron beams 300 entering the beam separator 214 from below, the force due to the electric field and the force due to the magnetic field both act in the same direction, and the multi-secondary electron beams 300 are bent diagonally upward and separated from the multi-primary electron beams 20.

The multi-secondary electron beam 300, which is bent obliquely upward and separated from the multi-primary electron beam 20, is further bent by the deflector 218 and projected onto the multi-detector 222 while being refracted by the electromagnetic lenses 224 and 226. The multi-detector 222 detects secondary electrons emitted from the substrate 101 by irradiation with the multi-primary electron beam 20. Specifically, the multi-detector 222 detects the projected multi-secondary electron beam 300. Reflected electrons and secondary electrons may be projected onto the multi-detector 222, or the reflected electrons may diverge midway and the remaining secondary electrons may be projected onto the multi-detector 222. The multi-detector 222 has a two-dimensional sensor. Then, each secondary electron of the multi-secondary electron beam 300 collides with a corresponding area of the two-dimensional sensor, generating electrons, and secondary electron image data is generated for each pixel. In other words, a detection sensor is arranged for each primary electron beam of the multi-primary electron beam 20 in the multi-detector 222. Then, the corresponding secondary electron beams emitted by irradiation of each primary electron beam are detected. Therefore, each of the multiple detection sensors of the multi-detector 222 detects the intensity signal of the secondary electron beam for the image caused by the irradiation of the primary electron beam for which it is responsible. The intensity signal detected by the multi-detector 222 is output to the detection circuit 106.

3 is a diagram showing an example of a plurality of chip regions formed on a semiconductor substrate in the first embodiment. In FIG. 3, when the substrate 101 is a semiconductor substrate (wafer), a plurality of chips (wafer dies) 332 are formed in a two-dimensional array in an inspection region 330 of the semiconductor substrate (wafer). A mask pattern for one chip formed on an exposure mask substrate is transferred to each chip 332 by, for example, reducing it to 1/4 by an exposure device (stepper, scanner, etc.) not shown. The region of each chip 332 is divided into a plurality of stripe regions 32 with a predetermined width, for example, in the y direction. The scanning operation by the image acquisition mechanism 150 is performed, for example, for each stripe region 32. For example, while moving the stage 105 in the -x direction, the scanning operation of the stripe region 32 is relatively advanced in the x direction. Each stripe region 32 is divided into a plurality of rectangular regions 33 in the longitudinal direction. The movement of the beam to the target rectangular region 33 is performed by collective deflection of the entire multi-primary electron beam 20 by the main deflector 208.

Figure 4 is a diagram for explaining the scanning operation of the multi-beam in the first embodiment. The example of Figure 4 shows the case of a 5 x 5 array multi-primary electron beam 20. The irradiation area 34 that can be irradiated by one irradiation of the multi-primary electron beam 20 is defined as (x-direction size obtained by multiplying the inter-beam pitch in the x direction of the multi-primary electron beam 20 on the substrate 101 surface by the number of beams in the x direction) x (y-direction size obtained by multiplying the inter-beam pitch in the y direction of the multi-primary electron beam 20 on the substrate 101 surface by the number of beams in the y direction). It is preferable to set the width of each stripe area 32 to the same size as the y-direction size of the irradiation area 34, or a size narrower by the scan margin. The example of Figure 4 shows the case where the irradiation area 34 is the same size as the rectangular area 33. However, this is not limited to this. The irradiation area 34 may be smaller than the rectangular area 33. Or it may be larger. Then, each beam of the multi-primary electron beam 20 is irradiated into a sub-irradiation region 29 surrounded by the inter-beam pitch in the x direction and the inter-beam pitch in the y direction where the beam is located, and scans (scans) the sub-irradiation region 29. Each primary electron beam 10 constituting the multi-primary electron beam 20 is assigned to one of the different sub-irradiation regions 29. Then, at the time of each shot, each primary electron beam 10 irradiates the same position in the assigned sub-irradiation region 29. The movement of the primary electron beam 10 in the sub-irradiation region 29 is performed by collective deflection of the entire multi-primary electron beam 20 by the sub-deflector 209. This operation is repeated to sequentially irradiate one sub-irradiation region 29 with one primary electron beam 10. Then, when the scanning of one sub-irradiation region 29 is completed, the irradiation position is moved to an adjacent rectangular region 33 in the same stripe region 32 by collective deflection of the entire multi-primary electron beam 20 by the main deflector 208. This operation is repeated to sequentially irradiate the stripe region 32. When scanning of one stripe region 32 is completed, the irradiation position moves to the next stripe region 32 by moving the stage 105 and/or by collective deflection of the entire multi-primary electron beam 20 by the main deflector 208. As described above, a secondary electron image for each sub-irradiation region 29 is acquired by irradiating each primary electron beam 10. By combining these secondary electron images for each sub-irradiation region 29, a secondary electron image of the rectangular region 33, a secondary electron image of the stripe region 32, or a secondary electron image of the chip 332 is constructed.

As shown in FIG. 4, each sub-irradiation area 29 is divided into a number of rectangular frame areas 30, and a secondary electron image (image to be inspected) in units of frame area 30 is used for inspection. The example in FIG. 4 shows a case where one sub-irradiation area 29 is divided into, for example, four frame areas 30. However, the number of divisions is not limited to four. It may be divided into other numbers.

It is also preferable to group, for example, a plurality of chips 332 arranged in the x direction into the same group, and divide each group into a plurality of stripe regions 32 with a predetermined width in the y direction. Movement between stripe regions 32 is not limited to each chip 332, and can also be performed for each group.

Here, when the multi-primary electron beams 20 are irradiated onto the substrate 101 while the stage 105 is moving continuously, the main deflector 208 performs a tracking operation by collective deflection so that the irradiation position of the multi-primary electron beams 20 follows the movement of the stage 105. Therefore, the emission position of the multi-secondary electron beams 300 changes from moment to moment relative to the central axis of the orbit of the multi-primary electron beams 20. Similarly, when scanning within the sub-irradiation region 29, the emission position of each secondary electron beam changes from moment to moment within the sub-irradiation region 29. The deflector 218 collectively deflects the multi-secondary electron beams 300 so that each secondary electron beam whose emission position has changed in this way is irradiated into the corresponding detection region of the multi-detector 222.

Figure 5 is a flow chart showing the main steps of the inspection method in embodiment 1. In Figure 5, the inspection method in embodiment 1 performs a series of steps including a representative area scanning step (S102), a representative area design image creation step (S104), an initial filter calculation step (S106), a filter setting step (S108), a scanning step (S202), a design image creation step (S204), a reference image creation step (S206), an alignment step (S220), a comparison step (S224), a determination step (S230), a feature extraction step (S232), a search step (S234), and a selection step (S236).

In the representative area scanning step (S102), first, the control computer 110 selects a representative area (representative frame) to be used to find the coefficients of the first filter function (initial filter) when inspecting the substrate 101. As shown in FIG. 4, the stripe area image of each stripe area 32 is divided into frame images 31 of a plurality of frame areas 30, for example, and inspection is performed for each frame image 31. Here, a representative frame area to be used to calculate the coefficients of the filter function is selected from a chip 332 on the substrate 101. The position of the representative frame area to be selected may be arbitrarily set in advance by the user. Then, the position of the selected representative frame area may be set in the control computer 110. Alternatively, a predetermined selection condition may be set, and the control computer 110 may select the position of the representative frame area.

Then, the image acquisition mechanism 150 acquires a secondary electron image of the stripe region 32 that includes the representative frame region of the substrate 101. Specifically, it operates as follows.

First, the stage 105 is moved to a position where the stripe region 32 including the representative frame region can be scanned. The image acquisition mechanism 150 acquires an image of the stripe region 32 including the representative frame region of the substrate 101 on which a graphic pattern is formed. Here, the multi-primary electron beam 20 is irradiated onto the stripe region 32 including the representative frame region of the substrate 101, and the multi-secondary electron beam 300 emitted from the substrate 101 due to the irradiation of the multi-primary electron beam 20 is detected to acquire a secondary electron image of the stripe region 32 including the representative frame region. As described above, the reflected electrons and secondary electrons may be projected onto the multi-detector 222, or the reflected electrons may diverge midway and the remaining secondary electrons (multi-secondary electron beam 300) may be projected.

As described above, the multi-secondary electron beams 300 emitted from the substrate 101 due to irradiation with the multi-primary electron beams 20 are detected by the multi-detector 222. The secondary electron detection data (measurement image data: secondary electron image data: inspected image data) for each pixel in each sub-irradiation area 29 detected by the multi-detector 222 is output to the detection circuit 106 in the order of measurement. In the detection circuit 106, the analog detection data is converted to digital data by an A/D converter (not shown) and stored in the chip pattern memory 123. The obtained measurement image data is then transferred to the filter calculation circuit 134 together with information indicating each position from the position circuit 107. The measurement image data is stored in a storage device (not shown) in the filter calculation circuit 134.

In the representative area design image creation step (S104), first, the expansion circuit 111 (an example of a design image creation unit) creates a representative design image by expanding the design pattern to be placed in the representative frame based on the design pattern data that is the basis for pattern formation on the substrate 101. Specifically, the design pattern data is read from the magnetic disk device 109 through the control computer 110, and each graphic pattern of the representative frame defined in the read design pattern data is converted into binary or multi-value image data (image expansion) to create the representative design image.

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

When the information of the design pattern that becomes such graphic data is input to the expansion circuit 111, it is expanded to data for each graphic, and graphic codes and dimensions indicating the graphic shapes of the graphic data are interpreted. Then, binary or multi-value design image data is expanded as a pattern arranged in a grid with a grid of a predetermined quantization dimension as a unit, and output. In other words, the design data is read, the inspection area is virtually divided into grids with a predetermined dimension as a unit, the occupancy rate of the graphic in the design pattern is calculated for each grid, and n-bit occupancy data is output. For example, it is preferable to set one grid as one pixel. Then, if one pixel has a resolution of 1/2 ⁸ (=1/256), a small area of 1/256 is allocated for the area of the graphic arranged in the pixel, and the occupancy rate in the pixel is calculated. Then, a representative design image of 8-bit occupancy data is created for each pixel. The data of the representative design image is output to the filter calculation circuit 134. The data of the representative design image is stored in a storage device (not shown) in the filter calculation circuit 134.

As an initial filter calculation step (S106), the filter calculation circuit 134 calculates a filter function (initial filter) or the coefficients of the filter function for filtering the representative design image based on the measurement image (representative frame image) of the representative frame area of the substrate 101 and the representative design image.

Figure 6 is a diagram for explaining an example of a method for calculating a filter function in the first embodiment. For example, as shown in Figure 6 (a), an unknown coefficient matrix a (i, j) (one example of coefficients) consisting of k x k elements, which is less than the number of pixels in the frame area 30, is obtained. For example, a 15 x 15 coefficient matrix a (i, j) is obtained for an image of the frame area 30 consisting of 512 x 512 pixels. The sum of the products of the pixels of k x k pixels and the coefficient matrix a (i, j) centered on the target pixel d (i, i) of the representative design image and divided by the number of pixels N (= k x k) is used to obtain a coefficient matrix a (i, j) that is closest to the target pixel r (i, i) of the representative frame image corresponding to the target pixel d (i, i). Such a relational expression (1) is shown below.

As shown in FIG. 6(b), while the pixel of interest is moved within the representative frame 32, the relational expression (1) is calculated each time. Then, the coefficient matrix a(i,j) that best satisfies the relational expression (1) defined using the unknown coefficient matrix a(i,j) obtained for each of all pixels within the representative frame 32 is obtained. The number of elements k×k of the coefficient matrix a(i,j) may be set appropriately. If it is too small, the accuracy will deteriorate, and if it is too large, the calculation time will be long. Also, when the pixel of interest moves within the representative frame 32, if it is close to an end, there may not be enough surrounding pixels on the end side. In such a case, the calculation may be performed using the surrounding pixels and the number of pixels N from which values can be obtained.

In the filter setting step (S108), the coefficient matrix a(i,j) (an example of an initial filter) obtained by the filter calculation circuit 134 is output to the reference image creation circuit 112 as an initial filter, and is set in the reference image creation circuit 112 as a filter function or coefficients of the filter function.

Here, as described above, the image quality of the measurement image (SEM image) obtained by scanning the substrate 101 with an electron beam is expected to change little by little as time passes from the time when the scanning operation for the substrate 101 is started. In other words, the image quality is expected to change between the measurement image obtained by scanning earlier and the measurement image obtained by scanning after time has passed. For example, the influence of charging of the substrate 101 can be mentioned. Therefore, in a die-database inspection in which a measurement image and a reference image, both of which are defined in grayscale, are compared, if the initial filter obtained in the initial filter calculation step (S106) is continued to be used to create a reference image, there is a problem that the difference between the measurement image and the reference image becomes larger over time, and pseudo defects tend to occur frequently. Therefore, it is required to create a reference image so as to follow the change in the image quality of the measurement image. Therefore, in the first embodiment, the filter function is switched midway as the inspection progresses. At that time, instead of performing the learning process from the representative area scanning step (S102) to the initial filter calculation step (S106) again to create a new filter function, many filter functions with a proven track record are accumulated and selected from them for use. In the first embodiment, the characteristics of an image to be filtered by a filter function are extracted and classified, and a database 130 is constructed that can search for filter functions based on the characteristics. The specific operation is described below.

Figure 7 is a block diagram showing an example of the internal configuration of the filter update circuit in the first embodiment. In Figure 7, the filter update circuit 132 includes a determination unit 60, a reference image creation unit 61, a feature extraction unit 62, a table creation unit 63, a search unit 64, a reference image candidate creation unit 66, a sum of squared differences (SSD) calculation unit 68, a selection unit 69, and storage devices 65, 67, 70, 71, and 72 such as magnetic disk devices. Each "~ unit" such as the determination unit 60, the reference image creation unit 61, the feature extraction unit 62, the table creation unit 63, the search unit 64, the reference image candidate creation unit 66, the SSD calculation unit 68, and the selection unit 69 includes a processing circuit, and this processing circuit includes an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. In addition, each "~ unit" may use a common processing circuit (the same processing circuit). Alternatively, different processing circuits (separate processing circuits) may be used. The input data or calculation results required for the determination unit 60, reference image creation unit 61, feature extraction unit 62, table creation unit 63, search unit 64, reference image candidate creation unit 66, SSD calculation unit 68, and selection unit 69 are stored in a memory (not shown) or memory 118 each time.

Prior to the inspection process, a number of filter functions that have been used in the past and have a proven track record are stored, and a specific feature is extracted from each filter function. The stored filter functions are preferably those that have been used in the past to inspect another substrate to be inspected by the same inspection device 100, but are not limited to this. Filter functions that have been used in the past to inspect another substrate to be inspected by another inspection device may also be included. The features of each filter function are extracted from a filtered image (reference image) obtained by applying each filter function to a reference design image of a region of a predetermined size in which a reference pattern is arranged. A specific description will be given below. Note that for filter functions used in another inspection device, the features of the filter function may be extracted by the other inspection device. Alternatively, the features of each filter function of a number of filter functions that have been used in the past and have a proven track record may be extracted offline and input to the inspection device 100.

Figure 8 is a diagram showing an example of a reference design image in embodiment 1. In Figure 8, a reference pattern 14 having a size of 5 x 5 pixels is placed in the center of a 15 x 15 pixel area in the reference design image 12. In the example of Figure 8, a gradation value of 255 is defined for the 5 x 5 pixels with a pattern. A gradation value of 0 is defined for the remaining pixels without a pattern.

The reference image creation unit 61 filters the reference design image 12 with each stored filter function to create a reference image for each filter function. Next, the feature extraction unit 62 extracts features from the reference image for each filter function obtained after the filter processing.

9 is a diagram showing an example of a reference image after filtering in the first embodiment. The feature extraction unit 62 extracts the maximum gradation value Max and the minimum gradation value Min of the reference pattern after filtering, which are indexes indicating the brightness of the pattern, as an example of a feature for the cross-sectional profile in the x direction on the center line of the reference pattern in the reference image and the cross-sectional profile in the y direction, respectively. In addition, the feature extraction unit 62 extracts the distances dx1 and dx2 between the middle gradation position (h/2) of the reference pattern after filtering, which are indexes indicating the center position of the pattern, and the reference position outside the pattern, as another example of a feature for the cross-sectional profile in the x direction on the center line of the reference pattern in the reference image. One end of the reference image in the x direction is set as reference point 1, and the other end is set as reference point 2, and these two reference points 1 and 2 are used as reference positions outside the pattern in the x direction. Similarly, for the cross-sectional profile in the y direction, the distances dy1 and dy2 between the middle gradation position (h/2) of the reference pattern after filtering, which serves as an index of the center position of the pattern, and a reference position outside the pattern are extracted as another example of a feature. One end of the reference image in the y direction is set as reference point 1, and the other end is set as reference point 2, and these two reference points 1 and 2 are used as reference positions outside the pattern in the y direction. In addition, for the cross-sectional profile in the x direction on the center line of the reference pattern in the reference image, the feature extraction unit 62 extracts the area Sx of the reference pattern after filtering as another example of a feature. Similarly, for the cross-sectional profile in the y direction, the area Sy of the reference pattern after filtering is extracted as another example of a feature.

Figure 10 is a diagram showing an example of the characteristics of a reference image in embodiment 1. In the example of Figure 10, the gradation values constituting the cross-sectional profile in the x direction on the center line of the reference pattern are shown. In the example of Figure 10, the maximum gradation value Max is 180 and the minimum gradation value Min is 20. Therefore, since the intermediate gradation is 80 (= (180-20)/2), it can be seen that the intermediate gradation position is located within a pixel showing a gradation value of 80. Between reference point 1 and the pixel (gradation value 80) containing the intermediate gradation position on the left side of the pattern, there are four pixels with gradation values of 20, 20, 30, and 50 defined. The distance for one pixel is defined as 1, and the distance of the sub-pixel unit within a pixel is defined as (1-V/256). Therefore, the distance dx1 between reference point 1 and the left intermediate gradation position of the pattern can be calculated as dx1 = 4 + (1-80/256) = 4.6875. Between reference point 2 and the pixel (tone value 80) containing the intermediate tone position on the right side of the pattern, there are four pixels defined with tone values of 20, 20, 30, and 50. Therefore, dx2 = dx1 = 4.6875. It is also preferable to define the area Sx as the sum of the tone values of the cross-sectional profile in the x direction. In the example of Figure 10, Sx = 1060. The same can be done when extracting each feature from the cross-sectional profile in the y direction.

Figure 11 is a diagram showing another example of the characteristics of the reference image in the first embodiment. In the example of Figure 10 described above, since there is a pixel defined as intermediate gradation 80, it can be seen that the intermediate gradation position is located within a pixel of intermediate gradation 80. However, as shown in the example of Figure 11, for example, when the maximum gradation value Max is 200, the minimum gradation value Min is 20, and the intermediate gradation is 90 (= (200-20) / 2), there is no pixel of gradation value 90. In this case, when there is no pixel of the same gradation, the distance dx1 can be found by linear interpolation. In the example of Figure 11, two points are interpolated: point A in a pixel of continuous gradation value 80 close to intermediate gradation 90 and point B in a pixel of gradation value 100. The coordinates (x, y) of each point are defined as (distance dx1, gradation value V). In this case, at point A, x = dx1 = 4.6875 as described above, and y = gradation value V = 80. Therefore, the coordinates of point A are (4.6875, 80). By performing the same calculation for point B as for point A, the coordinates of point B (5.6093, 100) can be obtained. Point X, which is a mid-tone 90, exists on the line connecting points A and B. Therefore, by using the coordinate values of points A and B to find the coefficients a and b of the equation of this line y = ax + b, we get a = 21.6967 and b = -21.7033. Therefore, the distance dx1 of point X can be found to be 5.1484 (= (90 + 21.7033) / 21.6967).

Next, the table creation unit 63 uses the characteristics of each filter function extracted in the above manner to create a hierarchical, searchable hierarchical table to be used in the database 130.

FIG. 12 is a diagram showing an example of a hierarchical table in the first embodiment. In FIG. 12, the hierarchical table 11 is composed of a first layer in which combinations M1, M2, M3, ... of maximum gradation Max and minimum gradation Min indicating the brightness of the image are defined from the top layer side, a second layer in which combinations D1, D2, D3, ... of distances dx1, dx2, dy1, dy2 indicating the center position are defined, a third layer in which combinations S1, S2, S3, ... of areas Sx and Sy indicating the area are defined, and a fourth layer in which actual filter functions F1, F2, F3, ... are defined. The tables of the second layer are defined dependently on each combination M of the first layer. Similarly, the tables of the third layer are defined dependently on each combination D of the second layer. Similarly, the tables of the fourth layer are defined dependently on each combination S of the third layer. The tables of the first layer are, for example, arranged in descending order from the largest maximum gradation Max. The tables of the second layer are, for example, arranged in descending order from the largest distance dx1. The third layer table is arranged, for example, in descending order of area Sx. The fourth layer table can be any type.

The database 130 is constructed using the created hierarchical table 11. As a result, the database 130 stores information on multiple filter functions (first image processing filters) that can be searched using feature information indicating the features.

In the above example, the maximum gradation Max, minimum gradation Min, distances dx1, dx2, dy1, dy2, and areas Sx and Sy are used as features, but there is no need to use all of these. It is also possible to use only the maximum gradation Max and minimum gradation Min as features. Or, it is also possible to use only the distances dx1, dx2, dy1, dy2 as features. Or, it is also possible to use only the areas Sx and Sy as features.

After the above pre-inspection processing is completed, inspection of the substrate 101 to be inspected begins.

In the scanning step (S202), the image acquisition mechanism 150 acquires multiple electron optical images (secondary electron images) by scanning the substrate 101 on which multiple graphic patterns are formed with an electron beam. Here, the substrate 101 on which multiple graphic patterns are formed is irradiated with a multi-primary electron beam 20, and a multi-secondary electron beam 300 emitted from the substrate 101 due to the irradiation of the multi-primary electron beam 20 is detected, thereby acquiring a secondary electron image of the substrate 101. As described above, the reflected electrons and secondary electrons may be projected onto the multi-detector 222, or the reflected electrons may diverge midway and the remaining secondary electrons (multi-secondary electron beam 300) may be projected.

As described above, the multi-secondary electron beams 300 emitted from the substrate 101 due to irradiation with the multi-primary electron beams 20 are detected by the multi-detector 222. The secondary electron detection data (measurement image data: secondary electron image data: inspected image data: electron optical image data) for each pixel in each sub-irradiation area 29 detected by the multi-detector 222 is output to the detection circuit 106 in the order of measurement. In the detection circuit 106, the analog detection data is converted to digital data by an A/D converter (not shown) and stored in the chip pattern memory 123. The obtained measurement image data is then transferred to the comparison circuit 108 together with information indicating each position from the position circuit 107.

Figure 13 is a configuration diagram showing an example of the configuration in the comparison circuit in the first embodiment. In Figure 13, the comparison circuit 108 includes storage devices 52 and 56 such as magnetic disk devices, an alignment unit 57, and a comparison unit 58. Each "~ unit" such as the alignment unit 57 and the comparison unit 58 includes a processing circuit, and this processing circuit includes an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. In addition, each "~ unit" may use a common processing circuit (the same processing circuit). Alternatively, different processing circuits (separate processing circuits) may be used. The input data or the calculated results required in the alignment unit 57 and the comparison unit 58 are stored in a memory (not shown) or memory 118 each time.

The comparison circuit 108 compares the inspection image, which is composed of at least a part of the secondary electron image for each sub-irradiation area 29, with the reference image. For example, the secondary electron image (electron optical image) for each frame area 30 is used as the inspection image. For example, the sub-irradiation area 29 is divided into four frame areas 30. For example, an area of 512 x 512 pixels is used as the frame area 30. Specifically, for example, it operates as follows.

In the design image creation step (S204), the unfolding circuit 111 reads out the design pattern data from the storage device 109 through the control computer 110, and converts each figure pattern defined in the read out design pattern data into binary or multi-value image data. It is preferable that the unfolding circuit 111 creates a design image of the corresponding frame area 30 in accordance with the acquisition order of the measurement images acquired in the scanning step.

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

When the design pattern data that becomes such figure data is input to the expansion circuit 111, it is expanded to data for each figure, and the figure code and figure dimensions indicating the figure shape of the figure data are interpreted. Then, it is expanded into binary or multi-value design image data as a pattern arranged in a grid with a grid of a predetermined quantization dimension as a unit, and output. In other words, the design data is read, the inspection area is virtually divided into grids with a predetermined dimension as a unit, the occupancy rate of the figure in the design pattern is calculated for each grid, and n-bit occupancy data is output. For example, it is preferable to set one grid as one pixel. Then, if one pixel has a resolution of 1/2 ⁸ (=1/256), a small area of 1/256 is assigned to the area of the figure arranged in the pixel, and the occupancy rate in the pixel is calculated. Then, it is created as 8-bit occupancy data. Such grids (inspection pixels) can be aligned with the pixels of the measurement data.

In the reference image creation step (S206), the reference image creation circuit 112 (reference image creation unit) performs filter processing on the design image corresponding to the target electron optical image for each of the multiple electron optical images using a predetermined image processing filter, thereby creating a reference image corresponding to the target electron optical image. Specifically, it operates as follows. The reference image creation circuit 112 performs filter processing on the design image data of the design pattern, which is the image data of the figure, using the filter function F set in the filter setting step (S108). This allows the design image data, which is the design side image data with the image intensity (grayscale value) being a digital value, to be matched to the image generation characteristics obtained by irradiation with the multi-primary electron beams 20. The image data of the created reference image is output to the comparison circuit 108. During the period from the first design image created for the first time after the start of the inspection process until the predetermined condition is satisfied, the initial filter is applied to the design image of each frame area 30.

The measurement image (inspection image) input into the comparison circuit 108 is stored in the storage device 56. The reference image input into the comparison circuit 108 is stored in the storage device 52.

In the alignment step (S220), the alignment unit 57 reads out the corresponding frame image 31 (electron optical image) and the reference image for each frame region 30 from the storage device, and aligns the two images in sub-pixel units smaller than a pixel. For example, the alignment can be performed using the least squares method. It is preferable to set the pixel size to an area of approximately the same size as each beam size of the multiple primary electron beams 20.

In the comparison step (S224), the comparison unit 58 compares the target frame image 31 (electro-optical image) with a reference image corresponding to the target frame image 31 for each frame region 30. The comparison unit 58 compares the two for each pixel according to a predetermined judgment condition, and judges the presence or absence of a defect such as a shape defect. For example, if the gradation value difference for each pixel is greater than the judgment threshold value Th, it is judged to be a defect. The comparison result is then output. The comparison result may be output to the storage device 109, monitor 117, or memory 118, or may be output from a printer not shown.

In the above example, pixel comparison is performed, but this is not the only option. For example, the contours of the graphic patterns in the frame image 31 and the reference image may be extracted, and a defect may be determined if the distance between the contours is greater than a determination threshold.

In the judgment step (S230), the judgment unit 60 judges whether a predetermined condition is satisfied after the filter function is set. For example, it judges whether a predetermined time has elapsed after the start of the scan performed after the filter function is set (or updated). Alternatively, it judges whether a preset number of stripe regions have been scanned after the filter function is set (or updated). Alternatively, it judges whether defects equal to or greater than the threshold value are detected in the same frame region 30 (or stripe region 32) in the comparison step after the filter function is set (or updated). In either case, it corresponds to the stage where a comparison process of some of the frame images 31 out of the multiple frame images 31 is performed. While the condition is not yet satisfied, the currently set filter function is continued to be used, and the reference image creation step (S206) is performed, and each step up to the comparison step (S224) is performed. If it is judged that the predetermined condition is satisfied, the process proceeds to the feature extraction step (S232). Also, if it is judged that the predetermined condition is satisfied, the creation of the subsequent reference images is temporarily halted.

In the feature extraction step (S232), first, the reference image creation unit 61 reads out the currently set filter function (e.g., an initial filter) and filters the reference design image 12 with the read out filter function to create a reference image using the filter function. Next, the feature extraction unit 62 extracts features from the reference image obtained after filtering. Specifically, the above-mentioned maximum gradation Max, minimum gradation Min, distances dx1, dx2, dy1, dy2, and areas Sx and Sy are extracted. The features of the reference image after filtering using the currently set filter function are indicated as maximum gradation Max', minimum gradation Min', distances dx1', dx2', dy1', dy2', and areas Sx' and Sy'.

In the above example, the feature extraction step (S232) is performed after the determination step (S230), but this is not limited to the above. When a predetermined filter function is set by the filter setting step (S108), it is also preferable to perform the feature extraction step (S232) without waiting for other steps to be performed. By performing the feature extraction step (S232) before the determination step (S230), the search step (S234) can be performed after the determination step (S230), thereby shortening the processing time.

In the search step (S234), when a comparison process of some of the frame images 31 has been performed, in other words, when it is determined that a predetermined condition is satisfied in the determination step (S230), the search unit 64 uses the database 130 to search for a plurality of filter functions (first image processing filters) having characteristics similar to (or close to) the characteristics of the currently used filter function (image processing filter). Specifically, the search unit 64 inputs the maximum gradation Max', minimum gradation Min', distances dx1', dx2', dy1', dy2', and areas Sx' and Sy', which are characteristics extracted from the reference image after filtering by the currently set filter function, into the database 130.

Figure 14 is a diagram for explaining similar filter functions in embodiment 1. In the example of Figure 14, for example, a cross-sectional profile in the x direction of a reference image is shown. In Figure 14, the maximum gradation Max, minimum gradation Min, and distances dx1 and dx2 are almost the same, but the rise and fall of the profile are slightly different. This difference is expressed as a difference in area Sx. It is preferable to use the sum of squared differences as an index showing such a difference. The same applies to the case of showing the difference between the maximum gradation Max, minimum gradation Min, and distances dx1 and dx2.

Figure 15 is a diagram for explaining the method of searching in the first embodiment. The database 130 searches from the top of the hierarchical table 11 in order. Specifically, the database 130 calculates the following formula (2) showing the sum of squared differences for the brightness of the first layer, using the maximum gradation Max and minimum gradation Min of each set M and the maximum gradation Max' and minimum gradation Min' of the reference image after filtering with the currently set filter function.

Then, extract the combination M that minimizes ΔM.

Next, for the center position of the second layer subordinate to the extracted combination M, the following formula (3) showing the sum of squared differences is calculated using the distances dx1, dx2, dy1, and dy2 of each set D and the distances dx1', dx2', dy1', and dy2' of the reference image after filtering using the currently set filter function.

Then, extract the combination D that minimizes ΔD.

Next, for the area of the third layer that is dependent on the extracted combination D, the following formula (4) is calculated, which shows the sum of squared differences, using the areas Sx, Sy of each set S and the areas Sx', Sy' of the reference image after filtering using the currently set filter function.

Then, extract the combination M that minimizes ΔS.

Then, the filter functions of the fourth layer that are dependent on the extracted combination S are extracted. Here, if the number of filter functions finally obtained is less than the lower threshold, for example, m combinations S with smaller ΔS are extracted, and filter functions of the fourth layer that are dependent on each combination S are extracted. Then, N filters are extracted from the filter functions that have smaller ΔS. Alternatively, a step amount may be set for the extraction of N filters. For example, if the step amount is 3, the first filter function, the fourth filter function, the seventh filter function, ..., that have smaller ΔS are extracted. Here, if the number of filter functions finally obtained when the filter functions of the fourth layer that are dependent on all the combinations S are extracted is less than the lower threshold, for example, m combinations D with smaller ΔD are extracted, and filter functions of the fourth layer that are dependent on each combination D are extracted. Through this process, a predetermined number of filter functions (second image processing filters) having characteristics similar to the characteristics of the filter function (image processing filter) currently being used are extracted from the multiple filter functions (first image processing filters) stored in the database 130. If the number of filter functions in the fourth layer that are dependent on the extracted combination S exceeds an upper threshold, the number of filter functions may be limited to the upper threshold, for example. This can prevent the processing time in the process described below from becoming too long.

Here, just because the features are similar (or have similar features) does not necessarily mean that they match the changed image quality state of the measured image at the time of the search. Therefore, in the first embodiment, a filter function that matches the current state is selected from the multiple searched filter functions.

In the selection step (S236), first, the reference image candidate creation unit 66 creates multiple reference image candidates by filtering the design image to be used for the actual inspection with the multiple filter functions found that have similar (or close) features in the reference image. The design image used here is preferably, for example, the design image used for the reference image created at the same time that it was determined in the determination step (S230) that the specified conditions were satisfied. For example, it is preferably the design image used for the last reference image before pausing.

Next, the SSD calculation unit 68 calculates the sum of squared differences between each of the obtained multiple reference image candidates and the corresponding frame image 31 (measurement image). It also calculates the sum of squared differences between the reference image filtered with the filter function currently used for the design image used to create the reference image candidate and the corresponding frame image 31 (measurement image).

Then, the selection unit 69 selects a filter function (second image processing filter) that creates a reference image that matches more closely with the reference image created using the filter function currently being used for comparison with the remaining frame images 31 of the plurality of frame images 31, from among a plurality of filter functions (second image processing filters) that have similar (or close) features in the reference image. Here, the filter function used to create the reference image candidate that has the smallest sum of squared differences is selected.

Then, the process returns to the filter setting step (S108), and the selected filter function is output to the reference image creation circuit 112 and set in the reference image creation circuit 112 as the filter function or the coefficient of the filter function. This updates the filter function.

After the filter function has been updated in the above manner, the reference image creation circuit 112 uses the selected filter function to perform filter processing on the design image corresponding to the target frame image 31 among the remaining frame images 31 for which comparison processing has not yet been performed, thereby creating a reference image corresponding to the target frame image 31. For example, if comparison processing has been performed up to the kth frame image, filter processing can be performed on the design images corresponding to the (k+1)th and subsequent frame images.

Then, for the (k+1)th and subsequent frame regions 30 for which comparison processing has not yet been completed, an alignment process (S220) is performed between the reference image created with the updated filter function and the measurement image of the corresponding frame region 30, and a comparison process (S224) is performed.

In addition, when inspecting one substrate 101, the filter function does not necessarily need to be updated just once. It is preferable to perform multiple updates in the same procedure.

As described above, according to the first embodiment, it is possible to quantitatively and efficiently search for a better filter function from the so-called neighborhood of the initial filter. Therefore, it is possible to create a reference image that follows the change in image quality of the measurement image obtained by scanning the substrate 101 with the electron beam.

Embodiment 2.
In the first embodiment, a case was described in which one reference image is selected from a plurality of reference image candidates using a plurality of filter functions having similar characteristics in the reference image, but in the second embodiment, a configuration is described for creating a reference image that can be more accurately adapted to a changed image quality state than in the first embodiment. The configuration of the inspection device 100 is the same as in Fig. 1. Also, the flowchart of the inspection method is the same as in Fig. 5. Contents other than those specifically described below may be the same as in the first embodiment.

The selection process (S236) is similar in that multiple reference image candidates are created, and for each reference image candidate, the sum of squared differences is calculated between the corresponding frame image 31 (measurement image).

Figure 16 is a diagram showing an example of multiple reference image candidates and a composite reference image in embodiment 2. The selection unit 69 in embodiment 2 selects two or more filter functions (second image processing filters) that create a reference image with a high degree of match from multiple filter functions (second image processing filters) searched from the database 130. For example, the filter function used to create reference image candidate 1 with the smallest sum of squared differences and the filter function used to create reference image candidate 1 with the second smallest sum of squared differences are selected. In the example of Figure 16, two reference image candidates 1 and 2 are shown. The two reference image candidates 1 and 2 show a case in which the pixel in the middle left and the pixel in the bottom right are different.

Then, the process returns to the filter setting step (S108), and the two or more selected filter functions are output to the reference image creation circuit 112 and set in the reference image creation circuit 112 as the filter functions or the coefficients of the filter functions.

After the filter function is updated as described above, the reference image creation circuit 112 uses the two or more selected filter functions (second image processing filters) to perform filter processing on the design image corresponding to the target frame image 31 among the remaining frame images 31 for which comparison processing has not yet been performed, thereby creating two or more reference image candidates corresponding to the target frame image 31. Then, the reference image creation circuit 112 creates a reference image by combining two or more reference image candidates. In the example of FIG. 16, of the different pixel values of the two reference image candidates 1 and 2, the pixel value of reference image candidate 1 is selected for the pixel in the middle left row, and the pixel value of reference image candidate 2 is selected for the pixel in the lower right row, and a reference image is created by combining both. As for which reference image candidate pixel value to use, the one that matches the frame image 31 more closely is selected. For example, when there is no filter function suitable for both the x and y directions, it is particularly effective because it is possible to combine the filter processing results using two filter functions suitable for each direction.

Here, in the reference image creation step (S206), if it is difficult to perform filtering with two filter functions at the same time, when creating the (k+1)th reference image, for example, the design image is filtered with the filter function used to create reference image candidate 1, and when creating the subsequent (k+2)th reference image, for example, the same design image is filtered with the filter function used to create reference image candidate 2. Then, the reference image creation circuit 112 combines the (k+1)th reference image candidate and the (k+2)th reference image candidate.

Then, for the (k+1)th and subsequent frame regions 30 for which comparison processing has not yet been completed, an alignment process (S220) is performed between the reference image created with the updated filter function and the measurement image of the corresponding frame region 30, and a comparison process (S224) is performed.

As described above, by extracting pixel values that are close to the pixel values of the target frame image 31 from two or more reference image candidates and synthesizing them, it is possible to obtain a reference image that is closer to the target frame image than in embodiment 1.

Embodiment 3.
In the third embodiment, a configuration will be described in which further features are added in addition to the features used in the first and second embodiments. The configuration of the inspection device 100 is the same as that shown in Fig. 1. Also, the flow chart of the inspection method is the same as that shown in Fig. 5. The contents other than the points specifically described below may be the same as those of the first or second embodiment.

Figure 17 is a block diagram showing an example of the internal configuration of the filter update circuit in the third embodiment. In Figure 17, the filter update circuit 132 is similar to Figure 7 except that a rank-by-rank figure number measurement unit 74 is further arranged in the filter update circuit 132. Each "~ unit" such as the judgment unit 60, the reference image creation unit 61, the feature extraction unit 62, the rank-by-rank figure number measurement unit 74, the table creation unit 63, the search unit 64, the reference image candidate creation unit 66, the SSD calculation unit 68, and the selection unit 69 includes a processing circuit, and this processing circuit includes an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. In addition, each "~ unit" may use a common processing circuit (the same processing circuit). Alternatively, different processing circuits (separate processing circuits) may be used. The input data or calculation results required for the judgment unit 60, reference image creation unit 61, feature extraction unit 62, rank-by-rank graphic number measurement unit 74, table creation unit 63, search unit 64, reference image candidate creation unit 66, SSD calculation unit 68, and selection unit 69 are stored in a memory (not shown) or memory 118 each time.

In the third embodiment, the rank of the micro-graphics in the design image used to obtain the filter function and the number of micro-graphics of that rank are used as the features of each of the multiple filter functions that have been used in the past. Therefore, for the multiple filter functions that have been used in the past, the rank-by-rank graphic number measurement unit 74 measures the micro-graphics of the ranks described below and their numbers from the design images used when the filter functions were calculated in the learning process such as the representative area scanning process (S102) to the initial filter calculation process (S106). If the filter functions that have been used in the past have been used in another inspection device, it is sufficient to measure them in the other inspection device.

Figure 18 is a diagram for explaining the micrographic rank in the third embodiment. In Figure 18, rank 1 indicates a pixel group in which three pixels aligned in the x direction or y direction are "0, V, 0" or "V, 0, V". The gradation value V is > 0. In other words, a pixel group indicating the presence of a pattern and a pixel group indicating the absence of a pattern on both sides is a micrographic of rank 1. Rank 2 indicates a pixel group in which four pixels aligned in the x direction or y direction are "0, V, V, 0" or "V, 0, 0, V". In other words, a pixel group indicating the presence of a pattern and a pixel group indicating the absence of a pattern on both sides is a micrographic of rank 2. Rank 3 indicates a pixel group in which five pixels aligned in the x direction or y direction are "0, V, V, V, 0" or "V, 0, 0, 0, V". In other words, a pixel group indicating the presence of a pattern and a pixel group indicating the absence of a pattern on both sides is a micrographic of rank 3. The ranks are defined in the same manner up to a certain rank. The smaller the rank value, the smaller the pattern.

Next, the table creation unit 63 creates a hierarchical, searchable hierarchical table for use in the database 130, which adds a hierarchy of micro-figure rank features in addition to the features from the first to third layers described above.

Figure 19 is a diagram showing an example of a hierarchical table in the third embodiment. In Figure 19, the hierarchical table 11 is composed of the first to third layers shown in Figure 12, the fourth layer, which is the next lower layer, in which combinations r1, r2, r3, ... of micro-figure ranks and the number of each rank are defined, and the fifth layer, in which actual filter functions F1, F2, F3, ... are defined. The tables in the fourth layer are defined depending on each combination S in the third layer. The tables in the first layer are, for example, arranged in descending order from the largest maximum gradation Max. The tables in the second layer are, for example, arranged in descending order from the largest distance dx1. The tables in the third layer are, for example, arranged in descending order from the largest area Sx. The tables in the fourth layer are, for example, arranged in descending order from the largest number of top ranks. The tables in the fifth layer may be arbitrary.

The database 130 is constructed using the created hierarchical table 11. As a result, the database 130 stores information on multiple filter functions (first image processing filters) that can be searched using feature information indicating the features.

As the feature extraction step (S232), the reference image creation unit 61 first creates a reference image using the currently set filter function (e.g., an initial filter), and the feature extraction unit 62 extracts the maximum gradation Max', minimum gradation Min', distances dx1', dx2', dy1', dy2', and areas Sx', Sy' as features from the reference image obtained after the filter process. In addition, the number of figures per rank measurement unit 74 measures the micro-figure rank and the number of each rank from the design image used when the initial filter was calculated in the learning process from the representative area scanning step (S102) to the initial filter calculation step (S106) described above.

In the search step (S234), if it is determined in the determination step (S230) that the predetermined conditions are satisfied, the search unit 64 uses the database 130 to search for a plurality of filter functions (first image processing filters) having characteristics similar to (or close to) the characteristics of the currently used filter function (image processing filter). Specifically, the search unit 64 inputs the maximum gradation Max', minimum gradation Min', distances dx1', dx2', dy1', dy2', areas Sx', Sy', and the micro rank and the number of them, which are characteristics extracted from the reference image after filtering by the currently set filter function, into the database 130.

Then, the database 130 extracts combinations starting from the top layer and extracts combination M that has the smallest ΔS for the third layer.

Then, from among the combinations r1, r2, r3, ... in the fourth layer that are subordinate to the extracted combination S, the combination r with the greatest number of rank 1s is extracted with priority. If rank 1 does not exist, the combination r with the greatest number of the next highest ranks is extracted with priority.

Alternatively, when the comparison process for the kth reference image has been completed, it is also preferable to preferentially extract the combination r that has the closest number of top micro ranks present in the design image for creating the next k+1th reference image.

Then, extract the fifth layer filter function that depends on the extracted combination r.

By using the micro-graphic rank, it becomes possible to search for filter functions stored in database 130 that are most suitable for micro-patterns that are difficult to detect defects.

As the amount of data stored in database 130 increases, there are cases where the combination of areas Sx and Sy for two different filters is the same. The same thing can happen between two or more filters. Therefore, by adding the micro-figure rank as one of the search elements, it is possible to distinguish between filters even when the aforementioned combination of areas is the same.

The micro-figure rank is an index that takes the same design image as the SEM image used to calculate the filter function as input, counts the number of patterns with relatively small dimensions among the patterns present in the image, and ranks them taking into account the size of the patterns. Since the pattern shape and number of patterns in the SEM image differ depending on the sample being inspected, the number of counted micro-figures is unlikely to be duplicated.

In general, when using inspection equipment to check for defects in samples, the smaller the dimensions of a micropattern, the more difficult it is to detect defects. One reason for this is that it is difficult to create a reference image that resembles an SEM image, which results in a large difference in the gradation values of each pixel between the two images, resulting in frequent occurrence of so-called false defects and the burying of signals indicating defects. The reason it is difficult to generate a reference image is that the smaller the pattern, the fewer the total number of pixels required to represent the pattern, i.e., the less information there is. With less information, for example, it becomes difficult to predict the pattern shape and position.

In addition, to create a reference image suitable for a micropattern, it is essential to select a filter suitable for the micropattern. To calculate the filter, it is desirable that the SEM image input for learning contains a large number of micropatterns, in other words, that the amount of information on the micropatterns is large. By using the micrographic rank, it is possible to search for a filter from the database, taking into account the amount of information on the micropatterns at the time of learning.

The contents of each subsequent step are the same as those in embodiment 1 or 2.

As described above, according to the third embodiment, it is possible to create a reference image that matches the measurement image for smaller figures.

In the above description, the series of "circuits" includes a processing circuit, which may include an electric circuit, a computer, a processor, a circuit board, a quantum circuit, or a semiconductor device. Each "circuit" may use a common processing circuit (the same processing circuit). Alternatively, different processing circuits (separate processing circuits) may be used. The program for executing the processor may be recorded on a recording medium such as a magnetic disk device, a magnetic tape device, a FD, or a ROM (read-only memory). For example, the position circuit 107, the comparison circuit 108, the expansion circuit 111, the reference image creation circuit 112, the stage control circuit 114, the lens control circuit 124, the blanking control circuit 126, the deflection control circuit 128, the filter update circuit 132, and the filter calculation circuit 134 may be composed of at least one of the processing circuits described above.

The above describes the embodiment with reference to specific examples. However, the present invention is not limited to these specific examples. In the example of FIG. 1, a case is shown in which multiple primary electron beams 20 are formed by a shaping aperture array substrate 203 from one beam irradiated from an electron gun 201, which serves as one irradiation source, but the present invention is not limited to this. Multiple primary electron beams 20 may be formed by irradiating primary electron beams from multiple irradiation sources.

In addition, although descriptions of device configurations, control methods, and other aspects that are not directly necessary for explaining the present invention have been omitted, the required device configurations and control methods can be appropriately selected and used.

All other electron beam inspection devices and electron beam inspection methods that incorporate the elements of the present invention and that can be modified as appropriate by a person skilled in the art are included within the scope of the present invention.

10 Primary electron beam 20 Multi-primary electron beam 22 Hole 29 Sub-irradiation area 30 Frame area 31 Frame image 32 Stripe area 33 Rectangular area 34 Irradiation area 52, 56 Storage device 57 Alignment section 58 Comparison section 60 Judgment section 61 Reference image creation section 62 Feature extraction section 63 Table creation section 64 Search section 65, 67, 70, 71, 72 Storage device 66 Reference image candidate creation section 68 SSD calculation section 69 Selection section 74 Rank-by-rank figure number measurement section 100 Inspection device 101 Substrate 102 Electron beam column 103 Inspection chamber 105 Stage 106 Detection circuit 107 Position circuit 108 Comparison circuit 109 Storage device 110 Control computer 112 Reference contour position extraction circuit 114 Stage control circuit 117 Monitor 118 Memory 120 Bus 122 Laser length measurement system 123 Chip pattern memory 124 Lens control circuit 126 Blanking control circuit 128 Deflection control circuit 130 Database 132 Filter update circuit 134 Filter calculation circuit 142 Driving mechanism 144, 146, 148 DAC amplifier 150 Image acquisition mechanism 160 Control system circuit 201 Electron gun 202 Electromagnetic lens 203 Shaping aperture array substrate 205, 206, 207, 224, 226 Electromagnetic lens 208 Main deflector 209 Sub-deflector 212 Collective blanking deflector 213 Limiting aperture substrate 214 Beam separator 216 Mirror 218 Deflector 222 Multi-detector 300 Multi-secondary electron beam 330 Inspection area 332 Chip

Claims (12)

a reference image creation unit that creates a reference image corresponding to the target electron optical image by performing a filter process on a design image corresponding to the target electron optical image for each of a plurality of electron optical images using a predetermined image processing filter;
a comparison unit for comparing the electro-optical image of the object with the reference image corresponding to the electro-optical image of the object;
a database storing information on a plurality of first image processing filters that can be searched for using feature information indicating features;
a search unit that searches for a plurality of second image processing filters having predetermined characteristics based on characteristics of an image processing filter currently selected as the predetermined image processing filter from among the plurality of first image processing filters using the database;
a selection unit that selects, from among the plurality of second image processing filters, a second image processing filter that creates a reference image that matches the electro-optical image better than a reference image created using a currently selected image processing filter;
1. An electron beam inspection apparatus comprising: The electron beam inspection device according to claim 1, characterized in that the feature of each first image processing filter is the area of the reference pattern after filtering when the image of the reference pattern is filtered by the first image processing filter. The electron beam inspection device according to claim 1 or 2, characterized in that the characteristic of each first image processing filter is the distance between the intermediate gradation position of the reference pattern after filtering when the image of the reference pattern is filtered by the first image processing filter and the reference position outside the pattern. An electron beam inspection device according to any one of claims 1 to 3, characterized in that the characteristics of each first image processing filter are the maximum and minimum gradation values of the reference pattern after filtering when an image of the reference pattern is filtered using the first image processing filter. An electron beam inspection device according to any one of claims 1 to 4, characterized in that the feature of each first image processing filter is the rank of the micro-figures in the design image used to acquire the first image processing filter and the number of micro-figures of the rank. the selection unit selects, from among the plurality of second image processing filters, two or more second image processing filters that generate a reference image having a high degree of match;
The electron beam inspection device described in any one of claims 1 to 5, characterized in that the reference image creation unit uses the two or more selected second image processing filters to perform filter processing on a design image corresponding to the target electron optical image among the remaining electron optical images of the multiple electron optical images , thereby creating two or more reference image candidates corresponding to the target electron optical image, and creating a reference image by combining the two or more reference image candidates. The electron beam inspection device described in any one of claims 1 to 5, characterized in that the reference image creation unit creates a reference image corresponding to the target electron optical image by performing filter processing on a design image corresponding to the target electron optical image among the remaining electron optical images of the multiple electron optical images on which comparison processing has not yet been performed, using a selected second image processing filter. creating a reference image corresponding to the electron-optical image of the target by filtering a design image corresponding to the electron-optical image of the target for each of the plurality of electron-optical images using a predetermined image processing filter;
comparing the electro-optical image of the object with the reference image corresponding to the electro-optical image of the object and outputting the result;
a step of searching for a plurality of second image processing filters having predetermined characteristics based on the characteristics of an image processing filter currently selected as the predetermined image processing filter from among the plurality of first image processing filters, using a database storing information on a plurality of first image processing filters that can be searched using characteristic information indicating the characteristics;
selecting a second image processing filter from among the plurality of second image processing filters that produces a reference image that matches better than a reference image produced using a currently selected image processing filter for comparison with the electro-optical image;
1. An electron beam inspection method comprising: The electron beam inspection method according to claim 8, characterized in that the feature of each first image processing filter is the area of the reference pattern after filtering when the image of the reference pattern is filtered by the first image processing filter. The electron beam inspection method according to claim 8 or 9, characterized in that the characteristic of each first image processing filter is the distance between the intermediate gradation position of the reference pattern after filtering when the image of the reference pattern is filtered by the first image processing filter and the reference position outside the pattern. An electron beam inspection method according to any one of claims 8 to 10, characterized in that the characteristics of each first image processing filter are the maximum and minimum gradation values of the reference pattern after filtering when an image of the reference pattern is filtered using the first image processing filter. An electron beam inspection method according to any one of claims 8 to 11, characterized in that a reference image corresponding to the target electron optical image is created by performing filtering processing on a design image corresponding to the target electron optical image among the remaining electron optical images of the plurality of electron optical images for which comparison processing has not yet been performed, using a selected second image processing filter.

Images