JP3524853B2 - Pattern inspection apparatus, pattern inspection method, and recording medium - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection apparatus, a pattern inspection method, and a recording medium, and more specifically, for example, a semiconductor (LSI) or a liquid crystal panel and masks thereof created according to design data ( The present invention relates to a pattern inspection device, a pattern inspection method, and a recording medium for inspecting a fine pattern such as a reticle).

[0002]

2. Description of the Related Art A die-to-die (die to dice) method is used for inspecting a wafer pattern in a manufacturing process of a semiconductor integrated circuit or for inspecting a mask for forming the pattern.
e) An optical pattern inspection device using a method called comparison is used. This inspection method is a method of finding a defect by comparing images obtained from the same position of a die to be inspected and its neighboring die with each other.

On the other hand, in order to inspect a mask called a reticle without a proximity die, a die-to-database (d
ie to database) A method called comparison is adopted. That is, a method is used in which CAD data is converted into an image format and used as a substitute for a proximity die, and the same inspection as described above is performed. This technique is disclosed in, for example, US Pat.
No. 3702 "Automated photomask inspection apparat
However, in this method, the rounded part of the corner of the actual pattern formed on the wafer is recognized as a defect, and as a countermeasure, the rounded part in the image obtained from the CAD data is rounded. This is avoided by a method such as pre-processing that gives the result: When die-to-database comparison inspection is performed in such a situation, pattern deformation that does not need to be determined as a corner defect is recognized as a defect. However, even if the above-mentioned pre-processing is performed, if it is set to ignore the pattern deformation of the corner, there is a dilemma that it is impossible to recognize the minute defect existing in other than the corner.

At present, the mask needs to be exactly matched with the CAD data, so that the inspection by the die-to-database comparison method is put to practical use. However, the pattern transferred onto the wafer is allowed to be deformed within a range in which the electrical characteristics and the like are guaranteed, and the pattern is actually deformed to a considerable extent due to differences in exposure conditions.

Further, the above-mentioned pattern inspection method of the die-to-die comparison system cannot detect a defect called a systematic defect which is commonly generated in all the dies on a wafer due to a mask defect or the like. That is,
This is because the same defect has occurred in both the die to be inspected and the adjacent die to be compared, and therefore the difference cannot be recognized by comparing the two.

Therefore, a matching inspection between CAD data and a wafer image has been proposed although it has not been put into practical use due to problems in calculation cost and the like. For example, NEC
Technical report Vol. 50 No. There is "Automatic failure point tracing method of logic LSI using electron beam tester" of 6/1997. This document describes a method using projections of wiring edges on the X and Y axes, a method focusing on wiring corners, and a method applying a genetic algorithm. Further, as a method adopted in this document, a matching method in which a closed region is extracted after linearly approximating an edge and the closed region is used is described. However, none of these methods can achieve a speed that can be used for high-speed inspection, and furthermore, it is impossible to perform matching while detecting the deformation amount of the pattern.

Further, at present, an automatic defect type classification (Auto Defect Clasing) is performed by comparing an image including a defect (defect image) with an image of a corresponding adjacent die (reference image).
sification: ADC) is used. However,
The unevenness of brightness of the reference image affects the recognition accuracy. In some cases, the inside and outside of the pattern cannot be identified only from the image. In such a case, it is often difficult to distinguish between a short circuit and a defect. In addition, since it is not possible to obtain information as to which pattern the defect destroys, it is impossible to classify the fatal defect into the pattern and the defect that is not so.

In the inspection method using the die-to-die comparison, there is an error in the position of the defect due to the stage accuracy of the inspection device and the optical system accuracy, and the error is about 10 times or more than the wiring pattern width. large. Due to this, even if the defect position is projected on the pattern (design pattern) to be formed, the defect position of the pattern cannot be accurately specified.

[0009]

In recent years, the pattern width of an integrated circuit has become about the light source wavelength used in the exposure process or less than that. For such pattern formation, the optical proximity effect correction is performed. (Optical Proximity
Correction: OPC) pattern is used. This is a technique of forming a mask by adding an OPC pattern to design data and exposing the mask to bring an actual pattern on a manufactured wafer close to the design data.

Whether the OPC pattern is effectively acting as a correction on the wafer pattern depends on whether the conventional die
It cannot be inspected by two-die comparison. Therefore, there is a demand for a method for solving the problem, for example, a method capable of performing comparative verification between a wafer pattern and design data in consideration of the allowable pattern deformation amount.

Further, for example, a system-on-chip (SO
In the high-mix low-volume production seen in C), a short delivery time is required. In such a case, even if a systematic defect is found in the final inspection, electrical inspection, it may not be possible to meet the short delivery time. As a countermeasure against this, there is a demand for monitoring the difference from the design data at each stage of the exposure process. Therefore, there is a demand for an inspection method in which a pattern deformation that does not affect the electrical characteristics is set as an allowable pattern deformation amount and the design data and the wafer pattern can be compared and verified while considering the deformation within the allowable pattern deformation amount.

At present, as a pattern deformation evaluation, a design check is performed by a lithography simulator or the like. In order to verify the validity of this simulation, a means for comparing and examining the pattern (simulation pattern) output by the lithography simulator and the actual pattern is required.

Further, it is becoming more and more important to improve the circuit design technique by obtaining the pattern deformation amount with respect to the design data.

By the way, at present, a CD-S is used for controlling the pattern line width of a wafer in the manufacturing process of a semiconductor integrated circuit.
EM (Critical Dimension Scanning Electron Microsc
ope) is used. This CD-SEM automatically measures the line width of a linear pattern at a designated position for each transfer unit of a stepper called a shot by using a line profile. This length measurement is carried out at several locations for several shots on several wafers per lot to determine whether the transfer function of the stepper is normal.
It can be managed in m units.

In addition to the line width, the shrinkage of the wiring end and the position of the isolated pattern are important for the management of the circuit pattern, but the automatic length measuring function of the CD-SEM is one-dimensional and only the length such as the line width. I can't measure. Therefore, the measurement of these two-dimensional shapes is performed by the operator visually observing an image obtained from the CD-SEM or another microscope.

Optical proximity correction (OPC) plays an important role not only in ensuring the line width of a linear pattern but also in forming the shape of a corner or an isolated pattern. Further, due to the improvement in operating frequency, in addition to the gate line width, it is now important to control the shape of the tip and the root of the gate wiring pattern called an end cap or field extension.

The shape measurement of such a two-dimensional pattern is
It is important not only in the sampling inspection in the manufacturing process but also in the trial production stage. Especially, in the trial production stage, the inspection of the pattern formation on the entire wafer surface is required.

However, as described above, the management of the two-dimensional shape is currently performed by human work, and automation is required in terms of accuracy and productivity.

Therefore, an object of the present invention is to perform a comparative inspection of a pattern image to be inspected and a reference pattern in real time.

Another object of the present invention is to allow matching while allowing a shape difference within an electrically allowable range.

Another object of the present invention is to perform stable defect detection.

Further, another object of the present invention is to enable quantitative and high-speed automatic measurement of a two-dimensional pattern (inspection target pattern image) which has been visually performed so far.

[0023]

To achieve the above object, in a first aspect of the present invention, a pattern inspection apparatus according to the present invention is a pattern inspection apparatus for inspecting a pattern to be inspected by comparing it with a reference pattern. A storage means for storing the reference pattern, an input means for inputting an image of the inspection target pattern, and an edge of the input image of the inspection target pattern and an edge of the stored reference pattern. By doing so, an inspection means for inspecting the inspection target pattern and an output means for outputting the inspection result are provided.

Here, the inspection means may match the inspection target pattern image with the reference pattern by comparing the edge of the inspection target pattern image with the edge of the reference pattern. it can.

Here, the matching may be performed by expanding the edge of the pattern image to be inspected.

Here, the matching can be performed by expanding the edge of the reference pattern.

Here, the matching may be performed by using the sum of products of the amplitude of the edge of the pattern image to be inspected and the amplitude of the edge of the reference pattern in each pixel as an evaluation value.

Here, the matching is performed by using the sum of the inner products of the edge vector of the pattern image to be inspected and the edge vector of the reference pattern in each pixel or the sum of the absolute values of the inner products as the evaluation value. The vector may have the amplitude of the edge as its magnitude and the direction of the edge as its direction.

Here, the matching may be performed by changing the weighting for each part of the reference pattern.

Here, the inspection means may associate the edge of each pixel of the reference pattern with the edge of each pixel of the actual pattern image.

Here, the association may be performed in consideration of the distance between the edge of each pixel of the reference pattern and the edge of each pixel of the pattern image to be inspected, and the direction difference between both edges. it can.

Here, the inspecting means may form an area on the basis of the edge of the inspection object pattern image for which the correspondence cannot be made, and recognize the area as a defective area.

Here, the inspecting means forms an area based on the edges of the pattern image to be inspected for which the correspondence can be made, and detects an area in which the luminance distribution is nonuniform. However, the area can be recognized as a defective area.

Here, the inspecting means may determine the defect type based on the geometrical characteristic amount of the defect area.

Here, the inspecting means may determine the defect type based on a feature amount relating to the brightness of the defect area.

Here, the inspection means may calculate a pattern deformation amount of the inspection target pattern with respect to the reference pattern.

Here, the pattern deformation amount is at least 1 of the positional displacement amount, the magnification variation amount, and the line width thickening amount.
Can be included.

Here, the inspecting means may add a pattern attribute to the reference pattern.

Here, the inspection means takes a profile on the inspection target pattern image, detects a predetermined point for each profile, performs curve approximation on the detected point, and detects the edge of the inspection target pattern image. And can be.

In the second aspect of the present invention, the pattern inspection method according to the present invention is a pattern inspection method for inspecting by comparing an inspection target pattern with a reference pattern, and an input for inputting an image of the inspection target pattern. A step, an inspection step of inspecting the inspection target pattern by comparing an edge of the input image of the inspection target pattern with an edge of the reference pattern stored in a storage unit, and outputting the inspection result And an output step to perform.

In a third aspect of the present invention, the recording medium according to the present invention is a computer-readable recording medium in which a program for causing a computer to execute a pattern inspection method for comparing an inspection target pattern with a reference pattern is inspected. In the pattern inspection method, an input step of inputting an image of the inspection target pattern is compared with an edge of the input image of the inspection target pattern and an edge of the reference pattern stored in a storage unit. As a result, an inspection step for inspecting the inspection target pattern and an output step for outputting the inspection result are provided.

According to the above configuration, the comparison inspection between the inspection target pattern image and the reference pattern can be performed in real time.

Further, the matching can be performed by allowing the shape difference within the electrically allowable range.

Further, stable defect detection can be performed.

Further, the two-dimensional pattern (pattern image to be inspected) which has been visually observed can be quantitatively and automatically performed at high speed.

[0046]

BEST MODE FOR CARRYING OUT THE INVENTION Preferred embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a diagram showing an example of a theoretical pattern based on the design data, and FIG. 2 is a diagram showing an example of a pattern (actual pattern) actually manufactured based on the design data. As shown in FIG. 2, the actual pattern has a short circuit defect, a defect due to particle adhesion, or deformation within the allowable deformation amount. Therefore, the pattern theoretically obtained based on the design data will be somewhat different.

The pattern inspection apparatus according to the present embodiment is an inspection target pattern (for example, a pattern as shown in FIG. 2).
Is compared with a reference pattern (for example, the pattern as shown in FIG. 1) to be inspected.

FIG. 3 is a diagram showing an outline of the inspection process performed by the pattern inspection apparatus according to this embodiment. In the inspection process, first, the first edge is detected from the image of the pattern to be inspected. Next, the inspection target pattern image and the reference pattern are matched by comparing the first edge with the edge of the first reference pattern. As a result of the matching, the shift amount S ₁ is obtained, the shift amount S ₁
To shift the first reference pattern. Then, the inspection target pattern (actual pattern) is inspected by comparing the first edge with the edge of the shifted first reference pattern. In this first inspection, the amount of pattern deformation is obtained and defects are detected. The shift amount S ₂ is obtained as one of the pattern deformation amounts.

Next, in order to detect the second edge from the inspection target pattern image, the corresponding second reference pattern is shifted by the shift amount S ₁ + S ₂ . A profile is obtained on the inspection target pattern image using the shifted second reference pattern, and the second edge is detected. Then, the inspection target pattern is inspected by comparing the second edge with the edge of the shifted second reference pattern. Also in this second inspection, the pattern deformation amount is obtained and the defect is detected. The shift amount S ₃ is obtained as one of the pattern deformation amounts.

FIG. 4 is a diagram showing an example of the hardware configuration of the pattern inspection apparatus in this embodiment. The pattern inspection apparatus according to this embodiment includes a main controller 1, a storage device 2,
Input / output control unit 3, input device 4, display device 5, printing device 6
And an image generation device 7.

The main control section 1 is composed of a CPU and the like, and controls the entire apparatus as a whole. The main control unit 1 has a storage device 2
Are connected. The storage device 2 can take the form of a hard disk, a flexible disk, an optical disk, or the like. In addition, the main control unit 1 is provided with an input device 4, such as a keyboard and a mouse, input data, via the input / output control unit 3,
A display device 5 such as a display for displaying the calculation result and the like, and a printing device 6 such as a printer for printing the calculation result and the like are connected.

The main control section 1 includes a control program such as OS (Operating System), a program for pattern inspection,
Further, it has an internal memory (internal storage device) for storing required data and the like, and the pattern inspection is realized by these programs and the like. These programs can be stored in a floppy (registered trademark) disk, a CD-ROM, or the like, and can be read by a memory, a hard disk, or the like and executed before execution.

FIG. 5 is a diagram showing a functional block diagram of the pattern inspection apparatus in this embodiment. Reference pattern generation unit 11, search unit 12, output unit 13, and defect recognition unit 1
4 is realized by a program. The basic database 21, the recipe database 22, and the defect type reference database 23 are provided in the storage device 2.

The core database 21 is provided externally, and the pattern inspection apparatus passes the LAN to the core database 21.
May be accessed.

(Recipe) Before the inspection, first, a set of inspection parameters called a recipe is set. The parameters are the distance (pixel interval) on the actual pattern between pixels at the time of image acquisition of the image of the pattern to-be-inspected, which is the object of inspection, and 512 × 512 or 1
There are number of pixels such as 024 × 1024. From these values, the distance (image size) on the actual pattern of the image to be processed at one time can be grasped. In addition, parameters for edge detection and parameters for recognizing defects are set.

Design data is used as data to be compared with the pattern image to be inspected. As the design data, for example, CAD layout data in GDS format,
Leia fusion and fracturing can be used. In the present embodiment, the bundle of line segments obtained by this processing is
The recipe database 2 is clipped as a reference pattern by clipping in a rectangular area whose length is the image size plus the stage error and the maximum parallel movement amount of the pattern.
Store in advance in 2. When the stage error is negligible compared to the maximum translation amount of the pattern, the absolute coordinate value of the pattern deformation can be measured. In the present embodiment, in consideration of the error amount of the stage and the maximum parallel movement amount of the pattern, the reference pattern is processed to be larger than the inspection target pattern image, but instead, the inspection target pattern image is larger than the reference pattern. You may make it process it.

The corners may be rounded for the reference pattern. As shown in FIG.
The design data is a polygon with an acute angle (dotted line in the figure), while the circuit pattern actually formed has rounded corners. Therefore, a circle, an ellipse, a straight line, or a curve described by another method may be applied to the corner portion to correct the pattern so that it approximates the actual pattern.

If design data is used for the reference pattern, it becomes a defect inspection for performing comparison inspection with the pattern to be realized. In this case, an allowable amount that does not affect the electrical characteristics is set as the allowable pattern deformation amount. This allowable pattern deformation amount can be set for each wiring attribute, and can be made variable depending on whether the pattern is complicated or not.

A curve forming the outline of the exposure pattern obtained by the litho simulator on the reference pattern (solid line in FIG. 50)
With, it becomes possible to inspect defects while verifying the correctness of the simulation. The output data of the litho simulator is a light intensity distribution obtained by optical simulation. The contour curve is obtained from this distribution. The allowable pattern deformation amount in this case sets an error allowable as a simulation.

In this embodiment, design data is used as the reference pattern.

FIG. 6 is a flow chart showing an example of the recipe registration processing in this embodiment. First, the operator uses the input device 4 and instructs the reference pattern generation unit 11 to search parameters for design data (here, parameters that specify the type of sample to be inspected and the process), inspection mode, and image acquisition parameters (inspection area). , Image size, number of pixels, slot number for identifying the wafer, and optical system adjustment parameters), and parameters for edge detection and inspection (step S202).

The following information is set as parameters for edge detection and inspection.

(R1) Desired amount of pattern deformation (R2) Limit of allowable pattern deformation on the − side and + side, and limit of allowable directional difference of edges (R3) Edge detection parameter empirically determined from image quality (R4) Extraction rule for automatically recognizing pattern attributes (corner, straight line part, end point, isolated pattern, etc.) (R5) Length of profile acquisition section, interval between profile acquisition section, profile acquisition section, profile An interval for checking the brightness value in the acquisition section, and a method of taking a profile (whether the threshold method is used, etc.) (R6) The flag reference pattern generation unit 11 that determines whether or not to make the profile acquisition section variable at the time of measurement, The basic database 21 is searched using the design data search parameters (the type of the sample to be inspected and the process) as keys, and the design data is acquired. (Step S204). The basic database 21 is a database that stores design data (CAD data) for the inspection target pattern image.

Next, the reference pattern generator 11 generates a reference pattern based on the design data (step S20).
6).

When it is necessary to perform shrink processing (processing for changing the magnification), size processing (processing for changing the line width), etc. on the design data so as to be most suitable for the position of the edge detected from the pattern image to be inspected. There is. In addition, since the positions of the edges to be detected are generally different between the first edge detection and the second edge detection, if necessary, two types of reference patterns for the first edge detection and the second edge detection are used. prepare.

Since the inspection is performed for each inspection unit area obtained by dividing the input inspection area by the image size, the reference pattern is also generated accordingly. The tests include serial tests and random tests.

FIG. 8 is a diagram for explaining the sequential inspection. The examination area is usually determined as the sum of rectangles. That is, the inspection area is set not as a unit of the entire surface of the wafer but as a plurality of areas designated by rectangles (a short rectangle on the upper side and a long rectangle on the lower side as shown in FIG. 8). In order to inspect an area at high speed, sequential scanning is performed for each inspection unit area. A reference pattern is created for each inspection unit area.

FIG. 9 is a diagram for explaining the random inspection. In random inspection, a certain area is not inspected sequentially but in pinpoint inspection. In FIG. 9, the inspection is performed only on the inspection unit areas 301 to 304.

FIG. 10 is a diagram showing an example of the reference pattern, and FIG. 11 is a diagram showing an example of converting the reference pattern of FIG. 10 into an edge vector for each pixel. In FIG. 10, the reference pattern (dotted line) is shown with sub-pixel accuracy. Usually, the edge direction of the reference pattern is parallel to the horizontal direction (x direction) or the vertical direction (y direction) of the pixel. The edge of the reference pattern, like the edge of the pattern image to-be-inspected, also has information about the starting point (sub-pixel accuracy), direction, and amplitude for each pixel. In this embodiment, all the amplitudes of the edges of the reference pattern are set to 1
I have to.

As shown in FIG. 12, the reference pattern may include a curve. To convert a reference pattern including a curve into an edge vector, for example, the center of the pixel 2
The tangent line 26 at the point 262 on the reference pattern closest to 61
There is a method of using 3 as an edge vector.

Next, the reference pattern generator 11 stores the reference pattern, the type of the sample to be inspected, the process, the inspection mode, the image acquisition parameters, and the parameters for edge detection and inspection in the recipe database 2.
2 is registered (step S208). These data are called a recipe, which is a set of inspection parameters, and are managed by using the type, process, and inspection mode as keys.

(Inspection Process) FIG. 13 is a flowchart showing an example of the inspection process in this embodiment. First, the operator inputs the recipe search parameters (here, product type, process, and inspection mode) to the inspection unit 12 via the input device 4 (step S302).

The inspection unit 12 searches the recipe database 22 by using the recipe inspection parameter as a key, and retrieves the recipe (step S304). Then, in order to acquire the inspection target pattern image (optical image, electron beam image, focused ion beam image, probe microscope image, etc.),
Instructing the image generation device 7 of image acquisition parameters,
Instructing slot transportation, alignment, and adjustment of the optical system (step S306). Alignment means CA
It is a function of obtaining a conversion coefficient between the coordinate system used by the D data and the coordinate value for managing the actual wafer observation position. This is embodied in CAD navigation. In CAD navigation, after alignment, the coordinate values to be observed on the CAD data are converted into coordinate values for managing the observation position of the actual wafer, the field of view of the imaging device is moved to that position, and an image of that position is obtained. The method is well known.

As the image generating device 7, a wafer defect inspection device, a CD-SEM or various microscopes which are usually used can be used.

The image generation device 7
The inspection target pattern image (and its center position) is output to the inspection unit 12 (step S308).

(First Edge Detection) Next, the inspection unit 12
Performs the first edge detection from the inspection target pattern image (step S310). As the edge detection, for example, there are the following two methods.

(A) One is a method suitable when there is a contrast between the inside of the pattern and the background. Most of such images can detect edges by binarization.
If the contrast is relatively unclear, the edge cannot be clearly detected. At this time, for example, [Reference 1]: R.
M. Haralick, “Digital step edges from ZERO crossin
g of second directional derivatives ”, IEEE Trans.
Pattern Anal. Machine Intell., Vol. PAMI-6, No.1, p
The edge can be obtained by applying the method disclosed in p.58-68,1984. According to this method, the inflection point of the edge portion can be obtained with an accuracy of about 1/10 of the pixel unit.

(B) The other is that only the edges are bright,
A method that can be used when there is no contrast between the inside of the pattern and the background, for example [Reference 2]: “Cartan Stege
r. An unbiased detector of curvilinear structure
s ”, IEEE Trans. Pattern Anal. Machine Intell., 20
(2), the edge is calculated by the method disclosed in February 1998. According to this method, the peak of the edge portion can be obtained with an accuracy of about 1/10 of the pixel unit. However, in this method, the edge direction has only a value of 0 to 180 degrees. That is, the inside of the pattern cannot be specified.

Even if an edge is obtained by the above method using an edge amplitude image obtained by applying a differential filter (for example, Sobel filter or bandpass filter) to an image having a contrast between the inside of the pattern and the background. Good. In this case, the inside of the pattern can be discriminated and the edge direction can be specified.

Since these methods use a window having a somewhat large size, not only an accuracy of about 1/10 of a pixel unit can be obtained, but also the edge direction is stable. This means that it is not always necessary to connect edges to perform linear approximation.

In the edge detection in step S310, the amplitude and direction of the edge are obtained in pixel units from the inspection target pattern image. The amplitude takes a larger value as the edge becomes clearer. (A) In the case of an image in which there is a contrast between the inside of the pattern and the background, for example, the absolute value of the primary differential value of the image is used as the amplitude, and the secondary differential value of the image is obtained using the method of the above-mentioned Document 1. The zero-cross point of can be the edge position. On the other hand, in the case of (B) an image in which only the edges are bright, the sign inversion value (absolute value) of the secondary differential value of the image is used as the amplitude, and the primary differential value of the image is obtained, for example, by using the method of the above-mentioned Document 2. The zero-cross point of can be the edge position. In either case, the edges are obtained with subpixel accuracy. In the case of the image of (A), the direction from 0 degree to 360 degrees can be defined, but in the case of the image of (B), 0 degree to 1 degree can be defined.
Only directions up to 80 degrees can be defined. This is because the inside of the pattern cannot be identified from the local information in the image (B).

FIG. 14 is a diagram showing an example of an image (inspection target pattern image) having a contrast between the inside of the pattern and the background, and FIG. 15 is a diagram showing edges detected from the image of FIG. is there. In FIG. 14, the brightness value is shown for each pixel. As shown in FIG. 15, the edge is detected for each pixel, and information on the starting point (sub-pixel accuracy), the direction (0 to 360 degrees), and the amplitude is obtained for each pixel. As described above, the amplitude has a larger value as the edge is clearer.

FIG. 16 is a view showing an example of an image (inspection target pattern image) in which only the edge (B) is bright, and FIG.
FIG. 8 is a diagram showing edges detected from the image of 3A in FIG. 7. Also in FIG. 16, the brightness value is shown for each pixel. Further, as shown in FIG. 17, the edge is detected for each pixel, and information on the starting point (sub-pixel accuracy), the direction (0 to 180 degrees), and the amplitude is obtained for each pixel.

(Matching) Next, the inspection section 12 expands the edge of the pattern image to be inspected to obtain an expanded edge (step S312). In this embodiment,
It is expanded by the allowable pattern deformation amount that is allowed in terms of electrical characteristics. At this stage, the allowable pattern deformation amount is a positive integer. This value is a value obtained by converting the larger absolute value of the − side limit and the + side limit of the (R2) allowable pattern deformation amount into an integer. By expanding the pattern by the permissible pattern deformation amount, it is possible to allow and match the shape difference within an electrically permissible range.

FIG. 18 is a diagram showing an example of the edge amplitude of a one-dimensional pattern image to-be-inspected, and FIG. 19 is a diagram showing an example of expanding the edges of FIG. 18 and 19
In order to simplify the explanation, an example is shown in which the edge amplitude of each pixel in one dimension has a scalar value. To treat deformations within the allowable pattern deformation amount equally,
A maximum value filter having a window with a size twice the allowable pattern deformation amount is applied. The maximum value filter is to obtain the maximum value of the values of each pixel in the window that is in the vicinity of the target pixel, and use that value as the value of the pixel after filtering. In FIG. 19, the edge of FIG. 18 is expanded by 2 pixels to the left and right. This is because the allowable pattern deformation amount is 2
This is an example in the case of pixels.

On the other hand, assume that the edges of the reference pattern are as shown in FIG. When the matching evaluation value (degree) is obtained from FIG. 19 and FIG. 20, the matching evaluation value is the same even at the current position even if the inspection target pattern image is shifted by 1 pixel or 2 pixels to the left and right. Become.

In order to avoid this, as shown in FIG. 21, weighting may be performed for expansion. This means that the smaller the allowable pattern deformation amount, the better. To achieve the expansion of FIG. 21, 0.5, 0.75,
A smoothing filter of 1.0, 0.75, 0.5 may be used. In the case of FIG. 21, the evaluation value decreases if the inspection target pattern image shifts even one pixel to the left or right.

Here, as shown in FIG. 22, it is assumed that the edge of the reference pattern is wide by 2 pixels. When the evaluation value is obtained from FIGS. 21 and 22, the same evaluation value is obtained even at the current position, even if the inspection target pattern image is deviated by 1 pixel to the left and right.

In order to avoid this, weighting and expansion may be performed as shown in FIG. In order to realize the expansion shown in FIG. 23, a smoothing filter of 0.5, 0.9, 1.0, 0.9, 0.5 (FIG. 24) may be used. The coefficients of the smoothing filter should be obtained experimentally.

From the above, expansion as shown in FIG. 23 is desirable, but expansion as shown in FIG. 19 and FIG. 21 can also be used from the viewpoints of processing speed, edge congestion, and the like.

FIG. 25 is a diagram showing an example of the edge amplitude of the two-dimensional pattern image to be inspected, and FIGS.
FIG. 26 is a diagram showing an example of expanding the edge of FIG. 25. Figure 2
At 5, all amplitude values are 0 except at 20. FIG. 26 shows the result when the same expansion as FIG. 19 was performed, and FIG. 27 shows the result when the same expansion as FIG. 23 was performed.

FIG. 28 is a diagram showing an example of edge vectors of a two-dimensional pattern image to be inspected, and FIGS.
0 is a diagram showing an example in which the edge vector of FIG. 28 is expanded. FIG. 29 shows the result when the same expansion as that in FIG. 19 was performed, and FIG. 30 shows the result when the same expansion as that in FIG. 23 was performed. Expansion is performed for each of the x and y components.

The inspection section 12 compares the expanded edge (the edge obtained by expanding the edge of the pattern image to be inspected) with the edge of the reference pattern, and performs the pixel-by-pixel matching between the pattern image to be inspected and the reference pattern. (Step S314).

In the present embodiment, as will be described later, since matching is performed with sub-pixel precision, here, matching is performed in pixel units for the purpose of speeding up. Therefore, FIG. 31 shows FIG. 11 in pixel units.

In the matching in the present embodiment, the reference pattern is vertically and horizontally shifted pixel by pixel with respect to the pattern image to be inspected, and the position where the evaluation value F ₀ is maximum is set as the matching position (FIG. 32). In the present embodiment, the sum of the amplitudes of the dilated edges in the pixels where the edges of the reference pattern are present is set as the evaluation value F ₀ as follows.

[0097]

[Equation 1]

Here, E (x, y) is an edge vector having the amplitude of the expanding edge as its magnitude and the direction of the expanding edge as its direction. The size of E (x, y) is 0 at a place where no edge exists. R (x + xs, y + ys) is an edge vector having the edge direction of the reference pattern as its direction. However, the size of R (x + xs, y + ys) is 1 at the place where the edge exists and 0 at the place where the edge does not exist. Here, (xs, ys) is the shift amount S ₁ of the edge of the reference pattern.

In the calculation of F ₀ , if only pixels for which R (x, y) is not 0 are stored, the calculation can be performed at high speed and the storage area can be reduced. Sequential Similarity Detect (SSDA) using the sum of pixel amplitude values as an evaluation function.
The calculation is further accelerated by using the truncation of the high-speed calculation used in the ion Algorithm).

FIGS. 33 and 34 are views obtained by superposing FIGS. 29 and 31. In FIG. 33, pixel 2
54 corresponds to the pixel 251 in FIG. 29 and the pixel 252 in FIG. 31. In FIG. 34, the pattern image to-be-inspected is shifted by 1 pixel to the right and 1 pixel below from the state of FIG. Therefore, pixel 255 corresponds to pixel 251 in FIG. 29 and pixel 253 in FIG. When the evaluation value F ₀ is used, the evaluation value becomes higher as the degree of overlap of pixels having edges increases. When the evaluation value F ₀ is used, expansion processing as shown in FIGS. 25 to 27 may be performed. The evaluation value F ₀ can be applied to both images (A) and (B).

In the present embodiment, the evaluation value F ₀ is used, but other evaluation values can be used. For example, in the case of an image having a contrast between the (A) inside of the pattern and the ground it is considered to use the following evaluation value F _a.

[0102]

[Equation 2]

Further, for example, in the case of an image in which only the edge (B) is bright, the following evaluation value F _b may be used.

[0104]

[Equation 3]

When using the evaluation value F _a or F _b ,
The expansion process as shown in FIGS. 8 to 30 may be performed. However, in the case of performing expansion as shown in FIG. 29, expansion is performed for both the maximum positive value and the maximum negative value,
Select the one with a larger value in the inner product calculation.

[0106] When comparing the evaluation value F ₀ and an evaluation value F _a and F _b, the evaluation value F ₀ is directed to the high-speed processing for data is a scalar. On the other hand, the evaluation values _Fa and _Fb are effective, for example, in the case shown in FIG. That is, the evaluation values _Fa and _Fb
When using, the edge vector of the vertical line portion of the reference pattern (FIG. 35A) and the inspection target pattern image (FIG.
The inner product of the horizontal line and the edge vector in (b)) is 0.
Since it is close to, the part 101 and the part 102 are well matched. However, when the evaluation value F ₀ is used, the direction is irrelevant, and only the amplitude is used for the determination, so that the part 101 and the part 103 may match.

Next, when the evaluation values F _a and F _b are compared, for example, as shown in FIG. 36, when F _a is used when the distance between the wirings 111 and 113 and the distance between the bases 112 and 114 are the same, , Which gives better results than F _b because it can be distinguished which is the line.

In the present embodiment, the edges of the pattern image to-be-inspected are expanded to perform matching, but instead, the edges of the reference pattern may be expanded to perform matching.

The weighting can be changed depending on the position of the edge of the reference pattern. This is done by the following procedure.

In FIG. 37, (a) shows an example of the reference pattern, and (b) shows an example of the reference pattern (solid line) and the inspection target pattern image (dotted line) of (a). The reference pattern shown in FIG. 37 (a) is a periodic pattern, but there is one gap. When such a reference pattern and the pattern image to-be-inspected are matched, as shown in FIG. 37 (b), even if the two patterns are deviated, they match except for the gap portion. It gets expensive. Therefore, it is conceivable to increase the weighting of the gap portion so that the matching evaluation value is greatly reduced when the gap of the inspection target pattern image and the gap of the reference pattern do not match.

As the weighting procedure, first, the pattern period is obtained by the autocorrelation method. Next, by comparing the original pattern and the pattern shifted by one period, the original pattern which is not in the pattern shifted by one period is obtained. Then, the obtained pattern is recognized as a unique pattern,
The degree (weighting) contributing to matching is made stronger than the other patterns. An empirical value (1 or more) is used for the amplitude of the reference pattern to express the degree of contribution. As this value, a fixed value or a fixed value / a ratio of unique patterns in all patterns is effective.

When matching is performed and the shift amount S ₁ = (xs, ys) that gives the maximum evaluation value is obtained, the reference pattern is shifted by S ₁ . Subsequent processing is performed in this shifted state.

The shift amount S ₁ can be output to the display device 5 and the printing device 6 as the inspection result.

After the matching is completed, the inspection target pattern image is binarized. Binarization is performed by determining the presence or absence of the edge amplitude with one of the edge detection parameters (threshold value) in the recipe. As the binarization method, the number of pixels corresponding to the edge of the reference pattern × p (usually about 0.9 to 1.1) is set to 1,
There is also a method (p-tile method) of binarizing the edge image of the inspection target pattern image. The above threshold value or p may be set as the parameter of (R3).

(First Inspection) Next, the inspection unit 12 performs the first inspection.
Conduct an inspection. Specifically, the pattern deformation amount is calculated and the defect is detected.

The inspection unit 12 first associates the edges of the inspection target pattern image with the edges of the reference pattern (step S318).

The position of the edge is handled with subpixel accuracy. Therefore, the distance between edges is also obtained with sub-pixel accuracy. The direction is determined as a value of 0 to 360 degrees, with the right direction as 0 degree.

For each edge pixel of the reference pattern, the edge pixel of the pattern image to-be-inspected within the distance of the allowable pattern deformation amount corresponding to (R2) is searched. Then, among the detected edges, those whose direction difference from the edge of the reference pattern is less than or equal to the allowable direction difference of the edge of (R2) are associated as an edge within the allowable deformation. That is, in the present embodiment, the distance between the edge of the pattern image to-be-inspected and the edge of the reference pattern, and the directions of both edges are considered in the correspondence. The associated vector d (x, y) between both edges can be used to obtain the pattern deformation amount.

When there are a plurality of candidates for association, the candidate with a small distance and a small direction difference is preferentially associated.

FIG. 38 is a diagram showing an example of the correspondence between the edges of the pattern image to-be-inspected and the edges of the reference pattern. In FIG. 38, an edge is indicated by an arrow to indicate the direction. In the example of FIG. 38, in each pixel including the edge of the reference pattern, the edges of the pattern image to be inspected are searched in the direction perpendicular to the edge direction from the center of the edge of the reference pattern to make the correspondence. . If an edge of the pattern image to be inspected whose distance is within the permissible pattern deformation amount and the difference in direction is equal to or less than the permissible direction difference between the edges is found, both edges are associated with each other.
In FIG. 38, the vector d between the associated edges is
(X, y) is shown for reference.

In FIG. 39, (a) shows an example of the edge of the reference pattern, and (b) shows an example of the edge of the inspection target pattern image corresponding to the reference pattern of (a). An example of associating both edges will be described with reference to FIG. In this example,
The allowable pattern deformation amount is one pixel. Further, the allowable difference in the direction of edges is 60 degrees. For example, when an edge of the inspection target pattern image corresponding to the edge 81 of the reference pattern is searched for, the edge 68 is within the distance of the allowable pattern deformation amount of the edge 81, and the direction difference is less than the allowable direction difference of the edge. Therefore, it is recognized as an edge corresponding to the edge 81. As for the edge 84 of the reference pattern, the edge 70 is recognized as the edge of the corresponding pattern image to-be-inspected. At this time, with respect to the edge 82 of the reference pattern, the edge 61 is not within the distance of the allowable pattern deformation amount. The edge 64 is not within the distance of the allowable pattern deformation amount, and the direction difference is also larger than the allowable direction difference of the edge. Although the edges 66 and 69 are within the distance of the allowable pattern deformation amount, the directional difference is larger than the allowable directional difference of the edges. Therefore, the edge corresponding to the edge 82 cannot be found.
Similarly, the edge 83 cannot be found.

The example of FIG. 39 does not distinguish between the inside and the outside of the pattern, and the direction has only a value of 0 to 180 degrees, but a method of distinguishing between the inside and outside of the pattern is also possible. is there. For example, if the edge direction is determined so that the inside of the pattern is always placed on the right hand, then FIG. 39 (a)
40 becomes as shown in FIG. 40, and the association can be performed more strictly.

Next, the inspection section 12 detects defects (step S320). When a defect is detected, the defect information (here, the defect position, the size information, and the image) is output to the defect type recognition unit 14 (steps S322 and S322).
4).

The defect type recognizing unit 14 determines the defect type based on the defect information and the information of the defect type reference database 23 (step S326). That is, the feature amount is obtained from the given image, and the defect type reference image database is obtained. The defect type is determined by collating with the feature amount of the image accumulated in (1). The defect type recognition unit 14 outputs the defect information and the defect type to the display device 5 and the printing device 6 via the output unit 13 (step S328). Here, the defect type reference database 23
Is a registration of already acquired images for each defect type.

As a method of recognizing a defective area, a method (recognition method A) of recognizing an area from an edge of an inspection target pattern image that cannot be associated and recognizing this as a defective area can be considered. This is effective for detecting a defect having a clear edge. However, since it is weak against the detection of an unclear edge defect, in such a case, the area is recognized from the edge of the associated pattern image to be inspected, and the distribution of pixel brightness values in the area is non-uniform. A method for recognizing such a portion as a defective area (recognition method B)
Is suitable. That is, the defect is recognized from the abnormality in the brightness value distribution.

In the recognition method A, the pixels of the edges (for example, the edges 61 to 67, 69 and 75 in FIG. 39 (a)) of the pattern image to be inspected which cannot be associated are recognized as defects. The inspection unit 12 expands these pixels and connects the pixels. A process called morphology is known as a process for expanding such a bitmap (binarized image). Next, the pixels connected by the labeling process are individually recognized as one block area. Here, the labeling process is a method of writing the same value to pixels connected in four neighborhoods or eight neighborhoods to generate a connected pixel group. By giving different values to the unconnected pixels, the connected pixel groups can be distinguished. A foreign substance is recognized in the unit that can be separated as the region of the lump, and its outer shape is recognized. Pixels inside from the outline are painted. With these pixels as defects, the centroid and size of the defects are calculated.

In the recognition method B, the edges of the inspected pattern images that have been associated with each other are connected to form a region. In each of the inner and outer regions, the part excluding the boundary (edge) is obtained as a block of pixels. For the inner and outer regions of the block of pixels, the pixel brightness value is obtained from the initially obtained inspection target pattern image. These values can be expected to have a normal distribution if there are no defects. That is, it is possible to detect defective pixels by applying a quality control method. In the normal case, there should be little variation in brightness in each of the inner area and the outer area. Therefore, it is possible to detect an area in which the luminance distribution is non-uniform among the areas in the pattern image to be inspected and recognize the area as a defective area. The obtained defective pixels are recognized as a cluster, and the center of gravity and size are calculated.

FIG. 41 is a diagram showing an example of a pattern image to be inspected. The broken line 201 indicates the edge of the inspection target pattern image. Solid lines 202 and 203 on both sides of the broken line 201
Is a line segment obtained by thickening an edge by a specified width, and a solid line 202,
The portion surrounded by 203 is recognized as an edge area. Base 2
04 and the luminance value of the inside 205 of the pattern form a normal distribution.

As shown in FIG. 42, there is a high possibility that the portion D exceeding about ± 3σ is a foreign matter. D includes noise, but the noise exists relatively uniformly in the area,
The foreign substance is solidified and exists. A binarized map in which pixels having a brightness value of D are 1 and pixels having brightness values other than that are 0 is created. A block of pixels with 1 less than or equal to a specified size (eg, 2 × 2 pixels) (eg,
The block of pixels 207) of FIG. 41 is erased. A median filter or morphology filter can be used.
This size is an empirical value considering the size of the foreign matter to be detected. A lump of pixels with 1s left (eg
The pixel block 206) in FIG. 41 is regarded as a foreign substance.

The defect type recognition unit 14 can perform automatic defect type classification as follows. That is, the geometric feature amount of a block of pixels recognized as a defect is obtained. As a result, it is possible to grasp the shape characteristics of defects such as roundness and slenderness, and it is possible to judge that they are foreign matter if they are round and scratches if they are slender. Pixels recognized as defective are divided into three parts, the inner side, the outer side, and the boundary of the pattern. For each of these portions, the feature amount using the pixel brightness value of the initially obtained inspection target pattern image is obtained. With the feature amount obtained here, for example, when the foreign substance is determined from the geometric feature amount, it is possible to determine whether the foreign substance is a metal piece or an organic substance (for example, a human stain). That is, if the foreign matter is a metal, it is bright because it is strongly reflected, and if it is an organic matter, it is dark and the type can be determined. Further, when the variation in the brightness of the pixel recognized as the foreign matter inside the pattern is large, it is determined that the foreign matter is likely to be present on the pattern, and conversely, the variation in the brightness is small. In this case, it is determined that there is a high possibility that the foreign matter exists under the pattern. This is a difficult process in the conventional die-to-die method. Using these feature quantities, the defect type is determined by a well-known classification method. As the classification method, it is effective to perform a comparison with the defect type reference image database by the k shortest distance method to make a determination.

Such automatic defect type classification is performed by the ADC (Automatic Defe) of the conventional optical method and SEM method.
ct Classification), but according to the method of the present invention that uses design data, the inside and outside of the pattern can be clearly distinguished, so that the feature amount of each part can be accurately captured and the classification accuracy can be improved. improves.

Next, the inspection unit 12 obtains the pattern deformation amount from the relationship between the edges of the associated inspection target pattern image and the edges of the reference pattern (step S33).
0). The pattern deformation amount is obtained for a portion where no defect is detected as a result of defect detection. Then, the pattern deformation amount is output via the output unit 13 to the display device 5 and the printing device 6.
(Step S332).

As the pattern deformation amount, a pattern deformation amount obtained from the entire image and a pattern deformation amount obtained for each pattern attribute can be considered.

As the pattern deformation amount obtained from the entire image, for example, a positional deviation amount, a magnification variation amount, and a line width thickening amount can be considered.

The displacement amount is obtained as an average value of the vectors d (x, y) between the associated edges. This is the shift amount (correction amount) S ₂ with subpixel precision of S ₁ = (xs, ys). Based on the shift amount S ₂ , the reference pattern that is shifted by the pixel-by-pixel matching is shifted by the correction amount, so that the matching with sub-pixel accuracy can be performed.

In order to obtain the magnification variation amount in the x direction, the regression line is obtained by approximating the x component of the vector d (x, y) relating to the vertical reference pattern by the regression line D (x). Then, the gradient of the regression line is set as the magnification variation amount in the x direction. The same applies to the magnification variation amount in the y direction.

In FIG. 43, (a) shows an example of the edge of the reference pattern (broken line) and the edge of the pattern image to be inspected (solid line), and (b) shows y = y between the edges shown in (a).
Vector d (x, y ₀₎ in the ₀ to x components of the regression line D
An example approximated by (x) is shown. When the x component of the vector d (x, y ₀ ) is approximated by the regression line D (x) = ax + b, the slope a corresponds to the magnification change amount. In the example of FIG. 43A, it can be seen that the pattern of the inspection target pattern image is larger than the reference pattern as a whole.

In FIG. 44, (a) shows another example of the edge of the reference pattern (broken line) and the edge of the pattern image to be inspected (solid line), and (b) shows y between the edges shown in (a).
An example in which the x component of the vector d (x, y ₀ ) at = y ₀ is approximated by the regression line D (x) is shown. In the example of FIG. 44 (a),
In addition to the pattern of the inspection target pattern image being larger than the reference pattern as a whole, the width of the line is thicker. In FIG. 44A, the reference pattern lines (wirings) 121, 122, and 123 correspond to the pattern lines 124, 125, and 126 of the inspection target pattern image, respectively.

The amount of thickening of the line width in the x direction is, for example, sig
n (x, y ₀ ) · {x component of d (x, y ₀ ) −D (x)}
Can be obtained as an average value of. Where sign
(X, y ₀ ) takes -1 if the position of (x, y ₀ ) is the left end of the line, and takes 1 if it is the right end of the line. Regarding the amount of thickening of the line width, sign (x, y ₀ ) ·
By obtaining the variance of the x component of {d (x, y ₀ ) −D (x)}, the line width variation index can be obtained.

Next, the pattern deformation amount obtained for each pattern attribute will be described. The attributes of the pattern may include a corner 171, a long wiring 172, a tip 173, an isolated pattern 174, etc. (FIG. 45). The pattern deformation amount relating to the attribute of the pattern includes, for example, the above-mentioned amount of positional deviation, the amount of change in magnification, and the amount of thickening of the line width, as well as the amount of deformation of the characteristic amount such as area, perimeter, circularity, moment, radius of curvature Can be considered.

The attribute of the pattern can be automatically added to the reference pattern. However, the attributes can be added manually. Add (extract) pattern attribute
The rule for doing this is set as (R4) when creating the recipe.

FIG. 46 is a diagram showing the amount of positional deviation of the tip. As shown in FIG. 46A, the amount of positional deviation of the tip is the distance from the edge 164 of the reference pattern to the edge 163 of the pattern image to-be-inspected (perpendicular to the edge of the reference pattern). As the amount of positional deviation of the tip, for example, the distance between the edge of the inspection target pattern image 163 closest to the edge 164 of the reference pattern and the edge 164 of the reference pattern can be measured.

Further, as shown in FIG. 46 (b), the average value, the maximum value, the minimum value, the median value, the standard deviation, etc. of a plurality of distances measured for the section 157 having an arbitrary width are set as the positional deviation of the tip. It may be a quantity.

Although the positional deviation amount of the tip has been described with reference to FIG. 46, the positional deviation amount can be similarly measured for a long wiring, a corner, a part where attributes are combined with each other, and the like. Further, for example, for a corner, the amount of positional deviation in a direction having a half angle of the corner or a designated angle can be measured.

FIG. 47 is a diagram showing the amount of displacement of the center of gravity of an isolated pattern. The amount of displacement of the center of gravity is determined by the center of gravity 16 of the edge 160 of the reference pattern (which constitutes an isolated pattern).
2 and the center of gravity 161 of the edge 159 of the pattern image to be inspected (constituting an isolated pattern).

Further, in FIG. 47, it is conceivable to measure the deformation amount of the feature amount (area, perimeter, circularity, moment, etc.) of the isolated pattern. That is, it is possible to measure the difference between the feature amount of the edge 160 of the reference pattern and the feature amount of the edge 159 of the inspection target pattern image.

In FIG. 48, (a) shows an example of an edge corner of a reference pattern, and (b) shows an example of an edge corner of an inspection target pattern image. The corners of the edge 166 of the reference pattern shown in FIG. 48 (a) are rounded. As the radius of curvature of the corner, for example, the major axis, the minor axis or the radius obtained by least-squares approximation of the curve of the corner with an ellipse or a circle can be used. By obtaining the radius of curvature of the corner of the edge 166 of the reference pattern and the radius of curvature of the corner of the edge 165 of the pattern image to-be-inspected, the deformation amount of the radius of curvature of the corner can be obtained.

It is also possible to apply the above inspections simultaneously (with one image pickup) to a plurality of places within one image pickup range (within the field of view), instead of performing them one by one.

The inspection item is selected according to (R1) the desired pattern deformation amount of the above-mentioned recipe item.

The pattern attribute extraction rule (the above (R
4)) are various, an example of which will be described with reference to FIG. The corner is extracted as the vicinity of the contact point of the two line segments that come into contact with each other at a predetermined angle (90 degrees, 270 degrees, etc.). The long wirings are separated as line widths and are extracted as two parallel line segments having a length equal to or longer than a designated length. The tip is a line segment with the length of the line width.
It is extracted as a contact portion with an angle of 0 degree.
The isolated pattern is extracted as a closed figure having a predetermined area or less.

(Second Edge Detection) The inspection section 12 again detects an edge from the pattern image to be inspected for a portion where no defect is detected as a result of the defect detection (step S334).

The edge detection of the pattern image to be inspected is performed by obtaining the profile on the pattern image to be inspected based on the second reference pattern. Where the second
As the reference pattern of, the reference pattern when the position of the point Q in FIG. 53 is considered as an edge is used. On the other hand, as the first reference pattern, for example, in the case of an image in which only the above-mentioned (B) edge is bright, a reference pattern in which the position of the point P is considered to be an edge is used. Therefore, the second
The reference pattern and the first reference pattern are generally different.

Before the edge detection of the pattern image to-be-inspected, the second reference pattern is shifted by the above-mentioned shift amount S ₁ + S ₂ . Subsequent processing is performed in this shifted state.

Various methods such as a threshold method and a linear approximation method are disclosed for obtaining the edge position from the profile. In the present embodiment, the threshold method among them is used to calculate C.
The line width measurement performed by D-SEM is applied to a two-dimensional pattern (inspection target pattern image). However, the same processing can be performed by replacing the threshold method with another method such as a linear approximation method. Here, the straight line approximation method is a method of approximating a profile by a straight line and specifying an edge position using an intersection.

There are two possible methods for edge detection. One of them is a method of presetting the profile taking direction and position with respect to the second reference pattern.

In the present embodiment, when the profile acquisition section is set in advance, it is performed when the recipe is created as described above. In this case, the (R6) profile acquisition section of the above-mentioned recipe item is made variable and the flag as to whether or not to determine at the time of measurement is turned off, and the profile acquisition section is set in advance for the second reference pattern. become.

The section for acquiring a profile (profile acquisition section) is the length of the profile acquisition section of (R5),
Based on the interval between the profile acquisition section and the profile acquisition section, for example, as shown in FIG.
Is set in the vertical direction of the second reference pattern (the double line in the figure). Second shown in FIG. 49
The reference pattern (1) has been corrected by rounding the corners, as already described with reference to FIG.
Further, instead of the above-mentioned second reference pattern, as shown in FIG. 50, it is also possible to use a curve (solid line in the figure) forming the outer shape of the exposure pattern obtained by the litho simulator.

As the second edge detection processing, at the position (section) corresponding to the above-mentioned profile section in the pattern image to-be-inspected, the profile is determined based on the interval for checking the brightness value in the profile acquisition section of (R5). create. The interval is usually an arbitrary value equal to or smaller than the pixel interval, and the length of the profile section is an arbitrary length longer than the deformation allowance of the pattern. The profile is created using a method such as bilinear interpolation, spline interpolation, or Fourier series.

FIG. 51 is an enlarged view of a part (portion B) of FIG. 49, and FIG. 52 is an enlarged view of a part (portion C) of FIG. The double line in the figure is the profile acquisition section, the intersection of the grid shows the pixel position, and the black dot shows the position where the luminance value of the inspection target pattern image is examined.

The bilinear interpolation method is (0,
0) (0,1) (1,0) (1,1) pixel luminance values I (0,0), I (0,1), I (1,
0), I (1,1), and positions (x, y), (0 <
The brightness value I (x, y) of a point in x ≦ 1, 0 <y ≦ 1) is calculated by the following calculation formula.

[0161]

## EQU00004 ## I (x, y) = [I (0,0) (1-x) + I
(1,0) x] (1-y) + [I (0,1) (1-x)
+ I (1,1) x] y

From the profile thus obtained, the threshold value method is applied to determine the second edge position. Figure 5
As shown in FIG. 3, the maximum brightness value V and its position P in the obtained profile are obtained. The threshold value T is a value obtained by multiplying the maximum brightness value V by a coefficient k specified in advance, and the intersection of the straight line of brightness value = threshold value T and the profile curve is obtained. At these intersections, an intersection Q that is located on the outer side of the pattern from the point P and is closest to the point P is obtained. This intersection Q is obtained for all profiles and is set as the second edge position.

The coefficient k plays a role in determining the second edge position. That is, since the cross-sectional shape of the actual wiring formed on the wafer is trapezoidal, the edge position is controlled by the coefficient k whether the trapezoid is controlled at the upper side, the lower side, or the middle side. be able to.

After obtaining the above-mentioned edges, curve approximation (including polygonal approximation) is performed based on them to obtain the second edge. The simplest method is a method of simply connecting as polygonal lines, but as a method of smoothly connecting using the least square method, for example, the following method can be used. That is, as shown in FIG. 54 (a), T. Pavlidis and SLHo
rowitz: “Segmentation of plane curves”, IEEE Tr
The split fusion method disclosed in ans. on Computers, vol. C-23, no. 8 Aug., 1974 can be used. In addition to this, it is also possible to use curve approximation by smoothing plane data using the least squares method and the two-dimensional spline function as shown in FIG. 54 (b). The former can be processed at high speed, but it is not flexible for those containing many rounded shapes. On the other hand, the latter has the characteristics of satisfying high speed and having flexibility. In addition to these, various methods such as a method using a Fourier descriptor are disclosed, and these methods can be replaced.

The above curve approximation can be performed even after the first edge detection.

Next, as another method different from this, there is a method of making the profile acquisition section variable and determining it at the time of edge detection. That is, as shown in FIG. 55 (a), this is a method of setting the profile acquisition section in the vertical direction of the first edge of the detected inspection target pattern image. According to this method, as shown in FIG. 55 (b), even if the first edge (solid line) of the inspection target pattern image is deviated from the second reference pattern (dotted line), the profile acquisition section is specified. , Edges can be detected.
This method is easier to follow the deformation of the pattern than the above method. After setting the profile acquisition section, the same processing as the above method is performed.

The result of the second edge detection can be output to the display device 5 and the printing device 6.

The detected second edge is, for example, as shown in FIG.
The edge vector for each pixel can be obtained by using the method described above. This edge vector corresponds to the edge vector obtained by the binarization process obtained before the first inspection.

(Second Inspection) After the second edge detection as described above, the inspection unit 12 performs the second inspection (step S336).

This inspection is the same processing as the above-mentioned first inspection.
Therefore, the defect is detected and the pattern deformation amount is obtained. This
The amount of positional deviation (shift
Amount) S ₃Is the above S₂Corresponding to. S obtained here
₃In the above S₁And S₂Is the second criterion
All between the pattern and the pattern of the pattern image to be inspected
The shift amount.

The inspection result is output to the display device 5 and the printing device 6 via the output unit 13 (step S33).
8).

If the above process is performed for all the inspection unit areas, the inspection process is terminated, and if not, the process returns to step S308 (step S340).

(Other inspections) In the case of an SEM having a function of electromagnetically observing a part of a low-magnification image with a high-magnification image, it is possible to measure a pattern that cannot be completely covered by a high-magnification image. That is, it means that the edge position obtained in the high-magnification image can be accurately converted into the edge position obtained in the low-magnification image. The same relationship as this may be realized by a high precision stage. For example, in FIG. 56, the positions 182 and 183 on the pattern 181 of the inspection target pattern image are respectively represented by the high-magnification image 1
84, 185, and then converted to a position on the low-magnification image 187 to obtain the width 1 of the pattern 181 of the inspection target pattern image.
When 86 is obtained, the length can be measured more accurately than when only the low-magnification image 187 is obtained.

(Inclination, magnification adjustment) In the above inspection method, the inclination of the pattern image to be inspected, before the inspection or at an appropriate time during the inspection, is utilized by utilizing the method of the pattern deformation amount. The magnification can be adjusted. That is, the inspection target pattern image and the reference pattern of the portion suitable for adjustment are acquired. Affine transformation is performed to obtain some pattern images to be inspected whose inclination and magnification that can be candidates have been changed. The obtained pattern image for inspection is compared with the reference pattern, and the pattern image for inspection having the smallest pattern deformation amount is selected. The inclination and magnification of the selected inspection target pattern image are registered as correction amounts. Instead of performing the affine transformation on the inspection target pattern image, the method of performing the affine transformation on the reference pattern may be replaced.

The affine transformation means a primary transformation using the coefficients a to f.

[0176]

[Equation 5] X = ax + by + c Y = dx + ey + f

By using the information such as the pattern deformation amount, the position and size of the defect region, the defect type, the statistical amount of the pattern deformation amount and the image obtained by the inspection method of the present invention, the circuit of the defect region can be obtained. It is possible to analyze the influence degree of, the degree of influence on the circuit in the preceding and following processes, and the analysis of optimization parameters such as exposure conditions.

The pattern inspection apparatus according to this embodiment can be considered as an apparatus for performing pattern matching, if attention is paid to the portion that outputs the shift amount.

Although one example of the present invention has been described above, various other modifications are possible. For example, it is easy to transform the acquired image data into an offline input processing type via an external input device such as a magneto-optical disk or a magnetic tape, or via a LAN such as Ethernet. It is also possible to adopt a hybrid method in which a typical die in a wafer is inspected by the method of the present invention and then the other die are inspected by die-to-die comparison.
Furthermore, the image generation method may be another method, and the design data is not limited to CAD and may be another method. In the present embodiment, the inspection result and the like are output to the display device 5 and the printing device 6, but may be output to an image database, a simulator, a recording medium, or the like, or may be output to another computer via a network. You may make it transmit (output).

[0180]

As described above, according to the present invention, the comparison inspection between the inspection target pattern image and the reference pattern can be performed in real time.

Further, it is possible to perform matching while allowing a shape difference within an electrically allowable range.

Further, stable defect detection can be performed.

Furthermore, the two-dimensional pattern (inspection target pattern image), which has been visually observed so far, can be quantitatively and automatically performed at high speed.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a theoretical pattern based on design data.

FIG. 2 is a diagram showing an example of a pattern actually manufactured based on design data.

FIG. 3 is a diagram showing an outline of inspection processing performed by the pattern inspection apparatus according to the embodiment of the present invention.

FIG. 4 is a diagram showing a hardware configuration example of a pattern inspection apparatus according to an embodiment of the present invention.

FIG. 5 is a diagram showing a functional block diagram of the pattern inspection apparatus in the embodiment of the present invention.

FIG. 6 is a flowchart showing an example of recipe registration processing according to the embodiment of the present invention.

FIG. 7 is a diagram showing an example of correction of a reference pattern.

FIG. 8 is a diagram for explaining sequential inspection.

FIG. 9 is a diagram for explaining a random inspection.

FIG. 10 is a diagram showing an example of a reference pattern.

11 is a diagram showing an example in which the reference pattern of FIG. 10 is converted into an edge vector for each pixel.

FIG. 12 is a diagram showing an example in which a reference pattern including a curve is converted into an edge vector.

FIG. 13 is a flowchart showing an example of inspection processing according to the embodiment of the present invention.

FIG. 14 is a diagram showing an example of an image (inspection target pattern image) having a contrast between the inside of the pattern and the background.

FIG. 15 is a diagram showing edges detected from the image of FIG.

FIG. 16 is a diagram illustrating an example of an image (inspection target pattern image) having only a bright contour.

17 is a diagram showing edges detected from the image of FIG.

FIG. 18 is a diagram showing an example of edge amplitude of a one-dimensional pattern image to-be-inspected.

FIG. 19 is a diagram showing an example in which the edge of FIG. 18 is expanded.

FIG. 20 is a diagram showing an example of the amplitude of an edge of a one-dimensional reference pattern.

FIG. 21 is a diagram showing another example in which the edge of FIG. 18 is expanded.

FIG. 22 is a diagram showing another example of the amplitude of the edge of the one-dimensional reference pattern.

23 is a diagram showing another example in which the edge of FIG. 18 is expanded.

FIG. 24 is a diagram showing an example of a smoothing filter.

FIG. 25 is a diagram showing an example of the amplitude of edges of a two-dimensional pattern image to-be-inspected.

FIG. 26 is a diagram showing an example of expanding the edge of FIG. 25.

FIG. 27 is a diagram showing another example in which the edges of FIG. 25 are expanded.

FIG. 28 is a diagram showing an example of edge vectors of a two-dimensional inspection target pattern image.

FIG. 29 is a diagram showing an example of expanding the edge vector of FIG. 28.

FIG. 30 is a diagram showing another example in which the edge vector of FIG. 28 is expanded.

FIG. 31 is another diagram showing the reference pattern of FIG. 10 as edge vectors in units of pixels.

FIG. 32 is a diagram for explaining matching.

FIG. 33 is a view obtained by superposing FIGS. 29 and 31.

FIG. 34 is a diagram in which FIG. 29 and FIG. 31 are superimposed.

FIG. 35A is a diagram showing an example of a reference pattern, and FIG. 35B is a diagram showing an example of an inspection target pattern image.

FIG. 36 is a diagram showing an example in which the wiring spacing and the base spacing are the same.

FIG. 37A is a diagram showing an example of a reference pattern, and FIG. 37B is a diagram showing an example of the relationship between the reference pattern of FIG.

FIG. 38 is a diagram showing an example of edges of a pattern image to-be-inspected and edges of a reference pattern after performing matching.

FIG. 39A shows an example of an edge of a reference pattern, and FIG. 39B
FIG. 4 is a diagram showing an example of edges of a pattern image to be inspected.

[Fig. 40] Fig. 40 is a diagram illustrating another example of a method of giving direction information.

FIG. 41 is a diagram showing an example of an inspection target pattern image.

FIG. 42 is a diagram showing an example of frequency distribution with respect to luminance values.

43A shows an example of an edge of a reference pattern and an edge of a pattern image to-be-inspected, and FIG. 43B shows a vector d (x, y ₀ ) at y = y ₀ between the edges shown in FIG. _43A . It is a figure which shows the example which approximated the x component of the with the regression line D (x).

FIG. 44A shows another example of the edges of the reference pattern and the inspection target pattern image, and FIG. 44B shows the vector d (x, y at y = y ₀ between the edges shown in FIG. It is a figure which shows the example which approximated the x component of ( ₀ ) with the regression line D (x).

FIG. 45 is a diagram illustrating an example of a pattern attribute.

FIG. 46 is a diagram showing a positional displacement amount of a tip.

FIG. 47 is a diagram showing a positional shift amount of the center of gravity of an isolated pattern.

48A is a diagram showing an example of an edge corner of a reference pattern, and FIG. 48B is a diagram showing an example of an edge corner of an inspection target pattern image.

FIG. 49 is a diagram showing an example of a profile acquisition section.

FIG. 50 is a diagram showing a curve forming an outer shape of an exposure pattern obtained by a litho simulator.

51 is an enlarged view of a portion (portion B) of FIG. 49. FIG.

52 is an enlarged view of a portion (portion C) of FIG. 51. FIG.

FIG. 53 is a diagram showing an example of a profile.

FIG. 54 is a diagram showing an example in which curve approximation is performed based on a second edge position (point) to obtain a second edge.

55A shows another example of the profile acquisition section, and FIG. 55B shows the first edge and the second edge of the inspection target pattern image.
It is a figure which shows the example of the relationship with the reference pattern of.

[Fig. 56] Fig. 56 is a diagram illustrating an example of performing length measurement using a high-magnification image and a low-magnification image.

[Explanation of symbols]

1 Main Control Unit 2 Storage Device 3 Input / Output Control Unit 4 Input Device 5 Display Device 6 Printing Device 7 Image Generation Device 11 Reference Pattern Generation Unit 12 Inspection Unit 13 Output Unit 14 Defect Type Recognition Unit 21 Core Database 22 Recipe Database 23 Defect Type Reference database 61-70, 75, 81-84 Edge 101-103 Part 111, 113 Wiring 112, 114, 204 Base 121-126 Line 151 Inspection target pattern image 152 Reference pattern 157 Section 159, 163, 165 Inspection target pattern image Edges 160, 164, 166 Reference pattern edges 161 Center of gravity of edges of inspection target pattern image 162 Center of gravity of edges of reference pattern 171 Corner 172 Long wiring 173 Tip 174 Isolated pattern 181 Patterns of inspection target pattern image 182, 183 Positions 184, 185 High-magnification image 186 Width of pattern of pattern image to be inspected 187 Low-magnification image 201 Broken lines 202, 203 Solid line 205 Inside pattern 206, 207 Pixels of pixels 251-255 Pixels 261 Pixels center 262 Pixels closest to center Point 263 on the pattern Tangent lines 301 to 304 Inspection unit area

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI H01L 21/66 G01B 11/24 K F (56) References JP-A-8-110305 (JP, A) JP-A-10-312461 (JP, A) JP 10-207917 (JP, A) JP 8-76359 (JP, A) JP 4-194702 (JP, A) JP 63-88682 (JP, A) Kaihei 7-43309 (JP, A) JP-A-2-191073 (JP, A) JP-A-6-66731 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 7 / 00-7/60 G01B 11/24 G01N 21/956 G03F 1/08 G06T 1/00 H01L 21/66

Claims (39)

(57) [Claims] 1. A pattern inspection apparatus for inspecting a pattern image to be inspected by comparing it with a reference pattern obtained from design data, wherein the design data is used to correct deformations that may occur in the pattern to be inspected. A unit for converting into a reference pattern, a storage unit for storing the reference pattern, an input unit for inputting an image of the inspection target pattern, a detection unit for detecting an edge of the inspection target pattern image, the inspection target pattern image An inspection means for inspecting the inspection target pattern by comparing the edge of the reference pattern with an edge of the reference pattern, and an output means for outputting the result of the inspection, the detection means for the inspection target pattern image. ,
A pattern inspection device characterized by detecting edges with sub-pixel accuracy. 2. The pattern inspection apparatus according to claim 1, wherein the detection unit detects an edge for each pixel in the inspection target pattern image. 3. The pattern inspection apparatus according to claim 1, wherein the inspection unit compares the edge of the inspection target pattern image with the edge of the reference pattern to detect the inspection target pattern image and the inspection target pattern image. A pattern inspection device characterized by performing matching with a reference pattern. 4. The pattern inspection apparatus according to claim 3, wherein the matching is performed by using a sum of products of an amplitude of an edge of the pattern image to be inspected and an amplitude of an edge of the reference pattern in each pixel as an evaluation value. A pattern inspection device characterized in that 5. The pattern inspection apparatus according to claim 3, wherein the matching calculates an evaluation value of matching in consideration of an edge direction of the inspection target pattern image and an edge direction of the reference pattern in each pixel. A pattern inspection apparatus characterized by being performed. 6. The pattern inspection apparatus according to claim 5, wherein the matching is a sum of inner products of an edge vector of the inspection target pattern image and an edge vector of the reference pattern in each pixel, or an absolute value of the inner product. Is performed as the evaluation value, the edge vector has the amplitude of the edge as its magnitude, and the direction of the edge as its direction. 7. The pattern inspection apparatus according to claim 1, wherein the inspection means associates an edge of each pixel of the reference pattern with an edge of each pixel of the inspection target pattern image. A pattern inspection device characterized by. 8. The pattern inspection apparatus according to claim 7, wherein the correspondence is a distance between an edge of each pixel of the reference pattern and an edge of each pixel of the inspection target pattern image, and a directional difference between both edges. A pattern inspection apparatus characterized in that 9. The pattern inspection apparatus according to claim 7, wherein the inspection unit forms an area based on an edge of the inspection target pattern image that cannot be associated with the area, and A pattern inspection apparatus characterized by being recognized as a defective area. 10. The pattern inspection apparatus according to claim 7, wherein the inspection unit forms an area based on an edge of the inspection target pattern image that can be associated with the area. A pattern inspection apparatus characterized by detecting an area having a non-uniform luminance distribution, and recognizing the area as a defective area. 11. The pattern inspection apparatus according to claim 9, wherein the inspection unit determines a defect type based on a geometric feature amount of the defect area. 12. The pattern inspection apparatus according to claim 9, wherein the inspection unit determines a defect type based on a feature amount related to the brightness of the defect area. . 13. The pattern inspection apparatus according to claim 7, wherein the inspection unit calculates a pattern deformation amount of the inspection target pattern with respect to the reference pattern. . 14. The pattern inspection apparatus according to claim 13, wherein the pattern deformation amount includes at least one of a positional deviation amount, a magnification variation amount, and a line width thickening amount. apparatus. 15. The pattern inspection apparatus according to claim 13 or 14, wherein the inspection means adds a pattern attribute to the reference pattern. 16. A pattern inspection apparatus for inspecting a pattern to be inspected by comparing it with a reference pattern, comprising: storage means for storing the reference pattern; input means for inputting an image of the pattern to be inspected; An inspection unit that inspects the inspection target pattern by comparing the edge of the pattern image with the edge of the reference pattern, and an output unit that outputs the inspection result, wherein the inspection unit is the inspection target pattern By comparing the edge of the image and the edge of the reference pattern, to perform the matching of the inspection target pattern image and the reference pattern, the matching, the edge of the inspection target pattern image, or the edge of the reference pattern , A pattern inspection apparatus characterized by performing weighting and expanding. 17. A pattern inspection device for inspecting an inspection target pattern image by comparing it with a reference pattern obtained from design data, wherein the design data is used to correct deformations that may occur in the inspection target pattern. By converting the reference pattern, storage means for storing the reference pattern, input means for inputting the image of the inspection pattern, by comparing the edge of the inspection pattern image and the edge of the reference pattern An inspection unit that inspects the inspection target pattern, and an output unit that outputs the inspection result, wherein the inspection unit compares the edge of the inspection target pattern image with the edge of the reference pattern, The inspection target pattern image and the reference pattern are matched, and the matching is performed by the inspection target pattern. Pattern inspection apparatus the direction of the direction and an edge of the reference pattern of the edge of the emission image in consideration and performs by calculating the evaluation value of the matching. 18. A pattern inspection device for inspecting an inspection target pattern image by comparing it with a reference pattern obtained from design data, wherein the design data is used to correct deformations that may occur in the inspection target pattern. By converting the reference pattern, storage means for storing the reference pattern, input means for inputting the image of the inspection pattern, by comparing the edge of the inspection pattern image and the edge of the reference pattern An inspection unit that inspects the inspection target pattern, and an output unit that outputs the inspection result, wherein the inspection unit compares the edge of the inspection target pattern image with the edge of the reference pattern, Matching the inspection target pattern image and the reference pattern, the matching is the reference pattern For each part, a pattern inspection apparatus and performs changing the degree contribute to matching. 19. A pattern inspection apparatus for inspecting a pattern to be inspected by comparing it with a reference pattern, comprising: a storage unit that stores the reference pattern; an input unit that inputs an image of the pattern to be inspected; A detection unit that detects an edge of a pattern image, an inspection unit that inspects the inspection target pattern by comparing an edge of the inspection target pattern image with an edge of the reference pattern, and an output that outputs the inspection result. And a means for creating a profile by checking a brightness value in a profile acquisition section in the inspection target pattern image, and detecting a predetermined point for each profile,
A pattern inspecting apparatus, wherein a curve approximation is performed on the detected point to obtain an edge of the inspection target pattern image. 20. A pattern inspection method for inspecting an inspection target pattern image by comparing it with a reference pattern obtained from design data, wherein the design data is used to correct deformations that may occur in the inspection target pattern. A step of converting into a reference pattern; an input step of inputting an image of the inspection target pattern; a detection step of detecting an edge of the inspection target pattern image; an edge of the inspection target pattern image; An inspection step of inspecting the inspection target pattern by comparing with an edge of the reference pattern, and an output step of outputting the inspection result, the detection step, for the inspection target pattern image, sub-pixel A pattern inspection method characterized by accurately detecting edges. 21. The pattern inspection method according to claim 20, wherein the detecting step detects an edge for each pixel in the inspection target pattern image. 22. The pattern inspection method according to claim 20, wherein the inspection step compares the inspection target pattern image with the edge of the inspection target pattern image by comparing the edge of the inspection target pattern image with the edge of the reference pattern. A pattern inspection method characterized by performing matching with a reference pattern. 23. The pattern inspection method according to claim 22, wherein the matching is performed by using a sum of products of an amplitude of an edge of the pattern image to be inspected and an amplitude of an edge of the reference pattern in each pixel as an evaluation value. A pattern inspection method characterized by the above. 24. The pattern inspection method according to claim 22, wherein in the matching, an evaluation value of the matching is calculated in consideration of an edge direction of the inspection target pattern image and an edge direction of the reference pattern in each pixel. A pattern inspection method characterized by being performed. 25. The pattern inspection method according to claim 24, wherein the matching is a sum of inner products of an edge vector of the inspection target pattern image and an edge vector of the reference pattern in each pixel, or an absolute value of the inner product. Is performed as the evaluation value, the edge vector has the amplitude of the edge as its magnitude, and the direction of the edge as its direction. 26. The pattern inspection method according to claim 20, wherein the inspection step comprises:
A pattern inspection method, wherein an edge of each pixel of the reference pattern is associated with an edge of each pixel of the inspection target pattern image. 27. The pattern inspection method according to claim 26, wherein the association is a distance between an edge of each pixel of the reference pattern and an edge of each pixel of the inspection target pattern image, and a direction difference between both edges. A pattern inspection method characterized in that 28. The pattern inspection method according to claim 26, wherein in the inspection step, an area is formed based on an edge of the inspection target pattern image that cannot be associated with the area, and the area is defined. A pattern inspection method characterized by recognizing a defective area. 29. The pattern inspection method according to claim 26, wherein the inspection step comprises:
Areas are formed based on the edges of the pattern image to be inspected that can be associated, the areas in which the luminance distribution is non-uniform are detected, and the areas are recognized as defective areas. A pattern inspection method characterized by: 30. The pattern inspection method according to claim 28, wherein the inspection unit determines a defect type based on a geometric feature amount of the defect area. 31. The pattern inspecting method according to claim 28, wherein the inspecting unit determines a defect type based on a feature amount related to the brightness of the defect area. . 32. The pattern inspection method according to claim 26, wherein the inspection unit calculates a pattern deformation amount of the inspection target pattern with respect to the reference pattern. . 33. The pattern inspection method according to claim 32, wherein the pattern deformation amount includes at least one of a positional deviation amount, a magnification variation amount, and a line width thickening amount. Method. 34. The pattern inspection method according to claim 32 or 33, wherein the inspection step adds a pattern attribute to the reference pattern. 35. A pattern inspection method for inspecting an inspection target pattern by comparing it with a reference pattern, comprising an input step of inputting an image of the inspection target pattern, an edge of the inspection target pattern image, and storing in a storage means. By comparing the edge of the reference pattern that has been performed, an inspection step of inspecting the inspection target pattern, and an output step of outputting the result of the inspection, the inspection step, the edge of the inspection target pattern image By comparing the edge of the reference pattern with the reference pattern, the matching of the inspection target pattern image and the reference pattern, the matching, for the edge of the inspection target pattern image, or the edge of the reference pattern, weighting. The pattern inspection method is characterized in that it is performed after being expanded. 36. A pattern inspection method for inspecting an inspection target pattern image by comparing it with a reference pattern obtained from design data, wherein the design data is corrected by correcting a deformation that may occur in the inspection target pattern. The step of converting into a reference pattern, the input step of inputting the image of the inspection target pattern, the edge of the inspection target pattern image, and the edge of the reference pattern stored in the storage means, the inspection An inspection step for inspecting a target pattern, and an output step for outputting the result of the inspection, wherein the inspection step compares the edge of the inspection target pattern image with the edge of the reference pattern to obtain the inspection target. The pattern image and the reference pattern are matched, and the matching is performed on the inspection target pattern. A pattern inspection method, which is performed by calculating a matching evaluation value in consideration of an edge direction of a turn image and an edge direction of the reference pattern. 37. A pattern inspection method for inspecting an inspection target pattern image by comparing it with a reference pattern obtained from design data, wherein the design data is corrected by correcting deformation that may occur in the inspection target pattern. The step of converting into a reference pattern, the input step of inputting the image of the inspection target pattern, the edge of the inspection target pattern image, and the edge of the reference pattern stored in the storage means, the inspection An inspection step for inspecting a target pattern, and an output step for outputting the result of the inspection, wherein the inspection step compares the edge of the inspection target pattern image with the edge of the reference pattern to obtain the inspection target. The pattern image and the reference pattern are matched, and the matching is performed by the reference pattern. The pattern inspection method is characterized in that the degree of contribution to matching is changed for each part of the pattern. 38. A pattern inspection method for inspecting an inspection target pattern by comparing it with a reference pattern, the input step of inputting an image of the inspection target pattern, and a detection step of detecting an edge of the inspection target pattern image. An inspection step of inspecting the inspection target pattern by comparing an edge of the inspection target pattern image with an edge of the reference pattern stored in a storage unit; and an output step of outputting a result of the inspection. In the detecting step, a profile is created by checking the brightness value in the profile acquisition section of the pattern image to be inspected, a predetermined point is detected for each profile, and curve approximation is performed on the detected point to perform the inspection. A pattern inspection method, characterized in that it is an edge of a target pattern image. 39. A computer-readable recording medium recording a program for causing a computer to execute the pattern inspection method according to any one of claims 20 to 38.