JP3466286B2 - Pattern inspection method and pattern inspection device - 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 method and a pattern inspection apparatus used directly in this pattern inspection method, and more particularly to one using image processing.

With the automation of industry due to the widespread use of computers, there is an increasing number of cases in which a pattern inspection method using image processing is used to perform a pattern inspection of a printed wiring board or the like in the assembly and inspection steps.

In this pattern inspection method, it is necessary to perform more accurate inspection in order to reduce the inspection time by eliminating the inspection relying on manpower.

[0004]

2. Description of the Related Art FIG. 21 is a block diagram showing two examples of a conventional pattern inspection apparatus.

The pattern inspection apparatus shown in FIG. 21A comprises a radial length measuring means 202, a radial coding means 203, a defect extracting means 204 and a defect information outputting means 205.

The radial length measuring means 202 is a detection optical system 2
The image data of the inspection pattern 201 of the object to be inspected cut out with 00 is measured in eight directions according to the continuity of pixels.

The radial coding means 203 detects the approximate center of the pattern from the length measurement values in the eight directions and performs radial coding according to preset parameters.

The defect extracting means 204 uses the inspection pattern 20.
In the window at the predetermined position of 1, the radial code generated by the radial encoding means 203 is compared with the defective code in the defect dictionary, and if the result of this comparison is that the radial code exists in the defect dictionary, this The code is extracted as a defect extraction code.

As a result, defect information is output from the defect information output means 205, and the inspection pattern 201 is judged to be a defect pattern.

Further, the pattern inspection apparatus of FIG. 21B has a position aligning means 212 and a difference pattern extracting means 21.
3, differential pattern length measuring means 214, and defect extracting means 215
And defect information output means 216.

The alignment means 212 aligns the inspection pattern 201 of the inspection object with the CAD pattern 211.

The difference pattern extraction means 213 is configured to detect the aligned inspection pattern 201 and CAD pattern 211.
Extract the difference pattern with.

The difference pattern length measuring means 214 measures the length of the difference pattern.

The defect extracting means 215 extracts the measured difference pattern as a defect extraction area according to a predetermined size.

As a result, the defect information output means 216 outputs defect information, and the inspection pattern 201 is judged to be a defect pattern.

[0016]

However, as shown in FIG.
In the conventional pattern inspection apparatus shown in (A), firstly, when a defect and a normal shape have the same shape, both of them are detected, and the defect is excessively detected.

More specifically, as shown in FIG. 22A, it is impossible to distinguish between the wire breakage 300 and the normal termination 301.
Both were detected. Further, the short circuit 302 and the normal T branch could not be distinguished, and both were detected.

Second, there is a problem that it is difficult to detect macroscopic defects.

More specifically, as shown in FIG. 22B, a peeled portion 306 is formed on the land with lead 305, but since the radial code is the same as 304,
When the peeled portion 306 had a large area, it could not be detected.

Thirdly, there is a problem that it is impossible to inspect a pattern exceeding the length measurement limit.

More specifically, as shown in FIG. 22C, when the land portion of the land with lead 307 exceeds the length measurement limit, the center cannot be detected and the defect cannot be detected.

On the other hand, the pattern inspection apparatus shown in FIG. 21B has a fourth problem that it is difficult to detect micro defects.

More specifically, as shown in FIG. 23 (A), in the difference pattern 312 between the inspection pattern 310 and the CAD pattern 311, if the defect judgment size is set to 2 × 2 pixels or more, disconnection that causes a micro defect will occur. The defective portion 313 is also missed.

In order to prevent this, the judgment size is set to 1 ×
When set to 1 pixel, the difference pattern 31 that does not become a defect
The pattern edge portion 314 of No. 2 was also detected as a defect, and the defect was detected excessively.

Fifth, there is a problem that the defect detection performance changes depending on the defect shape or the matching position.

More specifically, as shown in FIG. 23B, the inspection standard (defect determination size) is set to 1 of the design pattern width W.
A defect of / 3 or more should be considered as a defect.

In the case of (a), the width of the chip 320a is 3
Since it is W / 8, which is longer than W / 3, it is detected as a defect.

In the case of (b), the chips 321a and 321 are missing.
Since the width of b is W / 4, which is shorter than W / 3, it is not detected as a defect.

In the case of (C), the width of the chip 322a is W
Although it is not detected as a defect at / 8, it is detected as a defect because the width of the chip 322b is 3W / 8, which is longer than W / 3.

As described above, the defect detection performance depends on the defect shape of the inspection pattern or the matching position between the inspection pattern and the CAD pattern.

Therefore, the present invention has been made in view of the above points, and detects both micro defects and macro defects, prevents excessive detection of defects, and prevents changes in defect detection performance. It is an object of the present invention to provide a pattern inspection method and a pattern inspection apparatus capable of performing the above.

[0032]

According to a first aspect of the present invention, there is provided a pattern inspection method for detecting a defect occurring in an inspection pattern of an object to be inspected, wherein the reference size is set in accordance with a predetermined reference size set in advance. Includes a micro defect detection process for detecting defects of small size, and a macro defect detection process for detecting defects of size above the reference size
The size is the maximum that exists in the reference pattern of the above-mentioned inspection object.
The minimum allowable value for thin patterns is also used to detect the above micro defects.
The process includes a first micro defect primary treatment process and a first micro defect treatment process.
The second micro-defect 1 including the defect secondary treatment step.
The next processing step is the image data of the inspection pattern of the inspection object.
Data based on the pixel continuity and the inspection code
The defect code of the generated and registered defect pattern and
Compared with the inspection code, size more than the above standard size
Inspection code that matches the defect code corresponding to a small defect in
In this configuration, the code is extracted as a defect candidate extraction code.
The first microdefect secondary processing step is applied to the above inspection pattern.
The image data of the reference pattern at the corresponding position
Generate a reference code by encoding based on continuity, and
All reference codes within the specified range based on the position
And the defect candidate extraction code are compared with each other to obtain the reference code.
The defect candidate extraction code that does not match the
Detect defects by extracting them as true defect codes corresponding to pits
It is characterized by having a configuration .

[0033]

[0034]

[0035]

[0036]

Further, in the invention of claim 2 , the microdefect detecting step is constituted by a second microdefect primary processing step and a second microdefect secondary processing step. In the next processing step, the image data of the inspection pattern of the inspection object is coded based on the continuity of pixels to generate an inspection code, and a category code indicating the type of pattern shape corresponding to various codes is set to the inspection code. The inspection category code is obtained by searching based on the above, and the category code of the defect pattern registered in advance is compared with the above inspection category code, and the inspection category code that matches the category code corresponding to the defect having a size smaller than the above reference size is obtained. The second micro-defect secondary processing step is performed by extracting as a defect candidate extraction code, and the reference pattern at the position corresponding to the inspection pattern is used. Image data is encoded based on pixel continuity to generate a reference code, and the category code indicating the type of pattern shape corresponding to each code is searched based on the above reference code to obtain the reference category code. Comparing all of the reference category codes within the predetermined range with the position as a reference and the defect candidate extraction category codes, a defect candidate extraction category code that does not match the reference category code corresponds to an unacceptable true defect. The pattern inspection method is characterized in that the true defect code is extracted and the defect is detected.

According to a third aspect of the present invention, the second micro-defect secondary processing step is performed by encoding image data of the reference pattern at a position corresponding to the inspection pattern based on pixel continuity. Generate a code, search for a category code indicating the type of pattern shape corresponding to each code based on the reference code to obtain a reference category code, and set a limited range according to the position and the type of the shape of the pattern. All or some of the reference category codes in the above and the defect candidate extraction category code are compared, and the defect candidate extraction category code that does not match the reference category code is set as the true defect code corresponding to the unacceptable true defect. The pattern inspection method is characterized in that it is replaced with a configuration for extracting and detecting defects.

According to the fourth aspect of the present invention, in the second micro defect secondary processing step, the relationship between the defect candidate extraction category code and the target category is classified, and the pattern shape corresponding to the classification is classified. Stores various upper category codes indicating types, based on the relationship between the defect candidate extraction category code and a specific category that is a target in the category distribution in a predetermined area centered on the defect candidate extraction category code, It is characterized in that the upper category code corresponding to this relationship is selected and output, and based on the upper category code, the true defect code corresponding to the unacceptable true defect is determined and the defect is detected. This is a pattern inspection method.

Further, in the invention of claim 5 , the second micro-defect secondary processing step detects a plurality of categories radially distributed from the center of the defect candidate extraction category code, and detects the category. By correlating a plurality of categories with each other and processing them as data of a hierarchical network structure, an upper category is output, and a true defect code corresponding to an unacceptable true defect is determined based on the upper category code. The pattern inspection method is characterized in that it is replaced with a configuration for detecting defects.

According to a sixth aspect of the invention, in the macro defect detecting step, the inspection pattern of the inspection object and the reference pattern are aligned, and the inspection pattern of the inspection object and the reference pattern are aligned. A difference pattern with a pattern is extracted, the difference pattern is measured, the measured length value is compared with the reference size, and an area of the reference size or more is extracted as a defect extraction area to detect a defect. It is a pattern inspection method characterized by being configured to detect.

[0042]

Further, the invention of claim 7 relates to the above macro defect.
The detection process includes a macro defect primary treatment process and a macro defect secondary process.
The above-mentioned macro defect primary treatment process
Alignment of the inspection pattern with the reference pattern
The inspection pattern of the above-mentioned inspected object
Pattern and the difference pattern between the above reference pattern and
The difference pattern is measured and the measured value is
Comparing with the reference size, defective areas larger than the reference size
The macro defect secondary processing step is configured to extract a candidate extraction region, and a pattern of a critical region in which a defect cannot be accepted is created based on the reference pattern, and a pattern of a critical region at a position corresponding to the defect candidate extraction region is created. The pattern inspection method is characterized in that the defect is detected by extracting a case where a position corresponding to the defect candidate extraction area exists in the critical area as a true defect area with reference to.

Further, the invention of claim 8 is a pattern inspection method, wherein the fatal area is provided with a constant width in each of an outer direction and an inner direction from a peripheral edge of each pattern of the reference pattern. .

The ninth aspect of the present invention is the pattern inspection method, wherein the constant width of the critical area is set to different values in the outer direction and the inner direction.

Further, the invention of claim 10 is a pattern inspection method, characterized in that the constant width of the critical area is changed and set in advance according to the type of the pattern shape.

Further, the invention of claim 11 measures the pattern interval of the reference pattern of the fatal area, and extracts an area in which the pattern interval is smaller than a predetermined value as a first area. Then, the pattern width of the reference pattern is measured, and an area in which the pattern width is smaller than a predetermined value is extracted as the second area.

[0048]

[0049]

[0050]

[0051]

The twelfth aspect of the invention is to detect an object to be inspected.
Pattern inspection system for detecting defects in inspection patterns
In, in accordance with a predetermined reference size set in advance,
A micro that detects defects smaller than the reference size
Defect detection means and defects larger than the above standard size
Equipped with macro defect detection means to detect
Is the thinnest existing in the reference pattern of the object to be inspected.
The minimum allowable value of the pattern is set, and the microdefect detecting means is used.
The first microdefect primary treatment means and the first microdefect
The first micro-defect primary treatment is constituted by a second treatment means.
Image data of the inspection pattern of the inspection object
Generate inspection code by encoding based on pixel continuity
Inspection code generating means and a defect pattern registered in advance
The defect code of the
Defect code for defects smaller than sub-size
Extract inspection code that matches with as defect candidate extraction code
More becomes configured and micro defect candidate extracting means, said first micro defects secondary processing means, encoding based on the continuity of the pixel image data of the reference pattern in the position corresponding to the test pattern reference to A reference code generating means for generating a code is compared with all the reference codes within a predetermined range based on the position and the defect candidate extraction code, and a defect candidate extraction code that does not match the reference code cannot be accepted. A pattern inspection apparatus characterized by comprising a microdefect extraction means for extracting a defect by detecting it as a true defect code corresponding to a true defect, and a microdefect extraction means.

In the thirteenth aspect of the present invention, the microdefect detecting means comprises a second microdefect primary processing means and a second microdefect secondary processing means, and the second microdefect 1 Next processing means, inspection code generation means for generating inspection code by encoding image data of the inspection pattern of the inspection object based on the continuity of pixels, and category code indicating the type of pattern shape corresponding to various codes. By comparing the inspection category code with the inspection category code acquisition unit that obtains the inspection category code by searching the inspection category code and the category code of the defect pattern registered in advance, and detects a defect having a size smaller than the reference size. And a micro-defect candidate extraction means for extracting an inspection category code that matches the corresponding category code as a defect candidate extraction code. The second micro-defect secondary processing means, and a reference code generation means for generating a reference code by encoding image data of the reference pattern at a position corresponding to the inspection pattern based on pixel continuity. ,
Reference category code acquisition means for obtaining a reference category code by searching a category code indicating a type of pattern shape corresponding to each code based on the reference code, and all the reference categories within a predetermined range based on the position A micro defect extracting means for comparing a code with the defect candidate extraction category code and extracting a defect candidate extraction category code that does not match the reference category code as a true defect code corresponding to an unacceptable true defect to detect a defect. And a pattern inspection apparatus having the following configuration.

According to a fourteenth aspect of the present invention, the second micro-defect secondary processing means encodes the image data of the reference pattern at a position corresponding to the inspection pattern on the basis of the continuity of the pixels to form a reference. Reference code generation means for generating a code, reference category code acquisition means for obtaining a reference category code by searching a category code indicating a type of pattern shape corresponding to each code based on the reference code, the position and the pattern Depending on the type of shape, all or some of the reference category codes in the limited range and the defect candidate extraction category code are compared, and the defect candidate extraction category code that does not match the reference category code, Micro defect extraction means for detecting defects by extracting as true defect codes corresponding to unacceptable true defects Is a pattern inspection apparatus being characterized in that instead of the more becomes configurations.

According to a fifteenth aspect of the invention, the second micro-defect secondary processing means classifies the relationship between the defect candidate extraction category code and the target category, and the pattern shape corresponding to this classification is obtained. Stores various upper category codes indicating types, and the above defect candidate extraction category code,
A higher category conversion means for selecting and outputting a higher category code corresponding to this relationship based on the relationship with a particular category of interest in a category distribution within a predetermined area centered on the defect candidate extraction category code; The pattern inspection apparatus is characterized in that it is replaced with a microdefect extracting means that determines a true defect code corresponding to an unacceptable true defect based on the upper category code and detects the defect.

According to a sixteenth aspect of the present invention, the second micro-defect secondary processing means has a radial category detecting means for detecting a plurality of categories radially distributed from the center of the defect candidate extraction category code. And, the detected plurality of categories are correlated with each other and processed as data of a hierarchical network structure, and a neural network that outputs a higher category and, based on the higher category code, correspond to unacceptable true defects. The pattern inspection apparatus is characterized in that the structure is replaced with a microdefect extraction means for determining a true defect code to detect a defect.

According to a seventeenth aspect of the present invention, the macro defect detecting means includes an aligning means for aligning the inspection pattern of the inspection object with a reference pattern, and an inspection of the aligned inspection object. A difference pattern extracting means for extracting a difference pattern between the pattern and the reference pattern, a difference pattern length measuring means for measuring the difference pattern, and comparing the measured length value with the reference size, A pattern inspection apparatus comprising: a macro defect extraction unit that detects a defect by extracting a region having a size equal to or larger than a reference size as a defect extraction region.

[0058]

According to the eighteenth aspect of the present invention, the macro defect secondary processing means includes a fatal area pattern creating means for creating a pattern of a fatal area in which a defect cannot be accepted based on the reference pattern, and the defect candidate extraction area. Macro defect extraction means for detecting a defect by extracting a case where a position corresponding to the defect candidate extraction region exists in the fatal region as a true defect region by referring to the pattern of the fatal region corresponding to This is a pattern inspection apparatus having a configuration.

A nineteenth aspect of the present invention is the pattern inspection apparatus, wherein the fatal area is provided with a constant width in each of the outer and inner directions from the peripheral edge of each pattern of the reference pattern. .

According to a twentieth aspect of the invention, there is provided a pattern inspection apparatus characterized in that the constant width of the critical area is set to different values in the outer direction and the inner direction.

According to a twenty-first aspect of the present invention, there is provided a pattern inspection apparatus characterized in that the constant width of the critical area is changed and set in advance according to the type of the pattern shape.

In the invention of claim 22 , the fatal area is measured by measuring the pattern interval of the reference pattern, and the area in which the pattern interval is smaller than a predetermined value is extracted as the first area. Then, the pattern width of the reference pattern is measured, and an area in which the pattern width is smaller than a predetermined value is extracted as a second area, and the area is extracted from the first area and the second area. The pattern inspection apparatus is characterized in that at least one of the above is calculated and provided.

[0064]

Since the pattern inspection method according to the first aspect of the present invention is configured to include the micro defect detection step and the macro defect detection step, in the micro defect detection step, according to a predetermined reference size set in advance, detecting a small defect in size than the reference size, the macro defect detection process, acts to detect defects of the reference size or larger size, the upper
The reference size exists in the reference pattern of the above-mentioned inspection object.
The minimum allowable value for the thinnest pattern
Defects smaller than the minimum value are detected and
It acts to detect defects with a size above a small value, and
The image data of the inspection pattern of the inspection object is used as pixel continuity.
Generate the inspection code by coding based on the
The defect code of the defective pattern and the above inspection code are compared.
Compared to defects smaller than the above standard size
The inspection code that matches the defect code
As a micro defect.
The defect detection process includes the first microdefect primary treatment process and the first microdefect primary treatment process.
Since it is composed of the micro defect secondary treatment process, the first Miku
(B) In the defect primary treatment process, the inspection pattern of the above-mentioned inspection object
Image data is encoded and detected based on pixel continuity.
Inspection code is generated and the defect of the registered defect pattern
Compare the code with the above inspection code, and use the above standard size
Match the defect code corresponding to the smaller size defect
To extract the inspection code as a defect candidate extraction code.
In the first micro-defect secondary treatment step,
Image data of the reference pattern at the position corresponding to the inspection pattern.
Data is encoded based on pixel continuity and the reference code
Is generated and all within a predetermined range based on the above position
Comparing the reference code and the above defect candidate extraction code,
Allows defect candidate extraction code that does not match the reference code
Extract as a true defect code corresponding to a true defect that cannot
It acts to detect defects.

[0065]

[0066]

[0067]

[0068]

[0069]

Further, in the pattern inspection method according to the second aspect of the present invention, the microdefect detecting step is constituted by the second microdefect primary processing step and the second microdefect secondary processing step. In the microdefect primary processing step, the image data of the inspection pattern of the inspection object is coded based on the continuity of the pixels to generate an inspection code, and a category code indicating the type of pattern shape corresponding to each code is generated. Search based on the above inspection code to get inspection category code,
The category code of the defect pattern registered in advance is compared with the inspection category code, and the inspection category code matching the category code corresponding to the defect having a size smaller than the reference size is extracted as a defect candidate extraction code. .

In the second micro defect secondary treatment step,
Image data of the reference pattern at the position corresponding to the inspection pattern is coded based on pixel continuity to generate a reference code, and a category code indicating the type of pattern shape corresponding to each code is based on the reference code. To obtain a reference category code, compare all the reference category codes in the predetermined range with the above-mentioned position with the defect candidate extraction category code, and detect the defect candidate extraction category that does not match the reference category code. It acts to extract the code as a true defect code corresponding to an unacceptable true defect and detect the defect.

In the second micro-defect secondary processing step of the pattern inspection method according to the third aspect of the present invention, the image data of the reference pattern at the position corresponding to the inspection pattern is coded based on the continuity of pixels. To generate a reference code, to obtain a reference category code by searching the category code indicating the type of pattern shape corresponding to various codes based on the reference code, depending on the position and the type of the shape of the pattern, All or some of the reference category codes within a limited range are compared with the defect candidate extraction category codes, and defect candidate extraction category codes that do not match the reference category code are identified as true defects corresponding to unacceptable true defects. It acts to detect defects by extracting them as defect codes.

In the second micro-defect secondary processing step of the pattern inspection method according to the fourth aspect of the present invention, the relationship between the defect candidate extraction category code and the target category is classified and the classification is dealt with. Various upper category codes indicating types of pattern shapes are stored, and the relationship between the defect candidate extraction category code and a specific category that is a target in a category distribution in a predetermined area centered on the defect candidate extraction category code is stored. Based on this, the upper category code corresponding to this relationship is selected and output, and the true defect code corresponding to the unacceptable true defect is determined and the defect is detected based on the upper category code.

In the second micro defect secondary processing step of the pattern inspection method according to the fifth aspect of the present invention, a plurality of categories radially distributed from the center of the defect candidate extraction category code are detected, By processing the detected plurality of categories as data of a hierarchical network structure by associating them with each other, a higher category is output, and a true defect code corresponding to an unacceptable true defect is output based on the higher category code. It acts to judge and detect defects.

In the macro defect detecting step of the pattern inspection method according to the sixth aspect of the present invention, the inspection pattern of the inspection object and the reference pattern are aligned, and the inspection pattern of the aligned inspection pattern of the inspection object. The difference pattern between the reference pattern and the reference pattern is extracted, the difference pattern is measured, the measured length value is compared with the reference size, and an area of the reference size or more is extracted as a defect extraction area. Acts to detect defects.

Further, in the pattern inspection method according to the invention of claim 11, the macro defect detecting step is constituted by the macro defect primary processing step and the macro defect secondary processing step. The inspection pattern and the reference pattern of the inspection object are aligned, the difference pattern between the inspection pattern and the reference pattern of the aligned inspection object is extracted, and the difference pattern is measured,
This length measurement value is compared with the reference size, and an area having a size equal to or larger than the reference size is extracted as a defect candidate extraction area.

In the macro defect secondary processing step, the true defect area corresponding to the unacceptable true defect is extracted from the defect candidate extraction area to detect the defect.

In the macro defect secondary processing step of the pattern inspection method according to the seventh aspect of the present invention, a pattern of a critical area in which a defect cannot be accepted is created based on the reference pattern and corresponds to the defect candidate extraction area. By referring to the pattern of the critical area of the position, the case where the position corresponding to the defect candidate extraction area exists in the critical area is extracted as a true defect area to detect the defect.

[0079] The pattern inspection method according to the invention of claim 8, 9, 10 lethal regions, in each of outwardly and inwardly from the peripheral edge of the pattern of the reference pattern,
Since it is provided with a constant width, the length measurement of the reference pattern is unnecessary.

Further, in the pattern inspection method according to the invention of claim 11 , the critical area is measured for the pattern interval of the reference pattern, and the area where the pattern interval is smaller than a predetermined value is first Area, the pattern width of the reference pattern is measured, and an area having a pattern width smaller than a predetermined value is extracted as a second area, and the first area and the second area are extracted. At least one of the areas is calculated and provided.

Therefore, the first region acts as a dead region where patterns may come into contact with each other, and the second region acts as a dead region where patterns may be broken.

[0082]

[0083]

[0084]

[0085]

[0086] Also, the pattern inspection apparatus according to the invention of claim 12, the first micro defects secondary processing means, since the more becomes configured as a reference code generating means and micro defect extraction means, the invention of claim 1 It acts so as to be directly used for implementing the pattern inspection method according to the present invention.

In the pattern inspection apparatus according to the thirteenth aspect of the present invention, the microdefect detecting means is used as the second microdefect 1
Since the secondary processing means and the second micro-defect secondary processing means are provided, the secondary processing means acts directly to implement the pattern inspection method according to the second aspect of the invention.

In the pattern inspection apparatus according to the fourteenth aspect of the present invention, the second micro defect secondary processing means is constituted by the reference code generating means, the reference category code acquiring means and the micro defect extracting means. It acts so as to be directly used for carrying out the pattern inspection method according to the invention of claim 3 .

[0089] Also, the pattern inspection apparatus according to the invention of claim 15, the second micro defects secondary processing means, since the more becomes constituting an upper category conversion means and micro defect extraction means, the invention of claim 4 It acts so as to be directly used for implementing the pattern inspection method according to the present invention.

In the pattern inspection apparatus according to the sixteenth aspect of the present invention, the second micro-defect secondary processing means is composed of the radial category detection means, the neural network and the micro-defect extraction means. It acts so as to be directly used for carrying out the pattern inspection method according to the fifth invention.

In the pattern inspection apparatus according to the seventeenth aspect of the present invention, the macro defect detecting means is constituted by the aligning means, the difference pattern extracting means, the difference pattern length measuring means and the macro defect extracting means. It acts so as to be directly used for implementing the pattern inspection method according to the invention of claim 6 .

[0092]

[0093] Also, the pattern inspection apparatus according to the invention of claim 18, the macro-defect secondary treatment unit, a fatal region pattern creating means, and the macro defect extraction means, since the more becomes configuration, the invention of claim 7 It acts so as to be directly used for implementing the pattern inspection method according to the present invention.

[0094] Also, the pattern inspection apparatus according to the invention of claim 19, the critical region, in each of the outward and inward from the periphery of each pattern of the reference pattern, so providing a constant width, claim 8 The present invention operates so as to be directly used for implementing the pattern inspection method according to the invention.

Further, in the pattern inspection apparatus according to the invention of claim 20 , the constant width of the critical area is set to have different values in the outward direction and the inward direction. Therefore, the pattern according to the invention of claim 9 Acts as it is used directly to carry out the test method.

[0096] Also, the pattern inspection apparatus according to the invention of claim 21, the constant width of the critical region, depending on the type of the pattern shape, in advance, since the setting is changed, the invention of claim 10 It acts as if it were used directly to implement such a pattern inspection method.

The pattern inspection apparatus according to the invention of claim 22 measures the pattern interval of the reference pattern in the critical area, and determines an area in which the pattern interval is smaller than a predetermined value. The first pattern is extracted as the first region, the pattern width of the reference pattern is measured, and the region in which the pattern width is smaller than a predetermined value is extracted as the second region. Since at least one of the two areas is calculated and provided,
It acts so as to be directly used for carrying out the pattern inspection method according to the invention of claim 11 .

[0098]

1 is a block diagram of the first embodiment of the present invention.

The pattern inspection apparatus according to the first embodiment detects a defect occurring in an inspection pattern such as a wiring pattern of a printed wiring board to be inspected.

The structure of the pattern inspection apparatus according to the first embodiment is such that the micro defect detecting means 3 and the macro defect detecting means 1 are
And 03.

The microdefect detecting means 3 comprises an inspection code generating means 4 composed of a radial length measuring means 5a and a radial coding means 5b, and a microdefect extracting means 6.

The radial length measuring means 5a checks the continuity of the pattern image in the image data of the inspection pattern 2 such as the printed wiring board wiring pattern created by the detection optical system 1,
Radial length measurement is performed by obtaining the length measurement value according to this continuity.

The detection optical system 1 has a photographing device, a signal processing circuit, a quantizing circuit, an image memory and a clipping circuit.

In this detection optical system 1, a video signal of an inspection pattern such as a printed wiring board obtained by a photographing device is subjected to analog processing such as amplification by a signal processing circuit and supplied to the quantization circuit as an analog signal. To do.

The analog signal is converted into a digital signal by the quantizing circuit. The digital signal is written in the image memory under the control of the CPU (central processing unit).

The cutout circuit cuts out the inspection pattern 2 to be image data from the image data in the image memory in a window having a size in consideration of the size of the pattern to be inspected and the size of the image.

In the wiring pattern as the inspection pattern 2, copper residue 2a, disconnection 2b, chip (micro) 2c and chip (macro) 2d may occur.

The radial length measuring means 5a uses the inspection pattern 2
The center of each pattern is set to the length measuring point O, and the length is measured from the length measuring point O in 8 directions every 45 °. The broken line X is the measurement result.
In the length measurement, the length (the number of pixels) of continuous pixels of the same data as the length measurement point O is calculated. By this length measurement, length measurement results in eight directions are obtained.

The radial coding means 5b detects the approximate center of the pattern from the length measurement value obtained by the radial length measuring means 5a, performs radial coding according to preset parameters, and generates an inspection code.

For example, by setting a plurality of threshold values, it is determined which range the length measurement value is within the threshold value, and coding is performed according to the determination result obtained from the length measurement values in eight directions to generate an inspection code. To do.

As a result, the inspection code generating means 4 encodes the image data of the inspection pattern 2 on the basis of the continuity of the pixels, and each of the copper residue 2a, the disconnection 2b, the chip (micro) 2c and the chip (macro) 2d. Generate an inspection code.

The micro defect extraction means 6 compares the inspection code with the defect code of the defect pattern registered in the defect dictionary in advance.

The defect dictionary has a radial code 5b.
A copper residue 2a, which is a typical defect generated by the above, and becomes a code of a defect pattern having a size smaller than the allowable minimum value W,
Each defect code such as the disconnection 2b and the chipping (micro) 2c is registered in advance.

The defect code corresponding to the missing (macro) 2d is not registered in the defect dictionary. This is because the pattern width D1 is equal to the width D2 and the radial codes are also equal.

The allowable minimum value W is a reference pattern (CAD (Computer Aided Design) data 1 such as a wiring pattern of a printed wiring board having a predetermined reference size set in advance.
This is the minimum allowable value of the thinnest pattern existing in the CAD pattern 102) created from 01.

The reference pattern may be a pattern obtained from a wiring pattern of a printed wiring board which has been confirmed to have no defects.

On the CAD pattern 102, a bitmap pattern 102a at the design stage is formed.

The allowable minimum width W is α% of the minimum line width of the bitmap pattern 102 in the CAD pattern 12.
It is decided on the basis of the case where it does not become smaller than the above.

For example, the minimum line width is 150 μm,
When α is 33, the allowable minimum value W is about 100 μm. The defects having a size of less than 100 μm are detected by the micro defect detecting means 3.

Therefore, the inspection codes for the copper residue 2a, the wire break 2b, the chip (micro) 2c and the chip (macro) 2d are respectively registered to the copper residue 2a, the wire 2b and the chip 2c registered in the defect dictionary. It is compared with the defect code.

As a result, the inspection codes of the copper residue 2a, the disconnection 2b, and the chip (micro) 2c, which are the inspection codes corresponding to the defect codes smaller than the allowable minimum value W, are extracted as the micro defect extraction codes. To be done.

Further, the inspection code of the defect (macro) 2d does not match the defect code corresponding to the defect having a size smaller than the allowable minimum value W, and therefore is not extracted as the micro defect extraction code.

Then, each microdefect extraction code of the copper residue 2a, the disconnection 2b, and the chipped (micro) 2c is supplied to the defect information output means 7 as microdefect information.

Therefore, the defect information output means 7 outputs each micro defect information of the copper residue 2a, the disconnection 2b, and the chip (micro) 2c.

Next, the macro defect detecting means 103 comprises a positioning means 104, a difference pattern extracting means 105, a difference pattern length measuring means 106 and a macro defect extracting means 107.

The alignment means 104 uses the inspection pattern 2
And the CAD pattern 102 are aligned.

The difference pattern extraction means 105 extracts the difference pattern between the aligned inspection pattern 2 and CAD pattern 102.

As a result, patterns corresponding to the copper residue 2a, the disconnection 2b, the chip (micro) 2c and the chip (macro) 2d are obtained.

The difference pattern length measuring means 106 is, for example,
The continuity of the extracted difference pattern image is checked by the same method as the radial length measurement 5a, and the length measurement value corresponding to this continuity is obtained.

As a result, length measurement values corresponding to the copper residue 2a, the wire break 2b, the chip (micro) 2c and the chip (macro) 2d are obtained.

The macro defect extracting means 107 compares each of the length measurement values with the permissible minimum value W and extracts an area equal to or larger than the permissible minimum value W as a macro defect extraction area.

As a result, the area of the chip (macro) 2d having the allowable minimum value W or more is extracted as the macro defect extraction area. Further, the areas of the copper residue 2a, the disconnection 2b, and the chipped (micro) 2c that are less than the allowable minimum value W are not extracted as macro defect extraction areas.

The macro defect extraction area of the defect (macro) 2d is recorded as defect information output means 7 in the defect information output means 7.
Is supplied to.

Therefore, the defect information output means 7 outputs the macro defect information of the missing (macro) 2d.

Therefore, the defect information output means 7 has the micro defect information of the copper residue 2a, the disconnection 2b and the chip (micro) 2c supplied from the micro defect detection means 3 and the chip defect supplied from the macro defect detection means 103. Both (macro) 2d macro defect information is output.

Therefore, it is possible to detect all the defects occurring in the inspection pattern 2 such as the wiring pattern of the printed wiring board.

Next, a second embodiment of the present invention shown in FIG. 2 will be described.

The pattern inspection apparatus according to the second embodiment prevents the detection of allowable defects.

The structure of the pattern inspection apparatus according to the second embodiment is such that the micro defect primary processing 8 serving as the first micro defect primary processing means and the micro defect secondary processing serving as the first micro defect secondary processing means. It comprises a micro defect detecting means 3 having a process 10 and a macro defect detecting means 103 having a macro defect primary process 108 and a macro defect secondary process 109.

The microdefect primary processing means 8 is composed of an inspection code generating means 4 including a radial length measuring means 5a and a radial coding means 5b, and a microdefect candidate extracting means 9.

The micro defect candidate extraction means 9 compares the inspection code generated by the radial coding means 5b of the inspection code generation means 4 with the defect code of the defect pattern registered in the defect dictionary in advance, and the allowable minimum value is obtained. An inspection code that matches a defect code corresponding to a defect having a size smaller than W is extracted as a micro defect candidate extraction code.

The micro defect secondary processing means 10 extracts the micro true defect code corresponding to the unacceptable true defect from the micro defect candidate extraction code, and does not extract the micro permissible defect code corresponding to the allowable defect.

Then, the micro true defect code is supplied to the defect information output means 7 as micro defect information.

Therefore, the defect information output means 7 outputs the micro defect information corresponding to the micro true defect code, and does not output the information of the allowable defect corresponding to the micro allowable defect code.

Next, the macro defect primary processing means 108
It is composed of the alignment means 104, the difference pattern extraction means 105, the difference pattern length measurement means 106, and the macro defect candidate extraction means 109.

The macro defect candidate extraction means 109 compares the length measurement value measured by the difference pattern length measurement means 106 with the above-mentioned allowable minimum value W, and determines the area above the allowable minimum value W as the macro defect candidate extraction area To extract.

The macro defect secondary processing means 110 extracts the macro true defect region corresponding to the unacceptable true defect from the macro defect candidate extraction region, and does not extract the macro permissible defect region corresponding to the allowable defect.

Then, the macro true defect area is supplied to the defect information output means 7 as macro defect information.

Therefore, the defect information output means 7 outputs the macro defect information corresponding to the macro true defect area, and does not output the allowable defect information corresponding to the macro allowable defect code.

Therefore, the defect information output means 7 has the micro defect information supplied from the micro defect detection means 3 and the macro defect information supplied from the macro defect detection means 103.
Are both output.

Therefore, it is possible to detect all the defects occurring in the inspection pattern 2 and prevent the detection of the allowable micro allowable defects and the allowable macro allowable defects.

Next, a third embodiment of the present invention shown in FIG. 3 will be described.

The pattern inspection apparatus according to the third embodiment shows a specific example of the micro defect secondary processing means 10 of the second embodiment, and prevents detection of an acceptable normal end 2f from the inspection pattern 2A. To do.

The radial length measuring means 5a checks the continuity of the pattern image in the image data of the inspection pattern 2A such as the wiring pattern of the printed wiring board created by the detection optical system 1, and measures the continuity of the pattern image. Obtain the long value and perform radial length measurement.

The inspection pattern 2A has the disconnection 2e and the normal termination 2f, and the respective length measurement values of the disconnection 2e and the normal termination 2f are obtained.

The radial encoding means 5b performs the radial encoding of the length measurement value obtained by the radial length measurement means 5a in accordance with preset parameters.

As a result, the inspection code generating means 4 encodes the image data of the inspection pattern 2A on the basis of the continuity of the pixels and generates the inspection codes of the disconnection 2e and the normal termination 2f.

The micro defect candidate extraction means 9 compares each inspection code of the disconnection 2e and the normal termination 2f with the defect code of the defect pattern registered in the defect dictionary in advance, and determines that the defect has a size smaller than the allowable minimum value W. An inspection code that matches the corresponding defect code is extracted as a micro defect candidate extraction code.

As a result, the micro defect candidate extraction codes of the disconnection 2e and the normal end 2f are extracted.

The micro defect secondary processing means 10 is composed of a reference code generating means 13 and a radial code comparing means 16.

The reference code generating means 13 comprises a positioning means 12, a radial length measuring means 14 and a radial coding means 15.

The alignment means 12 aligns the inspection pattern 2A with the CAD pattern 102A created from the CAD data 101.

A bitmap pattern 102b is formed on the CAD pattern 102A.

The radial length measuring means 14 checks the continuity of the pattern image in the image data of the aligned CAD pattern 102A, obtains the length measurement value according to this continuity, and performs the radial length measurement.

The radial coding means 15 performs radial coding on the length measurement values obtained by the radial length measurement means 14 according to preset parameters to generate a reference code.

As a result, the reference code generating means 13
A reference code corresponding to the normal termination 2f is generated.

The radial code comparing means 16 determines the reference code corresponding to the normal termination 2f at the corresponding position in the CAD pattern 102A and the position thereof as the micro-defect candidate extraction codes of the disconnection 2e and the normal termination 2f. As a center, comparison is made with all radial codes within a preset fixed range.

More specifically, as shown in FIG. 4, the radial code Cr for extracting the micro defect candidate of the inspection pattern 2A is set to C
It is compared with the radial code of each pixel in the (n × n) area of the AD pattern 102A.

Further, a micro defect candidate extraction radial code that does not match the reference code corresponding to the normal termination 2f, that is, a micro defect candidate extraction radial code of the disconnection 2e is extracted as a micro true defect code corresponding to an unacceptable true defect.

Further, the micro defect candidate extraction radial code that matches the reference code corresponding to the normal end 2f, that is, the micro defect candidate extraction radial code of the normal end 2f is not extracted.

Then, the micro true defect code is supplied to the defect information output means 7 as micro defect information.

Further, the defect information output means 7 has a disconnection 2e.
The micro defect information corresponding to the micro true defect code based on the above is output, and the information of the micro allowable defect based on the normal end 2f is not output.

Next, a fourth embodiment of the present invention will be described with reference to FIG. In this fourth embodiment, the second microdefect 1
Micro defect primary treatment means 20 as second treatment means, and second
The micro-defect secondary processing means 21 serving as the micro-defect secondary processing means for preventing the normal end 2f from being detected.

The microdefect primary processing means 20 comprises an inspection code generating means 4, an inspection category coding means 22, and a microdefect candidate extracting means 23.

The inspection code generation means 4 uses the inspection pattern 2
The image data of A is converted into a radial code based on the continuity of pixels, and each inspection code of the disconnection 2e and the normal termination 2f is generated.

The inspection category coding means 22 searches the category codes indicating the types of pattern shapes corresponding to the various codes shown in FIG. 6 based on the inspection codes of the disconnection 2e and the normal termination 2f.

As a result, each inspection category code of (F) disconnection and (K) normal termination is obtained.

The micro defect candidate extraction means 23 compares each inspection category code of (F) disconnection and (K) normal termination with the category code of the defect pattern registered in the defect dictionary in advance, and determines from the allowable minimum value W. An inspection category code that matches a category code corresponding to a small size defect is extracted as a micro defect candidate extraction code.

As a result, the micro defect candidate extraction codes of the disconnection 2e and the normal termination 2f are extracted.

The micro defect secondary processing means 21 comprises a reference code generating means 13, a reference category coding means 24 and a category code comparing means 25.

The reference code generating means 13 is composed of a position aligning means 12, a radial length measuring means 14, and a radial coding means 15, and CA at a position corresponding to the inspection pattern 2A.
The image data of the D pattern 102A is coded based on the continuity of pixels to generate a reference code corresponding to the normal end 2f.

The reference category coding means 24 searches the category code indicating the type of the pattern shape corresponding to the various codes shown in FIG. 6 based on the reference code corresponding to the normal termination 2f (K) of the normal termination. Get the reference category code.

In the category code comparison 25, the disconnection 2
Each of the micro defect candidate extraction codes of e and the normal termination 2f is a reference category code of the (K) normal termination at a corresponding position in the CAD pattern 102A, and a predetermined fixed range centered on the reference category code. Compare with all category codes in.

More specifically, as shown in FIG. 7, the microdefect candidate extraction category code Cc of the inspection pattern 2A is set to C
The category code of each pixel in the (n × n) area of the AD pattern 102A is compared.

Further, (K) a micro defect candidate extraction category code that does not match the reference category code of the normal termination 2f, that is, a micro defect candidate extraction category code of the disconnection 2e is extracted as a micro true defect code corresponding to an unacceptable true defect. To do.

Further, (K) a micro defect candidate extraction category code that matches the reference category code of the normal termination 2f,
That is, the micro defect candidate extraction category code of the normal terminal 2f is not extracted.

Then, the micro true defect code is supplied to the defect information output means 7 as micro defect information.

Further, the defect information output means 7 has a disconnection 2e.
The micro defect information corresponding to the micro true defect code is output, and the information of the micro permissible defect based on the normal termination 2f is not output.

Next, a modification of the fourth embodiment will be described with reference to FIG.

In this modification, the shape of the pattern extracted as the micro defect candidate extraction code from the micro defect candidate extracting means 23 is determined.

For example, (K) when the normal end is extracted as a micro defect candidate extraction code, the width of the pattern width,
Further, even when the patterns have different lengths, the detected position does not change in the thickness direction X1 of the pattern.

More specifically, as shown in the inspection pattern 2A in FIG. 8, the thick line a, the thin line b, the double line c, and the broken line d have different thicknesses and lengths, but the thick line e and the thin line are different from each other. f,
The detected positions of the double line g and the broken line h do not change depending on the thickness direction X1 of the pattern.

Therefore, the CAD pattern 102 to be compared
If the position is accurately aligned with A, the area for searching the category code to be compared in the CAD pattern 102A can be limited only in the length direction X2 of the pattern.

Next, a fifth embodiment of the present invention will be described with reference to FIG.

This fifth embodiment comprises the same microdefect primary means 20 as in the fourth embodiment and a microdefect secondary treatment 30 which is a second microdefect secondary treatment, and detects the copper residue 2a. To prevent.

The microdefects primary treatment means 20 is composed of copper residue 2
The defect candidate extraction category codes of a, disconnection 2b, and chip (micro) 2c are extracted, and chip (macro) 2d is not extracted.

The microdefect secondary processing means 30 is composed of an upper category conversion means 31 and a defect judgment means 32.

As shown in FIG. 10, the higher category conversion means 31 is composed of a category distribution storage section 31a, a category search section 31b, a registration rule storage section 31c and a higher category output section 31d.

The category distribution storage unit 31a is supplied from the inspection category coding unit 22 on condition that an extraction signal indicating that the micro defect candidate extraction code has been extracted by the micro defect candidate extraction unit 23 has been input. The inspection category code is stored as the distribution of the inspection category code.

When the micro-defect candidate extraction code and the distribution of the inspection category code are supplied to the CPU, the category searching unit 31b determines a specific target object in the distribution of the inspection category code centered on the micro-defect candidate extraction code. Search categories according to predetermined rules.

The registration rule storage unit 31c classifies the relationship between the micro defect candidate extraction code and the target category, and stores various upper categories indicating the pattern shapes corresponding to this class.

The upper category output unit 31d calculates the relationship between the searched specific category and the micro defect candidate extraction code, and applies from the plurality of upper categories stored in the registration rule storage unit 31c based on this relationship. The upper category is read out and supplied to the defect determination means 32.

The upper category corresponding to the relationship between the specific category B and the micro defect candidate extraction code A is shown in FIG.
There is one shown in (G) as an example.

In FIG. 11 (A), when the target specific category B exists within a certain distance from the micro defect candidate extraction code A, it is set as the higher category N1 (rule).

In FIG. 11B, when the pattern binarized image 1a is continuously present within a certain distance from the micro defect candidate extraction code A, the target specific category B exists, it is set as the upper category N2. (rule).

In FIG. 11C, in the case where the target specific category B exists in a specific direction (within the range of the elevation angle of the specific angle) and within a certain distance with respect to the micro defect candidate extraction code A, Upper category N3 (rule).

In FIG. 11D, if the specific category B of interest within a certain distance from the micro defect candidate extraction code A is the specific direction, it is set as the upper category N4 (rule).

In FIG. 11 (E), when the target specific category B and the other specific category C exist at a distance within a fixed range of n pixels from the micro defect candidate extraction code A, the higher category N5 is set. Yes (rule).

In FIG. 11 (F), when the target specific category B exists within a certain distance from the micro defect candidate extraction code A, it is set as the upper category N6 (rule).

However, if the microdefect candidate extraction code A exists further in the direction of the category B with respect to the microdefect candidate extraction code A, it is set to the higher category N7.

In FIG. 11 (G), when there are a predetermined number or more of micro defect candidate extraction codes A within a certain distance, it is classified as a higher category N8 (rule).

Each of the rules defined and described in FIGS. 11A to 11G is based on the rule and is configured by adding each element as shown in FIG. Further, in addition to the configuration shown in FIG. 12, each element can be combined with each other to create more rules.

In FIG. 12, in the continuity among the respective elements, when the pattern binary image 1a continuously extends, the reference direction is the direction of the specific category B to be the target for the micro defect candidate extraction code A, The code direction is a specific direction of the code to be referred to, a plurality of category references are a plurality of target specific categories B and C, and the extension is a micro defect candidate extraction code A prior to the target specific category B. A plurality of the same category indicates each element that a plurality of micro defect candidate extraction codes A exist within a certain range.

The category searching unit 31b is shown in FIG.
As shown in each of FIGS. 13A to 13C, within a certain distance by radial scanning in 16 directions (FIG. 13A), rectangular area scanning (FIG. 13B), circular area scanning (FIG. 13C), and the like. For example, it is searched whether or not the target specific categories B and C exist within a range of about 20 pixels from the center.

[0215] The upper category N stored in the registration rule storage unit 31c corresponding to the relationship between the searched specific categories B and C and the extracted micro defect candidate extraction code A.
1 to N8 are read out under the control of the registration rule reference circuit, and the upper categories N1 to N8 are output to the upper category output unit 31.
It is supplied to the defect determination means 32 from d.

This defect judging means 32 is of the upper category N.
The quality of the inspection pattern 2 is determined based on 1 to N8.

That is, the defect determining means 32 uses the copper residue 2
The quality of the inspection pattern 2 is judged based on the respective upper category codes corresponding to a, the disconnection 2b and the chip (micro) 2c, and the disconnection 2b and the chip 2c are provided as the micro true defect code corresponding to the unacceptable true defect. Is extracted and the copper residue 2a is not extracted.

[0218] Then, the micro true defect code is supplied to the defect information output means 7 as micro defect information.

Further, the defect information output means 7 has a disconnection 2b.
And micro defect information corresponding to each micro true defect code of the chip (micro) 2c is output to prevent the detection of the copper residue 2a.

Next, a sixth embodiment of the present invention will be described with reference to FIG.

The sixth embodiment comprises the same micro-defect primary processing means 20 as in the fourth embodiment and a micro-defect secondary processing means 40 serving as a second micro-defect secondary processing means.
It is intended to prevent detection of the copper residue 2a.

The microdefects primary treatment means 20 is used for the copper residue 2
The micro defect candidate extraction codes of a, disconnection 2b, and chip (micro) 2c are extracted, and chip (macro) 2d is not extracted.

The micro defect secondary processing means 40 is composed of a radial category detecting means 41, a neural network 42 and a defect judging means 43.

The radial category detecting means 41 is shown in FIG.
As shown in (A), it comprises a category distribution memory 41a, a category search unit 41b, and a normalization circuit 41c.

The micro defect candidate extraction means 23 extracts the micro defect candidate extraction code and supplies it to the category searching section 41b, and the inspection category codes from the inspection category coding means 22 are sequentially stored in the category distribution memory 41a.

The category searching unit 41b selects from among the plurality of inspection category codes stored in the category distribution memory 41a centering on the extracted micro defect candidate extraction code, as shown in FIG. Search for inspection category codes on radial lines.

This category code search is performed when a specific category code which is a target different from the central micro defect candidate extraction code is detected from the category distribution memory 41a.

Furthermore, when an unknown pixel is detected,
After that, the category code detected first is used. In this way, the category codes of ~ in Fig. 15B can be detected.

The central micro-defect candidate extraction code is shown as, and a total of 9 category codes can be searched.

Further, the search for the category code is performed by referring to FIG.
As shown in (C), the micro defect candidate extraction code may be radially divided into eight areas, and the area may be sequentially searched from the area closer to the micro defect candidate extraction code.

The respective category codes searched in this way are detected, and the normalization circuit 41c sorts out the data processing method regarding the rotation of the distribution and the inversion of the front and back of each category code.
The adjusted and normalized category code is supplied to the neural network 42.

The neural network 42 processes the sequentially supplied category codes as a hierarchical network structure. As a result, the upper category code is calculated and supplied to the defect determining means 43.

The operation of this upper category code is processed by a three-layer back propagation type hierarchical network structure.

That is, the defect judging means 43 judges and extracts each true defect code of the disconnection 2b and the chip (micro) 2c corresponding to the unacceptable true defect based on the upper category code, and extracts the copper residue 2a. do not do.

The micro true defect code is output to the defect information output means 7 as micro defect information.

Further, the defect information output means 7 has a disconnection 2b.
And the micro defect information corresponding to each micro true defect code of the chip (micro) 2c is output, and the information of the copper residual 2a is not output.

Note that a modified example such as shown in FIGS. 16A and 16B can also be adopted. FIG. 16 (A) shows a configuration in which a category is searched from a plurality of category distributions centering on each of defect candidate categories 1 to 9, and a neural network calculates and outputs a higher category code based on each searched category. You can also do it. FIG. 16B shows a plurality of neural networks 101-.
10n are provided in parallel, the above-mentioned categories are calculated based on a plurality of categories to which the respective neural networks 101 to 10n are respectively supplied, and are supplied to the higher order neural network. The upper neural network may be configured to calculate the highest category.

Next, a seventh embodiment of the present invention shown in FIG. 17 will be described.

The pattern inspection apparatus according to the seventh embodiment is a specific example of the macro-defect secondary processing means 110 of the second embodiment, and prevents detection of the copper residue 2g having an allowable size from the inspection pattern 2B. To do.

That is, the copper residue 2 g having a large size is a defect which is present in the vicinity of the conductor but has no practical harm and does not need to be detected.

Defects that must be really detected are copper residues (or pinholes) that exist between conductors (or voids) and may cause a short circuit (or disconnection), such as a large copper residue 2 g. ).

The macro defect primary processing means 108 includes the alignment means 104, the difference pattern extracting means 105, the difference pattern length measuring means 106, and the macro defect candidate extracting means 109.
And consists of.

More specifically, as shown in FIG. 18, the alignment means 104 of the macro defect primary processing means 108 uses the CA
The D pattern 102B and the inspection pattern 2B are aligned.

The difference pattern extraction means 105 extracts the difference pattern 118. Difference pattern length measuring means 106
Measures the difference pattern 118.

The macro defect candidate extraction means 109 compares the length measurement value measured by the difference pattern length measurement means 106 with the above-mentioned allowable minimum value W, and the difference measurement pattern 119 for the area above the allowable minimum value W. To extract.

As a result, the pixel 120 which becomes the macro defect candidate extraction region of the large size copper residue 2g and the pixel 121 which becomes the macro defect candidate extraction region of the large size copper residue h are extracted to the difference length measurement pattern 119.

The macro defect secondary processing means 110 is composed of a fatal area pattern forming means 111 and a fatal area pattern reference means 112.

The critical area pattern creating means 111 uses the CA
Based on the D pattern 102B, the fatal area pattern 115 including the fatal areas 116 and 117 in which defects cannot be tolerated.
To create.

The critical areas 116 and 117 are the peripheral edges 113a and 1 of the linear lead 113 of the CAD pattern 102B.
It is provided with a constant width d1 in each of the outward direction and the inward direction from 13a. In addition, the peripheral edge 114a of the land 114
Is provided with a constant width d1 in each of the outward direction and the inward direction.

The constant width d1 may have different values in the outer direction and the inner direction.

The critical area pattern reference means 112 refers to the critical area pattern 115 at the corresponding position for the pixels 120 and 121 extracted as the macro defect extraction area.

Since the pixel 121 corresponds to the critical area 116, it is extracted as a macro true defect area, and the pixel 120 does not correspond to the critical areas 116 and 117, so it is not extracted.

For this reason, it is possible to prevent the detection of the copper residue 2 g which is an allowable defect which can be accepted. That is, excessive detection of defects can be reduced.

Then, the macro true defect area corresponding to the remaining copper 2h is output to the defect information output means 7 as macro defect information.

Therefore, the defect information output means 7 can output the macro defect information corresponding to the copper residue 2h and prevent the detection of the copper residue 2g which is an allowable allowable defect.

Further, since the critical areas 116 and 117 are included, the defect detection performance due to the defect shape or alignment does not change.

Next, an eighth embodiment of the present invention shown in FIG. 19 will be described.

The pattern inspection apparatus according to the eighth embodiment is a specific example of the macro-defect secondary processing means 110 of the second embodiment, and detects copper residues 2g and 2i of an acceptable size from the inspection pattern 2C. Is to prevent.

That is, the copper residues 2g and 2i are defects which are present in the vicinity of the conductor but have no practical harm and do not need to be detected.

The defect that should be really detected is the copper residue that exists between the conductors like the copper residue 2g and may cause a short circuit. Alternatively, the pinhole exists in the conductor like the pinhole 2j and may cause disconnection.

The macro defect primary processing means 108 includes the alignment means 104, the difference pattern extracting means 105, the difference pattern length measuring means 106, and the macro defect candidate extracting means 109.
Composed of and.

More specifically, as shown in FIG. 20, the alignment means 104 aligns the CAD pattern 102B with the inspection pattern 2C.

The difference pattern extraction means 105 extracts the difference pattern 140. Difference pattern length measuring means 106
Measures the difference pattern 140.

The macro defect candidate extraction means 109 compares the length measurement value measured by the difference pattern length measurement means 106 with the above-mentioned allowable minimum value W, and determines the area of the allowable minimum value W or more as the difference measurement pattern 141. To extract.

As a result, a pixel 142 serving as a macro defect candidate extraction area with a copper residue of 2 g, a pixel 143 serving as a macro defect candidate extraction area of a copper residue h, and a pixel 144 serving as a macro defect candidate extraction area of a copper residue i. , The pixel 145 that is the macro defect candidate extraction region of the pinhole 2j is extracted in the differential length measurement pattern 141.

The macro defect secondary processing means 130 carries out a fatal area pattern creation 131 including a radial length measurement 132 and a fatal area pattern reference 136.

The critical area pattern creating means 131 uses the CA
Straight lead 113 and land 114 of D pattern 102B
Radial length measurement is performed by the radial length measuring unit 132 to extract a pattern interval between and, and a region where the pattern interval is smaller than a predetermined value is extracted as the first region 134.

Further, the width of the linear lead 113 of the CAD pattern 132 and the pattern width of the diameter of the land 114 are radially measured by the radial length measuring means 132, and the pattern width is smaller than a predetermined value. The second
Areas 135, 135 of

Further, at least one of the first area 134 and the second areas 135, 135 is calculated.

Since the radial length measuring means 132 is used, the first area 134 and the second area 13 are used.
5,135 can be obtained at the same time.

The fatal area pattern reference unit 136 uses the pixels 142, 143 extracted as the macro defect extraction area,
For the 144 and 145, the fatal area pattern 133 at the corresponding position is referred to.

The pixel 142 includes a first region 134 and a second region 134.
The areas 135 and 135 of No. 1 are not extracted as macro true defect areas.

Since the pixel 143 corresponds to the first area 134, it is extracted as a macro true defect area.

Since the pixel 144 does not correspond to the first area 134 and the second areas 135, 135, it is not extracted as a macro true defect area.

Since the pixel 145 corresponds to the second area 135, it is extracted as a macro true defect area.

Therefore, it is possible to prevent the detection of the copper residues 2g and 2i, which are tolerable defects. That is,
Excessive detection of defects can be reduced.

Then, each macro true defect area corresponding to the copper residue 2h and the pinhole 2j is supplied to the defect information output means 8 as macro defect information.

Therefore, the defect information output means 7 can output each macro defect information corresponding to the copper residual 2h and the pinhole 2j, and prevent the detection of the copper residuals 2g and 2i which are the allowable defects.

Further, since the first area 134 and the second areas 135, 135 are included, the defect detection performance due to the defect shape or alignment does not change.

[0280]

As described above, according to the invention of claim 1,
Since the configuration includes the micro defect detection step and the macro defect detection step, in the micro defect detection step, a defect having a size smaller than the reference size is detected according to a predetermined reference size set in advance, and the macro defect detection is performed. In the process, in order to detect defects of the size above the standard size , preset
Defects smaller than the specified standard size, and
Larger defects than smaller defects of this size
Both can be detected reliably, the above reference size,
The thinnest pattern present in the reference pattern of the object to be inspected
Since the minimum allowable value of the
Defects that are smaller in size and smaller than smaller defects.
Size is smaller than the allowable minimum value to detect large defects.
Both small defects and defects that are larger than the minimum allowable size
Can be reliably detected in the micro defect detection process.
Is the pixel of the image data of the inspection pattern of the inspection object.
Generate inspection code by coding based on continuity
The defect code of the registered defect pattern and the inspection code
Compared to the standard size,
The inspection code that matches the defect code that corresponds to the
Set in advance to detect defects by extracting as outgoing code
Ensures defects smaller than the specified standard size
The micro-defect detection process, which can detect
Micro defect primary treatment process and first micro defect secondary treatment process
Since it was composed of the following, in the first micro defect primary treatment process
Is the pixel of the image data of the inspection pattern of the inspection object.
Generate inspection code by coding based on continuity
The defect code of the registered defect pattern and the inspection code
Compared to the standard size,
Check the inspection code that matches the defect code
Extracted as a complementary extraction code, and the first micro defect secondary processing
In the process, the reference of the position corresponding to the inspection pattern
Code image data of pattern based on pixel continuity
Generate a reference code, and based on the above position,
All the reference codes in the circle and the above defect candidate extraction code
And the defect candidate extraction that does not match the reference code.
The output code is a true defect code that corresponds to an unacceptable true defect.
And the defect is detected. Therefore, acceptable
It is possible to prevent the detection of a micro allowable defect.

[0281]

[0282]

[0283]

[0284]

[0285]

[0286]

[0287]

[0288]

[0289]

[0290]

According to the second aspect of the invention, since the microdefect detection step is composed of the second microdefect primary treatment step and the second microdefect secondary treatment step, the second microdefect In the primary processing step, the image data of the inspection pattern of the inspection object is coded based on the continuity of pixels to generate an inspection code, and the category code indicating the type of pattern shape corresponding to each code is added to the inspection code. To obtain the inspection category code, compare the category code of the defect pattern registered in advance with the inspection category code, and the inspection category code that matches the category code corresponding to the defect having a size smaller than the reference size Is extracted as a defect candidate extraction code.

In the second micro defect secondary treatment step,
Image data of the reference pattern at the position corresponding to the inspection pattern is coded based on pixel continuity to generate a reference code, and a category code indicating the type of pattern shape corresponding to each code is based on the reference code. To obtain a reference category code, compare all the reference category codes in the predetermined range with the above-mentioned position with the defect candidate extraction category code, and detect the defect candidate extraction category that does not match the reference category code. The code is extracted as a true defect code corresponding to an unacceptable true defect to detect the defect.

Therefore, it is possible to prevent the detection of an allowable micro allowable defect.

According to the third aspect of the invention, in the second micro defect secondary processing step, the image data of the reference pattern at the position corresponding to the inspection pattern is coded based on the continuity of the pixels. A reference code is generated, a category code indicating the type of pattern shape corresponding to each code is searched based on the reference code to obtain a reference category code, and the reference category code is limited according to the position and the type of shape of the pattern. All or some of the reference category codes within the range are compared with the defect candidate extraction category codes, and defect candidate extraction category codes that do not match the reference category code are true defect codes corresponding to unacceptable true defects. To detect the defect.

Therefore, it is possible to prevent the detection of an allowable micro allowable defect. Further, since the inspection category code is compared with all the reference category codes in the limited range according to the position and the type of the shape of the pattern, the comparison time can be shortened as compared with the case without limitation. .

According to the invention of claim 4 , in the second micro defect secondary processing step, the relationship between the defect candidate extraction category code and the target category is classified, and the pattern shape corresponding to this classification is classified. Based on the relationship between the defect candidate extraction category code and a specific target category in the category distribution within a predetermined area centered on the defect candidate extraction category code. The upper category code corresponding to this relationship is selected and output, and the true defect code corresponding to the unacceptable true defect is determined based on the upper category code to detect the defect.

Therefore, it is possible to prevent the detection of an allowable micro allowable defect.

Further, according to the invention of claim 5 , in the second micro-defect secondary processing step, a plurality of categories radially distributed from the center of the defect candidate extraction category code are detected and the detection is performed. By processing the created multiple categories as data of a hierarchical network structure by relating them to each other, the upper category is output, and the true defect code corresponding to the unacceptable true defect is determined based on the upper category code. Detect defects.

Therefore, it is possible to prevent the detection of an allowable micro allowable defect.

According to the invention of claim 6 , in the macro defect detecting step, the inspection pattern of the inspection object and the reference pattern are aligned with each other, and the inspection pattern of the aligned inspection object and the inspection pattern of the inspection object are compared with each other. A difference pattern from the reference pattern is extracted, the difference pattern is measured, the measured length value is compared with the reference size, and an area having a size equal to or larger than the reference size is extracted as a defect extraction area to detect a defect. To detect.

Therefore, it is possible to reliably detect a defect having a size larger than the reference size.

[0302]

[0303]

[0304]

Further, according to the invention of claim 7 , in the macro defect secondary processing step, a pattern of a critical area in which a defect cannot be accepted is created on the basis of the reference pattern, and a position corresponding to the defect candidate extraction area is created. By referring to the pattern of the critical area, a defect is detected by extracting a case where a position corresponding to the defect candidate extraction area exists in the critical area as a true defect area.

Therefore, it is possible to prevent the detection of an allowable macro allowable defect. Further, since the critical area is included, the defect detection performance due to the defect shape or alignment does not change.

Further, according to the invention of claim 8.9.10 , since the fatal area is provided with a constant width in the outer and inner directions from the peripheral edge of each pattern of the reference pattern, the reference pattern of the reference pattern is formed. It is possible to eliminate the need for length measurement.

According to the eleventh aspect of the invention, the fatal area is measured by measuring the pattern interval of the reference pattern, and the area in which the pattern interval is smaller than a predetermined value is set as the first area. The pattern width of the reference pattern is extracted and the pattern width is extracted, and a region having a pattern width smaller than a predetermined value is extracted as a second region.
And at least one of the second area and the second area.

Therefore, the first area can be a dead area in which the patterns can come into contact with each other, and the second area can be a dead area in which the patterns can be broken.

[0310]

[0311]

[0312]

[0313]

[0314] Also, according to the invention of claim 12, the first micro defects secondary processing means, since the more becomes configured as a reference code generating means and micro defect extraction means according to the invention of claim 1 pattern It can be used directly to carry out the inspection method.

According to the thirteenth aspect of the present invention, the microdefect detecting means includes the second microdefect primary processing means and the second microdefect primary processing means.
The micro-defect secondary treatment means according to claim 2,
Can be directly used to implement the pattern inspection method according to the invention.

According to the fourteenth aspect of the present invention, the second micro defect secondary processing means is constituted by the reference code generating means, the reference category code acquiring means and the micro defect extracting means. It can be directly used for implementing the pattern inspection method according to the third invention.

[0317] Also, according to the invention of claim 15, the second micro defects secondary processing means, since the more becomes constituting an upper category conversion means and micro defect extraction means according to claim 4
Can be directly used to implement the pattern inspection method according to the invention.

[0318] Also, according to the invention of claim 16, the second micro defects secondary processing means, since more become configured as radial category detection means and a neural network and micro defect extraction means, the invention of claim 5 It can be directly used for implementing the pattern inspection method according to the present invention.

[0319] Also, according to the invention of claim 17, the macro-defect detecting means, since the more becomes constituting the alignment means and the difference pattern extracting means and difference pattern measuring means and the macro defect extraction means according to claim 6 Can be directly used to implement the pattern inspection method according to the invention.

[0320]

According to the eighteenth aspect of the invention, the macro defect secondary processing means is composed of the fatal area pattern forming means and the macro defect extracting means.
It can be directly used for implementing the pattern inspection method according to the seventh invention.

[0322] Also, according to the invention of claim 19, the critical region, in each of the outward and inward from the periphery of each pattern of the reference pattern, so providing a constant width, the invention of claim 8 It can be used directly to implement such a pattern inspection method.

[0323] Also, according to the invention of claim 20, the constant width of the critical region, since the different values between the outer direction and the inward direction, the pattern inspection method according to the invention of claim 9 It can be used directly for implementation.

Further, according to the invention of claim 21 , the constant width of the critical area is set by being changed in advance according to the type of the pattern shape. Therefore, the pattern inspection according to the invention of claim 10 It can be used directly to carry out the method.

According to the twenty-second aspect of the present invention, the fatal area is measured by measuring the pattern interval of the reference pattern,
An area in which the pattern interval is smaller than a predetermined value is extracted as a first area, the pattern width of the reference pattern is measured, and the pattern width is smaller than a predetermined value. Is extracted as a second area, and at least one of the first area and the second area is calculated and provided, and thus is directly used for implementing the pattern inspection method according to the invention of claim 11. You can

[Brief description of drawings]

FIG. 1 is a configuration diagram of a first embodiment of the present invention.

FIG. 2 is a configuration diagram of a second embodiment of the present invention.

FIG. 3 is a configuration diagram of a third embodiment of the present invention.

FIG. 4 is a diagram for explaining the radial code comparison of FIG.

FIG. 5 is a configuration diagram of a fourth embodiment of the present invention.

FIG. 6 is a diagram showing types of categories shown in FIG.

FIG. 7 is a diagram for explaining the category code comparison of FIG.

FIG. 8 is a diagram for explaining the category code comparison of FIG.

FIG. 9 is a configuration diagram of a fifth embodiment of the present invention.

10 is a schematic configuration diagram of a higher category conversion means in FIG.

FIG. 11 is a diagram showing upper categories.

FIG. 12 is a diagram showing rules of a category search unit.

FIG. 13 is a scanning mode diagram of a category search unit.

FIG. 14 is a configuration diagram of a sixth embodiment of the present invention.

15 is a diagram for explaining the micro defect secondary processing means of FIG. 14;

FIG. 16 is a diagram for explaining a modification of the sixth embodiment of FIG.

FIG. 17 is a configuration diagram of a seventh embodiment of the present invention.

FIG. 18 is a diagram for explaining the macro defect secondary processing of FIG. 17;

FIG. 19 is a configuration diagram of an eighth embodiment of the present invention.

FIG. 20 is a diagram for explaining the macro defect secondary processing of FIG. 19;

FIG. 21 is a configuration diagram of a conventional pattern inspection method.

FIG. 22 is an explanatory diagram of a problem of the pattern inspection method of FIG.

FIG. 23 is an explanatory diagram of a problem of the pattern inspection method of FIG.

[Explanation of symbols]

1 Detection optical system 2 inspection pattern 2A inspection pattern 2B inspection pattern 2C inspection pattern 2a Copper residue 2b disconnection 2c chip (micro) 2d chip (macro) 2e disconnection 2f Normal termination 2g copper residue 2h Copper residue 2i Copper residue 2j pinhole 3 Micro defect detection means 4 Inspection code generation means 5a Radial length measuring means 5b Radial coding means 6 Micro defect extraction means 7 Defect information output means 8 Micro defect primary treatment means 9 Micro defect candidate extraction means 10 Micro defect secondary treatment means 12 Positioning means 13 reference code generation means 14 Radial length measuring means 15 Radial coding means 16 Radial code comparison means 20 Micro defect primary treatment means 21 Micro defect secondary treatment means 22 Inspection category coding means 23 Micro Defect Candidate Extraction Means 24 Standard category coding means 25 Category Code Comparison Means 30 Micro defect secondary treatment means 31 Upper category conversion means 32 Defect determination means 40 Micro defect secondary treatment means 41 Radial category detection means 42 Neural network 43 Defect determination means 101 CAD data 102 CAD pattern 102A CAD pattern 102B CAD pattern 102a bitmap pattern 102b bitmap pattern 103 Macro defect detection process 104 alignment 105 Difference pattern extraction 106 Difference pattern length measurement 107 Macro defect extraction 108 Macro defect primary processing 109 Macro defect candidate extraction 110 Macro defect secondary processing 111 Create critical area pattern 112 Refer to fatal area pattern 115 Deadly area pattern 116,117 Deadly area 130 Macro defect secondary processing 131 Create critical area pattern 132 Radial length measurement 133 fatal area pattern 134 First Area 135 Second area 136 Refer to the fatal area pattern

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Gotoshi Ando 1015 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu Limited (56) References JP-A-63-32666 (JP, A) JP-A-61- 245007 (JP, A) JP 60-227431 (JP, A) JP 7-20061 (JP, A) JP 6-249792 (JP, A) JP 5-264240 (JP, A) JP 5-187842 (JP, A) JP 5-108800 (JP, A) JP 5-20433 (JP, A) JP 5-10890 (JP, A) JP 4-286971 (JP, A) JP-A-3-91875 (JP, A) JP-A-2-170279 (JP, A) Yoshikazu Kakimoto et al., Shu: Optical technology to support high-density mounting Printed board pattern inspection method, O plus E , Japan, New Technology Communications, 1992 Month, No. 148,99-103 Ando MamoruShun, printed circuit board inspection using the radial matching, Precision Engineering Journal, Japan, Japan Society for Precision Engineering, August 5, 1990, No. 668, 30-33 Hideo Okada et al., High-performance printed circuit board pattern inspection method with macroscopic view, Proceedings of IEICE Spring Conference 1994, Japan, The Institute of Electronics, Information and Communication Engineers, March 26, 1994, Vol. ． 1994, 7-140 (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 1/00 305 G01N 21/88 JISST file (JOIS)

Claims (22)

(57) [Claims] 1. A pattern inspection method for detecting a defect occurring in an inspection pattern of an inspection object, wherein a micro defect detection step of detecting a defect having a size smaller than the reference size according to a preset predetermined reference size. And a macro defect detection step of detecting a defect having a size equal to or larger than the reference size, wherein the reference size is an allowable minimum value of the thinnest pattern existing in the reference pattern of the inspection object, and the micro defect detection step Is constituted by a first micro-defect primary treatment step and a first micro-defect secondary treatment step, and the first micro-defect primary treatment step is performed by using the image data of the inspection pattern of the inspection object as pixels. The inspection code is generated by encoding based on the continuity of the inspection pattern, the defect code of the defect pattern registered in advance is compared with the inspection code, and the inspection is performed based on the reference size. An inspection code that matches a defect code corresponding to a small defect is extracted as a defect candidate extraction code, and the first micro-defect secondary processing step is performed in the image of the reference pattern at a position corresponding to the inspection pattern. The data is coded based on the continuity of pixels to generate a reference code, all reference codes within a predetermined range are compared with the defect candidate extraction code based on the position, and the reference code does not match. A pattern inspection method, wherein a defect candidate extraction code is extracted as a true defect code corresponding to an unacceptable true defect to detect the defect. 2. A pattern inspection method for detecting defects occurring in an inspection pattern of an object to be inspected, wherein a micro defect detection step of detecting a defect having a size smaller than the reference size according to a predetermined reference size set in advance. And a macro defect detection step of detecting a defect having a size equal to or larger than the reference size, wherein the reference size is an allowable minimum value of the thinnest pattern existing in the reference pattern of the inspection object, and the micro defect detection step Is constituted by a second micro-defect primary treatment step and a second micro-defect secondary treatment step, and the second micro-defect primary treatment step is performed by using the image data of the inspection pattern of the inspection object as pixels. The inspection code is generated by coding based on the continuity of, and the category code indicating the type of pattern shape corresponding to each code is searched based on the above inspection code. Then, the inspection category code is obtained, and the category code of the defect pattern registered in advance is compared with the above inspection category code, and the inspection category code that matches the category code corresponding to the defect having a size smaller than the reference size is extracted as the defect candidates. In the second micro-defect secondary processing step, the image data of the reference pattern at the position corresponding to the inspection pattern is encoded based on the continuity of pixels to generate a reference code. A category code indicating the type of pattern shape corresponding to each code is searched based on the reference code to obtain a reference category code, and all the reference category codes and the defect candidate extraction within a predetermined range based on the above position The defect candidate extraction category that does not match the reference category code is compared with the category code. A pattern inspection method characterized in that a Gori code is extracted as a true defect code corresponding to an unacceptable true defect to detect the defect. 3. The second micro-defect secondary processing step, wherein image data of the reference pattern at a position corresponding to the inspection pattern is encoded based on pixel continuity to generate a reference code, and various codes are generated. A category code indicating the type of pattern shape corresponding to is searched based on the above reference code to obtain a reference category code, and all or part of a limited range is obtained depending on the above position and the type of shape of the pattern. Comparing the reference category code with the defect candidate extraction category code, a defect candidate extraction category code that does not match the reference category code is extracted as a true defect code corresponding to an unacceptable true defect to detect a defect. The pattern inspection method according to claim 2, wherein the pattern inspection method is replaced with a configuration. 4. The second micro-defect secondary processing step classifies the relationship between the defect candidate extraction category code and the target category, and various upper category codes indicating the types of pattern shapes corresponding to this classification. Based on the relationship between the defect candidate extraction category code and a specific category of interest in the category distribution within a predetermined area centered on the defect candidate extraction category code, the upper category corresponding to this relationship. 3. The pattern inspection according to claim 2, wherein the code is selected and output, and the defect is detected by determining a true defect code corresponding to an unacceptable true defect based on the upper category code. Method. 5. The second micro-defect secondary processing step is performed by detecting a plurality of categories radially distributed from the center of the defect candidate extraction category code, and mutually detecting the plurality of detected categories. Instead of the configuration in which the upper category is output by processing the data as a hierarchical network structure in association with each other, the true defect code corresponding to the unacceptable true defect is determined based on the upper category code to detect the defect. The pattern inspection method according to claim 2, wherein 6. The macro defect detection step performs alignment between an inspection pattern of the inspection object and a reference pattern, and extracts a difference pattern between the aligned inspection pattern of the inspection object and the reference pattern. Then, the difference pattern is measured, the measured length value is compared with the reference size, and a region larger than the reference size is extracted as a defect extraction region to detect a defect. The pattern inspection method according to claim 1 or 2, which is characterized. 7.The macro defect detection process described above
It consists of a primary treatment process and a macro defect secondary treatment process, The macro defect primary treatment step, Positioning of the inspection pattern of the above-mentioned inspection object and the reference pattern
Do the The inspection pattern of the aligned object and the inspection pattern
Extract the difference pattern from the reference pattern, Measure the difference pattern, This length measurement value is compared with the reference size, and
Areas that are larger than the standard size are extracted as defect candidate extraction areas
Configuration, The macro defect secondary treatment step, Based on the above reference pattern, a critical area where defects cannot be tolerated
Area pattern, The pattern of the critical area at the position corresponding to the defect candidate extraction area
The position corresponding to the defect candidate extraction area is
If it exists in the critical area, it is extracted as a true defect area.
Claims characterized by being configured to detect defects1
Or 2 or 6Described pattern inspection method. 8. The pattern inspection method according to claim 7, wherein the lethal region is provided with a constant width in each of an outer direction and an inner direction from a peripheral edge of each pattern of the reference pattern. 9. The pattern inspection method according to claim 8, wherein the constant width of the critical area is set to different values in the outward direction and the inward direction. 10. The pattern inspection method according to claim 8, wherein the certain width of the critical area is set by being changed in advance according to the type of the pattern shape. 11. The fatal area is measured by measuring a pattern interval of the reference pattern, and an area in which the pattern interval is smaller than a predetermined value is firstly determined.
Area, the pattern width of the reference pattern is measured, and an area having a pattern width smaller than a predetermined value is extracted as a second area, and the first area and the second area are extracted. 8. The pattern inspection method according to claim 7, wherein at least one of the areas is calculated and provided. 12. A pattern inspection apparatus for detecting defects occurring in an inspection pattern of an object to be inspected, a micro defect detecting means for detecting a defect having a size smaller than the reference size according to a predetermined reference size set in advance. And a macro defect detecting means for detecting a defect having a size equal to or larger than the reference size, wherein the reference size is an allowable minimum value of the thinnest pattern existing in the reference pattern of the inspection object, and the micro defect detecting means Is constituted by a first micro-defect primary processing means and a first micro-defect secondary processing means, and the first micro-defect primary processing means makes the pixel data of the image data of the inspection pattern of the inspection object. The inspection code generating means for generating the inspection code by encoding based on the continuity of the inspection code and the defect code of the previously registered defect pattern are compared with the inspection code. And a microdefect candidate extracting means for extracting an inspection code that matches a defect code corresponding to a defect having a size smaller than the reference size as a defect candidate extracting code, and the first microdefect secondary processing means is provided. A reference code generating means for generating a reference code by encoding the image data of the reference pattern at a position corresponding to the inspection pattern based on the continuity of pixels, and all the references within a predetermined range based on the position. And a defect candidate extraction code that does not match the reference code, extracts a defect candidate extraction code as a true defect code corresponding to an unacceptable true defect, and detects a defect. A pattern inspection apparatus having a configuration. 13. A pattern inspection apparatus for detecting defects occurring in an inspection pattern of an object to be inspected, said micro defect detecting means for detecting a defect having a size smaller than said reference size in accordance with a predetermined reference size set in advance. And a macro defect detecting means for detecting a defect having a size equal to or larger than the reference size, wherein the reference size is an allowable minimum value of the thinnest pattern existing in the reference pattern of the inspection object, and the micro defect detecting means With a second micro-defect primary processing means and a second micro-defect secondary processing means. The inspection code generating means for generating the inspection code by encoding based on the continuity of, and the category code indicating the type of pattern shape corresponding to each code are described above. An inspection category code acquisition unit that obtains an inspection category code by searching based on the inspection code is compared with a category code of a defect pattern registered in advance and the inspection category code to deal with a defect having a size smaller than the reference size. A micro defect candidate extraction means for extracting an inspection category code that matches the category code as a defect candidate extraction code, and the second micro defect secondary processing means is used for the reference pattern at a position corresponding to the inspection pattern. A reference code generating means for generating a reference code by encoding the image data of 1) based on the continuity of pixels, and a category code indicating the type of pattern shape corresponding to each code are searched based on the reference code to obtain a reference category. A reference category code acquisition means for obtaining a code, and a predetermined position based on the above position All the reference category codes within the range of the above and the defect candidate extraction category code are compared, and the defect candidate extraction category code that does not match the reference category code is extracted as a true defect code corresponding to an unacceptable true defect. A pattern inspection apparatus comprising a microdefect extracting means for detecting a defect. 14. A reference code generator that encodes image data of the reference pattern at a position corresponding to the inspection pattern based on pixel continuity in the second micro-defect secondary processing means to generate a reference code. Means, a reference category code acquisition means for obtaining a reference category code by searching a category code indicating the type of pattern shape corresponding to each code on the basis of the reference code, according to the position and the type of shape of the pattern. , All or some of the reference category codes within a limited range are compared with the defect candidate extraction category codes, and defect candidate extraction category codes that do not match the reference category code correspond to unacceptable true defects. The structure is replaced with a micro defect extraction means for extracting defects as true defect codes and detecting defects. 14. The pattern inspection apparatus according to claim 13, wherein: 15. The second micro-defect secondary processing means classifies the relationship between the defect candidate extraction category code and the target category, and various upper category codes indicating the types of pattern shapes corresponding to this classification. Based on the relationship between the defect candidate extraction category code and a specific category of interest in the category distribution within a predetermined area centered on the defect candidate extraction category code, the upper category corresponding to this relationship. An upper category conversion unit that selects and outputs a code, and a microdefect extraction unit that determines a true defect code corresponding to an unacceptable true defect based on the upper category code and detects a defect, instead of the configuration. 14. The pattern inspection apparatus according to claim 13, wherein 16. The second micro-defect secondary processing means comprises: a radial category detecting means for detecting a plurality of categories radially distributed from the center of the defect candidate extraction category code, and the detected plurality of categories. By correlating the above categories with each other and processing them as data of a hierarchical network structure, the true network that outputs the upper category and the true defect code corresponding to the unacceptable true defect are determined based on the above category code. 14. The pattern inspection apparatus according to claim 13, wherein the pattern inspection apparatus is replaced by a microdefect extracting means for detecting a defect. 17. The macro defect detecting means comprises: an aligning means for aligning an inspection pattern of the inspection object with a reference pattern; and an alignment inspection pattern of the inspection object with the reference pattern. A difference pattern extracting means for extracting a difference pattern, a difference pattern length measuring means for measuring the difference pattern, and a length measurement value thus measured are compared with the reference size, and an area larger than the reference size is defective. The pattern inspection apparatus according to any one of claims 12 to 16, characterized in that the pattern inspection apparatus comprises a macro defect extraction unit that extracts a defect by extracting it as an extraction region. 18. A pattern inspection apparatus for detecting a defect occurring in an inspection pattern of an inspection object, a microdefect detecting means for detecting a defect having a size smaller than the reference size in accordance with a predetermined reference size set in advance. And a macro defect detecting means for detecting a defect having a size equal to or larger than the reference size, wherein the reference size is an allowable minimum value of the thinnest pattern existing in the reference pattern of the inspection object, and the macro defect detecting means Is composed of a macro defect primary processing means and a macro defect secondary processing means, and the macro defect primary processing means is an alignment means for aligning the inspection pattern of the inspection object with the reference pattern. A difference pattern extracting means for extracting a difference pattern between the aligned inspection pattern of the inspection object and the reference pattern; The difference pattern length measuring means for measuring the pattern, and the macro defect candidate extracting means for comparing the length measured value with the reference size and extracting the area equal to or larger than the reference size as the defect candidate extraction area. The macro defect secondary processing means includes a fatal area pattern creating means for creating a pattern of a fatal area in which a defect cannot be accepted based on the reference pattern, and a fatal area at a position corresponding to the defect candidate extraction area. A pattern comprising a macro defect extracting unit that refers to a pattern and detects a defect by extracting a case where a position corresponding to the defect candidate extraction region exists in the critical region as a true defect region. Inspection device. 19. The pattern inspection apparatus according to claim 18, wherein the critical area is provided with a constant width in each of an outer direction and an inner direction from a peripheral edge of each pattern of the reference pattern. 20. The pattern inspection apparatus according to claim 19, wherein the constant width of the critical area has different values in the outward direction and the inward direction. 21. The pattern inspection apparatus according to claim 19, wherein the certain width of the critical area is changed and set in advance according to the type of the pattern shape. 22. The pattern interval of the reference pattern is measured for the fatal area, and the area where the pattern interval is smaller than a predetermined value is firstly determined.
Area, the pattern width of the reference pattern is measured, and an area having a pattern width smaller than a predetermined value is extracted as a second area, and the first area and the second area are extracted. 19. The pattern inspection apparatus according to claim 18, wherein at least one of the areas is calculated and provided.