JP2003098117A - Apparatus for detecting defect of image and its defect detection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a high-speed and highly accurate apparatus for detecting defects of image without being affected by variations of good products, and its defect detection method. SOLUTION: The apparatus 30 for detecting defects of image is constituted of a data conversion means 2 for converting an image 1 to be inspected to digital image information, a first storage means 7 for storing digital image information, a second storage means 8 and a third storage means 9 for separately storing digital image information stored into the first storage means 7, a first processing means 10 for executing expansion filtering to digital image information stored into the second storage means 8, a second processing means 11 for carrying out contraction filtering to digital image information obtained by the first processing means 10, a third processing means 12 for carrying out contraction filtering to digital image information stored into the third storage means 9, a fourth processing means 13 for carrying out expansion filtering to digital image information obtained by the third processing means 12, and a fifth processing means for calculating a difference value of digital image information obtained by the second processing means 11 and the fourth processing means 13.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image defect detecting apparatus and a defect detecting method therefor, and in particular,
The present invention relates to a defect detection device and a defect detection method for detecting defects in a photomask pattern.

[0002]

2. Description of the Related Art Conventionally, for example, in the case of inspecting a photomask or the like used in semiconductor manufacturing, a main inspection method using image processing is to compare the same patterns at different positions among actual patterns. Two methods of die-to-database by comparing die-to-die with actual patterns and design data (usually CAD data) are known, for example, Japanese Patent Publication No.
No. 463 or Japanese Patent Laid-Open No. 8-76359.

However, considering the present and future miniaturization of patterns, there is a problem that cannot be solved by the above two inspection methods. In die-to-die, the dimensions of individual patterns are subtly different on the same photomask, that is, variations in non-defective products cannot be ignored. Normally, in the case of die-to-die, in order to compare patterns existing at distant positions on the same photomask, pseudo-defects caused by such variations in non-defective products (they are erroneously recognized as defects though they are originally good products) It is necessary to suppress the inspection sensitivity in order to suppress (defects caused).

Further, in the die-to database, it is necessary to generate a pattern image from the design data in order to compare the design data with the actual pattern image. If the pattern image generated from this design data does not extremely match the actual pattern image, a pseudo defect will occur. In the actual pattern image, there are variations of non-defective products even on the same photomask as shown above. In other words, even if an image is generated from design data with reference to a certain pattern, there will be some subtle mismatches for the above reasons.

Further, regardless of the die-to-die and the die-to-database, the image itself obtained by the influence of the vibration etc. on the stage carrying the photomask and the optical system element for sensing during the process of taking the image. The reproducibility of is not sufficient. Further, with the miniaturization of patterns, the size of defects required for detection is becoming smaller nowadays, and it is becoming more and more difficult to detect image defects due to mis-size or misalignment. .

[0006] In addition, as for such an image defect detecting apparatus or image defect detecting method, Japanese Patent Laid-Open No. 59-20037.
Although Japanese Patent Laid-Open No. 2 is found, such a publication merely describes a technique of compressing digital image information in order to simplify a conversion processing circuit when processing an image,
No method is disclosed for detecting the defect itself in the image.

Further, Japanese Patent Laid-Open No. 2-215118 and Japanese Patent Laid-Open No. 5-249656 disclose mask inspection devices, but only the point of using transmitted light is described. There is no description about the technique of enlarging and reducing the image itself. Furthermore, JP-A-8-7
Japanese Patent No. 6359 describes that a transmitted light and a reflected light are used in combination in the inspection of a mask.
Japanese Patent Laid-Open Publication No. 304997 discloses a method of creating image data obtained by enlarging and reducing one image data of two image data and comparing each image data with the other image data at the time of image inspection. However, in both cases, it was impossible to cancel the error due to the influence of the optical system and the vibration.

[0008]

SUMMARY OF THE INVENTION An object of the present invention is to improve the above-mentioned drawbacks of the prior art and to detect an image defect in an image including the same photomask and the like, which is not affected by variations in non-defective products. A high-speed and high-accuracy image defect detection apparatus and a defect detection method therefor are provided.

[0009]

In order to achieve the above-mentioned object, the present invention adopts the technical constitution as described below. That is, as a first aspect of the present invention, means for creating inspection image information from an inspection object,
First image processing means for sequentially applying an expansion filter and a contraction filter to the inspection image information in this order, and a contraction filter and an expansion filter for the inspection image information in this order. Second image processing means for sequentially applying and performing image processing, and image information output for outputting image information composed of a difference value of image information output from each of the first image processing means and the second image processing means. The expansion filter and the contraction filter are each configured by a spatial filter in which a plurality of pixels that handle luminance information are arranged in a matrix, and a central pixel of the spatial filter is an inspection pixel, and The other pixel having the inspection pixel as the center is defined as a neighboring pixel, and the expansion filter compares the luminance data of the inspection pixel of the filter with the inspection pixel. Of the luminance data of the prime and the luminance data of the neighboring pixels, it is configured to be the maximum luminance data, the contraction filter, the luminance data of the inspection pixel of the filter, the luminance data of the inspection pixel and the neighboring pixels It is characterized in that it is configured to be the minimum brightness data of the brightness data of.

Further, as a second aspect of the present invention,
Means for creating inspected image information from an inspected object, a plurality of pixels for handling luminance information for the multiple inspected image information are arranged in a matrix, and a central pixel is an inspected pixel, and the inspected pixel Is composed of a spatial filter in which other pixels centered at are the neighboring pixels, and the luminance data of the inspection pixel of the spatial filter is the maximum luminance of the luminance data of the inspection pixel and the luminance data of the neighboring pixel. Sequential application of an expansion filter process for data and a contraction filter process for making the brightness data of the inspection pixel of the filter the minimum brightness data of the brightness data of the inspection pixel portion and the brightness data of the neighboring pixels in this order. Or first image processing means for performing image processing by sequentially applying the contraction filter and the expansion filter in this order,
It is characterized by comprising image information output means for outputting image information consisting of a difference value between the image information output from the first image processing means and the inspected image information.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Since the image defect detecting apparatus and the image defect detecting method according to the present invention have the above-described technical configurations, for example, a photomask used in semiconductor manufacturing is inspected. That is, it is possible to perform high-speed inspection without being affected by variations in non-defective products that cause the generation of pseudo defects.

That is, in the present invention, the input image is duplicated (copied) in each of the two memories, and the two filters are subjected to contradictory filter processing a proper number of times. Then, a difference image is created from these two images, and if there is a pixel in the difference image that is equal to or larger than the threshold value for determining a defect, the pixel is regarded as a defect and the corresponding portion of the captured original image is displayed on a monitor or the like. It is a thing.

That is, in the present invention, since the two images to be finally compared are originally the same image, the problem of non-defective product variation can be ignored. Further, in the present invention, since two images to be compared are created from the same input image, two different images such as a comparison / reference image immediately before the inspection, which is performed by a die-to-die or a die-to-database, can be precisely generated. Since it is not necessary to perform alignment (alignment processing) and the inspection algorithm of the present invention is a combination of simple filtering processing, it is possible to perform inspection at a higher speed than conventional inspection machines.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Specific examples of an image defect detecting apparatus and an image defect detecting method according to the present invention will be described in detail below with reference to the drawings. That is, FIG. 1 is a block diagram showing a configuration of a specific example of the image defect detection apparatus 30 according to the present invention. In the figure, a data conversion unit 2 for converting an image 1 to be inspected into digital image information, Data conversion means 2
The first storage means 7 for storing the digital image information output from the second storage means 8 for copying the digital image information stored in the first storage means 7 and individually storing the copied digital image information.
And the third storage means 9 and the digital image information stored in the second storage means 8 using the expansion filter
The first processing means 10 for executing the processing, the second processing means 11 for executing the second processing by using the contraction filter for the digital image information obtained by the calculation result of the first processing means 10, Third processing means 12 for performing second processing on the digital image information stored in the third storage means 9 by using a contraction filter, third processing means 1
The fourth processing means 13 for executing the first processing by using the expansion filter on the digital image information obtained by the calculation result of 2, the second processing means 11 and the fourth processing means 1.
Fifth processing means 14 for calculating the difference value of each digital image information obtained from the calculation result of No. 3, and fourth storage means 15 for storing the calculation result of the fifth processing means 14.
Characteristic image defect detection device 3 composed of
0 is shown.

The image defect detecting device 30 according to the present invention.
Will mainly describe a photomask such as a photomask having a predetermined pattern used for manufacturing a semiconductor device or the like, but a defect detecting apparatus 30 for the image and a defect detecting method thereof according to the present invention. Needless to say, is not limited to this specific example and is used for detecting defects in all images.

Further, the data conversion means 2 according to the present invention.
However, the configuration is not limited to the specific example, and any configuration can be used as long as its shape can be detected from the inspection image 1 and converted into digital image information. That is, in FIG. 1, a light beam output from one side of the photomask 1 mounted on the inspection stage 6 as an image to be inspected is transmitted through the photomask 1 by an appropriate light source 4, for example, a laser light source. It is shown that the transmitted light is received by an arbitrary light receiving element 5.

In the present invention, a thin laser beam is output from the light source 4 and is appropriately scanned in, for example, one method (for example, the Y direction) while the inspection stage is moved in the other direction (for example, the X direction). By moving, it is possible to inspect all the patterns arranged on the photomask 1. The light receiving element may be configured to move in synchronization with the laser.

Further, according to the present invention, a light source 4 having such a size as to irradiate the entire photomask 1 at the same time is used, and the light receiving element 5 is provided with light such as that used in a CCD camera. It is also possible to employ light receiving means.
Further, it is desirable that the data converting means 2 according to the present invention is provided with, for example, an analog / digital converting means 3.

In the present invention, the data conversion means 2
The digital image information converted by the above is temporarily stored in the first storage means 7 which is a memory having an appropriate configuration, and then stored in the first storage means 7 using the appropriate copying means 20. The digital image information thus obtained is copied and stored in both the second storage means 8 and the third storage means 9.

Further, the filter treatment used in the present invention is an expansion filter treatment and a contraction filter,
The filter used for the filtering is a so-called spatial filter 21 as illustrated in FIG. 2, and the size thereof is not particularly limited, but for example, 3 × as shown in FIG.
3 can be generally used, and it can be arbitrarily used according to the shape, characteristics, etc. of the image to be inspected, such as 8 proximity method or 4 proximity method.

Other 5 × 5 or 7 × 7 spatial filters can be used as well. In the present invention, it is desirable to handle the internal luminance data of the information of the digital image information. Further, in the present invention, when the expansion filter processing is performed using the spatial filter 21 composed of 3 × 3 pixels, for example, the pixel P (x, y) at the center of the spatial filter 21 is inspected. It is defined as a pixel, and includes the pixel data of the center of the inspection pixel, that is, the pixel data of its own pixel P (x, y) and eight pixels called eight neighboring pixels forming the peripheral portion, The pixel data having the largest luminance data replaces the data of the central pixel P (x, y).

That is, in the expansion filter processing,
Processing for increasing bright data will be executed. On the other hand, in the contraction filter processing, conversely, the pixel data at the center of the inspection pixel, that is, the pixel P (x, y) of its own and eight neighboring pixels forming the peripheral portion thereof are called eight neighboring pixels. The pixel data having the smallest luminance data, including the pixel data of the pixel, is the pixel P (x,
The data of y) is replaced.

That is, in the contraction filter processing,
Processing for increasing the dark data will be executed. In the present invention, one of the same digital image information is subjected to the first process that is the expansion filter process, and then the second process that is the contraction filter process is executed, and the other is the same. The digital image information is characterized in that, conversely, the second process which is the contraction filter process is executed first, and then the first process which is the expansion filter process is executed.

That is, in the present invention, by adopting such a configuration, it becomes possible to accurately extract the defective portion from the same digital image information. In the present invention,
For the two identical digital image information, the first
In the case of executing the processing of 1 or the second processing, it is possible to execute each of the processing only once, but it is also preferable to repeat the processing a plurality of times.

The number of repetitions does not necessarily have to be the same, and the number of repetitions of the first processing and the second
It may be different from the number of repeated executions of the process. However,
It is desirable that the numbers of times of processing for different digital image information be the same. That is, in the present invention, for one of the inspection pixels, a pair of processes including a minimum expansion filter process and a contraction filter process,
Alternatively, the pair of processes including the contraction filter process and the expansion filter process are executed once, but this pair of processes may be repeated a plurality of times, and as shown in FIG. The expansion filter process may be executed a plurality of times and then the contraction filter process may be executed a plurality of times, or the contraction filter process may be executed a plurality of times and the expansion filter process may be executed a plurality of times.

In the present invention, as described above, the opposite process is individually performed on the same digital image information, and the difference value is obtained by subtracting the respective arithmetic processing results. It is possible to clearly detect the defective portion in the inspection image 1. The difference value is stored in the fourth storage means 15, and then the operator arbitrarily reads the digital image information stored in the fourth storage means and compares it with a predetermined reference value or experience. It is also possible to judge whether or not there is a defective portion of the image, and the reference value is read from the storage means 17 in which a predetermined threshold value is stored in advance, and the comparison / judgment means 16 makes a comparison / judgment. Then, it is also possible to automatically determine the presence or absence of the defective portion of the image.

Further, in the present invention, it is possible to provide an appropriate display means 19 as necessary and display and inform the inspection result in an appropriate type. In addition, in FIG.
Although an example in which the output information of the determination means 16 is output to the display means 19 is described, the information of the fourth storage means 15 storing the calculation result of the subtraction means 14 is displayed in the display means 19.
Needless to say, it may be configured to output to.

In the present invention, the first processing means 1
0 and the second processing means 11 are provided with a first auxiliary storage means 22 for temporarily storing digital image information which is the calculation result of the first processing means 10, and the second auxiliary storage means 22. The digital image information which is the calculation result of the third processing means 12 is temporarily provided between the third processing means 12 for executing the processing of 4 and the fourth processing means 13 for executing the first processing. It is desirable that a second auxiliary storage means 23 for storing is provided.

With such a configuration, when the expansion filter process or the contraction filter process is performed, the newly formed digital image information is the original digital image stored in the second or third storage means 8 or 9. Since it is different from the image information, if the newly processed digital image information is stored again in the second or third storage means 8 or 9,
Since the contents of the digital image information are different, a problem occurs in the subsequent processing. Therefore, the temporary storage means 22, 23 is temporarily stored until the processing of all the pixels of the digital image information is completed.
This is because it must be stored in.

Then, the following filter processing is performed using the digital image information stored in the first and second auxiliary storage means 22 and 23, and the result is the second or the original storage means. It can be stored again in the third storage means 8 and 9. That is, in the present invention, as shown in FIG. 1, the first processing means 10 and the second processing means 11 are connected.
And a first auxiliary storage means 22 for temporarily storing the digital image information which is the calculation result of the first processing means 10, and the third processing means 12 and the fourth processing means 12.
Second image processing unit 13 which temporarily stores the digital image information which is the calculation result of the third processing unit 12
It is desirable that the auxiliary storage means 23 is provided.

In the present invention, the image information arithmetically processed by the second processing means 11 and the fourth processing means 13 is stored again in the second or third storage means 8, 9. Instead, a third auxiliary storage means 8 ′ is provided between the second processing means 11 and the fifth processing means 14, and the fourth processing means 13 and the fifth processing means 14 are combined. It is also desirable to separately provide the fourth auxiliary storage means 9'in between.

The fifth processing means 14 used in the present invention for calculating the difference value consisting of the subtraction means does not specify its configuration in particular, and conventionally known subtraction means can be used. . Further, in the present invention, the presence or absence of the defective portion of the image is judged by comparing the digital image information constituted by the difference value stored in the fourth storage means 15 with a predetermined reference value. The comparison determination means 16 and the display means 19 for displaying the output of the comparison determination means 16 are provided, and the reference value is, for example, the fifth storage means 17.
It is desirable to be stored in.

In the present invention, by displaying the image of the difference value as described above, not only the presence / absence of a defect in the image to be inspected but also the existence position and size of the defect part are displayed. Therefore, the information can be effectively used when the inspection image is corrected in a later process. Further, in the present invention, the difference value image information can display the magnitude of the difference value with brightness.

For example, since the magnitude of the difference value can be displayed in 256 levels of luminance value, the magnitude of the difference value and the luminance value are made to correspond to each other and the predetermined level of the luminance value is thresholded. By setting it as a value, it is possible to recognize a portion of the difference value image that exceeds the threshold value as a defective portion. Further, by displaying the difference value image in color, it becomes possible to easily recognize even a small image.

In the present invention, as shown in FIG. 1, not only the output information from the comparison / determination means 16 but also the information from the first storage means 7 is displayed on the display means 19 via the wiring L. Since it is configured to be input, the coordinate position in the difference value image can be recognized at the same time, so that the position of the defective portion can be easily confirmed. In the case of using the spatial filter 21 as shown in FIG. 2 used in the present invention, the wiring formed by a pattern shape with many straight lines of 0 degrees or 90 degrees, or the shape of a cell, etc. Effective for analysis.

However, most of the actual pattern shapes are composed of straight lines of 0 degree or 90 degrees, and therefore, by using the spatial filter 21 as shown in FIG. 2, most wiring patterns or It is possible to inspect the cell shape. In addition, it may be unsuitable for detecting a rhombus or a pattern having an oblique line such as 45 degrees or 135 degrees.

In such a case, it is necessary to use the spatial filter 21 to adopt the 4-neighbor system or to adjust the weighting of each pixel constituting the spatial filter 21. In addition, in the case of an image to be inspected, which has many patterns with other special diagonal lines, for example, patterns with a diagonal line of 22.5 degrees, it is necessary to adopt an appropriate spatial filter of 5 × 5 or larger. There is.

Even when using such a spatial filter,
It is necessary to adjust the weighting of each pixel. Of course, the 5
An appropriate spatial filter having a size of × 5 or more can be used for inspecting a wiring pattern or a cell shape which is a pattern composed of straight lines of 0 ° or 90 °.

In such a case, there is an advantage that the processing time is shorter than the case where the 3 × 3 spatial filter is used. However, in the filtering process of the present invention, there are cases where it does not work well due to the relationship between the characteristics of the spatial filter used and the shape of the image to be inspected. In this case, using a specific spatial filter, when using a specific detection method, in the filtering process in advance,
A part having a pattern that cannot be detected is detected, its position information is stored in a predetermined storage means 18, and when the address of the position information is detected, the filter process is not executed. It is desirable to exclude it from the display operation after the subtraction processing or the determination operation.

In other words, in the present invention, the comparison / determination means 16 has the look-up table 18 in which the position information of the image portion where the defect cannot be determined is stored in advance depending on the type of the filter used. It is also desirable to be connected. Specifically, first, a pattern in which a pattern composed of a straight line of 0 degrees or 90 degrees and a pattern having a diagonal line of 45 degrees or 135 degrees are mixed is CAD.
Etc.

The image information obtained by the CAD has no defective portion. Next, when the expansion processing or the contraction processing is executed on the image information by the CAD using the spatial filter 21 in which the weighting of all pixels is constant, for example, by the spatial filter as shown in FIG. Since the abnormal reaction is shown at the position of the degree or 135 degree pattern, it is understood that the portion in which the abnormal reaction has occurred is not the pattern of the 0 degree or 90 degree system. Alternatively, it is stored in an appropriate storage means provided in the CPU, and when the actual expansion processing and contraction processing are executed, the position is masked and the processing is not executed. Is desirable.

That is, in the present invention, in the inspection of the photomask, the image is distorted due to the influence of minute vibration, and even if the same design value pattern of the same photomask is used, the actual pattern is not always the same size. In order to solve the problem that it is difficult to detect micro defects in photomasks, which are becoming denser and finer due to the influences of not only the shape but also the shape, two different images from one image are solved. I decided to make an image of the nature.

That is, for example, a concave filter is corrected by an expansion filter, and a contraction filter that emphasizes the concave defect is applied on the other hand. After performing such processing a suitable number of times, reverse filter processing is performed by the same number to restore the image size. In other words, if the expansion filter is applied twice, the contraction filter is applied twice. By performing the combination of the expansion filter to the contraction filter and the combination of the contraction filter to the expansion filter for one image, it becomes possible to detect concave defects, convex defects, dots, and pinhole defects.

In these filter processes, only defective portions are changed and non-defective products are maintained as non-defective products. Therefore, micro-defects are not affected by the above-mentioned micro-distortion due to micro vibration and variations in non-defective products. Can be detected. Further, in some cases, the edges of the inspection target area have to be classified into angle groups before the filtering process. At this time,
It is necessary to perform effective filtering on the classified angle groups.

In the defect inspection of the conventional photomask, in order to compare with the image actually taken from the photomask, the same pattern image at different positions of the same photomask or the image generated from the design data is used. That is, it is necessary to always prepare two images for comparison and reference in advance, and it is difficult to detect a minute defect due to an error generated by this.

However, according to the present invention, since two images for comparison and reference are created by filtering from one image taken from an actual photomask, the error which has prevented the detection of minute defects up to now. The factor has disappeared. Moreover, since the two images for comparison and reference are originally the same image, it is sufficient to simply compare them. Filter processing used in the process of making two images for comparison / reference from one image, edge angle detection processing that is performed if necessary before that, and a difference image of the comparison / reference image is obtained to detect defects. Since the processing is a simple operation, it can be processed at high speed.

That is, in the defect detecting method according to the present invention, when the image defect portion K is present in the inspected image 1 as shown in FIG. 4, first, (1) in FIG. 4), when the digital image information is expanded on the image 1 to be inspected by using the expansion filter which is the first processing means, as shown in (2) of FIG. Is transformed into a state in which the digital image information shown in (2) of FIG. 4 (A) is shrunk by using a shrinking filter which is the second processing means. ), The boundary part is slightly deformed.

On the other hand, first, in (1) of FIG. 4B, the digital image information 1 having the same defect portion K as that of (1) of FIG. When the digital image information is contracted using the filter, the defective portion K becomes large as shown in (2) of FIG. 4 (B). After that, when the digital image information shown in (2) of FIG. 4B is expanded by using the expansion filter which is the first processing means, the digital image information is expanded as shown in (3) of FIG. 4B. In addition, the defective portion K becomes smaller than (2) in FIG.

Therefore, if the difference value between the two is taken, it is possible to find the existence of the defective portion of the image. Furthermore, in the present invention, when detecting a digital image, it is necessary to detect the angle of the edge of the digital image information and store in a separate memory which edge has which inclination. There is. A known filter may be used to detect the angle of the edge.

The following filter processing may need to be performed for each angle classified by the above edge angle detection.
For example, an edge is detected with an edge detection filter of an appropriate size to obtain 0, 90, 180, 270 degrees and 45, 135, 225,
The filter processing may be performed by classifying each group into 315 degrees or the like and using the effective filter for each group.

The structure and operation of the image defect detecting apparatus 30 according to the present invention will be described in detail below with reference to FIG. For example, the photomask 1 set on the inspection stage 6 is positioned. Next, the illumination 4 is irradiated from the upper surface of the photomask, the transmitted light is received by the light receiving element 5, and is converted into digital image data by the A / D converter 3.

At this time, it is assumed that the inspection stage is moved to the next inspection position while being controlled so that the sampled image does not move. The image (8 bits, gray scale) thus collected is stored in the first memory 7. Further, the stored image is further copied to the second storage means 8 and the third storage means 9 as shown in FIG.

The digital image information copied to the second storage means 8 is processed by the expansion filter processing using the first processing means 10, and the expansion-processed digital image information is stored in the first auxiliary storage means 22. To store. The digital image information stored in the first auxiliary storage unit 22 is expanded again by the expansion filter processing in the first processing unit 10, and the digital image information is overwritten in the second storage unit 8.

The digital image information stored in the second storage means 8 is contracted by the second processing means 11 to generate contracted digital image information, which is stored in the first auxiliary storage means 22. Overwrite. Next, the digital image information stored in the first auxiliary storage means 22 is contracted again by the contraction filter processing by the second processing means 11, and the contracted digital image information is stored in the second storage means 8. To overwrite.

In this embodiment, the last storage means may be a storage means other than the second storage means 8, for example, the above-mentioned third auxiliary storage means 8 '. is there.
In this example, the following processing is also performed in parallel with this.
That is, as shown in FIG. 3 (B), the digital image information copied to the third storage means 9 is reduced by the third processing means 12 using the contraction filter to reduce the digital image information. 2 to the auxiliary storage means 23.

The digital image information stored in the second auxiliary storage means 23 is contracted again by the contraction filter process using the third processing means 12, and the digital image information is stored in the third storage means 9 again. To overwrite. The digital image information stored in the third storage unit 9 is then expanded by the expansion processing by the fourth processing unit 13 to overwrite the digital image information expanded in the second auxiliary spare storage unit 23. .

Next, the second auxiliary spare storage means 2
The digital image information stored in No. 3 is overwritten in the third storage unit 9 with the digital image information expanded by the expansion processing by the fourth processing unit 13 again. In this specific example, the last storage means may use a storage means different from the third storage means 9, for example, the above-mentioned fourth auxiliary storage means 9 '.

After that, the subtraction means 14 creates the difference digital image information between the digital image information in the second storage means 8 and the digital image information in the third storage means 9. The brightness value of the difference digital image information thus obtained is compared with a preset threshold value that is determined to be a defect. For the portion determined to be defective, the defective portion of the memory 1 is displayed on a monitor or the like.

In the present invention, when the image quality is not good, the noise may be removed in the range necessary for the filter processing. Next, the structure of the image defect detection method according to the present invention will be described. As described above, the image defect detecting method according to the second aspect of the present invention is, for example, a defect detecting method for determining the presence or absence of an image defect in a predetermined inspection image,
At least two duplicates are created from digital image information of the inspected image obtained by digitally processing the inspected image, and an expansion filter is used for one of the duplicated digital image information. The second processing means using the processing means and the contraction filter in this order to obtain the first processing information, and the second copy using the contraction filter for the other duplicated digital image information.
The first processing means using the expansion filter and the first processing means using the expansion filter in this order to obtain the second processing information, and then the first processing means.
Is a method of detecting an image defect, in which a difference value between the second processing information and the second processing information is calculated, and whether the image defect exists in the image to be inspected is determined from the result.

In the defect detecting method according to the present invention, it is desirable that the first process and the second process are repeated once or plural times. Further, in the present invention,
It is desirable to compare the difference value with a predetermined reference value determined in advance to determine whether or not an image defect exists in the image to be inspected. Further, in the defect detection method according to the present invention, when determining whether or not an image defect is present in the image to be inspected, depending on the kind of filter used in the inspection, It is also desirable to refer to the information stored in advance that there is position information of the image part where the defect cannot be judged.

Here, a specific example of the data processing method according to the selection of the filter in the present invention and a specific example of the step of comparing the difference value with an appropriate reference value will be described with reference to the flowchart of FIG. . That is, first, after the start, the image to be inspected is selected in step (S1),
In 2), a filter suitable for the inspected image is selected, and a pixel position inappropriate for processing with the selected filter, that is, a pixel position where a defective portion of the image cannot be identified is determined in advance. Then, the position information is stored in the sixth storage means 18 in FIG. 1, for example.

Next, in step (S3), copy image information is created from the inspected image information and step (S4).
Then, the expansion / contraction filter processing as described above is executed. Then, in step (S5), the difference value is calculated in the calculating means 14 from the two types of image information subjected to the expansion / contraction filter processing, and the result is stored in the fourth storing means 15 in step (S6). Store.

Then, in step (S7), as shown in FIG.
The fourth storage means 1 in the determination means 16 in
The difference value data stored in 5 is sequentially read out for each pixel position, and in step (S8), the pixel positions of the individual difference value data that are sequentially read out are stored beforehand and should be excluded. It is determined whether it corresponds to the pixel position,
If the read position data (i = n, j = m in Xi, Yj) of the difference value data corresponds to the pixel position to be excluded, it is excluded from the subsequent processing steps in step (S9). To be done.

If the position data of the read difference value data does not correspond to the pixel position to be excluded, the difference value data is held in step (S10). Next, in step (S11), it is determined whether or not the difference value data of the read inspection image is the last one, and if NO, step (S12).
, The next difference value data is selected, and the position data is incremented by 1 (i = n + 1, j = m + for Xi and Yj).
1), returning to step (S8), the above operation is repeated.

Then, if the difference value data is the last position data in step (S11), all the held difference value data are converted into luminance data in step (S13), and in some cases, In step (S14), the display means 19 displays the result, and in step (S15), the result is compared with a reference value stored as a threshold value in the fifth storage means 17, In step (S16), if the brightness data obtained from the difference value is smaller than the reference value of the brightness data, in step (S17) the inspection image has no defective portion. However, if the luminance data obtained from the difference value is larger than the reference value of the luminance data in step (S16),
In step (S18), it is determined that the image to be inspected has a defective portion, and in step (S19), the difference value data is displayed on the display means 19 together with the position information, and the process ends.

If the method of detecting defects in the image according to the present invention is shown as another aspect, the first step of converting the image to be inspected into digital image information, the digital image information in the first step.
Second step of storing the digital image information stored in the first storage means in the second storage means and the third step of storing the digital image information stored in the first storage means, respectively. A fourth step of performing a first process on the digital image information stored in the means using an expansion filter,
A fifth step of performing a second process by using a contraction filter on the digital image information obtained in the fourth step, and using a contraction filter for the digital image information stored in the third storage means. And a sixth step of executing the second processing, a seventh step of executing the first processing by using an expansion filter on the digital image information obtained in the sixth step, and a fifth step of the same. And an eighth step of calculating a difference value from the respective processing information obtained in the seventh step, and a ninth step of storing the difference value obtained in the eighth step in the fourth storage means. This is a method for detecting a defect in a configured image.

In the above, it is also desirable to appropriately repeat the fourth step and the fifth step, and it is also desirable to appropriately repeat the sixth step and the seventh step. The present invention further includes a tenth step of comparing the difference value with a predetermined reference value determined in advance to determine whether or not an image defect exists in the image to be inspected. Is preferred.

Further, in the present invention, in the tenth step, depending on the type of filter used for the inspection, there is position information of the image portion where the defect of the image cannot be judged. Eleventh, referring to information that stores things in advance
It is also desirable to include the step of. In the present invention, the number of repetitions of the first process using the expansion filter and the second process using the contraction filter is not particularly limited, but the number of repetitions of the pattern in the image to be inspected is not limited. It may be limited by the size, interval, etc. Especially when contraction processing is performed, as a result of processing, if adjacent patterns come into contact with each other or overlapping portions occur, subsequent processing becomes difficult. From this perspective,
The number of repetitions of each process is limited.

The selection criterion of the spatial filter used in the present invention is not particularly limited,
Preferably, the shapes and sizes of the wiring patterns or cells in the image to be inspected, the intervals between them, the angles of the constituent patterns (there are many 0 degree / 90 degree patterns, 45 degree /
It is possible to select an appropriate spatial filter from information such as whether there are many 135 degree system patterns or other patterns such as 22.5 degree system patterns.

At the same time, naturally, the distribution of weighting in each pixel portion of the selected spatial filter, the number of times of repetition described above, etc. are naturally incorporated in the selection elements. Therefore, it is preferable to easily and automate the selection of the spatial filter by preparing a look-up table that defines the relationship between a predetermined pattern and the spatial filter to be used, based on a large number of simulation results. Things are possible.

Next, another aspect of the image defect detection method according to the present invention will be described. In other words, in this specific example, the size of defects required for detection is becoming smaller and smaller with the miniaturization of patterns now.For example, for defects due to miss size or defects due to misalignment, the detection is much more difficult. Is getting difficult. For example, when it is desired to obtain a pattern having a plurality of normal cell shapes created by using CAD as shown in FIG. 5, for example, as shown in FIG. 6, four patterns at the center of the pattern are provided. When the cell is larger than the surrounding pattern, it is difficult to detect the defect by the conventional image defect detecting method.

However, by using the image defect detecting method according to the present invention, the image defect can be detected efficiently and accurately. That is, FIG.
The image with the defect shown in FIG.
The expansion filter which is a spatial filter as shown in FIG. 7 is used to execute the first process as described above to obtain image data as shown in FIG. 7A, and then the contraction filter similar to that described above. By executing the second process by using, the image data as shown in FIG. 7B can be obtained.

On the other hand, the same image having the same defective portion shown in FIG. 6 is first subjected to the second process using the same contraction filter as described above for the data, so that the data shown in FIG. After the image data as shown in A) is obtained, the same expansion filter as described above is used to execute the first process, whereby the image data as shown in FIG. 8B is obtained. become.

In this example, the image data is
When the expansion filter is applied first, the large size defect is connected to the surrounding pattern. Therefore, even when the contraction filter is applied later, the connected portion does not return to the original shape. On the other hand, if the image data having the defective portion is first processed by the contraction filter, the shape of the connection portion does not disappear even after the expansion processing is performed, and therefore, the shapes of FIG. 7B and FIG. By calculating the difference value of the image data of (B), the position of the defective portion in the image as shown in FIG. 9 is displayed.

The present invention has an effect that not only the image defects of the mis-size type, but also the image defects based on the positional deviation can be easily detected. Next, another specific example of the image defect detection apparatus and the image defect detection method according to the present invention will be described in detail with reference to FIG. That is, as the present specific example, the above specific example is one in which one image to be inspected is copied, and the image data to which each inspected image is subjected to expansion filter processing / contraction filter processing and contraction filter processing. / The basic technical idea was to create image data that has been subjected to expansion filter processing and calculate the difference value between the two, but in this specific example, expansion is performed on the original inspection image. The image data subjected to any one of the filter processing / contraction filter processing or the contraction filter processing / expansion filter processing is compared with the original inspection image, and the difference value is calculated.

Specifically, a means for creating a duplicate image to be inspected from one image to be inspected, and an expansion filter and a contraction filter are sequentially applied to the duplicate image to be inspected in this order, or a contraction filter is used. A first image processing unit that sequentially applies an expansion filter in this order to perform image processing, or performs image processing by repeating the image processing a plurality of times, and an image output from the first image processing unit. An image defect detecting device comprising image information output means for outputting image information consisting of information and a difference value between the inspected image, and a step of preparing one inspected image,
Creating a duplicate inspection image from the inspection image,
For the duplicate inspection image, an expansion filter and a contraction filter are sequentially applied in this order to perform image processing, or a contraction filter and an expansion filter are sequentially applied in this order to perform image processing. First time to repeat
Image processing step, the image information output from the first image processing step and the image to be inspected are compared, and a difference value calculating step of calculating a difference value between the image information and the image information including the difference value is displayed. An image defect detection method comprising an output step.

That is, in this example, after copying the original image data of the image to be inspected stored in the first storage means 7 by the copying means 20, the copied image data is The same image processing as the first image processing means A described in FIG. 1 is performed, and the result is compared with the original image data of the inspection image stored in the first storage means 7 and the calculation is performed. The means 14 calculates the difference value data, and the subsequent data processing is the same as the processing shown in FIG. 1 and described in the above concrete example.

Therefore, the image data (1) of the original image to be inspected in FIG. 4 is compared with the image data (3) processed by the first image processing means A, and the difference value is calculated. Will be Similarly, after the original image data of the image to be inspected stored in the first storage means 7 is copied by the copying means 20, the copied image data is converted into the second image data described in FIG. The same image processing as that of the image processing means B is performed, and the result is stored in the first storage means 7.
The difference value data is calculated by the calculating means 14 in comparison with the original image data of the image to be inspected stored in FIG. 1, and the subsequent data processing is shown in FIG. The same process as the process described above is executed.

Therefore, the image data (1) of the original image to be inspected in FIG. 4 is compared with the image data (3) processed by the second image processing means B, and the difference value is calculated. Will be The difference value data obtained by such a method is not so different from the case of the above-mentioned specific example, and it has been proved that it has sufficient practicality.

Still another aspect of the image defect detecting method according to the present invention relates to a defect detecting method for judging the presence / absence of an image defect in a predetermined image to be inspected, in which the image to be inspected is digitally processed. At least two duplicates are created from the digital image information of the inspected image obtained by the above-mentioned method, and one of the duplicated digital image information uses the first processing means using the expansion filter and the contraction filter. The first processing information is obtained by applying the second processing means in this order to obtain the first processing information, and the second processing means using the contraction filter and the expansion filter are used for the other duplicated digital image information. Processing means for obtaining the second processing information in this order, and then the difference value between the first and second processing information is calculated. From the result, there is an image defect in the inspection image. A recording medium recording a program for executing each processing of determining whether Luke the computer.

[0081]

The effect of the present invention is that, for example, when inspecting a photomask used in semiconductor manufacturing, high-speed inspection can be performed without being affected by variations in non-defective products that cause pseudo defects. The present invention, as described above,
The input digital image information is copied (copied) into each of the two memories, and the two digital image information items are subjected to the contradictory filtering process a suitable number of times. Image information is created, and if there is a pixel in the difference digital image information that is equal to or greater than a threshold value for determining a defect, then that pixel is regarded as a defect, and the corresponding portion of the captured original digital image information is displayed on a monitor or the like.

That is, since the two pieces of digital image information finally compared in the present invention are originally the same digital image information, the problem of non-defective product variation can be ignored. The present invention also compares 2 from the same input digital image information.
Since the digital image information of one sheet is created, it is not necessary to precisely align (alignment processing) two different digital image information such as comparison / reference digital image information immediately before the inspection which is performed by a die-to-die or a die-to database. Further, since the inspection algorithm of the present invention is a combination of simple filtering processes, it is possible to detect minute defects faster than conventional inspection machines.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a specific example of an image defect detection apparatus according to the present invention.

FIG. 2 is a diagram showing a specific example of a spatial filter used in the image defect detection apparatus according to the present invention.

FIG. 3 is a flowchart showing a specific example of an operation procedure in the image defect detecting method according to the present invention.

FIG. 4 is a diagram showing an example of a change in an image defect portion in the image defect detection method according to the present invention.

FIG. 5 is a diagram showing a configuration example of a CAD pattern of a normal shape used in still another specific example of the image defect detection method according to the present invention.

6 is a diagram showing a configuration example of a pattern shape of an image having a defective portion manufactured from the CAD pattern in FIG.

FIG. 7 is a diagram showing an operation result in still another specific example of the image defect detection method according to the present invention,
7A is a pattern obtained when the image pattern of FIG. 6 is subjected to the first processing, and FIG. 7B is shown in FIG.
It is a pattern obtained when the second processing is performed on the image pattern of (A).

FIG. 8 is a diagram showing an operation result in still another specific example of the image defect detecting method according to the present invention,
FIG. 8A is a pattern obtained when the image pattern of FIG. 6 is subjected to the second processing, and FIG.
It is a pattern obtained when the first processing is performed on the image pattern of (A).

FIG. 9 is a diagram showing an operation result in still another specific example of the image defect detection method according to the present invention, and is a diagram of FIG. 7 (B) and FIG. 8 (B). It is a pattern diagram formed by taking a difference value.

FIG. 10 is a flowchart illustrating an operation procedure in a specific example of the image defect detection method according to the present invention.

FIG. 11 is a block diagram for explaining an example of an image defect detection apparatus for executing the method of another specific example of the image defect detection method according to the present invention.

[Explanation of symbols]

1 ... Inspected image 2 ... Data conversion means 3 ... A / D converter 4 ... Light source 5 ... Light receiving element 6 ... Inspection stage 7 ... First storage means 8 ... Second storage means 8 '... Third auxiliary storage means 9 ... Third storage means 9 '... Fourth auxiliary storage means 10 ... First processing means 11 ... Second processing means 12 ... Third processing means 13 ... Fourth processing means 14 ... Difference value calculating means 15 ... Fourth storage means 16 ... Comparison determination means 17 ... Threshold storage means, fifth storage means 18 ... Image defect determination impossible position storage means 19 ... Display means 20 ... Copying means 21 ... Spatial filter 22 ... First auxiliary storage means 23 ... Second auxiliary storage means 30 ... Image defect detection device

Front page continuation (51) Int.Cl. ⁷ Identification code FI theme code (reference) H01L 21/027 H01L 21/30 502P F term (reference) 2G051 AA51 AB02 BA10 CA04 CB02 DA07 EA08 EA12 EA14 EB01 EC06 ED08 ED14 2H095 BD04 BD24 5B057 AA03 BA02 CA12 CA16 CE06 CF01 CF02 DA03 DB02 DC32 DC36 5L096 BA03 EA02 FA14 GA08 GA51 GA55 JA11

Claims (4)

[Claims] 1. A means for creating image information to be inspected from an object to be inspected, and a first image processing means for applying image processing to the image information to be inspected by sequentially applying an expansion filter and a contraction filter in this order. , Second image processing means for sequentially applying a contraction filter and an expansion filter to the image information to be inspected in this order, the first image processing means, and the second image processing means, respectively. Image information output means for outputting image information composed of difference values of image information output from the expansion filter and the contraction filter, each of which is a space in which a plurality of pixels for handling luminance information are arranged in a matrix. The pixel is composed of a filter, the central pixel portion of the spatial filter is an inspection pixel, and the other pixels centered on the inspection pixel are neighboring pixels; The filter is configured such that the luminance data of the inspection pixel of the filter is the maximum luminance data of the luminance data of the inspection pixel and the luminance data of the neighboring pixels, and the contraction filter is the inspection pixel of the filter. The image defect detection device is characterized in that the luminance data of (1) is the minimum luminance data of the luminance data of the inspection pixel and the luminance data of the neighboring pixels. 2. A step of creating image information to be inspected from an object to be inspected, a plurality of pixels for handling luminance information are arranged in a matrix for the image information to be inspected, and a central pixel is an inspection pixel, In addition to configuring the other pixel portion centering on the inspection pixel with a spatial filter that is a neighboring pixel, the luminance data of the inspection pixel of the spatial filter,
Of the brightness data of the inspection pixel and the brightness data of the neighboring pixels, the expansion filter processing to obtain the maximum brightness data, the inspection pixel data of the spatial filter, the brightness data of the inspection pixel and the brightness data of the neighboring pixels Of which
A first image processing step of sequentially applying contraction filter processing to obtain the minimum luminance data in this order, and an image obtained by sequentially applying the contraction filter processing and expansion filter processing to the inspected image information in this order. An image comprising a second image processing step of processing, and a step of calculating a difference between image information output from each of the first image processing step and the second image processing step. Defect detection method. 3. A means for generating image information to be inspected from an object to be inspected, a plurality of pixels for handling luminance information for the multiple image information to be inspected are arranged in a matrix, and a central pixel is set as an inspection pixel. In addition, the pixel is composed of a spatial filter in which other pixels centered on the inspection pixel are used as neighboring pixels, and the luminance data of the inspection pixel of the spatial filter is converted into the luminance data of the inspection pixel and the luminance data of the neighboring pixel. Of the brightness data of the inspection pixel and the brightness data of the neighboring pixels, the expansion filter processing to obtain the maximum brightness data and the brightness data of the inspection pixel of the filter,
The first image processing means for applying image processing by sequentially applying contraction filter processing to obtain the minimum luminance data in this order or sequentially applying the contraction filter and expansion filter in this order, and the first image. An image defect detecting device comprising: image information output means for outputting image information consisting of a difference value between the image information output from the processing means and the inspected image information. 4. A step of creating image information to be inspected from an object to be inspected, a plurality of pixels for handling luminance information are arranged in a matrix for the image information to be inspected, and a central pixel is an inspection pixel, In addition, the other pixel portion centering on the inspection pixel is configured by a spatial filter having neighboring pixels, and the luminance data of the inspection pixel is maximum among the luminance data of the inspection pixel and the luminance data of the neighboring pixel. Expansion filter processing configured to obtain the luminance data of the pixel and contraction configured to set the data of the inspection pixel to the minimum luminance data of the luminance data of the inspection pixel and the luminance data of the neighboring pixels. A first image processing step in which image processing is performed by sequentially applying filter processing in this order, or image processing is performed by sequentially applying the contraction filter and the expansion filter in this order. And comparing the image information output from the first image processing step with the inspected image information,
And a difference value calculating step of calculating the difference value.