JP2002005850A - Defect inspection method and apparatus, mask manufacturing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To extract defects at a higher accuracy, without having to position image data nor to store a plurality of quality image data. SOLUTION: A defect extraction image data are found as the difference between an original defect image data obtained by photographing the side 10a of a glass substrate 10 and a defect shift image data obtained by shifting the original defect image data by pixel units in the direction determined by a pattern formed on the side 10a of the glass substrate 10. A defect part 19 such as a flaw, a ship or a crack, existing on the side 10a of the glass substrate 10, is extracted by searching for luminance in the defect extraction image data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a glass substrate or an IC for a mask used in an exposure process when manufacturing a semiconductor device.
The present invention relates to a defect inspection method and an apparatus for inspecting the appearance of a surface of a mold, an IC chip, and the like, and a method of manufacturing a mask.

[0002]

2. Description of the Related Art For example, in a method of inspecting the appearance of an object to be inspected such as an IC chip or a printed circuit board, image data of a non-defective object to be inspected is stored in advance, and the image data and the IC chip or the printed circuit board are imaged. A defect on an IC chip or a printed circuit board is extracted from a difference from image data obtained by the method.
For example, FIG. 12 is a schematic diagram of image data obtained by imaging an IC chip with a camera. In order to perform the appearance inspection of the IC chip surface, for example, FIG.
Non-defective image data 2 shown in FIG. 3 is stored in advance. In this case, in order to extract a small defect 3 on the surface of the IC chip, it is necessary to increase the resolution of the image, and accordingly, the field of view of a camera for imaging the IC chip or the like becomes narrow. For this reason, the non-defective image data 2 is stored in advance not only in the inspection area 1 but also a plurality of non-defective image data divided so as to cover the entire surface of the IC chip.

However, for the inspection area 1, the non-defective image data 2 of the area 1 is read out and aligned with the inspection area 1 of the image data obtained by the image pickup by the camera. And the image data 2 of the non-defective product are obtained. If the defect 3 exists in the inspection area 1, the defect 3 is extracted from the difference image data as shown in FIG. Hereinafter, in the same manner as described above, for each inspection area, a difference between the image data in each inspection area and the non-defective image data is obtained, and defects are extracted.

[0004]

However, the above-mentioned appearance inspection method has the following problems. Unless the non-defective image data 2 is aligned with the inspection area 1 of the image data obtained by the imaging of the camera with high accuracy, for example, noise occurs near the edge and the defect 3 cannot be determined. Further, a plurality of good image data 2 must be stored in advance.
If the illumination state of the IC chip surface changes or if there is uneven illumination, the defect 3 cannot be correctly determined from the difference image data.
Further, when the number of non-defective image data 2 is large, it takes a lot of trouble to extract a defect.

Accordingly, an object of the present invention is to provide a defect inspection method and apparatus capable of extracting defects with high accuracy without aligning image data and storing a plurality of non-defective image data. .

Another object of the present invention is to provide a mask manufacturing method capable of manufacturing a mask by performing accurate defect extraction without aligning image data and storing a plurality of non-defective image data. With the goal.

[0007]

According to the present invention, image data obtained by imaging an object to be inspected and this image data are shifted by a predetermined area unit in a direction determined by a pattern formed on the object to be inspected. A defect inspection method characterized by obtaining image data of a difference from image data and extracting a defect on the inspection object by searching for luminance in the difference image data.

Further, according to the present invention, with respect to image data obtained by imaging an object to be inspected, the degree of coincidence of image data in these small areas or the degree of coincidence of each of the small areas is shifted while shifting two adjacent small areas. A defect inspection method characterized by determining one or both of the positional deviations and identifying a defect on the inspection object based on one or both of the degree of coincidence and the positional deviation.

The present invention also relates to a method of combining image data obtained by imaging an object to be inspected and image data obtained by shifting the image data by a predetermined area in a direction determined by a pattern formed on the object to be inspected. The image data of the difference is obtained, and a defect on the object to be inspected is extracted by searching for the luminance in the difference image data. Thereafter, the two small regions adjacent to the image data are shifted while the small regions are shifted. One or both of the degree of coincidence of each image data and the position shift of each of the small areas are determined, and the defect on the inspection object is identified based on one or both of the degree of coincidence and the position shift. This is a defect inspection method characterized by the following.

Further, according to the present invention, an object to be inspected is imaged by an image pickup means, and image data obtained by the image pickup means is subjected to image processing using the defect inspection method of the present invention. A defect inspection apparatus for inspecting a defect on a body.

According to the present invention, there is provided a mask manufacturing method for manufacturing a mask by forming a circuit pattern of a semiconductor device on a glass substrate, wherein image data obtained by imaging the glass substrate and the image data Finding image data of a difference from image data shifted by a predetermined area unit in a direction determined by the circuit pattern formed on the glass substrate, and extracting a defect on the glass substrate by searching for luminance in the difference image data A mask manufacturing method.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (1) Hereinafter, a first embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram of a defect inspection apparatus. The glass substrate 10 to be inspected is a glass substrate of a mask used in an exposure step when manufacturing a semiconductor device. This mask is manufactured by forming a circuit pattern of a semiconductor device on a glass substrate 10. Such a glass substrate 10
Are those in which the size of defects such as scratches, chips, cracks, etc. is smaller than a predetermined size, especially the side surface 1 of a plate-like glass substrate.
There is a demand for a defect having a high quality in which the size of a defect generated in Oa is smaller than a predetermined size. These defects such as scratches, chips, and cracks are caused by remaining abrasive in a polishing process for manufacturing the glass substrate 10. Then, defects such as scratches and chips generated on the side surface 10a of the glass substrate 10,
In particular, since cracks may progress, if the size is not less than a predetermined size, for example, 200 to 300 microns, the glass substrate 10 is regarded as a defective product. Therefore, in reality, 50 to 50
There is a need to extract defects as large as 100 microns. The edge of the side surface 10a of the glass substrate 10 is chamfered 10b.

At a position facing the side surface 10a of the glass substrate 10, a camera 11 is provided as imaging means. The camera 11 is a CCD camera or a television camera, and captures an image of the side surface 10a of the glass substrate 10 and outputs an image signal.

An illuminating device 12 is provided on a side of the glass substrate 10 opposite to the side surface 10a. The illumination device 12 illuminates the side surface 10a of the glass substrate 10 uniformly.

The personal computer 13 inputs an image signal output from the camera 11 through the image input board 14, stores the image signal in an internal image memory as image data, and processes the image data to perform image processing.
It has a function of extracting a defect existing on the 0 side surface 10a.

The personal computer 13 has the functions of an image shifter 15, a difference image calculator 16 and a defect extractor 17 for extracting defects such as scratches, chips, cracks, etc., present on the side surface 10a of the glass substrate 10. Have.

The image shifting means 15 converts the image data obtained by the image pickup by the camera 11 into the side surface 10 of the glass substrate 10.
Since the chamfer 10b is applied in a direction determined by the pattern formed on the a, here, the chamfer 10b is plain, the image data is divided into a predetermined area unit, specifically a pixel unit, in a direction perpendicular to the direction of the chamfer 10b. , For example, by about five pixels. In addition, as a predetermined area unit,
The sub-pixels can be shifted in pixel units, and the processing time becomes long, but highly accurate defect detection can be performed.

The difference image calculating means 16 is provided with the image shifting means 1.
5, the image data (hereinafter, referred to as defect original image data) which is different from the image data before being shifted (hereinafter, referred to as defect original image data) and the image data after being shifted (hereinafter, referred to as defect shifted image data).
(Referred to as defect extraction image data).

The defect extracting means 17 includes a difference image calculating means 16
Has a function of searching for the luminance in the defect extraction image data obtained by the above and extracting defects such as scratches, chips, cracks, etc., present on the side surface 10a of the glass substrate 10.

The personal computer 13 monitors the original defect image data stored in the image memory, the defect shifted image data shifted by the image shifting means 15, or the defect extracted image data obtained by the difference image calculating means 16. 18 is provided.

Next, the operation of the device configured as described above will be described.

The side surface 10a of the glass substrate 10 is illuminated with illumination light from the illumination device 12 to be uniformly illuminated.

The camera 11 images the side surface 10a of the glass substrate 10 which is uniformly illuminated, and outputs an image signal.

The personal computer 13 inputs an image signal output from the camera 11 through the image input board 14, and stores the image signal in the internal image memory as defect original image data.

FIG. 2 is a schematic diagram of defect original image data.
In the glass substrate 10, there is no pattern or pattern on the surface,
If there is no defect such as a scratch, chip, or crack, the defect original image data has a uniform brightness. When a defect is present, the brightness of the defective portion is not uniform as compared with its surroundings, and is observed as a defect. In the defect original image data shown in FIG. 2, the chamfer 10b of the glass substrate 10 is shown, and a scratch defect portion (defect portion)
19 is shown.

In order to extract such a defective portion 19, the image shifting means 15 converts the original image data stored in the image memory into a direction determined by the pattern formed on the side surface 10 a of the glass substrate 10, in this case, a plain image. Is chamfered 10b, and is shifted, for example, by about 5 pixels in a direction (X direction) perpendicular to the direction of the chamfer 10b.
FIG. 3 is a schematic diagram of the defect shifted image data after the defect original image data is shifted by about 5 pixels in the X direction.

Next, the difference image calculating means 16 is a defect extraction image which is a difference between the defect original image data (FIG. 2) before shifting by the image shifting means 15 and the defect shifted image data (FIG. 3) after shifting. Ask for data. FIG. 4 shows a schematic diagram of the defect extraction image data.

Next, the defect extracting means 17 searches the luminance in the defect extracted image data obtained by the difference image calculating means 16 and extracts a defective portion 19 present on the side surface 10a of the glass substrate 10. Further, the defect extracting means 17 specifies the position and size of the defective portion 19 by binarizing the defect extracted image data.

Thereafter, the circuit pattern of the semiconductor device is formed on the surface of the non-defective glass substrate 10, and a mask used in an exposure process when the semiconductor device is manufactured is manufactured.

As described above, in the first embodiment, the original defect image data obtained by imaging the side surface 10a of the glass substrate 10 and the original defect image data
The defect extraction image data, which is a difference from the defect-shifted image data shifted in units of pixels in the direction determined by the pattern formed on the side surface 10a, is obtained, and the luminance in the defect-extracted image data is searched for. Side 1 of
Since the flaw defect portion 19 existing on 0a is extracted, the defect portion 19 can be extracted with high accuracy without aligning the image data and storing a plurality of non-defective image data. Then, the glass substrate 10 determined to be non-defective
Can be manufactured as a mask used in an exposure process when manufacturing a semiconductor device.

If the number of shifted pixels of the original defect image data at the time of obtaining the defective shifted image data is changed according to the size of the defective portion 19, a small scratch defect portion 19, for example, 5
The defective portion 19 having a size of 0 to 100 microns can be extracted.

In the defect extraction image data shown in FIG. 4, the chamfer 10b at the edge of the glass substrate is also extracted as the extraction portion 10c. However, since the coordinate positions of these chamfers 10b are known in advance, these extraction portions 10c Can be identified as not a defect.

(2) Next, a second embodiment of the present invention will be described with reference to the drawings. The same parts as those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof will be omitted.

FIG. 5 is a block diagram of the defect inspection apparatus. The personal computer 13 is provided on the side surface 10 of the glass substrate 10.
In order to extract a defect such as a scratch, a chip, or a crack existing in a, a function of the small area calculating means 20 and a function of the defect identifying means 21 are provided.

As shown in FIG. 6, the small area calculating means 20 shifts two adjacent small areas A and B from the defect The pattern matching of each of the image data is performed to determine the degree of coincidence between the small areas A and B or the small areas A and B.
Has a function of obtaining one or both of the positional deviations. Note that the degree of coincidence between the small areas A and B is calculated, for example, by performing normalized correlation.

The defect discriminating means 21 comprises a small area calculating means 20
Has a function of identifying the flaw / defect portion 19 present on the side surface 10a of the glass substrate 10 based on one or both of the degree of coincidence and the positional deviation obtained by the above.

Next, the operation of the device configured as described above will be described.

The side surface 10a of the glass substrate 10 is illuminated with illumination light from the illumination device 12 to be uniformly illuminated.

The camera 11 captures an image of the side surface 10a of the glass substrate 10 that is uniformly illuminated, and outputs an image signal.

The personal computer 13 inputs an image signal output from the camera 11 through the image input board 14, and stores the image signal in the internal image memory as defect original image data shown in FIG.

The small area calculating means 20 sets two adjacent small areas A and B as shown in FIG. 6 for the image portion corresponding to, for example, the chamfer 10b in the defect original image data obtained by the imaging by the camera 11. , These small areas A and B
Is shifted in the direction in which the chamfer 10b is formed (Y direction).

At the same time, the small area calculating means 20 performs pattern matching of each image data in the small areas A and B while shifting the small areas A and B, and
Either one or both of the degree of coincidence of B and the displacement of these small areas A and B are obtained.

As described above, when one or both of the degree of coincidence and the displacement of each of the small areas A and B are obtained, if the pattern patterns in these small areas A and B are coincident, each of the small areas A and B is obtained. The degree of coincidence of B is high and shows a constant value, and no displacement occurs between the small areas A and B.

However, as shown in FIG.
If the defective portion 22 exists above, for example, one of the small areas A
Is a pattern of the chamfer 10b having no defect, whereas the other small area B has a pattern in which the defective portion 19 is present.
As a result, the degree of coincidence between the small areas A and B is reduced.

Further, while the pattern of the chamfer 10b having no defect in one of the small areas A is a pattern having the defective portion 19 in the other small area B, these small areas A,
By performing pattern matching between B, the position of the other small area B is shifted in the X direction. As a result, a displacement occurs between the small areas A and B.

Therefore, the defect identification means 21 is located on the side surface 10 a of the glass substrate 10 based on one or both of the degree of coincidence and the positional shift between the small areas A and B obtained by the small area calculation means 20. Defective portions 19 such as scratches, chips, and cracks are identified.

As described above, in the second embodiment, two small areas A and B adjacent to each other are shifted with respect to the original defect image data obtained by imaging the side surface 10a of the glass substrate 10. Either or both of the coincidence and the displacement of each image data in the regions A and B are obtained, and the defective portion 19 on the side surface 10a of the glass substrate 10 is determined based on either or both of the coincidence and the displacement. As described above, the defect existing in the edge portion of the glass substrate 10 without aligning the image data and storing a plurality of non-defective image data as in the first embodiment. The portion 19 can be extracted with high accuracy.
Further, since the comparison is made between two adjacent small areas A and B, the influence of uneven illumination is small. Then, the glass substrate 10 determined to be non-defective can be manufactured as a mask used in an exposure step when manufacturing a semiconductor device.

When a defect 19 is present at the edge of the glass substrate 10, since there is a lot of noise at the edge, even if the defect is extracted in the first embodiment, the defect 19 and the noise remain. May not be easily identified.

For example, FIG. 7 shows an edge portion (chamfer 10
FIG. 4B is a schematic diagram of defect original image data in which a defective portion 19 exists in FIG. The defect original image data is shifted by a predetermined number of pixels with respect to the defect original image data by using the defect extraction method of the first embodiment, and the difference between the defect shifted image data and the defect original image data is determined. Even when the extracted image data is obtained, the flaw / defect portion 19 can be extracted as in the case of the defect extracted image data shown in FIG. 8, but noise occurs at the edge portion.

Therefore, with respect to the edge portion of the glass substrate 10, if the degree of coincidence or displacement of each image data in two adjacent small areas A and B is obtained by using the second embodiment, It is possible to reliably identify a defective portion at an edge portion where noise occurs.

In other words, by combining the first embodiment and the second embodiment, the defect original image data obtained by imaging the side surface 10a of the glass substrate 10 and the defect original image Defect extraction image data which is a difference from the defect shifted image data in which the data is shifted pixel by pixel in a direction determined by the pattern formed on the side surface 10a of the glass substrate 10 is searched for, and the luminance in the defect extraction image data is searched. A defect portion 19 such as a scratch, a chip, or a crack existing on the side surface 10a of the glass substrate 10 is extracted, and two adjacent neighboring defect original image data obtained by imaging the side surface 10a of the glass substrate 10 are extracted. While shifting the small areas A and B, one or both of the coincidence and the displacement of each image data in the small areas A and B are obtained. If the defective portion 19 on the side surface 10a of the glass substrate 10 is identified based on one or both of them, the defective portion 19 such as a scratch, chip, or crack can be extracted with high precision with respect to the side surface 10a of the glass substrate 10. In addition, it is possible to reliably identify a defective portion at an edge portion where noise occurs.

The present invention is not limited to the above-described first and second embodiments, and can be variously modified in the implementation stage without departing from the gist of the invention.

Further, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some components are deleted from all the components shown in the embodiment, the problem described in the column of the problem to be solved by the invention can be solved, and the problem described in the column of the effect of the invention can be solved. In the case where a certain effect can be obtained, a configuration from which this configuration requirement is deleted can be extracted as an invention.

For example, the first and second embodiments may be modified as follows.

In the first embodiment, the case where the pattern for manufacturing the mask is applied to the appearance inspection of the side surface 10a of the plain glass substrate 10 has been described. Alternatively, the present invention may be applied to a visual inspection of a test object on which a uniform pattern is formed.

As an object to be inspected having such a uniform pattern formed thereon, for example, one having a plurality of lines 22 formed at predetermined intervals in the X direction as shown in FIG. A plurality of marks 22 are formed in the XY (vertical and horizontal) directions for each of the pitches p1 and p2, and a plurality of lines 24 intersect each other as shown in FIG.

For the inspected object shown in FIG. 9, defect original image data obtained by imaging the inspected object is shifted in units of pixels in the Y direction to obtain defect shifted image data. By searching for the luminance in the defect extraction image data, which is the difference between the data and the defect shifted image data, a defect portion such as a scratch, a chip, or a crack existing on the inspection object is extracted.

For the inspection object shown in FIG. 10, defect-shifted image data is obtained by shifting the defect original image data obtained by imaging this inspection object by p1 pitch in the X direction or p2 pitch in the Y direction. By searching for the luminance in the defect extracted image data which is the difference between the defect original image data and the defect shifted image data,
A defective portion such as a chip or a crack is extracted.

For the inspection object shown in FIG. 11, defect-shifted image data is obtained by shifting the defect original image data obtained by imaging the inspection object by n1 pitches in the X direction or n2 pitches in the Y direction. By searching for the luminance in the defect extracted image data which is the difference between the defect original image data and the defect shifted image data,
A defective portion such as a chip or a crack is extracted.

Next, another feature of the present invention will be described.

According to the present invention, an image pickup means (11) for picking up an image of an object to be inspected and image data obtained by the image pickup means are shifted by a pixel unit in a direction determined by a pattern formed on the object to be inspected. An image shifting unit (15); a difference image calculating unit (16) for obtaining image data of a difference between the image data before shifting by the image shifting unit and the image data after shifting; A defect extracting means (17) for searching for a luminance in the obtained difference image data to extract a defect on the inspection object.

According to the present invention, an image pickup means (11) for picking up an image of an object to be inspected, and image data obtained by image pickup by the image pickup means, each image in each of these small areas being shifted by shifting two adjacent small areas. A small area calculating means (20) for obtaining one or both of the degree of coincidence of the data and the positional deviation of each of the small areas; and either one of the degree of coincidence or the positional deviation obtained by the small area calculating means or A defect identification unit (21) for identifying a defect on the inspection object based on both of them.

According to the present invention, an image pickup means (11) for picking up an image of an object to be inspected and image data obtained by the image pickup by the image pickup means are shifted in pixel units in a direction determined by a pattern formed on the object to be inspected. An image shifting unit (15); a difference image calculating unit (16) for obtaining image data of a difference between the image data before shifting by the image shifting unit and the image data after shifting; A defect extracting means (17) for searching for the luminance in the obtained difference image data to extract a defect on the object to be inspected, and two adjacent small pixels for image data obtained by imaging by the imaging means. A small area calculating means (20) for determining one or both of the degree of coincidence of each image data in these small areas and the positional shift of each small area while shifting the area; A defect identification unit (21) for identifying a defect on the object to be inspected based on one or both of the determined degree of coincidence and the position shift. is there.

[0065]

As described above in detail, according to the present invention, a defect inspection method and a defect inspection method capable of extracting defects with high accuracy without aligning image data and storing a plurality of non-defective image data. Equipment can be provided.

Further, according to the present invention, there is provided a mask manufacturing method capable of manufacturing a mask by performing accurate defect extraction without aligning image data and storing a plurality of non-defective image data. it can.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a first embodiment of a defect inspection apparatus according to the present invention.

FIG. 2 is a schematic diagram of defect original image data in the defect inspection apparatus according to the first embodiment of the present invention.

FIG. 3 is a schematic diagram of defect-shifted image data in the defect inspection apparatus according to the first embodiment of the present invention.

FIG. 4 is a schematic diagram of defect extraction image data in the defect inspection device according to the first embodiment of the present invention.

FIG. 5 is a configuration diagram showing a second embodiment of the defect inspection apparatus according to the present invention.

FIG. 6 is a schematic diagram for explaining the operation of the small area calculation means in the second embodiment of the defect inspection apparatus according to the present invention.

FIG. 7 is a schematic diagram of defect original image data in which a defect portion exists at an edge portion of a glass substrate.

FIG. 8 is a schematic diagram of defect extraction image data in which a defect portion exists at an edge portion of a glass substrate.

FIG. 9 is a view for explaining a defect inspection operation for an object to be inspected having another pattern to be inspected by the defect inspection apparatus according to the present invention.

FIG. 10 is a diagram for explaining a defect inspection operation for an inspected object having another pattern to be inspected by the defect inspection apparatus according to the present invention.

FIG. 11 is a view for explaining a defect inspection operation for an inspected object having another pattern to be inspected by the defect inspection apparatus according to the present invention.

FIG. 12 is a schematic diagram of image data obtained by imaging an IC chip with a camera for explaining a conventional appearance inspection method.

FIG. 13 is a schematic diagram of non-defective image data in a conventional appearance inspection method.

FIG. 14 is a schematic diagram of difference image data in a conventional appearance inspection method.

[Explanation of symbols]

 10: Glass substrate 10a: Side surface of glass substrate 10b: Chamfer of glass substrate 11: Camera 12: Illumination device 13: Personal computer 14: Image input board 15: Image shifting means 16: Difference image calculation means 17: Defect extraction means 18: Monitor 19: Defect part 20: Small area calculation means 21: Defect identification means

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI theme coat ゛ (Reference) H01L 21/027 H01L 21/30 502V F-term (Reference) 2F065 AA49 AA56 BB02 CC18 DD03 DD04 FF42 GG01 HH02 JJ03 JJ26 QQ04 QQ13 QQ21 QQ24 QQ37 QQ38 RR02 2G051 AA56 AB07 CA03 CA04 EA11 EA14 EC06 ED07 ED12 ED30 2H095 BD04 BD17 BD19 BD23 BD24 5B057 AA03 DA03 DB02 DC22 DC32 DC34

Claims (5)

[Claims] An image data of a difference between image data obtained by imaging an object to be inspected and image data obtained by shifting the image data by a predetermined area in a direction determined by a pattern formed on the object to be inspected. A defect on the object to be inspected by searching for luminance in the difference image data. 2. An image data obtained by imaging an object to be inspected, wherein two adjacent small areas are shifted and any one of a matching degree of each image data in the small areas or a positional shift of each of the small areas is determined. A defect inspection method characterized by determining one or both of them, and identifying a defect on the inspection object based on one or both of the degree of coincidence and the displacement. 3. An image data of a difference between image data obtained by imaging an object to be inspected and image data obtained by shifting the image data by a predetermined area in a direction determined by a pattern formed on the object to be inspected. And extract the defect on the inspection object by searching for the luminance in the difference image data,
Thereafter, while shifting two adjacent small areas with respect to the image data, one or both of the coincidence of each image data in these small areas and the positional deviation of each of the small areas are obtained. A defect inspection method comprising: identifying a defect on the inspection object based on one or both of the positional shifts. 4. An object to be inspected is imaged by an image pickup means, and image processing is performed on the image data obtained by the image pickup means using the defect inspection method according to claim 1, 2 or 3, and A defect inspection device for inspecting a defect on an inspection object. 5. A mask manufacturing method for manufacturing a mask by forming a circuit pattern of a semiconductor device on a glass substrate, wherein image data obtained by imaging the glass substrate and the image data are stored on the glass substrate. Extracting image data of a difference from image data shifted by a predetermined area unit in a direction determined by the formed circuit pattern, and extracting a defect on the glass substrate by searching for luminance in the difference image data. A method for manufacturing a mask, comprising: