JP2003337013A - Method and apparatus for inspection of defect - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To inspect a defect in such a way that a size decision region having a defect detection sensitivity according to a size of a defect part is set automatically according to a density of a pattern. <P>SOLUTION: By a sensitivity creation part 9, reference image data is converted into a lightness according to a width of the pattern, and sensitivity image data having each size decision region of each defect detection sensitivity according to the lightness is created. The defect part is extracted by a defect size creation part 10 on the basis of a difference between to-be-inspected image data found by a difference computing part 8 and the reference image data, the size of the defect part is converted into a lightness so as to create defect- size image data, the defect-size image data is compared with the sensitivity image data by a decision part 11, and a defect with respect to the defect part is decided. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a defect inspection by comparing inspected image data obtained by imaging an inspected object such as a chip component such as a semiconductor chip having a pattern formed thereon with reference image data. The present invention relates to a defect inspection method and an apparatus therefor.

[0002]

2. Description of the Related Art A fine circuit pattern is formed on a chip component such as a semiconductor chip. This pattern is
The pattern formation density is different in each area on the chip.

In such a pattern inspection method, an image (hereinafter referred to as reference image data) acquired by picking up a pattern of a non-defective product is stored in advance, and a pattern to be inspected is picked up and the image (hereinafter, an image to be inspected) is stored. (Referred to as data), and the difference between the inspected image data and the reference image data is obtained. Next, a portion having a large difference value is extracted from the difference image data, and the size of the area where the extracted portion is continuous is obtained. Next, the size of the extracted portion is compared with a preset threshold for defect size determination, and the defect of the extracted portion is determined.

In this pattern inspection method, a threshold for defect size determination is designated for each area of the pattern. Then, the defect determination is generally performed by each threshold value specified for each area.

[0005]

Since the patterns formed on the chip component have different densities in each area, the shape of each area is complicated and the number of areas formed is large. Therefore, it is necessary to perform a work for designating a threshold for defect size determination for each of these areas, which takes a huge amount of time, and the work efficiency for designating the threshold is lowered.

Therefore, the present invention provides a defect inspection method and apparatus capable of performing a defect inspection by automatically setting a size determination region having a defect detection sensitivity corresponding to the size of a defective portion according to the pattern density. With the goal.

[0007]

SUMMARY OF THE INVENTION The present invention is a defect inspection method for inspecting a defect on an object to be inspected on the basis of image data to be inspected and reference image data acquired by imaging an object to be inspected on which a pattern is drawn. In the first step, the reference image data is converted into the lightness according to the width of the pattern, and the sensitivity image data having each size determination region of each defect detection sensitivity according to the lightness is created; By comparing the defect size image data and the sensitivity image data with the second step of extracting the defect part based on the reference image data and converting the size of the defect part into the brightness to create the defect size image data And a third step of making a defect determination on a defective portion.

According to the present invention, in a defect inspection apparatus for inspecting a defect of an object to be inspected on the basis of image data to be inspected and reference image data obtained by imaging an object to be inspected with a pattern, the reference image data is Sensitivity creating means for creating a sensitivity image data having each size determination region of each defect detection sensitivity corresponding to the brightness, which is converted into the brightness according to the width of the pattern, and based on the difference between the inspected image data and the reference image data. Defect size is extracted, and the size of the defective part is converted into lightness to create defect size image data, and the defect size image data and the sensitivity image data are compared to determine the defect for the defective part. A defect inspecting apparatus, comprising: a determining unit for performing the inspection.

The sensitivity creating means in the defect inspection apparatus according to the present invention creates horizontal scan image data and vertical scan image data by converting the pattern widths of the reference image data in the horizontal direction and the vertical direction, respectively.

The sensitivity creating means in the defect inspection apparatus of the present invention scans the reference image data linearly in the horizontal direction and the vertical direction, respectively, and measures the number of consecutive pixels of the white level or the black level to measure the pattern width. Is calculated and converted into the brightness according to this pattern width.

The sensitivity creating means in the defect inspection apparatus of the present invention creates sensitivity image data by adjusting the brightness according to the size required to detect a defective portion.

The sensitivity creating means in the defect inspection apparatus of the present invention extracts a small brightness of the brightness of the horizontal scan image data and the vertical scan image data, and creates sensitivity image data having a defect detection sensitivity using the brightness. To do.

The sensitivity creating means in the defect inspection apparatus of the present invention uses the lightness of the pattern corner portion, whichever is smaller among the lightness of the horizontal scan image data and the lightness of the vertical scan image data.

The sensitivity creating means in the defect inspection apparatus of the present invention extracts a portion where the brightness of the horizontal scan image data and the brightness of the vertical scan image data are both large as a pattern corner portion, and corrects the brightness of the pattern corner portion. .

The defect size creating means in the defect inspection apparatus of the present invention divides the defective portion in the defect size image data, obtains each size for each of the divided defective portions, and converts these sizes into lightness.

The defect size creating means in the defect inspection apparatus of the present invention labels each divided defect portion and reproduces each divided defect portion using these labeling results.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a defect inspection apparatus. A circuit pattern 2 is formed on the chip component 1. The chip component 1 is a non-defective product placed on the table 3 in order to obtain reference image data. Further, an uninspected chip component 1 for pattern defect inspection is placed on the table 3.

Above the chip component 1, a CCD camera 4 is provided.
Is provided. The CCD camera 4 takes an image of the circuit pattern 2 of the chip part 1 and outputs the image signal. The CCD camera 4 may be replaced with a television (TV) camera or a line sensor camera. When the line sensor camera is used, the line sensor camera is scanned above the circuit pattern 2 or the table 3 is moved to acquire the entire image of the circuit pattern 2.

The image processing unit 5 receives the image signal output from the CCD camera 4, stores the image data of the non-defective chip component 1 as reference image data in the reference image memory 6, and also checks the uninspected chip component 1. Of the image data is sequentially input, and the image data is stored in the image memory 7 as the image data to be inspected.

The difference calculating section 8 reads out the reference image data stored in the reference image memory 6 and the inspected image data stored in the inspected image memory 7, and outputs the reference image data and the inspected image data. It has a function of obtaining a difference, that is, a difference in gray level, and acquiring this difference image data.

The sensitivity creating section 9 reads out the reference image data stored in the reference image memory 6, converts the reference image data into the brightness corresponding to the width of the pattern, and detects the defect detection sensitivity corresponding to the brightness. It has a function of creating sensitivity image data having each size determination area.

Specifically, the sensitivity creating section 9 binarizes the reference image data with a predetermined threshold value, and scans the binarized reference image data linearly in the horizontal and vertical directions. Then, the number of continuous pixels of the white level or the black level is measured to obtain the pattern width, and the pattern width is converted into the lightness to create the horizontal scan image data and the vertical scan image data.

For example, FIG. 2 is a schematic diagram of binarized reference image data, and there are a white level portion w and a black level portion b by binarization. The white level portion w is a metal wiring pattern portion of the circuit pattern 2 in the chip component 1, and the black level portion b is an insulator portion.

The sensitivity creating section 9 scans the reference image data in the horizontal direction x, continuously detects the white level part w at the scan position x ₁ , for example, then detects the black level part b, and again detects the white level. Level part w, black level part b,
The white level portion w and the black level portion b are detected.

Next, the sensitivity creating section 9 determines each white level portion w.
And each black level portion b are detected, and the number of consecutive white level or black level pixels is measured to obtain the pattern width.

Next, the sensitivity creating section 9 creates a horizontal scan image data as shown in FIG. 3 by converting it into a lightness corresponding to the pattern width.

For example, the white level portion w detected first at the scan position x ₁ is 5 pixels in succession, then the black level portion b is 2 pixels, the white level portion w is 2 pixels, and then the black level portion b. Is 2 pixels, then the white level part w is 2 pixels,
Next, when the black level portion b is continuously detected by two pixels, the sensitivity creating unit 9 determines, for example, the brightness “5” and the next black level with respect to the first detected white level portion w as shown in FIG. The brightness “2” for the part b, the brightness “2” for the next white level part w, the brightness “2” for the next black level part b, and the brightness “2” for the next white level part w. , And the brightness “2” is set for the next black level portion b.

Although the lightness here is the number of pixels detected continuously for the sake of clarity, it is not limited to this and may be converted into any lightness value corresponding to the pattern width. .

Similarly to the above, the sensitivity creating section 9 scans the reference image data in the vertical direction y to create vertical scan image data as shown in FIG. In this vertical scan image data, for example, the white level portion w detected first at the scan position y ₁ is continuously detected by 8 pixels, and then the black level portion b is continuously detected by 6 pixels.
Sets the brightness “8” for the first detected white level portion w and the brightness “6” for the next black level portion b as shown in FIG.

Further, the sensitivity creating section 9 adjusts each lightness in the horizontal scan image data shown in FIG. 3 and the vertical scan image data shown in FIG. 4 to each lightness corresponding to a size required to detect a defective portion. Then, the sensitivity image data as shown in FIG. 5 is created. For example, when it is desired to detect a defective portion with a size equal to or larger than ½ of the pattern width, the defect detection sensitivity is set to ½ of “5”. The value is set to 2.5 ", but here, the decimal point is cut off, and the defect detection sensitivity is set to" 2 "or higher.

In this way, lateral sensitivity image data having defect detection sensitivities of "2" or higher, "3" or higher, and "1" or higher as shown in FIG. 5 are created.

The defect detection sensitivity has been described with respect to the case where it is desired to detect a defective portion with a size of ½ or more of the pattern width. However, the defect detection sensitivity is not limited to this, and the brightness may be changed according to the size required for detecting the defective portion. May be adjusted.

Further, the sensitivity creating section 9 has a portion where both the brightness of the horizontal scan image data shown in FIG. 3 and the brightness of the vertical scan image data shown in FIG. 4 are large, for example, both brightness are larger than a preset threshold value. It has a function of extracting a portion of the pattern corner portion as a pattern corner portion and correcting the brightness of the pattern corner portion.

FIG. 6 shows typical pattern corner portions C _{1 to} C ₅ that have been extracted. These pattern corner portions C ₁
Lightness That defect detection sensitivity -C ₅ is corrected. This defect detection sensitivity correction is performed by, for example, the pattern corner portion C ₁
The lightnesses corresponding to the widths in the horizontal direction and the vertical direction in the pattern corner portion C ₁ are obtained, and these lightnesses are corrected as the defect detection sensitivities in the horizontal direction and the vertical direction, respectively.

The defect detection sensitivity corresponds to the length of each of the horizontal and vertical widths in the pattern corner portion C ₁ when it is desired to detect a defective portion with a size of ½ or more of the pattern width. One half of each lightness may be corrected as each defect detection sensitivity in the horizontal and vertical directions.

If the defect detection sensitivity in the pattern corner portion C ₁ is not corrected in this way, for example, in the pattern corner portion Cs as shown in FIG. 7, the pattern width is set to be long in the horizontal and vertical directions. The defect portion G is detected by the defect detection sensitivity of high brightness.

In this case, for example, if the defect detection sensitivity is size "2" or more and the size of the defective portion G is "1",
The defective portion G is not detected even if it is a defect having a size to be detected.

[0039] Therefore, by correcting the defect detection sensitivity in the pattern corner portion C _1, it can be reliably detected defective portions G existing in the pattern corner portion C _1.

The defect size creating section 10 receives the difference image data acquired by the difference calculating section 8, extracts a defect portion based on the difference image data, converts the size of the defect portion into the lightness, and detects the defect size. It has the function of creating image data.

In this case, the defect size creating section 10 includes a defect portion (hereinafter, referred to as a white defect) existing on a white level portion when the image data to be inspected is binarized by a preset threshold value, and a black portion. It is divided into a defect portion existing on the level portion (hereinafter, referred to as black top defect).

Next, the defect size creating unit 10 performs labeling on the white top defect and the black top defect, respectively, measures the sizes of the white top defect and the black top defect, and determines the white top defect. The respective widths (x direction) and heights (y direction) of the black top defects are obtained.

Here, the defective portion G may exist on the boundary between the white level portion and the black level portion (exist over the pattern edge) as shown in FIG.

When measuring the size of such a defect portion G, the defect size creating section 10 divides the defect portion G according to the pattern shape, and for example, in the defect portion G shown in FIG. It has a function of dividing, obtaining each size for each of the divided defect portions G ₁ and G ₂ , and converting these sizes into lightness.

Further, the defect size creating section 10 has a function of labeling each of the divided defective portions G ₁ and G ₂ .

The judging section 11 receives the defect size image data created by the defect size creating section 10 and the sensitivity image data created by the sensitivity creating section 9, and compares the defect size image data with the sensitivity image data. ,
For example, the defect size image data has a function of performing defect determination on the defect portion G by extracting a portion brighter than the sensitivity image data.

Further, the judging section 11 includes the defect size creating section 10
Reading the results of the labeling for each defective portion G _1, G ₂ divided by, has a split function of reproducing the defect portion G _1, G ₂ was using these labeling results.

Next, the operation of the apparatus configured as described above will be described with reference to the defect inspection flowchart shown in FIG.

First, a good chip component 1 is placed on the table 3. CCD camera 4 is a good chip component 1
The circuit pattern 2 is imaged and the image signal is output.
This image signal is sent to the image processing unit 5.

The image processing unit 5 inputs the image signal output from the CCD camera 4 and stores the image data of the non-defective chip component 1 in the reference image memory 6 as reference image data.

If the reference image data of the chip component 1 to be inspected is stored in the reference image memory 6 in advance, it is not necessary to obtain the reference image data of the good chip component 1.

Next, the uninspected chip components 1 are sequentially placed on the table 3. The CCD camera 4 images the circuit pattern 2 of the uninspected chip component 1 and outputs the image signal. This image signal is sent to the image processing unit 5.

The image processing unit 5 inputs the image signal output from the CCD camera 4 and receives each uninspected chip component 1
Each image data is sequentially input, and these image data are stored in the inspected image memory 7 as inspected image data.

Next, in step # 1, the sensitivity creating section 9 reads out the reference image data stored in the reference image memory 6 and converts the reference image data into the brightness corresponding to the width of the pattern.

FIG. 10 shows a specific algorithm for defect inspection. First, in step # 1-1, the sensitivity creating unit 9 binarizes the reference image data with a predetermined threshold value.

Next, the sensitivity creating section 9 performs step # 1-2.
In, the binary reference image data is scanned linearly in the horizontal direction, the number of consecutive pixels at the white level or the black level is measured to obtain the pattern width, and the brightness is converted according to the pattern width. Then, horizontal scan image data as shown in FIG. 3 is created.

At the same time, the sensitivity creating section 9 performs step #
In 1-3, the binarized reference image data is scanned linearly in the vertical direction, the number of consecutive pixels of the white level or the black level is measured to obtain the pattern width, and the pattern width is determined according to this pattern width. The brightness is converted into vertical scan image data as shown in FIG.

Next, in step # 2, the sensitivity creating section 9 adjusts the brightness in the horizontal scan image data shown in FIG. 3 to the brightness corresponding to the size required to detect the defective portion, and the sensitivity in the horizontal direction. Create image data.

For example, when it is desired to detect a defective portion with a size equal to or larger than ½ of the pattern width, the sensitivity creating section 9
Sets the defect detection sensitivity in the horizontal scan image data shown in FIG. 3 to "2.5" which is a half of "5". Here, the decimal point is rounded down and the defect detection sensitivity is set to "2" or more. To As a result, the sensitivity creating unit 9 is
The sensitivity image data in the lateral direction having the defect detection sensitivities of “2” or higher, “3” or higher, and “1” or higher is created as shown in FIG.

At the same time, the sensitivity creating section 9
The brightness in the vertical scan image data shown in FIG. 4 is adjusted to the brightness according to the size required to detect the defective portion to create the sensitivity image data in the vertical direction.

In step # 2-1, the sensitivity creating section 9 compares the brightness of the horizontal scan image data shown in FIG.
A part where both the brightness of the vertical scan image data is large, for example, a part where both brightnesses are larger than a preset threshold value is extracted as pattern corner parts C _{1 to} C ₅ as shown in FIG. 6, for example.

Next, the sensitivity creating section 9 performs step # 2-2.
At each corner C of the pattern ₁~ C₅The brightness of
That is, the defect detection sensitivity is corrected. That is, the sensitivity creation unit 9
Is, for example, the pattern corner portion C₁Lateral width at
The brightness corresponding to the
Correct as a degree.

At the same time, the sensitivity creating section 9 proceeds to step # 2.
In -3, the brightness of the pattern corner portions C _{1 to} C ₅ , that is, the defect detection sensitivity is corrected. That is, the sensitivity creating unit 9 obtains the brightness corresponding to the vertical width of the pattern corner portion C ₁ , for example, and corrects this brightness as the vertical defect detection sensitivity.

The defect detection sensitivities of the pattern corner portions C _{1 to} C ₅ are, for example, in the horizontal direction and the vertical direction in the pattern corner portion C ₁ when it is desired to detect a defect portion having a size of ½ or more of the pattern width. One half of each lightness corresponding to the length of each width may be corrected as each defect detection sensitivity in the horizontal and vertical directions.

The binarized reference image data, the horizontal scan image data and the vertical scan image data are
They are displayed on a display not shown as a width preview and a height preview, respectively.

On the other hand, in step # 3, the difference calculation unit 8 reads out the reference image data stored in the reference image memory 6 and the inspection image data stored in the inspection image memory 7, respectively, and refers to them. The difference between the image data and the inspected image data, that is, the difference in gray level is obtained, and this difference image data is acquired.

Next, in step # 4, the defect size creating section 10 receives the difference image data acquired by the difference calculating section 8, extracts the defective portion based on this difference image data, and then executes the next step # 5. At, the size of the defect portion is converted into the lightness to create defect size image data.

Here, the defect size creating section 10 binarizes the image data to be inspected by a preset threshold value, and exists on the white level portion from the image data to be inspected binarized in step # 4-1. The white defect on the black level portion is extracted, and at the same time, the black defect on the black level portion is extracted in step # 4-2.

Next, in step # 5-1, the defect size creating section 10 performs labeling on the over-white defect, measures the size of the over-white defect in the next step # 5-2, and then performs step ##. In step 5-3, the horizontal direction of the white defect is obtained, and in step # 5-4, the vertical direction is obtained.

Then, the defect size creating unit 10 converts the horizontal direction and the vertical direction of the white defect to each lightness.

At the same time, the defect size creating section 10 labels the black top defect in step # 5-5, measures the size of the black top defect in the next step # 5-6, and then executes step # 5. In -7, the horizontal direction of the black top defect is obtained, and in step # 5-8, the vertical direction is obtained.

Then, the defect size creating section 10 converts the horizontal direction and the vertical direction of the black top defect into respective lightnesses.

Here, the defective portion G may exist on the boundary between the white level portion and the black level portion as shown in FIG.

When measuring the size of such a defective portion G, the defect size creating section 10 divides the defective portion G according to the pattern shape, and for example, in the defective portion G shown in FIG. It is divided into _two defective portions G ₁ and G ₂ .

Then, the defect size creating section 10 obtains each size for each of these divided defect parts G ₁ and G ₂ .
Convert these sizes to lightness.

Accordingly, the defect size image data in the horizontal and vertical directions of the white defect and the defect size image data in the horizontal and vertical directions of the black defect are created.

The defect size creation unit 10 matches the width of the white top defect and the width of the black top defect in step # 5-9, sends the combined defect size image data to the determination unit 11, and detects the white top defect. Step # 5 between the vertical direction and the vertical direction of the black top defect
In -10, the defect size image data thus matched is sent to the judgment unit 11.

Next, in step # 6, the judging section 11 receives the defect size image data created by the defect size creating section 10 and the sensitivity image data created by the sensitivity creating section 9, and receives these defect size images. By comparing the data with the sensitivity image data and extracting a portion where the defect size image data is brighter than the sensitivity image data, the defect determination for the defect portion G is performed.

In this case, the judgment unit 11 compares the defect size image data in which the width of the white defect and the width of the black defect are combined with the lateral sensitivity image data in step # 6-1 to determine the defect size. A portion where the image data is brighter than the sensitivity image data is determined as a defective portion G.

At the same time, the determination unit 11 compares the defect size image data in which the vertical direction of the white defect and the vertical direction of the black defect are combined with the vertical sensitivity image data in step # 6-2, and the defect A portion in which the size image data is brighter than the sensitivity image data is determined as a defective portion G.

In the case of such defect determination, for example, sensitivity image data for detecting a defective portion with a size equal to or larger than ½ of the pattern width as shown in FIG. It is possible to reliably detect the defective portion G having a size equal to or larger than one-half.

By changing the size of the sensitivity image data, it is possible to reliably detect the defective portion G having a size equal to or smaller than half the pattern width.

Further, as shown in FIG. 7, even if a defective portion G exists in the pattern corner portions C _{1 to} C ₅ , this defective portion G can be reliably detected with a size of, for example, ½ or more of the pattern width. it can.

[0084] Further, even if the defect portion G is present on the boundary between the white level portions and the black level portion as shown in FIG. 8, each defective portion G ₁ in accordance with this defect G pattern shape,
G ₂ 2 divided, since converted into brightness corresponding to each size for each of these defective portions G _1, G _2, of one or more sizes of half of these defective portions G _1, G ₂ each, for example, the pattern width The defective portion G can be reliably detected.

Therefore, as shown in FIG. 11, if the size of the defective portion Ga existing on the pattern of the black level portion is, for example, ½ or more of the pattern width, this defective portion Ga
Is determined as a defect (NG).

On the other hand, if the size of the defective portion Gb existing on the boundary between the white level portion and the black level portion, that is, the portion G ₁ existing on the black level portion is, for example, ½ or less of the pattern width. , The defective portion G ₁ is not judged as a defect (OK). Similarly, if the size of the portion G ₂ existing on the white level portion is, for example, ½ or less of the pattern width, this defective portion G ₂ is not determined as a defect (O
K).

However, if the size of the portion G ₁ existing on the black level portion is, for example, ½ or more of the pattern width, this defective portion G ₁ is judged as a defect (N
G). Similarly, if the size of the portion G ₂ existing on the white level portion is, for example, ½ or more of the pattern width, this defective portion G ₂ is determined as a defect (NG).

Therefore, even if the sizes of the defective portion Ga and the defective portion Gb are the same, the defect determination result differs depending on the presence of the defective portions Ga and Gb.

Next, the judgment section 11 moves to step # 7-1 and combines the defect judgment result in step # 6-1 and the defect judgment result in step # 6-2.

Next, in step # 7-2, the judging section 11 sorts out the white top defects labeled in step # 5-1 and the black top defects labeled in step # 5-5. .

At this time, the judging section 11 reads the labeling result of each of the defect portions G ₁ and G ₂ divided by the defect size creating section 10, and uses each of the labeling results to divide each of the defect portions G 1. ₁ and G ₂ are reproduced.
As a result, the defect portion G existing on the boundary between the white level portion and the black level portion as shown in FIG. 8 is reproduced, for example, only the defect portion G ₂ on the white level portion as shown in FIG. Instead, the defective portions G ₁ and G ₂ that have been divided can be reproduced as a defective portion Gb.

As described above, in the above-mentioned one embodiment,
Convert the reference image data to the brightness according to the width of the pattern,
Sensitivity image data having each size determination area of each defect detection sensitivity according to this brightness is created, while the defect portion is extracted based on the difference between the inspected image data and the reference image data, and the size of this defect portion is determined. Since the defect size image data is created by converting it to the brightness and the defect size image data is compared with the sensitivity image data to judge the defect for the defective portion, the sensitivity image data according to the pattern width can be automatically created. , The defective portion G on the pattern can be reliably determined.

In this case, even if the circuit pattern 2 on the chip component 1 has a different density for each area, the shape of each area is complicated, or the number of areas formed is large, these areas are not formed. It is possible to automatically determine a defective portion without performing a work of designating a threshold for defect size determination for each.

Therefore, the defect inspection of the circuit pattern 2 formed on the chip component 1 can improve the work efficiency without requiring an enormous amount of time for the work.

Further, if the size of the sensitivity image data is changed to an arbitrary size, it is possible to detect a defective portion G having a size equal to or less than half the pattern width or various sizes.

[0096] Also, even if there is a defect portion G in the pattern corner portion C ₁ -C ₅ as shown in FIG. 7, is corrected brightness i.e. defect detection sensitivity of each pattern corner portion C ₁ -C _5, The defective portion G existing in these pattern corner portions C _{1 to} C ₅ can be reliably detected, for example, with a size of ½ or more of the pattern width.

[0097] Further, even if the defect portion G is present on the boundary between the white level portions and the black level portion as shown in FIG. 8, each defective portion G ₁ in accordance with this defect G pattern shape,
G ₂ 2 divided, since converted into brightness corresponding to each size for each of these defective portions G _1, G _2, of one or more sizes of half of these defective portions G _1, G ₂ each, for example, the pattern width The defective portion G can be reliably detected.

[0098] In addition, since the labeling for each defective portion G _1, G _2, which are the defective portion and dividing, organizing each defective portion can be easily, and the defect was divided G _1, G
₂ can be reproduced faithfully.

The present invention is not limited to the above-described one embodiment, and can be variously modified at the stage of implementation without departing from the spirit 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 constituent elements are deleted from all the constituent elements shown in the embodiment, the problem described in the section of the problem to be solved by the invention can be solved, and it is described in the section of the effect of the invention. When the effect of being obtained is obtained, a configuration in which this constituent element is deleted can be extracted as an invention.

For example, in the above-described embodiment, the reference image data and the inspected image data are binarized, but the present invention is not limited to this, and the multi-valued image data such as ternary or quaternary image data is converted. Good.

Although the defective portion is extracted based on the difference between the reference image data and the inspected image data, the invention is not limited to this, and the absolute value of the difference or the minimum min-max method may be used.

The defect detection sensitivity of the sensitivity image data is not limited to a size equal to or more than half the pattern width. Also, for example, 1
When it is desired to have 6 levels of sensitivity, the pattern width may be converted to 16 gradations. Furthermore, 0 to 256 is set to 0 to 128 in order to make the sensitivity of 1/2 width, and the way of assigning values when divided into 16 sections is, for example, 0.8, 16, 24, 3
2,40,48,56,64,72,80,88,9
6, 104, 112, 120, 128 may be used.

Further, in order to obtain the sensitivity image data, scanning is performed in the horizontal direction and the vertical direction, but the scanning direction may be an arbitrary direction, for example, an oblique direction.

As for the defect detection sensitivity in the sensitivity image data, the respective lightnesses in the horizontal scan image data shown in FIG. 3 and the vertical scan image data shown in FIG. 4 are compared, and the smaller lightness is set as the defect detection sensitivity. One sheet of sensitivity image data may be created.

[0106]

As described above in detail, according to the present invention, a defect can be inspected by automatically setting a size determination area having a defect detection sensitivity corresponding to the size of a defective portion according to the pattern density. An inspection method and its apparatus can be provided.

[Brief description of drawings]

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

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

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

FIG. 4 is a schematic diagram of vertical scan image data in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 5 is a schematic diagram of sensitivity image data in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 6 is a diagram for explaining a correcting action of the defect detection sensitivity in the pattern corner portion in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 7 is a diagram for explaining overlooking of a defective portion at a pattern corner portion in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 8 is a diagram for explaining a detection function of a defect portion existing on a boundary between white and black level portions in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 9 is a flowchart of defect inspection in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 10 is a diagram showing a specific algorithm for defect inspection in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 11 is a diagram for explaining the operation of defect determination when a defect portion exists on the boundary between a white level portion and a black level portion in the embodiment of the defect inspection apparatus according to the present invention.

FIG. 12 is a diagram showing a defective portion extracted into an incomplete shape.

[Explanation of symbols]

1: Chip parts 2: Circuit pattern 3: Table 4: CCD camera 5: Image processing unit 6: Reference image memory 7: Inspected image memory 8: Difference calculator 9: Sensitivity creation section 10: Defect size creation section 11: Judgment unit

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G01B 11/24 FF term (reference) 2F065 AA56 BB02 CC25 DD06 FF04 JJ02 JJ03 JJ19 JJ25 JJ26 QQ04 QQ24 QQ25 QQ31 RR09 2G051 AA51 AB02 AC21 CA03 CA04 EA08 EA16 EA24 4M106 AA02 CA39 DB04 DB20 DJ18 5B057 AA03 BA02 CA08 CA12 CA16 CE11 DA03 DA08 DC04 DC32 5L096 AA06 BA03 FA54 FA59 GA08 HA07

Claims (10)

[Claims] 1. A defect inspection method for inspecting a defect of the inspected object based on the inspected image data obtained by imaging the inspected object on which a pattern is drawn and the reference image data, wherein the reference image data is A first step of converting to brightness according to the width of the pattern and creating sensitivity image data having each size determination region of each defect detection sensitivity according to this brightness; the inspected image data and the reference image data A second step of extracting a defective portion based on the above, converting the size of the defective portion into lightness to create defect size image data, and comparing the defect size image data and the sensitivity image data with each other. A third step of performing a defect determination on a defective portion, and a defect inspection method comprising: 2. A defect inspection apparatus for inspecting a defect of the inspected object based on the inspected image data acquired by imaging the inspected object on which a pattern is drawn and the reference image data, Sensitivity creating means for creating a sensitivity image data having each size determination region of each defect detection sensitivity according to the brightness, which is converted into a brightness according to the width of the pattern, and the inspected image data and the reference image data. The defect size is extracted based on the difference, the size of the defect part is converted into the lightness to create defect size image data, and the defect size image data is compared with the sensitivity image data and the defect size image data. A defect inspecting apparatus comprising: a determining unit that determines a defect with respect to a defective portion. 3. The sensitivity creating means creates horizontal scan image data and vertical scan image data by converting the pattern widths of the reference image data in the horizontal direction and the vertical direction into the brightness, respectively. The defect inspection apparatus according to claim 2. 4. The sensitivity creating means scans the reference image data linearly in the horizontal direction and the vertical direction, respectively, and measures the number of consecutive white level or black level pixels to obtain a pattern width. The defect inspection apparatus according to claim 2 or 3, wherein the brightness is converted into a brightness according to a pattern width. 5. The defect inspection apparatus according to claim 2, wherein the sensitivity creating unit creates the sensitivity image data by adjusting the brightness according to a size required to detect the defective portion. . 6. The sensitivity creating means extracts the lightness that is smaller among the lightnesses in the horizontal scan image data and the vertical scan image data, and uses the lightness to obtain the sensitivity image data having the defect detection sensitivity. The defect inspection apparatus according to claim 4, wherein: 7. The sensitivity creating means uses the lightness of the pattern corner portion, whichever is smaller of the lightness of the horizontal scan image data and the lightness of the vertical scan image data. The defect inspection apparatus according to claim 7. 8. The sensitivity creating means extracts, as a pattern corner portion, a portion where both the lightness of the horizontal scan image data and the lightness of the vertical scan image data are large and corrects the lightness of the pattern corner portion. The defect inspection apparatus according to claim 3, wherein: 9. The defect size creating means divides the defect portion in the defect size image data, obtains each size for each of the divided defect portions, and converts these sizes into lightness. The defect inspection apparatus according to claim 2. 10. The defect size generating means labels each of the divided defect portions, and reproduces each of the divided defect portions using the labeling result. Inspection device.