JP3781467B2 - Pattern inspection method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention is a capacitor, IC, marking patterns such as the chemical, a pattern that is particularly stamped on curved surfaces such as a capacitor, to a pattern inspection査手method for inspecting a defect.
[0002]
[Prior art]
Conventionally, as a technique for inspecting a seal pattern, as shown in FIG. 16, a technique is known in which a difference area between a non-defective image and an inspection target image is obtained and quality is determined based on the size of the area of the difference area.
[0003]
Furthermore, since the stamp pattern has a variation in the thickness of the line depending on the state of the seal, as shown in FIG. 17, by providing an area that does not calculate a difference area called a dead zone in several pixels around the non-defective image, FIG. As described above, the defect inspection is performed while allowing a pattern having a variation in thickness.
[0004]
[Problems to be solved by the invention]
However, when the target is a pattern stamped on a curved surface such as a capacitor, each pattern is deformed or tilted due to distortion of the stamped surface. If the dead zone is increased in order to allow the deformation and inclination, there is a problem that a defect that cannot be detected occurs as shown in FIG.
[0005]
The present invention is much variation in are stamped in on curved surface line thickness, such as capacitor, relative prone pattern variations, to provide a pattern inspection査手methods that can detect only critical pattern defects Objective.
[0006]
[Means for Solving the Problems]
Pattern inspection method according to claim 1 includes the steps of creating a good distance data whose value is the shortest distance to each pixel of the non-defective pattern image from the pixels of the outermost periphery of the good product pattern, most of the inspection object pattern a step of creating a test object distance data whose value is the shortest distance from the outer circumference of the pixel to each pixel of the image of said object pattern, and extracts the difference region in the image of the image with the inspection object pattern of the non-defective pattern Then, each value of the non-defective product distance data and the inspection target distance data in the difference area is compared with a predetermined determination threshold value, and if at least one of the values exceeds the determination threshold value, it is determined as defective. And a step of determining .
[0007]
The pattern inspection method according to claim 3 , wherein a difference area is extracted from an image of a non-defective pattern and an image of a pattern to be inspected, and pixels at the outermost periphery of the difference area are outermost peripheries of the good pattern and the inspection target pattern, respectively. The difference area is classified according to the number of pixels shared with the pixel, and the area of the difference area is determined by the determination threshold according to the classification, or the distance from the outermost peripheral pixel of the difference area is determined by the determination threshold or, characterized in that it changed.
[0008]
3. In the pass / fail determination for each classification according to claim 2, the outline of each difference area obtained by extracting the difference area from the image of the non-defective pattern including symbols and characters and the image of the inspection target pattern is the non-defective pattern / inspection target pattern. For an area that is not shared with either contour line, if the maximum length of each of the areas in the X-axis direction and the Y-axis direction exceeds the determination threshold, the area is determined to be defective. According to this, the “hole crushing” defect shown in FIG. 20 can be detected with high accuracy.
[0009]
3. In the pass / fail determination for each classification according to claim 2, the outline of each difference area obtained by extracting the difference area from the image of the non-defective pattern including symbols and characters and the image of the inspection target pattern is the non-defective pattern / inspection target pattern. For an area that shares only one outline with one of the contour lines, the distance from the contour line that the value of each pixel in the area shares with either one to the shortest distance to that pixel An image is created, and a pass / fail judgment is made if there is a pixel with a distance that exceeds the judgment threshold. Next, the shortest distance from the contour line that does not share the value of each pixel in the area to the pixel A distance image is created, and if there is a pixel whose distance exceeds the determination threshold, it is determined as defective. According to this, the “chip” and “smear” defects shown in FIG. 21 can be detected with high accuracy.
[0010]
3. In the pass / fail determination for each classification according to claim 2, the outline of each difference area obtained by extracting the difference area from the image of the non-defective pattern including symbols and characters and the image of the inspection target pattern is the non-defective pattern / inspection target pattern. For a region shared with a plurality of one of the contour lines, the value of each pixel in the region is twice the shortest distance from the contour line that is not shared with either contour line to that pixel. A distance image is created, and if there is a pixel whose distance exceeds the determination threshold, it is determined as defective. According to this, the “pattern break” and “pattern combination” defects shown in FIG. 22 can be detected with high accuracy.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the pattern quality inspection method of the present invention will be described based on specific embodiments.
[0012]
[First Embodiment]
1 to 5 show a first embodiment.
1 shows a non-defective product pattern, FIG. 2 shows an inspection target pattern, and FIG. 3 shows a difference area obtained from the non-defective product pattern and the inspection target pattern. A and B in FIG. 3 are labels of the difference area.
[0013]
FIG. 4 shows only a region that overlaps the difference region extracted from a non-defective product distance image created from a non-defective product pattern. The number in the figure is the shortest distance from the contour line of the good product pattern to the pixel.
[0014]
FIG. 5 shows only the region overlapping the difference region extracted from the inspection target distance image created from the inspection target pattern, and the number in the figure is the shortest distance from the contour line of the inspection target distance image to the pixel.
[0015]
The above calculation processing can be realized by a microcomputer. The microcomputer determines that the distance value between the distance images in the difference areas shown in FIGS. 4 and 5 is larger than a preset maximum distance determination threshold value.
[0016]
If the maximum distance determination threshold is set to an intermediate value between the maximum values of the distances between label A and label B, “missing / bleeding” within the thickness variation range of label A is regarded as a non-defective product. Only “chips / bleeding” that seem to be defects can be detected.
[0017]
[Second Embodiment]
6 to 15 show [Second Embodiment].
As in the first embodiment, calculation processing and pass / fail judgment can be realized by a microcomputer. The microcomputer of the [second embodiment] is configured as follows.
[0018]
FIG. 6 shows a non-defective pattern, and FIG. 7 shows a pattern to be inspected. The difference area obtained by extracting the difference area from the non-defective pattern and the inspection target pattern is given the same label and different points belonging to the same area. FIG. 8 shows the areas distinguished from each other by attaching other labels to the areas. Each number of 1-4 is the label number. Next, classification and determination are performed for each label.
[0019]
FIG. 9 is an enlarged view of the label 1. Since the contour line of the label 1 is not shared with the contour lines of both the non-defective pattern and the inspection target pattern, it is determined that the defect is a “hole crushing” defect. Since it is the size that affects the quality of the “hole crushing” defect, if the maximum length in the X-axis direction and the maximum length in the Y-axis direction are larger than a predetermined size determination threshold, it is determined as defective. .
[0020]
FIG. 10 is an enlarged view of the label 2. Since the contour line of the label 2 is shared only with the contour line of one inspection target pattern, it is determined that the defect is a “missing / bleeding” defect. It is the shortest distance from the shared outline and the non-shared outline of each pixel in the region that affects the quality determination of the “missing / bleeding” defect. FIG. 11 shows a distance image showing the shortest distance from the shared contour line of each pixel to the pixel, and FIG. 12 shows a distance image showing the shortest distance from the non-shared contour line to the pixel. . If the maximum distance in the distance images of FIGS. 11 and 12 is larger than a preset maximum distance determination threshold, it is determined as defective.
[0021]
FIG. 13 is an enlarged view of the label 3. Since the contour line of the label 3 is shared only with the contour line of one non-defective pattern, it is determined that the defect is a “missing / bleeding” defect. If the distance is larger than the maximum distance determination threshold, it is determined as defective.
[0022]
Note that if the maximum distance determination threshold value used when determining a defect with a “missing / bleeding” defect is set to an intermediate value between the maximum distance values of the label 2 and the label 3, the “missing” • “Bludge” is regarded as a non-defective product, and only “missing / bleeding” that seems to be a serious defect of the label 4 can be detected.
[0023]
FIG. 14 is an enlarged view of the label 4. Since the outline of the label 4 is shared with the outlines of the two good patterns, it is classified as a “pattern cut / pattern combination” defect. It is the distance between contour lines that are not shared that affects the quality determination of the “pattern cut / pattern combination” defect.
[0024]
However, in the method for obtaining the distance image, as shown in FIG. 15, the distance between the non-shared contour lines cannot be obtained, and only the distance between the pixels between the two non-shared contour lines can be obtained. .
[0025]
Therefore, by calculating twice the distance from the non-shared contour line to the farthest pixel, the distance between the non-shared contour lines is approximated.
Actually, a distance image from a non-shared outline is obtained, and if twice the maximum distance among the distance values is larger than the maximum distance determination threshold used in the pass / fail determination of the “missing / bleeding” defect, it is defective. Is determined.
[0026]
The size determination threshold value and the maximum distance determination threshold value vary depending on the size of the target pattern, the allowable degree of deformation, the allowable degree of inclination, and the allowable degree of variation in pattern thickness. Once determined, highly reliable and highly accurate stamp pattern inspection is possible. In a specific example, in the marking pattern inspection on the capacitor, parameters are set so that the inclination is allowed up to 9 degrees and the thickness variation is allowed up to 2 pixels, and each determination threshold value is calculated according to the size of each pattern. It is confirmed that a good pattern quality inspection result can be obtained by performing the inspection.
[0027]
【The invention's effect】
According to pattern inspection method according to claim 1, wherein the steps of creating a good distance data whose value is the shortest distance to each pixel of the non-defective pattern image from the pixels of the outermost periphery of good products pattern inspection object pattern A step of creating inspection object distance data whose value is the shortest distance from the outermost peripheral pixel to each pixel of the inspection target pattern image, and extracting a difference area between the non-defective pattern image and the inspection target pattern image after, in the differential region, and values of the non-defective distance data and said object distance data is compared with the determination threshold value determined in advance, if at least one of the values exceeds the determination threshold value since chromatic and determining a failure, variation in the imprinted have line thickness on curved surfaces is large, relative prone pattern of deformation, only the critical pattern defects It can be out.
[0028]
According to the pattern inspection method of the third aspect , the difference area is extracted from the image of the non-defective pattern and the image of the inspection target pattern, and the outermost peripheral pixels of the difference area correspond to the non-defective pattern and the inspection target pattern, respectively. The difference area is classified according to the number of pixels shared with the outermost peripheral pixel, and the area of the difference area is determined by the determination threshold according to the classification, or the distance from the outermost peripheral pixel of the difference area is determined by the determination threshold. Even if the pattern is deformed or there is a variation in the pattern thickness, the pattern is stamped on the curved surface and there are many variations in the line thickness. Pattern inspection can be performed with high accuracy.
[Brief description of the drawings]
1 is a non-defective pattern sample used in the first embodiment. FIG. 2 is an inspection target pattern sample used in the first embodiment. FIG. 3 is a first embodiment. Difference area between used good pattern and inspection target pattern [FIG. 4] A diagram showing the distance from the outline of the good pattern used in the difference area between the good pattern and inspection target pattern used in the first embodiment. 5 is a diagram showing the distance from the contour of the pattern to be inspected in the difference area between the non-defective pattern and the pattern to be inspected used in the first embodiment. FIG. 6 is used in the second embodiment. Non-defective product pattern sample [FIG. 7] Sample of target inspection pattern used in [second embodiment] [FIG. 8] Same area of difference area between good product pattern and target inspection pattern in [second embodiment] Image labeled on [Figure 9] In the second embodiment, an image obtained by enlarging the difference area of the label 1 [FIG. 10] In the [second embodiment], an image obtained by enlarging the difference area of the label 2 [FIG. 11] [Second embodiment] In the second embodiment, the distance image from the contour line shared with the pattern to be inspected in the difference area of the label 2 [FIG. 12] In the second embodiment, from the non-shared contour line of the difference area of the label 2 FIG. 13 is an image obtained by enlarging the difference area of the label 3 in the second embodiment. FIG. 14 is an image obtained by enlarging the difference area of the label 4 in the second embodiment. FIG. 15 is a distance image from an unshared outline of the difference area of label 4 in [Second Embodiment]. FIG. 16 is an explanatory diagram of a conventional pattern quality inspection method. Illustration of 【Figure 1 Shows a "collapse hole" pattern inspection method according to the pattern defect type of conventional illustration of a pattern inspection technique [19] illustrates a conventional pattern inspection technique [20] present invention.
FIG. 21 shows “missing / bleeding” of the type of pattern defect by the pattern quality inspection method of the present invention.
FIG. 22 shows “pattern breakage / pattern combination” of the types of pattern defects according to the pattern quality inspection method of the present invention.

Claims (4)

A step of creating a good distance data whose value is the shortest distance to each pixel of the image of the non-defective patterns from the pixels of the outermost periphery of the good product pattern,
A step of creating a test object distance data whose value is the shortest distance to each pixel of the image of the inspection object pattern from the pixels of the outermost periphery of the inspection object pattern,
After extracting the difference area between the image of the good product pattern and the image of the inspection target pattern ,
Each value of the non-defective product distance data and the inspection object distance data in the difference area is compared with a predetermined determination threshold, and if at least one of the values exceeds the determination threshold, it is determined as defective. Process and
A pattern inspection technique. The judgment threshold is
In the difference area, the maximum distance value from the outermost peripheral pixel of the good product pattern to each pixel of the image of the good product pattern,
Set to an average value of the maximum distance value from the outermost peripheral pixel of the inspection target pattern to each pixel of the image of the inspection target pattern in the difference area.
The pattern inspection method according to claim 1. Extract the difference area between the non-defective pattern image and the inspection target pattern image,
The difference area is classified according to the number of pixels that the outermost peripheral pixels of the difference area share with the outermost peripheral pixels of the non-defective pattern and the inspection target pattern,
Depending on the classification, whether the area of the difference area is determined by a determination threshold or whether the distance from the outermost pixel of the difference area is determined by a determination threshold is changed.
Pattern inspection technique. If all the pixels on the outermost periphery of the difference area are shared with the pixels on the outermost periphery of the non-defective pattern, if the maximum lengths of the difference area in the X-axis direction and the Y-axis direction exceed the determination threshold, Judged as bad
When at least one of the outermost peripheral pixels of the difference area is not shared with the outermost peripheral pixel of the non-defective pattern, the outermost peripheral pixel of the difference area is defined as the non-defective pattern and the inspection target pattern. For an area shared with only one of the outermost peripheral pixels, distance data that is the shortest distance from the shared outermost peripheral pixel to the respective pixels in the difference area; and Create distance data that is the shortest distance from the outermost peripheral pixel that is not shared in the difference area to each pixel, and determine that there is a pixel with a distance that exceeds the determination threshold,
For an area shared by two or more outermost peripheral pixels of at least one of the non-defective pattern and the inspection target pattern, the value of each pixel in the difference area is not shared. Distance data that is twice the shortest distance from each pixel to each pixel is created, and if there is a pixel whose distance exceeds the determination threshold, it is determined as defective.
The pattern inspection method according to claim 3.

Images