JP3756264B2 - Method for detecting defects in a sample having a periodic pattern - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for detecting a pattern defect of a sample having a periodic pattern such as a shadow mask.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, various methods are known for detecting defects in industrial products having periodic openings such as shadow masks, that is, using unit openings as unit patterns.
For example, for defects that are visible from the direction perpendicular to the sample surface, a method for recognizing patterns by examining video signals taken with a camera, and comparing defects with patterns obtained by similarly photographing patterns without defects. There are known a detection method, a method using a diffraction phenomenon of light having a periodic pattern when irradiated with coherent light, and the like. However, these methods have a problem that a lot of time is spent for inspection.
[0003]
For this reason, as a method of reducing the inspection time by the applicant of the present application, as shown in FIG. 8A, the transmitted light transmitted through the sample is photographed using a CCD line sensor camera or the like, and the obtained image data ( Signal) and a method of processing the image data (signal) obtained by photographing the specularly reflected light from the sample using a CCD line sensor camera or the like as shown in FIG. 8B. ing.
The method shown in FIG. 8A is for detecting a large or small defect or pinhole in the hole (opening) 890 of the sample, and FIG. 8B shows an egress defect (see FIG. 8) in the hole (opening) 890 of the sample. (B) is for detecting 850) shown in (b).
However, in the case of the method shown in FIGS. 8A and 8B, the captured image is caused by mismatch between the pitch between the unit patterns of the periodic pattern to be captured and one pixel pitch of a CCD line sensor camera or the like. There is a case where a frequency component that cannot be ignored is carried on the signal (image data), which causes a problem that the sensitivity of defect detection is lowered.
It should be noted that the image signal (image data) that has been captured cannot be ignored due to a mismatch between the pitch between the unit patterns of the periodic pattern imaged by the linear area imaging unit and the 1 pixel pitch of the linear area imaging unit. When a frequency component is carried, this frequency component is called moire.
[0004]
On the other hand, Japanese Patent Laid-Open No. 5-302820 discloses a method for detecting an Eggle defect in a hole (opening) of a sample and not affected by moire.
In this method, oblique light that has passed through a sample is photographed by a CCD line sensor camera or the like, and the obtained data is processed to detect a defect. As shown in FIG. It was broken.
However, in the case of this method, as shown in FIG. 9B, since light is transmitted at every pitch of the opening (pattern) 890 of the sample, the angle at which the CCD line sensor camera 801 looks at the sample 807 (elevation angle θ and It is necessary to adjust the optimum angle, that is, the optimum elevation angle.
In the case of this method, when inspected with transmitted light in a specific direction, as shown in FIG. 9B, even if an egress defect 851 is detected, the egress defect 852 is not detected. For this reason, as shown in FIG. 9C, in order to detect the egress in various directions (positions), it is necessary to inspect with transmitted light in the directions. That is, an inspection is required for each inspection region (for each direction).
As described above, the method of photographing oblique light transmitted through the sample with a line CCD line sensor camera or the like shown in FIG. 9 is not affected by moire and can detect an egre defect, but it increases the inspection tact. It was.
[0005]
[Problems to be solved by the invention]
As described above, any method for detecting defects in industrial products having periodic holes (also referred to as openings) such as a shadow mask has a problem of increasing the time spent on inspection or is affected by moire. There is a problem that the defect detection sensitivity is lowered, and this countermeasure has been demanded.
Under such circumstances, the present invention is designed to detect defects in industrial products having a periodic opening such as a shadow mask, that is, using unit openings as unit patterns, and unit patterns are periodically arranged. Similar to the method shown in FIGS. 8 (a) to 8 (b), it is intended to provide a defect detection method which is performed by specularly reflected light or transmitted light passing through the sample from the illumination means and is not affected by moire.
[0006]
[Means for Solving the Problems]
A defect detection apparatus for a sample having a periodic pattern according to the present invention is a defect detection method for detecting a defect in a sample having a periodic pattern, which moves at a constant speed, in a direction perpendicular to the moving direction of the sample. Illuminating a linear area of the sample with a linear light source, and imaging the specularly reflected light or transmitted light that has passed through the sample from the linear illumination with the linear area imaging means, and obtained by the linear area imaging means An image data processing step for detecting a defect of the sample from the obtained image data, wherein the image data processing step is differentiating the first image data obtained directly or indirectly by the linear region imaging means. A differential operation process using a filter is performed to obtain second image data, and a comparison process for comparing the obtained second image data with a predetermined slice level is performed to detect a defective portion. , Linear area Pitch between unit patterns of the periodic pattern to be photographed by the shadow means, corresponds to an integral multiple of one pixel pitch of the linear areas photographing means, wherein the differential filter is a pitch between the unit pattern of the periodic pattern And the size and coefficient for emphasizing the defective portion in accordance with the ratio of the linear region photographing means to one pixel pitch .
In the above, the linear area photographing means is a CCD line sensor camera.
In addition, the comparison process described above is a binarization process that binarizes the obtained second image data at a predetermined slice level.
In the above, the sample is a shadow mask or a color filter.
[0007]
Note that a sample having a periodic pattern is a sample in which unit shapes are periodically repeated, for example, an industrial product in which openings are periodically repeated such as a shadow mask, and an opening having a unit shape such as a shadow mask is 1 Samples having a periodic pattern are treated as two patterns and are arranged repeatedly.
In addition to the shadow mask, units with a certain optical property shape (this is called a unit pattern), such as a color image pickup tube decomposition filter, photomask, and Fresnel lens, are regularly arranged one-dimensionally and two-dimensionally. Examples of a sample having a periodic pattern include industrial products that have a periodic pattern, or industrial products in which unit patterns are repeatedly arranged while gradually changing the optical properties, shape, and arrangement pitch in one-dimensional and two-dimensional directions. The same applies to plain paper, film, iron plate, coated surface, and the like.
In addition, specularly reflected light means reflected light in which the incident light on the sample and the reflected light from the sample are opposite to each other at the same angle with respect to the surface perpendicular to the sample surface. Data obtained by photographing by the photographing means and data processed by the processing means are collectively referred to as image data.
[0008]
[Action]
The defect detection apparatus for a sample having a periodic pattern according to the present invention is configured as described above to detect a defect of an industrial product in which unit patterns moving at a constant speed are periodically arranged from the illumination means to the sample. Due to the mismatch between the pitch between the unit patterns of the periodic pattern to be photographed and the one-pixel pitch of the linear region photographing means, the photographed image signal (image data) is It is possible to provide a defect detection method in which frequency components that cannot be ignored, that is, moire does not occur.
Specifically, a linear region of the sample in a direction perpendicular to the moving direction of the sample is illuminated with a linear light source, and regular reflection light or transmitted light that has passed through the sample from the linear illumination is captured by the linear region imaging means. An image data processing step of detecting a defect of the sample from the image data obtained by the linear area imaging means, wherein the image data processing means is directly or indirectly by the linear area imaging means. The first image data obtained in the first step is subjected to a filtering process such as a differential operation process to obtain second image data, and the second image data obtained is compared with a predetermined slice level. The pitch between the unit patterns of the periodic pattern photographed by the linear area photographing means corresponds to an integer multiple of one pixel pitch of the linear area photographing means. By being It has achieved a record.
Specifically, the entire area is simplified by using a CCD line sensor camera as the linear area photographing means, and is particularly effective when the sample is a shadow mask.
[0009]
【Example】
The defect detection method for a sample having a periodic pattern according to the present invention will be described with reference to the drawings.
FIG. 1 is a flowchart showing the steps of the defect detection method of this embodiment, and FIG. 2 shows the relationship between the pitch of the periodic pattern to be photographed and the pixels of the photographing means, which is a characteristic part of this embodiment. It is a figure.
FIG. 3 is a schematic diagram of an apparatus configuration for carrying out the present embodiment, FIG. 4 is a diagram for explaining a photographing mechanism, and FIG. 4B is a diagram of A1 in FIG. The cross section in A2 is shown.
1, FIG. 2, FIG. 3 and FIG. 4, 100 is a defect detection device, 110 is a sample (shade mask), 120 is a CCD line sensor camera, 130 is a linear light source (fluorescent lamp), 130A is a linear region, 140 Is a transfer device, 150 is a camera rotation control unit, 160 is a camera movement unit, 170 is a signal processing unit, 180 is a control unit, 190 is a hole (opening), 191 is a hole arrangement direction, 192 is a taper portion, and 200 is Normal direction.
The apparatus shown in FIG. 3 is placed on a transport device 140 and a linear light source 130 placed in a direction perpendicular to the transport direction with respect to a sample (shade mask) 110 having a periodic pattern that is transported sequentially. This is a defect detection device that performs illumination, images the transmitted light from the sample 110 with the CCD rachiin sensor camera 120, processes the obtained signal data (image data) with the signal processing unit 170, and detects defects. . In this embodiment, unlike the method shown in FIG. 9, the optimal elevation angle adjustment and the inspection for each inspection region (for each direction) are not required.
[0010]
In this embodiment, the defect detection of the shadow mask in which holes (openings) are periodically arranged is detected by the apparatus shown in FIG. 3 in the manner of imaging shown in FIG. 4 and as shown in FIG. This is a defect detection method in a case where defect detection is performed by setting the relationship between the pitch of the periodic pattern and the pixels of the imaging means, processing the captured image data.
First, as shown in FIG. 2, the relationship between the pitch of the holes (openings) 190 of the shadow mask to be photographed and the pixels of the CCD line sensor camera 120 is set. (Fig. 1 (S10))
The pitch of the holes (openings) 190 of the shadow mask to be photographed is set so as to correspond to an integer multiple of one pixel pitch of the CCD line sensor camera 120.
In this embodiment, as shown in FIG. 2, the ratio of the hole (opening) of the photographed shadow mask, that is, the pitch of the photographed periodic pattern to the pixel pitch of the CCD line sensor camera is 3: 1. .
[0011]
Next, while placing a shadow mask (sample) on the apparatus shown in FIG. 3, the linear region on the surface of the sample is illuminated with a linear light source, and the specularly reflected light passing through the sample from the linear illumination is captured by the linear region imaging means. Images are taken, and image data Ni (D _ij ) corresponding to scan (scan) is sequentially obtained as shown in FIG. (Fig. 1 (S20))
Here, Ni (D _ij ) is image data corresponding to the i-th scan (scanning), and D _ij means a data value in the j-th pixel of the CCD line sensor camera in the i-th scan.
[0012]
Next, as shown in FIG. 5, for all pixels of the CCD line sensor camera, as shown in FIG. 5, i-th scan data Ni, i + 1-th scan data Ni + 1, i + 2-th scan data Ni + 2 Is integrated to generate first image data Si. (Fig. 1 (S30))
The integration processing is performed in order to reduce the influence of random noise by performing integration processing on a plurality of scan (scan) image data.
As shown in FIG. 6 (a), the accumulated first image data Si is data on a large undulation having a small frequency component. There is no impact.
From the first image data Si, a frequency that cannot be ignored in the captured image signal (image data) due to a mismatch between the pitch between the unit patterns of the periodic pattern to be captured and the one-pixel pitch of the linear region imaging means. Ingredients have been removed.
The cause of the undulation shown in FIG. 6A is that the pitch between the unit patterns of the periodic pattern to be photographed is not perfectly matched with the one pixel pitch of the linear region photographing means.
Needless to say, the integration is not particularly limited to the image data for three scans (scanning).
[0013]
Next, the first image data is subjected to filter processing such as differential calculation processing to obtain second image data in which the defective portion is emphasized. (Fig. 1 (S40))
In this embodiment, since the ratio of one pitch (minimum pitch) of the periodic pattern photographed as shown in FIG. 2 to the pixel pitch of the CCD line sensor camera is 3: 1, as a filter, FIG. The differential filter shown in a) was used.
When the ratio of one pitch (minimum pitch) of the captured periodic pattern to the pixel pitch of the CCD line sensor camera is 2: 1, the filter shown in FIG. Should be used.
Further, the filter process is a process of calculating a sum of products with respect to the accumulated image data (first image data) using a filter having an element array.
Here, using the image data (signal data) Si (Dij) obtained from the photographing means (CCD line sensor camera 120) and the spatial filter table W (Ak), Si (Dij) and W (Ak) are used. ) Is called filter processing, and W (Ak) used for this processing is called a filter, and is generally called a differential filter or a spatial filter.
As shown in FIG. 6B, the second image data obtained by the filter process is data in which the defective portion is emphasized.
[0014]
Further, binarization processing for binarizing the second image data at a predetermined slice level is performed to extract only defective portions. (Fig. 1 (S50))
The binarized data is shown in FIG. 6C, and only defective portions are detected.
In this way, a frequency component that cannot be ignored is not generated in the captured image signal (image data) due to a mismatch between the pitch between the unit patterns of the periodic pattern to be captured and the one-pixel pitch of the linear region imaging means. As a result, defect detection can be reliably performed.
[0015]
The apparatus shown in FIG. 3 used in this embodiment will be further described.
In FIG. 3, the sample (shade mask) 110 to be inspected is moved in the direction of the arrow by the transport device 140 and sequentially passes below the CCD line sensor camera 120.
Further, as shown in FIG. 4, the rotation drive control unit 150 adjusts so that the scanning direction of the CCD line sensor camera 120 and the arrangement direction 191 of the holes (openings) 190 of the sample (shade mask) 110 are parallel to each other. To do.
Although a fluorescent lamp is used as the linear light source 130, the CCD line sensor camera 120 converts the transmitted light that has passed through the sample (shade mask) 110 by irradiating the linear region 130 A of the sample in a direction orthogonal to the traveling direction of the sample 110. It was set to be able to shoot with.
[0016]
In the apparatus shown in FIG. 3, the CCD can be adjusted so that the field of view of imaging can be adjusted and the pitch between the unit patterns of the periodic pattern of the sample 110 is an integral multiple of one pixel pitch of the linear area imaging means. The field of view of the line sensor camera 120 is adjusted.
In the case of the present embodiment, the pitch between the unit patterns of the periodic pattern of the sample 110 to be photographed is set to 3 times the one pixel pitch of the linear region photographing means.
[0017]
The signal processing unit 170 processes image data obtained by photographing with the CCD line sensor camera 120 to detect defects, and integrates image data obtained by scanning (scanning) shown in FIG. Integration processing, filter processing for obtaining second image data by multiplying the image data (first image data) obtained by the integration, and obtaining the second image data at a predetermined slice level. A binarization process for binarization is performed.
[0018]
In the above embodiment, the method shown in FIG. 8A is not affected by moire, but also in the method shown in FIG. It goes without saying that the influence of moire can be eliminated by setting the minimum pitch) to an integral multiple of the pixel pitch of the CCD line sensor camera.
[0019]
【effect】
In the present invention, as described above, the defect detection of an industrial product in which unit patterns such as a shadow mask are periodically arranged is detected by specularly reflected light or transmitted light passing through the sample from the illumination means, and the period of photographing. Provided a defect detection method in which a frequency component that cannot be ignored, that is, moire does not occur in a photographed image signal (image data) due to a mismatch between a pitch between unit patterns of a characteristic pattern and a pixel pitch of a linear region photographing means As a result, defect detection is made more reliable.
In addition, unlike the method of photographing oblique light transmitted through the sample shown in FIG. 9, it is not necessary to perform optimum elevation angle adjustment or inspection for each inspection region (for each direction), and the defect detection method is excellent in terms of workability. Is possible.
[Brief description of the drawings]
FIG. 1 is a flowchart showing steps of a defect detection method according to the present embodiment. FIG. 2 is a view showing a relationship between a pitch of a periodic pattern to be photographed and pixels of a photographing means. Schematic diagram of apparatus configuration for carrying out FIG. 4 is a diagram for explaining the mechanism of photographing. FIG. 5 is a diagram for explaining signal integration. FIG. 6 is an image data after integration and an image after filtering. FIG. 7 is a diagram showing data. FIG. 7 is a diagram showing a differential filter. FIG. 8 is a diagram for explaining a conventional defect detection method. FIG. 9 is a diagram for explaining a conventional defect detection method.
100 Defect detection device 110 Sample (shade mask)
120 CCD line sensor camera 130 Linear light source (fluorescent lamp)
130A Linear area 140 Conveying device 150 Camera rotation movement control unit 160 Camera movement unit 170 Signal processing unit 180 Control unit 190 Hole (opening)
191 Hole arrangement direction 192 Taper part 195 Hole pitch (unit pattern minimum pitch)
200 Normal direction 801 CCD line sensor camera 802 Linear light source 803 Lens 805 Rotating stage 806 Uniaxial moving stage 807 Sample 808 Holding column 809 Camera controller and signal processing device 810 Controller 811 Control computer 830 Transmitted light 835 Regular reflected light 850, 851, 852 Aegle defect 890 hole (opening)
892 Taper

Claims (4)

A defect detection method for detecting defects in a sample having a periodic pattern that moves at a constant speed, illuminating a linear region of the sample in a direction perpendicular to the moving direction of the sample with a linear light source, A step of photographing the specularly reflected light or transmitted light that has passed through the sample from the linear illumination by the linear region photographing unit, and an image data processing step of detecting a defect of the sample from the image data obtained by the linear region photographing unit; And the image data processing step performs a differential operation process using a differential filter on the first image data obtained directly or indirectly by the linear region photographing means, to obtain a second image. A unit of periodic pattern which is obtained by obtaining a data and performing a comparison process comparing the obtained second image data with a predetermined slice level to detect a defective portion and is photographed by a linear region photographing means pattern Pitch, equivalent to an integral multiple of one pixel pitch of the linear areas photographing means, said differential filter, and the pitch between the unit pattern of the periodic pattern, and one pixel pitch of the linear region imaging means A defect detection method for a sample having a periodic pattern, characterized in that the size and coefficient for emphasizing a defect portion are set in accordance with the ratio .   2. The defect detection method for a sample having a periodic pattern according to claim 1, wherein the linear area photographing means is a CCD line sensor camera.   3. A defect detection method for a sample having a periodic pattern, wherein the comparison process according to claim 1 is a binarization process for binarizing the obtained second image data at a predetermined slice level.   4. The defect detection method for a sample having a periodic pattern according to claim 1, wherein the sample is a shadow mask or a color filter.

Images