JP4357640B2 - A method for inspecting a defect of a quasi-periodic pattern. - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for inspecting a quasi-periodic pattern having array elements whose array pitch (interval) or array direction (tilt) is changed based on a certain rule in the same plane.
[0002]
[Prior art]
Conventionally, an inspection method for detecting defects in a periodic pattern of a product in which array elements are repeatedly arranged at a predetermined array pitch to form a periodic pattern based on a predetermined rule in the same plane, or the same Based on a predetermined rule in the plane, the array elements are repeatedly arranged while changing the array pitch and the array direction little by little to detect a defect in the pseudo period pattern of the product forming the pseudo period pattern. As an inspection method, visual inspection was performed by an inspector.
However, in the visual inspection, there is a problem of inspector fatigue and a problem of variation in the determination result for each inspector.In these periodic pattern and pseudo periodic pattern inspection, instead of visual inspection, There was a need for automated inspection.
In recent years, in products such as shadow masks that have pseudo-periodic patterns with through-holes as array elements, with increasing mass production and higher product accuracy, in particular, inspections with high defect detection accuracy have been carried out. There was a need for automation.
[0003]
On the other hand, as an alternative to visual inspection by the applicant of the present application, defect detection of a periodic pattern or a pseudo-periodic pattern is performed by extracting a defective portion from a video signal obtained from an imaging means such as a CCD. A method and an apparatus for implementing it have been proposed.
Moire generated because the interval of the periodic pattern arrangement in the video signal obtained by the imaging means does not coincide with or does not become an integer multiple of the pixel arrangement of the imaging means causes a pseudo defect, which significantly reduces the detection sensitivity of the defect. Therefore, in these defect detection methods, the optical system is adjusted so that the interval of the arrangement of the periodic patterns in the obtained video signal coincides with the arrangement of the imaging means pixels or becomes an integral multiple, and the focus is set so that moire does not occur. The method of shifting was adopted.
[0004]
However, the method of shifting the focus leads to a decrease in detection sensitivity due to a decrease in resolution.
In addition, the method of adjusting the optical system so that the interval of the periodic pattern array in the obtained video signal is equal to or an integer multiple of the pixel array of the imaging means can be adjusted to a certain area when inspecting the pseudo periodic pattern. Even if the optical system is adjusted and the pseudo defect does not occur, under this adjustment, the interval of the arrangement of the periodic patterns in the video signal obtained in other areas is the same as the pixel arrangement of the video means or an integral multiple. Instead, pseudo defects may occur.
[0005]
[Problems to be solved by the invention]
As described above, the pseudo-period of the product in which the array elements are repeatedly arrayed to form a pseudo-periodic pattern while changing the array pitch and the array direction little by little based on a predetermined rule in the same plane. In the inspection for detecting pattern defects, there has been a demand for an inspection method in which detection of pseudo defects is small, the cost of the inspection apparatus is increased, and the handling of each type is not troublesome.
The present invention, corresponding to this, provides a method for detecting a defect in a pseudo-periodic pattern, in which the detection of pseudo defects is small, the cost of the inspection apparatus does not increase, and the handling of each type does not take time. It is something to try.
[0006]
[Means for Solving the Problems]
The method for inspecting a defect of a quasi-periodic pattern according to the present invention forms a quasi-periodic pattern by repeatedly arranging array elements while gradually changing the array pitch and the array direction based on a predetermined rule in the same plane. A method for detecting a defect in a pseudo-periodic pattern of a product, which is obtained by an imaging means using a CCD element or the like, and a video signal of the product's pseudo-periodic pattern is subjected to A / D conversion processing, After obtaining digitized image data corresponding to a video signal, for the obtained image data, (a) an array element pixel as a pixel area occupied by each array element corresponding to a predetermined array pitch and array direction An area is determined, the value of each pixel corresponding to the array element pixel area is added, and an integrated value is obtained for each array element pixel area, and (b) based on the result of the integration process , For each array element pixel area, enhancement processing such as differentiation processing for emphasizing the difference in integrated value for each array element pixel area, and (c) a predetermined slice for the result obtained by the enhancement processing step compared to the level, a series of processing consisting of a defect candidate extracting process of extracting the defect candidate, performs a plurality of times, which performs defect detection processing step of extracting a defect portion from the extraction results obtained, The defect detection processing step divides the digitized image data into several areas in advance, and then, for each of the divided areas, array element pixel regions corresponding to the array element arrangement of the areas, respectively. The integration process, the emphasis process, and the defect candidate extraction process are sequentially performed, and the obtained defect part candidates are combined and extracted as a defect part .
Further, according to the defect inspection method of the pseudo periodic pattern of the present invention , the array elements are repeatedly arranged while changing the array pitch and the array direction little by little based on a predetermined rule in the same plane. This is an inspection method for detecting a defect in a pseudo-periodic pattern of a product being formed, and an A / D conversion process is performed on a video signal of the pseudo-periodic pattern of the product obtained by an imaging means using a CCD element or the like. Then, after obtaining digitized image data corresponding to the video signal, for the obtained image data, (a) an array as a pixel area occupied by each array element corresponding to a predetermined array pitch and array direction An integration process of the array element pixel area that determines an element pixel area, adds the values of each pixel corresponding to the array element pixel area, and obtains an integrated value for each array element pixel area; On the basis of each of the array element pixel areas, an emphasis process such as a differentiation process for emphasizing the difference in the integrated value for each array element pixel area, and (c) Compared to the slice level, a series of processing consisting of defect candidate extraction processing for extracting defect portion candidates is performed a plurality of times, and a defect detection processing step for extracting defect portions from the obtained extraction results is performed . In the defect detection processing step, a plurality of array element pixel areas corresponding to an array having a predetermined array pitch and array direction are determined in advance, and each array pixel area is determined for the entire digitized image data. Use each of the integration process, the enhancement process, and the defect candidate extraction process, and from the result obtained by the defect candidate extraction process, an area where the number of detected defective part candidates is less than a predetermined number is set as the array pixel area. As defect detection zone to respond that, it extracts a defective portion candidates compartment region as a defect part, the combined defect of the defect detection zone, and extracts the defect in the entire quasi-periodic pattern It is characterized by.
In any of the above, the product is a shadow mask.
[0007]
[Action]
By adopting such a configuration, the pseudo periodic pattern defect inspection method of the present invention has less detection of pseudo defects, does not increase the cost of the inspection apparatus, and does not require time and effort for each type of product. Thus, it is possible to provide a defect detection method for a pseudo-periodic pattern.
Specifically, after a product pseudo-periodic pattern video signal obtained by an imaging means using a CCD element or the like is A / D converted to obtain digitized image data corresponding to the video signal Then, for the obtained image data, (a) an array element pixel area as a pixel area occupied by each array element corresponding to a predetermined array pitch and array direction is determined, and each pixel corresponding to the array element pixel area (B) Integration processing for each array element pixel area for each array element pixel area based on the result of the integration process. A defect candidate extraction process for extracting a defect candidate by comparing with a predetermined slice level with respect to a result obtained by an enhancement process such as a differential process for emphasizing a difference in values and (c) the enhancement process step A series of The management, performed a plurality of times, which performs obtained extraction result defect detection process for extracting a defective portion from step, the defect detection process is divided, the digitized image data, in advance some areas After that, for each divided area, an array element pixel area corresponding to the array element array of each area is determined, and the integration process, the enhancement process, and the defect candidate extraction process are sequentially performed. In addition, the defect detection processing step is performed in advance by arranging the defect candidate candidates, and the defect detection processing step includes an array element pixel area corresponding to an array having a predetermined array pitch and array direction. Are determined, and each of the array pixel areas is used for the entire digitized image data, and the integration process, the emphasis process, and the defect candidate extraction process are performed, respectively. From the results obtained, the area where the number of detected defective part candidates is less than a predetermined number is extracted as a defective part detection area corresponding to the array pixel area, and the defective part candidate in the area is extracted as a defective part. This is achieved by combining the defect parts in each defect part detection area and extracting them as defect parts in the entire pseudo-periodic pattern .
Specifically, by determining the array element pixel area corresponding to the array pitch (interval) and array direction (tilt) of the array elements of the pseudo periodic pattern, the image data of each array element of the pseudo periodic pattern is determined from the integrated value. It is assumed that you can be surely obtained.
In addition, by performing enhancement processing such as differentiation processing for each array element pixel area, it is possible to emphasize the difference in the integrated value for each array element pixel area, and the defect candidate extraction process can reliably Candidates can be extracted.
[0008]
More specifically, in the defect detection processing step, the digitized image data is divided into several areas in advance, and each divided area corresponds to the arrangement of array elements in the area. By determining the array element pixel area, sequentially performing integration processing, enhancement processing, defect candidate extraction processing, and extracting the obtained defect portion candidates together as a defect portion, or In the detection processing step, a plurality of array element pixel areas corresponding to arrays having a predetermined array pitch and array direction are determined in advance, and each array pixel area is used for the entire digitized image data. An integration process, an emphasis process, and a defect candidate extraction process are performed, and an area in which the number of detected defective part candidates is less than a predetermined number is determined from the result obtained by the defect candidate extraction process. As a corresponding defect detection area, the defect candidate in the area is extracted as a defect, and the defect in each defect detection area is extracted as a defect in the entire pseudo-periodic pattern. This has been achieved.
That is, the entire quasi-periodic pattern is divided into a plurality of predetermined areas having different arrangement pitches (intervals) and arrangement directions (tilts) of array elements, and the corresponding video signals are processed under conditions suitable for each area, Each defect candidate is detected, and the defect portion is detected together with the defect candidate detected for each area, so that it is possible to detect the defect portion with less extraction of pseudo defects.
[0009]
In particular, it is possible to cope with inspection of a product having a through hole as a pseudo periodic pattern, such as a shadow mask, and it is possible to cope with mass production and high accuracy.
[0010]
According to the method for inspecting a defect of a quasi-periodic pattern of the present invention, when the quasi-periodic pattern is within the field of view of the imaging unit, the image signal of the field of view of one imaging unit is simply image-processed for each area. In addition, there is no need to arrange an optical system, and enormous adjustment time is not required when changing the product type.
In the case where the pseudo periodic pattern is large, or in the case of inspection that requires resolution, that is, the detection of the pseudo defect is extremely difficult, this can be dealt with by performing the same processing using a plurality of imaging means.
[0011]
DETAILED DESCRIPTION OF THE INVENTION
An example of an embodiment of a pseudo periodic pattern defect inspection method of the present invention will be described with reference to the drawings.
FIG. 1 shows a flow of a first example of an embodiment of a defect inspection method for a quasi-periodic pattern according to the present invention. FIG. 2 (a) is a schematic configuration diagram of an apparatus for carrying out this, and FIG. ) Is a schematic configuration diagram of a modified example of the apparatus, FIG. 3 is a diagram for explaining the integration process, FIG. 4 is a diagram for explaining the areas into which the pseudo-periodic pattern is divided, and FIG. 5 is a defect candidate extraction process It is a figure for demonstrating.
In FIG. 1, S110 to S222 indicate processing steps.
1 to 5, 110, 111, and 112 are imaging means, 120 and 121 are illumination means, 130 is an image processing unit, 140 is an A / D conversion unit, 180 and 181 are inspection samples (inspected objects), 310. Is a pixel field of view (pixel region), 321, 322, 323 are array element regions, 321 A, 322 A, 323 A are array element pixel regions, X 1, X 2, X 3 are X direction lengths of the array element pixel regions, Y 1, Y 2, Y3 is the length of the array element pixel region in the Y direction, ΔX is the pitch in the X direction, and ΔY is the pitch in the Y direction.
[0012]
First, a first example of the embodiment will be given.
This example uses a device shown in FIG. 2 (a) to detect a defect in a quasi-periodic pattern of a plate-like metal plate product having a quasi-periodic pattern in which fine through-holes are arranged as array elements. In the case of the case, in the condition determination process (S110), a sample is used in advance, the process condition and the area of the pseudo-periodic pattern corresponding to the sample are determined, and this defect is determined in the product defect detection process (S120). The defect detection is performed based on the processing conditions and areas.
In addition, since it is a condition that an array element pixel area corresponding to each detection area is determined and used when detecting a defect, the processing based on the array element pixel area for each specific detection section is referred to here. Then, it says processing conditions.
[0013]
The processing will be described with reference to FIGS.
First, a condition is obtained by dividing the image data obtained at the time of inspection from the sample (defect sample) into areas corresponding to the processing conditions, and obtaining the processing conditions corresponding to the areas. (S110)
First, a product sample is used as an inspection sample 180, and the entire pseudo-periodic pattern is photographed by the imaging means 110 (S111) to obtain a video signal. (S112) Next, A / D conversion is performed on the video signal (captured image data) obtained by photographing (S113) to obtain digitized image data. (S114)
Note that a predetermined defect is provided in a predetermined position in the sample in advance.
In this example, a CCD camera is used as the imaging unit 110, and a video signal (captured image) is obtained by photographing with the transmitted light emitted by the illumination unit 120 arranged on the opposite side of the inspection sample 180 from the imaging unit 110.
Then, the video signal obtained by photographing by the imaging unit 110 is A / D converted by the A / D conversion unit 140 to obtain digitized image data.
[0014]
The obtained video signal (captured image) includes array information that corresponds to the array of pseudo-periodic patterns, and whose array pitch (interval) and array direction are different. When a defect is detected by the same process, many pseudo defects are extracted. Therefore, here, the area of the image data is divided into several areas of the pseudo-periodic pattern in which the intervals and inclinations of the array elements are constant to some extent.
Specifically, the digitized image data area is divided into N sections in the following manner in accordance with the array element information of the pseudo periodic pattern.
Detect defect candidates under each of N predetermined processing conditions, select areas where the number of defect candidates is equal to or less than a predetermined number, with each N areas filling the entire area of the image data, This area is a defect candidate detection area.
If necessary, the entire area of the image data can be divided into N areas corresponding to the N processing conditions, except for the overlapping part of the areas.
4A shows an example of image data division when N is 2 and the corresponding area is 2, and FIG. 4B shows an example when N is 3. FIG. It is.
[0015]
In the processing condition n (S116), the array element pixel area SAn is determined from the array pitches Pxn and Pyn and the array direction θn, and the value of each pixel in each array element pixel area is added to obtain an integrated value. Do.
Next, if there is a difference in the integrated value for each array element pixel area SAn, a secondary differentiation process is performed as an emphasis process in order to emphasize it. (S117) The secondary differentiation process is performed using a spatial filter for each array element pixel area SAn.
Compared with a predetermined slice level, defect candidates are extracted, and the processing condition n and the obtained detection area n are stored with an area having a predetermined number or less of defect candidates as a defect detection area n that meets this processing condition. To do.
In this way, n is performed from 1 to N.
In this way, each detection area n corresponding to the process n can be obtained.
[0016]
Here, detection of defect candidates for each processing condition will be described with reference to FIG.
FIG. 3 shows a pixel field of view (pixel region) 310 that is a part of digitized image data and corresponding array elements 321, 322, and 323 of the pseudo-periodic pattern of the product so that the positional relationship thereof can be understood. In the enlarged schematic view, a part of a predetermined area corresponding to a predetermined processing condition is shown.
In FIG. 3, reference numerals 321A, 322A, and 323A denote array element pixel regions.
As shown in FIG. 3, the array is inclined by θ with respect to the x direction of the figure, and each array element 321, 322, 323 is a rectangular shape with pixels of X1 × Y1, X2 × Y2, and X3 × Y3, respectively. , And can be surrounded by the array element pixel regions 321A, 322A, 323A.
In an area where the arrangement pitch (interval) and arrangement direction (inclination) of the arrangement elements 321, 322, and 323 are somewhat constant, X1 = X2 = X3 and Y1 = Y2 = Y3.
Further, it is assumed that the center positions of the array elements are separated by ΔX and ΔY.
[0017]
Let the array element 322 be the array element of interest.
For example, the sum of the pixel values of the array element of interest is defined as SUM (also referred to as an integrated value), the sum of the pixel values of the surrounding array elements 321 and 322 is defined as SUM, and a function of the distance from the array element, respectively. Assuming that SUM (−ΔX, −ΔY) and SUM (+ ΔX, + ΔY) are expressed, the differential value (Bibunti) obtained by subjecting the SUM data to the second order differential processing in the array element 322 of interest is expressed by the following equation (1): It is expressed as
Bibunti = SUM (−ΔX, −ΔY) + SUM (+ ΔX, + ΔY)
-2 x SUM (1)
In order to compare the array element of interest with surrounding array elements, the calculation of the above equation (1) is performed on the entire predetermined area.
Only when the sum of the values of the pixels of the array element of interest is different from the sum of the values of the pixels of the surrounding array elements, the differential value (Bibunti) that is the operation value is significantly different from that of the other.
Thus, it is possible to detect a defect candidate by comparing with a predetermined value (slice level).
For example, as shown in FIG. 5 (a), when SUM data (integrated value) is obtained, the data of the secondary differential processing of the SUM data (integrated value) is as shown in FIG. 5 (b). Become.
5A is removed, and only the defect candidate signal A0 in FIG. 5A is emphasized and obtained as the defect candidate signal A1 in FIG. 5B.
As shown in FIG. 5B, the defect candidate signal A1 is significantly larger or smaller than the normal part.
By setting a slice level SLl for extraction so that only such a signal value can be extracted and comparing it with a predetermined slice level, the location of the defect candidate can be recognized.
Here, such processing is performed on the entire pseudo-periodic pattern under the processing condition n, and only a predetermined area of the pseudo-periodic pattern in which the number of defect candidate candidates to be extracted is a predetermined number or less corresponds to the processing condition n. Detection area n.
[0018]
Next, the defect detection of the quasi-periodic pattern is performed on the product as follows. (S210)
First, the entire quasi-periodic pattern of the product is photographed by the photographing means 110 (S211), and a video signal (captured image data) is obtained. (S212)
Next, the obtained video signal (photographed image data) is A / D converted to obtain digitized image data. (S213)
Next, in the condition setting process (S110), a defect candidate is detected for the detection area n of the corresponding image data for each predetermined processing condition n. (S217 to S218)
Here, the obtained defect part candidates for each detection area n are used as defect parts for each detection area n, and information on the defect positions is accumulated. (S219)
Masking is performed on areas other than the target area of the processing conditions, and defect candidates are detected.
In the defect detection, as in the case of the condition determination process (S110), the integration process and the secondary differentiation process are performed, and then the obtained defect part candidate is detected as a defect as it is compared with a predetermined slice level. Is.
And the defective part detected in each process condition is put together and it is set as the defective part of a product.
[0019]
Next, a second example of the embodiment will be described.
In the second example, the condition determination process S110 shown in FIG. 1 is applied to defect detection as it is. In the process S110 shown in FIG. 1, product shooting is performed instead of sample shooting (S111). Is similar to the condition determination process in the first example of the embodiment, and accumulates information on the defect position in the area n when the detection area n corresponding to the processing condition n is obtained.
That is, in each processing condition n, the defect part candidate in the detection area n is directly used as a defect part, and all the processing conditions n and the defect part in the corresponding detection area n are combined and detected as a defect. .
Since each process overlaps with the first example of the embodiment, description thereof is omitted.
[0020]
(Modification)
In the first example and the second example, the inspection object is photographed from the field of view of one CCD camera. However, in the case where the inspection object is large or inspection that requires resolution, a plurality of imaging means are installed. Then, determine the area to be processed for each camera, apply the same processing as the first example and the second example to the determined area for each camera, detect defects in the area, A method for detecting defects by combining defect detection data for each camera is also applicable.
For example, as shown in FIG. 2B, the inspection sample 180 is photographed by the second imaging means 111 and 112.
The A / D conversion unit 140 performs A / D conversion on the video signals from the two imaging units 111 and 112, but each A / D conversion unit is provided for each imaging unit 111 and 112. Needless to say, you can have it.
[0021]
【The invention's effect】
As described above, the present invention can provide a method for detecting a defect in a pseudo-periodic pattern, in which the detection of pseudo defects is small, the cost of the inspection apparatus is not increased, and the handling of each type is not troublesome. It was.
In particular, it is effective for detecting defects in pseudo-periodic patterns of plate-like metal plate products having pseudo-periodic patterns in which fine through-holes are arranged as array elements, such as shadow masks. It can cope with accuracy.
[Brief description of the drawings]
FIG. 1 is a flowchart of an example of an embodiment of a defect inspection method for a quasi-periodic pattern according to the present invention. FIG. 2A is a diagram illustrating an embodiment of a defect inspection method for a quasi-periodic pattern according to the present invention. FIG. 2B is a schematic configuration diagram of a modified example of the apparatus for carrying out the example.
FIG. 3 is a diagram for explaining an integration process. FIG. 4 is a diagram for explaining an area for dividing a periodic pattern. FIG. 5 is a diagram for explaining a defect candidate extraction process.
110, 111, 112 Imaging means 120, 121 Illumination means 130 Image processing unit 140 A / D conversion unit 180, 181 Inspection sample (inspection object)
310 Pixel field of view (pixel area)
321, 322, 323 Array element area 321 A, 322 A, 323 A Array element pixel area X 1, X 2, X 3 X direction length Y 1, Y 2, Y 3 Array element pixel area Y direction length ΔX X direction pitch ΔY Y direction pitch

Claims (3)

Based on a predetermined rule in the same plane, the array elements are repeatedly arrayed while gradually changing their array pitch and array direction to detect a defect in the pseudo-periodic pattern of the product forming the pseudo-periodic pattern. A digital signal corresponding to the video signal is obtained by performing A / D conversion processing on the video signal of the pseudo-periodic pattern of the product obtained by the imaging means using a CCD element or the like. After that, for the obtained image data, (a) an array element pixel area as a pixel area occupied by each array element corresponding to a predetermined array pitch and array direction is determined, and the array element pixel area corresponds to the array element pixel area The value of each pixel is added to obtain an integrated value for each array element pixel area, and (b) an array element pixel area for each array element pixel area based on the result of the integration process. A defect for extracting a defect candidate by comparing with a predetermined slice level with respect to a result obtained by an enhancement process such as a differentiation process and (c) an enhancement process step for emphasizing a difference of each integrated value a series of processing comprising the candidate extraction process is performed a plurality of times, which performs defect detection processing step of extracting a defect portion from the extraction results obtained, the defect detection processing step, digitized image data Is divided into several areas in advance, and for each of the divided areas, an array element pixel region corresponding to the array element array of each area is determined, and in order, integration processing, enhancement processing, defect candidate extraction A method for inspecting a defect of a quasi-periodic pattern, characterized by performing processing and extracting the obtained defect portion candidates together as a defect portion . Based on a predetermined rule in the same plane, the array elements are repeatedly arrayed while gradually changing their array pitch and array direction to detect a defect in the pseudo-periodic pattern of the product forming the pseudo-periodic pattern. A digital signal corresponding to the video signal is obtained by performing A / D conversion processing on the video signal of the pseudo-periodic pattern of the product obtained by the imaging means using a CCD element or the like. After that, for the obtained image data, (a) an array element pixel area as a pixel area occupied by each array element corresponding to a predetermined array pitch and array direction is determined, and the array element pixel area corresponds to the array element pixel area The value of each pixel is added to obtain an integrated value for each array element pixel area, and (b) an array element pixel area for each array element pixel area based on the result of the integration process. A defect for extracting a defect candidate by comparing with a predetermined slice level with respect to a result obtained by an enhancement process such as a differentiation process and (c) an enhancement process step for emphasizing a difference of each integrated value A series of processing consisting of candidate extraction processing is performed a plurality of times, and a defect detection processing step for extracting a defective portion from the obtained extraction result is performed , and the defect detection processing step is performed in advance with a predetermined arrangement pitch, A plurality of array element pixel areas corresponding to the array having a set in the array direction are determined, and each array pixel area is used for the entire digitized image data. From the result obtained by the defect candidate extraction process, the area where the number of defect candidate detections is less than a predetermined number is defined as a defect part detection area corresponding to the array pixel area, and the defect part in the area Weather Extracts a as a defect portion, the combined defect of the defect detection zone, defect inspection method of quasi-periodic pattern, characterized in that to extract as a defect portion in the entire quasi-periodic pattern. 3. The method for inspecting a defect of a quasi-periodic pattern according to claim 1, wherein the product is a shadow mask.

Images