JP2000028333A - Method for detecting pattern defect and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain highly precise defect detection without detecting a unitization error as a defect. SOLUTION: Pattern matching by a 'shaking method' is operated to a master pattern M and an object pattern P being multivalue images having gradation in this pattern defect detecting method. The master pattern M is 'shaken' to the object pattern P by using the width of one pixel or less as a unit. The master pattern M and the object pattern P don't show any satisfactory coincidence in (a), and a large gradation difference is caused in the both patterns as shown in (b). When the master pattern M is shifted from the object pattern P to the right side in a figure by 0.4 picture elements, further satisfactory coincidence is generated in the both pattern as shown in (c), and the gradation difference is reduced in the both patterns as shown in (d).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern defect detecting method and a pattern defect detecting method for detecting a defect by comparing a pattern such as a printed wiring board, an IC mask pattern, and a lead frame with a master pattern for each area of a predetermined size. It relates to a detection device.

[0002]

2. Description of the Related Art FIG. 18 is a view for explaining an outline of processing of a pattern defect inspection by a conventional comparison method.

Conventionally, as a method for inspecting a pattern defect of a printed wiring board or the like, a defect inspection method called a comparison method has been known. In this method, an object pattern that is a pattern to be inspected and a master pattern that is a reference pattern are superimposed and compared. At this time, these object patterns and master patterns are given as multi-valued image data having gradation by first performing A / D conversion of image data obtained by an image pickup device such as a CCD. Then, the object pattern or the master pattern as such multi-valued image data is binarized by a predetermined threshold value to obtain binary image data, and then superimposed on each other. D was detected.

[0004]

By the way, when a multi-valued image is binarized as shown in FIG. 18, the edge portion of the pattern is not shown in the binary image data as shown in FIG. May have a quantization error Q of one pixel or less. However, since these are not originally defects, it is not desirable to detect them as defects. Therefore, conventionally, the quantization error Q is not determined to be a defect by setting the window to have a size of at least one pixel such as 2 × 2 pixels or 3 × 3 pixels as the minimum unit of defect detection.

However, this method has a problem that a defect of about one pixel cannot be detected.

The present invention is intended to overcome the above-mentioned problems in the prior art, and a pattern defect detection method and a pattern defect detection method capable of detecting a defect with high accuracy without detecting a quantization error as a defect. It is an object to provide a detection device.

[0007]

In order to achieve the above object, a method according to a first aspect of the present invention compares a two-dimensional object pattern with a two-dimensional master pattern for each area of a predetermined size. A pattern defect detection method for performing defect detection, comprising two-dimensionally extending an inspection area of an object pattern and an expansion area extending around a required number of pixels around an area of a master pattern to be positionally corresponding to the inspection area. A plurality of master areas that are displaced by a predetermined amount by a unit amount are set, and the multi-valued image signals of the plurality of master areas and the multi-valued image signal of the inspection area are compared for each master area to detect defects. (A) For each of a plurality of master areas, for each pixel corresponding to the master area and the inspection area, the gradation difference of the multi-level image signal is calculated. A step of obtaining a tone difference absolute value that is a pair value, and (b) comparing the tone difference absolute value for each of the plurality of master areas with a predetermined threshold value, and is equal to or greater than the threshold value in all master areas. If the tone difference absolute value is included, the inspection area is determined to be defective. If all tone difference absolute values are less than the threshold value in one or more of the master areas, the inspection area is defective. Determining that there is no data.

According to a second aspect of the present invention, there is provided a pattern defect detecting apparatus for detecting a defect by comparing a two-dimensional object pattern with a two-dimensional master pattern for each area having a predetermined size. Means for inputting a multi-valued image signal of the object pattern, means for storing a multi-valued image signal of the master pattern, a multi-valued image signal in the inspection area of the object pattern, and a master pattern to be positionally associated with the inspection area Extracting means for respectively extracting multi-level image signals in a plurality of master areas which are two-dimensionally displaced by a predetermined amount by a unit amount over an extended area extended around the required number of pixels around the area of Inspection means for inputting a multi-level image signal in the area and a multi-level image signal in the inspection area, wherein the inspection means comprises a plurality of master areas; For each of the pixels corresponding to the master area and the inspection area, the absolute value of the gray level difference, which is the absolute value of the gray level difference of the multi-level image signal, is obtained, and the floor for each of the plurality of master areas is determined. The absolute value of the difference is compared with a predetermined threshold value. If the absolute value of the tone difference is equal to or greater than the threshold value in all the master areas, it is determined that the inspection area is defective, and one or more master areas are defective. When all the tone difference absolute values are less than the threshold value in the area, it is determined that there is no defect in the inspection area.

According to a third aspect of the present invention, there is provided a pattern defect detecting apparatus according to the second aspect, wherein:
The unit amount is smaller than the size of one pixel, and the extracting means interpolates the multi-valued image signal of the master pattern to obtain a multi-valued image signal of a plurality of master areas displaced by a predetermined amount by the unit amount. It is characterized by the following.

According to a fourth aspect of the present invention, there is provided the pattern defect detecting apparatus according to the third aspect, wherein:
The interpolation is performed in each direction of the two-dimensional orthogonal axis.

According to a fifth aspect of the present invention, there is provided the pattern defect detecting apparatus according to any one of the second to fourth aspects, wherein the inspection means comprises a multi-level image signal in the inspection area. And maximum difference holding means for obtaining and holding the maximum value among the tone difference absolute values in all the pixels corresponding to the multi-level image signals of each of the plurality of master areas for each of the plurality of master areas, Minimum value selecting means for selecting the minimum value among the maximum values for the plurality of master areas; determining that there is no defect when the minimum value is less than the threshold value; And determination means for performing the determination of (1).

According to a sixth aspect of the present invention, there is provided the pattern defect detecting apparatus according to any one of the second to fourth aspects, wherein the inspection means comprises a multi-valued image signal in the inspection area. And a gradation difference calculating means for calculating a gradation difference absolute value of a pixel corresponding to each of the multi-valued image signals of the plurality of master areas, and for each of the plurality of master areas, all the gradation difference absolute values are A match determination unit that determines that the master area is a pattern match when the difference is less than the threshold, and determines that the master area is a pattern mismatch when at least one tone difference absolute value is equal to or greater than the threshold; A comparison result holding means for holding a comparison result of each of the plurality of master areas by the match determination means, and a defect in the inspection area when all the comparison results held in the comparison result holding means do not match the pattern. A judgment Rino, characterized in that the examination zone if at least one comparison result pattern matching in which and a result determining means for determining no defects.

[0013]

Embodiments of the present invention will be described below with reference to the drawings.

<1. Principle of Embodiment> FIG. 1 is a diagram schematically illustrating a “wobble method” in a pattern defect detection method according to an embodiment of the present invention. In this embodiment, the inspection area P0 of the object pattern is centered on the area M0 of the master pattern corresponding to the center, and the area (area) is expanded two-dimensionally into an expansion area extending around the required number of pixels. A plurality of master areas, each of which is displaced by a predetermined amount for every / 5 (0.2) pixel, are set, and the master areas M11 to MBB are suffixed in hexadecimal).
The defect is detected by comparing the inspection area P0 with the inspection area P0.
In this case, the setting range of the extension area is a range that can absorb a displacement error due to a displacement or distortion of the inspection object,
FIG. 1 exemplifies a case where the image is displaced by one pixel vertically, horizontally, and horizontally around an area M0 (= master area M66).

However, the multi-level image signal of the original master pattern is in pixel units. On the other hand, this embodiment is different from the first embodiment in that a multi-valued image signal deviated (oscillated) by less than one pixel is generated by interpolation.

FIG. 2 is a diagram for explaining the concept of a multi-valued image signal deviated in units of less than one pixel in this embodiment. As shown in FIG. 2, this embodiment will exemplify generation of an image of a master pattern in a state of being displaced by 0.2 pixels.

In this embodiment, in order to obtain a deviated image signal in units of less than one pixel, an image signal deviated in such a manner is obtained by using linear interpolation based on a gradation signal of an adjacent pixel. ing.

Hereinafter, an image signal (gradation value) at the pixel position i of the base pattern of the base is represented by Gi, and an image signal (gradation value) corresponding to the pixel i deviated in units of less than one pixel.
Is denoted by Hi. At this time, the formula for obtaining Hi is as follows.

Hi = INT {Gi−1 · (1-x) + Gi · x} (1) where x = 0.2, 0.4, 0.6, 0.8 For example, the floor of adjacent pixels When the tone values are Gi-1 = 100 and Gi = 60 and x = 0.2, Hi = 92 from the above equation.

In this embodiment, an image signal of a master pattern in which x = 0 or 1 is required to perform the swinging of ± 1 pixel. No interpolation is necessary since the image signal can be used as it is. Equation (1) is an equation for one-dimensional displacement as a pixel position, but this is performed two-dimensionally. That is, an interpolated image is generated using the equation (1) for each of the main scanning direction and the sub-scanning direction.

By setting a plurality of master areas M11 to MBB deviated in units of less than one pixel in this manner, even if the inspection object is displaced by less than one pixel, the inspection area P0 is not positioned. A corresponding master pattern will be present in any of the master areas M11 to MBB. In the example of FIG. 1A, it is assumed that the object pattern P corresponding to the inspection area is shifted by 0.8 pixels upward and 0.8 pixels leftward from the reference position. The master pattern that exactly matches the position of the area P0 is not the master area M66 that should originally match at this position, but the master pattern that is 0.8 pixels downward and 0.8 pixels to the right from the master area M66. Zone M
AA.

After setting such a master area group, an image signal in the inspection area P0 shown in FIG. 1A and an image symbol in each of the master areas M11 to MBB shown in FIG. And are respectively compared. In other words, defect inspection windows W and W 'having a predetermined pixel size, in this embodiment, one pixel size, are set in each area, and the inspection windows W and W' are set to correspond to each other in the entire area while being made to correspond to each other. Scanning is performed to obtain the absolute value of the gradation difference of each pixel in the defect inspection windows W and W ′ of the master pattern and the object pattern.

FIG. 3 is a diagram showing an example in which a comparison is made by the "shake method" using a displaced image signal in units of less than one pixel. In FIGS. 3 (a) to 3 (d), the horizontal axis represents the pixel. 3 (a) and 3 (c), the vertical axis represents the gradation of the image signal. In FIGS. 3 (b) and 3 (d), the vertical axis represents the image signal of FIGS. 3 (a) and 3 (c), respectively. Is the absolute value of the gradation difference. However, the pixel positions are actually distributed two-dimensionally, but are represented here as one-dimensional for simplicity of explanation.

FIG. 3A shows a case where a displacement of less than one pixel occurs between the object pattern P and the master pattern M, more precisely, a case where a displacement of about 0.4 pixel occurs. This shows a state in which both are superimposed. Here, since the defect inspection windows W and W 'have a size of one pixel, the floors of both patterns are moved while moving the windows W and W' along the horizontal axis representing the pixel position in FIG. When the absolute value of the difference is obtained, the result is as shown in FIG. As can be seen from FIG. 3B, there are many gradation differences between both patterns even though the object pattern has no defect. Then, as is clear from FIG. 3A, the object pattern P and the master pattern M are
Even if both patterns are relatively displaced in pixel units,
In this case, the state in FIG. 3A is the state in which the two patterns match the most, so that the gradation difference does not become smaller than in FIG. 3B.

For this reason, in this embodiment, by shaking in units of less than one pixel by interpolation,
As shown below, this problem has been overcome.

FIG. 3C shows a case where the master pattern is interpolated to the right in the figure by about 0.4 pixel with respect to the positional relationship between the two patterns in FIG. It shows how it was. As can be seen from FIG. 3D, which shows the absolute value of the difference between the two patterns in FIG. 3C, the difference between the two patterns is smaller than that in FIG. 3B. It can be seen from the positional relationship between the object pattern P and the master pattern M that the two patterns match.

In this embodiment, the master area M having a width of less than one pixel is shaken as described above.
The absolute value of the tone difference, which is the absolute value of the tone difference between each of the pixels 11 to MBB and the inspection area P0, is obtained for all the pixels in all the positional relationships (master areas M11 to MBB). If the master area M11 to MBB includes a gradation difference absolute value that is equal to or greater than a predetermined threshold (hereinafter, referred to as a "determination threshold"), it is determined that the inspection area P0 has a defect, and one or more masters If the absolute values of all the gradation differences are less than the determination threshold in the sections M11 to MBB, it is determined that there is no defect in the inspection area P0.

However, as a specific method of this determination, in the first and second embodiments, the master areas M11 to MBB are used.
, The maximum absolute value of the gray level difference (hereinafter referred to as “maximum gray level difference”) among all the pixels in the area is determined. If the minimum value is smaller than the determination threshold value, it is determined that there is no defect.

The fact that the minimum maximum gradation difference is equal to or larger than the judgment threshold value means that all the master areas include the gradation difference absolute value which is equal to or larger than the judgment threshold value.
When the minimum maximum gradation difference is smaller than the judgment value, it means that all the gradation difference absolute values are smaller than the judgment threshold value in at least one master area.

In the third and fourth embodiments, the absolute value of the gradation difference is compared with the judgment threshold value for all the pixels in each of the master areas M11 to MBB. For example, if the result of pattern matching is held in that area, and if at least one pixel is equal to or greater than the determination threshold, the result of pattern mismatch is held in that area, and the pattern does not match for all master areas M11 to MBB In the inspection area P0, it is determined that there is a defect in the inspection area P0, and if the pattern matches at least one of the master areas M11 to MBB, it is determined that there is no defect in the inspection area P0.

According to the method of the present embodiment described above, since the comparison inspection is performed based on the multi-valued image signals of the object pattern and the master pattern, no quantization error occurs and a small defect of one pixel unit is obtained. , And highly accurate defect detection can be performed.

Hereinafter, a description will be given of a pattern defect detection apparatus according to some embodiments using such a method and pattern defect detection using the apparatus.

<2. First Embodiment >><< 2-1. Device Configuration >> A pattern defect inspection device having a circuit configuration configured to perform pattern defect detection corresponding to the example of FIG. 1 will be described below.

FIG. 4 is a diagram for explaining an inspection area in the object pattern and the master pattern according to the first embodiment. In this detection device, as shown in FIG. 4, an inspection area A1 having a size of 508.times.508 pixels is included in an object pattern and a master pattern.
., A′11, A′12... Are set, and defect detection is performed for each inspection area. Each inspection area corresponds to the inspection area P in the object pattern.
0, which corresponds to the area M0 (master area M66) in the master pattern.

FIG. 5 shows a schematic configuration diagram of a pattern defect detection device. As shown in FIG. 1, the apparatus has an XY table 5 capable of holding an inspection object 4 such as a printed wiring board with a positioning accuracy within a predetermined range, and the XY table 5 in the X direction (main scanning direction). ) Is performed by the X-direction drive unit including the motor 6, and the drive in the Y direction (sub-scanning direction) is performed by the Y-direction drive unit including the motor 7.

Above the XY table 5, eight CCD line sensors 8 are installed at predetermined intervals on the X direction line. (However, in FIG. 5, only one device is shown for the sake of convenience, and the others are not shown.) Each of these CCD line sensors 8 has 2048 light-receiving elements constituting pixels arranged in a line on the X-direction line. Is used. The reading of the image of the printed wiring board 4 by these eight CCD line sensors 8 is performed by moving the XY table 5 in the Y direction by the motor 7 while scanning the CCD line sensor 8 in the X direction. Scanning is performed in a band shape with a scanning width a, and scanning on the outward path side is performed. When the forward scan is completed, the XY table 5 is moved by the motor 6 in the X direction by a width slightly smaller than the scan width a so as to overlap the inspection area, and then the XY table 5 is moved to-(minus). ) Move in the Y direction to perform scanning on the return side. As a result, the entire area of the printed wiring board 4 is scanned, and an image is read. FIG.
(a) is an object pattern P obtained by capturing image data on the outward path by one CCD line sensor 8.
It is shown.

The analog image signal read by the CCD line sensor 8 is amplified by a buffer amplifier 9 and then converted into a digital image signal by an A / D converter 10.

A switching circuit 13 is connected to the output terminal side of the A / D conversion unit 10. The switching circuit 13 is connected to the contact 13a when a master pattern is input and to the contact 13b when an object pattern is input. Each is switched.

That is, when the master pattern is input, in other words, when the printed wiring board 4 serving as the inspection reference is arranged on the XY table 5 and the image is read by the CCD line sensor 8, the A / D The multi-level image signal output in time series from the conversion unit 10 is stored in the memory 15 through the contact 13a. On the other hand, when an object pattern is input, in other words, when the printed wiring board 4 to be inspected is arranged on the XY table 5 and its image is read by the CCD line sensor 8, A
A multi-level image signal output in time series from the / D conversion unit 10 is input to one input terminal of the defect inspection unit 16 through the contact 13b.

The defect inspection unit 16 (details of which will be described later) performs a defect inspection process based on the image signal, that is, object data, and master data called from the memory 15 at the same time. In addition,
The defect inspection unit 16 includes a delay unit 16a and a comparison unit 16b.

Further, a bus line BL is provided in the pattern defect detection device, and the bus lines BL are connected to the motors 6 and 7, the A / D conversion unit 10, the switching circuit 13, and the comparison unit in the defect inspection unit 16. 16b, along with the CPU 17,
The monitor 18 and the operation unit 19 are electrically connected.

CPU 17 includes motors 6 and 7, A / D converter 10, switching circuit 13, and comparison unit 1 in defect inspection unit 16.
6b is controlled while synchronizing.

The monitor 18 displays a setting input by an operator, a defect inspection result, and the like.

The operation unit 19 is an input means such as a keyboard and a mouse on which the operator who confirms the display on the monitor 18 performs various settings and instruction inputs.

FIG. 6 is a diagram showing a schematic circuit configuration of the comparison section 16b of the defect inspection section 16 of the pattern defect inspection apparatus. The multi-valued image data (object data OD) of the inspection area P0 in the object pattern obtained by scanning the inspection object is input to the defect inspection unit 16 in time series. At the same time, the multi-valued image data (master data MD) of area M0 in the master pattern called from memory 15 in synchronization with the input of object data OD.
a) is input in chronological order.

The input object data OD is input to each of the comparison detection blocks B11 to B33. The comparison detection blocks B11, B21, B31, B12, B22, B32, B1
3, B23 and B33 respectively correspond to the master area M11 of FIG.
Of the master areas M11, M16, M1B, M61, M66, M6B, MB1, MB6, M shifted from each other by an integer pixel
It is provided corresponding to BB.

On the other hand, the master data MDa input to the defect inspection unit 16 is stored in the shift register S in the delay unit 16a.
The master data MDa, MDb, and MDc are shifted by one line at a time for each line in the horizontal direction of the comparison detection blocks B11 to B33.
And a delay circuit (not shown) inserted for each stage in the vertical direction delays by one pixel for each stage in the vertical direction. As a result, the master data given to each of the comparison detection blocks Bij (i = 1 to 3, j = 1 to 3) are all two-dimensionally displaced by one pixel.

FIG. 7 shows the comparison detection block Bij (i =
FIG. 3 is a diagram illustrating a specific circuit configuration of (1-3, j = 1-3). Since each of the comparison detection blocks Bij has the same configuration, only the circuit of one block is shown here.

The comparison detection block Bij includes an image interpolation unit Ca
And a detailed comparison block Cij (i = 1 to 5, j = 1 to 5).

For the input master data MD,
In the image interpolation unit Ca, x = 0, 0.2, 0.4,.
A total of 25 types of interpolated image signals deviated by 6, 0.8 pixels are generated and transmitted to the detailed comparison block Cij. Note that among these, the image shifted by 0 pixel in the main scanning and sub-scanning directions outputs the input master data MD as it is.

FIG. 8 is a diagram showing a specific circuit configuration of the detailed comparison block Cij.

When the object data OD and the master data are input to the detailed comparison block Cij by the MD, the gradation difference calculation circuit Da calculates the absolute value of the difference between the gradations and sends it to the selection output circuit Db.

In the selection output circuit Db, before the absolute value of the gradation difference described later is input, the maximum value of the absolute value of the gradation difference stored in the maximum gradation difference holding circuit Dc is read in advance. Are compared. Then, only those larger than the predetermined value are output.

The maximum gradation difference holding circuit Dc is a selection output circuit D
The values output from b are sequentially held as maximum values,
The finally obtained maximum gradation difference, which is the maximum absolute value of the gradation difference in the detailed comparison block Cij, is output.

In this way, the maximum gradation difference output by each detailed comparison block Cij shown in FIG. 7 is collected by the minimum difference selection circuit Cb, and the minimum difference selection circuit Cb outputs the minimum value of the minimum maximum value. Obtain and output a difference candidate.

In this way, the minimum / maximum gradation difference candidates output from each comparison / detection block Bij shown in FIG. 6 are collected by the minimum difference determination circuit Ba, and the minimum value among the minimum / maximum gradation difference candidates is calculated. Is obtained and further transmitted to the defect determination circuit Bb.

The defect judgment circuit Bb compares the minimum / maximum gradation difference with the judgment threshold value. If the difference is smaller than the judgment threshold value, it is judged that there is no defect in the inspection area P0. If so, it is determined that there is a defect, and any of those results is output as an inspection result signal.

<< 2-2. Processing Procedure >> Next, the procedure of the defect detection processing by the above-described pattern defect detection apparatus will be described with reference to FIG. 9 which is a flowchart showing the procedure.

First, the capture of an image serving as a master pattern is started (step S1). Hereinafter, an image is read by the CCD line sensor 8, and the obtained analog image signal is converted into a digital image signal by the A / D converter 10.

Next, it is determined whether or not the mode is the master pattern input mode (step S2). If the mode is the master pattern input mode, the flow advances to step S3 to store the input master data in the memory 15 (step S2).
3). Then, when the input of the object pattern is completed, a comparative inspection is performed (step S4).

FIG. 10 is a flowchart for explaining in detail the procedure of the comparative inspection process in step S4 of FIG. Less than,
The processing in step S4 will be described in more detail with reference to FIG.

First, the image is divided into blocks of a predetermined size (FIG. 10: step S41).

Next, it is determined whether the following processing has been completed for all blocks (FIG. 12: step S4).
2) If the processing has been completed, the comparison inspection processing in step S4 is completed; if the following processing has not been completed, step S43 is performed.
Proceed to.

Next, the object data of the inspection area P0 of the object pattern corresponding to the designated block and the master data in consideration of the amount of deviation are cut out and read (FIG. 10: step S43).

Next, a shaking process is performed, and an inspection (comparison) is performed in all the positional relationships shown in FIG. 1 (FIG. 1).
0: Step S44).

FIG. 11 is a flowchart for explaining the swinging process of step S44 in FIG. 10 in detail.

First, the master area M deviated by interpolation
Obtain master data of 11 to MBB (FIG. 11: Step S
441).

In the first embodiment, interpolation is performed on the master pattern. FIG. 12 is a diagram for explaining how a defect detection error occurs when interpolation is performed on an object pattern.

Consider a case where a defect inspection of an object pattern having a defect d0 as shown in FIG. 12B is performed on a master pattern as shown in FIG. That is, the object pattern on the inspection object often has a minute defect of about one pixel. FIG. 12 (c) shows such an object pattern that is deviated in units of less than one pixel by the interpolation as described above. FIG. 12C shows 0.2, 0.6,.
8 shows an object pattern deviated by eight images. As shown in FIG. 12B, in FIG. 12B, the defect d0 is larger than the determination threshold value. Therefore, the inspection result indicating that there is a defect should normally be obtained. However, in each interpolation image of FIG. As a result, none of the defects d1, d2, and d3 exceed the determination threshold, so that they are not recognized as defects, and the inspection result is "no defect".

As described above, when the object data is interpolated and shifted, the possibility of overlooking a defect of about one pixel increases. On the other hand, since there are few patterns that change in units of one pixel in the master pattern, it is unlikely that the above-mentioned defect is overlooked if interpolation is performed on the master pattern. Therefore, in the apparatus according to the first embodiment, a deviated image is obtained by interpolating the master pattern. As a result, it is possible to suppress the occurrence of oversight of defects as described above, and to perform defect detection with higher accuracy.

Next, it is determined whether or not the inspection has been completed for the master data of all the master areas M11 to MBB (FIG. 11: step S442). If the inspection has been completed, the maximum floor obtained from all the inspection results has been obtained. A minimum / maximum gradation difference which is a minimum value among the differences is obtained (FIG. 11: step S44).
3) Conversely, if not completed, the next master area M11
.About.MBB and the object data of the inspection area P0 are compared (FIG. 11: step S444).

Here, the comparison in step S444 is to calculate the absolute value of the gradation difference for each pixel, and the maximum gradation difference is saved (FIG. 11: step S44).
7).

Returning to FIG. 10, next, at step S443
It is determined whether the minimum / maximum gradation difference obtained in (FIG. 11) is equal to or greater than a determination threshold (FIG. 10: Step S4).
5) If YES, the inspection result of the designated block is determined to be "defective" (FIG. 10: step S46), and if NO, the inspection result is determined to be "no defect" (FIG. 10: step S47).

When the above steps have been performed for all the blocks, the processing in step S4 ends.

Returning to FIG. 9, it is next determined whether the inspection result is "defective" or "no defect" (step S5), and the inspection result is determined to be "no defect" (no defect in all blocks). If so, the process returns to step S1, and if it is determined that there is a defect (there is a defect in any block), the process proceeds to step S6.

Next, an image of the defective portion is displayed on the monitor 18 according to the inspection result, and the operator confirms it (step S6). Then, the operator confirms whether the defect is a defect or the like and takes action.

Next, it is determined whether the inspection has been completed for all the inspected objects (object patterns) (step S7). If the inspection has not been completed, the process returns to step S1, and if completed, the defect inspection is performed. finish.

As described above, according to the first embodiment, binarization of an image signal is performed in order to detect a defect by comparing the multi-value image signals of the object pattern and the master pattern. Since there is no need to include a quantization error, it is possible to detect a defect of about one pixel without detecting the quantization error as a defect, so that highly accurate defect detection can be performed. In addition, since the defect is detected by the "shake method", it is possible to detect the defect with high accuracy without performing accurate positioning of the inspection object.

Further, since the "shake method" having a width smaller than the size of one pixel can be performed by interpolating the multi-valued image data of the master pattern, it is possible to achieve higher precision without performing accurate positioning. Defect detection can be performed.

Further, since the inspection is performed using the master pattern interpolated in each direction of the two-dimensional orthogonal axis, even if there is a two-dimensional displacement, accurate positioning can be performed without performing accurate positioning. Defects can be detected.

Further, since interpolation is performed for the master pattern, compared with the case where the object pattern is interpolated,
Even when a defect of about one pixel occurs in the object pattern, erroneous detection can be suppressed and highly accurate defect detection can be performed.

Furthermore, in each of the master areas M11 to MBB, the inspection accuracy can be easily changed by changing the setting of the judgment threshold value in order to maintain the maximum gradation difference. can do.

<3. Second Embodiment> The pattern defect inspection apparatus according to the first embodiment uses a window of one pixel size, but the pattern defect inspection apparatus according to the second embodiment uses a window of 2 × 2 pixels. Is used. Therefore, in the apparatus according to the second embodiment, the detailed comparison block Cij of each comparison detection block Bij of the comparison unit 16b in the defect inspection unit 16 has a different configuration. Further, the object data OD is directly used as the object data OD1 in the delay unit 16a of the defect inspection unit 16 in FIG. 5, and is delayed by one line by a delay circuit (not shown) and input to the comparison unit 16b as the object data OD2. Configuration. The other configuration is the same as that of the device of the first embodiment.

FIG. 13 is a diagram showing a specific circuit configuration of the detailed comparison block Cij in the second embodiment. Hereinafter, pattern defect detection by the apparatus according to the second embodiment will be described with reference to FIGS. 13, 6 and 7.

When the object data OD1 and the master data MD1 are input to the detailed comparison block Cij, the tone difference calculating circuit Da1 calculates the absolute value of the tone difference,
The value is transmitted to the addition circuit De, and the value is latched by the latch circuit Dd1.

When the next object data OD1 and master data MD1 are input to the gradation difference calculation circuit Da1, the absolute value of the gradation difference between the two data is calculated and transmitted to the addition circuit De. Dd1 is latched. At this time, the data of the previous pixel that has been latched by the latch circuit Dd1 is output to the addition circuit De. As a result, the absolute value of the gradation difference between the object data OD1 for two pixels continuous in the main scanning direction and the master data MD1 has been input to the addition circuit De.

Similarly, the object data OD2 and the master data M1 are delayed by one line with respect to the object data OD1 and the master data MD1, respectively.
D2 are sequentially input to the gradation difference calculation circuit Da2, and the absolute values of those gradation differences are transmitted to the addition circuit De.
The absolute value of the gradation difference between the latched object data OD2 and master data MD2 is transmitted to the addition circuit De.

As described above, the absolute value of the gradation difference between the two data corresponding to the window of 2 × 2 pixels is input to the addition circuit De, and the addition circuit De calculates the sum of these and outputs the sum to the selection output circuit Db. Send.

When the sum of the absolute values of the gradation differences is input to the selection output circuit Db, the maximum value of the sum of the absolute values of the gradation differences stored in the maximum gradation difference holding circuit Dc is read out. Is compared. Then, the smaller data is sent to the maximum gradation difference holding circuit Dc. In this way, the maximum gradation difference holding circuit Dc sequentially holds the values output from the selection output circuit Db as the maximum value, and finally obtains the level in each window in the finally obtained detailed comparison block Cij. The maximum gradation difference, which is the maximum value of the sum of the absolute values of the differences, is output.

The shift of the data in the shift register to the right as described above means that the inspection windows W and W ′ shown in FIG. 1 are scanned to the right. The maximum gradation difference output at the end of scanning of the entire area of the input pattern is the maximum gradation difference obtained in the positional relationship between the corresponding object pattern and the master pattern in FIG. The output from the block Cij is collected by the minimum difference selection circuit Cb (see FIG. 9), where the minimum and maximum gradation difference candidates are output, and further output from all the comparison detection blocks Bij (see FIG. 6). Are collected by the minimum difference determination circuit Ba, and the minimum one of the minimum and maximum gradation differences is obtained therefrom. Then, the obtained minimum and maximum gradation difference is transmitted to the defect determination circuit Bb,
Then, the defect is detected and compared with the determination threshold value in the same manner as in the first embodiment. Thus, the comparison processing of the inspection area P0 and the master areas M11 to MBB shown in FIG. 1 has been performed. Then, such a determination is made for all blocks of all the images to be inspected, thereby performing a defect inspection of the entire object to be inspected.

With this configuration, this apparatus performs defect detection with a window size of 2 × 2 pixels.
By the way, in one pixel unit such as the defects d1 to d3 in FIG. 12C, a defect having a plurality of pixel sizes whose gradation difference does not exceed the threshold value may exist in the object pattern. This is likely to occur when the center of a defect is located at a pixel boundary when an analog image signal is digitized. In such a case, the device of the first embodiment has a window of one pixel, so that such defects d1 to d3 cannot be detected. However, in the second embodiment, the window size is 2 pixels. × 2 pixels, and the sum of the absolute value of the difference between the gradation values of the object data and the master data in each pixel in the window is calculated, and the sum is compared with the judgment threshold to detect a defect. Such defects d1 to d3 can also be detected.

In addition to the above effects, according to the second embodiment, it is possible to obtain effects other than the effect that "a defect of about one pixel can be detected" in the first embodiment. it can.

<4. Third Embodiment> The apparatus according to the first embodiment obtains the maximum value of the absolute value of the gradation difference in each block Cij, and compares the minimum value, the minimum maximum gradation difference, with the judgment threshold value. However, in the third embodiment, the results obtained by comparing the absolute value of the gradation difference with the judgment threshold value in each comparison detection block Bij are held, A defect detection result is obtained by collecting and performing a logical product operation. Therefore, in the apparatus according to the third embodiment, the comparison unit 1 in the defect inspection unit 16
6b is different. The other configuration is the same as that of the device of the first embodiment, and the window is also one pixel size. Hereinafter, the configuration of the comparison unit 16b in the device according to the third embodiment and the defect detection processing using the same will be described.

FIG. 14 is a diagram showing a schematic circuit configuration of the comparison unit 16b of the defect inspection unit 16 of the pattern defect inspection apparatus according to the third embodiment, and FIG. 15 is a diagram showing the comparison unit 16b of FIG.
FIG. 16 is a diagram showing a specific circuit configuration of the comparison detection block Bij (i = 1 to 3, j = 1 to 3) of FIG.
5 is a diagram showing a specific circuit configuration of a detailed comparison block Cij in FIG.

As shown in FIG. 16, the detailed comparison block Cij
Calculates the absolute value of the gradation difference between the input master data MD and object data OD, and transmits the absolute value to the determination circuit Df.

The judgment circuit Df compares the transmitted signal with the judgment threshold value. If the signal is less than the judgment threshold value, it is determined that the object data OD and the master data MD match, and "0" is judged. If so, it is determined that the two data do not match, and "1" is transmitted to the OR gate Dg as a result signal.

The OR gate Dg receives the result signal from the decision circuit Df and the result signal held in the result holding circuit Dh, and performs a logical OR operation on the result signals. Dh is held.
Once the result signal of “1” is input, the result holding circuit Dh holds the state until the inspection (comparison) of the block to be inspected is completed.

In FIG. 15, the result signals obtained from the detailed comparison blocks Cij are collected by an AND gate Cc to perform an AND operation, and the result signal is output to an AND gate B.
c (FIG. 14).

In FIG. 14, the result signals output from the respective comparison detection blocks Bij are collected by an AND gate Bc and subjected to an AND operation, and the result is output as a final inspection result signal.

As a result, the AND operation of the result signal obtained by the comparison processing between the inspection area P0 shown in FIG. 1 and all of the master pattern areas M11 to MBB is performed. If it is determined that both patterns match (result signal "0") by comparison with any one of M11 to MBB, the inspection result signal is also "0", and it is determined that there is no defect in the inspection area P0. Conversely, if both patterns are determined to be inconsistent (result signal "1") in all comparisons of both areas, the inspection result signal is also "1", and it is determined that the inspection area P0 has a defect. Will be. Then, such a determination is made for all the blocks of all the images to be inspected, thereby performing the defect inspection of the entire object to be inspected.

As described above, according to the third embodiment, it is possible to obtain an effect other than the effect that “the device can be maintained with good maintenance” in the first embodiment. In addition, since only the one-bit result signal is held in each of the master areas M11 to MBB, the capacity of the holding circuit can be reduced.

<5. Fourth Embodiment> The pattern defect inspection apparatus according to the third embodiment uses a window of one pixel size, and each detailed comparison block Cij outputs only a result signal. In the pattern defect inspection apparatus of the form (1), a window having 2 × 2 pixels is used, and each detailed comparison block Cij outputs only a result signal. Therefore, in the apparatus according to the fourth embodiment, the configuration of each detailed comparison block Cij of each comparison detection block Bij of the comparison unit 16b in the delay unit 16a of the defect inspection unit 16 is different. The other configuration is the same as that of the device according to the third embodiment.

FIG. 17 is a diagram showing a specific circuit configuration of the detailed comparison block Cij in the fourth embodiment. Hereinafter, FIG. 17 (and FIG. 14 and FIG. 15)
The circuit configuration of the device according to the embodiment and the pattern defect detection by the device will be described.

When the object data OD1 and the master data MD1 are input to the detailed comparison block Cij, the tone difference calculating circuit Da1 calculates the absolute value of the tone difference,
This is transmitted to the judgment circuit Df1.

The decision circuit Df1 compares the obtained absolute value of the gradation difference with the decision threshold value. "1" is latched by the latch circuit Dd1 as a result signal, and the result signal is directly transmitted to the AND gate Dk.

The result signal obtained by comparing the next object data OD1 with the master data MD1 is a latch circuit D.
When it is input to d1, a result signal based on both data is similarly obtained and transmitted to the AND gate Dk, and is also latched by the latch circuit Dd1. At this time, the result signal of the previous pixel that has been latched by the latch circuit Dd1 is output to the AND gate Dk. As a result, object data OD1 for two pixels continuous in the main scanning direction
And the master data MD1 are compared, a signal is input to the AND gate Dk.

Similarly, the object data OD2 and the master data M1 are delayed by one line with respect to the object data OD1 and the master data MD1, respectively.
D2 are sequentially input to the gradation difference calculation circuit Da2, and the absolute values of those gradation differences are transmitted to the determination circuit Df2. A part of the result signal similar to the above obtained by the determination circuit Df2 is directly transmitted to the AND gate Dk, and the others are latched by the latch circuit Dd2, and are also latched by the latch circuit Dd2. The result signal of the previous pixel is sent to AND gate Dk.

As described above, the result signal for each pixel of both data corresponding to the window of 2 × 2 pixels is AN
The data is input to the D gate Dk and subjected to an AND operation, and a result signal representing the operation result is output to the holding circuit Dj. Once the result signal of “1” is input, the holding circuit Dj holds the state until the inspection of the block to be inspected is completed.

As described above, a result signal is output while the inspection windows W and W ′ shown in FIG. 1 are scanned rightward. Therefore, the result signal output at the end of scanning of the entire pattern input to the detailed comparison block Cij is the result signal obtained in the positional relationship between the corresponding object pattern and the master pattern in FIG. All detailed comparison blocks C
ij, and are collected by an AND gate Cc (see FIG. 15) to perform an AND operation. Further, they are output from all the comparison detection blocks Bij (see FIG. 14) and collected by the AND gate Bc. Then, an AND operation is further performed to output a final inspection result signal.

As a result, the comparison processing of the inspection area P0 shown in FIG. 1 with all the master areas M11 to MBB has been performed. Then, such a determination is made for all blocks of all the images to be inspected, thereby performing a defect inspection of the entire object to be inspected.

As described above, according to the fourth embodiment, it is possible to obtain an effect other than the effect that “the device can be easily maintained” in the second embodiment.

<6. Modifications> In the first to fourth embodiments, an example of the pattern defect detection device and the pattern defect detection processing by the pattern defect detection device has been described, but the present invention is not limited to this.

In the first and third embodiments, the window has a size of one pixel. In the second and fourth embodiments, the window has a size of 2 × 2 pixels. It may be three or more pixels in size,
Further, the shape need not be a square like 2 × 3 pixels. In addition, assuming that both of the circuit configurations of the first and second embodiments are provided, the window may be switched between a one-pixel window and a 2 × 2 pixel window through the operation unit 19.

In the first to fourth embodiments, the pattern defect detection processing is performed by the circuit in the defect inspection unit 16. However, the processing may be performed by software by providing a CPU, a memory, and the like. .

In the first to fourth embodiments, 0.
The comparison test by the “shake method” is performed by using the master areas M11 to MBB deviated in units of two pixels.
It may be a unit of fluctuation from 0 or more pixels such as a 0.1 pixel unit.

Further, in the first to fourth embodiments, the interpolation of the master pattern is performed by linear interpolation. However, any interpolation method such as polynomial interpolation and spline interpolation can be used.

[0117]

As described above, according to the present invention, the defect detection is performed by comparing the multi-valued image signals of the object pattern and the master pattern. , Image signal 2
Since it is not necessary to perform the value conversion, and it is possible to detect a defect of about one pixel without detecting a quantization error as a defect, highly accurate defect detection can be performed.

The multi-valued image signals of a plurality of master areas two-dimensionally displaced by a predetermined amount per unit amount over the extended area of the master pattern and the multi-valued image signals of the inspection area of the object pattern are divided into a plurality of pieces. Since the defect detection is performed by comparing each of the master areas, highly accurate defect detection can be performed without performing accurate alignment of the inspection object.

According to the third aspect of the present invention,
By interpolating the multi-valued image signal of the master pattern, multi-valued image signals of a plurality of master areas displaced by a predetermined amount by a unit amount smaller than the size of one pixel are obtained, and the inspection areas of the object area and those are obtained. Since the comparison is made with the multi-valued image signal in the above, even more accurate defect detection can be performed without performing accurate positioning.

[0120] In particular, according to the invention of claim 4,
Inspection is performed using multi-valued image signals of a plurality of master pattern areas obtained by performing interpolation in each direction of a two-dimensional orthogonal axis. Therefore, even when there is a two-dimensional displacement, accurate positioning is performed. The defect can be reliably detected without performing.

According to the fifth aspect of the present invention,
The maximum value among the absolute values of the tone difference is obtained and held for each of the plurality of master areas, the minimum value among the maximum values is selected, and the minimum value is compared with the threshold value to determine the defect. Since the detection is performed, the inspection accuracy can be easily changed by changing the setting of the threshold value, so that an apparatus having good maintainability can be provided.

According to the sixth aspect of the present invention,
Since the comparison result for each of the plurality of master areas is held and the defect is detected based on the held comparison result, the storage capacity can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram schematically illustrating a “wobble method” of pattern defect detection according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining the concept of a multilevel image signal deviated in units of less than one pixel according to the embodiment.

FIG. 3 is a diagram illustrating an example in which pattern matching is performed using a deviation image signal in units of less than one pixel.

FIG. 4 is a diagram illustrating inspection areas in an object pattern and a master pattern according to the first embodiment.

FIG. 5 is a diagram illustrating a schematic configuration of a pattern defect detection device according to the first embodiment.

FIG. 6 is a diagram illustrating a schematic circuit configuration of a comparison unit in the pattern defect detection device according to the first embodiment.

FIG. 7 is a specific circuit diagram of a comparison detection block according to the first embodiment.

FIG. 8 is a diagram illustrating a specific circuit configuration of a detailed comparison block according to the first embodiment.

FIG. 9 is a flowchart illustrating a procedure of a defect detection process according to the first embodiment.

FIG. 10 is a flowchart illustrating the comparison inspection processing procedure of FIG. 9 in detail;

11 is a flowchart illustrating the pattern matching processing of FIG. 10 in detail.

FIG. 12 is a diagram for explaining occurrence of a detection error when an object pattern is interpolated.

FIG. 13 is a diagram illustrating a specific circuit configuration of a detailed comparison block according to the second embodiment.

FIG. 14 is a diagram illustrating a schematic circuit configuration of a comparison unit of a defect inspection unit of the pattern defect inspection device according to the third embodiment.

15 is a diagram illustrating a specific circuit configuration of a comparison detection block of the comparison unit in FIG. 14;

16 is a diagram showing a specific circuit configuration of a detailed comparison block in FIG.

FIG. 17 is a diagram illustrating a specific circuit configuration of a detailed comparison block according to the fourth embodiment.

FIG. 18 is a diagram for explaining an outline of processing of a pattern defect inspection by a conventional pattern matching method.

[Explanation of symbols]

Reference Signs List 8 CCD line sensor (input means) 15 memory 16 defect inspection unit 16a delay unit (extraction unit) 16b comparison unit B11 to B33 comparison detection block (comparison unit with Ba and Bb) Ba minimum difference determination circuit Bb defect determination circuit C11-C55 Detailed comparison block Ca Image interpolation unit Cb Minimum difference selection circuit Da, Da1, Da2 Gray-level difference calculation circuit Db selection output circuit Dc Maximum gray-level difference holding circuit De Adder circuit Df, Df1, Df2 determination circuit Dh result holding circuit M11-MBB master area P0 inspection area W, W 'defect inspection window

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Hiroyuki Onishi 4-chome Tenjin Kitamachi 1-chome, Horikawa-dori-Terauchi, Kamigyo-ku, Kyoto F-term (reference) 2F065 AA49 AA56 BB02 CC01 CC17 CC25 FF01 FF04 FF61 JJ02 JJ05 JJ25 MM03 MM22 PP12 QQ00 QQ03 QQ11 QQ12 QQ23 QQ24 QQ25 QQ27 QQ36 QQ39 QQ47 RR09 SS11

Claims (6)

[Claims] 1. A pattern defect detection method for performing defect detection by comparing a two-dimensional object pattern with a two-dimensional master pattern for each area having a predetermined size, comprising: an inspection area of the object pattern; Setting a plurality of master areas that are two-dimensionally displaced by a predetermined amount by a unit amount over an extended area extending around the required number of pixels around the area of the master pattern that should correspond to the position; The multi-valued image signal of the inspection area and the multi-valued image signal of the inspection area are compared for each of the master areas to perform defect detection. (A) For each of the plurality of master areas, Obtaining a tone difference absolute value that is an absolute value of a tone difference of a multi-level image signal for each pixel corresponding to the area and the inspection area; The absolute value of the gray level difference for each of the star areas is compared with a predetermined threshold value. Determining that there is a defect, and determining that there is no defect in the inspection area if all of the absolute values of the tone difference are less than the threshold value in one or more of the master areas. Pattern defect detection method. 2. A pattern defect detection apparatus for detecting a defect by comparing a two-dimensional object pattern with a two-dimensional master pattern for each area having a predetermined size, wherein a means for inputting a multi-valued image signal of the object pattern is provided. Means for storing a multi-valued image signal of the master pattern; a multi-valued image signal in an inspection area of the object pattern; and a required pixel centering on the area of the master pattern which should correspond in position to the inspection area. Extracting means for respectively extracting multi-valued image signals in a plurality of master areas displaced by a predetermined amount by a unit amount in a two-dimensional manner over an extended area extending around a number of areas, and a multi-valued image in the plurality of master areas Signal, and an inspection means for inputting a multi-level image signal in the inspection area, wherein the inspection means comprises: For each of them, a tone difference absolute value that is an absolute value of a tone difference of a multi-valued image signal is obtained for each pixel corresponding to the master area and the inspection area, Comparing the tone difference absolute value and a predetermined threshold value, if all the master areas include the tone difference absolute value that is equal to or greater than the threshold value, it is determined that the inspection area is defective, 1
The pattern defect detection device according to claim 1, wherein when all the absolute values of the tone differences are less than the threshold value in the master area, it is determined that there is no defect in the inspection area. 3. The pattern defect detection device according to claim 2, wherein the unit amount is smaller than the size of one pixel, and the extracting unit interpolates a multi-valued image signal of the master pattern. A pattern defect detection device for obtaining multi-valued image signals of a plurality of master areas displaced by the predetermined amount by a unit amount. 4. The pattern defect detection device according to claim 3, wherein the interpolation is performed in each direction of a two-dimensional orthogonal axis. 5. The pattern defect detection device according to claim 2, wherein said inspection means comprises: a multi-valued image signal in said inspection area; and a multi-level image signal in each of said plurality of master areas. Maximum difference holding means for obtaining and holding the maximum value of the absolute values of the tone differences in all the pixels corresponding to the value image signal for each of the plurality of master areas; and A minimum value selecting means for selecting a minimum value among the maximum values; and determining the absence of the defect when the minimum value is less than the threshold value, and determining the presence of the defect when the minimum value is equal to or more than the threshold value. A pattern defect detection device, comprising: a determination unit for performing a determination. 6. The pattern defect detection device according to claim 2, wherein said inspection means comprises: a multi-level image signal in said inspection area; and a multi-level image signal in each of said plurality of master areas. A tone difference calculating unit for determining the tone difference absolute value of the pixel corresponding to the value image signal; and for each of the plurality of master areas, all the tone difference absolute values are less than the threshold value. In the case, the master area is determined to be a pattern match, and when at least one of the tone difference absolute values is greater than or equal to the threshold value, the master area is determined to be a pattern mismatch. A comparison result holding unit for holding a comparison result by the match determination unit for each of the areas; and if all the comparison results held in the comparison result holding unit do not match the pattern, The inspection area is provided with a result determining means for performing the determination of the presence of the defect, and when at least one of the comparison results indicates the pattern coincidence, determining the absence of the defect in the inspection area. Pattern defect detection device.