JP2003083907A - Defect inspecting method and its apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a defect inspecting method and apparatus with easy sensitivity adjustment capable of both inspecting locational deviation in an image to be inspected, image distortion, etc., by considering brightness between the images and of reducing misreports and highly sensitively detecting defects only by setting a threshold value in defect inspection for comparing the image to be inspected with a reference image and detecting the defects from their difference. SOLUTION: Correction is performed in such a way as to reduce the difference between the image to be inspected and the reference image according to the contrast of the image. On a micro locational deviation part, reduction in the occurrence of misreports and highly sensitive inspection are implemented by a single low threshold value on a whole area to be inspected by a means for especially intensively computing overall brightness between the images as necessary. In the case that an object to be inspected is a semiconductor wafer, by the computing means for previously computing the brightness difference between the images caused by the difference of a film thickness in the wafer, it is possible to avoid the occurrence of misreports due to brightness irregularities without reduction in sensitivity and easily perform sensitivity adjustment by the single threshold value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention compares an image of an object such as a semiconductor wafer, a TFT, or a photomask obtained by using light or a laser with a reference image stored in advance, and based on the difference, a fine image is obtained. The present invention relates to a defect inspection method and apparatus for inspecting pattern defects, foreign matters, and the like. In particular, the present invention relates to a defect inspection method and apparatus suitable for performing visual inspection of semiconductor wafers.

[0002]

2. Description of the Related Art As a conventional technique for detecting a defect by comparing a detected image of an object to be inspected with a reference image, Japanese Patent Application Laid-Open No. 05-2005
The method described in JP-A-264467 is known.

In this method, a sample to be inspected in which repetitive patterns are regularly arranged is sequentially picked up by a line sensor, compared with an image with a time delay of the repetitive pattern pitch, and the non-matching portion is detected as a pattern defect. It is a thing. However, in reality, there are vibrations of the stage, inclination of the target object, and the like, and the positions of the two images are not necessarily aligned. Therefore, the image captured by the sensor and the image delayed by the repeated pattern pitch It is necessary to find the amount of displacement. Then, after aligning the two images based on the obtained positional deviation amount, the difference between the images is taken, and when the difference is larger than a specified threshold value, it is a defect, and when the difference is smaller than a non-defect. judge.

[0004]

As described above, according to the conventional technique, the positional deviation detection between the detected image and the reference image to be inspected → the alignment → the image comparison, that is, the difference calculation between the two images → the defect extraction Although the defect inspection is performed according to the procedure described above, if the correct positional deviation amount is not obtained when the positional deviation is detected, the positioning at the wrong position is performed. in this case,
In the aligned image, the difference between the aligned images is not large in the region where the change amount of brightness is small, but the difference is large in the region where the change amount of brightness is large.
For example, 11 in FIG. 1 is an example of a detected image, 12 is an example of a reference image, 1a is a uniformly bright background area, and 1b is an area with a dark pattern on a bright background. In addition, detected image 1
1 has a defect 1c. In the image of this example, position 1D
The waveform of the luminance value at -1D 'is as shown in FIG. Here, when the positional deviation amounts of 11 and 12 are correctly obtained, the difference image after the positioning of 11 and 12 is as shown in FIG. The difference image is an image displayed as a grayscale difference according to the difference value at each corresponding position of the detected image and the reference image. If it is assumed that a portion whose difference value is equal to or larger than the specific threshold value TH is a defect, the defect 1c of the detected image 11 in FIG.
Only detected. However, when the misregistration amounts of 11 and 12 are calculated incorrectly, the difference image after the alignment is shown in FIG.
In a region where the change in the luminance value is large, such as the edge of the pattern in the region 1b, the difference value becomes large even with a slight positional deviation, and is detected as a defect. This should not be originally detected as a defect. In other words, it is false information. Conventionally, the threshold TH has been increased as one method for avoiding the occurrence of false information as shown in FIG. This lowers the sensitivity, and a defect having a difference value equal to or less than the edge portion cannot be detected. As another method, the threshold TH is set high in the high contrast portion where false information is likely to occur, and the threshold TH is set low in the low contrast portion where false information is unlikely to occur. Therefore, the sensitivity adjustment becomes complicated.

Further, when the object to be inspected is a semiconductor wafer, even if the amount of misalignment is correctly obtained, if there is a difference in film thickness within the wafer, detection is performed as shown in 4a and 4b of FIG. A difference in brightness occurs between the image 11 and the reference image 12 in the same pattern, and the difference value is as large as 4c. This is also a false alarm, and the threshold TH must be increased in order to prevent detection.
Alternatively, there is no choice but to set separate threshold values for the area with uneven brightness and the area without uneven brightness.

In order to solve such a problem of the prior art, an object of the present invention is to perform a comparative inspection in which a detected image of an inspection object is compared with a reference image and a defect is detected from the difference.
It is an object of the present invention to provide a defect inspection method and an apparatus thereof that reduce false alarms due to color unevenness, positional deviation, and image distortion, and realize highly sensitive defect inspection.

Further, another object of the present invention is to perform a defect inspection for inspecting a semiconductor wafer, which reduces false reports due to uneven brightness of a pattern caused by a difference in film thickness and realizes a highly sensitive defect inspection. A method and an apparatus therefor are provided.

[0008]

In order to achieve the above object, the present invention performs brightness matching in a non-defective portion in an inspection for detecting a defect from a difference between two images by comparing the two images. It is characterized by One of its effective configurations is characterized by detecting defects with high sensitivity at a low threshold value even if there is a slight positional deviation. As another configuration, if the inspection target is a semiconductor wafer, even if there is a difference in brightness in the same pattern between images due to the difference in film thickness within the wafer, the brightness can be adjusted in advance by adjusting the brightness. The feature is that a defect inspection with a low threshold value and high sensitivity is performed regardless of the presence or absence of unevenness.

Further, in the inspection for detecting a portion where the difference in brightness between the two images is equal to or more than a specific threshold value as a defect,
It is characterized in that correction is performed according to the contrast of the image so that the difference between the detected image of the inspection object and the reference image is reduced. As an effective configuration, even if there is a slight misalignment or image distortion, the difference becomes large, and in the high-contrast area where false detection (false alarm) is likely to occur, the difference between the images should be reduced in advance so that the difference becomes small. It is characterized by performing a defect inspection with a low threshold value and a high sensitivity in the final defect detection processing by matching the heights.

By providing these structures, it is possible to realize a highly sensitive defect inspection method and a defect inspection apparatus in which a false alarm does not occur even if the threshold value is low with respect to the entire inspection target area.

Further, the present invention is based on the positional deviation amount detecting process for detecting the positional deviation amount between the detected image of the object to be inspected and the reference image serving as a reference, and the positional deviation amount calculated in the positional deviation amount detecting process. Position adjustment process for performing the position adjustment process for the detected image and the reference image on a pixel-by-pixel basis, and the feature amount of each pixel is calculated in each of the detected image and the reference image aligned in the position adjustment process. The image is decomposed into a plurality of categories according to the feature amount calculation process of each pixel and the correlation of the corresponding pixel in the feature amount of each pixel calculated in each image in the feature amount calculation process of each image. Image decomposition process, a correction coefficient calculation process for calculating a correction coefficient for matching the brightness of the detected image and the reference image for each category decomposed in the image category decomposition process, and the correction coefficient operation. The brightness correction process of correcting the brightness value in each pixel in at least one image using the correction coefficient for each category calculated in the process to adjust the brightness, and the brightness correction process in the brightness correction process. A defect inspection method comprising a defect detection process of comparing a detected image and a reference image that have been aligned with each other to calculate a difference value and detect a defect or defect candidate with a predetermined determination threshold value. Is.

Further, according to the present invention, in the correction coefficient calculation step, a correction coefficient for adjusting the brightness is calculated by using a scatter diagram showing the relationship between the brightness value of the detected image and the brightness value of the reference image. Is characterized by.

[0013]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to FIGS. 1 to 13.

As an embodiment, taking a defect inspection method in an optical appearance inspection apparatus for semiconductor wafers as an example, FIG. 5 shows an example of the configuration of the apparatus. The appearance inspection apparatus mounts a sample (an object to be inspected such as a semiconductor wafer) 51, a stage 52 for moving the same, a light source 501, and illumination for converging light emitted from the light source 501 and illuminating the sample 51 as illumination light. Optical system 502, sample 51
Objective lens 5 for forming an optical image obtained by being reflected from
03, and a detector 53 having an image sensor 504 for receiving the formed optical image and converting it into an image signal according to the brightness. The image processing unit 55 calculates a defect candidate in the wafer, which is a sample, based on the image detected by the detection unit 53. Then, the image processing unit 5
Reference numeral 5 denotes an AD conversion unit 54 that converts an input signal from the detection unit 53 into a digital signal, and a preprocessing unit 50 that performs image correction such as shading correction and dark level correction from the digital signal.
5. A delay memory 506 that stores a comparison target digital signal as a reference image signal, and a position shift amount between the digital signal (detected image signal) detected by the detection unit 53 and the reference image signal of the delay memory 506 is detected. Displacement amount detector 5
07, by using the calculated positional deviation amount, the positional deviation between the detected image signal and the reference image signal is corrected and both image signals are compared, and a portion having a difference value larger than a specific threshold value is defective. The image comparison unit 508 that outputs the candidates and the feature extraction unit 509 that calculates the coordinates and the feature amount of the defect candidate are configured. The overall control unit 56 has a user interface unit 5 having a display unit and an input unit that receives a change in inspection parameters (such as a threshold value used in image comparison) from a user and displays detected defect information.
10, a storage device 511 that stores the feature amount and image of the detected defect candidate, and a CPU that performs various controls. The mechanical controller 512 drives and controls the stage 52 based on a control command from the overall control unit 56. The image processing unit 55, the detection unit 53 and the like are also drive-controlled based on a command from the overall control unit 56.

As shown in FIG. 6, a semiconductor wafer 51 to be inspected has a large number of chips of the same pattern arranged regularly. The inspection apparatus shown in FIG. 5 compares images at the same positions of two adjacent chips, for example, the areas 61 and 62 in FIG. 6, and detects the difference as a defect.

The operation will be described below.
Controls the mechanical controller 512 to continuously move the sample semiconductor wafer 51 by the stage 52. In synchronization with this, the images of the chips are sequentially captured by the image sensor 504 of the detection unit 53. The image sensor 504 of the detection unit 53 outputs the input signal to the image processing unit 55.

In the image processing unit 55, first, the input analog signal is converted into a digital signal by the AD conversion unit 54, and the preprocessing unit 505 performs shading correction, dark level correction and the like. Further, the pre-processing unit 505 performs processing for improving S / N by removing noise and edge enhancement as necessary. However, the image quality improvement processing by the S / N improvement can be performed later.

The position shift detecting section 507 includes a preprocessing section 50.
5, the image signal of the chip to be inspected (detected image signal), and the reference image signal output from the delay memory 506, which is delayed by the time the stage moves by the chip interval, that is, one of the chips to be inspected. The image signal of the previous chip (reference image signal) is input as a set. The image signals of these two chips, which are sequentially input in synchronization with the movement of the stage, do not become signals at exactly the same location when the stage is vibrated or the wafer set on the stage is tilted. Therefore, the positional deviation detection unit 507 calculates the positional deviation amount between two images that are continuously input. At this time, the detected image signal and the reference image signal are continuously input, but the positional deviation amount is calculated sequentially for each processing unit with a specific length as one processing unit. The following processing is also performed for each processing unit.

The image comparison unit 508 aligns the images using the calculated positional deviation amount, compares the detected image signal with the reference image signal, and detects a region where the difference value is larger than a specific threshold value. Output as a candidate.

The feature extraction unit 509 edits each of the plurality of defect candidates by deleting small ones as noise, merging neighboring defect candidates as one defect, and the like. , Size, etc. are calculated and output as a final defect. These pieces of information are stored in the storage device 511. In addition, it is presented to the user via the user interface unit 510.

Here, when the image comparison unit 508 obtains defect candidates from simple difference values, they are not always true defects. An example will be described below.

The detected image 11 and the reference image 12 shown in FIG.
Both images have a dark pattern of crosses in a uniformly bright background. Further, only the detected image 11 has the defect 1c.
The graph in FIG. 1B shows the positions 1D-1D of the detected image 11.
It is the waveform of the luminance value in '. As shown in this graph, the image has an area 1a with high brightness and little change in brightness, that is, low contrast, and an area 1b with large change in brightness at the edge portion of the pattern, that is, high contrast area 1b. FIG. 2A shows an image of the difference value at each corresponding position when the positional deviation detection unit 507 calculates the correct positional deviation amount for these images and the positional adjustment is performed, that is, the difference is It is an image in which a small area is displayed dark and a large area is displayed brightly. FIG. 2B is a waveform of the difference value at the positions 1D-1D '. If an area having a difference value equal to or larger than the threshold value TH is a defect, in this case, only the area of the defect 1c is detected.

On the other hand, FIGS. 3A and 3B show an image of a difference value and a waveform of the difference value when the misregistration amount is erroneously calculated and misalignment is performed. In the low-contrast region where the change amount of brightness is small like the region 1a, the difference value does not increase even if the positions of the detected image 11 and the reference image 12 are slightly shifted, but the change amount of brightness is large like the region 1b. It is shown that in the high-contrast region, the difference value becomes large even with a slight displacement. When a region having a threshold value TH or more is set as a defect candidate, a position where the difference value becomes large due to such positional displacement is also detected. These should not be detected by nature. In the following, what is not such a true defect is described as a false alarm.

Conventionally, in order to avoid detection of a false alarm due to displacement, one method is to increase the threshold value from TH to TH2 as shown in FIG. 3B. Cannot be detected. That is, in order to avoid the occurrence of false information, the sensitivity is lowered and the inspection is performed. As another method, as shown in FIG. 3C, the threshold value is set to TH2 in the high contrast area and TH is set in the low contrast area. It becomes necessary to adjust the sensitivity for the user.

When the film thickness of the semiconductor wafer 51 is not uniform, the inspection target image and the reference image have a difference in brightness. For example, three lined crosses 4a and 4b in FIG. 4 are corresponding patterns in the inspection target image 11 and the reference image 12, but the brightness is greatly different due to the difference in film thickness (hereinafter, referred to as brightness. It is described as unevenness, and the detected image 11
There is a defect 1c only in. FIG. 4A is an image of the difference value at each corresponding position when the positional deviation detection unit 507 calculates the correct positional deviation amount for these images and the positioning is performed. Even in the case of the pattern, the difference value becomes large in the portion with uneven brightness. Figure 4 (b) shows
It is a waveform of the difference value in position 1D-1D '. If a region having a difference value equal to or larger than the threshold value TH is a defect, a cross pattern having a large difference value due to uneven brightness is also detected in addition to the defect 1c. These are false reports. In order to avoid detection of false information due to such uneven brightness, the threshold value is increased from TH to TH2 as in the case of occurrence of false information due to positional deviation, and inspection is performed with low sensitivity, or uneven brightness is detected. A threshold value is set to TH2 in a certain area, and a threshold value is set to TH in a portion where there is no unevenness in brightness. .

On the other hand, in the present invention, the image comparison unit 508
In, before calculating the difference between the detected image and the reference image,
The brightness of the images is adjusted in advance. Figure 7
Is an embodiment of the outline of processing. First, based on the position shift amount calculated by the position shift detection unit 507, the inspection image F (i,
j) and the reference image G (i, j) are aligned in pixel units (S70). The feature amount of each pixel is calculated for the image after this alignment (S71), and the target image is decomposed into a plurality of images according to the feature amount (S72). The group of pixels after decomposition is described as a category below. Next, the brightness (luminance value: gradation value) F of the detected image and the reference image for each category F
A correction coefficient for combining (i, j) and G (i, j) is calculated (73). The brightness is adjusted by correcting the brightness value of one image so as to be close to the brightness value of the other image for each category using this correction coefficient (74). Then, the difference is calculated between each pixel (i, j) corresponding to the corrected detected image F ′ (i, j) and the reference image G (i, j) (S75), and the threshold value calculated for each pixel is calculated. The larger one is extracted as a defect candidate (S76).

Next, one embodiment of the processing procedure of brightness adjustment from S71 to S74 will be described in detail. Here, the detected image F (i, j) and the reference image G (i, j) to be compared are
Among the detected images, the brightness value (brightness) is corrected. First, in S70, the detected image F (i, j) and the reference image G (i,
The feature amount of each pixel (i, j) is calculated using j) (S71). There are many feature amounts such as brightness, contrast, difference in brightness (shading difference) between the detected image and the reference image, and frequency. Hereinafter, the case where contrast is used as the feature amount will be described as an example. First, the contrast is calculated for each of all pixels in the target area (S71). There are various operators for contrast calculation, but part 1
One is a range filter. As shown in FIG. 8, the contrast C (i, j) at the coordinate position (i, j) in the target area is defined as the maximum brightness Max (A, B, C, D) in the neighboring area. The difference is the minimum value Min (A, B, C, D). When the filter size is 2 × 2, if the brightness value at (i, j) is A and the brightness in the vicinity is B, C, and D, the calculation formula is (Formula 1).

C (i, j) = Max (A, B, C, D) -Min (A, B, C, D) (Equation 1) Further, depending on the image quality, the influence of noise is obtained instead of the range filter. A reducing percentile filter may be used. Further, the contrast C (i, j) at the coordinate position (i, j) in the target area may be calculated by the second derivative. In this case, as shown in FIG.
Using, the calculation according to (Equation 2) is performed.

Dx = B + H−2 × E, Dy = D + F−2 × E C (i, j) = Max (Dx, Dy) (Equation 2) In addition to this, in order to obtain the luminance change amount in the neighborhood. Various calculation methods can be adopted. In this way, the detected image F (i,
j) the contrast FC (i, j) of each pixel (i, j)
j), the contrast GC (i, j) of each pixel (i, j) in the reference image G (i, j) is calculated, and the detected image F (i, j) is calculated.
j) and the corresponding pixel (i, j) of the reference image G (i, j)
The average value (Equation 3), the difference (Equation 4), or the larger one (Equation 5) is used to integrate the contrasts of the two images, and the contrast value at each pixel (i, j) is unique. To decide. Then, according to the uniquely determined contrast value C (i, j), the image is divided into several categories as contrast categories (S72). Hereinafter, what is decomposed into several stages is referred to as a contrast category. As a result, a portion where the brightness is uniform (low contrast area) such as the area 1a, and a portion where the brightness is sharply changed (high contrast area) like the pattern edge portion of the area 1b.
Are gradually decomposed into contrast categories.

C (i, j) = (FC (i, j) + GC (i, j)) / 2 (Equation 3) C (i, j) = | FC (i, j) -GC (i, j) ) | (Equation 4) C (i, j) = Max (FC (i, j), GC (i, j)) (Equation 5) Next, a correction coefficient for adjusting the brightness is set for each contrast category. Calculate (S73). An example is shown in FIG.
First, for pixels belonging to the same contrast category, a scatter diagram is created with the horizontal axis representing the luminance value X in the detected image and the vertical axis representing the corresponding luminance value Y in the reference image. Then, an approximate straight line is obtained from the scatter plot. Figure 10
101 is an approximate straight line obtained from a scatter diagram of pixels belonging to a certain contrast category. There are various methods for calculating the approximate straight line, and as an example thereof, there is a least-squares approximation (a method for obtaining a straight line that minimizes the total sum of distances from each point). Then, the slopes a and Y of the calculated approximate straight line
The intercept b becomes the correction coefficient for the contrast category (S73).

The brightness value F (i, j) of the detected image is corrected using the correction coefficient calculated in this way, and the brightness is adjusted (S74). In reality, the brightness value of the detected image is F
If it is (i, j), the corrected detected image F ′
(I, j) is calculated from the slope a of the approximate straight line and the Y intercept b (equation 6).
Calculate with. This (Equation 6) is a brightness adjustment by correcting the brightness value of one image so as to be close to the brightness value of the other image for each contrast category, which is a feature of the present invention.

F ′ (i, j) = a × F (i, j) + b (Equation 6) Then, the corrected brightness value F ′ (i, j) of the detected image with the adjusted brightness. And the brightness value G of the reference image
The difference of (i, j) is obtained by the difference D (i, j) of (Equation 7) (S75), and a portion having a difference value larger than the set threshold value TH is set as a defect candidate (S76).

D (i, j) = F ′ (i, j) −G (i, j) (Equation 7) As for the correction of the brightness of the detected image, as shown in (Equation 6), the scatter diagram is inclined by 45 degrees. , Y intercepts are placed on a straight line, which is equivalent to rotating (rotation amount: gain) and shifting (shift amount: offset) the luminance value of each pixel in the scatter diagram. Figure 1
4 shows the operation. And the difference value D (i, j)
Is equivalent to the distance from the straight line after conversion. This means that the closer the distance to the straight line is, the smaller the difference value after the correction is, that is, the stronger the correction is. Further, the threshold value TH for detecting defects is set outside the scatter diagram after conversion, as shown in FIG. Therefore, in order to set the threshold value TH to be low and perform a highly sensitive inspection, FIG.
The scatter plot after conversion needs to be slim as shown in.

Therefore, in the present invention, the scatter diagram for each contrast category is decomposed into smaller categories, and an approximate straight line is obtained for each decomposed small category. One example is shown in FIG.
This will be described using 5. 151 is a scatter diagram in a certain contrast category. This is decomposed with a feature amount different from the contrast. Here, an example in which the grayscale difference between the uncorrected detected image and the reference image is used as the feature amount will be described. First, for the pixel of interest, the shade difference: F (i, j) -G (i, j) is calculated, and the scatter diagram area is further decomposed into several shade difference categories according to the set step size of the shade difference. To do. This is 1
This is equivalent to slicing the scatter plot in parallel with a straight line having an inclination of 45 degrees, such as 52. Next, approximate straight lines 152-1 to 152-N are calculated for each of the sliced small categories.
Then, as described above, each of the approximate straight lines 152-1 to 15-15
Correction is performed using the 2-N slope and the y-intercept. Since the slope a and the y-intercept b, which are correction coefficients, are determined depending on which small category the pixel of interest belongs to, the correction coefficient at F (i, j) is a.
If (i, j) and b (i, j), then the corrected F '
(I, j) is represented by (Equation 8).

F ′ (i, j) = a (i, j) × F (i, j) + b (i, j) (Equation 8) In this way, the image is decomposed using still another feature amount, By making the scatter diagram slim, the scatter diagram after correction is also 15
As shown in FIG. 3, it becomes slim and the threshold value TH can be set low.

Here, the grayscale difference is used as the feature amount for further slicing the scatter diagram, but another feature amount such as brightness information of the detected image and the reference image may be used. Further, instead of slicing the scatter diagram, a plurality of straight lines may be simultaneously calculated from one scatter diagram by using Hough transform or the like as shown in FIG. Further, the correction coefficient in each category may be calculated not by linear approximation but by polynomial curve approximation.

Here, if all the pixels are corrected, even the defects will be corrected. Therefore, in the present invention, another characteristic amount is used in the scatter diagram in order to detect a normal area such as color unevenness with strong brightness and to detect abnormalities (defect candidates) without much correction. By distinguishing between normal and abnormal areas, the brightness adjustment method is changed. For example, as the feature amount for identifying the defect and the color unevenness, the texture information of the inspection image and the reference image, the grayscale difference,
Although there are various contrasts, frequencies, etc., here, description will be made by taking an example of identification using frequency (number of pixels). As a general characteristic, the color unevenness (normal area) has a large frequency due to the characteristics that it is spread over a wide range, repeatedly occurs, or occurs on the entire surface of a certain pattern. On the other hand, the frequency of defects (non-normal areas) is low. By utilizing this, the defect and the color unevenness are identified, and the method of adjusting the brightness is changed. FIG. 17
171 is a scatter plot of a certain contrast category sliced by the grayscale difference, and 172 is a graph showing the number of pixels for each category after slicing (hereinafter, referred to as frequency distribution). Where the grayscale difference is small, it is normal. There is a high possibility that a portion or a portion having a large difference in shade has a defect or uneven color. Then, as the grayscale difference increases, the frequency also decreases. Therefore, it is discriminated whether normal or abnormal according to the frequency. For example, as shown by 173, if the frequency difference is equal to or larger than the threshold value THN, the normal category such as color unevenness is set, and if the frequency is equal to or lower than THN, the abnormal category is set as indicated by 173. The threshold THN is preset as a parameter. Further, as shown in FIG. 18, the standard deviation σ can be calculated from the frequency distribution, and the point where N × σ can be set as the threshold value.

In this way, after the categories are discriminated between normal and abnormal categories, correction (brightness matching) is performed according to the discrimination result. The method is shown in FIG. First, for a category determined to be normal, an approximate straight line 101 is obtained using the data in the category as in 191. Accordingly, the brightness of the data in this category is adjusted (192). In addition, linear approximation is not performed in the category that is determined to be abnormal, and brightness is adjusted using the approximate straight line 101 of the closest normal category (193). As a result, in the abnormal category, since the distance from the approximate straight line becomes large, the adjustment of the brightness becomes weak, and the difference value after correction remains uncorrected to some extent (1
94). Further, when it is clear that the category is the defect category, it is possible not to adjust the brightness at all.
In this case, the defect category has gain = 1 and offset =
You can set 0. In this way, in the present invention, it is possible to detect a minute defect while suppressing the noise signal to a low level.

Furthermore, these methods can be changed according to each contrast category. As an example,
As shown in FIG. 20, in the low contrast category, the scatter diagram tends to vary as compared with the high contrast category. In particular, the color unevenness between chips caused by the difference in film thickness is
Most of them are in the low contrast category. However, when the color unevenness is particularly strong, the slice is divided into the same small categories with all the contrasts, and normal / abnormal is discriminated (2
01), the brightness may not be completely matched in the low contrast portion and may remain as a false information. In such a case, by slicing more finely in the low contrast portion, the number of approximate straight lines can be increased, the correction can be strengthened more than in the high contrast category, and false information can be reduced (202).

As another example, as described above, the pattern edge portion is included in the category of high contrast, but in such a place, the difference value is small even if the position is slightly displaced or the image is distorted. It becomes large and is likely to be false information. Therefore, in the high-contrast region, as shown in FIG. 21, the frequency threshold value THN for discriminating between normal and abnormal can be set smaller than that in the low contrast category. As a result, the categories that are determined to be normal are increased, and the matching is performed in a wide range.

As described above, the sensitivity adjustment according to the contrast is performed by the slicing method and the frequency threshold value THN.
Can be done in several ways, such as changing These may be used properly depending on the sample, or may be used depending on the image quality and the feature amount of the image. Also, of course, sensitivity adjustment On / O
FF switching is also possible. Further, it is possible to constantly monitor the accuracy of the positional deviation detection and make the brightness adjustment stronger than usual in the high contrast area only when the correct positional deviation amount is not calculated.

Here, the kind of contrast calculation filter, the filter size, the number of divisions into the contrast category, the step size, and the number of slices by the difference in density, the step size, etc. can be flexibly changed by defining them in the lookup table. is there.

An example in which the scatter diagram is slimmed by performing the image decomposition based on the contrast and the density difference has been described above. However, as other feature values, the brightness value of the detected image, the brightness value of the reference image,
Image decomposition may be performed using frequency regions, color information, texture information, brightness variance values, and the like. In short, it is within the scope of the present invention if the image can be decomposed into regions having the same characteristics and the scatter diagram can be slimmed.

Also, an example has been shown in which an image is decomposed by using two feature amounts and the degree of fitting is changed by further using the frequency. However, even if the scatter diagram is slimmed from three or more feature amounts. Good. For example, as described above, in a high-contrast region such as a pattern edge, the density difference becomes large even if there is a slight positional deviation. In other words, the scatter diagram becomes wider if there is a displacement. Therefore, the spread of the scatter diagram data is checked, the positional deviation amount that minimizes the spread is calculated in subpixel units, and the brightness is adjusted using this scatter diagram. That is, the brightness adjustment and the positional shift detection in subpixel units are simultaneously performed, so that the comparison image is optimized so that the highest sensitivity inspection can be performed.

Although the positional deviation amount is naturally detected in two dimensions below, a case in which the positional deviation amount is detected in one dimension (X-axis direction) will be described for simplicity of explanation. As shown in FIG. 11, it is assumed that two images aligned in pixel units are displaced by α in subpixel units (−0.5 <α
<0.5 pixels). At this time, in order to align the detected image f and the reference image g with each other by ½ × α and perform the alignment in sub-pixel units, linear interpolation is performed (Equation 9) (Equation 1).
Images F (x) and G (x) shown in 0) may be created.

F (x) = f (x) + α / 2 · (f (x) −f (x−1)) (Equation 9) G (x) = g (x) + α / 2 · (g (x + 1) ) -G (x)) (Equation 10) Further, the detected image is the image F ′ after the adjustment of the brightness.
(X) is represented by (Equation 11).

F ′ (x) = gain (x) · F (x) + offset (x) (Equation 11) Here, in order to simultaneously calculate the sub-pixel positional deviation amount and the brightness correction coefficient of the detected image and the reference image. Can be obtained by calculating a coefficient that minimizes the sum of the brightness differences between the two images after the position correction and the brightness correction. That is, it is sufficient to solve the three equations (Equation 12) to (Equation 14). As described above, according to the present invention, in the inspection in which two images (detection image and reference image) are compared and a defect is detected from the difference value, the difference between the images becomes large due to uneven brightness or positional deviation. The brightness is adjusted in advance for the area. Further, in the region having the feature as the defect, the brightness adjustment is weakened or the brightness adjustment is not performed. FIG. 12D is a waveform of the difference value after the brightness adjustment according to the present invention when the correct positional deviation amount is not detected in FIG. FIG. 13D is the same as FIG.
7 is a waveform of a difference value after the brightness adjustment according to the present invention when there is uneven brightness. In both cases, the brightness is originally matched to the area that should not be detected as a defect, and the difference value becomes small. On the other hand, since the brightness adjustment is weakened in the defective portion, the difference value does not become so small. For this reason, conventionally, the threshold value is set to TH2 for all areas, or two threshold values TH and TH2 are set to avoid the occurrence of false alarms. However, according to the present invention, the sensitivity is not lowered. It is possible to avoid the occurrence of false alarms due to displacement and uneven brightness, and to enable high-sensitivity inspection and easy sensitivity adjustment only at the low threshold TH3.

Furthermore, according to the present invention, the optical visual inspection apparatus can usually achieve a detection sensitivity of 100 nm. Further, by using the present invention, it is possible to obtain a detection sensitivity of 50 nm when the image quality is good.

Although one embodiment of the present invention has been described with reference to a comparative inspection image in an optical appearance inspection apparatus for semiconductor wafers, electron beam pattern inspection and DU.
It is also applicable to visual inspection using V as a light source. In this case, detection sensitivity of 30 to 70 nm can be achieved. Also,
The object to be inspected is not limited to the semiconductor wafer,
If the defect detection is performed by comparing the images, for example, a TFT substrate, a photomask, a printed board, or the like can be applied.

[0050]

As described above, according to the present invention,
In particular, it is possible to reduce the occurrence of false information by adjusting the brightness for a high-contrast portion and brightness unevenness where false information is likely to occur due to misalignment or image distortion.

Further, according to the present invention, by adjusting the brightness, it becomes possible to set a low threshold value as a result, and it is possible to realize a high sensitivity inspection.

Further, according to the present invention, it is possible to reduce the occurrence of false alarms and detect defects by only setting the threshold value, and it is possible to easily adjust the sensitivity.

Further, according to the present invention, the detection sensitivity of 5 can be obtained by applying to the comparison inspection in the optical appearance inspection apparatus.
0 nm can be achieved. Also, electron beam pattern inspection and D
A detection sensitivity of 30 to 70 nm can be achieved by applying the appearance inspection using UV as a light source.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a detected image of an inspection target and a luminance waveform at that time.

FIG. 2 is a diagram showing an example of a conventional threshold value setting method when a misregistration amount is correctly calculated.

FIG. 3 is a diagram showing an example of a conventional threshold value setting method when a misregistration amount is erroneously calculated.

FIG. 4 is a diagram showing an example of a detected image of an object to be inspected when there is uneven brightness between comparison chips and a conventional threshold value setting method.

FIG. 5 is a configuration diagram showing an embodiment of a defect inspection apparatus according to the present invention.

FIG. 6 is a diagram showing a semiconductor wafer to be inspected.

7 is a diagram showing a processing flow in an image comparison unit of the image processing unit shown in FIG.

FIG. 8 is a diagram illustrating an example of a contrast calculation method for a pixel of interest (i, j), which is a feature amount calculation for each pixel according to the present invention.

FIG. 9 is a diagram illustrating another embodiment of a contrast calculation method for a pixel of interest (i, j), which is a feature amount calculation for each pixel according to the present invention.

FIG. 10 is a diagram for explaining an example of an approximate straight line calculation from a scatter diagram according to the present invention.

FIG. 11 is a diagram showing a positional relationship between a detected image f and a reference image g after pixel unit alignment according to the present invention.

FIG. 12 is a diagram for explaining an example of a threshold value setting method according to the present invention when a misregistration amount is erroneously calculated.

FIG. 13 is a diagram for explaining an embodiment of a threshold value setting method when there is uneven brightness between comparison chips according to the present invention.

FIG. 14 is a diagram for explaining an example of a behavior of brightness adjustment by linear approximation of a scatter diagram according to the present invention.

FIG. 15 is a diagram for explaining an embodiment of slimming a scatter diagram according to the present invention.

FIG. 16 is a view for explaining another embodiment of slimming a scatter diagram according to the present invention.

FIG. 17 is a diagram for explaining an embodiment of a frequency-based category discrimination method according to the present invention.

FIG. 18 is a diagram for explaining an example of a method of calculating a frequency threshold value according to the present invention.

FIG. 19 is a diagram showing a difference in correction between a normal category and an abnormal category according to the present invention.

FIG. 20 is a diagram for explaining an embodiment of a scatter diagram for each contrast and linear approximation by contrast according to the present invention.

FIG. 21 is a diagram for explaining an example of a method of setting a frequency threshold value by contrast according to the present invention.

[Explanation of symbols]

11 ... detected image, 12 ... reference image, 51 ... sample, 52 ...
Stage, 53 ... Detection unit, 501 ... Light source, 502 ... Illumination optical system, 503 ... Objective lens, 504 ... Image sensor, 55 ... Image processing unit, 54 ... AD conversion unit, 505 ... Pre-processing unit, 506 ... Delay memory, Reference numeral 507 ... Positional deviation amount detection unit, 508 ... Image comparison unit, 509 ... Feature extraction unit, 56 ...
Overall control unit 510 ... User interface unit 51
1 ... Storage device, 512 ... Mechanical controller, 10
1 ... Approximate straight line.

Continued front page    (72) Inventor Takashi Okabe             292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa             Inside the Hitachi, Ltd. production technology laboratory F term (reference) 2G051 AA51 AB02 BA05 CA03 EA12                       EA14 EB01 ED07 ED08                 4M106 AA01 BA04 CA39 CA50 DB04                       DB20 DJ11 DJ18 DJ21 DJ23                 5L096 BA03 EA12 EA14 EA35 FA06                       FA17 FA34 FA37 FA69 GA04                       GA08 HA07 JA11

Claims (11)

[Claims] 1. A position shift amount detecting step of detecting a position shift amount between a detected image to be inspected and a reference image serving as a reference, and the detected image using the position shift amount calculated in the position shift amount detecting step. The amount of positional deviation between the reference image and the reference image is corrected, and the brightness of at least one of the images is adjusted so that the difference between the corrected detected image and the non-defective portion of the reference image is reduced, and the brightness is adjusted. A defect inspection method comprising: a defect detection process of comparing a detected image and a reference image that have been combined and detecting a defect or a defect candidate from the difference. 2. A position shift amount detecting step of detecting a position shift amount between a detected image to be inspected and a reference image serving as a reference, and the detected image using the position shift amount calculated in the position shift amount detecting step. The amount of positional deviation between the reference image and the reference image is corrected, and the brightness of at least one of the images is adjusted to match the brightness in the portion where the brightness unevenness occurs between the corrected detected image and the reference image. A defect inspection method comprising: comparing a detected image and a reference image that have been matched with each other and detecting a defect or a defect candidate from the difference. 3. A position shift amount detecting step of detecting a position shift amount between a detected image to be inspected and a reference image serving as a reference, and the detected image using the position shift amount calculated in the position shift amount detecting step. Position difference between the reference image and the reference image is corrected, and the brightness of at least one of the images is adjusted in the region where the image distortion is generated between the corrected detected image and the reference image to obtain the brightness. And a reference image, and a defect detecting step of detecting a defect or a defect candidate from the difference between the detected image and the reference image. 4. A position shift amount detecting step of detecting a position shift amount between a detected image to be inspected and a reference image serving as a reference, and the detected image using the position shift amount calculated in the position shift amount detecting step. Position difference amount between the reference image and the reference image is corrected, and the brightness of at least one image is adjusted according to the contrast of the image so that the difference between the corrected detected image and the reference image is reduced, and the brightness is adjusted. A defect inspection method comprising: comparing a detected image and a reference image that have been matched with each other and detecting a defect or a defect candidate from the difference. 5. An amount of positional deviation between a detected image to be inspected and a reference image serving as a reference is detected, and when the amount of positional deviation is detected,
A positional deviation amount detecting step of monitoring the positional deviation amount detection accuracy, and correcting the positional deviation amount between the detected image and the reference image using the positional deviation amount calculated in the positional deviation amount detecting step, When the monitored positional deviation amount detection error is large, the brightness of at least one of the images is adjusted so that the difference between the non-defect portion of the corrected detected image and the reference image in the area where the positional deviation occurs is small. A defect inspection method comprising: a defect detection process of combining a detected image and a reference image, the brightness of which is adjusted, and detecting a defect or a defect candidate from the difference. 6. A position shift amount detecting step of detecting a position shift amount between a detected image to be inspected and a reference image serving as a reference, and the detected image based on the position shift amount calculated in the position shift amount detecting step. A registration process of performing registration processing between the reference image and the reference image on a pixel-by-pixel basis, and a feature of each pixel for calculating a feature amount of each pixel in each of the detected image and the reference image aligned in the alignment process Quantity calculation step, and an image category that decomposes the image into a plurality of categories according to the correlation at the corresponding pixel in the feature value at each pixel calculated for each image in the feature value calculation step for each image A decomposition process, a correction coefficient calculation process for calculating a correction coefficient for matching the brightness of the detected image and the reference image for each category decomposed in the image category decomposition process, and a correction coefficient calculation process. The brightness correction process of correcting the brightness value at each pixel in at least one image using the correction coefficient for each category to adjust the brightness, and the brightness correction process in the brightness correction process. 2. A defect inspection method, comprising: comparing a detected image with a reference image, calculating a difference value, and detecting a defect or a defect candidate with a predetermined determination threshold value. 7. A correction coefficient for adjusting the brightness is calculated in the correction coefficient calculation step using a scatter diagram showing the relationship between the brightness value of the detected image and the brightness value of the reference image. The defect detection method according to claim 6. 8. The misregistration amount detecting step detects the misregistration amount in subpixel units by using a scatter diagram showing the relationship between the luminance value of the detected image and the luminance value of the reference image, and in the aligning step. The defect detection method according to claim 6 or 7, wherein the alignment processing is performed in subpixel units. 9. A stage on which a sample to be inspected is mounted, an irradiation optical system for irradiating the sample with a beam, a detector for detecting the beam from the sample and converting it into a signal, and a detector of the detector. AD that converts the output signal into a detected digital image
A conversion circuit, a memory that stores the conversion circuit, a positional deviation amount detection unit that detects the positional deviation amount between the detected digital image and the reference digital image that serves as a reference, and the positional deviation amount detected by the positional deviation amount detection unit. An alignment unit that aligns the detected digital image and the reference digital image based on the above, and at least one image according to the contrast of the difference in brightness between the detected digital image and the reference digital image aligned by the alignment unit. An image in which a defect or defect candidate is detected by comparing a reference digital image with a brightness correction unit that is aligned with respect to each other, and a detected digital image that is aligned by the alignment unit and is corrected by the brightness correction unit. An image processing unit having a comparison unit, and an output monitor for displaying a defect detection result detected by the image processing unit. Defect inspection apparatus. 10. A stage on which a sample to be inspected is mounted, an irradiation optical system for irradiating the sample with a beam, a detector for detecting the beam from the sample and converting it into a signal, and a detector of the detector. AD that converts the output signal into a detected digital image
A conversion circuit, a memory that stores the conversion circuit, a position deviation amount detection unit that detects the position deviation amount between the detected digital image of the inspection target and the reference digital image that is the reference, and the position deviation calculated by the position deviation amount detection unit. An alignment processing unit that performs alignment processing of the detected digital image and the reference digital image on a pixel-by-pixel basis based on the amount of each of the detected digital image and the reference digital image aligned by the alignment processing unit. A feature amount in a pixel is calculated, the image is decomposed into a plurality of categories according to the correlation in the corresponding pixel in the feature amount in each pixel calculated in each of the images, and each of the decomposed categories is divided. A correction coefficient for matching the brightness of the detected image with that of the reference image is calculated, and the calculated correction coefficient for each category is used for each pixel in at least one of the images. A brightness correction unit that corrects the brightness value to adjust the brightness, and a difference value is calculated by comparing the detected digital image and the reference digital image whose brightness has been adjusted by the brightness correction unit. And a defect inspection device configured to detect a defect or a defect candidate with a predetermined determination threshold value. 11. A scatter diagram showing a relationship between a brightness value of a detected digital image and a brightness value of a reference digital image, in a brightness correction unit of the image processing unit, for calculating a correction coefficient for adjusting brightness. 11. The defect detecting device according to claim 10, wherein the defect detecting device is configured to be used.