JP2009097959A - Defect detecting device and defect detection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To adjust the detection sensitivity of defect detection, performed by using image signals, respectively acquired from each imaging pixel, corresponding to each S/N ratio of the imaging pixel, in defect detection wherein a gray level difference is detected, between a pixel inside one unit pattern and a pixel inside the other unit pattern, corresponding to the pixel between two unit patterns, appearing repeatedly to correspond to a repeating pattern on a photographed image acquired by imaging a sample surface, where the repeating pattern is formed by an imaging element formed by arraying a plurality of imaging pixels, and to a case when the gray level difference exceeds a detection threshold, where the pixels from which the gray level difference is detected represent a defect. <P>SOLUTION: The gray level difference is corrected, corresponding to the S/N ratio of an output signal of the imaging pixels generating respectively each gray level value from which the gray level difference is detected. Alternatively, the detection threshold to be compared with the gray level difference is corrected corresponding to the S/N ratio of the output signal of the imaging pixels generating respectively each gray level value, from which the gray level difference is detected. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

The present invention relates to a defect inspection apparatus and a defect inspection method for detecting a different portion as a defect of the pattern by comparing an inspection image obtained by imaging the same pattern to be inspected and a reference image. .
More specifically, in a captured image obtained by imaging the surface of a sample on which a repetitive pattern is formed, pixels corresponding to each other between two unit patterns among a plurality of unit patterns that repeatedly appear corresponding to the repetitive pattern A defect inspection apparatus and a defect inspection that detect a gray level difference between them and determine that a defect candidate exists in a portion indicated by a pixel in which the gray level difference is detected when the detected gray level difference exceeds a predetermined detection threshold The present invention relates to a technique for adjusting defect detection sensitivity by correcting a detected gray level difference or a detection threshold.

The manufacture of semiconductor devices such as semiconductor wafers, photomask substrates, and liquid crystal display panels consists of a large number of man-hours, and the occurrence of defects in final and intermediate processes is inspected and fed back to the manufacturing process. Is important from the viewpoint of improving yield. In order to detect defects in the middle of the manufacturing process, images obtained by imaging the pattern formed on the surface of a sample such as a semiconductor wafer, a photomask substrate, a liquid crystal display panel substrate, or a liquid crystal device substrate are obtained. A pattern defect inspection for detecting defects existing on the surface of a sample by inspecting the pattern is widely performed.
In the following description, a semiconductor wafer defect inspection apparatus that inspects a defect of a pattern formed on a semiconductor wafer will be described as an example. However, the present invention is not limited to this, and can be widely applied to defect inspection apparatuses for inspecting semiconductor devices such as a semiconductor memory photomask substrate, a liquid crystal device substrate, and a liquid crystal display panel substrate. .

FIG. 1 shows a block diagram of a defect inspection apparatus similar to that proposed by the applicant of the present application in Japanese Patent Application No. 2003-188209 (the following Patent Document 1). The defect inspection apparatus 1 is provided with a stage 11 movable in a three-dimensional direction, and a sample stage (chuck stage) 12 is provided on the upper surface of the stage 11. A wafer 2 as a sample to be inspected is placed on the sample stage 12 and fixed.
In addition, an imaging unit 14 for capturing an optical image of the surface of the wafer 2 is provided above the sample stage 12. An image sensor (imaging device) such as a one-dimensional or two-dimensional CCD sensor (preferably a one-dimensional TDI sensor) is used for the imaging unit 14, and an optical image of the surface of the wafer 2 imaged on the light receiving surface is used. Convert to electrical signal. Here, assuming that the imaging unit 14 uses a one-dimensional TDI sensor as an imaging device, the configuration example of FIG. 1 will be described below.

  By moving the imaging unit 14 and the wafer 2 relatively by moving the stage 11, the imaging unit 14 is scanned in the X direction or the Y direction with respect to the wafer 2 to obtain a two-dimensional image of the surface of the wafer 2. As the illumination optical system for illuminating the wafer 2, a bright field illumination optical system or a dark field illumination optical system is used.

The image signal output from the imaging unit 14 is stored in the image storage unit 21 after being converted into a multi-value digital signal (gray level value).
On the wafer 2, as shown in FIG. 2, a plurality of dies (chips) 3 are repeatedly arranged in a matrix in the X and Y directions. As shown in FIG. 3, the surface of the wafer 2 is illustrated by the imaging unit 14 including the one-dimensional TDI sensor 141 in which N imaging pixels P <b> 1 to PN are arranged one-dimensionally in the illustrated “line direction”. It is assumed that an image of the area 101 including the wafers 3a and 3b is acquired by scanning along the “scanning direction”. Hereinafter, the term “scanning direction” used for the one-dimensional TDI sensor 141 in this specification refers to a direction in which an imaging target is scanned by the TDI sensor, and “line direction” refers to a direction orthogonal to the “scanning direction”. .

The same pattern is formed on each die 3a, 3b,. For this reason, when the wafer 2 is imaged, a pattern generated in the area where each die is imaged is used as a unit pattern, and a repeated pattern in which these unit patterns repeatedly appear in the captured image of the wafer 2. Accordingly, the gray level values of the corresponding portions of the captured images of each die are essentially the same value. In the example of FIG. 3, the gray level values of the pixels 111 and 112 corresponding to each other in the captured image obtained by capturing the die 3 a and the die 3 b, that is, the pixels 111 and 112 where the images of the same portion of the die 3 a and the die 3 b are formed are essentially the same. Value.
Therefore, when a difference in gray level values (gray level difference) between corresponding locations (pixels 111 and 112) in the captured images of the two dies that are supposed to be the same is detected, compared to a case where there is no defect in both dies. If one die is defective, the gray level difference becomes large. By detecting such a large gray level difference, defects existing on the die can be detected (die-to-die comparison).

Further, when a repeated pattern such as a memory cell is formed in one die, a defect is detected even if a gray level difference is detected between images obtained by capturing corresponding portions that should be the same in the repeated pattern. Can be detected (cell-to-cell comparison).
In die-to-die comparison, it is common to compare images obtained by imaging two adjacent dies (single detection). This does not tell which die is defective. Therefore, a comparison is further made with dies adjacent to different sides, and when the gray level difference of the same portion becomes larger than the threshold value again, it is determined that the die is defective (double detection). The same applies to the cell-to-cell comparison.

  Returning to FIG. 1, the defect inspection apparatus 1 includes a difference detection unit 22. The difference detection unit 22 is input with gray level signals at locations that are originally identical to each other in the captured image stored in the image storage unit 21. ) Is calculated.

When an output signal of the imaging unit 14 that is a one-dimensional TDI camera is captured while the imaging unit 14 is scanned relative to the wafer 2, a two-dimensional image of the wafer 2 is accumulated in the image storage unit 21.
When performing die-to-die comparison, images of respective portions corresponding to each other of two adjacent dies are extracted from the image storage unit 21 and input to the difference detection unit 22. One of these two images is an inspection image, and the other is a reference image, and the difference detection unit 22 calculates a gray level difference signal between pixels at corresponding positions between the inspection image and the reference image. The gray level difference signal is input to the detection threshold value determination unit 23 and the defect detection unit 24.
In the case of performing cell-to-cell comparison, as the inspection image and the reference image, the images of the portions corresponding to each other in the two adjacent cells are extracted from the image storage unit 21, and the difference detection unit 22 The gray level difference signals between the pixels at the corresponding locations between are calculated.
In order to improve the inspection speed, parallel processing is usually performed in the image comparison processing. In this case, the captured image obtained by imaging the wafer 2 is divided into small partial images of a predetermined size called “frames”, and two frames that are originally identical to each other are used as an inspection image and a reference image, respectively. .

The detection threshold value determination unit 23 determines a detection threshold value by a predetermined statistical process based on the distribution of the gray level difference detected by the difference detection unit 22 and outputs the detection threshold value to the defect detection unit 24. The defect detection unit 24 compares the gray level difference input from the difference detection unit 22 with the detection threshold value determined by the detection threshold value determination unit 23 to detect a defect included in the inspection image. That is, when the gray level difference signal exceeds the detection threshold, the defect detection unit 24 includes one of the inspection image and the reference image including a defect at the pixel position where the gray level difference signal is calculated. Judge.
Then, the defect detection unit 24 creates defect information including information such as the position and size of the detected defect, the gray level difference between the inspection image and the reference image, and the gray level value of these images. Then output.

JP 2004-177397 A

In the above-described pattern defect inspection, a captured image of the sample surface is used. For imaging of the sample surface, an image sensor that includes a plurality of arranged imaging pixels and generates a one-dimensional image signal or a two-dimensional image signal is used. .
In general, there is a variation in the S / N ratio of an output signal between a plurality of imaging pixels provided in one imaging element. This will be described with reference to FIG. The one-dimensional TDI sensor 141 includes N imaging pixels P1 to PN. Similar to the above-described pattern defect inspection, a two-dimensional image obtained by scanning the wafer with the one-dimensional TDI sensor 141 is acquired, and a gray level difference ΔGL between corresponding pixels of the two dies is detected. Is considered to be detected as a defect when the gray level difference ΔGL exceeds the detection threshold.

  Reference numeral 120 shown in FIG. 4 is a histogram of the gray level difference ΔGL between the gray level values obtained by the imaging pixel P2 having a relatively large S / N ratio of the output signal, and reference numeral 121 is the S of the output signal. 10 is a histogram of a gray level difference ΔGL between gray level values obtained by an imaging pixel P (N−1) having a relatively small / N ratio. The histogram 121 of gray level differences detected from the gray level value of the imaging pixel P (N−1) having a relatively small S / N ratio is detected from the gray level value of the imaging pixel P2 having a relatively large S / N ratio. In addition, the distribution is wider than the gray level difference histogram 120, indicating that the noise level of the gray level value of the imaging pixel P (N-1) is larger than the gray level value of the imaging pixel P2.

The detection threshold that is compared with the gray level difference ΔGL in the defect detection process is to prevent a pseudo defect (that is, a defect detected in the defect inspection even though there is no true defect in the portion of the sample) as much as possible. And it is decided not to miss the true defect as much as possible.
Therefore, the detection threshold value compared with the gray level difference ΔGL obtained from the gray level value having a large noise level as shown in the histogram 121 is relatively similar to the detection threshold value Th shown in the figure so that the pseudo defect is not detected as much as possible. The detection threshold, which should be set to a large value and the noise level as shown in the histogram 120 is compared with the gray level difference ΔGL obtained from the small gray level value, is illustrated so as not to miss the true defect as much as possible. It should be set to a relatively small value such as the detection threshold Th ′.

  In the conventional pattern defect inspection, in order to suppress the occurrence of pseudo defects, the detection threshold Th that matches the noise level of the gray level value output from the imaging pixel having a small S / N ratio of the output signal is set to S of the output signal. It has also been used for defect detection in a pixel region where a gray level value can be obtained from a small imaging pixel having a relatively large / N ratio. For this reason, the defect detection sensitivity is sacrificed in the pixel region that should have been able to detect the defect using the detection threshold Th ′ having higher sensitivity.

  In view of the above problems, the present invention uses image signals obtained from the respective imaging pixels in accordance with the S / N ratio of the output signals of the respective imaging pixels constituting the imaging device used for pattern defect detection. An object is to appropriately set the detection sensitivity of defect detection to be performed.

  In order to achieve the above object, in the present invention, a unit that repeatedly appears corresponding to a repetitive pattern in a captured image obtained by imaging the surface of a sample on which a repetitive pattern is formed by an image sensor in which a plurality of image pickup pixels are arranged. A gray level difference that is a difference between gray level values of a pixel in one unit pattern of two of the patterns and a pixel in another unit pattern corresponding to the pixel is detected, and the gray level difference is detected. When the gray level difference exceeds the predetermined detection threshold, the detected gray level difference is detected in the defect detection in which the pixel detecting the gray level difference is detected as a defect candidate. The two gray level values in which the gray level difference is detected are corrected according to the S / N ratio of the output signal of each imaging pixel that has been generated. Alternatively, a predetermined detection threshold value to be compared with the detected gray level difference is corrected according to the S / N ratio of the output signal of each imaging pixel that has generated two gray level values from which the gray level difference has been detected. .

  At this time, the gray level difference is corrected so that the gray level difference detected from the gray level value generated by the imaging pixel having a smaller S / N ratio is further reduced. As a result, the gray level difference detected from the gray level value of the imaging pixel having a relatively small S / N ratio (that is, the noise level is high) is corrected so as to be less likely to exceed the detection threshold. The detection sensitivity is relatively low in a pixel region having a relatively high gray level value. On the other hand, the detection sensitivity is relatively high in a pixel region having a gray level value with a relatively low noise level.

  When correcting the detection threshold, the detection threshold is corrected so as to increase as the detection threshold is compared with the gray level difference detected from the gray level value generated by the imaging pixel having a smaller S / N ratio. This makes it difficult for the gray level difference detected from the gray level value of the imaging pixel having a relatively small S / N ratio (that is, the noise level is high) to exceed the detection threshold value. The detection sensitivity of the pixel area is relatively low. On the contrary, the detection sensitivity of the pixel region having a gray level value with a relatively low noise level is relatively high.

A line sensor such as a one-dimensional TDI sensor or a two-dimensional image sensor such as a two-dimensional CCD sensor may be used as the above-described imaging element in which a plurality of imaging pixels are arranged. In the case of a line sensor, two of the unit patterns that repeatedly appear corresponding to the repetitive pattern, each gray level value of a pixel in one unit pattern and the pixels in the other unit pattern corresponding to this pixel are , Both may be obtained by the same imaging pixel. For example, this occurs when the moving direction of the one-dimensional TDI sensor for scanning the sample and the arrangement direction of the unit patterns are the same.
In such a case, depending on the S / N ratio of the output of the imaging pixel, the gray level difference between the gray level values of each of the two pixels obtained from the imaging pixel or the contrast with the gray level difference is compared. The detection threshold value may be corrected.

Also, two of the unit patterns that repeatedly appear corresponding to the repetitive pattern, and each gray level value of a pixel in one unit pattern and a pixel in the other unit pattern corresponding to this pixel varies depending on different imaging pixels. May be obtained.
In such a case, the gray level value of each of the two pixels respectively obtained from these imaging pixels in accordance with the values such as the average value of the S / N ratio and the mean square value of the outputs of the two different imaging pixels. It is sufficient to correct the gray level difference between the two or the detection threshold value compared with the gray level difference.

  According to the present invention, when there are variations in the S / N ratio of the output signals of a plurality of imaging pixels constituting an imaging device used for pattern defect detection, defect detection is performed using image signals obtained from these imaging pixels, respectively. Detection sensitivity can be appropriately set for each pixel in accordance with the S / N ratio of the output signal of the image pickup pixel, so that the image signal obtained from the image pickup pixel having a large S / N ratio can be obtained. In this portion, it becomes possible to detect a defect with a higher detection sensitivity than before.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. FIG. 5 is a block diagram of the defect inspection apparatus according to the first embodiment of the present invention. The defect inspection apparatus 1 has a configuration similar to the configuration described with reference to FIG. 1, and the same reference numerals are given to the same components and the description of the same functions is omitted.
The imaging unit 14 of the defect inspection apparatus 1 includes the one-dimensional TDI sensor 141 described with reference to FIG. 4, and the one-dimensional TDI sensor 141 is an imaging in which N imaging pixels P1 to PN are arranged on a straight line. It is an element.

  In the defect inspection apparatus 1, the S / N ratio measuring unit 30 that measures the S / N ratio of the output signals of the imaging pixels P1 to PN of the one-dimensional TDI sensor 141 and the imaging pixels P1 to PN are measured. An S / N ratio storage unit 31 for storing the S / N ratio, and a difference detection unit according to the S / N ratios of the output signals of the imaging pixels P1 to PN stored in the S / N ratio storage unit 31. And a gray level difference correction unit 40 that corrects the gray level difference detected by 22.

The gray level difference correction unit 40 includes a correction value calculation unit 42 and a correction parameter determination unit 41.
The correction value calculation unit 42 determines the gray level difference output from the difference detection unit 22 based on the S / N ratio of the output signals of the imaging pixels P1 to PN that have generated the gray level value from which the gray level difference is detected. Correction is performed according to a predetermined correction formula.
The correction parameter determination unit 41 outputs a parameter (correction parameter) that defines a predetermined correction formula used by the correction value calculation unit 42 as an output signal of each of the imaging pixels P1 to PN stored in the S / N ratio storage unit 31. It is determined according to the S / N ratio.

FIG. 6 is an operation flowchart of the defect inspection apparatus 1 shown in FIG.
In step S <b> 11, the S / N ratio measurement unit 30 preliminarily starts the defect detection process for the wafer 2 by the defect inspection apparatus 1, and the signal intensity and noise intensity of a captured image obtained by capturing a predetermined test pattern with the imaging unit 14 in advance. Is detected for each pixel, and the S / N ratio of the output signal of each imaging pixel P1 to PN is measured.
In step S <b> 12, the S / N ratio of the output signal measured for each imaging pixel P <b> 1 to PN by the S / N ratio measurement unit 30 is stored in the S / N ratio storage unit 31.

In step S13, the wafer 2 is scanned by the one-dimensional TDI sensor 141 by relatively moving the imaging unit 14 having the one-dimensional TDI sensor 141 and the wafer 2 as shown in FIG. A two-dimensional image is obtained. When the wafer 2 is scanned by the one-dimensional TDI sensor 141, an image of a band-shaped area surrounded by a dotted line indicated by reference numeral 101 having the same width as the width of the one-dimensional TDI sensor 141 is acquired, and the wafers 3a to 3c are acquired in the band-shaped areas. Is present.
A captured image obtained by capturing the band-like region 101 is indicated by 102. In the captured image 102, unit patterns 3s to 3u appear corresponding to the dies 3a to 3c repeatedly formed on the wafer 2, respectively. The captured image 102 is stored in the image storage unit 21.

  In step S <b> 14, the frames 103 and 104 corresponding to two adjacent unit patterns 3 s and 3 t in the captured image 102 are extracted from the image storage unit 21 and input to the difference detection unit 22. One of these two images is an inspection image and the other is a reference image. The difference detection unit 22 calculates a gray level difference between each pixel 111 included in the inspection image and the corresponding pixel 112 in the reference image.

  As shown in FIG. 7, in this configuration example, the wafer 2 is scanned by the one-dimensional TDI sensor 141 along the arrangement direction of the dies 3a to 3c, and the wafers 3a and 3b arranged along this scanning direction are imaged and acquired. The gray level difference between corresponding pixels between the unit pattern images 3s and 3t thus detected is detected. Accordingly, the gray level values of the two pixels that have detected the gray level difference are acquired from the same imaging pixel, and the noise levels thereof are equal.

The gray level difference is input to the correction value calculation unit 42 of the gray level difference correction unit 40, and the correction value calculation unit 42 corrects the gray level difference in a subsequent step S16 according to a predetermined correction formula. In step S15, therefore, the correction parameter determination unit 41 determines a correction parameter that defines the predetermined correction formula.
As shown in FIG. 8, one imaging pixel P2 of the imaging pixel 141 has a relatively large S / N ratio of its output signal, and another certain imaging pixel P (N-1) has the S / N of its output signal. Assume that the ratio is relatively small. Therefore, the histogram 131 of the gray level difference ΔGL obtained by the difference detection unit 22 detecting the difference between the gray level values obtained from the imaging pixel P2 has a small variance and is obtained from the imaging pixel P (N−1). The histogram 132 of the gray level difference ΔGL obtained by detecting the difference between the gray level values has a large variance.

When the correction parameter determination unit 41 corrects the gray level difference ΔGL in accordance with the predetermined correction formula, the correction is performed for the gray level difference ΔGL detected from the gray level value generated by the imaging pixel having a smaller S / N ratio of the output signal. The correction parameter of the correction formula is determined so as to be further reduced by.
Therefore, the dispersion of the histogram 133 of the corrected gray level difference ΔGL2 obtained by correcting the gray level difference obtained by detecting the difference between the gray level values obtained from the imaging pixel P2 and the imaging pixel P (N−1). The difference between the variance of the histogram 131 and the variance of the histogram 132 is the difference between the variance of the histogram 134 of the corrected gray level difference ΔGL2 obtained by correcting the gray level difference obtained by detecting the difference between the gray level values. Can be made smaller.

  Therefore, as shown in the corrected gray level difference histograms 133 and 134 of FIG. 8, the corrected gray level difference ΔGL2 obtained for the imaging pixels P2 and P (N−1) is compared with the same detection threshold Th and detected. When a pixel whose threshold gray Th exceeds the corrected gray level difference ΔGL2 is detected as a defect candidate, the probability that a pseudo defect occurs in the pixel whose gray level value is obtained by the imaging pixel P2 and the imaging pixel P (N−1). The difference between the probability of occurrence of a pseudo defect in a pixel from which a gray level value is obtained can be made smaller than in the past. This means that the detection sensitivity in the pixel where the gray level value can be obtained by the imaging pixel P2 having a larger S / N ratio of the output signal can be increased as compared with the conventional case.

The predetermined correction formula used by the correction value calculation unit 42 to correct the gray level difference ΔGL obtained by detecting the difference between the gray level values obtained from the imaging pixels Pi (i = 1 to N) is, for example, You may define like following Formula (1).
ΔGL2 = αi × ΔGL + βi (i = 1 to N) (1)
Here, ΔGL2 is the gray level difference after correction, αi and βi are correction parameters determined for each imaging pixel, and the suffix i indicates the pixel number of the imaging pixels P1 to PN.
Here, the correction parameters αi and βi are determined as a function of the S / N ratio (SN) of the output signal of the imaging pixel Pi. Examples of graphs of functions for determining the correction parameters αi and βi are shown in FIGS. 9A and 9B. When the S / N ratio (SN) of the output signal of the imaging pixel Pi is smaller, the correction parameter αi of this example is set so that the reduction amount from the gray level difference ΔGL before correction to the gray level difference ΔGL2 after correction becomes large. And βi are defined to be monotonically increasing functions with respect to the S / N ratio (SN) in the domain.

  In step S16, the correction value calculation unit 42 corrects the gray level difference ΔGL detected by the difference detection unit 22 in step S14 by the equation (1) defined by the correction parameters αi and βi determined in step S15. A gray level difference ΔGL2 is determined. The corrected gray level difference ΔGL2 is input to the detection threshold value determination unit 23 and the defect detection unit 24.

  In step S <b> 17, the detection threshold value determination unit 23 performs predetermined statistical processing on each corrected gray level difference obtained by correcting each gray level difference detected between each pixel of the inspection image for one frame and each pixel of the reference image. The detection threshold is determined and output to the defect detection unit 24. FIG. 10 is a flowchart for explaining the processing content of the detection threshold value determination step S17 of FIG. 6, and FIG. 11 is an explanatory diagram of the detection threshold value determination process shown in FIG. Note that the detection threshold value determination process described below with reference to FIGS. 10 and 11 is merely an example, and the detection threshold value determination process used in the present invention is not limited to this.

In step S21, the detection threshold value determination unit 23 inputs the gray level difference, and in step S22, a histogram of the gray level difference is created as shown in FIG. In step S23, the detection threshold value determination unit 23 calculates the cumulative frequency of gray level differences from this histogram.
In step S24, the detection threshold value determination unit 23 assumes that the input gray level difference follows a predetermined distribution, and the cumulative frequency calculated in step S23 so that the cumulative frequency is linearly related to the gray level difference. Convert. At this time, the cumulative frequency is converted on the assumption that the gray level difference follows a certain distribution such as normal distribution, Poisson distribution, or chi-square distribution. This conversion cumulative frequency is shown in FIG.

In step S25, an approximate line (y = ax + b) indicating the relationship between the gray level difference and the conversion cumulative frequency is derived according to the conversion cumulative frequency converted in step S24 (see (C) in FIG. 11).
In step S26, the detection threshold Th is determined from the parameters a and b of the approximate line and the sensitivity setting parameter (fixed value).

Here, VOP and HO are set as fixed sensitivity setting parameters in the approximate straight line between the gray level difference and the conversion cumulative frequency, and the cumulative frequency P1 corresponding to the cumulative probability (p) (p is multiplied by the number of samples). )), And a gray level difference obtained by moving VOP in the vertical axis direction and HO in the horizontal axis direction from that point is set as a detection threshold Th. Therefore, the detection threshold Th is a predetermined calculation formula Th = (P1−b + VOP) / a + HO (2)
Is calculated by

  In step S18, the defect detection unit 24 compares the correction gray level difference ΔGL2 output from the gray level difference correction unit 40 with the detection threshold value determined by the detection threshold value determination unit 23 to detect the correction gray level difference signal ΔGL2. If the threshold value is exceeded, it is determined that either the inspection image or the reference image includes a defect at the pixel position where such a corrected gray level difference ΔGL2 is detected.

  In the above description, the image sensor included in the imaging unit 14 of the defect inspection apparatus 1 is a line sensor, and the gray level values of two pixels whose difference detection unit 22 detects a gray level difference are acquired from the same imaging pixel. Explained the case. However, the imaging unit 14 may employ a two-dimensional image sensor as the image sensor. In such a case, the difference detection unit 22 may be generated by imaging pixels having different gray level values of the two pixels that detect the gray level difference.

Moreover, even if the image sensor included in the image capturing unit 14 is a line sensor, the difference detection unit 22 may generate two image pixels having different gray level values for detecting a gray level difference. This is because, for example, when the scanning direction of the wafer 2 by the imaging unit 14 is different from the arrangement direction of the dies 3 or the corresponding pixels in which the gray level difference is detected by the difference detection unit 22 as exemplified below are captured. This also occurs when they are not arranged in the same direction as the scanning direction by the unit 14.
For example, as shown in FIG. 12, by moving the imaging unit 14 having the one-dimensional TDI sensor 141 and the wafer 2 relative to each other, the imaging unit 14 scans the wafer 2 and surrounds it with a dotted line denoted by reference numeral 101. Consider a case where an image of a band-like region is acquired. Wafers 3a to 3d are present in the belt-like region 101. A captured image obtained by capturing the band-like region 101 is indicated by 102. In the captured image 102, unit patterns 3s to 3v appear corresponding to the dies 3a to 3d repeatedly formed on the wafer 2, respectively.
Here, instead of the two dies 3a and 3b arranged in the scanning direction, the gray of the corresponding pixels 111 and 112 of the two unit patterns 3s and 3v obtained by imaging the two dies 3a and 3d arranged in the direction orthogonal to the scanning direction. If the inspection procedure is determined so that the difference detection unit 22 detects the level difference, the imaging pixels for obtaining the gray level values of the pixels 111 and 112 are different.

The imaging device of the imaging unit 14 includes a plurality of arranged imaging pixels P1 to PN, and the difference detection unit 22 detects the gray level difference ΔGL. The imaging pixels Pi and Pj (i, i, j = 1 to N), a predetermined correction formula used by the correction value calculation unit 42 to correct such a gray level difference ΔGL is defined as the following formula (3), for example. You can do it.
ΔGL2 = αij × ΔGL + βij (i, j = 1 to N) (3)
Here, ΔGL2 is a gray level difference after correction, αij and βij are correction parameters determined for each imaging pixel, and suffixes i and j indicate pixel numbers of the imaging pixels P1 to PN.

  Here, the correction parameters αij and βij are the average value ((SNi + SNj) / 2) of the S / N ratio (SNi and SNj) of the output signals of the imaging pixels Pi and Pj, the mean square value,

  As a function of Then, when the S / N ratio (SNi and SNj) of the output signals of the imaging pixels Pi and Pj is smaller, the correction is made so that the amount of reduction from the gray level difference ΔGL before correction to the gray level difference ΔGL2 after correction becomes large. The parameters αij and βij may be defined so as to be a monotonically increasing function with respect to the average value or the mean square value of the S / N ratio in the domain of definition.

  FIG. 13 is a block diagram of a defect inspection apparatus according to the second embodiment of the present invention. If the S / N ratio of the output signal of the imaging pixel Pi (i = 1 to N) does not change, it is not necessary to change the correction parameters αi, βi, αij, and βij even if the type of the wafer 2 to be inspected changes. Therefore, in the second embodiment shown in FIG. 13, when the S / N ratio measurement unit 30 measures the S / N ratio of the output signal of the imaging pixel Pi (i = 1 to N) of the imaging unit 14 in advance before the inspection. The correction parameter determination unit 41 determines correction parameters αi, βi, αij, and βij. Then, the determined correction parameters αi, βi, αij, and βij are stored in the correction parameter storage unit 43, and are read from the correction parameter storage unit 43 for use in later inspection.

  When the difference detection unit 22 detects a gray level difference between gray level values acquired from different imaging pixels, the two pixels from which the gray level difference is detected by changing the type of the wafer 2 to be inspected. The combination of the pixel numbers i and j of the imaging pixels that generate the gray level value changes. Therefore, the correction parameters αij and βij may be determined in advance for all combinations of pixel numbers i and j and stored in the correction parameter storage unit 43.

  FIG. 14 is an operation flowchart of the defect inspection apparatus 1 shown in FIG. The same reference numerals are given to the same processing steps as those in the operation flowchart shown in FIG. In step S11, after the S / N ratio measurement unit 30 measures the S / N ratio of the output signals of the imaging pixels P1 to PN, in step S31, the correction parameter determination unit 41 determines based on the measured S / N ratio. Correction parameters αi, βi, αij and βij are determined. The correction parameter determination method itself is the same as the determination method performed in step S15 of FIG. In step S32, the correction parameters αi, βi, αij and βij determined in step S31 are stored in the correction parameter storage unit 43.

  Thereafter, steps S13, S14, and S16 to S18 are executed in the same manner as steps S13, S14, and S16 to S18 shown in FIG. However, in step S16, the correction value calculation unit 42 determines the correction gray level difference ΔGL2 by a correction formula defined by the correction parameters αi and βi, αij and βij stored in the correction parameter storage unit 43 in step S32.

FIG. 15 is a block diagram of a defect inspection apparatus according to a third embodiment of the present invention. The defect inspection apparatus 1 also has a configuration similar to the configuration described with reference to FIG. 1, and the same reference numerals are given to the same components and the description of the same functions is omitted.
In the defect inspection apparatus 1 according to the present embodiment, the S / N ratio measuring unit 30 that measures the S / N ratio of the output signal of each imaging pixel of the imaging device included in the imaging unit 14 and the S that is measured for each imaging pixel. The S / N ratio storage unit 31 that stores the / N ratio, and the detection threshold value determination unit 23 determines according to the S / N ratio of the output signal of each imaging pixel stored in the S / N ratio storage unit 31. A detection threshold value correction unit 50 that corrects the detected threshold value is provided. The functions of the S / N ratio measurement unit 30 and the S / N ratio storage unit 31 may be the same as those of the embodiment shown in FIGS.

The detection threshold correction unit 50 includes a correction value calculation unit 52 that calculates a correction detection threshold obtained by correcting the detection threshold determined by the detection threshold determination unit 23, and a correction parameter determination unit 51.
The correction value calculation unit 52 corrects the detection threshold value determined by the detection threshold value determination unit 23 for each gray level difference sequentially output by the difference detection unit 22, and the gray level difference and the defect detection unit 24 respectively correct the detection threshold value. A corrected detection threshold value to be compared is calculated. The correction value calculation unit 52 performs a predetermined correction on the detection threshold value determined by the detection threshold value determination unit 23 based on the S / N ratio of the output signal of the imaging pixel that has generated the gray level value that detects each gray level difference. By correcting according to the equation, each correction detection threshold value to be compared with each gray level difference is calculated.
The correction parameter determination unit 51 sets a parameter (correction parameter) that defines a predetermined correction formula used by the correction value calculation unit 52 as the S / N of each output signal of each imaging pixel stored in the S / N ratio storage unit 31. Determined according to N ratio.

FIG. 16 is an operation flowchart of the defect inspection apparatus 1 shown in FIG.
In step S11, the S / N ratio measurement unit 30 measures the S / N ratio of each output signal of the plurality of imaging pixels provided in the imaging device of the imaging unit 14, and in step S12, the measured output signal The S / N ratio is stored in the S / N ratio storage unit 31.
In step S <b> 13, the imaging unit 14 and the wafer 2 are relatively moved, whereby the imaging unit 14 scans the wafer 2 to obtain a two-dimensional image of the surface of the wafer 2. In step S14, the difference detection unit 22 calculates a gray level difference between two corresponding pixels.
In step S <b> 17, the detection threshold value determination unit 23 performs predetermined statistical processing on each gray level difference detected between each pixel of the inspection image for one frame and each pixel of the reference image to determine a detection threshold value. . The processes in steps S11 to S14 and S17 are the same as the processes described above with reference to FIG.

  Now, it is assumed that the imaging unit 14 includes a one-dimensional TDI sensor 141 as shown in FIG. 4, and in step S13, as shown in FIG. 7, the one-dimensional TDI sensor 141 follows the arrangement direction of the dies 3a to 3c. It is assumed that the image is acquired by scanning the wafer 2. Further, the gray level difference obtained in step S14 is obtained by detecting a difference between corresponding pixels between the unit pattern images 3s and 3t obtained by imaging the wafers 3a and 3b arranged along the scanning direction. Shall be. In this case, the gray level values of the two pixels that have detected the gray level difference are acquired from the same imaging pixel and the noise levels thereof are equal.

  As shown in FIG. 17, an imaging pixel P2 of the imaging pixel 141 has a relatively large S / N ratio of its output signal, and another imaging pixel P (N-1) has an S / N of its output signal. Assume that the ratio is relatively small. The dispersion of the gray level difference ΔGL obtained by detecting the difference between the gray level values obtained from the imaging pixel P2 is small, and the difference between the gray level values obtained from the imaging pixel P (N−1) is small. The variance of the histogram 141 of the gray level difference ΔGL obtained by detection increases.

The correction value calculation unit 52 corrects the detection threshold value determined by the detection threshold value determination unit 23 according to a predetermined correction formula in the subsequent step S42. In step S41, the correction parameter determining unit 51 determines a correction parameter that defines the predetermined correction formula.
When the correction parameter determination unit 51 corrects the detection threshold according to the predetermined correction formula, the detection threshold compared with the gray level difference ΔGL detected from the gray level value generated by the imaging pixel having a relatively small S / N ratio. The correction parameter of the correction formula is determined so as to be further increased by the correction.

Accordingly, as shown in histograms 140 and 141 in FIG. 17, a detection threshold Th1 that is compared with a gray level difference obtained by detecting a difference between gray level values obtained from the imaging pixel P2 having a relatively large S / N ratio. Is smaller than the detection threshold Th2 compared with the gray level difference obtained by detecting the difference between the gray level values obtained from the imaging pixel P (N-1) having a relatively small S / N ratio. Since the correction is performed, the defect detection sensitivity is relatively high in the pixel region where the gray level value obtained from the imaging pixel P2 is obtained.
Therefore, conventionally, the defect detection is detected with the same detection sensitivity as the detection sensitivity in the pixel in which the gray level value is obtained by the imaging pixel P (N-1) having a relatively small S / N ratio, and the gray level is detected by the imaging pixel P2. In a pixel from which a value is obtained, defect detection is performed with higher detection sensitivity than in the past.

The correction value calculation unit 52 uses the correction threshold Th to be compared with the gray level difference ΔGL obtained by detecting the difference between the gray level values obtained from the imaging pixel Pi (i = 1 to N). For example, the predetermined correction formula may be defined as the following formula (4).
Th2 = γi × Th + δi (i = 1 to N) (4)
Here, Th2 is a detection threshold value after correction, γi and δi are correction parameters determined for each imaging pixel, and the suffix i indicates the pixel number of the imaging pixels P1 to PN.

  Here, the correction parameters γi and δi are determined as a function of the S / N ratio (SN) of the output signal of the imaging pixel Pi. Examples of graphs of functions that determine the correction parameters γi and δi are shown in FIGS. 18A and 18B. When the S / N ratio (SN) of the output signal of the imaging pixel Pi is smaller, the correction parameters γi and δi of this example are set so that the amount of increase from the detection threshold Th before correction to the detection threshold Th2 after correction increases. Is defined to be a monotonically decreasing function with respect to the S / N ratio (SN) in its domain.

In step S42, the correction value calculation unit 52 corrects the detection threshold Th determined by the detection threshold determination unit 23 in step S17 by the equation (4) defined by the correction parameters γi and δi determined in step S41. A detection threshold Th2 is determined. The correction detection threshold Th2 is input to the defect detection unit 24.
In step S43, the defect detection unit 24 compares the gray level difference ΔGL output from the difference detection unit 22 with the correction detection threshold Th2 output from the correction value calculation unit 52, and the gray level difference GL is corrected and detected. When the threshold Th2 is exceeded, it is determined that either the inspection image or the reference image includes a defect at the pixel position where such a gray level difference GL is detected.

In the defect detection apparatus 1 according to the third embodiment and the fourth embodiment described below, the imaging element of the imaging unit 14 includes a plurality of imaging pixels P1 to PN arranged, and the difference detection unit 22 determines the gray level difference ΔGL. When the gray level values of the two pixels to be detected are generated by different imaging pixels Pi and Pj (i, j = 1 to N), the correction value calculation unit 52 compares with such a gray level difference ΔGL. A predetermined correction formula used for correcting the detection threshold Th may be defined as the following formula (5), for example.
Th2 = γij × Th + δij (i, j = 1 to N) (5)
Th2 is a detection threshold value after correction, γij and δij are correction parameters determined for each image pickup pixel, and suffixes i and j indicate pixel numbers of the image pickup pixels P1 to PN.

  Here, the correction parameters γij and δij are the average values ((SNi + SNj) / 2) of the S / N ratios (SNi and SNj) of the output signals of the imaging pixels Pi and Pj, the mean square value,

  As a function of When the S / N ratio (SNi and SNj) of the output signals of the imaging pixels Pi and Pj is smaller, the correction parameter γij is set so that the amount of increase from the detection threshold Th before correction to the detection threshold Th2 after correction increases. And δij may be defined so as to be a monotonically decreasing function with respect to the average value or the mean square value of the S / N ratio in the domain of definition.

FIG. 19 is a block diagram of a defect inspection apparatus according to a fourth embodiment of the present invention. In the fourth embodiment shown in FIG. 19, when the S / N ratio measurement unit 30 measures the S / N ratio of the output signal of the imaging pixel Pi (i = 1 to N) of the imaging unit 14 in advance before the inspection, The correction parameter determination unit 51 determines correction parameters γi, δi, γij, and δij. Then, the determined correction parameters γi, δi, γij, and δij are stored in the correction parameter storage unit 53, and are read from the correction parameter storage unit 53 and used in the subsequent inspection.
When the difference detection unit 22 detects a gray level difference between gray level values acquired from different imaging pixels, correction parameters γij and δij are determined in advance for all combinations of pixel numbers i and j. You may memorize | store in the correction parameter memory | storage part 53. FIG.

  FIG. 20 is an operation flowchart of the defect inspection apparatus 1 shown in FIG. The same reference numerals are assigned to the same processing steps as those in the operation flowchart shown in FIG. In step S11, after the S / N ratio measurement unit 30 measures the S / N ratio of the output signals of the imaging pixels P1 to PN, in step S51, the correction parameter determination unit 51 determines based on the measured S / N ratio. Correction parameters γi, δi, γij and δij are determined. The correction parameter determination method itself is the same as the determination method performed in step S41 of FIG. In step S52, the correction parameters γi, δi, γij and δij determined in step S51 are stored in the correction parameter storage unit 53.

  Thereafter, steps S13, S14, S17, S42 and S43 are executed in the same manner as S13, S14, S17, S42 and S43 shown in FIG. However, in step S42, the correction value calculation unit 52 determines the correction detection threshold Th2 using a correction formula defined by the correction parameters γi, δi, γij, and δij stored in the correction parameter storage unit 53 in step S52.

The present invention relates to a defect inspection apparatus and a defect inspection method for comparing a reference image with an inspection image obtained by imaging the same pattern to be inspected and detecting a different portion as a defect of the pattern. Is available.
Preferably, a pattern formed on the surface of a sample such as a semiconductor wafer, a photomask substrate, a liquid crystal display panel substrate, or a liquid crystal device substrate is imaged, and the image obtained thereby is inspected on the sample surface. It can be used for pattern defect inspection to detect existing defects.

It is a block diagram of the conventional defect inspection apparatus. It is a figure which shows the arrangement | sequence of the die | dye on a semiconductor wafer. It is a figure which shows two corresponding pixels which detect a gray level difference in a defect detection process. It is explanatory drawing of the image pick-up element which has an imaging pixel from which S / N ratio of an output signal differs. 1 is a block diagram of a defect inspection apparatus according to a first embodiment of the present invention. It is an operation | movement flowchart of the defect inspection apparatus shown in FIG. 6 is a diagram illustrating a first example of an image captured by the imaging unit 14 on the wafer 2. FIG. It is explanatory drawing of the correction | amendment of the gray level difference by the gray level difference correction part 40. FIG. (A) is a graph showing an example of the correction parameter αi, and (B) is a graph showing an example of the correction parameter βi. It is a flowchart explaining a detection threshold value determination process. (A)-(C) is explanatory drawing of the detection threshold value determination process shown in FIG. 6 is a diagram illustrating a second example of an image captured by the imaging unit 14 on the wafer 2. FIG. It is a block diagram of the defect inspection apparatus by 2nd Example of this invention. It is an operation | movement flowchart of the defect inspection apparatus shown in FIG. It is a block diagram of the defect inspection apparatus by 3rd Example of this invention. It is an operation | movement flowchart of the defect inspection apparatus shown in FIG. It is explanatory drawing of correction | amendment of the detection threshold value by the detection threshold value correction | amendment part 50. FIG. (A) is a graph showing an example of the correction parameter γi, and (B) is a graph showing an example of the correction parameter δi. It is a block diagram of the defect inspection apparatus by 4th Example of this invention. It is an operation | movement flowchart of the defect inspection apparatus shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect detection apparatus 2 Wafer 3, 3a-3d Die 3s-3v Unit pattern 102 Captured image 111, 112 pixel 141 Image pick-up element P1-PN Imaging pixel

Claims (10)

An imaging unit including an imaging element in which a plurality of imaging pixels are arranged;
Two of the unit patterns that repeatedly appear corresponding to the repeated pattern in the captured image acquired by imaging the surface of the sample on which the repeated pattern is formed by the imaging unit, and a pixel in one of the unit patterns A gray level difference detection unit that detects a gray level difference that is a difference between respective gray level values of pixels in another unit pattern corresponding to the pixel;
The gray level difference detected by the gray level difference detection unit is corrected according to the S / N ratio of the output signal of each imaging pixel that has generated the two gray level values from which the gray level difference has been detected. A gray level difference correction unit to perform,
A defect detection unit that determines that a defect candidate exists in a portion indicated by a pixel in which the gray level difference is detected when the gray level difference corrected by the gray level difference correction unit exceeds a predetermined detection threshold. ,
The gray level difference correction unit corrects the gray level difference so that a gray level difference detected from a gray level value generated by an imaging pixel having a smaller S / N ratio is further reduced. Defect detection device. An imaging unit including an imaging element in which a plurality of imaging pixels are arranged;
Two of the unit patterns that repeatedly appear corresponding to the repeated pattern in the captured image acquired by imaging the surface of the sample on which the repeated pattern is formed by the imaging unit, and a pixel in one of the unit patterns A gray level difference detection unit that detects a gray level difference that is a difference between respective gray level values of pixels in another unit pattern corresponding to the pixel;
A defect detection unit that compares the gray level difference with a predetermined detection threshold and determines that a defect candidate exists in a portion indicated by a pixel in which the gray level difference is detected when the gray level difference exceeds the predetermined detection threshold When,
The predetermined detection threshold value to be compared with each gray level difference detected by the gray level difference detection unit, and each of the imaging pixels that have generated the two gray level values from which each gray level difference has been detected. A detection threshold value correction unit for correcting the output signal according to the S / N ratio of the output signal,
The detection threshold correction unit corrects the detection threshold so that a detection threshold compared with a gray level difference detected from a gray level value generated by an imaging pixel having a smaller S / N ratio increases. A defect detection apparatus characterized by the above. The image sensor is a line sensor that outputs a one-dimensional image signal,
The defect detection apparatus according to claim 1, wherein the gray level difference detection unit detects a difference between two gray level values generated by the same imaging pixel as the gray level difference. The image sensor is a line sensor that outputs a one-dimensional image signal,
The defect detection device according to claim 1, wherein the gray level difference detection unit detects a difference between two gray level values respectively generated by different imaging pixels as the gray level difference. The image sensor is a two-dimensional image sensor that outputs a two-dimensional image signal,
The defect detection device according to claim 1, wherein the gray level difference detection unit detects a difference between two gray level values respectively generated by different imaging pixels as the gray level difference. The surface of the sample on which the repetitive pattern is formed is imaged by an imaging device in which a plurality of imaging pixels are arranged,
Two of the unit patterns that repeatedly appear corresponding to the repeated pattern in the captured image acquired by imaging with the imaging element, a pixel in one unit pattern, and a pixel in another unit pattern corresponding to this pixel The gray level difference that is the difference between the gray level values of
Correcting the detected gray level difference according to the S / N ratio of the output signal of each imaging pixel that has generated the two gray level values from which the gray level difference has been detected,
When the corrected gray level difference exceeds a predetermined detection threshold, it is determined that a defect candidate exists in a portion indicated by the pixel from which the gray level difference is detected,
When correcting the gray level difference in accordance with the S / N ratio, the gray level difference detected by the gray level value generated from the gray level value generated by the imaging pixel having the smaller S / N ratio is further reduced. A defect detection method comprising correcting a level difference. The surface of the sample on which the repetitive pattern is formed is imaged by an imaging device in which a plurality of imaging pixels are arranged,
Two of the unit patterns that repeatedly appear corresponding to the repeated pattern in the captured image acquired by imaging with the imaging element, a pixel in one unit pattern, and a pixel in another unit pattern corresponding to this pixel The gray level difference that is the difference between the gray level values of
The gray level difference is compared with a predetermined detection threshold, and when the gray level difference exceeds the predetermined detection threshold, it is determined that a defect candidate exists in a portion indicated by a pixel where the gray level difference is detected,
The predetermined detection threshold value to be compared with each gray level difference is set according to the S / N ratio of the output signal of each imaging pixel that has generated the two gray level values from which each gray level difference is detected. A defect detection method characterized by determining a detection threshold value to be compared with a gray level difference detected from a gray level value generated by an imaging pixel having a smaller S / N ratio. The image sensor is a line sensor that outputs a one-dimensional image signal,
The defect detection method according to claim 6 or 7, wherein a difference between two gray level values generated by the same imaging pixel is detected as the gray level difference. The image sensor is a line sensor that outputs a one-dimensional image signal,
8. The defect detection method according to claim 6, wherein a difference between two gray level values respectively generated by different imaging pixels is detected as the gray level difference. The image sensor is a two-dimensional image sensor that outputs a two-dimensional image signal,
8. The defect detection method according to claim 6, wherein a difference between two gray level values respectively generated by different imaging pixels is detected as the gray level difference.

Images