JP2001148017A - Device for inspecting substrate - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress the deterioration of the signal quality of each pixel data due to a phase shift and to detect a minute foreign matter with respect to the pixel size with high accuracy even when the relation between the pixel size of a photodetector and the repeating pitch of an electrode pattern is not integer multiple and a phase shift exists. SOLUTION: When a pattern pitch of 300 μm and an image size of 14 μm, the number of pattern-corresponding pixels obtained by dividing the pattern pitch by the image size is about 21.4. When the number of pixels are multiplied by 5 as a 1st integer number, the value 107 is obtained. Therefore, the relative positional relation between a pseudo pattern and the pixels becomes almost the same phase by forming the pseudo pattern with five patterns. A difference data preparing means 30 uses pixels adjacent to each other of the pseudo pattern and prepares difference data V. A threshold preparing means 50 prepares a threshold T at a time when a defect is detected from the data V. Defect detecting means 80 and 90 detect a defect by comparing the data V with the value T.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a substrate inspection apparatus for detecting whether or not a foreign substance has adhered to a substrate having a predetermined repetitive pattern, for example, a glass substrate in a manufacturing process of a liquid crystal display. The present invention relates to a substrate inspection apparatus in which correction of pixel data at the time of detecting a defect is improved.

[0002]

2. Description of the Related Art A liquid crystal display (LCD: Liqui)
d Crystal Display) is a CRT (C
Since it can be made thinner and lighter than an Anode Ray Tube, a CTV (Color Tele Tube) can be used.
vision) and OA equipment, etc., and the screen size is increased to 10 inches or more.
Higher definition is being promoted. The liquid crystal display has TN (Twisted Nematic)
Type, STN (Super Twisted Neati)
c) type and TFT (Thin Film Trans)
isor) type. TFT-type liquid crystal displays use a TF patterned on each pixel electrode substrate.
T is provided.

[0003] A substrate inspection apparatus detects whether or not a minute foreign substance exists on the pixel electrode substrate. A conventional substrate inspection apparatus irradiates a laser beam from obliquely above a glass substrate provided with an electrode pattern, observes scattered light of foreign matter from above, and detects the presence of foreign matter. The substrate inspection apparatus detects the presence of foreign matter and various defects by utilizing the fact that an electrode pattern or the like formed on a substrate has a repetitive property (periodicity). FIG. 1 is a diagram showing a positional relationship among a CCD light receiving element 10, an imaging lens 11, and a substrate 12 when a glass substrate inspection is performed by a pattern comparison method. As is clear from the figure, the electrode pattern on the substrate 12 forms an image on the CCD light receiving element 10 via the imaging lens 11. In the pattern comparison method, since electrode patterns (TFT patterns and wiring patterns) provided on a substrate are repeatedly arranged, attention is paid to the periodicity of the patterns, and adjacent electrode patterns are sequentially compared with each other. ,
This is a method of detecting foreign matter and various defects based on the difference image (shading information).

FIG. 2 is a diagram showing a schematic configuration of a conventional substrate inspection apparatus. The CCD light receiving element 10 is
It is about 0 pixels, and the size per pixel is about 1
4 [μm]. The CCD light receiving element 10 receives scattered light and reflected light from the substrate 12. The image data output from the CCD light receiving element 10 is converted into digital data (shading information) of about 8 bits or 10 bits by the A / D converter 20 and taken into the FIFO memory 22.
At this time, the inspection stage (not shown) is about 25 [μm].
As it moves, the image data of the next area is taken into the FIFO memory 22 from the CCD light receiving element 10. At the same time, the first image data f (x) of the previous area is output from the FIFO memory 22, and the second image data g (x) delayed by the pattern pitch is output from the delay circuit (shift register) 24 at the next stage. Sub-pixel correction circuit 2
6 is output.

Generally, the pattern pitch P [μm] is C
The size a of one pixel of the CD light receiving element 10 (about 14 [μ
m]), but having a phase shift ε [μm]. Therefore, the absolute value of the difference between the first image data f (x) and the second image data g (x) is equal to this phase shift ε.
, A noise having a difference signal level ΔV. Therefore, there is a problem that it is difficult to detect a defect having the same level as the difference signal level ΔV. Therefore, conventionally, a process called sub-pixel correction is performed to obtain a phase shift ε.
To reduce the difference signal level ΔV. That is, as shown in FIG.
An edge of the pattern pitch P [μm] is located at the first pixel, which becomes a phase shift ε [μm]. That is, C
Assuming that the size of one pixel of the CD light receiving element 10 is a and the number of pixels (integer) corresponding to the pattern pitch P is j, the pattern pitch P is expressed as P = a × j + ε.

The sub-pixel correction circuit 26 corrects the luminance data of each pixel from the CCD light receiving element 10, that is, the image data f (x) and g (x) in consideration of the phase shift ε [μm]. . As a result, the aforementioned difference signal level ΔV is reduced. That is, the sub-pixel correction circuit 26 outputs the first image data f from the FIFO memory 22.
(X) and the second image data g (x) from the delay circuit (shift register) 24, respectively, and perform a predetermined calculation (sub-pixel correction) process on them. That is, for the i-th first image data f (i), the i-th first image data f (i-1) and f (i + 1) are used as in the following equation using the first and second image data f (i-1) and f (i + 1). The luminance data F (i) of one image data f (i) is obtained. F (i) = f (i) + (e / 2) × ｛f (i−1) −f
(I + 1)｝ The same phase correction is performed on the second image data g (i). However, in order to reverse the correction direction, the first image data f
The sign is partially reversed from (i). G (i) = g (i) − (e / 2) × ｛g (i−1) −g
(I + 1)｝ Here, e is k × (ε / a). Here, k is a coefficient of 1 or less determined by the characteristics of the lens.

[0007] The above equation represents the first image data f (x) from the FIFO memory 22 corresponding to each pixel of the CCD light receiving element 10.
And the second image data g (x) from the delay circuit (shift register) 24 (arithmetic processing), thereby performing sub-pixel correction, and correcting the first image data F
(X) and the second image data G (x) are used as a difference image calculation circuit 28.
Is output to By this sub-pixel correction processing, the influence of the difference signal level ΔV is eliminated, and an accurate value can be obtained as the comparison difference signal of the adjacent pattern.

[0008] The difference image operation circuit 28 is a first image data F
The absolute value V of the difference between (x) and the second image data G (x) is obtained and output to the defect determination circuit 29. The defect determination circuit 29 makes a defect determination according to whether the absolute value V of the difference from the difference image calculation circuit 28 is larger than a threshold value. The absolute value V of this difference is a comparison difference signal between adjacent patterns, and is normally zero under ideal conditions. However, if the absolute value of the difference is larger than the threshold value, it is determined to be a defect, and it is possible to determine whether it is a pinhole defect or a pattern defect based on the sign change of the difference.

[0009]

In the above-described sub-pixel correction, the luminance data of the i-th image data f (i) is obtained by using the preceding and succeeding image data f (i-1) and f (i + 1). This is an effective correction processing method when the values of these three consecutive pixel data change smoothly. However, if the ratio of signal change due to foreign matter to be detected with respect to the pixel size is large,
Conversely, when the sub-pixel correction is performed, there is a problem that the signal is blurred, noise or the like is generated, the S / N is deteriorated, and the signal quality of the pixel data is reduced.

The present invention has been made in view of the above points, and when the relationship between the pixel size of the light receiving element and the repetition pitch of the electrode pattern is not an integral multiple but has a phase shift, each phase shift is caused by this phase shift. It is an object of the present invention to provide a substrate inspection apparatus capable of suppressing a decrease in signal quality of pixel data and detecting a minute foreign substance with respect to a pixel size with high accuracy.

[0011]

According to a first aspect of the present invention, there is provided a substrate inspection apparatus, comprising: a lens system for forming an image of a pattern having a predetermined pitch formed on a substrate; Light-receiving element means composed of a plurality of pixels that receive and receive light, and differential data between pixels such that the relative positional relationship between the pixel and the pattern formed on the light-receiving element is substantially in phase. Data creating means for sequentially creating and outputting the data, threshold creating means for creating a threshold based on the difference data output from the difference data creating means, and output from the difference data creating means Defect detecting means for detecting a defect by comparing the difference data with a threshold value created by the threshold value creating means. Even if the relationship between the pixel size of the light receiving element means and the pattern repetition pitch is not an integral multiple and has a phase shift, the pixel size of the light receiving element means can be set to an integer for a plurality of patterns by using a plurality of patterns And the relative positional relationship between the pattern and the pixel is substantially in phase. Therefore, in the present invention, instead of detecting a defect based on a comparison difference signal between adjacent patterns as in the related art, a pixel in which the relative positional relationship between the pattern and the pixel is substantially in phase Difference data between them is created by difference data creation means. Then, a threshold value at the time of detecting a defect is created from the difference data by using a threshold value creating means, and the defect detection means compares the difference data with the threshold value to detect a defect.
As a result, the signal quality of each pixel data is prevented from deteriorating due to the phase shift, and it is possible to detect minute foreign matter with respect to the pixel size with high accuracy.

According to a second aspect of the present invention, in the substrate inspection apparatus according to the first aspect, the difference data creating unit is configured to reduce the number of pixels corresponding to the pattern obtained by dividing a pitch of the pattern by a size of the pixel. A set of the number of the patterns corresponding to the first integer when the value of the number of pixels becomes the value closest to the integer by multiplying by the smallest arbitrary first integer as one pseudo pattern, The difference data is sequentially created by using adjacent patterns. This specifically describes how the difference data creating means determines pixels such that the relative positional relationship between the pattern and the pixels is substantially in phase. For example, when the pattern pitch is 300 μm and the pixel size is 14 μm
m], the number of pixels corresponding to the pattern obtained by dividing the pattern pitch by the pixel size is about 2
1.4. When the number of pixels is multiplied by 5 as a first integer, the value becomes 107, so that a pseudo pattern is formed by five patterns. The difference data creation means creates difference data using pixels of adjacent ones of the pseudo pattern.

According to a third aspect of the present invention, in the substrate inspection apparatus according to the first or second aspect, each of the pixels of the light receiving element is divided by dividing the lens system into a plurality of blocks. Divided according to the block, and among the difference data output from the difference data creating means, a plurality of pieces of data corresponding to the divided pixels of the light receiving element means are used as calculation difference data, and the calculation difference data is A variance value is obtained based on the variance value, and the threshold value is created based on the variance value. This is a specific example of how the threshold value creating means creates the threshold value. Making the threshold value common to all the pixels of the light receiving element means increases the influence of a magnification error on the light receiving element surface due to aberration of the lens system. The data is divided into blocks, and a variance value of the difference data is obtained for each block, and a threshold value is created based on the variance value.

According to a fourth aspect of the present invention, in the substrate inspection apparatus according to the third aspect, the threshold value generating means multiplies the variance value by a predetermined coefficient, and multiplies the multiplied value by an average value of the differential data for calculation. And the threshold value is created by adding the laser power distribution correction value. This is a more specific example of how the threshold value creating means creates the threshold value.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to the drawings. FIG. 4 is a diagram showing a schematic configuration of the substrate inspection apparatus according to the present invention. The difference between the substrate inspection apparatus according to the embodiment of the present invention and that shown in FIG. 2 is that the relative data relationship between the pixels of the CCD light receiving element 10 and the repetition pitch of the electrode pattern is in phase with the difference data creation circuit 30. The difference data between the pixels is generated, a threshold value is generated based on the difference data output from the difference data creation circuit 30, and a defect is determined based on whether the difference data is larger than the threshold value. The point is to do it.

The CCD light receiving element 10 has about 5000 pixels, and the size per pixel is about 14 [μm].
It is. Therefore, the effective field of view of the CCD light receiving element 10 is 70
[Mm]. The CCD light receiving element 10 receives the reflected light from the substrate 12 as shown in FIG. CCD light receiving element 1
The image data output from 0 is converted by the A / D converter 20 into digital image data (shading information) of about 8 bits or 10 bits.
The data is taken into the IFO memory 31 and the latch circuit 37.

The difference data creation circuit 30 includes a FIFO memory 31, latch circuits 32-39, a selector 3A and a difference image calculation circuit 3B. FIFO memory 3
1 latches digital image data output from the A / D converter 20 at the input timing of the write enable signal Wenb, and latches the digital image data captured at the input timing of the read enable signal Renb.
Output to 2. The latch circuits 32 to 36 are connected in series and function as a shift register for sequentially shifting the digital image data output from the FIFO memory 31.
Therefore, the digital image data captured by the latch circuit 32 is sequentially transferred to the latch circuits 33 to 36, and the latched digital image data is output from each of the latch circuits 32 to 36 to each selection terminal of the selector 3A. You. The selector 3A selects one of the outputs of the latch circuits 32 to 36 based on a selection signal (not shown) and outputs the selected output to the difference image calculation circuit 3B as the second image data G (x).

On the other hand, the latch circuits 37 to 39
The digital image data passes through the path of the memory 31, the latch circuits 32-36 and the selector 3A, and adjusts the timing with the output digital image data. The digital image data taken into the latch circuit 37 is sequentially latched by the latch circuits 38 and 39.
And outputs the first image data F (x) to the difference image calculation circuit 3B. The difference image calculation circuit 3B is provided with a selector 3A.
, And calculates the absolute value V of the difference data between the second image data G (x) output from the latch circuit 39 and the first image data F (x) output from the latch circuit 39.
And output to the line memory 70.

The threshold value creation circuit 50 creates a threshold value T based on the absolute value V of the difference data output from the difference image calculation circuit 3B. How the threshold value creation circuit 50 creates the threshold value T will be described later. The threshold value memory 60 stores a plurality of threshold values T created by the threshold value creation circuit 50, and the optimum threshold value T
Is output to the arithmetic circuit 80. The line memory 70 has a FI
An absolute value V of the difference data output from the difference image calculation circuit 3B is input, and the absolute value of the difference data is set to a value corresponding to the processing time of the threshold value creation circuit 50 and the threshold value memory 60. V is delayed and output to the arithmetic circuit 80. The arithmetic circuit 80 subtracts the threshold value T output from the threshold value memory 60 from the absolute value V of the difference data output from the line memory 70, and calculates the subtracted value (V−
T) is output to the defect determination circuit 90. Defect determination circuit 90
Determines a defect based on the subtraction value (VT). In other words, if the subtraction value (V−T) is a positive value, it means that the absolute value V of the difference data is larger than the threshold value T, and the defect determination circuit 90 determines that the defect is not defective. On the other hand, if the subtraction value (V−T) is a negative value, it means that the absolute value V of the difference data is smaller than the threshold value T, and the defect determination circuit 90 is defective (foreign matter). Is present).

FIG. 5 is a diagram showing the concept of how the difference data creation circuit 30 generates the absolute value V of the difference data. In the drawing, squares shown by dotted lines are work patterns WP1 to WP8, which correspond to electrode patterns formed on the substrate, and whose repetition period is P [μm]. On the other hand, the CCD light receiving element 10 is a linear sensor provided in a line from the pixel A1 to the pixel Z1, and relatively moves in the direction from the pixel A1 to the pixel A5 by the movement of the inspection stage (not shown). It is supposed to. Therefore, the pixels A1 to Z1, A2 to Z2,
A3 to Z3, A4 to Z4, and A5 to Z5 form a pixel matrix.

The pixel matrix and the work pattern W
Looking at the relationship with P1 to WP8, the pixel A1 is located at the upper left corner of the work pattern WP1, and the pixel D1 is located at the upper left corner of the work pattern WP2. However, only the work pattern WP1 overlaps the pixel A1, whereas the work pattern WP1 and the work pattern WP2 overlap the pixel D1. That is, the relative positional relationship between the pixel A1 and the work pattern WP1,
The relative positional relationship between the pixel D1 and the work pattern WP2 is different.

In FIG. 5, the pixel A1 and the work pattern WP
1. The relative positional relationship between the pixel N1 and the work pattern WP5 is the same. Similarly, the pixel D1 and the work pattern WP
2. Relative positional relationship between pixel Q1 and work pattern WP6, pixel G1 and work pattern WP3, pixel T1 and work pattern WP7, relative position between pixel K1 and work pattern WP4, relative position between pixel X1 and work pattern WP8. The relationships are the same. The difference data creation circuit 30 according to the present embodiment is configured to create difference data based on pixel data of a pixel corresponding to a work pattern having the same relative positional relationship between the pixel and the work pattern as described above. Have been.

For example, when the relative positions of the pixels A1 to Z1 and the work patterns WP1 to WP8 are as shown in FIG. 5, pixel data of the pixels A1 to Z1 are sequentially output from the A / D converter 20. You. At this time, the write / read timing of the FIFO memory 31 is adjusted so that the selector 3A outputs the pixel data of the pixel A1 in synchronization with the timing at which the pixel data of the pixel N1 is output from the latch circuit 39, and the selector 3A And latch circuits 32-36
Select Thereby, the difference image calculation circuit 3B includes:
The pixel data G (N1) of the pixel N1 output from the latch circuit 39 is captured as the first image data F (x),
The pixel data F (A1) of the pixel A1 output from the selector 3A is taken in as the second image data G (x).
The difference image calculation circuit 3B outputs the pixel data F (A
The pixel data G (N1) of the pixel N1 is subtracted from 1), and the absolute value of the difference data V = | F (A1) -G (N1) |
Is output to the threshold value creation circuit 50 and the line memory 70. Hereinafter, in the same manner, the difference image calculation circuit 3B
To the pixel O from the pixel data F (B1) to F (D1)
It subtracts the pixel data G (O1) to G (Q1) of 1 to Q1, and outputs the absolute value V of the difference data. That is,
The FIFO memory 31, the latch circuits 32 to 36, and the selector 3A delay the pixel data by 13 pixels and output the delayed pixel data to the difference image calculation circuit 3B.

FIG. 5 shows a case where three to four pixels correspond to one work pattern for the sake of convenience of explanation, but in practice, 20 to 2 pixels correspond to one work pattern.
Five pixels correspond. For example, if the pixel size a is 14 [μm] and the repetition pitch P of the work pattern is 300 [μm], one work pattern corresponds to 300 ÷ 14 ≒ 21.4 pixels. As described above, when the number of pixels b corresponding to one work pattern is not an integer but includes a decimal point, the number of pixels b is multiplied by the smallest arbitrary integer c to obtain a value closest to the integer. For example, the number of pixels b = 21.4 and the integer c =
Multiplying by 5 gives a value of 107. This means that the number of pixels corresponding to the five work patterns is just an integer of 107. Accordingly, in such a case, the differential data creation circuit 30 delays the pixel data by 107 pixels by the FIFO memory 31, the latch circuits 32-36 and the selector 3A, and outputs the second image output from the selector 3A. The absolute value V of the difference data between the data G (x) and the first image data F (x) output from the latch circuit 39 is output. As a result, the relative positional relationship between the pixel and the work pattern becomes substantially the same, and it is not necessary to perform the correction process due to the phase shift.

Actually, however, there is a slight phase shift, and a specific portion of the imaging lens is reduced due to a magnification error in the light receiving element surface due to aberration of the imaging lens, mounting error, inclination of the substrate, and the like. Distortion and expansion distortion appear, and a phase shift occurs at a comparison position between two work patterns to be subjected to difference data. Therefore,
In this embodiment, in consideration of such an influence of the phase shift, the threshold value is determined based on a plurality of work patterns to be subjected to difference data by using the threshold value creation circuit 50.

FIG. 6 is a diagram showing the concept of how the threshold value creation circuit 50 creates a threshold value. In the figure, work patterns 1a to 1e, 2a to 2
e, 3a to 3e,..., 8a to 8e correspond to the electrode patterns formed on the substrate and have a repetition period of 30.
0 [μm], and the size of one pixel of the CCD light receiving element 10 is 14 [μm]. That is, FIG. 6 illustrates a case where the pixel size a is 14 [μm] and the repetition pitch P of the work pattern is 300 [μm]. Therefore,
As described above, one work pattern is 300 ÷ 14
# 21.4 pixels correspond. When the number of pixels corresponding to one work pattern is 21.4, the number of pixels corresponding to five work patterns is just an integral multiple of 107.

Therefore, the relative positional relationship between the first pixel to the 21.4th pixel from the left of the CCD light receiving element 10 and the work patterns 1a, 2a, 3a,..., 8a is the same. . Similarly, 4 from the 21.4th pixel
2. Work patterns 1b, 2b, 3 up to the 8th pixel
, 8b, 6
4. Work patterns 1c, 2c, 3 up to the second pixel
c,..., and 8c, 8
5. Work patterns 1d, 2d, 3 up to the 6th pixel
,..., 8d, 1
Work patterns 1e, 2e, 3 up to the 07th pixel
, 8e have the same relative positional relationship.

The difference data creation circuit 30 according to this embodiment is configured to create difference data based on pixel data of a pixel corresponding to a work pattern having the same relative positional relationship between the pixel and the work pattern. FIG. 6 shows the relative positional relationship between each pixel and each work pattern.
In such a case, the selector 3A outputs the pixel data of the first pixel in synchronization with the timing at which the pixel data of the 108th pixel is output from the latch circuit 39 in FIG. That is, from the latch circuit 39 of FIG.
8) The write / read timing of the FIFO memory 31 and the latch circuits 32-36 are set so that the selector 3A outputs the pixel data of the n-th pixel in synchronization with the timing at which the pixel data of the n-th pixel is output. . As a result, the difference image calculation circuit 3B outputs the first image data F of the (n + 108) th pixel output from the latch circuit 39.
The absolute value V of the difference data obtained by subtracting the second image data G (n) of the n-th pixel output from the selector 3A from (n + 108) is output.

FIG. 6 shows a case where five work patterns are used as one sample and eight samples smp1 to smp8 are used. This is because, as shown in FIG. 3, the edge of the pattern pitch is CC due to the distortion of the imaging lens.
This is because a position shift occurs on the D light receiving element. That is, when the imaging by the imaging lens 11 is ideally performed without distortion, the phase shift amount ε has the same value at any position, but in reality, the light receiving element due to the aberration of the imaging lens 11 The imaging lens 11
3B, the edge of the pattern pitch appears at the j-th position of the CCD light-receiving element 10 as in the reduction distortion of FIG. 3B, or the j + 2th position of the CCD light-receiving element 10 as in the expansion distortion of FIG. 3C. , And the phase shift amount ε also varies.

Therefore, in this embodiment, as shown in FIG. 1, the imaging lens 11 is divided into a plurality of blocks B1 to Bn, and one block is divided into eight samples as described above. Accordingly, each pixel of the CCD light receiving element 10 is also divided. For example, the imaging lens 11
Is divided into six areas of blocks B1 to B6,
If the number of pixels of the CD light receiving element 10 is about 5136, the number of pixels corresponding to one block of the imaging lens 11 is about 8
It becomes 56 pieces. These 856 pixels are the number of pixels corresponding to one block of the imaging lens 11. Therefore, since the number of pixels of one sample is 107, eight samples smp1 to smp8 exist in one block as shown in FIG.

The threshold value creation circuit 50 creates a threshold value based on the eight samples smp1 to smp8. FIG.
In the case of, one sample s with five work patterns
mp is formed, and one block is formed by eight samples smp1 to smp8.
, The absolute value V of 856 difference data per block is input. The threshold value creation circuit 50 creates a threshold value from the absolute values V of the 856 difference data. For example, the sum 絶 対 V of the absolute value V of the difference data of eight samples and the sum ΣW of the square of the absolute value V are calculated. Sum of absolute value V squared ΣW divided by 8 ΣW / 8
Is subtracted from the sum 絶 対 V of the absolute value V by 8 to calculate the variance δ by calculating the square root thereof. That is, the variance δ is δ = SQRT (ΣW / 8−ΣV / 8).

At this time, since there is a case in which the absolute value V of the difference data may correspond to the data of the foreign substance to be detected, the maximum value is determined from the absolute values V of the eight difference data. The data indicating the second maximum value may be excluded from the calculation processing in advance, and the calculation processing may be performed based on the absolute values V of the six difference data. Further, when the absolute value V of the eight differential data is far from other values, the value may be excluded.

When the variance value δ is determined, the threshold value creation circuit 50 multiplies the variance value δ by a coefficient Ka of 3 to 6, and divides the sum of the absolute values V by 8 to obtain a value ΔV. / 8
Then, the final threshold value T1 is created by adding the laser power distribution correction values Kb = 0 to 64. The threshold value T1 created in this way is stored in the threshold value memory 60 as a threshold value used for determining a defect between each work pattern of the block B1. Block B
Similarly, the threshold values T2 to T6 used for the defect determination between the work patterns 2 to B6 are obtained based on the absolute values V of the difference data of the blocks B2 to B6. The arithmetic circuit 80 uses the threshold value T obtained in this way to subtract from the absolute value V of the difference data outputted from the line memory 70 a target value for determining whether or not a defect, that is, a foreign substance exists. (VT) is assigned to the defect determination circuit 90.
Output to As a result, the substrate inspection apparatus according to this embodiment can accurately detect the presence of a foreign substance sufficiently smaller than the pixel size.

In the above-described embodiment, the case where the threshold value is obtained based on the absolute values V of the eight difference data has been described. However, the threshold value is obtained based on the absolute values V of other numbers of difference data. You may do so. In addition, it goes without saying that the present invention can be applied to masks, reticles, and wafer sugars having a predetermined repetition pattern other than the glass substrate.

[0035]

According to the present invention, even when the relationship between the pixel size of the light receiving element and the repetition pitch of the electrode pattern is not an integral multiple and has a phase shift, the signal quality of each pixel data is degraded due to the phase shift. And it is possible to detect minute foreign matter with respect to the pixel size with high accuracy.

[Brief description of the drawings]

FIG. 1 is a diagram showing a positional relationship among a light receiving element (CCD) 10, an imaging lens 11, and a substrate 12 when a glass substrate inspection is performed by a pattern comparison method.

FIG. 2 is a diagram showing a schematic configuration of a conventional substrate inspection apparatus.

FIG. 3 is a diagram showing how a pattern pitch edge appears on a CCD light-receiving element due to distortion of an imaging lens and appears.

FIG. 4 is a view showing a schematic configuration of a substrate inspection apparatus according to the present invention.

FIG. 5 is a diagram showing the concept of how the difference data creation circuit of FIG. 4 generates the absolute value V of the difference data.

6 is a diagram showing the concept of how the threshold value creation circuit of FIG. 4 creates a threshold value.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... CCD light receiving element 20 ... A / D converter 30 ... Difference data creation circuit 31 ... FIFO memory 32-39 ... Latch circuit 3A ... Selector 3B ... Difference image calculation circuit 50 ... Threshold creation circuit 60 ... Threshold memory 70: line memory 80: arithmetic circuit 90: defect determination circuit

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2G051 AA84 AB01 AC21 BA10 CA03 CB01 CB05 EA08 EA11 EA12 EB01 EC03 ED07 5B057 AA04 BA15 BA19 CA08 CA18 CB08 CB18 CE09 CH11 DA03 DA04 DA07 DA08 DB02 DB09 DC09 5L096 AA07 BA EA43 FA69 GA08

Claims (4)

[Claims] A lens system configured to form an image of a pattern having a predetermined pitch formed on a substrate; a light receiving element unit including a plurality of pixels configured to receive an image of the pattern through the lens system; Difference data creating means for sequentially creating and outputting difference data between pixels such that the relative positional relationship between the pattern formed on the light receiving element and the pixels is substantially in phase, and the difference data Threshold value creating means for creating a threshold value based on the difference data output from the creating means; and a threshold created by the difference data output from the difference data creating means and the threshold value creating means. A substrate inspection apparatus, comprising: a defect detection unit configured to detect a defect by comparing the value with a value. 2. The method according to claim 1, wherein the difference data creating unit assigns the smallest arbitrary first integer to the number of pixels corresponding to the pattern obtained by dividing the pitch of the pattern by the size of the pixel. A group of the number of the patterns corresponding to the first integer in the case where the value of the number of pixels is multiplied by the value closest to the integer is regarded as one pseudo pattern, and the adjacent ones of the pseudo patterns are used as the pseudo pattern. A substrate inspection apparatus for sequentially creating difference data. 3. The difference data according to claim 1, wherein the threshold value creating unit divides each pixel of the light receiving element unit according to the block by dividing the lens system into a plurality of blocks. Of the difference data output from the creating means, a plurality of pieces of data corresponding to the divided pixels of the light receiving element means are used as calculation difference data, and a variance value is obtained based on the calculation difference data. A substrate inspection apparatus, wherein the threshold value is created based on a threshold value. 4. The method according to claim 3, wherein the threshold value generating unit multiplies the variance value by a predetermined coefficient, and adds an average value of the calculation difference data and a laser power distribution correction value to the multiplied value. A substrate inspection apparatus for generating the threshold value.