JPH05346404A - Pattern inspection system - Google Patents

Abstract

PURPOSE:To enhance detection accuracy by avoiding the missing of incorrect interval when a circular pattern is juxtaposed to a linear pattern. CONSTITUTION:The pattern inspection system comprises means for generating data binary encoded depending on presence/absence of wiring pattern on a sample to be inspected, means for developing the binary encoded data two- dimensionally and storing the data thus developed, and means for setting a plurality of measuring lines extending radially in a same plane from an arbitrary starting point in a two-dimensional plane and measuring the length of each measuring line based on the continuity of data in each measuring direction. The pattern inspection system further comprises means for operating difference of measured length between two measuring lines having 180 deg. shifted directions, means for operating the sum of measured lengths of two measuring lines, and means for making a decision whether the arbitrary point is set at the center of wiring pattern or the center of the interval between two adjacent wiring patterns based on the outputs from the difference operating means and the sum operating means.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspecting apparatus for inspecting a wiring pattern for binarization based on binary data of the wiring pattern, and in particular, it can be applied to any pattern shape by using a "radial length measuring sensor". , And to an improved technique of a pattern inspection device with increased versatility.

[0002]

2. Description of the Related Art Generally, the quality inspection of a printed wiring pattern, an IC pattern, etc. is an essential element for improving the reliability of a system product using the same, but the pattern shape becomes complicated and fine. As a result, the limit of visual inspection is being reached, and the development of automatic inspection technology is an urgent task.

As one conventional example of the automatic inspection technique, a wiring pattern is optically read and converted into an electric signal, which is then binarized and taken into a memory circuit. There is a method of detecting an abnormality in the pattern shape by quantitatively determining the degree of overlap with the reference pattern. However, this method has a fatal defect that it cannot be widely applied to patterns of various shapes because it requires a specific reference pattern. That is, in this conventional method, it is necessary to newly recreate the reference pattern each time the shape of the target defect pattern is changed, and the versatility is poor.

JP-A-63-229571 (hereinafter, referred to as
Another conventional example of the automatic inspection technique is described in Japanese Patent Laid-Open Publication No. 2003-242242. According to this, by using the "radial length measuring sensor", a highly versatile pattern inspection apparatus can be realized without using the above-described reference pattern. Here, the radial length measuring sensor will be described.

When the surface of the sample to be inspected is photographed by optical means, there is a difference in brightness between the wiring pattern portion and the exposed portion of the substrate. Therefore, the wiring pattern is binarized by binarizing the image data with an appropriate threshold value. It is possible to generate binarized data depending on the presence or absence of. For simplification of description, if a portion with a pattern (high luminance portion) is a logical H and a portion without a pattern (low luminance portion) is a logical L, a two-dimensional distribution pattern of the binarized data (hereinafter, From the two-dimensional pattern, it is possible to distinguish between the presence of a wiring pattern (logic H) and the absence of a wiring pattern (logic L).

The radial length measuring sensor has a plurality of bit memory cell columns arranged radially along a virtual line called a length measuring line, and FIG. 7 shows an example thereof. In this example,
Eight measuring lines 1 to 45 at intervals of 45 ° centering on the point O
8 is set, and 1-5, 2-6, 3-7, or 4-4 have a relationship in which their 180 ° directions are different from each other.

Now, when the point O is made to correspond to an arbitrary position (P for convenience) of the two-dimensional pattern, the logic H or the logic L is stored in the memory cell at the point O. It can be seen that the logic H is located on the wiring pattern, and the logic L is located outside the wiring pattern. At this time, the logic H or the logic L is also stored in the memory cell columns of the respective length measuring lines 1 to 8 in the same manner. Therefore, by checking the continuity between the logic of the point O and the logic of the respective length measuring lines, The distance from P to the edge of the wiring pattern is set in multiple directions (45 ° in the example of FIG. 7).
Can be measured in 8 directions).

That is, the point O is a logic H (wiring pattern).
If the measurement line n (n is 1 to 8) up to the m-th bit is also a logical H, the distance from the point O to the m-th bit (O-
From m), the length from the “arbitrary point on the wiring pattern” to the edge of the wiring pattern in the direction of the length measurement line n can be known. Alternatively, the point O is the logic L (board) and the length measurement line n
Similarly, if the bits up to the m'th bit are logical L, the length (O-m ') from the "arbitrary point on the substrate" to the edge of the wiring pattern in the direction of the length measurement line n can be similarly known. it can.

Here, by applying the value for each length measurement line (hereinafter, length measurement value) obtained by the radial length measurement sensor to a predetermined conditional expression, the center of the wiring pattern or the center of the interval between the adjacent wiring patterns is determined. Can be detected. The predetermined conditional expression according to the above publication is as in the following expression (1). The difference between _{P 1 ≧ N 1 ...... (1} ) P 1 is the length measurement value r _{n + 8} of the length measurement values r _n and 180 ° opposite length measuring line n + 4 of the measuring line _n (| r n -r n _{+ 8} ｜)
It is the number of pairs of length measurement values that are less than or equal to a predetermined measurement margin, in other words, the number of length measurement value pairs whose lengths are balanced.
That is, according to the equation (1), when there are N ₁ or more pairs of length measurement values r _n and r _{n + 4} that are balanced in a substantially equal value,
It is detected as the center of the wiring pattern or the center of the space between the adjacent wiring patterns. It should be noted that N ₁ is a constant (called a balance number) for setting the detection sensitivity, and is set to, for example, about ¼ of the total number of length measurement lines of the radial length measurement sensor.

[0010]

However, in such a conventional pattern inspection apparatus, the "difference" between the length measurement values r _n and r _{n + 4} different in 180 ° direction is taken, and the difference is a predetermined measurement. Since the center is detected by the number of length measurement value pairs that fall within the margin, for example, when a circular pattern and a linear pattern are adjacent to each other, it may not be possible to detect the center of the interval. In this inspection method, a defect cannot be detected unless it is detected as the center, so that a defect in a defective interval occurs.

FIG. 8 is a conceptual diagram of measurement when the center can be detected.
, 9 is a linear pattern, 10 is a circular pattern
Is. The center O of the radial length measuring sensor has patterns 9 and 1
It is located slightly off from the center X of the closest distance between 0s.
It One of the length measurement lines 1, 4, 5, 6 and 8 shown by the solid line is
The distance from the point O to the edge of the pattern that is in contact with the pattern
Measured value r according to separation₁, R_Four, R_Five, R₆And r₈But
can get. On the other hand, the length measurement lines 2, 3 and 7 indicated by broken lines are
The measurement value r ₂, R
₃And r₇Is the measurement limit value (maximum number of bits in the memory cell column
Equivalent to) will be a constant value determined by. Here, 180 degrees
Measured values with different directions (r₁And r_Five, R_FourAnd r ₈) Is measured
If it is below the fixed margin,
Point O is detected as the center of the interval between patterns 9 and 10.
Be done.

On the other hand, FIG. 9 is a conceptual diagram of measurement when the center cannot be detected. As shown in the figure, when the center O ′ of the radial length measurement sensor and the center X of the closest distance between the patterns 9 and 10 substantially coincide with each other, only a pair of length measurement values may be obtained (see FIG. In the above example, r ₁ and r ₅ ), and in this case, the above expression (1) is not satisfied, so that the point O ′ is not recognized as the center of the interval. Such a problem is likely to occur especially when the number of length measurement lines is reduced, which is an obstacle to realizing a low-cost radial length measurement sensor with a small number of length measurement directions.

If the balance number N ₁ of the above equation (1) is set to "1", the center can be detected even in the case of FIG. 9, but if the balance number is made small, the sensitivity becomes too high,
A new problem of over-detection occurs in which the center is detected even in a place where measurement is unnecessary. [Purpose] Therefore, according to the present invention, by adding a “sum” of a pair of length measurement values different in 180 ° direction to a condition, for example,
An object of the present invention is to avoid missing a defective interval when a circular pattern and a linear pattern are adjacent to each other and improve detection accuracy.

[0014]

In order to achieve the above-mentioned object, the present invention provides data for generating binarized data according to the presence or absence of a wiring pattern on a sample to be inspected as shown in the principle diagram of FIG. A generation unit, a storage unit that two-dimensionally develops and stores the binarized data, and a plurality of length measurement lines that extend in the two-dimensional plane in a radial direction starting from an arbitrary point located in the two-dimensional plane are set. Then, the measuring means for measuring the length measurement value for each length measurement line from the continuity of the data in each length measurement line direction, and the difference between the length measurement values of the two length measurement lines of which the directions are different by 180 degrees are calculated. Difference calculation means, sum calculation means for calculating the sum of the length measurement values of the two length measurement lines, and the arbitrary point based on the output of the difference calculation means and the output of the sum calculation means, or the center of the wiring pattern, or Judgment to determine whether or not to use the center of the space between adjacent wiring patterns A step, and each operation value of the two length-measuring lines starting from the arbitrary point detected as the center is coded, and the coded information is collated with the normal coded information in the dictionary to perform the wiring. It is characterized by inspecting a pattern spacing defect.

[0015]

In the present invention, the "difference" and "sum" of a pair of length measurement values having different 180 ° directions are obtained, and the center is detected based on these "difference" and "sum". Here, the “sum” of the pair of length measurement values represents the width of the wiring pattern or the interval between the adjacent patterns. Therefore, when the “difference” between the pair of length measurement values is within a predetermined margin and the “sum” of the length measurement values is smaller than the predetermined value, for example, the outermost contact between adjacent patterns may be detected. it can.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. 2 to 6 are views showing an embodiment of the pattern inspection apparatus according to the present invention. First, the configuration will be described. In FIG. 2, a lens 20, a light source 21 and an optical sensor (for example, C
The line sensor 23 composed of the CD) 22 is the sample 24 to be inspected.
By scanning above, the two-dimensional pattern on the sample 24 is converted into an electric signal. This signal is subjected to required processing by a signal processing circuit (not shown), a filter circuit, and a two-dimensional pattern conversion circuit, and converted into binary data according to the presence / absence of a wiring pattern by a binary circuit 25. After that, it is stored in the storage circuit 26. Therefore, the respective parts from the line sensor 23 to the binarization circuit 25 integrally function as data generation means for generating binarized data according to the presence / absence of the wiring pattern on the inspected sample 24, and the storage circuit. Reference numeral 26 serves as a storage unit that two-dimensionally expands and stores the binarized data.

The binarized data in the storage circuit 26 is subjected to a required preprocessing by the preprocessing circuit 27, and thereafter, a radial length measuring sensor (hereinafter, may be abbreviated as a length measuring sensor) as a measuring means.
The measured values in a plurality of directions which are transferred to the sensor 28 and measured by the sensor 28 are sent to the center detection circuit 29. The output from the center detection circuit 29 is coded by the coding circuit 30 and then sent to the defect judgment circuit 31 where it is compared with a non-defective code stored in the dictionary to judge a pattern defect.

FIG. 3 is a block diagram of the center detecting circuit 29, which is an example corresponding to the length measuring sensor 28 which outputs length measuring values (r _{1 to} r ₈ ) in eight directions. 40 is a balance determination unit, 50
Is a proximity determination unit. Each of these parts has a pair of length measurement values (r ₁ -r ₅ , r ₂ -r ₆ , ...
₄ -r _8, i.e. r _n and r _{n + 4)} is input,
Balance determination circuits 41, 42, and 4 for each pair
3 and proximity determination circuits 51, 52, 53.

Further, the balance determination circuits 41 to 43 have a difference value r _dif between the length measurement values r _n and r _{n + 4} (hereinafter, the length measurement difference) r _{dif.
(R dif = | r n -r} n + 4 |) difference calculator for calculating the (differential calculation means) 41a, measurement difference r _dif and a predetermined reference value (M;
Balance margin) and when compared to "r _dif ≦ M" comparator 41b for outputting a balance signal S ₄₁ to S ₄₃ becomes the logic H is provided. In addition, the proximity determination circuit 51-53, measured Sum of long values r _n and r _{n + 4} (hereinafter, sum of measurement) r _add
A sum calculator (sum calculator) 51a for calculating (r _add = r _n + r _{n + 4} ) and a length measurement sum r _add are compared with a predetermined reference value (L ₁ ; proximity determination value) to obtain “r _add ”. ≦ L ₁ "comparator 51b outputs a proximity signal S ₅₁ to S ₅₃ becomes the logic H is provided at.

That is, the balance determination unit 40 determines whether or not the difference between the two measured values (r _n , r _{n + 4} ) in the opposite direction is within the margin M, in other words, whether or not there is a balance. In addition, in the proximity determination unit 50, the
Whether the sum of the length measurement value is below the predetermined value L _1, whether the line width of the wiring pattern in other words the predetermined value L ₁ or less, or the spacing between adjacent patterns is close to or less than a predetermined value L ₁ is To be judged. Therefore, the signals S _{41 to} S ₄₃ are 3-bit codes representing the balance number, and the signals S ₅₁ to ₅₃ are 3-bit codes representing the proximity number.

These 3 + 3 bit signals S _{41 to} S ₄₃ , S
₅₁ to S ₅₃ is, R of about condition determining circuit (determination means)
The central signal S _{60 which} is sent to the OM ₆₀ and becomes a logic H when the predetermined discrimination condition described later is satisfied is output from the ROM ₆₀ . FIG. 4 is a block diagram of the encoding circuit 30 corresponding to one direction of the length measuring sensor. 70 is a balance encoding unit, 80 is a length encoding unit, 90 is an OF encoding unit, and 100 is a data conversion unit. Is.

The balance code section 70 has a length measurement value pair (r
_n , r _{n + 4} ) difference calculator 70a, a comparator 70b that compares the output of the difference calculator 70a with a predetermined margin M to determine the balance of the length measurement value pair, the length measurement value pair (R
_n , r _{n + 4} ) comparing the respective lengths of _n , r _{n + 4} ) to determine the directionality, and a gate for passing the output (direction) of the comparator 70c when the length measurement value pairs are not balanced. 70d is provided, and 2-bit data of a balance bit and a direction bit is generated.

The length coding unit 80 adds the length measurement value pair (r _n , r _{n + 4} ) to obtain a length measurement sum.
a, a comparator 80b for comparing the first coded threshold value SL ₁ with the length measurement sum, and a comparator 80c for comparing the second coded threshold value SL ₂ with the length measurement sum. It is generated by 2-bit data representing the size of. OF coding unit 9
0 is two comparators 90a that compare each of r _n and r _{n + 4} forming a pair of length measurement values with a predetermined OF threshold value (a value corresponding to the maximum length measurement value of the radial length measurement sensor). 90b is provided, and 2-bit data indicating whether each is OF (overflow) is generated.

The data conversion circuit 100 is provided with the above-mentioned components.
Each 2-bit data is received and 4-bit data
Data (per one direction) is generated and output. Figure 5
r₁-R_FiveTo r_Four-R₈Up to 4 directions, all code
FIG. 4 is a configuration diagram including a coding circuit, and 4 from each coding circuit.
Bit data (16-bit data in total) is the central signal
No., that is, S retrieved from the ROM 60 of FIG.₆₀To
Therefore, it passes through the gate circuit 101 which is turned on / off, and as shown in FIG.
It is sent to the defect determination circuit 31 shown. Where the encoding times
The road is balanced by two measuring lines with different 180 ° directions.
If it is not balanced,
Is the direction of the longer measuring line, whether it is OV or not
Judgment, “sum” of 2 length-measuring lines is 3 length areas (SL ₁Since
Lower area, SL₁Greater than and SL₂The following areas,
SL₂Larger area).
Bit data generated for each judgment result
Code together. That is, in the coded information
Is the information indicating the "sum" of the measured values, which is not found in the conventional example.
Will be included.

Next, the operation will be described. The following expression (2) is a first example of the determination condition used in this embodiment. [P ₁ ≧ N ₁ ] or [P ₂ ≧ N ₂ ] (2) where: P ₁ is the number of length measurement value pairs for which (| r _n −r _{n + 4} | <M) holds, that is, The number of balanced length measurement pairs.

P ₂ is a well-balanced length measurement value pair, and r _n + r _{n + 4} <L ₁ holds, that is, the number of length measurement value pairs for which proximity determination has been made. : N ₁ is a prescribed number of balanced pairs (for example, 2): N ₂ is a prescribed number of adjacent pairs (N ₁ or less, for example 1) The above conditional expression (2) is for P ₁ ≧ N ₁ and P ₂ ≧ N ₂ . 2 Includes judgment conditions. According to the former condition, the center is detected when the number of balanced length-measurement value pairs is equal to or more than the specified number N ₁ , and according to the latter condition, the well-balanced and proximity-measured length-measurement value pairs are detected. The center is detected when the number of n is greater than the predetermined value N ₂ .

That is, in the present embodiment, in addition to the balance determination based on the length measurement "difference" as in the conventional case, the proximity determination based on the length measurement "sum" is also used, so that a pair of balanced length measurement values can be obtained. Even if the number is small, for example, about one, it is possible to detect the minimum line width center of the wiring pattern, the closest approach center of the adjacent wiring pattern, and the like from the sum of the length measurements. this is,
This is because the coded information of this embodiment includes the information of the length measurement “sum”, and the coding ability is higher than that of the conventional example described above.

The following expression (3) is a second example of the discrimination condition used in this embodiment. [P ₁ ≧ N ₁ ] and [P ₃ ≧ 1] (3) where: P ₃ is the number of length measurement value pairs for which (r _n + r _{n + 4} <L ₂ ) holds. According to this conditional expression (3), when the number of balanced measurement value pairs is N ₁ or more and at least _one measurement value pair having a default value L ₂ or less exists. The center is detected. In other words, the center detection is not performed when the length measurement sum of all length measurement value pairs is L ₂ or more. As a result, it is possible to remove from the detection target the portions having the pattern intervals that are not so close to each other.

The following expression (4) is a third example of the discrimination condition used in this embodiment, and is a combination of the first conditional expression (2) and the second conditional expression (3). .. {[P ₁ ≧ N ₁ ] or [P ₂ ≧ N ₂ ]} and [P ₃ ≧ 1] (4) The following expression (5) is a fourth example of the determination condition used in this embodiment. Yes, this is a determination method of weighting according to the distance.

[P_L× W_L+ P_S× W_S≧ N₃] (5) However,: P_SIs  (| R_n-R_{n + 4}│ <M And r_n+ R_{n + 4}<L₂)  The number of length measurement value pairs that hold, that is, P above₂alike
Of the balanced length value pairs, the length measurement sum is the default value L
₂Number of shorter length measurement pairs (hereinafter
Number).

: P_LIs  (| R_n-R_{n + 4}│ <M And r_n+ R_{n + 4}≧ L₃)  The number of length measurement pairs for which
Of the length measurement value pairs, the length measurement sum is the default value L₃Measuring longer than
The number of value pairs (hereinafter, the number of length measurement value pairs).

: W_LIs the length measurement value P_LWeight given to
It is a value. : W_SIs the short measurement value P_SIs a weight value given to.  (W_S> W_L): N₃This is the center judgment value. According to this conditional expression (5), the length / short ratio of the length measurement value pair
Weighted value and center judgment value N₃Compared to
Therefore, the weight value W_L, W_SBy adjusting the
The output characteristics can be freely tuned.

The following expression (6) is a fifth example of the discrimination condition used in this embodiment, and is a combination of the third conditional expression (5) and the fourth conditional expression (6). .. [P _L × W _L + P _S × W _S ≧ N ₃ ] and [P ₃ ≧ 1] (6)

[0034]

According to the present invention, since the coding ability is enhanced by adding the "sum" of a pair of measured values having different 180 ° directions to the condition, for example, a circular pattern and a linear pattern are adjacent to each other. In such a case, the failure of the center can be avoided, and the detection accuracy can be improved.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is an overall configuration diagram of an embodiment.

FIG. 3 is a block diagram of a center detection circuit.

FIG. 4 is a block diagram of a coding circuit corresponding to one direction of the length measuring sensor.

FIG. 5 is a configuration diagram including coding circuits for all directions of the length measurement sensor.

FIG. 6 is a block diagram of a defect determination circuit.

FIG. 7 is a conceptual diagram of a radial length measurement sensor.

FIG. 8 is a conceptual diagram of measurement when the center can be detected.

FIG. 9 is a conceptual diagram of measurement when the center cannot be detected.

[Explanation of symbols]

 23: Line sensor (data generation means) 24: Sample to be inspected 25: Binarization circuit (data generation means) 26: Storage circuit (storage means) 28: Radial length measuring sensor (measurement means) 41a: Difference calculator (difference) Calculation means) 51a: Sum calculator (sum calculation means) 60: ROM (determination means)

Claims (2)

[Claims] 1. A data generating means for generating binarized data according to the presence / absence of a wiring pattern on a sample to be inspected, a storage means for two-dimensionally developing the binarized data and storing the binarized data. Measuring means for setting a plurality of length measurement lines extending in the radial direction on the same plane starting from an arbitrary point located in the dimension plane and measuring the length measurement value for each length measurement line from the continuity of data in each length measurement line direction. A difference calculation means for calculating a difference between the length measurement values of two length measurement lines whose directions are different by 180 degrees, and a sum calculation means for calculating a sum of the length measurement values of the two length measurement lines; A center for the wiring pattern or a center for the interval between the adjacent wiring patterns based on the output of the computing means and the output of the sum computing means; Each of the two length-measuring lines starting from the given arbitrary point Encoding calculated value, pattern inspection apparatus characterized by by matching the normal coding information of the coded information and the dictionary to check the spacing defect defect of the wiring pattern. 2. A step of generating binarized data according to the presence / absence of a wiring pattern on a sample to be inspected and developing and storing the binarized data two-dimensionally, and a position in the two-dimensional plane. Set a plurality of length measurement lines that extend in the same plane in the radial direction starting from an arbitrary point, and measure the length measurement value for each length measurement line from the continuity of the data in each length measurement line direction; and Among them, a step of calculating a difference and a sum of length measurement values of two length measurement lines whose directions are different by 180 degrees, and the arbitrary point is set to the center of the wiring pattern based on the output of the difference calculation means and the output of the sum calculation means, or A step of deciding whether or not to use the center of a space between adjacent wiring patterns, a step of coding each calculated value of the two length-measuring lines starting from the arbitrary point detected as the center, and the coded information And normal in the dictionary Pattern inspection method characterized by comprising the steps of collating the coded information to check the spacing defect defect of the wiring pattern.