JP2001074414A - Method for aligning pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To align while absorbing the local distortion of an inspecting workpiece, even when a pattern which serves as a positioning mark is not present. SOLUTION: A division region is set in a mask pattern, and a virtual line perpendicular to the pattern is set as a second positioning mark in each division region. The division region of a measuring pattern is extracted in correspondence with the division region of the mark pattern, and a correlation value with the mask pattern in a predetermined range about a position is calculated in correspondence with the second positioning mark. The position with the highest correlative value is deemed the second positioning mark of the measuring pattern, and the mutual second position marks of the measuring pattern and the mask patterns are aligned, so that the mask pattern and the measuring pattern are aligned by each division region (step 109).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection method for inspecting a pattern formed on a tape carrier or the like, and more particularly to an alignment method for aligning a master pattern and a pattern to be measured.

[0002]

2. Description of the Related Art Conventionally, a TAB (Tape Automated Bonus) has been used as a mounting technique suitable for a demand for a higher pin count of ICs and LSIs.
ding) method is known. In the TAB method, a copper foil pattern formed on a polyimide tape carrier (TAB tape) is bonded to electrodes of an IC chip to form external leads. The copper foil pattern is formed by attaching a copper foil to a tape carrier with an adhesive and etching this.

In such a tape carrier, a pattern is visually inspected by a human using a microscope after the pattern is formed. However, visually inspecting a fine pattern requires skill and has a problem of overworking the eyes. Therefore, as an alternative to the visual inspection, a technique has been proposed in which a pattern formed on a tape carrier or the like is imaged by a TV camera and automatically inspected (for example, JP-A-6-273132 and JP-A-7-273132).
No. 110863).

FIGS. 16 and 17 show Japanese Patent Application Laid-Open No. 6-273132.
FIG. 1 is a diagram for explaining a conventional inspection method for detecting a disconnection described in Japanese Patent Application Laid-Open Publication No. HEI 10-125, 1988. A master pattern created by imaging the pattern to be measured determined as a non-defective product is registered as a set of straight lines indicating pattern edges. The pattern to be measured is input as a set of edge data (edge coordinates) indicating a pattern edge extracted from a grayscale image obtained by capturing the pattern. Then, the extracted edge data n1, n2, n3.
.. and the line of the master pattern are associated. In order to perform this association, as shown in FIG. 16, continuous straight lines A1 and A2, A2 and A3,.
Bisecting lines A2 'and A that bisect each of the corners
Ask for 3 '.

The bisectors A2 ', A3',... Divide the master pattern straight lines A1, A2, A3,.
Thus, the edge data n1, n2, n3,... Of the pattern to be measured present in each area are respectively associated with the straight lines A1, A2, A3,. become. For example, in FIG. 16, the edge data n1 to n3 are associated with the straight line A1, and the data n4 to n6 are associated with the straight line A2.
Next, the edge data of the pattern to be measured is compared with the master pattern to check whether the pattern to be measured is disconnected. This inspection is realized by a labeling process of tracing the pattern edge by tracing the connected edge data n1 to n9 of the pattern to be measured, as shown in FIG. At this time, since the edge data is not connected at the disconnected portion due to the disconnection generated at the leading end of the pattern to be measured, there is no edge data corresponding to the straight lines A3 to A5 of the master pattern. Thus, the disconnection of the pattern to be measured can be detected.

FIG. 18 is a view for explaining a conventional inspection method for detecting a short circuit described in JP-A-6-273132. In this inspection method, first, in an inspection area 20 in which a master pattern and a pattern to be measured are cut out to a predetermined size, edge data connected to the pattern to be measured is tracked. Thereby, each edge data of the pattern to be measured is sequentially labeled as n1 to n18. However, the edge data n8 and n17 are not registered in the master pattern Mb composed of two opposing straight lines similarly to the master pattern Ma composed of two opposing straight lines indicating a pattern edge. Thus, a short circuit of the pattern to be measured can be detected.

FIG. 19 is a view for explaining a conventional inspection method for detecting a defect or a protrusion described in Japanese Patent Application Laid-Open No. Hei 7-10863. In this inspection method, first, a perpendicular line perpendicular to the center line L is drawn, and the length between intersections at which the perpendicular line intersects the straight lines A1 and A2 indicating the edges of the master pattern is obtained in advance as the width W0 of the master pattern. Next, in the actual inspection, the edge data n of the pattern to be measured
Then, the distance between the facing edge data is obtained by lowering the perpendicular to the center line L of the master pattern. This distance is the width W of the pattern to be measured, and by comparing this with the width W0 of the master pattern, a defect or a protrusion in the pattern to be measured can be detected.

However, in the pattern inspection apparatus using such an inspection method, there is a problem that the pattern inspection takes a long time because the detailed inspection by comparison with the master pattern is performed by software over the entire pattern to be measured. . Therefore, a pattern inspection apparatus capable of performing inspection in a short time has been proposed (for example, Japanese Patent Laid-Open No. 10-1419).
No. 30). In the pattern inspection apparatus described in JP-A-10-141930, a defect candidate of a pattern to be measured is detected by hardware, and only a predetermined small area including the detected defect candidate is inspected by software. Defects can be inspected faster than before.

In the pattern inspection apparatus using the above inspection method, it is necessary to align the master pattern and the measured pattern in order to compare the measured pattern captured by the camera with the master pattern. This alignment has been performed by matching the position of a positioning mark provided in advance on the master pattern with the position of the corresponding positioning mark of the pattern to be measured.

[0010]

However, the above-described alignment method has a problem that when a tape carrier or the like on which a copper foil pattern is formed is used as an inspection work, a normal inspection may not be performed. there were. The reason why such a problem occurs will be described with reference to FIG.
The tape carrier is a thin film made of polyimide and has flexibility. For this reason, a local distortion exists in the tape carrier imaged by the camera, and this distortion causes a displacement of the measured pattern from the master pattern. As shown in FIG. 20, there is no independent pattern that can be a positioning mark near the center of the pattern formed on the tape carrier. Therefore, in the conventional alignment method, an independent pattern provided around the tape carrier is used as the positioning mark. However, such a positioning method cannot correct a position shift occurring in a portion other than the peripheral portion due to local distortion of the tape carrier. Therefore, the positional deviation of the measured pattern from the master pattern is detected as a defect in the above-described inspection. The present invention has been made in order to solve the above-described problems, and it is easy to cause local distortion, and even if it is an inspection work such as a tape carrier that does not have an independent pattern that can be a positioning mark, it can be used locally. It is an object of the present invention to provide an alignment method capable of absorbing a local distortion of an inspection work by performing an accurate alignment.

[0011]

According to the pattern alignment method of the present invention, the entirety of the master pattern and the pattern to be measured is aligned by aligning the positions of the first positioning marks with respect to the entire area of the pattern to be measured and the master pattern. After performing the alignment of the master pattern, setting a plurality of divided areas in the master pattern, and setting at least four virtual lines orthogonal to the pattern as the second positioning marks are performed for each divided area. The divided region of the pattern to be measured corresponding to the divided region of
The correlation value with the master pattern is calculated for a predetermined range centered on the position corresponding to the positioning mark, and the peak position having the highest correlation value is regarded as the second positioning mark of the pattern to be measured. By aligning the positions of the second positioning marks of the patterns, the alignment of the master pattern and the pattern to be measured is performed for each divided area. In the conventional alignment method, if there is a local distortion in the inspection work, even if the alignment with the master pattern has been completed as a whole, a deviation from the master pattern has occurred locally. This deviation may be detected as a defect candidate. In this case, even if the local distortion of the inspection work is within the allowable range, the secondary inspection is performed, so that there is a problem that an inspection time is required. On the other hand, in the present invention, if the local distortion of the inspection work is within the allowable range, the local distortion of the inspection work is absorbed by the alignment of each divided region, and the divided region of the master pattern and the pattern to be measured are absorbed. Delicate displacement of the divided areas can be corrected.

In one embodiment of the pattern alignment method according to the present invention, the peak position is determined for each line parallel to the second positioning mark of the master pattern within the predetermined range, and then the peak position is determined. The amount of deviation between the position and the second positioning mark of the master pattern is obtained, and the average value of the peak positions is obtained from the remaining lines excluding the lines with the maximum and minimum deviations, and this average position is measured. This is regarded as a second positioning mark of the pattern. Further, one configuration example of the pattern alignment method of the present invention is such that, in the divided region of the master pattern and the corresponding divided region of the pattern to be measured, the coordinates of the second positioning mark and the measured pattern and the master pattern are determined. By determining the coordinate conversion formula between them and converting the divided region of the master pattern by the coordinate conversion formula, the positioning of the master pattern and the pattern to be measured is performed for each divided region. In one configuration example of the pattern alignment method according to the present invention, the deviation between the result of inputting the coordinates of the second positioning mark of the measured pattern into the coordinate conversion formula and the coordinates of the corresponding positioning mark of the master pattern is marked. It is necessary to exclude the second positioning mark whose deviation is larger than a predetermined threshold from both the pattern to be measured and the master pattern and to obtain the coordinate conversion formula again. The coordinate conversion formula is determined by iterating until.

[0013]

Next, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a first embodiment of the invention.
And FIG. 2 is a block diagram of a pattern inspection apparatus used in this inspection method. In FIG. 2, a pattern inspection apparatus includes a sprocket 2 on a supply side that sends out a tape carrier 1 on which a copper foil pattern is formed,
A sprocket 3 on the winding side for winding the tape carrier 1, a line sensor camera 4 for imaging the tape carrier 1, a primary inspection for detecting a defect candidate of the pattern to be measured, and output of address information indicating a position of the defect candidate And a second image processing apparatus 6 that obtains an error between a measured pattern and a master pattern in a predetermined area including a defect candidate based on the address information and performs a secondary inspection of the measured pattern. And a host computer 7 for controlling the entire apparatus, and a display device 8 for displaying inspection results.

First, a master pattern created in advance before inspection will be described. Host computer 7
Are read by a magnetic disk drive (not shown) (step 101 in FIG. 1) and read out, for example, design value data (hereinafter referred to as CAD data) of a tape carrier created by a CAD (Computer Aided Design) system and written on a magnetic disk. The pattern edge data is extracted from the CAD data. The edge data is a set of pixels “1” indicating a pattern edge. Then, the host computer 7 paints the area surrounded by the pixel “1” indicating the pattern edge with “1”, and examines the pattern painted with the pixel “1” (the background other than the pattern is “0”). The first master pattern is used as a reference (step 102).

As described above, in the present embodiment, in order to create an accurate master pattern, CAD data used as a master in manufacturing the tape carrier 1 is used. Next, the operator of the pattern inspection apparatus specifies three or more first positioning marks in the first master pattern M for performing the overall alignment for one IC (step 1).
03).

FIG. 3 is a view showing a first master pattern M for one IC, and Tm represents a first positioning mark. The operator displays the screen shown in FIG.
Is specified as a first positioning mark Tm on a video such as that described above. The host computer 7 stores the position of the designated first positioning mark Tm.

Next, the operator sets a plurality of divided areas Em in the first master pattern M for one IC as shown in FIG. 4A, and sets each divided area Em as shown in FIG. , Four or more second positioning marks Fm (Fm1 to Fm4) are designated (step 104). That is, the operator sets a virtual line orthogonal to the pattern as the positioning mark Fm on the image as shown in FIG. 4B displayed on the screen of the display device 8. Therefore, FIG.
In the example of (b), the positioning mark Fm is set in the horizontal direction (X direction) with respect to the pattern in the vertical direction (Y direction).

The host computer 7 stores the set position and size of each of the divided areas Em, and also stores information on the second positioning mark Fm, such as the direction, length and center coordinates of the positioning mark Fm, and the positioning mark Fm. Fm
The grayscale data ("0" or "1") of the upper master pattern M
) Is stored. Note that the size of each divided region Em may not be constant.

Next, the host computer 7 converts the first master pattern into a second master pattern for detecting a defect, a pinhole or a disconnection, and a third master pattern for detecting a protrusion, scattering or short circuit as follows. (Step 105). FIG. 5 is a diagram for explaining a method of creating the second and third master patterns, and shows a part of the first master pattern. In FIG. 5, for the sake of simplicity, the first master pattern is represented only by a straight line representing a pattern edge, and the second and third lines are represented by a straight line representing a pattern edge and an oblique line representing the inside thereof. Although the master pattern is shown, the actual first to third master patterns are pattern edges and the inside thereof are filled with a pixel “1”.

First, as shown in FIG. 5A, the first master pattern is contracted in a direction perpendicular to its center line,
A second master pattern M1 is created. This is because opposing straight lines A1 and A1 indicating both edges of the first master pattern
4 (center line is L1) and the distance between A2 and A3 (center line is L2) are narrowed to narrow the first master pattern.

The accuracy of defect detection by the second master pattern M1 is determined by how much the first master pattern is contracted. For example, when it is desired to recognize a defect when there is a defect exceeding 1/5 of the width of the first master pattern, the width of the second master pattern M1 is set to 3/5 of the width of the first master pattern. What is necessary is just to reduce it. It goes without saying that the detection accuracy may be determined in pixel units or actual dimensions. Thus, the second master pattern M1 for detecting a defect, a pinhole, or a disconnection is created.

Subsequently, as shown in FIG. 5 (b), the first master pattern is expanded in a direction perpendicular to the center line to create a third master pattern M2. This corresponds to the opposite straight line A5 indicating both edges of the first master pattern.
And A8 (center line is L3), A6 and A7 (center line is L
4), A9 and A12 (center line is L5) and A10 and A1
1 (the center line is L6), and the first master pattern can be made thicker by widening the respective intervals. However, what actually becomes the third master pattern M2 is an area obtained by logically inverting the result of the expansion processing, that is, a first master pattern Ma composed of straight lines A5 to A8 and a first master pattern composed of straight lines A9 to A12. This is an area sandwiched between two patterns generated by expanding the pattern Mb.

The accuracy of defect detection by the third master pattern M2 is determined by how much the first master pattern is expanded. For example, when it is desired to recognize a defect when there is a defect exceeding 1/5 of the width of the first master pattern, the width of the third master pattern M2 is set to 7/5 of the width of the first master pattern. It should just be expanded as follows. The fact that the detection accuracy may be determined in pixel units or actual dimensions is the same as in the second master pattern. Thus, the third master pattern M2 for detecting protrusions, scattering, or short circuits is created.

Next, the inspection of the pattern to be measured will be described. A tape carrier 1 wound on a supply-side reel (not shown) is conveyed toward a take-up side by a supply-side sprocket 2 and a take-up-side sprocket 3.
It is taken up on a take-up reel (not shown). The line sensor camera 4 captures an image of the tape carrier 1 being transported. The first image processing device 5 digitizes the grayscale image output from the camera 4 and temporarily stores it in an internal image memory (not shown) (step 106). Camera 4
Is a line sensor in which pixels are arranged in the X direction. Therefore, by moving the camera 4 in the Y direction by a transport mechanism (not shown), two-dimensional image data is stored in the image memory.

Subsequently, the image processing device 5 binarizes the grayscale image of the pattern to be measured stored in the image memory (step 107). The density image data of the pattern to be measured includes the copper foil pattern and the other background (base material such as a base film), but since there is a density difference between the copper foil pattern and the background, the copper foil pattern Is set as a threshold value, the copper foil pattern is converted to “1” and the background is converted to “0”. In this way, it is possible to obtain a pattern to be measured in which the pattern edge and the inside thereof are filled with the pixel “1”.

Next, the image processing apparatus 5 aligns the entire measured pattern of one IC subjected to the binarization processing and the master pattern of one IC (step 108).
FIG. 6 is a diagram for explaining this positioning method.
The image processing apparatus 5 stores the first master pattern M (FIG. 6A) for one IC in which the first positioning mark Tm is set as described above and the position information of the positioning mark Tm in the host computer 7. Sent by Image processing device 5
Is a search for an area corresponding to the first positioning mark Tm in the pattern to be measured P for one IC stored in the image memory, thereby obtaining the first pattern corresponding to the positioning mark Tm as shown in FIG. The positioning mark Tp is detected.

Then, the image processing apparatus 5 obtains the distances DXp and DXm between the two positioning marks arranged in the X direction for each of the measured pattern P and the master pattern M. The distance between marks is the distance between the centers of gravity of two positioning marks. Subsequently, the image processing apparatus 5 calculates an enlargement / reduction rate (DXp / DXm) from the obtained distance between marks, and the distance between marks of the master pattern matches the distance between marks of the pattern to be measured based on the enlargement / reduction rate. The master pattern M is enlarged or reduced in all directions as described above.

Next, the image processing apparatus 5 determines the distances DYm and DYp between the two positioning marks arranged in the Y direction for each of the master pattern M 'subjected to the enlargement / reduction correction and the pattern P to be measured, as shown in FIG. , As shown in FIG. Then, the relative speed between the line sensor camera 4 and the tape carrier 1 is adjusted so that the distance between marks of the pattern to be measured matches the distance between marks of the master pattern, and the tape carrier 1 is imaged again. The image resolution in the Y direction is determined by the pixel size of the line sensor camera 4 and the relative speed. Therefore, by changing the moving speed of the line sensor camera 4, the image resolution in the Y direction can be adjusted, and the distance between marks can be matched.

Next, the image processing device 5 uses the positioning mark positions of the measured pattern P 'obtained by imaging in this manner and the positioning mark positions of the enlarged / reduced corrected master pattern M' as shown in FIG. Pattern P ', M'
Is obtained, and the master pattern M ′ is rotated so as to eliminate the angle shift. Finally, the image processing apparatus 5 aligns the positions of the master pattern M ′ and the pattern to be measured P ′ so that the mark positions match each other.

As described above, in the present embodiment, the image resolution in the Y direction is adjusted by changing the capture speed of the camera 4 with respect to the image resolution in the X direction determined by the number of pixels of the line sensor camera 4. , Vertical (Y) and horizontal (X) can be set to 1: 1. Therefore, the pattern to be measured, which is a good product but has an elongation within the allowable range with respect to the standard, can be matched with the master pattern, and the position of the pattern is automatically corrected for variations in the pattern position during formation. It can be performed.

Next, the image processing apparatus 5 performs positioning of the pattern to be measured and the master pattern for each divided region (step 109). FIG. 7 is a flowchart showing a positioning method for each divided area, and FIG. 8 is a diagram for explaining the positioning method. In the alignment processing in step 109, the image processing apparatus 5
8A, one divided region Em is cut out from the first master pattern M for one IC, and a divided region Ep corresponding to the divided region Em is cut out from the pattern to be measured P for one IC shown in FIG. (Step 201 in FIG. 7).

Subsequently, the image processing device 5 sets the divided area Em
The second positioning mark Fm (Fm1-Fm
A search is made for a second positioning mark in the divided area Ep corresponding to 4) (step 202). FIG. 9 is a flowchart illustrating this search method, and FIG. 10 is a diagram illustrating a search range of the second positioning mark in the divided area Ep. As shown in FIG. 10, the search range of the second positioning mark in the divided region Ep of the pattern to be measured is a range of coordinate positions ± 8 pixels corresponding to the positioning mark Fm in a direction parallel to the positioning mark Fm. The direction perpendicular to the positioning mark Fm is a range of a total of five lines of ± 2 lines at the coordinate position corresponding to the positioning mark Fm.

In the search processing in step 202, the image processing apparatus 5 sets the start value of the search position coordinates ys in a direction perpendicular to the positioning mark Fm (step 301 in FIG. 9). Since the start value at this time is the minimum value in the search range shown in FIG. 10, it becomes the coordinate position -2 corresponding to the positioning mark Fm.

Further, the image processing device 5 sets a start value of the search position coordinate xs in a direction parallel to the positioning mark Fm (step 302). The starting value at this time is
The coordinate position (center coordinate of the mark Fm) corresponding to the positioning mark Fm is -8. Next, the image processing device 5 calculates a normalized correlation value COR (xs) between the density data of the master pattern on the positioning mark Fm and the density data of the pattern to be measured at the search position by the following equation (step 303). .

[0035]

(Equation 1)

In equation (1), masterlength
th is the length of the positioning mark Fm, master (i)
Is the density data of the master pattern on the positioning mark Fm (i is 1, 2, 3,..., Masterlength)
h) and gray (i + xs) are grayscale data of the pattern to be measured at a point where the coordinates in the direction perpendicular to the mark Fm are the search position coordinates ys and the coordinates in the direction parallel to the mark Fm are i + xs.

According to steps 301 and 302, the current search position coordinates xs and ys are both start values. Correlation value C
After calculating OR (xs), the image processing device 5 determines whether or not the search position coordinate xs in the direction parallel to the positioning mark Fm has reached the end value (step 304). Since the end value at this time is the maximum value in the search range shown in FIG. 10, the coordinate value (the mark F
(center coordinate of m) +8.

If the search position coordinate xs has not reached the end value, the image processing apparatus 5 increases the search position coordinate xs by 1 (step 305), and calculates the correlation value COR (xs) again (step 303). Such a correlation value COR (x
The calculation of s) is repeated until the end value of the search position coordinates xs is reached. After the search position coordinate xs reaches the end value, the image processing device 5 calculates the maximum correlation value COR (xsm) of the calculated correlation values COR (xs) and the correlation value COR (xsm−1) of xsm ± 1 pixel. ), COR (xsm + 1), a position where the correlation value COR (xs) takes a local maximum value (hereinafter, referred to as a position).
(Referred to as a peak position) is calculated with an accuracy of a pixel or less (step 306).

FIG. 11 is a diagram for explaining a method of calculating the peak position of the correlation value COR (xs). In the present embodiment, the correlation values COR (xsm) and COR (xsm−
1), COR (xsm + 1) is assumed to be quadratic. Thereby, the following equation is established.

[0040]

(Equation 2)

[0041]

(Equation 3)

[0042]

(Equation 4)

From the equations (2), (3) and (4), the coefficients a, b and c can be obtained as follows.

[0044]

(Equation 5)

[0045]

(Equation 6)

[0046]

(Equation 7)

On the other hand, the point where the slope obtained by differentiating ax ² + bx + c with respect to x becomes 0 is the peak position, so the following equation is established.

[0048]

(Equation 8)

From equations (5) to (8), X at the peak position
The coordinate x can be obtained as in the following equation.

[0050]

(Equation 9)

In this way, the X coordinate x of the peak position can be obtained with the precision of the pixel or less by the equation (9). Next, the image processing apparatus 5 determines the amount of deviation between the center coordinate of the positioning mark Fm and the X coordinate x of the peak position, that is, the mark F
It is determined whether or not the absolute value of the shift amount in the direction parallel to the mark Fm, which is obtained by subtracting the X coordinate x of the peak position from the center coordinate of m, is 6 pixels or more (step 307). Further, the image processing device 5 determines whether or not the maximum value (correlation value at the peak position x) of the correlation value COR (xs) is 0.5 or less (Step 308).

When the absolute value of the shift amount in the direction parallel to the positioning mark Fm is 6 pixels or more, or when the maximum value of the correlation value COR (xs) is 0.5 or less, the image processing device 5 The current search line at coordinates ys is
Is determined as a line for which no positioning mark could be detected (step 309). The line that could not be detected is likely to be searched for another position that does not correspond to the positioning mark Fm when the absolute value of the shift amount is 6 pixels or more, and the correlation value COR ( When the maximum value of xs) is 0.5 or less, it is because the correlation with the positioning mark Fm is insufficient.

Next, the image processing apparatus 5 sets the search position coordinates y
It is determined whether or not s has reached the end value (step 31).
0). Since the end value at this time is the maximum value in the search range shown in FIG. 10, it is the coordinate position +2 corresponding to the positioning mark Fm. When the search position coordinate ys has not reached the end value, the image processing device 5 increments the search position coordinate ys by 1 (step 311), and performs the processing of steps 302 to 309 again. Such calculation of the peak position is repeated until the search position coordinate ys reaches the end value.

After the search position coordinate ys reaches the end value, the image processing device 5 calculates the coordinates of the second positioning mark Fp in the divided area Ep corresponding to the second positioning mark Fm (step 312). ). This positioning mark F
For the coordinates in the direction perpendicular to the mark Fm of p, the mark F
The center coordinates of m are used as they are. Further, the image processing device 5
Calculates the average value of the X-coordinate x of the peak position from three lines parallel to the mark Fm excluding the line with the largest amount of deviation and the line with the smallest amount of deviation in the search range, and calculates this average value as the positioning mark Fp The coordinates in the direction parallel to the mark Fm. With such a coordinate calculation method, position correction based on a coordinate conversion formula described later is performed only in a direction parallel to the mark Fm. If an undetected line exists in the process of step 309, it is determined that the positioning mark Fp could not be detected because the average value of three lines cannot be obtained.

Step 202 (301 to 301) as described above
The processing of 12) is performed on all the second positioning marks Fm (Fm1 to Fm4) set in the divided area Em. In the present embodiment, only the positioning mark Fm parallel to the X axis is described, but the same applies to the case where the positioning mark Fm is parallel to the Y axis. However, if the mark Fm is parallel to the Y axis, step 202
In the above description regarding the search position coordinates xs and ys
Are replaced, the coordinate x of the peak position becomes y, and xsm
Becomes ysm.

Thus, as shown in FIG. 8C, the positioning marks Fm (Fm1 to Fm4) set in advance in the divided area Em of the master pattern M correspond to the positioning marks Fp (Fp1 to Fp4) corresponding thereto. In the corresponding divided region Ep of the pattern P to be measured as shown in FIG.

Next, the image processing device 5 calculates the coordinate conversion formula between the pattern to be measured and the master pattern by using the coordinates of the positioning marks Fm1 to Fm4 and the coordinates of the corresponding positioning marks Fp1 to Fp4. Is obtained by the least squares method (step 203 in FIG. 7). Xm = αXp + βYp + γ Ym = δXp + εYp + ζ (10)

In equation (10), Xm and Ym are the X and Y coordinates of the master pattern, Xp and Yp are the X and Y coordinates of the pattern to be measured, and α, β, γ, δ, ε, and ζ are constants. Next, the image processing device 5 performs positioning marks Fp1 to Fp4.
The coordinates of an arbitrary positioning mark, for example, the mark Fp1, are substituted into the coordinate conversion formula of Expression (10) as Xp and Yp, and the coordinates Xm and Ym are calculated. Then, a deviation between the coordinates of the positioning mark Fm1 corresponding to the positioning mark Fp1 substituted into the coordinate conversion formula and the calculated coordinates Xm, Ym is obtained for each X, Y coordinate. The calculation of such a deviation is performed for each positioning mark (step 204).

Subsequently, the image processing apparatus 5 determines whether each of the calculated deviations is larger than a predetermined threshold (Step 205).
When the distortion of the divided region Ep of the pattern to be measured is within the allowable range,
In addition, it is determined that the derived coordinate conversion formula is appropriate, and the coordinate conversion of the master pattern in the divided area Em is performed using this coordinate conversion formula (step 206).

When all the deviations are larger than the predetermined threshold value, the distortion of the divided area Ep of the pattern to be measured is out of the allowable range, and the tape carrier 1 to be inspected is defective. Is determined (step 207).
On the other hand, when the deviation smaller than the threshold value and the deviation larger than the threshold value are mixed, the image processing apparatus 5 excludes the positioning mark having the deviation larger than the threshold value, for example, the mark Fp4 and the corresponding positioning mark Fm4. (Step 208), the remaining positioning marks Fm1 to Fm3, Fm
From the coordinates of p1 to Fp3, the coordinate conversion equation of equation (10) is obtained again (step 203).

Steps 203 to 205, 2 as described above
Steps 07 and 208 are repeated until each deviation becomes equal to or less than a predetermined threshold value. In this manner, the coordinate conversion equation of equation (10) is determined, and the master pattern can be converted in step 206. Using a coordinate transformation equation such as equation (10) is a so-called affine transform
n), whereby the positional deviation between the divided region Em and the divided region Ep can be corrected. In addition,
Since the second and third master patterns are created from the first master pattern, the alignment between the first to third master patterns and the pattern to be measured is performed once using the first master pattern. Good.

Further, in order to obtain the coordinate conversion equation of equation (10), at least three second positioning marks are required for each of the master pattern and the pattern to be measured. However, since the accuracy of the coordinate conversion formula deteriorates at every three points, a second positioning mark of at least four points is designated, and the second positioning mark whose deviation is larger than the threshold value is determined by the coordinate conversion formula. It is excluded from derivation. Therefore, the second positioning mark is 3 for both the master pattern and the pattern to be measured.
If the deviation does not fall below the threshold value even if the points are point by point, it is not possible to determine the coordinate conversion equation by setting the second positioning mark to two points at a time. Is determined to be defective.

After the positioning process in step 109 is completed, the image processing apparatus 5 sets the divided area of the pattern to be measured, the second and third master patterns corresponding to the divided area, and the fourth and fifth master patterns described later. (Step 110)
To 113). The inspections in steps 110 to 113 are simultaneously performed by the hardware of the image processing apparatus 5.

First, the inspection (step 110) by comparison with the second master pattern M1 will be described. FIG.
FIG. 2 is a view for explaining this inspection method. In the example of FIG. 12, the portion other than the satin-finished pattern NP is the pattern to be measured P. As shown in FIG. 12, the image processing device 5 compares the divided area Ep of the pattern to be measured P with the corresponding divided area Em of the second master pattern M1 whose position has been corrected by the coordinate conversion formula. However, what is actually compared is the pattern NP obtained by logically inverting the pattern P to be measured and the second master pattern M1.

Pattern NP and second master pattern M1
The result of this logical product differs depending on whether the pattern P to be measured has a defect or a disconnection. For example, when the pattern to be measured P has “1” as its value and the master pattern M1 has “1”, if the pattern to be measured P has no loss or disconnection, the pattern NP and the master pattern M1 Since there is no overlap, the result of this logical product is “0”.

On the other hand, if there is a defect C in the pattern P to be measured as shown in FIG. 12, the pattern NP and the master pattern M1 overlap in this portion, and the result of the logical product is "1".
Becomes This is the same when the pattern to be measured has a pinhole H or a disconnection. In this way, it is possible to detect a defect, a pinhole or a disconnection in the pattern to be measured. Then, the image processing device 5 determines that the result of the logical product is “1”.
And the position recognized as a defect candidate (C, H in FIG. 12)
Position) is stored.

Next, the inspection (step 111) by comparison with the third master pattern M2 will be described. FIG.
FIG. 3 is a diagram for explaining this inspection method. As shown in FIG. 13, the image processing device 5 compares the divided area Ep of the pattern to be measured P with the corresponding divided area Em of the third master pattern M2 whose position has been corrected by the coordinate conversion formula. Similarly to the above, when the logical product of the measured patterns Pa and Pb and the third master pattern M2 is obtained, the result of the logical product differs depending on whether the measured patterns Pa and Pb have a protrusion or a short circuit. That is, the pattern to be measured P
If there is no protrusion or short circuit in a and Pb, the result of the logical product is “0”.

On the other hand, if there is a protrusion K in the pattern Pa to be measured as shown in FIG.
Since “a” and the master pattern M2 overlap, the result of the logical product is “1”. Similarly, if there is a short circuit S between the patterns Pa and Pb to be measured, the result of the logical product becomes “1”. The same applies to the case where a scattering exists in the pattern to be measured. In this way, it is possible to detect a protrusion, a scatter or a short circuit of the pattern to be measured. Then, the image processing device 5 stores the position (the position of K and S in FIG. 13) at which the result of the logical product becomes “1” and is recognized as a defect candidate.

Next, the inspection (step 112) by comparison with the fourth master pattern will be described. FIG. 14 is a diagram for explaining this inspection method. FIG. 14 (a)
In the pattern to be measured P shown in FIG. 5, when the splash F exists between the area corresponding to the second master pattern M1 and the area corresponding to the third master pattern M2, the splash F cannot be detected by the above inspection. The fourth master pattern is for detecting such scattering F.

The image processing device 5 expands the second master pattern M1 in a direction perpendicular to the center line L (FIG. 14).
(B)). At this time, the image processing device 5 expands the master pattern M1 by a predetermined number of pixels. Subsequently, the image processing apparatus 5 calculates the logical product of the result of expanding the master pattern M1 and the pattern P to be measured. In this case, since all the pixels of the pattern P to be measured in the area corresponding to the master pattern M1 after the expansion processing (the area overlapping the pattern M1 in FIG. 14B) are “1”, the master pattern M1 after the expansion processing is used. Are all "1" in the area corresponding to. Therefore, the result of the logical product is the same as the master pattern M1 after the expansion processing,
(FIG. 14 (c)).

Subsequently, the image processing device 5 expands the fourth master pattern M3 in a direction perpendicular to the center line L (FIG. 14D). At this time, the image processing device 5 expands the master pattern M3 by a predetermined number of pixels. Then, the image processing device 5 takes the logical product of the result of expanding the master pattern M3 and the pattern P to be measured. In this case, the area corresponding to the master pattern M3 after the expansion processing (FIG. 14)
Since all the pixels of the pattern P to be measured in the area overlapping the master pattern M3 in (d) are “1”, the result of the logical product in the area corresponding to the master pattern M3 after the expansion processing is all “1”. . Therefore, the result of the logical product is the same as the master pattern M3 after the expansion processing, and is set as a new fourth master pattern M3 (FIG. 14E).

When the dilation process and the logical product process are repeated, the size of the fourth master pattern M3 approaches the size of the pattern P to be measured.
(F), as shown in FIG. 14 (g), it becomes the same as the pattern to be measured P (however, if the scattering F exists in the pattern to be measured P, it is not completely the same). And
Even if the same processing is repeated thereafter, the fourth master pattern M3 does not become larger than the pattern to be measured P.

The reason is that the pixel outside the edge of the pattern P to be measured is “0”. Therefore, even if the master pattern M3 shown in FIG. This is because the result of the logical product outside the edge of P does not become “1”. The scatter F exists in the pattern P to be measured.
Is not connected to the pattern to be measured P, so that the pixel “1” due to the scattering F does not appear in the fourth master pattern M3.

Next, the image processing device 5 executes the final fourth
The exclusive OR of the master pattern M3 and the pattern to be measured P is calculated. The result of the exclusive OR differs depending on whether or not the measured pattern P has scattering. When there is no scattering in the pattern to be measured P, since the pattern to be measured P and the master pattern M3 are the same, the exclusive OR results are all “0”.

On the other hand, if there is a scatter in the pattern P to be measured, there is a difference between the pattern P to be measured and the master pattern M3, and there is a pixel whose exclusive OR result is "1". Become. In this way, it is possible to detect a splatter present in an area that does not correspond to the second and third master patterns M1 and M2. Then, the image processing apparatus 5 stores the position where the result of the exclusive OR is “1” and recognized as a defect (the position of F in FIG. 14).

Next, the inspection (step 113) by comparison with the fifth master pattern will be described. FIG. 15 is a diagram for explaining this inspection method. FIG. 15 (a)
When a pinhole H exists between the area corresponding to the second master pattern M1 and the area corresponding to the third master pattern M2 in the pattern to be measured P shown in FIG. Can not. The fifth master pattern is for detecting such a pinhole H.

First, the image processing device 5 logically inverts the third master pattern M2. Thereby, the pixel “1”
Becomes “0” and the pixel “0” becomes “1”, the result of logically inverting the master pattern M2 is shown in FIG. 15B in which the pattern edge and the inside thereof are filled with the pixel “1”.
A pattern M4 as shown in FIG.

Subsequently, the image processing device 5 sets the pattern M4
In the direction perpendicular to the center line L (FIG. 15)
(C)). At this time, the image processing device 5 sets the pattern M4
Is reduced by a predetermined pixel. Then, the image processing device 5
Is the result of contracting the pattern M4 and the pattern to be measured P
And the logical sum of The result of this OR is the same as the pattern M4 after the contraction processing, and this is set as a fifth master pattern M5 (FIG. 15D).

Next, the image processing apparatus 5 contracts the fifth master pattern M5 in a direction perpendicular to the center line L (FIG. 15 (e)). At this time, the image processing device 5 contracts the master pattern M5 by a predetermined pixel. And
The image processing device 5 calculates the logical sum of the result of contracting the master pattern M5 and the pattern P to be measured. The result of the logical sum is the same as the master pattern M5 after the contraction processing, and is set as a new fifth master pattern M5 (FIG. 15 (f)).

When such a contraction process and a logical sum process are repeated, the size of the fifth master pattern M5 approaches the size of the pattern P to be measured, and finally, the size of the pattern shown in FIG.
As shown in (g), the pattern becomes the same as the pattern to be measured P (however, when the pattern to be measured P has a pinhole H, it is not completely the same). Then, even if the same processing is repeated thereafter, the fifth master pattern M5
Does not become smaller than the pattern to be measured P.

The reason is that since the pattern to be measured P is filled with the pixel “1”, even if the master pattern M5 shown in FIG. Is not "0". In addition,
Although the pinhole H exists in the pattern to be measured P, but the pinhole H is not connected to the edge of the pattern to be measured P, the pixel “0” due to the pinhole H does not appear in the fifth master pattern M5. .

Next, the image processing apparatus 5 sets the final fifth
The exclusive OR of the master pattern M5 and the pattern to be measured P is calculated. The result of the exclusive OR differs depending on whether the pattern P to be measured has the pinhole H or not. When there is no pinhole in the pattern to be measured P, the pattern to be measured P and the master pattern M5 are the same, and the exclusive OR results are all “0”.

On the other hand, when the pinhole H exists in the pattern P to be measured, there is a difference between the pattern P to be measured and the master pattern M5, and there is a pixel whose exclusive OR result is "1". Will be. Thus, the second and third
Pinholes H present in areas not corresponding to the master patterns M1 and M2 of FIG. Then, the image processing device 5 stores the position (the position H in FIG. 15) at which the result of the exclusive OR is “1” and the defect is recognized.

After performing the primary inspection as described above, the image processing apparatus 5 outputs the stored position of the defect candidate as address information. When a defect candidate is detected by the first image processing device 5 (YES in step 114), the second image processing device 6 determines a predetermined size smaller than the divided area around the defect candidate at the position indicated by the address information. A second inspection of the pattern to be measured is performed by comparing the pattern to be measured and the first master pattern to determine an error in the region having the size of (1) (step 115). This inspection method is the same as the above-described conventional method shown in FIGS.

The processing of steps 109 to 115 is repeated until there is no unchecked divided area (step 11).
6) Perform for each divided area. Note that the second positioning mark Fm does not have to be set in all divided areas of the master pattern. For example, in the peripheral portion near the first positioning mark Tm, the positional deviation between the master pattern and the pattern to be measured is small, so that the alignment for each divided region may be omitted. That is, the image processing apparatus 5 omits the alignment for each of the divided regions Ep of the measured pattern corresponding to the divided region Em of the master pattern in which the second positioning mark Fm is not set.

The comparison inspection between each of the second to fifth master patterns and the pattern to be measured can be realized by hardware.
Since only a predetermined area including the detected defect candidate is inspected by comparing the measured pattern requiring a long processing time with the first master pattern, the measured pattern can be inspected at a higher speed than before.

When a local distortion is present in the tape carrier 1, a deviation from the local master pattern occurs in the conventional pattern inspection method, and this deviation is detected as a defect candidate. Even if the one local distortion is within the allowable range, the secondary inspection is performed, and the inspection time is reduced. On the other hand, in the present embodiment, if the local distortion of the tape carrier 1 is within the allowable range (that is, the deviation is equal to or less than the threshold), the tape carrier 1 is aligned by the coordinate conversion formula for each divided area. Since one local distortion is absorbed, this distortion is not detected as a defect candidate.

In FIGS. 8A and 8B, there is no overlap between the divided areas. However, the actual divided areas are set so that the left, right, top, and bottom overlap with other divided areas. Thus, it is possible to determine whether or not the distortion at the boundary between the divided regions is within the allowable range based on the deviation of the plurality of adjacent divided regions, and it is possible to inspect the connection of the divided regions. Further, in the present embodiment, the inspection of the next divided area is performed after the inspection of one divided area is completed. However, if a plurality of divided areas are inspected in parallel, it is impossible to perform a higher-speed inspection. Needless to say.

[0089]

According to the present invention, after the entire position of the master pattern and the pattern to be measured are aligned, a plurality of divided areas are set in the master pattern, and at least four virtual lines orthogonal to the pattern are set. Is set for each divided area, a divided area of the pattern to be measured corresponding to the divided area of the master pattern is extracted, and a predetermined area centered on the position corresponding to the second positioning mark is set. The peak value having the highest correlation value is regarded as the second positioning mark of the measured pattern, and the positions of the second positioning marks of the measured pattern and the master pattern are determined. By matching, the positioning of the master pattern and the pattern to be measured can be performed for each divided region. As a result, even in the case of an inspection work such as a tape carrier in which local distortion is likely to occur and there is no independent pattern that can serve as a positioning mark, local alignment of the inspection work is performed. Such a distortion can be absorbed, and a fine displacement between the divided region of the master pattern and the divided region of the pattern to be measured can be corrected, so that a normal inspection can be performed. In addition, if the local distortion of the inspection work is within the allowable range, the distortion will not be detected as a defect candidate, and the secondary inspection due to the distortion will not be performed, thereby shortening the inspection time. can do.

After the peak position is determined for each line parallel to the second positioning mark of the master pattern, the amount of deviation between the peak position and the second positioning mark of the master pattern is determined. The average value of the peak position is obtained from the remaining lines excluding the line and the minimum line, and the average position is regarded as the second positioning mark of the pattern to be measured, whereby the position of the second positioning mark of the pattern to be measured is determined. It can be determined with high accuracy.

In the divided area of the master pattern and the corresponding divided area of the pattern to be measured,
The coordinate conversion formula between the pattern to be measured and the master pattern is determined from the coordinates of the positioning mark, and the divided area of the master pattern is converted by the coordinate conversion formula to obtain the position of each divided area between the master pattern and the pattern to be measured. Matching can be realized.

Further, the second coordinate of the pattern to be measured is expressed by the coordinate conversion formula.
A deviation between the result of inputting the coordinates of the positioning mark and the coordinates of the corresponding positioning mark of the master pattern is obtained for each mark, and a second positioning mark having a deviation larger than a predetermined threshold value is determined for the pattern to be measured and the master pattern. By excluding from both sides and recalculating the coordinate conversion formula until all the deviations become equal to or less than a predetermined threshold value, and determining the coordinate conversion formula, the accuracy of the coordinate conversion formula can be increased. Accurate positioning can be performed. When all the deviations are larger than the predetermined threshold value, the distortion of the divided area of the pattern to be measured is out of the allowable range, and it can be determined that the inspection work is defective.

[Brief description of the drawings]

FIG. 1 is a flowchart of a pattern inspection method using an alignment method according to a first embodiment of the present invention.

FIG. 2 is a block diagram of a pattern inspection apparatus.

FIG. 3 shows a first master pattern for one IC and a first master pattern.
FIG. 4 is a diagram showing a positioning mark.

FIG. 4 is a diagram showing a divided area of a first master pattern and a second positioning mark for each divided area.

FIG. 5 is a diagram for explaining a method of creating second and third master patterns.

FIG. 6 is a diagram for explaining a method of aligning an entire pattern to be measured and a master pattern.

FIG. 7 is a flowchart showing a method of aligning a pattern to be measured and a master pattern for each divided region.

FIG. 8 is a diagram for explaining a method of aligning a pattern to be measured and a master pattern for each divided region.

FIG. 9 is a flowchart showing a second positioning mark search method.

FIG. 10 is a diagram showing a search range for a second positioning mark in a pattern to be measured.

FIG. 11 is a diagram for explaining a method of calculating a peak position of a correlation value.

FIG. 12 is a diagram for explaining an inspection method based on comparison with a second master pattern.

FIG. 13 is a diagram for explaining an inspection method based on comparison with a third master pattern.

FIG. 14 is a diagram for explaining an inspection method based on comparison with a fourth master pattern.

FIG. 15 is a diagram for explaining an inspection method based on comparison with a fifth master pattern.

FIG. 16 is a diagram for explaining a conventional inspection method for detecting disconnection.

FIG. 17 is a view for explaining a conventional inspection method for detecting disconnection.

FIG. 18 is a view for explaining a conventional inspection method for detecting a short circuit.

FIG. 19 is a view for explaining a conventional inspection method for detecting a defect or a protrusion.

FIG. 20 is a diagram for explaining a problem in a conventional alignment method.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Tape carrier, 2 ... 3 sprocket, 4 ... Line sensor camera, 5 ... 1st image processing apparatus, 6 ... 2nd image processing apparatus, 7 ... Host computer, 8 ... Display device.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shuzo Matsuno 3-20-1, Nishi-Shimbashi, Minato-ku, Tokyo F-Term in Nippon Avionics Co., Ltd. (Reference) 2F065 AA03 AA56 BB13 BB27 CC01 CC02 DD11 DD19 EE00 FF04 FF61 JJ02 JJ25 MM03 QQ05 QQ24 QQ25 QQ29 QQ31 QQ39 QQ41 QQ42 RR10 SS13 TT02 5B057 AA11 BA30 CC03 DA03 DA07 DA08 DC16 5L096 BA03 CA02 EA17 FA06 FA32 FA69 FA76 GA19 GA51 HA08

Claims (4)

[Claims] 1. A pattern inspection method for inspecting a measured pattern by comparing an image of a measured master pattern serving as a reference with an image of the measured pattern captured by a camera, wherein the position of the master pattern and the measured pattern are aligned. A positioning method comprising: aligning the positions of the first positioning marks with each other over the entire area of the pattern to be measured and the master pattern; A plurality of divided areas are set, and at least four virtual lines orthogonal to the pattern are set as second positioning marks for each divided area. The divided area of the pattern to be measured corresponding to the divided area of the master pattern To a predetermined range centered on the position corresponding to the second positioning mark. Calculating the correlation value between the measured pattern and the master pattern, assuming the peak position having the highest correlation value as the second positioning mark of the measured pattern, and aligning the positions of the second positioning marks of the measured pattern and the master pattern with each other. A position alignment method between the master pattern and the pattern to be measured for each divided area. 2. The pattern alignment method according to claim 1, wherein said peak position is determined for each line parallel to said second positioning mark of said master pattern within said predetermined range. The amount of deviation of the master pattern from the second positioning mark is obtained, and the average value of the peak positions is obtained from the remaining lines excluding the line with the maximum deviation and the line with the minimum deviation. A method of positioning a pattern, wherein the method is regarded as a second positioning mark. 3. The pattern alignment method according to claim 1, wherein the pattern to be measured and the master pattern are divided by the coordinates of each second positioning mark in the divided region of the master pattern and the corresponding divided region of the pattern to be measured. Determining the coordinate conversion formula between the two, and converting the divided region of the master pattern by the coordinate conversion formula, thereby performing alignment of the master pattern and the pattern to be measured for each divided region. . 4. The pattern alignment method according to claim 3, wherein a deviation between a result of inputting the coordinates of the second positioning mark of the pattern to be measured in the coordinate conversion formula and the coordinates of the corresponding positioning mark of the master pattern is calculated. For each mark, it is necessary to exclude the second positioning mark whose deviation is larger than a predetermined threshold value from both the pattern to be measured and the master pattern and to obtain the coordinate conversion formula again. A pattern alignment method characterized by determining the coordinate conversion formula by repeating the following until: