JPH08181053A - Position detecting method - Google Patents

Abstract

PURPOSE: To enhance the detecting accuracy of a pattern to be an aligning reference by executing a masking on the part of a template image, conducting the masking even on the part corresponding to the masked part, obtaining a correlation value of non-masked areas, and extracting a candidate image as a corresponding image. CONSTITUTION: An exposure area is imaged, and the part 45a of a template image 45 previously stored from the imaged image 50 is masked. Simultaneously, the part 53a corresponding to the masked part of the template image of the candidate image 53 listed as the candidate of the extracted corresponding image is also masked. Then, the non-masked area 4b of the image 45 is compared with the non-masked area 53b of the candidate image to obtain correlation value of the non-masked areas, and when the correlation value is a predetermined value or more, the image 53 is extracted as the corresponding image.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an alignment method,
For example, the present invention relates to a positioning method that can be suitably applied to an exposure apparatus that forms a large-sized large-sized liquid crystal display element by exposing a small-sized exposed area by connecting and overlapping the small-sized exposed areas on a large-sized substrate.

[0002]

2. Description of the Related Art Conventionally, a step-and-repeat type projection exposure apparatus has been used as one of exposure apparatuses for forming a large-area liquid crystal display element or the like on a substrate. This is to form a large-area device as a whole by exposing (joint exposure) a plurality of small-area exposure regions with their ends connected to each other.

However, since the projection optical system has an image forming characteristic such as distortion, the pattern formed on the connecting portion (connecting portion) between the exposure areas having a small area is likely to be deviated. For this reason, the pattern that should originally be formed may be discontinuous at the connecting portion. If this deviation is large (that is, if the joining accuracy is low), there will be a difference between the performance of the element formed in the joint portion and the performance of the element formed in the other portion, resulting in a joint portion. The contrast difference and the color unevenness were generated on the screen corresponding to, and these were sometimes perceived by human eyes.

In general, each element is formed by sequentially exposing a second layer pattern and a third layer pattern on a base pattern. Also in this case, if the overlay accuracy of the pattern which is overlaid on the underlying pattern and exposed is poor, the performance of the element is deteriorated as in the case described above.
The accuracy of joining between adjacent exposure areas like this,
In order to improve the overlay accuracy of the patterns formed on the same area, conventionally, by detecting the deviation of the mark exposed at the same time as the exposure of the pattern at a typical position outside the area of the exposed pattern, A method is known in which a correction parameter is obtained so that this deviation amount is minimized.

For example, when a glass plate as a photosensitive substrate is divided into four exposure areas and connected for exposure, the exposure areas are connected by using a reticle in which a vernier pattern is arranged in the vicinity of the outer periphery of the area where the pixel pattern is drawn. Alternatively, the vernier patterns are overlapped and exposed so that the vernier patterns overlap each other. Then, the vernier pattern images at a plurality of positions formed on the plate are read visually or by a dedicated measuring machine, and the correction parameter (2 for the photosensitive substrate or the like is used so as to minimize the read value on average). The amount of scaling, the amount of shift, the amount of rotation, etc. in the dimensional direction are calculated.

However, since the mark (vernier pattern) used for alignment (correction) does not form a pixel pattern, these marks cannot be formed in the area where the pixel pattern forming the liquid crystal display element is formed. , I had no choice but to make it around it. However, when connecting the exposure areas and exposing them in superposition,
Since the alignment state of the regions corresponding to the connecting portion and the alignment state of the peripheral region are different, it was not possible to accurately connect or overlap the patterns. In particular, it was not possible to perform sufficient alignment with respect to the image (for example, the central portion of the pixel pattern region) located at a position apart from the vernier pattern formed in the peripheral position. In addition, in order to improve the alignment accuracy, it is possible to take measures such as increasing the number of vernier patterns arranged around the substrate, but the periphery of the substrate is easily affected by the manufacturing process, and it is arranged around the substrate. Since the damage to the vernier pattern is relatively large, we could not expect a sufficient improvement in accuracy.

[0007]

Therefore, instead of using the vernier pattern arranged around the substrate as a reference for alignment, the vernier pattern is matched with a reference image (template image) registered in advance from the pixel patterns to be exposed. A method of extracting an image, detecting the position of the substrate based on the position of the image, and performing alignment has been proposed. In that case, images are taken at a plurality of locations including the connecting portion of the adjacent first and second regions, the positions of the images corresponding to the template image are detected at each of the plurality of imaged locations, and those positions are detected. By performing the alignment so as to minimize the deviation of, the joint exposure accuracy and the overlay exposure accuracy are improved.

As described above, when the pixel pattern forming the liquid crystal display element is used as the position detecting pattern, there is an advantage that a special pattern such as a vernier pattern is not required on the substrate. For example, the manufacturing process is likely to be affected as described below. (1) In the case where another pattern exists close to the characteristic pattern collected as the template image or the image to be processed. (2) In the case where the edge of the pattern collected as the template image or the image to be processed has an inclination, or the like, the outline of the pattern becomes an ambiguous blurred image on the image. (3) In the case where an offset of the brightness difference is applied to the entire sampled image data due to the difference in the manufacturing process or the difference of the image capturing device, and the difference in the brightness of the screen occurs.

In the above case, there are problems such as erroneous position detection, poor position detection accuracy, and errors. The present invention has been made in view of the above problems of the conventional position detection method, and a pattern serving as a positioning reference when exposing small-area exposure regions by joining or overlapping. It is an object of the present invention to provide a new and improved alignment method capable of increasing the detection accuracy of the, and further increasing the connection accuracy and the overlay accuracy.

[0010]

In order to solve the above-mentioned problems, the present invention takes an image on a photosensitive substrate (5) having an exposure area to which an original image pattern formed on an exposure original plate (R) is transferred. In order to detect the position of the photosensitive substrate (5) by detecting the information (53, 56) corresponding to the image information regarding the template (45, 47) serving as the reference from the image information (50, 55) obtained as a result. , Image the exposure area,
Corresponding images (5) corresponding to template images (45, 47) stored in advance from the captured images (50, 55).
3, 56), first, part of the template image (45, 47) (45a, 47a) is masked, and the candidate images (53, 56) listed as candidates for the corresponding image to be extracted are extracted. Of these, the portions (53a, 56a) corresponding to the masking portions of the template image are also masked. Then, the non-masking area (45b, 47b) of the template image (45, 47)
And a non-masking area (53b, 56b) of the candidate image are compared to obtain a correlation value between the non-masking areas, and when the correlation value is equal to or more than a predetermined value, the candidate image (53, 56) is extracted as a corresponding image. ing.

Before performing a masking process on a part of the image in order to extract the corresponding image of the template image as described above, all images of the template image (45, 47) and all images of the candidate image (53, 56), the correlation value between all the images is obtained, and when the correlation value between all the images is lower than a predetermined value, the masking process may be performed. When performing the masking process, the part of the template image to be masked (45
It is also possible to preset a, 47a). Alternatively, the template image is previously set in a plurality of regions (46a-
46c, 48a to 48d), the candidate image is also divided into a plurality of regions corresponding to the template image, and the corresponding divided regions are compared to obtain a correlation value between the divided regions, The lowest value (46
It is also possible to apply a masking treatment to c, 48a).

According to another aspect of the present invention, image information (50, 55) obtained by picking up an image on a photosensitive substrate (5) having an exposure area to which an original image pattern formed on an exposure original plate (R) is transferred. ) As a reference template (45,
Information corresponding to the image information regarding (47) (53, 56)
To detect the position of the photosensitive substrate (5), the exposure area is imaged, and the captured image (50, 5
5) template images (45,
When extracting the corresponding image (53, 56) corresponding to 47), first, the average value of the gray level of the predetermined area (49) included in the template image and the template image of the candidate image listed as the candidate of the corresponding image are extracted. The average value of the gray level of the area (57) corresponding to the predetermined area is compared. The difference between the average gray level of the template image and the average gray level of the candidate image is corrected as an offset error (49a) to obtain the correlation value between the template image and the candidate image, and the correlation value is not less than a predetermined value. In this case, the candidate image is identified and extracted as the corresponding image.

Further, the present invention is also applied to the case where the correction amount of the joint error or the overlay error at the time of performing the alignment is determined by using the corresponding image extracted by the above method.
The first case relates to a connection error, and first, a reference is made from image information obtained by capturing an image on a photosensitive substrate (5) having an exposure area to which an original image pattern formed on an exposure original plate (R) is transferred. To detect the position of the photosensitive substrate (5) by detecting information (28A, 28B) corresponding to the image information (27) relating to the template that becomes the first and second exposure areas adjacent to each other in the exposure area. Imaging is performed at a plurality of points including the connecting portion 24A of the regions (for example, 22A and 22B).

Next, at each of the plurality of imaged locations, the first and second exposure areas (for example, 22A and 2A) are formed.
2B) corresponding images (28A, 28B) corresponding to the template image (27) are extracted by the above method. Further, the difference between the position (X1, Y1) of the corresponding image in the first exposure area (22A) and the position (X2, Y2) of the corresponding image in the second exposure area (22B) is calculated as the first and second differences.
Of the exposed areas (for example, 22A and 22B) (2
4A) as a connection error (XST, YST). Then, a plurality of imaged locations (28A, 28
B) A plurality of joint errors (X
The correction amount at the time of alignment is determined so that (ST, YST) becomes the average minimum.

The second case relates to the overlay error and is obtained by first imaging the photosensitive substrate (5) having the exposure area to which the original image pattern formed on the exposure original plate (R) is transferred. From the image information, the information (3
5A, 35B, 37A, 37B) to detect the position of the photosensitive substrate (5) in order to detect the position of the photosensitive substrate (5).
And 22B) including the connecting portion 24A are imaged at a plurality of locations.

Next, at each of the plurality of imaged locations, the first and second exposure areas are respectively replaced by the first area.
First corresponding images (35A, 3A) corresponding to the first template image (34) of the layer pattern (31A, 31B).
5B) and the second of the patterns of the second layer (32A, 32B).
The second corresponding images (37A, 37B) corresponding to the template image (36) of (3) are extracted by the above method. Further, the difference (X1-X2, Y1-Y2) between the position of the first corresponding image (35A) and the position of the second corresponding image (37A) in the first exposure area (22A), and the second exposure area (22
The difference (X1'-X2 ', Y1'-Y) between the position of the first corresponding image (35B) and the position of the second corresponding image (37B) in B).
2 ') is determined as the overlay error (XOV, YOV) generated in the joint portion (24A) of the first and second exposure areas (for example, 22A and 22B). Then, the correction amount at the time of alignment is determined so that the plurality of overlay errors (XOV, YOV) obtained at each of the plurality of imaged locations that have been imaged are minimized on average.

[0017]

First, the corresponding image (5) corresponding to the template image (45, 47) stored in advance from the exposure pattern
The exposure area (50, 55) including the area (3, 56) is imaged. Next, of the candidate images (53, 56) listed as candidates to be extracted as the corresponding images, a portion (53a) in which another pattern is present close to the candidate image or a portion with a blurred edge portion ( Masking is applied to portions that may cause erroneous detection or poor detection accuracy, such as 56a). At the same time template images (45, 4
Masking is also applied to the corresponding portions (45a, 47a) of 7). Then, the non-masking regions of the template image and the candidate image (45b and 53b, 47b and 56b).
The correlation value is obtained by comparing b). As described above, according to the present invention, when obtaining the correlation value, the portion (masking region) that causes erroneous detection or poor detection accuracy is excluded from the comparison target in advance, so that the accurate correlation value can be obtained. Then, when the correlation value is equal to or larger than the predetermined value, the candidate image (53, 56) is extracted as the corresponding image. As a result,
The alignment pattern can be detected with high accuracy, and the alignment accuracy between the original image pattern and the photosensitive substrate can be improved.

Before performing a masking process on a part of the image to extract the corresponding image of the template image as described above, first, all the images of the template image and the candidate image (45 and 53, 47 and 56). Can also be compared to obtain the correlation value. Then, when the correlation value between all the images is lower than a predetermined value, it is assumed that an area that causes erroneous detection or poor detection accuracy in the image, that is, an area to be masked and excluded from the comparison target is included. Judgment is made and masking processing is performed.

When it is possible to predict a region (a part where another pattern exists in the vicinity of the candidate image or a blurred part) that causes erroneous detection or poor detection accuracy, that part (45a, 45a, 47a) can be preset as a masking area. Alternatively, when it is not possible to predict the area that causes erroneous detection or poor detection accuracy, the template image and the candidate image are automatically divided into a plurality of areas (46a to 46c, 48a to 48).
d), the corresponding divided areas are compared with each other to obtain the correlation value, and it is determined that there is an area that causes erroneous detection or poor detection accuracy in the portion (46c, 48a) having the lowest correlation value, It is also possible to apply a masking process to the area.

Further, when the difference in the image capturing device causes an offset of the brightness difference in the entire sampled image data, resulting in a difference in brightness and darkness of the screen, the following correction operation is performed. First, from the exposure pattern, the corresponding image (5
The exposure area (50) including 7) is imaged. Next, the average values of the gray levels of the predetermined area of the template image (49) and the predetermined area (57) of the candidate image that is listed as a candidate for the corresponding image are compared. The predetermined area in which the average gray levels are compared may be the entire image or a part thereof. Then, when the correlation value between the template image and the candidate image is obtained and the candidate image is extracted as the corresponding image based on the correlation value, the difference between the template image and the average gray level of the candidate image is corrected as an offset error. By performing (49a), the accuracy of extraction can be improved.

When determining the correction amount of the connection error or the overlay error when performing the alignment using the corresponding images extracted by the above method, the following operation is performed.
In the case of correcting the connection error, first of all, the adjacent first of the exposure patterns of the small area with respect to the pattern of the large area.
Also, the joint portion (24A) of the second exposure area (for example, 22A and 22B) is imaged at a plurality of locations at the same time. Then, an image (28) corresponding to the template image (27) is obtained from each of the plurality of images obtained by imaging by the masking correction method or the brightness offset error correction method.
A, 28B) are detected in plural. Next, the first exposure area (2
The position (X1, Y1) of the image (28A) detected from the area corresponding to 2A) and the position (X1, Y2) between the corresponding images (28B) in the area corresponding to the second exposure area (22B). Difference between the first and second exposure areas (22A
And 22B), a plurality of joint errors (XST, YST) generated in the joint portion (24A) are obtained. Further, the correction amount at the time of alignment is determined so that these connection errors (XST, YST) are averagely minimized. As a result, it is possible to perform the joining based on the correction amount having a higher accuracy than in the conventional case.

When the overlay error is corrected, first, the first and second exposure regions (for example, 22) adjacent to the large-area pattern in the small-area exposure pattern are adjacent to each other.
The joint portion (24A) of A and 22B) is imaged at a plurality of locations at the same time. The images (35A, 35B, 37A, 37B) corresponding to the first and second template images (34, 36) set in advance for the first layer and the second layer are selected from the plurality of images obtained by imaging. ) Is detected by the masking method or the luminance offset error correction method. Next, the position difference (X1) of the images (35A, 37A) detected from the area corresponding to the first exposure area (22A)
-X2, Y1-Y2) and the image (35B, 37B) detected in the area corresponding to the second exposure area (22B).
Difference (X1'-X2 ', Y1'-Y2') between the two positions of the exposure area (for example, 22A and 22B) at the joint (24A) between the overlapping error (XOV, YOV) As a plurality of overlapping errors (XOV, Y
The correction amount for alignment is determined so that the OV) becomes the minimum on average. As a result, it is possible to perform the joining based on the correction amount having a higher accuracy than in the conventional case.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT An embodiment of the alignment operation of a projection exposure apparatus according to the present invention will be described in detail below with reference to the accompanying drawings. (1) Overall Configuration of Projection Exposure Apparatus In FIG. 1, reference numeral 1 denotes a projection exposure apparatus for manufacturing a large-sized liquid crystal display device that uses image processing for alignment as a whole, and exposure light emitted from a light source 2 such as an ultra-high pressure mercury lamp. Is condensed by the elliptical mirror 3 and then is incident on the condenser lens system 4 via an optical integrator or the like.

The exposure light properly condensed by the condenser lens system 4 illuminates the reticle R with a substantially uniform illuminance.
The exposure light projects the pattern of the reticle R onto the photosensitive substrate, that is, each shot area on the glass plate 5 coated with the photosensitive agent, via the projection optical system PL. After that, a large-sized liquid crystal display element is manufactured by sequentially stacking a plurality of layers of circuit patterns on the glass plate 5 and exposing them.

Here, the glass plate 5 is held on the Z stage 6, and the Z stage 6 is placed on the XY stage 7. The XY stage 7 positions the glass plate 5 in a plane (XY plane) perpendicular to the optical axis of the projection optical system PL, and the Z stage 6 positions the glass plate 5 in the projection optical system PL.
Positioning is performed in the optical axis AX (Z axis) direction. Incidentally, a θ table for rotating the glass plate 5 is interposed between the Z stage 6 and the glass plate 5. A reference mark assembly 8 having various alignment marks formed thereon is fixed near the glass plate 5 on the Z stage 6. A movable mirror 9 for distance measurement in the X and Y directions is fixed near the reference mark assembly 8.

The moving mirror 9 is irradiated with a laser beam from the laser interferometer 10, and the laser interferometer 10 receives the reflected light reflected by the mirror surface so that the position of the XY stage 6 is constantly measured. There is. Drive device 1
Reference numeral 1 drives the XY stage 7 based on the coordinate values and the like measured by the laser interferometer 10. When the reticle R is aligned, the alignment microscope 12 irradiates the alignment mark RM near the pattern area of the reticle R with the alignment light via the mirror 13. The reflected light from the alignment mark RM is reflected by the mirror 13 and returned to the microscope 12. Then, for example, the reticle R is aligned by adjusting the position of the reticle R based on the position of the image of the alignment mark RM that is re-imaged inside the alignment microscope 12.

Further, the alignment mark RM of the reticle R and the alignment mark in the reference mark assembly 8 on the Z stage 6 are simultaneously observed by the alignment microscope 12, and the reticle R is aligned according to the positional relationship between the images of both. good. Further, it is possible to simultaneously observe the alignment mark of the reticle R and the alignment mark on the glass plate 5 with the alignment microscope 12 to obtain the positional relationship between them.

The autofocus detection system comprises a light transmitting system 14 and a light receiving system 15, and the light transmitting system 14 includes the glass plate 5
The image of the detection pattern such as the slit pattern is projected obliquely to the optical axis AX of the projection optical system PL. The reflected light of the image of the detection pattern re-images the image of the detection pattern in the light receiving system 15. The height of the exposed surface of the glass plate 5 is obtained from the amount of positional deviation of the image of the re-formed detection pattern, and the height of the exposed surface of the glass plate 5 is determined by the Z stage 6 to be the best focus for the projection optical system PL. It is configured to be set to a position.

An alignment optical system for image processing is arranged beside the projection optical system PL. The alignment optical system is configured to guide the reflected light reflected from the observation area on the glass plate 5 to the relay lens 18 via the objective lens 16 and the mirror 17. Then, this reflected light is focused on the image pickup surface of a CCD camera 19 using a charge-coupled image pickup device (CCD), and an image of the pattern of the observation region is formed on the image pickup surface.

The pattern matching device 20 takes in a video signal (imaging signal) output from the CCD camera 19 as an image to be processed. The pattern matching device 20 stores the template image in the memory and obtains the coordinates of the pattern that matches the template by pattern matching. By using this coordinate value, the amount of deviation that occurs during joint exposure or overexposure is obtained, and is given as a correction parameter to the drive device 11 during the next and subsequent exposures,
It is designed to improve the alignment accuracy.

Further, the correction parameter obtained here is configured to be able to correct the image forming characteristic of the projection lens and be fed back to the reticle alignment system. In addition,
The image stored as the template image is a part of the pixel pattern formed on the glass plate 5. (2) Method of Improving Detection Accuracy by Processing Image Data In position detection by pattern matching as described above, when a pixel pattern that constitutes a liquid crystal display element is used as a pattern, a special pattern such as a vernier pattern is formed on a glass plate. Although it has the advantage that it is not necessary, it is affected by the manufacturing process. For example, in the following cases.

First, as shown in FIG. 2A, a characteristic pattern 52 in an image 50 collected as an image to be processed.
When another pattern 51 is present in the vicinity of, the composite image 53 of the characteristic pattern 52 and the other pattern 51 is taken up as a candidate image, and the image and the template image 45 serving as a reference (Fig. 2 (b)) will be pattern-matched. Then, since the correlation value with the template image 45 is low due to the existence of another pattern 51, the pattern matching device 20 may determine that the image is unsuitable even though the image is originally correct. .

Further, as shown in FIG. 3 (a), there is a large blur variation on the image because of a large variation in inclination at a part 51A of the edge of the pattern 51 in the image 55 collected as the image to be processed. In the case of the image, the correlation value between the template image 47 (FIG. 3B) and the candidate image 56 becomes low, and the pattern matching device 20 makes the image unsuitable even though it is an originally correct image. May be determined to be.

Further, as shown in FIG. 4A, when the difference in the manufacturing process and the difference in the image capturing device cause an offset of the brightness difference in the entire sampled image data, resulting in a difference in the brightness of the screen. In addition, due to the difference in gray level between the template image 49 and the candidate image 57, the correlation value between the template image 49 and the candidate image 57 is low, and the pattern matching device 20 considers that the image is unsuitable even though the images are originally correct. There is a risk of making a decision.

Conventionally, due to these reasons, there have been problems such as erroneous position detection, poor position detection accuracy, and stop of execution of the projection exposure apparatus due to error occurrence. The coping method will be described below with reference to the flowcharts shown in FIGS. 2, 3, and 5 to 8.

5 to 7 for a masking processing method applicable when another pattern exists near the candidate image to be extracted first, or when an unclear portion exists at the edge of the candidate image, etc. While explaining. First, in step 151, the template image 45 previously recorded from the exposure area by the CCD camera 19 or the like,
Candidate image 5 considered to be a candidate for corresponding image corresponding to 47
3, 56 are captured as video signals (imaging signals). Next, at step 152, template images 45, 4
All images of No. 7 and all images of candidate images 53 and 56 are compared to obtain a correlation value. Then, in step 153, when the correlation value indicates a predetermined value or more, the candidate image 5
Since it can be determined that Nos. 3 and 56 correspond to the template images 45 and 47, the process proceeds to step 154, and the candidate images 53 and 56 are taken into the pattern matching device 20 and the position detection by pattern matching can be performed.

On the other hand, in step 153,
If the correlation value does not reach the predetermined value, the extracted candidate image is erroneous, or another pattern exists near the candidate image, or there is an unclear part in the candidate image. Since it is considered that the correlation value showed a low value although it was originally a correct image, step 161
In the following, image data processing (masking processing) based on the present embodiment is performed.

First, in the design stage, when it is assumed in advance that another part exists in the vicinity of the candidate image or the edge of the candidate image has an inclination and that part is blurred and unclear. Advances from step 161 to step 162, and masks the portions 53a and 56a of the candidate images 53 and 56 which are considered to be the cause of erroneous detection. Template images 45 and 47 at the same time
Of the parts 45a and 47a (FIG.
Masking is also applied to (c) and FIG. 3 (c). After performing the masking process in this way, step 1
In step 71, the non-masking portions of the template image and the candidate image are compared with each other in the correlation values of 45b and 53b and 47b and 56b. Then, in step 172, it is determined again whether the correlation value is equal to or more than the predetermined value. If the correlation value is equal to or more than the predetermined value, it can be determined that the candidate images 53 and 56 correspond to the template images 45 and 47. Therefore, the process proceeds to step 173, and the candidate image is taken into the pattern matching device 20 and the position can be detected by the pattern matching. If the correlation value does not reach the predetermined value even though the masking process is performed in step 172, it can be determined that the extracted candidate image is erroneous, and thus the process proceeds to step 174 and another The candidate image is extracted, and the flow from step 152 onward is repeated.

On the other hand, if the portion causing the erroneous detection is not known in advance, the process proceeds from step 161 to step 163 and, as shown in FIGS. And the template image are divided into a plurality of corresponding regions (46a to 46c, 48a to 48d), and a correlation value is obtained for each region. Although the image is divided into three in the example of FIG. 2D and the image is divided into four in the example of FIG. 3D, it goes without saying that the number of divided images and the divided shape can be set arbitrarily. Then, in the region having the lowest correlation value (for example, the region of 46c in the example of FIG. 2D, and the region of 48a in the example of FIG. 3D), another pattern or a blurred portion that causes erroneous detection is present. Since it is considered to be present, masking is applied to the area in step 164. After performing the masking process in this way, the process proceeds to step 171, and the correlation values of the unmasked portions of the template image and the candidate image are compared. Then, in step 172, it is determined again whether the correlation value is equal to or more than the predetermined value. If the correlation value is equal to or more than the predetermined value, it can be determined that the candidate images 53 and 56 correspond to the template images 45 and 47. Therefore, the process proceeds to step 173, and the candidate image is taken into the pattern matching device 20 and the position can be detected by the pattern matching. If the correlation value does not reach the predetermined value even though the masking process is performed in step 172, it can be determined that the extracted candidate image is erroneous, and thus the process proceeds to step 174 and another The candidate image is extracted, and the flow from step 152 onward is repeated.

As described above, according to this embodiment,
As shown in FIGS. 2 (c) and 3 (c), a proximity pattern or an unclear portion that causes erroneous detection is removed from the comparison target by masking processing, and then the correlation value between the template image and the candidate image is calculated. Because it compares, like the conventional method,
It is possible to avoid erroneous detection such that an originally correct candidate image is truncated, and improve the position detection accuracy.

In addition to the methods shown in the above-mentioned embodiments, those skilled in the art have come up with various image processing methods for removing the proximity pattern or the unclear portion which causes the erroneous detection from the comparison object by the masking process. Is clear to get
It is understood that those methods are also included in the scope of the technical idea of the present invention described in the claims. For example, if there is a difference in the tendency that erroneous detection may occur in a plurality of processed images on the same substrate, some template images that have been partially deleted (masked) are prepared, and which processed image is to be processed. It is also possible to use a template image having a high correlation value.

Next, an image data processing method applicable to a case where an offset of the brightness difference is placed on the entire sampled image data due to a difference in the manufacturing process or a difference in the image capturing device, and a difference in brightness of the screen appears. Will be described with reference to FIGS. 4 and 8. First, in step 181, a candidate image 57 considered as a candidate for a corresponding image corresponding to the template image 49 previously recorded from the exposure area by the CCD camera 19 or the like is displayed as a video signal (imaging signal).
Take in as. Next, at step 182, the average gray level of the template image 49 and the candidate image 5
The average values of 7 gray levels are obtained, and the two are compared, and the difference is calculated. Next, in step 182, the difference between the average gray levels is set as an offset error and stored in the memory. Then, in step 184, when the candidate image and the template image are actually compared to obtain the correlation value, the stored offset error is added. In this way, in this embodiment, the offset error caused by the brightness difference between the template image and the candidate image is corrected, and then the correlation values of both images are compared. And
If it is determined in step 185 that the correlation value is equal to or greater than the predetermined value, it can be determined that the candidate image 57 corresponds to the template image 49, and thus step 18
Proceeding to step 6, the candidate image is taken into the pattern matching device 20 and the position can be detected by pattern matching. In step 185, if the correlation value has not reached the predetermined value even though the luminance offset error has been corrected, it can be determined that the extracted candidate image is erroneous, and the process proceeds to step 187. Another candidate image is extracted again, and the flow from step 182 onward is repeated.

As described above, according to the embodiment shown in FIG. 8, there is a difference in brightness between the candidate image acquired by the CCD camera 19 and the template image stored in advance due to deterioration of the optical system or clouding. Even if it occurs, it is possible to correct the difference in brightness as an offset error.Therefore, as in the conventional method, even though the image is originally a correct candidate image, it is truncated due to the difference in brightness. It is possible to avoid erroneous detection that would otherwise occur, and improve the position detection accuracy.

When taking the average value of the gray level, the average value of the entire template image and the candidate image may be taken, or the average value of only a predetermined area may be adopted. . It is also possible to adopt a configuration in which a plurality of stages, for example, two stages, are provided as criteria for determining the correlation value, and the brightness offset error is corrected only when the correlation value is between the low stage and the high stage. is there.
In that case, first, the template image and the candidate image are compared as they are without correcting the luminance offset error, and if the criterion of the stage in which the correlation value is high is cleared, the candidate image is sent to the pattern matching device 20. Take in. Then, when the criterion of the high correlation value cannot be cleared, the judgment is made based on the criterion of the low stage, and when the standard of the low stage cannot be cleared, the candidate image is truncated as unsuitable. . However, when the standard with a high correlation value cannot be cleared but the standard with a low correlation value is cleared, it is possible to add a luminance offset error and make a judgment again based on the high correlation value. it can. With such a configuration, it is possible to prevent a candidate image having a luminance offset error, which is originally correct, from being truncated by only one determination, and improve the accuracy of position detection. be able to. (3) Connection Error Measuring Method Next, a case will be described in which the above-described highly accurate pattern detection method is applied to the connection error measuring method. In the case of this example, the pattern matching device 20 obtains the value of the correction parameter used at the time of alignment based on the actual shift amount of the pixel pattern formed in the connecting portion.
Incidentally, it is assumed that the large-sized liquid crystal display element is formed by connecting and exposing four reticle pattern images on one glass plate 5 as shown in FIG.

In the figure, reference numerals 22A to 22D indicate exposure areas where the pattern formed on each reticle is projected and exposed. In addition, 24A to 24D indicate these four exposure areas 22A.
22D, the boundary line between adjacent regions is shown.
The pattern matching device 20 uses the four boundary lines 24.
The areas on A to 24D are captured as an image to be processed at the time of detecting the shift amount of the connecting portion.

Here, of the four boundary lines 24A to 24D, the boundary line 24A and the pixel patterns 25A and 2 on both sides thereof are formed.
5B is moved so as to enter under the alignment microscope, that is, the objective lens 16. Next, as shown in FIG. 10A, an image 26 captured by the CCD camera 19 is displayed.
Is taken into the pattern matching device 20 as an image to be processed.

Subsequently, as shown in FIG. 10B, a pattern in which the template registered in advance matches the image 27 is searched from the image 26 to be processed. At this time, according to the present embodiment, the masking process described with reference to FIGS. 5 to 7 or the luminance offset error correction described with reference to FIG. 8 is performed to extract the corresponding image. Corresponding images can be extracted with a much higher accuracy than the method of. In the example shown in FIG. 10, two images matching the template image 27 are present on both sides of the boundary line 24A, but the patterns closest to the boundary line 24A are extracted as matching patterns (candidate images) 28A and 28B, respectively. .

When the matching patterns 28A and 28B are specified, the coordinate value (X, Y) corresponding to the representative point P0 of the template image 27 is obtained for each matching pattern 28A and 28B. Here, the coordinates of the matching pattern 28A in the exposure area 22A on the left side with respect to the boundary line 24A are defined as (X1, Y1), and the coordinates of the matching pattern 28B existing in the exposure area 22B on the right side are defined as (X2, Y2).

In this way, when the coordinates of the matching patterns 28A and 28B located on both sides of the boundary line 24A are obtained, the right side exposure area 22B with the matching pattern 28A formed in the left side exposure area 22A as a reference. The connection error (XST, YST) is calculated by the following equation: XST = X2-X1-XPT (1) YST = Y2-Y1-YPT (2) Here, XPT indicates the pitch of the pixel patterns 25A and 25B arranged in the lateral direction (X direction) at regular intervals, and when the expression (1) becomes 0, it means that the pixel pattern is formed as designed. Further, YPT represents the difference in the position of the pixel pattern in the Y direction, and in this case YPT = 0.
Therefore, if YST obtained from equation (2) is 0, there is no connection error, but if this YST is not 0 by design (that is, YPT ≠ 0), it will be compared with this design value. .

When the pattern matching device 20 completes the measurement process of the connection error for this area, the pattern matching device 20 sequentially moves the XY stage 7 in parallel, and similarly for the other areas on the boundary line 24A, the connection errors (XST, YST). ) Is measured. After that, the joint error (XS
The correction parameter is determined by using the least square method or the like so that the value of (T, YST) becomes the minimum on average. Similarly, with respect to the other three boundary lines 24B to 24D, the connection error is similarly measured over a plurality of points to determine the correction parameter.

Since this correction parameter value is a value obtained from the actual error amount occurring at the connecting portion, it is an accurate value that reflects reality as compared with the conventional one. According to the above configuration, the connecting portion 24 of the two exposure areas 22A and 22B
The pixel pattern 28A formed in the area by imaging A
And 28B are extracted with high accuracy as a pattern for detecting the deviation amount by using the masking processing method and the luminance offset error correction method as shown in FIGS. 5 to 8, and the connection error is obtained using these patterns, By determining the correction parameter value for alignment by this value,
It is possible to further improve the accuracy of joining as compared with the conventional method. (4) Method of measuring overlay error in joint portion Next, a method of improving detection accuracy by image data processing such as the above-described masking processing and luminance offset error correction method is applied to the first pattern formed in the lower layer and its A case in which the method is applied to a method of measuring an overlay error for reducing an overlay error between the second pattern of the upper layer formed on the upper layer and between adjacent regions will be described with reference to FIG. Also in this example, the exposure areas 22A and 22B
The pixel pattern in the vicinity of the boundary line 24A will be described as an example.

First, the XY stage 7 is driven by the driving device 11
Driven by the boundary line 24A and the exposure area 2 on both sides thereof.
2A and 22B pixel patterns are alignment microscopes,
That is, it is moved so as to enter under the objective lens 16.
Next, as shown in FIG. 11A, the image 33 captured by the CCD camera 19 is captured by the pattern matching device 20 as an image to be processed.

The same repetitive pattern is formed in the left and right exposure areas 22A and 22B so as to extend over the two areas, and the pixel patterns 32A and 32A of the second layer are formed on the underlying pixel patterns 31A and 31B. It is assumed that 32B is formed in an overlapping manner. Next, the pattern matching device 20 uses the pixel patterns 31A and 3
Template image 34 registered in advance for 1B
A pattern matching with is retrieved from the imaging screen 33.
At this time, according to the present embodiment, the masking process described with reference to FIGS. 5 to 7 and the luminance offset error correction described with reference to FIG. 8 are performed to extract the corresponding image. Corresponding images can be extracted with much higher accuracy than the method described above. In the example shown in FIG. 11, the patterns on the side close to the boundary line 24A are extracted as matching patterns (candidate images) 35A and 35B, respectively.

Then, with respect to the matching patterns 35A and 35B extracted as described above, the coordinate values (X, Y) which form a layer at the representative point P1 of the template image 34 are obtained for the matching patterns 35A and 35B, respectively. Here, the exposure area 2 on the left side of the boundary line 24A
Set the coordinates of 2A matching pattern 35A to (X1, Y1)
And the matching pattern 35 of the exposure area 22B on the right side.
The coordinates of B are (X1 ', Y1').

Similarly, the masking process described with reference to FIGS. 5 to 7 and the luminance offset error correction described with reference to FIG. 8 are used to preregister the pixel patterns 32A and 32B of the second layer. Template image 3
A pattern matching 6 is searched from each of the areas 22A and 22B, and the pattern on the side close to the boundary line 24A is extracted as matching patterns 37A and 37B.

When the matching patterns 37A and 37B are specified, the coordinate value (X, Y) corresponding to the representative point P2 of the template image 36 is obtained for each matching pattern 37A and 37B. Here, the left exposure area 22
The coordinates of the matching pattern 37A of A are (X2, Y2), and the matching pattern 37B of the exposure area 22B on the right side.
Let (X2 ', Y2') be the coordinates of.

In this way, the matching patterns 35A, 35B and 37A located on both sides of the boundary line 24A,
When the coordinates of 37B are obtained, the difference between the overlapping amounts of the matching patterns 35B and 37B in the right-side exposure area 24B with respect to the overlapping amount of the matching patterns 35A and 37A formed in the left-side exposure area 24A (ie, the overlapping amount). Move to the measurement of the alignment error).

This overlay error (XOV, YOV) is calculated by the following equation: XOV = (X1'-X2 ')-(X1-X2) (3) YOV = (Y1'-Y2')-(Y1-Y2) … (4) can be obtained. Here, the first term of the expressions (3) and (4) indicates the amount of overlap of the pattern formed in the right exposure area 22B, and the second term is the left exposure area 22B.
The overlapping amount of the pattern formed in A is shown.

When the pattern matching device 20 finishes the process of measuring the connection error for this portion, the XY stage 7 is sequentially moved in parallel, and the overlay error (XOV, YOV) is similarly applied to other regions on the boundary line 24A. ) Is measured. After that, the correction parameter is determined by using the least square method or the like so that the values of the overlay error (XOV, YOV) obtained over multiple points become the average minimum. Another 3
Similarly, with respect to the boundary lines 24B to 24D of the book, the overlay error is measured at multiple points to determine the correction parameter.

According to the above configuration, the overlapping amount of the first layer pattern and the second layer patterns 31A and 32A is set to the patterns 35A and 37A formed in the connecting portion 24A of the exposure regions 22A and 22B. 35B and 37B are directly obtained, and the value of the correction parameter used at the time of alignment is determined based on the difference in the overlap amount of each exposure area, so that the overlay error in the joint portion is further improved compared to the conventional case. Can be made smaller. Further, when extracting the patterns 35A, 37A and 35B, 37B,
Masking the part that causes false detection,
Alternatively, since the brightness offset error is corrected, the alignment pattern can be extracted with higher accuracy than ever before. (4) Other Embodiments In the above-described embodiment, the connection error or the overlay error of the pixel pattern of the right exposure area 22B is determined with the pixel pattern of the left exposure area 22A as the reference with the boundary line 24A interposed therebetween. Although the case has been described, the present invention is not limited to this, and conversely, the exposure area 2 on the left side with respect to the pixel pattern of the exposure area 22B on the right side across the boundary line 24A is used as a reference.
The connection error or the overlay error of the 2A pixel pattern may be obtained.

Further, in the above-mentioned embodiment, the boundary line 24
When the pixel patterns of the left and right two exposure areas 22A and 22B are imaged at a time with A sandwiched therebetween and the coordinates (x, y) of the matching pattern that matches the template image 27 or 34, 36 for each exposure area 22A, 22B are obtained. However, the present invention is not limited to this, and the pattern matching processing for one exposure area and the pattern matching processing for the other area may be alternately performed. That is, after the coordinates of the matching pattern are obtained for one of the exposure areas, the XY stage 7 may be moved to obtain the coordinates of the matching pattern for the other exposure area. In this case, the connection error or the overlay error is obtained by taking the movement amount of the XY stage 7 into consideration.

Further, in the above-mentioned embodiment, the case where the joint error and the overlay error are measured for the joint portions of the exposure regions to be joint-exposed has been described. However, the present invention is not limited to this, and the joint accuracy and the overlay error are not limited thereto. The connection error and the overlay error may be measured with emphasis on the direction in which high accuracy is required. For example, when manufacturing a liquid crystal display element or the like, high accuracy is required for joining in a direction orthogonal to the extending direction of the wiring pattern, that is, in the vertical direction of the screen, and therefore, a joining error or an overlaying error in this direction is required. It suffices to determine the correction parameter so that the component has an average minimum value.

Further, in the above-mentioned embodiments, the case of manufacturing a large-sized liquid crystal display element has been described, but the present invention is not limited to this, and can be applied to the case of manufacturing a memory pattern, another circuit pattern or the like. Further, in the above-described embodiment, the reflected light reflected from the observation area on the glass plate 5 is directly passed through the objective lens 16 and the mirror 17 as the alignment optical system for image processing to the relay lens 18 directly.
However, the present invention is not limited to this, and a light transmissive conjugate index plate may be provided at a position conjugate with the glass plate between the mirror 17 and the relay lens 18, Further, it may be electrically switched whether to optically pass through the conjugate index plate. Here, the conjugate index plate is a plate in which index marks are drawn, and the drawing surface and the exposure surface of the glass plate 5 are formed in a conjugate relationship. Then, the alignment may be performed based on the index image formed on the image pickup surface of the CCD camera 19 and the pattern image of the observation area.

By using this index mark, the positional relationship between the index mark and the alignment optical system can be kept normal even if the optical axis of the alignment optical system drifts.

[0065]

As described above, according to the present invention, when the image corresponding to the template image is detected from the patterns actually formed in the exposure area, it is possible to identify the portion that causes erroneous detection. Detection accuracy is improved by performing masking processing or software correction processing such as correction of luminance offset error. Then, the correction amount is determined so that the connection error and the overlay error between the exposure areas obtained by the position of the pattern detected with high accuracy in this way are minimized on average.
As a result, it is possible to realize a positioning method with higher accuracy in connection and superposition as compared with the conventional method.

[Brief description of drawings]

FIG. 1 is a schematic view showing a projection exposure apparatus to which a positioning method according to the present invention can be applied.

FIG. 2 is an explanatory diagram showing a first case to which the correction method for pattern detection according to the present invention can be applied.

FIG. 3 is an explanatory diagram showing a second case to which the correction method for pattern detection according to the present invention can be applied.

FIG. 4 is an explanatory diagram showing a third case to which the correction method for pattern detection according to the present invention can be applied.

FIG. 5 is a flowchart showing a procedure of a pattern detection correction method according to the present invention.

FIG. 6 is a flowchart showing a procedure of a pattern detection correction method according to the present invention.

FIG. 7 is a flowchart showing a procedure of a pattern detection correction method according to the present invention.

FIG. 8 is a flowchart showing a procedure of a pattern detection correction method according to the present invention.

FIG. 9 is a schematic plan view for explaining the joint exposure.

FIG. 10 is a schematic plan view for explaining a method for measuring a connection error.

FIG. 11 is a schematic plan view for explaining a method of measuring overlay error.

[Explanation of symbols]

1 Projection Exposure Device 5 Photosensitive Substrate (Glass Plate) 20 Pattern Matching Device 22A to 22D Exposure Area 45, 47, 49 Template Image 45a, 47a Masking Area 50, 55 Imaging Area (Processing Image) 53, 56, 57 Imaging Pattern ( Candidate image) Reference number = 94P01207 Page (27 /
26)

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location 525 W

Claims (7)

[Claims] 1. By detecting information corresponding to image information relating to a template serving as a reference from image information obtained by imaging a photosensitive substrate having an exposure area to which an original image pattern formed on an exposure original plate is transferred. In the method for detecting the position of the photosensitive substrate, when the exposure area is imaged and a corresponding image corresponding to a template image stored in advance is extracted from the captured image, masking is performed on a part of the template image. At the same time, masking is performed also on a portion corresponding to the masking portion of the template image among the candidate images listed as candidates for the corresponding image, and a non-masking area of the template image and a non-masking area of the candidate image are provided. If the correlation value between the non-masking areas is calculated and the correlation value is equal to or more than a predetermined value, Wherein the candidate image and extracts, as the corresponding image, the position detection method. 2. The template image is compared with all the candidate images to obtain a correlation value between all images, and when the correlation value between all images is lower than a predetermined value, the template A part of the image is masked, and a part of the candidate image corresponding to the masking part of the template image is also masked, so that a non-masking region of the template image and a non-masking region of the candidate image are formed. 2. The position detecting method according to claim 1, wherein a correlation value between the non-masking areas is obtained by comparison, and the candidate image is extracted as the corresponding image when the correlation value is equal to or more than a predetermined value. 3. The position detecting method according to claim 1, wherein the masking portion of the template image is preset as a predetermined portion of the template image. 4. The template image is divided into a plurality of regions, the candidate image is also divided into a plurality of regions corresponding to the template image, and the corresponding divided regions of the template image and the candidate image are compared with each other. 3. The position detecting method according to claim 2, wherein the correlation value between the divided areas is obtained, and the masking portion of the template image is set to a portion having the lowest correlation value among the divided areas. 5. By detecting information corresponding to image information regarding a reference template from image information obtained by imaging a photosensitive substrate having an exposure area to which an original image pattern formed on an exposure original plate is transferred. In the method for detecting the position of the photosensitive substrate, when the exposure area is imaged and a corresponding image corresponding to a template image stored in advance is extracted from the captured image, gray of a predetermined area included in the template image is extracted. The average value of the levels and the average value of the gray levels of the regions corresponding to the predetermined region of the template image of the candidate images listed as candidates for the corresponding image are compared, and the average value of the gray levels of the template image and the candidates are compared. The difference from the average gray level of the image is corrected as an offset error and The correlation value between the candidate image, and extracts the candidate image when the correlation value is greater than a predetermined value as the corresponding image, the position detecting method. 6. A large-area pattern is formed by connecting a small-area original image pattern formed on an exposure original plate to adjacent first and second exposure areas on a photosensitive substrate and exposing the same. In the method of aligning the original image pattern with the photosensitive substrate, imaging is performed at a plurality of locations including a connecting portion of the first and second exposure areas adjacent to each other in the exposure area, and at each of the plurality of imaging locations that are imaged, The corresponding image corresponding to the template image is extracted from each of the first and second exposure areas by the method according to claim 1, 2, 3, 4 or 5, and the corresponding image in the first exposure area is extracted. The difference between the position of the corresponding image and the position of the corresponding image in the second exposure region was obtained as a joint error generated at the joint portion of the first and second exposure regions, and the image was taken. A position detection method, characterized in that a correction amount at the time of alignment is determined so that the plurality of connection errors obtained at each of the plurality of imaging locations are averagely minimized. 7. A large-area pattern is formed by connecting a small-area original image pattern formed on an original exposure plate to adjacent first and second exposure regions on a photosensitive substrate and exposing the original pattern. In the method of aligning the original image pattern with the photosensitive substrate, imaging is performed at a plurality of locations including a connecting portion of the first and second exposure areas adjacent to each other in the exposure area, and at each of the plurality of imaging locations that are imaged, The first pattern of the first layer is formed from the first and second exposure regions, respectively.
First corresponding image corresponding to the second template image of the second layer pattern and the second corresponding image corresponding to the second template image of the second layer pattern.
The corresponding image of claim 1 is extracted by the method according to any one of claims 1, 2, 3, 4 or 5, and the position of the first corresponding image and the position of the second corresponding image in the first exposure area are extracted. Difference and the difference between the position of the first corresponding image and the position of the second corresponding image in the second exposure area, the superposition occurring in the joint portion of the first and second exposure areas. A position detecting method, wherein a correction amount at the time of alignment is determined so that the plurality of overlay errors obtained at each of the plurality of imaged locations that are imaged are averaged to be the minimum.