JP5324309B2 - Alignment apparatus, alignment method, and semiconductor device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an alignment technique capable of further improving registration. <P>SOLUTION: An alignment device calculates misregistration between both objects based on a superimposed image being an overlapping image of first and second patterns arranged on both the objects, and corrects the misregistration by driving both the objects relatively. The superimposed image has a first periodic concentration change (moire) generated by superimposing a first line group (pitch p1) of the first pattern and a second line group (pitch p2) of the second pattern, and also has a second periodic concentration change generated by superimposing a second line group of the first pattern and a first line group of the second pattern. A calculation means of the alignment device obtains two approximate curves approximating the two periodic concentration changes by image processing, and calculates misregistration between both the objects based on the misregistration between the two approximate curves. <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to an alignment technique for aligning two objects.

  As a technique for detecting a positional deviation between two objects, for example, a technique described in Patent Document 1 exists. Patent Document 1 describes a technique for detecting misalignment of two separate masking steps in semiconductor device manufacturing. Specifically, there is described a technique for detecting misalignment between two mask patterns (hereinafter also simply referred to as masks) printed at different layer levels using a moire generation line pattern.

  More specifically, two types of line patterns arranged at different pitches are arranged on two different objects (different layers). The first line pattern arranged at the first pitch is arranged on the first object, and the second line pattern arranged at the second pitch is arranged on the second object. Then, when these two line patterns (the first line pattern and the second line pattern) are overlapped, a periodic density change (moire fringes) in a predetermined direction occurs.

  This moire fringe is generated at a normal position (normal position) when the two objects are arranged without being displaced from each other. On the other hand, when two objects are arranged with a slight positional shift, moire fringes are generated at positions shifted from the normal position. It should be noted that the moire fringe shift appears by enlarging the minute shift between the two objects. In other words, it is possible to detect a relatively small shift (more minute shift) between objects based on a relatively large moiré fringe shift (actually a small shift).

  Then, by observing the moire fringes, specifically, by observing the deviation of the moire fringes from the normal position, deviations of the two types of line patterns are detected. In Patent Document 1, it is mainly described that moire fringes are optically observed and visually compared.

JP 2000-12459 A

  By the way, in order to accurately detect a shift between two types of line patterns in the above technique, it is required to accurately know the shift amount from the normal position of the moire fringes.

  However, it is not always easy to accurately know the amount of deviation of the moire fringes from the normal position.

  For example, in order to obtain the amount of deviation from the normal position of the moire fringe appearing in the X direction, the position of the moire fringe is obtained by obtaining the center position in the X direction of the high density portion (black portion in short) in the moire fringe. Can be considered. However, it is not always easy to accurately obtain the center position of the high density portion. The error related to the position of the moire fringe (the center position of the high density portion) has a great influence on the detection accuracy of the positional deviation between the two types of line patterns.

  Accordingly, an object of the present invention is to provide an alignment technique that can further improve accuracy.

  In order to solve the above-mentioned problems, the invention of claim 1 is an alignment apparatus that performs alignment on both objects of a first object and a second object, and is arranged on the first object. Image acquisition means for acquiring a first superimposed image that is an overlapping image of the first pattern and the second pattern arranged on the second object, and a predetermined value based on the first superimposed image. A calculating unit that calculates a positional deviation between the two objects in a plane; and a position correcting unit that relatively drives the two objects to correct the positional deviation, and includes the first pattern and the The second pattern is arranged in parallel at a first line group arranged in parallel at a first pitch in the first direction and at a second pitch different from the first pitch in the first direction. Each having a second line group The first superimposed image includes a first periodic density change generated by superimposing the first line group of the first pattern and the second line group of the second pattern. And having a second periodic density change generated by superimposing the second line group of the first pattern and the first line group of the second pattern in the first direction. The calculating means includes a first approximate curve that approximates the first periodic density change and a second approximate curve that approximates the second periodic density change. Are calculated based on pixel values at a plurality of positions in the first direction of the first superimposed image, respectively, and a positional shift between the first approximate curve and the second approximate curve In the first direction between the two objects. And calculates a first deviation that.

  According to a second aspect of the present invention, in the alignment apparatus according to the first aspect of the invention, the first pattern and the second pattern are parallel at a third pitch in a second direction different from the first direction. A third line group to be arranged; and a fourth line group arranged in parallel at a fourth pitch different from the third pitch in the second direction. Has a third periodic density change in the second direction generated by superimposing the third line group of the first pattern and the fourth line group of the second pattern. In addition, a fourth periodic density change generated by superimposing the fourth line group of the first pattern and the third line group of the second pattern is also applied to the second direction. And the calculating means includes the third means A plurality of third approximate curves approximating the periodic density change and a fourth approximate curve approximating the fourth periodic density change, respectively, in the second direction of the first superimposed image. Based on the pixel value at the position of the second object, and based on the positional deviation between the third approximate curve and the fourth approximate curve, the second direction between the objects in the second direction is determined. 2 is calculated.

  According to a third aspect of the present invention, in the alignment apparatus according to the second aspect of the invention, the image acquisition means includes a third pattern disposed on the first object and a second pattern disposed on the second object. A second superimposed image that is an overlapping image with the fourth pattern is also acquired, and the third pattern and the fourth pattern are the fifth line group, the sixth line group, and the seventh line group, And the fifth line group and the sixth line group are line groups arranged in parallel in the first direction, respectively, and the seventh line group. And the eighth line group are line groups arranged in parallel in the second direction, respectively, and the arrangement pitch of a plurality of lines in the fifth line group and the sixth line group It is different from the arrangement pitch of multiple lines. The arrangement pitch of the plurality of lines in the seventh line group and the arrangement pitch of the plurality of lines in the eighth line group are different values, and the second superimposed image has the first direction and In the second direction, each has two periodic density changes generated by superimposing the third pattern and the fourth pattern, and the calculating means is 2 in the first direction. A third positional shift in the first direction between the two objects is calculated based on a positional shift between two approximate curves representing two periodic density changes; Based on a positional shift between two approximate curves representing two periodic density changes, a fourth positional shift in the second direction between the two objects is calculated, and the first position is calculated. The displacement, the second displacement, and the Based on the positional deviation of 3 and said fourth position deviation, and calculates the positional deviation in the rotational direction between the two objects together.

  According to a fourth aspect of the present invention, in the alignment apparatus according to the second aspect of the invention, the first pattern of the first object and the second pattern of the second object are the same for exposure. A pattern generated using a mask, wherein the first periodic change in density and the second periodic change in a state where the first object and the second object are arranged to face each other A density change is generated in the first direction, and the third periodic density change and the fourth periodic density change are generated in the second direction. .

  According to a fifth aspect of the present invention, in the alignment apparatus according to any one of the second to fourth aspects, the third line group and the fourth line group are the first pattern and the second line group. Each of the first and second patterns is arranged opposite to each other across a first boundary line passing through the center in the first direction and extending in the second direction. The first non-moire area has a moiré area, and the first non-moire area is included in the first superimposed image according to a positional deviation between the first pattern and the second pattern in the first direction. In the vicinity of the first boundary line, the first pattern group and the second pattern are regions formed by overlapping the same pitch line groups, and the first line group and the second pattern The line group refers to the first pattern and the front In each of the second patterns, the second superimposed patterns are arranged opposite to each other across a second boundary line that passes through the center in the second direction and extends in the first direction. The second non-moiré area is further provided, and the second non-moiré area is in accordance with a positional deviation in the second direction between the first pattern and the second pattern. In the vicinity of the second boundary line in the image, the first pattern and the second pattern are regions formed by overlapping line groups of the same pitch, and the calculation means includes the first The length of the non-moire area in the first direction and the length of the second non-moire area in the second direction are obtained, and the positional deviation in the first direction between the objects and the both are determined. Position in the second direction between objects And calculating a rough value related to this, and the position correcting means performs a rough alignment operation for correcting the positional deviation by relatively driving both the objects based on the rough value, and the image acquiring means, After the rough alignment operation, the first superimposed image is acquired again, and the calculation means is configured to calculate the first approximate curve and the first approximation curve based on the first superimposed image acquired after the rough alignment operation. A second approximate curve is obtained, and a positional deviation between the two objects in a predetermined plane is further calculated, and the position correction means relatively drives both the objects after the rough alignment operation. The positional deviation between the two objects is corrected.

  A sixth aspect of the present invention is the alignment apparatus according to any one of the second to fifth aspects, wherein the third line group and the fourth line group are the first pattern and the second line group. And the first line group and the second line are arranged opposite to each other across a first boundary line extending through the center in the first direction and extending in the second direction. A group is arranged to face each other across a second boundary line extending through the center in the second direction and extending in the first direction in each of the first pattern and the second pattern. The first superimposed image is a non-moire area generated in accordance with a positional deviation between the first pattern and the second pattern, and the first boundary in the first superimposed image. Near the line or with the second boundary In the first pattern and the second pattern have a non-moire area formed by overlapping line groups having the same pitch, and the calculating means is based on the shape of the non-moire area. The positional deviation in the rotation direction within the predetermined plane is calculated as one of the positional deviations between the objects in the predetermined plane.

  The invention of claim 7 is the alignment apparatus according to any one of claims 1 to 4, wherein the first object and the second object are the first pattern and the second object. A rough alignment pattern pair is provided separately from the pattern, and the calculation means uses the rough alignment pattern pair to calculate a rough value regarding a positional deviation between the two objects in the predetermined plane, The position correcting unit performs a rough alignment operation for correcting the positional deviation by relatively driving both the objects based on the approximate value, and the image acquiring unit performs the first alignment after the rough alignment operation. The superimposed image is acquired, and the position correcting unit relatively drives both the objects after the rough alignment operation, and is calculated based on the first superimposed image. And correcting the deviation.

  According to an eighth aspect of the present invention, in the alignment apparatus according to any one of the first to seventh aspects, the first superimposed image is an image captured in an out-of-focus state.

  According to a ninth aspect of the present invention, in the alignment apparatus according to any one of the first to eighth aspects of the present invention, the image acquisition means transmits the both objects in a state where the both objects are arranged in close proximity to each other. The first superimposed image and the second superimposed image are obtained using transmitted light that is transmitted.

  According to a tenth aspect of the present invention, in the alignment apparatus according to any one of the second to ninth aspects, the first pattern and the second pattern are the first direction and the second pattern, respectively. It has a substantially rhombus shape whose direction is a diagonal direction.

  The invention according to claim 11 is an alignment method for performing alignment on both objects of the first object and the second object, and a) a first pattern arranged on the first object; Obtaining a first superimposed image that is an overlap image with a second pattern arranged on the second object; b) based on the first superimposed image, both the objects in a predetermined plane A step of calculating a positional deviation between the objects, and c) a step of correcting the positional deviation by relatively driving both the objects, wherein the first pattern and the second pattern include: A first group of lines arranged in parallel at a first pitch in one direction and a second group of lines arranged in parallel at a second pitch different from the first pitch in the first direction. Each of the first superimposed images Has a first periodic density change in the first direction generated by superposition of the first line group of the first pattern and the second line group of the second pattern. In addition, a second periodic density change generated by superimposing the second line group of the first pattern and the first line group of the second pattern is also applied to the first direction. And b-1) a first approximate curve that approximates the first periodic density change and a second approximate curve that approximates the second periodic density change. Respectively, obtaining based on pixel values at a plurality of positions in the first direction of the first superimposed image, and b-2) mutual relationship between the first approximate curve and the second approximate curve. The first direction between the two objects based on a displacement between them; Characterized by a step of calculating a first positional deviation definitive.

  The invention of claim 12 is a semiconductor device produced by using the alignment method according to the invention of claim 11.

  A thirteenth aspect of the invention is an alignment apparatus that performs alignment of both a first object and a second object, the first pattern disposed on the first object and the first object. Image acquisition means for acquiring a first superimposed image, which is an overlapping image with a second pattern arranged on two objects, and based on the first superimposed image, both the objects in a predetermined plane. A calculation means for calculating a positional deviation between the first pattern and a position correction means for correcting the positional deviation by relatively driving both the objects, and the first pattern and the second pattern are respectively The first superimposed image is generated by superimposing line groups having different pitches in both the first pattern and the second pattern. First And a second periodic density change in the first direction, and third and fourth lines generated by superimposing line groups of different pitches in both the first pattern and the second pattern. The calculation unit has a periodic density change in a second direction different from the first direction, and the calculation means includes a first approximate curve that approximates the first periodic density change and the second period. A second approximate curve that approximates a typical density change, a third approximate curve that approximates the third periodic density change, and a fourth approximate curve that approximates the fourth periodic density change. Respectively, based on the pixel values at a plurality of positions in the first superimposed image, and based on the positional deviation between the first approximate curve and the second approximate curve, A first misalignment between the objects in the first direction; Calculating a second positional shift in the second direction between the objects based on the positional shift between the third approximate curve and the fourth approximate curve. Features.

  The invention of claim 14 is an alignment method for performing alignment on both objects of a first object and a second object, and a) a first pattern arranged on the first object; Obtaining a first superimposed image that is an overlap image with a second pattern arranged on the second object; b) based on the first superimposed image, both the objects in a predetermined plane A step of calculating a positional shift between the objects, and c) a step of correcting the positional shift by relatively driving both the objects, wherein the first pattern and the second pattern are respectively , Having two line groups arranged at different predetermined pitches, and the first superimposed image is generated by superimposing line groups having different pitches in both the first pattern and the second pattern. Ru 3rd and 4th which have the 1st and 2nd periodic density change in the 1st direction, and are generated by superposition of the line group of a different pitch in both the 1st pattern and the 2nd pattern. And b-1) a first approximation curve that approximates the first periodic concentration change. A second approximate curve approximating the second periodic density change, a third approximate curve approximating the third periodic density change, and a fourth approximate curve approximating the fourth periodic density change. 4 approximate curves based on pixel values at a plurality of positions in the first superimposed image, and b-2) the first approximate curve and the second approximate curve, respectively. Based on the misalignment between each other, the first between the two objects Calculating a first misalignment in a direction; b-3) based on a misalignment between the third approximate curve and the fourth approximate curve, the second misalignment between the objects. And calculating a second positional shift in the direction of 2.

  According to the invention described in claims 1 to 14, the relative positional deviation between the first periodic density change and the second periodic density change is used to make an emergency between the two objects. It is possible to calculate a very small positional deviation. In particular, by obtaining two approximate curves based on pixel values at a plurality of positions, the position can be more accurately compared with the case where the positional deviation of the periodic density change is obtained at the position of the feature point of each density change. It is possible to calculate the deviation.

  In particular, according to the second aspect of the invention, it is possible to perform accurate positioning by correcting the positional deviation in the two translational directions.

  In particular, according to the third aspect of the present invention, it is possible to perform accurate positioning by correcting the positional deviation in the rotational direction.

  In particular, according to the invention described in claim 4, since the first pattern and the second pattern are generated by the same exposure mask (single mask), the two patterns are different from each other. The error is reduced compared to the case of generating.

  In particular, according to the invention described in claim 5, it is possible to detect a relatively large misalignment in the translation direction with the first and second patterns without using a separate pattern for rough alignment. is there.

  In particular, according to the sixth aspect of the present invention, it is possible to detect a positional deviation in the rotational direction using a set of two-dimensional moire patterns.

  In particular, according to the invention described in claim 8, it is possible to obtain a leveling effect without using an image processing filter.

  In particular, according to the invention described in claim 9, since both objects are arranged close to each other and aligned, it is possible to bring both objects closer together while maintaining good alignment accuracy.

It is a figure which shows schematic structure of an alignment apparatus (joining apparatus). It is sectional drawing of an alignment apparatus. It is a figure which shows the position of the alignment mark on a to-be-joined object of an upper side. It is a figure which shows the position of the alignment mark on a lower to-be-joined object. It is a figure which shows a rough alignment mark (upper side). It is a figure which shows a rough alignment mark (lower side). It is a figure which shows the two-dimensional pattern for a moire production | generation. FIG. 8 is an enlarged view of the vicinity of the central region of the two-dimensional pattern in FIG. 7. It is a figure which shows each pitch of eight line groups contained in a two-dimensional pattern. It is a figure which shows a mode that the top and bottom right and left of an upper side to-be-joined object are reversed. It is a figure which shows a mode that an upper to-be-joined object and a lower to-be-joined object oppose. It is a figure which shows typically a mode that one pattern inverts and opposes the other pattern. It is a figure which shows a mode that a pattern pair overlaps. It is a figure which shows the pattern for one moire generation. It is a figure which shows the pattern for the other moire generation. It is a figure which shows the moire which generate | occur | produces by superimposition of two patterns. It is a figure which shows the position shift of the moire resulting from the position shift of both patterns. It is a figure which shows the position shift of the moire resulting from the position shift of both patterns. It is a figure which shows the pattern for one moire production | generation for a dual line comparison method. It is a figure which shows the pattern for the other moire production | generation for dual line comparison methods. It is a figure which shows the moire which generate | occur | produces by superimposition of two patterns. It is a figure which shows the position shift of the moire resulting from the position shift of both patterns. It is a figure which shows the position shift of the moire resulting from the position shift of both patterns. It is a figure which shows the position shift of the moire resulting from the position shift of both patterns. It is a conceptual diagram which shows the periodic density change (moire) in a superimposition image. It is a figure which shows the approximate curve of two moire. It is a figure which shows the superimposition image in the state in which both patterns overlap without a position shift. It is a figure which shows each division area in the superimposed image of FIG. It is a figure which shows the superimposition image in the state which both patterns shifted | deviated to the X direction. It is a figure which shows the superimposed image in the state from which both patterns shifted | deviated to both the X direction and the Y direction. It is a conceptual diagram which shows the state which both the to-be-joined objects have shifted | deviated from each other. It is a figure which shows the geometrical relationship at the time of deriving | leading-out rotation deviation. It is a flowchart which shows joining operation | movement. It is a flowchart which shows fine alignment. It is a figure which shows the state which two patterns have shifted | deviated comparatively largely to the X direction. It is a figure which shows the superimposed image containing a non-moire area | region (high-intensity area | region). It is a figure which shows the state in which two types of patterns are superimposed on the upper and lower sides. It is a figure which shows the generation principle of a non-moire area | region (high-intensity area | region). It is a figure which shows the generation principle of a non-moiré area | region (low-intensity area | region). It is a figure which shows the superimposed image containing a non-moire area | region (low-intensity area | region). It is a figure which shows the state in which two types of patterns are superimposed on the upper and lower sides. It is a figure which shows another superimposed image. It is a figure which shows another superimposed image. It is a figure which shows the superimposed image with a rotation gap. It is a figure which shows another superimposition image with a rotation gap. It is a figure which shows another superimposed image with a rotation gap. It is a figure which shows the pattern for a moire generation concerning a modification. It is a figure which shows a mode that an upper pattern and a lower pattern oppose. It is a figure which shows the superimposition image in the state in which two patterns of a rhombus overlap without a position shift. It is a figure which shows the superimposed image in the state from which both patterns shifted | deviated to both the X direction and the Y direction. It is a figure which shows another superimposed image. It is a figure which shows another superimposed image. It is a figure which shows the superimposed image with a rotation gap. It is a figure which shows another superimposition image with a rotation gap. It is a figure which shows another superimposed image with a rotation gap. It is a figure which shows the pattern for a moire generation concerning a modification. It is a figure which shows the pattern for a moire generation concerning another modification. It is a figure which shows the pattern for a moire generation concerning another modification.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<1. First Embodiment>
<1-1. Device>
1 and 2 are diagrams showing an alignment apparatus 1 according to the present invention. FIG. 1 is a vertical cross-sectional view of the alignment apparatus 1, and FIG. 2 is a II cross-sectional view of the alignment apparatus 1. In each figure, directions and the like are shown using an XYZ orthogonal coordinate system for convenience.

  The alignment apparatus 1 performs an alignment operation (alignment operation) between the object to be bonded 91 and the object to be bonded 92 in a chamber (vacuum chamber) 2 under reduced pressure, and a surface activation process is performed on the facing surface. It is an apparatus for joining both objects to be joined 91 and 92. This alignment apparatus 1 is also expressed as a joining apparatus. According to this apparatus 1, it is possible to solid-phase-bond both the to-be-joined objects 91 and 92. FIG. Note that various materials (for example, silicon (semiconductor wafer) or the like) are used as the objects to be bonded 91 and 92. Moreover, both the to-be-joined objects 91 and 92 are expressed also as an object of alignment operation | movement, respectively.

  The alignment apparatus 1 includes a vacuum chamber 2 that is a processing space for both the objects to be bonded 91 and 92, and a load lock chamber 3 connected to the vacuum chamber 2. The vacuum chamber 2 is connected to a vacuum pump 5 via an exhaust pipe 6 and an exhaust valve 7. The vacuum chamber 2 is put into a vacuum state by reducing (reducing pressure) the pressure in the vacuum chamber 2 in accordance with the suction operation of the vacuum pump 5. Further, the exhaust valve 7 can adjust the degree of vacuum in the vacuum chamber 2 by the opening / closing operation and the exhaust flow rate adjusting operation.

  Both objects 91 and 92 are held in the load lock chamber 3 by the clamping chuck 4 c at the tip of the introduction rod 4 and then moved into the vacuum chamber 2. Specifically, the upper article 92 is held at the tip of the introduction rod 4, moved in the X direction to a position immediately below the head 22, and then held by the head 22. Similarly, the lower workpiece 91 is moved toward the stage 12 in the X direction while being held at the tip of the introduction rod 4 and is held by the stage 12.

  Both the head 22 and the stage 12 are installed in the vacuum chamber 2. The head 22 is moved (translationally moved) in the X direction and the Y direction by the alignment table 23, and is rotated in the θ direction (rotation direction about the Z axis) by the rotation drive mechanism 25. The head 22 is driven by the alignment table 23 and the rotation drive mechanism 25 based on a position detection result by a position recognition unit 28 and the like described later, and alignment operations in the X direction, the Y direction, and the θ direction are executed.

  The head 22 is moved (lifted / lowered) in the Z direction by the Z-axis lifting / lowering drive mechanism 26. The Z-axis raising / lowering drive mechanism 26 can control the applied pressure at the time of joining based on a signal detected by a pressure detection sensor (not shown). Note that the X direction, the Y direction, and the Z direction are directions orthogonal to each other.

  The alignment apparatus 1 also includes a position recognition unit 28 that recognizes the positions of the objects to be joined 91 and 92.

  As shown in FIG. 2, the position recognizing unit 28 includes light sources 28a and 28b, and imaging units (cameras) 28c and 28d that acquire a light image related to the object to be bonded as image data. Further, both the objects to be bonded 91 and 92 are provided with position identification pattern marks (hereinafter also simply referred to as patterns or marks). Here, four position identification patterns MT2, MT4, AT2, and AT4 (see FIG. 3) are provided on one workpiece 92, and four position identification patterns MT1, MT3 are provided on the other workpiece 91. , AT1, AT3 (see FIG. 4). Here, the four patterns AT1 to AT4 are patterns for rough alignment (rough positioning), and the other four patterns MT1 to MT4 are patterns for fine alignment (high precision positioning). The alignment operation using these patterns will be described in detail later.

  The alignment operation (alignment operation) of the objects to be bonded 91 and 92 is executed by recognizing the positions of the patterns attached to the objects to be bonded 91 and 92 by the position recognition unit (camera or the like).

  As shown in FIG. 2, the position recognition unit 28 uses the image data related to the transmitted light from the light sources 28 a and 28 b in a state where the objects to be bonded 91 and 92 face each other. Recognize the position of Specifically, each light emitted from the light sources 28a and 28b arranged on the outer side of the vacuum chamber 2 is transmitted through a window portion (not shown) of the vacuum chamber 2, and thereafter is reflected by the mirrors 28e and 28f. Reflected, the direction of travel is changed and travels downward. Each of the lights further passes through both the objects to be bonded 91 and 92, and also passes through the window portion 2b (FIG. 1) and reaches the imaging portions 28c and 28d, respectively. The position recognizing unit 28 acquires, as image data, the optical images (images related to transmitted light) related to the objects 91 and 92 acquired in this way. The superimposed images of the patterns MT1 and MT2 and the superimposed images of the patterns PAT1 and AT2 are respectively acquired by the imaging unit 28c. The superimposed images of the patterns MT3 and MT4 and the superimposed images of the patterns AT3 and AT4 are respectively acquired by the imaging unit 28d. Further, the position recognition unit 28 recognizes the position of the mark based on the image data. In addition, as the light sources 28a and 28b, light (for example, infrared light) that passes through both the workpieces 91 and 92 and the stage 12 is used.

  Thereafter, the head 22 is driven in two translational directions (X direction and Y direction) and a rotational direction (θ direction) by the alignment table 23 and the rotation drive mechanism 25 based on the position information recognized by the position recognition unit 28. Is done. Thereby, both the to-be-joined objects 91 and 92 are moved relatively.

  In this way, the alignment operation is executed.

  Further, such a position recognition operation and an alignment driving operation are repeatedly executed. According to this, the drive error at the time of driving the head 22 is gradually reduced, and a more accurate alignment operation is executed.

  The device 1 further includes a controller 100. The controller 100 is a control unit that controls the alignment operation and the bonding operation.

  The controller 100 (FIG. 1) includes an image acquisition control unit 101, a position deviation calculation unit (also simply referred to as a calculation unit) 102, and a position correction control unit 103.

  The image acquisition control unit 101 controls the image acquisition unit 31 including the imaging units 28c and 28d. The image acquisition unit 31 acquires an overlapping image (superimposed image) GC of both patterns respectively arranged on the surfaces of two opposing objects.

  The calculation unit 102 calculates a positional deviation between the objects to be joined 91 and 92 within a predetermined plane based on the superimposed image GC acquired under the control of the image acquisition control unit 101.

  The position correction control unit 103 controls the position correction unit 33 configured to include the drive mechanism that drives the alignment table 23 and the rotation drive mechanism 25, and relatively drives both the workpieces 91 and 92 to position the position correction unit 33. Correct the deviation.

<1-2. Outline of alignment>
In the first embodiment, the alignment apparatus 1 performs rough alignment with the patterns shown in FIGS. 5 and 6 and performs fine alignment with the pattern shown in FIG. 7 (moire generation two-dimensional pattern).

  More specifically, first, a rough alignment operation is performed using the pattern CT shown in FIG. 5 and the pattern DT shown in FIG.

  The pattern CT is a cross-shaped pattern mark, and the pattern DT is a pattern mark in which four squares are arranged in a grid. The pattern CT is disposed on the bonding surface of the upper workpiece 92, and the pattern DT is disposed on the bonding surface of the lower workpiece 91. For example, the size (length) H2 of the pattern CT and the size (length) H1 of the pattern DT are each about several hundred μm (micrometers). In addition, the gap GP (see FIG. 6) between the protruding portion of the cross in the pattern CT in the vertical and horizontal directions and the four squares of the pattern DT is, for example, several tens of μm. The patterns CT and DT are also expressed as a rough alignment pattern pair.

  When the workpieces 91 and 92 are correctly positioned, the four squares of the pattern DT are arranged so as not to overlap with the cross portion of the pattern CT, and the four squares are all the same length from the cross portion. Placed apart.

  In the rough alignment, the alignment is performed so that the cross portion of the pattern CT is arranged at the center of the four squares of the pattern DT. As a result, it is possible to keep the positional errors of the workpieces 91 and 92 within a predetermined range (for example, about several μm).

  Furthermore, in this embodiment, by using a pattern (two-dimensional pattern for generating moire) MT as shown in FIG. 7, the misalignment of both the objects 91 and 92 is recognized again, and even higher precision alignment (fine Alignment) is performed.

  Position recognition using this pattern MT is performed using moire fringes (hereinafter also simply referred to as moire) generated by superimposing two types of line groups arranged at different pitches. Hereinafter, position recognition using this pattern MT will be described in detail from the basic principle.

<1-3. Basic Principle of Misalignment Detection Using Moire>
14 to 16 are diagrams for explaining the principle of occurrence of moire. First, moire generated when two types of line groups LG1 and LG2 are superimposed will be described.

  FIG. 14 is a diagram showing the pattern PT1, and FIG. 15 is a diagram showing the pattern PT2. The patterns PT1 and PT2 are also expressed as moire generation patterns.

  As shown in FIG. 14, the pattern PT1 has a line group LG1 arranged in parallel at a pitch p1. On the other hand, as shown in FIG. 15, the pattern PT2 has a line group LG2 arranged in parallel at a pitch p2. Here, the pitch p2 is larger by a minute amount Δp than the pitch p1 (p2 = p1 + Δp). For example, the pitch p1 is 19 μm (micrometer) and the pitch p2 is 20 μm (Δp = 1 μm). Further, each of the plurality of lines included in each line group LG1, LG2 has a predetermined width (for example, half the value of the pitch p2).

  FIG. 16 is a diagram showing a superimposed image of the pattern PT1 and the pattern PT2. As shown in FIG. 16, in the overlapping portion of the pattern PT1 and the pattern PT2, the shading changes in the horizontal direction of the drawing, and so-called moire occurs.

  The moire period PD is expressed by the following equation (1).

  The expression (1) is obtained by calculating the arrangement length (p1 × (n + 1)) of (n + 1) lines arranged at the pitch p1 and the arrangement length (p2 ×) of n lines arranged at the pitch p2. n) are derived from the fact that both are equal to the arrangement period PD.

  In addition, when the pattern PT2 is shifted by Δp leftward from the pattern PT1 from the state of FIG. 16, the moire is shifted rightward by the distance p2.

  Similarly, as shown in FIG. 17, when the pattern PT2 is shifted to the left by the distance (2 × Δp) from the state of FIG. 16, the moire is shifted to the right by the distance (2 × p2). Specifically, in FIG. 16, the leftmost straight line of the pattern PT1 and the leftmost straight line of the pattern PT2 exist at the same position in the horizontal direction, whereas in FIG. 17, the third straight line from the left of the pattern PT1. And the third straight line from the left of the pattern PT2 are present at the same position in the horizontal direction. That is, when the pattern PT2 is shifted 2 × Δp to the left with respect to the pattern PT1, the moire (for example, the central portion of the high concentration region) is shifted to the right by the distance (2 × p2) as shown in FIG. .

  Further, as shown in FIG. 18, when the pattern PT2 is shifted to the right by a distance (2 × Δp) from the state of FIG. 16, the moire is shifted to the left by a distance (2 × p2).

  Thus, the moire shift amount ΔD1 is a value obtained by multiplying the relative shift amount Δd of both patterns by the value (p2 / Δp) (see Expression (2)).

  Here, a change rate MG (also referred to as an enlargement rate) of the moire shift amount ΔD1 with respect to the pattern relative shift amount Δp is expressed by the following equation (3). For example, when p1 = 19 μm and p2 = 20 μm, MG = 20.

  Then, according to Equation (4) derived from Equation (2), it is possible to obtain the deviation amount Δd between the pattern PT2 and the pattern PT1 based on the moire deviation amount ΔD1 from the state of FIG.

  That is, by using the moire deviation amount ΔD1, it is possible to obtain the relative deviation amount Δd between the patterns PT1 and PT2 as a value of (1 / MG) of the value ΔD1. In this way, it is possible to detect a relatively small shift (more minute shift) between the patterns PT1 and PT2 based on a relatively large moiré stripe shift (actually a small shift). .

<1-4. Misalignment detection principle using moire (dual detection)>
In the above method (also referred to as the first method), when the moire position (see FIG. 16) in a state where there is no positional deviation between the patterns PT1 and PT2 is known, the above-described principle is used. It is possible to derive the deviation amount Δd due to. In other words, in the first method, for example, it is necessary to acquire the normal position information of the moire based on information other than the superimposed image GA of the two moire generation patterns (information outside the superimposed image). However, in general, it is not easy to obtain the position (absolute position) of the moire in the state of FIG.

  Therefore, in the following, a technique (hereinafter also referred to as the second technique) that can be easily applied even when the absolute position of moire (see FIG. 16) in a state where there is no positional deviation is not known is shown in FIGS. 23 will be described.

  In this second method, as shown in FIGS. 19 and 20, two patterns PL and PU in which two types of patterns PT1 and PT2 are provided in the vertical direction (vertical direction) in the figure are used. In the pattern PL and the pattern PU, line patterns PT1 and PT2 are provided on opposite sides in the vertical direction. Specifically, in the pattern PL, the pattern PT1 is provided on the upper side in the vertical direction, and the pattern PT2 is provided on the lower side in the vertical direction. On the other hand, in the pattern PU, the pattern PT2 is provided on the upper side in the vertical direction, and the pattern PT1 is provided on the lower side in the vertical direction. Here, the pattern PT1 is a pattern in which a plurality of parallel lines (line group LG1) is arranged at a pitch p1, and the pattern PT2 is a pattern in which a plurality of parallel lines (line group LG2) is arranged at a pitch p2. Pattern. The pitch p2 is larger by a minute amount Δp than the pitch p1.

  FIG. 21 is a diagram illustrating a state (superimposed image GB) in which the pattern PL and the pattern PU are accurately positioned and superimposed at the reference position (here, the leftmost line position). That is, FIG. 21 shows a state in which both patterns PL and PU are superimposed with no positional deviation.

  Specifically, in the upper half region PH1, a periodic density change generated by the superposition of the line group LG1 of the pattern PL and the line group LG2 of the pattern PU occurs in the left-right direction of the drawing. In the lower half region PH2, a periodic density change generated by the superposition of the line group LG2 of the pattern PL and the line group LG1 of the pattern PU occurs in the left-right direction of the drawing. In FIG. 21, the same moire occurs in the upper half region PH1 and the lower half region PH2.

  On the other hand, FIG. 22 is a diagram showing a state (superimposed image GB) in which the pattern PU is shifted to the right by the minute amount Δp with respect to the pattern PL and the patterns PU and PL are superimposed. In FIG. 22, the minute movement amount Δd = Δp.

  At this time, in the upper half area PH1, the pattern PT2 in the pattern PU moves a slight amount Δp to the right with respect to the pattern PT1 in the pattern PL. Therefore, as shown in FIG. 22, the moire in the upper half area PH1 moves a small amount ΔD1 to the left as compared with FIG. This movement amount ΔD1 is equal to the value p2 from the equation (2).

  Further, in the lower half region PH2, the pattern PT1 in the pattern PU moves by a slight amount Δp to the right relative to the pattern PT2 in the pattern PL. In other words, the pattern PT2 in the pattern PL moves a slight amount Δp to the left relative to the pattern PT1 in the pattern PU. Therefore, as shown in FIG. 22, the moire in the lower half region PH2 moves to the right by a small amount ΔD1 compared to FIG. This movement amount ΔD1 is equal to the value p2 from the equation (2).

  In this way, when the pattern PU and the pattern PL are displaced relative to each other, the moire in the upper half area PH1 and the moire in the lower half area PH2 move in opposite directions. Further, when the pattern PU and the pattern PL have a relative positional shift of the shift amount Δd, the moire of the upper half region PH1 and the moire of the lower half region PH2 are shifted from each other by the shift amount ΔD1 of the equation (2). Therefore, as shown in FIG. 22, the moire in the upper half area PH1 and the moire in the lower half area PH2 are shifted by a deviation amount ΔD2 that is twice as large as ΔD1 (see Expression (5)).

  Here, an enlargement ratio MG2 as defined in Expression (6) is defined. For example, when p1 = 19 μm and p2 = 20 μm, MG2 = 40.

  When such a value MG2 is used, as shown in Expression (7), the deviation amount ΔD2 between the moire in the upper half area PH1 and the moire in the lower half area PH2 is equal to the deviation amount Δd between the pattern PU and the pattern PL ( MG2) expressed as a doubled value. For example, when MG2 = 40, ΔD2 is 40 times Δd.

  On the contrary, according to the equation (8) derived from the equation (7), the deviation amount Δd between the pattern PT2 and the pattern PT1 is the relative deviation amount ΔD2 between the moire in the upper half region PH1 and the moire in the lower half region PH2. Is calculated based on

  FIG. 23 is a diagram illustrating a state (superimposed image GB) in which the pattern PU is shifted to the right by a minute amount (3 × Δp) and the patterns PU and PL are superimposed on each other. In FIG. 23, the minute movement amount Δd = (3 × Δp).

  At this time, as shown in Expression (7), the deviation amount ΔD2 between the moire in the upper half region PH1 and the moire in the lower half region PH2 is a value (MG2) times the deviation amount Δd between the pattern PU and the pattern PL. It is.

  Therefore, conversely, by observing and obtaining the shift amount ΔD2, both patterns PU. The relative displacement amount Δd of PL can be calculated. For example, when MG2 = 40, Δd = ΔD2 / MG2 = ΔD2 / 40. More specifically, when ΔD2 = 1 μm (micrometer), Δd = 1/40 μm = 25 nm (nanometer).

  FIG. 24 is a diagram showing a state (superimposed image GB) in which the pattern PU is shifted to the left by a minute amount Δp and the patterns PU and PL are superimposed on the pattern PL. In FIG. 24, the minute movement amount Δd = Δp.

  Similar to the above, by observing and acquiring the deviation amount ΔD2, both patterns PU. The relative displacement amount Δd of PL can be calculated.

  In FIG. 24, as can be seen from comparison with FIG. 22, the direction of the deviation between the moire in the upper half region PH1 and the moire in the lower half region PH2 is reversed. Therefore, it is possible to know the direction of deviation of the pattern PU with respect to the pattern PL based on the direction of deviation between the moire of the upper half area PH1 and the moire of the lower half area PH2.

  As described above, according to the second method, the superimposed images GB of the patterns PT1 and PT2 are formed in the upper half region PH1 and the lower half region PH2, which are shifted in opposite directions, and the superimposed image GB is formed. Moire occurs at the top and bottom of each. Then, by obtaining the direction and magnitude of the deviation between the moire in the upper half area PH1 and the moire in the lower half area PH2, it is possible to know the direction and magnitude of the deviation of the pattern PU with respect to the pattern PL.

  In particular, according to the second method, the moire position (see FIG. 16) in a state where there is no positional deviation between the patterns PL and PU does not necessarily need to be known in advance. That is, the first method needs to refer to information on the normal position of the moire, whereas the second method does not need information on the normal position. That is, according to the second method, information on the normal position of the moire based on information (also referred to as external information) other than the superimposed image GB of the two moire generation patterns is not required. In other words, it is possible to know the direction and amount of misalignment between the two moire generation patterns based only on the superimposed image GB of the moire generation pattern (in short, only internal information). Specifically, by comparing the upper half region PH1 and the lower half region PH2 in the superimposed image GB with each other (in other words, referring to the internal information mutually), desired information (both patterns PL, PU (The direction and amount of misregistration). Therefore, it is possible to easily know the positional deviation between the two objects.

  When grasping the position of the moire, the center position of the high density portion of the moire may be determined by visual observation or the like. However, the periodic density changes of the upper and lower moires are respectively determined as follows. It is preferable to obtain.

  FIG. 25 is a conceptual diagram showing a periodic density change (solid line) of one of the upper and lower moire in the superimposed image GB. In FIG. 25, the horizontal axis indicates the position X of the superimposed image GB in a predetermined direction (for example, the X direction), and the vertical axis indicates the density (pixel value) V at each position X of the superimposed image GB. Yes.

  Specifically, first, pixel values V (for example, 256 gray scale values) at a plurality of moire positions X are acquired. Then, based on a plurality of pixel values relating to different positions X, the periodic density change (pixel value change) of moire in the X direction is approximated by a curve having a predetermined shape. For example, the periodic density change may be approximated by a sine curve curve (also expressed as a cosine curve curve) by regression processing. In addition, the broken line in FIG. 25 has shown the approximate curve by a sine curve.

  Here, the pixel value of the pixel at each position X is an average value (simple value) of a plurality of pixels including the pixel at each position X and several peripheral pixels in the X direction of the pixel in the previous stage of the approximation process. An average value or a weighted average value) may be acquired. More specifically, smoothing filter processing (blurring filter processing) may be performed on the original pixel column in the X direction related to moire to change the pixel value of the pixel column. Each pixel in the pixel row after the change has an average pixel value of neighboring pixels of the pixel at each position X. Note that the present invention is not limited to this, and the maximum value among the pixel values of the plurality of pixels including the pixel value of the pixel at each position X and the pixel values of the surrounding pixels, or the median value of the plurality of pixels, You may make it determine as a pixel value in each position.

  By the approximation process as described above, for example, one approximate curve relating to the upper half region PH1 is obtained.

  Similarly, an approximate curve for the other moire (for example, lower half region PH2) is also obtained.

  As a result, approximate curves (two approximate curves in total) for each of the upper half region PH1 and the lower half region PH2 are obtained.

  FIG. 26 shows an approximate curve AL1 that approximates a periodic density change (ie, moire) in the upper half region PH1 and an approximate curve AL2 that approximates a periodic density change (ie, moire) in the lower half region PH2. FIG. For example, the approximate curve AL1 represents a density change on the straight line LN1 (see FIG. 22) of the upper half region PH1, and the approximate curve AL2 is on the straight line LN2 (see FIG. 22) of the lower half region PH2. This expresses the change in the concentration.

  Then, based on the positional deviation (ΔD2) between the two approximate curves AL1, AL2, the positional deviation (Δd) in a predetermined direction between the two objects is calculated based on the equation (8). Note that the positional shift between the two approximate curves AL1 and AL2 is also expressed as the phase shift between the two sine curves AL1 and AL2.

  According to this, based on the positional shift between the approximate curves AL1 and AL2 regarding two moires (periodic density changes), the positional shift between the patterns PU and PL (and thus both the objects 91 and 92). ) Can be grasped.

  In particular, by determining the position of the moire by obtaining approximate curves based on pixel values at a plurality of positions, the position of a certain feature point of each density change (for example, the black area of the moire) can be detected by human eyes or the like. Compared with the case where the position of the moire is specified at the center position, it is possible to calculate the positional deviation more accurately. Further, if the position of moire is specified by the position of one feature point, the position accuracy of the one feature point is restricted depending on the pixel resolution of image processing. On the other hand, by specifying the moire position based on the measurement information at a plurality of positions, the measurement error is reduced as compared with the case where the moire position is specified at the position of a certain point. Thus, by obtaining two approximate curves based on pixel values at a plurality of positions, compared to the case of obtaining the positional deviation of the periodic density change at the position of one feature point of each density change, Further, it is possible to calculate the positional deviation more accurately.

  In addition, the superimposed image GB does not necessarily have to be a clear image acquired in a focused state. For example, the approximate curves AL1 and AL2 may be obtained using a superimposed image GB captured in a slightly out-of-focus state (out-of-focus state) rather than in a completely in-focus state. According to this, the same leveling effect can be obtained without using an image processing filter (leveling filter or the like). In this case, it is particularly preferable to acquire pixel values at a plurality of moire positions X as gray scale values.

  In the second method, the density change on the straight line in the predetermined direction in the upper half area PH1 of the superimposed image GB is compared with the density change on the straight line in the predetermined direction in the lower half area PH2 of the superimposed image GB. Is done. In other words, periodic density changes (approximate curves) on two straight lines (upper and lower) are compared with each other. Therefore, this second method is also referred to as “dual line comparison method”. In particular, a method of comparing two moires using two approximate curves is also referred to as a “dual approximate curve comparison method”.

<1-5. Misalignment detection principle using moire (two-dimensional)>
In this embodiment, basically, the second method described above is employed. However, in order to obtain a positional shift in two directions (for example, the X direction and the Y direction) parallel to the facing surfaces of the workpieces 91 and 92, the second method is related to one direction from two directions. A technique (also referred to as a third technique) extended to a thing is adopted.

  FIG. 7 is a diagram showing a moire generation pattern MT used in the present embodiment, and FIG. 8 is an enlarged view of the vicinity of the central region RC of the pattern MT. FIG. 9 is a diagram showing eight segmented regions RG1 to RG8 of the pattern MT.

  As shown in FIG. 7 and the like, the pattern MT has a substantially square shape. Further, as shown in FIG. 9, the pattern MT is divided into eight regions RG1 to RG8 that radially extend from the center of the pattern MT (specifically, an eight-divided region in which the central angle is divided into 45 degrees). The Each region RG1 to RG8 has a substantially right triangle shape. Of the four regions divided by the two center lines CY and CX extending in the X direction and the Y direction, a plurality of lines (line group LG1) are arranged at a pitch p1 in the upper left region and the lower right region, respectively. A plurality of lines (line group LG2) are arranged at a pitch p2 in the lower left area and the upper right area, respectively. This pattern MT has a point-symmetric configuration with respect to its center point CP. For example, the sizes (lengths) Hx and Hy in the X and Y directions of the pattern MT are about 2 mm, the pitch p1 is 19 μm (micrometer), and the pitch p2 is 20 μm (Δp = 1 μm). ).

  More specifically, out of the two equally divided regions RG1 and RG2 included in the upper left region of the pattern MT, the region RG2 that is relatively close to the central axis CY extending in the left-right direction (X direction) in the drawing has upper and lower portions in the drawing. A plurality of lines extending in the direction (Y direction) are arranged at a pitch p1 in the left-right direction (X direction) in the figure. The other region (region relatively close to the central axis CX extending in the vertical direction) RG1 of the two equally divided regions included in the upper left region of the pattern MT has a plurality of portions extending in the left-right direction (X direction) in the drawing. The lines are arranged at a pitch p3 (here, the same as the pitch p1) in the vertical direction (Y direction) in the figure.

  Similarly, among the two equally divided regions RG5 and RG6 included in the lower right region of the pattern MT, a plurality of lines extending in the Y direction have a pitch p1 in the X direction in the region RG6 relatively close to the central axis CY. Arranged. In the other region (region relatively close to the central axis CX) RG5 of the two equally divided regions included in the lower right region of the pattern MT, a plurality of lines extending in the X direction have a pitch p3 (here) Is the same as the pitch p1).

  In addition, among the two equally divided regions RG3 and RG4 included in the lower left region of the pattern MT, a plurality of lines extending in the Y direction are arranged at a pitch p2 in the X direction in the region RG3 that is relatively close to the central axis CY. The In the other region (region relatively close to the central axis CX) RG4 of the two equally divided regions included in the lower left region of the pattern MT, a plurality of lines extending in the X direction have a pitch p4 in the Y direction (here, (Same as the pitch p2).

  Similarly, among the two equally divided regions RG7 and RG8 included in the upper right region of the pattern MT, a plurality of lines extending in the Y direction are arranged at a pitch p2 in the X direction in the region RG7 relatively close to the central axis CY. Is done. In the other region (region relatively close to the central axis CX) RG8 of the two equally divided regions included in the upper right region of the pattern MT, a plurality of lines extending in the X direction have a pitch p4 in the Y direction (here, (Same as the pitch p2).

  In addition, the line group LG1 in the region RG2 and the line group LG2 in the region RG3 are arranged to face each other across the boundary line CY. The boundary line CY is a boundary line (center line) that passes through the center in the Y direction and extends in the X direction. Similarly, the line group LG1 in the region RG6 and the line group LG2 in the region RG7 are arranged to face each other across the boundary line CY.

  In addition, the line group LG3 (here, the line group LG1) in the region RG1 and the line group LG4 (here, the line group LG2) in the region RG8 are arranged to face each other across the boundary line CX. The boundary line CX is a boundary line (center line) that passes through the center in the X direction and extends in the Y direction. Similarly, the line group LG4 (here, the line group LG2) in the region RG4 and the line group LG3 (here, the line group LG1) in the region RG5 are arranged to face each other across the boundary line CX. Note that the boundary lines CX and CY do not have to be actually drawn lines, but may be virtual lines.

  FIG. 3 is a diagram showing the positions of the patterns MT2, MT4, AT2, AT4 provided on the upper workpiece 92. FIG. 4 shows the patterns MT1, MT3 provided on the lower workpiece 91. As shown in FIG. It is a figure which shows the position of AT1, AT3.

  As shown in FIG. 4, the same pattern MT as FIG. 7 is provided as patterns MT1 and MT3 on the bonding surface of the lower workpiece 91. On the other hand, as shown in FIG. 3, the same pattern MT as FIG. 7 is provided as patterns MT2 and MT4 on the bonding surface of the upper object 92. These four patterns MT1 to MT4 are all the same, and are created on the surfaces (hymen) of both objects 91 and 92 using the same (exposure) mask pattern (also FIG. 10). reference).

  Then, as shown in FIGS. 10 and 11, the left and right sides of the upper object 92 are inverted, and the bonding surface of the upper object 92 and the bonding surface of the lower object 91 face each other. . As shown in FIG. 11, in the opposed state of both the objects to be joined 91 and 92, the pattern AT1 and the pattern AT2 face each other, the pattern AT3 and the pattern AT4 face each other, and the pattern MT1 and the pattern MT2 face each other. The pattern MT3 and the pattern MT4 are opposed to each other. In FIGS. 10 and 11, in order to clearly show the arrangement of the patterns MT1 to MT4 and AT1 to AT4, these patterns are expressed in a deformed manner.

  FIG. 12 is a diagram schematically illustrating a state in which the pattern MT2 on the workpiece 92 is reversed and faces the pattern MT1 on the workpiece 91. In FIG. 12, the state before inversion is shown on the left side of the white arrow, and the state after inversion is shown on the right side of the white arrow. In FIG. 12, the arrangement pitch (specifically, p1 and p2) of lines in each region is also shown.

  As shown on the right side of the white arrow in FIG. 12, when the state in which the pattern MT1 and the pattern MT2 are overlapped after the pattern MT2 is reversed is viewed from above in the vertical direction, different pitch portions of the patterns MT1 and MT2 overlap each other. Specifically, the line group having the pitch p1 of the pattern MT1 and the line group having the pitch p2 of the pattern MT2 overlap, and the line group having the pitch p2 of the pattern MT1 and the line group having the pitch p1 of the pattern MT2 overlap.

  FIG. 13 is a schematic diagram showing in detail how both the patterns MT1 and MT2 overlap in a state in which the objects to be bonded 91 and 92 are opposed to each other. Specifically, as shown in FIG. 13, the region RG1 of the pattern MT1 and the region RG8 of the pattern MT2 overlap, and the region RG2 of the pattern MT1 and the region RG7 of the pattern MT2 overlap. In addition, the region RG3 of the pattern MT1 and the region RG6 of the pattern MT2 overlap, and the region RG4 of the pattern MT1 and the region RG5 of the pattern MT2 overlap. Similarly, the region RG5 of the pattern MT1 and the region RG4 of the pattern MT2 overlap, and the region RG6 of the pattern MT1 and the region RG3 of the pattern MT2 overlap. In addition, the region RG7 of the pattern MT1 and the region RG2 of the pattern MT2 overlap, and the region RG8 of the pattern MT1 and the region RG1 of the pattern MT2 overlap.

  Another set of pattern pairs (MT3, MT4) has the same relationship as the above-described pattern pair (MT1, MT2) (the same relationship as the relationship shown in FIGS. 12 and 13). The usage of the pattern pair (MT3, MT4) will be described in detail later.

  As a result of the inversion operation as described above, for example, a superimposed image GC as shown in FIGS. 27 and 28 is acquired. The superimposed image GC shown in FIGS. 27 and 28 corresponds to a state in which the patterns MT1 and MT2 are overlapped without any displacement. In FIG. 28, each of the divided areas RH1 to RH8 obtained by dividing the superimposed image GC into eight is further superimposed on the superimposed image GC. Further, the superimposed image GC is shown as an image obtained by inverting the left and right of the images actually acquired by the imaging units (cameras) 28c and 28d for convenience of illustration.

  The superimposed image GC has a plurality (eight in this case) of periodic density changes (moire).

  The superimposed image GC has four moire in the X direction. Specifically, the superimposed image GC includes a moire on the straight line LN1 in the region RH2, a moire on the straight line LN1 in the region RH7, a moire on the straight line LN2 in the region RH3, and a straight line LN2 in the region RH6. With moire.

  Among these, the moire on the straight line LN1 in the region RH2 is arranged in the X direction at the pitch p2 in the line group LG1 arranged in the X direction at the pitch p1 in the region RG2 of the pattern MT1 and the region RG7 in the pattern MT2. It is generated by overlapping with the line group LG2 (see also FIG. 13). Further, the moire on the straight line LN1 in the region RH7 is arranged in the X direction at the pitch p1 in the line group LG2 arranged in the X direction at the pitch p2 in the region RG7 of the pattern MT1 and the region RG2 in the pattern MT2. Generated by overlapping with the line group LG1. Thus, each moire on the straight line LN1 is generated by superimposing two types of line groups arranged in the X direction at different pitches p1 and p2 in the patterns MT1 and MT2.

  Similarly, each moire on the straight line LN2 is also generated by superimposing two types of line groups arranged in the X direction at different pitches p1 and p2 in the patterns MT1 and MT2. Specifically, the moire on the straight line LN2 in the region RH3 is arranged in the X direction at the pitch p1 in the line group LG2 arranged in the X direction at the pitch p2 in the region RG3 of the pattern MT1 and the region RG6 in the pattern MT2. Generated by overlapping with the line group LG1. Further, the moire on the straight line LN2 in the region RH6 is arranged in the X direction at the pitch p2 in the line group LG1 arranged in the X direction at the pitch p1 in the region RG6 of the pattern MT1 and the region RG3 in the pattern MT2. Generated by overlapping with the line group LG2.

  Further, the superimposed image GC has four moire also in the Y direction. Specifically, the superimposed image GC includes a moire on the straight line LN3 in the region RH1, a moire on the straight line LN3 in the region RH4, a moire on the straight line LN4 in the region RH5, and a straight line LN4 in the region RH8. With moire. In each moire on the straight lines LN3 and LN4, two types of line groups arranged in the Y direction are superimposed at different pitches p3 and p4 (here, p3 = p1, p4 = p2) in the patterns MT1 and MT2. Respectively.

  Specifically, the moire on the straight line LN3 in the region RH1 has a pitch between the line group LG3 (LG1) arranged in the Y direction at the pitch p3 (p1) in the region RG1 of the pattern MT1 and the region RG8 of the pattern MT2. It is generated by overlapping with the line group LG4 (LG2) arranged in the Y direction at p4 (p2). Moire on the straight line LN3 in the region RH4 is a line group LG4 (LG2) arranged in the Y direction at a pitch p4 (p2) in the region RG4 of the pattern MT1 and a pitch p3 (p1) in the region RG5 of the pattern MT2. It is generated by overlapping with the line group LG3 (LG1) arranged in the Y direction. The moire on the straight line LN4 in the region RH5 is the line group LG3 (LG1) arranged in the Y direction at the pitch p3 (p1) in the region RG5 of the pattern MT1 and the pitch p4 (p2) in the region RG4 of the pattern MT2. It is generated by overlapping with the line group LG4 (LG2) arranged in the Y direction. Moire on the straight line LN4 in the region RH8 is a line group LG4 (LG2) arranged in the Y direction at a pitch p4 (p2) in the region RG8 of the pattern MT1 and a pitch p3 (p1) in the region RG1 of the pattern MT2. It is generated by overlapping with the line group LG3 (LG1) arranged in the Y direction.

  FIG. 29 is a diagram showing a superimposed image GC in a state where both patterns MT1 and MT2 are shifted by Δd (for example, 1/40 μm = 25 nm) in the X direction. In FIG. 29, the moire moves in the left (−X) direction in the regions RH2 and RH6 in the superimposed image GC compared to FIG. 27, and the moire in the regions RH3 and RH7 in the superimposed image GC compared to FIG. Moving to the right (+ X).

  Therefore, for example, by comparing the moire in the region RH2 and the moire in the region RH3, the amount of deviation Δd between the patterns MT1 and MT2 can be calculated. Specifically, the moire on the straight line LN1 in the region RH2 is compared with the moire on the straight line LN2 in the region RH3 to obtain a value ΔX (= ΔD2), and the two patterns MT1 and MT2 are calculated based on the equation (8). What is necessary is just to calculate the amount of deviation Δd (Δdx) in the X direction between each other. For example, when ΔX = 40 μm (micrometer), it is calculated as Δdx = 40/40 μm = 1.0 μm (micrometer) (when MG2 = 40). When ΔX = 1 μm (micrometer), Δdx = 1/40 μm = 25 nm (nanometer) is calculated (when MG2 = 40). However, the present invention is not limited to this. For example, the moire in the region RH6 and the moire in the region RH7 are compared to obtain the value ΔX, and the amount of deviation Δd (Δdx) in the X direction between the patterns MT1 and MT2 is calculated. You may make it do.

  FIG. 30 is a diagram illustrating a superimposed image GC in a state where both patterns MT1 and MT2 are shifted by Δdx (for example, 1.0 μm) in the X direction and are shifted by Δdy (for example, 2.0 μm) in the Y direction. In FIG. 30, the moire is moved by ΔX / 2 in the left (−X) direction in the regions RH2 and RH6 in the superimposed image GC, and the moire is in the regions RH3 and RH7 in the superimposed image GC. Compared with, it moves ΔX / 2 in the right (+ X) direction. In addition, the moire in the regions RH1 and RH5 in the superimposed image GC moves by ΔY / 2 in the downward (−Y) direction compared to FIG. 27, and the moire in the regions RH4 and RH8 in the superimposed image GC compared to FIG. Is moved by ΔY / 2 in the upward (+ Y) direction.

  Therefore, for example, by comparing the moire in the region RH2 and the moire in the region RH3, the amount of deviation Δdx in the X direction between the patterns MT1 and MT2 can be calculated. Specifically, the moire on the straight line LN1 in the region RH2 is compared with the moire on the straight line LN2 in the region RH3 to obtain a value ΔX (= ΔD2), and the two patterns MT1 and MT2 are calculated based on the equation (8). A shift amount Δdx (= Δd) between them may be calculated. However, the present invention is not limited to this. For example, the moire on the straight line LN2 in the region RH6 and the moire on the straight line LN1 in the region RH7 are compared to obtain the value ΔX, and the shift amount Δdx between the patterns MT1 and MT2 is obtained. May be calculated.

  Further, for example, by comparing the moire in the region RH1 and the moire in the region RH8, the amount of deviation Δdy in the Y direction between the patterns MT1 and MT2 can be calculated. Specifically, the moire on the straight line LN3 in the region RH1 is compared with the moire on the straight line LN4 in the region RH8 to obtain a value ΔY (= ΔD2), and the two patterns MT1 and MT2 are calculated based on the equation (8). A shift amount Δdy (= Δd) between them may be calculated. However, the present invention is not limited to this. For example, the moire on the straight line LN4 in the region RH5 and the moire on the straight line LN3 in the region RH4 are compared to obtain the value ΔY, and the deviation amount Δdy between the patterns MT1 and MT2 is obtained. May be calculated.

  For example, when ΔX = 40 μm (micrometer) and ΔY = 80 μm, Δdx = 40/40 = 1.0 μm and Δdy = 80/40 = 2.0 μm are calculated (when MG2 = 40). . When ΔX = 1 μm (micrometer) and ΔY = 2 μm, Δdx = 1/40 = 0.025 μm = 25 nm (nanometer) and Δdy = 2/40 = 0.05 μm = 50 nm. (When MG2 = 40).

  In the moire comparison operation described above, it is preferable to use an approximate curve obtained based on the pixel values of a plurality of pixels in the superimposed image GC.

  For example, the following four approximate curves AL1, AL2, AL3, AL4 may be used. The approximate curve AL1 is a curve that approximates moire on the straight line LN1 in the region RH2, and the approximate curve AL2 is a curve that approximates moire on the straight line LN2 in the region RH3. The approximate curve AL3 is a curve that approximates moire on the straight line LN3 in the region RH1, and the approximate curve AL4 is a curve that approximates moire on the straight line LN4 in the region RH8. Each approximate curve AL1, AL2, AL3, AL4 is acquired based on each pixel value V at a plurality of positions X as described above (see FIG. 25).

  Based on the positional deviation (ΔD2) between the two approximate curves AL1 and AL2, the positional deviation (Δdx) in the X direction between the patterns MT1 and MT2 is calculated based on the equation (8). . Further, based on the positional deviation (ΔD2) between the other two approximate curves AL3 and AL4, the positional deviation (Δdy) in the Y direction between the patterns MT1 and MT2 is calculated based on the equation (8). Is done.

  According to this, based on the positional deviation of the approximate curve regarding two moires (periodic density change), the positional deviation between the two patterns MT1 and MT2 (and hence both the objects 91 and 92) is grasped. can do. Specifically, it is possible to perform accurate positioning by correcting misalignment in two translational directions (X direction and Y direction).

  In this embodiment, another set of patterns MT3 and MT4 is used. That is, two sets of pattern pairs (MT1, MT2) and (MT3, MT4) are used. As a result, it is possible to correct the positional deviation (Δθ) in the rotational direction. Hereinafter, such a technique will be described.

  Specifically, as described above, the positional deviation amount (Δdx, Δdy) = (Δxa, Δya) is calculated based on the superimposed image GC (also referred to as GC1) of the pair of patterns (MT1, MT2). . Here, as shown in FIG. 31, the positional deviation amounts (Δxa, Δya) are the positional deviation amounts (Δdx, Δdy) of the pattern MT2 with respect to the pattern MT1 in the vicinity of the center position Q1 of the pattern MT1. In other words, the positional deviation amounts (Δxa, Δya) are deviation amounts of the central position Q3 of the pattern MT2 with respect to the central position Q1 of the pattern MT1. In FIG. 31, for convenience of illustration, both the objects to be bonded 91 and 92 are shown as being relatively displaced from each other. It is very minute.

  Similarly to the above, based on the superimposed image GC (also referred to as GC2) of another set of pattern pairs (MT3, MT4), a positional deviation amount (Δdx, Δdy) = (Δxb, Δyb) is calculated. Here, as shown in FIG. 31, the positional deviation amounts (Δxb, Δyb) are positional deviation amounts (Δdx, Δdy) of the pattern MT4 with respect to the pattern MT3 in the vicinity of the center position Q2 of the pattern MT3. In other words, the positional deviation amounts (Δxb, Δyb) are deviation amounts of the central position Q4 of the pattern MT4 with respect to the central position Q2 of the pattern MT3.

  More specifically, based on the positional deviation ΔD2 between the two approximate curves AL11 and AL12 that approximate the two moire in the X direction in the superimposed image GC2, respectively, in the X direction between the patterns MT3 and MT4. A positional deviation (Δxb) is calculated. Similarly, based on the positional deviation ΔD2 between the two approximate curves AL13 and AL14 that approximate the two moires in the Y direction in the superimposed image GC2, the positional deviation in the Y direction between the patterns MT3 and MT4 ( Δyb) is calculated. The approximate curves AL11 to AL14 may be acquired in the same manner as the approximate curves AL1 to AL4, respectively.

  Based on these positional deviations Δxa, Δya, Δxb, Δyb, a positional deviation Δθ in the rotational direction between the workpieces 91, 92 is calculated.

  Specifically, the positional deviation Δθ in the rotational direction is calculated based on the following equation (9). The value L is the distance between the point Q1 and the point Q2, that is, the distance between the patterns MT1 and MT2.

  This equation (9) is derived from the geometric relationship as shown in FIG.

  Thus, since the positional deviation Δθ in the rotational direction is calculated, it is possible to correct the positional deviation in the rotational direction and perform accurate positioning.

  Further, based on the positional deviations Δxa and Δxb, a positional deviation Δx in the X direction between the workpieces 91 and 92 is calculated based on the following equation (10).

  Further, based on the positional deviations Δya and Δyb, a positional deviation Δy in the Y direction between the workpieces 91 and 92 is calculated based on the following equation (11).

  As described above, based on the positional deviation amounts Δxa, Δya, Δxb, Δyb at the two points, the positional deviation in the translational direction (X direction and Y direction) between the two objects 91, 92 is obtained. Therefore, it is possible to perform more accurate positioning than in the case of obtaining the positional deviation in the translation direction at one point.

<1-6. Detailed operation>
Next, the joining operation according to the first embodiment will be described with reference to the flowcharts of FIGS. FIG. 33 is a flowchart showing an outline of the joining operation according to the first embodiment, and FIG. 34 is a flowchart showing the fine alignment operation in the joining operation. These operations are executed under the control of the controller 100.

  When both the workpieces 91 and 92 are brought into the opposed state in the vacuum chamber 2, first, rough alignment operation is executed in step S10 (FIG. 33). As described above, in the first embodiment, the rough alignment operation is performed using the pattern CT (FIG. 5) and the pattern DT (FIG. 6).

  In step S10, the calculation unit 102 (FIG. 1) determines the rough positioning values (both approximate values) regarding the positional deviations (Δx, Δy, Δθ) between the workpieces 91 and 92 in a plane parallel to the XY plane. Is calculated based on the superimposed image of the pattern CT and the pattern DT. Then, the position correction unit 33 performs a rough alignment operation for correcting the displacement by relatively driving both the workpieces 91 and 92 based on the approximate value.

  Specifically, the alignment is performed so that the center portion of the cross of the pattern CT is arranged at the center portion of four squares of the pattern DT (see FIG. 6). More specifically, an error related to a gap GP (for example, about several tens of μm) between the protruding portion of the cross in the pattern CT in the vertical and horizontal directions and the four squares of the pattern DT is set to a predetermined value (for example, about several μm) or less. Rough alignment is performed (see FIG. 6). As a result, it is possible to keep the positional errors of the workpieces 91 and 92 within a predetermined range (for example, about several μm). In this rough alignment, it is preferable that the position error of both the workpieces 91 and 92 be within the upper limit value of the position error detection by the moire described above. For example, the position error of both the workpieces 91 and 92 is set to a value smaller than a value of 1/2 of the moire period PD and 1 / MG2 (for example, 380 / (2 × 40) = about 4.8 μm). It is preferable.

  Next, a fine alignment operation is executed in step S20. That is, the fine alignment operation is executed after the rough alignment operation. In step S20, the fine alignment operation is executed by using the pattern MT (FIG. 7) as described above.

  Specifically, as shown in FIG. 34, first, in step S21, the image acquisition unit 31 superimposes a superimposed image GC1 related to one set of patterns (MT1, MT2) and a superimposed set related to another set of patterns (MT3, MT4). An image GC2 is acquired.

  Thereafter, in step S22, the calculation unit 102 calculates the positional deviations Δxa and Δya based on the superimposed image GC1 using the third method. Similarly, in step S23, the calculation unit 102 calculates the positional deviations Δxb and Δyb based on the superimposed image GC2 by using the third method described above.

  Next, in step S24, the calculation unit 102 calculates Δx, Δy, Δθ based on the equations (9) to (11). That is, based on the positional deviation (Δxa, Δya) calculated based on the superimposed image GC1 and the positional deviation (Δxb, Δyb) calculated based on the superimposed image GC2, the relative relationship between the workpieces 91 and 92 is relative to each other. The positional deviations Δx, Δy, Δθ are further calculated.

  In step S25, it is determined whether or not the calculated deviation amounts Δx, Δy, Δθ are within a predetermined allowable error range. As the allowable error, for example, a value of several hundred nm (nanometer) to several tens of nm (nanometer) can be set.

  When it is determined that the deviation amounts Δx, Δy, Δθ are not within the predetermined allowable error range, the process proceeds to step S26. In step S26, a position shift correction operation by the position correction unit 33 is executed to correct the calculated shift amounts Δx, Δy, Δθ. Specifically, the head 22 is driven by the translation drive mechanism and the rotation drive mechanism, and the position of the article 92 (that is, the relative position of both the articles 91 and 92) is corrected. Then, the process returns to step S21.

  On the other hand, when it is determined that the deviation amounts Δx, Δy, Δθ are within the predetermined allowable error range, the fine alignment operation is terminated and the process proceeds to step S30.

  According to the above operation, until the deviation amounts Δx, Δy, Δθ are within the allowable error range, the superimposed image GC1, GC2 acquisition operation (step S21) and the positional deviation calculation operation (steps S22 to S24) are aligned. Drive operation (step S26) is repeatedly executed. According to this, the driving error at the time of driving the head 22 is gradually reduced, and a more accurate fine alignment operation is executed.

  Thereafter, in step S30 (FIG. 33), a joining operation is performed. Specifically, the head 22 is moved (lifted / lowered) in the Z direction by the Z-axis lifting / lowering driving mechanism 26 to join the workpieces 91 and 92 together. According to this, both objects 91 and 92 can be joined in the state where both objects 91 and 92 were positioned with very high accuracy.

  In particular, in the above embodiment, since a transmission optical system is used as the imaging optical system, it is not necessary to insert an imaging device (camera) in the gap between the objects to be joined 91 and 92. Therefore, in a state where the objects to be bonded 91 and 92 are disposed in close proximity to each other, a superimposed image is acquired using transmitted light transmitted through both the objects to be bonded 91 and 92, and an alignment operation is performed based on the superimposed image. And the joining operation can be performed. According to this, since the driving distance in the Z direction at the time of joining can be shortened, it is possible to join the objects to be joined 91 and 92 while maintaining good alignment accuracy.

  Further, in particular, according to the above-described embodiment, even when infrared light having a wavelength longer than that of ultraviolet light and visible light is used, a very high-precision position shift detection operation and a very high-precision alignment operation (fine alignment operation) ) Can be performed.

<2. Second Embodiment>
In the first embodiment, the case where the rough alignment patterns AT1 to AT4 are used has been exemplified. In the second embodiment, a case in which rough alignment is also performed using the fine alignment patterns PT1 to PT4 without using the rough alignment patterns AT1 to AT4 is illustrated. In addition, this 2nd Embodiment is a modification of 1st Embodiment, and demonstrates below centering around difference with 2nd Embodiment.

  In the second embodiment, a displacement detection operation based on the principle described below and a drive operation for correcting the detected displacement are performed as a rough alignment operation instead of step S10 in FIG. .

  Below, the position shift detection operation used in the rough alignment operation of the second embodiment will be described.

  FIG. 35 is a diagram showing a situation in which the two patterns MT1 and MT2 are relatively large in the X direction (for example, the amount of deviation Δdx = 50 μm).

  In such a situation, as shown in FIG. 35, an elongated (strip-shaped) non-moire region (a region where no moire occurs) extending in the Y direction of the superimposed image GC exists in the central portion CR in the X direction. .

  36 to 38 are diagrams for explaining such a non-moire region. In FIG. 37 and FIG. 38, the upper pattern line is shown in light color to distinguish the upper pattern from the lower pattern. The same applies to FIGS. 39 and 41 described later.

  FIG. 36 shows a superimposed image in a state where the two patterns PL and PU shown in FIG. 19 and FIG. 20 are respectively rotated by 90 degrees and further shifted by a distance Wx (= Δdx) in the X direction. Yes. In FIG. 36, there is a non-moire region similar to FIG. 35 and 36, the pattern PL and the pattern MT1 correspond to each other, and the pattern PU and the pattern MT2 correspond to each other.

  FIG. 37 and FIG. 38 are diagrams for explaining the principle of generation of such a non-moire region. FIG. 37 is a sectional view conceptually showing the superposition of two patterns PL and PU.

  As shown in FIG. 37, the lower pattern PL has a line group LG1 on the left side (−X side) and a line group LG2 on the right side (+ X side). On the other hand, the upper pattern PU has the line group LG2 on the left side (−X side) and the line group LG1 on the right side (+ X side). As shown in FIG. 38, the line group LG1 is arranged at a pitch p1 in the Y direction (the direction perpendicular to the paper surface in FIG. 37), and the line group LG2 is arranged at the pitch p2 in the Y direction.

  The upper pattern PU is overlapped with the lower pattern PL while being shifted by a distance Wx in the + X direction.

  In the region MR on both sides of the central region CR in the center in the X direction, moire similar to that in FIG. 21 occurs. Each region MR is also expressed as a moire generation region (or moire region).

  On the other hand, in the central region CR at the center in the X direction, the line group LG1 does not exist, and only the line group LG2 of the pattern PU and the line group LG2 of the pattern PL are superimposed.

  36 and 38 show a case where the Y-direction position of each line of the line group LG2 of the pattern PU and the Y-direction position of each line of the line group LG2 of the pattern PL are the same. In this case, both patterns PU. Since a plurality of lines respectively included in each line group LG2 of PL overlap each other and are arranged at a pitch p2, a regular striped pattern is generated by being elongated in a band shape. The center region CR is also expressed as a non-moire region (a region where moire does not occur). The band-like striped region CR is recognized as a region having constant luminance (constant density) when viewed macroscopically. In particular, the central region CR in FIGS. 36 and 38 is acquired as a bright region (high luminance region) compared to the moire region MR. In these drawings, the boundaries (virtual boundary lines) BL1 and BL2 between the strip-shaped central region CR and the moire regions MR on both sides thereof are parallel to each other.

  Similarly, in the superimposed image GC of FIG. 35, a constant luminance region is generated by superimposing the line group LG2 (LG4) having the same pitch in both the patterns MT1 and MT2 in the vicinity of the center line Cx. Specifically, a band-shaped central region CR extending in the Y direction exists as a non-moire region (a region where moire does not occur) having a constant luminance.

  39 and 40 show a case where the Y-direction position of each line of the line group LG2 of the pattern PU and the Y-direction position of each line of the line group LG2 of the pattern PL are shifted by a distance (p2 / 2). Has been. In this case, both patterns PU. Since a plurality of lines included in each line group LG2 of PL are alternately arranged at a pitch (p2 / 2), a black region (solid region) is generated by being elongated in a band shape. 39 and 40 is also recognized as a region having a constant luminance (a constant density) when viewed macroscopically. In particular, the central region CR in FIGS. 39 and 40 is acquired as a dark region (low luminance region) compared to the moire region MR.

  Thus, the luminance of the non-moire region CR varies depending on the relative shift amount in the Y direction between the line group LG2 of the pattern PU and the line group LG2 of the pattern PL. In other words, the luminance of the non-moire region CR varies depending on the relative shift amount in the Y direction of the two line groups LG2 arranged at the same pitch p2. However, each non-moire region CR is common in that it has uniform brightness when viewed macroscopically.

  FIG. 41 is a diagram showing a state in which both patterns PU and PL are shifted in the opposite direction to FIG. That is, it is a conceptual diagram showing a state in which the upper pattern PU is overlapped with the lower pattern PL shifted by a distance Wx in the −X direction.

  Even in such a state, moire occurs in the regions MR on both sides of the central region CR at the center in the X direction.

  In this case, the line group LG2 does not exist in the central region CR in the center in the X direction, and only the line group LG1 of the pattern PU and the line group LG1 of the pattern PL are superimposed. Accordingly, a constant luminance region formed by two line groups LG1 arranged at the same pitch p1 is formed in the central region CR. Similarly to the above, the luminance of the central region CR varies depending on the relative shift amount in the Y direction of the two line groups LG1.

  As described above, in the superimposed image GC, the line group LG1 (or LG2) having the same pitch in both the patterns MT1 and MT2 is provided near the center line Cx in accordance with the positional deviation in the X direction between the patterns MT1 and MT2. It has non-moire areas formed so as to overlap each other (see also FIG. 35).

  The calculation unit 102 calculates the width Wx in the X direction of the central region CR based on the superimposed image GC in FIG. Specifically, the calculation unit 102 obtains the actual length in the X direction corresponding to one pixel, and obtains the value Wx, that is, the shift amount by obtaining the number of pixels corresponding to the width Wx of the central region CR. An approximate value of Δdx is calculated. Thereafter, a rough alignment operation is performed to reduce the shift amount Δdx. As the shift amount Δdx, it is preferable to calculate an average value of lengths (widths) in the X direction of the central region CR1 at a plurality of positions Y.

  In the second embodiment, an X-direction misalignment detection operation and the like are executed in this way.

  Also, with respect to the Y direction, the same operation as the positional deviation detection operation in the X direction is performed.

  In the second embodiment, the misalignment detection operation based on the principle as described above and the drive operation for correcting the detected misalignment are executed as a rough alignment operation instead of step S10 in FIG.

  FIG. 42 is a diagram showing another superimposed image GC. In FIG. 42, the above-described two patterns MT1 and MT2 are superimposed images in a situation where the two patterns MT1 and MT2 are relatively large in both the X direction and the Y direction (for example, the shift amount Δdx = 50 μm and the shift amount Δdy = 100 μm). GC is shown.

  When the superimposed image GC as shown in FIG. 42 is acquired during rough alignment, the above-described misregistration detection operation in the X direction and the like may be performed using the non-moire region CR1 included in the superimposed image GC. .

  Also, with respect to the Y direction, the same operation as the positional deviation detection operation in the X direction is performed. Specifically, a misregistration detection operation in the Y direction or the like may be performed using the non-moire region CR2 included in the superimposed image GC of FIG. Specifically, the calculation unit 102 calculates the width Wy in the Y direction of the central region CR2. Specifically, the calculation unit 102 obtains the actual length in the Y direction corresponding to one pixel and the number of pixels corresponding to the width Wy of the central region CR2, thereby obtaining the value Wy, that is, the shift amount. An approximate value of Δdy is calculated. Then, a rough alignment operation that reduces the shift amount Δdy is performed. As the shift amount Δdy, it is preferable to calculate an average value of lengths (widths) in the Y direction of the central region CR2 at a plurality of positions X.

  According to this, it is possible to calculate the positional deviation amounts Δdx and Δdy in the X direction and the Y direction in FIG.

  FIG. 43 is a diagram showing still another superimposed image GC. FIG. 43 shows a superimposed image GC in a situation where the above two patterns MT1 and MT2 are further shifted in the X and Y directions (for example, the shift amount Δdx = 200 μm, the shift amount Δdy = 200 μm). ing. Also in this case, it is possible to execute rough alignment by performing a misalignment detection operation in the X direction and the Y direction using the non-moire region included in the superimposed image GC in the same manner.

  As described above, in the second embodiment, rough alignment is performed using the fine alignment patterns MT1 to MT4 as described above. Therefore, it is possible to detect a relatively large displacement in the translation direction by using the patterns MT1 to MT4 without using a separate pattern for rough alignment. For example, there is a positional deviation (for example, several tens to several hundreds of μm) exceeding 1/2 of the moire period PD and a value of 1 / MG2 (for example, 380 / (2 × 40) = about 4.8 μm). Even in this case, the positional deviation can be detected only by the patterns MT1 to MT4. In other words, it is possible to perform rough alignment using only the fine alignment patterns PT1 to PT4 without using the separate rough alignment patterns AT1 to AT4.

  In the second embodiment, after step S20, the same operation as in the first embodiment is performed. That is, the operation of acquiring the superimposed images GC1 and GC2 is executed again, and the approximate curves AL1 to AL4 and AL11 to AL14 are obtained based on the superimposed images GC1 and GC2, and the positional deviations Δxa, Δya, Δxb, Δyb are obtained. Is required. Then, based on the positional deviations Δxa, Δya, Δxb, Δyb, the positional deviation amounts Δx, Δy, Δθ are obtained, and an operation for correcting the positional deviation amounts Δx, Δy, Δθ is executed.

<3. Third Embodiment>
In the first and second embodiments, two translations of the objects to be joined 91 and 92 are performed based on the two superimposed images GC1 and GC2 related to the two pairs of patterns (MT1, MT2) and (MT3, MT4). The case where the positional deviation (Δx, Δy) in the direction and the positional deviation Δθ in one rotational direction are obtained is illustrated.

  On the other hand, in the third embodiment, based on one superimposed image GC1 related to one set of pattern pairs (MT1, MT2), the positional deviations (Δx, Δy) of the two objects 91, 92 in two translational directions. And a case where both the rotational displacement of one rotation direction Δθ are obtained.

  In the third embodiment, the alignment operation is executed using only one set of pattern pairs (MT1, MT2). In addition, this 3rd Embodiment is a modification of 2nd Embodiment, and, below, it demonstrates centering around difference with 2nd Embodiment.

  First, as in the second embodiment, rough alignment is performed using the superimposed image GC. However, in the third embodiment, the positional deviation amounts Δdx and Δdy are obtained using one superimposed image GC1, and a driving operation for correcting the positional deviation amounts Δdx and Δdy is executed.

  In the rough alignment operation according to the third embodiment, the positional deviation Δθ in the rotational direction is also obtained based on the single superimposed image GC1 as follows. Then, a driving operation for correcting the positional deviation Δθ is executed.

  FIG. 44 is a diagram illustrating a certain superimposed image GC (GC1). FIG. 44 shows a superimposed image GC in a situation where the two patterns MT1 and MT2 are shifted by a minute angle Δθ (here, 3 degrees) in the rotation direction θ. FIG. 44 shows a case where there is no deviation in the X direction and the Y direction.

  In the superimposed image GC of FIG. 44, boundary lines BL1 and BL2 between the center region CR and the moire region MR are skewed.

  Therefore, in the third embodiment, the superimposed image GC1 is subjected to image processing, and the angle (skew angle) formed by the two boundary lines BL1 and BL2 between the central region CR and the moire regions MR on both sides thereof is calculated. In other words, the rotational deviation amount Δdθ between the two patterns MT1 and MT2 is calculated based on the shape of the non-moire region CR.

  FIG. 45 is a diagram showing another superimposed image GC1. In the superimposed image GC1 of FIG. 45, the two patterns MT1 and MT2 are relatively large in both the X direction and the Y direction (for example, the shift amount Δdx = 50 μm, the shift amount Δdy = 100 μm) and in the rotation direction θ. It was acquired in a situation where a small angle (here, 3 degrees) is shifted.

  FIG. 46 is a diagram showing still another superimposed image GC1. In the superimposed image GC1 in FIG. 46, the two patterns MT1 and MT2 are further shifted in both the X direction and the Y direction (for example, the shift amount Δdx = 200 μm, the shift amount Δdy = 200 μm) and are minute in the rotation direction θ. It is obtained in a situation where the angle (here 3 degrees) is deviated.

  44 to 46, the calculation unit 102 performs image processing on the superimposed image GC1, and the angle (oblique angle) formed by the two boundary lines BL1 and BL2 between the central region CR and the moire regions MR on both sides thereof. Line angle). Thereby, the rotational deviation amount Δdθ between the two patterns MT1 and MT2 is calculated.

  Here, when the rotational deviation amount Δθ exists during rough alignment, the calculation unit 102 calculates the width Wx in the X direction of the central region CR1 and the width Wy in the Y direction of the central region CR2 as follows. You just have to ask for it. Specifically, the width (length in the X direction) of the central region CR1 may be obtained at a plurality of positions Y, and the average value may be obtained as the value Wx (Δdx). Similarly, the width (length in the Y direction) of the central region CR2 may be obtained at a plurality of positions X, and the average value thereof may be obtained as the value Wy (Δdy). When the two boundary lines BL1 and BL2 intersect (see FIG. 44), the average value is calculated by adding the signs after reversing the positive and negative signs related to the width before and after the intersection of the boundary lines BL1 and BL2. What should I do? The same applies to the second embodiment.

  As described above, in the rough alignment operation according to the third embodiment, the positional deviation amounts Δdx, Δdy, Δθ are obtained based on the single superimposed image GC1, and the positional deviation amounts Δdx, Δdy, Δθ are corrected. A driving operation is performed.

  Thereafter, fine alignment is performed based on the same principle as in the first and second embodiments. However, in the fine alignment in step S20 according to the third embodiment, rough alignment is executed using only one superimposed image GC1.

  Specifically, based on the superimposed image GC1, a small deviation amount Δxa in the X direction and a small deviation amount Δya in the Y direction are obtained. For the angle in the rotation direction, the same operation as the detection of the shift amount during the rough alignment is performed. That is, also in the fine alignment, the shift angle Δθ is calculated based on the shape of the central region CR1 (or CR2) in the superimposed image GC1, specifically, the angles of the boundary lines BL1 and BL2.

  Then, the head 22 is driven (micro-driven) in the X direction, the Y direction, and the θ direction in order to correct the positional deviations of these deviation amounts Δxa, Δya, Δθ. That is, the position deviation correction operation is executed.

  As described above, in the third embodiment, the positional deviation Δθ in the rotational direction is calculated based on the shapes of the non-moire regions CR1 and CR2. Therefore, unlike the first and second embodiments, it is not necessary to use two sets of pattern pairs (MT1, MT2) and (MT3, MT4) when determining the positional deviation Δθ in the rotational direction. That is, it is possible to detect a positional deviation in the rotational direction only by one set of pattern pairs (MT1, MT2).

  In the third embodiment, the case where the rotational deviation Δθ is calculated based on the shapes of the central regions CR1 and CR2 in the superimposed image GC1 in both rough alignment and fine alignment is illustrated, but the present invention is not limited to this. . For example, only in one of rough alignment and fine alignment, the rotational deviation Δθ may be calculated based on the shapes of the central regions CR1 and CR2 in the superimposed image GC1 and the deviation Δθ may be corrected.

<4. Modified example>
<Modification regarding 4-1.2 dimensional moire pattern>
In each of the above-described embodiments, the case where a substantially square pattern MT as shown in FIG. 7 is used is illustrated, but the present invention is not limited to this. For example, a pattern MS as shown in FIG. 47 may be used.

  The pattern MS in FIG. 47 has a substantially rhombus shape in which the X direction and the Y direction are diagonal directions, and corresponds to a pattern in which the four corners of the substantially square pattern MT shown in FIG. 7 are deleted.

  The pattern MS has a point-symmetric configuration with respect to its center point. The pattern MS has a line group LG1 in which a plurality of lines are arranged at a pitch p1 in each of the upper left region and the lower right region. In the upper left region, two line groups LG1 arranged in different directions (X direction and Y direction) are divided into two radial regions and arranged. The same applies to the lower right region. The pattern MS includes a line group LG2 in which a plurality of lines are arranged at a pitch p2 in each of the lower left region and the upper right region. In the lower left region, two line groups LG2 arranged in different directions (X direction and Y direction) are divided into two radial regions and arranged. The same applies to the upper right region.

  Then, such a pattern MS is formed on each of the objects to be bonded 91 and 92, and the pattern MS on the object to be bonded 92 is inverted as shown in FIG. Oppose to MS. As a result, as in the above embodiments, the line groups arranged at different pitches face each other. Specifically, the line group having the pitch p1 of the upper pattern MS and the line group having the pitch p2 of the lower pattern MS overlap, and the line group having the pitch p2 of the upper pattern MS and the pitch p1 of the lower pattern MS are overlapped. The line group overlaps. As a result, moire occurs in the superimposed image of the two upper and lower patterns MS.

  49 to 55 are diagrams showing examples of the superimposed image GS of the two patterns MS as described above. FIG. 49 is a diagram showing a superimposed image generated in a state in which two patterns MS2 are overlapped without any positional deviation, as in FIG. FIG. 50 is a superimposed image showing the degree of shift similar to FIG. 51 corresponds to FIG. 42, and FIG. 52 corresponds to FIG. Similarly, FIG. 53 corresponds to FIG. 44, FIG. 54 corresponds to FIG. 45, and FIG. 55 corresponds to FIG.

  By using this pattern MS, it is possible to obtain the same effects as those of the above embodiments. In particular, when the substantially diamond-shaped pattern MS is used, the exclusive area on the surfaces of the objects to be joined 91 and 92 can be reduced as compared with the substantially square-shaped pattern MT.

<Another Modification Regarding 4-2.2 Dimensional Moire Pattern>
Further, as a pattern for detecting misalignment in two different directions using the “dual line comparison method” (also referred to as a two-dimensional pattern for generating moire or a two-dimensional moire pattern), as shown in FIGS. It is not limited to patterns.

  For example, a two-dimensional moire pattern as shown in FIG. 56 may be used.

  The two-dimensional moire pattern shown in FIG. 56 has line groups LG1 arranged at a pitch p1 inclined by 45 degrees with respect to the X direction in the upper left area and the lower right area, and 45 in the opposite direction to the X direction. The line group LG2 which is inclined at a pitch p2 is provided in the lower left region and the upper right region. Then, a superimposed image obtained by inverting the upper moire pattern shown on the left side of FIG. 56 and superposing it on the lower moire pattern shown on the right side of FIG. 56 is acquired. Also in the superimposed image, the line group LG1 having the pitch p1 of the upper pattern and the line group LG2 having the pitch p2 of the lower pattern overlap, and the line group LG2 having the pitch p2 of the upper pattern and the pitch p1 of the lower pattern are overlapped. The line group LG1 overlaps. As a result, moire occurs in the superimposed images of the upper and lower two patterns. In FIG. 56, each line group LG1, LG2 is schematically shown.

  Such an upper moire pattern and a lower moire pattern each have two line groups LG1 and LG2 arranged at different predetermined pitches.

  The superimposed image of the upper moire pattern and the lower moire pattern is obtained by expressing two periodic density changes generated by superimposing line groups LG1 and LG2 having different pitches in both the upper moire pattern and the lower moire pattern. Have in the direction. Then, based on the positional deviation in the X direction between the two approximate curves approximating these two periodic density changes, the positional deviation in the X direction between the objects 91 and 92 is acquired. .

  Similarly, the superimposed image has two other periodic density changes in the Y direction generated by superimposing line groups LG1, LG2 having different pitches in both the upper moire pattern and the lower moire pattern. Then, based on the positional deviation in the Y direction between the two approximate curves approximating these two periodic density changes, the positional deviation in the Y direction between the workpieces 91 and 92 is acquired. .

  However, when the two-dimensional moire pattern of FIG. 56 is used, the original generation direction of moire is inclined 45 degrees with respect to the X direction and the Y direction. Although the periodic density change (moire) also exists in the X direction and the Y direction as described above, the movement amount of the moire in the X direction and the Y direction is about 0. 0 of the movement amount in the original generation direction of the moire. It is reduced to 7 times (= 1 / √2). Therefore, the effect of improving accuracy is reduced. In other words, from such a viewpoint, it is preferable to employ a pattern as shown in FIGS.

  Alternatively, a two-dimensional moire pattern as shown in FIG. 57 may be used. 57 and 58 (described below), the line groups LG1 and LG2 are schematically shown as in FIG.

  On the right side in the lower moire pattern PL shown on the right side of FIG. 57, the line group LG2 arranged at the pitch p2 and the line group LG1 arranged at the pitch p1 are respectively in the C direction. It is arranged to face each other. On the left side in the lower moire pattern PL, the line group LG1 arranged at the pitch p1 and the line group LG2 arranged at the pitch p2 are arranged so as to face each other along the D direction. Yes. Here, the C direction and the D direction are directions inclined by 45 degrees with respect to the X direction and the Y direction, respectively. However, the present invention is not limited to this, and the C direction and the D direction may be the same as the X direction and the Y direction, respectively.

  The upper moire pattern PU shown on the left side of FIG. 57 has the same configuration.

  Then, a superimposed image obtained by inverting the upper pattern PU shown on the left side of FIG. 57 and superposing it on the lower pattern PL shown on the right side of FIG. 57 is acquired. Also in this superimposed image, moire occurs in a portion where parallel line groups having different pitches overlap.

  In the upper moire pattern PU shown on the left side of FIG. 57, the C direction and the D direction are interchanged at the time of inversion. Then, the line group LG1 (lower left side before inversion) of the upper moire pattern PU overlaps the line group LG1 (lower right side) arranged in the C direction of the lower moire pattern PL. The line group LG1 (upper left side before inversion) of the upper moire pattern PU overlaps the line group LG2 (upper right side) arranged in the C direction. As a result, this superimposed image has two periodic density changes in the C direction generated by superimposing line groups LG1, LG2 having different pitches in both the upper moire pattern and the lower moire pattern. Then, based on the positional deviation in the C direction between the two approximate curves approximating these two periodic density changes, the positional deviation in the C direction between the workpieces 91 and 92 is acquired. .

  Similarly, the line group LG1 (upper right side before inversion) of the upper moire pattern PU overlaps the line group LG1 (upper left side) arranged in the D direction of the lower moire pattern PL, and the lower moire pattern PL The line group LG1 (lower right side before inversion) of the upper moire pattern PU overlaps the line group LG2 (lower left side) arranged in the D direction. As a result, the superimposed image also has two periodic density changes in the D direction generated by superimposing the line groups LG1 and LG2 having different pitches in both the upper moire pattern and the lower moire pattern. Then, based on the positional deviation in the D direction between the two approximate curves approximating these two periodic density changes, the positional deviation in the D direction between the two objects 91 and 92 is acquired. .

  However, the two-dimensional moire pattern in FIG. 57 is relatively large in size. In other words, according to the patterns shown in FIGS. 7 and 47, the pattern size can be further reduced. That is, from the viewpoint of size reduction, it is preferable to employ a pattern as shown in FIG.

  Alternatively, a two-dimensional moire pattern as shown in FIG. 58 may be used.

  58, the line group LG2 arranged in the X direction at the pitch p2 and the line group LG1 arranged in the X direction at the pitch p1 are arranged to face each other on the left side in the lower moire pattern PL of FIG. . On the right side in the lower moire pattern PL, the line group LG1 arranged in the Y direction at the pitch p1 and the line group LG2 arranged in the Y direction at the pitch p2 are arranged to face each other.

  58, the line group LG2 arranged in the X direction at the pitch p2 and the line group LG1 arranged in the X direction at the pitch p1 face each other on the right side before the inversion of the upper moire pattern PU in FIG. Has been. Further, on the left side before the inversion of the upper moire pattern PU, the line group LG1 arranged in the Y direction at the pitch p1 and the line group LG2 arranged in the Y direction at the pitch p2 are arranged to face each other. .

  58 may be used by inverting the upper pattern PU of FIG. 58 and overlaying it on the lower pattern PL of FIG. Also in the superimposed image, moire occurs in a portion where parallel line groups having different pitches overlap.

  However, in FIG. 58, the upper pattern PU and the lower pattern PL are not the same, and the left and right are inverted before overlapping. Therefore, for example, in the exposure process included in the semiconductor manufacturing process, each two-dimensional moire pattern is created using different (for exposure) mask patterns (also simply referred to as masks). In this way, when the upper and lower pattern pairs are manufactured with separate masks, the total manufacturing error is more likely to occur than when the upper and lower pattern pairs are manufactured with the same mask, and the moiré positional accuracy is ensured. Hard to do.

  On the other hand, when the two-dimensional moire pattern MT according to each of the above embodiments is employed, the same two-dimensional surface is formed on the surfaces of both the objects to be bonded 91 and 92 by the exposure operation using the same (exposure) mask pattern. Moire patterns (MT1 to MT4) are created (see FIG. 10). Then, the top and bottom of the workpieces 91 and 92 are reversed and the upper and lower patterns are overlapped to generate the moire as described above. As described above, when the two-dimensional moire pattern MT according to the above-described embodiment is used, the two-dimensional moire patterns MT1 to MT4 can be formed on the surfaces of the workpieces 91 and 92 using the same mask. is there. In particular, since the upper and lower pattern pairs (for example, (MT1, MT2), etc.) are generated by the same mask (single mask), the total manufacturing error is smaller than when both patterns are generated by different masks. Can be reduced. In short, it is possible to reduce the manufacturing error to ½. Therefore, it is easy to ensure the positional accuracy of the moire. It is also possible to reduce the cost required for manufacturing an exposure mask. Similarly, from these viewpoints, it is possible to adopt the patterns (FIGS. 47, 56, and 57) according to the modifications other than FIG. 58, and the two-dimensional moire pattern MT according to the modification of FIG. It is preferable to do.

  In each of the above-described embodiments and modifications, the pitches p1 and p2 of the two line groups LG1 and LG2 that generate moire in the X direction and the two line groups LG3 and LG4 that generate moire in the Y direction are used. The case where pitches p3 and p4 are the same is illustrated. For example, in FIGS. 7 and 9, the pitch p3 of the line group LG3 (for example, the pitch in the regions RG1 and RG5) is the same as the pitch p1 of the line group LG1, and the pitch p4 of the line group LG4 (for example, the pitch in the regions RG4 and RG8). ) Is the same as the pitch p2 of the line group LG2. However, the present invention is not limited to this. For example, the pitch p3 of the line group LG3 may be different from the pitch p1 of the line group LG1. Further, the pitch p4 of the line group LG4 may be different from the pitch p2 of the line group LG2. More specifically, in the regions RG1 and RG5 in FIG. 7, a line group arranged in parallel in the Y direction at a pitch (for example, 29 μm) different from the pitch p1 is provided, and in the regions RG4 and RG8 in FIG. 30 μm) may be provided in a group of lines arranged in parallel in the Y direction.

  In the first embodiment and the like, the case where the same pattern MT is used as the two pairs of patterns (MT1, MT2) and (MT3, MT4) is exemplified. In other words, the case where the four line groups LG1 to LG4 in the pattern pair (MT1, MT2) are respectively employed in the four line groups LG5 to LG8 in the pattern pair (MT3, MT4) is illustrated. However, the present invention is not limited to this.

  For example, different patterns may be used for one set of pattern pairs (MT1, MT2) and another set of pattern pairs (MT3, MT4). More specifically, the pitches p5 to p8 of the line groups LG5 to LG8 in the pattern pair (MT3, MT4) are different from the pitches p1 to p4 of the line groups LG1 to LG4 in the pattern pair (MT1, MT2). A value may be adopted. Alternatively, the pattern MT in FIG. 7 may be adopted as the pattern pair (MT1, MT2), and the pattern MS in FIG. 47 may be adopted as the pattern pair (MT3, MT4).

<4-3. Other>
Moreover, in each said embodiment, although the case where this invention was applied to joining operation | movement with alignment was illustrated, it is not limited to this. For example, the present invention may be applied to alignment technology in nanoimprint. More specifically, when the position of the mold (transparent mold) and the position of the substrate on which the circuit pattern is to be formed are aligned, the above-described pattern MT and the like are arranged on the mold and the substrate, respectively Such an alignment operation may be performed.

DESCRIPTION OF SYMBOLS 1 Alignment apparatus 2 Vacuum chamber 28a, 28b Light source 28c, 28d Image pick-up part 91, 92 Target object (to-be-joined object)
AL1-AL4, AL11-AL14 Approximate curve (sine curve approximate curve)
BL1, BL2 border CR Central part (non-moire area)
CT, DT rough alignment pattern CX, CY boundary line (center line)
GA, GB, GC, GC1, GC2 Superimposed image LG1-LG8 Line group MR Moire region MT, MT1-MT4 Two-dimensional moire pattern (moire generation pattern)
p1-p8 pitch

Claims (14)

An alignment apparatus for performing alignment on both objects of a first object and a second object,
Image acquisition means for acquiring a first superimposed image, which is an overlapping image of the first pattern arranged on the first object and the second pattern arranged on the second object;
Calculation means for calculating a positional shift between the two objects in a predetermined plane based on the first superimposed image;
Position correcting means for relatively driving the two objects to correct the positional deviation;
With
The first pattern and the second pattern include a first line group arranged in parallel at a first pitch in a first direction, and a second line different from the first pitch in the first direction. Each having a second line group arranged in parallel at a pitch of
The first superimposed image is
Having a first periodic density change in the first direction generated by superimposing the first line group of the first pattern and the second line group of the second pattern;
A second periodic density change generated by superimposing the second line group of the first pattern and the first line group of the second pattern also has the first direction in the first direction. ,
The calculation means includes a first approximate curve that approximates the first periodic density change and a second approximate curve that approximates the second periodic density change, respectively, Based on the pixel values at a plurality of positions in the first direction of the image, and based on the positional deviation between the first approximate curve and the second approximate curve, An alignment apparatus that calculates a first positional deviation in the first direction between the first position and the second position. The alignment apparatus according to claim 1,
The first pattern and the second pattern include a third group of lines arranged in parallel at a third pitch in a second direction different from the first direction, and the first pattern and the second pattern in the second direction. Each having a fourth line group arranged in parallel at a fourth pitch different from the pitch of 3,
The first superimposed image is
Having a third periodic density change in the second direction generated by superimposing the third line group of the first pattern and the fourth line group of the second pattern;
A fourth periodic density change generated by superimposing the fourth line group of the first pattern and the third line group of the second pattern is also provided in the second direction. ,
The calculating means includes a first approximate curve for approximating the third periodic density change and a fourth approximate curve for approximating the fourth periodic density change, respectively. Based on the pixel values at a plurality of positions in the second direction of the image, and based on the positional deviation between the third approximate curve and the fourth approximate curve, A second positional shift in the second direction is calculated. The alignment apparatus according to claim 2,
The image acquisition unit also acquires a second superimposed image that is an overlapping image of a third pattern arranged on the first object and a fourth pattern arranged on the second object. ,
The third pattern and the fourth pattern have a fifth line group, a sixth line group, a seventh line group, and an eighth line group, respectively.
The fifth line group and the sixth line group are line groups arranged in parallel in the first direction, respectively.
The seventh line group and the eighth line group are each a line group arranged in parallel in the second direction,
The arrangement pitch of the plurality of lines in the fifth line group is different from the arrangement pitch of the plurality of lines in the sixth line group,
The arrangement pitch of the plurality of lines in the seventh line group is different from the arrangement pitch of the plurality of lines in the eighth line group,
The second superimposed image is
In the first direction and the second direction, respectively, two periodic density changes generated by superposition of the third pattern and the fourth pattern,
The calculating means includes
Based on the positional deviation between the two approximate curves representing the two periodic density changes in the first direction, the third positional deviation in the first direction between the two objects is calculated. ,
Based on the displacement between the two approximate curves representing the two periodic density changes in the second direction, a fourth displacement in the second direction between the objects is calculated. ,
Based on the first displacement, the second displacement, the third displacement, and the fourth displacement, a rotational displacement between the two objects is calculated. An alignment device. The alignment apparatus according to claim 2,
The first pattern of the first object and the second pattern of the second object are patterns generated using the same exposure mask,
In a state where the first object and the second object are arranged to face each other, the first periodic density change and the second periodic density change are generated in the first direction. And the third periodic density change and the fourth periodic density change are generated in the second direction. The alignment apparatus according to any one of claims 2 to 4,
The third line group and the fourth line group are a first line extending through the center in the first direction and extending in the second direction in each of the first pattern and the second pattern. Placed across the border of 1,
The first superimposed image has a first non-moire area,
The first non-moire area is in the vicinity of the first boundary line in the first superimposed image in accordance with a positional deviation between the first pattern and the second pattern in the first direction. An area formed by overlapping a line group of the same pitch of both the first pattern and the second pattern;
The first line group and the second line group include a first line extending through the center in the second direction and extending in the first direction in each of the first pattern and the second pattern. Placed across the border of two,
The first superimposed image further includes a second non-moire area,
The second non-moire region is located near the second boundary line in the first superimposed image in accordance with a positional deviation between the first pattern and the second pattern in the second direction. An area formed by overlapping a line group of the same pitch of both the first pattern and the second pattern;
The calculation means obtains a length of the first non-moire region in the first direction and a length of the second non-moire region in the second direction, and the first between the two objects. And calculating a rough value regarding the positional deviation in the direction of the second direction and the positional deviation in the second direction between the two objects.
The position correction means performs a rough alignment operation for correcting the positional deviation by relatively driving the both objects based on the approximate value,
The image acquisition means acquires the first superimposed image again after the rough alignment operation,
The calculation means obtains the first approximate curve and the second approximate curve based on the first superimposed image acquired after the rough alignment operation, and calculates the mutual relationship between the two objects in a predetermined plane. Further calculate the misalignment between
The position correction means, after the rough alignment operation, relatively drives the two objects to correct a positional deviation between the two objects. The alignment apparatus according to any one of claims 2 to 5,
The third line group and the fourth line group are a first line extending through the center in the first direction and extending in the second direction in each of the first pattern and the second pattern. Placed across the border of 1,
The first line group and the second line group include a first line extending through the center in the second direction and extending in the first direction in each of the first pattern and the second pattern. Placed across the border of two,
The first superimposed image is a non-moire region generated according to a positional deviation between the first pattern and the second pattern, and the first boundary line in the first superimposed image A non-moire region formed by overlapping a group of lines having the same pitch in both the first pattern and the second pattern in the vicinity or in the vicinity of the second boundary line;
The calculating means calculates a positional deviation in the rotation direction in the predetermined plane based on a shape of the non-moire area as one of positional deviations between the objects in the predetermined plane. A featured alignment device. The alignment apparatus according to any one of claims 1 to 4,
The first object and the second object have a rough alignment pattern pair separately from the first pattern and the second pattern,
The calculation means uses the rough alignment pattern pair to calculate a rough value regarding a positional deviation between the two objects in the predetermined plane,
The position correction means performs a rough alignment operation for correcting the positional deviation by relatively driving the both objects based on the approximate value,
The image acquisition means acquires the first superimposed image after the rough alignment operation,
The position correction means drives the both objects relatively after the rough alignment operation, and corrects a positional shift calculated based on the first superimposed image. In the alignment apparatus in any one of Claim 1 thru | or 7,
The alignment apparatus according to claim 1, wherein the first superimposed image is an image captured in an out-of-focus state. The alignment apparatus according to any one of claims 1 to 8,
The image acquisition means acquires the first superimposed image and the second superimposed image using transmitted light that passes through both objects in a state where the objects are close to each other. A featured alignment device. The alignment apparatus according to any one of claims 2 to 9,
The alignment apparatus, wherein the first pattern and the second pattern each have a substantially rhombus shape in which the first direction and the second direction are diagonal directions. An alignment method for performing alignment on both objects of a first object and a second object,
a) obtaining a first superimposed image that is an overlapping image of the first pattern arranged on the first object and the second pattern arranged on the second object;
b) calculating a positional deviation between the two objects in a predetermined plane based on the first superimposed image;
c) correcting the displacement by relatively driving both objects;
With
The first pattern and the second pattern include a first line group arranged in parallel at a first pitch in a first direction, and a second line different from the first pitch in the first direction. Each having a second line group arranged in parallel at a pitch of
The first superimposed image is
Having a first periodic density change in the first direction generated by superimposing the first line group of the first pattern and the second line group of the second pattern;
A second periodic density change generated by superimposing the second line group of the first pattern and the first line group of the second pattern also has the first direction in the first direction. ,
Said step b)
b-1) A first approximated curve that approximates the first periodic density change and a second approximated curve that approximates the second periodic density change, respectively. Obtaining based on pixel values at a plurality of positions in the first direction;
b-2) calculating a first positional shift in the first direction between the two objects based on a positional shift between the first approximate curve and the second approximate curve. When,
An alignment method comprising: A semiconductor device produced using the alignment method according to claim 11. An alignment apparatus for performing alignment on both objects of a first object and a second object,
Image acquisition means for acquiring a first superimposed image, which is an overlapping image of the first pattern arranged on the first object and the second pattern arranged on the second object;
Calculation means for calculating a positional shift between the two objects in a predetermined plane based on the first superimposed image;
Position correcting means for relatively driving the two objects to correct the positional deviation;
With
Each of the first pattern and the second pattern has two line groups arranged at different predetermined pitches,
The first superimposed image is
Having first and second periodic density changes in the first direction generated by superimposing lines of different pitches in both the first pattern and the second pattern;
A third direction and a fourth periodic density change generated by superimposing lines having different pitches in both the first pattern and the second pattern are different from the first direction. Have
The calculating means includes
A first approximation curve that approximates the first periodic density change, a second approximate curve that approximates the second periodic density change, and a third that approximates the third periodic density change. And the fourth approximate curve that approximates the fourth periodic density change based on pixel values at a plurality of positions in the first superimposed image, respectively,
Calculating a first positional shift in the first direction between the two objects based on a positional shift between the first approximate curve and the second approximate curve;
A second positional shift in the second direction between the two objects is calculated based on a positional shift between the third approximate curve and the fourth approximate curve. Alignment device. An alignment method for performing alignment on both objects of a first object and a second object,
a) obtaining a first superimposed image that is an overlapping image of the first pattern arranged on the first object and the second pattern arranged on the second object;
b) calculating a positional deviation between the two objects in a predetermined plane based on the first superimposed image;
c) correcting the displacement by relatively driving both objects;
With
Each of the first pattern and the second pattern has two line groups arranged at different predetermined pitches,
The first superimposed image is
Having first and second periodic density changes in the first direction generated by superimposing lines of different pitches in both the first pattern and the second pattern;
A third direction and a fourth periodic density change generated by superimposing lines having different pitches in both the first pattern and the second pattern are different from the first direction. Have
Said step b)
b-1) A first approximate curve that approximates the first periodic density change, a second approximate curve that approximates the second periodic density change, and the third periodic density change. Obtaining a third approximate curve to be approximated and a fourth approximate curve to approximate the fourth periodic density change based on pixel values at a plurality of positions in the first superimposed image, respectively. When,
b-2) calculating a first positional shift in the first direction between the two objects based on a positional shift between the first approximate curve and the second approximate curve. When,
b-3) calculating a second positional shift in the second direction between the two objects based on a positional shift between the third approximate curve and the fourth approximate curve. When,
An alignment method comprising: