JP3099376B2 - Substrate alignment apparatus and alignment method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus and a method for aligning a substrate by photoelectrically detecting an alignment mark formed on the substrate such as a semiconductor wafer or a plate for a liquid crystal display.

[0002]

2. Description of the Related Art Conventionally, in alignment of wafers and plates, a method of photoelectrically detecting an alignment mark formed at a predetermined position on a substrate through a microscope objective lens has been generally used. There are roughly two types of photoelectric detection methods. One is a light beam scanning method in which a mark is relatively scanned with a spot such as a laser beam, and the scattered or diffracted light generated in the mark is received by a photomultiplier or a photodiode. The magnified image of the illuminated mark is captured with a TV camera (vidicon tube or CCD),
There is a method using the image signal. In either case, the obtained photoelectric signal is subjected to waveform processing to determine the center position of the mark.

The light beam scanning system and the imaging system have completely different configurations in their respective scanning systems. Here, both are assumed to be electro-optical scanning devices. Among such scanning devices, Japanese Patent Application Laid-Open No. 61-128106 discloses a method of detecting a mark position by moving a wafer stage one-dimensionally with respect to a laser beam spot.
JP-A-60-130742, JP-A-60-256
The technology disclosed in Japanese Patent Application Publication No. 002 and the like is known.

Further, as a method of detecting a mark position within a scanning range of a one-dimensional scanning beam after positioning a wafer stage at a design value, Japanese Patent Application Laid-Open Nos. 56-12729 and 57-138134 disclose a method. And the like are known. Japanese Patent Application Laid-Open No. Sho 57-1426 discloses an apparatus using an imaging method as an electro-optical scanning apparatus.
No. 12, JP-A-60-178628, JP-A-60-225428, and the like are known.

In these prior arts, monochromatic light is used as a scanning beam or mark illumination light mainly for the following two reasons. In a projection type exposure apparatus (stepper), in a type in which a wafer mark is detected via a projection optical system, a single-wavelength illumination light or a laser beam is used to avoid large chromatic aberration of the projection optical system. A monochromatic laser beam is used to focus on a minute spot for high-brightness, high-resolution detection. When the monochromatic illumination light (or beam) is used as described above, a relatively large S / N ratio can be obtained. However, in a wafer handled by an exposure apparatus, a photoresist layer having a thickness of about 0.5 μm to 2 μm is generally formed on the entire surface of the wafer. Therefore, an interference phenomenon due to monochromaticity has occurred, which has been erroneously detected at the time of detecting the mark position, or has resulted in an unclear image.

Therefore, in recent years, in order to reduce the interference phenomenon caused by the resist, it has been proposed to increase the wavelength of the illumination light or increase the bandwidth thereof. For example, in an imaging type scanning device, illumination light is generated from a halogen lamp or the like, and its wavelength bandwidth is about 300 nm (excluding the photosensitive area to the resist)
Then, the coherence of the light reflected by the surface of the resist and the surface of the wafer almost disappears, and a clear image can be detected. Accordingly, in the imaging method, an extremely high-precision alignment sensor which is not affected by the resist can be obtained only by whitening the illumination light (widening the band) and achromatizing the imaging optical system.

[0007] Using the whitened illumination light,
When a mark on a wafer is detected by an imaging unit (a vidicon tube or a CCD), the stage on which the wafer is mounted needs to be positioned so that the mark is positioned within the observation visual field (within the imaging range) of the alignment optical system. is there. That is, the wafer on the stage has a position error with respect to the moving coordinate system of the stage depending on the accuracy of mechanical pre-alignment. For this reason, even if the stage is moved with the design value with respect to the field of view of the alignment optical system, it is not necessarily guaranteed that the stage will fall within the imaging range.

In general, the detection range (imaging range) of a high-precision fine alignment sensor is narrow, so that it is necessary to detect a rough mark position for moving the stage to a position where fine alignment can be performed. This rough mark position detection is called search alignment, and various methods have been proposed. As one method of search alignment using the imaging means, the magnification of the alignment optical system can be switched between low magnification and high magnification, and a wide range on the wafer is imaged at a low magnification at a time, and the image signal ( A method of detecting a mark by analyzing a video signal) can be considered. According to this method, it is not necessary to shift the wafer stage for search alignment, and only the analysis time of the video signal is required, so that the speed is increased.

[0009]

However, when trying to view a wide range by lowering the magnification of the alignment optical system, various circuit patterns other than marks on the wafer enter the imaging range, making it difficult to analyze image signals. Problems arise. In particular, when the illumination light for mark detection is whitened, a part of the edge of the circuit pattern on the wafer surface or the boundary edge of the street line (geometric step structure) can be clearly detected without the influence of interference fringes. ,
An edge of such a non-mark pattern may be erroneously detected as a mark position.

Further, lowering the observation magnification of the alignment optical system lowers the resolution (actual size on a wafer per pixel) at the time of image analysis, so that the width of the alignment mark in the measurement direction is several μm. In the case where the level is reduced to a small extent, there is a possibility that the contrast of the signal becomes insufficient.
That is, the waveform change (amplitude) corresponding to the mark on the video signal becomes weak, and the waveform processing becomes difficult.

As a search alignment using the image pickup means, the wafer stage is moved by a predetermined amount in the X direction or Y direction.
There is also a method in which video signals for one screen are taken in by shifting in the direction, and analysis of whether or not a mark exists is repeated. In this case, since the alignment optical system can use a high magnification, the resolution of the video signal is high,
There is an advantage that it can be used for fine alignment as it is. However, on the other hand, the captured video signal for one screen is analyzed to confirm the presence of the mark edge,
When the stage does not exist, the stage is shifted by a fixed amount, so that there is a problem that the speed is lacking.

An object of the present invention is to provide an alignment apparatus and method for searching a mark position with high throughput and high accuracy.

[0013]

According to the present invention, there is provided a substrate having an alignment mark (BPMn) formed with an optical or geometrical difference with respect to a surface. (W), a stage (ST) movable two-dimensionally, a coordinate measuring device (IFX, IFY) for measuring position coordinates of the stage (ST),
Illumination system (7, 8, 9, 9A, 9B, 9C) for illuminating a local area of a size that can include a mark (BPMn) on a substrate
Imaging means (2) for imaging a local area and outputting an image signal according to a change in lightness and darkness in the local area, and mark position detection for detecting a position of the mark within an imaging range based on the image signal Means (30) for moving the stage (ST) on the basis of the mark position detected by the mark position detecting means (30) and the measurement value of the coordinate measuring device (IFX, IFY). In the apparatus for aligning (W) with respect to a predetermined reference position, the mark position detecting means includes a first waveform of an image signal obtained by one-dimensional scanning (scanning line SL) in a local area by the imaging means (2). Storage circuit for storing data (43)
A shift movement control circuit (52, 54, step 1108) for controlling the movement of the stage (ST) so as to shift the stage (ST) in the one-dimensional scanning direction by a predetermined distance smaller than the imaging range (VPAx); The second waveform data of the image signal obtained by the one-dimensional scanning of the imaging means (2) is compared with the first waveform data stored in the storage circuit (43), and a mutually overlapping portion (WD) The first and second waveform data are joined together after correction so that the waveforms of the first and second waveforms substantially coincide with each other.
Waveform synthesis circuit (steps 1000 to 101)
6) and the position (IUX (i) of a waveform portion equivalent to the waveform (BWn) that changes with specific characteristics according to at least one edge of the mark (BPMn) based on the composite waveform data U ′ (I). ) To extract an edge position extraction circuit (steps 1120 to 1126). Then, of the extracted edge positions, a selection circuit (steps 1128 to 1128) for selecting an edge position that satisfies the design condition
1136), and the mark position is specified based on the selected edge position.

[0014]

[Operation] In the present invention, a large area on a wafer is C
When capturing an image with a two-dimensional image sensor such as a CD or a vidicon tube, or a one-dimensional image sensor, the wafer is shifted by a certain amount in the scanning line direction with respect to the image sensor to capture an image signal, and an overlapping area on a screen. , The levels and amplitudes of the image signals are corrected and then joined. For this reason, the magnification of the optical system for forming an image of the mark pattern on the wafer surface on the image sensor can be increased, and the size on the wafer per pixel (pixel) on the image signal is sufficient for the accuracy. High resolution. Further, in the present invention, in order to recognize the mark position on the wafer after creating data of the image signal waveform combined into one, the image signal for each screen before combining is analyzed to recognize the mark. High-speed processing becomes possible as compared with the system. Further, in the present invention, at least one slope portion (one edge) of a bottom waveform (or a peak waveform) appearing at a portion corresponding to a mark position (edge position) in a synthesized image signal waveform is detected. Are determined to meet predetermined conditions (contrast value, design distance, interval, etc.).
Therefore, it is not necessary to detect the slope portions on both sides of the bottom waveform (peak waveform) in the algorithm, so that the number of data to be processed is reduced, and there is an advantage that the time for the arithmetic processing for mark recognition is reduced.

[0015]

Next, the configuration of a projection exposure apparatus provided with an alignment apparatus according to an embodiment of the present invention will be described with reference to FIG. In FIG. 1, an image of a pattern area PA on a reticle R is formed and projected on a wafer W via a projection lens 1. The wafer W is mounted on a stage ST that moves in the X and Y directions by a step-and-repeat method, and the coordinate position of the stage ST is measured by the laser interferometers IFX and IFY. Reticle R is aligned with the apparatus (optical axis AX of projection lens 1) by positioning reticle alignment marks RM1 and RM2 provided on both sides of pattern area PA with respect to reticle alignment microscopes RA1 and RA2. You.

The apparatus according to the present embodiment is applied to a wafer alignment sensor which detects only marks on the wafer W by an off-axis method. The wafer alignment sensor includes a prism-shaped mirror 12 disposed immediately below the projection lens 1 and an objective lens 1.
1, a beam splitter 10, an imaging lens 5, a conjugate index plate 4, an imaging lens 3, and a two-dimensional imaging device 2 such as a CCD.

Further, in order to illuminate the mark area on the wafer W, a light source 8 such as a halogen lamp,
, An optical fiber 7 for guiding light of a broadband wavelength (the photosensitive area of the wafer W is cut), a condenser lens 9, an illumination field stop 9A, a total reflection mirror 9B, a lens system 9C, and a beam splitter 10 described above. An illumination optical system is provided.

In the above configuration, the wafer W is disposed optically conjugate with the index plate 4 with respect to the combined system of the objective lens 11 and the imaging lens 5, and the index plate 4 and the light receiving surface of the CCD 2 are connected to the imaging lens 3 Are conjugated with respect to Therefore C
The CD 2 simultaneously captures an enlarged image of the mark on the wafer W and an enlarged image of the fixed (reference) mark on the index plate 4. The exit end face of the fiber 7 of the illumination optical system is relayed to the pupil plane (aperture stop position) between the objective lens 11 and the lens system 9C as a secondary light source image, and performs Koehler illumination on the wafer W. Further, the field stop 9A is conjugated to the wafer W by a combined system of the objective lens 11 and the lens system 9C,
The aperture image of the field stop 9A is projected on the wafer W.

In this embodiment, at least the objective lens 1
1, achromatizing is performed on each of the imaging lens 5 and the imaging lens 3, and deterioration of the imaging characteristics due to chromatic aberration is suppressed. Further, in the apparatus of the present embodiment, the reference mark FM is provided on the stage ST, the projection point of the index mark on the index plate 4 in the wafer alignment sensor onto the wafer W, and the reticle alignment mark RM on the reticle R.
It is used to measure the distance (baseline) between the projection points of RM1 and RM2.

Next, a circuit for processing a video signal from the CCD 2 in FIG. 1 will be described with reference to FIG. CCD2
Denotes a two-dimensional image sensor, in which pixels are arranged in a horizontal scanning direction and a vertical scanning direction.
In D2, the measurement direction that crosses the edge of the mark on the wafer W is made to coincide with the horizontal scanning direction. Well, CC
From D2, a composite video signal in which a horizontal synchronization signal and a vertical synchronization signal are mixed is obtained. This video signal is
The signal is sent to an analog-digital converter (ADC) 42 via a pre-processing circuit 40 such as a frequency filter or an AGC.
At the same time, the video signal from the CCD 2 is sent to a control circuit 44 including a synchronization signal separation circuit, a clock generation circuit, and the like.

The control circuit 44 generates a clock signal S based on the horizontal synchronizing signal of the CCD 2 so that one clock pulse is generated for each pixel electrical scan (read scan).
Output CL. This clock signal SCL is output from the CCD 2
And a memory (RAM) 43 for detecting whether or not the electrical scanning of the pixel falls within the sampling range (the number of horizontal scanning lines in the vertical direction) in one frame, and a memory (RAM) 43 for storing output data of the ADC 42. It is sent to an address counter 48 which outputs an address value.

Accordingly, in the RAM 43, digital waveform data for a designated number of lines from a predetermined horizontal scanning line of the CCD 2 is stored. The waveform data in the RAM 43 is
Address bus AB managed by processor 50
The processor 50 is connected to the US and the data bus D-BUS.
And a predetermined waveform processing calculation is performed. Address bus A-BUS and data bus DB of processor 50
A stage controller 52 for controlling the stage ST is connected to the US, and the controller 52 inputs the coordinate measurement values of the interferometers IFX and IFY and inputs the stage ST.
Of the drive motor 54 is controlled.

The control unit 30 as the mark position detecting means shown in FIG. 1 includes the members 40 to 50 shown in FIG. 2, and the symbol WSC represents a control signal to the controller 52, which is a signal A- in FIG. BUS and D-BUS are supported. Reference symbol RSC in FIG. 1 represents a bus line for controlling a reticle stage (not shown).

Next, a preferred mark shape and arrangement for this embodiment will be described with reference to FIGS. 3, 4, and 5. FIG. FIG. 3 shows a shot arrangement on the wafer W, and the projected image of the pattern area PA of the reticle R is aligned with each of the shot areas SA. At the time of exposure, the wafer stage ST is precisely stepped such that the center of each shot area SA coincides with the center of the pattern area PA of the reticle R.

In each shot area SA, a wafer mark MDx for alignment in the X direction and a wafer mark MDy for alignment in the Y direction are formed. In this embodiment, it is assumed that the marks MDx and MDy are detected by off-axis type wafer alignment sensors (2 to 12 in FIG. 1). Each mark MD
As shown in FIG. 5A, x and MDy are such that three diffraction grating bar marks BPM1, BPM2, and BPM3 are parallel to each other at intervals hb and hc in a street area having a width LD between shot areas SA. The multi-marks are arranged in a row.

FIG. 5A shows a mark MDx for the X direction.
Only the bar mark BPMn extends in the Y direction. In addition, as shown in FIG.
n is a pattern in which minute rectangular patterns formed in a convex shape (or a concave shape) with respect to the wafer base are arranged at a constant pitch along the center line CL. Further, in the present embodiment, it is assumed that the center line CL of the center mark BPM2 among the three bar marks BPM1, BPM2, and BPM3 faces the center of the shot area SA.

Here, the three bar marks BPMn are provided so as to be substantially directed toward the center of the shot area SA, but may be provided along the direction in which the street line extends. In this case, the longitudinal direction of the three bar marks BPMn, that is, the non-measurement direction can be made as long as the length of one side of one shot area SA at most, so that the pre-alignment error in the non-measurement direction is almost several. mm,
This allowable amount is a value that does not cause any problem in practical use.

FIG. 4 shows the arrangement of the index marks TL and TR on the conjugate index plate 4. Each of the index marks TL and TR is composed of a plurality of fine lines formed of a chrome layer on a transparent glass plate. In the figure, IF indicates an image field of the objective lens 11, and a rectangular area PPA indicates an imaging area of the CCD 2. In the case of fine alignment, two index marks TL, T
The wafer stage ST is positioned such that the mark MDx (or MDy) is sandwiched between R.

Further, as shown in FIG. 6, the conjugate index plate 4 of this embodiment has, for the X direction, a region VPBx including index mark groups TL and TR sandwiching the wafer mark MDx, and a region VPBx above (non-measurement direction). And a transparent area VPAx set for search alignment. CCD2 has an area VP
It has an imaging range that covers Ax and VPBx. Although FIG. 1 shows only the alignment optical system for the X direction, the alignment optical system for the X direction and the Y direction is
In the case where the conjugate index plate 4 is arranged separately from the conjugate index plate 4, the conjugate index plate 4 is also separately provided for the X direction and the Y direction.

In this embodiment, when the mark MDx (or MDy) on the wafer W is detected by search alignment, the area VPAx on the conjugate index plate 4 is used.
In the case of fine alignment after the mark MDx is found, the area VPBx is used. That is, a mode (search alignment) for processing an image signal obtained when the area VPAx is horizontally scanned within the imaging range of the CCD 2 and a mode for processing an image signal obtained when the area VPBx is horizontally scanned ( And fine alignment).

Now, the wafer W in a mechanically pre-aligned state is placed on the wafer stage ST so that one mark MDx on the wafer W enters the visual field of the objective lens 11 based on the design value. If the pre-alignment accuracy is good when the stage ST is moved, FIG.
The three bar marks BPM1, BPM2 and BPM2 shown in FIG.
BPM3 is located substantially at the center of region VPAx in FIG. When the pre-alignment error is large, the region VPA
Within x, the mark MDx is shifted in the X direction and the Y direction. Pre-alignment failure in the Y direction of the mark MDx is dealt with by lengthening the Y direction (non-measurement direction) of the bar mark BPMn.

FIG. 7A shows a state after completion of pre-alignment.
, Any mark MDx (BPM1 to BP) on the wafer W
M3) was specified and the wafer stage ST was sent at the design value
Sometimes conjugate imaged by CCD2 of alignment optical system
The positional relationship between the mark on the index plate 4 and the wafer mark MDx
FIG. 7B shows an example of the area in the index plate 4 at that time.
Each of N (where N ≧ 1) scanning lines SL passing through VPAx
The image signal is transferred to pixels (pixels) at the same horizontal position.
Image signal waveform obtained by averaging in the vertical direction
Show. In FIG. 7B, the horizontal axis represents the scanning position of the scanning line.
The pixel PX ₀~ PX_nAddress (A in RAM 43)
Dress value).

The control circuit 44 has one scanning line SL.
Is divided, for example, by 1024 by the clock signal SCL, and pixels per one pulse of the clock signal SCL are allocated. Therefore, the image (video) shown in FIG.
The signal waveform is stored in the RAM 43 in the order of address as a digital value of the signal level for each pixel. The actual averaged waveform is obtained by the processor 50 by reading out image signal data for N scanning lines once stored in the RAM 43 and performing an arithmetic operation. The image signal waveform obtained from the CCD 2 has simple bottom waveforms BW1, BW2, and BW3 at the positions of the bar marks BPMn because the interference phenomenon in the resist layer is reduced by using the broadband illumination light. When the entire mark MDx falls within the area VPAx at a time in the pre-aligned state as shown in FIG. 7A, the video signal waveform of FIG. 7B is analyzed to determine the specific bar of the mark MDx. If the pixel position of the mark, for example, BPM2, is obtained, the bar mark position for the index marks TL, TR on the index plate 4 can be specified based on the coordinate position of the wafer stage ST and the pixel position of the bar mark BPM2 at that time. Therefore, fine alignment of the mark MDx can be performed by immediately moving the wafer stage ST and moving the mark MDx between the index marks TL and TR. However, the distance in the scanning line direction of the area PPA that can be imaged at a time has a resolution of 0.2 to 0.3 μm per pixel (on the wafer), and
Assuming that the number of pixels of the scanning line is 1024, it is limited to about 200 μm to 300 μm on the wafer W. Of course, if the magnification of the combining system of the objective lens 11 and the imaging lens 5 in the alignment optical system is reduced, the size per pixel increases, and the range that can be observed at a time can be expanded. However, in that case, the resolution becomes coarse, and the contrast corresponding to the mark image on the video signal waveform deteriorates. Therefore, the magnification of the synthesis system (magnification from the wafer W to the CCD 2) cannot be set to a certain value or less, and the observation range of one pixel is limited on the wafer.

In the present embodiment, among the video signals of one screen imaged by the CCD 2, the video signal waveforms of N scanning lines in the area VPAx are averaged, stored as one waveform data, and then stored as one waveform data. The stage ST is moved by a certain distance in the measurement direction (horizontal scanning direction). Then, one waveform data is created for the next imaging screen in the same procedure, and one piece of combined waveform data is created by splicing with the waveform data obtained on the previous screen. In this way, by sequentially joining the waveform data obtained for each screen in the RAM 43, the waveform data of the video signal as if the search range was expanded in the horizontal scanning direction (mark measurement direction) of the CCD 2 can be obtained. To get.

The generation of the composite waveform data as described above is performed by R
Any number of screens can be used as long as the capacity of the AM 43 permits, but the size on the wafer in the horizontal scanning direction of one screen is 20 mm.
In the case of about 0 to 300 μm, by joining two screens or about three screens, it is possible to sufficiently cope with deterioration of prealignment accuracy. Thus, next, a method of creating composite waveform data by screen joining will be described. FIG. 8A shows a CCD.
2 shows an example of an image that is first captured, in which only two of the three bar marks BPM1 to BPM3 (mark MDx) enter the imaging range (area VPAx) due to deterioration of prealignment accuracy. It is assumed that the other one (BPM3) is out of the imaging range. At this time, the averaged waveform data of the N scanning lines stored in the RAM 43 is as shown in FIG. Further, the wafer stage ST is shifted in the scanning line direction so that the second imaging region VPAx is shifted from the first imaging region VPAx in FIG.
Positioning is performed so as to overlap by the width WD in the scanning line direction as shown in FIG. The management of the width WD is extremely accurate because the positioning of the stage ST is controlled by the measured values of the interferometers IFX and IFY. The overlap width WD is, for example, 3
Since the largest one of the intervals hb and hc of the multi-bar mark BPMn, that is, hb <hc is used here, the value is smaller than the sum of the interval hc and the width hm of the bar mark BPMn itself (see FIG. 5B). It is largely determined.

That is, the width WD is set so that WD ≧ hc + hm, and the number of pixels in the horizontal scanning line direction (10
24) or less, preferably about 1/8. Of course, in the case of a single mark other than the multi-bar mark, it is theoretically sufficient to secure the overlap width WD by the width hm of the single mark. However, in practice, the joint state at the time of waveform synthesis is improved. Therefore, the overlap width WD must be at least two or three times the width hm.

The image pickup area VPAx after the shift as described above is as shown in FIG. 8B, and the averaged waveform data obtained from the second image pickup area VPAx is shown in FIG.
(D). The waveform data in FIG.
The data is stored in the RAM 43 in an address area different from the waveform data of FIG. As shown in FIG. 8B, by setting the overlapping width WD to be equal to or more than hc + hm, the bar mark BPM imaged for the first time is displayed in the second imaging region VPAx.
In addition to 2, the third bar mark BPM3 also enters. Therefore, as shown in FIG. 8C and FIG.
Waveform data including W1 and BW2, and bottom waveform BW2,
The waveform data including BW3 is separately stored. However, the end address value of the waveform data in FIG. 8C, the start address value of the waveform data in FIG.
Since the difference (the number of points) of the address value with respect to D is predetermined, the two waveform data can be easily combined into one continuous waveform data.

FIG. 8E shows a state where two waveform data are combined and stored in another address area in the RAM 43. Here, the waveform data of FIG. 8D is used as it is, and the waveform data of FIG. 8C is synthesized by removing the portion of the overlap width WD. In the synthesized waveform data of FIG. 8E, the spliced portion of the waveform is SPL, but if the two waveform data are spliced as they are, the respective signal intensities and amplitudes (contrasts) of the two waveform data are different. At the same time, a large level change occurs in the joint SPL, and three bottom waveforms B that should have substantially the same amplitude
W1, BW2, and BW3 become irregular. Especially C
If the output stage of CD2 has AGC, the stage S
When the average luminance in the screen changes with the shift movement of T, the signal level greatly changes.

Therefore, the averaged waveform data of FIG. 8C extracted from the first imaging area VPAx (for convenience, the left screen is associated with FIG. 8) and the second imaging area VPAx
It is necessary to successfully splice the averaged waveform data of FIG. 8D extracted from PAx (for convenience, the right screen is associated with FIG. 8). In the present embodiment, the description is limited to the case where the waveform data obtained from two screens is connected. However, the same method can be applied to the case where the waveform data for three screens or more is connected.

FIG. 9 schematically shows enlarged two waveform data in the overlap width WD. FIG. 9A shows the waveform data on the left screen, and the maximum value in the overlap width WD is LLp. , The minimum value (bottom value) generated in the bottom waveform BW2 is LL
b. Similarly, FIG. 9B shows waveform data on the right screen, where the maximum value in the overlap width WD is LRp and the bottom waveform BW2
The minimum value (bottom value) occurring in the portion is LRb. FIG.
The processor 50 shown in FIG. 4 calculates the maximum values LLp and LRp and the minimum value Lp from the two waveform data in the RAM 43, respectively.
Lb and LRb are obtained by calculation. Further, the processor 50 obtains the midpoint levels LLm and LRm of the amplitudes of the bottom waveforms of the right screen and the left screen, respectively, by the following equations. LLm = (LLp + LLb) / 2 (1) LRm = (LRp + LRb) / 2 (2) Here, the waveform data of the overlapping portion of the left screen (FIG. 9A)
The intensity at each pixel (address) location that forms
(1) to U (k) (where k is the total number of pixels PX _n -PX ₀ of the waveform data on the left screen), and the intensity and amplitude of the waveform data on the left screen are corrected. The intensity at each pixel position of the corrected waveform data is represented by U ′ (1) to
Assuming that U ′ (k), the processor 50 performs a correction operation based on the following equation, using I as a variable from 1 to k. U ′ (I) = (U (I) −LLm) (LRp−LRb) / (LLp−LLb) + LRm (3) As a result, the waveform of the overlapping portion becomes substantially equal, and the joint portion SPL is extremely smooth. become.

FIG. 10 shows a flowchart of the processing of equation (3). In FIG. 10, the processor 50 determines in steps 1000, 1002, and 1004 the maximum value LL at the overlapping portion on the two waveform data as described above with reference to FIG.
p, LRp, minimum values LLb, LRb, and midpoint level L
Lm and LRm are detected and stored. Then step 100
At 6, the processor 50 secures an address area in the RAM 43 for storing corrected waveform data obtained by correcting the waveform data of FIG. 8C. Then, in step 1008, the variable I is set to 1, and in step 1010, (LRp-LRb)
/ (LLp-LLb) is calculated, and the result is stored in a variable D
Input to S. Further, a final correction value U ′ (I) is obtained in step 1012 as (U (I) −LLm) × DS + LRm. After that, in step 1014, the correction value U ′ is stored in the area of the correction waveform data secured in the RAM 43.
(I) is transferred, and it is determined in step 1016 whether or not the variable I has reached k. If not, the variable I is incremented and the process returns to step 1010.

By the above steps 1000 to 1016, the corrected waveform data of the original signal obtained from the left screen is stored in the RAM 43. After that, the processor 50
The corrected waveform data and the waveform data on the right screen (FIG. 8D) are combined into one waveform data. There are several methods for creating the composite waveform data. One of them is
As shown in FIG. 8E, the overlapping portion (W
This is a method in which the data in D) is deleted and all the waveform data on the right screen in FIG. The second method is a method in which the center of the overlapping portion in the corrected waveform data and the center of the overlapping portion in the waveform data of FIG. The third method is to take out data at every other pixel from the overlapping portion in the corrected waveform data, and to take every other pixel from the overlapping portion in the waveform data of FIG. This is a method of extracting data and synthesizing it.

In any of the methods, the synthesized waveform data can be smoothly joined as if they were captured as one image. FIG. 11 shows an example of the synthesized waveform data thus synthesized, and the processor 50 searches for the mark MDx (bar mark BPMn) based on the synthesized waveform data. FIG. 11A shows synthesized waveform data in which a horizontal axis indicates a scanning position (address value) and a vertical axis indicates a signal level, and FIG. 11B shows the synthesized waveform data using a numerical filter technique. Processor 5
0 represents the differential waveform data calculated. This differential waveform data will be described later in detail. Before the above-described waveform splicing process, the processor 50 performs the smoothing process of the original waveform data shown in FIGS. 8C and 8D.
More on this later. When combining original waveform data of two or more screens, the original waveform data from any screen may be corrected.

Next, the overall operation of mark position detection in this embodiment will be described with reference to the flowchart of FIG. In step 1100 of FIG.
0 selects a search alignment sequence set in advance by the operator. The selection of the sequence means whether or not to synthesize the waveform data, and also how many screens of the waveform data are to be synthesized. Assuming that the realistic maximum number of screens to be combined is 3, the processor 50 sets the combined screen number fs (0 to
3) is read and sequence setting is performed. If fs is 0, it is determined in step 1102 that the search alignment itself has not been set, and the sequence in FIG. 12 is canceled, and the process proceeds to, for example, a fine alignment sequence. If it is determined in step 1102 that it is not 0, it is determined in next step 1104 whether or not fs> 3 is true. If it is true, the error processing routine ER is executed.
Jump to R.

Next, at step 1106, the processor 50
Is the optimal stage S corresponding to the number of screens fs (1-3)
The position coordinates of T are calculated. Here, when fs is 1, the position of the wafer stage ST may be calculated based on design layout information of the mark MDx. When fs is 2 or 3, two or three stage positions are calculated on the design value such that the bottom waveform of the mark MDx appears almost at the center of the synthesized waveform data. At this time, an amount corresponding to the overlap width WD in the horizontal scanning line direction is also taken into account at the same time. Next, in step 1108, the processor 50 controls the motor 54 via the stage controller 52, and positions the stage ST at the coordinate position determined in step 1106.

The processor 50 then proceeds to step 111
At 0, the image signal waveform (N lines) in the area VPAx on the wafer W is taken into the RAM 43 by the CCD 2. Further, in step 1112, it is determined whether or not there is another screen to be captured.
Return to 08, stage ST so that other screens can be imported
Move the position of.

When the above sequence is completed, the RAM 4
3 stores N pieces of original waveform data for a maximum of three screens. Then, the processor 50 adds and averages the N original waveform data for each screen in step 1114 to create averaged waveform data, and stores the averaged waveform data in the RAM 43. Further, in step 1116, the processor 50 smoothes the averaged waveform data by using a numerical filter, and in step 1118, performs 1st processing by the processing shown in FIG.
Create synthetic waveform data for a book. However, if signal waveforms for two or more screens are not captured, step 1118 will be skipped. However, even when only the averaged waveform data for one screen is used in the subsequent processing, it will be referred to as synthesized waveform data in order to unify the description. The smoothing in step 1116 is for removing high-frequency components included in the original signal waveform. FIG.
FIG. 13A shows the data group of the averaged waveform data in the RAM 43 in the order of addresses in a discrete manner. On the other hand, the correlation calculation with the parabolic numerical filter FNa as shown in FIG. Execute for each (address).

Next, the processor 50 executes the sequence shown in FIG. 14. At this stage, the synthesized waveform data as shown in FIG. 11A is stored in the RAM 43. The sequence of FIG. 14 is immediately executed by the processor 50 after the sequence of FIG. The processor 50 smoothes and differentiates the composite waveform data U ′ (I) stored in the RAM 43 in step 1120.
This differential operation is performed by, for example, a correlation operation between a numerical filter FNb having a constant slope as shown in FIG. 13C and the synthetic waveform data U ′ (I), and FIG.
In the case of the composite waveform data U ′ (I) shown in FIG.
The differential waveform data V (I) as shown below is obtained. This data V (I) is also stored in the RAM 43.

Next, the processor 50 proceeds to step 1122
The peak (bottom) position IUX (i) considered to be the mark center is obtained by analyzing the synthesized waveform data U ′ (I). Here, i represents the number of peak positions (an integer equal to or greater than zero) found in the composite waveform data U ′ (I). At this time, the processor 50 determines the value U ′ (I) of U ′ (I) that satisfies the following two relations at the same time.
Find P (i) and IUX (i). U '(I) <U' (I-1), U '(I-2), U' (I-3) ... (4) U '(I) ≤ U' (I + 1), U '(I + 2 ), U '(I + 3) (5) This is the work of finding the bottom waveforms BW1, BW2, BW3, etc. from the composite waveform data U' (I) of FIG. This is the center position (address value) of the waveform BWn, and UP (i) is the contrast intensity (level) of each bottom waveform BWn.

Next, the processor 50 proceeds to step 1124.
Then, using the differential waveform data V (I), only one of the differential peak positions generated by the bottom waveform portion having the peak UP (i) in the composite waveform data U ′ (I) is obtained. First, the processor 50 calculates a value Vx of I that satisfies the following two relational expressions at the same time, and a differential contrast value Vy corresponding to the value Vx. V (I)> V (I−1), V (I−2) (6) V (I) ≧ V (I + 1), V (I + 2) (7) Furthermore, the processor 50 ), When a set of values Vx and Vy meeting the conditions of the expression (7) are found, the contrast value Vy exists within a certain set range, and (Vx-I
UX (i)) is also checked to see if it is within a certain setting range. That is, the check by the relational expressions (6) and (7) is performed as shown in FIG.
(I) Find the positive peak in. FIG.
(A) shows one bottom waveform BWn in the composite waveform data U ′ (I), and FIG. 15 (B) shows differential waveform data corresponding to the bottom waveform. As shown in FIG. 15B, the positive peak on the differential waveform corresponds to the position of the up slope (rising portion increasing from the bottom value) in the bottom waveform BWn. Therefore, the condition that the differential contrast value Vy exists within a certain range means that it has been checked whether or not the amplitude of the bottom waveform BWn is large and the rising edge is steep. When the above condition is satisfied, the processor 50 registers the differential contrast value Vy as VP (j) and registers the value Vx as IVX (j). In this case, the IV corresponding to IUX (i) or UP (j)
It is also conceivable that X (j) or VP (j) cannot be found. At that time, IUX found on the composite waveform data
(I), it is determined that UP (i) does not correspond to the mark, and it is excluded from the subsequent processing. Conversely, IVX
(J) or a plurality of VP (j) may be found for one IUX (i). In this case, IUX (i)
Only the IVX (j) data appearing at the position closest to is selected, and the other IVX (j) data is excluded from the subsequent processing. However, since the numbers i and j need to correspond to each other on a one-to-one basis, if the numbers i and j are excluded from the target of the subsequent processing, the numbers i or j paired with the numbers i and j
Is also excluded.

Next, in step 1126 and thereafter, the processor 50 starts a recognition operation using the design information of the three bar marks BPMn of the mark MDx. In step 1126,
A plurality of data IUX (i) selected so far,
Arbitrary three data are selected as a set from UP (i). Here, a1, a2, and a3 are three bar marks B
Assuming that 1 ≦ a1 <a2 <a3 ≦ i, corresponding to each of the PMn, the processor 50 proceeds to the next step 112
8, using the mark intervals hb and hc shown in FIG.
Whether UX (a2) -IUX (a1) is within a certain set range with respect to interval hb, and IUX (a3)-
It is determined whether the IUX (a2) is within a certain set range with respect to the interval hc. If the combination of the three data IUX (i) determined in step 1128 does not meet the mark interval condition, step 1126 is performed again.
Go back and make another combination.

Next, the processor 50 determines in step 1130 that three data IUX (a1), IUX (a2), IUX
The contrast values UP (a3) corresponding to each of UX (a3)
1) It is determined whether the difference between UP (a2) and UP (a3) is within a certain range. This is three marks B
This is because all of the PMn have the same shape and dimensions, so that the three bottom waveforms BWn should also have substantially the same shape. Here, three contrast values UP (a1), UP (a2),
When UP (a3) is not uniform, the previous step 1126 is executed.
Return to and select another combination.

If it is determined in step 1130 that the mutual intensity of the three contrast values is within a certain range, the processor 50 determines in step 1132 the three data IUX (a).
1), IUX (a2), IUX (a3), and IVX
Based on (a1), IVX (a2) and IVX (a3), a check is made on the mutual difference between the widths of the three marks. Here, the value of IVX (a1) -IUX (a1),
IVX (a2) -IUX (a2) value and IVX (a
3) Each value of -IUX (a3) is determined, and it is determined whether or not the difference between these three values is within a certain range. As described above with reference to FIG. 15, the position of IVX (i) corresponds to almost the middle of the up slope of the bottom waveform of the composite waveform data U ′ (I), and the position IUX (i) is zero on the differential waveform. Corresponds to cross points. Therefore, IVX (j) -I
The value of UX (i) corresponds to about の of the width of the mark BPMn in the scanning direction. This means that the position Vx and IUX
(I) is nothing more than checking the condition that the vehicle is approaching within a certain range.

Finally, the processor 50 proceeds to step 1134.
And the three data VP (a1), VP (a2), VP (a
3) It is determined whether or not the mutual level difference is within a certain range. Here, it is checked whether or not each of the up slopes of the selected three bottom waveforms is aligned at a predetermined rise. As described above, steps 1128, 1130, and 11
The combination of the three data that passed through all 32 and 1134 is registered in the processor 50 as one mark candidate with the weighted evaluation value in the next step 1136. The weighting here is based on the previous step 1128,
This is performed based on the condition determination in 1130 and 1132.
First, in the determination of the mark interval in step 1128, As
Are used as coefficients to set evaluation values A1 and A2, and the following equation is calculated. A1 = As (IUX (a2) -IUX (a1) -hb) ... (8) A2 = As (IUX (a3) -IUX (a2) -hc) ... (9) Further, determination of the mark width in step 1132 Then, B1 = MAX (IVX (a1) -IUX (a1), IVX (a2) -IUX (a2), IVX (a3) -IUX (a3)) -MIN (IVX (a1) -IUX (a1), IVX (a2) −IUX (a2), IVX (a3) −IUX (a3)) (10) The evaluation value B1 is calculated.

In step 1130, the processor 50 calculates α = MAX (UP (a1), UP (a2), UP (a3)) − MIN (UP (a1), UP (a2), UP (a3)) ( 11) is calculated, and the level of the up slope position and the bottom value U in the composite waveform data determined by IVX (j) are calculated.
The difference from P (i) is determined. That is, aa = U '(IVX (a1))-UP (a1) bb = U' (IVX (a2))-UP (a2) cc = U '(IVX (a3))-UP (a3) Do. Then, the following equation is calculated. G = (α × 100) / (MAX (aa, bb, cc) + MIN (aa, bb, cc)) Finally, the following equation is calculated as the evaluation function value Z. Z = (A1 + A2) ・ IMP + B11IWP + G ・ IUP (12) Here, IMP, IWP, and IUP are weighting factors, each of which is set to a value of about 0 to 4.

Further, in step 1134, an evaluation relating to the differential contrast value VP (j) is also performed. Where aa '
= VP (a1), bb '= VP (a2), cc' = VP
Given (a3), the processor 50 calculates the value β of the following equation. β = (MAX (aa ′, bb ′, cc ′) − MIN (aa ′, bb ′, cc ′)) × 100 ÷ (MAX (aa ′, bb ′, cc ′) + MIN (aa ′, bb ′, cc ')) This value β is effective as an evaluation when the contrast of the mark signal portion of the composite waveform data is low, and the evaluation function value Z is as follows. Z = (A1 + A2) ・ IMP + B1 ・ IWP + α ・ IUP + β ・ IVP (13) where IVP is also a weighting factor and takes a value of 0-4.

The above four weighting factors IMP, IWP, I
UP and IVP are set in advance according to the shape and surface state of the wafer mark to be detected. Well, processor 5
In step 1136, 0 registers three mark position candidates together with the evaluation function value Z. And the next step 1
At 138, it is checked whether or not the remaining three combinations of bottom waveforms remain.
Returning to step 6, if there is no other combination, the next step 11
At 40, it is checked whether or not a registered mark candidate exists. Here, when there is no mark candidate, the process proceeds to error processing. If there is one or more mark candidates in step 1140, in step 1142, the candidate having the smallest evaluation value Z or within a certain range is selected and determined as the position of one mark MDx.

According to the above sequence, the three bar marks BPMn as shown in FIG. 5 or FIG. 7 can be searched almost certainly. As described above, in the embodiment of the present invention, three bar marks BPMn are used for search alignment, but one bar mark may be used. Further, each bar mark does not need to be in the form of a diffraction grating, and may be a simple line having continuous edges. However, in that case, it is better that the line width is narrow to some extent. Further, detection of a mark for search alignment is disclosed in
As disclosed in Japanese Patent No. 19726, the rotation (θ) error of the entire wafer W is performed at a plurality of locations on the wafer W,
After correcting the position and rotation by measuring the translation error in the X and Y directions, the position of one mark in the X or Y direction may be measured again to correct the translation error (offset).

Even if the mark has the same shape as that of FIG. 5, the signal waveform level shown in FIG. 7B may be reversed between the mark position and the periphery depending on the state of the wafer surface. In this case, by inverting the waveform data stored in the RAM 43, the waveform at the mark position can be made to have a bottom shape, so that the algorithm described in this embodiment can be applied as it is.

FIG. 16A shows a modification of the search mark, in which wide bar marks BPMa, BPMc and narrow bar mark BPMb are arranged in the direction of the scanning line SL at a constant interval. Is the case. In addition, bar mark BPM
It is assumed that b is located between the bar marks BPMa and BPMc. Since the width of the narrow bar mark BPMb is equal to the width of the three marks BPMn, a single bottom waveform C is formed on the image signal as shown in FIG. However, the bar marks BPMa and BPMc on both sides have bottom waveforms A and B or D and E at the edge positions on both sides.
When such a waveform is obtained, in the search alignment of the present embodiment, the center of the center bar mark BPMb is treated as an original mark for fine alignment, and the marks BPMa and BPMc on both sides thereof are treated for recognition. It is good to decide.

At the time of search alignment, if three bottom waveforms are selected from the five bottom waveforms A to E, the algorithm of this embodiment can be applied as it is. 3
It is preferable that the two bottom waveforms are selected at different intervals. Examples of the combination are listed below: bottom waveform [A, B, C], bottom waveform [A, B, D], bottom waveform [A, B, E], bottom waveform [B, C, E], bottom waveform Waveforms [C, D, E], bottom waveforms [B, D, E],
Bottom waveforms [A, D, E]. The interval between the bottom waveforms in any one of these combinations (hb,
(corresponding to hc) is set in the processor 50, so that processing can be performed with the same algorithm.

Next, the weighting method will be described with reference to FIG.
An example will be described with reference to FIG. FIG. 17A shows an example of the composite waveform data U ′ (I). The two bottom waveforms BWa and BWb appearing here are, for example, marks BP in FIG. 16A.
It is likely to appear when the interval between the left and right edges of Ma is slightly narrower than in the case of FIG. Among these bottom waveforms BWa and BWb, BWa corresponds to the left edge of the mark BPMa and BWb corresponds to the right edge. The bottom waveform is differentiated to obtain differentiated waveform data V as shown in FIG. When (I) is obtained, the differential contrast value VPa corresponding to the up slope of the bottom waveform BWa is small,
The differential contrast value VPb corresponding to the up slope of the bottom waveform BWb increases. Therefore, equation (1)
Since the evaluations by 1) and (12) may be inappropriate, the evaluation by the expression (13) is more preferable, and the weighting factor I
VP must be zero.

Other weighting factors IMP, IWP, IUP
It is better to change appropriately according to the signal waveform. For example, in the case of a very noisy signal waveform, the coefficient IMP
Should be set to the most important (3 to 4) and the coefficient IVP should be set to 0. Next, another method of search alignment will be described with reference to FIG. FIG. 18 shows an example of an image signal waveform when an image is taken to search for a street line itself formed on the wafer W as a kind of mark. In FIG. 18, the scanning direction is set to be orthogonal to the direction in which the street line extends. Usually, in a street line having no pattern such as an alignment mark, a feature appearing on a signal waveform is the boundary portion ASM between the street line and a circuit pattern (chip) region. The center of the street line is the CCL, and the image signal waveform often has good symmetry about the center CCL. In the case of such a waveform, the inner bottom waveforms BSb and BSc near the center CCL have a substantially symmetrical shape, and the bottom waveforms BSa and BSa located outside the bottom waveforms BSb and BSc.
BSd also has good symmetry. Therefore, the bottom value UP
When checking the magnitudes of (i) and the differential contrast value VP (j), it is desirable to compare using a combination of the bottom waveforms BSa and BSd or a combination of the bottom waveforms BSb and BSc. In this case, the positive peak on the differential waveform of the bottom waveform BSb is compared with the negative peak on the differential waveform of the bottom waveform BSc, or the bottom waveform BSa
Is compared with a positive peak on the differential waveform of the bottom waveform BSd. Thus, depending on the shape of the mark (synthetic waveform), the positive peak (up slope) or the negative peak (down slope) of the differential peak
It is desirable to use differently.

[0064]

As described above, according to the present invention, the image signal splicing process is performed by comparing the contrast at the overlapping portion, so that the signal waveform at the splicing portion is distorted and the image signal for each screen is distorted. Without being affected by differences in contrast,
There is an effect that it can be used for subsequent waveform processing. Also, since only the slope position of at least one of the bottom (or peak) waveforms corresponding to the marks is used, the processing time is shortened and the memory is reduced.

[Brief description of the drawings]

FIG. 1 is a perspective view showing a configuration of a projection exposure apparatus to which an alignment apparatus according to an embodiment of the present invention is applied.

FIG. 2 is a block diagram illustrating a configuration of a signal processing system in the alignment apparatus of the present embodiment.

FIG. 3 is a plan view showing the arrangement of alignment marks and shot (chip) regions on a wafer.

FIG. 4 is a plan view showing an arrangement relationship between an image field and an imaging range of an alignment system.

FIG. 5A is a plan view showing an arrangement of alignment marks in a street line on a wafer. (B) It is an enlarged view of an alignment mark.

FIG. 6 is a plan view showing a mark arrangement of a conjugate index plate in the alignment optical system.

FIGS. 7A and 7B are diagrams illustrating an example of a state of an index plate when capturing a wafer mark and an example of an image signal waveform obtained by capturing the image.

8A, 8B, 8C, 8D, and 8E are waveform diagrams showing how image signals obtained for a plurality of screens are spliced.

FIGS. 9A and 9B are diagrams showing an example of a waveform of an overlapping portion at the time of joining signals.

FIG. 10 is a flowchart when the splicing of signals is executed by software.

FIGS. 11A and 11B are diagrams showing a combined signal waveform and its differential waveform.

FIG. 12 is a flowchart showing an operation at the start of search alignment.

FIG. 13 is a diagram illustrating an example of a numerical filter that performs smoothing (smoothing) and differentiation of a signal waveform.

FIG. 14 is a diagram showing an algorithm for analyzing a synthesized signal waveform to determine a mark position.

FIG. 15 is a waveform chart for explaining application of an algorithm.

FIG. 16 is a diagram showing another example of a mark shape which can be similarly detected by the algorithm according to the embodiment of the present invention.

FIG. 17 is a waveform diagram for explaining a peak break in a signal waveform.

FIG. 18 is a waveform chart showing an example of an image signal when a street line area is imaged.

[Description of Signs of Main Parts]

R: Reticle 1, W: Wafer ST: Wafer stage, IFX, IFY: Interferometer MDx, MDy: Wafer alignment mark BPMn: Bar mark, VPAx: Image pickup area SL: Horizontal scanning line, BWn: Bottom waveform WD: Overlapping part, U '(I): Synthetic waveform data V (I): Differential waveform data 1: Projection lens, 2: CCD 4: Conjugate index plate, 8: Light source 11, Objective lens, 42: ADC 43: RAM, 50: Processor

Claims (14)

(57) [Claims] 1. A stage movable two-dimensionally while holding a substrate having an alignment mark formed with an optical or geometrical difference with respect to a surface, and a position of the stage. And a lighting system for illuminating a local area of a size that can include the mark on the substrate, and an image signal corresponding to a change in brightness in the local area by imaging the local area. And a mark position detecting means for detecting a position of the mark within an image pickup range based on the image signal, and the mark position detected by the mark position detecting means and the position measuring device. In the apparatus for aligning the substrate with respect to a predetermined reference position by moving the stage based on the measurement value, the mark position detection unit may be configured to detect the mark position by the imaging unit. A storage circuit for storing first waveform data of the image signal obtained by one-dimensional scanning in a local area; and moving the stage in the direction of the one-dimensional scanning so as to be shifted by a predetermined distance smaller than the imaging range. A shift movement control circuit for controlling; and after the shift movement, comparing second waveform data of the image signal obtained by one-dimensional scanning of the imaging means with first waveform data stored in the storage circuit. And a waveform synthesizing circuit for correcting the waveforms at the mutually overlapping portions so that they substantially coincide with each other, and then splicing the first and second waveform data to form one synthesized waveform data; An edge position extraction circuit for extracting a position of a waveform portion equivalent to a waveform that changes with specific characteristics according to at least one edge of the mark, based on A selection circuit that selects an edge position that satisfies a design condition from among the edge positions, and specifies the mark position based on the selected edge position. 2. An image of an alignment mark on a substrate is taken by using a predetermined image pickup device, and the position of the alignment mark is detected based on image data output from the image pickup device to perform the alignment of the substrate. In the alignment method, a step of imaging the first region on the substrate by the imaging device to obtain first image data; and relatively shifting the substrate and an imaging range of the imaging device in a predetermined direction. Capturing a second area on the substrate by the imaging device to obtain second image data; and joining the first image data and the second image data to form one composite image data And detecting a position of the alignment mark in the predetermined direction based on the composite image data. Alignment wherein the alignment is performed of the substrate Zui. 3. The alignment method according to claim 2, wherein a part of the first area and the second area overlap in the predetermined direction. 4. The image processing apparatus according to claim 1, wherein, of the first image data and the second image data, image data of the overlapping area is deleted from one of the image data and joined to the other image data. 4. The alignment method according to claim 3, wherein image data is created. 5. The composition by joining the center of the image data of the overlapping area in the first image data and the center of the image data of the overlapping area in the second image data. 4. The alignment method according to claim 3, wherein image data is created. 6. The first image data and the second image data are interpolated with each other and the first image data and the second image data are interpolated with each other. 4. The alignment method according to claim 3, wherein the combined image data is created by matching. 7. A step of extracting a waveform equivalent to a waveform that changes with predetermined characteristics corresponding to an edge of the alignment mark from synthesized waveform data of the synthesized image data in the predetermined direction; Selecting a waveform that satisfies a predetermined condition as a waveform corresponding to the edge of the alignment mark from among them; and obtaining edge position information of the alignment mark from the selected waveform; based on the edge position information 3. The alignment method according to claim 2, wherein the alignment of the substrate is performed. 8. A waveform corresponding to the alignment mark is selected from the extracted waveforms based on a mutual interval between the extracted waveforms.
4. The alignment method according to 1. 9. The alignment method according to claim 7, wherein a waveform corresponding to the alignment mark is selected from the extracted waveforms based on a contrast of the extracted waveform. 10. A value corresponding to the line width of the alignment mark is obtained from the extracted waveform, and a waveform corresponding to the alignment mark is selected from the extracted waveforms based on the line width corresponding value. The alignment method according to claim 7, wherein: 11. The alignment method according to claim 7, wherein a waveform corresponding to the alignment mark is selected from the extracted waveforms based on a slope of a slope portion of the extracted waveform. . 12. An evaluation value corresponding to each of the plurality of conditions set as the predetermined condition is obtained for each of the extracted waveforms, and the alignment is selected from the extracted waveforms based on the weighted evaluation value. The alignment method according to claim 7, wherein a waveform corresponding to the mark is selected. 13. The plurality of conditions include: a distance between the extracted waveforms; a contrast of the extracted waveform; a line width value of a mark obtained from the extracted waveform; 13. The alignment method according to claim 12, further comprising a slope of a slope portion. 14. An edge position information of the alignment mark is obtained based on synthesized waveform data of the synthesized image data in the predetermined direction and differentiated waveform data obtained by differentiating the synthesized waveform data. 3. The alignment method according to claim 2, wherein the alignment of the substrate is performed based on the information.