JP3003694B2 - Projection exposure equipment - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an exposure apparatus used for manufacturing semiconductor elements, liquid crystal elements and the like, and more particularly to an apparatus for aligning a substrate to be exposed in a step-and-repeat (or step-and-scan) type exposure apparatus. is there.

[0002]

2. Description of the Related Art In an exposure apparatus of this type, a circuit pattern drawn on a master called a mask or a reticle is printed on a resist layer of a semiconductor wafer and developed to form a resist pattern of a desired circuit. ing. In general, in the manufacture of semiconductor devices, since several to several tens of circuit patterns are superimposed, it is necessary to accurately superimpose a circuit pattern already formed on a wafer and an optical image of a circuit pattern to be exposed. is there. Various devices necessary for the superposition are called a positioning device or an alignment device. This alignment apparatus is indispensable to an exposure apparatus that performs the above-described overlay exposure, and has been improved in recent years so that high-speed processing can be performed with higher precision. In addition, it can be said that the alignment apparatus is roughly composed of three element technologies. One is an alignment optical system for optically detecting an alignment mark formed in advance on the wafer and obtaining a photoelectric signal corresponding to the profile of the mark, and the other two are optical signals for the alignment. A signal processing system which electrically processes the alignment mark with respect to the original position by using an appropriate algorithm, and precisely positions the position of the wafer or the position of the mask (or reticle) in accordance with the obtained shift amount. This is a positioning mechanism for correcting.

In recent years, these three element technologies, namely, an alignment optical system as an optical technology, a signal processing system as an electronic technology (information processing technology), and a positioning mechanism as a precision machine technology have been organically and highly dimensional. Combined reduced projection exposure apparatuses (steppers) have come to be widely used in semiconductor manufacturing plants. This stepper focuses and projects an optical image of a reticle circuit pattern onto a partial area (called a shot area) on a wafer by a high-resolution projection lens (numerical aperture 0.35 to 0.5). Since the number of exposure shots is, for example, about 15 × 15 mm square,
The stage on which the wafer is mounted is stepped two-dimensionally in the x and y directions to expose the entire surface of the wafer. In the case of such a stepper, there are roughly three types of practical forms of the alignment optical system incorporated therein. The first method is a TTR (Thro) for simultaneously observing (or detecting) an alignment mark formed on a reticle and an alignment mark on a wafer via a projection lens.
The second method is a TTL (Through The Lens) method that detects only an alignment mark on a wafer via a projection lens without detecting an alignment mark of a reticle at all.
s) The third method is an off-axis (Off-Axis) method in which only an alignment mark on a wafer is detected through a microscope objective lens separately provided from the projection lens by a certain distance.

In the above-mentioned TTR system and TTL system, since the mark on the wafer is detected via a projection lens, the irradiation light for illuminating the wafer mark is either a coherent laser beam (single wavelength) or g of a mercury lamp used for exposure. Line or i
It was limited to the line spectrum (quasi-monochromatic). This is due to the chromatic aberration of the projection lens, and the exposure light (g-line or i-line)
This is because the aberration is designed to be the best with respect to the line (1). Such a TTR system is disclosed in, for example, JP-A-57-138134 and JP-A-57-14261.
And the TTL system is disclosed in, for example,
No. 0-130742, JP-A-61-44429 and JP-A-61-128106.

On the other hand, in the Off-Axis method, since there is no limitation by the projection lens as described above, the illumination light of the wafer mark may be any light, but the light which does not expose the resist layer on the wafer may be used. desirable. Such an Off-Axis method is disclosed in, for example,
No. 2823, JP-A-57-19726 and the like.

In general, in the TTR system and the TTL system, a mark on a wafer is detected via a projection lens. Therefore, a mark attached to an arbitrary shot area of a shot arrangement on a wafer can be relatively freely moved by moving a wafer stage. Can be detected. In contrast, in the Off-Axis method,
Due to the movement stroke of the wafer stage, it has been usual to detect only marks (for example, two or three marks) attached to a predetermined shot area on the wafer. Therefore, in order to freely detect a mark in an arbitrary shot area on a wafer while taking advantage of the illumination light of the Off-Axis type alignment optical system, the movement stroke of the wafer stage may be increased. . In a general stepper, since the coordinate position of the wafer stage is measured by a laser light wave interferometer (interferometer), an increase in the moving stroke is caused by a large moving mirror (optical square) fixed on the wafer stage. And the size of the base on which the wafer stage is mounted is increased.
The advantage of improving the mark detection accuracy obtained by the free illumination form of the ff-Axis system is insignificant.

[0007]

SUMMARY OF THE INVENTION As described above, Off
In the Axis method, if the movement stroke of the wafer stage is increased, a mark attached to an arbitrary shot area on the wafer can be detected.
-Mark detection may take time depending on the arrangement of the Axis type alignment optical system and its illumination form. Since the Off-Axis type alignment optical system is far away from the pattern projection area of the projection lens, the wafer stage is excessively moved by that distance, that is, the reduction in throughput caused by the arrangement is essentially avoided. Therefore, how to shorten the mark detection time is a problem.
As mentioned above, the method of relatively scanning the mark on the wafer with the spot light (slit shape) of the focused laser beam and photoelectrically detecting the scattered and diffracted light from the edge of the mark can minimize the mark detection time. . According to this method, marks can be detected with a practically sufficient S / N ratio due to the high brightness of the laser beam. One major problem is that since the laser beam is generally a single wavelength, the entire surface of the wafer is 1 to 5 μm.
The resist layer covering with a thickness of m causes phenomena such as thin film interference (multiple interference), which causes scattering from the mark edge and unexpected distortion in the light amount distribution of the diffracted light (synonymous with waveform distortion on the photoelectric signal). Is to give. This distortion of the light amount distribution does not always occur in any wafer, and differs depending on the thickness and type of the wafer base and the resist layer. Even when the distortion occurs, the distortion method generally changes in various ways. . Accordingly, such a single-wavelength laser beam or a quasi-monochromatic (emission line spectrum of a mercury lamp or the like) is used to generate the laser beam.
Simply adopting it as mark illumination light (spot light or uniform illumination light) for the f-Axis type alignment optical system has no advantage over the prior art, and lowers throughput due to unnecessary movement of the wafer stage. Only the inconvenience such as remains.

When light of a broad wavelength distribution having a certain bandwidth (about 200 nm) is used as mark illumination light of the Off-Axis method, the adverse effect of the resist layer is drastically reduced, and mark detection accuracy is improved. Therefore, the Off-Axis type alignment optical system
If each of the marks attached to a plurality of shot areas at arbitrary positions on the wafer is detected under broadband illumination light, highly accurate alignment can be expected.

However, as described above, even in the case of a single wavelength or a quasi-monochromatic color, the distribution of the light amount from the mark edge is not always distorted. Considering, for example, that the influence of the mark position detection algorithm can be reduced, changing all the mark detection to the Off-Axis type alignment optical system of the broadband illumination requires a wafer having no distortion or a distortion. This means that even for less affected wafers, the throughput must be reduced,
Hana is irrational.

Accordingly, an object of the present invention is to provide a positioning apparatus capable of detecting a mark position with high accuracy while minimizing a decrease in throughput.

[0011]

In the present invention, two types of alignment sensors for detecting alignment marks on a substrate via a resist layer, namely, a first mark position detecting means and a second mark position detecting means, are provided. Is provided. First
Mark position detecting means is a single wavelength or quasi-monochromatic first mark.
And a first position information (AP1, AP3) relating to the position of the first mark.
, AP4, etc.).

The second mark position detecting means includes a light transmitting system for irradiating the second mark on the substrate with the second light having a wavelength distribution wider than the first light, and the second position with respect to the position of the second mark. The information (AP2) is output. Here, the first mark and the second mark may share the same mark pattern on the substrate, or may be a dedicated mark pattern separated by a predetermined distance in advance.

[0013] A position determining means (eg, EGA operation unit 502) for determining the overall position information of the substrate by using both the first position information and the second position information is provided. As a result, the region to be exposed on the substrate and the pattern region on the mask are relatively aligned.

According to the present invention, when a mark on a substrate such as a wafer is detected through a resist layer, light of a single wavelength or quasi-monochromatic light is applied, and reflected light (scattered light, scattered light, Diffracted light, specular reflected light)
Paying attention to the fact that the mark is slightly distorted due to the optical distortion, the mark detection is performed by using the illumination light of the broadband wavelength which is hardly affected by the interference. However, performing all mark detection on the wafer with an alignment sensor having illumination light of such a broadband wavelength improves the accuracy of mark detection, but is generally disadvantageous in terms of throughput. In particular, in a projection exposure apparatus, when detecting a wafer mark via a projection optical system, illumination light of a wide wavelength band cannot be used due to chromatic aberration, and an alignment sensor (Off-Axis type) separately provided outside the projection optical system. I have to become. For this reason, movement of the wafer or the like for performing mark detection by the Off-Axis alignment sensor is indispensable, and the throughput is reduced accordingly. In the present invention,
Since the second mark position detecting means is used only for some of a plurality of marks to be detected on the wafer, a decrease in throughput is suppressed to a small extent.

[0015]

FIG. 1 shows a configuration of a projection type exposure apparatus according to an embodiment of the present invention and a configuration of a positioning (alignment) apparatus incorporated therein. Illumination light for exposure (g-line or i-line from a mercury lamp or ultraviolet pulse light from an excimer laser light source) IL irradiates the pattern area PA of the reticle R with a uniform illuminance distribution via the condenser lens CL. The illumination light IL passes through the pattern area PA and enters, for example, both sides (or one side) of the telecentric projection lens PL and reaches the wafer W. Here, the aberration of the projection lens PL is corrected for the wavelength of the illumination light IL, and the reticle R and the wafer W are conjugate to each other under the wavelength. The illumination light IL is Koehler illumination, and is formed as a light source image at the center of the pupil EP of the projection lens PL. The reticle R is held on a reticle stage RS which can be finely moved two-dimensionally, and the reticle R is detected by a reticle alignment system formed of a mirror 16, an objective lens 17, and a mark detection system 18 around the reticle alignment mark. As a result, the projection lens PL is positioned with respect to the optical axis AX. On the other hand, the wafer W is placed on the wafer stage ST that moves two-dimensionally by the drive system 13, and the coordinate values of the wafer stage ST are sequentially measured by the interferometer 12. The stage controller 14 controls the movement and positioning of the wafer stage ST by controlling the drive system 13 based on coordinate measurement values and the like from the interferometer 12. On the wafer stage ST, a reference mark FM used for a baseline measurement or the like described later is provided.

In this embodiment, a TTL type alignment optical system serving as first mark position detecting means is provided. The beam LB from the laser light source 1 is He-
Red light such as a Ne laser, which is insensitive to the resist layer on the wafer W. This beam LB passes through a beam shaping optical system 2 including a cylindrical lens and the like, and a mirror 3
a, the light enters the objective lens 6 via the lens system 4, the mirror 3 b, and the beam splitter 5. The beam LB emitted from the objective lens 6 is reflected by a mirror 7 inclined at 45 ° below the reticle R, and enters the periphery of the field of view of the projection lens PL in parallel with the optical axis AX. Then, the beam LB irradiates the wafer W vertically through the center of the pupil EP of the projection lens PL. Here, the beam LB is condensed as slit-like spot light SP0 in the space in the optical path between the objective lens 6 and the projection lens PL by the function of the beam shaping optical system 2. Then, the projection lens PL re-images the spot light SP0 on the wafer W as a spot SP. The mirror 7 is fixed outside the periphery of the pattern area PA of the reticle R and within the field of view of the projection lens PL. Therefore, the slit-shaped spot light SP formed on the wafer W is located outside the projected image of the pattern area PA.
In order to detect a mark on the wafer W using the spot light SP, the wafer stage ST is moved horizontally with respect to the spot light SP. When the spot light SP relatively scans the mark, regular reflected light, scattered light, diffracted light, and the like are generated from the mark, and the light amount changes depending on the relative position between the mark and the spot light SP. The light information travels backward along the light transmission path of the beam LB, and is reflected by the projection lens PL, the mirror 7, the objective lens 6, and the beam splitter 5, and the light receiving element 8
Reach The light receiving surface of the light receiving element 8 is the pupil E of the projection lens PL.
It is located on a plane EP ′ substantially conjugate to P, has an insensitive area for specularly reflected light from the mark, and receives only scattered light and diffracted light.

FIG. 2B shows the distribution of optical information from the mark on the wafer W on the pupil EP (or pupil image plane EP '). Above and below (in the y direction) the specularly reflected light D0 extending in the x direction at the center of the pupil EP in the x direction, a positive first-order diffracted light + D1, a second-order diffracted light + D2, and a negative first-order diffracted light -D1, respectively.
, The second-order diffracted light -D2 is arranged, and the right and left (x
Direction), scattered light Dr from the mark edge is located.
Since this is described in detail in the above-mentioned Japanese Patent Application Laid-Open No. 61-128106, the detailed description is omitted, but the diffracted light ± D1 and ± D2 are generated only when the mark is a diffraction grating mark. Therefore, as shown in FIG. 2C, the light receiving element 8 has four independent light receiving surfaces 8a and 8a within the pupil image plane EP '.
b, 8c, 8d, the light receiving surfaces 8a, 8b receive the scattered light Dr, and the light receiving surfaces 8c, 8d receive the diffracted light ± D1, ± D
2 are arranged to receive light. When the numerical aperture (NA) of the projection lens PL on the wafer side is large and the third-order diffracted light generated from the diffraction grating mark also passes through the pupil EP, the light receiving surfaces 8c and 8d receive the three-dimensional light. Good size. Each photoelectric signal from the light receiving element 8 together with the position measurement signal PDS from the interferometer 12 is
(Laser Step Alignment) The information is input to the arithmetic unit 9 and the mark position information AP1 is created. The LSA calculation unit 9 samples and stores the photoelectric signal waveform from the light receiving element 8 when the wafer mark is scanned with respect to the spot light SP based on the position measurement signal PDS, and analyzes the waveform to analyze the mark to obtain the mark. Information AP1 is output as the coordinate position of the wafer stage ST when the center coincides with the center of the spot light.

In the above, the laser light source 1, beam shaping optical system 2, mirrors 3a and 3b, lens system 4, beam splitter 5, objective lens 6, mirror 7, light receiving element 8, LS
The A operation unit 9 and the projection lens PL constitute first mark position detection means for the wafer W. FIG. 2
The solid line shown in the optical path of the TTL type alignment optical system in the middle represents the image forming relationship with the wafer W, and the broken line is the pupil EP
Represents a conjugate relationship with

Next, an Off-Axis type alignment system as a second mark position detecting means will be described. Light generated from the halogen lamp 20 is condensed on one end surface of the optical fiber 22 by the condenser lens 21. The light passing through the fiber 22 passes through a filter 23 that cuts the photosensitive wavelength (short wavelength) region of the resist layer and the infrared wavelength region, and reaches a half mirror 25 via a lens system 24. The illumination light reflected here is reflected on mirror 2
After being substantially horizontally reflected at 6, the light enters the objective lens 27, and is further reflected by a prism (mirror) 28 fixed around the lower part of the barrel of the projection lens PL so as not to block the field of view of the projection lens PL. The wafer W is irradiated vertically. Although not shown here, an appropriate illumination field stop is provided at a position conjugate with the wafer W with respect to the objective lens 27 in the optical path from the exit end of the fiber 22 to the objective lens 27. The objective lens 27 is a telecentric system, and an image of the exit end of the fiber 22 is formed on a surface 27a of the aperture stop (same as the pupil), and Koehler illumination is performed. The optical axis of the objective lens 27 is set to be vertical on the wafer W, so that the mark position does not shift due to the tilt of the optical axis when detecting the mark.

The reflected light from the wafer W passes through the objective lens 28 and the half mirror 25, and is imaged on the index plate 30 by the lens system 29. The index plate 30 is arranged conjugate with the wafer W by the objective lens 27 and the lens system 29, and linearly extends in a rectangular transparent window extending in the x and y directions, respectively, as shown in FIG. Index mark 30a, 3
0b, 30c, and 30d. Therefore, the image of the mark of the wafer W is formed in the transparent window of the index plate 30, and this wafer mark image and the index marks 30a, 30b, 30c, 30
d means CC via relay systems 31 and 33 and mirror 32
An image is formed on an image sensor 34 such as a D camera. The video signal from the image sensor 34 is input to an FIA (field image alignment) operation unit 35 together with the position measurement signal PDS from the interferometer 12. The FIA operation unit 35 determines the shift of the mark image with respect to the index marks 30a to 30d based on the waveform of the video signal, and calculates the image of the wafer mark from the stop position of the wafer stage ST represented by the position measurement signal PDS. 30d
The information AP2 about the mark center detection position of the wafer stage ST when it is accurately located at the center is output.

With the above configuration, the illumination light of the wafer W passing through the filter 23 illuminates a local area (smaller than a shot area) including a mark on the wafer W with almost uniform illuminance, and has a wavelength range of about 200 nm. Is determined by the width. the above,
FIA operation unit 3 from halogen lamp 20 in code order
The members up to 5 form a second mark position detecting means. Further, since light having a wavelength band of about 200 nm passes through the telecentric imaging optical system including the objective lens 27, the lens system 29, and the relay systems 31 and 33, it is necessary to correct chromatic aberration corresponding to the light. For this purpose, the achromatic technique of the microscope lens may be used as it is.

Further, the numerical aperture (NA) of the objective lens 27 on the wafer side is preferably smaller than the numerical aperture of the projection lens PL. This is the objective lens 27, lens system 29
Is relatively large, and chromatic aberration correction is also required. Therefore, the numerical aperture of the objective lens 27 is equal to or larger than the numerical aperture of the current general projection lens, about 0.45 to 0.5. In this case, even if a certain working distance (interval to the wafer surface) of the objective lens 27 is ensured, it is advantageous in that the diameter of the objective lens can be prevented from becoming large. In this embodiment, the prism 28
The observation field of view of the objective lens 27 is partially cut into the lower surface of the lens barrel of the projection lens PL so as to be as close as possible to the field of view of the projection lens PL. In general, this type of stepper includes a focus sensor that accurately detects the distance (deviation) between the image plane of the projection lens PL and the surface of the wafer W, the surface of a shot area on the wafer W, and the image formation of the projection lens PL. And a leveling sensor for detecting a relative inclination with respect to the surface.
The focus sensor and the leveling sensor irradiate a light beam in the infrared region from obliquely onto the wafer W on which the projection field of view of the projection lens PL exists, and determine the shift of the light receiving position of the reflected light to perform focusing and leveling. It is configured. Therefore, when observing the mark on the wafer W via the objective lens 27 and considering the throughput, when the mark on the wafer W is brought into the observation visual field of the objective lens 27, the focus sensor is operated to adjust the focus ( It is preferable to perform vertical movement of the Z stage incorporated in the wafer stage ST. However, since the region where the focus sensor detects the surface of the wafer W is displaced from the observation field of view of the objective lens 27,
If the wafer W has minute unevenness or warpage between the two lenses, accurate focusing cannot be performed on the objective lens 27 having a large numerical aperture. However, the numerical aperture of the objective lens 27 is set to one of the numerical aperture of the projection lens.
If it is set to about ２〜 to ／, the depth of focus in practical use becomes large, and the mark can be observed almost without being affected by minute unevenness or warpage of the wafer. Further, when the numerical aperture is reduced, even if the perpendicularity of the optical axis of the objective lens 27 to the wafer W, that is, the so-called telecentricity, is slightly deteriorated, the partial position (for example, the center and the four corners) in the observation field of view is reduced. It is also advantageous that the image quality does not change abruptly in each case.

In the configuration shown in FIG. 1, the TTL alignment system (1, 2, 3a, 3b, 4, 5, 6, 7, 8) is used.
Although only one set is shown, another set is provided in a direction orthogonal to the paper surface, and a similar spot light is formed in the projection image plane. Extension lines in the longitudinal direction of these two spot lights are directed to the optical axis AX. In the configuration shown in FIG.
Since the detection center (center of the index plate 30) of the f-Axis type alignment system is far from the center of the projection lens, a straight line connecting the measurement position of the interferometer 12 and the center of the projection lens, that is, a length measurement axis (center of the length measurement beam) Line) minimizes Abbe error (off-axis error due to stage tilt). Although only one set is shown here,
For example, as shown in JP-A-56-102823, one set of alignment systems is provided on each of the X measurement axis and the Y measurement axis.

However, when the alignment system cannot be provided on each of the measurement axes due to the problem of the arrangement of the optical system, or when observing the marks in both xy directions with one set of the optical system, the Abbe error is large and the accuracy is sufficient. Can not be obtained. Therefore, in such a configuration, a stage tilt angle measurement sensor (yawing sensor) for measuring the tilt (small rotation in the xy plane) of the stage ST is attached to the stage, and Abbe error at the time of alignment is removed by correction calculation. Means may be used. That is, the mark detection position in the direction including the Abbe error may be corrected based on the yaw amount of the stage ST. When performing alignment, it is necessary to perform global alignment for wafer combination and fine alignment for high-accuracy alignment. This global alignment is described in, for example, Japanese Patent Application Laid-Open No. 60-13074.
As disclosed in Japanese Unexamined Patent Application Publication No. 2 (KOKAI), there is a method of mixing a TTL alignment system and an Off-Axis alignment system. In the apparatus of the present embodiment, the TTL that normally has a fast processing speed is used.
A sequence for performing global alignment by detecting three or two marks on the wafer by the alignment system is adopted. However, alignment may not be performed properly depending on the thickness and type of the wafer underlayer and the resist layer (especially when mark detection is not successful). Therefore, global alignment is performed using an alignment system using a broad-band illumination wavelength of the Off-Axis system. Means for switching the sequence so as to execute the alignment are provided. In this case, the sequence is switched by determining the mark detection time and the magnitude and distortion of the mark detection signal when performing global alignment by the TTL type alignment system.

Next, an alignment system of the TTL system, Of
The f-Axis type alignment system and the main control system 50 that performs overall control of the stage controller 14 and the like will be described. The main control system 50 is provided with position information PD from the interferometer 12.
It is assumed that S is always input. Alignment (AL
G) The data storage unit 501 can input both the mark position information AP1 from the LSA operation unit 9 and the mark position information AP2 from the FIA operation unit 35.

EGA (Enhancement Global
The alignment unit 502 calculates the position of each mark based on the mark position information stored in the ALG data storage unit 501.
The actual coordinate value of the shot array on the wafer is calculated by a statistical calculation method, and the calculation result is sent to the sequence controller 506. For details, see JP-A-61-4
It is disclosed in Japanese Patent No. 4429.

Exposure (EXP) shot map data section 5
Numeral 03 stores design values of shot array coordinate values to be exposed on the wafer, and these design values are stored in the EGA operation unit 502.
Is sent to the sequence controller 506. An alignment (ALG) shot map data unit 504 stores shot array coordinate values (design values) to be aligned on the wafer, and these coordinate values are stored in the EGA operation unit 502.
Is sent to the sequence controller 506. The correction data section 505 stores various data for alignment, data for correcting positioning for an exposure shot, and the like. These correction data are stored in the ALG data storage section 5.
01 and the sequence controller 506. The sequence controller 506, based on each of the above data,
A series of procedures for controlling the movement of the wafer stage ST at the time of alignment or at the time of exposure by the step-and-repeat method is determined.

The operation of the main control system 50 will be described later in detail. FIG. 3 shows one shot area Sn on the wafer W and the alignment marks MXn, M on the wafer.
FIG. 9 is a diagram showing an arrangement relationship with Yn, and shows one shot region Sn;
Are surrounded by a scribe line SCL, and marks MXn and MYn are formed at respective central portions of two orthogonal sides of the scribe line SCL. SC is the center point of the shot area Sn and the optical axis AX of the projection lens PL passes during exposure. Each of the marks MXn and MYn is the center SC
Is located on lines CX and CY extending in the x direction and the y direction, respectively, with respect to the origin. The mark MXn is used for detecting the position in the x direction, and the mark MYn is used for detecting the position in the y direction.

FIG. 4A is an enlarged view of the mark MXn, and includes five linear patterns P1, P2, P extending in the y direction.
3, P4 and P5 are arranged at a substantially constant pitch in the x direction. FIG. 4B shows a cross-sectional structure of the mark MXn in the x direction. Here, five linear patterns P1 to P5 are formed in a convex shape protruding from the base of the wafer W, and the upper surface thereof is formed by a resist layer PR. Coated. As shown in FIG. 3, a line CX parallel to the y-axis passing through the center SC of the shot area Sn passes through the center of the width of the central linear pattern P3 of the mark MXn. The same applies to the mark MYn. The mark MYn is composed of five linear patterns, and the center line of the central linear pattern coincides with the line CY.

In this embodiment, such marks MXn, MXn,
MYn and TTL alignment system and Off-Axi
Detection is common to the s-type alignment system. FIG.
5A shows a state of the mark MXn detected by the imaging element 34 of the Off-Axis alignment system, and FIG. 5B shows a waveform of an image signal at that time. As shown in FIG. 5A, the mark MXn to be detected is positioned between the index marks 30a and 30b on the index plate 30,
The precise position XA of the wafer stage ST at that time is obtained in advance. The image sensor 34 electrically scans the images of the five linear patterns P1 to P5 of the mark MXn and the index marks 30a and 30b along the scanning line SL. At this time, for example, since only one scanning line is disadvantageous in terms of the S / N ratio, the level of the image signal obtained by a plurality of horizontal scanning lines entering the video sampling area VSA indicated by the broken line is determined for each pixel in the horizontal direction. It is better to perform averaging on. As shown in FIG. 5B, the image signal has rising and falling waveform portions corresponding to the index marks 30a and 30b on both sides, and these positions (positions on pixels) XR1 and XR2 are obtained. , The midpoint position XR0 is obtained.

On the other hand, since the photographing element 34 photoelectrically detects the bright field image of the mark MXn, the five linear patterns P1
At each of the left and right step edges P5 to P5, the light returning to the objective lens 27 is extremely reduced due to light scattering. For this reason,
The left and right edges of the linear patterns P1 to P5 are imaged as black lines. Therefore, the waveform on the image signal has a bottom BL1 at a position corresponding to the left edge and the right edge.
, BR1,..., BL5, BR5.

The FIA calculation unit 35 calculates the position Xm in the x direction of the center (line CX) of the mark MXn (patterns P1 to P5) based on such a waveform. More specifically, the FIA operation unit 35 uses the pattern P
After calculating the center position of each of 1 to P5 based on the left and right edge positions (bottom BLn, BRn), the respective positions of the five linear patterns P1 to P5 are added and divided by 5 to obtain the center. The mark position in the x direction that should be obtained is detected.

Then, the FIA operation unit 35 calculates the difference ΔX between the position XR0 obtained previously and the mark measurement position Xm.
= XR0 -Xm, and outputs a value obtained by adding the position XA when the wafer stage ST is positioned and the difference .DELTA.X as the mark position information AP2. FIG. 1 shows only one image sensor 34. In a solid-state imaging device such as a CCD, it is possible to have the same pixel density in both the horizontal scanning direction and the vertical scanning direction. However, it is generally easy to capture and process image signals in the horizontal scanning direction. Therefore, as shown in FIG. 6, the optical path from the relay lens 31 after the index plate 30 to the image pickup device is divided into two by a beam splitter or a switching (movable) mirror 32 ′.
Providing relay lenses 33x and 33y in each optical path,
An image sensor 34x for imaging the mark MXn together with the index marks 30a and 30b, and a mark MYn for the index mark 30
An imaging element 34y for imaging is provided together with c and 30d. The horizontal scanning directions of the imaging elements 34x and 34y are arranged in directions orthogonal to each other. In this way, the mark M
Xn and MYn are detected with the same resolution. Of course, one image sensor may generate an image signal waveform corresponding to each of the marks MXn and MYn from each of the horizontal and vertical scanning lines. In the case of a single imaging element, an image rotator may be provided in the imaging optical path after the index plate 30 to rotate the image by 90 degrees between the marks MXn and MYn.

Further, in the FIA operation unit 35, the bottom B of the image signal waveform corresponding to the marks MXn and MYn is stored.
In order to detect Ln and BRn at high speed, a circuit for binarizing each pixel at a predetermined slice level is also provided. For example, as shown in FIG. 7, an arbitrary slice level SV1,
Using one of SV2 and SV3, the bottom waveforms BLn and BRn are binarized, and the bottom position from the center in the scanning direction (x direction or y direction) of the binarization transformation is X.
Find Bn. The slice level SV1 is adjusted to the upper part of the bottom waveform, the slice level SV2 is adjusted to the middle part, and the slice level SV3 is adjusted to the lower part. Which slice level is selected is empirically determined by a wafer process, a wafer base, and the like.

Next, how marks are detected by the spot light SP of the TTL type alignment system will be described with reference to FIGS. FIG. 8A shows a shot region Sn on a wafer.
8B shows an arrangement of a conventional diffraction grating mark MK and a spot light SP provided in conjunction with FIG. 8B. FIG. 8B shows the spot light SP extending in the y direction and the mark MK relatively scanned in the x direction. 3 shows the intensity distribution of the diffracted light when the above-mentioned operation is performed. The diffraction grating mark MK is formed by forming a minute rectangular pattern (convex or concave) with a duty ratio of 1: 1 at a constant pitch in the y direction.
The width of the mark MK in the x direction is set substantially equal to the width of the spot light SP. When the spot light SP overlaps the mark MK, depending on the wavelength and the lattice constant of the laser spot light,
Higher-order diffracted light is generated at a predetermined diffraction angle in a plane perpendicular to the plane of FIG. 8A and spread in the y direction. Since this diffracted light is generated when there is a periodic structure pattern having a step edge having a constant pitch in the longitudinal direction of the spot light in the irradiated portion of the spot light SP, the S / N ratio is extremely good. Therefore, the photoelectric signal waveform obtained from this diffracted light is
The spot light SP has a waveform similar to the intensity distribution in the width direction (for example, Gaussian distribution), and is binarized at an appropriate slice level SV4 as shown in FIG. The center position can be specified. However, on the other hand, since a minute rectangular pattern is adopted as the grid element, deformation due to the wafer process tends to occur.

On the other hand, when the multi-mark MXn as shown in FIG. 9A is scanned by the SP of the slit-like spot light, unlike the case of the diffraction grating, the left and right of the linear pattern extending in the y-direction are different. The scattered light generated at the edge is photoelectrically detected. In this case, the width of the spot light SP in the x direction is smaller than the width of the linear pattern, for example, １ ／ or less. Since the multimark MXn (same for MYn) is used to detect an edge that extends continuously in parallel with the longitudinal direction of the spot light SP, the generation of scattered light is smoothed in the direction in which the edge extends (y direction). (Averaging) is performed.
As shown in FIG. 9C, the distribution of the scattered light from each edge has peaks PK1 and PK1 on the left and right sides of one linear pattern.
PK5, PK5 '. If the positions of these peak waveforms PKn, PKn' in the x direction are arithmetically averaged, the center position of the mark MXn can be obtained. However, in such a method of detecting the edge scattered light, the spot light S
Since P is a laser beam (single wavelength), the resist layer PR near the edge of the linear pattern as shown in FIG.
Phenomena such as unevenness in the direction in which scattered light is generated due to thickness unevenness, and the fact that the received light amount of scattered light itself fluctuates significantly due to interference (thin film interference) by the resist layer PR due to a single wavelength. Has been recognized. For this reason, as shown in FIG. 9C, the peak waveforms at each edge are not always completely uniform, but rather often become complex waveforms. Processing such a waveform is not always easy, but not very difficult. The problem with such a waveform is that the position where the edge scattered light is generated is affected by the thickness unevenness of the resist layer PR and apparently easily shifts in the measurement direction as described above. This causes a large error in determining the center position of the mark MXn. This is almost always
This is caused by the interference caused by the spot light SP having a single wavelength. On the other hand, if the illumination light is made to have a broadband wavelength distribution as shown in FIG. 5, no interference occurs in the resist layer, and scattered light and specularly reflected light at the step edge are extremely reduced.

Incidentally, the photoelectric signal waveform when the mark MXn (or MYn) is relatively scanned with the spot light SP is not necessarily all distorted as shown in FIG. In some cases, even for wafers of the same process, the thickness of the resist layer may be different for each wafer due to the management of the resist layer, and even in the same wafer, the waveform may be different between the central portion and the peripheral portion.

The signal waveforms shown in FIG. 8 and FIG.
It is stored in a waveform memory in the SA operation unit 9 by a digital sampling technique in response to a position measurement pulse (for example, every 0.02 μm) from the interferometer 12. Next, a typical alignment sequence of this embodiment will be described.
Here, an enhancement global alignment (EGA) system for achieving both high throughput and high alignment accuracy will be described. Details are disclosed in JP-A-61-44429. A brief description will be given.

In the EGA system, a plurality of wafers (3 to
9) The positions of the marks MYn and MXn of the shot areas Sn are measured, and based on the measured values, the minute coordinates in the running coordinate system of the stage ST of the wafer W, that is, in the xy coordinate system defined by the interferometer 12. The rotation error θ, the degree of orthogonality w of the shot array (or the movement of the stage ST) on the wafer, and the scaling errors Rx and R due to linear minute expansion and contraction of the wafer.
The parameters relating to y and the minute positional deviation of the wafer in the x and y directions, ie, the offset errors Ox and Oy, respectively, are obtained by least square approximation. Then, the shot array coordinates in design are converted into shot array coordinates to be actually exposed (stepping position coordinates of the wafer stage ST) through these parameters, and the pattern area PA of the reticle R is transferred to each shot area Sn on the wafer. Are superposed and exposed.

Here, the shot array coordinate value in the design is represented by (D
xn, Dyn) and the coordinate value of the wafer to be positioned by actual stepping is (Fxn, Fyn), the following equation (1) is established.

[0041]

(Equation 1)

Here, A and O are called transformation matrices and are approximated by the following equations (2) and (3).

[0043]

(Equation 2)

Assuming that the position coordinate value (RFxn, RFyn) of the shot area obtained by position measurement (sample alignment) of each mark Mxn, Myn of a plurality of shot areas on the wafer is the actual exposure. Power coordinates (Fxn, Fyn) and measured values (RFxn, RF
yn), that is, an address error (Ex)
n, Eyn) = (RFxn-Fxn, RFyn-Fy)
The values of the parameters of the transformation matrices A and O are determined by calculation so that n) is minimized. Thus, the transformation matrices A, O
Is determined, the stepping position (Fxn, Fyn) of the wafer stage ST is determined based on the above equation (1).
, And the stage ST may be positioned there and exposed.

The suffix n is the number of the shot on the wafer. Here, associating with the main control system 50 shown in FIG. 1, the shot array coordinate value (Dx
n, Dyn) are stored in the EXP shot map data section 503, the coordinate values (design values) of the shot area for obtaining the actual measurement values (RFxn, RFyn) are stored in the ALG shot map data section 504, and the equation (1) ),
(2), (3) and the least squares approximation for determining the transformation matrices A and O are stored in the EGA calculation unit 502.

In the above-mentioned EGA system, conventionally, sample alignment was performed using only the spot light (single wavelength) of the alignment system of the TTL system. However, as described with reference to FIG. Therefore, in this embodiment, Of using the broadband illumination light
Sample alignment is also performed in an f-Axis type alignment system. Among the parameters θ, w, Rx, Ry, Ox, and Oy in the calculation of the EGA method, parameters that are considered to be inaccurate in the measurement of the single-wavelength spot light SP, in particular, parameters Rx and R related to scaling
y was replaced with a similar parameter obtained using an Off-Axis type alignment system.

FIG. 10 shows sample alignment shot areas S1, S2, S3, S4, S5, and S5 on the wafer W.
The arrangement of S6, S7 and S8 is shown, and each shot area S1 is shown.
To S8 are marks MXn, MXn, as shown in FIG.
MYn is formed. FIG. 10 representatively shows only the marks MX1 and MY1 of the shot area S1, but the same applies to the other shot areas S2 to S8. Here 8
Although the sample alignment shots are shown, it is important in the EGA method that the arrangement of the plurality of sample shots is not on the same straight line. This is necessary when the number of sample shots is as small as three or four, for example. Further, as is clear from FIG. 10, it is preferable to allocate the sample shots so that the array coordinate values of the sample shots do not overlap in the x and y directions as much as possible. For example, only the shot area S8 matches the x coordinate value of the shot area S1, and no one matches the y coordinate value.

The sample shot areas S1 to S
The coordinate value 8 is stored in the ALG shot map data section 504. Therefore, the sequence controller 506
After the wafer W is roughly (eg, about ± 2 μm) aligned and mounted on the wafer stage ST, the ALG
The respective coordinate values of the shot areas S1 to S8 of the sample alignment are input from the shot map data section 504, and the marks MXn, MXn,
The wafer stage ST is sequentially moved to a relative scanning position with respect to the MYn spot light SP. In the normal alignment mode, the mark MX1 in the shot area S1 is relatively scanned in the x direction by the spot light SP extending in the y direction, and the mark MY1 is relatively scanned in the y direction by the spot light SP extended in the x direction. This operation is performed in the shot areas S2 to S8.
Until clockwise. The eight measured values (RFx1, RFy1) to (RFx8, R
Fy8) is an ALG data storage unit 50 as information AP1.
1 is temporarily stored.

Next, the sequence controller 506
The stage controller 14 is controlled so that the marks MXn and MYn of the above-described eight shot areas S1 to S8 are located at the projection positions of the index plate 30 of the Off-Axis type alignment system by the objective lens 27 and the like. At this time, in the present embodiment, as described above, only the marks which dominantly affect the scaling errors Rx and Ry are selected from the parameters to be calculated by the EGA method, and the sample alignment is executed again.

The scaling errors Rx and Ry are
Here, for example, the marks MY2, MY3, MX4, M
Only the eight marks X5, MY6, MY7, MX8 and MX1 are selected. That is, in FIG. 10, when the wafer W is divided into four blocks of an upper region, a lower region, a right region, and a left region, the scaling amount Rx is expansion and contraction in the x direction. The mark M in the x direction of the shot areas S1, S8, S4, S5 included in the block
Xn only is sample-aligned and the scaling amount R
Since y is an expansion and contraction in the y direction, the shot areas S2, S3, S6 and S7 included in the upper and lower blocks are used.
The sample alignment is performed only on the mark MYn in the y direction.

As described above, information on each mark position to be sample-aligned by the Off-Axis type alignment system is stored in the ALG shot map data section 504 in advance. Therefore, the sequence controller 50
Reference numeral 6 denotes a relative distance (base) in the wafer plane between the spot light SP of the TTL alignment system stored in the correction data unit 505 and the detection center (center of the index plate 30) of the Off-Axis alignment system. Line amount) and
Based on the information on the sample alignment position obtained by the TTL type alignment system obtained earlier, the eight marks are sequentially positioned in the window of the index plate 30 as shown in FIG.
As shown in FIG. 3B, the position of each mark is measured again by calculating the amount of positional deviation. The measured values RFy2, RFy2, RFy3, RFy4 of the eight mark positions thus measured
, RFy5, RFy6, RFy7, RFy8, RFx1
From the FIA operation unit 35 as the information AP2
The data is stored in the G data storage unit 501.

Next, the EGA calculation unit 502 determines each parameter based on the above equations (2) and (3).
Here, in this embodiment, two types of calculation methods are possible. 1
First, the parameters Rx, Ry, w, θ, Ox, Oy are based only on the result of sample alignment by the TTL method.
Is obtained, only the scaling amounts Rx ′ and Ry ′ are independently obtained by using the sample alignment result by the Off-Axis method, and the scaling amounts Rx and Ry previously obtained are obtained.
It is a method to replace it. The other is to replace them at the stage of actual measurement of the sample alignment result,
In this method, the parameters Rx, Ry, w, θ, Ox, and Oy are directly obtained. That is, in the latter case, the eight measured values are (RFx1 ', RFy1), (RFx2, RFy2).
'), (RFx3', RFy3), (RFx4, RF
y4 '), (RFx5', RFy5), (RFx6, R
Fy6 '), (RFx7, RFy7'), (RFx8
', RFy8'), and the calculation of the EGA method is performed based on this value. The actual calculation for each parameter is
For example, it is disclosed in detail in Japanese Patent Application Laid-Open No. 62-84516, and is obtained by the following equations (4) and (5). here,
The transformation matrix A is represented as follows.

[0053]

(Equation 3)

A = Rx, B = −Ry (w + θ),
It is assumed that C = Ry · θ, D = Ry, E = Ox, and F = Oy.

[0055]

(Equation 4)

By solving equations (4) and (5), A, B,
Find C, D, E, F. Then, further, Rx = A,
Ry = D, θ = C / D, w = −B / AC / D, Ox =
If E and Oy = F are obtained, each parameter is obtained. In the expressions (4) and (5), the number of addition operation terms including the actually measured value of ΣRFxn is eight in this embodiment, and when the measured value is replaced, the value is used as it is.

When the parameters Rx 'and Ry' are independently obtained from the eight actually measured values detected by the Off-Axis method as in the first calculation method, the above equation (4) is used.
The parameters A (Rx) and D (Ry) may be calculated using (5). In this case, the measured value of the mark in the x direction is
RFx1 ', RFx3', RFx5 ', RFx8'
These are substituted for RFxn on the right side of the equation (4), and the measured values of the mark in the y direction are RFy2 ', RFy4
, RFy6 ′, and RFy7 ′, and these may be substituted into the right side RFyn of Expression (5). Then, the parameters A and D thus obtained may be replaced with the parameters previously obtained by the TTL measurement.

When the conversion matrices A and O are calculated as described above, the values are stored in the EGA operation unit 502, and the EGA operation unit 502 further calculates the shot to be exposed based on the equation (1). Array position (Fxn, Fy
n), and sequentially calculates the calculated values.
06 is output. As a result, the sequence controller 506 outputs a command for stepping control of the wafer stage ST to the stage controller 14. For this reason, the wafer stage ST is provided for each shot area Sn on the wafer W.
Are sequentially stepped so as to coincide with the projected image of the pattern area PA of the reticle R, and the overlay exposure is performed.

As described above, in this embodiment, both the TTL type and the off-axis type are used for one wafer, but the scaling amount is generally the same between a large number of wafers subjected to the same process. Often indicates. Therefore, when processing wafers for one lot (usually 25 wafers) continuously, the first few wafers (about 3 to 5 wafers) are processed using both the TTL method and the Off-Axis method. As in the embodiment, a sequence of the EGA method for minimizing the measurement error of the scaling amount is adopted, and for the subsequent wafer processing, the average value of the scaling amounts Rx and Ry obtained so far is fixed. Other parameters (O
x, Oy, θ, etc.) only in the TTL method or Off-A
The determination may be made by sample alignment using only one of the xis methods.

Further, when it is predetermined to use both the TTL method and the Off-Axis method for a certain wafer, the marks MX2 and MX3 of the sample alignment shots S1 to S8 shown in FIG. , M
Y4, MY5, MX6, MX7, MY8, MY1 are TT
Sample alignment using L-type alignment system and E
It is used to determine only the parameters Ox, Oy, θ, and w of the GA, and the other marks are sample-aligned by an Off-Axis type alignment system, and the parameters Rx, Ry
The mark to be measured may be assigned in advance so as to be used for the determination. In this case, the total measurement time of the EGA type sample alignment becomes considerably short,
A decrease in throughput is suppressed.

Next, a second embodiment of the present invention will be described with reference to FIG. This embodiment is different from the first embodiment in how to determine the EGA sample alignment shot and how to use the alignment system. In general, when comparing the TTL alignment system and the Off-Axis alignment system shown in FIG. 1, the minute fluctuation of the distance (TTL base line) of the spot light SP with respect to the projection point at the center of the reticle R pattern. Is often more stable than a slight change in the distance (Off-Axis base line) between the center (off-axis system optical axis) of the index plate 30 and the projection point of the pattern center of the reticle R.

Therefore, as shown in FIG. 11, a total of 12 sample alignment shots S, 4 shots at the center on the wafer W and 2 shots at each of the upper, lower, left and right portions, are provided.
Set 1 to S12. The marks MX1, MY1, MX2, MY2, MX3, MY3, MY3 formed in the four central shot areas S1, S2, S3, S4 respectively.
X4 and MY4 perform position measurement with a single wavelength spot light of an alignment system of the TTL system, and calculate only offset errors Ox and Oy among the parameters of the EGA system from the result. Next, the remaining marks MX5, MY5 to MX12 and MY12 of the remaining eight shot areas S5 to S12 are turned off.
Position measurement is performed by an alignment system for white light illumination in the Axis system, and the remaining parameters Rx, Ry, w, and θ in the EGA system are calculated from the results.

In this manner, a plurality of marks at the center of the wafer, which can maintain the offset error relatively well, are replaced with a TTL having a stable baseline variation with respect to the reticle R.
Since the sample alignment is performed by the alignment system of the system, the determined parameters Ox and Oy have extremely high accuracy. Further, from the measurement information of the marks in the shot areas S5 to S12 located around the wafer W, the parameter R which is likely to cause a measurement error due to the interference of the resist layer is obtained.
Since x, Ry, θ, and w are obtained using an alignment system for white light (broadband) illumination, there is no influence of interference, and these parameters Rx, Ry, w, and θ are also highly accurate.

Next, a third embodiment of the present invention will be described with reference to FIG. The present embodiment employs a TTR (through-the-reticle) system for detecting a wafer surface from above a reticle R via a projection lens, and simultaneously scans a mark RM of the reticle R and a wafer mark with a slit-shaped spot light. Things. As shown in FIG. 12, a transparent window-shaped mark RM is formed at the outermost periphery (corresponding to a scribe line on a wafer) of the pattern area PA of the reticle R.
The alignment optical system of the TTR system includes a total reflection mirror 60, an objective lens 61 and a focusing element 62, relay lenses 63 and 65, a mirror 64, a vibration mirror 66, a mirror driving system 67, a beam splitter 68, and relay systems 69 and 70. ,
Polarizing beam splitter 71 and light receiving elements 72 and 73
It consists of. The laser beam LB that has passed through the laser light source 1 and the beam shaping optical system 2 contains two polarized components at a wavelength longer than the exposure wavelength (for example, Ar ion laser light). This beam LB is reflected by the beam splitter 68, reciprocally deflected by the vibrating mirror 66, and enters the bifocal element 62 through the relay lenses 65 and 66. This 2
The focusing element 62 is a device in which a crystalline substance such as quartz is bonded to ordinary optical glass (or quartz), and gives different refracting powers to ordinary rays and extraordinary rays. Therefore, the beam LB that has passed through the bifocal element 62 enters the telecentric objective lens 61 at different divergence angles (or convergent angles) depending on the polarization components (P-polarized light and S-polarized light). The beam emitted horizontally from the objective lens 61 is bent by the mirror 60 in parallel with the optical axis AX of the projection lens, and enters the reticle R perpendicularly. At this time, due to the difference in the polarization components, one of the polarization components, for example, the beam LBp of the P-polarized light, forms an image as a slit-shaped spot light SPr in the plane of the mark RM of the reticle R, and the other polarization component,
For example, the S-polarized beam LBs forms a slit-shaped spot light SPw in a plane Iw in the space above the reticle R and forms an image. Here, mark (pattern) surface of reticle R and surface Iw
Is set to be equal to the amount of chromatic aberration on the reticle side of the projection lens caused by the wavelength of the beam LB. In FIG. 12, the beam LBp emitted from the objective lens 61 is indicated by a solid line, and the beam LBs is indicated by a broken line. Here, the bonding surface of the bifocal element 62 is determined so as to coincide with the front focal plane of the objective lens 61, that is, the plane EP ′ conjugate with the pupil EP of the projection lens PL. Further, the origin of the beam deflection of the vibration mirror 66 and the bifocal element 62 (pupil conjugate plane EP ′) are defined by the relay lenses 63 and 65.
Are conjugated to each other. The spot light SPr scans the mark RM one-dimensionally by the action of the vibrating mirror 66, and scattered light, specularly reflected light, etc. from the mark RM (edge of the window) are reflected by the mirror 60, the objective lens 61, and the focusing element 6.
2. After passing through the relay systems 63 and 65, the vibrating mirror 66, and the beam splitter 68, they reach the relay systems 69 and 70. Since the spot light SPr is P-polarized light, the mark RM
Is mainly P-polarized light, and almost 100% of the light is transmitted by the polarization beam splitter 71 and reaches the light receiving element 72.

On the other hand, the S-polarized beam LBs that has passed through the transparent window of the mark RM of the reticle R is conjugated to the spot light SPw (spot light SP
w ′) to form an image. At this time, the spot light S
Pw ′ also scans the wafer mark one-dimensionally by the action of the vibrating mirror 66. The scattered light, specularly reflected light, and diffracted light from the wafer mark return to the optical path of the beam LBs, pass through the transparent window of the mark PM, and return to the polarization beam splitter 71 in the same manner. Since the optical information from the wafer mark is mainly S-polarized light, almost 100% is reflected by the polarizing beam splitter 71 and reaches the light receiving element 73.

The driving system 67 of the oscillating mirror generates the spot light S
An up / down pulse signal is output for each unit scan amount of Pr, SPw ′, and is output to the TTR calculation unit 75. The TTR calculation unit 75 digitally samples the photoelectric signal waveforms from the light receiving elements 72 and 73 in response to the up / down pulse signal, and based on the two signal waveforms, the mark RM of the reticle R and the wafer mark ( For example, the relative displacement amount with respect to (MXn, MYn, MK) is obtained. Then, based on the displacement amount and the stop position of the wafer stage ST, an actual measurement value of the wafer mark is obtained,
This is sent to the ALG data storage unit 501 shown in FIG. 1 as information AP3.

When the beam LB incident on the beam splitter 68 is limited to only the S-polarized component and is incident on the relay lens 65 by bypassing the vibrating mirror 66, the TTL alignment system shown in FIG. Exactly the same mark detection becomes possible. That is, in this case, since only the stationary spot light SPw (SPw ′) is produced, the wafer is scanned with this spot light SPw ′, and the photoelectric signal waveform from the light receiving element 73 is converted into the position pulse from the interferometer 12. Digital sampling is performed in response to the signal PDS. In this case, the position of the wafer stage ST where the spot light SPw 'coincides with the center of the wafer mark is sent to the ALG data storage unit 501 as information AP3.
It is clear that this method has exactly the same sequence as the first embodiment.

In FIG. 12, the light receiving elements 72, 73
Are both arranged conjugate with the pupil conjugate plane EP ′, and FIG.
The light receiving surface is configured as shown in FIG. An image conjugate plane F0 is formed between the relay systems 69 and 70, where an image of the mark RM of the reticle R and an image of the wafer mark are simultaneously formed. In the case where a TTR type alignment system as in this embodiment is provided, the mark position information AP3 from the TTR operation unit 75 may be used instead of the information AP1 from the LSA operation unit 9 shown in FIG.

As described above, in the stepper in which the TTR type alignment system of this embodiment is incorporated in place of the TTL type alignment system of FIG. In addition to the die-by-die alignment method that performs alignment by using an EGA method that is used in combination with an Off-Axis alignment system that uses white light illumination, various alignment modes can be used freely. Further, since the bifocal element 62 is arranged and used in a pupil conjugate, a non-photosensitive wavelength longer than the exposure wavelength can be selected for the beam LB, and the relative lateral displacement between the two spot lights SPr and SPw is almost ignored. be able to. Further, since the beam LB can select a wavelength having a small absorption with respect to the resist layer, the optical characteristics (reflectance, refraction, and refraction) of the resist layer from the time when the spot light SPw ′ starts scanning on the wafer for mark detection. Rate, etc.) hardly changes. Therefore, errors in mark position detection due to changes in the optical characteristics of the resist layer are extremely small.

Therefore, prior to the mark detection by the die-by-die method, marks attached to several shot areas on the wafer are detected by an off-axis white light alignment system, and then the same mark is detected. Is detected by the spot light SPw 'of the TTR system, and the mark signal waveforms or mark positions obtained by both of the detection methods are compared, and a minute offset error of the mark position detected by the single-wavelength spot light SPw' is compared. Is stored in the correction data section 505 of the main control system 50 in advance. Then, when performing die-by-die alignment by the alignment system of the TTR method, the position of the wafer mark may be corrected by the value stored in the correction data unit 505 to perform the alignment. This can be similarly applied to the TTL type alignment system.

FIG. 13 is a diagram showing an Off state according to the fourth embodiment of the present invention.
13 shows a modification of the alignment system of the -Axis system. FIG.
In the Off-Axis system shown in (1), only image detection is illustrated, but a laser spot can be projected onto a wafer by using the objective lens 27 and the like. Therefore, in the configuration of FIG. 1, the mirror 26 is changed to a beam splitter or a dichroic mirror 80. After that, a light receiving system including a relay lens 81, a beam splitter 82, a relay lens 87, and a light receiving element 87, and a laser light source 8
6. Beam shaping optical system including cylindrical lens 8
5, a light transmission system including a mirror 84 and a lens system 83 are provided. Here, the light receiving surface of the light receiving element 87 is specularly reflected light, diffracted light,
The light is divided so as to receive the scattered light independently of each other, and is arranged by the relay systems 81 and 87 so as to be conjugate with the stop surface (front focal plane) 27a of the objective lens 27. The laser beam from the laser light source 86 passes through the telecentric objective lens 27 and the prism 28 and forms an image as a slit-like spot light on the wafer W along an optical path indicated by a solid line.

In this embodiment, since a stationary spot light is formed on the axis of the objective lens 27, the wafer stage S
By scanning T, a photoelectric signal waveform corresponding to the wafer marks MXn and MYn is obtained from the light receiving element 88.
Further, the white illumination light passes through the mirror 80 through the objective lens 27.
However, when the wavelength band used and the wavelength of the beam from the laser light source 86 overlap, a switch or the like may be provided so that only one of them is turned on.

This embodiment is suitable as an Off-Axis alignment system for an excimer stepper or the like, in which the optical performance of the projection lens is greatly restricted among the steppers. Further, since the mark detection position by the spot light and the mark detection position of the index plate 30 by the white light substantially coincide with each other, there is an advantage that a decrease in throughput is almost prevented. Of course, the relative distance between the position of the spot light and the positions of the index marks 30a to 30d is
It is necessary to measure in advance using the reference mark FM.

As described above, in each embodiment of the present invention,
An autofocus system may also be provided for the Axis alignment system, but the cost is high, so the mark measurement area on the wafer is focused using the autofocus system for the projection lens PL. Thereafter, the Z-stage may be sent below the Off-Axis objective lens 27 with the Z stage fixed. At this time, as shown in the shot arrangement shown in FIG. 11, the surrounding four shot areas are set to two adjacent shot areas (for example, shots S5 and S6 or shots S7 and S8), so that both shot areas are set. Since the regions can have the same focus, the throughput is improved. However, in this case, the wafer surface needs to be kept parallel to the projection image plane by the leveling stage. If a digital micrometer or the like that accurately reads the height position of the Z stage in the optical axis direction is incorporated, focus adjustment is performed when detecting a mark on a wafer with a TTL or TTR alignment system. The height position of the Z stage at that time is stored, and when the mark of the same shot is detected by an Off-Axis type alignment system, the focus is controlled by open control so as to reproduce the stored height position of the Z stage. Matching may be performed. Further, instead of the single-wavelength illumination light (spot light), the same quasi-monochromatic light as the exposure light, such as g-line and i-line, is applied to the slit plate, and the slit image by the light is projected onto the wafer via the projection lens. The same applies to an alignment system for image formation and projection.

Further, the alignment system using the single-wavelength illumination light described in each embodiment detects the mark as a slit-like spot light, but is not limited to this. The same applies to a system in which a local region including a mark on a wafer is uniformly illuminated with a single-wavelength illumination light, the reflected light is formed into a mark image, and the mark image is photoelectrically detected. Further, as another alignment method using single-wavelength illumination light, it can be similarly used in a double diffraction grating alignment method. This alignment method uses the T shown in FIG.
It can be incorporated into a TL alignment system, the TTR alignment system shown in FIG. 12, or the Off-Axis alignment system shown in FIG.
The configuration shown in FIG.

FIG. 14 shows an example in which the laser beam is incorporated in an alignment system of the Off-Axis system. A single-wavelength parallel laser beam LB0 contains linearly polarized components different from each other and is incident on the frequency shifter 100. The frequency shifter 100 receives a high-frequency signal having a modulation frequency of 2f, and emits a parallel beam LB1 having a frequency difference of 2f between two linearly polarized components of the beam LB0. This beam LB1 is converted by the Wollaston prism 90 into two
LBs and LBp, respectively, and enter the lens system 91. The beams LBs, LBp
Is the lens system 9 after being converged once at the pupil (aperture) position of the system
2 (corresponding to the objective lens 27 in FIG. 13), and the marks MXn on the wafer W
Are simultaneously irradiated. The Wollaston prism 90 is arranged conjugate with the wafer W with respect to the lens systems 91 and 92.
However, if left as is, the polarization directions of the two beams incident on the wafer W from two directions are different, and no interference fringes are generated on the wafer W. Therefore, an input / 2 plate 97 for aligning two equivalent beams to one polarization component is provided by, for example, a beam LB.
It is provided in the optical path of p and is converted into a beam LBs of the other polarization component. By doing so, the same position on the wafer W is
The two beams illuminating from the directions are both S-polarized beams LBs, and the linear pattern P1 of the mark MXn (the one shown in FIG. 4 can be used as it is) on the wafer W according to the wavelength of the beam LBs and the intersection angle between them. One-dimensional interference fringes are generated in the array direction of .about.P5. FIG. 15 shows the relationship between the waveform FW of the intensity distribution of the interference fringes and the cross-sectional shapes of the linear patterns P1 and P2 of the mark MXn. Here, the pitch of the waveform FW in the x direction is exactly １ ／ of the pitch of the gratings P1, P2,.
Set to be double. Since this interference fringe has a frequency difference of 2f between the beams from two directions, the frequency 2
Flowing at f. For this reason, since the interference fringe irradiates the mark MXn, interference light is generated from the mark MXn in a substantially vertical direction, passes through the lens system 92, is reflected by the small mirror 93 on the axis, and passes through the pupil relay system 94. Light receiving element 95
Reach The light receiving element 95 receives the interference light that changes in the amount of light in a sine wave shape at the frequency 2f, and outputs a corresponding photoelectric signal to the heterodyne arithmetic unit 96. The heterodyne operation unit 100 receives the original signal of the modulation frequency 2f as a reference signal and obtains a phase difference (within ± 180 °) of the photoelectric signal of the frequency 2f with respect to the reference signal. This phase difference is the amount of misalignment of the mark MXn with respect to the reference position in the x direction, here ± 1/4 of the lattice patterns P1, P2.
It corresponds to the amount within the pitch. Therefore, the heterodyne arithmetic unit 100 outputs precise position information AP4 of the mark MXn based on the position information PDS of the wafer stage ST when the wafer W is stopped and the deviation amount within ± 1/2 pitch. This information AP4 is sent to the ALG data storage unit 501 as a mark position measurement value using light of a single wavelength.

When the configuration shown in FIG. 14 is incorporated into the TTL system, the objective lens 6 shown in FIG. 1 is used as a lens system 92 so that the surface of the spot light SP0 in FIG. 1 corresponds to the wafer surface in FIG. What is necessary is just to assemble an optical system. In the case of further incorporating into the TTR system, a pair of the objective lens 61 and the bifocal element 62 in FIG. 12 is provided instead of the objective lens 92 in FIG. 14, and the wafer surface in FIG. May be made to correspond to the surface Iw. When there is no frequency difference between the two beams irradiating the wafer from two directions, the interference fringes FW become stationary, and in this case, the wafer is slightly moved.

In the above-described diffraction grating alignment method, since the coherence of the laser beam is used, an interference phenomenon in the resist layer can occur exactly in the same manner as described above.

[0079]

As described above, according to the present invention, a plurality of alignment marks formed on one substrate are illuminated with a single wavelength or quasi-monochromatic illumination light to detect the mark position, and further to provide broadband illumination. The position of the mark is detected by irradiating with light, and the positioning and alignment of the substrate are performed using the results of both detections. Therefore, the influence of the resist layer is reduced, and extremely high alignment accuracy can be obtained.

The present invention is not limited to the projection type exposure apparatus,
The present invention can be applied to exposure apparatuses of other types, such as proximity and contact types, in exactly the same manner.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a projection type exposure apparatus according to a first embodiment of the present invention.

FIGS. 2A, 2B, and 2C are diagrams showing the configuration of each part of the apparatus shown in FIG.

FIG. 3 is a diagram showing an arrangement of shot areas on a wafer.

FIGS. 4A and 4B are diagrams showing a plane and a cross section of a mark shape on a wafer.

FIGS. 5A and 5B are diagrams showing mark detection by an Off-Axis type alignment system. FIGS.

FIG. 6 is a diagram showing a configuration of an Off-Axis alignment system provided with two image sensors.

FIG. 7 is a diagram illustrating a state of detecting a bottom of an image signal.

8A and 8B are diagrams showing a state of a photoelectric signal waveform when a conventional diffraction grating mark is scanned with a spot light as an alignment mark.

FIGS. 9A, 9B, and 9C are diagrams showing a state of a photoelectric signal waveform when the mark shown in FIG. 4 is scanned with spot light.

FIG. 10 is a diagram showing an arrangement of shot areas on a wafer to be aligned in the first embodiment.

FIG. 11 is a diagram showing an array of sample alignment shots in a second embodiment.

FIG. 12 is a diagram showing a configuration of a TTR type alignment system according to a third embodiment of the present invention.

FIG. 13 shows an Off-Axi according to a fourth embodiment of the present invention.
FIG. 2 is a diagram showing a configuration of an s-type alignment system.

FIG. 14 is a diagram showing a configuration of a double diffraction grating alignment system according to a fifth embodiment.

FIG. 15 is a diagram illustrating a state of a diffraction grating and distribution of interference fringes.

[Explanation of symbols]

R: reticle, PL: projection lens, W: wafer, EP: pupil, MXn, MYn: mark, S
n: shot area, 8: light receiving element, 9: L
SA arithmetic unit, 20 ... halogen lamp, 23
..Filter, 27 ... Objective lens, 30 ... Index plate, 34 ... Image sensor, 35 ... FIA operation unit, AP1, AP2, AP3, AP4 ... Mark detection position information, 501 ..ALG data storage unit, 502
... EGA operation unit

Claims (9)

(57) [Claims] An illumination optical system for irradiating illumination light to a mask,
A projection exposure apparatus, comprising: a projection optical system having aberration corrected with respect to the wavelength of the illumination light. The projection exposure apparatus has an objective optical system different from the projection optical system, and the first beam and the second beam have different wavelengths from the illumination light. In order that the detection position of the grating mark on the substrate almost coincides with the beam,
The first and second beams are irradiated onto the substrate via the objective optical system, and the first and second beams are used to detect the position information of the grid marks, respectively, so that the first and second beams are detected.
And an off-axis type alignment optical system for independently receiving a plurality of lights generated from the lattice mark by irradiation of the second beam. 2. The projection exposure apparatus according to claim 1, wherein the alignment optical system receives light generated from the lattice mark by interference with the movement of the substrate. 3. The method according to claim 1, wherein the first beam is a single-wavelength laser beam, and the alignment optical system has a photoelectric element that receives interference light obtained from the lattice mark by irradiation of the first beam. 3. The projection exposure apparatus according to claim 2, wherein: 4. The apparatus according to claim 3, wherein said first beam comprises two laser beams, and said alignment optical system generates interference fringes on said substrate by irradiating said two laser beams. 3. The projection exposure apparatus according to claim 1. 5. The projection exposure apparatus according to claim 1, wherein the alignment optical system has a photoelectric element for independently detecting a plurality of lights generated from the lattice mark by the irradiation of the first beam. . 6. The apparatus according to claim 3, wherein said second beam is white light, and said alignment optical system has an image sensor for detecting light generated from said grid mark by irradiation of said second beam. The projection exposure apparatus according to any one of claims 1 to 5. 7. The alignment optical system according to claim 1, wherein the alignment optical system has a beam splitter or a dichroic mirror that combines the first beam and the second beam and guides the combined beam to the objective optical system. The projection exposure apparatus according to claim 1. 8. The beam splitter or the dichroic mirror, wherein the alignment optical system is configured to irradiate the light generated from the grid mark by the irradiation of the first beam and the light generated from the grid mark by the irradiation of the second beam. 8. The projection exposure apparatus according to claim 7, wherein the light is separated by: 9. The projection exposure apparatus according to claim 1, wherein said illumination light is excimer laser light.