JPH0521314A - Projection aligner - Google Patents

Abstract

PURPOSE:To improve base line measurement accuracy and to improve treatment velocity by presetting measurement values of two sets of interferometers equal when measuring a position of a reticle using a reference board when a base line is measured. CONSTITUTION:A reference board FP having a reference mark FM2 which registers with a mark RM on a reticle R and a reference mark FM1 which registers with a detection center point of an off axis alignment system OWA is provided on a wafer stage WST. When a base line is measured, position deviation amount of the reticle R and the reference board FP is obtained keeping the wafer stage WST stationary, and a position deviation amount of a detection central point of the off alignment system OWA and the refence board FP is also obtained. Furthermore, inside counters are mutually preset to make measurement values of interferometers (IFX,IFY1) used during off axis alignment and interferometers (IFX, IFY2) used during exposure equal at a position of the wafer stage WST when a base line is measured.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a projection exposure apparatus for exposing a photosensitive layer coated on a substrate such as a semiconductor wafer or a glass plate for liquid crystal, and particularly, it precisely manages a baseline of an off-axis type alignment system. The present invention relates to a projection exposure apparatus having a function to perform.

[0002]

2. Description of the Related Art Conventionally, projection exposure apparatuses equipped with an off-axis alignment system (hereinafter referred to as steppers for convenience) have been disclosed in JP-A-53-56975 and JP-A-56.
As disclosed in Japanese Laid-Open Patent Publication No.-134737, a reference mark plate is fixed on a wafer stage that holds a photosensitive substrate (hereinafter, referred to as a wafer) and moves two-dimensionally by a step-and-repeat method. However, the reference mark plate is used to control the distance between the off-axis alignment system and the projection optical system, that is, the so-called baseline amount. FIG. 1 is a diagram schematically showing the principle of baseline measurement disclosed in each of the above publications. In FIG. 1, the main condenser lens ICL uniformly illuminates the reticle (mask) R during exposure. The reticle R is held by the reticle stage RST, and this reticle stage RST is moved so that the center CC of the reticle R coincides with the optical axis AX of the projection lens PL. On the other hand, wafer stage WST
A reference mark FM equivalent to the alignment mark formed on the surface of the wafer is attached above the reference mark FM.
When the stage WST is positioned so that is at a predetermined position within the projection field of the projection lens PL, the mark RM of the reticle R and the reference mark are made by the alignment system DDA of the TTL (through the lens) system provided above the reticle R. FM and FM are detected at the same time. The distance La between the mark RM and the center CC of the reticle R is a predetermined value in design, and the distance between the projection point of the mark RM and the projection point of the center CC on the image plane side (wafer side) of the projection lens PL is: La /
It becomes M. Here, M is the magnification of the projection lens PL when the reticle side is viewed from the wafer side, and M = 5 in the case of the ⅕ reduction projection lens.

Outside the projection lens PL (outside the projection field of view)
For off-axis wafer alignment system O
WA is fixed. The optical axis of the wafer alignment system OWA is parallel to the optical axis AX of the projection lens PL on the projection image plane side. Inside the wafer alignment system OWA, a target mark TM that serves as a reference for aligning the mark on the wafer or the reference mark FM is provided on the glass plate, and the projection image plane (wafer surface or reference mark FM The plane) is almost conjugate with the plane.

The baseline amount BL is shown in FIG.
As described above, the reticle mark RM and the reference mark FM are aligned.
Position WST of the stage WST when₁And the indicator
When the mark TM and the reference mark FM are aligned
Position X of mushroom stage WST ₂And laser interferometer
Measure the difference (X₁-X₂) Is calculated by
It This baseline amount BL will be used later as a mark on the wafer.
The wafer with the wafer alignment system OWA.
What becomes the reference amount when it is sent directly under the shadow lens PL
Is. That is, one shot on the wafer
The distance between the center of the area) and the mark on the wafer is XP,
Wafer stay when mark matches index mark TM
X position of the WST₃Then, the shot center and reticle
In order to match the center CC of the wafer, the wafer stage W
It suffices to move ST to the position of the following equation.

X ₃ -BL-XP or X ₃ -BL + XP In principle, this calculation formula only represents a one-dimensional direction, and it is necessary to consider it in two dimensions in reality. Furthermore, TTL alignment system DDA The calculation method also differs depending on the arrangement of (that is, the mark RM) and the arrangement of the wafer alignment system OWA.

In any case, after the mark position on the wafer is detected by using the off-axis type wafer alignment system OWA, the pattern of the reticle R is immediately formed on the wafer by feeding the wafer stage WST by a predetermined amount. The shot area can be accurately overlapped and exposed.

[0007]

In the prior art as described above, the detection center point (center of the index mark TM) of the off-axis alignment system OWA and the projection point of the mark RM of the reticle R by the projection lens PL. When measuring the positional relationship (baseline amount BL) with, the relative distance is obtained by moving the wafer stage WST and using a laser interferometer. Therefore, due to inevitable factors such as the running accuracy of wafer stage WST and the air fluctuation of the laser beam optical path of the laser interferometer, there is a limit to the improvement of the accuracy of the baseline measurement. Further, the movement of the wafer stage WST for positioning the fiducial mark FM in the detection area of the TTL alignment system DDA, and the off-axis alignment system OWA of the fiducial mark FM.
Wafer stage WS for positioning at the detection center point of
It is necessary to move T, and there is a limit to increasing the speed of the baseline measurement processing.

Further, in the conventional stepper, the extension line of the measuring axis (beam optical axis) of the laser interferometer for measuring the position of the wafer stage WST is set so as to intersect the optical axis of the projection lens in both the X and Y directions. Only set, off
When detecting various marks with the axis alignment system OWA, it is sometimes difficult to always realize the mark detection direction in which the Abbe error (sine error) becomes zero. Therefore, a set of laser interferometers that makes the Abbe error zero with respect to the optical axis of the projection lens and a laser interferometer that makes the Abbe error zero with respect to the detection center point of the off-axis alignment system OWA. It is also conceivable to provide In this case, the two laser interferometers are turned off.
It will be used by switching between the stage position measurement at the time of wafer alignment using the axis alignment system OWA and the stage position measurement at the time of projection exposure, but consider the consistency (unity) of the values in both position measurement. Otherwise, as a matter of course, it will cause an error.

The present invention has been made in view of the above conventional problems, and an object of the present invention is to obtain a projection exposure apparatus which has an improved baseline measurement accuracy and an improved processing speed.

[0010]

According to the present invention, a reference mark FM ₂ aligned with a mark RM on a reticle R and an off-axis alignment system O are provided on a wafer stage WST.
A reference plate FP formed with a reference mark FM ₁ matching the detection center point of the WA is provided. During the baseline measurement, the amount of positional deviation between the reticle R and the reference plate FP is obtained while the wafer stage WST is stationary, and at the same time, the positional deviation between the detection center point of the off-axis alignment system OWA and the reference plate FP is detected. I tried to find the amount.

Furthermore, a pair of interferometers (IFs) that satisfy the Abbe error with respect to the off-axis alignment system OWA.
X, IFY ₁ ) and a pair of interferometers (IFX, IFY ₂ ) that satisfy the Abbe error with respect to the projection optical system, and the two sets of the above-mentioned two sets at the position of the wafer stage WST during the baseline measurement. The internal counters were configured so that they could be preset to each other so that the interferometer measurements were equal.

[0012]

When measuring the position of the reticle R using the reference plate FP at the time of baseline measurement, presetting such that the measurement values of the two interferometers are equal to each other, two measurements for the same direction, for example, the Y direction are performed. An imaginary line connecting the reference points of the interferometer is precisely parallel to the reflecting surface of the Y-direction moving mirror (IMy) on the wafer stage.

Therefore, after presetting, no error will occur even if one of the two sets of interferometers is selected and used as it is for position control of the wafer stage.

[0014]

2 is a perspective view showing the construction of a projection exposure apparatus according to an embodiment of the present invention, in which the same members as those in the conventional apparatus of FIG. 1 are designated by the same reference numerals. In FIG. 2, on the reticle R, a pattern area PA in which a circuit pattern or the like to be exposed is formed on the wafer and reticle marks RM ₁ and RM ₂ for alignment are provided. The reticle marks RM ₁ and RM ₂ are photoelectrically detected via the objective lenses 1A and 1B of the TTL alignment system, respectively.
The reticle stage RST can be moved two-dimensionally (in the X, Y, and θ directions) by a drive system such as a motor (not shown in FIG. 2), and its movement amount or movement position is three laser interferometers IRX. , IRY, IRθ. The rotation amount of the reticle stage RST about the Z axis (coordinate axis parallel to the optical axis AX) is obtained by the difference between the measurement values of the interferometers IRY and IRθ, and the parallel movement amount in the Y axis direction is the interference system IR.
It is obtained by the averaging value of the measured values of Y and IRθ, and the parallel movement amount in the X-axis direction is obtained by the interferometer IRX.

In this embodiment, the second TTL alignment system for detecting the mark on the wafer W via only the projection lens PL is provided separately for the X direction and the Y direction. The second TTL alignment system for the X direction is composed of the mirror 2X fixed between the reticle stage RST and the projection lens PL, the objective lens 3X, etc., and the second TTL alignment system for the Y direction is the same. It is composed of a mirror 2Y and an objective lens 3Y which are arranged in parallel.

In this embodiment, the first TTL alignment system including the objective lenses 1A and 1B is hereinafter referred to as a TTR (through the reticle) alignment system, and the objective lenses 3X,
The second TTL alignment system containing 3Y will be simply referred to as the TTL alignment system. Now, on the two sides of the wafer stage WST on which the wafer W is mounted, the movable mirror IMx that reflects the beam from the laser interferometer IFX and the beams from each of the laser interferometers IFY ₁ and IFY ₂ are reflected. The movable mirror IMy is fixed. The beam from the interferometer IFX is perpendicular to the reflecting surface of the movable mirror IMx extending in the Y direction, and the extension line of the beam is orthogonal to the extension line of the optical axis AX of the projection lens PL. The beam from the interferometer IFY ₂ is perpendicular to the reflecting surface of the movable mirror IMy extending in the X direction, and the extension line of the beam is also orthogonal to the extension line of the optical axis AX. The beam from the other interferometer IFY ₁ is perpendicular to the reflecting surface of the moving mirror IMy and parallel to the beam of the interferometer IFY ₂ .

Further, the off-axis type wafer alignment system includes a reflection prism (or mirror) 4A and an objective lens 4 fixed near the lower end of the projection lens PL.
B etc. Light receiving system 4 for wafer alignment system
C includes a conjugate optotype mark TM therein, and a CCD camera takes an image of a mark on the wafer imaged on the optotype mark plate through the prism 4A and the objective lens 4B. In the present embodiment, the optical axis of the objective lens 4B which falls on the wafer stage WST via the prism 4A and the optical axis A of the projection lens PL.
It is set such that X and X are separated from each other by a certain distance only in the X direction, and there is almost no positional difference in the Y direction.

Further, the wafer stage W of the objective lens 4B
The extension line of the optical axis falling on ST is orthogonal to each of the extension line of the beam of the interferometer IFX and the extension line of the beam of the interferometer IFY ₁ . The arrangement of such an interferometer is described in detail in JP-A-1-
It is disclosed in Japanese Patent No. 309324. On the wafer stage WST, a reference plate FP provided with two reference marks FM1 and FM2 for baseline measurement is fixed. The reference plate FP is arranged on a corner of the wafer stage WST surrounded by the two movable mirrors IMx and IMy, and forms a light-shielding layer such as chromium on the surface of a transparent material having a low expansion coefficient such as a quartz plate. A part of the reference marks FM1 and FM2 is etched. The fiducial mark FM1 is an off-axis type wafer alignment system (4A, 4A,
B, 4C) and the fiducial mark FM2 is TT
It can be detected by the R alignment system (1A, 1B) or the TTL alignment system (2X, 3X; 2Y, 3Y).

The distance between the reference marks FM1 and FM2 in the X direction is accurately made with submicron accuracy. However, if there is a residual placement error amount, that value is precisely measured in advance and used as a device constant. Shall be required. FIG. 3 is a plan view showing the arrangement of each member on the wafer stage WST. The wafer W is mounted on the wafer holder WH which can be minutely rotated on the wafer stage WST and is vacuum-sucked.
In this embodiment, the linear notch OF of the wafer W is mechanically pre-aligned so as to be parallel to the X axis and then placed on the wafer holder WH.

As shown in FIG. 3, the center of the diameter (optical axis AX) of the lower end of the lens barrel of the projection lens PL and the field of view of the objective lens 4B are arranged as close as possible. In this way, when the projection lens PL and the reference plate FP are arranged, the wafer W
Has moved most from the position directly below the projection lens PL to the diagonally lower right in the figure, so that the wafer W can be loaded and unloaded in this state. This arrangement is disclosed, for example, in JP-A-63-224326.

FIG. 4 shows the reference mark FM on the reference plate FP.
1 is a plan view showing a detailed mark arrangement of FM1 and FM2. In FIG. 4, the intersection of a straight line LX parallel to the X axis and a straight line LY ₂ parallel to the Y axis is the center of the reference mark FM ₂ , and at the time of baseline measurement, the intersection is almost the optical axis AX of the projection lens PL. Match. In this embodiment, a light emitting cross-shaped slit mark IFS is arranged on the intersection, and the illumination light having the same wavelength as the exposure light illuminates only the local region ISa including the light emission slit mark IFS from the back side of the reference plate FP. Further, reference marks FM2A and FM2B corresponding to the respective positions of the reticle marks RM ₁ and RM ₂ are provided at two symmetrical positions on the straight line LX with respect to the light emission slit mark IFS. The marks FM2A and FM2B are reference plates FP.
The upper chromium layer is etched with a cross slit, and the mark FM2A is aligned with the reticle mark RM _1, and the mark FM2B is aligned with the reticle mark RM ₂ .

Center of light-emitting slit mark IFS (intersection point)
The circular area PIF having the origin is the projection visual field area of the projection lens PL, and in the case of the present embodiment, the mark area LIMx detectable by the TTL alignment system (2X, 3X) for the X direction shown in FIG. 2 is the visual field area. It is arranged on the straight line LY ₂ in the PIF and has a TTL alignment system (2Y,
3Y) two marks LIMy and LS detectable by
My is arranged on the straight line LX of the visual field region PIF. The detailed arrangement relationship of each mark will be described later, but in this embodiment, two TTL alignment systems 1A,
1B respectively detect the reticle marks RM ₁ and RM ₂ and the reference marks FM2A and FM2B at the same time, the TTL alignment system (2X, 3X) for the X direction detects the mark LIMx and detects the mark LIMx for the Y direction. Each mark FM2 so that a TTL alignment system (2Y, 3Y) can be created.
A, FM2B, LIMx, and LIMy were arranged.

On the other hand, a straight line LY ₁ set apart from the straight line LY ₂ by a certain distance in the X direction is parallel to the Y axis, and an off-axis alignment system is provided on the intersection of the straight line LY ₁ and the straight line LX. A reference mark FM1 having a size that can be included in the visual field MIF of the objective lens 4B is formed. The mark FM1 has an X mark to enable two-dimensional alignment.
It is an aggregate of a plurality of line patterns provided in parallel with each of the Y direction and the Y direction. As is clear from the above description, in the reference plate FP, the straight line LY ₁ coincides with the center line (measurement axis) of the beam of the interferometer IFY ₁ as much as possible in the XY plane,
The straight line LY ₂ is fixed on the wafer stage WST so that it coincides with the center line (measurement axis) of the beam of the interferometer IFY ₂ as much as possible (that is, as much as possible without causing rotational deviation).

Further, two fiducial marks FM are placed at symmetrical positions on the straight line LX across the intersection of the straight lines LX and LY _1.
2C and FM2D are provided. Fiducial mark FM2
C and FM2D are cross-shaped slit patterns having exactly the same shape and size as the reference marks FM2A and FM2B, and the spacing in the X direction is also the same as the spacing between the marks FM2A and FM2B. The mark LSMx in FIG. 4 is detected by the TTL alignment system (2X, 3X) for the X direction, and is provided at the same position as the X coordinate value of the reference mark FM2B.

FIG. 5 shows the reference mark FM2 on the reference plate FP.
Only the respective mark arrangements on the side are enlarged, and the projection lens P
Emitting slit mark I at the center of projection field region PIF of L
A state in which the intersection point of FS is matched is shown. In FIG. 5, the positional relationship between the outer shape of the reticle R and the outer shape of the pattern area PA that are ideally positioned in that state is represented by a two-dot chain line. Mark LIMx for TTL alignment system,
LIMy is located at the outermost periphery of the projection visual field PIF because the mirrors 2X and 2Y at the tip of the TTL alignment system are arranged so as not to block the projection area of the pattern area PA. In this state, the reference mark FM2A can be aligned with the reticle mark RM ₁ , but the reticle mark R
As shown in FIG. 6, M ₁ (RM ₂ is the same) is composed of a double slit mark RM ₁ y extending in the X direction and a double slit mark RM ₁ x extending in the Y direction, and these marks RM ₁ y and RM ₁ x are formed as a dark part in the transparent part surrounded by the rectangular light shield SB. Among the cross-shaped slits of the reference mark FM2A, the slit extending in the X direction is sandwiched by the double slit mark RM ₁ y, and the slit extending in the Y direction is sandwiched by the double slit mark RM ₁ x, The ideal alignment has been achieved.

Here, the distance K _{1 in} the X direction between the center of the reference mark FM2A and the center of the mark LIMy and the distance K _{2 in} the X direction between the center of the light emitting slit mark IFS and the center of the mark LSMy are shown in FIG. Emitting slit mark I shown in
X when the FS scans the reticle mark RM ₁ in the Y direction
The difference is set by the offset amount ΔXk (converted value on the wafer side) in the direction. That is, K ₁ = K ₂ + Δ
Xk or is set to K ₁ = K ₂ -ΔXk,.

Further, the center position in the X direction of the mark LSMx that can be detected by the TTL alignment system for the X direction coincides with the center position in the X direction of the reference mark FM2B. This is a distance K in the X direction between the center points of the two reference marks FM2A and FM2B and the center of the light emission slit mark IFS.
₃ is a condition that holds when both are equal. The position of the mark LSMx in the Y direction is almost the same as the position of the mark LIMx in the Y direction, but strictly speaking, the emission mark IFS
And the center of the mark LIMx centered in the Y direction between the K ₄ of when the interval between the Y-direction between a center of the mark LSMx emitting mark IFS was _{K 5, K 4 = K 5} + ΔY
k or K ₄ = K ₅ −ΔYk.
(Note, K _4, K ₅ is omitted). Here, ΔYk is shown in FIG.
As shown in, the light emission slit mark IFS is an offset amount in the Y direction when the double slit mark RM ₁ x of the reticle mark RM ₁ is scanned in the X direction.

Next, the detailed structure of the TTR alignment system (1A) will be described with reference to FIG. Reticle mark R
A total reflection mirror 100 is obliquely installed at an angle of 45 ° above M ₁ , and the optical axis of the horizontally arranged objective lens 101 is made perpendicular to the reticle R. This TTR alignment system switches between a beam splitter 102, a light source 103 for generating light having an exposure wavelength, and blocking and passing of illumination light for coaxial epi-illumination. The shutter 104, the optical fiber 105 that guides the illumination light, the condenser lens 106 that condenses the illumination light from the exit end of the optical fiber 105 to uniformly illuminate the illumination field stop 107, and the illumination light from the field stop 107 is a Koehler. It has a self-illumination system composed of a lens system 109 that transmits light to the objective lens 101 under illumination conditions. Thus
The objective lens 101 illuminates only the inside of the light-shielding band SB in which the mark RM _{1 of the} reticle R is formed. As a result, the reflected light from the mark RM ₁ is reflected by the mirror 100 and the objective lens 1.
It is reflected by the beam splitter 102 via 01 and enters the imaging lens 110. The image light flux of the mark RM ₁ is divided into two by the half mirror 111, and the imaging lens 110
The images are magnified and imaged on the respective imaging surfaces of the CCD camera 112X for detecting the X direction and the CCD camera 112Y for detecting the Y direction. The CCD cameras 112X and 112Y are arranged such that the horizontal scanning lines with respect to the magnified image of the mark RM ₁ are orthogonal to each other.

At this time, the mark RM₁Of the light-shielding band SB including
Directly below the inner area, the reference mark FM2A on the reference plate FP
Is positioned, CCD112X, 112Y is the reference mark
The cross-shaped slit of FM2A is imaged as a black line. Picture
The image processing circuit 113X uses the image from the CCD camera 112X.
Digital waveform processing of the image signal, Y of the reference mark FM2A
Direction slit and reticle mark RM₁The dub
Ruslit mark RM ₁X direction with x (horizontal scan line direction)
Direction) position shift amount. The image processing circuit 113Y is C
Digital waveform processing of image signal from CD camera 112Y
Then, the slit extending in the X direction of the reference mark FM2A
And the reticle mark RM₁Double slit mark RM
₁Find the amount of misalignment with y in the Y direction (horizontal scanning line direction)
It The main control system 114 includes processing circuits 113X and 113Y.
Required reference mark FM2A and reticle mark RM₁
The amount of misalignment in the X and Y directions with and is outside the preset allowable range.
When, the drive system 115 of the reticle stage RST is
By controlling, the position of the reticle R is corrected.

The drive system 115 uses the three interferometers IRX, IRY, and IRθ shown in FIG.
The position (X, Y, θ) before correction of ST is detected, and after correction, the measurement values to be detected by the three interferometers IRX, IRY, IRθ are calculated. Therefore, drive system 11
5 is such that the measured values of the three interferometers IRX, IRY, and IRθ are the measured values that should be detected after correction,
The reticle stage RST is positioned by position servo control. Further, the main control system 114 uses the wafer stage WS.
The drive system 116 that controls the movement of T based on the measured value of the interferometers IFX, IFY ₁ , or IFY ₂ is also controlled.

Now, in the TTL alignment system 1A shown in FIG. 7, the illumination light from the light emitting mark IFS on the reference plate FP is projected onto the projection lens PL, the transparent portion inside the light blocking band SB of the reticle R, the mirror 100, A light emitting mark light receiving system for detecting through the objective lens 101, the beam splitter 102, the lens system 109 and the beam splitter 108 is provided. This light emitting mark light receiving system comprises a lens system 120 and a photoelectric sensor (photomultiplier) 121,
The light receiving surface of the photoelectric sensor 121 is the pupil E of the projection lens PL.
P and the pupil plane between the objective lens 101 and the lens system 109 are arranged to be conjugate with each other. The photoelectric sensor 121 photoelectrically detects a transmitted light amount that changes when the light emission mark IFS scans the reticle mark RM ₁ (or RM ₂ ) and outputs a photoelectric signal SSD according to the change. This photoelectric signal S
SD processing is performed by digitally sampling a signal waveform in response to an up / down pulse (for example, one pulse for each 0.02 μm moving amount) output from interferometers IFX and IFY ₂ in accordance with the scanning of wafer stage WST, This is done by storing in memory.

Next, an example of the configuration of the TTL alignment system (2Y, 3Y) in FIG. 2 will be described with reference to FIG. The TTL alignment system used in this example is He-Ne.
The red light from the laser light source 130 is used as the mark illumination light to prevent the resist layer of the wafer W from detecting the mark reflected light and preventing the resist layer from being exposed to light. Further, this TTL alignment system incorporates two alignment sensors having different mark detection principles, and the objective lens 3Y is shared so that the two alignment sensors can be selectively used. Such a configuration is disclosed in JP-A-2-272305 or JP-A-2-2.
Since it is disclosed in detail in Japanese Patent No. 83011, a brief description will be given here.

The He-Ne laser light from the laser light source 130 is split by the beam splitter 131 and reaches the shutters 132A and 132B which are complementarily opened and closed. In FIG. 8, the shutter 132A is open and the shutter 132B is closed, and the laser light is a two-beam interference alignment (L
The light enters the light transmitting system 133A of the IA) system. This light transmission system 1
33A splits the incident beam into two laser beams and outputs the two laser beams with a constant frequency difference using an acousto-optic modulator. In the case of FIG. 8, the two laser beams output from the light transmitting A 133A are arranged parallel to each other in the direction perpendicular to the paper surface of the figure. This 2
The laser beam of the book is reflected by the half mirror 134 and further divided into two by the beam splitter 135. The two laser beams reflected by the beam splitter 135 intersect on the diaphragm APA on the wafer conjugate plane by the objective lens 3Y. The two parallel laser beams that have passed through the diaphragm APA are reflected by the mirror 2Y, enter the projection lens PL, and intersect again on the wafer W or the reference plate FP. One-dimensional interference fringes are formed in the area where the two laser beams intersect, and the interference fringes flow in the pitch direction of the interference fringes at a speed according to the frequency difference between the two beams.

Therefore, the mark LIM shown in FIGS.
If y and LIMx are diffraction gratings parallel to the interference fringes, interference beat light whose intensity changes at a beat frequency corresponding to the frequency difference is generated from the diffraction grating-shaped marks LIMx and LIMy. When the pitches of the diffraction gratings of the marks LIMx and LIMy and the pitch of the interference fringes are set to have a certain fixed relationship, the interference beat light is vertically generated from the wafer W or the reference plate FP, and two light beams are projected through the projection lens PL. The mirror 2Y, the diaphragm APA, and the objective lens 3Y are returned in this order along the optical path of the transmitted light beam. The interference beat light partially passes through the beam splitter 135 and reaches the photoelectric detector 139.
The light receiving surface of the photoelectric detector 139 is the pupil plane EP of the projection lens PL.
Is almost conjugate with. Further, a plurality of photoelectric elements (photodiodes, phototransistors, etc.) are separately arranged on the light receiving surface of the photoelectric detector 139, and the interference beat light is located at the center of the photoelectric detector 139 (center of the pupil plane). The light is received by the element. The photoelectric signal becomes a sinusoidal AC signal having a frequency equal to the beat frequency and is input to the phase difference measuring circuit 140.

Further, the two light-transmitting beams transmitted through the beam splitter 135 are crossed as a parallel light flux on the transmissive reference grating plate 137 by the inverse Fourier transform lens 136. Therefore, one-dimensional interference fringes are formed on the reference grating plate 137, and the interference fringes flow in one direction at a speed according to the beat frequency. The photoelectric element 138 is a reference grid plate 137.
Either the interference light of the ± first-order diffracted light or the interference light of the 0th-order light and the second-order diffracted light that is coaxially generated from is received. The intensity of these interference lights also changes sinusoidally at a frequency equal to the beat frequency, and the photoelectric element 138 uses the AC signal having a frequency equal to the beat frequency as a reference signal for the phase difference measuring circuit 1.
Output to 40.

The phase difference measuring circuit 140 includes a photoelectric element 138.
Based on the reference signal from the reference, the phase difference Δφ (within ± 180 °) of the AC signal from the photoelectric detector 139 is obtained, and the mark LIMy on the reference plate FP corresponding to the phase difference Δφ.
The information SSB of the positional deviation amount of the (or equivalent mark on the wafer) in the Y direction, that is, the grating pitch direction is output to the main control system 114 in FIG. The resolution of the positional deviation detection is determined by the relationship between the pitch of the mark LIMy and the pitch of the interference fringes irradiated on this mark, and the resolution of the phase difference detection circuit. The phase difference detection resolution is ± 1 °. Then, when the grating pitch Pg of the mark LIMy is 8 μm and the pitch Pf of the interference fringes is Pg / 2, the positional deviation detection resolution is represented by ± (1 ° / 180 °) × (Pg / 4), It becomes ± 0.01 μm.

The main control system 114 of FIG. 7 servo-controls the drive system 116 of the wafer stage WST on the basis of the positional deviation information SSB from such a high resolution LIA TTL alignment system, and controls the reference plate FP. Mark of LIM
Wafer stage WST can be servo-locked so that y is always driven in a fixed positional relationship with reference grid plate 137.

However, when the servo lock is performed, it is sufficient that the phase difference between the signals from the photoelectric element 138 and the photoelectric detector 139 is stable at a predetermined value.
It is not necessary to convert the phase difference into the positional shift amount, and the servo lock can be performed only by detecting the change amount of the phase difference.
Another detection method of the TTL alignment system is, as disclosed in Japanese Unexamined Patent Publication No. 2-233011, cited above, for a slit-like laser spot light extending in a direction orthogonal to the mark detection direction. The signal level obtained by scanning the mark and photoelectrically detecting the diffraction and scattered light generated from the mark is used as the wafer stage W for the mark scanning.
In this method, digital sampling is performed in response to the up / down pulses from the interferometers IFX and IFY ₂ that accompany the movement of ST.

Laser step alignment (L
SA) type light transmitting system 133B includes a shutter 132A.
Is closed and the laser beam is incident when the shutter 132B is open. The incident beam is shaped by the action of the beam expander and the cylindrical lens into a slit shape in which the beam cross section of the converging point extends in one direction, and the beam splitters 134 and 135, the lens system 3Y and the mirror 2 are formed.
It is incident on the projection lens PL via Y. At this time, aperture AP
A is conjugate with the wafer surface (the surface of the reference plate FP) under the wavelength of the He-Ne laser light, and the beam is condensed here in a slit shape. In the case of the TTL alignment system shown in FIG. 8, the beam spot created by the LSA method is
It is shaped like a slit extending in the X direction at a stationary position within the projection visual field PIF. When the mark LSMy on the reference plate FP crosses the beam spot by scanning the wafer stage WST in the Y direction, the diffracted light or scattered light generated from the mark LSMy is projected onto the projection lens PL, the mirror 2Y, the objective lens 3Y, and The light reaches the photoelectric detector 139 via the beam splitter 135 and is received by peripheral photoelectric elements other than the central photoelectric element. The photoelectric signal from this photoelectric element is LS
The signal is input to the A processing circuit 142, and is digitally sampled in response to the up / down pulse signal UDP from the interferometer IFY ₂ (or IFY ₁ ) for the wafer stage WST. The processing circuit 142 stores the digitally sampled signal waveform in a memory, and performs high-speed waveform processing using digital calculation to convert the waveform in the memory from the center point of the slit-shaped spot light of the LSA method in the Y direction and the Y of the mark LSMy.
The Y coordinate value of wafer stage WST when the center point of the direction precisely matches is calculated and output as mark position information SSA. This information SSA is the main control system 114 in FIG.
And is used for drive control of drive system 116 of wafer stage WST.

In the LSA processing circuit 142, a memory for digitally sampling the photoelectric signal SSD from the photoelectric sensor 121 shown in FIG. 7 in response to the up / down pulse signal UDP and a signal waveform in the memory are processed at high speed. A projection lens PL having a circuit and a reticle mark RM ₁
The coordinate value of the wafer stage WST when the projection image by the light emitting mark IFS coincides with the reticle mark RM ₁
Is output to the main control system 114 as the projection position information SSC.

Next, the detailed configuration of the off-axis alignment system OWA will be described with reference to FIGS.
FIG. 10 shows the configuration of the off-axis alignment system OWA, IMP represents the wafer surface or the surface of the reference plate FP, and the image of the surface area located in the visual field MIF of the objective lens 4B is the prism mirror 4A and the objective. An image is formed on the index plate 4F via the lens 4B, the mirror 4C, the lens system 4D, and the half mirror 4E. The light for illuminating the surface IMP is transmitted through the half mirror 4E to the lens system 4D and the mirror 4
Surface I through C, objective lens 4B, and prism 4A
Proceed to MP. The illumination light has a band width of about 300 nm in a wavelength range where the sensitivity of the resist layer of the wafer is extremely low.

As shown in FIG. 9, the index plate 4F has index marks TMX ₁ , TMX ₂ , and TM formed on the transparent glass by a plurality of (for example, four) line patterns formed by a light-shielding portion.
Y ₁ and TMY ₂ are formed. FIG. 10 shows a state where the intersection of the straight lines LX and LY ₁ set on the reference plate FP and the center of the index plate 4F coincide with each other. Index mark TM
X ₁ and TMX ₂ are the reference marks FM ₁ on the reference plate FP.
The index mark TMY ₁ , which is provided so as to sandwich in the direction,
TMY ₂ is provided so as to sandwich the reference mark FM ₁ in the Y direction.

Now, each index mark on the optotype plate 4F and the reference mark
Semi-mark FM₁Image of (or mark on wafer)
Two via image forming lens 4G and half mirror 4H
The images are enlarged and formed on the CCD cameras 4X and 4Y. CCD
The imaging area of the camera 4X is the area in FIG. 9 on the optotype plate 4F.
Area 40X is set, and the imaging area of the CCD camera 4Y is
It is set in the area 40Y. And water from CCD camera 4X
The flat scan line is the index mark TMX₁, TMX₂Line of
The CCD camera 4Y is set in the X direction orthogonal to the turn.
The horizontal scanning line of is the index mark TMY ₁, TMY₂Line of
It is defined in the Y direction orthogonal to the pattern. CCD camera
The image signal from each of 4X and 4Y has a signal level for each pixel.
Digital sampling circuit, multiple horizontal scan lines
Time to convert and average the image signal (digital value) obtained for each
Road, index mark TM and reference mark FM₁X direction with, Y
Waveform processing of a circuit that calculates each misregistration amount in the direction at high speed
The information of the amount of positional deviation processed by the circuit is the main control of FIG.
The information SSE is sent to the system 114.

In the case of this embodiment, the detection center point of the off-axis alignment system OWA is, for example, the X of two index marks TMX ₁ and TMX _{2 in} the X direction.
In the Y direction, it is a bisecting point of the two index marks TMY ₁ and TMY _{2 in} the Y direction. However, in some cases, two index marks TMX ₁ , TM
Of X ₂ , for example, the center point in the X direction of only the mark TMX ₂ may be the detection center.

FIG. 11 is an enlarged view of the reference mark FM ₁ formed on the reference plate FP, in which a plurality of line patterns extending in the Y direction are arranged at a constant pitch in the X direction.
A plurality of line patterns extending in the X direction are arranged in the Y direction at a constant pitch to form a two-dimensional pattern. When detecting the position of the reference mark FM _{1 in} the X direction, C
The image signal from the CD camera 4X is analyzed by the waveform processing circuit, and the average position of each detection position (pixel position) of the plurality of line patterns arranged in the X direction is set as the position of the reference mark FM1 in the X direction, and the index mark TMX ₁ , TMX _{2 from} the center position may be obtained. Reference mark F for Y direction
The CCD camera 4Y also similarly detects M _{1 and} the amount of positional deviation.

By the way, as described above with reference to FIG.
The arrangement of various marks on the reference plate FP detected by the TR alignment system and the TTL alignment system is defined in a fixed positional relationship.
2 will be described. FIG. 12 is an enlarged view of each mark located on the straight line LX, the mark LIMy is a diffraction grating in which grating elements are arranged at a constant pitch (for example, 8 μm) in the Y direction, and the mark LSMy is shown enlarged in a circle. Pitch PS of a minute square dot pattern in the X direction
It is a two-dimensional lattice pattern in which xs are arranged and a pitch PSy is arranged in the Y direction. Mark LSMy is LS for Y direction
The beam spot is detected by the beam spot of the A-type TTL alignment system, the beam spot extends in a slit shape in the X direction, and the beam width in the Y direction is substantially equal to the dimension of the dot pattern in the Y direction. The pitch PSx in the X direction contributes to the generation of diffracted light at the time of mark detection, and the pitch PSy in the Y direction is for arranging a plurality of lattice marks in the Y direction to form a multi-mark. Therefore, when multi-marking is not required, only an example of dot pattern groups arranged on the straight line LX is required.

The pitch PSx in the X direction is uniquely determined by the wavelength of the beam spot and the required diffraction angle of the first-order diffracted light, but the pitch PSy in the Y direction is PSx.
It should be equal to or greater than. As described with reference to FIG. 5, the distance K ₁ between the center point of the mark LIMy in the X direction and the center point of the reference mark FM2A in the X direction, the center point of the emission mark IFS in the X direction and the mark L
The distance K _{2 from} the center point of SMy in the X direction is K ₁ = K ₂
There is a relationship of ± ΔXk. This condition is that the center of the mark detection area (irradiation area of interference fringes) of the LIA type TTL alignment system and the center point (beam spot) of the mark detection of the LSA type TTL alignment system in this embodiment are substantially the same. However, it is not necessarily limited to the above conditions.

The TTL alignment system described with reference to FIG. 8 has the same structure for the X direction, and position information in the X direction of each mark is sent to the main control system 114. Next, the baseline measurement and various alignment operations by the apparatus (stepper) of this embodiment will be described.
Before that, the correction for the mounting error of the reference plate FP on the wafer stage WST will be described. Of the mounting errors of the reference plate FP, what affects the final accuracy is the residual rotation error of the reference plate FP in the coordinate system XY.

Conventionally, it has also been proposed to fix a reference plate of this type on a wafer stage via a metal which can be finely adjusted with a set screw or the like (for example, Japanese Patent Laid-Open No. 55-55).
No. 135831). However, considering the variation with time, the method of fixing the reference plate through the fine adjustment mechanism would be extremely disadvantageous in terms of accuracy stabilization. For this reason, it is desirable that the reference plate FP be fixed on the wafer stage so as not to be finely moved (on the order of nm).

Whichever fixing method is used, in this embodiment
Is to obtain the residual rotation error amount of the reference plate FP in advance.
I chose The residual rotation error referred to here is, for example, as shown in FIG.
And the straight line LX set on the reference plate FP shown in FIG.
It means the degree of parallelism with the reflecting surface of the movable mirror IMy. Wafer
Interferometer IF controls all coordinate positions of stage WST.
X, IFY₁(Or IFY ₂) Is the standard,
Each reflecting surface of the movable mirrors IMx and IMy is a reference for coordinate position measurement.
It can be said that Therefore, with the reflecting surface of the moving mirror IMy
The parallelism with the straight line LX on the reference plate FP becomes a problem. Also
As a mounting error, it is orthogonal to the reflecting surface of the movable mirror IMy.
Between the Y direction and the X direction orthogonal to the reflecting surface of the movable mirror IMx.
Regarding the parallel displacement in each direction, the wafer stage WST
Since it can be handled by positioning, there is almost no problem.

The residual rotation error of the reference plate FP is shown in the figure.
It may be determined by self-measurement with a stepper
Alternatively, it may be obtained by trial baking using a wafer. This
Here, a method based on self-measurement will be described as an example. Figure
Of the stepper alignment sensors shown, Y
Two interferometers IFY that have the direction of mark detection
₁, IFY₂Satisfies the Abbe condition for either one of
What you do is an off-axis wafer alignment
In the present embodiment, the interferometer IFY
₁Of the alignment system OWA in the Y direction
The mark detection function shall be used. First, on the reference plate FP
Of the two fiducial marks FM2A and FM2D respectively in the Y direction
Coordinate position is measured by off-axis alignment system OWA.
Measure. Therefore, as shown in FIG.
Turn off the bar mark extending in the X direction of the mark FM2D.
View of objective lens 4B of axis alignment system OWA
The index mark TMY shown in FIG. 9 located in the field₁, T
MY₂The amount of positional deviation in the Y direction between that time,
Index mark TMY₁, TMY₂To either one of
The bar mark extending in the X direction of the reference mark FM2D is aligned.
May be arranged. In addition, in FIG.
The moving mirror IMy and the straight line LX on the reference plate FP are rotated by θf.
It is shown as exaggerated and is exaggerated.

In any case, the index marks TMY ₁ , TM
A deviation amount ΔYFd in the Y direction of the reference mark FM2D based on Y ₂ is detected based on the image signal from the CCD camera 4Y in FIG. The positional deviation amount is obtained as the information SSE input to the main control system 114 in FIG. 7. At the same time, interferometer IFY _1, IFY ₂ measurements YA1, YA2 when detects the reference mark FM2D the objective lens 4B is stored in the main control system 114.

Next, the wafer stage WST is moved in the X direction by a fixed amount Lfp, and the bar mark extending in the X direction of the reference mark FM2A is moved relative to the index marks TMY ₁ and TMY ₂ of the off-axis alignment system OWA. Position. The state at this time is shown in FIG. The fixed amount Lfp at that time is set equal to the design distance in the X direction between the reference marks FM2A and FM2D.

Similarly, the shift amount ΔYFa of the reference mark FM2A in the Y direction and the measured values YB ₁ and YB ₂ of the interferometers IFY ₁ and IFY ₂ are obtained. With the above operation, the measurement work is completed, and thereafter, the residual rotation error θf is obtained by calculation. First, the wafer stage WST is moved in the X direction by a fixed amount L.
If yawing does not occur when moved by fp, the rotation error θf ′ can be approximately calculated by the following equation.

Θf ′ = (YA ₁ −ΔYFd) − (YB ₁ −ΔYFa) / Lfp = (YA ₁ −YB ₁ ) + (ΔYFa−ΔYFd) / Lfp (1) However, when yawing occurs Is a minute rotation error Δ of the wafer stage WST due to the yawing.
This means that θy is included in the equation (1). Therefore, the true residual rotation error θf is given by the following equation.

Θf = θf′−Δθy (2) The rotational error Δθy due to yawing is obtained by Δθy≈ (YA ₁ −YA ₂ ) / LB− (YB ₁ −YB ₂ ) / LB (3). Here, LB is two interferometers IFY ₁ , I
It is the distance in the X direction between the respective measurement axes of FY ₂ .

If the same measurement is performed using the reference mark FM2C instead of the reference mark FM2D, the design distance in the X direction between the reference marks FM2A and FM2C is the interval LB in the X direction between the interferometers IFY ₁ and IFY _2. Therefore, the movement of the wafer stage WST by a fixed amount Lfp is also Lfp = LB. Therefore, the reference mark FM2
When A and FM2C (or FM2B and FM2D) are used, the equation (3) is as follows.

Δθy≈ (YA ₁ −YB ₁ ) + (YB ₂ −YA ₂ ) / Lfp (4) Therefore, from equations (1), (2), and (4), the residual rotation error θf is θf≈ θf′−Δθy = (ΔYFa−ΔYFd) − (YB ₂ −YA ₂ ) / Lfp (5)

That is, of the two reference marks used for measurement
Two interferometers IFY with intervals in the X direction₁, IFY₂X
When the distance is equal to the direction, the interferometer I used as a reference
FY ₁Measured value (YA₁, YB₁) Without monitoring
Will be good too. As described above, the reference plate FP is transferred.
The residual rotation error θf for the moving mirror IMy is calculated.
Then, this value is stored in the main control system 114. The reference plate F
There are four reference marks along the straight line LX on P.
Therefore, using any two reference marks among them, residual rotation
The error may be obtained and the average value thereof may be taken. example
Rotation obtained by fiducial marks FM2A and FM2C
Error θf₁And the fiducial marks FM2B and FM2D
Rotation error θf₂And the average value (θf₁+ Θf
 ₂) / 2 is the residual rotation error of the reference plate FP. Furthermore
On the straight line LX, marks LIMy, LSMy, IF
S, FM₁Is provided, any of these
Detects two or more by off-axis alignment system OWA
Then, the mark position in the Y direction may be measured. In any
The distance in the X direction between the two marks to be measured is
It is desirable to be as large as possible to secure the degree.

The method of measuring the residual rotation error by self-measurement described above is an example, and other methods by self-measurement are also possible. This will be described later in the operation sequence. Further, although the above measurement method is for obtaining θf, since θf is detected as an offset during wafer alignment by the off-axis alignment system OWA, a method for obtaining θf by examining the vernier after exposure can be considered. That is,
Overlay exposure is performed on the test wafer using the off-axis alignment system OWA, and the residual mounting error θ is read by reading the vernier in the resist pattern after development for checking overlay accuracy in the X and Y directions.
f can be obtained.

Next, the operation of the baseline measurement by the apparatus of this embodiment will be described. The operation described here is a typical operation, and some modified operations will be collectively described later. 14 and 15 are flowcharts for explaining a typical sequence, and the sequence is mainly controlled by the main control system 114.

First, the reticle R stored in a predetermined storage is automatically or manually conveyed and loaded on the reticle stage RST depending only on mechanical positioning accuracy and delivery accuracy (step 500). In this case, loading accuracy of reticle R, double slit mark RM ₁ x and the size of the window region of the reticle mark shown in FIG. 6 (inner light-shielding band SB) of about 5mm square, R
If the length of M ₂ y is set to about 4 mm, it is desirable that it is ± 2 mm or less.

Next, the main control system 114 preliminarily roughly aligns the position of the reticle R so that the marks RM ₁ and RM ₂ of the reticle R can be normally detected by the TTL alignment systems 1A and 1B. Perform reticle search. This reticle search includes step 5 in FIG.
04 and 506, the SRA method and the IFS method 2
In step 502, which mode is selected is selected. Pre-alignment by the IFS method in step 504 refers to reticle stage R as shown in FIG.
With the position of ST fixed, the wafer stage WST is moved in a large stroke (for example, several mm) in the X and Y directions so that the light emission mark IFS searches for a position where the reticle mark RM ₁ or RM ₂ is likely to exist. Position the reticle marks RM ₁ and RM ₂ with the interferometers IFX and IF.
This is a method in which the reticle stage RST is roughly detected based on Y ₂ , the amount of deviation of the detected position from the designed position is obtained, and the reticle stage RST is finely moved by relying on the interferometers IRX, IRY, and IRθ for the reticle stage RST. .

On the other hand, the pre-alignment by the SRA method in step 506 is executed as follows.
The plain surface of the reference plate FP is placed directly below the position where the reticle marks RM ₁ and RM ₂ are likely to exist, and in that state the TTR
CCD camera 11 using alignment systems 1A and 1B
The pattern on the reticle R is imaged by 2X and 112Y (FIG. 7), and the image signal waveform corresponding to the horizontal scanning line in one screen is captured in the memory. Next, the reticle stage RST is moved in the X direction or the Y direction by a certain amount by the drive system 115 based on the measurement values of the interferometers IRX, IRY, IRθ, and then the image signal waveform of the second screen is captured from the CCD camera. Connect to the signal waveform on the first screen. After that, the combined image signal waveforms are analyzed, the respective positions of the reticle marks RM ₁ and RM ₂ are calculated, the amount of deviation from the designed position is calculated, and then the position of the reticle stage RST is moved.

Regardless of which search mode is selected,
Mark R of Le R₁, RM₂Each center of two TTR
CCD sensors provided in each of the alignment systems 1A and 1B
About a few μm in the center of the imaging area of the camera 112X, 112Y
Pre-alignment can be performed with a degree of accuracy. Next, the main control system 11
4 is the reticle alignment operation from step 508
Before entering, but before that, two fiducial marks FM2A, FM
2B are designed in the visual field PIF of the projection lens PL.
Drive system 116 so that it comes to the position, interferometers IFX, IFY ₂
(Or IFY₁) Is controlled according to the measured value of
Position the WST. Wafer stage WST
When placed, the fiducial marks FM2A, (FM2B)
Reticle mark RM₁(RM₂) Was generally aligned with
In this state, the images are taken by the CCD cameras 112X and 112Y.
At this stage, the processing circuits 113X and 113Y in FIG. 7 are operated.
Then, the reticle mark R for the reference mark FM2A
M₁Of X in the X and Y directions (ΔXR₁, ΔYR₁)
And the reticle mark RM for the reference mark FM2B₂
Of X in the X and Y directions (ΔXR₂, ΔYR₂) And
measure.

Next, at step 510, each positional deviation amount is allowed.
It is judged whether it is within the acceptable value, and if it is outside the allowable value
If so, go to step 512. At this time, two reticles
Le Mark RM₁, RM₂As apparent from the shape and arrangement of
The alignment of the reticle R in the X direction is
Each reticle for each center point of FM2A and FM2B
RM₁, RM₂Of the center points of the reticle center CC
When it is deviating toward the right, when it is deviating in the opposite direction
If it is negative, the shift amount in the X direction ΔXR₁And ΔXR ₂The polarity of
Is achieved by making the absolute value equal to the absolute value.

Similarly, the reticle R is aligned in the Y and θ directions.
As for the liment, each reticle mark RM₁, RM₂Heart of
Positive when the point shifts in the positive direction of the Y axis of the stationary coordinate system
And the deviation amount in the Y direction ΔYR₁And ΔYR₂Polarity and absolute value of
This is achieved by making and equal. Θ direction of reticle R
The amount of deviation ΔθR (in the rotation direction) is determined by the reticle mark R
M ₁, RM₂Let Lrm be the interval in the X direction of
Deviation amount ΔYR₁, ΔYR₂From (actual size on reticle)
It is calculated by the following formula.

ΔθR = sin ⁻¹ ((ΔYR ₁ −ΔYR ₂ ) / Lrm) ≈ (ΔYR ₁ −ΔYR ₂ ) / Lrm (6) However, since the interval Lrm is constant for any reticle, The evaluation of the deviation amount of the reticle R is simply Δ
It is only necessary to find out the magnitude of the absolute value of YR ₁ -ΔYR ₂ .
From the above, when the deviation amount of the reticle R in the X, Y, and θ directions is larger than the allowable value, the reticle stage RST is finely moved in step 512. At this time, X direction, Y
Reticle stage RS
The amount of deviation (ΔXR ₁ , ΔYR
₁ ), (ΔXR ₂ , ΔYR ₂ ), the position of the reticle stage RST is calculated by three interferometers IR.
Finely move to the position to be corrected while monitoring with X, IRY, and IRθ. This drive method is called a so-called open control method, and if the control accuracy of the drive system 115 and the positioning accuracy of the reticle stage RST are sufficiently high and thus stable, one position deviation measurement (step 508) and 1 The reticle R can be accurately aligned with the target position by only performing the position correction once (step 512). However, since it is necessary to confirm whether or not the target position is accurately aligned by the position correction, the main control system 114
Repeats the operation from step 508 again.

By the above steps 508 to 510,
The reticle R has two reference marks FM2 on the reference plate FP.
It means that the A and FM2B are aligned with the designed coordinate position. Now, although the reticle R is aligned with the reference marks FM2A and FM2B in this way, the reference plate FP has a constant residual rotation error θf with respect to the reflecting surface of the movable mirror as described above with reference to FIG. Therefore, strictly speaking, the reticle R after alignment is rotated by θf with respect to the reflecting surface of the movable mirror.

Then, in step 512, the reticle stage
Reticle mark RM when finely moving RST₁The standard of
The alignment position with respect to the FM2A
Ox ₁, ΔOy₁) Make sure you have an offset
Curumark RM₂To the reference mark FM2B of
Ment position is further (ΔOx₂, ΔOy₂) Offset
Set to have Where X-direction offset Δ
Ox₁, ΔOx₂Can be zero, and the offset in the Y direction
To ΔOy₁, ΔOy₂Is defined as follows.

ΔOy ₁ = Lrm · θf / 2 ΔOy ₂ = −Lrm · θf / 2 Therefore, when the reticle R is aligned with the reference plate FP as a reference, the final condition considering the mounting error (θf) of the reference plate FP is: , The amount of positional deviation when each mark is detected by the TTR alignment system is as follows.

X direction: ΔXR ₁ = ΔXR ₂ → 0 Y direction: ΔYR ₁ → ΔORy ₁ , ΔYR ₂ → ΔORy ₂ To set the final alignment position with these offsets, the reticle interferometers IRX, IRY, An open control method using IRθ may be used, or a positional deviation amount obtained from the processing circuits 113X and 113Y of the TTR alignment system may be used as a deviation signal, and the final positional deviation amount may be used as a target value to control the reticle stage RST in a closed loop. You may drive with.

By the way, the residual rotation error θf of the reference plate FP
In addition to the method described above with reference to FIG. 13, there is a method using the reticle marks RM ₁ and RM ₂ and a TTR alignment system. Before the step 508 in the flowchart of FIG. 14, the method is performed by using the TTR method such that the reticle marks RM ₁ and RM ₂ and the reference marks FM2C and FM2D are used.
This can be done by adding the step of aligning and.

That is, when the pre-alignment of the reticle R is completed in step 504 or 506, the reticle R is set with an accuracy within ± several μm, so the reticle marks RM ₁ and RM ₂ are used as temporary reference points. The coordinate positions of the reference marks FM2C and FM2D are measured. At this time, since the reticle marks RM ₁ and RM ₂ are located almost symmetrically with respect to the optical axis AX of the projection lens PL in the X direction, the Y of the reticle mark RM ₁ detected by the TTR alignment system 1A and the reference mark FM2C is detected.
Direction displacement amount ΔY2C and TTL alignment system 1
Strictly speaking, each of the positional deviation amount ΔY2D between the reticle mark RM ₂ and the reference mark FM2D detected by B in the Y direction includes an Abbe error. However, the arithmetic mean value ΔYRC [(ΔY2C +) that represents the amount of deviation in the Y direction between the center point of the reticle R and the center point of the reference mark FM1.
When ΔY2D) / 2] is obtained, the Abbe's error component is canceled by the arithmetic mean. Therefore, if the measurement value Yfm ₁ of the interferometer IFY ₂ when the deviation amounts ΔY2C and ΔY2D are detected by the TTR alignment systems 1A and 1B is stored and the value of YF ₁ = Yfm ₁ −ΔYRC is obtained, the reticle R The Y coordinate value YF ₁ of the center point of the reference mark FM1 (the midpoint in the X direction between the reference marks FM2C and FM2D) with the center point as the reference is obtained.

Regarding the X direction, based on the shift amount ΔX2C detected by the TTR alignment system 1A and the shift amount ΔX2D detected by the TTR alignment system 1B, the directionality (positive / negative) of the shift is considered, The deviation amount ΔXRC [(ΔX2C−ΔX2D) / 2] between the center point of the reticle R and the center point of the reference mark FM1 in the X direction may be calculated. At this time, the X coordinate position of wafer stage WST is detected as Yfm ₁ by interferometer IFX, and XF
By calculating ₁ = Yfm ₁ −ΔXRC, X of the center point of the reference mark FM1 with the center point of the reticle R as a reference is calculated.
The coordinate value XF ₁ is obtained.

The coordinate value (XF
₁ and XF ₂ ) are the reference marks FM1 from the reflecting surfaces of the interferometers IFX ₂ and the moving mirrors IMy and IMx based on IFX.
The value includes the distance to the center point of. Then, wafer stage WST is moved to step 508 in FIG.
To execute. As described above, in step 508,
First, the reticle marks RM ₁ and RM ₂ and the reference marks FM 2A and FM 2B by the TTR alignment systems 1A and 1B.
The respective positional deviation amounts of and are calculated. Reticle mark R
The amount of displacement of the reference mark FM2A with respect to M ₁ is (ΔX
R ₁ , ΔYR ₁ ) and the positional deviation amount of the reference mark FM2B with respect to the reticle mark RM ₂ is (ΔXR ₂ , ΔY
R ₂ ). At this time, the fiducial marks FM2A and FM2B are not required at step 508 in FIG.
The coordinate values (Xfm ₂ , Yfm ₂ ) of wafer stage WST when detected by the R alignment system are interferometers IFX, I.
Memorize from FY ₂ .

From the above measurement results, the main control system 114 determines the coordinate values (XF ₂ , YF ₂ ) of the center point of the reference mark FM2 (the midpoint in the X direction of the marks FM2A and FM2B) with the center point of the reticle R as the reference. ) Is calculated as follows. XF ₂ = Xfm ₂ − (ΔXR ₁ −ΔXR ₂ ) / 2 YF ₂ = Yfm ₂ − (ΔYR ₁ −ΔYR ₂ ) / 2 The coordinate values (XF ₂ and YF ₂ ) are interferometers IFY ₂ and IF.
From each reflecting surface of the movable mirrors IMy and IMx with X as a reference,
The value includes the distance to the center point of the fiducial mark FM2.

Therefore, the wafer stage WST is moved from the detection position of the reference marks FM2C and FM2D to the reference mark FM2.
The mounting error θf ′ of the reference plate FP including the yawing amount Δθy when the reference plate FP is moved to the detection positions of A and FM2B is
It is calculated by the following formula. θf ′ ≈ YF ₁ −YF ₂ / XF ₁ −XF ₂ (7) In this case, the measured values of IFY ₁ of the two interferometers and the interferometer I
Since the amount of change in the difference from the measured value of FY ₂ corresponds to the yawing amount Δθy, the true mounting error θf can be obtained by correcting the yawing amount Δθy.

While the above calculation is being performed, the main control system 114 executes the following steps 510 and 512.
That is, as described above, determining the mounting error θf of the reference plate FP in the sequence of FIG. 14 is performed by calculating the positional deviation amount (ΔXR ₁ , ΔY) initially measured in step 508.
Only R ₁ ), (ΔXR ₂ , ΔYR ₂ ) are required. Next, the main control system 114 executes the operation from step 516 shown in FIG. In step 516, the position of the reference plate FP is set to the interferometers IFX and IFY for the wafer stage WST.
_The servo lock is selected based on the measurement value of ₂ (or IFY ₁ ) or the LIA method of the TTL alignment system is selected. If the servo lock using the interferometer is selected, the process proceeds to step 518, the coordinate value of the wafer stage WST at the time when the reticle alignment is achieved is stored, and the interferometers IFX, IFX are stored.
Drive system 116 of wafer stage WST is set so that the measured value of FY ₂ (or IFY ₁ ) always matches the stored value.
Servo control. When the LIA type servo lock is selected, the process proceeds to step 520, and the shutters 132A and 132B shown in FIG. 8 are set to the states shown in the figure,
Interference fringes are irradiated onto the marks LIMx and LIMy on the reference plate FP. Then, the phase difference measuring circuit 140 servo-controls wafer stage WST so that the phase difference with respect to the reference signal in each of the X direction and the Y direction is always a predetermined value. In the case of the LIA method, the two marks LIMx and LIMy on the reference plate FP are aligned with the reference lattice plate 138 fixed inside the TTL alignment system.

The servo lock of wafer stage WST is
Interferometer IFX, IFY₂(Or IFY₁)
Based on the TTL alignment system even in the interferometer mode
It is possible with LIA mode with almost the same accuracy, but experiment
According to simulations and simulations, LIA mode interferes better
It is confirmed to be more stable than the meter mode. one
Generally, a moving straw in the X and Y directions of the wafer stage WST
Is larger than the diameter of the wafer, for example, 30 cm or more
Is necessary. Therefore, the interferometers IFX and IFY ₂from
The optical path length of the laser beam exposed to the atmosphere is several + cm or less
Local refractive index fluctuations are generated in the air above and between them.
Even if the wafer stage WST is strictly stationary
However, the value of the counter inside the interferometer is 1 / 100μ
It varies in the order of m to 1/10 μm. Therefore the interferometer
If the servo is locked so that the count value remains constant,
The position of wafer stage WST changes due to the fluctuation of the folding rate.
For example, it may move slightly within a range of ± 0.08 μm.
It Refractive index fluctuations are due to the light of the laser beam from the interferometer.
A mass of air with a temperature difference slowly passed through the passage.
It occurs when there is an accident. This interferometer for wafer stage
Is more stable than the LIA method because it has environmental disadvantages.
May lack sexuality. The beam used in the LIA method is
Cover to prevent exposure to the atmosphere
Can be provided, and the exposure of the beam is avoided
Space between reticle and projection lens, and projection lens and wafer
Since the space with c is only a few cm at most, the refractive index
Fluctuations are unlikely to occur.

From the above, if the position servo of the reference plate FB (wafer stage WST) can be performed by using the TTL alignment system while the reference marks FM2A, FM2B are detected by the TTR alignment system, that position should be minimized. It is preferable to do so. Next, the main control system 114
In step 522, reference mark detection using the TTR alignment system and the off-axis alignment system at the same time is performed. Generally, when the reticle stage RST is finely moved to the target position and alignment is achieved in the previous step 510, the reticle stage RST is fixed by vacuum suction or the like to the column side which is the base thereof. At the time of this suction, the reticle stage RST may be laterally displaced by a minute amount. This lateral shift is a slight one, but it is one of the error factors in baseline management, and it is necessary to fully recognize it. The recognition is CC of TTR alignment system.
Step 5 again using D cameras 112X and 112Y
08 measurement operation, or interferometers IRX, I
This can be done by monitoring the amount of change in the measured values of RY and IRθ from the time when the reticle alignment was achieved. However, in the present embodiment, since the lateral displacement is also managed as the baseline amount, only the lateral displacement amount need not be individually calculated.

By the way, at the stage of step 522, already
Within the detection area of off-axis alignment system OWA
The reference mark FM1 on the reference plate FP is located. There
In the main control system 114, the off-axis control shown in FIG.
Targets using CCD cameras 4X and 4Y of the alignment system
X of the optotype mark TM and the reference mark FM1 in the plate 4F,
The positional deviation amount (ΔXF, ΔYF) in the Y direction is actually measured on the wafer.
Calculate as the size. At the same time CCD of TTR alignment system
Reticle mark R using cameras 112X and 112Y
M₁Between the reference mark FM2A and the reference mark FM2A (ΔXR₁,
ΔYR₁) And reticle mark RM₂And fiducial mark FM
Positional deviation from 2B (ΔXR ₂, ΔYR₂) And the wafer
Measure as the actual size of the side. At this time, the TTR system is also off
・ The Axis system also uses a CCD camera as a photoelectric sensor.
Image signal corresponding to the captured mark image
Capture waveforms to memory Match timing as much as possible
In this way, the processing circuits 113X and 113Y are controlled. Was
However, a CCD camera generally outputs an image signal for one frame.
Outputs every 1/30 second, so TTR method and off-act
Strictly capture the image signal with the cis method in frame units
There is no need to sync to. That is, the signals are almost the same at the same time.
You only have to import the data, for example within a few seconds (preferably
1 second or less) will be sufficient. The reference plate FP
If the position of is locked by an interferometer, TTR
Image signal waveform acquisition and off-axis method
The capture of the image signal waveform at
More than the time it takes for the wafer stage position to change due to
Need to have short intervals.

Next, the main control system 114 releases the servo lock of the wafer stage WST in step 524 and moves to the operation of step 526, and detects each mark on the reference plate FP by using the LSA method and the IFS method at the same time. Therefore, the movement (scanning) of wafer stage WST is started. This step 526 moves the wafer stage WST so that the light emission slit mark IFS scans the reticle mark RM ₁ two-dimensionally as described above with reference to FIGS. 6 and 5. The light emission slit mark IFS is positioned so as to have the positional relationship shown in FIG. At this time, the slit-shaped beam spot extending in the X direction by the LSA method of the TTL alignment system is displaced in the Y direction with respect to the mark LSMy on the reference plate FP. When the wafer stage WST is scanned in the Y direction from that state, the photoelectric signal from the LSA type photoelectric detector 139 and the photoelectric signal SS from the IFS type photoelectric element 121 are detected.
Both waveforms of D and D are as shown in FIG. FIG.
(A) is a detection waveform of the mark LSMy captured on the memory by the LSA method.
Since My has five diffraction grating patterns, five peaks are generated on the signal waveform. The processing circuit 142 shown in FIG. 8 obtains the barycentric position of each of the five peak waveforms, and calculates the average value as the Y coordinate position YLs of the mark LSMy.

On the other hand, the signal SSD obtained by the IFS system
16B, the reticle mark RM ₁
The double slit mark RM ₁ y of FIG. The processing circuit 142 obtains the center point of each of the two bottom waveforms in the signal waveform of FIG. 16B, and calculates the center point as the center coordinate position YIf in the Y direction of the projected image of the double slit mark RM ₁ y. .

Similarly, the light emission slit mark IFS is moved as shown by the arrow in the X direction in FIG. 6 to scan the double slit mark RM ₁ x of the reticle mark RM ₁ . At this time, the slit-shaped spot by the LSA method of the TTL alignment system for the X direction is the mark LS on the reference plate FP.
Simultaneous scanning by Mx gives the same waveform as in FIG. At this time, the X coordinate value of the mark LSMx detected by the LSA method for the X direction is XLs, and the LFS
Double slit mark RM ₁ x detected by the method
The X coordinate value of is XIf.

As shown in FIG. 16, coordinate positions TLs and X
The difference from If is the baseline amount in the Y direction between the detection center point of the TTL alignment system by the LSA method for the Y direction and the projection point of the center CC of the reticle R. Next, main control system 1
In step 528, 14 performs calculation for obtaining the baseline amount. The parameters required for this calculation are shown in Fig. 1.
As shown in the table in FIG. 7, it is divided into the measured value measured and a constant value predetermined in design. In the actual measurement values in the table of FIG. 17, “TTR-A” refers to the TTR alignment system 1A in FIG. 2, and “TTR-B” refers to the TTR alignment system 1B. Further, the actual measurement value by each alignment system is displayed by dividing the positional deviation amount or the mark position in the X direction and the Y direction. On the other hand, as the design constant value, the center point of the reference mark FM1 and the reference mark FM
2A distances (ΔXfa, ΔYfa) in the X and Y directions, X and Y of the center point of the reference mark FM1 and the reference mark FM2B.
The respective distances (ΔXfb, ΔYfb) in the direction are precisely measured in advance with reference to the straight line LX and stored.

The main control system 114 has constant values ΔXfa and ΔXfa.
Based on, the X direction distance LF between the bisector of the line segment connecting the central points of the reference marks FM2A and FM2B and the central point of the reference mark FM1 is calculated. LF = (ΔXfa, ΔXfb) / 2 (8) Next, the main control system 114 causes the X-direction deviation amount ΔXR ₁ obtained by TTR-A and the X-direction deviation amount ΔX obtained by TTR-B.
The half of the difference ΔXcc from R ₂ is obtained as the dimension on the wafer side.

ΔXcc = (ΔXR ₁ −ΔXR ₂ ) / 2 (9) where ΔXR ₁ and ΔXR ₂ are reticle marks RM ₁ and R
It is assumed that when M ₂ is displaced in the direction of the reticle center with respect to each of the reference marks FM2A and FM2B, it is a positive value, and when it is displaced in the reverse direction, it is a negative value. When the value ΔXcc obtained by the equation (2) is zero, the projection point of the center CC of the reticle R is precisely aligned with the bisector in the X direction of the center points of the two reference marks FM2A and FM2B. Will be.

Next, the main control system 114 determines the center CC of the reticle R based on the measured value ΔXF and the calculated values LF and ΔXcc.
On the XY coordinate plane and the center point in the X direction of the index plate 4F of the off-axis alignment system OWA (the bisecting point between the index marks TMX ₁ and TMX ₂ ) on the XY coordinate plane. The distance BLOx in the X direction from the projection point is calculated as the X direction baseline amount for the off-axis alignment system OWA. BLOx = LF−ΔXcc−ΔXF (10) where ΔXF is the reference mark FM1 with respect to the bisector of the index marks TMX ₁ and TMX _{2 in} the X direction.
It is assumed that a positive value is obtained when the shift is detected in the direction of L (reference marks FM2A, FM2B), and a negative value is detected when the shift is detected in the opposite direction.

Next, main control system 114, based on the measured values ΔYR ₁ and ΔYR ₂ , a line segment connecting the projection point of center point CC of reticle R and the center points of fiducial marks FM2A and FM2B. A deviation amount ΔYcc in the Y direction from the bisector (which is on the straight line LY ₂ ) is obtained. ΔYcc = (ΔYR ₁ −ΔYR ₂ ) / 2 (11) where ΔYR ₁ and ΔYR ₂ are reticle marks R
Reference marks FM2A, F corresponding to M ₁ and RM ₂ , respectively
With respect to M2B, a positive value is taken when the Y direction is deviated in the positive direction of FIG. 4 (upward in the plane of FIG. 4), and a negative value is taken when deviated in the reverse direction. This shift amount Ycc is determined by the projection point of the center CC of the reticle R and the reference marks FM2A, FM.
It becomes zero when the bisector of the line segment connecting the center points of 2B exactly coincides.

Further, the main control system 114 has a constant value ΔYfa,
Based on ΔYfb, a deviation amount ΔYf ₂ in the Y direction between the bisector of the line segment connecting the center points of the reference marks FM2A and FM2B and the center point of the reference mark FM1 is obtained. ΔYf ₂ = (ΔYfa−ΔYfb) / 2 (12) Based on the above calculated values ΔYcc, ΔYf ₂ and the measured value ΔYF, the main control system 114 determines the projection point of the center CC of the reticle R and the off-axis. -Center point in the Y direction of the index plate 4F of the alignment system OWA (index marks TMX ₁ and TMX ₂
Distance BLOy in the Y direction from the projected point of
Is calculated as the Y-direction baseline amount of the off-axis alignment system OWA.

BLOy = ΔYcc−ΔYf ₂ −ΔYF (13) By the above calculation, the off-axis alignment system O
The baseline amount of WA (BLOx, BLOy) is obtained, and then the main control system 114 obtains the baseline amount (BLTx, BLTy) of the LSA type TTL alignment system. The baseline amount BLTy of the LSA type TTL alignment system for the Y direction is the amount of deviation in the Y direction between the center point of the slit beam spot in the Y direction, the center point of the reticle R, and the projection point of the center CC of the reticle R. Yes, and is calculated by the following formula.

BLTy = YIf−YLs (14) Similarly, the baseline amount BLTx of the LSA type TTL alignment system for the X direction is the center point of the slit-shaped beam spot in the X direction and the center CC of the reticle R. Is the amount of deviation in the X direction from the projected point of, and is calculated by the following equation. BLTx = YIf−YLs (15) However, in the values obtained by the equations (14) and (15), the arrangement error ΔYsm in the Y direction between the center of the light emission mark IFS and the mark LSMy on the reference plate FP, and the light emission Since the X-direction placement error ΔXsm between the mark IFS and the mark LSMx is included, if these errors cannot be ignored, they are stored as constant values in advance, and equations (14) and (15) are respectively calculated as It may be changed to (14 ') and (15').

BLTy = YIf−YLs−ΔYsm (14 ′) BLTx = XIf−XLs−ΔXsm (15 ′) With the above sequence, the baseline measurement is completed, and the wafer pre-aligned on the wafer stage WST. W is placed. A plurality of exposed areas, that is, shot areas onto which the pattern area PA of the reticle R is projected are two-dimensionally arranged on the wafer W. And each shot area has an off-axis alignment system OW
A or TTL alignment system (2X, 3X; 2Y, 3
The alignment mark detected by Y) is formed in a fixed relationship with the center point of the shot area.
In many cases, the alignment marks on those wafers are provided in the street lines. Since a number of methods or sequences are conventionally known as an actual wafer alignment method, description of these methods and sequences will be omitted here, and only basic wafer alignment will be described.

FIG. 18 shows the arrangement of shot areas and marks on the wafer W. The distance between the center SCn of the shot area SAn and the mark WMx for the X direction in the X direction is ΔXwm, and the center SCn and the Y mark for the Y direction. The width between the Y direction and WMy is defined as ΔYwm in design. First, when the off-axis alignment system OWA is used, the mark WMx of an arbitrary shot area SAn is within the detection area of the off-axis alignment system OWA and the index mark TMX ₁ ,
The wafer stage WST is positioned so as to be caught in TMX ₂ . Here, the marks WMx and WMy are assumed to be multi-line patterns like the reference mark FM1.

Then, the main control system 114 reads the coordinate position Xm of the positioned wafer stage WST in the X direction from the interferometer IFX. Further, the image signal from the CCD camera 4X in the off-axis alignment system OWA is processed to detect the shift amount ΔXp in the X direction between the center point of the index plate 4F and the center point of the mark WMx. Next, the wafer stage WST is moved so that the mark WMy of the wafer is sandwiched by the index marks TMX ₁ and TMX _{2 of} the off-axis alignment system.
To position. The coordinate position Ym in the Y direction at this time is read from the interferometer IF ₁ . Then, the amount of deviation ΔYp in the Y direction between the center point of the index plate 4F and the center point of the mark WMy is obtained by imaging with the CCD camera 4Y.

After the above mark position detection is completed, the coordinate position of the wafer stage WST for aligning the center SCn of the shot area SAn with the projection point of the center CC of the reticle R at the time of exposure (only by calculation of the following equation) Xe, Y
e) is required. Xe = Xm- [Delta] Xp + (BLOx- [Delta] Xwm) (16) Ye = Ym- [Delta] Yp + (BLOy- [Delta] Ywm) (17) The mark WM is used in the LSA type TTL alignment system.
When detecting x and WMy, the mark W by the LSA method is used.
With the detection positions of Mx and WMy being Xm and Ym, the stage coordinate position at the time of exposure can be obtained by the following equation.

Xe = Xm + BLTx−ΔXwm (18) Ye = Ym + BLTy−ΔYwm (19) In the above description, the off-axis alignment system OWA.
Of the interferometers IFX, I even at the detection center point in the stationary coordinate system of
Since both measurement values of FY ₁ are determined to be orthogonal, if the measurement values of the two interferometers IFX and IFY ₁ are used for two-dimensional mark position detection using the off-axis alignment system OWA, Wafer stage WST coordinates at detection, positions Xm, Ym, and mark position deviation amount ΔX
This means that p and ΔYp do not include the Abbe error.

Therefore, an off-axis alignment system
When using OWA to detect wafer marks and fiducial marks
Interference that satisfies the Abbe condition for the projection lens PL
Total IFY₂Not the alignment system OWA.
Interferometer IFY that satisfies the bbbe condition ₁Is important to use
It However, the Abbe condition for the projection lens PL
Interferometers IFX and IFY₂And off axis a
Interferometer I that meets Abbe conditions for Liment system OWA
FX, IFY₁To switch and use as is,
Interferometer IFX for two Y directions₁, IFY₂Each internal power
Unter reset (or preset)
You need to do it with and. From the conclusion, the projection lens P
At the same time that the fiducial mark FM2 is detected via L,
A reference marker via the F-axis alignment system OWA
When measuring the FM1 and performing baseline measurement
Two interferometers IF at the Hastage WST stop position
Y₁, IFY₂The value of each internal counter of
Preset equal to the value. Therefore, as mentioned above
In the sequence of FIGS. 14 and 15, two interferometers are used.
IFY₁, IFY₂That the preset operation of
Due to the mounting error θf of the reference plate FP described above.
A correction calculation of the baseline amount is required. From there
The specific example will be described below.

First, steps 508 and 51 in FIG.
Reticle alignment is performed by 0 and 512. At this time, considering the mounting error θf of the reference plate FP, as described above, the X of the reticle marks RM ₁ and RM ₂ is taken into consideration.
Direction alignment position is a ΔXR ₁ = ΔXR _2, and set to be forced to zero, Y direction alignment position within _{ΔYR 1 → ΔOy 1, ΔYR 2} → ΔO
Set to be driven to y ₂ . That is, the reticle R is aligned such that the line connecting the _two reticle marks RM ₁ and RM ₂ is parallel to the reflecting surface of the movable mirror IMy.

Thereafter, the measurement of the baseline error is started, but after the reticle alignment is achieved, the servo lock works so that the wafer stage WST does not move. Considering in the servo lock state, at that time, an interferometer I that satisfies the Abbe condition for the projection lens is used.
There is an error of Ly (Δθa + Δθr) between the measured value Le of FY ₂ and the measured value Lf of the interferometer IFY ₁ that satisfies the Abbe condition for the off-axis alignment system. Here, Ly is two interferometers IFY ₁ and IFY ₂
Is the interval of each measurement axis in the X direction, and the rotation error Δθa is
The ideal position of the reflecting surface of the movable mirror KMy (ideal X
It is a slight rotation error from the axis. Also, the rotation error Δθr
Is the ideal position (X axis) of the reflecting surface of the movable mirror IMy generated when the wafer stage WST reaches the predetermined origin position.
It is a small rotation error from. These errors Δθa, Δθ
Although r cannot be directly measured alone, normally, when wafer stage WST reaches the origin position, interferometers IFY ₁ , IF
By resetting (or presetting) the internal counter of Y ₂ at the same time, it is possible to measure the change in the combined value of Δθa and Δθr from the reset position. That is, the amount of change in Δθa + Δθr can be measured as the yawing amount based on the reset position.

Therefore, when the position of the wafer stage WST is being monitored or controlled by the interferometer IFY ₂ satisfying the Abbe condition for the projection lens, the measured value Lf of the other interferometer IFY ₁ is naturally Lf-
An error of Le = Ly (Δθa + Δθr) is included, and the measured value Lf of the interferometer IFY ₁ cannot be directly incorporated into the baseline amount measurement as a true value.
Alternatively, the control of wafer stage WST cannot be directly transferred to the control of interferometer IFY ₁ .

Therefore, at the time of baseline measurement, the reference plate FP
And the measurement value Lf of the interferometer IFY ₁ when the wafer stage WST is servo-locked by positioning
The difference from the measured value Le of ₂ is ΔLyw [Ly (Δθa + Δθ
r)], the internal counter of the interferometer IFY ₁ is changed (preset) from the measured value Lf to the measured value Le. By doing so, in the subsequent control, the interferometer IFY ₂ used for positioning the wafer stage WST during exposure.
Even if the control based on ( ₁ ) is switched to the control based on the interferometer IFY ₁ used at the time of off-axis alignment, no trouble occurs.

The state at this time is exaggeratedly shown in FIG.
In FIG. 19, two fiducial marks FM2A and FM2B
The line LX connecting the lines is rotated by an error θf with respect to the line Lrc parallel to the reflecting surface of the movable mirror IMy. When the reticle R is aligned, the reticle mark RM ₁ becomes the reference mark F.
The reticle mark RM ₂ is positioned offset by ΔO y _{1 with} respect to M2A, and the reticle mark RM ₂ is ΔO with respect to the reference mark FM2B.
Since the position is offset by y ₂ , the line segment connecting the reticle marks RM ₁ and RM ₂ is eventually parallel to the line Lrc. In FIG. 19, since the line Lrc is defined to pass through the reticle center CC, the reticle marks RM ₁ , RM ₂ and the center CC are located on the line Lrc.

Now, in this state, two interferometers IFY ₁ ,
Although IFY ₂ is preset to the same count value Le, the reference of the _two interferometers IFY ₁ and IFY ₂ after preset changes to the reference line Lir ′ as shown in FIG. In FIG. 19, a line Lir indicates the interferometers IFY ₁ , IFY when the wafer stage WST reaches the origin position, for example.
Indicates the standard when ₂ is preset to the same value. That is, it can be considered that the interferometers IFY ₁ and IFY ₂ measure the distance from the virtual reference line Lir or Lir ′ to the movable mirror IMy. Therefore, immediately after presetting, the reference line Lir ′, the reflecting surface of the movable mirror IMy, and the line Lrc
Are parallel to each other. By the way, when the yawing of the wafer stage WST is obtained from the difference between the measurement values of the _two interferometers IFY ₁ and IFY ₂ after presetting, the yawing amount reference is changed to the line Lir ′ in FIG. That is, a line parallel to the reflecting surface of the movable mirror IMy when the reference plate FP is positioned immediately below the projection lens PL and the off-axis alignment system during the baseline measurement,
It becomes the standard for the subsequent yaw amount measurement.

Further, in the baseline measurement, the amount of positional deviation (ΔXF, ΔYF) between the reference mark FM1 and the index mark TM is obtained by the off-axis alignment system as shown in FIG. In FIG. 19, Of is a detection center point defined by the index mark TM of the off-axis alignment system. Here, the true baseline amount is determined by the positional relationship between the center point CC of the reticle R and the detection center point Of, but if the mounting error θf of the reference plate FP is extremely small, the baseline amount in the X direction is Figure 17
The constant value ΔXfa (distance between FM1 and FM2A) and constant value ΔXfb (distance between FM1 and FM2B), the amount of deviation of the center point CC in the X direction during reticle alignment, and the amount of deviation detected by the off-axis alignment system. Determined by ΔXF.

That is, the two fiducial marks FM2A, F2
Assuming that the distance on the line LX between the midpoint of the M2B in the X direction and the center of the reference mark FM1 is LF, LF is given by the above equation (8).
Similarly to, LF = (ΔXfa + ΔXfb) / 2. Further, the deviation amount ΔXcc in the X direction with respect to the center point of the reference mark LM2 which remains during the reticle alignment is the measured values ΔXR ₁ and ΔXR _{2 in} FIG.
Therefore, similarly to the above equation (9), ΔXcc = (ΔXR ₁ −ΔXR ₂ ) / 2 can be obtained.

Therefore, the true baseline amount BL in the X direction
Ox is obtained by BLOx = LF−ΔXcc−ΔXF, as in the above equation (10). On the other hand, with respect to the baseline amount BLOy in the Y direction, since a sine error (amount of deviation in the Y direction) caused by the mounting error θf occurs, the accuracy cannot be guaranteed by the above-described equation (13).

Consider again here with reference to FIG. First, if the two interferometers IFY ₁ and IFY ₂ are preset, it does not matter which interferometer is used to control the position of the wafer stage. For example, Reticle R
The wafer stage WST is moved by a distance LF in the X direction (strictly, LF−ΔXcc) so as not to change the measurement value of the interferometer IFY ₁ from the state in which a specific point on the wafer is positioned immediately below the center point CC of If moved, the specific point on the wafer will be located at the point Pc in FIG.
Therefore, the true baseline amount BLOy in the Y direction to be managed
Is the distance in the Y direction between the detection center point Of and the point Pc of the off-axis alignment system.

Since the mounting error θf of the reference plate FP has been obtained, the deviation amount ΔTfc between the point Pc and the reference mark FM1 in the Y direction is ΔYfc≈ (LF-under the condition that θf is sufficiently small. ΔXcc) · θf (20) Therefore, the above equation (13) is changed, and the baseline amount BLOy in the Y direction in consideration of the mounting error θf is given by the following equation.

BLOy = ΔYcc−ΔYf ₂ −ΔYfc−ΔYF (21) Note that ΔYcc and ΔYf ₂ are expressed by the above equations (11) and (1), respectively.
It is obtained from 2). As described above, the two interferometers IFY ₁ and IFY ₂ are preset to the same value at the time of the baseline measurement by the reference plate FP, and the base line amount is corrected according to the mounting error θf. By performing the alignment of the reticle R with respect to the reference mark on the reference plate FP in the line measurement state, all the error factors are canceled out, and the accuracy is remarkably higher than that of the conventional baseline measurement. .

When reading the stop position of the wafer stage WST with the interferometer IFY ₁ during the baseline measurement operation, the measurement values of the internal counter are sampled several tens of times within about 1 second, and these are measured. By averaging,
The error due to fluctuation is experimentally 0.03 μm to 0.01
It is reduced to about 2 μm. Also, as shown in FIG.
When the off-axis alignment system OWA is used to detect the alignment marks WMx, WMy, etc., the positioning of the wafer stage WST is controlled by the interferometers IFY ₁ and IFX. At that time, the yawing of the wafer stage WST may occur. There is. However, the yawing at this time does not affect the final alignment accuracy (overlay accuracy of the reticle and the shot on the wafer) after presetting the _two interferometers IFY ₁ and IFY ₂ .

FIG. 20 shows an example of realizing the mutual presetting of the _two interferometers IFY ₁ and IFY ₂ described in FIG. 19. Here, it is assumed that the mutual presetting is realized by hardware, but the same function is realized by software. The calculation of can be achieved with exactly the same idea. In FIG. 20, an up-down counter (UDC) 200 is an internal counter of the interferometer IFY ₁ , and reversibly counts the up-pulse UP1 and the down-pulse DP1 generated as the wafer stage WST moves in the Y direction. Up-down counter (UDC) 20
Reference numeral _{2 is} an internal counter of the interferometer IFY ₂ and similarly counts the up pulse UP2 and the down pulse DP2 reversibly. The count values of the UDCs 200 and 202 are output to the main control system 114 as parallel 24-bit Y coordinate values DY ₁ and DY ₂ , for example. Latch circuit (LT) 204, 20
6 inputs count values DY ₁ and DY ₂ , respectively, and latch pulses S ₁ a and S ₁ b from the main control system 114.
When it receives the count value, it continues to hold the count values DY ₁ and DY ₂ . Here, the output value of the LT 204 is the UDC 202
Is applied as a preset value to the UDC 200 and the output value of the LT 206 is applied to the UDC 200 as a preset value. U
DCs 200 and 202 are set to preset values in response to load pulses S ₁ b and S ₂ b from main control system 114, respectively.

[0114] As mentioned earlier, if the preset interferometer IFY ₂ measurements Le a (DY ₂₎ to the interferometer IFY _1, is output latch pulse S ₂ a relative LT206, predetermined time (uSec order ) The load pulse S ₁ b is output to the UDC 200 with a delay. Of course, Figure 2
The circuit of 0, the inverse of the preset is possible, it is also possible to preset measured value Lf of the interferometer IFY ₁ a (DY ₁₎ to the interferometer IFY _2. Since coordinate measurement using an interferometer is relative, interferometers IFY ₁ , IFY ₂
Instead of the preset, the UDCs 200 and 202 may be simultaneously reset to zero, or simultaneously reset to a constant value independent of the measured values Le and Lf.

By the way, the operation of the baseline measurement exemplified above is performed after the precise reticle alignment is completed, as shown in FIGS. 14 and 15.
The baseline measurement may be performed when the reticle is roughly aligned. For example, in step 504 or 506 in FIG. 14, the reticle mark R
Until M ₁ and RM ₂ come to a position where they can be detected by the TTR alignment systems 1A and 1B, SRA method or IFS
Method to roughly align the reticle. Then run the step 522 in step 508 and FIG. 15 in FIG. 14 at the same time, the reference mark FM2A and positional deviation between the reticle mark _{RM 1 (ΔXR 1, ΔYR 1} ), the reference mark FM2B the reticle mark RM positional deviation amount between _{2 (ΔXR 2, ΔYR 2)} , and the positional deviation amount between the index mark of the reference marks FM1 and off-axis alignment system (ΔXF, ΔYF) Request.

At this time, the reference plate FP is servo-locked in the interferometer mode or the LIA mode. Try it several times to find the average value. This averaging reduces the amount of error that occurs randomly. In this way, when each positional deviation amount is obtained, the center CC (or mark RM ₁ , mark RM ₁ ,
RM ₂ ) projection point and off-axis alignment system O
The relative positional relationship with the WA detection center point is known. further,
The position (rough alignment position) of reticle stage RST in this state is determined by interferometers IRX, IRY, IR.
It is read from the measured value of θ and stored. It is desirable to average this reading as well.

Then, the position displacement amount (ΔXR
₁ , ΔYR ₁ ), (ΔXR ₂ , ΔYR ₂ ), (ΔXF,
Based on ΔYF) and a preset constant value, the detection center point of the off-axis alignment system OWA coincides with the center of the reference mark FM1 (ΔXF = 0, ΔYF =
0) the projection point of the reticle center CC and the center point of the reference mark FM2 (marks FM2A and FM).
The position shift amount (X, Y, θ direction) with respect to 2B is divided. After that, the reticle stage RST is slightly moved from this rough alignment position stored by the interferometers IRX, IRY, and IRθ by this positional deviation amount. In this way, the reticle R is precisely aligned with the detection center of the off-axis alignment system OWA, and thereafter the main control system 114 continues the sequence from step 524 in FIG.

As described above, when there is a sensor (interferometer or alignment system) capable of highly accurately measuring the position change amount of reticle stage RST (that is, reticle R) over a relatively long range (for example, ± several mm). 1 can store the rough alignment position, detect the reference marks for baseline measurement, and finely align the reticle R after that.
Throughput can be improved over the sequence of 4 and 15.

In the embodiment of the present invention, the LIA type TTL is used.
Although the alignment system is used for the servo lock of the reference plate FP, the LIA type TTL alignment system itself needs to perform baseline management with the center CC of the reticle R. If the TTL alignment system of the LIA method is used to detect the mark on the wafer W, the reticle marks RM ₁ and RM ₂ and the reference mark FM 2A detected by the TTR alignment systems 1A and 1B,
When the FM2B and the FM2B are precisely matched, the phase errors Δφx and Δφy of the marks LIMx and LIMy detected by the LIA TTL alignment systems 3X and 3Y, respectively.
Should be stored as a value corresponding to the baseline error amount with respect to the center CC of the reticle R.

Next, a modification of this embodiment will be described.
Step 50 in the sequence described above with reference to FIGS.
In Nos. 8 to 512, the TTR alignment systems 1A and 1B were used to achieve the reticle alignment completely, but the operation can be omitted to some extent.
As shown in FIG. 2, in the apparatus of the present embodiment, the positional deviation of the reticle R in the X, Y, and θ directions can be detected by interferometers IRX and IR.
Step 50 because the Y and IR θ are being monitored successively.
When the projection point coordinates of the reticle marks RM ₁ and RM ₂ are detected by the interferometer on the wafer stage side by the IFS system search operation of _{No. 4} , the reticle R in the X, Y and θ directions is calculated by the coordinate values. Alternatively, the reticle stage RST may be finely moved by relying on an interferometer on the reticle side so as to correct the amount of deviation from the designed arrangement. In this case, the reticle side interferometers IRX, IR
Sufficiently high Y, IRθ measurement resolution (for example, 0.005
μm), the positioning of the reticle R will be performed extremely accurately.

Further, the off-axis
The alignment system OWA is a static alignment method in which the mark detection is performed while the wafer stage WST is stationary. The same effect can be obtained even if the scanning alignment method is used. For example, when the off-axis alignment system OWA adopts a method of projecting a laser beam spot on the wafer W into a slit shape and detecting a mark on the wafer by scanning the stage WST,
When the wafer stage WST is moved so that the reference mark FM1 on the reference plate FP crosses the beam spot, at the same time, the light emission mark IFS changes the reticle mark RM ₁ ,
Alternatively, the arrangement of each mark on the reference plate FP may be determined so as to scan the RM ₂ .

Further off-axis alignment system O
When the LIA method is incorporated in the WA and the reference mark FM1 on the reference plate FP is made to have the same diffraction grating as the marks LIMx and LIMy, the reference mark FM1 detected by the off-axis alignment system OWA becomes the off-axis alignment. Wafer stage WST can be servo-locked based on the detection result of the phase difference measuring circuit so that it is always aligned with the reference grating for LIA in the system. In this case, with the detection center of the off-axis alignment system OWA precisely aligned with the center of the fiducial mark FM1, the TTR alignment systems 1A, 1B
The baseline amount can be calculated only by obtaining the positional deviation amounts between the reference marks FM2A and FM2B and the reticle marks RM ₁ and RM ₂ .

As a TTL alignment system, CC
By using the D camera 4 to capture both the mark image on the wafer or the reference plate FP and the image of the index mark provided in the optical path of the TTL alignment system, the mark can be detected by detecting the amount of positional deviation. Alternatively, a method of detecting the position of may be used. In the case of this method, the projection point of the center point (detection center point) of the index mark in the optical path of the TTL alignment system to the wafer side and the center of the reticle marks RM ₁ and RM ₂ (or the center CC of the reticle) are projected. The baseline amount may be managed between the points.

By the way, the IFS system shown in this embodiment has been explained as a stage scan, that is, a scanning alignment system, but it may be a static alignment system. For that purpose, the light emission mark IFS on the reference plate FP is changed from a slit shape to a rectangular light emitting surface,
Double slit RM ₁ y of the reticle mark shown in FIG.
(Or RM ₁ x) a rectangular light-emitting surface that is sufficiently larger than the width of the double slit is positioned, and a mark RM ₁ y is applied from above the reticle R using a TTR alignment system or the like.
If the (or RM ₁ x) portion is imaged by a CCD camera or the like, an image signal having the same waveform as the waveform shown in FIG. 16B can be obtained. At this time, if the index mark is not in the TTR alignment system, the CCD
It is also possible to obtain the shift amount of the double slit mark RM ₁ y (or RM ₁ x) with reference to a specific pixel position of the camera. Further, in this method, the projection point at the center of reticle mark RM ₁ (or RM ₂ ) is calculated based on the deviation amount and the coordinate value of wafer stage WST when the rectangular light emitting surface is positioned. As shown in FIG. 21, a light-shielding slit pattern SSP for measuring a deviation amount from the double slit mark RM ₁ y (RM ₁ x) is provided in a part of the rectangular light emitting surface PIF. , TTR
The light emitting surface PIF may be imaged by a CCD camera of the alignment system and the amount of positional deviation between the dark line due to the double slit mark RM ₁ y and the dark line due to the slit pattern SSP may be obtained.

FIG. 22 shows the reference on the wafer stage WST.
Arrangement of plate FP and arrangement of off-axis alignment system
Shows a modified example of the off-axis alignment system
The position of the objective lens 4B is within the plane of the drawing in FIG.
It is under L. This position is on the front side of the device.
Corresponds to the wafer loading direction. Marks in Figure 22
Interferometer I for position measurement of wafer stage WST
FY, IFX₁, IFX ₂Except for the others, the others are those of Figure 3.
Is the same as. In the case of FIG. 22, the optical axis position of the projection lens PL
And detecting off-axis alignment system OWA
The line segment connecting the heart (almost the optical axis position of the objective lens 4B) is
Since it is parallel to the Y axis, there is only one interferometer IPHY in the Y direction.
And X-direction interferometer IFX₁, IFX₂Was two.
In accordance with this, the position of each mark on the reference plate FP is changed.
The center points of the reference marks FM1 and FM2.
The connecting line segment is parallel to the Y axis.

Also in the case shown in FIG. 22, the marks on the wafer are removed by the off-axis alignment system OWA.
Alternatively, when detecting the reference mark FM1 or the like, interferometers IFX ₁ and Ify satisfying the Abbe condition are used, and interferometers IFX ₂ and 1F are used for wafer stage positioning during exposure.
Y is used. That is, when the mark is detected by the off-axis alignment system OWA, the interferometer I
The position coordinate value in the X direction measured by FX ₁ is the interferometer IF.
It is associated with the position coordinate value measured by X ₂ . This correspondence is made by interferometer IFY ₁ ,
It is performed in exactly the same way as the mutual preset between IFY ₂ .

Although the exposure apparatus described in the above embodiments is a stepper that exposes the projected image of the pattern area PA on the reticle R onto the wafer W by the step-and-repeat method, the present invention is a reticle. The same can be applied to a step scan type exposure apparatus that simultaneously scans the wafer in a direction perpendicular to the optical axis of the projection optical system. Also S
The same alignment system can be applied to an X-ray aligner using an X-ray source such as OR, an X-ray stepper, and the like.

[0128]

As described above, according to the present invention, the baseline measurement can be performed without being influenced by various precisions of the substrate stage, so that the accuracy of the baseline measurement can be improved. In addition, alignment of the reticle (mask) and baseline measurement can be performed almost at the same time, the stage is moved to check the mask rotation error (error in the θ direction), and the stage is moved for baseline measurement. Since there is no need to do so, there is an effect that the total processing speed is improved.

Further, according to the present invention, the reticle alignment and the baseline measurement can be performed almost at the same time. Therefore, even if the sequence for performing the baseline measurement is set every wafer exchange, the throughput is not deteriorated. The long-term drift of the baseline, the position drift of the reticle holder due to the irradiation of the reticle with the exposure light, and the like can be confirmed and corrected at high speed.

According to the embodiment, the TTL alignment system or the TTR alignment system (second mark detecting means)
While the position of the reference plate is servo-locked using, the mark on the reference plate is off-axis alignment system (first
Since it is detected by the mark detection means) and the baseline is measured, there is no need to use an interferometer for measuring the position of the substrate stage as in the conventional method, and measurement is caused by the air fluctuation (refractive index fluctuation) in the optical path of the interferometer. The error can be reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing a state of baseline measurement in a conventional projection exposure apparatus.

FIG. 2 is a perspective view showing a configuration of a projection exposure apparatus according to an embodiment of the present invention.

FIG. 3 is a plan view showing the arrangement of reference mark plates on the wafer stage.

FIG. 4 is a plan view showing the arrangement of various marks on the reference mark plate.

FIG. 5 is a plan view showing an arrangement relationship between an image field of a projection lens, a reticle pattern, and a reference mark.

FIG. 6 is a diagram showing an example of the shape of a reticle alignment mark.

FIG. 7 is a diagram showing the configuration of a TTR alignment system.

FIG. 8 is a diagram showing a configuration of a TTL alignment system.

FIG. 9 is a diagram showing a pattern arrangement of an index plate of an off-axis alignment system.

FIG. 10 is a diagram showing a configuration of an off-axis alignment system.

FIG. 11 is an enlarged view showing a reference mark FM1 on a reference mark plate.

FIG. 12: Reference marks FM2 and LI on the reference mark plate
Enlarged view of M and LSM

FIG. 13 is a diagram illustrating an error in taking a reference mark plate on a wafer stage and a measuring method therefor.

FIG. 14 is a view for explaining a typical sequence of this device.

FIG. 15 is a view for explaining a typical sequence of this device.

FIG. 16 is a diagram showing an example of a waveform of a photoelectric signal detected by an LSA system and an ISS system.

FIG. 17 is a diagram summarizing constant values and measured values required for baseline management.

FIG. 18 is a plan view showing the arrangement of shot marks and wafer marks on the wafer.

FIG. 19 is a view for explaining the principle of mutual presetting of two Y-direction interferometers.

FIG. 20 is a circuit block diagram showing an example for presetting an interferometer.

FIG. 21 is a view showing another pattern example of the light emitting marks on the reference mark plate.

FIG. 22 is a plan view showing another arrangement of the off-axis alignment system.

[Explanation of symbols for main parts]

R Reticle W Wafer PL Projection lens RST Reticle stage WST Wafer stage 1A, 1B TTR alignment system 2X, 3X X-direction TTL alignment system 2Y, 3Y Y-direction TTL alignment system OWA Off-axis alignment system FP Reference plate FM1 off- reference marks for the axis alignment system FM2 TTR reference mark IFX for alignment systems, laser interferometer RM _1, RM ₂ reticle mark 200, 202 up-down counter for IFY wafer stage

Claims (2)

[Claims] 1. A mask stage for holding a mask, a projection system for projecting a pattern of the mask, a substrate stage for two-dimensionally moving while holding a photosensitive substrate in an image plane of the projection system, and the projection system. An alignment system for detecting a mark on the photosensitive substrate, and a detection center point of the alignment system for measuring the coordinate position of the substrate stage, which has a detection center point at a position apart from the optical axis of A pair of first interferometers having two measurement axes orthogonal to each other and a pair of second interferometers having two measurement axes orthogonal to each other at the optical axis position of the projection system.
An interferometer, and after obtaining a baseline amount by measuring the relative positional relationship between the coordinates of the specific point on the mask that can be projected by the projection system and the coordinates of the detection center point of the alignment system, An apparatus that aligns the photosensitive substrate with an alignment system and moves the substrate stage based on the alignment result and the baseline amount to position the photosensitive substrate at an exposure position by the projection system. And a second fiducial mark that can be detected by the alignment system and a second fiducial mark that can be set in a unique positional relationship with a specific point of the mask when the first fiducial mark is positioned at the detection center point. A reference plate on which the base plate is formed; and the base plate based on the arrangement of the first and second reference marks on the reference plate. At the stop position of the substrate stage positioned when measuring the in amount, the measurement value of the first interferometer and the measurement value of the second interferometer are set to be equal to one of the measurement values. A projection exposure apparatus comprising: setting means. 2. A projection system for image-forming and projecting a pattern of a mask onto a photosensitive substrate, a substrate stage for two-dimensionally moving while holding the photosensitive substrate, and the projection system for detecting marks on the photosensitive substrate. First mark detection means having a detection center point set at a predetermined position outside the projection visual field region, and a mark on the mask at a predetermined position inside the projection visual field region,
Alternatively, second mark detecting means for detecting a pattern on an object located in the projection image plane of the projection system, and detection by the first mark detecting means for measuring the coordinate position of the substrate stage. A set of first interferometers having two measurement axes orthogonal to each other at a center point, and a set of second interferometers having two measurement axes orthogonal to each other at the optical axis position of the projection system are provided. In the projection exposure apparatus, the projection exposure apparatus is provided on the substrate stage, and is provided at a position corresponding to a designed arrangement relationship between a projection point of the mark of the mask by the projection system and a detection center point of the first mark detection means. 1
A reference plate having a reference mark and a second reference mark formed thereon;
For the baseline measurement, the first reference mark is located near the detection center point of the first mark detecting means, and the second reference mark is located near the projection point of the mask mark. Control means for positioning the substrate stage; when the positioning is performed, the current position of the stage measured by the first interferometer and the current position of the stage measured by the second interferometer are recognized as the same position. As described above, the projection exposure apparatus is provided with a correction unit that corrects the coordinate measurement value by at least one of the first interferometer and the second interferometer.