JPH08124830A - Projection exposure device - Google Patents

Abstract

PURPOSE: To provide a correcting element, which controls an axial chromatic aberration on an alignment light path, to the pupil of a projection optical system, to make a plane apart from a mark plane and a photosensitive substrate surface conjugate with each other to an alignment light path, and to detect optical data of a mask separately from light reflected from the mask. CONSTITUTION: Alignment beams LBW1 and LBW2 irradiating a wafer mark are made to impinge on a projection optical system PL intersecting each other in a space D above a reticule R and deflected by a correcting optical element (phase gratings GX1 and GX2 on a correcting plate PC) to form an image on a wafer mark WMx. Interference light BTw of π primary diffracted light generated from the wafer mark WMx is deflected by the correcting optical element and made to impinge on a photoelectric element 26W through the intermediary a visual field diaphragm 24W. The visual field diaphragm 24W is possessed of an opening located at a point conjugate with the intersecting point P' (conjugate point of the wafer mark WMx) of the alignment beams LBW1 and LBW2 so as to remove stray light which going toward the optical element 26W.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an alignment device for aligning and exposing a reticle (mask) pattern onto a pattern formed on a photosensitive substrate through a projection optical system. In particular, the present invention relates to an alignment apparatus for irradiating a photosensitive substrate with illumination light having a wavelength different from that of exposure illumination light via a projection optical system.

[0002]

2. Description of the Related Art Conventionally, the projection optical system of this type of projection exposure apparatus has various chromatic aberrations as well as chromatic aberrations so that the circuit pattern of the reticle is imaged on the photosensitive substrate with the best image quality at the wavelength of the illumination light for exposure. It has been corrected. Therefore, the most ideal way to align the reticle and the photosensitive substrate is
The alignment mark on the reticle and the alignment mark on the photosensitive substrate are illuminated with illumination light having the same wavelength as the exposure illumination light (ultraviolet light having a wavelength of 365 nm or 248 nm),
This is a method of photoelectrically detecting both marks through an alignment optical system arranged above the reticle. However, the mark detection using the exposure wavelength does not always give a good mark detection result due to the influence of the photoresist layer on the photosensitive substrate. This is mainly because a large distortion occurs in the signal waveform and the like when the mark is photoelectrically detected due to the absorption of the exposure wavelength of the photoresist layer and the phenomenon of thin film interference.

Therefore, in a projection exposure apparatus (stepper) capable of performing projection exposure in the submicron region in recent years, exposure is performed for the purpose of reducing the addition of optical distortion due to the photoresist layer and avoiding unnecessary exposure of the photoresist. An alignment method in which a mark on a photosensitive substrate is detected using a wavelength sufficiently longer than the wavelength, for example, red light of a He-Ne laser as illumination light has become mainstream.

There are roughly two types of alignment apparatus using He-Ne laser light. One of them is an alignment optical system, and an illumination beam from a He-Ne laser is projected onto a photosensitive substrate only through a projection optical system. TTL for projecting and detecting the reflected light (including scattered and diffracted light) from the mark on the photosensitive substrate by the alignment optical system only through the projection optical system.
(Through the lens) method. This TTL method is H
Since it is not necessary to conjugate the reticle and the photosensitive substrate to each other at the wavelength of 633 nm of the e-Ne laser, it is not necessary to consider the axial chromatic aberration and the chromatic aberration of magnification that occur in the projection optical system, and the alignment optical system can be assembled relatively easily. I was able to. However, since the TTL method only detects the position of the mark on the photosensitive substrate, the distance between the projection position of the center point of the circuit pattern of the reticle and the mark detection position of the alignment optical system of the TTL method, a so-called baseline is sometimes set. There is also a problem in that it is necessary to measure accurately and manage the baseline value.

Another method using the He-Ne laser light is to arrange a positive single lens (correction lens) having a small diameter in the center of the pupil plane (Fourier transform surface) of the projection optical system to obtain a projection optical system. Only with respect to the optical path having an extremely small numerical aperture of the system, the reticle and the photosensitive substrate are conjugated with each other under the wavelength of the He-Ne laser light, and the reflected light from the mark on the photosensitive substrate is passed through the projection optical system and the correction lens. An image is formed on the reticle, and the mark image on the photosensitive substrate and the mark formed on the reticle are simultaneously detected by the alignment optical system T
It is a TR (through the reticle) method.

Such a TTR alignment method is disclosed in, for example, US Pat. No. 5,100,237 (hereinafter referred to as USP.'237).
Is disclosed in. Further, in this publication, a one-dimensional diffraction grating is provided as a mark on a reticle and a mark on a photosensitive substrate, and these diffraction gratings are irradiated with two illumination beams having symmetrical incident angles to form a diffraction grating on the reticle. Of the diffraction grating of the photosensitive substrate and the interference light generated from the diffraction grating of the photosensitive substrate and returning to the reticle side through the projection optical system and the correction lens, by photoelectrically detecting the pitch of the diffraction grating between the reticle and the photosensitive substrate. A method of directional alignment is disclosed.

FIG. 1 schematically shows the configuration of the projection optical system disclosed in USP.'237 and the optical path of an illumination beam for alignment, and FIG. 2 is shown in USP.'237. FIG. 6 is a perspective view showing a positional relationship between marks on a reticle R and marks on a photosensitive substrate (hereinafter, referred to as a wafer W). FIG. 2 (A)
As shown in FIG. 5, a light-shielding band (hatched portion) is formed around the circuit pattern region on the reticle R, and a reticle mark RMx by a diffraction grating and a transparent window WI are provided in a part of the light-shielding band. The reticle mark RMx is formed by engraving minute lines extending in the Y direction at a constant pitch in the X direction, and the windows WI are arranged side by side in the Y direction so as to be closer to the reticle mark RMx on the reticle center side. .

Further, as shown in FIG. 2B, a street line STL having a width ys is provided between the shot areas on the wafer W, and the wafer mark forming area MA is provided in the street lines STL. Is set. In this area MA, a lattice-like wafer mark WMx in which minute lines extending in the Y direction are engraved at a constant pitch in the X direction.
Are formed in an arrangement corresponding to the window WI on the reticle R. The wafer mark WMx is formed in a half area on the center side of the shot area in the mark forming area MA, and the other half area is a blank area.

With respect to such a mark arrangement, as shown in FIGS. 2 (A) and 1 (A), it has a thickness (diameter) including both the reticle mark RMx and the window WI. Two coherent illumination beams (He-Ne laser) L
B ₁ and LB ₂ are made incident such that they intersect at the mark surface of the reticle R. The incident angles of the _two beams LB ₁ and LB ₂ are determined so as to be symmetrical with respect to the optical axis AXa of the alignment optical system, and the crossing angle is set to the reticle mark RMx.
Is determined so as to generate an interference fringe having a pitch of Pgr / 2 with respect to the pitch Pgr of. Therefore, the reticle mark RM
From x, the −1st order diffracted light generated by the incidence of the beam LB ₁ and the + 1st order diffracted light generated by the incidence of the beam LB ₂ proceed coaxially in a direction parallel to the optical axis AXa. The two first-order diffracted lights are interference lights BT that interfere with each other.
r, and its intensity corresponds to the positional relationship in the X direction between the interference fringes and the reticle mark RMx.

On the other hand, 2 which has passed through the window WI on the reticle R
As shown in FIG. 1 (A), the book beams LB ₁ and LB ₂ are incident on a bilateral telecentric projection lens system composed of a front lens group GA and a rear lens group GB.
Then, the two beams LB ₁ and LB ₂ are refracted by an amount corresponding to the axial chromatic aberration by the correction lens CL arranged at the center of the pupil plane EP of the projection lens system, and intersect on the wafer W. The correction lens CL is set to have a diameter as small as possible, for example, 10% or less with respect to the effective diameter of the pupil plane EP, in order to reduce the influence on the image forming light flux at the time of exposure.
If there is no correction lens CL, the two beams LB ₁ ,
LB ₂ cannot cross on the wafer W as shown by the broken line.

Now, the two beams LB ₁ and LB ₂ intersecting on the wafer W generate interference fringes on the wafer mark WMx. If the pitch of the wafer mark WMx is Pgw,
The pitch of the interference fringes on the wafer is set to Pgw / 2.
Therefore, from the wafer mark WMx, interference light BTw of ± first-order diffracted light is generated in a direction parallel to the optical axis AXa of the alignment optical system as shown in FIG. This interference light B
Tw returns to the position of the window WI of the reticle R via the projection lens system and the correction lens CL, passes through this window WI, becomes parallel to the interference light BTr, and enters the alignment optical system. The intensity of the interference light BTr changes according to the relative position in the pitch direction between the interference fringes on the wafer W and the wafer mark WMx.

Therefore, the reticle mark RMx and the wafer mark are obtained by photoelectrically detecting the interference lights BTr and BTw obtained through the alignment optical system and obtaining the level change of each photoelectric signal or the phase difference of each photoelectric signal. The relative position error with WMx in the X direction is known. Note that FIG. 1B is a view of the optical path of FIG. 1A viewed from a direction (X direction) orthogonal to each other. When viewed from this direction, _two beams LB ₁ and LB ₂ reaching the wafer are shown. Wafer mark WM
The interference light BTw from x appears to overlap. Also, the pupil plane E
If a single lens CL is inserted in P, axial chromatic aberration can be corrected, but correction of lateral chromatic aberration is insufficient in most cases. Therefore, USP.'237 also discloses that a lens element for correcting chromatic aberration of magnification, a prism and the like are added in the alignment optical path between the reticle R and the projection lens system.

When a constant frequency difference Δf (for example, several tens of KHz) is given between the _two irradiation beams LB ₁ and LB _{2 in} the above-described configurations of FIGS. 1 and 2, the interference lights BTr and BTw have a frequency Δf. The intensity of the two photoelectric signals is modulated into an AC signal (sine wave) having a frequency Δf. Therefore, the relative positional deviation amount between the reticle mark RMx and the wafer mark WMx is obtained from the phase difference between the two photoelectric signals.

Further, in Japanese Unexamined Patent Publication No. 5-160001, instead of the correction lens CL shown in FIG. 1, the phase grating is provided only at the position in the pupil plane EP through which the illumination beams LB ₁ and LB ₂ and the interference light BTw pass. Has been disclosed to correct both axial chromatic aberration and lateral chromatic aberration.

[0015]

However, in the above-mentioned conventional techniques, on the other hand, the great advantage that the axial chromatic aberration is almost completely corrected in the optical path of the illumination beam for alignment or the detection beam is obtained. It has been discovered that the advantage is, on the contrary, a great drawback for the alignment method in which the diffraction grating-like mark is illuminated by the interference fringes of two beams.

In general, the alignment optical system has a field stop in a plane where the illumination beams LB ₁ and LB ₂ intersect with each other, that is, in a plane conjugate with the mark plane of the reticle R (and the wafer W). The interference lights BTr and BTw from the RMx and the wafer mark WMx are photoelectrically detected independently. Here, when the illumination beams LB ₁ and LB ₂ toward the wafer mark WMx pass through the window WI of the reticle R, specular reflection light is generated from both the mark surface of the reticle R and the other surface (glass surface) opposite to the mark surface. . Further, these reflected lights are reflected by the illumination beam LB.
₁ , the light path of LB ₂ goes backward to reach the field stop, and the light generated from the edge of the opening of the field stop is mixed as noise into the interference light BTw from the wafer mark WMx. In particular, the interference light BTw is often interference light of ± first-order diffracted light generated in the same direction from the wafer mark WMx, for example, the optical axis direction. Therefore, the intensity of the interference light BTw is smaller than that of the regular reflection light from the reticle R, that is, the noise component included in the interference light BTw is relatively large. This becomes particularly remarkable when the phase grating disclosed in Japanese Patent Laid-Open No. 5-160001 is used instead of the correction lens CL in FIG.

From the above, the noise component from the reticle R is superimposed on the interference light BTw from the wafer mark WMx, and the mixing of the noise component results in the wafer mark WMx.
Therefore, the accuracy of position detection will decrease. Originally, it is possible to detect the position on the order of nanometers by the alignment method using the interference fringes, but due to the influence of the above noise component, the detection accuracy of the order cannot always be obtained. That is, in the system that measures the positions of the wafer mark WMx and the reticle mark RMx by the interference lights BTw and BTr, a slight amount of stray light is mixed with the original signal light, which interferes with the original signal light. It gives more or less distortion to the signal light and causes a problem that the distortion affects the phase angle in the position measurement based on the phase difference, that is, in the heterodyne alignment method.

Therefore, according to the present invention, when an alignment illumination beam having a wavelength different from that of the exposure light is projected on the mask and the photosensitive substrate via the projection optical system, the light information from the mask or the photosensitive substrate is reflected by the mask. An object of the present invention is to provide a projection exposure apparatus capable of detecting without mixing and performing highly accurate alignment.

[0019]

In order to achieve the above object, the present invention basically adopts a configuration as shown in FIG.
That is, an illumination system that irradiates the mask (R) on which the original image pattern is formed with exposure light (I ₁ ), and a projection optical system (PL) that has the best imaging performance under the exposure light (I ₁ ). When, the exposure light (I ₁₎ different from the wavelength of the alignment beam (L
B ₁ and LB ₂ ) irradiate the first grating mark (reticle mark RMx) formed on the mask (R) and the transparent portion (window WI) in the vicinity of the first lattice mark RMx and project the transparent portion (WI). From the second grating mark (WMx), the beam irradiation means (including the objective lens OB) for irradiating the second grating mark (wafer mark WMx) formed on the photosensitive substrate (W) via the optical system (PL). Diffracted light (BTw) that is generated and travels through the projection optical system (PL) and the transparent portion (WI);
Has a light receiving system (20, 22, 23, 24W, 24R) having a photoelectric detector (26W, 26R) for receiving the diffracted light (BTr) generated from the lattice mark (RMx), and the output of the photoelectric detector Position detection means for detecting relative displacement between the mask (R) and the photosensitive substrate (W) based on the signal is provided.

In the present invention, a part of the beam irradiating means is provided with two first crossing marks on the first grating mark (RMx).
The alignment beams (LBR ₁ and LBR ₂ ) intersect with each other in a space separated by a distance D from the pattern surface of the mask so as to pass only the transparent portion (WI) without irradiating the first grating mark (RMx) 2 Beam generating member (2) for generating a second alignment beam (LBW ₁ , LBW ₂ ) of the book
The focus element DF and the like) are provided, and the field stop (24W) is arranged at a position conjugate with the intersection of the two second alignment beams in the light receiving system.

Further, according to the present invention, two second alignment beams (LB) are formed in the pupil plane (EP) of the projection optical system (PL).
W ₁ and LBW ₂ ) are arranged in a portion through which the second alignment beams (LBW ₁ and LBW ₂ ) pass and the projection optical system (P
A plurality of optical elements (GX ₁ , GX ₂ ) for deflecting the second lattice mark (WMx) so as to intersect each other via L) are provided. Then, these optical elements (GX ₁ , GX ₂ ) cause the original axial chromatic aberration Lc occurring on the mask side at the wavelength of the alignment beam to be separated by the distance D for at least the optical path of the second alignment beam (LBw ₁ , LBw ₂ ).
It is characterized by correcting up to.

The beam widths of the two second alignment beams (LBW ₁ , LBW ₂ ) reflected by the mask (R) on the field stop (24W) and their crossing angles are W 1 and θ, The aperture width of the field stop (24 W) is set to W2 and the intersection (P) of the two second alignment beams and the field stop (2 W).
4W) and an optical system (OB, 10, 14,
22) where β is the magnification, the amount of the interval D is D ≧ (W1−W2) / {β ² × 4 · sin (θ / 2)}.

[0023]

In general, in the TTR alignment method, when detecting the optical information from the alignment mark on the substrate, in order to eliminate stray light from the projection optical system, the field of view is conjugate to the alignment mark on the substrate in the light receiving system. Provide a diaphragm.
Here, when the axial chromatic aberration of the projection optical system is corrected for the alignment beam having a wavelength different from that of the exposure light, a point on the mask that is conjugate with the alignment mark on the substrate is conjugate with the field stop. Therefore, a part of the two alignment beams that illuminate the alignment mark on the substrate are specularly reflected by the surface of the mask and intersect on the field stop. In order to effectively remove stray light, this field stop has an opening of substantially the same size as the alignment mark on the substrate, that is, smaller than the alignment beam and the regular reflection beam from the mask surface. Therefore, these beams are diffracted and scattered by the aperture edge of the field stop. In particular, when an opening edge exists in a portion where the two light beams specularly reflected on the mask surface overlap, the light diffracted by the opening edge becomes noise,
It is mixed in the optical information (diffracted light) generated from the alignment mark on the substrate.

Therefore, in the present invention, as shown in FIG. 3, two alignment beams (LBW) intersecting on the wafer W are crossed.
_1, LBW _2), the projection phase gratings GX ₁ correction plate disposed in the pupil plane EP of the optical system PL (the transparent glass plate) on PC, GX
It is obtained by diffracting and deflecting at ₂ and further, the crossing position on the reticle R side of the _two alignment beams (LBW ₁ , LBW ₂ ) directed to the wafer W through the window WI of the reticle R is intentionally set to the window WI. Distance D from the position in the optical axis AX direction
The alignment beam (LB) is shifted even though correction optical elements such as phase gratings (GX ₁ , GX ₂ ) are provided.
The axial chromatic aberration Lc is separated by the distance D under the wavelength of (W ₁ , LBW ₂ ).
I left it alone. That is, on the optical paths of the two alignment beams that irradiate the wafer mark, the correction optical elements (GX ₁ and GX ₂ ) are designed so that the reticle R and the wafer W are conjugated with the axial chromatic aberration amount D. And two alignment beams for irradiating the reticle mark are made to enter so as to intersect with each other on the reticle.

In this way, the wafer mark (WM
x) two alignment beams are specularly reflected by the reticle surface as they pass through the window WI of the reticle R 2
The light flux does not intersect on the field stop, or the aperture edge does not exist in the portion where the two light fluxes specularly reflected by the reticle surface overlap. This can prevent noise components from the reticle from being mixed in the diffracted light generated from the alignment mark on the wafer. In particular, since the diffracted light from the wafer is often weaker than the regular reflection light from the reticle, the effect of the present invention is extremely large. Further, since the amount of the axial chromatic aberration Lc (that is, the distance D) to be left is set to be small, it is possible to minimize the error caused by the vibration and tilt of the alignment optical system, and thus the highly accurate reticle. Can be aligned with the wafer.

[0026]

EXAMPLES Examples of the present invention will now be described. In each of the following examples, the correction lens disclosed in USP.'237 or the phase grating disclosed in Japanese Patent Laid-Open No. 5-160001 is projected. It is arranged on the pupil plane EP of the optical system PL, and is corrected so that the axial chromatic aberration Lc is left by a predetermined amount (interval D) under the wavelength of the alignment beam. That is, two alignment beams (L
BW ₁ and LBW ₂ ) intersect at a spatial position away from the reticle R by a distance D in the optical axis AX direction of the projection optical system PL. Further, for example, it has a bifocal element or the like, or is disclosed in JP-A-5-114544 or JP-A-5-15.
By adopting the alignment beam transmission system disclosed in Japanese Patent No. 2190, apart from the two alignment beams intersecting on the wafer mark WMx at the spatial position separated from the reticle R by the distance D, on the reticle mark RMx. Two intersecting alignment beams (LBR ₁ and LBR ₂ )
Is generated and a total of four alignment beams are used.

First, the principle of the present invention will be described with reference to FIG. In FIG. 4A, two alignment beams LBW ₁ and LBW ₂ for irradiating the wafer mark WMx are shown for the sake of easy understanding, and _two alignment beams (LBR ₁ , LBR ₁ , LB
R ₂ ) is omitted in the figure. In addition, the diffracted light (interference light) BTw generated from the wafer mark is obtained by the embodiment described later (FIG. 5, FIG.
Photoelectric element 26 via a plurality of optical elements described in FIG.
Although it is incident on W, it is schematically shown by omitting a part thereof.

In FIG. 4A, two alignment beams LBW ₁ and LBW for irradiating the wafer mark are shown.
₂ intersects at a spatial position of a distance D above the reticle R (a point P conjugate with the wafer mark), and then enters the projection optical system PL through the window WI of the reticle R as shown in FIG. The light is deflected by the correction optical element (the phase gratings GX ₁ and GX ₂ on the correction plate PC in FIG. 3) arranged on the pupil plane EP, and an image is formed (crossed) on the wafer mark WMx. Wafer mark W
Interference light BTw of, for example, ± 1st-order diffracted light generated from Mx in substantially the same direction is corrected optical element (phase grating G in FIG. 3).
X ₃ ), the light is deflected through the window WI, the objective lens OB, the beam splitter 16, the lens system 22, and the field stop 24W.
The light enters the photoelectric element 26W via the. Field stop 24W
Has an opening at a position P ′ which is conjugate with an intersection point (position conjugate with the wafer mark WMx) P of the two alignment beams LBW ₁ and LBW ₂ , and is generated from the projection optical system PL or the like toward the photoelectric element 26W. Stray light is removed. Therefore, the photoelectric element 26W can photoelectrically detect only the interference light BTw from the wafer mark WMx by the field stop 24W.

Now, a part of the two alignment beams LBW ₁ and LBW ₂ for irradiating the wafer mark WMx is specularly reflected on the surface of the reticle R (both the pattern surface and the glass surface), and as shown in FIG. As shown by the dotted line, the specularly reflected lights DL ₁ and DL ₂ travel toward the photoelectric element 26W. However, in FIG. 4A, the two alignment beams L
BW ₁ and LBW ₂ intersect at a point P above the reticle R,
Further, the field stop 24W is arranged at a position P'which is conjugate with the intersection P. Therefore, the regular reflection light DL from the reticle R
₁ and DL ₂ intersect at a point Q that is a distance D ′ away from the field stop 24W on the reticle R side. It should be noted that a plurality of optical elements between the intersection P and the field stop 24W (see FIG.
In (A), if the lateral magnification of the composite system including the objective lens OB and the lens system 22) from the intersection P to the conjugate point P ′ is β, the distance D ′ is expressed as D ′ = 2 · D × β ^2. To be done.

At this time, even if the two specular reflection lights DL ₁ and DL ₂ are separated on the field stop 24W, or even if the two specular reflection lights DL ₁ and DL ₂ intersect, the field stop 24W.
In the above, the overlapping portion does not include the opening edge, in other words, if the width of the overlapping portion of the two specular reflection lights on the field stop 24W is smaller than the opening width, even if the specular reflection light is the opening edge. Even if it is diffracted by and mixed in the interference light BTw,
The beat signal due to the specular reflection light is not detected by the photoelectric element 26W, and the specular reflection light diffracted at the opening edge does not become noise.

FIG. 4 (B) is incident on the field stop 24W,
The states of specular reflection lights DL ₁ and DL ₂ having a width W1 are shown.
These specularly reflected lights DL ₁ and DL ₂ partially overlap with each other on the field stop 24W, and a part thereof passes through the opening having the width W2. Here, the reason why the regular reflection light from the reticle R does not become noise is that the field stop of the overlapping portion of the _two regular reflection lights DL ₁ and DL ₂ (double hatched portion in FIG. 4B) is as described above. The width on 24 W is smaller than the opening width W2, that is, the distance D ′ of the intersection Q of the _two specular reflection lights DL ₁ and DL _{2 from} the field stop 24 W (conjugate point P ′) is as follows. It is when you are satisfied with the relationship.

D ′ ≧ (W1−W2) / {2 · sin (θ / 2)} (1) Here, since D ′ = 2 · D × β ² , the equation (1) is as follows. expressed. D ≧ (W1−W2) / {β ² × 4 · sin (θ / 2)} (2) Therefore, in the embodiment of the present invention, to satisfy the expression (2),
The position of the intersection P of the _two alignment beams LBW ₁ and LBW ₂ (distance D from the reticle surface) is set. Further, the two alignment beams LBW ₁ and L L are aligned by the amount corresponding to the axial chromatic aberration amount Lc (= D) to be left in the projection optical system PL under the wavelength of the alignment beam.
A correction optical element is formed on the pupil plane EP so that BW ₂ is deflected.

Two alignment beams LBW ₁ and LB
In the case of the heterodyne method in which W ₂ has a constant frequency difference Δf, two specular reflection lights DL ₁ and DL from the reticle R are used.
_{The fact} that the opening edge does not exist in the overlapping portion on the field stop 24W of _{No. 2} means that the specular reflection lights DL ₁ and DL ₂
Even if diffracted light or scattered light is generated at the opening edge,
It means that the diffracted and scattered light does not have the beat frequency of the frequency difference Δf. That is, in the heterodyne system, noise light components having the same beat frequency as the interference light BTw are also eliminated. Thereby, the specular reflection light D
Even if L ₁ and DL ₂ are diffracted and scattered at the opening edge of the field stop 24W and are mixed in the interference light BTw, the specular reflection light does not become noise in the photoelectric element 26W.

Generally, in order to improve the effect of removing the stray light described above, it is necessary to make the aperture width W2 of the field stop 24W as small as possible. Further, since the two alignment beams LBW ₁ and LBW ₂ irradiate the wafer mark WMx with a margin, the beam diameter (cross-sectional area) thereof is set larger than the size (formation region) of the wafer mark WMx. . Further, for example, the aberration of the alignment optical system, or a dichroic mirror obliquely provided above the reticle R, which reflects one of the exposure light and the alignment light and transmits the other, or two alignments in the heterodyne system. An acousto-optic modulator (AOM) that gives a predetermined frequency difference between the beams LBW ₁ and LBW ₂ or a plurality of laser beams having different wavelengths is incident as disclosed in JP-A-6-82215. The specularly reflected beam from the reticle R spreads on the field stop 24W by a diffraction grating plate or the like that generates two alignment beams for each wavelength and emits them toward the AOM. From the above, each of the above equations (1) and (2) is positive on the right-hand side, and both are valid.

For example, the lateral magnification β of the above-mentioned combining system (OB, 22) is 1, the projection magnification of the projection optical system PL is ⅕, the wafer mark pitch is 4 μm, and the beam width W1 on the field stop 24W. 700 μm, aperture width W2 of the field stop 24W
Is 500 μm, and the alignment beam wavelength λ is 633 n.
If m, the crossing angle θ of the regular reflection beam on the field stop 24W is θ = 3.63 °, and D = D ′ from the equations (1) and (2).
/2=1.58 mm. Therefore, the two alignment beams LBW ₁ and LBW ₂ cross each other at a position P that is 1.58 mm or more away from the surface of the reticle R.

At this time, the intersection P may be separated by 1.58 mm or more below the pattern surface of the reticle R (on the projection optical system side) or above the glass surface of the reticle R (on the exposure light source side). The intersection P may be separated by 1.58 mm or more. Further, particularly when the distance D is set to the minimum value, that is, when the intersection P is separated by 1.58 mm, the error caused by the vibration or the tilt of the alignment optical system can be minimized, and the accuracy is high. It is possible to align the reticle R with the wafer W.

Here, since the specularly reflected light is generated from both surfaces (pattern surface and glass surface) of the reticle R, the reticle R
When the intersection P is displaced from the pattern surface of the reticle R to the upper side (the glass surface side) or the intersection point P is displaced from the glass surface of the reticle R to the lower side (the pattern surface side) of the reticle R, the two Expression of intersection P of alignment beam
It may be necessary to shift the distance further than the distance D of (2). This is because the intersection P from one of the pattern surface and the glass surface is expressed by the formula (2).
This is because the distance to the intersection P on the other surface may be shorter than the distance D described above depending on the thickness of the reticle R even if the distance D is more than the distance D.

Therefore, for example, the thickness of the reticle R is 6 mm.
When the refractive index is 1.5, the optical thickness of the reticle R (optical path length 4 mm) is added to set the intersection P to 5.58.
Keep at least mm. At this time, the distance D (≧ 5.58 mm) to the intersection P may be the distance from either the pattern surface of the reticle R or the glass surface. This is because if the value of the distance D is determined in this way, the intersection P is always the distance D of the equation (2) from both the pattern surface and the glass surface.
This is because they will be separated from each other.

However, depending on the thickness of the reticle R, the inside of the reticle R, that is, between the pattern surface and the glass surface, has the above-mentioned intersection P.
Can be set. Depending on the thickness of the reticle R, the intersection P can be calculated from both the pattern surface and the glass surface.
This is because the distance D in (2) (more accurately, the value obtained by multiplying the distance D in Expression (2) by the refractive index of the reticle R) is used. In the above example, since the optical thickness of the reticle R is 4 mm, the midpoint of the reticle R is 2 mm from both the pattern surface and the glass surface.
mm away. Therefore, the above equations (1) and (2) are satisfied, and the intersection P can be set inside the reticle R (for example, the middle point).

The intersection P of the two alignment beams
The position (interval D) is set appropriately with the correction amount (corresponding to the power in the correction lens and the pitch in the phase grating) of the correction optical element arranged on the pupil plane EP of the projection optical system PL. Can be changed arbitrarily. In the above description, the two alignment beams LBW _1, by setting the intersection P of LBW ₂ above the reticle R, viewing the specular reflected light DL _1, DL ₂ intersection Q, as shown in FIG. 4 (A) Front of iris 24W (reticle side)
However, the field stop 24
The intersection Q may be set behind W. At this time, the intersection P of the two alignment beams is set below the reticle R (on the side of the projection optical system). Further, although the transmission type reticle that transmits the exposure light and the alignment beam is described in FIG. 4A, the present invention is directly applied to a reflection type reticle that reflects the exposure light and the alignment beam on the front surface or the back surface. Then, the same effect can be obtained.

Next, each embodiment of the present invention will be described based on the above-mentioned basic configuration (principle). The first embodiment projects the correction plate PC disclosed in Japanese Patent Laid-Open No. 5-160001. The correction plate P is arranged at the center of the pupil of the optical system PL.
The axial chromatic aberration due to C is corrected to a distance D above the reticle R, and further, by using a bifocal element DF or the like, the two beams intersecting on the reticle mark RMx and the distance D above the reticle R are separated. Spatial position and wafer mark WMx
It is configured to use a total of four beams with two beams intersecting at the top. Further, as described above, the distance D is set to satisfy the equation (2). Further, an example of a projection exposure apparatus suitable for applying the present invention is disclosed in Japanese Patent Application Laid-Open No. 4-7814, and as disclosed in this publication also in this embodiment, a dichroic is provided above the reticle R. The exposure light emitted from the illumination system including the light source for exposure is reflected by the dichroic mirror and irradiates the reticle R, and the light beam transmitted from the alignment system passes through the dichroic mirror and reticle marks and It shall be configured to illuminate the wafer mark.

FIG. 3 shows a projection optical system P according to the first embodiment.
It is the figure which showed typically the structure of L, and the optical path of the alignment beam. In FIG. 3, exposure light I ₁ (for example, wavelength 365n
Two beams for injecting _two alignment beams LB ₁ and LB ₂ by injecting a laser beam (for example, a He-Ne laser having a wavelength of 633 mm) having a different wavelength from the i-line of m or an excimer laser having a wavelength of 248 nm or 193 nm). The generation member is omitted in the figure. A two-beam generating member for slightly differentiating the frequencies of the _two alignment beams LB ₁ and LB ₂ is disclosed in, for example, Japanese Patent Application Laid-Open No. 4-45512,
It is disclosed in JP-A-4-7814.

In FIG. 3, two beams LB ₁ and LB
₂ enters the bifocal element DF in parallel with the optical axis of the light transmitting system (DF, OB, etc.). The bifocal element DF is disclosed in, for example, Japanese Patent Application Laid-Open Nos. 2-227602 and 2-23150.
As disclosed in Japanese Patent No. 4 publication, two beams L
B ₁ and LB ₂ are divided into four beams by the birefringence effect. The four beams emitted from the bifocal element DF pass through the objective lens OB to become parallel light beams, and the two parallel beams (for example, p deflection) LBR ₁ and LBR ₂ emitted from the objective lens OB are reticle marks. The remaining two parallel beams (eg, s-deflection) LBW that intersect on RMx
₁ and LBW ₂ are spatial positions distant from the pattern surface of the reticle R by a distance D (that is, by the correction plate PC under the wavelengths of the beams LBW ₁ and LBW ₂ ).
Window WI of reticle R after crossing
It enters the projection optical system PL through the inside.

Two parallel beams LBW ₁ and LBW
₂ is converged as a beam waist at a decentered position in the pupil plane EP by the front lens group GA of the projection optical system PL, and at the same time, a phase grating G on a correction plate (quartz or the like) PC
X _1, a predetermined amount in a predetermined direction by the GX ₂ (phase grating GX
₁ , the diffraction angle of the 1st-order diffracted beam generated from GX ₂ ) is deflected, and enters the rear lens group GB as a divergent beam. Then, the two beams LBW ₁ and LBW ₂ pass through the rear lens group GB to become tilted parallel beams and intersect (image) on the wafer mark WMx.

Now, of the plurality of diffracted lights generated from the wafer mark WMx, for example, ± first-order diffracted lights generated along the optical axis AX of the projection optical system PL (interference light BT of FIG. 4A).
w) passes through the rear lens group GB of the projection optical system PL, is diffracted by the phase grating GX ₃ on the correction plate PC, and becomes a parallel light flux through the front lens group GA. Further, the interference beam (parallel beam) passes through the window WI of the reticle R and passes through the reticle R.
Of two light-transmitting beams L separated by a distance D from the pattern surface of
The light enters the objective lens OB through the intersection of BW ₁ and LBW ₂ . On the other hand, the ± first-order diffracted light (interference light BTr in FIG. 5) generated from the reticle mark RMx in the optical axis direction of the light transmission system also passes through the objective lens OB, and together with the interference light BTw from the wafer mark WMx, the light reception system shown in FIG. To be photoelectrically detected.

[0046] Here, as the imaging state of the light beam I ₂ (wavelength lambda ₁₎ in FIG. 3, the light beam I ₂ generated from a point on the wafer W
However, the distance Lc between the point on the reticle R side where the image is formed on the reticle R side by the projection optical system PL without being corrected by the correction optical elements (PC, CL) (the original axial chromatic aberration of the projection optical system PL). Amount) is larger than the interval D to be set, the phase grating (GX ₁ , G on the correction plate PC
X ₂ , GX ₃ ) or the correction lens CL in FIG. 1 has an action (for example, a positive refracting power) of bending the beam inward (on the optical axis side). On the contrary, when the wavelength λ ₁ of the light flux I ₂ is close to or coincides with the wavelength λ _{0 of the} exposure light,
Since the original axial chromatic aberration amount Lc is smaller than the interval D to be set, the correction lens CL and the phase gratings (GX ₁ , GX ₂ ,
GX ₃ ) has a function of bending the beam outward (for example, negative refracting power).

Next, the light transmitting system of the alignment apparatus for realizing the system of FIG. 3 will be described with reference to FIGS. 6 to 9. Here, instead of creating two light-transmitting beams for the reticle mark RMx and two light-transmitting beams for the wafer mark WMx by the bifocal element DF in FIG. 3, they are separated in advance inside the light-transmitting system. 4 light-transmitting beams LBR ₁ and L
BR ₂ , LBW ₁ and LBW ₂ shall be created.
In FIG. 6, members having the same functions and functions as those in FIG. 4 are designated by the same reference numerals.

Now, in FIG. 6, the reticle mark RM
x and the window WI have the same shape and arrangement as those in FIG. 2, the two beams LB that irradiate the reticle mark RMx.
R ₁ and LBR ₂ and the two beams LBW ₁ and LBW ₂ for irradiating the wafer mark are the Z axis parallel to the optical axis AXa of the objective lens OB of the alignment system and the X axis extending in the pitch direction of the mark RMx. They intersect in the ZX plane including and.
Here, in order to simplify the following description, the objective lens OB
From the projection optical system PL, or a space having a plane conjugate with the reticle in the air is called an image space for convenience, and the space from the front lens group GA to the rear lens group GB of the projection optical system PL or the pupil plane EP. A space having a plane conjugate with is in the air is called a pupil space for convenience.

On the other hand, the optical path from the objective lens OB to the reticle R and the projection optical system PL is Z-Y as shown in FIG. 6B.
When viewed on a plane, the two beams LBR ₁ , LBR ₂ and 2
The book beams LBW ₁ and LBW ₂ are also separated in the Y direction corresponding to the respective center positions of the reticle mark RMx and the window WI. Also, four beams LBR ₁ and LB
R _2, LBW _1, LBW ₂ is within Z-Y plane through the reticle plane to the vertical.

These four beams are as shown in FIG.
Beam splitter 16 and lens system 1 from the laser light source side
4, incident on the objective lens OB via the reflecting mirror 12 and the lens system 10. In FIG. 6, the Fourier transform plane EP ₁
Is formed in the pupil space between the objective lens OB and the lens system 10 and is conjugate with the pupil plane EP in the projection optical system PL. Also, the surface R'in the image space between the lens system 10 and the lens system 14 is conjugate with the reticle surface, where Z-
Two beams LBR ₁ and LBR ₂ intersect in the X plane. The plane P ₀ ′ in the same image space is the wafer plane P
A ₀ each a conjugate, the two beams LBW _1, LBW ₂ intersect in the Z-X plane.

Beam splitter 16 arranged in pupil space
Are generated vertically from the reticle mark RMx, and include the objective lens OB, the lens system 10, the reflecting mirror 12, and the lens system 14.
The interference light BTr passing therethrough is reflected to the light receiving system and is generated perpendicularly from the wafer mark WMx, and the rear lens group GB of the projection optical system PL and the correction element (correction lens CL,
Alternatively, GX ₃ ) in the case of a phase grating, the front lens group GA, the window WI of the reticle R, and the objective lens OB to lens system 14
The interference light BTw passing therethrough is reflected to the light receiving system. Further, EP ₁ 'in FIG. 6 is a plane conjugate with the Fourier transform plane EP ₁ .

The four beams LBR ₁ , LBR ₂ and L
Since BW ₁ and LBW ₂ are relayed so as to be parallel light fluxes in the image space, they are both convergent and divergent light fluxes in the pupil space (Fourier transform plane). next,
The system in the previous stage that realizes the beam optical path as shown in FIG.
The laser light source will be described with reference to FIG. As shown in FIG. 7, the beam LB from the He—Ne laser light source is transmitted to the first AOM (acousto-optic modulator) 50 by Raman-Nass (Ram
an-Nath) The light is incident so as to satisfy the diffraction condition. AOM5
0 is driven by an appropriate high frequency signal (for example, a frequency of 80 MHz) and generates ± first-order diffracted beams LBa and LBb and a zero-order beam LB ₀ having a diffraction angle according to the frequency f ₁ . The pitch direction (direction of traveling wave) of the modulation grating of the AOM 50 is determined so that the first-order diffracted beams LBa and LBb spread in the Z-Y plane.

Now, the first-order diffracted beams LBa, LBb
Passes through the spatial filter 52 that cuts the 0th-order beam LB ₀ and other higher-order beams, and then intersects at the plane P ₀₂ by the lens system 54. This plane P ₀₂ is conjugate with the plane P ₀₃ where the diffraction point of the AOM 50 is located. Therefore, the beam LB
Is a parallel light beam, both the + 1st-order diffracted beam LBa and the -1st-order diffracted beam LBb reaching the surface P ₀₂ are parallel light beams. If the frequency of the beam LB is f ₀ , +1
The frequency of the next-order diffracted beam LBa becomes f ₀ + f ₁ , and −1
The frequency of the next-order diffracted beam LBb becomes f ₀ −f ₁ .

By the way, the diffraction point of the second AOM 56 is located on the surface P _02, and the two primary beams LBa and LBb are A
It is incident on the OM 56 so as to satisfy the condition of Bragg diffraction. However, the pitch direction of the modulation grating of AOM56 (direction of traveling wave) is 45 ° with respect to that of AOM50.
It is set to the position rotated only about the optical axis. The reason why the AOM 56 is rotated by 45 ° is that the first-order diffracted beam and the 0th-order beam generated by the two beams LBa and LBb simultaneously incident on the AOM 56 can be separated most in space. Is. That is, the beam L
When Ba and LBb are incident on the AOM 56, the respective 0th-order beams thereof spread and advance in the Z-Y plane. For example, the + 1st-order diffracted beam from the beam LBb and the beam LBa.
-1st-order diffracted beam from Z is orthogonal to Z-Y plane.
-Expand and proceed in the X plane. Therefore, the first-order diffracted beam and the 0th-order beam from the AOM 56 are transferred to the lens system 58.
When passing through the relay system by the lens system 62 and the lens system 62, the spatial filter 60 is arranged on the Fourier transform plane EP ₃ 'in the relay system, and the apertures on the spatial filter 60 are symmetrical in the X direction with the optical axis interposed therebetween. Beam L
The + _1st order diffracted beam made from Bb can be extracted as beam LB ₁ , and the −1st order diffracted beam made from beam LBa can be extracted as beam LB ₂ .

Therefore, the drive frequency of the AOM 56 is set to f ₂
Then, the frequency of the beam LB ₁ is (f ₀ −f ₁ ) + f ₂
And the frequency of the beam LB ₂ becomes (f ₀ + f ₁ ) −f ₂ . Therefore, the frequency difference Δf between the _two beams LB ₁ and LB ₂ becomes 2 (f ₁ −f ₂ ), and f ₁ = 80 MHz, f
_{When 2} = 80.050 MHz, a beat frequency of Δf = 50 KHz is obtained. The two beams LB ₁ and LB ₂ thus obtained intersect at the plane P ₀₁ , but the plane P ₀₁ is the plane P _01.
Since it is conjugated with ₀₂ , it is also conjugated with the plane P ₀₃ . However, the interference fringes generated when an object is placed on the surface P ₀₁ have a pitch in the X direction because the beams LB ₁ and LB ₂ are in the Z-X plane, and have a speed corresponding to the frequency Δf. It is flowing in the X direction. On the other hand, the modulation grating in the first-stage AOM 50 has a pitch in the Y direction, and apparently flows in the Y direction at a speed corresponding to the frequency f ₁ .

As described above, with the configuration of FIG. 7, two basic beams LB ₁ and LB ₂ having the frequency difference Δf for the heterodyne system can be obtained. This configuration is 2
A method using a plurality of AOMs in parallel is called a tandem method, and has a great advantage over the parallel method when the alignment beams have multiple wavelengths. This will be described in detail later.

FIG. 8 shows a four-beam generation system following FIG.
8A is an optical path seen in the Z-Y plane.
8 (B) is an optical path seen in the Z-X plane. Shown in Figure 7
I did not do it, but two beam LB₁, LB₂Crossed
Face P₀₁Beam on the reticle or wafer
A field stop 64 that determines the shape and size (width) is arranged.
You. The field stop 64 has a reticle mark RMx and
Has a transparent opening similar to the outer shape of the wafer mark WMx.
And the two beams LB that passed through the aperture₁, LB ₂Is
Beam splitting with split surface parallel to Z-X plane
The light is obliquely incident on one surface of the mirror 66. By this
From the beam splitter 66, two beams LB₁, LB₂
Of the four beams LB, each of which is divided by about 1/2 intensity
R₁, LBR₂, LBW₁, LBW₂(Parallel light respectively
A bundle emits in the Z-Y plane in parallel with the optical axis. Four
The beam is relayed to the next stage by the lens system 68.
However, the lens system 68 has a surface P₀₁Fourier transform plane EP for
₂Converge each of the four beams in the beam waist
You. As is apparent from FIG. 8, it was observed in the ZX plane.
When the Fourier transform plane EP₂'In the pupil space where
Room LBR₁And LBR₂Is symmetrical about the optical axis,
Located parallel to each other. This is beam LBW₁When
LBW₂Is the same. This beam LBR₁,
LBR₂And beam LBW₁, LBW₂And Y direction separation
The two beams L incident on the beam splitter 66 are
B₁, LB₂Is translated in the Y direction in FIG.
It can be changed by
A rotatable parallel flat plate between the beam splitter 64 and the beam splitter 66.
Place the plate glass and make its rotation axis parallel to the X axis.
Yes. Beam LBR by beam splitter 66₁, LB
R₂And beam LBW₁, LBW₂The separation from
As shown in (B), the mark RMx on the reticle and the window WI are
It is set corresponding to the interval in the Y direction.

FIG. 9 shows an optical path length correction system (also shown in FIG. 8).
2 optics) or two optics for the reticle.
Room LBR₁, LBR₂The intersection position in the image space of
Two beam LBW for wafer₁, LBW₂In the image space of
A certain amount of crossing position in the optical axis direction (result
It works by shifting the distance D) on the wheel. Figure 9 (A)
A view of the optical path length correction system seen in the Z-Y plane, and FIG.
Two beam LBW ₁, LBW₂The optical path of Z-X plane
The view seen inside, and FIG. 9 (C) are mainly two beams LB.
R₁, LBR₂It is the figure which looked at the optical path of in the ZX plane.

As shown in FIG. 9A, the four beams L
BR ₁ , LBR ₂ , LBW ₁ and LBW ₂ are lens systems 70
Incident on the split prism 72 through the two beams LB
W ₁ and LBW ₂ are total reflection surfaces 72A in the split prism 72.
Is reflected at a right angle by, and is bent at a right angle again by the mirror 74,
In the Z-Y plane, it travels parallel to the optical axis of the lens system 70. The two beams LBW ₁ and LBW ₂ intersect with each other on the plane P _0W ′ by the lens system 70 when viewed in the ZX plane as shown in FIG. 9B. Then, the beams LBW ₁ and LBW ₂
Passes through the synthetic prism 76 (substantially the same as the prism 72) and enters the lens system 82. Lens system 82
The front-side focal position of is the surface R ′ in FIG.
Is shifted from the surface P _0W 'in the optical axis direction, Z-X
When viewed in a plane, the two beams LBW ₁ and LBW ₂ that have passed through the lens system 82 do not become parallel to the optical axis and proceed with a slight convergence tendency. Here, there is an image space between the lens systems 70 and 82, the surface P _0W ′ is a surface conjugate with both the surface of the wafer W and the surface P _{0 in} FIG. 6, and the surface R ′ is of the reticle R. It is a surface conjugate with the pattern surface. Therefore, between the lens systems 70 to 82, the beams LBW ₁ and LBW are
Each of ₂ is a parallel light flux.

On the other hand, the reticle incident on the split prism 72
2 beam LBR for Le₁, LBR₂Is the Z-Y plane
When viewed inside, the light becomes parallel to the optical axis and is transmitted as it is.
The light enters the optical block 78 for path length correction. This beam
LBR₁, LBR₂Is as shown in FIGS. 9 (A) and 9 (C).
Originally, the surface P_r0It was set to intersect at '
The optical block 78 extends the optical path length,
After being bent at a right angle by the beam 80, the surface R ′ in FIG.
Cross at. Beam LBR reflected by mirror 80 ₁, L
BR₂Is at a right angle on the total reflection surface 76A of the composite prism 76.
It is bent and becomes parallel to the optical axis and enters the lens system 82.
You. At this time, two beam LBR₁, LBR₂Is
The parallel light flux between the lens systems 70 and 82 is
Because it intersects at R ', after passing through the lens system 82
This also goes parallel to the optical axis.

The four beams that have passed through the lens system 82 described above are incident on the system shown in FIG. That is, FIG.
The Fourier transform plane EP ₁ 'in (A) is E in FIG. 6 (B).
The four beams are arranged so as to be at the same point as P ₁ 'and the four beams are the beam splitter 16, the lens system 14, ... In FIG.
The light is transmitted in the order of. FIG. 10 shows 4 shown in FIG. 8 and FIG.
This is a modified example in which the beam generation system and the optical path length correction system are collectively realized in one image space. And FIG. 10 (A) is Z
FIG. 10B shows the system from the field stop 64 to the lens system 82 in the −Y plane, and FIG. 10B shows the system in the ZX plane.

Two beams L passing through the field stop 64
B ₁ and LB ₂ are split into two by a beam splitter 84, and the two beams horizontally transmitted through this splitter 84 also pass through the combining prism 76 as they are and the lens system 8
2 and is emitted as beams LBR ₁ and LBR _{2 in} the same state as in FIG. On the other hand, the two beams reflected by the beam splitter 84 are reflected by the mirrors 85 and 86 as shown in the figure, and further, the total reflection surface 76 of the combining prism 76.
The beam LB is reflected by A, enters the lens system 82 in parallel with the optical axis in the Z-Y plane, and is in the same state as in FIG.
Eject as W ₁ and LBW ₂ .

In the system shown in FIG. 10, the beams LBW ₁ and LBW ₂ have a longer physical optical path as they pass through the mirrors 85 and 86. That is, in FIG. 10B, the plane P
The field stop 64 of ₀₁ is conjugated with the mark RMx of the reticle R, but the beam LB incident on the lens system 82 is
The intersection of W ₁ and LBW ₂ is located on the surface P ₀₁ ′ closer to the laser light source than the surface P ₀₁ due to the optical path routed by the mirrors 85 and 86. Therefore, the distance in the optical axis direction between the planes P ₀₁ and P ₀₁ 'corresponds to the distance D shown in FIGS.

According to the system shown in FIG. 10, the relay system including the lens systems 68 and 70 in the system shown in FIGS. 8 and 9 is omitted, so that the light transmitting section of the alignment apparatus becomes compact. As shown in FIG. 10A, a parallel plate-shaped optical block 87 having a variable thickness is inserted between the beam splitter 84 through which the _two beams LBR ₁ and LBR ₂ pass and the combining prism 76. , Where beam LBR ₁ ,
The intersection of LBR _{2 in} the image space and the beams LBW ₁ , L
The distance in the optical axis direction from the intersection position of BW _{2 in} the image space may be finely adjusted. Also, if you want to change the optical path length greatly,
In FIG. 10A, the mirrors 85 and 86 may be integrally moved in parallel in the Y direction.

As described above, the light transmitting section of the alignment apparatus according to the present embodiment is constituted by FIGS. 6 to 10, and FIG.
In the pupil space from the objective lens OB to the lens system 10, the wafer mark beam LBW ₁ ,
The LBW ₂ is tilted with respect to the optical axis AXa in the ZX plane, and the reticle mark beams LBR ₁ and LB
R ₂ is parallel to the optical axis AXa in the ZX plane, but the relationship may be reversed. Rather, the opposite is the movement of the objective lens OB according to the mark arrangement (optical axis A
This is advantageous when considering the movement of the position of Xa on the reticle R in the X and Y directions. Specifically, the rear focal plane of the objective lens OB is made to coincide with the plane P ₀ in FIG. 6, and the beams LBW ₁ and LBW ₂ passing through the pupil space of the system are sandwiched with the optical axis in the ZX plane. Beams may be transmitted so as to be parallel to each other.

This will be further described with reference to FIG. FIG. 11 is a perspective view showing a system from the objective lens OB to the Fourier transform plane (pupil conjugate plane) EP ₁ , in which the positions of the mark RMx on the reticle R and the window WI are different for each reticle. In order to deal with this, the objective lens OB and the mirror 90 are integrally movable in the Y direction (radiation direction within the visual field of the projection optical system PL).

As shown in FIG. 11, the two beams LBW ₁ and LBW _{2 which} are parallel to the optical axis AXa pass through the Fourier transform plane EP ₁ and are bent at a right angle by the mirror 90 to form the objective lens OB.
When incident on the objective lens OB and the lens system 10, since the two beams LBW ₁ and LBW ₂ are an afocal system, the objective lens OB and the mirror 90 are
, The two beams LBW ₁ ,
The intersection position of LBW ₂ always exists on the plane P ₀ . However, the two beams LBW ₁ , L in the pupil space including the plane EP ₁
Unless BW ₂ is parallel to the optical axis AXa, two beams LB are moved every time the objective lens OB and the mirror 90 are moved.
The crossing position of W ₁ and LBW ₂ slightly moves in the vertical direction with respect to the plane P ₀ . That is, the conjugate relationship changes. The amount of fluctuation is two beams LBW ₁ in the pupil space,
Although it depends on the inclination of LBW ₂ with respect to the optical axis AXa, if the objective lens OB is a telecentric system, the variation is the focal length of the objective lens OB (objective lens OB
The distance from the main surface to the reticle) Lf is determined by the ratio of the distance D. If D / Lf is small, the amount of change in the conjugate relationship with respect to the unit movement amount of the objective lens OB and the mirror 90 in the Y direction is also small.

Now, in FIG. 11, the beam LBW for the wafer is used.
₁ and LBW ₂ are set to be an afocal system, so that the above-described change in the conjugate relationship does not occur. This means that the conjugate relationship of the reticle beams LBR ₁ and LBR ₂ varies. However, since the crossing angle of the _two beams LBR ₁ and LBR ₂ is extremely small on the reticle, the two beams LBR ₁ and LB 2
The length (depth) of the intersection space of R ₂ in the optical axis AXa direction becomes extremely large, and the fluctuation amount of the conjugate relation on the reticle can be ignored in practical use.

Even if it cannot be ignored, the beam LBR
₁ , the irradiation areas on the reticle marks RMx of LBR ₂ are gradually separated in the X direction (pitch direction), and the width of the irradiation areas (range where interference fringes are generated) overlapping each other in the X direction is small. Beam LBR ₁ , LBR
The aperture shape of the field stop 64 may be set so that the sectional shape of _{2 is} a rectangle long in the X direction on the reticle. In addition, in the case of the reticle mark RMx, since it is made by etching the chrome layer, stable grating performance can be obtained, so that the beams LBR ₁ and LBR ₂ can be obtained.
Even if the width in the X direction (X pitch direction) of the interference fringe generation range is reduced and the number of lattice lines included in the range is reduced to about half (for example, 20 to 10), The interference light BTr can be satisfactorily obtained.

However, if the beam LB for the wafer
When the conjugate relation changes in W ₁ and LBW ₂ , naturally the entire width in the pitch direction of the interference fringes formed on the wafer mark WMx decreases, and the inside of the wafer mark WMx that contributes to the generation of the interference light BTw is naturally reduced. The number of grids also decreases. In general, a wafer mark does not have a stable grid performance as that of a reticle mark, and therefore it is desirable to capture as many grid lines as possible in terms of accuracy. For the above reasons, it is advantageous that the afocal system is achieved for the wafer beams LBW ₁ and LBW ₂ incident on the objective lens OB.

Next, the structure of the light receiving system shown in FIG. 6 will be described with reference to FIG. In FIG. 5, the same function as in FIG.
The same members are designated by the same reference numerals. As described above, the interference light BTr from the reticle mark RMx and the interference light BTw from the wafer mark WMx are both the optical axis AXa of the objective lens OB in the image space in this embodiment.
And reaches the beam splitter 16 via the lens system 10, the mirror 12, and the lens system 14. The interference lights BTr and BTw split by the beam splitter 16
Is the Fourier transform plane (the conjugate plane of the pupil plane EP) as shown in FIG.
It passes through EP ₄ ′, is reflected by the mirror 20, is subjected to the inverse Fourier transform by the lens system 22, and is split into two by the beam splitter 23. Then, on the optical path that has passed through the beam splitter 23 and on the surface P ₀ (two beams LBW ₁ and LBW ₂
A field stop 24W is arranged at a plane (including a crossing point P), that is, at a position conjugate with the wafer surface (crossing point P'in FIG. 4A). The field stop 24W has an opening corresponding to the size of the wafer mark WMx (or an interference fringe generation region on the wafer mark), extracts only the interference light BTw, and cuts other light (for example, interference light BTr). To do. Extracted interference light B
Tw is received by the photoelectric element 26W, and an AC signal SSw having a beat frequency (for example, 50 KHz) is obtained.

In this embodiment, the distance D between the surface P ₀ and the pattern surface of the reticle R is expressed by the equation as described above with reference to FIG.
(2) is satisfied, and the field stop 24W is installed at a position conjugate with the surface P ₀ . Therefore, even if the specularly reflected light generated from the surface of the reticle R reaches the field stop 24W, there is no opening edge in the overlapping portion, so that even if the specularly reflected light hits the opening edge and diffracted light or scattered light is generated, The diffracted and scattered light does not have a beat frequency with a frequency difference Δf. Therefore, even if the specular reflection light is diffracted and scattered at the aperture edge of the field stop 24W and is mixed in the interference light BTw, the photoelectric element 26W does not detect a noise component having the same beat frequency as the interference light BTw, and in the heterodyne system. The specularly reflected light does not become noise.

On the other hand, a field stop 24R is arranged at a position conjugate with the reticle surface (mark surface) on the optical path in the direction reflected by the beam splitter 23. The field stop 24R has an opening corresponding to the size of the reticle mark RMx (or the interference fringe generation region on the reticle mark RMx), extracts only the interference light BTr, and cuts other light (for example, the interference light BTw). To do. The extracted interference light BTr is the photoelectric element 26R.
An AC signal SSr having a beat frequency (for example, 50 KHz) is received by.

Therefore, two AC signals (detection signals) SS
If the phase difference between w and SSr is measured, the reticle mark RMx and the wafer mark WMx in the X direction (pitch direction).
The relative positional deviation amount of is determined within a range of ± 1/4 of the mark pitch. For example, the pitch of the reticle mark RMx is 10 μm, the reduction ratio 1 / M is 1/5, and the pitch of the wafer mark WMx is 2 μm.
Assuming that 80 ° can be detected with a resolution of about 2 °, ± 0.5 μm corresponds to ± 180 ° on the wafer.
The change in phase difference of 2 ° is 0.5 (2/180) ≈ 0.00
This corresponds to a positional deviation amount of 56 μm. An apparatus for obtaining the relative positional deviation amount between the reticle R and the wafer W based on the above-mentioned phase difference is disclosed in Japanese Patent Application Laid-Open No. Hei 4-7814, etc., and therefore is not shown in any of the drawings. is there.

Now, in order to measure the phase difference between two AC signals (sine waves) SSr and SSw having the same frequency, two signal waveforms are digitalized at the same timing in response to a clock having a frequency sufficiently higher than the frequency. It is desirable to perform Fourier transform or Fourier integration by a computer (processor) after sampling and loading in the waveform memory. In this case, the greater the number of cycles (number of waves) that are taken into the waveform memory and used for Fourier calculation,
Although the measurement accuracy is improved by the effect of reducing random noise and the like, the calculation time becomes longer accordingly.

Generally, in the case of Fourier calculation, a sine wave and a cosine wave of a reference frequency (here, a beam frequency) are used in the calculation process to calculate the phase component of the signal SSr with respect to the reference frequency and the phase component of the signal SSw with respect to the reference frequency. After being calculated individually, the signal SSr,
The phase difference of SSw is calculated. Therefore, when the phase component of the signal SSr with respect to the reference frequency is obtained, the reticle mark RMx has a stable lattice performance, and therefore the number of cycles of the signal SSr used for calculation is reduced within a range where a predetermined accuracy can be obtained. , When the phase component of the signal SSw with respect to the reference frequency is obtained, the shape deterioration of the wafer mark WMx,
The number of cycles of the signal SSw used for the calculation is variable according to the degree of step (concavo-convex), surface roughness, decrease in reflectance, and the like. In this way, if the wafer mark WMx is well formed, the calculation time can be shortened, and if the lattice performance of the wafer mark WMx is deteriorated, the calculation time will be longer, but the measurement accuracy will be improved. It has the advantage of not being dropped.

Such a decrease in the wafer mark lattice performance can also be expected by obtaining the amplitude of the signal SSw from the photoelectric element 26W in the waveform memory. Therefore, if the amplitude is small, the Fourier calculation is performed. It suffices to increase the number of cycles of the signal SSw used for. Wafer mark W
Since M is provided for each shot area on the wafer W, the lattice performance may change depending on the position of the mark WM on the wafer W. Therefore, it is possible to increase or decrease the number of cycles used for calculation depending on the amplitude of the signal SSW, not only can the measurement accuracy of the mark position be substantially the same regardless of the position on the wafer, but also disclosed in, for example, Japanese Patent Laid-Open No. 61-44429. As described above, the global alignment (EGA) method in which the marks WM of a plurality of shots are sequentially measured in advance has an advantage that the processing time can be shortened while maintaining high measurement accuracy.

The above-mentioned TTR type alignment apparatus is, for example, a step-and-repeat type projection exposure apparatus as disclosed in Japanese Patent Laid-Open No. 4-45512, or Japanese Patent Laid-Open No. 4-307720. Incorporated in a step-and-scan type scanning projection exposure apparatus as disclosed in US Pat. When incorporated in such an apparatus, it may be necessary to measure the position of the reticle R alone or the wafer W alone for the convenience of the system.

In that case, as disclosed in Japanese Patent Laid-Open No. 4-45512, the _two light-transmitting beams LB ₁ and LB ₂ shown in FIGS. 6 to 10 and the light-transmitting beam LB are shown.
_Two of R ₁ and LBR ₂ , or the transmitted light beam LBW ₁ ,
At least one set of two LBW ₂ is branched from the light transmission path by a beam splitter and intersects on the reference grating plate, and the interference light (± 1st order diffracted light) generated from the reference grating plate is received by the photoelectric element. , To obtain the reference signal (beat frequency). Then, if the phase difference between the reference signal and the signal SSr is obtained, the positional deviation amount of the reticle mark RMx with respect to the reference lattice plate can be known, and the alignment of the reticle alone with the reference lattice plate can be performed. Similarly, on the wafer side, a single unit can be aligned with the reference lattice plate. Next, a second embodiment of the present invention will be described with reference to FIGS.
As disclosed in Japanese Patent Laid-Open No. 215, the alignment beam is made to have multiple wavelengths. Here, even if the number of wavelengths is increased, a plurality of light-transmitting beams having extremely different wavelengths are not simultaneously transmitted, but for example, a He-Ne laser having a wavelength of 633 nm and a semiconductor laser (LD having an oscillation spectrum at a wavelength of 670 to 710 nm) ) And, the wavelength is 40-80n
At least two wavelengths separated by about m are used.

In order to increase the number of wavelengths, the beam LB incident on the first-stage AOM 50 shown in FIG. 7 is converted into a parallel beam (633 nm) from a He-Ne laser light source and a parallel beam (for example, from a semiconductor laser light source). 690 nm) and a beam splitter or the like to coaxially combine them. In this way, the light is incident on the reticle mark RMx or the wafer mark WMx.
Book beams LBR ₁ , LBR ₂ , or LBW ₁ , LBW
₂ is 2 wavelengths both, the crossing angle of the two beams to produce interference fringes can be made different by an amount corresponding to the wavelength difference (690-633nm). It should be noted that beams from different laser light sources do not interfere with each other because they do not have temporal coherency and spatial coherency at all, and interference fringes due to the intersection of the He-Ne laser beam and the semiconductor laser beam do not occur. ..

Next, referring to FIG. 12, the beam transmission to the wafer mark WMx when the number of wavelengths is increased will be described. In FIG. 12, for the wafer mark WMx having a pitch in the X direction, the wavelength λ ₁ (He-Ne laser)
Two beams LBW ₁ and LBW ₂ are incident at an incident angle ± θ ₁ , and two beams LBW ₁ ′ and LBW ₂ ′ having a wavelength λ ₂ (semiconductor laser) are incident at an incident angle ± θ ₂ . Since these incident angles θ ₁ and θ ₂ are determined with reference to the pitch (diffraction angle) of the modulation grating in the first-stage AOM 50, the two beams LBW ₁ and LBW ₂ with incident angles ± θ ₁ The interference fringes generated by the intersection and the interference fringes generated by the intersection of the two beams LBW ₁ ′ and LBW ₂ ′ with an incident angle of ± θ ₂ are superposed at exactly the same pitch with zero phase difference. are doing.

Therefore, the two beams LBW ₁ and LBW ₂
The interference beat light BTw generated from the wafer mark WMx by the irradiation of the laser beam and the interference beat light BTw 'generated from the wafer mark WMx by the irradiation of the _two beams LBW ₁ ′ and LBW ₂ ′ are coaxial with each other in the vertical direction. Go to
Photoelectric detection is simultaneously performed by the photoelectric element 26W in FIG. At this time, the signal SSw from the photoelectric element 26W has a single AC waveform having a beat frequency.

As described above, at least the wafer mark W
By making the two beams for irradiating Mx multi-wavelength, the phase measurement error due to the influence of thin film interference in the photoresist layer coated on the wafer W with a substantially uniform thickness (about 1 μm) is reduced. Further, for the multi-wavelength conversion, a beam from a light source having a broad intensity distribution within a predetermined bandwidth (for example, 40 to 100 nm) may be used. Stage A
If it is incident on the OM 50 as a parallel light beam, the incident beam continuously exists on the wafer mark WMx between the incident angles θ ₁ and θ _{2 in} FIG. 12, for example. However, in this case as well, 1 corresponding to the pitch of the wafer mark WMx is formed on the wafer.
Only two interference fringes are generated.

When the above multi-wavelength is implemented, the axial chromatic aberration (or the chromatic aberration of magnification) is slightly different depending on the wavelength range (bandwidth), and 2 depending on the wavelength.
Due to the different crossing angles of the two beams, the diameter (size) or the interval of the two beams irradiating the wafer mark WMx on the correction plate PC has a spread corresponding to the wavelength range.

FIG. 13 shows two beams LBW ₁ and LBW having a wavelength of 633 nm directed to the wafer mark WMx.
₂ and two beams of wavelength 690 nm LBW ₁ ', LBW
Shows an example of distribution on the phase grating A1, A2 for sending and ₂ '. The beams LBW ₁ and LBW ₂ have elliptical beam cross-sectional shapes that are long in the Y direction, and are arranged on the pupil plane EP at intervals Xh in the X direction corresponding to the incident angle ± θ ₁ . And the beam L
BW ₁ ', LBW _2' is the beam cross section is a long ellipse in the Y-direction, incidence angle _{± θ 2 (θ 2> θ} 1) interval in the X direction on the pupil plane EP corresponds to Xd (Xd > Xh).

When the laser is used as the light source, the shape of the irradiation area on the wafer (the field stop 6 in FIG. 8 or 10)
4) is rectangular, the beam cross section (spot) on the phase grating for light transmission (A1, A2) becomes a long ellipse in the direction orthogonal to the rectangle due to the influence of diffraction. The overall shape of each of the optical phase gratings is a rectangle, and the width in the interval direction is determined according to the difference between the intervals Xh and Xd corresponding to the increase in wavelength and the spot size. Further, when the difference between the _two wavelengths λ ₁ and λ ₂ is large, the intervals Xh and Xd of the respective light-transmitting beams on the pupil plane EP are separated without overlapping with each other, so that the light-transmitting beams of the respective wavelengths are separated from each other. It is better to provide a phase grating for light transmission, and in this case, it is desirable to slightly change the pitch of the Fresnel grating for each phase grating for each wavelength.

FIG. 14 schematically shows the incident state of a light beam for a wafer mark on a reticle R with _two wavelengths λ ₁ and λ ₂ . The center lines of the _two light-transmitting beams LBW ₁ and LBW _{2 having} a wavelength λ ₁ = 633 nm intersect in a plane P _0b slightly lower than the plane P ₀ (on the reticle side) and have a wavelength λ ₂ = 690 nm. Light beam LB of book
The center lines of W ₁ ′ and LBW ₂ ′ are emitted from the objective lens OB so as to intersect with each other in a plane P _0a slightly above the plane P ₀ (on the side of the objective lens OB). At this time, the distances Da and Db between the surfaces P _0a and P _{0b and} the reticle pattern surface are D = (Da
+ Db) / 2 is set. Therefore, the plane P ₀ is conjugate with the wafer plane at the average wavelengths of the two wavelengths 633 nm and 690 nm, and will be called the average conjugate plane.

At this time, assuming that the respective transmitted beams of wavelengths λ ₁ and λ ₂ pass through the same transmitting phase grating on the correction plate PC, the axial chromatic aberration is corrected (the conjugate between the wafer surface and the surface P _0). )
Is designed for the average wavelength, so the distance (Da-Db) between the surfaces P _0a and P _0b is the amount of chromatic aberration corresponding to the difference between the _two wavelengths λ ₁ and λ ₂ . If the amount of chromatic aberration due to the difference is sufficiently small on the reticle side (for example, 1 mm or less), the intersection plane P _0b of the beams LBW ₁ and LBW ₂ of the wavelength λ ₁ is obtained.
And the crossing plane P _0a of the beams LBW ₁ ′ and LBW ₂ ′ having the wavelength λ ₂ may be matched with the average conjugate plane P ₀ . However, in this case, the crossing plane of the beams LBW ₁ and LBW _{2 and} the crossing plane of the beams LBW ₁ ′ and LBW ₂ ′ on the wafer surface are projected in the optical axis AXa direction with respect to the differential chromatic aberration amount on the reticle side ( It will be shifted by the square of 1 / M). That is,
The amount of differential chromatic aberration on the reticle side is 1 mm, and the projection magnification is 1 /
5, the difference chromatic aberration amount on the wafer side is 1000/2.
5 = 40 μm. Further, when the amount of differential chromatic aberration on the reticle side is relatively large, the intersection planes are shifted on the reticle side for each light-transmitting beam of each wavelength as shown in FIG. 14, and the conjugate planes of these intersection planes are formed on the wafer surface. It is desirable that they match.

Since the light-transmitting beam for the reticle mark does not necessarily have to have multiple wavelengths, either the beam having the wavelength λ ₁ or the beam having the wavelength λ ₂ is selected and irradiated onto the reticle mark. As described above, a color filter, a dichroic mirror, or the like may be provided in the light transmission system. Alternatively, the respective beams of wavelengths λ ₁ and λ ₂ are applied to the reticle mark at the same time, and in the optical path to the photoelectric element 26R that receives the interference light BTr from the reticle mark in the light receiving system of FIG. A filter for extracting either λ ₁ or λ ₂ may be provided.

Further, as shown in FIG. 14, a beam L having a wavelength λ ₁
BW ₁ , LBW ₂ and beam LBW ₁ ′, LBW of wavelength λ _2
In order to shift the crossing planes P _0b and P _0a with _{2'in the} optical axis direction, various lens systems of the alignment system and the objective lens OB are used.
Then, chromatic aberration commensurate with it should be generated. An example of the objective lens system having an image forming point on a different plane depending on the wavelength is disclosed in, for example, Japanese Patent Laid-Open No. 59-42517.

[0091]

As described above, according to the present invention, a correction element for controlling the axial chromatic aberration in the optical path through which the beam for alignment passes is provided at or near the pupil plane of the projection optical system, and the pattern surface of the mask is provided. Since the surface spaced apart from the (mark surface) by a certain amount in the optical axis direction and the surface of the photosensitive substrate are conjugated with respect to the alignment optical path, the light reflected by the surface of the mask and incident on the light receiving system becomes a noise component. It is possible to prevent the light from entering the alignment light from the substrate mark. Therefore, the distortion of the detection signal of the substrate mark is particularly reduced, and the effect of improving the accuracy of the relative displacement measurement between the mask and the photosensitive substrate is obtained.

[Brief description of drawings]

FIG. 1 is a diagram illustrating an alignment method in a conventional projection exposure apparatus.

FIG. 2 is a perspective view schematically showing mark arrangement on a reticle and mark arrangement on a wafer.

FIG. 3 is a diagram illustrating the principle of an alignment method according to each embodiment of the present invention.

FIG. 4 is a diagram schematically showing a basic configuration of an alignment optical system according to an example of the present invention.

FIG. 5 is a diagram showing a configuration of a light receiving system combined with an alignment optical system according to an embodiment of the present invention.

FIG. 6 is a diagram showing a part of the configuration including an objective lens in the alignment optical system according to the embodiment of the invention.

7 is a perspective view showing a system for forming a heterodyne type transmitted light beam supplied to the alignment optical system of FIG. 6;

FIG. 8 is a diagram showing a system of a four-beam generation member which makes two light-transmitting beams into four, which is created by the system of FIG.

FIG. 9 is a view showing a system of a bifocal member for supplying the four beams created by the system of FIG. 8 to the system of FIG. 6 by giving an optical path difference between the reticle beam and the wafer beam.

FIG. 10 is a view showing a modified example in which the systems of FIGS. 8 and 9 are combined to simultaneously perform four-beam formation and two-focus formation.

FIG. 11 is a perspective view showing a state when the objective lens of the alignment optical system is made movable.

FIG. 12 is a diagram showing a state of beam incidence when a light-transmitting beam to a mark on a wafer has multiple wavelengths.

FIG. 13 is a diagram showing the arrangement of multi-wavelength transmitted beams that pass through a correction element.

FIG. 14 is a diagram illustrating an optical path on a reticle of a multi-wavelength transmitted beam.

[Explanation of symbols]

R ... Reticle PL ... Projection optical system EP ... Pupil plane W ... Wafer RMx ... Reticle mark WI ... Window WMx ... Wafer mark PC ... Correction plate GX ₁ , GX ₂ , GX ₃ ; A 1, A 2, A 3 ... Phase grating LBW ₁ , LBW ₂ ... Light-transmitting beam for wafer LBR ₁ , LBR ₂ ... Light-transmitting beam for reticle BTw ... Wafer mark interference light BTr ... Reticle mark interference light

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location G03F 9/00 H H01L 21/30 522 D

Claims (6)

[Claims] 1. An illumination system for irradiating a mask on which an original image pattern is formed with exposure light, a projection optical system for projecting the original image pattern on a photosensitive substrate, and an alignment beam having a wavelength different from that of the exposure light for the mask. Beam irradiation for irradiating the first lattice mark formed on the first substrate and the transparent portion in the vicinity thereof and for irradiating the second lattice mark formed on the photosensitive substrate through the transparent portion and the projection optical system. Means, and a light receiving system having a photoelectric detector for receiving diffracted light generated from the second grating mark and traveling through the projection optical system and the transparent portion, and diffracted light generated from the first grating mark. Based on the output signal of the photoelectric detector,
In a projection exposure apparatus including a position detection unit that detects a relative positional deviation between the mask and the photosensitive substrate, the beam irradiation unit includes two first alignment beams intersecting on the first lattice mark. And two second alignment beams that intersect each other in a space separated from the surface of the mask by a predetermined distance D so as to pass only the transparent portion without irradiating the first grating mark. A generating member; the light receiving system has a field stop installed at a position conjugate with an intersection of the two second alignment beams; and a plane of the pupil of the projection optical system or a plane in the vicinity thereof. Is arranged in a portion through which the two second alignment beams pass, and the two second alignment beams are deflected so as to intersect with each other on the second grating mark via the projection optical system. A plurality of optical elements; correction of the original axial chromatic aberration amount Lc of the projection optical system generated on the mask side at the wavelength of the alignment beam by the plurality of optical elements until it becomes substantially equal to the distance D; A characteristic projection exposure apparatus. 2. The plurality of optical elements are composed of a phase grating formed on a surface of a transparent plate arranged in a pupil plane of the projection optical system or a plane in the vicinity thereof, and diffracted by the phase grating. 2. The apparatus according to claim 1, wherein the pitch and the diffraction direction of the phase grating are determined so that the diffracted beam of the second alignment beam intersects on the second grating mark. 3. The second of the two reflected by the mask
A beam width of the alignment beam on the field stop and its intersection angle are W1 and θ, an aperture width of the field stop is W2, and a crossing point of the two second alignment beams and the field stop are in a conjugate relationship. The amount of the interval D is D ≧ (W1−W2) / {β ² × 4 · sin (θ / 2)}, where β is the magnification of the optical system. The apparatus according to item 2 or item 3. 4. The beam generating member generates a plurality of sets of two second alignment beams for irradiating the second grating mark with a plurality of light beams having different wavelengths, and the plurality of optical elements. The axial chromatic aberration amount is corrected so that each of the intersections of the plurality of sets of two second alignment beams is different in the optical axis direction of the projection optical system. The described device. 5. The beam generating member irradiates the second grating mark with two second alignment beams of a first wavelength and two second alignment beams of a second wavelength different from the first wavelength. The optical element is configured to correct the amount of axial chromatic aberration at the average wavelength of the first wavelength and the second wavelength until it becomes substantially equal to the distance D.
The device according to item. 6. An illumination system for irradiating a mask on which an original image pattern is formed with exposure light, a projection optical system for projecting the original image pattern on a photosensitive substrate, and a light having a wavelength different from that of the exposure light.
A first alignment beam formed on the mask and a transparent portion in the vicinity thereof are irradiated with a book alignment beam, and a first alignment mark formed on the photosensitive substrate is provided via the transparent portion and the projection optical system. Beam irradiating means for irradiating two grating marks, diffracted light generated from the second grating mark and traveling through the projection optical system and the transparent portion, and diffracted light generated from the first grating mark are received. In a projection exposure apparatus having a light receiving system having a photoelectric detector, and position detecting means for detecting relative positional deviation between the mask and the photosensitive substrate based on an output signal of the photoelectric detector. When the axial chromatic aberration amount of the projection optical system generated at the wavelength of the alignment beam is Lc, it is arranged at a position where the two alignment beams pass in the pupil plane of the projection optical system, and The axial chromatic aberration amount in order to conjugate the space surface, which is separated from the surface of the mask in the optical axis direction by a predetermined distance D, with the photosensitive substrate via the projection optical system at the wavelength of the alignment beam. Lc to Lc-D
A correction optical element for correcting only the above is provided, the light receiving system has a field stop in a plane conjugate with the space plane, and a beam width on the field stop of the two alignment beams reflected by the mask. , And its intersection angle is W
1 and θ, the aperture width of the field stop is W2, and the magnification of the optical system that conjugates the space plane and the field stop is β, the amount of the interval D is D ≧ (W1−W2 ) / {Β ² × 4 · sin (θ / 2)}, which is a projection exposure apparatus.