JPH02227603A - Position detecting device - Google Patents

Abstract

PURPOSE:To adjust the angle of intersection of beams to the best angle by providing a 1st optical member which displaces the two beams in the array direction of a diffraction grating and a 2nd optical member which varies the interval between the two beams in a pupil plane. CONSTITUTION:The two parallel beams LB1 and LB2 from a frequency shifter 12 are reflected by a beam splitter 14, passed through a pupil relay system 17A, and transmitted through a beam splitter 20, and then the beams are transmitted through parallel plane glass 50 (capable of slanting) and made incident on a bifocal optical system 21. This optical system 21 is constituted by combining a birefringent material 21b and a telecentric objective 21a for a microscope, etc., in one body and given different power according to the polarized components of the beams LB1 and LB2. Further, parallel plane glass plates 54 and 55 which constitute the 2nd optical member are arranged in the pupil spaces of the beams LB1 and LB2 and the inclination of glass plates are controlled individually by a driving control system 56 to adjust the distances of two spot light beams SP1 and SP2 from the optical axis AX4 independently. Then the 1st and 2nd optical members are adjusted according to the state of a detection signal.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a position detection device using a diffraction grating formed on an object such as a semiconductor element, and is particularly used during alignment of an exposure apparatus for lithography. The present invention relates to a device that detects the relative positional relationship between a mask having an original pattern and a substrate to which the original pattern is transferred.

[Conventional technology]

In recent years, projection exposure devices (steppers) have come into widespread use as devices for transferring fine patterns of semiconductor elements and the like onto semiconductor wafers with high resolution. Conventionally, in this type of stepper, the reticle (synonymous with mask) and one silane on the wafer)8! As a so-called alignment method, TTR (through-the-reticle) simultaneously detects alignment marks formed around the circuit pattern on the reticle and alignment marks formed around the circuit pattern on the wafer. An apparatus having an alignment system of this type is known.

This alignment method detects both the mark on the reticle and the mark on the wafer with high precision, calculates the amount of relative positional deviation between them, and then adjusts the reticle so that this amount of deviation is corrected.
Or the wafer is slightly moved. In general, in a projection exposure apparatus, in order to image a reticle pattern on a wafer with high resolution, the projection optical system uses exposure illumination light (for example, wavelength 4
Currently, chromatic aberrations are well corrected only for 36 nm g-line, 365 n- wavelength i-line, 248 n- wavelength KrF excimer laser light, etc.).

This means that in the alignment optical system that detects the reticle mark and the wafer mark via the projection optical system,
This means that the light for mark illumination is limited to a wavelength that is the same as, or very close to, the wavelength of the exposure light.

A resist layer is formed on the surface of the wafer used in the exposure process, and marks on the wafer are detected through the resist layer during alignment. For this resist layer, in order to enable pattern formation with higher resolution, it has been considered to adopt a multilayer resist structure or the like that has a high absorption rate and low transmittance for exposure light. In this case, the illumination light for alignment is attenuated before reaching the mark on the wafer, and the reflected light from the mark (regularly reflected light, scattered light, diffracted light, etc.) is also attenuated. A problem arises in that the mark is not recognized by the alignment optical system with a sufficient amount of light, reducing mark detection accuracy.

In addition, when alignment illumination light is irradiated onto a mark on a wafer for alignment, the resist layer in that area is naturally exposed to light, and when various processes are performed after development, the mark on the wafer is destroyed. Although this is not an essential problem, it also causes the problem that it cannot be used for alignment during overlapping exposure of the next layer.

Therefore, based on the TTR type separate wavelength alignment system (a system in which the illumination light for alignment coincides with the wavelength of the exposure light) disclosed in JP-A No. 63-153820, for example, it is possible to form an alignment system on a wafer or reticle. Japanese Patent Application Laid-Open No. 2004-1999 has developed a method that optically detects one-dimensional diffraction grating marks and uses the pitch information to detect the position of the wafer or reticle with high resolution (a fraction of the pitch to a few tenths of the pitch). 63-283
This is proposed in Publication No. 129.

Various methods have been proposed and put into practical use for position detection using diffraction grating marks. The method disclosed in Japanese Patent Application Laid-Open No. 63-283129 is one in which a one-dimensional interference pattern is created by simultaneously irradiating a diffraction grating mark with substantially parallel laser beams from two directions, and this interference pattern is used to create a one-dimensional interference pattern. This method attempts to specify the position of a diffraction grating mark using interference fringes, and is also called interference fringe alignment method because it uses interference fringes.

There are two more methods for this type of interference fringe alignment method.
The heterogain method gives a constant frequency difference to the , and the homodyne method has no frequency difference.

In the homodyne method, stationary interference fringes are created parallel to the diffraction grating, and in order to detect the position, the diffraction grating (object) must be slightly moved in its pitch direction, and the position of the diffraction grating is determined using the interference fringes as a reference. On the other hand, in the heterodyne method, due to the frequency difference (beat frequency) between the two laser beams, the interference fringes flow at high speed in the pitch direction at the beat frequency, and the position of the diffraction grating is determined based on the interference fringes. It cannot be determined, but is determined based mainly on the temporal element (phase difference) accompanying the high-speed movement of the interference fringes.

[Problem that the invention seeks to solve]

In the position detection system that adopts the interference fringe alignment method described above,
As a result of experiments, it has become clear that the intersection angle formed by the beams irradiating the diffraction grating from two directions and the symmetry of the incident angles of the two beams are extremely important. If the symmetry of the intersection angle or angle of incidence fluctuates even slightly, it will worsen interference fringe linearity, telecentricity, etc., and since the measurement resolution is originally high, the ratio of detection error to measurement resolution will be large. It becomes a problem. The higher the measurement resolution, the more accurate alignment is possible, but if the detection error rate is large, the alignment accuracy will eventually be limited by the width of the error amount, making it difficult to use the interference fringe alignment method. The benefits you had will be gone.

There are various factors that can disrupt the intersecting angle and symmetry of the two beams and make them unstable, but the typical problems are problems caused by the device structure and problems caused by the object (grid mark) during position detection. It can be divided into

Problems on the equipment side include the fact that in exposure equipment, etc., the alignment system's objective lenses and mirrors are forced to have a structure that moves in response to changes in mark position, and that the light source for creating two beams is Various optical elements are interposed in the optical path up to the diffraction grating, and manufacturing errors and assembly errors may occur to a greater or lesser extent among them.

The problem during measurement on the object side is mainly due to the symmetry of the angle of incidence, but within the area where the two beams intersect, the partial surface of the diffraction grating to be measured is exposed to light every time it is measured. It is a slight displacement in the axial direction. This means that when diffraction grating marks at a plurality of points on a wafer are sequentially measured as in alignment of a wafer, a sine error occurs for each mark. Many of the causes are due to the flatness (warpage, curvature, etc.) of the wafer surface.

Among the conventional TTR alignment systems, those that focus a laser beam into a slit, relatively scan bar marks, etc. on the wafer, and photoelectrically detect the scattered and diffracted light do not suffer from similar problems. However, this effect was small compared to the overall detection accuracy, which takes into account noise errors, waveform distortion, etc. during mark detection, and was hardly a concern. In addition, in the method of focusing a laser beam onto a slit-shaped spot, the shape of the beam cross section is elongated in the pupil plane of the projection lens or alignment objective lens in a direction perpendicular to the longitudinal direction of the slit of the spot on the wafer. Become. Therefore, in the detection direction of the slit-shaped mark (width direction of the slit-shaped spot), the numerical aperture of the beam is larger than in the direction orthogonal thereto. Generally, if the numerical aperture of the beam is increased, the diameter (radiance) of the spot light can be made smaller and the detection resolution can be improved, but the width of the beam waist in the optical axis direction is correspondingly smaller. In addition, in the method of focusing the beam onto a spot, the wavefront of the beam changes in the optical axis direction across the center of the beam waist. Therefore, if the mark is displaced in the optical axis direction with respect to the beam waist due to the flatness of the wafer surface and is scanned, the original sharp photoelectric signal waveform cannot be obtained.

The present invention has been made in view of the above-mentioned problems, and is an object of the present invention to simultaneously maintain stably the intersection angle and the symmetry of the incident angle of the two beams in a two-beam interferometric position detection device used in the interference fringe/alignment method. With the goal.

[Means for solving problems]

In the present invention, we focus on the fact that the beams that irradiate the diffraction grating on the object from two directions via the objective optical system are separated by a predetermined interval within the pupil plane of the objective optical system, and the two beams passing through the pupil plane A first optical member was provided in the optical path between the beam-generating light source and the objective optical system, and the first optical member was displaced by the same amount in the grating arrangement direction within the pupil plane while keeping the interval substantially constant.

Furthermore, in the present invention, a second optical member that changes the interval between the two beams in the pupil plane by displacing the positions of the two beams in the pupil plane in the grating arrangement direction is moved from the light source to the objective optical member. It was installed in the optical path to the system.

The first optical member is for adjusting the symmetry of the angle of incidence of the two beams on the object, and includes a distance adjusting means for displacing the distance between the object and the objective optical system in the optical axis direction, and a diffraction grating. The first optical member is controlled by means for adjusting the first optical member according to the phase difference of the detection signal when the diffracted light (interference light) generated from the first optical member is photoelectrically detected through the objective optical system. Further, the second optical member adjusts the intersection angle of the two beams at the object plane, and automatic control is possible by detecting the amplitude (modulation degree) of the detection signal.

[Effect]

In the present invention, compared to conventional imaging type or beam spot type position detection devices, the two-beam interference type has an effective detection range in the optical axis direction several times to several tens of times more. In other words, the wavefront created by the intersection of two beams within the effective detection range is approximately constant at each position in the optical axis direction, and the pitch of the interference fringes is well preserved. I'm paying attention.

This will be explained with reference to FIG.

In Fig. 2(A), two parallel beams LB,, LB, are projected on a virtual plane IP so as to intersect exactly, and a one-dimensional diffraction grating WP (duty is 1:l) is projected onto the plane IP. ) are arranged in parallel, showing two beams LB,,LB
, the wavelength of , λ, the pitch of the diffraction grating WP is P, and the optical axis of the objective optical system that projects the two beams LB+ and LBm is the plane IP.
If AXa is perpendicular to
In order to generate TL, it is necessary to satisfy equation (1).

・・・・・・(1) If this formula (1) is satisfied, there will be two beams LB.

, LBt intersect, one-dimensional interference fringes IF with a pitch of 1/2P exist in any plane parallel to the plane IP.
w is created.

Each bright and dark fringe of this interference fringe IFw is set to be parallel to each grating element (bar pattern) of the diffraction grating WP.

The above was explained in the case of the homodyne method where there is no frequency difference Δf between the two beams LB and %LB, but in the case of the heterodyne method where there is a difference Δf, the interference fringes IFW are at velocity V in the direction of the arrow (lattice arrangement direction) However, the interference fringe I
The pitch preservation property of Fw is exactly the same as that of the homodyne method. Here, the speed ■ is the pitch of the interference fringe IFw P''(-
1/2P), it is expressed by equation (2).

■−Δf−P'・・・・・・・・・・・・・・・・・・
(2) However, if the symmetry of the incident angles θa and θb is disrupted, the generation direction of the interference light BTL will also be tilted, which may cause problems during photoelectric detection.

This is because the first-order light of the higher-order diffracted light generated from the diffraction grating WP is vertically transmitted or reflected by the irradiation of the beam LB, and the first-order light of the higher-order diffracted light generated by the irradiation of the beam LB8 is vertical. This is because interference light BTL is created by transmitting or reflecting from will not be obtained.

The present invention focuses on the fact that the two-beam interference method has an extremely wide and stable detection range in the optical axis direction, and achieves high-precision detection by exploiting the symmetry of the incident angles of the two beams LB, , and LBI. It is something to do.

Furthermore, if the intersection angle θ (θa + θb) of the two beams LB, LBt changes, the pitch of the interference fringes IFw changes, which greatly changes the linearity with respect to the diffraction grating pitch, so the position detection accuracy decrease. In particular, in the homodyne method, since a static intensity value of the interference light BTL is detected, this influence is large.

This will be explained with reference to FIG. 2(A). In Fig. 2 (A), WP is a diffraction grating, IFw is a velocity V
The intensity distribution of interference fringes moving at The total amount of diffracted light Ae from all of is the size of interference light BTL to be photoelectrically detected (instantaneous value)
FIG. 2(A) shows a case where the pitch of the interference fringes IFw slightly deviates from the ideal condition of being 1/2 of the pitch of the diffraction grating WP (a case where the linearity deteriorates). Therefore, the distribution of the diffracted light Ae is not flat in the grating arrangement direction, but is a distribution that fluctuates in a sinusoidal manner. The photodetector used for photoelectric detection receives the total amount of diffracted light Ae from each edge, so the amplitude of the photoelectric signal becomes smaller than under ideal conditions, as shown in Figure 2 (B).
) are signal waveforms under ideal conditions (signals DSr, DSw)
Figure 2 (C) shows the signal waveform when the conditions deviate from the ideal conditions.As is clear from Figures 2 (B) and (C), when the conditions deviate from the ideal conditions, the signal waveform [(P -P value) becomes smaller, and an offset Vof of the DC component is added.

It is necessary to suppress the decrease in signal amplitude and the occurrence of offset Vof as much as possible. Further, in the present invention, the crossing angle of the two beams LB, BL, and BL is maintained stably.

〔Example〕

Next, FIG. 1 shows a machine in which a position detection device according to an embodiment of the present invention is incorporated into a projection exposure device (stepper)
The explanation will be given with reference to FIGS. 3, 4, 5, 6, and 7, but the basic configuration is based on Japanese Patent Application Laid-Open No. 63-28312.
This is similar to that disclosed in Publication No. 9.

In FIG. 1, a reticle I having a predetermined circuit pattern and a diffraction grating mark for alignment is held on a two-dimensionally movable reticle stage 2. Each pattern on the reticle 1 is imaged onto a wafer 4 under exposure light by a projection lens 3 which is telecentered on both sides. However, this projection lens 3 has a wavelength of illumination light for exposure (gli,
i*, etc.), and the reticle 1 and the wafer 4 are arranged so as to be conjugate to each other with respect to the exposure wavelength. Further, a diffraction grating mark similar to the grating mark formed on the reticle 1 is also formed on the wafer 4. Now, the wafer 4 is moved two-dimensionally in the xSY direction using the step-and-repeat method, and is attracted to the stage 5 that moves slightly in the Z direction. When the transfer exposure of the reticle l to one shot area on the wafer 4 is completed, Stepped to the next shot position. A part of the reticle stage 2 includes a laser beam interferometric length measuring device (hereinafter referred to as an interferometer) 4 for detecting the position of the reticle l in the horizontal plane in the X direction, the X direction, and the rotation (θ) direction.
A movable mirror 6 that reflects the laser beam from 3 is fixed. This interferometer 43 has three laser beams for length measurement in order to independently detect positions in the X direction, X direction, and θ direction, but some of them are omitted here to simplify the explanation. . The movement stroke of the reticle stage 2 is several millimeters or less, and the detection resolution of the interferometer 43 is set to, for example, about 0.01 μm.

On the other hand, a movable mirror 7 that reflects a laser beam from an interferometer 45 for detecting the position of the wafer 4 in the X direction and the X direction within the horizontal plane is fixed to a part of the wafer stage 5. This interferometer 45 also has two laser beams for length measurement in order to independently detect the positions in the X direction and the X direction, but some of them are omitted here for the purpose of simplifying the explanation. A drive motor 42 drives the reticle stage 2 in the X, X, and θ directions, and a drive motor 46 moves the wafer stage 5 in two dimensions and along the Z axis.

By the way, the illuminator for exposure consists of 30 mercury lamps and 3 elliptical mirrors.
1. Manual lens group 32 including condensing lenses and interference filters, optical integrator (fly's eye lens)
33, a mirror 34, a main condenser lens 35, a gicroic mirror 22, and the like. The dichroic mirror 22 is installed obliquely at 451 above the reticle 1, and reflects the exposure light from the condenser lens 35 vertically downward, thereby uniformly illuminating the reticle 1. This dichroic mirror 22 has a 90° angle with respect to the wavelength of the exposure light.
% or more, and has a transmittance of 50% or more for the wavelength of illumination light for alignment (longer wavelength than exposure light).

Next, the alignment system of this stepper will be explained. Irradiation light (beam) LB for alignment is emitted from a laser light source 10, enters a two-beam frequency shifter 12 including two acousto-optic modulators (AOM), and includes linearly polarized light having different frequencies and orthogonal to each other. It is converted into two beams LB, , LBI. The frequency difference is 2
High frequency signal SF, , Sr1 that drives two AOMs
It corresponds to the difference in frequency.

two parallel beams LB from the frequency shifter 12,
, LBI is reflected by the beam splitter 14, passes through the pupil relay system 17A, passes through the beam splinter 20, and then passes through the parallel plane glass 50 (which can be tilted) as the first optical member of the present invention. The light enters the focusing optical system 21. The bifocal optical system 21 integrates a birefringent material (a lens of crystal, calcite, etc.) 21b arranged conjugately with the pupil of the alignment system, that is, the pupil EP of the projection lens 3, and a telecentric objective lens 21a for a microscope or the like. The polarization components of the beams LB, , LB (with respect to the dichroic mirror 22, the light with a polarization plane parallel to the plane of the paper in the figure is P polarization, and the light with a polarization plane perpendicular to it is S polarization). It gives different powers depending on the polarization.

Here, it is assumed that the laser light source 10 oscillates orthogonally linearly polarized laser light. Therefore, the beam LB consisting of one polarized light (for example, P polarized light) component emitted from the bifocal optical system 21,
, LB, are imaged (intersected) at a focal point 26a in the space above the reticle l, and beams LB, , LB3 consisting of the other polarized light component (for example, S polarized light) are focused at a focal point 27a that coincides with the pattern surface on the lower surface of the reticle l. A surface that forms an image (intersects) and is conjugate with the other focal point of the bifocal optical system 21, that is, the focal points 26a and 27a on the laser light source 10 side, exists within the two-beam frequency shifter 12. Here, the distance between the two focal points 26a and 27a of the bifocal optical system 21 in the optical axis direction corresponds to the amount of chromatic aberration on the reticle l side of the projection lens 3 at the wavelength of the alignment laser beam. The focal plane 26a in this space is made conjugate by the projection lens 3 with an imaging plane 26b that coincides with the surface of the wafer 4, and the focal plane 27a (reticle pattern plane) is directed spatially downward from the surface of the wafer 4 by the projection lens 3. It becomes conjugate with the imaging plane 27b which is far away.

The distance between the imaging planes 26b and 27b is the same as that of the wafer 4 of the projection lens 3.
The image forming surface 26b corresponds to the amount of chromatic aberration on the side.
and 27b is DW, the distance between focal planes 26a and 27a is D'', and the projection magnification of the projection lens 3 is 1/M.
(Usually M is one of 1.2.5.5.10)
Then, there is a relationship of -D r -M'' ・Dw for boat fishing.The farther the wavelength of the alignment laser light is from the wavelength of the exposure light, the more Dw and Dr will change depending on the aberration characteristics of the projection lens 3. The depth of focus of this type of projection lens is extremely shallow, about ±lam, and the distance Dw can reach the order of 200 mm, depending on the wavelength of the alignment illumination light. It is desirable that the wavelength of the laser beam (laser beam) be almost insensitive to the resist coated on the wafer 4, but this condition is not necessarily met in the present invention. This is because an extremely large aberration occurs due to the wavelength of the illumination light, and there is a risk that a large distortion may be added to the optical information itself, especially from the diffraction grating mark on the wafer 4. Therefore, the optimum It is more important to give priority to determining the alignment illumination light. Therefore, if the alignment illumination light illuminates the resist for a long time (for example, more than 1 minute), it will be exposed to light (fading will occur after development). In some cases, the wavelengths have weak sensitivity such as LB, LB, and LB.In this case, a shutter is provided in the light transmission path of the two beams LB, LB, and LB, which are opened during alignment.

Of the laser beams for alignment, two P-polarized beams LB, , LB are incident on the diffraction grating mark portion of the reticle 1 at the focal plane 27a from two directions at angles formed by the beams LB, , LB. and forms an image. In addition, the two S-polarized beams LB, LB, from the focal plane 26a that have passed through the transparent part of the reticle 1 are transmitted to the focal plane 26a through the projection lens 3.
The beam LB is applied to the diffraction grating mark portion of the wafer 4 at b.
.

The light is incident from two directions at the angles formed by and LB, and an image is formed. The reflected diffracted light (interference light BTL) from the diffraction grating mark on the reticle L is reflected by the beam brick 20 via the dichroic mirror 22.2, the focusing optical system 21, and the parallel plane glass 50, and then passes through the pupil relay system 17B. Only the diffracted light (BTL) traveling on the axis is filtered by a spatial filter 23 placed on the pupil conjugate plane (Fourier plane).
Further, the light reaches a photoelectric detector 25 through a condensing lens 24 . Further, the reflected diffracted light (interference light BTL) from the diffraction grating mark on the wafer 4 returns to its original optical path via the projection lens 3.
Dichroic mirror 2 passes through the transparent part of reticle l.
2.2 focal optical system 21, parallel plane glass 50, beam splitter 20, pupil system 17B, spatial filter 23
, and a photoelectric detector (light receiving element) through the condensing lens 24.
Reach 25. The spatial filter 23 is arranged at a position conjugate with the pupil plane of the alignment optical system, that is, at a position substantially conjugate with the pupil (exit pupil) of the projection lens 3, and filters specularly reflected light (zero-order light) from the reticle 11 or the wafer 4. It is designed to block only light that is diffracted perpendicularly to the diffraction grating of the reticle l or the wafer 4 (in the normal direction of the surface). In front of the photoelectric detector 25, a reticle 1 and a wafer 4 are connected via a bifocal optical system 21, a pupil relay system 17B1, and a lens 24.
An aperture plate 25 is provided which is arranged conjugately with each of the aperture plates 25 and 25.

Now, the photoelectric signal DS ", DS" obtained from the photoelectric detector 25
Since the interference fringes created by the beams LB, , LB, which irradiate the reticle l or wafer 4 from two directions are irradiated onto each diffraction grating mark so that they flow in the pitch direction, w is a high-frequency signal SF. , , The two beams LB, , LB, from the two-beam frequency shifter 12, which become sinusoidal alternating current signals (beat frequency) with a frequency corresponding to the difference Δf between SFs, pass through the beam splinter 14 and pass through the pupil ( An image is formed (crossed) on a reference diffraction grating 18 by a lens system (inverse Fourier transform lens) 16 that converts a free-grained surface into an image plane, and this reference diffraction grating 18 is fixed on the device. .

The beams LB and LB8 are also incident on this diffraction grating 18 from two directions at predetermined crossing angles. The photoelectric detector (light receiving element) 19 receives diffracted light (or interference light) other than the zero-order light that has passed through the reference diffraction grating 18, and outputs a sinusoidal photoelectric signal DR.

This photoelectric signal DR consists of two beams LB1, LB! The frequency is proportional to the difference frequency of 0, which becomes the reference beat signal.
The phase detection system 40 inputs the reticle-side photoelectric signal DSr from the photoelectric detector 25, the wafer-side photoelectric signal DSw, and the photoelectric signal DR from the photoelectric detector 19, and detects both signals DSr based on the signal DR. , DSw waveforms are detected. The detected phase difference (±180°) uniquely corresponds to the amount of relative positional deviation within 1/2 of the grating pitch of the diffraction grating marks formed on the reticle l and the wafer 4, respectively. The main control system 41 controls the drive motor 42.46 based on information on the detected phase difference (positional deviation amount), position information from each of the interference systems 43.45 obtained via the servo system 44, etc. Relative positioning (alignment) between the reticle 1 and the wafer 4 is performed.In the explanation of FIG.
are inclined to each other in a plane perpendicular to the plane containing.

The parallel plane glass 50 is tilted and driven by the drive control system 52 so that the two beams LB, LBt, LBt move in parallel in the same direction within a plane that includes the optical axis (AXa) of the objective lens 21b.

In the above overall configuration, part of the alignment optical system,
In particular, the bifocal optical system 21 is movable to any position according to the arrangement of the alignment marks on the reticle l, making it possible to detect marks regardless of the arrangement of the marks. I will discuss this in detail later. Furthermore, since the exposure light and the alignment illumination light are separated by the guichroic mirror 22 provided obliquely above the reticle 1, mark detection is possible even during the exposure operation. This means that even if the alignment between the reticle 1 and the wafer 4 goes out of alignment due to some disturbance during exposure, it can be detected immediately.

Furthermore, it also means that the positioning servo between the reticle stage 2 and the wafer stage 5 can be executed in a closed loop based on the phase difference information from the phase detection system 40 even during the exposure operation. Therefore, the line width of the exposed resist pattern will not increase due to slight image blurring. Note that the light source of the exposure light may be replaced with an excimer laser light source or the like other than the mercury lamp.

Next, using Figure 3, the detailed configuration of only the alignment system,
and the principle of alignment will be schematically explained. In FIG. 3, the gicroic mirror 22 and beam splitter 14 are omitted for the sake of simplicity, and the same members as in FIG. 1 are given the same reference numerals. Furthermore, although the position of the parallel plane glass 50 in FIG. 3 is different from that in FIG. 1, it may be either position in this embodiment.
The two beams LB, LB, which are emitted from the pupil relay system 17A and enter the pupil relay system 17A, become parallel light fluxes inside the pupil relay system 17A, and exit after intersecting by − degrees. At this time, the principal rays LA, , LAt of the beams LB, LB, LB, emitted from the optical axis AXa of the objective lens 21a, etc. become parallel to each other, and form a spot in the pupil plane with the imaging light beam converged. Become.

By the way, the polarization directions of the beams LB+ and LBg are separated into P polarized light and S polarized light by the bifocal optical system 21, and the beams LB+ and LBg are separated into P polarized light and S polarized light,
When focusing on 6a and 27a, the light intensity (light amount) of P polarized light and S (a light) is adjusted to be a predetermined ratio.Normally, the light that reaches wafer 4 has more loss. So,
The amount of light directed to the wafer 4 is increased. for that purpose,
The bifocal element 21b may be rotated by a predetermined angle around the optical axis AXa, or a λ/2 plate may be inserted between the laser light source lO and the dual beam frequency thicker 12, and the bifocal element 21b may be rotated by a predetermined angle around the optical axis. What is necessary is to adopt a structure in which the light is rotated, that is, the ratio of the amount of light between the polarized light component reaching the reticle l and the polarized light component reaching the wafer 4 can be adjusted to an optimal one. Now, if the parallel plane glass 50 is arranged to be exactly perpendicular to the optical axis AXa, two beams LB
, , LB, are imaged as spots at two point-symmetric points across the optical axis AXa in the pupil plane of the telecentric bifocal optical system 21, that is, the birefringent material 21b, by the action of the pupil relay system 17A. The beam LB is incident on the birefringent material 21b and is separated into a P-polarized beam LB, , and an S-polarized beam LB Im by the polarization component, and
The light reaches the reticle l as a parallel light beam tilted by an angle determined by the spot position 1 in the pupil plane with respect to the optical axis AXa of the focusing optical system 21.

Similarly, the beam LB is also separated into a P-polarized beam LB, , and an S-polarized beam LB, and the optical axis AXa of the objective lens 21a is
The beams LBIF and LBIS are formed into parallel light fluxes at symmetrical angles with each other in between, and reach the reticle 1. Regarding P-polarized light, since the focal point 27a and the conjugate image plane (IP') at the intersection inside the pupil relay system 17A are conjugate, the P-polarized beams LBIP, LB*r become almost parallel beams at the diffraction grating mark RM. In FIG. 3, the lattice arrangement direction of the mark RM is the left-right direction in the plane of the paper, and the direction I'14 from the optical axis AXa of each of the beams LB+p and LBgr is also the same as in FIG. Defined within the paper. The reticle l has a diffraction grating mark RM as shown in Fig. 4(a).
and a transparent window P0 are formed, and the beam LBIF
, LB□ illuminate the reticle 1 with a size that covers the mark RM and the window P0.

The mark RM shown in FIG. 4(a) is used for position detection in the X direction (grating arrangement direction), and the diffraction grating mark WM on the wafer 4 is also similar to this as shown in FIG. 4(b). It corresponds to The mark WM is determined to be aligned with the position of the window Po of the reticle 1 during alignment (or during exposure).

Now, the almost parallel S-polarized beams LB, ..., LBms exiting the bifocal optical system 21 are imaged (intersected) at a focal point 26a in space, and then transmitted through the window P of the reticle 1 and projected. The light is once focused as a spot light by the pupil EP of the lens 3, and then imaged so as to be incident on the diffraction grating mark WM of the wafer 4 from two different directions. This is because, for S-polarized light, the focal point 26a (wafer conjugate plane) and the intersecting plane IP' inside the pupil relay system 17A are conjugate. A nearly parallel S-polarized beam LB emitted from the projection lens 3,...
Each of LBx- is incident at a symmetrical angle with respect to the grating arrangement direction of the diffraction grating mark WM. S reached wafer 4
The angle θ formed by the principal rays of the polarized beams LB,..., LBx-
Even if it is large, it does not exceed the numerical aperture of the exit (wafer) side of the projection lens 3. Note that the reticle 1 and the wafer 4 are arranged conjugately with respect to the intersecting plane IP' inside the pupil system 17A. Therefore, the beam LB passing through the intersecting plane IP''
, , LB, are parallel light beams, each beam LB, , , LBl, , LB, , on the reticle and wafer, respectively.
SLB, both become parallel beams of light.

As described above, the beam L is applied to the mark RM from two different directions.
When B IF, LB zp is irradiated, mark R
Interference fringes occur on M, but the interference fringes move (flow) in the grating arrangement direction of the mark RM in response to the difference frequency between the two beams LB, , LB, and the mark R.
Diffracted light 104 is generated from M along the optical axis AXa of the bifocal optical system 21, but due to the movement of the interference fringes, the diffracted light 104 becomes a beat wavefront that periodically repeats changes in brightness and darkness. Therefore, the signal DSr from the photoelectric detector 25 becomes a sinusoidal alternating current signal corresponding to the period of the change in brightness and darkness.

The above is exactly the same in the relationship between the diffraction grating mark WM on the wafer 4 and the S-polarized beams LB, .
05 is generated vertically, which travels along the chief ray of the projection lens 3 and passes through the window P0 of the reticle 1 to the photodetector 2.
Reach 5.

Now, although it is assumed that the photoelectric detector 25 is arranged conjugately with each of the marks RM and WM via the bifocal optical system 21, in reality, as shown in FIG. A 25° aperture plate as shown in FIG. 4(C) is provided at a conjugate position, and the diffracted light 104 and 105 transmitted through the aperture AP SAs of the 25° aperture plate is configured to be photoelectrically detected. Here, aperture A is
For example, the aperture A is for taking out a diffraction image of the diffracted light 104 from the mark RM of the reticle 1, and the aperture A is for taking out the diffraction image of the diffracted light 105 from the mark WM of the wafer l. Therefore, by separately providing the light receiving surface of the photoelectric detector 25 after each aperture At, As, the position of the reticle l by the mark RM and the mark WM can be detected.
It becomes possible to detect the position of the wafer 4 independently. still,
Aperture A has a P-polarized beam L B lp, , L
Mark RM of reticle 1 illuminated by B zp
An image of S-polarized light LB, , LB is formed at the same time.
,.

The reflected and diffracted light also enters as background noise. For this reason, it is preferable to provide a polarizing plate for passing P-polarized light in aperture A, and a polarizing plate for passing S-polarized light in aperture A. In this way, each of the two photoelectric detectors 25
Crosstalk, where light from the wafer and light from the reticle are mixed, is sufficiently reduced.

Now, the diffracted lights 104 and 105 become periodic (sinusoidal) brightness information, and the resulting photoelectric signals DSr and DSw become sinusoidal alternating current signals even if the reticle 1 or wafer 4 is stationary. Therefore, in this case, using the photoelectric signal DR from the photoelectric detector 19 shown in FIG. 1 as a reference signal, the phase difference φr between the diffracted light 104 from the mark RM and the photoelectric signal DSr is detected by the phase detection system 40. . Similarly, the phase difference φW between the photoelectric signal DSw of the diffracted light 105 from the mark WM and the reference signal DR is detected. Then, if we calculate the difference between the phase differences φr and φW,
You can see the amount of deviation in direction. This detection method is called the optical heterogain method, and as long as the reticle 1 and wafer 4 are within the positional error range (prealignment accuracy) of 1/2 of the grating pitch P, the positional deviation can be detected with high resolution even in a stationary state. Can be detected. Therefore, it is convenient to apply closed-loop position servo to prevent minute positional deviations from occurring while exposing the pattern of the reticle 1 to the resist of the wafer 4. In this detection method, alignment is completed by moving reticle l or wafer 4 so that φ"-φW becomes zero (or a predetermined value), and then reticle l and wafer 4 are moved relative to each other at that alignment position. A servo lock can be applied to prevent movement.

Next, the configuration of the two-beam frequency shifter 12 will be explained with reference to FIGS. 5, 6, and 7.

As shown in FIG. 5, a parallel beam LB (orthogonal linearly polarized light) from the laser light source 10 is reflected by a mirror 70, and a polarized beam splitter 71 converts the P-polarized component beam LBP and S
It is divided into a polarized beam LB, and a polarized beam LB. Beam LBP
is reflected by the mirror 72, and the AOM (acousto-optic modulator)
73, the beam LB is polarized beam splinter 7
1 and enters the AOM 74. The AOM 73 is driven by a high-frequency signal S F l having a frequency fl, and outputs, as a beam LB, primary light polarized by a diffraction angle determined by the frequency f1. AOM74 has frequency rx
1 driven by a high frequency signal SF8 of (ft = ft - Δf) and deflected by the diffraction angle determined by its frequency f.
The secondary light is output as beam LB. The zero-order light of the beam incident on each AOM is blocked by a slit or the like placed at an appropriate position. Furthermore, the drive frequency f+,
The relationship between fi and the difference frequency Δf is ft >>Af, ft
>>Af is desirable, and the upper limit of Af is determined by the responsiveness of the photoelectric detector 19.25.

Now, the beam LB from the AOM 74 is reflected by the mirror 75 and enters the polarizing beam splitter 76, and the beam LB from the AOM 74
The beam LBP from the LBP enters the polarizing beam splitter 76 from a direction perpendicular thereto. Here, the polarizing beam splitter 76 does not combine the two beams LBF and LB completely coaxially, but combines them parallel to each other so as to be spaced apart by a certain amount. In the case of this embodiment, this interval is between the two beams LB that irradiate the reticle L and the wafer 4, and
, LB, and the intersection angle θ is defined.

The two parallel beams LB, LB, then enter the optical system shown in FIG. Beams LBP (p-polarized light) and LB, (S-polarized light) have a frequency difference of only Δ1, but in this example, the reticle is irradiated with two P-polarized beams with a frequency difference, and the wafer is irradiated with two P-polarized beams. Two S with frequency difference
It is necessary to irradiate with a polarized beam. In other words, the system in FIG. 6 consists of one S-polarized beam LB with frequency f, and
, one P-polarized beam LBP and two to four beams LBIP, LBIP, LBl! , LB, .

The two beams LBs, LB, have a frequency f using a 1/2 wavelength plate 117 with a polarization direction of 45°, polarization beam splitters 11B, 119, etc. as shown in FIG.

P polarized beam LB, P and S polarized beam LB0 are coaxially combined to form beam LB, and frequency f8(r+ −
Af) P polarized beam LB port and S polarized beam LBffi
fi is coaxially combined to form a beam LBI which exits the polarizing beam splitter 119.

As shown in FIG. 6, when the sl light beam LB, (frequency f,) is incident on the 172-wavelength plate 117, its polarization direction is rotated by 45°. Therefore, in the polarizing beam splitter 11B, the P-polarized beam LB□ and the S-polarized beam LB
It is divided into t and t. Similarly, the polarization direction of the P-polarized beam LBP (frequency f,) that has passed through the 172 wavelength plate 117 is rotated by 450, so that the beam splitter 11
In B, P-polarized beam LBIF and S-polarized beam LB+
− is divided into Each beam passes through a polarizing beam splitter 1 through a right angle prism 120, 121 with a metal reflective surface.
191 are combined again into two parallel beams LB,, LB
Inject as I. Two LB,, LB! are located symmetrically across the optical axis AXa of the alignment optical system.

The two beams L B'+ and L B z then enter an optical system as shown in FIG. 7, for example. This optical system receives two beams LBt and LBt and enters the plane (image conjugate plane) IP.
a prism 122 that intersects at a predetermined angle at m;
Two parallel beams LB, , L that intersect at the zero plane IPm including the lens 123 whose plane IPm coincides with the front focal plane
B□ is focused on the rear focal plane (pupil conjugate plane) of the lens 123 as spots SP1, SP, and then diverges and enters the pupil relay system 17A. Beam LB as shown in FIG.

Assuming that the principal ray of beam LB is LA1, and the principal ray of beam LB is LA8, in this embodiment, the principal rays ILA, , LA□ are parallel to the optical axis AXa even between the lens 123 and the pupil relay system 17A. . And spot sp, , sp,
Parallel plane glasses 54 and 55 are arranged so as to be tiltable independently of each other in the vicinity of the second optical member of the present invention. In the case of FIG. 7, the central axis of rotation of the parallel plane glasses 54, 55 is perpendicular to the plane of the paper.

Furthermore, the front focal plane of the pupil relay system 17A is
In order to match the planes 1 and SPx, the spots sp, , sp□ shown in FIG. 7 are relayed by the pupil relay system 17A to the pupil plane where the birefringent material 21b is located, as shown in FIG. 3.

Parallel plane glasses 54 and 55 each have two beams LB,
, LB and Yo (portions where the beams are convergent and divergent) and are separately tilt-controlled by the drive control system 56. As a result, two spot lights SP
, , SP from the optical axis AXa can be adjusted independently. Note that in this embodiment, two beams L B+ , L
Bt (Parallel plane glasses 54.55 were provided for each D, but since there is a parallel plane glass 50 for adjusting the degree of telecentricity, only one of the parallel plane glasses 54.55 can be tilted, and the other is fixed. You can leave it as is.

Next, referring to FIG. 8, a configuration is shown in which a part of the alignment optical system is made movable to correspond to changes in the mark position. It is marked with a sign. Marks RM +, R on reticle 1
M x has a different position depending on the reticle,
Only one of them is formed within the same reticle. In the actual alignment optical system, the objective lens 21a,
? J[The bifocal optical system 21 using the refractive material 21b is arranged horizontally, and a mirror M for vertically bending the optical axis AXa is provided at the tip of the objective lens 21a. This mirror M
1. The objective lens 21a and the birefringent material 21b are fixed to a holding metal fitting 62 so as to move together in the horizontal direction of the optical axis AXa, and the metal fitting 62 is moved in the horizontal direction by the drive control system 60. , and the difference from FIGS. 1 and 3 is that a mirror M is provided between the birefringent material 21b and the parallel plane glass 50.
8 and a relay system 17C. Relay system 1
7C connects the pupil plane EP where the birefringent substance 21b is located to the plane BP near the parallel plane glass 50. Note that IPm' is an image conjugate plane, which is conjugate with the first intersecting plane IPm of the reticle 11, wafer 4, and two beams LB, LB, 0M:, relay system 17C.
is fixed to the main body of the device, and the birefringent material 21b and the relay system 1
7C varies depending on the position of the mark RM, but since the optical path therebetween is an afocal system, the image conjugate plane I formed inside the relay system 17C
Pm' is always conjugated with the reticle 1 and the wafer 4.

Now, the reticle is converted and the mark changes from RM to RM,
(optical axis AXa), the drive control system 60
moves the hardware 62 horizontally to align the optical axis AXa with AXa'
match.

In this case, there is no need to be too strict about the - number position,
Two beams LB, , LB, mark RM3 and window P
Therefore, it is sufficient to control the feeding of the hardware 62 using an inexpensive position monitoring system such as an encoder or a position sensor.

In this way, when a new reticle is set and the observation position of the alignment optical system is set, a reference mark plate is installed on a part of the wafer stage 5 at approximately the same height as the wafer 4, as shown in FIG. FM, mark R on the reticle
It comes to the projection position of M. On the reference mark board FM,
A diffraction grating mark completely similar to the diffraction grating mark WM on the wafer 4 is formed. Therefore, that mark is hereinafter referred to as the fiduciary mark.

The diffraction grating element at the center of the mark RM□ on the reticle extends linearly in the y direction toward the optical axis AX of the projection lens 3, as shown in FIGS. Also, the mark parallel to the mark RM is positioned directly below the mark RM (under the window PO).

In this state, a servo lock is applied by the interferometer 45, motor 46, etc. so that the position of the wafer stage 5 at least in the X direction is fixed.

Next, two stages (not shown) included in the wafer stage 5
Move up and down in the Z (optical axis AX) direction within a predetermined range,
At that time, the change in the phase difference φW between the signal DSw and the signal DR detected by the phase detection system 40 is sampled for each minute displacement amount of the Z stage in two directions.
It is carried out by At this time, although not shown in FIG. 9, the distance between the projection lens 3 and the fiducial mark plate FM is determined by a known oblique incident light type focus detection system or a potentiometer that monitors the drive of the Z stage. In addition, in FIG. 9, the distance between the two beams LB1.
．． LBffi is inclined in the plane of the paper with respect to the optical axis AXa of the alignment optical system on the fiducial mark plate FM,
In reality, it is tilted in a plane perpendicular to the plane of the paper. In this way, 2
By monitoring the phase difference φW at each position of the stage in two directions, a characteristic VT as shown in FIG. 1θ is obtained. 1st
In figure 0, the horizontal axis represents the position of the fiducial mark plate FM in two directions, the vertical axis represents the phase difference φW, and the positions Z1, Zl, Z
It is assumed that the fiducial mark plate FM approaches the projection lens 3 in the order of 1 and Z4.

The intersection angle θ of the two beams LB, , LB, , is approximately 9″ in the case of the numerical value given above, and the beam diameter is 1
When the width is about 10 μm, the width of the detection area in two directions is 30 μm.
There is also 0 μm. Therefore, the distance between positions 21 and 24 is 30
It can be made to about 0I1m. In addition, in the case of this embodiment, the phase difference φ
Since W can be detected continuously within the range of ±180°,
Correspondingly, displacement in the X direction can be measured within a range of ±1/2 pitch. The grating pitch of the fiducial mark is 8 pm, and the resolution of the phase detection system 40 is 0.
2', the measurement resolution of positional deviation is 0.
It can be as much as 0.044 μm. In reality, due to the influence of noise, the practical resolution is approximately 0.01 μm (0.4° in phase). Therefore, if positions Z and Z4 are 300tI
m, the difference Δ between the phase difference φW and φw4 at that time is
Assuming that φW is 0.8 @ (approximately 0.02 pm in terms of lateral deviation), the slope of the characteristic VT, that is, the principal line of the two beams, which is the bisector of the two beams LB, , and LBl, is The inclination of the light ray from the normal line (the telecentric inclination in the plane including the optical axis AX of the projection lens 3) is calculated as follows, where ΔX is the amount of lateral shift corresponding to the difference ΔφW, AX/ l Z+ Zs l −0,02/300−
It is calculated as 6.7X10-'.

This value is for the two beams LB1 and LB for alignment.
It represents the telecentricity (parallelism of optical axis AX and optical axis AXa on the wafer side) of the bisector of , , and the smaller it is,
Indicates good telecentricity.

Also, between zl and 24, the phase difference φw and φW4 are 1.
Assuming that the degree changes by 180' (1/2 pitch), the telecentric degree is ΔX/lZ+-Z, l :4/300:1333Xl
It becomes O-'.

After determining the degree of telecentricity as described above, the main control system 4
1 determines whether the degree of telecentricity is within the allowable range.

If the telecentricity has deteriorated, the main control system 41 calculates the amount of inclination of the parallel plane glass 50 based on the difference between the target telecentricity and the measured telecentricity, and tilts the parallel plane glass 50 by that amount. A command for tilting is sent to the drive control system 52.
Output to.

The inclination of the parallel plane glass 50 is such that the beam LB formed on each pupil plane in the alignment optical system and the projection lens 3,
, LB, are determined to be shifted by the same amount in the pitch direction of the diffraction grating (or interference fringes) while keeping the interval constant.

By this adjustment, on the wafer side of the projection lens 3, the incident angles θa and θb of the two beams LB and LB□ on the object match extremely accurately, and the symmetry of the incident angles is maintained.

FIG. 11 shows the vicinity of the reference grating plate 18 for creating the reference signal DR shown in FIG. The beams LB, LB, and LB□ are incident at a predetermined intersection angle, and the first-order diffracted light (beat signal) that has perpendicularly passed through the reference grating plate 18 is incident on the light receiving element 19. The signal DR is sent to the phase detection system 40 and is also sent to the modulation degree detection system 4 via the switch SS.
Sent to 20. This modulation degree detection system 420 (hereinafter referred to as PP detection system 420) detects the peak-to-peak value (p-p value) of the beat signal (25 to 7). Note that the switch SS receives the measurement signal DS in addition to the reference signal DR.
r, DSw are selectively selected and switched so that P-P 411 can be detected, but all signals OR,
, DSr, and DSw separately for each P-P detection system 4.
20 may be provided and detect the P-P value of each signal at the same time.

This P-P detection system 420 receives signals DR and DSr.

By detecting the modulation degree (amplitude) of DSw, it is used to know the variation in the intersection angle of the two beams LB, LB□. As explained earlier in Fig. 2, two beams L
When the intersection angle on each grid of B, LB, changes slightly,
Correspondingly, the pitch of the interference fringes on the grating also changes minutely.

Therefore, when adjusting the parallel plane glass 50 for telecentricity adjustment, the P-P detection system 420 simultaneously sends the signal DSw (
or DSr), and confirm that the P-P value is large (does not change).

When adjusting the degree of telecentricity, the fiducial mark plate FM on the wafer stage is used, so there are no unstable factors (resist layer, deformation of the mark, etc.) like the mark WM on the wafer. The P-P value is the laser beam LB, , LB.

As long as the light intensity does not fluctuate, good reproducibility can be expected over a long period of time. Therefore, when the reference mark plate FM is used to obtain the signal DSw, the PP detection system 420 detects the P-P value of the signal DSw, and if it is less than a predetermined value, the Drive control system 56 shown in Figure 7
The parallel plane glasses 54 and 55 are adjusted through the parallel plane glasses 54 and 55 to slightly change the distance between the two beams LB, , LB, in the pupil plane, so as to obtain a predetermined P-P value.

The interval change in the pupil plane of the beams LB, , LB, is the difference between the two beams LB, on the object plane (diffraction grating).

By changing the intersection angle of LB, the pitch of the interference fringes IFw is changed.

It is efficient to perform this fine adjustment of the intersection angle together with the telecentricity adjustment. Further, fine adjustment of the intersection angle can be easily carried out by the hill climbing method, etc., in which the plane-parallel glass 54, 55 serving as the second optical member is continuously tilted within a certain range, and the point at which the P-P value is maximized is found within that range. can.

Furthermore, the P-P detection system 420 monitors the signal DSw during the wafer alignment operation by switching the switch SS, and detects the signal chain of the detected mark position based on the P-P value at the time of mark detection. This is because the marks WM attached to multiple shot areas on the wafer may be significantly damaged by the process depending on the position on the wafer (particularly the peripheral area). , it is also effective in knowing this in advance.

Therefore, marks WM of 3 to 9 shot areas on the wafer
(Enhancement Global Alignment) (D-
In the G-A) method, the position measurement values of the measured marks WM whose P-P values at the time of detection are aligned within a certain range can be used for statistical processing.

As described above, in this embodiment using the stepper shown in FIG. 1, only the alignment systems 1 & IN are shown, but by providing 2 to 3 similar alignment systems, the reticle 1 and the wafer 4 can be moved in the x, y directions. and rotational alignment is possible. Further, in this embodiment, the reference signal D
Since the phase is detected based on R, for example, when exposing the first layer on wafer 4 (first print), the phase difference between signal DSr and signal DR obtained from mark RM of reticle L will be detected during the exposure operation. so that it always has a constant value,
You can also servo-lock the reticle stage.

Next, a second embodiment of the present invention will be described with reference to FIG. 12 as well as FIGS. 5 and 6.

FIG. 2 shows a modification of the arrangement of parallel plane glasses 54 and 55 for adjusting the intersection angle of the two beams LB+1LBI;
Between OM73 and beam splitter 76, and AOM7
A convex lens 400a is provided between the 4 and the beam subrifter, respectively.
400b, and parallel plane glasses 54, 55 are provided between the convex lenses 400a, 400b and the beam splitter 76.

Then, the parallel plane glass shown in FIG. 7 is omitted,
The prism 122 is changed to a convex lens 402. These convex lenses 400a, 400b, and 402 are all connected due to the focal length relationship, and the diffraction point (deflection origin) of the AOM 73.74 is the image conjugate plane IPm in FIG. 12 (FIG. 7). It is set to be almost conjugate. The optical system shown in FIG. 6 is similarly arranged between the beam splitter 76 and the convex lens 402. Further, Pe becomes a pupil conjugate plane, and the two beams LBF and LBs are focused as a spot within the plane Pe by the action of convex lenses 400a and 400b. Therefore, the beams LB, LB, LB, emitted from the convex lens 402 become parallel light fluxes and intersect at the plane IPm.

In this embodiment as well, since the parallel plane glasses 54 and 55 are arranged in the pupil space, the pupils EP of the two beams LB, LB,
, the interval within can be changed arbitrarily, and the intersection angle can be adjusted.

Incidentally, when the diffraction point of the AOM73.74 is conjugated with the image conjugate plane IPm and the arrangement shown in FIG. 5 is adopted as in this embodiment, the drive signal SF to the AOM73.74,
If the frequencies of , SF, are increased or decreased by the same amount (about 10% to several tens of percent) while keeping the difference Δ[ constant, the two beams tBs and LB.

The position within the plane Pe will be shifted by the same amount in the same direction. This means that fine adjustment of the degree of telecentricity is possible with the AOM73.74, similar to the parallel plane glass 50. Therefore, telecentricity adjustment is performed using two AOMs.
The crossing angle can be adjusted by increasing or decreasing the drive signals SF, , SF, 73, 74, and by correcting the inclination of two (or one) parallel plane glasses 54, 55. In either adjustment, the information (phase information and amplitude information) of the signal DSw from the photoelectric detector 25 is used.

Next, the configuration of a stepper according to a third embodiment of the present invention will be explained using FIGS. 13(A) and CB).

13(A) and (B) only show the rear part of the pupil relay system 17B shown in FIG. 3, and the other configurations are the same as in FIG. 3. In this embodiment, the pupil relay system 17B A divided light receiving element 30 is placed at the position of the spatial filter 23 provided after the
0R1300W was directly placed to receive diffracted light (beat signal) 104.105 at the pupil plane of the system. 13th
As shown in Figure (A), a polarizing beam splitter 310 is installed obliquely after the relay system 17B, and since the diffracted light 105 from the wafer 4 is an S-polarized component, it is reflected here and reaches the light receiving element 300W. On the other hand, diffracted light 1 from reticle l
Since 04 is a P-polarized component, the beam splitter 310
and reaches the light receiving element 300R. Light receiving element 300
R1300W has a light-receiving surface as shown in FIG. 13(B), for example, and a surface 300a that receives the diffracted light 104 or 105 is formed in the center, and the upper and lower (or left and right)
Surfaces 300b and 300C that receive the zero-order light Lo are formed on the surfaces 300b and 300C. These three light receiving surfaces 300a, 300b, 3
00c are electrically insulated and each outputs a photoelectric signal individually. In addition, the light receiving surface 3 (10a, 300b, 3
00c is arranged substantially conjugate with the pupil plane of the system, that is, the pupil EP of the projection lens 3.

Therefore, the photoelectric signal on the light receiving surface 300a of the light receiving element 300R becomes the signal DSr, and the photoelectric signal on the light receiving surface 300a of the light receiving element 300W becomes the signal DSr.
The photoelectric signal of a becomes the signal DSw, which is used for phase detection in exactly the same way as in the previous embodiment.

As is clear from the optical path shown in Fig. 3, all of the diffracted light 104.1058 (and the 0th order light LO) are converged on a minute spot on the pupil plane of the system, so the light receiving surface 300
The areas of a, 300b, and 300c can be made extremely small, which is advantageous for improving the responsiveness of the light receiving element. Further, the numerical aperture (N) of the diffracted light beams 104 and 105 on the pupil plane.

A,) is quite small (for example, 0.03 or less), so by making the light-receiving surface 300a a slightly larger area than the exact spot size of the diffracted light 104 and 105, the light from the light-receiving elements 300R and 300W is The positioning accuracy in the axial direction and in the direction orthogonal to the optical axis does not need to be so strict, and the device manufacturing becomes easier. Moreover, the parallel plane glass 50 has two beams LB+, LB
The parallel plane glass 50
No matter how tilted, the relative positional relationship between the diffracted light beams 104 and 105 and the light receiving surface 300a does not change.

Next, a fourth embodiment of the present invention will be described with reference to FIG. 14. Here, an example will be described in which zero-order light L0, which is not used for original grating mark detection, is used.

FIG. 14(A) is a diagram showing the state of the O-order beam Lo generated from the mark WM (or reference mark FM) on the wafer. The +1st-order diffracted light 15-1 generated by the irradiation of the beam LB, . On the other hand, the O-order beam LO, which travels in exactly the opposite direction to the beams LB, . The O-order beam Lo, which travels in exactly the opposite direction to the beam LBts, is the specularly reflected light IS- of the beam LB, .
and the second-order diffracted light 2S generated by the irradiation of the beam LBis
-8 interfered.

Therefore, when a diffraction grating exists within the intersection area of the two beams LB, ..., LB, s, the photoelectric signal from the light receiving surface 300b that receives the θ-order beam LO and the photoelectric signal from the light receiving surface 300c are as follows. Both result in alternating current signals having beat frequencies as shown in FIG. 14(B). However, since the intensity of the specularly reflected light and the second-order diffracted light differ greatly, the waveform has a relatively large offset vof. This waveform can be considered to have a low degree of modulation, similar to that shown in FIG. 2(C).

Therefore, the photoelectric signals from the light receiving surfaces 300c and 300b are manually input to the modulation degree detection system 420 shown in FIG.
-P (a) and the offset value Vof (or average peak value), it is possible to estimate the light amount ratio between the specularly reflected light 2S- (IS-@) and the second-order diffracted light IS-□ (2S-1).

If it is determined from the two photoelectric signals that the light amount ratio is different, it can be determined that some irregularity has occurred in the mark WM. In that case, it is expected that a difference in light amount will occur between the first-order diffracted lights 15-1 and 23-5 of the diffracted light 105, and therefore the degree of modulation of the output signal DSw will also change.

When estimating the irregularity of the mark WM, it is advantageous to use higher-order diffracted light, so it is preferable to use two O-order beams Lo.

Next, a fifth embodiment of the present invention will be described with reference to FIG. 15. In this embodiment, a position detection device (alignment system) that passes through a proximity exposure device in which a mask Ms and a wafer W are brought close to each other. Let's talk about. have different frequencies,
and two parallel beams LB, , LB of the same polarization component
, condense light onto the pupil conjugate plane via lenses LG, , LG, respectively. Parallel plane glasses 54, 55 are provided in the vicinity of the pupil conjugate plane so as to be tiltable. The parallel plane glasses 54 and 55 have different thicknesses and function as the second optical member of the present invention. Two beams LB.

, LB O are further reflected by a right-angle prism mirror PM and combined so that the chief rays are separated by a certain distance and become parallel.
It enters the relay system 17A. I of relay system 17A! (
In the image space), a 1/2 wavelength plate 117 and a polarizing beam splitter 11B similar to those shown in FIG. 6 are arranged.

The S-polarized beams LB, 1, and LB reflected by the polarizing beam splitter 11B each become parallel light beams and enter the reference grating plate 18 at a predetermined intersection angle. The light receiving element 19 receives the ±1st order diffracted light (IS-12S-
, ), and is arranged so as not to receive the O-order beam Lo. The P-polarized beam LBIP% LBtP transmitted through the polarizing beam splitter 118 becomes a parallel light beam, intersects at the image conjugate plane IPm, and then exits from the relay system 17A, is reflected by the beam splitter 20, and is parallelized for telecentricity adjustment. The light enters the telecentric objective lens 21a for alignment via the flat glass 50 (first optical member). At this time, beams LB, P, L
Btp becomes a spot light (beam waist position) at or near the pupil plane EP of the objective lens 21a, and from the objective lens 21a, it becomes a parallel light flux and proceeds to the mask Ms. The beam LBIP, L emitted from the objective lens 21a
In this embodiment, each chief ray of B□ is connected to the mask Ms and the wafer W.
It is set to intersect approximately in the middle of the proximity gap between the two countries. The proximity gap varies depending on the type of light source of the exposure device, the exposure energy irradiation system, etc., but it is known to be set between 10 μm and 500 μm for boat fishing. If the gap is narrow, the intersection fiJf region of the two beams LB1p and LB□ (in the optical axis direction)
However, if the gap is widened, the marks RM and WM definitely exist within the intersection area. The question is whether or not to do so. However, this problem can be easily solved by simply increasing the diameters of the beams LBIP and LBtp that reach the mask Ms, and the length of the intersection region in the optical axis direction can be set relatively freely. Here, too, the positional relationship between the mark RM and the mark WM is the same as that shown in FIG. 4, and the grating pitch of both marks is the same (.

Now, diffracted light (beat signal) from marks RM and WM 1
04.105 is the objective lens 21a, parallel plane glass 50
．． Beam splitter 20. It reaches the pupil conjugate spatial filter 23 via the relay system 17B. Here, O-order beam L
o is blocked, and only the diffracted lights 104 and 105 enter the lens system (
The light passes through the inverse Fourier transform lens) 24 and enters the beam splitter BS. Here, the diffracted lights 104 and 105 are divided into two in light quantity, and reach the light receiving elements 25A and 25B via aperture plates 25A'' and 25B', respectively, which serve as field stops, as shown in FIG.

Here, the aperture plate 25A° is the mark RM of the mask Ms.
The aperture plate 25B° is arranged conjugately with the mark WM on the wafer W and transmits only the image of the mark WM. The reason why the aperture plate 25 is divided into two parts for the mask Ms and the wafer W is that when the gap between the mask Ms and the wafer W is large and the magnification of the objective lens is large, the depth of focus of the objective lens 21a alone can be reduced. This is because it is difficult to make the images of both marks on the same imaging plane. Of course, Figures 1 and 3
If a bifocal optical system is employed as shown in the figure, both 7-dimensional images can be formed on the same plane. Further, if the gap is small, the aperture plates 25A'' and 25B' may be used in common.

In this embodiment as well, the telecentricity adjustment can be performed in exactly the same way as in the previous embodiment by using the fiducial mark plate FM instead of the mask Ms or the wafer W. However, it is also possible to use the marks RM of the mask MS without using the fiducial mark plate FM. To do this, the mask holder holding the mask Ms may be moved up and down in the Z-axis direction.

Furthermore, in order to finely adjust the intersection angle of the two beams LBIF and LBtr, the inclination of the parallel plane glasses 54 and 55 may be adjusted as appropriate. In this case, it is preferable to use the thick parallel plane glass 54 for coarse adjustment and the thin parallel plane glass 54 for fine adjustment.

By the way, the parallel plane glass 50 is a prism mirror PM.
and the relay system 17A.

Even in that case, the parallel plane glass 50 is the lens system LG,
, LGx and the relay system 17A, the same effect can be obtained.

This similarly applies to the cases of FIGS. 1 and 3. Moreover, since the light-receiving elements 25A and 25B are arranged substantially image conjugate, the transmitted light beams LB, .

Even if the parallel plane glass 50 is not provided in the common optical path of LB□ and the diffracted lights 104 and 105, no lateral shift occurs in the diffracted lights 104 and 105 that reach the light receiving elements 25A and 25B.

Although the parallel plane glass 50 is the same as in the case of FIG. 1, it is desirable to arrange it on the pupil conjugate plane or a position as close to it as possible. Similarly, it goes without saying that the objective lens 21a can be configured to be movable in the optical axis direction together with the reflective mirror at the tip in the afocal system according to changes in the observation position of the marks RM and WM.

Next, a sixth embodiment of the present invention will be described with reference to FIG. 16. In this embodiment, two beams L incident on the diffraction grating are
B, The intersection angle of LBI was configured to be finely adjusted using two AOMs 73 and 74 in the frequency shifter 12.

The basic arrangement of AOM73, 74, etc. in Fig. 16(A) is the same as in Fig. 5 above, but here AOM73 and AOM73,
OM74 is the center line j parallel to the Z axis! 180” around c
Place them in a rotated relationship. Therefore AOM? Assuming that the frequencies of the drive signals SF, , SF, to 3.74 are almost the same, the first-order light (beam LB
P , LBs ) is deflected at a diffraction angle symmetrical about the center line 1c in the xz plane in FIG. 16(A).

The beams emitted from the AOMs 73 and 74 pass through lens systems 400a and 400b, respectively, and reach spatial filters 410a and 410b. Lens system 400a, 40
The front focal points of 0b almost coincide with the diffraction points (deflection points) in AOM73 and 74, and the lenses 400a,
The optical axis AXa of the light beam 400b (which coincides with the optical axis of the objective lens system 21 through subsequent synthesis) is determined to be identical to the zero-order light Do. Furthermore, spatial filter 41Oa
, 410b are arranged at the rear focal planes (corresponding to Pe in the 12th drawing) of the lens systems 400a and 400b, respectively, and block the θ-order light Do and transmit only the primary light beams LBp and LBs. After the lens system 400a; 400b,
The chief rays of the beams LB, LB are parallel to the optical axis AXa, and the beam waists of the beams LB, LB are located at the apertures of each of the spatial filters 410a and 410b.

Therefore, the spatial filters 410a and 410b are approximately conjugate with the pupil EP of the objective lens system 21, or the two beams L
By making it conjugate with the optically theoretical waist position of B, , LB, the two beams LB, , LB that reach the diffraction grating can be made into substantially parallel beams. Also, the openings of the spatial filters 410a and 4tob are for the beams LB,...
A little flexibility is provided in the X direction so that the distances dp and ds between the principal ray of LBt and the optical axis AXa may vary slightly.

Now, FIG. 16(B) is a block diagram schematically showing an example of a generating circuit for drive signals SF, . . . SFI.

The drive signals SF, , SF, are generated by a frequency synthesizer part 411, 412 using a crystal oscillator Cx. The output frequency is set using counter 4, which sets the division ratio, etc.
14.416, and the set values CN, . . . CNI are given from the main control system 41 (FIG. 1), etc. Set value CN,
, CN8 are determined so that the frequency difference between the signals SF and , SF becomes 25. Therefore, drive signal SF
,,80.000 M7, 80 respectively as SFt
．． When a high frequency of 0.025 M7 is applied to AOM73.74, both the spacings ap and ds have the same design value. However, if the frequencies of the drive signals SF, SF8 are increased or decreased by a few percent to several tens of percent (in some cases, about twice as much), the distances dp and ds will both change by the same amount from the optical axis AXa relative to the design value. change in the opposite direction. Therefore, on the pupil plane in the alignment system or projection lens 3, the principal rays of the two beams LB, LBs change their positions symmetrically about the optical axis while remaining parallel, and as a result, the mark RMSWM
The angle of incidence of the two beams L B ls%L B Mushrooms, or the beams LB□ and LB□, which reach the angle changes symmetrically, allowing fine adjustment of the intersection angle.

Therefore, in the case of the configuration shown in FIG. 16(A), the modulation degree detection system 420 shown in FIG. 11 and the drive circuit shown in FIG. Using RM etc., two beams LB, , t
, am may be automatically adjusted (or self-checked).

In this case, instead of mechanical adjustment, simply drive signal S
Since it is an electrical adjustment that changes the frequency of F, , SF, it is extremely stable and fast.

As described above, in this embodiment, A in the frequency shifter 12
OM73.74 functions as the second optical member of the present invention. Since the intersection angle is finely adjusted using the AOM 73.74, it is possible to easily deal with fluctuations in the diffraction characteristics due to heat generation and aging of the AOM 73.74.

[Other variations]

(1) Irradiating the diffraction grating mark RM or WM2
The beams LB, , LBI may have different polarization directions. For example, the beam LB is a counterclockwise circularly polarized beam, and the beam L
If B is made into clockwise circularly polarized light, no interference fringes will occur on the grating mark, but if a birefringent element or the like is inserted in front of the light receiving element 19, 25, 300, a beam signal will be obtained in the same way.

(2) In order to obtain a frequency difference between the beams LB, LB, LB, the laser light source 10 may be a Zeeman laser.

(3) Even in the case of the homodyne method in which the frequency difference between the beams LB and LB is made zero, the degree of telecentricity and the angle of intersection can be adjusted in the same way.

(4) By using an AOM, it is possible to easily switch between the homodyne method and the heterogain method by simply changing the frequency of the drive signal, depending on the presence or absence of a frequency difference between the drive signals. The servo lock of the wafer stage during exposure using the hetero gain method can be selectively switched at high speed.

(5) In the TTR and TTL type alignment systems, the wavelengths of the two beams LB+ and LBm may be made close to the wavelength of the exposure light, and the bifocal element (lj [refractive substance) 21a] may be omitted.

In that case, an illumination field stop is arranged on the image conjugate plane in the light transmission path of the beams LB, LBI, to limit the irradiation area of the resist layer on the wafer to the mark WM.

(6) In the configuration shown in Figure 16, AOM73.7
When the diffraction angle of the beam drifts due to heat generation in 4,
The heat generation is monitored by a temperature sensor, and as the temperature rises, the frequency of the drive signals SF, , SF□ is automatically adjusted to generate beams LB, , LB.

It is also possible to keep the intersection angle constant.

〔Effect of the invention〕

As described above, according to the present invention, in an alignment optical system or the like that irradiates two beams so as to intersect on a diffraction grating, the positions of the two beams on the pupil plane of the objective optical system (projection optical system) are set by the same amount in the grating arrangement direction. Since the first optical member is provided to shift the distance between the two beams, the telecentricity of the two beams can be adjusted. Furthermore, since the second optical member is provided to change the distance between the two beams in the pupil plane, the diffraction grating of the two beams can be adjusted. The intersection angle at the top can be adjusted to the optimal one.

Therefore, the accuracy of position detection using a diffraction grating and interference fringes is
It has the advantage of being able to be stably obtained while maintaining the high resolution of tenons.

Furthermore, even if the interference fringes and the diffraction grating are relatively displaced largely in the optical axis direction, extremely high telecentricity is maintained.
The special effect of not interfering with position detection can also be obtained.

For this reason, the focusing (AF) operation during alignment, which is performed in conventional steppers, etc., can be omitted, so there is an advantage that the throughput is improved in sequences using the E-G-A method. In addition, in proximity exposure equipment (for example, an X-ray stepper using SOR) where the gap setting accuracy and the distance setting accuracy between the light source and the mask are relatively low, the mask and wafer are mostly set mechanically. With this, operations such as alignment and exposure can be immediately started, and throughput can be expected to improve as well.

Further, the first optical member and the second optical member receive one detection signal (
For example, it can be adjusted based on
There is no need to provide separate detection systems for telecentricity measurement and cross angle measurement, which is extremely simple.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the configuration of a stepper according to the first embodiment of the present invention, FIG. 2 (A) is a diagram of the principle of the present invention explaining the appearance of interference fringes and a diffraction grating, and FIG. 2 (B) (C) is a principle diagram showing an example of a signal waveform, Fig. 3 is a diagram explaining the main parts of the optical system shown in Fig. 1 in more detail, and Fig. 4 shows the diffraction grating marks and apertures in Fig. 3. A plan view showing the relationship between the plates,
Figures 5, 6, and 7 are diagrams showing the detailed structure of the frequency shifter in Figure 1, Figure 8 is a diagram showing the practical structure of the alignment system in Figure 1, and Figure 9 is a diagram showing the practical structure of the alignment system in Figure 1. FIG. 10 is a graph showing the measured value of telecentricity, FIG. 11 is a diagram showing the connection method of the modulation degree detection system, and FIG. 13A is a diagram showing an optical system according to a second embodiment of the present invention, FIG.
FIG. 14(B) is a diagram showing the configuration of a light receiving system according to a third embodiment of the present invention, and FIG. 14(A) and FIG. 14(B) are explanations of the principle of a fourth embodiment of the present invention. , a diagram showing how diffracted light is generated and the waveform of the beat signal of the zero-order beam; FIG. 15 is a diagram showing the configuration of the alignment device according to the fifth embodiment;
FIG. 16(A) is a perspective view showing the configuration of a frequency shifter according to the sixth embodiment, and FIG. 16(B) is a block diagram showing an example of a drive circuit suitable for the configuration of FIG. 16(A). . [Explanation of symbols of main parts] 1... Reticle, 3... Projection lens, 4. W...
Wafer, 5... Wafer stage, 10... Laser light source, 12... Frequency shifter 17A,, 17B... Relay system, 18... Reference grating plate, 19... Light receiving element, 21a...・Objective lens, 2
1b... Birefringent substance, 23... Spatial filter 25
．． 25A, 25B... Light receiving element, 40... Phase detection system, 41... Main control system, 50.54.55... Parallel plane glass, WP, RM, WM... Diffraction grating mark,
104.105... Interference beat light (diffraction light) from marks RM and WM, 73.74... Acousto-optic modulator (AOM), 300.
30OR, 300W... Light receiving element, 420... Modulation degree detection system, LB... Laser beam, LB,... P polarized beam, LB,... S polarized beam, L B I ・
...beam of frequency f, LB, ...beam of frequency f3, EP, EP, ...pupil plane, SF,,si;'x...AOM drive signal, DR.
...Reference signal, DSw, DSR...Measurement signal, AXa
...Optical axis of alignment system, IFw...Interference fringe, I
P', IPm...image conjugate plane, FM...reference mark plate.

Claims (3)

[Claims] (1) Diffraction grating formed on an object at a predetermined intersection angle
a beam generating means for producing two beams for irradiating from a direction; an objective optical system for irradiating the object at a predetermined intersection angle by passing the two beams at a predetermined distance apart in a pupil plane; and a photoelectric detector that receives diffracted light from a diffraction grating via the objective optical system, and detects the position of the object based on a detection signal from the photoelectric detector, wherein the beam generating means includes: The two beams are arranged in the optical path from the light source to the object, and the two beams are aligned in the grating arrangement direction of the diffraction grating in the pupil plane while keeping the interval between the two beams approximately constant in the pupil plane. The first to be displaced to
an optical member; a second optical member disposed in the optical path from the light source to the objective optical system and changing the interval between the two beams in the pupil plane, depending on the state of the detection signal; A position detection device that adjusts the first optical member and the second optical member. (2) The beam generating means includes two acousto-optic modulators for producing the two beams, and uses the two acousto-optic modulators as either the first optical member or the second optical member. 2. The apparatus according to claim 1, wherein the deflection origins of the two acousto-optic modulators are arranged substantially conjugate with the object with respect to the objective optical system. (3) A drive circuit that aligns the deflection planes of the beams emitted from the two acousto-optic modulators in the same plane and increases or decreases the frequency of the drive signal applied to each of the two acousto-optic modulators by approximately the same amount. 3. The device according to claim 2, characterized in that it comprises: