JPH0427683B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an alignment device for aligning a mask and a wafer when exposing a mask pattern onto a wafer, and particularly to an alignment device in a proximity exposure device. Regarding.

As shown in Japanese Patent Application Laid-Open No. 111280/1983, the conventional positioning method for proximity exposure focuses on both the mask and the wafer by narrowing down the laser beam with an objective lens with a small numerical aperture. A method is known in which a spot is formed and the spot is scanned to detect scattered light from alignment marks on a mask and a wafer.
However, if the laser beam is narrowed down using an objective lens with a small numerical aperture to focus on both the mask and the wafer as in this method, the laser spot will inevitably become large, making it difficult to obtain high resolution and highly accurate positioning. becomes difficult to do.

Therefore, as shown in Japanese Patent Publication No. 56-1777, a method can be considered in which a bifocal lens is used to focus the laser beam with an objective lens with a large numerical aperture, and the mask and wafer are focused separately. However, if a bifocal lens is simply used, the mark on the mask and the mark on the wafer will interfere with the light beam of the laser beam, and the disadvantage is that positioning cannot be performed with sufficient precision.

In addition, in order to perform positioning with high precision, the above method requires high precision in the scanning mechanism. For example, in order to scan with an amplitude of 100 μm and obtain a position accuracy of 0.01 μm, the scanning mechanism must be 0.01/100 = 0.01 % accuracy was required, and conventional investigation mechanisms were insufficient for highly accurate positioning.

The present invention consists of two parts facing each other like a mask and a wafer.
An object of the present invention is to provide an alignment device that enables alignment of two flat plates with high precision.

That is, in the alignment device of the present invention, at least one flat plate is optically transparent, and the first flat plate having the first mark for alignment and the second flat plate having the second mark for alignment are placed facing each other. In a device for adjusting the relative positions of both flat plates, images are formed on the first plane and the second plane facing each other with a predetermined gap, and the image formation positions are located on the spread of the first and second planes. a luminous flux emitting means for emitting a first luminous flux and a second luminous flux separated by a predetermined distance in a direction; a first flat plate and a second flat plate facing each other, and the first and second luminous fluxes are incident from a light-transmissive flat plate side; Based on the image formation state of each light beam on both flat plates,
a first control means for controlling the first and second flat plates to coincide with the first and second planes, respectively; the first light flux irradiates a first mark, and the second light flux irradiates a second mark; The apparatus further includes a second control means for relatively displacing the first flat plate and the second flat plate in the direction in which the first and second planes extend. In one embodiment, the light beam emitting means has an objective lens having a depth of focus smaller than a predetermined gap between the first plane and the second plane at the wavelengths of the first and second light beams, and the light beam is emitted through the objective lens. The first and second light beams are emitted.

Further, the light beam emitting means polarizes the coherent light beam to obtain the first and second light beams.
It is provided with a polarization separation optical system that separates the light beam into a light beam, and an optical path length correction optical system that makes the optical path length of the two separated light beams different in accordance with a predetermined gap between the first and second planes.

The first control means separates the reflected light of the first and second light beams irradiated onto each of the first and second flat plates by polarization, and individually receives each separated light beam,
The second control means includes a one-dimensional image sensor that detects the size of spot images of the first and second beams formed on the first and second flat plates, and the second control means includes:
It includes a photoelectric detector that separates optical information from the first and second marks on the first and second flat plates by polarization, receives the separated optical information individually, and detects each mark.

The present invention will be described in detail below along with side views of implementation.

1 to 5 are optical path diagrams showing an alignment optical system of a positioning apparatus according to an embodiment of the present invention.

In FIG. 1, a laser beam emitted from a laser light source 1 passes through a beam expander 2, a polarizing plate 3, and a quarter-wave plate 4 to become a parallel beam of light, and enters a cylindrical lens 5. At this time, the laser beam becomes circularly polarized light by the polarizing plate 3 and the quarter wavelength plate 4. The cylindrical lens 5 converts the light beam into a band-shaped light beam with an elongated cross section, and the light beam reaches the vibrating mirror 6 . Note that the focal position of the cylindrical lens 5 coincides with the center of the vibrating mirror 6. The laser beam emitted from the cylindrical lens 5 becomes a band-shaped spot light converged within the plane of the paper on the reflecting surface of the vibrating mirror 6. That is, the cylindrical lens 5 has no power in the plane of the paper of FIG. 1, but has power in a direction perpendicular to the plane of the paper. The light beam reflected by the vibrating mirror 6 spreads in the direction perpendicular to the plane of the paper in FIG.
Parallel flat glass that passes through a lens 7 whose focus coincides with the center of the vibrating mirror 6, and shifts the light flux within the plane of the paper.
It reaches the beam splitter 8 through HG. still,
The vibrating mirror 6 has a rotational axis C of vibration rotation perpendicular to the plane of the paper of FIG. 1, and causes the light beam reaching the lens 7 to minutely vibrate up and down within the plane of the paper. In addition, Figure 2 is an optical path diagram in the direction of arrow AA in Figure 1, and shows the oscillating mirror 6.
is not a reflective surface as a whole, but has an elongated reflective part 6a that matches the cross-sectional shape of the band-shaped spot light, and a transparent part 6b.
It is composed of. The reason why the reflecting portion 6a is made elongated in this way is to prevent the returning light from entering the laser light source 1 (backtalk).
This backtalk prevention is done by polarizing plate 3 and 1/4 wavelength plate 4.
It is also carried out by

Here, each light beam formed by the cylindrical lens 5, the vibrating mirror 6, and the lens 7 will be described in detail as shown in FIG. In FIG. 8, the light beam exiting the cylindrical lens 5 is converged on the reflecting surface of the vibrating mirror 6 into an elongated strip-shaped spot light SP.
After that, the spot light SP spreads at a certain angle,
When the light is incident on the lens 7, it becomes a light beam SP' extending in a direction perpendicular to the longitudinal direction of the spot light SP, as shown by the shaded area in the figure. Therefore, in order for the luminous flux SP' to vibrate minutely as shown by arrow a, the spot light
The spot light SP may be rotated back and forth around the rotation axis C of the vibrating mirror 6, which is orthogonal to the longitudinal direction of the SP.

Now, returning to the explanation of FIG. 1, the band-shaped light beam incident on the beam splitter 8 is split into two, and one of the light beams passes through the 1/2 wavelength plate 9 as the light beam BW for irradiating the wafer and is then sent to the mirror 10. It is reflected and reaches the beam splitter 12 through the optical path length correction prism 11. The other light split by the beam splitter 8 is used as the light flux BM for mask irradiation, which is 1/2
The light passes through a wavelength plate 13 and a parallel plate glass 14 (hereinafter referred to as harving glass 14) that can be tilted to shift the light beam, is reflected by a mirror 15, and reaches the beam splitter 12. In this embodiment, polarizing half prisms are used as the beam splitters 8 and 12 to separate the luminous flux by polarization in order to improve the S/N of the photoelectric signal. For this separation, a 1/2 wavelength plate 9,
After 13, polarizing plates 9 and 1 whose polarization directions are perpendicular to each other
3' is provided.

Now, the light flux incident on the beam splitter 12
BW and BM are combined again into one light beam, and the combined light beam reaches mirror 18 via lens 16 and half mirror 17. The mirror 18 reflects the light beam from the half mirror 17 downward from the plane of the paper.
It is installed at a 45° angle. Moreover, the mirror 15 is
The center of the light beam BM combined by the beam splitter 12 is eccentric with respect to the center of the light beam BW in accordance with a predetermined amount of deviation between the alignment mark on the wafer and the alignment mark on the mask, which will be described later. is set to .

Now, FIG. 3 is a diagram of the mirror 18 seen from the B direction, and FIG. 4 is a diagram of the mirror 18 seen from the C direction.
are respectively focused on the patterned surface of the mask 20 and the surface of the wafer 21, and an elongated band-shaped spot light is imaged on each surface. Assuming that the gap between the mask 20 and the wafer 21 (so-called broximity gap) is g, the numerical aperture of the objective lens 19 is set to be large so that the depth of focus of the objective lens 19 is sufficiently smaller than this gap g. Furthermore, the gap g is an eigenvalue optimally determined by the exposure apparatus. Therefore, the optical path length correction prism 11 is arranged so that the beams BM and BW are each formed into a band-shaped spot light image at a position separated by a gap g in the optical axis direction of the objective lens 19.
Adjustments are made by relatively moving the two trapezoidal prisms 11a and 11b in a direction perpendicular to the optical axis (vertical direction in the plane of the paper in FIG. 1). Furthermore, the prism 11a, 1
Due to the effect of the cylindrical lens 5, the light beam passing through 1b becomes convergent light within the plane of the paper, and becomes a parallel light flux in the direction perpendicular to the plane of the paper. Further, the lens 7 and the lens 16 constitute an afocal system, and the center of the vibrating mirror 6 is set to coincide with the focal position of the lens 7, and the pupil 19a of the objective lens 19 is set to coincide with the focal position of the lens 16. It is being Therefore, regardless of the presence or absence of the optical path length correction prism 11, the vibrating mirror 6 and the pupil 19a of the objective lens 19 are optically conjugate, forming a telecentric irradiation optical system.

When the band-shaped spot light imaged on the mask 20 and the wafer 21 by the light beam emitting means configured as described above irradiates the alignment mark on the mask 20 and the alignment mark on the wafer 21, each mark Diffraction light including zero-order light is generated as optical information. In addition, the mask 20 and the wafer 2
Even if there is no mark on the flat surface, reflected light will occur. This optical information is reversely incident on the objective lens 19, mirror 18, half mirror 17 and mirror 2.
2 and reaches the lens 23. lens 2
A part of the optical information that has passed through the lens 3 passes through the half mirror 24 and reaches the lens 25 . Note that the focal position of the lens 23 is determined to coincide with the pupil 19a of the objective lens 19. An afocal system is formed by the two sets of lenses and the lens 25, and a half mirror 24 is placed in the parallel optical path.

Note that the reversely incident light flux reaching the lens 23 is substantially parallel in the plane of the paper of FIG. 1, and becomes convergent light in the direction perpendicular to the plane of the paper. Therefore, the optical path lights of the lenses 23 and 25 are parallel light beams in the direction perpendicular to the plane of the drawing. The optical information reflected by the half mirror 24 is
The light enters a polarizing half prism 26 similar to the half prisms 8 and 12. The polarizing half prism 26 separates the zero-order light DMo from the Max 20 and the zero-order light DWo from the wafer 21 into two by polarization.
One-dimensional image sensors 27 and 2 such as CCD, respectively
8 to the light receiving surface. One-dimensional image sanse 2
7 and 28 are photoelectric converters for detecting the magnitude of the spot light on the mask 20 and the spot light on the wafer 21. The first control means uses the one-dimensional image sensors 27 and 28 to control the objective lens 1.
By detecting the positional deviation between the mask 20 and the wafer 21 with respect to the focal position of the lens 9, and moving the mask 20 and the wafer 21 relatively in the optical axis direction of the objective lens 19 (hereinafter referred to as the Z direction). , to accurately focus the spot light onto the mask 20 and wafer 21. This can also be used to automatically set the gap g.

Now, the optical information from the mask 20 and the wafer 21 that has passed through the lens 25 is reflected by the mirror 29.
The light reaches a spatial filter 30 which cuts out the zero-order lights DMo and DWo and transmits the diffracted lights of other orders. The first-order or higher-order diffraction light that has passed through the spatial filter 30 is condensed by a condenser lens 31 and enters a polarizing half prism 32 . The polarizing half prism 32 separates the diffracted light into diffracted light DM from the mask 20 and diffracted light DW from the wafer 21 by polarization, and guides them to light source detectors 33 and 34, respectively.

Figure 5 shows a spatial filter 30 and a condensing lens 31.
FIG. 4 is a diagram showing the arrangement of the photoelectric detector 34 and the relationship between the zero-dimensional DWo and the diffracted light DW. Here, the spatial filter 30 is placed at the focal point of the lens 25. Therefore, the pupil 19a of the objective lens 19 and the spatial filter 30 become conjugate, and the spatial filter 3
A pupil image of the objective lens 19 is formed on 0. still,
The spatial filter 30 may be a transparent plate provided with an elongated light shielding part for zero-dimensional cutting according to the shape of the spot light. In the case of the optical system shown in FIG. 1, of the laser beam incident on the beam splitter 8,
The P component in the polarization direction is transmitted through the beam splitter 8, and the S component is reflected. If 1/2 wavelength plate 9,1
3, the luminous flux BM becomes an S component, and mirror 1
After being reflected by the beam splitter 5, the beam is transmitted through the beam splitter 12 without being reflected. On the other hand, the light flux BW becomes a P component, and after being reflected by the mirror 10, it is reflected by the beam splitter 12 without being transmitted. Therefore, the 1/2 wavelength plates 9 and 13 are inserted to rotate the polarization directions of the beams BM and BW by 90 degrees, and the beams are directed from the beam splitter 12 to the left side of the figure as shown in Figure 1.
BM and BW are combined and ejected.

Note that the beam from the beam splitter 12 is directed downward in the figure.
When BM and BW are combined and injected, the above 1/
The two-wavelength plates 9 and 13 are no longer necessary.

Next, the relationship between the spot light imaged by the objective lens 19 and each mark on the mask 20 and wafer 21 will be explained based on FIGS. 6 and 7. Figure 6a is an enlarged view of Figure 4, and Figure 6b
is an enlarged view of FIG. Figure 7 shows objective lens 1.
9 shows the spot light and the arrangement of each mark within the field of view 19b of No. 9.

Now, the light beam BW enters the objective lens 19 such that its center coincides with the optical axis of the objective lens 19. The band-shaped spot light WS generated by the light beam BW passes through the transparent portion of the mask 20 and is focused on a diffraction grating pattern mark 21a formed on the wafer 21 with a width of about 5 μm and a length of about 100 μm. On the other hand, the luminous flux
The band-shaped spot light MS produced by the BM is imaged at a position offset by a distance d in the horizontal direction from the optical axis of the objective lens 19, and is focused on a grid-like pattern mark of approximately 5 μm in width and 100 μm in length formed on the mask 20. The image is formed on 20a. Further, the spot lights WS and MS are set to be parallel to each other, and their longitudinal direction also coincides with the direction in which the gratings of the marks 20a and 21a are arranged. Furthermore, the widths of the spot lights WS and MS are determined to be approximately equal to the widths of the marks 20a and 21a. The spot lights WS and MS vibrate simultaneously in a direction perpendicular to their longitudinal direction. Note that in this embodiment, the amplitude of this vibration is
It is determined to be approximately equal to the width of 1a. Therefore, in this embodiment, the diffracted light from the marks 20a and 21a
DM and DW are spot light MS, as shown in Figure 6b.
The light enters the objective lens 19 at an angle such that the light spreads in the longitudinal direction of the WS. Incidentally, by increasing the length h relative to the width of the spot lights MS and WS, the accuracy of position detection of the marks 20a and 21a is improved. This is because if the length h is large, a large amount of diffracted light will be generated, and the S/N of the detection signal will be improved. Furthermore, a similar mark 20b is formed on the mask 20 in parallel to the mark 20a. The distance between the marks 20a and 20b is determined by the distance between the marks 20a and 20b.
The luminous flux determined by the numerical aperture of 9 and the gap g
It is designed not to block the BW optical path. Furthermore, the mark 20b does not actually contribute to photoelectric mark detection. However, mark 20a
mark 20 so that the distance between and 20b is 2d.
If b is provided, it is extremely convenient for visual alignment.

In addition, in FIGS. 1 and 2, the objective lens 1
Among the 0th-order light beams DMo, DWo and diffracted light beams DM, DW that are incident on the mirror 18, the half mirror 17, and the beam splitter 1, the light beams DMo and DWo are
2 and 8, and reaches the vibrating mirror 6, and enters the laser light source 1 as backtalk. Therefore, as shown in FIG. 9, the band-shaped beams BM, BW and the beams DMo, DWo from the laser light source 1 are arranged so that the beams DMo, DWo pass through the transparent part 6b of the vibrating mirror 6.
slightly shifts the optical path. For this purpose, in this embodiment, the mirror 18 shown in FIGS. 1, 3, and 4 is
tilt slightly from Specifically, the position of the mirror 18 shown in FIG. 3 is slightly rotated within the plane of the paper. In this way, both the spot lights MS and WS in FIG. 7 are shifted slightly in the longitudinal direction within the field of view 19b. However, since this shift does not change the gap d between the spot lights MS and WS, it does not cause any error in detecting the marks 20a and 21a, and does not reduce the accuracy.

Next, a mask and wafer alignment apparatus equipped with an alignment optical system according to an embodiment of the present invention will be described with reference to FIG. FIG. 10 is a schematic diagram partially showing the configuration of the alignment device in the form of a block diagram. In FIG. 10, a stage 51 for two-dimensionally moving the wafer 21 is disposed on a base 50, and the stage 51 moves on the base 50 in an xy plane ( (A surface formed by the x direction in the figure and the direction perpendicular to this (front and back direction of the paper))
It is said that it is possible to move along the This stage 5
The coordinate position of 1 is detected by a laser interferometer 53. In addition, the wafer 21 is placed on the stage 51, and the wafer 21 is slightly rotated, tilted, or tilted relative to the stage 51.
A wafer holder 54 is provided to move the wafer up and down in the z-axis direction perpendicular to the xy plane. Each movement of the wafer holder 54 is performed by a driving section 55 provided on the stage 51. Further, a diffraction grating-shaped reference mark 56 (hereinafter referred to as FM 56) similar to the marks 20a and 21a is provided on the wafer holder 54 in order to calibrate the alignment optical system and verify the gap g between the mask 20 and the wafer 21. There is.

On the other hand, a drive unit 5 for adjusting the position of the mask 20 is provided at the tip of the column 57 extending from the base 50.
8 is provided. This drive unit 58 is the mask 5
Rotating the mask holder 59 holding 0,
It operates by moving it up and down in the z direction or tilting it. Further, an alignment optical system 60 is arranged above the mask 20 so that the objective lens 19 described above is located. The alignment optical system 60 is moved in the direction of the arrow by the drive unit 61 so that it can be taken in and out above the mask 20.
The drive unit 61 also detects the amount of insertion and removal.
It is assumed that the alignment optical system 60 includes the entire configuration of FIGS. 1 to 4, but only the objective lens 19, mirror 18, half mirror 17, lens 16, mirror 22, and lens 23 are moved as one unit for insertion and removal. You can do it like this. The detection device 62 includes one-dimensional image sensors 27 and 2 in the alignment optical system 60, that is, in the configuration shown in FIG.
8. Receives each photoelectric signal from the photoelectric detectors 33 and 34 as input to perform mark detection and focus detection. That is, the one-dimensional image sensor 27,2
In addition to detecting the size of the spot image on the mask or wafer based on the output signal of
Detects the state of image formation on the mask or wafer, that is, the positional shift. The detection device 62 also controls the inclination of the parallel flat glass HG shown in FIG. Also
ACT52, laser interferometer 53, drive unit 55, 5
8, 61 and the detection device 62 are controlled by a control means 63 such as a microcomputer that controls the whole. In this embodiment, these detection devices 62
and the respective functions of the control device 63 to serve as first and second control means.

Although only one alignment optical system is shown in FIG. 10, there are actually three alignment optical systems 60x, 60y, 6 as shown in FIG.
0θ is placed around the mask 20. 11th
Alignment optical system 60x, indicated by a circle in the figure,
60y and 60θ each represent the field of view of the objective lens 19, and each objective lens 19 is moved in and out of the mask 20 as shown by the arrow. Also in this case,
The mask 20 has four circuit pattern areas CP (hereinafter referred to as chips CP) separated by street lines SL as shown in the shaded area in the figure.
A so-called multi-die type can be used. Marks 20a and 20b shown in FIG. 7 are provided at three locations around the mask 20 on the street line SL, corresponding to the positions of the alignment optical systems 60x, 60y, and 60θ. Further, the marks 20a and 20b on the mask 20 are
It is set to extend radially from the center of the The marks 20a and 20b at three locations are observed simultaneously, the alignment optical system 60x detects the position of the mask 20 and the wafer 21 in the x direction, and the alignment optical system 60y similarly detects the position in the y direction. The system 60θ detects the rotation of the mask 20 and the wafer 21 together with the system 60y.

In addition, three alignment optical systems 60x, 60
y and 60θ are each equipped with the one-dimensional image sensors 27 and 28 shown in FIG.
Also detected.

In addition, as shown in Fig. 10, FM56 is used for alignment in the x and y directions as shown in Fig. 12.
A structure with two grid-like patterns extending in the direction is used. Further, the FM 56 may be of the type shown in FIGS. 13a and 13b so as to correspond to the imaging positions of the two laser beam spots MS and WS. In this case, the two parallel lattice patterns are spaced apart by a distance d and are arranged so as to be stepped by an amount corresponding to the gap g.
The use of such FM56 has the advantage that calibration of the alignment optical system and alignment operation can be performed extremely easily.

Now, in FIG. 10, an example of the detection circuit incorporated in the detection device 62 will be explained with reference to FIG. 14.

FIG. 14 is a block diagram of a circuit for detecting the mark 20a on the mask 20. Incidentally, the detection of the mark 21a on the wafer 21 is also performed by a circuit having almost the same configuration, so the explanation thereof will be omitted.

In FIG. 14, the diffracted light from the mark 20a
The output signal of the photoelectric detector 33 that receives DM is input to the preamplifier 100, and output as a voltage signal corresponding to the amount of DM light. The output signal of the preamplifier 100 is input to a phase detection circuit (PSD) 101, and is also input to a comparator 103 where it is compared with a reference voltage 102. The output signal of the PSD 101 is input to a low pass filter (LPF) 104, and output from the LPF 104 as a DC level signal corresponding to the positional deviation between the mark 20a and the spot light MS.

The oscillator (OSC) 105 outputs a rectangular wave signal _S1 and a sine wave signal _S2 of the same frequency, and outputs a rectangular wave signal S1 and a sine wave signal _S2.
is input to the PSD 101 as a reference signal and used to detect the phase difference with the output signal of the preamplifier 101, and the sine wave signal _S2 is power amplified by the power amplifier 106 and used as a drive signal for the vibrating mirror 6 in FIG. It will be done. The PSD 101 can generally be configured with a multiplier, etc., and since its output signal includes ripples (high frequency components), the LPF 10
As described above, only the DC level signal is extracted in step 4.

A comparator 103 compares the output signal of the preamplifier 101 with a reference voltage 102 and outputs the spot light.
This is to determine whether or not the MS and the mark 20a roughly overlap, and this result can be used for rough positioning. That is, as shown in FIG. 15a, the vibration of the vibrating mirror 6 is stopped and the mark 2 is
0a is scanned by the spot light MS, the maximum diffracted light is generated when the spot light MS coincides with the mark 20a, as shown in FIG.
A peak appears in the temporal change of the output signal of 00.
By comparing this with the reference voltage 102, the comparator 103 outputs a pulse signal as shown in FIG. 15c.

While this pulse signal is at the "H" level, it indicates that the spot light MS and the mark 20a are approximately overlapped, so that rough positioning can be achieved. Note that the output signal of the preamplifier 100 is passed through a low-pass filter to the comparator 1.
03, similar rough positioning will be achieved even if the vibrating mirror 6 is kept vibrating.

Next, in order to perform highly accurate alignment, the spot light MS is scanned with respect to the mark 20a while the vibrating mirror 6 is kept vibrating. At this time, the spot light MS is slightly vibrated on the mark surface by the vibrating mirror 6, but the amplitude is set approximately to the width of the mark 20a. The appearance of the mark 20a and the spot light MS in this case and the resulting signal waveform are shown in FIGS. 16a and 16b. That is, Figure 16a
When the spot light MS scans the mark 20a while vibrating minutely, the output signal of the LPF 104 becomes an S-curve signal as shown in FIG. 16b. That is, when the center of the spot light MS and the center of the mark 20a completely coincide, the phase difference between the output signal of the preamplifier 100 and the rectangular wave signal of the OSC 105 becomes zero. In addition, strictly speaking, preamplifier 10 at this time
The frequency component of the output signal of 0 is zero with respect to the oscillation frequency of the OSC 105, and only the frequency component is twice the oscillation frequency of the OSC 105. Furthermore, in this case, a waveform whose phase is inverted on both sides with respect to the center O of the mark 20a is obtained. Therefore, when the center of the mark 20a and the center of the spot light MS are completely aligned, the output of the LPF 104 becomes zero, and the shift When the zero point is aligned, a positive or negative signal appears at the output of the LPF 104 depending on the direction, and extremely accurate positioning is achieved by performing this zero point alignment.

Note that the mark 21a on the wafer 21 is also illuminated by a spot light.
It is detected by signal processing the output of the photoelectric detector 34 by the WS using a completely similar detection circuit.

Next, the overall operation of the apparatus according to this embodiment will be explained according to an actual positioning procedure.

First, as mentioned above, the mark 20a on the wafer 20 and the mark 21a on the mask 21 are illuminated by the spot light MS,
Since alignment is performed by WS, alignment errors will occur if the distance d is not accurately determined. In order to eliminate this error, the interval d can be determined by highly accurate adjustment, but this poses a problem in terms of long-term stability. Therefore, in this embodiment, the first
In the following method, two spot light MS,
Determine the WS interval.

FIG. 17 is a schematic diagram showing parts necessary for explaining this interval determination extracted from FIG. 10, and the same reference numerals indicate the same parts.

First, the stage 51 is moved to align the fiducial mark (FM) 56 shown in FIG. 12 with the spot light WS produced by the wafer alignment light beam BW. At this time, focus adjustment is performed using the one-dimensional image sensor 28 so that the focal position of the spot light WS matches the height of the FM 56. Also, the alignment with FM56 is done at the 14th point so that the center of the luminous flux BW and the center of FM56 match.
This is done by moving the stage 51 using the circuit shown in the figure. When the diffraction grating pattern of the FM 56 and the spot light WS match in this way, the spot light WS is moved above the wafer holder 54 by an amount corresponding to the proximity gap g, and further the mark 20a
The stage 51 is moved toward the mask alignment light beam BM side by an amount equal to the predetermined interval d between and 21a. This movement is performed while reading the measurement value of the laser interferometer 53, and therefore control on the order of 1/100 μm is possible.

Next, a spot light MS is applied to the center of the grid pattern of FM56 using a light flux BM for mask alignment.
Move the luminous flux BM so that the centers of BM coincide. The movement of the light beam BM is performed by rotating the harping glass 14 shown in FIG. 1 while zeroing the S-curve signal of the LPF 104 shown in FIG. 14.

By such a procedure, two spot optical MS
The distance between MS and WS is determined to be exactly equal to the distance d between the mark 20a on the mask 20 and the mark 21a on the wafer 21.
This is done on the basis of

In addition, during the above procedure, the 13th
If the one shown in the figure is used, there is no need to move the stage 51, and the setting of the interval d can be done in a shorter time accordingly.

Also, during the above procedure, the first
Alignment optical system 60y and 60θ shown in Figure 1
Parallelization is also performed. That is, adjustment is made so that the line segment connecting the longitudinal direction of the spot light of the alignment optical system 60y and the longitudinal direction of the spot light of the alignment optical system 60θ coincides with the x direction of the stage 51.

Specifically, the light flux of the alignment optical system 60y
After positioning the stage 56 so that the spot light WS generated by the BW coincides with the FM 56, the stage 56 is moved in the x direction so that the FM 56 is located directly below the alignment optical system 60θ. Then, so that the spot light WS by the light flux BW of the alignment optical system 60θ matches FM56,
The spot lights WS and MS are displaced in the y direction (direction of minute signals) within the field of view 19b of the objective lens 19 in the alignment optical system 60θ. This displacement is
This is done by rotating the harping glass HG shown in Figure 1.

As described above, the alignment optical system 60y
A line segment connecting the detection center (center of microvibration of the spot light) and the detection center of the alignment optical system 60θ is made to match (or be parallel to) the movement locus of the stage 51 in the x direction.

Now, as mentioned above, the two spot optical MSs,
When the centers of WS are set at distance d and gap g, the wafer 21 is placed on the wafer holder 54 and the mask 20 is supported on the mask holder 59, as shown in FIG. Note that, after the mask 20 is once placed on the wafer holder 54, it is removed from the wafer holder 54.
is transferred to the mask holder 59 by moving the mask upward (or moving the mask holder 59 downward).

Next, as shown in FIGS. 10 and 11, the alignment optical system 60 (60x, 60y, 60θ)
Extruded above 0. The positions to be drawn out correspond to the three mark positions on the mask 20.

In this way, each alignment optical system 60 detects the three marks 20a on the mask 20 by the spot light MS of the light beam BM, and the mask holder 59 is moved so that the center of minute vibration of the spot light MS coincides with the center of the mark 20a. It is moved by a drive section 58. At this time, focus detection is performed simultaneously at three positions of the alignment optical system 60, and the inclination of the mask 20 is also adjusted so that the marks 20a at the three positions are all focused on the spot light MS.

In this way, the mask 20 is positioned with respect to the light flux BM from each alignment optical system 60, and the mask holder 59 is thereafter fixed at that position.

On the other hand, the chip CP of the mask 20 is placed on the wafer 21.
It is assumed that pattern areas corresponding to the above are formed in a matrix shape, and alignment marks 21a are formed on the street lines.

Further, the wafer 21 usually has a flat notch, that is, a flat, and when placed on the wafer holder 54, the wafer 21 is positioned at a predetermined position on the wafer holder 54 using this flat. Therefore, if the arrangement of pattern areas within the wafer 21 is known in advance, the distance and direction from the FM 56 to each pattern area within the wafer 21 can be roughly determined. For this purpose, each alignment optical system 6
When the spot light WS of the luminous flux BW of 0 coincides with the FM56, the stage 5 is detected by the laser interferometer 53.
It is sufficient to measure and store the coordinate value of 1.

Therefore, the stage 51 is moved according to the coordinate values of the FM 56 so that the pattern area of the wafer 21 is located directly below the chip CP of the mask 20. As a result, the chip CP and the pattern area are aligned with an error of about several tens of μm.

And three alignment optical shapes 60x, 6
0y, 60θ, detection of the three marks 21a on the wafer 21 is started.

Next, the processing circuit shown in FIG.
Vibration center of spot light WS on 1 and mark 21a
The stage 51 is moved within the xy plane so that the center of As a result, the three marks 21a on the wafer 21 and the three marks 20a on the mask 20 are aligned with each other at intervals d,
The chip CP on the mask 20 and the pattern area on the wafer 21 are precisely overlapped.

When the mask 20 and wafer 21 are aligned in this way, the alignment optical system 60 is retracted,
Exposure illumination light, x-rays, etc. are irradiated from above the mask 20 to transfer the pattern of the mask 20 onto the photosensitive layer on the wafer 21. When the transfer is performed on a part of the wafer 21, the stage 51 moves a certain distance. And again the alignment optical system 60
is extended to a predetermined position above the mask 20. This position is the spot light of the luminous flux BM.
The center of vibration of the MS is determined to coincide with the mark 20a of the mask 20. However, since the alignment optical system 60 is mechanically extended, mechanical positional deviation occurs. Therefore, after pulling out the alignment optical system 60 to a predetermined position, the vibration center of the spot light MS and the mask 20 are aligned.
the first mark 20a so that the mark 20a of the
Rotate the harping glass HG shown in the figure.
After that, as described above, the mark 21 on the wafer 21 is
The stage 51 is moved and aligned so that a and the center of vibration of the spot light WS coincide with each other.
In this way, the pattern of the mask 20 is transferred to the entire area on the wafer 21 by repeating sequential exposure and moving the stage 51 a certain distance, ie, so-called step-and-repeat.

As described above, in this embodiment, a diffraction grating pattern is used as the alignment mark, and this mark is scanned with minute vibrations using a strip-shaped, elongated spot light to detect the diffracted light from the mark. and the center of the mark can be precisely aligned. Moreover, since the vibration amplitude of the spot light is made almost equal to the mark width, the vibration of the vibrating mirror 6 can also be reduced, and extremely high precision alignment can be achieved. Moreover, the mask 20 and the wafer 21 are
Objective lens 19 with a shallower depth of focus than the gap g
Two luminous fluxes BM, separated by a distance d, emitted from
Since mark detection is performed by BW, extremely high precision positioning can be achieved without the diffracted light from the mask 20 and the diffracted light from the wafer 21 interfering with each other. Furthermore, the beams BM and BW are separated into two by polarization, and the diffracted beams DM and
Since DW is also separated into two by polarization, even if the mark 20a of the mask 20 is located in the beam BW that irradiates the wafer 21 during alignment operation, the intensity of the diffracted light DW from the wafer 21 However, the alignment accuracy is not reduced.

In the above embodiment, the mask 20 and the wafer 21 are aligned each time a small area is exposed, but the invention is not limited to this. For example, after aligning the alignment optical systems 60y and 60θ to determine the position of the mask 20,
The alignment optical system 6 aligns marks 21a at two locations separated in the direction (for example, near the periphery of the wafer 21).
The rotation of the wafer 21 may be detected by moving the stage 51 in the x direction so as to detect each other at 0y. Then, the wafer holder 54 rotates by the amount of rotation.
Alternatively, exposure may be carried out sequentially after the mask holder 59 is rotated relatively to align the relative positions of the mask 20 and the wafer 21. Exposure may be performed using only so-called global alignment. Further, even if the wafer holder 54 or the mask holder 59 is rotated, the rotational error will not improve beyond the mechanical accuracy. Therefore, alignment optical system 6
When detecting two marks 21a separated in the x direction on the wafer 21 alternately at 0y, the position _y1 of the stage 51 in the y direction when one mark 21a is aligned with the detection center at 60y, and the position of the other mark 21a in the y direction. mark 21
From the position y ₂ of the stage 51 when a is aligned with the detection center of 60y, σ=y ₁ −y ₂ /x (where x is the distance between the two marks 21a in the x direction) is calculated. Thereby, a minute inclination angle σ between the line segment connecting the two marks 21a on the wafer 21 and the movement locus of the stage 51 in the x direction can be detected. Therefore,
When stepping the stage 51 for exposure, if the stage 51 is moved diagonally with respect to the xy coordinate axes according to the inclination angle σ, it is possible to correct the inclination angle σ, that is, the wafer rotation.
And-repeat exposure is possible, achieving high-accuracy and high-speed alignment. Furthermore, the harping glass HG in Fig. 1 does not require all three alignment optical systems 60x, 60y, and 60θ.
It is sufficient to provide at least the alignment optical system 60θ, and the same effect can be obtained.

Next, a second embodiment of the present invention will be described based on FIG. 18.

FIG. 18 is a modification of the circuit shown in FIG. 14, and blocks having the same functions and effects as those in FIG. 14 are given the same numbers. However, the 18th
The circuit shown in the figure is an embodiment suitable for detecting the diffracted light DW from the wafer 21.

Now, the difference between the circuit of FIG. 18 and that of FIG. 14 is that the output signal of the preamplifier 100 is input to a low pass filter (LPF) 200 and then applied to a comparator 103. The output signal of the LPF 104 is input to the comparator 201,
The zero cross point of the S curve signal from the PSD 101 is detected. AND gate 202 is comparator 1
03 and the output signal of the comparator 201, and when the center of the mark 21a on the wafer 21 and the vibration center of the spot light WS match, a detection pulse signal is output. A counter 204 is provided in the control means 63 of FIG. 10, and counts the interference fringes obtained from the laser interferometer 53, and latches and outputs the count value at the rising edge of the detection pulse signal from the AND gate 202.

Now, with this configuration, the stage 51 is vibrated with respect to the spot light WS that vibrates minutely, and the 19th
When the mark 21a is scanned in the direction of the arrow as shown in FIG. 19(a), the LPE 104 outputs an S curve signal as shown in FIG. 19(b). On the other hand, LPF200 is the first
A peak signal as shown in FIG. 9c is output, and the output signal of the comparator 103 is compared with the reference level 102 to become a pulse signal as shown in FIG. 19d. Further, the comparator 201 outputs a pulse signal which becomes high level only when the S curve signal is positive, as shown in FIG. 19e. Therefore, the AND gate 202 outputs a pulse signal as shown in FIG.
This coincides with the maximum point of the peak signal of LPF200. In this way, when the count value of the counter 204 is latched at the rising edge of the detection pulse signal of the AND gate 202, the position of the stage 51 when the vibration center of the spot light WS and the center of the mark 21a coincide, that is, the position of the mark 21a is determined. location is required. Thereafter, by returning the stage 51 to the latched count value, accurate alignment is achieved. According to this second embodiment, the spot light WS
Just by passing mark 21a once,
After that, alignment can be performed using the measured values of the laser interferometer 53, so high-speed alignment is possible. Furthermore, by ANDing the pulse signal at the zero cross point of the S curve signal of the PSD 101 and the pulse signal corresponding to the light intensity of the diffracted light DW,
Since the position of the center of the stage 1a is detected, when scanning the stage 51, accurate position detection can be performed even if the stage 51 is not moved at a constant speed.

Of course, this second embodiment is not only effective for detecting the mark 21a on the wafer 21;
It is also extremely effective when adjusting two spot light MS and WS using FM56.

In each embodiment of the present invention, as shown in FIG.
The two spot lights MS and WS are the field of view of the objective lens.
The case where they are arranged in parallel within 9b is described. However, the mask 20 and the wafer 21
In order to position the two spot lights MS and WS in one direction, it is only necessary that the longitudinal directions of the two spot lights MS and WS coincide, so for example, as shown in FIG. They may be placed on a line, or they may be shifted left and right (in a direction perpendicular to the longitudinal direction) from a position on a line, as shown in FIG. 20b. Of course, alignment marks may be provided on the mask 20 or the wafer 21 in accordance with the positions of the spot lights MS and WS.

Furthermore, in each embodiment, the two light beams BM and BW are irradiated perpendicularly onto the mask or wafer, but it is not necessary to do so, and the imaging position of the two light beams is spaced by the gap g between the mask and the wafer. The angle of incidence of the two light beams does not need to be 0° as long as they are separated by a predetermined distance d in the direction of the mask or wafer surface that is perpendicular to this direction. Furthermore, it goes without saying that the shape of the spot light is not limited to a band-like shape, but may also be circular.

As described above, according to the present invention, spot light is imaged on each of two facing flat plates with a gap g and a distance d, and based on the optical information from both flat plates, the relative relationship between the two flat plates is determined. Since the position (the direction of the gap and the alignment direction perpendicular to that direction) is controlled, the optical information from both flat plates does not interfere with each other, allowing extremely highly accurate positioning. Furthermore, since the two spot lights are imaged independently on each flat plate, the spot size becomes extremely small, and the alignment has the advantage of high precision and the ability to align the two flat plates at high speed.

[Brief explanation of drawings]

FIG. 1 is an optical path diagram showing an alignment optical system of a positioning apparatus according to an embodiment of the present invention, and FIGS. 2, 3, 4, and 5 are main parts of FIG. 1, respectively. Optical path diagram, Figure 6a is an enlarged optical path diagram of Figure 4, Figure 6b is an enlarged optical path diagram of Figure 3, and Figure 7 is a spot light within the field of view of the objective lens. Explanatory diagram showing the positional relationship between and marks, No. 8
The figure is a schematic explanatory diagram showing the state of the light flux at the vibrating mirror part, and FIG. 9 is a schematic explanatory diagram showing the contrivance of the vibrating mirror to prevent backtalk to the laser light source.
FIG. 10 is a schematic diagram partially showing an example of the configuration of the apparatus of the present invention in the form of a block diagram, FIG. 11 is a schematic plan view showing an example of the arrangement of the alignment optical system with respect to the mask and wafer, and FIG. 12 is a schematic diagram of the reference mark. FIG. 13a and b are enlarged plan views and front views showing another example of the reference mark, FIG. 14 is a block diagram showing an example of the detection circuit, and FIG. 15a is a vibrating mirror. Schematic diagram showing the state of marks and spot light when aligning without vibration, No. 15
Figures b and c are waveform diagrams showing the detection signal in the case of the previous figure,
Fig. 16a is a schematic diagram showing the state of the mark and spot light when aligning by vibrating the vibrating mirror, Fig. 16b is a waveform diagram showing the detection signal in the case of the previous figure, and Fig. 17 is a 2 FIG. 18 is a block diagram of a detection circuit according to another embodiment, and FIG. 19a is an example of the previous figure. Fig. 19 b, c, d, e, and f are waveform diagrams showing the detection signals in the case of the previous figure, and Fig. 20 a, b are schematic diagrams showing the state of the mark and spot light during alignment. FIG. 7 is an explanatory diagram showing different examples of the arrangement of two spot lights within the field of view of the objective lens. 1... Laser light source, 6... Vibrating mirror, 8,
12... Beam splitter, 11... Optical path length correction prism, 19... Objective lens, 20... Mask, 21... Wafer, 20a, 21a... Mark, 27, 28... One-dimensional image sensor, 30
... Spatial filter, 33, 34... Photoelectric detector, 51... Stage, 52, 53... Stage drive unit, 54... Wafer holder, 55... Wafer drive unit, 58... Mask drive unit, 59... Mask holder, 60, 60x, 60y, 60θ... Alignment optical system, 62... Detection device, 63... Control device, 100... Preamplifier, 101... Position detector, 103, 201... Comparator, 10
4,200...low pass filter, 105...oscillator, 106...power amplifier, 202...and gate, 204...counter.

Claims (1)

[Claims] 1. At least one flat plate is optically transparent, and a first flat plate having a first mark for alignment and a second flat plate having a second mark for alignment are faced to each other. , in a device for adjusting the relative positions of both flat plates, images are formed on a first plane and a second plane facing each other with a predetermined gap, and the image formation positions are in the direction in which the first and second planes extend. a luminous flux emitting means for emitting a first luminous flux and a second luminous flux separated by a predetermined distance; a first flat plate and a second flat plate facing each other;
When the first and second light beams are incident from the light-transmissive flat plate side, the first and second flat plates are aligned with the first and second planes, respectively, based on the image formation state of each light beam on both flat plates. a first control means for controlling the first flat plate and the second flat plate so that the first light beam irradiates the first mark and the second light beam irradiates the second mark;
and second control means for relatively displacing the flat plate in the direction in which the first and second planes extend. 2. The light beam emitting means has an objective lens having a depth of focus smaller than a predetermined gap between the first plane and the second plane at the wavelengths of the first and second light beams, and the light beam is emitted through the objective lens. 1. The positioning device according to claim 1, wherein the positioning device is configured to emit the second light beam. 3. The light beam emitting means includes a polarization separation optical system that separates the coherent light beam into two light beams by polarization in order to obtain the first and second light beams; 3. The positioning apparatus according to claim 2, further comprising an optical path length correction optical system that changes the optical path length according to a predetermined gap between the planes. 4 The first control means separates the reflected light of the first and second light beams irradiated onto each of the first and second flat plates by polarization, individually receives each separated light beam, and controls the first and second light beams. 1. The positioning device according to claim 3, comprising a one-dimensional image sensor that detects the size of the spot image of the first and second light beams formed on the second flat plate. . 5 The second control means includes a photoelectric device that separates optical information from the first and second marks on the first and second flat plates by polarization, receives the separated optical information individually, and detects each mark. 4. The alignment device according to claim 3, further comprising a detector.