JPH07119575B2 - Exposure equipment - Google Patents

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an exposure apparatus such as a semiconductor device having a fine pattern and a submicron rule of 1 micron or less.

2. Description of the Related Art Semiconductor devices have been densified more and more recently, and the size of the fine pattern of each element is less than 1 micron. The conventional alignment of the photomask and the LSI wafer at the time of manufacturing the LSI is performed by using the alignment mark provided on the wafer, by rotating the stage on which the wafer is mounted and by biaxial translation.
This was done by overlaying the marks on the photomask and the marks on the wafer, but the alignment accuracy is about ± 0.3 microns, and when forming submicron elements, the alignment accuracy is poor and not practical.

Therefore, in order to improve the alignment accuracy between the reticle and the wafer, a method using a two-beam interference method has been proposed, as disclosed in USP 4,771,180. In this method, as shown in FIG. 16, an alignment grating 121 provided on the reticle 120 is used for alignment between the reticle 120 having a circuit pattern and the wafer 110. The light beam 122 emitted from the light source for alignment is applied to the alignment grating 121. The light beam 122 emitted to the alignment grid 121 is
The diffracted light 112P diffracted at 121 is projected onto the wafer 110 mounted on the wafer stage 150 through the alignment optical system 130 and the reduction projection lens 140. Reference numeral 1 denotes light for exposing a circuit pattern, which projects a circuit pattern of a reticle onto a wafer through a lens 140. As shown in FIG. 17A, the alignment optical system 130 has a pair of Fourier transform lenses 131 and 132, a spatial filter 133, and three mirrors 134, 135 and 136. Alignment optical system 130
Is a pair of Fourier transform lenses 131 and 132, which diffracts the light that is incident on the alignment grating 121 on the reticle 120 and is diffracted.
Only the ± first-order lights 123 and 124 are selected by the spatial filter 133 and crossed at the back focal positions of the Fourier transform lenses 131 and 132 to generate interference fringes, which are projected onto the wafer 110 by the reduction projection lens 140. A positioning grating 111 is also provided on the wafer 110, and the light 200 diffracted by the grating 111 and interfered with each other passes through the reduction projection lens 140 and the positioning optical system 130, and a photodetector 160 (see FIG. 17). (A)
). Then, based on the interference light detected by the photodetector 160, the gratings 111 and 2 on the wafer 110 are
The misalignment with the light flux interference fringes 125 is detected, and the wafer 110 is moved with respect to the reticle 120 based on the detection result, whereby the wafer 110 and the reticle 120 are aligned. Then, after such alignment, the reticle 120
The upper circuit pattern is exposed on the wafer 110 by the exposure light L. As described above, in the method using the two-beam interference method, since the grating 121 on the reticle 120 and the grating 111 on the wafer 110 are aligned with each other through the two-beam interference fringes 125, the reticle 120 and the wafer 110 are aligned. And are aligned with high precision.

However, in such a method, the alignment optical system 130 must be provided between the reticle 120 and the reduction projection lens 140. This alignment optical system 130
A pair of Fourier transform lenses 131 and 132, two mirrors
A large number of optical members such as 134 and 136 are required, and the structure is complicated. Therefore, chromatic aberration correction, the grating 1 on the reticle 120
It is difficult to adjust the image formation position of the interference fringes 125 formed by 21 for optical adjustment. There is also a problem that the position of each optical member in the alignment optical system 130 changes with time. Further, since the alignment optical member 130 is fixed to the reticle 120,
When the circuit pattern 129 on the reticle 120 is exposed on the wafer 110, there is a problem that the alignment optical member 130 narrows the exposure range. As described above, this method has an advantage that the reticle 120 and the wafer 110 can be aligned with high accuracy. However, in an actual semiconductor process, since the circuit patterns are overlaid and exposed by a large number of different types of reticles, it is necessary to make the formation position of the alignment grating 121 different for each reticle. As shown in FIG. 18, the formation position of the grating 121 on the reticle 120 is the exposure area 128 of the circuit pattern.
Outside the area, and is limited to the areas 127a and 127b to which the light rays for alignment are irradiated. Therefore, there is a problem that the circuit pattern formed on the wafer 110 is limited. Further, the alignment between the reticle 120 and the wafer 110 is performed based on the change in the optical output of the interference light diffracted from the alignment grating 111 on the wafer 110 by moving the wafer stage 150. ing. For this reason,
Due to the vibration of the wafer stage 150, the S / N ratio of the detection signal of the interference light may decrease. There is also the problem of low throughput. The method of aligning the reticle and the wafer in the reduction projection exposure method using optical heterodyne interference instead of such a method is described in the 36th JSAP Spring Meeting of Applied Physics (1989), 1a-K-8, P. Reported at .560. In this method, as shown in FIG. 19, He-Ne
The laser light emitted from the dual-frequency laser 230 is emitted from the reticle 22.
The light radiated to the alignment grid 221 on 0 and the alignment grid 21 on the wafer 210 mounted on the wafer stage 250.
The light emitted to 1 is switched by the shutter 260. One of the lights switched by the shutter 260 is applied to the alignment grating 221 on the reticle 220, and the other light passes through the slit 222 provided in the reticle 221, then passes through the reduction projection lens 240, and the wafer. The alignment grating 211 on 210 is illuminated. Each alignment grid 2
The ± first-order light that has been applied to 11 and 221 and diffracted passes sequentially through the color correction lens 271, the lens 272, the polarization beam splitter 273, the lens 274, the polarizing plate 275, and the slit 276, and then enters the photoelectric conversion element 277. And this photoelectric conversion element 277
Based on the phase difference of each interference light detected in
The reticle 220 and the wafer 210 are aligned with each other.
In this method, since the light emitted to the alignment grating 221 on the reticle 220 and the light emitted to the alignment grating 211 on the wafer 210 are switched by the shutter 260, there is a problem that throughput is lowered. is there. Also,
With this method, it has been reported that light having a polarization plane unnecessary for alignment is mixed into the alignment light, and the alignment accuracy decreases.

Therefore, it is an object of the present invention to realize highly accurate and high-throughput alignment by simultaneously aligning a reticle and a wafer using the same interference fringe. Furthermore, by providing the alignment optical system on the reticle, the misalignment due to the displacement of the alignment optical system seen in the conventional example of FIG. 17 does not occur, and the reticle and the wafer can be aligned with high accuracy. The purpose is to achieve.

Means for Solving the Problems According to the present invention, two aligning light fluxes having slightly different frequencies and capable of interfering with each other are provided on an aligning grating provided at a predetermined position on a wafer. So that the interference fringes are formed through the window and the reduction projection lens provided on the reticle, and at the same time, the interference fringes are formed on the alignment grating on the alignment grating provided at a predetermined position on the reticle. The alignment optics that irradiate the light, the first photodetector that detects the heterodyne beat signal of the ± 1st-order light that is diffracted by the alignment grating on the reticle, and interferes A second photodetector for detecting the interfering ± 1st-order light heterodyne beat signals, and respective heters detected by the first photodetector and the second photodetector. A phase meter that detects the phase difference of the dyne beat signal and the phase difference of each heterodyne beat signal detected by this phase meter are detected, the relative position of the wafer to the reticle is measured based on this phase difference, and then the reticle Alternatively, there is provided an alignment optical device characterized by displacing the position of a wafer for alignment.

Operation According to the present invention, it is possible to simultaneously perform the alignment grating on the reticle and the alignment grating on the semiconductor wafer by using the same interference fringe. Therefore, the overlay accuracy and throughput are higher than in the conventional example in which the gratings of the reticle and the wafer are separately aligned. Further, according to the present invention, unlike the conventional example, a positioning optical system is not required between the reticle and the reduction projection lens. Therefore, there is no fluctuation in the overlaying position due to the aging of the alignment optical system, which is stable. Further, there is an advantage that the reduction projection exposure apparatus has a good productivity because there is no complicated adjustment of the alignment optical system as in the conventional example.

EXAMPLE As shown in FIG. 1, the alignment apparatus of the present invention is used to align the semiconductor wafer 10 and the reticle 20 in the reduction projection exposure method. The semiconductor wafer 10 is placed on the wafer stage 30, and the surface thereof is coated with photoresist. As shown in FIG. 2, a circuit pattern area 22 in which a predetermined circuit pattern is formed is provided in a rectangular shape in the center of the reticle 20. As shown in FIG. 1, a reduction projection lens 40 is provided between the reticle 20 and the wafer 10, and the circuit pattern area 22 on the reticle 20 is formed on the wafer 10 by the reduction projection lens 40. The image is reduced and projected. In this embodiment, the reticle 20 and the wafer 10 are aligned by the TTR (Through the lens) -on axis method. The wafer stage 30 can be moved in the XYZ directions, can be tilted, and can be rotated in the θ direction. As shown in FIG.
Two window groups 25, 25, 25 are provided respectively. Each window group
The reference numeral 25 is located outside the outer edges of the three linear circuit pattern areas 22 on the straight line. Each window group 25 has three windows 25a, and each window 25a in each window group 25 is arranged along each outer peripheral edge of the circuit pattern region 22. In each window group 25, the reticle 20 is an LSI of the wafer 10.
It is provided so that a projected image of each window group 25 is formed in the LSI chip area when the chip area is exposed. The reticle 20 includes the alignment grid group 21 and each window group 25.
Three alignment gratings 21a are provided on the outer side of the pair of windows 25a of each window group 25 in pairs. Each alignment grid group 21 has an X direction between the reticle 20 and the wafer 10,
It is used to detect misalignment in each of the Y direction and the θ direction. The light-shielding band 23 is on the lower surface of the peripheral portion of the reticle 20.
Are covered by. This shading zone 23 is the window 25 of each window group 25.
Alignment grid 2 of each alignment grid group 21 without covering a
It only covers 1a. The wafer 10 is also provided with three alignment grid groups 11 (see FIG. 1). Each alignment grid group 11 has three alignment grids 11a. Each of the alignment gratings 11a in each of the alignment grating groups 11 includes a window group 25 provided on the reticle 20.
The predetermined window 25a is provided so as to be located within the projection portion of the window 25a when projected onto the wafer 10 through the reduction projection lens 40. As shown in FIG. 1, two light beams 51 from an alignment optical system (not shown) are provided in each of the alignment grating group 21 and each window group 25 which form a pair of the reticle 20.
And 52 are illuminated. The frequency of each luminous flux 51 and 52 is
Slightly different, both beams 51 and 52 can interfere with each other. The light beams 51 and 52 intersect on the alignment grating 11a to form interference fringes 53. Then, each of the light beams 51 and 52 is diffracted by the alignment grating 21a of the reticle 20 and the interfered ± first-order light is applied to the photodetector 61 provided above the reticle 20. This photodetector 61 is a reticle
The optical heterodyne beat signal strength of the ± first-order light that is diffracted and interfered by the alignment grating 21a of 20 is detected. Also,
The light beams 51 and 52 with which the windows 25a of the window group 25 are irradiated pass through the reduction projection lens 40 after passing through the windows 25a.
Then, each of the light fluxes 51 and 52 is reduced and projected onto the wafer 10 by the reduction projection lens 40 to form an interference fringe 54 on the alignment grating 11a on the wafer 10. Each light beam is diffracted by the alignment grating 11a of the wafer 10 and the interfered ± first-order light passes through the reduction projection lens 40 and the window 25a of the reticle 20 and irradiates the photodetector 62 provided above the reticle 20. To be done.
This photodetector 62 detects the optical heterodyne beat signal intensity of the ± first-order light diffracted by the alignment grating 11a on the wafer 10. The output of each photodetector 61 and 62 is a phase meter 63.
The phase meter 63 controls the drive circuit 64 of the wafer stage 30 so that the phase difference between the optical heterodyne beat signals detected by the photodetectors 61 and 62 becomes zero. . The principle of the alignment device of the present invention will be described with reference to FIGS. 3 (a) and 3 (b).
The two light beams 51 and 52 are U ₁ and U ₂ , respectively, and their frequencies are f ₁ and f ₂ , respectively. When the incident angles of the two light beams 51 and 52 on the wafer 10 are θ, the pitch P of the interference fringes formed by the two light beams 51 and 52 is given by the following equation.

(Where λ is the wavelength of the interference light) If the alignment grating 11a is formed on the semiconductor wafer 10 at a pitch equal to the pitch of the interference fringes or at an integral multiple pitch, the alignment grating of the wafer 10 causes The diffracted ± first-order lights U ₁ (−1) and U ₂ (1) are represented by the following equations, respectively.

U ₁ (−1) = A (f ₁ ) exp {i (2πf ₁ t−δ)} U ₂ (+1) = A (f ₂ ) exp {i (2πf ₂ t + δ)} where δ is on the wafer This is the phase difference of the diffracted light when the alignment grating moves by Δx _W and is expressed by the following equation.

The intensity of the ± 1st order interference light is expressed by the following equation.

_{I W = | U 1 (-1} ) + U 2 (1) | 2 = A W 2 (f 1) + B W 2 (f 2) + 2A W (f 2) B W (f 2) cos {2π (f 1 −f ₂ ) t−2δ} = A _W ² (f ₁ ) + B _W ² (f ₂ ) + 2A _W (f ₂ ) B _W (f ₂ ) cos 2π {(f ₁ −f ₂ ) t −2 (Δx _W / P)} As is clear from this equation, the intensity of the heterodyne beat signal detected by the photodetectors 61 and 62 includes the two-beam interference fringes and the relative displacement Δx _W between the gratings on the wafer. Be done.
Similarly, the intensity I _R of the interference light that is incident on the alignment grating 21a on the reticle 20 and diffracted is expressed by the following equation.

^{_{I R = A R 2 (f}} 1) + B R 2 (f 2) + 2A R (f 2) B R (f 2) cos2π {(f 1 -f 2) t -2 (△ x R / P)} Therefore , The phase difference between the interference light from the wafer 10 and the interference light from the reticle is measured by the phase meter, and if the alignment is performed so that the phase difference becomes 0, the reticle 20 and the wafer 10 are separated by the interference fringes. Can be aligned. In the present invention, the two-beam interference fringes are generated in the area where the two beams intersect. Therefore, since the depth of focus is deep, it is easy to use an alignment wavelength different from the exposure wavelength of the reduction projection lens as the alignment light. FIG. 4 shows a phase change measured by the phase meter 63 when the wafer 10 is moved with respect to the reticle 20. The phase change measured by the phase meter 63 changes periodically at the pitch of the interference fringes. The wafer stage 30 on which the wafer 10 is placed is, as shown in FIG.
It has a direction coarse movement stage 31 and a Y direction coarse movement stage 32 mounted on the X direction coarse movement stage 31. On the Y direction coarse movement stage 32, an X-
Y-θ direction stage 33, Z-α-β direction drive stage 3
4, and the θ direction coarse movement stage 35 are sequentially mounted. Then, the wafer 10 is placed on the θ direction coarse movement stage 35. A laser interferometer length measuring device 36 for detecting the position of the Y-direction coarse movement stage 32 is provided on the Y-direction coarse movement stage 32. As described above, such a wafer stage 30 is driven by the output of the drive circuit 64, and the drive circuit 64 is controlled based on the detection result of the phase meter 63.
The operation of the wafer stage 30 by the drive circuit 63 will be described based on the flowchart shown in FIG. The wafer 10 is first placed on the θ direction coarse movement stage 35 (wafer loading), and the entire wafer stage 30 is moved to a position where the circuit pattern of the reticle 2 is exposed by the exposure light source (coarse alignment). ). This coarse alignment is, for example,
This is performed by moving the entire wafer stage 30 to the position where the rough alignment mark provided on the wafer 10 is observed by the global microscope. Then the wafer
The X-direction coarse movement stage 31, the Y-direction coarse movement stage 32, and the θ-direction coarse movement stage 35 are driven so that the alignment light 11a on the alignment grating 11a is irradiated with the alignment light (X , Y, and θ direction global alignment).
This global alignment is 0.5μ, which is the same as the global alignment performed in the normal wafer process.
It is performed with an accuracy of about m. At this time, the Y-direction coarse movement stage 32 is moved based on the detection result of the laser interferometer length measuring device 36 provided on the Y-direction coarse movement stage 32. Similarly, the X-direction coarse movement stage is also moved based on a laser interferometer (not shown). After that, the alignment grid 21a of each alignment grid group 21 on the reticle 20 and the wafer 10
When the alignment grating 11a in each of the grating groups 11 is irradiated with the two light beams 51 and 52, as described above, the position of the interference light by the alignment grating 21a of the reticle 20 and the interference light by the alignment grating 11a of the wafer 10 is increased. The phase difference is detected by the phase meter 63. Then, X− is set so that the phase difference becomes zero.
The Y-θ direction fine movement stage is driven. As a result, the reticle 20 and the wafer 10 are aligned. In this way, when the reticle 20 and the wafer 10 are aligned,
The exposure light is directly applied to the reticle 20, and the circuit pattern on the reticle 20 is exposed on the wafer 10. The error factors in the alignment between the reticle 20 and the wafer 10 in the present embodiment were analyzed, and the alignment accuracy (error) between the reticle 20 and the wafer 10 due to each error factor was examined, and the results shown in FIG. 7 were obtained. was gotten. 2 as an error factor
Measurement error of alignment grating, manufacturing error of alignment grating, position control error of X direction fine movement stage and Y direction fine movement stage 33, position control error of θ direction fine movement stage 35, manufacturing error of reticle itself by light flux interference sensor Can be considered. In this embodiment, by the TTR- _{on axis} method,
Since the circuit pattern of the reticle 20 is exposed on the wafer 10, there is no error (alignment sequence error) that occurs when the stage 30 is moved to the exposure position, and there are three axes in the X, Y, and θ directions. Since the alignment is performed, neither a positional deviation error due to a movement error of the θ direction coarse movement stage nor an error due to a positional deviation (reticle rotation) when fixing the reticle occurs. The overall error as a whole was 50 nm / 3σ.

FIGS. 8 (a) and 8 (b) show a histograb with overlay accuracy. The overlay accuracy of the conventional method for the alignment grating having the SiO ₂ step is 0.15 μm, but 55 nm / 3σ for the alignment method of the present invention. Furthermore, the overlay accuracy of 55nm / 3σ was achieved even for Al alignment gratings with surface roughness. FIG. 9 is a graph showing the stability of the interference fringe generation position used by the method of the present invention. According to the method of the present invention, the variation amount of the interference fringe generation position with respect to the environmental change is within 10 nm after 18 hours have passed, and therefore the alignment is performed with high accuracy even with respect to the environmental change. As described above, according to the method of the present invention, highly accurate alignment of 55 nm / 3σ can be realized, and since stable alignment can be performed against environmental changes, high throughput is possible, and 64-megabit DRAM etc. It is extremely effective for aligning highly integrated semiconductors. Such TTR- _on
rather than the _axis exposure system, TTL (Through the lens) - off
_When the circuit pattern of the reticle is exposed on the wafer by the _axis exposure method, the reticle 20 'shown in FIG. 10 is used. A square circuit pattern area 26 in which a circuit pattern is formed is formed in the center of the reticle 20 '. One window 27 is provided at a position outside the outer peripheral edge of the circuit pattern region 26. And
One alignment grid 28 is provided outside each window 27. As described above, the wafer 10 is provided with the alignment grid groups 11 each having three alignment grids 11a in the LSI chip area onto which the circuit pattern is projected. The lower surface of the outer peripheral edge of the reticle 20 'is a light-shielding band.
Covered by 29. The shading bands 29 cover each alignment grid 28 but not the windows 27. A wafer stage 90 shown in FIG. 11 is used in a TTL- _{off- axis} exposure method using such a reticle 20 '. The wafer 90 stage includes an X-direction coarse movement stage 91, an X-direction coarse movement stage 92, and an X-direction coarse movement stage 92.
The −Y-θ direction fine movement stage 93 is mounted in order.
Then, Z-α is placed on the XY-θ direction fine movement stage 93.
−β direction fine movement stage 94 and θ direction coarse movement stage 95
Are listed in order. And the θ direction coarse movement stage
The wafer 10 is placed on the 95. A laser interferometer length measuring device 96 for detecting the position of the XY-θ direction fine movement stage 93 is provided on the XY-θ direction fine movement stage 93. The alignment between the reticle 20 'and the wafer 10 is performed as follows. First, the entire wafer stage 90 is the wafer
With the circuit pattern of the reticle 20 ′ being substantially arranged on the position where the reticle 20 ′ can be projected, the laser reticle 20 is moved by a predetermined distance based on the laser interferometer 46 to detect a position on the wafer 10 for detecting a deviation in the X direction, for example. The matching grid 11a is the reticle 2
The 0'X detection window 26 is aligned with the position projected on the wafer 10 by the reduction projection lens. The positional deviation error in this case is about 0.5 μm. In this state,
Positioning light is projected, and as described above, the phase difference between the heterodyne beat signals is measured via the two-photometric interference fringes, and the amount of positional deviation between the reticle 20 'and the wafer 10 in the X direction is detected. Similarly, the entire wafer stage 90 is moved by a predetermined amount from the exposure position of the circuit pattern, and the alignment gratings 11a for detecting the deviation in the Y direction and the deviation in the θ direction on the wafer 10 are moved to the reticle 20 '.
Of the reticle 20 ′ and the wafer 10 are moved to respective positions corresponding to the Y direction deviation detection window 26 and the θ direction deviation detection window 26, respectively, and the Y direction deviation amount and the θ direction deviation amount. The amount of displacement is detected respectively. In this way, the X and Y directions of the reticle 20 'and the wafer 10 are
When the amount of positional deviation in each of the .theta. And .theta. Directions is detected, the entire wafer stage 90 is moved based on the laser interferometer length measuring device 96 and placed at the initial exposure position. Then, at this position, after the position of the wafer stage 90 with respect to the reticle 20 'is roughly adjusted, the XY position is corrected so as to correct the already-measured positional deviation amounts in the X, Y, and θ directions.
The −θ direction fine movement stage 93 is driven, and the wafer 10 is aligned with the reticle 20 ′. Wafer 10 is a reticle
Once aligned with 20 ', the circuit pattern of reticle 20' is exposed on wafer 10. In this embodiment, the stage moves to the exposure position after alignment,
The same thing can be done even if the reticle stage moves.
FIG. 12 shows the result of analyzing the error factors in the alignment between the reticle 20 'and the wafer 101 in this case. In this case, unlike the above-mentioned TTR- _{on- axis} exposure method, the entire wafer stage 90 moves in the X direction between the reticle 20 'and the wafer 10,
The entire wafer stage 90 is moved in the respective directions in order to detect the positional deviation amounts in the Y direction and the θ direction, respectively. Therefore, the movement of the wafer stage 40 causes an error (alignment sequence error) of 30 nm / 3σ. Other error factors are the same as in the case of the TTR- _{on- axis} exposure method, but the amount of movement error of the X-direction coarse movement table and the Y-direction coarse movement table is slightly increased. As a result, the total error as a whole is 62 nm / 3σ, and the TTR- _{on axis is}
Although the alignment accuracy is slightly lower than in the case of the exposure method, the formation position of the alignment grid on the wafer is not limited and the alignment grid can be formed at any position. Therefore, the exposure light source for exposing the circuit pattern onto the wafer can prevent the exposure region from being narrowed by the light for alignment. 3 shows another embodiment of the present invention. In this embodiment, as shown in FIG. 13, the alignment optical system 80 has a laser light source 81 that oscillates laser light of a predetermined wavelength. The laser light emitted from the laser light source 81 is split into two lights by the beam splitter 82, and the respective laser lights are given to the first acousto-optic element 83a and the second acousto-optic element 83b, respectively. Then, each acousto-optic element 83a and 8
The respective laser lights frequency-modulated by 3b are sequentially passed through the half mirror 84 and the lens group 85, and then are irradiated onto a predetermined alignment grating and window of the reticle 20 by the mirror 86. The interference light diffracted by the alignment grating on the reticle 20 and the interference light diffracted by the alignment grating on the wafer 10 are reflected by the mirror 86 and the lens group 85.
After passing through the half mirror 84, each photodetector 61 and 62
Given to. The output of each photodetector 61 and 62 is a phase meter 63.
Is given to. Then, the heterodyne beat signal of each interference light is detected by the phase meter 63. In this embodiment,
Since the polarization direction of the laser light emitted from the laser light source and split by the beam splitter 82 is constant, there is no mixing of different polarization planes in each laser light, which may adversely affect the phase characteristics of the heterodyne beat signal. There is no. Therefore, the reticle 20 and the wafer 10 are aligned with high accuracy. The alignment optical system 80 of the present embodiment is, for example, the 14th
It is adjusted as shown. Alignment optics 80
In adjusting the, a pair of alignment light emitted to the reticle 20 or a pair of parallel light emitted to the reticle 20 in the same path as the alignment light is adjusted by the half mirror 84, the lens group 85, and the mirror 86. The light reflected on the reticle 20 and reflected by the reticle 20 is applied to the half mirror 84 by the mirror 86 and the lens group 85. Then, the half mirror 84 irradiates the light to the mirror 87a, and the half mirror 8a reflects the light again.
4. The reticle 20 is illuminated by the lens group 85 and the mirror 86. The spot position of the light at this time is made to coincide with the spot position of the light emitted from each of the acousto-optic elements 83a and 83b and directly emitted onto the reticle 20 by the half mirror 84, the lens group 85, and the mirror 86. , Lens group
85 and mirror 86 are adjusted. In this way, when the mirror 86 and the lens group 85 are adjusted, a pair of parallel light beams are similarly transmitted to the half mirror 84, the lens group 85, and the mirror 86.
The reticle 20 is irradiated through. The reflected light reflected by the reticle 20 is guided to the half mirror 84 through a similar path, and an interference fringe generated by the interference between the light split by the half mirror 84 and the light incident on the half mirror 84 is generated. , CC
Captured by D image sensor 87b. Then, the interference fringe is displayed on the monitor television 87c, and the lens group 85 and the mirror 86 are adjusted so that the interference fringe becomes clear. 15th
Another method for adjusting the alignment optical system 80 will be described with reference to the drawings. In this adjusting method, the light reflected by the reticle 20 is split by the half mirror 84, and the image forming lens 87e is formed.
Are allowed to pass through. The imaging lens 87e
The light passing through is diffracted by the diffraction grating 87f and is applied to the CCD image sensor 87g. As the imaging lens 87e, one having the same optical ability as the imaging lens 85 used in the alignment optical system 80 is used, and the diffraction grating 87f is used.
Is the same as the reference grating 76 used in the alignment optical system 70 shown in FIG. Also, the imaging lens 87
The respective positions of e, the interference fringe grating 87f, and the CCD image sensor 87g are determined by the three-dimensional length measuring device. 2 reflected from reticle 20 and split by half mirror 84
The luminous flux is applied to the diffraction grating 87f through the imaging lens 87e, and the diffraction grating 87f forms a two-beam interference fringe.
Then, moire fringes are generated between the interference fringes and the diffraction grating 87f. The moiré fringes are captured by the CCD image sensor 87g and are displayed on the monitor TV 87h, and the mirror 86 and the imaging lens 85 are arranged so that the moiré fringes become one.
Is adjusted. As described above, in the present embodiment, since the alignment optical system 80 can be easily adjusted, the reticle 20 and the wafer 10 are aligned with extremely high accuracy.

Effects of the Invention The alignment optical device of the present invention can align a reticle and a wafer with high accuracy in a reduction projection exposure apparatus. The alignment grid provided on the reticle is reduced and projected on the wafer, and is usually five times the pitch of the alignment grid provided on the mask in the proximity exposure apparatus. It can be provided with high accuracy, and more accurate positioning can be performed. Since the pitch of the alignment grating on the reticle can be increased, the alignment optical system that irradiates the alignment grating with light does not need to have high performance, and an inexpensive alignment optical system can be used. Moreover, since the two beams are simultaneously irradiated to the alignment grating on the reticle and the alignment grating on the wafer, the alignment is performed with high throughput.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of an alignment optical device for aligning a reticle and a wafer in an exposure apparatus of the present invention, FIG. 2 is a plan view showing an example of a reticle, and FIG. Schematic diagram for explaining the principle of the optical device, third
FIG. 4B is a graph showing the heterodyne beat signal of the interference light from the wafer alignment grating and the interference light from the reticle alignment grating, and FIG. 4 is a graph showing the phase change with respect to the relative variation between the reticle and the wafer. The figure is a block diagram of the wafer stage, FIG. 6 is a flowchart for explaining the operation of the alignment optical device, and FIG. 7 is the TTR- _{on axis.}
FIGS. 8A and 8B are diagrams showing overlay accuracy with respect to a SiO ₂ step and an Al surface roughened lattice, and FIG. 9 is an interference diagram. Fig. 10 is a plan view showing another example of the reticle, Fig. 11 is a configuration diagram showing another example of the wafer stage, and Fig. 12 is a TTR- _{on axis} exposure method. FIG. 13 is a diagram showing error factors and errors of the alignment optical device, FIG. 13 is a configuration diagram showing another example of the alignment optical system,
FIG. 14 is an explanatory view of the adjustment method of the alignment optical system,
15 (a) and 15 (b) are explanatory views of another adjusting method of the alignment optical system, FIG. 16 is a configuration diagram showing an example of a conventional alignment optical device, and FIG. Enlarged view of part, No.
FIG. 17 (b) is an explanatory view of the principle, FIG. 18 is a plan view of a reticle used in the alignment optical device, and FIG. 19 is a configuration diagram showing still another example of the conventional alignment optical device. is there. 10 ... Semiconductor wafer, 20 ... Reticle, 25a ... Window, 30
...... Wafer stage, 11 ... Lattice group, 40 ... Reduction projection lens, 51, 52 ... Luminous flux, 61, 62 ... Photodetector, 63 ... Phase meter, 64 ... Driving circuit.

Claims (7)

[Claims] 1. An optical device for aligning the reticle and the wafer when reducing and projecting a circuit pattern provided on the reticle onto a wafer, which is slightly different in frequency from the exposure wavelength. Two light beams for alignment, which are not parallel to each other and may interfere with each other, are provided on the reticle so that interference fringes are formed on the alignment grating provided at a predetermined position on the wafer. A positioning optical system for irradiating through a window and a reduction projection lens, and at the same time, irradiating a positioning grating provided at a predetermined position on the reticle so that the interference fringes are formed on the positioning grating; A first photodetector that detects a heterodyne beat signal of ± first-order light diffracted by an alignment grating on a reticle and interferes with the alignment mark on the wafer. A second optical detector for detecting a heterodyne beat signal of ± 1 order light interference is diffracted by,
A phase meter for detecting the phase difference between the respective heterodyne beat signals detected by the first photodetector and the second photodetector, and detecting the phase difference between the respective heterodyne beat signals detected by the phase meter An exposure apparatus characterized in that the relative position of the wafer with respect to the reticle is measured based on this phase difference, and the position of the reticle or the wafer is changed to perform alignment. 2. The exposure apparatus according to claim 1, wherein the reticle and the wafer each have three alignment gratings for detecting the amount of positional deviation between the reticle and the wafer in the X direction, the Y direction, and the θ direction, respectively. 3. The exposure apparatus according to claim 1, wherein the wafer is aligned by detecting a positional deviation amount from the reticle at a position where a circuit pattern on the reticle is exposed on the wafer. 4. The wafer is moved in each direction from the position where the circuit pattern on the reticle is exposed on the wafer by a predetermined amount in each of the X direction, the Y direction, and the θ direction with respect to the reticle. 3. The exposure apparatus according to claim 2, wherein the amount of deviation is detected, and then the wafer is moved to the exposure position and moved by the amount of deviation in each direction for alignment. 5. The alignment optical system comprises a light source for irradiating light of a predetermined wavelength and a light of a predetermined wavelength emitted from the light source.
A beam splitter for splitting, a pair of acousto-optic elements for frequency-converting each light split by this beam splitter, and light that has been frequency-modulated by each acousto-optic element are made to enter one set of lens system, and 2. The exposure apparatus according to claim 1, further comprising means for irradiating the diffraction grating and the window of the reticle with the two-split light flux to generate a two-light flux interference fringe. 6. The alignment optical system has a heterodyne holography alignment optical system including a light source for alignment, a beam splitter, an acousto-optic device, a condenser lens, a mirror alignment grating, a photodetector, and a phase meter. A half mirror, a total reflection mirror, and an image sensor are provided between the condenser lens and an acousto-optic element, and collimated light having the same wavelength as that of the alignment light source is incident on the reticle surface through the condenser lens, and from the reticle surface. Reflected light by the total reflection mirror through the half mirror, and the incident light split by the half mirror and the reflected light are interfered with each other while observing interference fringes on the image sensor, the mirror and the condenser lens. 2. The exposure apparatus according to claim 1, wherein the position adjustment of the two light beams for alignment is performed. 7. The alignment optical system has a heterodyne holographic alignment optical system including a light source for alignment, a beam splitter, an acousto-optic device, a condenser lens, a mirror, an alignment grating, a photodetector, and a phase meter. Then
A half mirror, a condenser lens, and a diffraction grating are provided between the condenser lenses and the acousto-optic element, and two light beams emitted from the light source for alignment and split by the beam splitter are incident on the reticle surface and reflected from the reticle. The generated light is irradiated onto the reference grating through the imaging lens, the half mirror and the imaging lens to generate two-beam interference fringes, and the moire fringes generated between the two-beam interference fringes and the reference grating are observed. The exposure apparatus according to claim 1, wherein the mirror, the condenser lens, and the position of two light beams for position adjustment are adjusted.