JPH08316125A - Method and apparatus for projection exposing - Google Patents

Abstract

PURPOSE: To improve the resolution exceeding the diffraction limit by emitting the light from a light source to a mask, diffracting the pattern of the mask, diffracting the diffracted light through a projection optical system, and reproducing the pattern on a sample to be exposed. CONSTITUTION: A mask 1 is inserted between a projection optical system 2 and diffraction gratings A, B, and a diffraction grating C is inserted between the system 2 and a wafer 4. In this case, the gratings A, B, C are simultaneously phase gratings. The light R perpendicularly incident to the mask 1 is diffracted to zero order diffracted light R0, + primary diffraction light R1 and - primary diffracted light R1' on the mask surface. The light R0 arrives at a point A0 on the grating A, and the light diffracted in the - primary direction is diffracted to + primary direction at the point B0 on the grating B. Thereafter, it is diffracted at the point C0 on the grating C via the left end of the pupil 3 in ± primary direction, and arrived at two points Q, P on the image surfaces.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern forming method for forming a fine pattern of various solid state elements and a projection exposure apparatus used for the method.

[0002]

2. Description of the Related Art In order to improve the degree of integration and operating speed of solid-state elements such as LSI, circuit patterns are becoming finer. Further, in order to improve the characteristics of optical / electronic elements such as lasers, various quantum effect elements, dielectric / magnetic element, etc., miniaturization of patterns is desired. At present, the reduction projection exposure method, which is excellent in mass productivity and resolution performance, is widely used for forming these patterns. Since the resolution limit of this method is proportional to the exposure wavelength and inversely proportional to the numerical aperture (NA) of the projection lens, the resolution limit has been improved by shortening the wavelength and increasing the NA.

As a method for further improving the resolution of the reduced projection exposure method, various image improving methods such as a phase shift method, a modified illumination method (oblique incidence illumination method), and a pupil filter method have been applied. These are intended to effectively use the performance of the conventional optical system to the limit of the theoretical diffraction limit (cutoff spatial frequency = 2NA / λ). For these image improvement methods (often called super-resolution methods), for example,
Innovation of ULSI Lithography Technology, Chapter 1, pp. 34-49 (Science Forum, 1994,
(Tokyo).

On the other hand, as a method for improving the resolution of a microscope beyond the conventional diffraction limit, there are known some methods for expanding the spatial frequency band of an optical system. These spatial frequency band expansion methods are described in, for example, Applied Physics, Volume 37, No. 9, pp. 853 to 859 (1968).
Year). One of these methods scans two grating patterns directly above the object and the image (at least within the depth of focus) while maintaining a conjugate relationship with each other. Demodulation is performed by forming a pattern, passing this moiré pattern through the lens system, and superposing it on the second grating pattern on the image side. Since the moire pattern has a lower spatial frequency than the object and the first grating pattern, it can pass through the lens system. It has been filed to apply this method to a reduction projection exposure method. Generally, it is difficult to mechanically scan the grating pattern directly on the wafer. Therefore, a photochromic material is directly provided on the wafer, and interference fringes are superposed on the wafer to perform scanning so that the grating functions.

[0005]

However, the various conventional techniques described above have the following problems.

First, the shortening of the wavelength of the exposure light is caused by optics (lens).
ArF excimer laser (wavelength 1
(93 nm) is considered to be the limit. In addition, due to problems in lens design and manufacturing, it is considered that the NA of the projection optical system is limited to 0.6 to 0.7. However, the resolution limit of the conventional exposure method is generally 0.5λ / NA, and it is about 0.3λ / NA when the periodic phase shift method is used. Therefore, the above-mentioned limits for shortening the wavelength and increasing the NA are used. However, it is difficult to form a pattern of 0.1 μm or less. Further, since the mask pattern is limited in the above-mentioned periodic phase shift method, the actual critical dimension is further reduced for more general circuit patterns. or,
Although it is required to increase the exposure area with the increase in the scale of LSI, it is extremely difficult to simultaneously satisfy the requirements of the exposure field of the projection optical system and the high NA.

On the other hand, various spatial frequency band expansion methods for the purpose of exceeding the conventional diffraction limit target a microscope,
The purpose is to magnify a small object. Therefore, there is a problem that it is not necessarily suitable for forming a minute optical image required in optical lithography. For example, in the method using the moire pattern, a mechanism or an optical system for scanning the two gratings directly above the mask and the wafer while maintaining a conjugate relationship with each other becomes extremely complicated. Since the exposure of the resist is substantially performed with the evanescent light, there is a problem that light is attenuated in the wavelength range and it becomes difficult to expose a thick resist. Furthermore, there are no suitable materials when using photochromics. Therefore, there is a problem that it is not always practical when considering mass production of LSI.

It is an object of the present invention to provide a method for improving the resolution of a projection exposure method for forming a fine pattern of various solid-state elements beyond the conventional diffraction limit (cutoff spatial frequency). Specifically, the NA of the projection optical system
It is an object of the present invention to provide a novel projection exposure method that can obtain an effect substantially equivalent to that when the NA is substantially doubled at most without changing the above, and an exposure apparatus that enables this.

Another object of the present invention is to obtain the effect of improving the resolution without changing the structure of the conventional exposure apparatus and the optical system to a large extent, and only by making some modifications to the apparatus, and to obtain a large exposure. It is an object of the present invention to provide a projection exposure method suitable for mass production of LSIs that simultaneously satisfy the field and high resolution.

[0010]

The above object is to provide a projection optical system for a mask pattern (numerical aperture = N) using light of wavelength λ.
When a pattern is formed by forming an image on a substrate with A and a reduction ratio = 1: M, a spatial period P1 (where λ / (1 is provided between the substrate and the projection optical system, in parallel with the substrate. .42
(NA) ≦ P1 ≦ λ / NA is desirable), and an image of the mask pattern is reproduced near the substrate surface due to interference of light diffracted by the first diffraction grating. As described above, it is achieved by providing, between the projection optical system and the mask, two diffraction gratings, that is, a second diffraction grating and a third diffraction grating in order from the mask side in parallel with the mask. .

In order to faithfully reproduce the image of the mask pattern by the diffracted light of the first diffraction grating, the period directions of the first, second and third diffraction gratings are the same, and the first diffraction grating A spatial period P1, a spatial period P2 of the second diffraction grating,
The spatial period P3 of the third diffraction grating is approximately 1 / P3 = 1 /
Set so that the relationship of P2-1 / (M · P1) is satisfied.
The optical distance Z1 of the first diffraction grating from the substrate surface and the optical distances Z2 and Z3 of the second and third diffraction gratings from the mask surface are approximately (Z3-Z2) / P2. = (Z3 / M + Z1 · M) / P1. Furthermore, P2 ≦ 1 / (1
-2 · NA / M) is desirable. Further, the installation positions of the first, second, and third diffraction gratings, the film thickness of the transparent substrate of each diffraction grating, and the period of the second diffraction grating are set so that the aberration between the mask surface and the image plane is minimized. It is preferable to set so that In addition, the width is Z1 · N between the substrate and the first diffraction grating.
A second light-shielding pattern having a spatial period of about 2 · Z1 · NA or less, and a region substantially conjugate with the first light-shielding pattern immediately above or below the mask. It is preferable to limit the exposure area by providing the light shielding pattern. Further, if necessary, it is preferable that the limited exposure region is scanned and exposed on the substrate, or is exposed while being moved in steps. Each of these diffraction gratings is preferably a phase grating.

The diffraction grating is a one-dimensional diffraction grating, and the wavefront aberration of the projection optical system is line-symmetrical with the diameter in the direction perpendicular to the periodic direction of the diffraction grating on the pupil as an axis. It is preferable to correct aberration. Further, the present invention exerts a particularly great effect when a periodic type phase shift mask is used as the mask. Furthermore, it is desirable to limit the period and direction of the fine pattern or correct the pattern shape according to the period and direction of the diffraction grating as needed. Further, the space between the first diffraction grating and the substrate is filled with a liquid having a refractive index n larger than 1, and the NA of the projection optical system is
Is set in the range of 0.5 <NA <n / 2, it becomes possible to form a finer pattern.

[0013]

The present invention is equivalent to effectively increasing the NA by providing a diffraction grating between the final element of the projection optical system and the wafer and increasing the incident angle of the light beam incident on the wafer surface. It is about getting an effect. However, simply providing a diffraction grating between the lens and the wafer of the conventional optical system causes the diffracted light that should be concentrated at one point on the image plane to be scattered at different positions on the image plane.
It is very difficult to reproduce the mask pattern. Therefore, it is necessary to reconfigure the optical system so that an image faithful to the original mask pattern is reproduced as a result of interference. Moreover, from the viewpoint of practicality, it is preferable that these optical systems can use conventional masks without major modification of conventional projection optical systems. The present invention satisfies these requirements as described below.

In order to explain the operation of the present invention, the principle of imaging according to the present invention will be described in comparison with the conventional method. FIG. 1 shows an image formed in an optical system according to one embodiment of the present invention, and for comparison, a result obtained when a conventional mask or a phase shift mask is illuminated vertically or obliquely in a conventional projection exposure optical system. The appearance of the image is shown in FIGS. 2a, b, c and d. In both figures, paraxial imaging approximation was performed assuming a 2: 1 reduction optical system, coherent illumination, and a one-dimensional pattern.

First, when the mask is normally illuminated vertically by the conventional optical system (FIG. 2a), the light 22 vertically incident on the transmissive mask 21 is diffracted by the pattern on the mask, and the pupil of the projection optical system 23 out of the diffracted light. Light rays passing through 24 (inside the diaphragm 20) converge on the image plane 25 and interfere with each other to form a pattern. Here, when the pattern period that gives the maximum diffraction angle that can pass through the pupil is defined as the resolution limit, the resolution limit is λ
/ (2NA) (where NA = sin θ ₀ ). Further, when the periodic phase shift mask 26 is applied to this optical system, the 0th-order diffracted light disappears and diffracted light is generated symmetrically with respect to the optical axis 29 (dotted line in the figure) as shown in FIG. 2b.
Therefore, the maximum diffraction angle that can pass through the pupil is doubled,
The resolution limit improves to λ / (4NA).

Further, when oblique illumination is applied to the conventional optical system (FIG. 2C, for simplicity, it is assumed that the 0th order light 27 of the mask diffracted light passes through the left end of the pupil in the figure), the 0th order light of the mask diffracted light. Only one-sided component (+ 1-order light 28 in the figure) having a positive or negative diffraction angle with respect to the center passes through the pupil and converges on the image plane. Since the diffracted light having a diffraction angle twice that of the case of vertical incidence can pass through the pupil, the resolution limit is still λ / (4N
A). However, since only one side of the diffraction spectrum is used, for example, the resolution of the isolated pattern is the same as that in the case of vertical illumination, and even in the case of the periodic pattern, there is a problem that the contrast is lowered. Furthermore, if the mask is changed to the periodic type phase shift mask 26, a plurality of diffracted lights cannot pass through the pupil, so the pattern is not resolved (FIG. 2d).

Next, FIG. 1 shows the image formation in the optical system according to one embodiment of the present invention. The optical system of FIG. 1 has a diffraction grating A and a diffraction grating B between the mask 1 and the projection optical system 2 and a diffraction grating C between the projection optical system 2 and the wafer 4 in the conventional optical system of FIG. It is inserted. Here, the diffraction gratings A, B, and C are all phase gratings.

The light R vertically incident on the mask 1 is 0-order diffracted light R0, + 1st-order diffracted light R1, and -1st-order diffracted light R on the mask surface.
Diffracted into 1 '. The 0th-order light R0 is a point A0 on the diffraction grating A.
The light that has reached the point where the light is diffracted in the −1st-order direction is diffracted in the + 1st-order direction at the point B0 on the diffraction grating B, and then passes through the left end of the pupil 3 (inside the diaphragm 5) to the point on the diffraction grating C. The light is diffracted in the ± 1st order directions at C0 and reaches two points Q and P on the image plane, respectively.
Further, the + 1st order diffracted light R1 reaches the point A1 on the diffraction grating A, and the light diffracted in the −1st order direction is diffracted in the + 1st direction at the point B1 on the diffraction grating B, and then the right end of the pupil 3. After passing through, the light is diffracted in the ± first-order directions at the point C1 on the diffraction grating C and reaches the points Q and P on the image plane. On the other hand, the optical paths for the 0th-order light R0 ′ and the −1st-order diffracted light R1 ′ diffracted in the + 1st-order direction at the point A0 are symmetric with respect to the optical path of the two light rays described above with respect to the optical axis 6 (the one-dot chain line in the figure). . That is, both of them finally diffract in the ± first-order directions at the point C0 on the diffraction grating C and reach the points P and Q ′ on the image plane. Therefore, at point P, the 0th-order light and the + 1st-order and -1st-order rays diffracted by the mask intersect. Obviously, this does not depend on the mask diffraction angle. Therefore, the diffraction image is faithfully reproduced at the point P.

Compared to the conventional method (FIG. 2a), the same N
A. An optical system having a magnification can be used to pass diffracted light having a diffraction angle of 2 times through the pupil. Therefore, the same effect as when NA is substantially doubled can be obtained. Moreover, in the oblique illumination (FIG. 2B), only the positive or negative diffracted light centered on the 0th-order light can be reproduced on the image plane, whereas in the present invention, the diffracted light on both sides can be reproduced on the image plane. It is possible to improve the resolution of the isolated pattern, which was difficult with, and it is possible to obtain a large contrast with respect to the periodic pattern. further,
If a periodic phase shift mask is applied to this optical system (Fig. 3
a) As a result of the 0th-order diffracted light disappearing and the + 1st-order light R + and the -1st-order light R- having a diffraction angle twice the normal ones interfering, the minimum resolution becomes λ / (8NA). This is half of λ / (4NA), which is the theoretical limit when the periodic type phase shift mask or the oblique illumination is used, and the present invention enables a dramatic improvement in resolution. Further, FIG. 3b shows a state of image formation when oblique illumination is applied in the present optical system. The oblique illumination allows the diffracted light R1 ″ having a large diffraction angle with respect to only one side to pass through the pupil, and the resolution can be improved up to twice that at the time of vertical illumination, that is, λ / (8NA). If various illumination lights with different mask incident angles are used, the effect of partial coherent illumination can be obtained just as in the conventional optical system.

The principle of the present invention will be described below from the standpoint of Fourier diffraction theory (FIG. 4). In the following description, the magnification of the optical system is 1, and the diffraction grating is a one-dimensional phase grating ±
Only the first-order diffracted light is considered. When the pupil 3 is viewed from the point P on the image plane through the diffraction grating C, the pupil appears to be divided into two due to diffraction (FIG. 4a). In each pupil, a mask Fourier transform image that passes through the pupil at a specific angle can be seen. On the other hand, considering the mask side, the light diffracted by the mask is diffracted by the diffraction gratings A and B to form a plurality of mask Fourier transform images on the pupil. this house,
What has passed through the pupil at a certain angle will be visible in the pupil seen above (Fig. 4b). That is, in the case of FIG.
The right Fourier diffraction image of FIG. 4b is visible in the left pupil of FIG. 4a and the left Fourier diffraction image of FIG. 4b is visible in the right pupil of FIG. 4a. At this time, the conditions for the image to be reproduced correctly at the point P are the following two points.

(1) The spectrum of the same point on the mask can be seen through the two pupils.

(2) The two spectra are connected continuously at the contact points of the two pupils.

In other words, it is necessary to be able to see one continuous spectrum through multiple pupils.

As seen from the image, a plurality of pupils shifted by f ′ can be seen through the diffraction grating C, and a plurality of Fourier diffraction images also shifted by f ″ can be seen through the diffraction gratings B and A in each pupil. Then, the amplitude distribution u (x) of the true image is expressed by the following equation.

U (x) = F [Σp (f−f ′) · Σo (f−f ″)] f ′ = ± SC f ″ = ± (SA−SB−SC) where F [] is Fourier Transformation, p (f) is the pupil function, o
(f) is a mask Fourier diffraction image, x is real space coordinates, f is spatial frequency coordinates, SA, SB, and SC are diffraction angles sin of the diffraction gratings A, B, and C, and Σ is the sum for different diffraction orders. Represents Therefore, if SA = SB + SC, then f ″ = 0, and for both f ′ = ± SC, the terms f ″ = 0 can be obtained. That is, one spectrum o (f) can be seen through the two pupils p (f ± SC). Further, in order to obtain an image of the same point on the mask at the point P,
The distance between the mask surface and the diffraction gratings A and B, and the distance between the diffraction grating C and the ideal image plane, ZA, ZB, and ZC, respectively, may be set to SA. (ZB-ZA) = SC. (ZB + ZC).

Under the paraxial approximation, the reduction ratio M: 1,
When applied to an optical system with an image-side numerical aperture NA, the diffraction grating A,
B, C periods PA, PB, PC, mask surface and diffraction grating A,
Distances ZA and ZB between B, distance Z between the diffraction grating C and the ideal image plane
It can be seen that C should be set almost as follows.

1 / PA = 1 / PB-1 / (M.PC) (ZB-ZA) / PA = (ZB / M + M.ZC) / PC Further, in order to obtain a sufficient resolution improving effect according to the present invention. , Λ / NA ≦ PC ≦ √2 · λ / NA is preferable.

The diffraction gratings A and B are preferably phase gratings. When the diffraction gratings A and B are not perfect phase gratings and transmit 0th-order light, the effects of the conventional optical system and the oblique incidence optical system, which are inferior in resolution to this method, overlap the effects of this method. Therefore, the resolution may be deteriorated. On the other hand, the diffraction grating C may be a phase modulation grating or an amplitude intensity modulation grating. The period of the diffraction grating C is quite small, and the refractive index is 1.
Considering the silicon oxide film of No. 5, the cross-sectional aspect ratio of the lattice pattern is about 1. In this case, it is necessary to pay attention to the light scattering effect on the pattern cross section. In the case of a diffraction grating having a light-shielding pattern, the thickness of the light-shielding film can be made considerably thin, so that the influence of scattering can be reduced. However, as will be described later,
The exposure area can be widened by using the phase modulation grating.

When the substrate side of the diffraction grating B is filled with a liquid or the like having a refractive index n larger than 1, the wavelength of this region and the si of the diffraction angle are
n becomes 1 / n. Therefore, if the period of the diffraction grating B is made finer and the diffraction angle is made equal to that in the case where the liquid is not filled, only the wavelength becomes 1 / n, and the resolution also improves to 1 / n. In this case, on the mask side, it is necessary to increase the mask illumination angle so that diffracted light with a larger diffraction angle can pass through the pupil, but at this time, diffracted light with a smaller diffraction angle cannot pass through the pupil. Therefore, it is desirable to increase the diameter of the pupil accordingly. This can be rephrased as follows. When the refractive index between the diffraction grating B and the substrate is 1, even if the NA of the projection optical system used in the present invention is 0.5 or more, no improvement in resolution can be obtained. This is because the diffraction angle with respect to a light ray incident on the diffraction grating B having an angle θ of sin θ> 0.5 and a period λ / NA is 90 degrees or more, and it is localized as an evanescent wave on the surface of the diffraction grating and is not transmitted to the wafer. On the other hand, if the refractive index between the diffraction grating B and the substrate is n, then sin
The diffraction angle θ ′ of the light incident on the diffraction grating B (the 0th-order light passing through the edge of the pupil must have a period λ / NA in order to vertically enter the wafer) at an angle θ = NA is sin θ ′ = (Λ / PB + sin θ) / n = 2NA / n, and the condition for θ ′ <90 degrees is NA <n /
It becomes 2. That is, the present invention can be effectively applied to an optical system having a maximum NA = n / 2. Generally, the immersion optical system requires a special optical design, but when applied to the present invention as described above, no special lens is required. Therefore, if a gap between the diffraction grating B and the substrate is filled with water (refractive index of about 1.3) and exposed using a projection lens with an NA of about 0.6 that is normally used in semiconductor processes, the NA is substantially reduced. 1.
An effect equivalent to setting 2 is obtained. In this case, if a phase shift mask is used, the wavelength of the i-line of the mercury lamp (36
5 nm), a resolution of 0.1 μm or less can be obtained. In this method, since the incident angle of the light that interferes in the vicinity of the wafer is extremely large, the imaging performance strongly depends on the polarization state of the light. In general, light having a polarization state in which the electric field vector is perpendicular to the plane of incidence of light is desirable for forming a high-contrast image.

In all the above discussions, the paraxial approximation is assumed, and the refractive index of the substrate of the diffraction grating is set to 1. In practice, the effect of the refractive index of the substrate of the diffraction grating and the aberration caused by the diffraction grating are The impact needs to be carefully considered. Therefore, the installation position of each diffraction grating may be slightly changed. It goes without saying that it is preferable to match the periodic directions of the patterns of the plurality of diffraction gratings with sufficient accuracy.

Next, four points to be noted in the present invention will be described.

First, in the present optical system, the exposure area is generally limited as compared with the conventional exposure method. As can be seen from FIG. 1, two rays intersect at points Q and Q ′ on the image plane and interfere with each other to form an image. This image is a false image generated at a position where it should not be originally formed, and is generally not preferable. To avoid this, as shown in FIG.
Directly above 1 (between the wafer and the diffraction grating C), a light-shielding mask 5
It is desirable to provide two to block these spurious images.
The diffraction grating C and the light shielding mask 52 can be formed on both surfaces of the same quartz substrate 53 as shown in the figure. (It may be formed on a separate substrate.) At the same time, a masking blade that shields a region substantially conjugate with the above-mentioned light-shielding mask is provided immediately above or below the mask to provide mask illumination. It is preferable to limit the area to the conjugate area. The exposure area that can be transferred by one exposure is the distance between the true image (point P) and the false image (point Q) (approximately 2 · NA ·
ZB) repeatedly appears in a region corresponding to twice the above distance in a region corresponding to ZB). Therefore, when the area that can be exposed is smaller than the area to be exposed, it is desirable to scan the exposed area on the wafer as shown in FIG. 5b. At this time, if the reduction ratio of the optical system is M: 1, it is needless to say that the ratio of the mask scan speed to the wafer scan speed is also strictly set to M: 1. Regarding the method of synchronously scanning these exposure regions on the mask and the wafer, the method used in the existing exposure apparatus can be used as it is. On the other hand, when the exposed area is larger than the area to be exposed, that is, when the distance between the true image and the false image covers one chip, for example, the exposure can be performed without scanning. The size of the exposure area is the diffraction grating B
Depending on the installation position of the exposure area B, the width of one exposure region increases as the diffraction grating B is separated from the image plane. However, since the width of the non-transferable region also increases at the same time, the ratio of both remains almost 1: 1. In order to eliminate the influence of false images, it is desirable that the width W of the exposure area on the wafer be W ≦ NA · ZB. When an amplitude intensity modulation grating is used for the diffraction grating B, the 0th-order diffracted light of the grating forms another false image at the midpoint between the true image and the false image, so that the exposure region is a phase grating. It is almost half of the case of.

Second, this method generally lowers the exposure intensity. The light beam imaged on the wafer in this method uses only light of a specific diffraction order among the light beams diffracted by the diffraction grating inserted in the optical system. Therefore, the light intensity that contributes to the exposure decreases each time the light passes through the diffraction grating. Further, limiting the exposure area on the mask and the wafer as described above also causes a decrease in throughput. Therefore, in this method, it is desirable to take measures such as using a light source having a sufficiently strong intensity and using a resist material such as a chemically amplified resist having high sensitivity.

Third, as shown in the above description, in addition to the desired diffraction image of f "= 0 on the pupil, f" = ± 2 (SA
The Fourier transform image shifted by + SB) is generated. This means that the high-order spectrum of the mask pattern overlaps a substantially low spatial frequency region, which is generally not preferable. In order to avoid this in the optical system of FIG. 1, it is sufficient to set PA ≦ 1 / (1-2 · NA / M). In this case, the mask has a diffraction angle of 2 · NA / M
Diffraction grating A for the diffracted light (R1 in FIG. 1) diffracted by
This is because there is no diffracted light in the + 1st order direction (corresponding to the dotted line emitted from A1 in FIG. 1).

Fourth, in the optical system of the present invention, it is necessary to pay attention to the aberration associated with the introduction of the diffraction grating. The aberration generated by the diffraction grating will be described with reference to FIG. It is assumed that the light ray after passing through the mask is in a plane including the optical axis and the periodic direction of the diffraction grating (for example, one-dimensional pattern and coherent illumination). In order for the optical system of FIG. 6a to be aplanatic, for example, OX ₁ X ₂ X ₃ I, OY ₁ Y ₂ Y ₃ I, and OZ ₁ Z ₂ Z ₃ I.
The difference between the optical path lengths of the two must be zero. However, if there is a difference in optical path length between them, this becomes an aberration. Assuming that the projection optical system is an ideal optical system with no aberration, X ₂ X ₃ = Y ₂ Y ₃ = Z ₂ Z _{3, and} thus OX ₁ X ₂ + X ₃ I,
The difference between OY ₁ Y ₂ + Y ₃ I and OZ ₁ Z ₂ + Z ₃ I is the aberration. OX ₁ X ₂ X ₃ I to OZ ₁ Z ₂ Z ₃ I across the diameter of the pupil
6b is plotted with respect to the pupil radius coordinate s standardized with OY ₁ Y ₂ Y ₃ I as the reference, the solid line in FIG. It can be seen that the aberration w + (s) for a ray having an angle of + with respect to the optical axis after passing through the mask is generally asymmetric on the pupil. Similarly, the aberration w- (s) with respect to a ray having an angle of − with respect to the optical axis is symmetric with respect to w + (s) and the pupil due to the symmetry of the optical system. In the present invention,
Since it is necessary to cause the light diffracted in the + direction and the light diffracted in the − direction to interfere with each other on the wafer at the same time, it is necessary to simultaneously correct aberrations for both. However, as can be seen from FIG. 6b, since the on-pupil aberrations for the light diffracted in the + direction and the − direction do not match, it is theoretically difficult to correct them at the same time by the projection optical system. Therefore, it is preferable to correct these aberrations between the mask and the projection optical system or between the wafer and the substrate. This can generally be done in the following way.

If w + (s) and w- (s) are equal, this can be corrected by the projection optical system. Therefore, Δw
(s) = {w + (s)}-{w- (s)} is a sufficiently small amount δ compared with the wavelength on the pupil (in the range of −1 ≦ s ≦ 1 in FIG. 6).
It should be suppressed to. On the other hand, Δw ± (s) is the installation position and period of each diffraction grating, the thickness and refractive index of the substrate supporting the diffraction grating,
Parameters such as the relative positional relationship between the substrate and the diffraction grating xi (i
= 1, 2, ...). Therefore, the problem is reduced to finding xi satisfying Δw (s, xi) <δ in the range of −1 ≦ s ≦ 1. An example of actual optimization will be described in Examples. In any case, if the aberration with respect to the ray having an angle of ± with respect to the optical axis after passing through the mask is made symmetrical on the pupil in this way, this can be corrected in the projection optical system. It is more preferable if the aberration itself can be sufficiently suppressed by the method described above.

As described above, 1 is used as the mask pattern for simplicity.
Although a two-dimensional pattern exists in reality, or when partial coherent illumination is used, the light beam after passing through the mask is not confined within the plane including the optical axis and the periodic direction of the diffraction grating. Instead, head to various points on your eyes. In this case, the function Δw (s, t) = {w + (s, t)}-{w- (s, t)} of the two-dimensional coordinates (s, t) on the pupil is considered as Δw, and And xi satisfying Δw (s, t, xi) <δ
You should ask. This means that w ± (s, t) is s = 0 on the pupil.
It means to be as symmetrical as possible with respect to.

Further, in order to obtain the effect of the present invention in all directions, it is conceivable that each diffraction grating is a two-dimensional diffraction grating as shown in FIGS. 7A and 7B, for example. In this case, the apparent pupil shape is 4-fold symmetry. However, due to the above-mentioned circumstances, it is rather difficult to simultaneously perform aberration correction on two pupils that are perpendicular to each other, except when the NA of the optical system is small. For this reason, it is rather difficult to obtain the effect of the present invention on the mask equally in all directions, and it is more realistic to use the one-dimensional diffraction grating as shown in FIG. 8a, b and c show three typical diffraction gratings and an apparent pupil shape. In the case of FIG. 8a,
The substantial NA increases nearly twice for patterns in the x direction, but decreases for patterns in the y direction. Figure 8b
, The effective NA for the pattern in the x direction is √2
It is doubled and becomes 1 / √2 for the pattern in the y direction. In the case of FIG. 8c, the NA is √2 in both the x and y directions, but it is considered that the imaging performance in the directions other than the x and y directions remarkably depends on the pattern direction. In either case, it is desirable to impose restrictions on the pattern layout rule and the like on the mask depending on the direction.

In order to eliminate the pattern direction dependence of the imaging performance, multiple exposure may be carried out by rotating the conditions of FIGS. 8a, 8b and 8c by 90 degrees, for example. In particular, when this is applied to FIG. 8c, it is equivalent to suppressing the pattern direction dependence in directions other than the x and y directions and multiplying NA by √2 in both the x and y directions without sacrificing the image contrast. You can get a statue. However, when the diffraction grating is rotated by 90 degrees, the aberration characteristic is also rotated by 90 degrees. Therefore, aberration correction is performed using a pupil filter, and this is performed together with the diffraction grating.
It is desirable to take measures such as rotating once. If it is difficult to suppress aberrations, a slit filter may be provided in the pupil if necessary.

When the periodic type phase shift mask is completely coherently illuminated as shown in FIG. 3, the optical paths of the ± first-order light beams that interfere in the vicinity of the wafer are always symmetrical with respect to the optical axis, and the optical path lengths of the respective light paths are long. Are equal. Therefore, a fine pattern can be formed even if the optical system is not corrected for aberration. That is,
When the periodic phase shift mask is used under the complete coherent illumination, the two-dimensional diffraction grating as shown in FIG. 7 can be used, and the effect of the phase shift mask can be maximized regardless of the pattern direction. it can. When a mask pattern in which various patterns are mixed is transferred, only the fine periodic pattern is exposed by the above method, and then the other portions are exposed by the conventional exposure method.

Further, the above-mentioned aberration generally increases rapidly with the value of NA. For this reason, there is relatively no problem in an optical system having an NA of about 0.1 to 0.2. Therefore, when applied to a large-area exposure apparatus with a low NA and a low magnification, a reflection type soft X-ray reduction projection exposure apparatus, and the like, various restrictions as described above are alleviated.

As described above, according to the present invention, it can be said that the left and right sides of the Fourier diffracted image centering on the 0th-order diffracted light beam are passed through the pupils separately and are combined on the image side. This idea itself has already been applied to an optical microscope as discussed in the above-mentioned document, but an optical system configuration that can realize this on a reduction projection optical system has been devised so far. There wasn't. The present invention is nothing but the realization of this in a reduction projection exposure system. That is, in the optical system of FIG. 1, a diffraction grating is provided between the projection optical system and the wafer,
The optical system is configured so that the incident angle of the light beam incident on the wafer surface is increased and an image faithful to the original mask pattern is reproduced as a result of wafer surface interference.
The present invention can be applied to various projection optical systems such as a refractive optical system, a reflective optical system, and combinations thereof, a reduction optical system, and a unity magnification optical system. The exposure method for exposing the mask pattern onto the wafer using these optical systems can be applied to any of batch transfer, scanning method, step-and-repeat, step-and-scan and the like. Also, as is apparent from the above description, the present invention is based on a purely geometrical optical effect. Therefore, there is no problem caused by the use of evanescent light unlike the method using the moire fringes described above. Also, since the diffraction grating can be installed away from the wafer and there is no need for synchronous scanning,
Much easier to achieve.

[0043]

【Example】

(Embodiment 1) Based on the present invention, a scanning type KrF excimer laser projection exposure apparatus having NA = 0.45, light source wavelength λ = 248 nm and reduction ratio 4: 1 was modified as schematically shown in FIG. That is, a transparent quartz plate 103 having a phase grating pattern on both sides was inserted between the mask 101 placed on the mask stage 100 and the projection optical system 102. or,
A transparent quartz plate 1 having a light-shielding pattern on one side and a phase grating pattern on the other side between a wafer 105 placed on a wafer stage (sample stage) 104 and a projection optical system 102.
06 was inserted so that the side of the light shielding pattern faces the wafer. The light-shielding pattern is C with a width of 300 μm and a period of 1 mm.
The r pattern and the phase grating pattern are Si with a period = λ / NA
The oxide film pattern was used. The period of the phase grating pattern on the mask side transparent quartz plate 103 is four times that on the wafer side.
The Si oxide film thickness was set so that the phase of light transmitted through the portion where the film was present and the portion where the film was not present were shifted by 180 degrees. These patterns were formed using EB lithography in the same manner as in the so-called chromeless phase shift mask manufacturing process. The width of the mask is 1.2 m on the side of the illumination optical system 107.
transparent quartz plate 1 having a light-shielding pattern of m and cycle = 4 mm
08 is provided. The light-shielding region of the light-shielding pattern was set to be conjugate with the light-shielding pattern on the wafer-side transparent quartz plate 106.

Period of the phase grating on both sides of the transparent quartz plate 103,
The film thickness and the installation position of each transparent quartz plate were optimized using the optimization function of the ray tracing program so that the aberration on the pupil of the projection optical system in the meaning described in the section of action becomes axially symmetric. Further, the aberration correction filter 109 is inserted at the pupil position of the projection optical system for the above-mentioned axially symmetric aberration correction. Here, the aberration correction filter 109 mainly corrects astigmatism in a direction perpendicular to the periodic direction of the diffraction grating. The transparent quartz plate having these diffraction gratings and the aberration correction filter are both replaceable so that they can be quickly set at predetermined positions. Further, in order to accurately position the transparent quartz plate, the holder (not shown) of each quartz substrate has a fine movement mechanism (not shown), and the position of each quartz substrate is measured to obtain the desired position. Can be set to a position. Further, by monitoring the image with an autofocus monitor (not shown) provided on the wafer stage 104, the monitor result is fed back so that the optimum image forming characteristic can be obtained on the image plane. It is also possible to adjust the position. It should be noted that the projection optical system itself may perform aberration correction on the diffraction grating in advance, and in this case, an aberration correction filter is not necessary. The exposure was performed while synchronously scanning the mask and the wafer. The stage control system 110 synchronously scans the mask stage 100 and the wafer stage 104 at a speed ratio of 4: 1.

Using the above-mentioned exposure apparatus, masks having patterns of various sizes including a periodic type phase shift pattern were transferred onto a chemically amplified positive type resist. After the exposure, a predetermined development process was performed, and as a result of observation with a scanning electron beam microscope, a dimension of 90 nm (a period of 180 n was measured by a periodic phase shift mask in the periodic direction (x direction) of the phase grating.
The resist pattern of m) could be formed. On the other hand, the resolution in the direction perpendicular to the above direction (y direction) was about 140 nm (cycle 280 nm) using a phase shift mask. Then, next, when the same mask was exposed by exposing the same mask by rotating the three phase gratings and the aberration correction filter by 90 degrees, the resolutions in the x direction and the y direction were reversed.

In the above embodiment, the type of optical system, N
A, light source wavelength, reduction rate, resist, type and size of mask pattern, period of diffraction grating and light-shielding pattern, installation position, etc. are extremely limited, but these various conditions are not contrary to the gist of the present invention. Various changes can be made within the range.

(Embodiment 2) Next, an example in which the optical system is optimized so that the influence of the aberration caused by the introduction of the diffraction grating is minimized will be shown. In the optical system of FIG. 10, O and I are the mask surface and the image surface of the optical system in which the diffraction grating is introduced, Σ and Σ ′ are the mask surface and the image surface of the projection optical system in which the diffraction grating is not introduced, hi (i = 1
6) shows the distance in the figure. The diffraction gratings A, B, and C and the light-shielding pattern directly on the wafer were formed on both surfaces of the transparent quartz substrate as in Example 1. At this time, the lateral aberration w ± (s) for a ray having an angle of ± with respect to the optical axis after passing through the mask is expressed as a function of the normalized pupil radius coordinate s as follows.

W ± (s) = wu ± (s) + ws ± (s) wu ± (s) = C ₁ h ₁ + C ₂ (s ₁ ) h ₂ + C ₅ h ₅ + C ₆ h ₆ ws ± (s) = C ₃ h ₃ + C ₄ h _{4 C 1 = tan [(s ±} s 0) / M] / M, C 2 = tan [± (s 1 / n) - (s ± s 0) / (nM)] / M, C 3 = tan [s _{/ M] / M, C 4} = tan (s), C 5 = tan [(s ± s 0) / n], C 6 = tan (s ± s 0) where, wu is s = 0 on the pupil Component asymmetric with respect to w
s represents a symmetrical component. However, s ₀ = NA, s ₁ = λ / P
A. Assuming that s ₀ (NA), the reduction ratio M, and the refractive index n of the transparent quartz substrate are system-specific values, the above equation includes seven optimization parameters, hi (i = 1 to 6) and s ₁ . Therefore, these values were optimized by imposing seven constraint conditions on wu ± (s) and ws ± (s) in order to minimize the aberration. Table 1 shows an example of optimization results for some NAs. However, the aberration is represented by a wavefront aberration in units of h ₅ / λ.

[0049]

[Table 1]

As can be seen from the table, it was possible to sufficiently suppress the aberration even at NA = 0.4. Similar optimization can be performed for various arrangements such as when the diffraction gratings A and B are provided on different transparent substrates. Furthermore, by introducing a new transparent substrate or a new diffraction grating to increase the optimization parameters, it is possible to satisfy more severe aberration conditions.

(Embodiment 3) Next, an example in which a DRAM having a design rule of 0.1 μm is prepared by using the exposure apparatus shown in Embodiment 1 will be described. FIG. 11 shows the manufacturing process of the above-mentioned device focusing on the exposure process.

First, S having wells (not shown) formed therein
Isolation 202 and gate 2 on i substrate 201
03 was formed (Fig. 11a). The isolation and gate pattern were exposed by the exposure apparatus shown in Example 1 using a periodic type phase shift mask. Here, since it was predicted by simulation that a part in which the pattern shape was distorted was generated in the peripheral part of the periodic pattern, a mask for removing this unnecessary part was prepared. The above-mentioned mask was subjected to overexposure on the same resist film as that subjected to the above-mentioned exposure using a conventional exposure apparatus, and then developed to remove portions unfavorable for circuit performance. It should be noted that it is possible to deal with it by ignoring it in a circuit without removing the unnecessary portion.

Next, a capacitor 204 and a contact hole 205 were formed (FIG. 11b). The electron beam direct writing method was used for pattern exposure of the contact holes. Next, a first layer wiring 206, a through hole (not shown) and a second layer wiring 207 were formed (FIG. 11c). The first layer wiring (0.1 μmL / S) was exposed using the periodic phase shift mask and the exposure apparatus described in the first embodiment. However, here, the directions and dimensions of the diffraction gratings were changed to those shown in FIG. 9c, and the diffraction gratings were rotated by 90 degrees to perform multiple exposure.
At this time, the aberration correction filter 109 was also rotated 90 degrees together with the diffraction grating. As a result, 0.1 μmL /
S could be formed. Like the contact holes, the through holes were formed by the electron beam direct writing method. Subsequent multilayer wiring patterns and final passivation patterns are designed according to the rule of 0.2 μm, and were formed by a normal KrF excimer laser projection exposure method not using the present invention. The device structure, materials, etc. can be changed without being limited to those used in the above-mentioned embodiments.

(Embodiment 4) Next, as another embodiment of the present invention, an example in which the present invention is applied to manufacture of a distributed feedback (DFB) laser will be described. The exposure device has an NA of 0.5.
The ArF excimer laser reduction projection exposure apparatus of 1 was modified in the same manner as in Example 1 and used. In the manufacturing process of the conventional 1/4 wavelength shift DFB laser, a diffraction grating having a period of 140 nm, which was formed by using an electron beam drawing method or the like, was formed by using the periodic phase shift mask and the above-mentioned exposure apparatus. As a result, it becomes possible to manufacture a DFB laser having substantially the same performance as that manufactured by using the electron beam drawing method or the like in a shorter period of time.

[0055]

As described above, according to the present invention, when the mask is irradiated with light through the illumination optical system and the mask pattern is imaged on the substrate by the projection optical system to form the pattern, the substrate and A diffraction grating is provided in parallel with the substrate between the projection optical system, and the mask pattern image is reproduced between the projection optical system and the mask so that the image of the mask pattern is reproduced near the substrate surface due to the interference of the light diffracted by the diffraction grating. Alternatively, by providing a diffraction grating or an imaging optical system between the mask and the illumination optical system, it becomes possible to form a fine pattern that exceeds the resolution limit of the conventional exposure apparatus. Specifically, without changing the NA of the projection optical system, an effect substantially equivalent to doubling the NA can be obtained. As a result, a large exposure field and high resolving power can be obtained without significantly changing the basic configuration of the optical system of the conventional exposure apparatus, and the size can be reduced to 0.1 μm by using the reduced projection optical lithography suitable for mass production.
It is possible to manufacture m-class LSIs.

[0056]

[Brief description of drawings]

FIG. 1 is a schematic diagram geometrically showing the principle of image formation of one optical system according to the present invention.

FIG. 2 is a schematic diagram showing the principle of image formation by various conventional exposure methods.

FIG. 3 is a schematic diagram showing the principle of image formation when a phase shift mask or an oblique illumination method is applied to one optical system according to the present invention.

FIG. 4 is a schematic diagram showing the principle of image formation of one optical system according to the present invention in a diffractive optical manner.

FIG. 5 is a schematic view showing an example of a part of an optical system and an exposure method according to the present invention.

FIG. 6 is a schematic diagram showing characteristics of one optical system according to the present invention.

FIG. 7 is a schematic diagram showing an optical component used in the present invention and an effect obtained by the optical component.

FIG. 8 is a schematic diagram showing an optical component used in the present invention and an effect obtained thereby.

FIG. 9 is a schematic diagram showing a configuration of an exposure apparatus according to an embodiment of the present invention.

FIG. 10 is a diagram showing characteristics of another embodiment of the present invention.

FIG. 11 is a schematic view showing a device manufacturing process according to another embodiment of the present invention.

[Explanation of symbols]

1 ... Mask, 2 ... Projection optical system, 3 ... Pupil, 4 ... Wafer,
5, 20 ... Aperture, 6, 29 ... Optical axis, A, B, C ... Diffraction grating, R ... Light, R0, R0 '... 0th-order diffracted light, R1, R +,
R1 ″ ... + 1st order diffracted light, R1 ′, R −... −1st order diffracted light,
A0, A1 ... Points on diffraction grating A, B0, B1 ... Points on diffraction grating B, C0, C1, C1 '... Points on diffraction grating C,
Q, P, Q '... points on the image plane, 21 ... conventional transmissive mask,
22 ... Light, 23 ... Projection optical system, 24 ... Pupil, 25 ... Image plane,
26 ... Periodic phase shift mask, 27 ... 0th order light of mask diffracted light, 28 ... + 1st order light, 51 ... Image plane, 52 ... Shading mask, 53 ... Quartz substrate, O ... Point on mask, X ₁ , Y ₁ , Z
₁ ... Point on diffraction grating A, X ₂ , Y ₂ , Z ₂ ... Point on diffraction grating B, X ₃ , Y ₃ , Z ₃ ... Point on diffraction grating C, I ... Point on image plane, 100 ... mask stage, 101 ... mask, 102
... projection optical system, 103 ... transparent quartz plate, 104 ... wafer stage (sample stage), 105 ... wafer, 106 ... transparent quartz plate, 107 ... illumination optical system, 108 ... transparent quartz plate, 1
09 ... Aberration correction filter, 110 ... Stage control system,
201 ... Si substrate, 202 ... Isolation, 203
... gate, 204 ... capacitor, 205 ... contact hole, 206 ... first layer wiring, 207 ... second layer wiring.

Claims (23)

[Claims] 1. A step of preparing a mask, a step of irradiating the mask with light from a light source, a step of diffracting the pattern of the mask, a step of diffracting the diffracted light through a projection optical system, and a step of A projection exposure method comprising a step of reproducing a mask pattern and exposing. 2. The projection exposure method according to claim 1, wherein the step of diffracting is performed twice. 3. A light source, first and second diffracting means for irradiating a pattern on a mask with light from the light source, and diffracting the light from the mask, and projection for projecting the diffracted light onto a sample. A projection exposure apparatus comprising an optical system, a third diffracting means for diffracting the light from the projection optical system, and a sample table on which a sample is placed under the third diffracting means. . 4. The projection exposure apparatus according to claim 3, wherein the first and second diffracting means are phase gratings. 5. A mask is irradiated with light of wavelength λ emitted from a light source through an illumination optical system, and a pattern on the mask is imaged on a substrate by a projection optical system having a numerical aperture NA and a reduction ratio M: 1. In the method of forming a pattern on the substrate by performing the above, a first diffraction grating parallel to the substrate is provided between the substrate and the projection optical system, and the light diffracted by the first diffraction grating is included. A second diffraction grating and a third diffraction grating are arranged between the mask and the illumination optical system in parallel with the mask in order from the mask side so that an image of the mask pattern is reproduced near the substrate surface due to interference. The projection exposure method is characterized by providing two diffraction gratings. 6. The cut-off spatial frequency f of the optical system provided with the diffraction grating is higher than the cut-off spatial frequency f0 of the optical system not provided with the diffraction grating and is not more than twice the f0. Item 6. The projection exposure method according to item 5. 7. The projection exposure method according to claim 5, wherein the spatial period P1 of the first diffraction grating is in the range of λ / (1.42 · NA) ≦ P1 ≦ λ / NA. 8. The first, second, and third diffraction gratings have the same period direction, and the first diffraction grating has a spatial period P1, the second diffraction grating has a spatial period P2, and the third diffraction grating has a spatial period P2. 6. The projection exposure method according to claim 5, wherein the spatial period P3 substantially satisfies the relationship of 1 / P3 = 1 / P2-1 / (M · P1). 9. The optical distance Z1 from the substrate surface of the first diffraction grating and the optical distances Z2 and Z3 from the mask surfaces of the second and third diffraction gratings are approximately (Z3-Z2) / The projection exposure method according to claim 5, wherein the relationship of P2 = (Z3 / M + Z1 · M) / P1 is satisfied. 10. The respective installation positions of the first diffraction grating, the second diffraction grating, and the third diffraction grating, the first diffraction grating, the second diffraction grating, and the third diffraction grating. The film thickness of each transparent substrate on which the diffraction grating is provided, the period of the second diffraction grating, the mask surface and the image according to the NA and reduction magnification of the projection optical system and the positional relationship between each diffraction grating and the substrate. 6. The projection exposure method according to claim 5, wherein the aberration between the surfaces is set to be minimum. 11. The projection exposure method according to claim 5, wherein a spatial period P2 of the second diffraction grating satisfies P2 ≦ 1 / (1-2 · NA / M). 12. The projection exposure method according to claim 5, wherein the second and third diffraction gratings are phase gratings. 13. The projection exposure method according to claim 5, wherein the first diffraction grating is a phase grating. 14. Between the substrate and the first diffraction grating,
A first light-shielding pattern having a width in the one direction of Z1 · NA or less and a spatial period of approximately 2 · Z1 · NA is provided, and the first light-shielding pattern on the mask is directly above or directly below the mask.
A second light-shielding pattern that shields a region substantially conjugate with the light-shielding pattern of (1) to limit the exposure region, or scan the substrate to expose the limited exposure region, or The projection exposure method according to claim 5, wherein the exposure is performed while moving in steps. 15. The diffraction grating is a one-dimensional diffraction grating,
6. The wavefront aberration of the projection optical system is corrected so as to be line-symmetrical about the diameter of the diffraction grating on the pupil in the direction perpendicular to the periodic direction as an axis. Projection exposure method. 16. The projection exposure method according to claim 5, wherein the mask includes a periodic phase shift mask. 17. The projection exposure method according to claim 5, wherein the mask has a fine pattern in a specific direction according to the period and direction of the first diffraction grating. 18. The projection exposure method according to claim 5, wherein the mask has its pattern shape corrected in accordance with the period and direction of the first diffraction grating. 19. Between the first diffraction grating and the substrate,
The projection exposure method according to claim 5, wherein the projection optical system is filled with a liquid having a refractive index n larger than 1, and the NA of the projection optical system is set to a range of 0.5 <NA <n / 2. 20. An illumination optical system for irradiating a mask on a mask stage with light having a wavelength λ emitted from a light source, a numerical aperture NA for forming a pattern on the mask near the substrate surface on the substrate stage, and a reduction ratio M. In a projection exposure apparatus having a projection optical system of 1: 1, a first spatial period P1 (λ / (1.42 · N) parallel to the substrate is provided between the substrate and the projection optical system.
A) ≦ P1 ≦ λ / NA), and the mask so that the image of the mask pattern is reproduced near the substrate surface by the interference of the light diffracted by the first diffraction grating. Between the illumination optical system, in parallel with the mask, the second diffraction grating and the third diffraction grating are arranged in order from the mask side.
A projection exposure apparatus having a single diffraction grating. 21. The periodic directions of the first, second and third diffraction gratings are the same, and the spatial period P1 of the first diffraction grating is
21. The spatial period P2 of the second diffraction grating and the spatial period P3 of the third diffraction grating substantially satisfy the relationship of 1 / P3 = 1 / (M.P1) + 1 / P2. Projection exposure device. 22. The respective installation positions of the first diffraction grating, the second diffraction grating, and the third diffraction grating, the first diffraction grating, the second diffraction grating, and the third diffraction grating. The film thickness of each transparent substrate on which the diffraction grating is provided and the period of the second diffraction grating are determined according to the NA and reduction magnification of the projection optical system, the positional relationship between each diffraction grating and the substrate, and the mask surface and the image. 21. The projection exposure apparatus according to claim 20, wherein the aberration between the surfaces is set to be minimum. 23. Between the substrate and the first diffraction grating,
It has a light-shielding pattern whose width in the one direction is Z1 · NA or less and whose spatial period is approximately 2 · NA · Z1, or
21. The projection exposure apparatus according to claim 20, wherein the projection exposure apparatus has a function of performing exposure by scanning an exposure area limited by the light-shielding pattern on the substrate or moving the substrate in a stepwise manner.