JPH04273427A - Mask, exposure method and aligner - Google Patents

Abstract

PURPOSE:To expose and transcribe a fine isolated pattern by a method wherein one periodic pattern and another pattern which have been formed in different places on one original plate are overlapped and exposed to light and a projection-type aligner provided with a specific illumination optical system is used. CONSTITUTION:One pattern PA which is to be overlapped and exposed to light is formed in such a way that, as one example, three light-shielding patterns PA1, PA2, PA3 are formed in a line-and-space shape (at a duty of 1:1). A first exposure operation is executed to a resist layer on a wafer W. An isolated line pattern which is to be formed on the wafer W is originally the light- shielding pattern PA2. The other pattern PB which is to be overlapped and exposed to light is arranged instead of the pattern PA, and a second exposure operation is executed to the same position on the wafer W. At this time, the width Db of the pattern PB is decided to be narrower than the interval Da between the line patterns PA1, PA3 on both sides of the pattern PA and to be larger than the line width of the line pattern PA2.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for forming circuit patterns for semiconductors, and more particularly to a mask used in a lithography process, a projection exposure method using the mask, and an exposure apparatus.

0002

[Prior Art] When forming circuit patterns for semiconductors, etc.,
In the case of current miniaturized LSIs, an original plate (mask, reticle) consisting of a light-shielding part and a transparent part enlarged 5 to 10 times as large as the actual pattern is used on a semiconductor substrate (mask, reticle) using a reduction projection exposure device. A method is used in which an image is formed on a photosensitive film (resist) coated on the surface of a wafer or the like, and this is exposed to light.

[0003] Conventionally, the circuit pattern to be transferred is drawn on the reticle used, magnified by the magnification of the projection exposure device, and therefore the circuit pattern is transferred onto the wafer in the desired size. be done. Conventionally, in this type of apparatus, as shown in FIG.
A substantially circular aperture stop 100 disposed substantially conjugate with the pupil ep of L illuminates the reticle (mask) R as a luminous flux L11 passing through a circular region centered on the optical axis of the illumination optical system. Here, the solid line representing the luminous flux represents the chief ray of light emitted from one point.

At this time, the ratio of the numerical aperture of the illumination optical system to the reticle side numerical aperture of the projection optical system PL, the so-called σ value, is an aperture stop of 10.
It is determined by 0, and its value is generally about 0.3 to 0.6. The illumination light L11 is diffracted by a pattern PP patterned on the reticle R, and from the pattern PP, 0th-order diffracted light D0, +1st-order diffracted light Dr, and -1st-order diffracted light D
l occurs. Each of the diffracted lights is focused by the projection optical system PL, and generates interference fringes on the wafer (sample substrate) W. This interference fringe is an image of pattern PP.

At this time, the angle θ formed by the 0th-order diffracted light D0 and the ±1st-order diffracted lights Dr and Dl is determined by sin θ=λ/P (λ: exposure wavelength, P: pattern pitch). As the pattern pitch becomes finer, sin θ increases, and when sin θ becomes larger than the reticle-side numerical aperture (NAR) of the projection optical system, the ±1st-order diffracted lights Dr and Dl cannot enter the projection optical system PL.

At this time, only the 0th order diffraction D0 reaches the wafer W, and no interference fringes are generated. In other words, sinθ>
In the case of NAR, an image of pattern PP cannot be obtained,
It becomes impossible to transfer the pattern PP onto the wafer W. From the above, in the conventional exposure apparatus, the si
The pitch P that satisfies nθ=λ/P≒NAR is obtained by the following formula, and P≒λ/NAR

(1) This pitch P is the minimum pitch on the reticle of a pattern that can be transferred onto the wafer W.

Therefore, the minimum pattern width is about half this, about 0.5×λ/NAR, but in reality, due to the relationship with the depth of focus, the minimum pattern width is about 0.6×λ/NAR. There is. If we convert this to the image plane (wafer W) side of the projection optical meter, we get 0.6×λ/NAW (NAW is the numerical aperture of the projection optical system on the wafer side, and NAW = B・N
AR and B are the reduction ratios of the projection optical system.

This value will hereinafter be referred to as the resolution limit of the projection optical system. In order to transfer a finer pattern than the resolution limit of the projection optical system itself, a method has been proposed in which the reticle pattern itself is changed to increase the resolution. This is a method called phase shift, in which a so-called phase shifter film is formed on the reticle so that the phase of the transmitted light is shifted by π from that of other parts, and the interference effect between the phase shifter part and the transmitted light with other parts is (particularly the cancellation effect) to increase resolution and depth of focus.

[0009]

[Problems to be Solved by the Invention] In a conventional projection exposure apparatus, the fineness (size) of a circuit pattern that can be transferred onto a wafer is determined by setting the exposure wavelength to λ and the wafer-side numerical aperture of the projection optical system to NAW. , about 0.6×λ/NAW was the limit. This is due to a diffraction phenomenon that occurs because light is a wave, and therefore, in principle, resolution can be improved by making the exposure wavelength shorter.

However, when the wavelength is shorter than 200 nm, there are many problems such as there is no suitable optical material that transmits the wavelength and absorption by air occurs. Also, the numerical aperture N
AW is currently already at its technical limit, and no further large N.
The situation is such that it cannot be expected to become A. In addition, although it is possible to improve resolution and increase the depth of focus with exposure using the phase shift method, the manufacturing process of the reticle with a phase shifter used in the phase shift method is complicated, and therefore the incidence of defects is high. , and the manufacturing cost is also extremely high. In addition, there are many problems such as inspection methods and correction methods that have not yet been established, and many problems remain for practical use.

In the current LSI manufacturing process, film formation,
Photolithography (circuit pattern transfer) and etching cycles are usually repeated about 20 times. In addition, the thickness of each film is about 0.05 μm to 1 μm, so as the process progresses, a step of about several μm is formed on the wafer, and in order to pattern the upper and lower parts simultaneously, it is necessary to A projection exposure method is required.

On the other hand, a multi-focus exposure method is also known as one method for expanding the depth of focus. In the depth increasing method using this exposure method, a pattern that exists alone (the size of the pattern other than the pattern is 1
:3 or more, hereinafter referred to as isolated patterns), a larger practical depth of focus can be obtained than with normal exposure methods, but it is theoretically impossible to improve the resolution. .

The present invention has been made in view of the current situation, and provides an exposure method that uses a reticle (mask) consisting of a light-shielding part and a transmitting part as in the past, and that can obtain high resolution and a deep depth of focus, and The purpose is to provide equipment.

[0014]

[Means for Solving the Problems] To achieve the above object, the present invention provides an illumination light amount distribution for illuminating the reticle pattern within a plane of the illumination optical system which becomes the Fourier transform plane of the reticle pattern. A projection type exposure apparatus is used in which the illumination is concentrated in one or more arbitrary areas each centered on one or more arbitrary points other than the illumination optical axis.

[0015] Furthermore, the exposure of the photosensitive film formed on the surface of the substrate to be exposed (semiconductor wafer, etc.) is performed by dividing into a plurality of sub-exposures, and in each of the plurality of sub-exposures, the reticle is The positional relationship between the exposure target substrate and the exposure target substrate is such that they are at relatively different positions in a plane perpendicular to the optical axis of the projection optical system. At this time, the relative positional relationship between the reticle and the wafer is such that at least one area on the wafer is exposed by overlapping different patterns on the reticle in all of the plurality of sub-exposures. did.

[0016] At this time, at least one of the plurality of patterns subjected to overlapping exposure is made up of a group of patterns arranged approximately periodically. Or,
At least two of the plurality of patterns to be overlaid and exposed are made up of a group of substantially periodically arranged patterns, and the periodicity or directionality of these periodic patterns is different from each other.

Furthermore, the pattern group forming the periodic pattern is fine enough to be less than the resolution limit of the projection optical system, and the other patterns are patterns or patterns that are twice as large or more than the resolution limit. It consists of a group. Furthermore, in each of the plurality of sub-exposures,
The relative positional relationship between the reticle and the object to be exposed (such as a semiconductor wafer) is changed by moving a stage that holds the object to be exposed relative to the optical axis of the projection optical system.

[0018]

[Operation] In the plane of the illumination optical system that has a Fourier transform relationship with the reticle pattern used in the present invention, the illumination light beam that illuminates the reticle has one or more points centered on one or more points other than the optical axis. The use of area-focused projection exposure devices allows for increased resolution and depth of focus, especially for periodic reticle patterns.

This principle will be explained below with reference to the drawings. In conventional projection exposure equipment, exposure light that enters the reticle from above at various angles of incidence is used indiscriminately, and the 0th, ±1st, ±2nd, and
Each diffracted light beam of ... was transmitted through the projection optical system almost without limit and formed an image on the wafer. On the other hand, in the projection exposure apparatus of the present invention, as shown in FIG. 16, from the illumination light L10, the optical axis AX of the illumination optical system is A light shielding plate 100' having a transmitting part other than the central part through which the light passes is provided to prevent light fluxes that do not pass along the optical axis of the illumination optical system, that is, any two light fluxes that are obliquely incident on the reticle pattern in a specific direction and angle. L
12, L13, or 2n (n is a natural number) light beams are formed to illuminate the reticle pattern PP. The reticle pattern PP usually includes many patterns having a periodic structure in which light transmitting parts and light shielding parts are repeatedly formed at a predetermined pitch. And 0 generated in reticle pattern PP
The second-order diffracted light and the ±1st-order diffracted light are imaged onto the wafer W via the projection optical system PL.

That is, by making 2 or 2n beams incident on the pattern PP in a predetermined direction and angle depending on the fineness of the reticle pattern PP, optimal 0th-order diffracted light and ±1st-order light are generated. This makes it possible to image a pattern on the wafer W with high contrast and a large depth of focus, which could not be resolved sufficiently in the past. Here, the luminous flux L1 incident on the reticle R
2 and L13 are converted by the aforementioned light shielding plate 100' into two light beams having principal rays that are incident symmetrically and inclined at a predetermined angle ψ with respect to the optical axis AX of the projection optical system PL. Here again, the solid line representing the luminous flux represents the chief ray of light emitted from one point. First, based on FIG. 16, the illumination light L1
The diffracted light by 2 will be explained. The illumination light L12 is diffracted by the reticle pattern PP and becomes 0th order diffracted light D0,+
A first-order diffracted light Dr and a -first-order diffracted light Dl are generated. However, since the illumination light L12 is incident on the reticle pattern PP at an angle ψ with respect to the optical axis AX of the projection optical system PL, the 0th order diffracted light D0 is also incident on the optical axis A of the projection optical system PL.
It moves in a direction tilted by an angle ψ with respect to X. here,
The projection optical system PL is a double-sided telecentric system.

Therefore, the +1st-order light Dr travels in the direction of θP +ψ with respect to the optical axis AX, and the -1st-order diffracted light Dl travels in the direction of θP +ψ with respect to the optical axis AX.
Proceeds in the direction of θm −ψ with respect to X. At this time θP
, θm are sin(θP +ψ)−sinψ=λ/P, respectively.
...(2)si
n(θm +ψ)−sinψ=λ/P
...(3).

Assume that both the +1st-order diffracted light Dr and the -1st-order diffracted light Dl are now incident on the projection optical system PL. When the diffraction angle increases with the miniaturization of the reticle pattern PP, the +1st-order diffracted light Dr traveling in the direction of θP +ψ becomes unable to pass through the projection optical system PL (sin(θP +
ψ)>NAR). However, the illumination light L12 is on the optical axis A.
Since it is incident at an angle with respect to X, even at this diffraction angle, the −1st-order diffracted light Dl can pass through the projection optical system PL (sin(θm +ψ)<NAR).

Therefore, on the wafer W, the 0th order diffracted light D0
Interference fringes are generated by the two beams of the -1st-order diffracted light Dl and -1st-order diffracted light Dl. This interference fringe is an image of the reticle pattern PP, and when the reticle pattern PP is 1:1 line and space, it is about 9
The contrast is 0%, and it becomes possible to pattern the pattern PP on the dead wafer W whose surface is coated with resist. The resolution limit at this time is sin (θm + ψ) = NAR
…
...(4) Therefore, NAR + sinψ = λ/P P = λ/(NAR + sinψ)
...(
5) is the pitch of the minimum pattern that can be transferred on the reticle.

If sin ψ is set to about 0.5×NAR, the minimum pitch of the pattern on the reticle that can be transferred is P=λ/(NAR +0.5NAR)=2λ/3NA
R...(6). On the other hand, as shown in FIG. 15, in the case of a conventional exposure apparatus in which the illumination light is a luminous flux passing within a circular area centered on the optical axis AX of the projection optical system PL, the resolution limit is shown in equation (1). Tayo P
≒λ/NAR. Therefore, higher resolution than conventional exposure apparatuses can be achieved.

Considering the illumination light L13 in the same way, +
The first-order light Dr1 travels in the direction of θP1-ψ with respect to the optical axis AX, and the -1st-order diffracted light Dl1 travels in the direction of θ with respect to the optical axis AX.
Proceeds in the direction of m1+ψ. D01 represents the 0th order diffracted light. At this time, θP1 and θm1 are each sin(θm1+ψ)−sinψ=λ/P
...(7)si
n(θP1-ψ)+sinψ=λ/P
...(8),
The resolution limit is when sin(θP1-ψ)=NAR.

Therefore, as in the case of illumination light L12, (5
) is the minimum pitch of the pattern that can be transferred. By using both the illumination lights L12 and L13, the center of gravity of the light quantity is prevented from being biased with respect to the optical axis of the projection optical system. Next, exposure light is incident on the reticle pattern at a specific incident direction and angle, and the 0th-order diffracted light and 1st
The reason why the depth of focus increases by forming an imaged pattern on the wafer using the second-order diffracted light will be explained.

When the wafer is aligned with the focal position of the projection optical system, each diffracted light that leaves one point on the reticle and reaches one point on the wafer passes through which part of the projection optical system. They all have the same optical path length. Therefore, even when the 0th order diffracted light passes through approximately the center of the pupil plane of the projection optical system as in the conventional case, the 0th order diffracted light and other diffracted lights have the same optical path length and their mutual wavefront aberration is 0. However, if the wafer does not match the focal position of the projection optical system (defocus occurs), the optical path length of the high-order diffracted light that is incident obliquely becomes the focal point relative to the 0th-order diffracted light that passes near the optical axis. It is shorter in the front (away from the projection optical system) and longer in the rear of the focal point (closer to the projection optical system), and the difference therebetween corresponds to the difference in incidence angle. Therefore, each of the 0th-order, 1st-order, . . . diffracted lights mutually forms a wavefront aberration, causing blurring of the imaged pattern before and after the focal position. This wavefront aberration ΔW
is expressed by the following formula, ΔW=1/2×(NA)2 Δf Δf: Defocus amount NA: Distance from the center on the pupil plane expressed in numerical aperture. Therefore, approximately the center of the pupil plane is The penetrating 0th order diffracted light (
ΔW=0), the circumference of the pupil plane, radius r1 ([NA
]), ΔW=1/2
It has a wavefront aberration of ×r1 2 Δf, which lowers the resolution before and after the focal position, that is, the depth of focus.

On the other hand, in the case of the projection exposure apparatus used in the present invention, for example, as shown in FIG. 16, when two light beams are incident on the reticle pattern PP at an incident angle of θm = 2ψ,
0th-order diffracted light D0 and 1st-order diffracted light Dl from pattern PP
and pass through a position on the pupil plane ep that is approximately equidistant from the center (both have radius r2), and the 0th-order diffracted light D0
The wavefront aberrations of the first-order diffracted light Dl and the first-order diffracted light Dl are equal, and ΔW=1/
2×r2 2 Δf, the aberration of the 1st-order diffracted light relative to the 0th-order diffracted light becomes zero, and within a certain range, blur due to defocus disappears. Therefore, the depth of focus becomes larger by that much than the conventional device.

As described above, with the projection exposure apparatus used in the present invention, it is possible to increase the resolution and depth of focus, especially for periodic patterns. In the present invention, a plurality of different patterns on a reticle are overlapped and exposed on the same portion of a wafer to form one circuit pattern. Therefore, a pattern formed on a substrate to be exposed such as a semiconductor wafer is a composite image of a plurality of reticle patterns. This composite image is created by the logical sum of these multiple reticle patterns, assuming that each of the multiple reticle patterns is set to a "0" part (approximately light-shielding part) and a "1" part (generally transmitting part) according to their transmittances. is formed as. In other words, only the portion where all the reticle patterns are "0" portions (approximately light-shielding portions) is different from the other portions. If the photosensitive film (photoresist) used is a positive type (the light irradiated area is removed), the resist will remain only in the part where all the reticle patterns to be overlapped are "0" parts, and the resist will be removed in other parts. A so-called isolated line pattern or isolated island pattern is formed.

If the photoresist used is a negative type (a film remains in the light irradiated area), a so-called isolated space pattern or isolated hole pattern is formed. At this time, among the multiple reticle patterns that are superimposed,
At least one pattern has a substantially periodic arrangement, and the illumination optical system of the projection exposure apparatus used is
As mentioned above, it is assumed that the resolution and depth of focus of this approximately periodic arrangement pattern are adjusted to be maximum. Also, among the circuit patterns to be formed by superposition, particularly important ones are patterns consisting of periodic arrays, or by combining multiple patterns consisting of periodic arrays, it is possible to create a substantial focal point of the composite image. An increase in depth becomes possible.

Alternatively, even if the same projection optical system is used, by utilizing the property that the depth of focus becomes deeper as the line width of the pattern becomes thicker, the line width of patterns other than the pattern consisting of the periodic arrangement described above can be adjusted using the projection optical system. If it is more than twice the resolution limit of the system, all of the reticle patterns to be superimposed,
It becomes possible to have a sufficiently large depth of focus of the projected image.

[0032]

[Embodiment] FIG. 1 shows an embodiment of a projection exposure apparatus used in the present invention, in which illumination light (i-line or g-line) emitted by a mercury lamp 1 is focused by an elliptical mirror 2 to a second focal point. be done. A shutter 3 for blocking and releasing illumination light is arranged at the second focal point and is driven by a motor 4. The illumination light that has passed through the shutter 3 is made into a substantially parallel beam by a relay lens (collimator lens) 5 and enters a fly's eye lens 6. Each light beam from a secondary light source (a collection of point light source images) formed at each exit end of each element lens of the fly-eye lens 6 is
After being condensed by the relay lens 7 and reflected by the mirror M1, it is superimposed on the incident part 8A of the light splitter 8. Therefore, the intensity distribution of the illumination light beam is approximately uniform at the position of the incident portion 8A. In FIG. 1, only the light beam from one point light source on the optical axis among the point light source images appearing on the exit side of the fly's eye lens 6 is representatively illustrated. Now, the light splitter 8 splits the illumination light from the input section 8A into a plurality of output sections 8B and 8C with approximately equal intensity. The cross-sectional areas of the injection parts 8B and 8C are both equal, and their centers are eccentric from the optical axis AX by a predetermined amount. The detailed structure of this light splitter 8 will be described later. Incidentally, the exit sections 8B and 8C are arranged in a plane 9 that is conjugate with the exit pupil ep (or entrance pupil) of the projection lens PL, that is, in a Fourier transform plane, in order to perform projection exposure using Koehler illumination.

The diverging light flux emitted from one point on the emission sections 8B and 8C is turned into a substantially parallel light flux by the first relay lens (collimator lens) 10 and sent to the reticle blind ARB.
illuminate with uniform illuminance. In FIG. 1, for the sake of simplicity, only the traveling state of the light beams diverging from one point on the emission parts 8B and 8C is shown. The reticle blind ARB limits the illumination area on the reticle R by arbitrarily adjusting the edge positions of the four sides using the drive unit 11. In the case of Fig. 1, the illumination light flux passing through the rectangular aperture determined by the reticle blind ARB basically includes a plurality of parallel beams (corresponding to the number of exit sections 8B, 8C...) that are symmetrically inclined with respect to the optical axis AX. Only the light beam is included, and there is no parallel light beam traveling parallel to the optical axis AX. These parallel light beams are condensed by the second relay lens 12, reflected by the mirror M2, and then
The light enters the main condenser lens 13 and becomes a plurality of parallel light beams having different incident angles, and irradiates the reticle R. At this time, the pattern surface of the reticle R (the surface facing the projection lens PL) is connected to the reticle blind AR with respect to the composite system of the main condenser lens 13 and the second relay lens 12.
Since it is conjugate with B, a rectangular aperture image of the reticle blind ARB is projected onto the reticle R.

The illumination light passing through the transparent portion of the reticle R is projected onto the wafer W via the projection lens PL, and a pattern image of the reticle R is formed on the photoresist layer. Here, the broken line shown in the optical path from the reticle blind ARB to the wafer W in FIG. 1 is a ray that means pupil conjugate. As mentioned above, in this embodiment, by illuminating the reticle R with at least two symmetrically inclined light beams, the resolving power and depth of focus are enhanced even with a normal reticle pattern configuration. A light splitter 8 is newly provided to achieve this function.

The detailed behavior of the illumination light flux will now be explained with reference to FIG. Figure 2 shows the light splitter 8 in Figure 1.
This is a schematic representation of the system from the two emitting sections 8B and 8C to the reticle R, and particularly shows how the illumination light flux travels on the emitting section 8B side. First, it is assumed that the center of the exit end of the exit section 8B is eccentric from the optical axis AX by ΔH. The state in which each of the light beams generated from points P1 and P2 at both ends of the exit end travels to the reticle R is shown by a solid line. Further, dashed lines L1 and L2 shown in the optical path in FIG. 2 represent one ray of light flux from points P1 and P2, respectively, and are assumed to be parallel to the optical axis AX in the pupil space. As is clear from FIG. 2, the points P1 and P2 are imaged as points P1' and P2' on the plane EP' in the pupil space between the second relay lens 12 and the main condenser lens 13. Therefore, plane EP' is conjugate with plane 9 and is likewise a Fourier transform plane. Now, the light flux from point P1 (point P1') becomes an almost parallel light flux on the reticle R and irradiates within a certain area, and the light flux from point P2 (point P2') also becomes an almost parallel light flux and irradiates within a certain area. to illuminate. At this time, the incidence angles of the two parallel light beams reaching the reticle R are shifted by an angle Δθ. This angle Δθ is determined depending on the deviation between points P1 and P2 within the plane 9, that is, the size (or cross-sectional area) of the emission part 8B as an effective light source, and the cross-sectional area of the emission part 8B is If the angle becomes infinitely small, the angle Δθ will also become infinitely small. However, in reality, the injection part 8B
The light source image formed at the exit end of the light source has a certain size for practical purposes (however, it is sufficiently smaller than the pupil diameter), and the angle Δθ
It is difficult to make it completely zero, and it is not necessary to make it completely zero. Furthermore, since the end surface (light source image) of the emission section 8B has a two-dimensional size, the angle Δθ exists not only in the direction within the plane of the paper in FIG. 2 but also in a direction perpendicular to the plane of the paper. As described above with reference to FIG. 16, the center line of the light beam from the emission part 8B is inclined at an angle ψ with respect to the optical axis AX, and this angle ψ is related to the deviation amount ΔH, and ΔH=m· sinψ(
However, m is a constant). Further, as an example, the behavior of the illumination light flux from another emission part 8C arranged symmetrically with respect to the optical axis AX is such that the light flux from the emission part 8B is
It looks like it's folded back around X.

The configuration of the apparatus will be explained with reference to FIG. 1 again. Wafer W is placed on wafer stage WST that moves two-dimensionally and is driven by motor 20. The coordinate position of wafer stage WST is sequentially monitored by an interferometer 21 using a laser beam to monitor the relative distance between a movable mirror on stage WST and a fixed mirror attached to the lower end of projection lens PL.

On the other hand, the reticle R is also placed on a reticle stage RST that moves slightly in two dimensions, and is driven by a motor or the like (not shown). Alignment of the reticle R with respect to the optical axis AX is performed by the reticle alignment system 22, and direct alignment (die-by-die alignment) between the reticle R and the wafer W is performed using TTR (through-the-die alignment).
(reticle) alignment system 23. Furthermore, individual alignment of the wafer W is performed using TTL (Through Transmission Transmission), which detects marks on the wafer W through the projection lens PL.
The lens) alignment system 24 or projection lens P
OFF-AXIS (
(off-axis) alignment system 25. Alignment signals from each alignment system 22, 23, 24, 25 are collectively processed by a control device 30 and are sent to a wafer stage WST or a reticle stage RST.
Used for precise positioning. This control device 30 includes a motor 4 for opening/closing the shutter 3 and an injection section 8 of the light splitter 8.
B, a motor (not shown) that adjusts the positions of 8C, a drive unit 11 of the reticle blind ARB, and a wafer stage W.
It also controls the motor 20 for driving the ST. Although not shown in FIG. 1, a photometric element is provided at an appropriate position in the optical path of the illumination system from the mercury lamp 1 to the main condenser lens 13 to detect the amount of illumination light (exposure) to the reticle R. By combining this with an integrating circuit, automatic exposure control is performed to close the shutter 3 at an appropriate exposure amount.

Next, referring to FIGS. 3 and 4, the light splitter 8
The detailed configuration will be explained below. FIG. 3 shows the light splitter 8 viewed from the side, and FIG. 4 is a diagram of FIG. 3 viewed from the direction of the optical axis AX. The light beam from the incident part 8A is divided into four parts by the optical fiber 80, and each part has four exit parts 8.
Guided to B, 8C, 8D, and 8E. four injection parts 8B,
8C, 8D, and 8E each represent the pupil image range 9 on the pupil conjugate plane 9.
Variable length support rods 81B, 81C, 8 so as to be located within A
1D, supported by 81E. The four support rods 81B to 81E are movable members 82B, 82C, 82D, and 82E that are movable in the circumferential direction along the annular guide 83.
are combined with each other. The four support rods 81B to 81E are
Each of the emitting sections 8B to 8E is moved in the radial direction with respect to the optical axis AX. These support rods 81B to 81E and movable member 82B
-82E are driven by motors. Incidentally, the incidence section 8
The metal fitting that holds the optical fiber A is configured to be able to freely rotate around the optical axis AX, and has four injection parts 8.
The stress on the optical fiber 80 caused by the rotation of B to 8E is reduced.

Although the device used in this embodiment is provided with four injection parts 8B to 8E, the number is not limited to four, and may be two (in principle, one) or more. Bye. Further, in this embodiment, the illumination light beam is divided by optical fibers, but other members such as a diffraction grating or a polygonal prism may be used. Furthermore, another fly-eye lens may be added closer to the reticle R than the light splitter 8. At this time, the fly-eye lens may be a single lens or may be composed of a plurality of fly-eye lens groups.

Although the projection lens PL is here described as a refractive type, it may be a reflective type or a catadioptric type. Although the projection magnification may be arbitrary, a reduction system is desirable for exposing and transferring fine patterns. Further, although a mercury lamp is used as a light source, other light sources such as a bright line lamp or a laser light source may be used.

Next, an example of the principle of a reticle pattern and an exposure method used in the method according to the present invention will be explained with reference to FIG. FIG. 5 shows a schematic pattern example for forming an isolated line pattern (when using a positive type resist) or an isolated space pattern (when using a negative type resist) by the method according to the present invention. First, as shown in FIG. 5(A), one pattern PA to be overlaid and exposed includes, for example, three light-shielding line patterns PA1 and PA2.
, PA3 in line and space form (duty 1
:1). Then, the reticle R is positioned so that the optical axis AX passes through a specific point on this pattern PA (reticle), and the resist layer on the wafer W is exposed for the first time. In the case of FIG. 5A, the isolated line pattern that should originally be formed on the wafer W is the light-shielding pattern PA2. The apparatus shown in FIG. 1 is extremely effective in projecting a periodic pattern PA as shown in FIG. 5(A). As mentioned earlier, the apparatus shown in FIG.
By optimizing the position of 8E, even if the pattern PA becomes finer, the contrast will be higher (that is, the resolution will be higher),
It has the characteristic of increasing the depth of focus. If the original isolated line pattern PA2 to be formed on the wafer W
Even if only the line pattern PA2 is provided on the reticle R and projection exposure is performed using the apparatus shown in FIG. 1, sufficient resolution cannot be expected because only the line pattern PA2 generates almost no ±1st-order diffraction components. That is, the line patterns PA1 and PA3 on both sides shown in FIG. 5(A) act as dummy patterns for giving periodicity to the pattern PA as a whole.

Now, the exposure amount (dose amount) to the resist layer during the first exposure shown in FIG. The minimum exposure amount Ec) is set to be larger than the starting minimum exposure amount Ec). FIG. 6(A) shows the exposure dose distribution (similar to the intensity distribution) IA given to the positive resist layer by the first exposure in the cross-sectional direction. Figure 6(A)
The vertical axis represents the exposure amount, and the horizontal axis represents the position of the pattern PA in the periodic direction.

[0043] Here, a concrete numerical example will be given. now,
The line width of the line pattern PA2 to be formed on the wafer W is 0.3 μm (pitch 0.6 μm), the wavelength λ of the exposure illumination light is 365 nm (i-line), and the reduction magnification of the projection lens PL is 1/5. Then, from the theoretical formula for the diffraction angle, the incident angle ψ of the light beam generated from one exit part in the illumination optical system and illuminating the reticle R is expressed by the following formula, as explained with reference to FIG. sin2ψ=λ/5・P=0.365×10-6/(5
×0.6×10-6) ≒0.12
2 Therefore, the angle of incidence ψ is approximately 3.5°.

Next, as shown in FIG. 5B, the other pattern PB to be overlaid and exposed is placed in place of the pattern PA, and a second exposure is performed at the same position on the wafer W. At this time, the width Db of the pattern PB is set to be narrower than the interval Da between the line patterns PA1 and PA3 on both sides of the previous pattern PA, and larger than the line width of the line pattern PA2. That is, the pattern PB is shaped and sized so that only the image of the line pattern PA2 formed as a latent image on the wafer W is shielded from light, and the entire periphery thereof is a transparent portion. Therefore, in the case of a pattern shape as shown in FIG. 5, the width Db of the pattern PB is set in the range of 0.5P<Db<1.5P from the pitch P of the line pattern PA. The exposure amount distribution given to the resist layer only by this pattern PB becomes IB as shown in FIG. 6(B). Here, it is assumed that during the second exposure as well, an exposure amount larger than the minimum exposure amount Eth is given to the resist layer. When double exposure is performed on the wafer W in this way, the resist layer has the distribution IA shown in FIG. 6(A) and the distribution IA shown in FIG. 6(B).
The light amount distribution IS synthesized with the distribution IB is given as shown in FIG. 6(C). As is clear from FIG. 6(C), the image of the synthesized light intensity distribution IS (referred to as a composite image) is composed of both the periodic pattern PA and the non-periodic (isolated) pattern PB. Also, only the portion serving as a light shielding portion (line pattern PA2) is sufficiently dark, and all other portions are brighter than the minimum exposure amount Eth. Therefore, in the case of a positive resist, Fig. 6(D)
An isolated line pattern Pr in which a portion below the minimum exposure amount (threshold value) Eth remains after development as shown in
becomes. It is clear that this isolated line pattern Pr corresponds to the line pattern PA2 in FIG. 5(A).

In the above embodiment, the periodic pattern PA
It is possible to increase the resolution of periodic pattern PA by about 1.5 to 2 times compared to conventional exposure methods.
In order to leave only an isolated pattern from the transferred image (latent image),
Since double exposure is performed using another pattern PB, the resist pattern finally formed on the wafer W becomes an extremely fine isolated pattern. As mentioned above, the width Db of the non-periodic pattern PB is preferably about 2 to 3 times the width (pitch/2) of one line pattern of the periodic pattern PA. One line pattern of periodic pattern PA (PA2
) is 1/2 to 1/1.5 of the conventional resolution, the width of the aperiodic pattern PB is about twice the conventional resolution limit, and there is sufficient focus for the projection of the pattern PB. Depth is obtained.

In the case of a positive resist, the relationship between the exposure amount and the remaining film thickness is as shown in FIG. 17(A), but in the case of a negative resist, the relationship between the resist The maximum exposure amount Ec at which the resist is completely removed and the minimum exposure amount Eth at which the resist remains completely are determined. For this reason,
When the exposure shown in FIGS. 5A and 5B is performed using a negative resist, it is possible to form an isolated space pattern that is about 1.5 to 2 times finer than the conventional method.

FIGS. 7 and 8 both show examples of reticle patterns for forming an isolated island pattern (when using a positive type resist) and an isolated hole pattern (when using a negative type resist). In FIG. 7(A), a total of nine fine light-shielding patterns PAn, which are smaller than the resolution limit of the projection optical system (0.6×λ/NAR), are periodically arranged in three columns on the left and right and three rows on the top and bottom. , and a light-shielding pattern PAs, which is finer than the light-shielding pattern PAn, is arranged in the middle of the light-shielding pattern PAn. Among the periodic patterns shown in FIG. 7A, the resist pattern to be actually formed on the wafer W is the central light-shielding pattern PA2.

On the other hand, FIG. 7(B) shows the above light shielding pattern P.
It represents a light-shielding pattern PB that is about 2 to 2.5 times larger than An in both the vertical and horizontal directions. Here, each of the nine light-shielding patterns PAn is square, and the space widths in the horizontal and vertical directions are equal to the width of the patterns PAn. Further, if the interval between the light-shielding patterns PAn that sandwich the central light-shielding pattern PA2 on the left and right sides is Dax, and the interval between the light-shielding patterns PAn that sandwich the light-shielding pattern PA2 on the upper and lower sides is Day, then the horizontal width Dbx of the light-shielding pattern PB in FIG. 7(B) is Dbx<Dax, and the vertical width Dby is Dby<Day.

Similar to the case of FIG. 6, the projected image of the periodic pattern in FIG. 7A is also exposed with a higher resolution and a deeper depth of focus than before due to its periodicity. On the other hand, since the projected image of the pattern PB in FIG. 7B also has a sufficient size, it is exposed with a deep depth of focus. Therefore, an isolated island pattern PS5 (when using a positive resist) formed by a composite image by double exposure
is formed with a sufficient depth of focus as shown in FIG. 7(C). This periodic pattern PAn
The superposition of pattern PB is the center light-shielding pattern PA2 of the nine light-shielding patterns, and pattern PB.
It is assumed that the reticle R and the wafer W are aligned so that their centers coincide with each other, and multiple exposure is performed. The reason why the resist remains only in the portion (PA2) where both the periodic pattern PAn and the non-periodic pattern PB serve as a light shielding portion is the same as the example shown in FIG.

Note that the four corners of the non-periodic pattern PB are arranged so as to match the smaller periodic pattern PAs by overlapping. For this reason, when multiple reticle patterns are superimposed, all of them become light-shielding parts.
The periodic pattern PAs is sufficiently small compared to the resolution limit, and does not provide a sufficient light shielding effect on the projected image on the wafer W. As a result, no resist pattern is formed in the matching portion between the periodic pattern PAs and the non-periodic pattern PB. The periodic pattern PAs is an auxiliary pattern (corner emphasis) for making the shape of the isolated island pattern PS5 to be formed closer to a square. Therefore, even without the periodic pattern PAs, the isolated island pattern P
Formation of S5 is possible.

FIG. 8(A) shows line-and-space patterns PA1, PA2 having periodicity in the horizontal direction in the figure.
, PA3, and FIG. 8(B) shows a line-and-space pattern PB1, which has periodicity in the vertical direction in the figure.
PB2 and PB3 are shown. The hatched portions of these two patterns each represent a light shielding portion. These two periodic patterns PA and PB are both periodic patterns,
Therefore, in any pattern exposure, it is possible to obtain the high resolution and large depth of focus that are the characteristics of the present invention. A composite image (resist image) obtained by overlapping exposure of these two patterns PA and PB becomes an isolated island pattern PS6 shown in FIG. 8(C). Here, the periodic pattern PA
The distance between the left and right line patterns PA1 and PA3 is D
ax, and three line patterns PA1, PA2,
If the lengths of PA3 are both Day, then the upper and lower line patterns PB1 and PB3 of the other periodic pattern PB
The interval Dby is set as Dby>Day, and the length D of the three line patterns PB1, PB2, PB3 is
bx is set to Dbx<Dax. The resist pattern to be formed on the wafer W is a square portion formed by the intersection of the line patterns PA2 and PB2. In this overlapping exposure, multiple exposure is performed such that the center of the periodic pattern PA and the center of the periodic pattern PB coincide with each other at the same point on the wafer W. In addition, since only the intersection of the center line patterns PA2 and PB2 of each of the three light-shielding line patterns is required as a resist image, the length of each line and space ( long side dimension)
shall be shorter than three times the width (short side dimension). Alternatively, the length of each line and space pattern may be made longer than this, and a pattern equivalent to pattern PB in FIG. 7(B) may be prepared for overlapping exposure to avoid unnecessary overlapping of the light-shielding portions caused by this. (Unnecessary light-shielding parts are exposed by this.)
. This will be described in detail later as an example.

An example of forming an isolated island pattern has been described above with reference to FIGS. 7 and 8, but this is based on the premise that a positive type resist is used. If a negative resist is used, an isolated hole pattern can be formed by a similar method. In addition, in FIGS. 6, 7, and 8, three line patterns or three rows and three columns of dot patterns are given as examples of periodic patterns.
The number of lines or dots constituting the periodic pattern is not limited to this, and may be arbitrary, such as 5, 7, or 5 rows and 5 columns. Further, the line width ratio (duty) between the light-transmitting portion and the light-blocking portion of the periodic pattern is not limited to 1:1 and may be arbitrary. However, it has been experimentally found that better results can be obtained by making the line width of the light-shielding part about 20% larger than the line width of the light-transmitting part.

Furthermore, the periodicity of the periodic pattern does not need to be so strict. For example, in the case of a pattern forming a periodic pattern, that is, three lines, the two lines at both ends may be thinner than the pattern at the center. that's all,
Although the embodiments of various patterns that are overlapped and exposed (multiple exposure) according to the present invention have been described, the actual positional relationship of each reticle pattern on the reticle when these are applied,
and FIGS. 9 and 10 regarding an example of the overlapping method
Explain using. FIG. 9 shows, as an example, the previous FIGS. 5, 7,
9 is a pattern layout diagram of a reticle R in which both patterns PA and PB shown in FIG. 8 are drawn together. FIG. This reticle R is used for forming contact holes during the IC manufacturing process. In FIG. 9, the pattern formation area of the reticle R is surrounded by a light-shielding band SB,
On the outside is the mark RM1 for reticle alignment.
, RM2 are formed. In the pattern area, a pattern PA represented by black square dots and a pattern PB represented by white square dots are regularly arranged in the X and Y directions. The areas around these patterns PA and PB are all light transmitting parts. In this reticle, an isolated hole (contact hole) pattern is formed by overlapping exposure of two patterns PA and PB on a negative resist. In a memory IC, unit memory cells are generally arranged regularly (periodically) to form a memory cell group. Therefore, contact holes corresponding to each memory cell are also arranged periodically. Also in FIG. 9, pattern PA and pattern PB
The centers of are arranged with a pitch of Px in the X direction and a pitch of Py in the Y direction, and the pattern PB is
They are formed at a distance (Px/2, Py/2) from pattern PA, respectively.

Using such a reticle R, first exposure is performed on the wafer W at an appropriate exposure amount Eth or more. Therefore, each pattern PA,
The amount of light corresponding to PB is exposed. Next, the relative positions of reticle R and wafer W in the X and Y directions are determined by wafer stage WST or reticle stage RST (Px
/2, Py/2) (dimension on the reticle) and then performs the second exposure with an appropriate exposure amount Eth or more. In the case of Figure 9, move it diagonally upward to the right. In the case of the apparatus shown in FIG. 1, this movement is performed by motor 20 for driving wafer stage WST, interferometer 21, and control device 30.

As mentioned above, pattern PA and pattern P
Since the distance from B was (Px/2, Py/2),
By the first and second exposures described above, the pattern PA and the pattern PB are exposed to overlap on the wafer W. As is clear from the reticle pattern arrangement in FIG. 9, there are four relative movement directions for precisely overlapping the patterns PA and PB on the wafer W. That is, the reticle R or the wafer W can be moved in any one of the diagonally upper right, diagonally upper left, diagonally lower right, and diagonally lower left within the paper plane of FIG. The direction in which the relative movement is performed can be set arbitrarily. In addition, in the case of the pattern arrangement shown in FIG.
Each pattern PA existing in a row along one side in the direction is not double exposed with the pattern PB, and thus remains as a film. Since this remaining pattern is incomplete as a contact hole pattern, it is desirable that it not be formed as a resist image. Therefore, it is preferable to provide a margin (transparent part) of widths Px and Py between the outermost pattern PA arrangement in FIG. 9 and the light-shielding band SB. In this way, the blank area is double exposed with the latent image of the outermost pattern PA, and it is possible to prevent the formation of unnecessary remaining patterns.

When a negative resist is used as described above, the resist pattern P shown in FIG. 7(C) is formed on the wafer W.
Contact holes such as S5 or resist pattern PS6 in FIG. 8(C) have a pitch of Px/2M in the X and Y directions.
, Py/2M (where 1/M is the projection magnification). Although such a contact hole pattern has periodicity, since each hole pattern is arranged discretely, it can essentially be considered an isolated pattern. Even if a reticle pattern for hole formation is set and exposed, sufficient resolution cannot be obtained.

FIG. 10 shows the state of double exposure when the reticle pattern of FIG. 9 is composed of the pattern PA and pattern PB shown in FIG. 8. Pattern PA with periodicity in the X direction and pattern P with periodicity in the Y direction
By double exposure with B, a resist pattern PS6 that becomes a contact hole is formed. In the figure, patterns PA and PB indicated by broken lines are obtained by the first exposure, and patterns PA and PB indicated by solid lines are obtained by the second exposure.

FIG. 11 is an example showing another positional relationship of reticle patterns to be subjected to multiple exposure. In this example,
It is assumed that the reticle R and the wafer W are shifted twice and overlapping exposure is performed three times. As shown in FIG. 8 or FIG. 10, the periodic pattern PA in which diffracted light is likely to occur,
When using PB, considering the diffraction efficiency, each pattern P
It is preferable that the number of repetitions of A and PB lines and spaces is as large as possible. Therefore, three patterns PA, PB, P
C is formed on the same reticle. The pattern PA is formed in the shape of a diffraction grating with a duty ratio of approximately 1:1 having periodicity in the X direction, and has five light-shielding line patterns (hatched portions). The pattern PB is formed in the shape of a diffraction grating with a duty ratio of approximately 1:1 having periodicity in the Y direction, and has five light-shielding line patterns (hatched portions). Here, the center of pattern PA and the center of pattern PB are ΔY in the Y direction.
apart, the center of the third pattern PC is the pattern P
This is a square light shielding portion (hatched portion) formed apart from the center of B by a distance ΔX in the X direction. Further, the area AB around the pattern PC is approximately equal to the shape and area of the two patterns PA and PB when they are overlapped. Further, the peripheries of each pattern PA, PB, and PC are all transparent parts. In forming the contact hole pattern, a square overlapping portion formed by overlapping the light-shielding line pattern at the center of pattern PA and the light-shielding line pattern at the center of pattern PB is used. Pattern PC is pattern PA and PB
This is for erasing unnecessary overlapping parts (latent images) caused by superposition of the images.

After performing the first exposure on the wafer using such a reticle, from the first exposure position,
The second exposure is performed by shifting the relative positions of the reticle R and the wafer W in each in-plane direction on the reticle by (0, +ΔY). Thereafter, from the first exposure position, the relative positions of the reticle R and the wafer W in each in-plane direction are determined on the reticle (-ΔX,
-ΔY) and perform the third exposure.

FIG. 12 shows latent images PA' and PB' formed on a negative resist when pattern PA and pattern PB in FIG. 11 are overlapped and exposed. In FIG. 12, the broken lines represent the edges of the latent images of each pattern, and the plurality of isolated areas filled in black represent unexposed portions that were not exposed at all by the two overlapping exposures. Among these multiple isolated areas, the central square pattern P
S8 remains unexposed even after the third overlapping exposure with pattern PC, and other unexposed isolated areas located around pattern PS8 are exposed by the transparent portion (area AB) around pattern PC. completely exposed to light. Therefore, only the square pattern PS8 is finally formed on the wafer W. In the case of a negative resist, the pattern PS8 portion is removed by development, and the resist around it is almost completely left behind, so that a contact hole pattern is formed in the same way as explained in FIG. 10 above. .

FIG. 13 shows an example of forming a contact hole pattern by regularly forming a plurality of the pattern arrays shown in FIG. 11 on the reticle R. In FIG. 13, the pattern PA in FIG. 11 is represented by a circle, the pattern PB is represented by a square, and the pattern PC is represented by a triangle. Using such a reticle R, when the relative position with respect to the wafer W is similarly shifted twice and overlapping exposure is performed three times, the resist on the wafer W has a pitch in the X direction of ΔX,
A contact hole pattern group having a pitch of ΔY in the Y direction is formed.

In FIG. 11 or FIG. 13, three patterns, PA, PB, and PC, are formed in the reticle, but the number is not limited to this, and there may be two,
It is also possible to create four patterns or the like and perform overlapping exposure two times or four times each. Also, patterns PA, PB
, PC, etc. as one block, and a plurality of blocks may exist.

Furthermore, as shown in FIGS. 5, 7, and 11, when only isolated hole patterns are formed by overlapping exposure, the first exposure, second exposure, or third exposure The amount of irradiation light (exposure amount) may be the same for two or three times, but is preferably different for each exposure. For example, when exposing periodic patterns (PA, PB), it is preferable to use a larger amount of irradiation light than when exposing a non-periodic pattern (PC). In the above-mentioned exposure apparatus shown in FIG.
, and a irradiance meter (not shown). Control device 30
recognizes that the current exposure is the nth exposure at the top, calculates the irradiation light amount according to the output signal from the irradiance meter, and stops the exposure when it reaches the target value. A command is sent to the drive motor 4 of the shutter 3. In addition, Fig. 9,
As shown in FIGS. 10 and 13, when a plurality of sets of patterns PA, PB (or PC) are regularly provided on the same reticle, the first exposure, the second exposure, or the third exposure It is desirable that the amount of irradiation light be the same.

When moving the relative position between the reticle and the wafer using the method described in each of the above embodiments, in the apparatus shown in FIG. It was assumed that the movement occurred during the second exposure or between the second and third exposures. or,
Alternatively, the stage RST holding the reticle R may be moved, or both stages may be moved. However, many of the actual projection exposure apparatuses already have a wafer stage that moves two-dimensionally, and perform step-and-repeat or step-and-scan exposure using this. The method of moving based on the measured coordinate values by No. 21 is the easiest to implement even in conventional devices.

FIG. 14 is a diagram showing almost the entire surface of a reticle used in the exposure method according to the present invention. The reticle pattern area (pattern formation area) 92 exists. In the conventional exposure method, it was possible to draw a circuit pattern entirely within this reticle pattern area 92 and transfer it to the wafer W by exposure. However, in the present invention, the relative positional relationship between the reticle and the wafer is different between the first exposure and the second exposure, as shown in FIGS. 9 and 10, for example. In FIG. 14, the area used for the first exposure is within the broken line 93, and the area used for the second exposure is within the two-dot chain line 94.

Therefore, the area (shaded area) used both in the first and second exposures becomes an effective reticle pattern area when the present invention is used, which is smaller than in the conventional case. However, for example, in the current step-and-repeat projection exposure apparatus, the size of the reticle pattern area 92 (X0,
Y0) are both around 100 mm (dimensions on the reticle), and the relative positional deviation amount between the first exposure and the second exposure (Px/2, Py/2) shown in FIGS. 9 and 10 teeth,
It is approximately equal to the memory cell size, and is approximately 10 μm in dimension on the reticle (ΔX, ΔY in FIG. 14).

Therefore, the conventional reticle effective area 100
,000μm×100,000μm, in the present invention, (100,000-2×10)μm×(100,000μm×100,000μm
000-2×10) μm. The ratio of this value is 1:0
．． Therefore, the reticle pattern effective area reduced by the present invention is only about 0.04% of the area in the conventional case.

Although FIG. 14 specifically shows a step-and-repeat type projection exposure apparatus in which the effective area of the projection optical system is circular, other projection exposure apparatuses, such as a step-and-scan type The same applies to exposure devices, and the area of the reticle pattern area that decreases is still very small. In each of the above embodiments, when using the apparatus shown in FIG. optically (
The reticle stage RST or the wafer stage WST is servo-controlled so as to drive the mark on the reticle R or the mark on the wafer W with respect to the offset mark detection center. Good too. Furthermore, in the explanation so far, either the reticle R or the wafer W is shifted by (ΔX, ΔY), but both the reticle R and the wafer W may be shifted in opposite directions or in the same direction. . In a device that performs mirror image projection using a 1-magnification projection optical system, it is meaningless to shift the reticle and wafer by the same distance in opposite directions with respect to the optical axis, but it is meaningless to shift the reticle and wafer by the same distance in opposite directions or in the same direction. A similar shift is possible by moving the wafer. In addition, in the case of a reduction projection optical system (mirror image projection),
When the projection magnification is 1/M, it is meaningless to shift the wafer in the opposite direction by 1/M with respect to the reticle movement, but there is no point in shifting the wafer in the opposite direction (in the same direction). ), similar positional shifting is possible. In the case of reduced projection, even if the reticle R and wafer W remain stationary on the apparatus and only the projection optical system is moved in parallel in the direction perpendicular to the optical axis AX, the projection image and the wafer can be Relative position can be shifted.

[0069]

As described above, according to the present invention, one periodic pattern and another pattern provided at different locations in one original plate (mask) are overlapped and exposed, and a special By using a projection exposure apparatus having an illumination optical system, it becomes possible to perform exposure transfer of fine patterns, especially fine isolated patterns, which was difficult to realize in the past.

Furthermore, in the present invention, a plurality of patterns to be subjected to multiple exposure are formed on one mask instead of on separate masks, so the throughput is correspondingly high. At the same time, mask manufacturing costs are also reduced. Further, since the plurality of patterns to be superimposed are located close to each other in the same mask, the amount of misalignment due to mask manufacturing errors can be extremely reduced. Furthermore, since there are patterns to be overlapped on the same mask, no overlay error due to alignment error occurs.

Furthermore, the mask used in the present invention is a normal mask having only information on transmittance, and has the advantage that it is not necessary to provide a phase shifter. In the above embodiments, the projection exposure apparatus used is one in which the light intensity distribution of the illumination light flux is concentrated in a local area whose center is eccentric from the optical axis on the pupil plane in the illumination optical system. However, as a modification, it is also possible to use an exposure apparatus having an annular light amount distribution on the pupil plane (or the Fourier transform surface within the illumination optical system). in this case,
The outer diameter of the annular light amount distribution is preferably about 0.6 to 0.8 in terms of σ value, and the inner diameter is preferably about 0.3 to 0.6 in terms of σ value.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the overall configuration of a projection exposure apparatus used in the present invention;

FIG. 2 is an optical path diagram schematically representing the illumination system of the device in FIG. 1;

FIG. 3 is a side view showing the configuration of a light splitter;

FIG. 4 is a front view showing the configuration of a light splitter;

FIG. 5 is a perspective view illustrating the principle and operation of the exposure method of the present invention;

6 is a diagram illustrating the light amount distribution during pattern exposure obtained by the exposure method of FIG. 5,

FIG. 7 is a diagram illustrating a combination of reticle patterns according to another embodiment of the present invention;

FIG. 8 is a diagram illustrating a combination of reticle patterns according to another embodiment of the present invention;

9 is a pattern layout diagram of a reticle provided with the reticle pattern shown in FIG. 8,

FIG. 10 is a diagram showing the mode of multiple exposure when using the reticle of FIG. 9;

FIG. 11 is a diagram showing the arrangement of a reticle pattern according to another embodiment of the present invention;

FIG. 12 is a diagram showing how the reticle pattern in FIG. 11 is subjected to multiple exposure;

13 is a diagram showing an example of arranging the reticle pattern of FIG. 11 on an actual reticle,

FIG. 14 is a diagram illustrating a pattern effective area on a reticle using the present invention;

FIG. 15 is a diagram illustrating the configuration and exposure state of a conventional projection exposure apparatus;

FIG. 16 is a diagram illustrating the principle of the exposure apparatus used in the present invention,

FIG. 17 is a graph showing the relationship between the amount of exposure to the resist layer and the remaining film thickness.

[Explanation of symbols]

Claims (8)

[Claims] 1. An exposure method using a projection exposure apparatus that exposes and transfers a pattern on an original, such as a circuit pattern, to a substrate to be exposed via a projection optical system, the pattern surface on the original to be transferred. An arbitrary region centered on one or more arbitrary points other than the optical axis of the illumination optical system in the plane of the illumination optical system of the exposure apparatus that has a substantially Fourier transform relationship with respect to The amount of illumination light for illuminating the original is concentrated, and the exposure to one exposed area on the exposed substrate is divided into multiple exposures, and in each of the multiple exposures, the amount of illumination light for illuminating the original is divided into multiple exposures. The relative positional relationship with the substrate to be exposed in a plane perpendicular to the optical axis of the projection optical system is varied, and in the plurality of exposures, at least one area on the substrate to be exposed is In all of the plurality of exposures, different patterns on the original plate are exposed in a superimposed manner, and at least one of the plurality of patterns exposed in a superimposed manner is a group of substantially periodically arranged patterns. A pattern exposure method characterized by the following. 2. At least two of the plurality of patterns exposed in a superimposed manner are composed of a group of patterns arranged approximately periodically, and the periods or directions of the patterns are different from each other. The exposure method according to claim 1, characterized in that: 3. The substantially periodically arranged patterns consist of a group of patterns that are fine enough to be below the resolution limit of the projection optical system, and the other patterns are about twice as fine as the resolution limit or more. 3. The exposure method according to claim 1, wherein the exposure method comprises a large pattern or a group of patterns. 4. The relative positional relationship between the original and the substrate to be exposed is changed by moving a stage that holds the substrate to be exposed in a direction perpendicular to the optical axis of the projection optical system. The exposure method according to claim 1, claim 2, or claim 3, wherein: 5. An exposure apparatus that implements the exposure method according to claim 1. 6. A mask comprising an original pattern to be exposed on a photosensitive layer of a photosensitive substrate, including a first pattern having a periodic structure with a fineness that can be transferred to the photosensitive layer, and a specific pattern of the first pattern. A second pattern having a shape that can match only the portion is juxtaposed closely at a predetermined interval, and the photosensitive layer is exposed to light in at least two times, and the first and second exposures are separated. In the meantime, both the images of the first pattern and the second pattern and the photosensitive substrate are relatively moved by an amount corresponding to the predetermined interval, so that the images of the first pattern and the image of the second pattern are separated. A mask characterized in that the pattern shape of only a specific portion of the first pattern is formed on the photosensitive layer by overlapping exposure on the photosensitive layer. 7. A mask having an original image pattern is placed on the object plane side of a projection optical system of a projection exposure apparatus, a photosensitive substrate onto which the original image pattern is transferred is placed on an image plane side of the projection optical system, and the In the method of exposing a projected image of the original pattern to the photosensitive substrate by irradiating the mask with illumination light from an illumination system of an exposure device, the mask includes a first pattern having a fine periodic structure; A second pattern having a shape that can match only a specific portion of the first pattern is provided shifted by a predetermined interval; the illumination system of the projection exposure apparatus is arranged in a Fourier transform relationship with respect to the mask. The illumination light is set to be concentrated in a region within a certain plane and separated from the optical axis by an amount determined according to the periodicity of the first pattern, and the mask and the photosensitive substrate are positioned relative to each other. a first exposure step of exposing the first pattern and the second pattern to the photosensitive substrate after the alignment; after the first exposure step, according to a predetermined interval between the first pattern and the second pattern; The projected image of the mask and the photosensitive substrate are relatively shifted by an amount such that the image of the second pattern and the image of the first pattern already exposed on the photosensitive substrate overlap. and a second exposure step of exposing the first pattern and the second pattern on the mask to the photosensitive substrate. 8. A projection type exposure apparatus used in lithography, comprising: (a) a first pattern having a fine periodic structure; and a first pattern having a shape that can match only a specific portion of the first pattern; a mask stage that holds a mask formed with two patterns shifted by a predetermined interval on the object plane side of the projection optical system in the projection exposure apparatus; (b) on the image plane side of the projection optical system; a substrate stage that holds a photosensitive substrate; (c) the projected image such that the projected image of the first pattern and the projected image of the second pattern can be alternatively projected onto the same position on the photosensitive substrate; (d) an illumination system for illuminating the mask, the illumination system comprising a light source and a plurality of optical elements, and a shift means for shifting the relative position of the mask and the photosensitive substrate; The illumination system has a Fourier transform surface that is substantially conjugate with the pupil plane of the projection optical system, and the illumination system further directs illumination light from the light source from the optical axis in the Fourier transform surface to the period of the first pattern of the mask. (e) irradiating the mask with illumination light from the illumination system to form each projected image of the first pattern and the second pattern; The method further comprises an exposure operation control means for dividing the number of exposure operations on the photosensitive substrate into at least two times and operating the shift means between each exposure operation, and controlling the pattern shape of only a specific portion of the first pattern. An exposure device formed on the photosensitive substrate.