KR20230155617A - Exposure device and exposure method - Google Patents

Abstract

The exposure apparatus EX includes light from light sources 2A, 2B, and 2C that generate light containing a plurality of bright line wavelengths (g-line, h-line, i-line, etc.) to illuminate the mask substrate M. a wavelength selection unit (6A, 6B, 6C) that enters and extracts an illumination light flux limited to a predetermined wavelength width including at least one specific bright line wavelength (i-line) among a plurality of bright line wavelengths, and a diffusion angle of the illumination light flux A first illumination optical system having numerical aperture variable portions 8A, 8B, 8C to adjust, and an optical integrator for irradiating an illumination beam with uniform illuminance on a mask substrate with a numerical aperture corresponding to the diffusion angle ( A second illumination optical system (ILn) including a fly-eye lens system (FEn) is formed, and the wavelength selection unit excludes bright lines on the long wavelength side and bright lines on the short wavelength side that appear in the vicinity of a specific bright line wavelength (i line), A first wavelength selection element (interference filter (SWb)) that extracts the spectral component of a specific bright line wavelength and the low-brightness spectral component distributed at the bottom of the specific bright line wavelength is installed.

Description

Exposure device and exposure method {EXPOSURE DEVICE AND EXPOSURE METHOD}

The present invention relates to an exposure apparatus for transferring a mask pattern to a substrate, and an exposure method.

Conventionally, in a photolithography process for manufacturing electronic devices such as liquid crystal display elements, semiconductor elements, and thin film magnetic heads, illumination light from a light source is irradiated onto a transmissive or reflective mask substrate, and a device pattern formed on the mask substrate (electronic device An exposure device is used to project and expose transmitted or reflected light from a pattern on a substrate to be exposed, such as a plate coated with a photosensitive agent such as photoresist, via a projection optical system. As a conventional exposure apparatus, for example, as disclosed in Japanese Patent Application Publication No. 2012-049332, each illumination light from a light source such as two mercury lamps is bundled into a circular shape on the inlet side and a rectangular shape on the outlet side ( It is known to form an illumination system (illumination device) that performs Köhler illumination of the illumination area on the slit on a mask substrate with a uniform illuminance distribution by an integrator using a fly-eye lens optical system or the like after synthesizing the bundled fiber into a slit shape.

When using a mercury lamp (such as an ultra-high pressure mercury discharge lamp) as a light source, the discharge arc light of the mercury lamp contains a plurality of bright lines, and a specific bright line wavelength is selected among them to become the illumination light for exposure (illumination light for the mask substrate). there is. In the photolithography process, considering the light-sensitive wavelength characteristics of the photoresist and the optical performance (resolution, chromatic aberration characteristics) of the projection optical system, etc., among the bright line wavelengths of the mercury lamp, the g-line (center wavelength 435.835 nm), h in the ultraviolet wavelength range are used. Line (center wavelength 404.656 nm) and i-line (center wavelength 365.015 nm) are mainly used. The resolution (R), expressed as the minimum linewidth possible for projection exposure, is NAp for the numerical aperture on the image side (exposed substrate side) of the projection optical system, λ (nm) for the wavelength of the illumination light, and k (0 < k ≤ 1) for the process constant. , it is defined as R = k·(λ/NAp). In this regard, by using the i-line, which has the shortest wavelength among the three bright line wavelengths, projection exposure (high-resolution exposure) of a finer mask pattern becomes possible. However, in recent years, as the exposure process for negative-type photoresist layers (light-sensitive layers), which have lower sensitivity than positive-type photoresist layers (light-sensitive layers), has increased, it has become necessary to set longer exposure times, leading to a decrease in the number of processed sheets per unit time of exposed substrates. (Decreased productivity) is a concern.

According to a first aspect of the present invention, there is provided an exposure apparatus for project exposure of a mask pattern onto a photosensitive substrate, comprising: a light source generating light containing a plurality of bright line wavelengths to illuminate a mask; and light from the light source. a wavelength selection unit that enters and extracts an illumination light flux limited to a predetermined wavelength width including at least one specific bright line wavelength among the plurality of bright line wavelengths, and a numerical aperture variable portion that adjusts a diffusion angle of the illumination light flux. a first illumination optical system having a first illumination optical system, and an optical integrator for incident the illumination beam with the diffusion angle adjusted and irradiating the illumination beam with a uniform illuminance on the mask with a numerical aperture corresponding to the diffusion angle. A second illumination optical system comprising a second illumination optical system, wherein the wavelength selection unit includes a spectral component of the specific bright line wavelength and a bright line of the specific bright line wavelength, excluding long-wavelength side bright lines and short-wavelength side bright lines appearing in the vicinity of the specific bright line wavelength. An exposure apparatus equipped with a first wavelength selection element that extracts a low-brightness spectral component distributed in the skirt is provided.

According to a second aspect of the present invention, there is provided an exposure method for project exposure of a mask pattern onto a photosensitive substrate, comprising: a peak of at least one specific bright line wavelength among light from a light source that generates light containing a plurality of bright line wavelengths; In addition to the spectral components of the above-described specific bright line, the wavelength is selected to extract a low-intensity spectral component distributed at the bottom of the specific bright line wavelength, not including the bright line on the long-wavelength side and the bright line on the short-wavelength side that appears in the vicinity of the specific bright line wavelength, and A mirror projection method in which an illumination light beam of a wavelength-selected spectral component is irradiated on the mask at a uniform illuminance, and no chromatic aberration occurs in the wavelength of the low-brightness spectral component, or a mirror projection method in which chromatic aberration does not occur in the wavelength of the low-brightness spectral component. An exposure method is provided including projecting and exposing the pattern of the mask onto the substrate through a catadioptric projection optical system in which chromatic aberration is corrected.

According to the third aspect of the present invention, light with a spectral distribution including a specific bright line wavelength selected by the wavelength selection unit among the light including the bright line wavelength generated from the light source device is used to create a pattern for an electronic device by an illumination optical system. An exposure method in which an image of the pattern is projected and exposed onto a photosensitive substrate by a projection optical system that irradiates a supporting mask and enters a light flux for exposure generated from the mask, wherein the image of the pattern is projected and exposed by the wavelength selector from the light source device. To extract light of a first spectral distribution and light of a second spectral distribution having different wavelength bands from the generated light, and to perform Köhler illumination of the mask by the illumination optical system, on the pupil surface in the illumination optical system, Comprising forming a first light source image distributed in a two-dimensional range by light of the first spectral distribution and a second light source image distributed in a two-dimensional range by light of the second spectral distribution by overlapping them. An exposure method is provided.

According to the fourth aspect of the present invention, the mask pattern is illuminated with illumination light of a predetermined wavelength distribution, and the image of the mask pattern is generated by a projection optical system that enters the imaging light flux generated from the mask pattern and projects it on the substrate. An exposure method of projection exposure on a substrate, wherein a specific central wavelength of the wavelength distribution of the illumination light is λ, the numerical aperture of the projection optical system on the substrate side is NAp, and a process constant is k (0 < k ≤ 1), k When a projection image of a square or square hole pattern with a size close to the minimum resolvable line width dimension determined by the resolving power (R) defined by (λ/NAp) is projected onto the substrate, the hole is transformed into an elliptical shape. Setting the width of the wavelength distribution of the illumination light including the center wavelength λ so that the ratio of the minor axis length to the major axis length of the projected image of the pattern is 80% or more, preferably 90% or more, and the set width An exposure method is provided, including illuminating a mask on which a pattern for an electronic device is formed with illumination light of a wavelength distribution of and projecting and exposing the pattern for an electronic device on the substrate.

Fig. 1 is a perspective view showing the schematic configuration of a scanning type projection exposure apparatus according to the first embodiment.
FIG. 2 is a diagram showing the arrangement of optical members of a projection optical system mounted on the projection exposure apparatus shown in FIG. 1.
FIG. 3 is a perspective view schematically showing the overall configuration of an illumination device for irradiating illumination light for exposure to a mask substrate loaded in the projection exposure device shown in FIG. 1.
FIG. 4 is a perspective view schematically showing the configuration of the first illumination optical system from the mercury lamp to the light guide fiber (fiber bundle) in the lighting device shown in FIG. 3.
Figure 5 is a graph schematically showing an example of the wavelength characteristics (spectral distribution) of light generated by arc discharge of an ultra-high pressure mercury discharge lamp.
FIG. 6 is a graph schematically showing how light with a narrow wavelength width including the i line is selectively extracted from the wavelength characteristics (spectral distribution) shown in FIG. 5 by an i line-narrow band interference filter. am.
FIG. 7 schematically shows how light with a wide wavelength width including the i line and its bottom is selectively extracted from the wavelength characteristics (spectral distribution) shown in FIG. 5 by an i line-broadband interference filter. This is a graph shown as .
FIG. 8 schematically shows how light with a wide wavelength width including both the i line and the h line is selectively extracted from the wavelength characteristics (spectral distribution) shown in FIG. 5 by an i line + h line interference filter. It's a graph.
FIG. 9 is a perspective view schematically showing the overall configuration of the light guide fiber (fiber bundle) formed in the lighting device shown in FIG. 3 and the respective shapes of the entrance end and the exit end.
FIG. 10 is a perspective view schematically showing the configuration of a second illumination optical system in the illumination device shown in FIG. 3 that irradiates illumination light from the exit end of the fiber bundle to the illumination area on the mask substrate.
FIG. 11 is a diagram schematically showing the state of illumination light in the optical path from the exit end of the fiber bundle shown in FIG. 10 to the fly-eye lens system.
FIG. 12 is a diagram schematically showing an example of the arrangement of a plurality of point light source images formed for each fiber strand at the exit end of the fiber bundle shown in FIG. 11.
FIG. 13 is a diagram showing the arrangement of a plurality of point light source images formed at each emission end of a plurality of lens elements constituting the FlyEye lens system shown in FIG. 11.
FIG. 14 is a diagram schematically showing the state of illumination light in the optical path from the fly-eye lens system shown in FIG. 10 to the illumination area on the mask substrate.
FIG. 15 is a diagram explaining the operation of adjusting the numerical aperture (diffusion angle) of the illumination beam irradiated to the entrance end of the fiber bundle by the magnification variable portion (numerical aperture variable portion) shown in FIG. 4.
FIG. 16 is a diagram schematically showing the state of the light flux incident on the three fiber bundles on the incident side of the fiber bundle shown in FIG. 9 and the state of the illumination light flux emitted from the six fiber bundles on the emission side.
Fig. 17 is a schematic diagram of the optical path from the exit end of the fiber bundle to the incident surface of the FlyEye lens system as seen from the X direction (scanning movement direction).
FIG. 18 is a view in the XY plane of the circular areas (CFa, CFb, CFc) in FIG. 17 distributed on the incident surface of the FlyEye lens system.
Figure 19(A) is a diagram of the distribution of spot light (point light source image) formed on the emission surface of the FlyEye lens system as seen from the This is a diagram showing the distribution of spot light (point light source image) formed on a surface as seen from the Y direction (step movement direction).
Fig. 20 is a diagram schematically showing the orientation characteristics (diffusion angle characteristics) of the illumination light flux (Irn) irradiated to the point OP on the illumination area on the mask substrate.
FIG. 21 shows the i-line narrow interference filter (SWa) of FIG. 6, the i-line-broad interference filter (SWb) of FIG. 7, and the i-line wide interference filter (SWb) of FIG. 8 mounted on each of the three wavelength selection units 6A, 6B, 6C. This table summarizes examples of combinations of i line + h line-interference filter (SWc).
FIG. 22 is a graph schematically showing the wavelength characteristics of the illumination light flux of the mask substrate obtained by combining interference filters according to the combination code B2 in the table in FIG. 21.
FIG. 23 is a graph showing an example of the light absorption characteristics depending on the wavelength of a negative photoresist for explanation of Modification Example 3.
Fig. 24 is a cross-sectional view schematically showing the inclination that occurs in the edge portion (side wall) of the remaining resist image after development, for explanation of Modification Example 3.
FIG. 25(A) shows the configuration according to Modification Example 4 and is a diagram showing the shape of the aperture plate APa on which a ring-shaped light transmitting portion is formed, and FIG. 25(B) shows Modification Example 4. This figure shows the configuration and shows the shape of the aperture plate APb on which the tetrapolar light transmitting portion is formed.
FIG. 26 shows the configuration according to Modification Example 5, and is a diagram showing an annular aperture plate arranged in the wavelength selection section 6A of the first illumination optical system.
Fig. 27 is a diagram showing a schematic overall configuration of the exposure apparatus according to the second embodiment.
FIG. 28 is a graph showing detailed spectral characteristics obtained when the wavelength characteristics of the ultra-high pressure mercury discharge lamp shown in FIG. 5 above are measured using a spectrometer with high wavelength resolution.
Fig. 29 is a graph showing the relationship between the chromatic aberration characteristics of the projection optical system and the bright line wavelength of the i line of the mercury lamp.
Figure 30 is a graph explaining the difference in wavelength characteristics between a high-pressure mercury discharge lamp and an ultra-high pressure mercury discharge lamp.
Fig. 31 is a diagram explaining the deformation of the shape of the projected image obtained when hole patterns of different sizes formed on a mask are projected onto a substrate.
FIG. 32 is a diagram illustrating a method for calculating the flatness (ellipticity) when the projection image of a hole pattern is transformed into an elliptical shape.
FIG. 33 is a diagram explaining the tilt of the projection image of the hole pattern transformed into an elliptical shape shown in FIG. 32.
Figure 34 is a graph showing an example of the change characteristics of the refractive index of synthetic quartz that changes depending on the wavelength.
Fig. 35 is a diagram schematically explaining the state of image shift by a parallel flat quartz plate formed as an image shift optical member.
Figure 36 is a graph showing an example of the amount of difference due to the relative positional shift between the projected image by the i line and the projected image by the h line, which changes depending on the inclination angle of the quartz plate of the image shift optical member.

An exposure apparatus related to an aspect of the present invention will be described in detail below with reference to preferred embodiments and accompanying drawings. In addition, the aspects of the present invention are not limited to these embodiments and include those with various changes or improvements. In short, the components described below include those that can be easily assumed by those skilled in the art and those that are substantially the same, and the components described below can be appropriately combined. Additionally, various omissions, substitutions, or changes to the components may be made without departing from the gist of the present invention.

[First Embodiment]

FIG. 1 is a perspective view showing a schematic overall configuration of a scanning type projection exposure apparatus EX according to the first embodiment, and FIG. 2 is a partial projection optical system PLn mounted on the projection exposure apparatus EX of FIG. 1. ) This is a drawing showing the arrangement of the optical members. 1 and 2, the direction in which the Z axis of the Cartesian coordinate system XYZ extends represents the direction of gravity, and the direction in which the This indicates the scanning movement direction in which the plate P is moved for scanning exposure, and the direction in which the Y axis extends indicates the direction of the step movement of the plate P. The projection exposure apparatus EX of this embodiment includes a projection optical system having six catadioptric partial projection optical systems PL1 to PL6, and a flat mask substrate M and a photoresist layer (photoresist, etc.) applied thereto. It will be described as a step-and-scan type exposure apparatus that transfers the image of the pattern for an electronic device formed on the mask substrate M to the plate P while synchronously moving the flat plate P in the X direction. . In addition, since the projection exposure apparatus EX shown in FIGS. 1 and 2 is the same as the configuration disclosed in, for example, the pamphlet of International Publication No. 2009/128488 or Japanese Patent Application Publication No. 2010-245224, The description of the device configuration shown in Figures 1 and 2 will be simplified.

[Configuration of projection optical system]

Each of the six illumination areas IA1 to IA6 (see FIG. 1) set on the mask substrate M has a rectangular shape whose dimension in the X direction, which is the scanning direction, is shorter than the dimension in the Y direction, which is the step movement direction. It is set. Illumination light for exposure adjusted to a uniform illuminance distribution (for example, uniformity within ±5%) is projected onto each of the illumination areas (IA1 to IA6) from an illumination device described later. Each of the six illumination areas (IA1 to IA6) is set at a position on the object surface side of each of the six partial projection optical systems (PL1 to PL6). For example, when the pattern portion of the mask substrate M appears in the illumination area IA1, the transmitted light generated from the pattern portion is reflected by the upper reflective surface of the prism mirror PMa and enters the partial projection optical system PL1. . The partial projection optical system PL1 transmits transmitted light (imaging light flux, light flux for exposure) from the pattern portion into a lens system (Ga1, Ga2, Ga3) arranged along the optical axis AXa, and a concave mirror ( By reflecting from the lower reflecting surface of the prism mirror PMa via the first imaging system PL1a including Ga4), the intermediate image of the illuminated area IA1 is imaged at equal magnification on the intermediate image plane IM1.

As shown in FIG. 1, a field stop plate FA1 having a trapezoidal opening with both edge portions in the Y direction slanted is disposed on the intermediate image plane IM1. The imaging luminous flux that passes through the opening of the field stop plate FA1 is reflected by the reflecting surface on the upper side of the prism mirror PMb, and as shown in FIG. 2, the lens system Gb1, Gb2, It is reflected toward the direction (-Z direction) of the plate P on the lower reflecting surface of the prism mirror PMb via the second imaging system PL1b including the concave mirror Gb3) and the concave mirror Gb4. Accordingly, within the trapezoidal projection area EA1 set on the plate P, the intermediate image formed in the opening of the field stop plate FA1 is re-imaged and imaged at equal magnification. The partial projection optical system PL1 uses the first imaging system PL1a and the second imaging system PL1b to project an image of the pattern portion within the illumination area IA1 into the projection area EA1 in an equal magnification relationship. It forms an image telecentrically.

As shown in Fig. 2, the first imaging system PL1a is a catadioptric half-field type imaging system in which a concave mirror Ga4 is arranged in the pupil plane Epa, and the second imaging system PL1b is also a catadioptric type. , It is a catadioptric half-field type imaging system in which a concave mirror (Gb4) is placed on the pixel plane (Epb). The co-planes (Epa, Epb) are optically conjugated to each other, and on each of the co-planes (Epa, Epb), a light source image (secondary light source image) formed in the lighting device that illuminates the illumination area (IA1) is formed. . Additionally, in the imaging optical path of the partial projection optical system PL1, between the mask substrate M and the prism mirror PMa, the focus state of the image projected on the projection area EA1 on the plate P is finely adjusted. A focus adjustment optical member FC1 for adjustment is formed. Additionally, between the field stop plate (FA1) and the prism mirror (PMb), image shift optics are provided to independently finely adjust the position of the projection area (EA1) projected on the plate (P) in the X and Y directions, respectively. The member SC1 is formed, and between the prism mirror PMb and the plate P, a magnification adjustment is performed to finely adjust the size of the image of the pattern portion projected on the projection area EA1 within a range of ± several tens ppm. An optical member MC1 is formed. The focus adjustment optical member (FC1), the image shift optical member (SC1), and the magnification adjustment optical member (MC1) are disclosed, for example, in the pamphlet of International Publication No. 2013/094286, so details regarding their configuration and functions are provided. The explanation is omitted.

In this embodiment, as shown in FIG. 2, the partial projection optical system PL1 includes a first imaging system PL1a, a second imaging system PL1b, prism mirrors PMa, PMb, and a field stop plate FA1. , a focus adjustment optical member FC1, an image shift optical member SC1, and a magnification adjustment optical member MC1, and other partial projection optical systems PL2 to PL6 are also configured in the same manner. Therefore, each of the other partial projection optical systems PL2 to PL6 also forms an image of the pattern portion of the mask substrate M at equal magnification in each of the trapezoidal projection areas EA2 to EA6 set on the plate P. do. Accordingly, when the mask substrate M and the plate P are moved one dimensionally at the same speed in the The pattern parts are connected in the Y direction. In addition, the partial projection optical system (PL1 to PL6), the illumination areas (IA1 to IA6), and the projection areas (EA1 to EA6) described above are referred to as the partial projection optical system (PLn) and the illumination areas (EA1 to EA6), respectively, when there is no need to specifically distinguish them. It will also be called area (IAn) or projection area (EAn) (n = 1 to 6).

[Configuration of lighting device]

3 is a perspective view showing a schematic overall configuration of an illumination device for projecting illumination light for exposure to each of six illumination areas IA1 to IA6 set on the mask substrate M, and the orthogonal coordinate system XYZ is the above. It is set the same as Figures 1 and 2. In the lighting device according to the present embodiment, as disclosed in Japanese Patent Application Publication No. 2010-245224, three mercury lamps (short arc type ultra-high pressure mercury discharge lamps) (2A, 2B, 2C) with the same specifications are used as light sources. (light source device) shall be provided. The number of lamps in the light source device is determined depending on the number of partial projection optical systems PLn so that the illumination light projected to each of the illumination areas IAn has a desired illuminance value, but it may be two or more. Ultra-high pressure mercury discharge lamps produce bright lines in the ultraviolet wavelength range, such as the g-line (wavelength 435.835 ㎚), h-line (wavelength 404.656 ㎚), and i-line (wavelength), by setting the vapor pressure of the mercury enclosed in the discharge tube to 10 ⁶ Pa (Pascal) or higher. 365.015 ㎚) is generated with high brightness. Each light emitting point (arc discharge portion) of the mercury discharge lamps 2A, 2B, 2C is disposed at the position of the first focus of the ellipsoid mirrors 4A, 4B, 4C, respectively, and the ellipsoid mirrors 4A, 4B, 4C ) The light flux BM reflected from each inner reflection surface is condensed (converged) toward the position of each second focus of the elliptical mirrors 4A, 4B, and 4C.

The light flux BM radiating in the -Z direction from each of the elliptical mirrors 4A, 4B, and 4C is divided into spectral components in the ultraviolet wavelength range for exposure by a dichroic mirror DM disposed in front of the second focus ( For example, the short wavelength range (460 nm or less) is reflected in the +X direction, and the spectral components in the longer wavelength range are separated to be transmitted. Since the luminous flux in the ultraviolet wavelength region for exposure reflected from each of the dichroic mirrors (DM) has the narrowest luminous flux diameter at the position of each second focus of the elliptical mirrors 4A, 4B, and 4C, the second focus Rotary shutters 5A, 5B, and 5C are disposed at each of the positions. The light beam in the ultraviolet wavelength range for exposure that has passed through each of the rotary shutters 5A, 5B, and 5C diverges and enters the wavelength selection units 6A, 6B, and 6C. Each of the wavelength selection units 6A, 6B, and 6C is provided with a plurality of lens elements and an interference filter for wavelength selection, and transmits only the desired bright line wavelength portion of the incident light flux in the ultraviolet wavelength range for exposure. The interference filters formed in each of the wavelength selection sections 6A, 6B, and 6C determine the fineness (resolution) of the pattern of the mask substrate M to be exposed and the exposure amount to be given to the light sensitive layer of the plate P. Depending on the dose, it is installed interchangeably (switchable) with several different wavelength selection characteristics. The difference in the wavelength selection characteristics of the interference filter will be explained in detail later, but the wavelength characteristics (wavelength distribution) of the illumination light for exposure projected onto the illumination area (IAn) on the mask substrate (M) can be pattern exposed with higher resolution. It can be converted into characteristics suitable for pattern exposure by increasing illuminance to increase productivity. Therefore, the interference filter has the characteristic of transmitting any one of the bright line wavelength components of the g-line (wavelength 435.835 nm), h-line (wavelength 404.656 nm), and i-line (wavelength 365.015 nm), g-line, h-line, and i-line. It is prepared in advance to have a characteristic of transmitting two consecutive bright line wavelength components (g line + h line, or i line + h line), or a characteristic of transmitting all bright line wavelength components of the g line, h line, and i line. It is done.

The light flux emitted from each of the wavelength selection units 6A, 6B, and 6C is transmitted to three fiber bundles (light guide fibers, optical transmission elements) 12A, 12B, and 12C on the incident side of the rear light distribution unit 10. The illumination beams (BMa, BMb, BMc) incident on each are incident on magnification variable parts 8A, 8B, and 8C for adjusting the numerical aperture (maximum inclination angle of the main ray) or the radial dimension (diameter). Each of the magnification variable parts 8A, 8B, and 8C continuously adjusts the numerical aperture (NA) of the illumination beams (BMa, BMB, BMc) incident on each of the fiber bundles 12A, 12B, and 12C within a certain range. In order to do this, a plurality of lens elements are provided that can move in the direction of the optical axis. By each of the magnification variable parts 8A, 8B, and 8C, as a result, the optical axis of the light source image (secondary light source image) distributed in each pupil plane (Epa, Epb) of the partial projection optical system PLn shown in FIG. 2 The radial dimensions from (AXa, AXb) can be changed continuously. That is, in each of the magnification variable parts 8A, 8B, and 8C, the maximum numerical aperture of the partial projection optical system PLn is set to NAp, and the numerical aperture of the illumination beam projecting the illumination area IAn is set to NAi. , the illumination σ value (0 < σ ≤ 1), which is determined by NAi/NAp, which is the ratio of numerical apertures, can be adjusted. Therefore, each of the magnification variable parts 8A, 8B, and 8C is also called a numerical aperture variable part capable of continuously adjusting the illumination σ value (numerical aperture (NPi) of the illumination luminous flux). Additionally, the configuration shown in FIG. 3 from the elliptical mirror 4A to the magnification variable section 8A, the configuration from the elliptical mirror 4B to the magnification variable section 8B, and the configuration from the elliptical mirror 4C to the magnification variable section 8C. ) Each of the configurations up to is collectively referred to as the first illumination optical system, and details of its functions will be described later.

The light distribution unit 10 distributes the illumination light beams BMa, BMB, and BMc incident from each of the three incident-side fiber bundles 12A, 12B, and 12C, corresponding to each of the six illumination areas IAn. It is distributed to six fiber bundles (FG1 to FG6) on the emission side so as to be distributed to each of the second illumination optical systems (IL1 to IL6) arranged. Each of the second illumination optical systems (IL1 to IL6) uses the emission end of the fiber bundle (FG1 to FG6) as a light source image (secondary light source image composed of multiple point light sources), and each illumination area (IAn) is converted to a Köhler illuminate. In addition, each of the second illumination optical system (IL1 to IL6) and the fiber bundle (FG1 to FG6) described above is, in cases where there is no need to specifically distinguish, the second illumination optical system (ILn) and the fiber bundle (FGn) (n) ＝ It is also called 1 ~ 6).

[First illumination optical system]

FIG. 4 is a perspective view showing the detailed configuration of the first illumination optical system disposed in the optical path from the mercury lamp 2A shown in FIG. 3 to the fiber bundle 12A on the incident side, and the orthogonal coordinate system It is set the same as 3. In addition, the first illumination optical system from the mercury lamp 2B to the fiber bundle 12B on the incident side and the first illumination optical system from the mercury lamp 2C to the fiber bundle 12C on the incident side are the same as those in FIG. It is composed of: As shown in FIG. 4, the luminous flux BM immediately after ejection from the ejection opening (end in the -Z direction) of the ellipsoid mirror 4A along the optical axis AX1 is above the ellipsoid mirror 4A (+Z direction). The opening portion and the lower electrode portion of the mercury lamp 2A create an annular intensity distribution centered on the optical axis AX1, that is, a center-missing distribution in which the illuminance at the center is extremely low. The luminous flux BM is condensed toward the position PS1 of the second focus of the ellipsoid 4A where the rotating blade of the rotary shutter 5A is disposed, and the arc discharge occurring between the electrodes of the mercury lamp 2A Since it is distributed thinly and long in the direction of the optical axis AX1, it does not converge light into a point at the position PS1, but becomes a beam waist with a finite size (diameter).

The wavelength selection unit 6A includes a lens system (collimator lens) 6A1 that receives the light flux BM diverging from the second focus position PS1 and converts it into a substantially parallel light flux and has different wavelength selection characteristics. Maintaining two interference filters (wavelength selection member, wavelength selection element, band pass filter) (SWa, SWb), and switching so that one of the interference filters (SWa, SWb) is inserted and removed in the optical path. A lens system (6A2) that converges the light flux (BMa) transmitted through the slide mechanism (FX) and any of the interference filters (SWa, SWb) at the focal position (PS2) (a position optically conjugate with the position (PS1)) ) is formed. The slide mechanism FX has a configuration that facilitates easy attachment and detachment of the interference filters SWa and SWb. When using a third interference filter (wavelength selection member, wavelength selection element, band-pass filter) having wavelength selection characteristics different from any of the interference filters (SWa, SWb), the mercury lamp 2A is activated by the rotary shutter 5A. In a state in which the luminous flux BM is shielded, either one of the interference filters SWa or SWb may be detached from the slide mechanism FX, and a third interference filter may be mounted instead. Additionally, when the slide mechanism is not installed, a mount mechanism that allows the interference filters (SWa, SWb), etc. to be easily attached and detached is provided.

The light beam BMa emitted from the wavelength selection unit 6A becomes a beam waist at the focal position PS2 and then enters the magnification variable unit 8A in a divergent state. At the focus position PS2, a circular light source image is formed by a blurred image of the arc discharge portion (light emission point) of the mercury lamp 2A. The magnification variable section 8A has two lens systems 8A1 and 8A2 whose positions along the optical axis AX1 are adjustable. The luminous flux BMa diverging from the focal position PS2 and traveling by the lens systems 8A1 and 8A2 has a predetermined luminous flux diameter or a predetermined aperture on the incident end FBi of the fiber bundle 12A on the incident side. The light is concentrated to be projected into water. The entrance end FBi of the fiber bundle 12A is basically arranged to be optically conjugate with the focus position PS2, but by adjusting the positions of the lens systems 8A1 and 8A2 of the variable magnification section 8A, The conjugate relationship may be intentionally violated. The two lens systems 8A1 and 8A2 function as a variable-power relay system, and with a change in the numerical aperture of the illumination luminous flux BMa, the luminous flux BMa is consequently concentrated at the entrance end FBi of the fiber bundle 12A. The diameter of becomes smaller or larger with respect to the effective maximum diameter of the incident end FBi.

[Wavelength selection by interference filter]

Here, an example of wavelength selection using an interference filter that can be mounted on the slide mechanism FX of the wavelength selection unit 6A will be described with reference to FIGS. 5 to 8. FIG. 5 is a graph schematically showing an example of the wavelength characteristics (spectral distribution) of the luminous flux (BM) generated by the arc discharge of a mercury lamp (ultra-high pressure mercury discharge lamp). In addition, FIG. 6 is a graph schematically showing how light with a narrow wavelength width including the i line is selectively extracted from the spectral distribution of FIG. 5 by the i-line narrow interference filter (SWa), and FIG. 7 is This is a graph schematically showing how light with a relatively wide wavelength width, including the i line and the low-brightness portion at the bottom thereof, is selectively extracted from the spectral distribution of FIG. 5 by the i line-wide interference filter (SWb), and FIG. 8 schematically shows how light with a wide wavelength width including both the i line and the h line is selectively extracted from the spectral distribution in FIG. 5 by the i line + h line interference filter (SWc) (third interference filter). This is the graph shown. In all of the graphs in FIGS. 5 to 8, the horizontal axis represents the wavelength (nm), and the vertical axis represents the relative intensity (%). In addition, in the wavelength characteristics (spectral distribution) of the luminous flux (BM) from the ultra-high pressure mercury discharge lamp shown in Fig. 5 (and Figs. 6 to 8), each of the main bright lines, the g line, h line, i line, and j line, The spectral portion on the peak is shown as the wavelength when measured with a spectrometer that does not have a very high wavelength resolution, and the actual wavelength of the spectral portion on the peak is defined as the full width at half maximum (width that is half the intensity of the peak intensity). In one case, it is about several nm to dozens of nm.

In this embodiment, as shown in FIGS. 6 to 8, three types of interference filters (SWa, SWb, and SWc) are prepared. As shown in FIG. 6, the i line-narrowband interference filter (SWa) has a transmittance of 10% or more between the wavelength of about 354 nm and about 380 nm, and has a transmittance of 10% or more between the wavelength of about 359 nm and about 377 nm. It has wavelength selection characteristics with a transmittance of 90% or more. Therefore, the full width at half maximum of the wavelength width selected by the i-line narrow interference filter (SWa) is about 22 nm, including the bright line wavelength (365.015 nm) of the i-line. In addition, as shown in FIG. 7, the i-ray-wideband interference filter (SWb) has a transmittance of 10% or more between the wavelength of about 344 nm and about 398 nm, and has a transmittance of 10% or more between the wavelength of about 350 nm and about 395 nm. It has wavelength selection characteristics with a transmittance of 90% or more. Therefore, the full width at half maximum of the wavelength width selected by the i-line-wide-band interference filter (SWb) is approximately 49 nm, including the bright line wavelength (365.015 nm) of the i-line. Both the i-line narrow interference filter (SWa) and the i-line wide interference filter (SWb) select only the bright line wavelength band of the i line as the illumination light for exposure, but the i-line narrow interference filter (SWa) has better wavelength selection. Because the bandwidth is narrow, the monochromaticity of the i-line (narrow) shown by the hatched portion in Fig. 6 is better than the i-line (light) shown by the hatched portion in Fig. 7 selected by the i-line-wide interference filter (SWb), and the partial projection optical system The influence of chromatic aberration characteristics of (PLn) is reduced, and higher resolution pattern exposure becomes possible.

However, the amount of light of the i-line (narrow) obtained by the i-line narrow interference filter (SWa) (area of the hatched portion in FIG. 6) is the amount of light of the i-line (light) obtained by the i-line-broad interference filter (SWb) Since it is small compared to (the area of the hatched portion in FIG. 7), it becomes necessary to slightly reduce the moving speed of the mask substrate M and plate P during scanning exposure, resulting in a decrease in productivity. In contrast, the i line (light) obtained by the i line-wide interference filter (SWb) is the low-brightness bottom portion between the bright line wavelength (365.015 nm) of the i line to the h line located near the long wavelength side. , and a relatively strong peak wavelength located in the vicinity of the short wavelength side, it contains spectral components at the bottom of low brightness, enabling high-resolution pattern exposure while increasing the amount of light by several percent or more. This becomes possible and productivity can be improved. i The bandwidth of wavelength selection by the line-wide interference filter (SWb) (about 49 nm at full width at half maximum) is the contrast value of the pattern projection image of the minimum line width obtained based on the chromatic aberration characteristics of the partial projection optical system (PLn). is determined by The chromatic aberration of the partial projection optical system (PLn) includes magnification (horizontal) chromatic aberration and axial (vertical) chromatic aberration. For example, in a projection optical system specialized only for the bright line wavelength of the i line, the chromatic aberration is at the bright line wavelength of the i line. It is corrected to have chromatic aberration characteristics of approximately zero, and a tendency for the amount of aberration to increase (quadratic tendency) on the short and long wavelength sides. Additionally, in a projection optical system that allows the use of two bright line wavelengths, the i and h lines, the amount of chromatic aberration is set to approximately zero at a wavelength approximately halfway between the i and h lines, and the amount of chromatic aberration is set to approximately zero between the respective bright line wavelengths of the i and h lines. It is corrected for chromatic aberration characteristics that tend to reduce the rate of change of the vehicle.

When using two bright line wavelengths, the i line and the h line, an i line + h line-interference filter (SWc) is attached to the slide mechanism (FX), and the light i line + h line of the spectral distribution shown in the hatched part in FIG. 8 is obtained. Use . As shown in FIG. 8, the i line + h line interference filter (SWc) has a transmittance of 10% or more between the wavelength of about 344 nm and about 420 nm, and the wavelength of about 350 nm to about 415 nm. It has wavelength selection characteristics with a transmittance of 90% or more. Therefore, the full width at half maximum of the wavelength width selected by the i line + h line interference filter (SWc) is about 70 nm, including the bright line wavelength of the i line (365.015 nm) and the bright line wavelength of the h line (404.656 nm). In pattern exposure using two bright line wavelengths, i-line and h-line, the minimum line width that can be resolved becomes larger compared to pattern exposure using only i-line, but the i-line + h obtained by the i-line + h line-interference filter (SWc) The amount of light in the line (area of the hatched portion in FIG. 8) is overwhelmingly increased compared to the case of the i-line-narrow interference filter (SWa) in FIG. 6 or the i-line-broad interference filter (SWb) in FIG. It improves dramatically. Therefore, when the pattern of the mask substrate M to be projected and exposed on the photosensitive layer of the plate P does not contain a pattern with a critical line width with high fineness, the i line + h line-interference filter (SWc ) By using , highly productive pattern exposure becomes possible.

[Optical distribution unit (10)]

FIG. 9 shows the overall configuration of the fiber bundle as the light distribution section 10 formed in the lighting device shown in FIG. 3, the shape of each incident end FBi of the fiber bundles 12A, 12B, and 12C on the incident side, and It is a perspective view schematically showing the shape of each emission end (FBo) of the fiber bundles (FG1 to FG6) on the emission side, and the orthogonal coordinate system XYZ is set the same as in FIG. 3. Each incident end FBi of the fiber bundles 12A, 12B, and 12C on the incident side is formed by bundling a plurality of fiber wires so that the entire cross-sectional diameter is a circle of several tens of mm or more. Each of the plurality of fiber strands of the fiber bundles 12A, 12B, and 12C is arranged into substantially equal strands in each of the six fiber bundles FG1 to FG6 in the strand distribution section 10a in the optical distribution section 10. Distributed to include players. The shape of each exit end (FBo) of the fiber bundles (FG1 to FG6) is shaped to be a rectangle similar to the shape of the illumination area (IAn) on the mask substrate M by bundling a plurality of fiber wires. One fiber bundle (FGn) is bundled so that approximately the same number of fiber strands from each of the fiber bundles (12A, 12B, 12C) on the incident side are included. For example, if each of the fiber bundles (12A, 12B, 12C) consists of 120,000 fiber wires (total of 360,000 fibers), one fiber bundle (FGn) bundles 60,000 fiber wires. It is composed. Among the 60,000 fiber strands of the fiber bundle (FGn), approximately 20,000 each are composed of fiber strands from each of the fiber bundles 12A, 12B, and 12C on the incident side. Additionally, one fiber strand is a quartz fiber with an external (clad) diameter of approximately 0.2 mm.

The fiber wire emits a light flux from the exit end while maintaining the numerical aperture (convergence angle or divergence angle) of the light beam irradiated to the incident end. Therefore, the numerical aperture (convergence angle) when the illumination luminous flux BMa irradiated to the incident end FBi of the fiber bundle 12A becomes the illumination luminous flux BSa from the exit end FBo of the fiber bundle FGn and is emitted. or divergence angle) is equal to the numerical aperture of the illumination beam (BMa), and the illumination beam (BMb) irradiated to the incident end (FBi) of the fiber bundle 12B is the emission end (FBo) of the fiber bundle (FGn). The numerical aperture (convergence angle or divergence angle) when the illumination beam BSb is emitted from becomes the same as the numerical aperture of the illumination beam BMb, and is irradiated to the incident end FBi of the fiber bundle 12C. When the illumination beam (BMc) becomes the illumination beam (BSc) and is emitted from the exit end (FBo) of the fiber bundle (FGn), the numerical aperture (convergence angle or divergence angle) is the same as the numerical aperture of the illumination beam (BMc). I do it. Therefore, the numerical apertures (convergence angles) of the illumination beams (BMa, BMb, BMc) irradiated to each incident end (FBi) of the fiber bundles (12A, 12B, 12C) are set to NAia, NAib, and NAic, and NAia When each of the magnification variable parts 8A, 8B, and 8C shown in FIG. 3 (FIG. 4) is adjusted so that = NAib = NAic, the illumination luminous flux emitted from the exit end (FBo) of each fiber bundle (FGn) ( The numerical apertures (divergence angles) of BSa, BSb, and BSc) are the same. When each of the magnification variable parts (8A, 8B, 8C) is adjusted so that the respective numerical apertures (convergence angles) (NAia, NAib, NAic) of the illumination luminous flux (BMa, BMb, BMc) are different from each other, each fiber bundle (FGn) ) The respective numerical apertures (divergence angles) of the illumination beams (BSa, BSb, BSc) emitted from the exit end (FBo) have different values.

[Second illumination optical system (ILn)]

FIG. 10 shows the illumination luminous fluxes (BSa, BSb, BSc) from each exit end (FBo) of the six fiber bundles (FGn) (FG1 to FG6) shown in FIG. 3 (FIG. 9) on the mask substrate (M). It is a perspective view schematically showing the configuration of the second illumination optical system ILn (IL1 to IL6) that irradiates each illumination area IAn, and the orthogonal coordinate system XYZ is set the same as in Fig. 3 or Fig. 9. The second illumination optical system ILn is configured so that the multiple point light source images formed at the emitting end (FBo) of the fiber bundle (FGn) are the light source images of Köhler illumination, and the position of the front focus coincides with the emitting end (FBo). A first condenser lens system (CFn) (CF1 to CF6) arranged, a fly-eye lens system (FEn) (FE1 to FE6) whose entrance surface (poi) is set at the position of the rear focus of the condenser lens system (CFn), and a fly-eye lens system (FEn) (FE1 to FE6) The position of the front focus is set on the exit surface (epi) of the fly-eye lens system (FEn) so that the light source image (secondary light source image) formed on the exit surface (epi) of the lens system (FEn) is the light source image of Köhler illumination. , and a second condenser lens system (CPn) (CP1 to CP6) in which an illumination area (IAn) (IA1 to IA6) is set at the rear focal point.

The condenser lens system (CFn) and the condenser lens system (CPn) are arranged along the optical axis (AX2) parallel to the Z axis, and the optical axis (AX2) is aligned with the geometric center point of the rectangular exit end (FBo) of the fiber bundle (FGn). , is set to pass through the geometric center point within the XY plane of the FlyEye lens system (FEn). The FlyEye lens system (FEn) is a lens element with a rectangular cross-section with the Y direction as the long side and the It is constructed by joining the plurality of (Le) in the X and Y directions like stacking bricks. A convex surface (spherical lens) having a predetermined focal length is formed on each of the incident surface (poi) side and the exit surface (epi) side of the lens element Le. In addition, the exit surface (epi) of the fly's eye lens system (FEn) is at the position of the illumination pupil of the second illumination optical system (ILn), and the overall external shape range in the XY plane of the fly's eye lens system (FEn) is approximately The lighting is set to include the diameter of the pupil (circular).

In addition, the exit surface (epi) of the fly's eye lens system (FEn) is set to an optically conjugate relationship (imaging relationship) with the exit end (FBo) of the fiber bundle (FGn), and the entrance surface ( poi) is set to have an optically conjugate relationship (imaging relationship) with the illumination area IAn (pattern surface of the mask substrate M). Therefore, a plurality of point light source images formed at the emission end (FBo) of the fiber bundle (FGn) are reimaged on the emission surface (epi) side of each of the plurality of lens elements (Le) of the fly-eye lens system (FEn). And the illumination area IAn is illuminated (imaged) in a shape similar to the rectangular cross-sectional shape of the lens element Le.

Figures 11(A) and 11(B) schematically show the state of the illumination light flux in the optical path from the exit end (FBo) of the fiber bundle (FGn) to the fly-eye lens system (FEn) shown in Figure 10. This is a drawing, and the rectangular coordinate system XYZ is set the same as in FIG. 10. FIG. 11(A) is a view of the optical path viewed from the Y-axis direction (step movement direction), and FIG. 11(B) is a view of the optical path viewed from the X-axis direction (scanning movement direction). Here, a fine circular light emitting point (0.2 diameter of less than mm) is converted to spot light (point light source image) (SPa, SPb, SPc). Additionally, the numerical apertures of the illumination beams (BSa, BSb, BSc) from each of the spot lights (SPa, SPb, SPc) are made the same. Therefore, the diffusion angle from each central ray of the illumination beams (BSa, BSb, BSc) parallel to the optical axis AX2 between the exit end (FBo) of the fiber bundle (FGn) and the first condenser lens system (CFn) is Both directions and Y directions have the same angle (θbo).

Illumination luminous fluxes (BSa, BSb, BSc) from each of the plurality of spot lights (SPa, SPb, SPc) formed at the exit end (FBo) of the fiber bundle (FGn) are generated by the first condenser lens system (CFn), As shown in FIGS. 11(A) and 11(B), they all overlap on the incident surface (poi) of the fly-eye lens system (FEn), and the incident surface (poi) is illuminated with a uniform illuminance distribution. Accordingly, the fiber bundle (FGn) and the first condenser lens system (CFn) function as the first optical integrator for the fly-eye lens system (FEn).

FIG. 12 schematically shows an example of the arrangement of a large number of spot lights (point light source images) (SPa, SPb, SPc) formed for each fiber strand at the exit end (FBo) of the fiber bundle (FGn) shown in FIG. 11. This is the drawing shown, and the rectangular coordinate system XYZ is set the same as in FIG. 11. At the rectangular exit end (FBo) in the Y direction, there is a spot light (point light source image) (SPa) indicated by a white circle, a spot light (point source image) (SPb) indicated by a black circle, and a spot light indicated by a double circle. (Point light source image) (SPc) are arranged in equal numbers in uniform distribution in the X and Y directions. In Fig. 12, three spot lights (SPa, SPb, SPc) are shown to be distributed regularly (periodically) in the XY direction, but in reality, they are distributed randomly and densely. As previously exemplified, when each of the fiber bundles 12A, 12B, and 12C on the incident side is composed of 120,000 fiber wires, spot light (SPa, SPb) is applied to the emission end (FBo) of the fiber bundle (FGn). , SPc), each of which is about 20,000 randomly distributed. As an example, the ratio of the dimensions of the exit end (FBo) in the XY direction, the ratio of the dimensions of one lens element (Le) in the fly-eye lens system (FEn) in the When the ratios are all set to approximately 1:3 and the external diameter of the fiber wire is 0.2 mm, there are approximately 143 fiber wires in the X direction and approximately 420 fiber wires in the Y direction of the exit end (FBo) (the total number is 143 × 420 ≒ 60,000) are arranged. In this case, the X-direction dimension of the injection end FBo is approximately 28.6 mm (0.2 mm × 143), and the Y-direction dimension is approximately 84 mm (0.2 mm × 420).

FIG. 13 shows a plurality of point light source images (spot lights SPa', SPb', This is a diagram showing the arrangement state of SPc')), and the orthogonal coordinate system XYZ is set the same as in FIG. 11 (or FIG. 12). In Fig. 13, a plurality of spot lights SPa', SPb', and SPc' formed on the emission surface epi of each lens element Le are formed on the emission end FBo of the fiber bundle FGn. A large number of spot lights (SPa, SPb, SPc) are re-imaged, and about 60,000 spot lights (SPa', SPb', SPc') are formed for each lens element (Le). Accordingly, countless spot lights (SPa', SPb', SPc') corresponding to the number of lens elements (Le) x approximately 60,000 are distributed across the entire emission surface (epi) of the FlyEye lens system (FEn).

FIGS. 14(A) and 14(B) are diagrams schematically showing the state of illumination light in the optical path from the fly-eye lens system (FEn) shown in FIG. 10 to the illumination area (IAn) on the mask substrate (M). , and the rectangular coordinate system XYZ is set the same as in FIG. 10 (or FIG. 11). FIG. 14(A) is a view of the optical path from the fly-eye lens system (FEn) to the illumination area (IAn) as seen from the This is a diagram showing the optical path to the area IAn as seen from the Y direction (step movement direction). Among the countless spot lights (SPa', SPb', SPc') formed on the emission surface (epi) of the FlyEye lens system (FEn), as shown in FIG. 14(A), the one that is furthest from the optical axis AX2 in the Y direction The illumination luminous fluxes (BSa', BSb', BSc') radiating from the spot lights (SPa', SPb', SPc') located at a distance (ΔHy) are converted into parallel luminous fluxes by the second condenser lens system (CPn). At the same time, the central ray (main ray) is inclined by an angle θhy from the optical axis AX2 and is projected to the entire Y direction of the illumination area IAn. The illumination luminous fluxes (BSa', BSb', BSc') radiating from each of the different spot lights (SPa', SPb', SPc') lined up in the Y direction on the exit surface (epi) of the FlyEye lens system (FEn) are also shown. , is converted into a parallel light flux with respect to the Y direction by the second condenser lens system CPn, and is projected (overlaid) on the entire Y direction of the illumination area IAn.

Among the countless spot lights (SPa', SPb', and SPc') formed on the exit surface (epi) of the FlyEye lens system (FEn), as shown in FIG. 14(B), the one furthest from the optical axis AX2 in the The illumination luminous fluxes (BSa', BSb', BSc') radiating from the spot lights (SPa', SPb', SPc') located at a distance (ΔHx) and traveling are converted into parallel luminous fluxes by the second condenser lens system (CPn). At the same time, the central ray (main ray) is inclined by an angle θhx from the optical axis AX2 and is projected to the entire X direction of the illumination area IAn. The illumination luminous flux (BSa', BSb', BSc') radiating from each of the different spot lights (SPa', SPb', SPc') lined up in the X direction from the exit surface (epi) of the FlyEye lens system (FEn) , is converted into a parallel light flux with respect to the Accordingly, the fly-eye lens system (FEn) and the second condenser lens system (CPn) function as a second optical integrator that irradiates the illumination area (IAn) with illumination light with a uniform illuminance distribution.

The angle θhy, which is the maximum inclination angle in the Y direction, and the angle θhx, which is the maximum inclination angle in the set, the illumination luminous fluxes BSa', BSb', BSc' have an isotropic diffusion angle θi (= θhy = θhx) around the chief ray parallel to the axis AX2 and perpendicular to the illumination area IAn. . Accordingly, the numerical aperture (NAi) of the illumination beams (BSa', BSb', BSc') irradiated to the illumination area (IAn) becomes sin(θi). In addition, as is clear from Fig. 14(A) and Fig. 14(B), the circular irradiation area of the illumination beam (BSa, BSb, BSc) irradiated on the incident surface (poi) of the fly-eye lens system (FEn) If the radius from the optical axis (AX2) is reduced, among the countless spot lights (SPa', SPb', SPc') formed on the emission surface (epi) of the FlyEye lens system (FEn), the distance from the optical axis (AX2) is the furthest from the optical axis (AX2). Since ΔHx, ΔHy) also become shorter, the diffusion angle (θi) (＝ θhy = θhx) also becomes smaller, and as a result, the numerical aperture (NAi) of the illumination luminous flux (BSa', BSb', BSc') becomes smaller, and the illumination σ The value also becomes smaller.

[Function of magnification variable section (8A, 8B, 8C)]

15(A) and 15(B) show the fiber bundles 12A (12B, 12C) on the incident side by the magnification variable portion (numerical aperture variable portion) 8A (8B, 8C) shown in Fig. 4. This is a diagram explaining how to adjust the numerical aperture (diffusion angle) of the illumination beam (BMa) (BMb, BMc) irradiated to the incident end (FBi). 15(A) and 15(B), the focus position PS2 is the mercury lamp 2A (2B, 2C) that has passed through the wavelength selection section 6A (6B, 6C) as shown in FIG. 4. ) is the plane on which the luminous fluxes (BMa) (BMb, BMc) converge (converge) to the smallest diameter, and at the focal position (PS2), there is a blurred image of the arc discharge part of the mercury lamp (2A) (2B, 2C). A circular light source image (LDa) is formed. By the lens system 8A1 (8B1, 8C1) and the lens system 8A2 (8B2, 8C2), the light source image LDa is projected to the light source on the incident end FBi of the fiber bundle 12A (12B, 12C) on the incident side. It is reimaged as an image (LDb).

Let the lens system (8A1) (8B1, 8C1) be a negative power (refractive power), and the lens system (8A2) (8B2, 8C2) be a positive power (refractive power), as shown in Figure 15(A). Similarly, when the lens system (8A1) (8B1, 8C1) and the lens system (8A2) (8B2, 8C2) are arranged to be spaced apart, the numerical aperture (diffusion angle) of the luminous flux (BMa) (BMb, BMc) forming the light source image (LDb) ) (NAα) becomes the maximum, and the light source image LDa is reimaged to have the smallest diameter on the incident end FBi of the fiber bundle 12A (12B, 12C). In addition, when the lens system 8A1 (8B1, 8C1) and the lens system 8A2 (8B2, 8C2) are arranged in close proximity as shown in FIG. 15(B), the luminous flux (BMa) (BMb, The numerical aperture (diffusion angle) NAβ of BMc) is minimized, and the light source image LDb is reimaged to have the largest diameter on the entrance edge FBi of the fiber bundle 12A (12B, 12C). By appropriately adjusting the position of the two lens systems 8A1 (8B1, 8C1) and the lens system 8A2 (8B2, 8C2) in the direction of the optical axis AX1, the light flux (BMa) (BMb) forming the light source image (LDb) , BMc), the numerical aperture (diffusion angle) can be adjusted between the maximum NAα and the minimum NAβ. Additionally, in the case of FIG. 15(A), the diameter of the light source image LDb may be set to be slightly smaller than the effective diameter of the entrance end FBi of the fiber bundle 12A (12B, 12C), and FIG. 15(B) In the case of , the diameter of the light source image LDb may be set to be slightly larger than the effective diameter of the incident end FBi of the fiber bundle 12A (12B, 12C).

Within the incident end (FBi) of the fiber bundle (12A) (12B, 12C), a plurality of fiber wires present in the range where the light source image (LDb) is formed, that is, the range where the light flux (BMa) (BMb, BMc) is irradiated. The luminous flux (BMa) (BMb, BMc) incident from each incident end of is, as explained in Figures 9 and 11, each of the fiber wires located at the emission end (FBo) of the fiber bundle (FGn) on the emission side. From the spot light (SPa, SPb, SPc) formed at the exit end of , BSb, BSc) and enter the civil service examination.

Each luminous flux (BMa, BMb, When the numerical aperture (diffusion angle) of BMc) is made the same, the respective numerical apertures (Figure The angle (corresponding to θbo) shown in 11) becomes the same value. However, if the numerical aperture (diffusion angle) of each luminous flux (BMa, BMb, BMc) on the incident side is changed by each of the magnification variable parts 8A, 8B, and 8C, the three rays emitted from the exit end FBo The respective numerical apertures of the illumination beams (BSa, BSb, BSc) can also be different. This is explained by Figure 16.

FIG. 16 shows the states of the light fluxes (BMa, BMb, BMc) incident on the fiber bundles 12A, 12B, and 12C on the incident side of the fiber bundle shown in FIG. 9, and the states of the fiber bundles FGn (FG1 to FG6) on the emission side. ) This is a diagram schematically showing the state of the illumination beam (BSa, BSb, BSc) emitted from. In Fig. 16, the numerical aperture (diffusion angle) of the light flux BMa projected to the incident end FBi of the fiber bundle 12A is NAia, and the light flux projected to the incident end FBi of the fiber bundle 12B is BMb. ) is set to NAib, the numerical aperture (diffusion angle) of the luminous flux (BMc) projected to the incident end (FBi) of the fiber bundle 12C is set to NAic, and the relationship NAia > NAib > NAic is set. It is assumed that it is done. In this case, the illumination beam (BSa) radiating and traveling from each of the plurality of spot lights (SPa) formed at the emission end (FBo) of the fiber bundle (FG1) on the emission side becomes the numerical aperture (NAia), and the plurality of The illumination luminous flux (BSb) radiating from each of the spot lights (SPb) and traveling becomes the numerical aperture (NAib), and the illumination luminous flux (BSc) radiating and traveling from each of the multiple spot lights (SPc) is the numerical aperture (NAic). ) becomes. Similarly, illumination beams BSa, BSb, and BSc with different numerical apertures are simultaneously emitted from each emission end FBo of the other fiber bundles FG2 to FG6.

Figure 17 illustrates the difference in irradiation distribution of the illumination beams (BSa, BSb, BSc) radiating from the exit end (FBo) of the fiber bundle (FGn) on the incident surface (poi) of the fly-eye lens system (FEn). To this end, it is a schematic diagram of the optical path from the exit end (FBo) to the incident surface (poi) of the fly-eye lens system (FEn) as seen from the X direction (scanning movement direction), and the orthogonal coordinate system It is set to. In FIG. 17, the illumination beam BSa radiating from each of the plurality of spot lights SPa formed at the exit end FBo (corresponding to the pupil plane) of the fiber bundle FGn and traveling is the condenser lens system CPn. It is converted into a substantially parallel light flux and irradiated by overlapping a circular area (CFa) centered on the optical axis (AX1) in the incident surface (poi) of the FlyEye lens system (FEn). Likewise, the illumination luminous flux BSb radiating from each of the plurality of spot lights SPb and traveling is converted into a substantially parallel luminous flux by the condenser lens system CPn, and is transmitted to the incident surface poi of the fly-eye lens system FEn. ), the illumination beam (BSc), which overlaps and irradiates a circular area (CFb) centered on the optical axis (AX1) in It is converted into a substantially parallel light flux, and is irradiated by overlapping a circular area (CFc) centered on the optical axis (AX1) within the incident surface (poi) of the Fly's Eye lens system (FEn).

The exit end (FBo) of the fiber bundle (FGn) is placed at the position (coplanar) of the front focus of the condenser lens system (CPn), and the entrance surface (poi) of the fly-eye lens system (FEn) is at the rear focus of the condenser lens system (CPn). Since it is a Köhler illumination method arranged at the position of The entire area CFa is illuminated, the illumination luminous flux BSb from the spot light SPb is irradiated throughout the circular area CFb, and the illumination luminous flux BSc from the spot light SPc is circular. The entire area (CFc) is irradiated.

FIG. 18 is a view of the circular areas (CFa, CFb, CFc) in FIG. 17 distributed on the incident surface (poi) of the fly-eye lens system (FEn) as seen within the XY plane, and the orthogonal coordinate system XYZ is shown in FIG. 17 It is set the same as . Since the numerical aperture (diffusion angle) (NAia, NAib, NAic) of the illumination luminous flux (BSa, BSb, BSc) is in the relationship NAia > NAib > NAic, as shown in FIG. 18, centering on the optical axis (AX2) If the radius of the area (CFa) is Ria, the radius of the area (CFb) is Rib, and the radius of the area (CFc) is Ric, the relationship becomes Ria > Rib > Ric. Additionally, within the area CFc of the radius Ric, all three illumination beams BSa, BSb, and BSc are distributed overlapping each other, and from the radius Ric to the radius Rib within the area CFb. In the area of the annular zone, two illumination luminous fluxes (BSa, BSb) are distributed overlapping each other, and in the area of the annular zone from the radius Rib to the radius Ria within the area CFa, the illumination luminous flux BSa ) are distributed only. In addition, the circular area (CCA) indicated by a broken line in FIG. 18 represents the boundary range where the illumination σ value is 1.0 (NAi = NAp), and the respective numerical apertures (NAia, NAia, The maximum value of NAib, NAic) is set to be less than or equal to the numerical aperture corresponding to the radius of the area (CCA). In addition, the positions separated from the optical axis AX2 in the Y direction and the This corresponds to the distances (ΔHy, ΔHx) explained in Fig. 14(B).

As described above, by adjusting each of the magnification variable portions (numerical aperture variable portions) 8A, 8B, and 8C shown in Figs. 4 and 15, a circular shape is formed on the incident surface (poi) of the fly-eye lens system (FEn). The radius (Ria, Rib, Ric) of each area (CFa, CFb, CFc) of the three illumination beams (BSa, BSb, BSc) distributed as The numerous spot lights SPa', SPb', and SPc' formed in the epi) can be given an intensity distribution according to the radial distance from the optical axis AX2.

Figures 19(A) and 19(B) show the exit surface of the FlyEye lens system (FEn), corresponding to the respective areas (CFa, CFb, CFc) of the illumination beam (BSa, BSb, BSc) shown in Figure 18. (epi) shows an example of the intensity distribution (light source image) of countless spot lights (SPa', SPb', SPc') formed in the (illumination plane). FIG. 19(A) is a view of the FlyEye lens system FEn from the X direction (scanning movement direction), and FIG. 19B is a view of the FlyEye lens system FEn from the Y direction (step movement direction). The countless spot lights (SPa') formed on the exit surface (epi) of the FlyEye lens system (FEn) are circular areas (CFa) (radius (Ria)) where the illumination beam (BSa) on the entrance surface (poi) is irradiated. The countless spot lights (SPb') generated in the portion corresponding to ), and the countless spot lights (SPc') generated in the portion corresponding to the exit surface (epi) are formed in the circular area (CFc) (radius (Ric)) where the illumination beam (BSc) on the entrance surface (poi) is irradiated. )) occurs in the part corresponding to .

When the wavelength characteristics of the three illumination beams (BSa, BSb, BSc) are the same, for example, an interference filter mounted on each of the wavelength selection units 6A, 6B, and 6C shown in Figs. 3 and 4, In the case of the i-line narrow interference filter (SWa) shown in Fig. 6, the radius (Ric) of the exit surface (epi) of the fly-eye lens system (FEn) (co-plane of the illumination system) corresponds to the circular area (CFc). In this part, all three spot lights (SPa', SPb', SPc') having the i-line (narrow) spectral distribution shown in FIG. 6 overlap to form one another. In addition, in the portion corresponding to the annular area from the radius (Ric) to the radius (Rib) of the emission surface (epi) (coplanar of the illumination system), two spot lights (SPa') having the spectral distribution of the i line (narrow) are provided. , SPb') is formed, and in the portion corresponding to the annular region from the radius (Rib) to the radius (Ria) of the emission surface (epi) (co-plane of the illumination system), 1 having the spectral distribution of the i line (narrow) Only two spots (SPa') are formed. In addition, a plurality of spot lights (point source images) (SPa', SPb') formed on the exit surface (epi) (co-plane of the illumination system) of the FlyEye lens system (FEn) shown in Figures 19(A) and 19(B). , SPc') is not evenly distributed within the circular area (CCA) in Fig. 18, which means that for the purpose of explaining the function of the magnification variable section (numerical aperture variable section) 8A, 8B, 8C, the fiber bundle ( This is because the numerical apertures (NAia, NAib, NAic) of the luminous fluxes (BMa, BMb, BMc) incident on each of 12A, 12B, and 12C) were intentionally set to the relationship NAia > NAib > NAic. In normal pattern exposure, the numerical apertures (NAia, NAib, NAic) of the luminous flux (BMa, BMb, BMc) are set in the relationship NAia = NAib = NAic.

19(A) and 19(B), a plurality of spot lights (point source images) (SPa', SPb', SPc') is distributed, and the wavelength characteristics of the luminous fluxes (BMa, BMb, BMc) (illuminating luminous fluxes (BSa, BSb, BSc)), which are the sources of each spot light (SPa', SPb', SPc'), are distributed. In the same case, the illumination light irradiated to the illumination area IAn of the mask substrate M has characteristics of different illuminance depending on the numerical aperture, as shown in FIG. 20 . Figure 20 schematically shows the orientation characteristics (diffusion angle characteristics) of the illumination beam Irn irradiated to the point OP on the illumination area IAn, and since it is a telecentric illumination condition (Köhler illumination), the point OP The principal ray (Lpi) of the illumination luminous flux (Irn) passing through is perpendicular to the plane of the illumination area (IAn) (pattern plane of the mask substrate M). The illumination luminous flux (Irn) is oriented so that the diffusion angle (θia) from the chief ray (Lpi) corresponding to the numerical aperture (NAia) is the maximum numerical aperture. Within this diffusion angle (θia), within the diffusion angle (θic) corresponding to the numerical aperture (NAic), the illuminance of the illumination luminous flux (Irn) is the intensity obtained by adding the three illumination luminous fluxes (BSa, BSb, BSc). , from the diffusion angle (θib) corresponding to the numerical aperture (NAib) to the diffusion angle (θic), the illuminance of the illumination luminous flux (Irn) becomes the intensity of adding the two illumination luminous fluxes (BSa, BSb), and From the diffusion angle θia to the diffusion angle θib, the illuminance of the illumination beam Irn becomes the intensity of only one illumination beam BSa. In other words, among the total diffusion angles (θia in FIG. 20) of the illumination luminous flux (Irn), the intensity of the diffusion angle (θic in FIG. 20) near the center is high, and a distribution in which the intensity decreases as the diffusion angle increases is given. You can.

In addition, each of the illumination areas (IAn) (IAn) (IA1 to IA6) on the mask substrate (M) has a conjugate relationship (imaging relationship) with each of the projection areas (EAn) (EA1 to EA6) on the plate (P). Therefore, the imaging light flux (diffracted light) for exposure projected at any one point in the projection area EAn has the same orientation characteristics (diffusion angle characteristics) as those in FIG. 20 .

In this way, the luminous flux (BMa, BMb) projected by the magnification variable portion (numerical aperture variable portion) 8A, 8B, 8C to each incident end FBi of the fiber bundles 12A, 12B, 12C on the incident side. By adjusting each numerical aperture (diffusion angle) of BMc), the overall numerical aperture (NAia in FIG. 20) of the illumination luminous flux Irn projected onto the illumination area IAn on the mask substrate M is changed to change the illumination σ. The value can be changed or the illuminance distribution can be provided within the range of the diffusion angle corresponding to the total numerical aperture of the illumination luminous flux (Irn). In addition, the magnification variable portions (numerical aperture variable portions) 8A, 8B, and 8C adjust the respective diameters of the light beams BMa, BMB, and BMc to the incident end FBi of the fiber bundles 12A, 12B, and 12C. Since it can be made larger or smaller with respect to the diameter of ) can also be adjusted.

[Function of wavelength selection unit (6A, 6B, 6C)]

Each slide mechanism FX of the wavelength selection portions 6A, 6B, and 6C shown in FIGS. 3 and 4 above includes, for example, 3 slide mechanisms having wavelength selection characteristics as shown in each of FIGS. 6 to 8. You can select the wavelength by installing one of several interference filters (SWa, SWb, SWc) interchangeably. In this embodiment, the wavelength characteristics of the illumination light irradiating the illumination area IAn of the mask substrate M are changed to a mask by combining interference filters mounted on each of the three wavelength selection units 6A, 6B, and 6C. It is possible to adjust according to the characteristics (characteristics) of the pattern to be exposed on the substrate (M).

FIG. 21 shows the i-line-narrow interference filter (SWa) of FIG. 6, the i-line-broadband interference filter (SWb) of FIG. 7, and the i-line-broadband interference filter (SWb) of FIG. 8 mounted on each of the three wavelength selection units 6A, 6B, 6C. This table summarizes examples of combinations of i line + h line-interference filter (SWc). In the table in Figure 21, the column at the left end is a code that names the combination of three types of interference filters (SWa, SWb, SWc), and the three rows on the right are the wavelength spectrum i-line (narrow), i-line (optical), The number of ○ marks written in the i line + h line column indicates the number of mercury lamps generating that wavelength spectrum. In addition, in the following description using FIG. 21, each of the three magnification variable parts 8A, 8B, and 8C is projected to each incident end FBi of the fiber bundles 12A, 12B, and 12C on the incident side. It is assumed that each numerical aperture (NAia, NAib, NAic) of the luminous flux (BMa, BMB, BMc) is set to the same value.

In Figure 21, the filter combination codes A0, A1, A2, A3, A4, and T must be used in any one of the three wavelength selection units (6A, 6B, and 6C) as an i-line narrow interference filter (SWa). is a combination in which the i-line narrow interference filter (SWa) is not equipped in any of the three wavelength selectors 6A, 6B, 6C, and the code B0, B1, B2 of the filter combination is the i-line- It is a combination in which a wide interference filter (SWb) and an i-line + h-line-interference filter (SWc) are installed, and the code C0 of the filter combination is an i-line + h-line in all three wavelength selection units (6A, 6B, 6C). -This is a combination in which an interference filter (SWc) is installed. In these combinations, the code A0 is the illumination area IAn on the mask substrate M, since all three wavelength selectors 6A, 6B, 6C are equipped with i-line narrow interference filters (SWa). The light quantity of the illumination luminous flux (Irn) is obtained as a value obtained by multiplying only the light quantity of the i-line (narrow) spectral distribution (see FIG. 6) from each of the three mercury lamps 2A, 2B, and 2C by approximately three times, and is relatively High-resolution pattern exposure becomes possible under high illuminance. In addition, code A1 is equipped with an i-line narrow-band interference filter (SWa) in two of the three wavelength selection units (6A, 6B, 6C) and an i-line-broadband interference filter (SWb) in the remaining one. In one case, the amount of illumination luminous flux Irn in the illumination area IAn on the mask substrate M is the spectral distribution of the i line (narrow) from two of the three mercury lamps 2A, 2B, and 2C ( It is obtained by adding approximately twice the amount of light (see FIG. 6) and the amount of light of the spectral distribution (see FIG. 7) of the i line (light) from one of the three mercury lamps (2A, 2B, 2C), and the code Compared with the combination of A0, the amount of illumination luminous flux (Irn) increases by several percent while maintaining the performance of high-resolution pattern exposure.

In Fig. 21, the combination code T means that each of the three wavelength selection units 6A, 6B, and 6C is equipped with a separate interference filter (SWa, SWb, SWc), and the combination code B0 means 3 means that only the i-line-wide interference filter (SWb) is equipped in all of the wavelength selection units (6A, 6B, 6C), and the combination code C0 means that all of the three wavelength selection units (6A, 6B, 6C) are equipped with In all, it means that only the i-line + h-line-interference filter (SWc) is equipped. As is clear from the table in FIG. 21, for any combination code, the illumination luminous flux Irn irradiated to the illumination area IAn has almost the bright line component of the i line from each of the three mercury lamps 2A, 2B, and 2C. 100% included. However, depending on the combination method of the interference filters (SWa, SWb, SWc), for example, the intensity ratio of the bright line component of the i line and the bright line component of the h line may be different from the original intensity ratio of the mercury lamp (see Fig. 5). This can be done with one spectral distribution.

FIG. 22 is a graph schematically showing the wavelength characteristics of the illumination luminous flux (Irn) obtained by combination code B2 in the table in FIG. 21. In combination code B2, one of the three wavelength selection sections (6A, 6B, 6C) is equipped with an i-line wide interference filter (SWb), and each of the remaining two wavelength selection sections is equipped with an i-line + h-line interference filter. (SWc) is installed. In this case, the wavelength spectrum distribution of the illumination luminous flux (Irn) is obtained by adding the light quantity obtained by doubling the spectral distribution of the i line + h line shown in FIG. 8 and the light quantity of the spectral distribution of the i line (light) shown in FIG. 7. do. Therefore, in the case of combination code B2, the light quantity of the bright line component of the i line is three times the light quantity of one mercury lamp, the light quantity of the bright line component of the h line is twice the light quantity of one mercury lamp, and the light quantity of the illumination luminous flux (Irn) is Changing the light quantity balance of the bright line component of the i line and the bright line component of the h line in the wavelength spectrum distribution, that is, changing the spectral characteristics of the illumination luminous flux (Irn) from the trend of the spectral characteristics of light from the mercury lamp (see Figure 5). It is possible.

In addition to the above embodiment, a wavelength selection unit ( 6A, 6B, 6C), light with a spectral distribution including a specific bright line wavelength selected by an illumination optical system (FIG. 3) is illuminated on the mask substrate M carrying a pattern for an electronic device (IAn) ( IA1 to IA6) is irradiated and the pattern is imaged by a projection optical system (partial projection optical system (PL1 to PL6)) that enters the exposure light flux (imaging light flux) generated from the mask substrate M (illumination area IAn). When projecting and exposing a light-sensitive substrate (plate P), a wavelength band is selected from the light generated from the light source device by the wavelength selection unit, as shown in the combination codes A1 to A4, T, and B0 to B2 in FIG. 21. Light of a different first spectral distribution (e.g., a spectral component selected from the i-line narrow interference filter (SWa)) and light of a second spectral distribution (e.g., i-line + h-line-interference filter (SWc)) to extract at least two of the spectral components selected from the co-plane (emission plane (epi) of the fly-eye lens system (FEn)) in the illumination optical system to perform Köhler illumination of the mask substrate (M) by the illumination optical system, A first light source image distributed in a two-dimensional range by light of the first spectral distribution (e.g., a set of multiple point light source images (SPa') in FIG. 13), and 2 by light of the second spectral distribution. By overlapping and forming a second light source image distributed in a dimensional range (e.g., a set of multiple point light source images SPb' in FIG. 13), as explained in FIG. 20, a mask substrate M is formed. An exposure method in which the balance of wavelength and intensity (wavelength intensity characteristics) is varied depending on the angle within an angular range (incident angle (θia) in Figure 20) corresponding to the maximum numerical aperture of the illumination beam (Irn) irradiated. This becomes possible.

[Affiliation between wavelength selection section and magnification variable section]

In the above description, the angles of the light beams BMa, BMb, BMc projected to each incident end FBi of the fiber bundles 12A, 12B, and 12C are set by each of the magnification variable parts 8A, 8B, and 8C. Although the numerical apertures (NAia, NAib, NAic) were set to the same value, each numerical aperture (NAia, NAib, NAic) of the luminous flux (BMa, BMb, BMc) was set to different values, and the wavelength selection units 6A, 6B, 6C ) By varying the interference filters mounted on the It can be granted. For example, the numerical apertures (NAia, NAib, NAic) shown in Fig. 20 are expressed as the relationship NAia = NAib > NAic (a relationship in which the radius (Rib) shown in Figs. 18 and 19 is equal to the radius (Ria)) 21, an i-line narrow interference filter (SWa) is installed in each of the wavelength selection units 6A and 6B, and an i-line + h-line interference filter is installed in the wavelength selection unit 6C, as shown in the combination code A2 in FIG. Install the filter (SWc). In that case, on the emission surface (epi) of the fly-eye lens system (FEn) shown in FIG. 19, the i-line (narrow) spectral distribution ( In Fig. 6), countless spot lights (SPa', SPb') are lined up uniformly, and within the circular area (CFc) of radius Ric, there are also countless spots with a spectral distribution of i line + h line (Fig. 8). Spot light (SPc') is lined up evenly.

Therefore, according to the present embodiment, the secondary light source image (numerous spot lights SPa', SPb', The wavelength characteristics within the distribution range of the collective image (SPc') can be changed depending on the radial position from the optical axis AX2. In this case, the light ray within the range (numerical aperture (NAic)) of the diffusion angle θic from the principal ray Lpi of the illumination beam Irn illuminating the mask substrate M shown in FIG. 20 includes both the i line and the h line. The spectrum including the bright line wavelength is included, and rays within the annular range (numerical aperture (NAic ~ NAia)) between the diffusion angle (θic) and the diffusion angle (θia) (= θib) include the i line (light). Only the spectrum is included. In this way, when the wavelength characteristics of the secondary light source image formed on the illumination coplane of the second illumination optical system ILn are changed in the radial direction, the pattern formed on the mask substrate M becomes a halftone pattern or a phase shift pattern. It becomes possible to suppress deterioration in the quality of the projected image due to influences such as pattern manufacturing errors.

In general, halftone patterns and phase shift patterns are assumed to be used under irradiation of illumination light of a specific wavelength, and a shift layer whose film thickness is managed so that the amplitude transmittance at that specific wavelength is under a predetermined condition is used as a mask. It is manufactured by forming it on a substrate. However, if an error occurs in the film thickness, or if the numerical aperture (illumination σ value) of the illumination light is changed, the amplitude transmittance by the shifter layer will fluctuate (deteriorate) from the desired condition, and the contrast of the pattern image to be projected and exposed will change. Imaging performance is deteriorated, such that the target fineness is not obtained or the target fineness is not obtained. In the present embodiment, even when using a mask substrate with such a halftone pattern or phase shift pattern, the illumination optical system (second illumination optical system ILn) that illuminates the mask substrate M is formed two-dimensionally on the same plane. Since the wavelength characteristics (spectrum) of the light source image can be changed in the radial direction, even when an error occurs in the film thickness of the shifter layer or when the numerical aperture (illumination σ value) of the illumination light is changed, the amplitude of the shifter layer It becomes possible to suppress the decline in imaging performance resulting from fluctuation (deterioration) in transmittance.

[Variation 1]

As mentioned above, in the first embodiment, the three mercury lamps (2A, 2B, 2C) are ultra-high pressure mercury discharge lamps with the same specifications, and the bright line wavelength of the i line and the bright line wavelength of the h line are mainly used for pattern exposure, but additional The bright line wavelength of the g-line may be used for pattern exposure. In this case, a projection optical system with chromatic aberration corrected in a wide wavelength range including three bright lines: i-line, h-line, and g-line is used. In addition, as disclosed in Japanese Patent Application Laid-Open No. 2012-049332, a projection exposure device equipped with a mirror projection projection optical system combining a large concave mirror and a small convex mirror is also provided, including an illumination device according to the present embodiment ( A first illumination optical system and a second illumination optical system (ILn) can be applied. Since the projection optical system of the mirror projection method does not use lens elements with strong refractive power, chromatic aberration due to differences in the wavelength of the illumination light rarely occurs, and the three bright line wavelengths of the mercury lamp, i-line, h-line, and g-line, are used. It can be used easily. In addition, although the three mercury lamps (2A, 2B, and 2C) were ultra-high pressure mercury discharge lamps with the same specifications, due to the wavelength characteristics of the light from the arc discharge section, the ratio of the respective peak intensities of the i-line, h-line, and g-line was A high-pressure mercury discharge lamp different from the ratio shown in 5 may be combined with an ultra-high-pressure mercury discharge lamp, and in some cases, a short-arc type low-pressure mercury discharge lamp and an ultra-high-pressure mercury discharge lamp may be combined. The number of first illumination optical systems from the mercury lamp 2 to the fiber bundle 12 on the incident side may be 2 or more. For example, when the number of partial projection optical systems (PLn) is 6 or more, the illumination luminous flux ( In order to secure the illuminance of Irn), four mercury lamps (2A to 2D), four first illumination optical systems, and four fiber bundles (12A to 12D) on the incident side can be formed.

[Variation 2]

The interference filter that can be mounted on the wavelength selection units 6A, 6B, 6C shown in FIGS. 3 and 4 above is an i line-narrow interference filter (SWa), i, having the wavelength characteristics shown in each of FIGS. 6 to 8. There are three types: line-broad interference filter (SWb) and i-line + h line-interference filter (SWc). However, when the projection optical system (partial projection optical system (PLn)) can be used up to the g-line wavelength, the g-line narrow band is used. An interference filter, g-line-wideband interference filter, or ultra-wideband i-line + h-line + g-line - interference filter can be prepared and mounted on the slide mechanism (FX). Additionally, an h-line narrow interference filter or an h-line wide interference filter may be prepared so as to include only the bright line wavelength of the h line. When an interference filter containing only the bright line wavelength of the h line is prepared, for example, either the i line-narrow interference filter (SWa) or the i-line wide interference filter (SWb) is used in each of the wavelength selection units 6A and 6B. , and either the h line-narrow interference filter or the h line-broadband interference filter is mounted on the wavelength selection unit 6C. Then, each of the magnification variable parts 8A, 8B, and 8C is adjusted to determine the numerical aperture (NAia) of the luminous flux (BMa) (i line) incident on the fiber bundle 12A and the numerical aperture (NAia) incident on the fiber bundle 12B. The numerical aperture (NAib) of the luminous flux (BMb) (i line) is set to the same value so that the illumination σ value is a large value (for example, 0.7 or more), and the luminous flux (BMc) incident on the fiber bundle 12C is ( The numerical aperture (NAic) of line h) is set so that the relationship is NAia = NAib > NAic.

In this case, the secondary light source image (a set of countless spot lights (SPa', SPb', SPc') formed on the exit surface (epi) of the fly's eye lens system (FEn), which is the illumination coplanar of the second illumination optical system (ILn) Image) contains countless spot lights (SPa) containing only the bright line wavelengths of the i lines scattered throughout the area (CFa) (=CFb) of the radius (Ria) (=Rib) corresponding to the numerical aperture (NAia) (=NAib). ', SPb') and countless spot lights (SPc') containing only the bright line wavelength of the h line scattered only within the area (CFc) of the radius (Ric) corresponding to the numerical aperture (NAic). Accordingly, the secondary light source image formed on the exit surface (epi) of the FlyEye lens system (FEn) has an intensity that is approximately constant throughout the area (CFa) of radius (Ria) (corresponding to the maximum numerical aperture (NAia)). It is set as a wavelength distribution characteristic that includes the spectral component of the bright line wavelength of the h line distributed only within the region CFc of the inner radius Ric (<Ria), along with the spectral component of the bright line wavelength of the i line distributed.

In addition, when an h line-narrow interference filter or an h line-broad interference filter is prepared, 2 formed on the illumination coplane of the second illumination optical system ILn by adjusting each of the magnification variable parts 8A, 8B, and 8C. The wavelength distribution characteristics of the differential light source image may be set to the opposite setting as above. That is, the entire area (CFa) (corresponding to the maximum numerical aperture (NAia)) of the radius (Ria) of the secondary light source image formed on the illumination pupil (emission surface (epi)) of the second illumination optical system ILn. The spectral component of the bright line wavelength of the h line may be used, and only the area (CFc) of the inner radius (Ric) (< Ria) may be used as the spectral component of the bright line wavelength of the i line. Additionally, the interference filter is a band-pass filter that extracts spectral components of a predetermined wavelength width, but is composed of a low-pass filter that transmits wavelength components longer than the cutoff wavelength and a high-pass filter that transmits wavelength components shorter than the cutoff wavelength in series. The array may be mounted between the lens systems 6A1 and 6A2. In that case, prepare a low-pass filter whose cutoff wavelength is set to around 350 nm to 360 nm, a first high-pass filter whose cutoff wavelength is about 375 nm, and a second high-pass filter whose cutoff wavelength is about 395 nm, Install the 1 high-pass filter and the 2nd high-pass filter interchangeably. Accordingly, in the combination of the low-pass filter and the first high-pass filter, the spectral component of the i line (narrow) as shown in FIG. 6 is extracted, and in the combination of the low-pass filter and the second high-pass filter, the spectral component as shown in FIG. 7 The spectral components of the i line (light) as shown are extracted.

[Variation Example 3]

In the manufacturing stage of devices such as display panel substrates or circuit boards for electronic component mounting, or in the manufacturing stage of a fine metal mask (so-called stencil mask) mounted in a deposition apparatus to partition the deposition portion on the substrate to be processed, etc. There are cases where pattern exposure is performed on the negative photoresist layer (light-sensitive layer) applied to the plate P with a thickness several to ten times the normal thickness (0.5 to 1.5 μm). Although negative-type photoresists often have lower light sensitivity than positive-type photoresists, they have the characteristic that the area irradiated with the illumination luminous flux (Irn) for exposure becomes insoluble in the developer and leaves a residual film. Additionally, negative photoresists sometimes have large differences in sensitivity or absorption rate to the wavelength of the illumination luminous flux (Irn) for exposure. Figure 23 is a graph showing an example of the light absorption characteristics of a negative photoresist with the wavelength (nm) of the illumination light flux (Irn) taken on the horizontal axis and the normalized absorption rate (0 to 1) taken on the vertical axis. In the case of the photoresist in Figure 23, there is a peak of absorption around a wavelength of 320 nm, and the absorption rate has the characteristic of decreasing approximately linearly between the wavelengths of 320 nm and 450 nm (wavelength dependence of the absorption rate), and the bright line wavelength of the i line is 365. The absorption rate at nm is about 0.5, and the absorption rate at the bright line wavelength of 405 nm of the h line is about 0.15. The characteristics in Fig. 23 are just an example and vary greatly depending on the material of the resist. When the thickness of the negative photoresist layer having the characteristics as shown in Figure 23 is 10 ㎛ or more, when the pattern is exposed with an illumination beam (Irn) containing both the bright line wavelength of the i line and the bright line wavelength of the h line, the wavelength dependence of the absorptivity As a result, light at the bright line wavelength of the i line is largely absorbed in the surface portion of the resist layer, and a sufficient amount of light is not provided to the bottom side (plate P side) of the resist layer. In contrast, light with a bright line wavelength of the h line is less absorbed by the resist layer, so a sufficient amount of light is provided to the bottom side (plate P side) of the resist layer.

Since the resist layer has a large thickness and the absorption rate depends on the wavelength, the i-line + h-line interference filter (SWc) shown in Fig. 8 is installed in each of the wavelength selection sections 6A, 6B, and 6C ( Fig. 21 When the combination code C0 in the table is selected and the pattern of the mask substrate M is projected and exposed on the plate P, the edge portion (side wall) of the pattern (resist image) of the resist layer remaining after development is exposed to the plate. It can be made inclined rather than perpendicular to the surface of (P). Fig. 24 is a cross-sectional view schematically showing the inclination of the edge portion (side wall) of the resist image remaining after development. In Fig. 24, a negative resist layer Luv is formed on the surface of the plate P (here, a metal film such as nickel is formed on the surface) with a thickness RT (10 μm or more), and after development, the resist layer The unexposed portion (non-irradiated portion) of (Luv) is removed to form an opening (HL) sandwiched between the edge portions (Ewa, Ewb). When manufacturing a fine metal mask, a metal layer (nickel, copper, etc.) is deposited on the plate P exposed through the opening HL by electrolytic plating. The side walls that become the edge portions (Ewa, Ewb) of the resist layer (Luv) are formed here inclined toward the opening (HL) side, so-called inverse taper shape.

In this way, in order to control the inclination amount of the side wall that becomes the edge portion (Ewa, Ewb) of the resist image to a desired value, the amount of light of the bright line wavelength of the i line included in the illumination luminous flux (Irn) (slanted line in FIG. 6 or FIG. 7 The balance of the light quantity of the bright line wavelength of the h line (corresponding to the negative area) is adjusted by a combination of interference filters mounted on each of the wavelength selection units 6A, 6B, and 6C, or the magnification variable units 8A, 8B, By each of 8C), it becomes possible to independently adjust the numerical aperture of the illumination light flux of the bright line wavelength of the i line and the numerical aperture of the illumination light flux of the bright line wavelength of the h line included in the illumination light flux Irn. Additionally, imparting a desired inclination amount to the side wall that becomes the edge portion (Ewa, Ewb) of the resist image after development is not limited to negative photoresist, and is equally possible with positive photoresist.

When the resist layer (Luv) is used for masking in the plating process when manufacturing a fine metal mask or forming a wiring layer, use photoresists for plating sold by Tokyo Ohka Industry Co., Ltd. under the product names PMER P-CS series, PMER P- LA series, PMER P-HA series, PMER P-CE series, or naphthoquinone type or chemically amplified photoresist of PMER P-WE series and PMER P-CY series, negative type with product name PMER-N-HC600PY. Photoresist, etc. can be used. In addition, plating resists sold by San-Ei Chemical Co., Ltd. under the product names SPR-558C-1 and SPR-530CMT-A can also be used. In addition, it has an appropriate light absorption rate in the wavelength range of the illumination luminous flux (Irn) during pattern exposure, and can be used as an ultraviolet curing monomer/oligomer (epoxy acrylate, urethane acrylate, polyester acrylate), photopolymerization initiator, and photosensitizer. , an ultraviolet curable resin containing additives, etc. may be used as the photosensitive layer instead of the resist layer (Luv).

[Variation Example 4]

If only the i-line-narrow interference filter or the i-line-wide interference filter is used, and the illumination luminous flux (Irn) for exposure is light containing only the bright line wavelength of the i-line, high resolution during pattern exposure becomes possible; As the wavelength of light becomes shorter, the depth of focus (DOF) also decreases. Therefore, in order to suppress the reduction of DOF in a high resolution state, the shape of the light source image (secondary light source image) formed on the illumination pupil of the illumination optical system is irregular, or the position (area) is point symmetrical around the optical axis in the illumination pupil. ) In some cases, it is ubiquitous in the 4th superfine. In that case, an aperture plate (illumination aperture stop) on which an annular or tetrapolar light transmitting portion is formed is installed at or near the exit surface (epi) of the fly-eye lens system (FEn).

25(A) and 25(B) are diagrams schematically showing the shapes in the , and the Cartesian coordinate system XYZ is aligned with FIG. 18 above. The aperture plate APa is formed by removing a light-shielding layer such as chromium deposited on the surface of a parallel plate of quartz in an annular shape by etching to form an annular light transmitting portion TPa as shown in FIG. 25(A). Similarly, for the aperture plate APb, the light-shielding layer on the surface of the parallel plate of quartz is removed in four quadrants by etching, and the four quadrants of the A fan-shaped light transmitting portion (TPb) is formed on each of them. Additionally, the aperture plate APb may only be a light-shielding portion in which light-shielding bands extending in the X and Y directions intersect in a square shape at the position of the optical axis AX2.

[Variation Example 5]

In the wavelength selection section 6A (6B, 6C) included in the first illumination optical system shown in FIG. 4 above, a beam radiating from the position PS1 of the second focus of the elliptical mirror 4A (4B, 4C) proceeds. A lens system (collimator lens) 6A1 that receives the incident light beam BM and converts it into a substantially parallel light beam, and a lens system 6A2 that converges the substantially parallel light beam at the focal position PS2 are formed. In the optical path between the lens systems 6A1 and 6A2, one of the interference filters (SWa, SWb, SWc), etc. is installed, but a ring-shaped aperture plate as shown in Fig. 25(A) may also be installed. FIG. 26 is a diagram showing an annular aperture plate APa' arranged in the wavelength selection section 6A of the first illumination optical system, and the same members as those shown in FIG. 4 are given the same symbols.

In Fig. 26, the annular aperture plate APa' has a peripheral light-shielding layer that shields the outside of the outer ring diameter defined corresponding to the maximum diameter of the luminous flux BM, which has become a substantially parallel luminous flux by the lens system 6A1, and It is constructed by forming a circular central light-shielding layer on a quartz flat plate that shields the inside of the inner ring diameter centered on the optical axis AX1. The aperture plate APa', like the interference filter SWa (or SWb, SWc, etc.), is mounted on a slide mechanism FX and is removably installed in the optical path. The illumination luminous flux (BMa) passing through the annular light transmitting portion (TPa) of the annular aperture plate (APa') converges at the focal position (PS2) by the lens system 6A2 and then diverges again to reach the magnification variable section at the rear end. Head to (8A). The outer ring diameter of the aperture plate (APa') defines the maximum numerical aperture (NAd1) of the illumination luminous flux (BMa), and the inner ring diameter of the aperture plate (APa') is circular within the cross section of the illumination luminous flux (BMa). The numerical aperture (NAd2) of the central missing range where the intensity distribution becomes zero is specified.

The illumination luminous flux BMa, which has an annular intensity distribution by the aperture plate APa', is divided into the total amount when entering the incident end FBi of the fiber bundle 12A by the magnification variable part 8A at the rear stage. The numerical aperture is adjusted, and the light flux incident on each fiber element at the incident end (FBi) of the fiber bundle 12A maintains the ratio between the maximum numerical aperture (NAd1) and the numerical aperture (NAd2) of the missing center range. do. As explained in FIG. 16 above, since each fiber element transmits light while preserving the numerical aperture (diffusion angle) of the incident light, the light is emitted from each emission end (FBo) of the fiber bundle (FGn) on the emission side. The numerical aperture of the luminous flux BSa is equal to the numerical aperture of the luminous flux BMa incident from the incident end FBi of the fiber bundle 12A. Therefore, in the case of this modification, the luminous flux BSa emitted from each emission end FBo of the fiber bundle FGn (the divergent luminous flux from the spot light SPa formed at the emission end FBo) is the maximum. It has an annular distribution that maintains the ratio of the numerical aperture (NAd1) and the numerical aperture (NAd2) of the missing center range. The illumination beam BSa radiating from each of the plurality of spot lights SPa formed at each exit end FBo of the fiber bundle FGn and traveling is, as explained in FIG. 17 above, a fly-eye lens system ( They overlap on the incident surface (poi) of FEn), and since the illumination luminous flux (BSa) itself has an annular intensity distribution, on the incident surface (poi) of the fly-eye lens system (FEn), the maximum numerical aperture (NAd1) and It is overlapped with the distribution of annular phases while maintaining the ratio (ratio) of the numerical aperture (NAd2) of the missing center range. Similarly, in the optical paths of the other wavelength selection units 6B and 6C, a circular aperture plate APa' is removably installed.

As described above, in this modified example, at least one of the illumination beams (BSa, BSb, BSc) irradiated by overlapping the incident surface (poi) of the fly-eye lens system (FEn) is divided into a desired beam centered on the optical axis (AX2). This can be done by the intensity distribution of the annular phase with an annular contrast ratio. Accordingly, by adjusting the magnification variable portions 8A, 8B, and 8C, among the countless spot lights SPa', SPb', and SPc' formed on the emission surface (epi) of the fly's eye lens system (FEn), for example , the spot light SPa' and SPb' are distributed in the annular range outside the region CFc of the radius Ric shown in FIG. 18 or FIG. 19, and the spot light SPc' is distributed in the area of the annular band outside the area CFc of the radius Ric shown in FIG. 18 or FIG. 19. It can be distributed within the area (CFc) of the radius (Ric) shown in . At that time, by appropriately selecting a combination of interference filters mounted on each of the wavelength selection units 6A, 6B, and 6C, for example, a spot distributed in the annular range outside the area CFc of the radius Ric The light (SPa' and SPb') has an i-line (narrow) spectrum, and the spot light (SPc') distributed within the area (CFc) of the radius (Ric) has an h-line (narrow) spectrum. can do. That is, the wavelength characteristics of the light source image formed on the illumination co-plane of the second illumination optical system ILn (emission surface (epi) of the fly-eye lens system FEn) are determined by the distance from the optical axis AX2 (corresponding to the numerical aperture). Accordingly, it becomes possible to change to a completely different wavelength.

In addition, as shown in Fig. 26, the annular aperture plate APa' is installed in the optical path in the wavelength selection section 6A (6B, 6C), but is not located in the optical path in the magnification variable section 8A (8B, 8C). You can install it. Additionally, an aperture plate APb', which is the same as the four-pole aperture plate APb as shown in FIG. 25(B), may be mounted instead of the annular aperture plate APa' in FIG. 26. In that case, the illumination luminous flux (BSa) (or BSb, BSc) irradiated on the incident surface (poi) of the fly-eye lens system (FEn) has a four-point fan shape like the light transmitting portion (TPb) in Figure 25(B). overlaps in the area of . In addition, in this modification, the illumination beam BSa (or BSb, BSc) irradiated on the incident surface poi of the fly-eye lens system FEn has a normal circular shape including the optical axis AX2, and the optical axis ( Since it overlaps the annular or quadrupolar region that does not include AX2), the secondary light source image ( Compared to the case of shielding part of the spot light (SPa', SPb', SPc'), there is also an advantage that the loss of the amount of illumination light is suppressed to a small extent.

[Second Embodiment]

Fig. 27 is a diagram showing a schematic overall configuration of the exposure apparatus according to the second embodiment, and the Z axis of the orthogonal coordinate system XYZ is set to the direction of gravity. Since the detailed configuration of the exposure apparatus as shown in Fig. 27 is disclosed in, for example, the pamphlet of International Publication No. 2013/094286 and the pamphlet of International Publication No. 2014/073535, the following description of the device configuration will be simplified. In order to scan and expose a mask pattern on a flexible long sheet substrate FS, the exposure apparatus shown in FIG. 27 is curved into a cylindrical surface at a constant radius from a center line CC1 set parallel to the Y axis, and curved in the Y direction. A reflective pattern is formed on the outer peripheral surface having a predetermined length (length corresponding to the width of the sheet substrate FS in the Y direction), and a cylindrical mask DMM rotating around the center line CC1 is mounted. In addition, the exposure apparatus of FIG. 27 has an outer peripheral surface curved in a cylindrical shape at a constant radius from a center line CC2 parallel to the Y axis, and the sheet substrate FS is closely supported in the elongated direction by the outer peripheral surface, and the center line CC2 is parallel to the Y axis. ) A rotary drum (DR) that rotates around is installed. Between the cylindrical mask (DMM) spaced apart in the Z direction and the rotary drum (DR), there is a partial projection optical system (PL1) of odd-numbered equal magnification images (not shown) having a configuration substantially equivalent to that shown in FIG. 2 above. A partial projection optical system (PL3, PL5...) and a partial projection optical system (PL2) of even-numbered equal magnification images (and partial projection optical systems (PL4, PL6...), not shown) are formed.

And, a polarizing beam splitter (PBSa) for incident illumination is installed between the outer peripheral surface of the cylindrical mask (DMM) and each of the odd-numbered partial projection optical systems (PL1, PL3, PL5...). A quarter wave plate (or film body) is mounted on the surface of each polarizing beam splitter (PBSa) on the cylindrical mask (DMM) side. In each of the illumination areas IAn, which is set in an elongated rectangular shape in the Y direction on the outer peripheral surface of the cylindrical mask DMM, odd-numbered illumination lights having substantially the same configuration as the second illumination optical system ILn shown in FIG. 10 above are provided. Illumination beams for exposure from each of the two illumination optical systems (IL1, IL3, IL5...) and the even-numbered second illumination optical systems (IL2, IL4, IL6...) are projected through the polarizing beam splitters (PBSa, PBSb). The polarizing beam splitter (PBSa, PBSb) (and quarter wave plate) polarizes the illumination light flux directed to the illumination area (IAn) of the cylindrical mask (DMM) and the reflected light flux from the mask pattern appearing within the illumination area (IAn). It is separated according to the state. To do this, it is necessary to linearly polarize the illumination beam projected onto the polarizing beam splitter (PBSa, PBSb). Therefore, an appropriate position in the illumination optical path of the second illumination optical system ILn, for example, a position between the fiber bundle FGn on the emission side shown in FIG. 10 and the second condenser lens system CPn, or FIG. 4 A polarizing plate is installed in the wavelength selection units 6A, 6B, 6C of the first illumination optical system shown in , or at positions before and after the magnification variable units 8A, 8B, 8C.

If the center plane (CCp) is a plane parallel to the YZ plane, including the center line (CC1) of the cylindrical mask (DMM) and the center line (CC2) of the rotating drum (DR), then in the XZ plane (in the plane of Figure 27) When viewed, there is a set of odd-numbered partial projection optical systems (PL1, PL3, PL5...) and odd-numbered second illumination optical systems (IL1, IL3, IL5...), and even-numbered partial projection optical systems (PL2, PL4, PL6...) and even-numbered sets. A set of second illumination optical systems (IL2, IL4, IL6...) are arranged symmetrically with respect to the center plane (CCp). In addition, the main ray on the side of each cylindrical mask (DMM) of the partial projection optical system (PLn) that enters the reflected light flux of the pattern generated from each of the illumination areas (IAn) on the cylindrical mask (DMM) has an extension line of which is the center line (CC1). ), and the main ray of the imaging light flux projected to each of the projection areas EAn set on the sheet substrate FS from each rotating drum DR side of the partial projection optical system PLn is an extension thereof. It is set to point towards the center line (CC2).

In this embodiment, since the projection magnification of the partial projection optical system PLn is equal (1:1), the radius from the center line CC1 of the outer peripheral surface (pattern formation surface) of the cylindrical mask DMM and the rotating drum ( The radius (strictly, the radius plus the thickness of the sheet substrate FS) from the center line (CC2) of the outer peripheral surface of DR) is the same, and the cylindrical mask (DMM) and the rotary drum (DR) are rotated at the same rotation speed. Then, the sheet substrate FS is scanned and exposed to the reflected light flux from the device pattern formed by the high-reflection portion and the low-reflection portion on the cylindrical mask (DMM). At that time, on the pattern formation surface of the cylindrical mask (DMM), with respect to the illumination light flux from the second illumination optical system ILn, the high-reflection portion has a reflectance as high as possible, and the low-reflectivity portion has a reflectance as low as possible (ideally, the reflectance is A film body is formed by a single layer or multiple layers having zero). An example of a method for producing a reflective mask pattern is a first film body (metal thin film, etc.) with a high reflectance (e.g., 80% or more, preferably 90% or more) in the wavelength spectrum of the illumination light beam for exposure. After depositing on the entire pattern formation surface of the cylindrical mask (DMM), the second film body has a low reflectance (for example, 10% or less, preferably 5% or less) in the wavelength spectrum of the illumination light flux for exposure. (Metal thin film, dielectric multilayer film, etc.) is laminated on the surface of the first film body, and by patterning using photolithography or the like, the part of the second film body that becomes a low reflection part is left, and the part that becomes a high reflection part is removed by etching to replace the underlying surface. There is a method for exposing the first membrane body. Also, contrary to this method, a second film body having a low reflectivity is first deposited on the entire pattern formation surface of the cylindrical mask (DMM), and then a first film body having a high reflectivity is deposited on the surface of the second film body. A method may be used in which the layers are laminated, leaving a portion of the first film body that serves as a highly reflective portion, and removing the portion that serves as a low reflection portion by etching to expose the underlying second film body.

In addition, similar to the halftone method or phase shift method used in transmissive mask substrates, in the case of reflective patterns, fine steps corresponding to the wavelength of the illumination beam are created on the surface of the reflective film laminated on the pattern formation surface. It may be formed as a reflection type shifter pattern in which the amplitude and intensity of the reflected light generated from the upper and lower surfaces of the step are offset by a phase difference. In this case, a high-reflectivity film is uniformly formed on the entire pattern formation surface of the cylindrical mask (DMM), and the pattern portion that reduces reflected light on the surface of the film has an amplitude reflectance (amplitude reflectance) that provides a phase difference of 180 degrees to the reflected light. A diffraction grid-like or checker-flag-shaped uneven pattern consisting of fine steps (where . is set to zero) is formed. In the case of a step structure that provides a phase difference other than 180 degrees to the reflected light, the amplitude reflectance becomes a finite value other than zero, so an intermediate reflectance can be obtained.

In an exposure apparatus using a reflective mask as described above, variations in the reflectance of the reflective pattern may occur when the mask (cylindrical mask (DMM)) is replaced. Particularly in reflective shifter patterns, manufacturing errors in the fine steps formed on the surface of the film body cause a phenomenon in which the reflectance of the portion of the pattern where the intensity of reflected light is to be substantially zero is not sufficiently reduced. In addition, even in the case of a reflective pattern simply composed of a high-reflection portion and a low-reflection portion, when formed on the outer peripheral surface of the cylindrical mask (DMM) as shown in FIG. 27, the pattern surface is curved in the circumferential direction of the cylindrical mask (DMM). , the incident angle of the main ray of the illumination light beam changes slightly depending on the circumferential position within the illumination area (IAn), and there is a possibility that a difference in reflectance within the illumination area (IAn) may occur.

Therefore, in this embodiment as well, as described in the first embodiment and its modifications, the combination of interference filters mounted on each of the wavelength selection units 6A, 6B, and 6C is changed, or the magnification variable unit 8A is changed. , 8B, 8C), the respective diameters (numerical apertures) of the illumination beams (BSa, BSb, BSc) projected onto the incident surface (poi) of the FlyEye lens system (FEn) are changed, or the FlyEye lens system (FEn) is adjusted. By changing each area (shape) of the illumination beam (BSa, BSb, BSc) projected onto the incident surface (poi) of the eye lens system (FEn), unevenness or curvature of the reflectance due to manufacturing error of the reflective pattern is prevented. It becomes possible to reduce the unevenness of reflectance that may occur due to the pattern surface. In particular, as explained in Figure 20 or Figure 26, the intensity distribution and numerical aperture for each wavelength can be adjusted within the range of the maximum diffusion angle (maximum numerical aperture) of the illumination beam Irn projected onto the illumination area IAn. Since this is done, there is an advantage that correction can be easily performed even if the reflectance changes or unevenness occurs in the reflective mask pattern.

[i-line-broadband interference filter]

As shown in FIG. 6 above, a typical interference filter (SWa) for i-line is centered on the bright line wavelength of the i-line and filters the i-line spectrum with a bandwidth as narrow as possible (for example, ±10 nm width or less). It is set to extract (permeate). In contrast, the i-line wide interference filter (SWb) is set to extract (transmit) the i-line spectrum with as wide a bandwidth as possible, including only the bright line wavelength of the i-line. The bandwidth of the i-line-wide interference filter (SWb) is set depending on the chromatic aberration characteristics of the catadioptric partial projection optical system (PLn) (hereinafter simply referred to as the projection optical system) described in each of the above embodiments. FIG. 28 shows detailed spectral characteristics obtained by measuring the wavelength characteristics of light generated from the arc discharge portion of the ultra-high pressure mercury discharge lamp shown in FIG. 5 above using a spectroscope with higher wavelength resolution than the spectrometer that measured the wavelength characteristics in FIG. 5. . The main bright lines caused by mercury in the ultra-high pressure mercury discharge lamp are the g-line with a wavelength of 435.835 nm, the h-line with a wavelength of 404.656 nm, the i-line with a wavelength of 365.015 nm, and the j-line with a wavelength of 312.566 nm. However, the i-line is irradiated by other substances in the lamp. A bright line (Sxw) (wavelength is about 330 nm) also occurs between the bright line wavelength and the j line wavelength.

On the other hand, in the case of a catadioptric projection optical system (PLn) in which chromatic aberration is corrected mainly for the bright line wavelength of the i line, the chromatic aberration characteristics tend to be as shown, for example, in FIG. 29. Figure 29 is a graph of chromatic aberration characteristics with wavelength on the horizontal axis and the amount of chromatic aberration (magnification chromatic aberration, or axial chromatic aberration) on the vertical axis. When chromatic aberration is corrected at the bright line wavelength of the i line, the lens element constituting the projection optical system is made of two or more types of base materials with different dispersion or refractive index, and the amount of chromatic aberration is substantially zero at the bright line wavelength of the i line. It is optically designed to However, the chromatic aberration characteristic generates a large amount of chromatic aberration in the long wavelength region and the short wavelength region with respect to the bright line wavelength of the i line. Therefore, in terms of this chromatic aberration characteristic, the wavelength width (ΔWi) is set so that it is within the allowable amount (ΔCAi) as the amount of chromatic aberration and does not include prominent bright line wavelengths other than the i line. As shown in Figure 28, a bright line (Sxw) exists near the short wavelength side of the bright line wavelength of the i line, and an h line exists near the long wavelength side, but there is no prominent bright line between the wavelengths of 340 nm and 400 nm. . In this regard, the i-line-wide-band interference filter (SWb) is manufactured with the characteristic of having a transmittance of 90% or more between the wavelength of about 350 nm and about 390 nm. In other words, the i-line-wide-band interference filter (SWb) does not include the spectral component on the peak of the strong bright line that appears on the short-wavelength side and the long-wavelength side of the bright line wavelength of the i-line, but only the spectral peak and the bottom of the bright line wavelength of the i-line. It is manufactured with wavelength selection characteristics (transmission characteristics) that extract (transmit) low-brightness spectral components distributed in . Additionally, even when using a projection optical system in which chromatic aberration has been corrected for other bright line wavelengths (h-line or g-line), the h-line can be adjusted in the same way because it has, to a greater or lesser extent, the chromatic aberration characteristic as shown in Fig. 29. -You can produce a broadband interference filter or a g-line-broadband interference filter.

[Other variations]

As shown in Figs. 4 and 15, the maximum diffusion angles ( In order to control the maximum numerical aperture, a magnification variable section 8A, 8B, 8C is formed having two sets of lens systems 8A1, 8A2 whose positions in the direction of the optical axis AX1 are adjustable; 8A2) It may be possible to exchange at least one of the lens systems with another lens system to fixedly switch the magnification (numerical aperture). Additionally, as disclosed in U.S. Patent No. 5,719,704, between the front lens system 8A1 and the rear lens system 8A2 described above, two conical prism-shaped optical members (axicon optical systems) may be installed. do. At this time, when performing normal illumination, the two cone-shaped prism-shaped optical members are brought into close contact in the direction of the optical axis AX1, and when performing annular illumination, the optical axis of the two cone-shaped prism-shaped optical members is ( By adjusting the interval in the AX1) direction, the cross-sectional shape of the illumination beam BMa passing between the lens system 8A1 and the lens system 8A2 may be an annular shape with a variable size. In this case, as in Modification 5 explained using FIG. 26 above, it becomes unnecessary to arrange the annular aperture plate APa' in the optical path of the illumination light flux, so the utilization efficiency of the illumination light when performing annular illumination can be further improved.

In addition, in each of the above-described embodiments and modifications, a fly-eye lens system (FEn) is used as the optical integrator, but a micro lens array, a rod integrator, etc. may be used instead. In addition, in each of the above-described embodiments, mercury lamps (ultra-high pressure mercury discharge lamps) 2A, 2B, 2C were used as light source devices, but lamps of any other discharge type can be used. In addition, as a light source device, a laser light source such as a light-emitting diode (LED), solid-state laser, gas laser, or semiconductor laser, or a laser light of a heald light is amplified and the harmonics of the heald light (ultraviolet wavelength range) are converted by a wavelength conversion element. ) A laser light source that generates can also be used.

In addition, the laser light source that generates the harmonics of the heald light is, for example, a fiber amplifier laser light source as disclosed in Japanese Patent Application Publication No. 2001-085771, and pulse laser light with a central wavelength of 355 nm is used as the illumination beam. It can be used as. In that case, the wavelength conversion element for harmonic generation mounted on the fiber amplifier laser light source, etc. functions as the same wavelength selection unit (wavelength selection element) as the interference filter (SWa, SWb, SWc). Moreover, as a light source device, a mercury discharge lamp (ultra-high pressure mercury discharge lamp) and a laser light source may be used together. For example, light from a mercury discharge lamp, the spectral component of which includes the i-line (center wavelength 365 nm) extracted from the i-line-narrow interference filter (SWa), or the i-line-broad interference filter (SWb) and pulse laser light with a central wavelength of 355 nm emitted from a fiber amplifier laser light source may be used together.

When using such a laser light source, in order to set the diffusion angle (numerical aperture) of the illumination beam to be large, as an example, on the optical path of the laser beam emitted as a parallel beam from the laser light source, the pitch gradually increases with respect to the radial direction. A zone plate diffraction grating made of an initial material such as quartz with small concentric circular (zone plate) topological irregularities may be arranged. The minimum pitch of the zone plate diffraction grating is given by is set accordingly.

In each of the above-described embodiments, the exposure apparatus has been described as an example of a multi-lens scanning exposure apparatus having a plurality of partial projection optical systems PLn, but the exposure apparatus has a stationary mask substrate M and plate P. A step-and-repeat type exposure apparatus may be used in which the pattern of the mask substrate M is exposed and the plate P is sequentially moved in steps. The light source of the lighting device is not limited to three mercury lamps or three laser light sources, and may include one, two, or four or more light sources. In addition, in the above-described embodiment, six fiber bundles (FGn) having six emission ends (FBo) are used, but the second illumination optical system (ILn) is composed of one piece and the projection optical system (PLn) is also composed of one piece. In the case of an exposure apparatus, one fiber bundle (FGn) is sufficient.

In addition, the illumination luminous flux BMa created by one light source (such as a mercury lamp 2A) and one first illumination optical system (including a wavelength selection section 6A and a magnification variable section 8A) is divided into one In the case of projecting onto the mask substrate M through the second illumination optical system IL1 and projecting and exposing the pattern of the mask substrate M onto the plate P through one partial projection optical system PL1, Without forming the fiber bundles 12A to 12C, FGn, the illumination luminous flux BMa from the magnification variable section 8A is directly transmitted to the flyeye via the first condenser lens system CF1 of the second illumination optical system IL1. You may make it incident on the lens system (FE1).

According to the first embodiment, its modification, or second embodiment described above, a light flux (BM) emitted from each of at least two first light sources and two second light sources (two of the mercury lamps 2A to 2C) ) a first wavelength selection unit and a second wavelength selection unit (two of 6A to 6C) formed corresponding to each of the first light source and the second light source to extract a spectral distribution of a predetermined wavelength width from A wavelength selection element (interference filter (SWa, SWb, SWc), etc.) formed in each of the selection unit and the second wavelength selection unit to change the spectral distribution such as the wavelength range or wavelength to be extracted is exchangeably disposed in the optical path. a mechanism (a slide mechanism (FX) or a mount mechanism), and a numerical aperture variable section (8A) for each of the first illumination light flux extracted from the first wavelength selection section and the second illumination light flux extracted from the second wavelength selection section. A light synthesis member (fiber bundle (FGn)) for synthesizing light at a numerical aperture individually set by ~ 2 of 8C) to form a secondary light source image on the illumination coplane of an illumination optical system including an optical integrator ) is formed. Therefore, there are differences in the type of pattern on the mask substrate (binary mask, phase shift mask, halftone mask, etc.), the fineness of the pattern to be exposed, the amount of taper inclination given to the edge of the resist layer after development, or the reflective type. In the case of the mask pattern, different wavelength distribution characteristics (characteristics in which the intensity of each spectrum differs depending on the position in the illumination plane) are displayed within the distribution of the secondary light source image, depending on various conditions (exposure recipe) such as fluctuations in reflectance and non-uniformity. Alternatively, the wavelength distribution can be varied with a diffusion angle corresponding to the maximum numerical aperture of the illumination light flux to the mask substrate. In addition, by changing (switching) the wavelength distribution in the illumination coplane, the projection optical system itself generated by the energy of the imaging light beam passing through the projection optical system (partial projection optical system (PLn)) that projects and exposes the pattern of the mask substrate is It also becomes possible to control (suppress) irradiation variation (projection magnification variation, focus variation, aberration variation, etc.).

Below, the wavelength characteristics (spectral characteristics) of light from the mercury discharge lamp will be supplemented with reference to FIG. 30. In each embodiment and modification example, a short arc type ultra-high pressure mercury discharge lamp (2A, 2B, 2C) is mainly used, but as a light source of the pattern exposure apparatus for electronic devices, the mercury vapor pressure in the discharge tube (emission tube) is about High-pressure mercury discharge lamps of approximately 10 ⁵ Pa to 10 ⁶ Pa are also used. In general, an ultra-high pressure mercury discharge lamp increases the mercury vapor pressure in the discharge tube to about 10 ⁶ Pa to about 10 ⁷ Pa, thereby increasing the respective spectral widths of the i-line, h-line, and g-line at bright line wavelengths appropriate for photolithography. It is slightly wider than that, and the relative balance of the intensity of each peak of the i-line, h-line, and g-line is different for the high-pressure mercury discharge lamp. In the discharge tube of the mercury discharge lamp, for example, as disclosed in Japanese Patent Application Publication No. 2009-193768, in addition to mercury (0.15 mg/㎣ or more) whose mercury vapor pressure is 150 to 300 atm when lit, Approximately 13 kPa of argon gas (noble gas) and halogens such as iodine, bromine, and chlorine in the form of compounds with mercury and other metals are enclosed. In addition, compared to a high-pressure mercury discharge lamp, the light from an ultra-high pressure mercury discharge lamp has several percent, or 10%, relative to the peak value of the light intensity of each bright line wavelength, even in the wavelength range between the i-line, g-line, and h-line. A spectral distribution (lower edge) of low luminance of about ∼20% exists.

Figure 30 is a graph explaining the difference in wavelength characteristics of a high-pressure mercury discharge lamp and an ultra-high pressure mercury discharge lamp, Figure 30(A) shows an example of the wavelength characteristics of light from a high-pressure mercury discharge lamp, and Figure 30(B) shows an example of the wavelength characteristics of light from an ultra-high pressure mercury discharge lamp. In each of Figures 30(A) and 30(B), the horizontal axis represents the wavelength (nm), and the vertical axis represents the relative intensity (%) of the spectrum when the peak value of the intensity of the i line of the bright line wavelength is set to 100%. indicates. Although each wavelength characteristic in Figures 30(A) and 30(B) varies slightly due to differences in lamp manufacturers and differences in rated power of the lamps, the spectrum of the wavelength 350 to 400 nm including the wavelength of the i line (365 nm) When paying attention to the distribution, in the high-pressure mercury discharge lamp, as shown in Fig. 30(A), the bottom portion with a relative intensity of several percent or more, preferably 10% or more, hardly appears. On the other hand, in an ultra-high pressure mercury discharge lamp, the relative intensity is several percent or more, as shown in Fig. 30(B), and a bottom portion (low-intensity spectral component) of approximately 10% appears.

The degree of relative strength of the bottom of line i in Figure 30(B) can vary depending on the amount of mercury encapsulated in the discharge tube, the type and content of other rare gases or halogens, and the mercury vapor pressure, but is approximately several to 20%. It becomes. In addition, the spectral widths of the i-line, h-line, and g-line of the bright line wavelength tend to be slightly wider (thicker) in the ultra-high pressure mercury discharge lamp of Figure 30(B) compared to the high-pressure mercury discharge lamp in Figure 30(A). It is written as . In addition, in the wavelength characteristics shown in FIG. 5 above, the spectral component in the range of 350 to 400 nm, which is the bottom of the i-line wavelength (365 nm), has an intensity of about 20% relative to the peak intensity of the i-line. It is written as .

In this regard, the illumination fluxes (BSa, BSb, BSc) wavelength-selected using the i-ray-wide interference filter (SWb) shown in FIG. 7 above to include only the i-line among the bright line wavelengths of mercury are shown in FIG. 6 above. Compared to the wavelength-selected illumination flux using the i-line-narrow interference filter (SWa) shown, the amount of light energy is higher. Even if the relative intensity of the bottom portion of the i line (in the range of 350 nm to 365 nm and 365 nm to 400 nm) is about 10% of the peak intensity of the i line as shown in Figure 30(B), the resist of the plate P Since the amount of light energy (dose amount) per unit time given to the layer is increased by an amount determined by the product of the intensity of the bottom portion and the wavelength width, the exposure amount increases by approximately 20% (1.1 × 1.1 ≒ 1.2 times). do. Therefore, when the pattern of the mask substrate M is scanned and exposed to the resist layer of the plate P by the exposure apparatus EX shown in FIG. 1 above, the mask substrate M and the plate P are scanned and exposed. It becomes possible to increase the speed by about 20%, and as a result, it becomes possible to increase the productivity of the high-resolution pattern exposure process using the i-line by about 20%.

Therefore, the wavelength width at 10% of the relative intensity of the spectral distribution of the i-line radiated from the high-pressure mercury discharge lamp shown in Fig. 30(A), or the wavelength selection set in the i-line-narrow interference filter (SWa) shown in Fig. 6 The width (bandwidth) is set to BWi (nm), and the wavelength at the peak intensity is set to be the center of the spectrum distribution of i-rays emitted from an ultra-high pressure mercury discharge lamp (short wavelength) outside the wavelength (bandwidth) (BWi). (side and long-wavelength side), when the hem portion reaches the spectrum of nearby bright line wavelengths, the average relative intensity at the hem portion is about several percent to 10 percent (preferably 20 percent or more). , it is desirable to use an ultra-high pressure mercury discharge lamp in which the amount of mercury, mercury vapor pressure, atmospheric pressure or component amount of noble gas, component amount of halogen, etc. are adjusted. In addition, it is not limited to the spectral distribution of the i line, and even when using the spectral distribution of the h line or g line from an ultra-high pressure mercury discharge lamp, the relative intensity is several percent compared to the h line or g line from the same high pressure mercury discharge lamp. Since diffusion occurs in the bottom part, it is sufficient to prepare an h-line-broadband interference filter or a g-line-broadband interference filter with wavelength selection characteristics including the bottom part.

Next, additional comments will be made on modified examples of the mask. In each of the above embodiments and their modifications, an exposure apparatus using a transmissive or reflective mask substrate (or cylindrical mask) on which a mask pattern is fixedly formed (carried) is assumed, but a plurality of micro-order micro mirrors are used. By using a DMD (Digital Mirror Device) with two-dimensional arrays, etc., the angles of each mirror are switched at high speed according to the data (CAD data) of the pattern to be exposed, so that the For the variable mask type exposure apparatus that projects a pattern image (also called a maskless exposure apparatus because it does not use a fixed mask pattern), the illumination system (FIGS. 3 to 20, etc.) described in each embodiment is used in the same manner. It can be applied. In the variable mask type exposure apparatus, the projection area that can be formed on the plate P by one DMD is limited to a small rectangular area like the projection area EA1 shown in FIG. 1, so a plurality of DMDs are required. And, a plurality of projection lens systems are formed that project the reflected light from each DMD onto the plate P. In this case, each reflection surface of a plurality of DMDs (a surface on which a plurality of micro mirrors are arranged) is a pattern for an electronic device in the same form as the distribution of a plurality of micro mirrors whose light reflection direction is individually controlled according to CAD data. It contains. Then, each reflective surface of the plurality of DMDs is disposed at a position corresponding to the illumination areas (IA1 to IA6) shown in FIGS. 1 to 3, and the intensity distribution is uniformized to within ±2%, for example, of the illumination luminous flux ( (corresponding to BSa', BSb', and BSc' shown in Fig. 14).

Therefore, among the plurality of micro-mirrors of the DMD, the projection luminous flux (illumination luminous flux) reflected from the micro-mirrors whose angles are set so that the illumination luminous flux is incident on the projection lens system, and projected onto the plate P via the projection lens system, is shown in Figure 20. It can have the same characteristics as the orientation characteristics (diffusion angle characteristics) as described above. Furthermore, by combining the interference filters shown in FIG. 21, the wavelength distribution can be made different within the diffusion angle corresponding to the maximum numerical aperture of the projection light beam irradiated onto the plate P from each of the micromirrors of the DMD. . In addition, instead of DMD, a phase difference is given to the reflected light flux by shifting the selected micromirrors from among the reflective surfaces of a large number of two-dimensionally arranged micromirrors (usually all set on the same plane) in the direction perpendicular to the reflective surface. You may also use a spatial light modulator (SLM: Spatial Light Modulator).

Next, other types of projection exposure apparatus will be added. In the above embodiments and modifications, a so-called multi-lens exposure apparatus is provided, which includes a plurality of partial projection optical systems (PLn) (PL1 to PL6) and a plurality of second illumination optical systems (ILn) (IL1 to IL6) corresponding thereto. Although this has been assumed, even if the exposure apparatus is provided with a single projection optical system and a single second illumination optical system, the same function can be easily provided by only slightly changing the configuration in the above-described embodiment. Specifically, in the wire distribution unit 10a in the optical distribution unit 10 shown in FIG. 9, a plurality of fiber wires included in each of the fiber bundles 12A, 12B, and 12C on the incident side are divided into 6 Instead of distributing to each of the fiber bundles (FG1 to FG6), they are bundled to form a single fiber bundle, and the emission end (FBo) of the single fiber bundle is set to the shape of a single illumination area set on the mask substrate (M). Just mold it so that it becomes a square shape similar to .

Next, additional information will be given on modifications of the light source device. In the configuration of FIG. 3 above, a plurality (two or more) of mercury discharge lamps 2A, 2B, 2C are used as the light source device, but even when a single ultra-high pressure mercury discharge lamp is used, in the above embodiment You can easily have the same function just by slightly changing the configuration. Specifically, after the light from a single ultra-high pressure mercury discharge lamp is converted into a substantially parallel light beam by the lens system (collimator lens) 6A1 in FIG. 4, the wavelength range of the spectral component of the i line, for example, is not included. It forms a dichroic mirror that transmits light in a wavelength band including the spectral components of the h line and the g line, and reflects light in a short wavelength band including the spectral component of the i line. In addition, for the light transmitted through the dichroic mirror, a wavelength selection unit 6A as shown in FIG. 4 (or FIG. 3) (including, for example, an interference filter for extracting the spectral component of the h line) and A magnification variable section 8A is formed, and for light reflected from the dichroic mirror, a wavelength selection section 6B (i-line narrow interference filter (SWa) or i-line wide interference filter (SWb)) is applied to the light reflected from the dichroic mirror. includes) and forms a magnification variable portion 8B. In this way, as explained in FIG. 22, a two-dimensional diffusion (range) is formed on the illumination pupil plane (corresponding to the emission surface (epi) of the fly-eye lens system (FEn)) in the second illumination optical system (ILn). The wavelength characteristics of the light source image (a set of point light source images) can be varied according to the characteristics of the selected interference filter.

Next, the setting of the wavelength selection characteristic by the i-line-wide-band interference filter (SWb) will be added. As shown in FIG. 30(B) above, the width of the bottom of the spectral component of the i line from an ultra-high pressure mercury discharge lamp (for example, a width that is approximately 10% of the peak intensity at the center wavelength of the i line) ) has an area of more than twice the width of the bottom of the spectral component of the i line from the high-pressure mercury discharge lamp shown in FIG. 30(A). The partial projection optical system PL1 (PL2 to PL6) of the exposure apparatus EX as shown in FIGS. 1 and 2 above has reflectors Ga4 and Gb4 arranged in the coplane (aperture position) Epa and Epb. It is a catadioptric half-field type imaging system. Such an imaging system has the advantage that chromatic aberration correction becomes easier compared to an imaging system of a total refractive type (all optical elements are composed only of refractive elements such as lenses), and has a plurality of bright line spectra (e.g., i Even when the pattern of the mask M is projected and exposed on the substrate P using illumination light containing the line spectrum component and the h-line spectrum component, deterioration (image distortion) of the projected image due to chromatic aberration can be reduced. You can.

However, even if projection exposure is performed with illumination light of only a single i-line spectrum component narrowed through an i-line narrow interference filter (SWa), if the pattern formed on the mask M becomes fine, the projection optical system (PL1) ) Deformation occurs in the projection image (image intensity distribution) due to various aberrations of (PL2 to PL6). The deformation is most noticeable in a pattern of fine isolated squares (approximately square) called a hole pattern.

31 shows a square hole pattern formed on the mask M, and a projection image (light intensity distribution) obtained when the hole pattern is projected onto the substrate P using an i-line narrow interference filter (SWa). This is a diagram schematically showing the relationship between the shape and the shape of the resist image that appears due to development of the resist layer after exposure, and the X and Y axes correspond to the orthogonal coordinate system XYZ in FIGS. 1 to 3 above. Here, FIG. 31(A) is a projection obtained in the case of a hole pattern CHA formed on the mask M with a size Dx × Dy sufficiently larger than the minimum line width value resolved by the projection optical system PL1 (PL2 to PL6). The shape of the image (resist image) Ima is schematically shown, and Figure 31(B) is a projection image obtained in the case of a hole pattern CHB formed on the mask M with a size of about twice the minimum line width. (Resist image) The shape of (Imb) is schematically shown, and Figure 31(C) is a projection image (resist image) obtained in the case of a hole pattern (CHC) formed on the mask M with a size close to the minimum line width value. ) (Imc) is a schematic representation of the shape. In Figure 31, the hole patterns (CHA, CHB, CHC) are all formed as isolated transparent portions within the surrounding light-shielding portion indicated by hatching, but in the opposite case, that is, they are formed as isolated light-shielding portions within the surrounding transparent portion. It's okay. In addition, the resolution (R), expressed as the minimum linewidth value that can be projected by the projection optical system (PLn) (n = 1 to 6), is generally determined by the numerical aperture (NAp) on the image side of the projection optical system (PLn) and the wavelength of the illumination light ( λ) (nm), by the process constant (k) (0 < k ≤ 1), it is defined as R = k·(λ/NAp).

As shown in Fig. 31(A), even in the case of a large square hole pattern (CHA) with dimensions Dx Mainly due to the value of the numerical aperture (NAp) on the image side of the projection optical system (PLn), that is, MTF (Modulation Transfer Function), the image is not sufficiently resolved and becomes rounded. This phenomenon occurs even when illumination light (for example, laser light with a spectral width of less than 1 nm, etc.) with a very narrow spread of the spectral distribution with respect to the central wavelength (λ) is used. In particular, as shown in Fig. 31(C), in the case of a square hole pattern (CHC) with dimensions close to the minimum resolvable linewidth value of the projection optical system PLn, the shape of the projection image (resist image) Imc is approximately It becomes circular. Under the characteristics of such a projection optical system (PLn), an i-line-wide-band interference filter (SWb) as shown in FIG. 7 above is used to extract illumination light to broadly cover the bottom of the spectral distribution with respect to the center wavelength of the i-line. Then, when the hole pattern CHC shown in Fig. 31(C) is subjected to projection exposure, the projection image (resist image) Imc changes from a circular shape to an elliptical shape due to the chromatic aberration characteristic of the projection optical system PLn.

Figure 32 is an exaggerated illustration of the projection image Imc being transformed into an oval shape, and the broken line in Figure 32 shows the projection image Imc' being almost exactly circular. The circular diameter of the projection image Imc' can be theoretically estimated from the basic optical characteristics of the projection optical system PLn. The minor axis length in the Y-axis direction of the projection image (Imc) transformed into an elliptical shape due to the influence of chromatic aberration is CHy, the major axis length in the X-axis direction is CHx, and the flatness (ellipticity) (Δf) of the elliptical shape is Δf = CHy. When /CHx is set, it is desirable to set the flatness (ellipticity) (Δf) to 80% or more, preferably 90% or more, from the allowable range for device manufacturing. That is, the diffusion range of the bottom of the i-line spectral distribution extracted by the i-line wide interference filter (SWb) is that of the projection image (Imc) of the square hole pattern (CHC) with dimensions close to the minimum resolvable linewidth value. The shape deformation from a circle is determined to be 80% or more, preferably an ellipse with a flatness (ellipticity) of 90% or more.

In addition, in Fig. 32, in the transformation of the projection image Imc into an ellipse, the major axis direction is shown as the X direction and the minor axis direction is shown as the Y direction, but each direction of the major axis and minor axis is in the There are cases where it goes in a random direction. In FIG. 33, the major and minor axes of the projection image (Imc) of the hole pattern transformed into an elliptical shape are rotated by Δρ with respect to the X and Y axes. Therefore, in order to accurately judge the adequacy of the diffusion range of the bottom part of the i-line spectral distribution extracted by the i-line wide interference filter (SWb), a square with dimensions close to the minimum resolvable linewidth value is used by test exposure, etc. The projection image (Imc) of the hole pattern (CHC) is exposed to the substrate P, the resist image corresponding to the developed projection image (Imc) is observed with an inspection device or the like, and the projection image ( The shape of the resist image corresponding to Imc) is specified (determination of the major axis and minor axis directions), and the major axis length (CHx) in the major axis direction and the minor axis length (CHy) in the minor axis direction are measured. Then, it is sufficient to determine whether the flatness (ellipticity) (Δf) obtained from the measurement results is within the acceptable range (80% or more, preferably 90% or more).

However, as shown in FIG. 2 above, in the image space (immediately below the field stop plate FA1 disposed on the intermediate image plane IM1) in the imaging optical path of the projection optical system PLn (n = 1 to 6) , the image shift optical member SC1 is formed. The image shift optical member SC1, as disclosed, for example, in the pamphlet of International Publication No. 2013/094286, includes transparent parallel flat glass (quartz plate) tiltable within the XZ plane in FIG. 2 and orthogonal thereto. It consists of transparent parallel flat glass (quartz plates) that can be tilted in any direction. By adjusting the inclination amount of each of the two quartz plates, the pattern image in the projection area EA1 (EA2 to EA6) projected on the substrate P can be slightly shifted in any direction in the XY plane. In addition, the arrangement of the image shift optical member SC1 is not limited to immediately below the field stop plate FA1 shown in FIG. 2, and the focus adjustment optical member FC1 or magnification adjustment as another correction optical system disposed in the image space. It can be replaced with the arrangement of any of the optical members (MC1).

Each of the two parallel flat quartz plates constituting the image shift optical member SC1 has a high transmittance from the ultraviolet wavelength range (about 190 nm) to the visible wavelength range, but in the case of synthetic quartz, as an example, Figure 34 As shown, the refractive index tends to vary greatly depending on the wavelength in the short-wavelength range of 500 nm or less, especially from the vicinity of the wavelength of 400 to 300 nm toward the short-wavelength side. In Figure 34, the horizontal axis represents the wavelength (nm), and the vertical axis represents the refractive index of synthetic quartz. Therefore, for example, among the light from an ultra-high pressure mercury discharge lamp (or high-pressure mercury discharge lamp), both the i-line spectral component with a central wavelength of about 365 nm and the h-line spectral component with a central wavelength of about 405 nm When projecting the pattern of the mask M using illumination light comprising, an image projected on the substrate P with an i-line spectral component according to the tilt amount of the quartz plate of the image shift optical member SC1; A phenomenon occurs in which the image projected on the substrate P with the h-line spectrum component is slightly misaligned in the XY plane.

Figure 35 is a schematic diagram of the behavior of the imaging light flux in the quartz plate (SCx) that shifts the image in the This is a drawing shown in detail. In the quartz plate (SCx), the incident surface (Stp), where the imaging light beam passing through the opening of the field stop plate (FA1) enters, and the exit surface (Sbp), where the imaging light beam is emitted, are parallel to each other and face each other at a distance (thickness) Dpx. It is configured to rotate (tilt) around a rotation center line parallel to the Y axis. In Fig. 35, among the imaging beams, only the chief ray LPr of the imaging beam radiating from the image point Poc, which is imaged as an intermediate image at the center point of the opening of the field stop plate FA1, is shown, and the line Lss is the chief ray. (LPr) represents the normal line of the incident surface (Stp) at the point where it intersects the incident surface (Stp), and line LPr' represents an extension of the principal ray (LPr) before incident on the incident surface (Stp). For the initial attitude state in which the incident surface (Stp) of the quartz plate (SCx) is perpendicular to the incident chief ray (LPr) (a state in which the inclination of the quartz plate (SCx) is zero), the quartz plate (SCx) is positioned within the XZ plane. When inclined by the angle Δθx, according to Snell's law, the chief ray LPr is emitted parallel to the line LPr' extending from the emission surface Sbp in the X direction by the shift amount δx.

In general, the amount of shift (δx) of light due to the inclination of a parallel plate glass with a refractive index (nx) can be calculated as δx ≒ Dpx·Δθx(1-1/nx) by applying Snell's law, When the light flux contains an i-line spectral component (wavelength 365 nm) and an h-line spectral component (wavelength 405 nm), the refractive index of the quartz plate (SCx) shows slightly different values for each wavelength. So, the refractive index in the i-line spectral component (wavelength 365 nm) of the quartz plate (SCx) is ni, the refractive index in the h-line spectral component (wavelength 405 nm) is nh, and the i-line spectral component (wavelength 365 nm) is If the shift amount of the image due to the h-line spectral component (wavelength 405 nm) is δxi and the shift amount of the image due to the h-line spectral component (wavelength 405 nm) is δxh, the shift amount (δxi) is calculated as δxi ≒ Dpx·Δθx(1-1/ni) , the shift amount (δxh) is calculated as δxh ≒ Dpx·Δθx(1-1/nh). In this regard, if the difference amount of shift amount resulting from the difference in wavelength (color misalignment) is δx(i-h), the difference amount (δx(i-h)) is

δx(i-h) ≒ Dpx·Δθx[(1-1/ni) - (1-1/nh)]

It becomes.

As an example, the thickness (Dpx) of the quartz plate (SCx) is 10 mm, the refractive index (ni) in the i-line spectral component (wavelength 365 nm) is 1.4746, and the refractive index (ni) in the h-line spectral component (wavelength 405 nm) is 1.4746. When the refractive index (nh) is set to 1.4696 and the change in the difference amount (δx(i-h)) with respect to the angle (Δθx) (0° to 10°) is calculated, the linear characteristics are as shown in the graph shown in FIG. 36. In the graph of Figure 36, the horizontal axis represents the inclination angle (Δθx) [deg.] of the quartz plate (SCx), and the vertical axis represents the difference amount (δx(i-h)) [μm]. Since the quartz plate SCx is disposed in the image space where the intermediate image is formed, and the projection magnification of the projection optical system PLn is equal to (×1), the difference amount (δx(i-h)) in FIG. 36 is , becomes the relative positional deviation amount between the pattern image by the i-line spectral component and the pattern image by the h-line spectral component that are projected as is on the substrate P. For example, when the inclination angle (Δθx) of the quartz plate (SCx) is 5°, the amount of difference (δx(i-h)) due to color misalignment is about 2 μm in the X direction, causing distortion in the projected and exposed pattern. Or, an error may occur in the line width.

The effect of color misalignment caused by the quartz plate (SCx) described above occurs in the same way on another quartz plate (SCy) that slightly shifts the projected image in the Y direction, and the quartz plate (SCy) is parallel to the X axis. When tilted by the inclination angle (Δθy) from the initial state of being horizontal around the rotation center line, a difference amount (δy(i-h)) due to color shift in the Y direction is generated. In this way, when the illuminance is increased by illumination light containing both the i-line spectral component (wavelength 365 nm) and the h-line spectral component (wavelength 405 nm), the projection optical system PLn (n = 1 to 6) The image shift range by the image shift optical member SC1 required to maintain good connection precision between the pattern images projected on the substrate P, respectively, is the difference amount due to color shift (δx(i-h), δy There may be restrictions depending on the degree of (i-h)).

In contrast, as in each of the above embodiments and modifications, by using an i-line-wideband interference filter (SWb), the bottom of the i-line spectral component from the ultra-high pressure mercury discharge lamp is separated from the nearby long-wavelength side as shown in FIG. 7. It can be broadly extracted in a range that does not include the bright line component (h line) or the bright line component on the short wavelength side, and with respect to the illuminance when using the i line-narrow interference filter (SWa), it can be extracted in the order of several to several dozen percent. While increasing the illumination, the color misalignment error (error corresponding to the difference amount (δx(i-h), δy(i-h))) caused by the inclination of the quartz plate (SCx, SCy) of the image shift optical member (SC1) is reduced. It can be suppressed.

In addition, the flatness (ellipticity) (Δf) when the projection image (Imc) of the hole pattern (CHC) described in FIGS. 32 and 33 above is transformed into an elliptical shape, or the directionality in the XY plane of the major and minor axes also changes depending on the degree of inclination angle of the quartz plates (SCx, SCy) of the image shift optical member (SC1). Therefore, when the image shift optical member SC1 using tiltable parallel flat glass (quartz plate (SCx, SCy)) is formed, as in the projection optical system PLn shown in Fig. 2, the image shift optical member SC1 At the maximum tilt angle (Δθx, Δθy) of parallel flat glass (quartz plate (SCx, SCy)) corresponding to the nominal maximum amount of image shift by of the i-line-wide-band interference filter (SWb) so that the flatness (ellipticity) (Δf) of the projected image (Imc) of CHC) is 80% or more (preferably 90% or more) in theory or in real exposure images. You may set the wavelength selection range.

From the above, in the embodiment described in FIGS. 31 to 36, the mask pattern (transmissive or reflective) is illuminated with illumination light of a predetermined wavelength distribution (for example, light from an ultra-high pressure mercury discharge lamp), and the light generated from the mask pattern is illuminated. In a projection exposure method in which an image of a mask pattern is projected and exposed on a substrate by a projection optical system that enters an imaging light flux and projects it on the substrate, a specific center wavelength of the wavelength distribution of the illumination light is set to λ (e.g., the center of the i line). wavelength), the numerical aperture on the substrate side of the projection optical system is NAp, the process constant is k (0 < k ≤ 1), and the minimum line width dimension that can be resolved is determined by the resolution (R) defined as k·(λ/NAp). When a projection image of a square or square hole pattern of a size close to By setting the width of the wavelength distribution of the illumination light including the center wavelength (λ) (e.g., the wavelength width selected in the interference filter) so that it is 80% (0.8) or more, preferably 90% (0.9) or more. , high-resolution pattern exposure becomes possible while increasing the illuminance of the illumination light irradiated to the mask pattern. Additionally, the size of the hole pattern, converted to the size of the image projected on the substrate side, is set to be larger than the dimension determined by the resolution (R) and smaller than twice the resolution (R).

To put it another way, it is an interference filter that filters light from a light source (such as a mercury discharge lamp) that emits light including a plurality of bright lines into illumination light with a wavelength appropriate for projection exposure of the mask pattern, among the wavelength distribution of the illumination light. The central wavelength of a specific bright line is λ (e.g., the central wavelength of the i line), the numerical aperture on the substrate side of the projection optical system is NAp, and the process constant is k (0 < k ≤ 1), and k·(λ/NAp) When a projected image of a square or square hole pattern with a size close to the minimum resolvable line width dimension determined by the defined resolution (R) is projected onto a substrate, the major axis length (CHx) of the projected image of the hole pattern that is transformed into an elliptical shape An interference filter whose filtering wavelength is set such that the ratio (CHy/CHx) of the minor axis length (CHy) to is 80% (0.8) or more, preferably 90% (0.9) or more, is mounted on the illumination system of the exposure apparatus. do.

Claims (5)

An exposure method of illuminating a pattern on a mask with illumination light and projecting and exposing an image of the pattern onto a substrate via a projection optical system, comprising:
The specific center wavelength is λ, the numerical aperture on the substrate side of the projection optical system is NAp, and the process constant is k (0 < k ≤ 1), and the resolution (R) is determined by k·(λ/NAp). When a projection image of a square or square hole pattern of a size close to the minimum resolutionable line width dimension is projected onto the substrate, the ratio of the minor axis length to the major axis length of the projection image of the hole pattern deformed into an elliptical shape is 80% or more. A first step of setting the wavelength width of the illumination light including the center wavelength (λ) so that it is preferably 90% or more,
An exposure method comprising a second step of illuminating the pattern with the illumination light having a set wavelength and projecting and exposing the pattern on the substrate. According to claim 1,
In the first step, light from a light source is incident on the wavelength selection unit, and the illumination light with a set wavelength is transmitted,
The wavelength selection unit has a first wavelength selection element that transmits a first wavelength width, and a second wavelength selection element that transmits a second wavelength width different from the first wavelength width, and selects the first wavelength from the optical path of the light. An exposure method wherein the wavelength width of the illumination light is set to the first wavelength width or the second wavelength width by switching and arranging the element and the second wavelength selection element. According to claim 2,
The first wavelength width includes at least one bright line among a plurality of bright lines included in the light, and the second wavelength width excludes a bright line on a long wavelength side and a bright line on a short wavelength side adjacent to the at least one bright line. , an exposure method having a wavelength wider than the first wavelength. According to claim 3,
The wavelength selection unit further has at least one other bright line adjacent to the at least one bright line and a third wavelength selection element that transmits light of a third wavelength width including both of the at least one bright line,
By switching and arranging the first wavelength selection element, the second wavelength selection element, and the third wavelength selection element in the optical path, the wavelength width of the illumination light is set to the first wavelength width, the second wavelength width, or the third wavelength selection element. Exposure method set by wavelength. According to claim 3 or 4,
An exposure method wherein the at least one bright line is any one of an i line, an h line, and a g line.