Pattern exposure apparatus and device production method

Abstract

An exposure apparatus according to the present invention causes reflection light from a selected on-state micromirror in a spatial light modulation element to enter a projection unit and projects and exposes a pattern corresponding to drawing data onto a substrate. The apparatus comprises an illumination unit that emits first illumination light having a wavelength λ1 and second illumination light having a wavelength λ2 (λ2≠λ1), which are allowable in terms of chromatic aberration characteristics of the projection unit, to the spatial light modulation element at an entry angle corresponding to a multiple of the inclination angle of the on-state micromirror. When the diffraction angle of j1-th order main diffraction light that is based on the the wavelength λ1, that is generated from the on-state micromirror, and that enters the projection unit is denoted by θj1, and the diffraction angle of j2-th order main diffraction light that is based on the wavelength λ2, that is generated from the on-state micromirror, and that enters the projection unit is denoted by θj2, the difference between the wavelength λ1 and the wavelength λ2 or the entry angles are set such that the diffraction angle θj1 and the diffraction angle θj2 distribute with the optical axis of the projection unit therebetween.

Description

Pattern exposure apparatus and device manufacturing method

[0001] The present invention relates to a pattern exposure apparatus that exposes a pattern for an electronic device, and a method for manufacturing an electronic device using such a pattern exposure apparatus.

[0002] Conventionally, in the lithography process for manufacturing electronic devices (microdevices) such as liquid crystal or organic EL display panels and semiconductor elements (integrated circuits, etc.), a step-and-repeat projection exposure apparatus (a so-called stepper) or a step-and-scan projection exposure apparatus (a so-called scanning stepper (also called a scanner)) has been used. This type of exposure apparatus projects and exposes a mask pattern for the electronic device onto a photosensitive layer applied to the surface of an exposed substrate (hereinafter simply referred to as a substrate), such as a glass substrate, a semiconductor wafer, a printed wiring board, or a resin film.

[0003] Because fabricating a mask substrate on which the mask pattern is fixedly formed requires time and expense, there is known an exposure apparatus that uses a spatial light modulation element (variable mask pattern generator) such as a digital mirror device (DMD) in which a large number of micromirrors that are slightly displaced are regularly arranged instead of a mask substrate (see, for example, Patent Document 1). In the exposure apparatus disclosed in Patent Document 1, light from a light source 3 using a semiconductor laser with a wavelength of 405 nm or 365 nm is obliquely irradiated at an incident angle of 22 to 26° onto a digital mirror device (DMD) serving as a spatial light modulator 4 via an irradiation optical system 6, and light reflected from on-state pixel mirrors among the large number of pixel mirrors of the spatial light modulator 4 (DMD) is projected onto an exposure area of ​​an object W via a projection optical system 5 for exposure.

[0004] In the case of Patent Document 1, the tilt angle of the pixel mirrors (micromirrors) of the DMD is set to half the incident angle of the illumination light, which is 22 to 26 degrees. Because the numerous pixel mirrors (micromirrors) are arranged in a matrix at a constant pitch, they also function as an optical diffraction grating (blazed diffraction grating). In particular, when projecting and exposing fine patterns for electronic devices, and when illuminating the DMD with oblique light, the DMD's function as a diffraction grating (the direction of diffracted light and its intensity distribution) can degrade the imaging state of the pattern or reduce the light intensity (exposure amount) of the projected imaging light beam.

[0005] International Publication No. 2018 / 088550

[0006] According to a first aspect of the present invention, there is provided a pattern exposure apparatus that irradiates illumination light onto a spatial light modulation element having a number of micromirrors that are two-dimensionally arranged at a predetermined pitch and that are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to be incident on a projection unit, thereby projecting and exposing a pattern corresponding to the drawing data onto a substrate, the pattern exposure apparatus comprising: an illumination unit that irradiates the spatial light modulation element with first illumination light of a wavelength λ1 and second illumination light of a wavelength λ2 (λ2 ≠ λ1) at an incident angle that corresponds to twice the angle of inclination of the on-state micromirrors, and in which the difference between the wavelength λ1 and the wavelength λ2 is set so that the difference between the diffraction angle θj1 and the diffraction angle θj2 falls within a predetermined allowable range, where θj1 is the diffraction angle of the main diffracted light of order j1 that is generated from the on-state micromirror at the wavelength λ1 and reaches the substrate via the projection unit, and θj2 is the diffraction angle of the main diffracted light of order j2 that is generated from the on-state micromirror at the wavelength λ2 and reaches the substrate via the projection unit.

[0007] According to a second aspect of the present invention, there is provided a pattern exposure apparatus that irradiates illumination light onto a spatial light modulation element having a number of micromirrors that are two-dimensionally arranged at a predetermined pitch and are selectively driven based on drawing data, causes reflected light from selected micromirrors in the on-state of the spatial light modulation element to be incident on a projection unit, and projects and exposes a pattern corresponding to the drawing data onto the substrate, wherein the first illumination light has a wavelength λ1 that is allowable in terms of chromatic aberration characteristics of the projection unit, and the second illumination light has a wavelength λ2 (λ2≠λ1) that is also allowable in terms of chromatic aberration characteristics of the projection unit, The pattern exposure apparatus includes an illumination unit that irradiates the spatial light modulation element with an incident angle corresponding to a double angle of the tilt angle of the mirror, and when the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in an on-state at wavelength λ1 and incident on the projection unit is θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in an on-state at wavelength λ2 and incident on the projection unit is θj2, the difference between the wavelength λ1 and the wavelength λ2 or the incident angle is set so that the diffraction angle θj1 and the diffraction angle θj2 are distributed on either side of the optical axis of the projection unit.

[0008] According to a third aspect of the present invention, there is provided a pattern exposure apparatus that irradiates illumination light onto a spatial light modulation element having a number of micromirrors that are two-dimensionally arranged at a predetermined pitch and that are selectively driven based on drawing data, causes reflected light from selected micromirrors in an ON state of the spatial light modulation element to be incident on a projection unit, and projects and exposes a pattern corresponding to the drawing data onto the substrate, wherein a first illumination light having a wavelength λ1 that is allowable in terms of chromatic aberration characteristics of the projection unit and a second illumination light having a wavelength λ2 (λ2≠λ1) that is also allowable in terms of chromatic aberration characteristics of the projection unit are projected at an angle that is double the nominal tilt angle of the micromirrors in the ON state. and a pattern exposure apparatus is provided, the apparatus comprising: an illumination unit that irradiates the spatial light modulation element with light at a designed incident angle θα that is set so that, when the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in an on-state under the wavelength λ1 and incident on the projection unit is θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in an on-state under the wavelength λ2 and incident on the projection unit is θj2, the wavelength λ1 and the wavelength λ2 are set so that the diffraction angles θj1 and θj2 that are generated under the condition of the designed incident angle θα are distributed on one side of the optical axis of the projection unit.

[0009] According to a fourth aspect of the present invention, there is provided a device manufacturing method including the steps of forming a photosensitive layer on a substrate on which an electronic device is to be fabricated, preparing drawing data corresponding to a pattern for the electronic device, placing the substrate on which the photosensitive layer has been formed on a moving stage of a pattern exposure apparatus according to any one of the first to third aspects of the present invention and setting the drawing data in a drive control unit of the spatial light modulation element of the pattern exposure apparatus, and exposing the pattern onto the photosensitive layer of the substrate by synchronizing the movement of the substrate by the moving stage and the driving of the micromirrors of the spatial light modulation element to an on state and an off state based on the drawing data.

[0010] According to a fifth aspect of the present invention, there is provided a pattern exposure apparatus that irradiates illumination light onto a spatial light modulation element having a number of micromirrors that are two-dimensionally arranged at a predetermined pitch and are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to be incident on a projection unit, thereby projecting and exposing a pattern corresponding to the drawing data onto the substrate, the apparatus comprising: an illumination unit that irradiates the spatial light modulation element with illumination light having a predetermined wavelength width ±Δλ with respect to a center wavelength λo at an incident angle θα (θα>0°) corresponding to double the angle of inclination of the on-state micromirrors; and is a wavelength λ1, and the wavelength λo-Δλ on the short wavelength side of the illumination light is a wavelength λ2, the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in an on-state under light of wavelength λ1 and incident on the projection unit is θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in an on-state under light of wavelength λ2 and incident on the projection unit is θj2. In this pattern exposure apparatus, the wavelength width ±Δλ is set so that the overall distribution shape of the main diffracted light of order j1 and the main diffracted light of order j2 that appear at the pupil of the projection unit is deformed into an isotropic shape by the difference between the diffraction angle θj1 and the diffraction angle θj2.

[0011] 1 is a perspective view showing an outline of the external configuration of a pattern exposure apparatus EX according to the present embodiment; FIG. 2 is a diagram showing an example of the arrangement of projection areas IAn of a DMD 10 projected onto a substrate P by each projection unit PLU of a plurality of exposure module groups MU; FIG. 3 is a diagram explaining the state of splice exposure by each of four specific projection areas IA8, IA9, IA10, and IA27 in FIG. 2; FIG. 4 is an optical layout diagram showing the specific configuration of two exposure modules MU18 and MU19 lined up in the X direction (scanning exposure direction) as viewed in the XZ plane; FIG. 5 is a diagram schematically showing a state in which the DMD 10 and the illumination unit PLU are tilted by an angle θk in the XY plane; and FIG. 6 is a diagram explaining in detail the imaging state of the micromirrors of the DMD 10 by the projection unit PLU. The diagram is a schematic diagram of an MFE lens 108A as an optical integrator 108 as viewed from the exit surface side. 7 is a diagram schematically illustrating an example of the positional relationship between a point light source SPF formed on the exit surface side of the lens element EL of the MFE lens 108A of FIG. 7 and the exit end of the optical fiber bundle FBn. This diagram is a diagram schematically illustrating the appearance of a light source image formed on the pupil Ep of the projection unit PLU shown in FIG. 6. This is a light path diagram that is a simplified representation of the light path diagram of FIG. 6. This diagram is a diagram schematically illustrating the appearance of a light source image Ips formed on the pupil Ep by a component equivalent to zero-order light of the imaging light beam Sa from the DMD 10. Similar to FIG. 7, this is a schematic diagram of an elliptical light source surface of the MFE lens 108A of the optical integrator 108 as seen from the exit surface side. This diagram is a diagram schematically illustrating the behavior of the imaging light beam Sa on the optical path from the pupil Ep of the projection unit PLU shown in FIG. 6 to the substrate P. This is an enlarged perspective view of the state of a micromirror Ms of a part of the DMD 10 when power supply to the drive circuit of the DMD 10 is off. 16 is an enlarged perspective view of a portion of the mirror surface of the DMD 10 when the micromirrors Ms of the DMD 10 are in the on and off states. It shows a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane, illustrating a case where only one row of micromirrors Ms aligned in the Y' direction is in the on state. It is a view of the mirror surface of the DMD 10 as viewed in the X'Z plane, along the arrow a-a'. It is a diagram schematically illustrating, in the X'Z plane, the imaging state of the light (imaging light beam) Sa reflected from an isolated micromirror Msa as shown in FIG. 16, by the projection unit PLU.21. This is a graph showing a schematic representation of the point spread function Iea of ​​the diffraction image in the pupil Ep due to specularly reflected light Sa from an isolated micromirror Msa. This is a diagram showing a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane, illustrating a case where many micromirrors Ms adjacent in the X' direction are simultaneously turned on. This is a diagram showing the portion of the mirror surface of the DMD 10 as viewed in the X'Z plane, indicated by the arrows a-a'. This is a graph showing an example of the distribution of the angle θj of the diffracted light Idj generated from the DMD 10 in the state shown in FIGS. 20 and 21. This is a diagram showing a schematic representation of the intensity distribution of the imaging light beam in the pupil Ep when diffracted light is generated as shown in FIG. 22. This is a diagram showing the state of a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane when a line-and-space pattern is projected. This is a diagram showing the portion of the mirror surface of the DMD 10 as viewed in the X'Z plane, indicated by the arrows a-a'. 24 and 25. This figure shows an example of the distribution of the angle θj of the diffracted light Idj generated from the DMD 10 in the state shown in FIG. 26. This figure shows a schematic representation of the intensity distribution of the imaging light beam at the pupil Ep when diffracted light is generated as shown in FIG. 26. This figure shows a specific configuration of the optical path from the optical fiber bundle FBn in the illumination unit ILU shown in FIG. 4 or 6 to the MFE lens 108A. This figure shows a specific configuration of the optical path from the MFE lens 108A in the illumination unit ILU shown in FIG. 4 or 6 to the DMD 10. This figure shows an exaggerated view of the state of the point light source SPF formed on the exit surface side of the MFE lens 108A when the illumination light ILm incident on the MFE lens 108A is tilted within the X'Z plane. This figure shows a graph obtained by calculating the relationship between the wavelength λ and the telecentric error Δθt based on equation (2) or equation (3). This figure shows the wavelength-dependent characteristics of the telecentric error Δθt when the wavelength λ of the illumination light ILm is changed in the range of 280 nm to 450 nm. 1 is a diagram showing a schematic representation of wavelength distribution characteristics obtained by combining eight laser beams having a central wavelength of 343.333 nm and peak wavelengths shifted by 0.02 nm. 2 is a graph showing characteristics of telecentricity error in a wavelength λ range of 343.200 nm to 343.450 nm. 3 is a diagram showing a schematic representation of the distribution in the pupil Ep of ninth-order diffracted light Id9 from the DMD 10 that is incident on the projection unit PLU. 4 is a diagram showing an exaggerated representation of the distribution state of high-order diffracted light (referred to as j-th order diffracted light) that appears in the pupil Ep of the projection unit PLU when illumination light ILm having a wide wavelength width Δλ is used.39(A) and 39(B) are graphs showing the change in the elliptical ratio of the elliptical distribution of high-order diffracted light appearing in the pupil Ep of the projection unit PLU when the wavelength width Δλ of the illumination light is changed. FIG. 39(B) is a graph showing the relationship between the change in the elliptical ratio and the σ value of the elliptical distribution of high-order diffracted light from the DMD 10 in the pupil Ep of the projection unit PLU. FIG. 39(A) and 39(B) are graphs showing the change in the elliptical ratio and the σ value of the elliptical distribution of high-order diffracted light appearing in the pupil Ep of the projection unit PLU. FIG. 39(A) and 39(B) are graphs showing the change in the elliptical ratio and the σ value of the elliptical distribution of high-order diffracted light appearing in the pupil Ep of the projection unit PLU. FIG. 39(B) shows an example of the wavelength distribution characteristics of the illumination light ILm. FIG. 39(A) and 39(B) are graphs showing the change in the 9th-order diffracted light distribution H9c and the 8th-order diffracted light distribution H8c appearing in the pupil Ep of the projection unit PLU. 41 is a diagram showing an optical arrangement according to a modification of the embodiment of FIG. 40. FIG. 42 is an optical arrangement diagram further modified from the configuration of the modification of FIG. 42. FIG. 43 is an exaggerated diagram showing an example of the arrangement of each of the illumination areas Imf1 and Imf2 projected onto the plane of the incident end pff of the MFE lens 108A. FIG. 44 is an exaggerated diagram showing another example of the arrangement of each of the illumination areas Imf1 and Imf2 projected onto the plane of the incident end pff of the MFE lens 108A. FIG. 45 is a diagram for schematically explaining the angular state when diffracted light Idj from a large number of micromirrors Msa in the on state is incident on the projection unit PLU. FIG. 46 is a diagram showing the point spread function Iea and the distributions of 8th- to 10th-order diffracted light Id8, Id9, and Id10 that appear when the error angle Δθd of the micromirror Msa in the on state is zero. 49 is a graph showing a simulation of the point spread function when the tilt angle θd of the micromirror Msa in the on state is changed by ±0.5° as an error angle Δθd from the design tilt angle θo, with respect to the characteristics shown in Fig. 47. This graph shows the point spread function Iea when the wavelength λ1 is 343.333 nm and the point spread function IeaL when the wavelength λ2 is 355.000 nm, with the error angle Δθd of the micromirror Msa in the on state being zero. This graph shows the characteristics of the point spread functions Iea and IeaL when the error angle Δθd of the micromirror Msa in the on state is +0.5°, with respect to the state shown in Fig. 49.10 is a graph simulating the change in light intensity according to the error angle Δθd of ninth-order diffracted light Id9(λ1), Id9(λ2), and Id9(λ3) generated under each of three wavelengths λ1, λ2, and λ3.

[0012] A pattern exposure apparatus (pattern formation apparatus) according to an aspect of the present invention will be described in detail below with reference to the accompanying drawings, showing preferred embodiments. It should be noted that the aspects of the present invention are not limited to these embodiments and include various modifications or improvements. That is, the components described below include those that would be easily conceivable to a person skilled in the art and those that are substantially identical, and the components described below can be combined as appropriate. Furthermore, various omissions, substitutions, or modifications of the components can be made without departing from the spirit of the present invention. The same reference numerals are used throughout the drawings and the following detailed description for members and components that achieve the same or similar functions.

[0013] [Overall Configuration of Pattern Exposure Apparatus] FIG. 1 is a perspective view showing the outline of the external configuration of a pattern exposure apparatus (hereinafter also referred to simply as an exposure apparatus) EX according to this embodiment. The exposure apparatus EX is an apparatus that projects exposure light, the intensity distribution of which is dynamically modulated in space by a spatial light modulation element (digital mirror device: DMD), onto a substrate to be exposed. In a specific embodiment, the exposure apparatus EX is a step-and-scan projection exposure apparatus (scanner) that uses a rectangular (square) glass substrate used in display devices (flat panel displays) as the exposure target. The glass substrate is a substrate P for flat panel displays, with at least one side or diagonal length of 500 mm or more and a thickness of 1 mm or less. The exposure apparatus EX exposes a projected image of a pattern created by the DMD onto a photosensitive layer (photoresist) formed to a constant thickness on the surface of the substrate P. After exposure, the substrate P is transported from the exposure apparatus EX and sent to a predetermined process (such as a film formation process, etching process, or plating process) after a development process.

[0014] The exposure apparatus EX is equipped with a stage device that is made up of a pedestal 2 placed on active vibration isolation units 1a, 1b, 1c, and 1d (1d not shown), a surface plate 3 placed on the pedestal 2, an XY stage 4A that is movable two-dimensionally on the surface plate 3, a substrate holder 4B that holds a substrate P on a plane by suction on the XY stage 4A, and laser length measurement interferometers (hereinafter simply referred to as interferometers) IFX, IFY1 to IFY4 that measure the two-dimensional movement position of the substrate holder 4B (substrate P). Such a stage device is disclosed, for example, in U.S. Patent Publication No. 2010 / 0018950 and U.S. Patent Publication No. 2012 / 0057140.

[0015] In FIG. 1 , the XY plane of the Cartesian coordinate system XYZ is set parallel to the flat surface of the stage device's base 3, and the XY stage 4A is set to be able to translate within the XY plane. In this embodiment, the direction parallel to the X axis of the coordinate system XYZ is set as the scanning movement direction of the substrate P (XY stage 4A) during scan exposure. The movement position of the substrate P in the X-axis direction is sequentially measured by the interferometer IFX, and the movement position in the Y-axis direction is sequentially measured by at least one (preferably two or more) of the four interferometers IFY1 to IFY4. The substrate holder 4B is configured to be able to move slightly relative to the XY stage 4A in the Z-axis direction, which is perpendicular to the XY plane, and to be able to tilt slightly in any direction relative to the XY plane, thereby actively adjusting the focus and leveling (parallelism) between the surface of the substrate P and the imaging plane of the projected pattern. Furthermore, the substrate holder 4B is configured to be able to rotate slightly (θz rotation) about an axis parallel to the Z axis in order to actively adjust the tilt of the substrate P in the XY plane.

[0016] The exposure apparatus EX further includes an optical base 5 that holds multiple exposure (drawing) module groups MU(A), MU(B), and MU(C), and main columns 6a, 6b, 6c, and 6d (6d is not shown) that support the optical base 5 from the pedestal 2. Each of the multiple exposure module groups MU(A), MU(B), and MU(C) includes an illumination unit ILU that is attached to the +Z side of the optical base 5 and that receives illumination light from an optical fiber unit FBU, and a projection unit PLU that is attached to the -Z side of the optical base 5 and has an optical axis parallel to the Z axis. Each of the exposure module groups MU(A), MU(B), and MU(C) further includes a digital mirror device (DMD) 10 that serves as a light modulation unit that reflects illumination light from the illumination unit ILU in the -Z direction and causes it to enter the projection unit PLU. A detailed configuration of the exposure module group including the illumination unit ILU, DMD 10, and projection unit PLU will be described later.

[0017] A plurality of alignment systems (microscopes) ALG are attached to the -Z direction side of the optical surface plate 5 of the exposure apparatus EX, and detect alignment marks formed at a plurality of predetermined positions on the substrate P. A calibration reference unit CU is provided at the -X direction end of the substrate holder 4B to confirm (calibrate) the relative positional relationship in the XY plane of the detection fields of the alignment systems ALG, to confirm (calibrate) the baseline error between the projection positions of the pattern images projected from the projection units PLU of each of the exposure module groups MU(A), MU(B), and MU(C) and the positions of the detection fields of the alignment systems ALG, or to confirm the position and image quality of the pattern images projected from the projection units PLU. Although some of the modules are not shown in FIG. 1 , in this embodiment, each of the exposure module groups MU(A), MU(B), and MU(C) includes, for example, nine modules arranged at regular intervals in the Y direction, although the number of modules may be more or less than nine.

[0018] 2 is a diagram showing an example of the arrangement of projection areas IAn of the digital mirror device (DMD) 10 projected onto the substrate P by the projection units PLU of each of the exposure module groups MU(A), MU(B), and MU(C), with the Cartesian coordinate system XYZ set to the same as in FIG. 1. In this embodiment, the first row of exposure module groups MU(A), the second row of exposure module groups MU(B), and the third row of exposure module groups MU(C), which are spaced apart in the X direction, each comprise nine modules arranged in the Y direction. The exposure module group MU(A) comprises nine modules MU1 to MU9 arranged in the +Y direction, the exposure module group MU(B) comprises nine modules MU10 to MU18 arranged in the -Y direction, and the exposure module group MU(C) comprises nine modules MU19 to MU27 arranged in the +Y direction. All modules MU1 to MU27 have the same configuration, and when exposure module group MU(A) and exposure module group MU(B) are positioned opposite each other in the X direction, exposure module group MU(B) and exposure module group MU(C) are positioned back to back in the X direction.

[0019] 2, the shape of the projection areas IA1, IA2, IA3, ..., IA27 (sometimes represented as IAn, where n is 1 to 27) by each of the modules MU1 to MU27 is, as an example, a rectangle extending in the Y direction with an aspect ratio of approximately 1:2. In this embodiment, as the substrate P is scanned and moved in the +X direction, spliced ​​exposure is performed at the -Y direction end of each of the first row of projection areas IA1 to IA9 and the +Y direction end of each of the second row of projection areas IA10 to IA18. Then, areas on the substrate P that were not exposed by each of the first and second rows of projection areas IA1 to IA18 are spliced ​​and exposed by each of the third row of projection areas IA19 to IA27. The center point of each of the projection areas IA1 to IA9 in the first column is located on a line k1 parallel to the Y axis, the center point of each of the projection areas IA10 to IA18 in the second column is located on a line k2 parallel to the Y axis, and the center point of each of the projection areas IA19 to IA27 in the third column is located on a line k3 parallel to the Y axis. The distance in the X direction between lines k1 and k2 is set to a distance XL1, and the distance in the X direction between lines k2 and k3 is set to a distance XL2.

[0020] Here, assuming that the joint portion between the −Y direction end of projection area IA9 and the +Y direction end of projection area IA10 is OLa, the joint portion between the −Y direction end of projection area IA10 and the +Y direction end of projection area IA27 is OLb, and the joint portion between the +Y direction end of projection area IA8 and the −Y direction end of projection area IA27 is OLc, the state of the joint exposure will be described with reference to Figure 3. In Figure 3, the Cartesian coordinate system XYZ is set in the same way as in Figures 1 and 2, and the coordinate system X'Y' in projection areas IA8, IA9, IA10, and IA27 (and all other projection areas IAn) is set so as to be inclined at an angle θk with respect to the X and Y axes (lines k1 to k3) of the Cartesian coordinate system XYZ. That is, the entire DMD 10 is tilted by an angle θk in the XY plane so that the two-dimensional arrangement of the many micromirrors of the DMD 10 forms the coordinate system X'Y'.

[0021] 3, the circular area encompassing each of the projection areas IA8, IA9, IA10, and IA27 (and all other projection areas IAn) represents the circular image field PLf' of the projection unit PLU. At the joint OLa, the projected image of the micromirrors arranged diagonally (at an angle θk) at the end of the projection area IA9 in the -Y' direction is set to overlap with the projected image of the micromirrors arranged diagonally (at an angle θk) at the end of the projection area IA10 in the +Y' direction. At the joint OLb, the projected image of the micromirrors arranged diagonally (at an angle θk) at the end of the projection area IA10 in the -Y' direction is set to overlap with the projected image of the micromirrors arranged diagonally (at an angle θk) at the end of the projection area IA27 in the +Y' direction. Similarly, at the joint OLc, the projected image of the micromirror arranged diagonally (at an angle θk) at the end of the +Y' direction of the projection area IA8 and the projected image of the micromirror arranged diagonally (at an angle θk) at the end of the -Y' direction of the projection area IA27 are set to overlap.

[0022] [Configuration of the Illumination Unit] Figure 4 is an optical layout diagram showing the specific configuration of module MU18 in the exposure module group MU(B) shown in Figures 1 and 2 and module MU19 in the exposure module group MU(C) as viewed in the XZ plane. The Cartesian coordinate system XYZ in Figure 4 is set to be the same as the Cartesian coordinate system XYZ in Figures 1 to 3. As is clear from the arrangement of each module in the XY plane shown in Figure 2, module MU18 is shifted by a fixed distance in the +Y direction relative to module MU19, and they are installed back-to-back. Since the optical components in module MU18 and module MU19 are made of the same materials and configured identically, the optical configuration of module MU18 will be mainly described in detail here. Note that the optical fiber unit FBU shown in Figure 1 is composed of 27 optical fiber bundles FB1 to FB27 corresponding to the 27 modules MU1 to MU27 shown in Figure 2.

[0023] The illumination unit ILU of the module MU18 is composed of a mirror 100 that reflects illumination light ILm traveling in the −Z direction from the output end of the optical fiber bundle FB18, a mirror 102 that reflects the illumination light ILm from the mirror 100 in the −Z direction, an input lens system 104 that acts as a collimator lens, an illuminance adjustment filter 106, an optical integrator 108 that includes a micro fly's eye (MFE) lens, a field lens, etc., a condenser lens system 110, and an inclined mirror 112 that reflects the illumination light ILm from the condenser lens system 110 toward the DMD 10. The mirror 102, the input lens system 104, the optical integrator 108, the condenser lens system 110, and the inclined mirror 112 are arranged along an optical axis AXc that is parallel to the Z axis.

[0024] The optical fiber bundle FB18 is composed of a single optical fiber line or a bundle of multiple optical fiber lines. The numerical aperture (NA, also referred to as the divergence angle) of the illumination light ILm emitted from the output end of the optical fiber bundle FB18 (each optical fiber line) is set so that it can enter the downstream input lens system 104 without being eclipsed. The position of the front focal point of the input lens system 104 is designed to be the same as the position of the output end of the optical fiber bundle FB18. Furthermore, the position of the back focal point of the input lens system 104 is set so that the illumination light ILm from a single or multiple point light sources formed at the output end of the optical fiber bundle FB18 is superimposed on the incident surface of the MFE lens 108A of the optical integrator 108. Therefore, the incident surface of the MFE lens 108A is Koehler illuminated by the illumination light ILm from the output end of the optical fiber bundle FB18. In the initial state, the geometric center point of the output end of the optical fiber bundle FB18 in the XY plane is located on the optical axis AXc, and the chief ray (center line) of the illumination light ILm from the point light source at the output end of the optical fiber bundle is parallel to (or coaxial with) the optical axis AXc.

[0025] The illumination light ILm from the input lens system 104 is attenuated by an illuminance adjustment filter 106 to a value between 0% and 90%, and then passes through an optical integrator 108 (MFE lens 108A, field lens, etc.) and enters a condenser lens system 110. The MFE lens 108A is a two-dimensional array of numerous rectangular microlenses, each measuring several tens of micrometers square, and its overall shape is set to be approximately similar to the overall shape of the mirror surface of the DMD 10 (aspect ratio of approximately 1:2) in the XY plane. The position of the front focal point of the condenser lens system 110 is set to be approximately the same as the position of the exit surface of the MFE lens 108A. Therefore, each of the illumination light beams from the point light sources formed on the exit side of each of the numerous microlenses of the MFE lens 108A is converted into approximately parallel beams by the condenser lens system 110, reflected by an inclined mirror 112, and then superimposed on the DMD 10 to form a uniform illuminance distribution.

[0026] The exit surface of the MFE lens 108A generates a surface light source in which a large number of point light sources (light-converging points) are densely arranged two-dimensionally, and therefore functions as a surface light source member. Such an MFE lens 108A may be configured, for example, as disclosed in Japanese Patent Application Laid-Open No. 2004-045885, in which a plurality of cylindrical micro fly's eye lens elements, each formed by arranging a large number of cylindrical lenses on each of the entrance surface side and exit surface side of the illumination light, are arranged at predetermined intervals in the optical axis direction.

[0027] In module MU18 shown in FIG. 4 , optical axis AXc, which is parallel to the Z axis and passes through condenser lens system 110, is bent by tilted mirror 112 and reaches DMD 10. The optical axis between tilted mirror 112 and DMD 10 is referred to as optical axis AXb. In this embodiment, a neutral plane including the center points of each of the numerous micromirrors of DMD 10 is set parallel to the XY plane. Therefore, the angle between the normal to this neutral plane (parallel to the Z axis) and optical axis AXb is the angle of incidence θα of illumination light ILm with respect to DMD 10. DMD 10 is attached to the lower side of mount 10M, which is fixed to the support column of illumination unit ILU. Mount 10M is provided with a fine-motion stage combining a parallel link mechanism and an extendable piezoelectric element, such as that disclosed in International Patent Publication No. 2006 / 120927, to fine-tune the position and orientation of DMD 10.

[0028] Illumination light ILm irradiated onto micromirrors of DMD 10 that are in the On state is reflected in the X direction within the XZ plane so as to head toward the projection unit PLU. On the other hand, illumination light ILm irradiated onto micromirrors of DMD 10 that are in the Off state is reflected in the Y direction within the YZ plane so as not to head toward the projection unit PLU. As will be described in more detail below, DMD 10 in this embodiment is of a roll and pitch drive type that switches between the On state and the Off state by tilting the micromirrors in the roll direction and the pitch direction.

[0029] A movable shutter 114 is removably provided in the optical path between the DMD 10 and the projection unit PLU to block light reflected from the DMD 10 during non-exposure periods. As shown on the module MU19 side, the movable shutter 114 is rotated to an angular position where it is removed from the optical path during exposure periods, and as shown on the module MU18 side, it is rotated to an angular position where it is inserted obliquely into the optical path during non-exposure periods. A reflective surface is formed on the DMD 10 side of the movable shutter 114, and light reflected therefrom from the DMD 10 is irradiated onto a light absorber 115. The light absorber 115 absorbs light energy in the ultraviolet wavelength range (wavelengths of 400 nm or less) without re-reflecting it and converts it into heat energy. For this reason, the light absorber 115 is also provided with a heat dissipation mechanism (heat dissipation fins and a cooling mechanism). Although not shown in FIG. 4, the reflected light from the micromirrors of DMD 10, which are in the off state during the exposure period, is absorbed by a similar light absorber (not shown in FIG. 4) installed in the Y direction (a direction perpendicular to the plane of the paper in FIG. 4) with respect to the optical path between DMD 10 and projection unit PLU.

[0030] [Configuration of Projection Unit] The projection unit PLU attached to the underside of the optical surface plate 5 is configured as a double-telecentric imaging projection lens system composed of a first lens group 116 and a second lens group 118 arranged along an optical axis AXa parallel to the Z axis. The first lens group 116 and the second lens group 118 are each configured to move translationally in the direction along the Z axis (optical axis AXa) by a micro-motion actuator relative to a support column fixed to the underside of the optical surface plate 5. The projection magnification Mp of the imaging projection lens system formed by the first lens group 116 and the second lens group 118 is determined by the relationship between the array pitch Pd of the micromirrors on the DMD 10 and the minimum line width (minimum pixel dimension) Pg of the pattern projected within the projection area IAn (n = 1 to 27) on the substrate P.

[0031] As an example, if the required minimum line width (minimum pixel dimension) Pg is 1 μm and the micromirror array pitch Pd is 5.4 μm, the projection magnification Mp is set to approximately 1 / 6, taking into consideration the tilt angle θk in the XY plane of the projection area IAn (DMD 10) described above in Figure 3. The imaging projection lens system consisting of lens groups 116 and 118 inverts / reflects a reduced image of the entire mirror surface of DMD 10 and forms an image on the projection area IA18 (IAn) on the substrate P.

[0032] The first lens group 116 of the projection unit PLU is capable of slight movement in the direction of the optical axis AXa by an actuator in order to make fine adjustments (on the order of ±several tens of ppm) to the projection magnification Mp, and the second lens group 118 is capable of slight movement in the direction of the optical axis AXa by an actuator in order to make high-speed focus adjustments. Furthermore, in order to measure positional changes in the Z-axis direction of the surface of the substrate P with an accuracy of submicron or less, a plurality of oblique incidence focus sensors 120 are provided below the optical surface plate 5. The plurality of focus sensors 120 measure overall positional changes in the Z-axis direction of the substrate P, positional changes in the Z-axis direction of partial areas on the substrate P corresponding to each of the projection areas IAn (n = 1 to 27), partial tilt changes of the substrate P, etc.

[0033] As explained above in Figure 3, the illumination unit ILU and projection unit PLU described above require that the projection area IAn be tilted by an angle θk in the XY plane, and therefore the DMD 10 and illumination unit PLU in Figure 4 (at least the optical path portion of mirrors 102 to 112 along the optical axis AXc) are arranged so that they are tilted overall by an angle θk in the XY plane.

[0034] Figure 5 is a schematic diagram showing the state in the XY plane in which the DMD 10 and illumination unit PLU are tilted by an angle θk in the XY plane. In Figure 5, the Cartesian coordinate system XYZ is the same as the coordinate systems XYZ in Figures 1 to 4, and the coordinate system X'Y' for arranging the micromirrors Ms of the DMD 10 is the same as the coordinate system X'Y' shown in Figure 3. The circle containing the DMD 10 is the image field PLf on the object side of the projection unit PLU, and the optical axis AXa is located at its center. Meanwhile, the optical axis AXb, which is obtained by bending the optical axis AXc that passes through the condenser lens system 110 of the illumination unit ILU by the tilted mirror 112, is positioned so that it is tilted by an angle θk from a line Lu that is parallel to the X axis, when viewed in the XY plane.

[0035] [Imaging Light Path by DMD] Next, with reference to Figure 6, the imaging state of the micromirrors Ms of the DMD 10 by the projection unit PLU (imaging projection lens system) will be described in detail. The Cartesian coordinate system X'Y'Z in Figure 6 is the same as the coordinate systems X'Y'Z shown in Figures 3 and 5, and Figure 6 illustrates the optical path from the condenser lens system 110 of the illumination unit ILU to the substrate P. Illumination light ILm from the condenser lens system 110 travels along the optical axis AXc, is totally reflected by the inclined mirror 112, and reaches the mirror surface of the DMD 10 along the optical axis AXb. Here, the micromirror Ms located at the center of the DMD 10 is designated Msc, and the micromirrors Ms located on the periphery are designated Msp, and it is assumed that these micromirrors Msc and Msp are in the On state.

[0036] If the tilt angle of micromirror Ms in the On state is, for example, 17.5° with respect to the X'Y' plane (XY plane), as a standard value, then the angle of incidence θα (the angle of optical axis AXb from optical axis AXa) of illumination light ILm irradiated onto DMD 10 is set to 35.0° in order to make the chief rays of reflected light Sac and Sap from each of micromirrors Msc and Msp parallel to the optical axis AXa of projection unit PLU. Accordingly, in this case, the reflective surface of tilted mirror 112 is also tilted by 17.5° (= θα / 2) with respect to the X'Y' plane (XY plane). The chief ray Lc of reflected light Sac from micromirror Msc is coaxial with the optical axis AXa, and the chief ray La of reflected light Sap from micromirror Msp is parallel to the optical axis AXa, and the reflected light Sac and Sap enter projection unit PLU with a predetermined numerical aperture (NA).

[0037] The reflected light Sac forms a telecentric image of the micromirror Msc, which has been reduced by the projection magnification Mp of the projection unit PLU, on the substrate P at the position of the optical axis AXa. Similarly, the reflected light Sap forms a telecentric image of the micromirror Msp, which has been reduced by the projection magnification Mp of the projection unit PLU, on the substrate P at a position away from the reduced image ic in the +X' direction. As an example, the first lens system 116 of the projection unit PLU is composed of three lens groups G1, G2, and G3, and the second lens system 118 is composed of two lens groups G4 and G5. An exit pupil (also simply referred to as a pupil) Ep is set between the first lens system 116 and the second lens system 118. At the position of the pupil Ep, a light source image of the illumination light ILm (a collection of numerous point light sources formed on the exit surface side of the MFE lens 108A) is formed, resulting in a Koehler illumination configuration. The pupil Ep is also called the aperture of the projection unit PLU, and the size (diameter) of this aperture is one factor that determines the resolving power of the projection unit PLU. Note that the position of the pupil Ep corresponds to the position of the aperture stop of the projection unit PLU.

[0038] The specularly reflected light from the micromirror Ms of the DMD 10 in the On state is set to pass through without being blocked by the maximum aperture (diameter) of the pupil Ep, and the numerical aperture NAi (also referred to as the maximum numerical aperture NAi(max)) on the image side (substrate P side) in the equation R=k1·(λ / NAi), which expresses the resolution R, is determined by the maximum aperture of the pupil Ep and the distance of the rear (image side) focal point of the projection unit PLU (lens group G1 to G5 as an imaging projection lens system). Furthermore, the numerical aperture NAo (also referred to as the maximum numerical aperture NAo(max)) on the object plane (DMD 10) side of the projection unit PLU (lens group G1 to G5) is expressed as the product of the projection magnification Mp and the numerical aperture NAi, and when the projection magnification Mp is 1 / 6, then NAo=NAi / 6 [NAo(max)=NAi(max) / 6].

[0039] 6 and 4, the exit ends of the optical fiber bundles FBn (n = 1 to 27) connected to each module MUn (n = 1 to 27) are set by the input lens system 104 to be optically conjugate with the exit end side of the MFE lens 108A of the optical integrator 108, and the entrance end side of the MFE lens 108A is set by the condenser lens system 110 to be optically conjugate with the center of the mirror surface (neutral plane) of the DMD 10. As a result, the illumination light ILm irradiated onto the entire mirror surface of the DMD 10 has a uniform illuminance distribution (for example, intensity unevenness within ±1%) due to the action of the optical integrator 108. Furthermore, the surface light source (a collection of numerous point light sources SPF) on the exit end side of the MFE lens 108A and the plane of the pupil Ep of the projection unit PLU are set in an optically conjugate relationship by the condenser lens system 110 and the lens groups G1 to G3 of the projection unit PLU.

[0040] 7 is a schematic diagram of the MFE lens 108A of the optical integrator 108, viewed from the exit surface side. The MFE lens 108A is configured by densely arranging a large number of lens elements EL, each having a rectangular cross section extending in the Y' direction within the X'Y' plane, whose cross-sectional shape is similar to the shape of the entire mirror surface (image formation area) of the DMD 10, in the X' and Y' directions. Illumination light ILm from the input lens system 104 shown in FIG. 4 is irradiated onto the entrance surface side of the MFE lens 108A, forming a substantially circular irradiation area Ef. The irradiation area Ef has a shape similar to each exit end of the single or multiple optical fiber lines of the optical fiber bundle FB18 (FBn) shown in FIG. 4, and is designed to be a circular area centered on the optical axis AXc.

[0041] Of the numerous lens elements EL of the MFE lens 108A, those lens elements EL located within the illumination area Ef have point light sources SPF produced by illumination light ILm from the output ends of the optical fiber bundles FB18 (FBn) densely distributed within a substantially circular area on the output surface side of each. The circular area APh in Figure 7 represents the aperture range when an aperture stop having a circular opening is provided on the output surface side of the MFE lens 108A. The actual illumination light ILm is produced by the numerous point light sources SPF scattered within the circular area APh, and light from point light sources SPF outside the circular area APh is blocked.

[0042] 8A, 8B, and 8C are schematic diagrams illustrating an example of the positional relationship between a point light source SPF formed on the output surface side of the lens element EL of the MFE lens 108A in FIG. 7 and the output end of the optical fiber bundle FBn. The coordinate system X'Y' in each of FIG. 8A, 8B, and 8C is the same as the coordinate system X'Y' set in FIG. 7. FIG. 8A illustrates a case where the optical fiber bundle FBn is a single optical fiber line, FIG. 8B illustrates a case where two optical fiber lines are arranged in the X' direction as the optical fiber bundle FBn, and FIG. 8C illustrates a case where three optical fiber lines are arranged in the X' direction as the optical fiber bundle FBn.

[0043] The output end of the optical fiber bundle FBn and the output surface of the MFE lens 108A (lens element EL) are set in an optically conjugate relationship (imaging relationship), so when the optical fiber bundle FBn is a single optical fiber line, a single point light source SPF is formed at the center position of the output surface side of the lens element EL, as shown in Figure 8(A). When two optical fiber lines are bundled in the X' direction as the optical fiber bundle FBn, the geometric centers of the two point light sources SPF are formed at the center position of the output surface side of the lens element EL, as shown in Figure 8(B). Similarly, when three optical fiber lines are bundled in the X' direction as the optical fiber bundle FBn, the geometric centers of the three point light sources SPF are formed at the center position of the output surface side of the lens element EL, as shown in Figure 8(C).

[0044] If the illumination light ILm from the optical fiber bundle FBn has a high power and the point light source SPF converges on the exit surface of each lens element EL of the MFE lens 108A, which serves as a surface light source component or optical integrator, damage (such as clouding or burning) to each lens element EL may occur. In this case, the convergence position of the point light source SPF may be set in space slightly offset from the exit surface of the MFE lens 108A (the exit surface of the lens element EL). A configuration in which the position of the point light source (focusing point) is offset to the outside of the lens element in an illumination system using a fly's eye lens is disclosed, for example, in U.S. Pat. No. 4,939,630.

[0045] 9 is a schematic diagram showing the state of the light source image Ips formed on the pupil Ep in the second lens system 118 of the projection unit PL shown in FIG. 6 when the entire mirror surface of the DMD 10 is assumed to be a single plane mirror tilted by an angle θα / 2 so as to be parallel to the inclined mirror 112 shown in FIG. 6. The light source image Ips shown in FIG. 9 is a re-image of a number of point light sources SPF (which form a surface light source aggregated in a substantially circular shape) formed on the exit surface side of the MFE lens 108A. In this case, no diffracted light or scattered light is generated from the single plane mirror placed in place of the DMD 10, and only a light source image Ips consisting of specularly reflected light (zeroth-order light) is generated at the center of the pupil Ep, coaxial with the optical axis AXa.

[0046] 9, when the radius corresponding to the maximum diameter of the pupil Ep is re and the radius corresponding to the effective diameter of the light source image Ips as a surface light source is ri, the σ value representing the size (area) of the light source image Ips relative to the size (area) of the pupil Ep is σ=ri / re. The σ value may be changed as appropriate to improve the line width or density of the pattern to be projected and exposed, or the depth of focus (DOF). The σ value can be changed by providing a variable aperture stop at a position on the exit surface side of the MFE lens 108A or at a position of the pupil Ep between the first lens system 116 and the second lens system 118 (conjugate with the circular area APh in FIG. 7).

[0047] In this type of exposure apparatus EX, the pupil Ep of the projection unit PLU is often used at its maximum diameter, so changing the σ value is mainly performed with a variable aperture stop provided on the exit surface side of the MFE lens 108A. In that case, the radius ri of the light source image Ips is defined by the radius of the circular area APh in Figure 7. Of course, a variable aperture stop may be provided in the pupil Ep of the projection unit PLU to adjust the σ value and depth of focus (DOF).

[0048] However, when the neutral plane of DMD 10 is made perpendicular to the optical axis AXa of projection unit PLU and illumination light ILm is set at a relatively large angle of incidence θα (for example, θα≧20°), it has been found that the intensity distribution at pupil Ep of the imaging light flux caused by the light reflected from micromirror Msa (or Msc) of DMD 10 in the on state does not become the distribution of the light source image Ips defined by a circular contour as shown in Figure 9, but becomes elliptical. This will be explained with reference to Figure 10.

[0049] Figure 10 is a simplified optical path diagram of Figure 6, and the Cartesian coordinate system X'Y'Z is set to the same as in Figure 6. For ease of explanation, the tilted mirror 112 shown in Figure 6 has been omitted. In Figure 10, the tilt angle θd of the micromirror Msa of the DMD 10 in the on state is set to a design value of 17.5° with respect to the neutral plane Pcc. Therefore, the angle between the optical axis AXb passing through the MFE lens 108A and the condenser lens system 110 and the optical axis AXa of the projection unit PLU, i.e., the angle of incidence θα, is set to 35° in the X'Z plane.

[0050] Of the numerous point light sources SPF formed on the exit side of the MFE lens 108A, illumination light ILma and ILmb from two point light sources SPFa and SPFb located on the outermost periphery of the circular area APh shown in FIG. 7 within a plane parallel to the X'Z plane including the optical axis AXb illuminate the entire DMD 10 via the condenser lens system 110. Central rays LLa and LLb of illumination light ILma and ILmb, respectively, are parallel to the optical axis AXb until they enter the condenser lens system 110. Therefore, when the surface light source (collection of point light sources SPF) on the exit side of the MFE lens 108A is viewed from the DMD 10 side, its shape is a circle CL1.

[0051] Here, assuming that the reflective surfaces of the numerous micromirrors of the DMD 10 are all parallel to the neutral plane Pcc, the illumination light ILma and ILmb travel as specularly reflected light along an optical axis AXb' that is tilted with respect to the optical axis AXa at an angle (-θα) symmetrical to the optical axis AXb. Also, assume that the principal surface of the first lens group 116 of the projection unit PLU and the principal surface of the condenser lens system 110 are located on an arc Prr centered at the intersection of the neutral plane Pcc of the DMD 10 and the optical axis AXa. When viewed from the side indicated by the arrow Arw1, the specularly reflected light traveling along the optical axis AXb' appears as a circle CL2, similar to the surface light source (a collection of point light sources SPF) on the exit side of the MFE 108A.

[0052] However, when viewed from the side of arrow Arw2 parallel to the optical axis AXa of projection unit PLU, the specularly reflected light traveling along optical axis AXb' appears as an ellipse CL2' because it views the circular surface light source (a collection of point light sources SPF) on the exit side of MFE lens 108A at an angle. On the other hand, when projecting a pattern by driving DMD 10, reflected light (and diffracted light) generated from many on-state micromirrors Msa becomes imaging light beam Sa and enters first lens group 116 of projection unit PLU. Since the first lens group 116 and the condenser lens system 110 are arranged along separate optical axes AXa and AXb, each tilted at an angle θα, when the intensity distribution of the zeroth-order light-equivalent component of the imaging light beam Sa generated from the micromirror Msa of the DMD 10 in the on state (the distribution of the image of the point light source SPF) is viewed on the pupil Ep, it appears elliptical CL3 because the circular surface light source on the exit surface side of the MFE lens 108A is viewed obliquely.

[0053] When the distribution of the surface light source on the exit surface side of the MFE lens 108A is a perfect circle centered on the optical axis AXb, the intensity distribution of the ellipse CL3 of the imaging light beam Sa (component equivalent to zero-order light) formed on the pupil Ep of the projection unit PLU is compressed in the incident direction of the illumination light ILm when viewed in the X'Y' plane. Because the incident direction of the illumination light ILm to the DMD 10 is the X' direction in the X'Y' plane, the major axis of the intensity distribution of the ellipse CL3 is parallel to the Y' axis and the minor axis is parallel to the X' axis. If the dimension of the major axis of the intensity distribution of the ellipse CL3 is Uy' and the dimension of the minor axis is Ux', the ratio Ux' / Uy' of the ellipses is cos θα, depending on the incident angle θα of the illumination light ILm. Since the incident angle θα is twice the tilt angle θd of the micromirror Msa of the DMD 10 in the on state, the ratio Ux' / Uy' may be set to cos(2·θd). When the incident angle θα is set to 35°, the ratio Ux' / Uy' is approximately 0.82.

[0054] 11, like FIG. 9, is a schematic diagram illustrating the light source image Ips formed on the pupil Ep by the zeroth-order equivalent component of the imaging light beam Sa from the DMD 10, which has the greatest intensity. The light source image Ips (ellipse CL3) has a radial dimension in the Y' direction with the same radius ri as in FIG. 9, and a radial dimension in the X' direction with a radius ri' that is approximately 0.82 times smaller than the radius ri. Thus, if the intensity distribution (distribution of the light source image Ips) formed on the pupil Ep by the zeroth-order equivalent component of the imaging light beam Sa is anisotropic, differences in the imaging characteristics of the edge portion may occur depending on the direction of the edge in the X'Y' plane (i.e., the XY plane) of the pattern projected onto the substrate P via the projection unit PLU. Therefore, it is generally desirable for the intensity distribution formed on the pupil Ep by the zeroth-order equivalent component of the imaging light beam Sa to have an isotropic circular shape.

[0055] Therefore, in this embodiment, the circular region APh of the aperture shape of the aperture stop provided on the exit surface side of the MFE lens 108A described above in FIG. 7 is modified into an elliptical region APh' with its major axis in the X' direction and its minor axis in the Y' direction, as shown in FIG. 12. Like FIG. 7, FIG. 12 is a schematic diagram of the MFE lens 108A of the optical integrator 108 viewed from the exit surface side. The elliptical region APh' is obtained by rotating the ellipse CL3 of the light source image Ips formed on the pupil Ep of the projection unit PLU by 90° within the X'Y' plane. Furthermore, the ellipse ratio (minor axis dimension / major axis dimension) of the elliptical region APh' is set to cos θα, the same as the ratio of the ellipse CL3 shown in FIG. 10.

[0056] In this way, by making the effective overall shape (contour) of the surface light source (collection of point light sources SPF) formed on the exit surface side of the MFE lens 108A elliptical, the intensity distribution (light source image Ips) of the zeroth-order light equivalent component of the imaging light beam Sa formed on the pupil Ep of the projection unit PLU can be made circular, and the imaging characteristics (particularly the contrast characteristics of the edges) can be made uniform regardless of the direction in which the edge of the pattern extends within the X'Y' plane (within the XY plane).

[0057] [Telecentric Error During Projection Exposure] Next, we will explain telecentric errors that can occur in an exposure apparatus EX that uses a DMD 10 as in this embodiment, but before that, we will briefly explain one of the causes of telecentric errors using Figure 13. Figures 13(A) and 13(B) are diagrams that schematically show the behavior of the imaging light beam Sa on the optical path from the pupil Ep shown in Figure 6 to the substrate P via the second lens group 118. The Cartesian coordinate system X'Y'Z in Figures 13(A) and 13(B) is the same as the coordinate system X'Y'Z in Figure 6. For simplicity of explanation, we will assume that the entire mirror surface of the DMD 10 is a single plane mirror that is tilted by an angle θα / 2 parallel to the inclined mirror 112 in Figure 6. 13A and 13B, lens groups G4 and G5 are arranged along the optical axis AXa between the pupil Ep and the substrate P, and an elliptical light source image (surface light source image) Ips is formed within the pupil Ep as shown in Fig. 11. Note that the chief ray of reflected light (imaging light beam) Sa that passes through a point on the periphery in the X' direction of the light source image (surface light source image) Ips and enters the lens groups G4 and G5 is denoted as La.

[0058] 13A shows the behavior of reflected light (imaging light beam) Sa when the center (or center of gravity) of the light source image (surface light source image) Ips is accurately positioned at the center of the pupil Ep. The chief rays La of the reflected light (imaging light beam) Sa directed toward any one point within the projection area IAn on the substrate P are all parallel to the optical axis AXa, and the imaging light beam projected onto the projection area IAn is telecentric, i.e., the telecentric error is zero. In contrast, FIG. 13B shows the behavior of the reflected light (imaging light beam) Sa when the center (or center of gravity) of the light source image (surface light source image) Ips is shifted laterally by ΔDx in the X' direction from the center of the pupil Ep. In this case, the chief rays La of the reflected light (imaging light beam) Sa directed toward any one point within the projection area IAn on the substrate P are all tilted by Δθt with respect to the optical axis AXa. The tilt amount Δθt becomes a telecentric error, and as the tilt amount Δθt (i.e., the lateral shift amount ΔDx) becomes larger than a predetermined allowable value, the imaging condition of the pattern image projected onto the projection area IAn deteriorates.

[0059] [DMD Configuration] As previously described, the DMD 10 used in this embodiment employs a roll and pitch drive system. Its specific configuration will be described with reference to FIGS. 14 and 15. FIGS. 14 and 15 are enlarged perspective views of a portion of the mirror surface of the DMD 10. The Cartesian coordinate system X'Y'Z is the same as the coordinate system X'Y'Z shown in FIG. 6. FIG. 14 shows the state when power is turned off to the drive circuits provided below each micromirror Ms of the DMD 10. When power is turned off, the reflective surface of each micromirror Ms is set parallel to the X'Y' plane. Here, the arrangement pitch of each micromirror Ms in the X' direction is Pdx (μm), and the arrangement pitch in the Y' direction is Pdy (μm). In practice, these are set to a square where Pdx = Pdy. Furthermore, when the effective dimension ratio is η (η<1.0), the dimensions Lms of the micromirror Ms in the X' and Y' directions are Lms=η·Pdx=η·Pdy, where η is set to approximately 0.8 to 0.9.

[0060] FIG. 15 shows a state in which power is supplied to the drive circuit, resulting in a mixture of micromirrors Msa in the ON state and micromirrors Msb in the OFF state. In this embodiment, the micromirror Msa in the ON state is driven to tilt at an angle θd (= θα / 2) from the X'Y' plane around a line parallel to the Y' axis, and the micromirror Msb in the OFF state is driven to tilt at an angle θd (= θα / 2) from the X'Y' plane around a line parallel to the X' axis. Illumination light ILm is irradiated onto each of the micromirrors Msa and Msb along a principal ray Lp parallel to the X'Z plane (parallel to the optical axis AXb shown in FIG. 6). Note that the line Lx' in FIG. 15 is a projection of the principal ray Lp onto the X'Y' plane and is parallel to the X' axis.

[0061] The angle of incidence θα of illumination light ILm on DMD 10 is the tilt angle with respect to the Z axis in the X'Z plane, and from the perspective of geometric optics, reflected light (imaging light beam) Sa travels in the -Z direction, approximately parallel to the Z axis, from the micromirror Msa in the ON state, which is tilted in the X' direction by an angle θα / 2. On the other hand, reflected light Sg reflected by the micromirror Msb in the OFF state is generated in the -Z direction, non-parallel to the Z axis, because the micromirror Msb is tilted in the Y' direction. In Figure 15, if line Lv is a line parallel to the Z axis (optical axis AXa) and line Lh is a projection of the chief ray of reflected light Sg onto the X'Y' plane, reflected light Sg travels in a tilted direction within the plane containing lines Lv and Lh.

[0062] [Imaging State by DMD] In projection exposure using the DMD 10, pattern exposure is performed by rapidly switching each of the numerous micromirrors Ms between an on-state tilt and an off-state vertical tilt based on pattern data (drawing data) using the operation shown in Figure 15, while scanning the substrate P in the X direction at a speed corresponding to the switching speed. However, depending on the fineness, density, or periodicity of the projected pattern, the telecentricity of the imaging light beam projected from the projection unit PLU (first lens group 116 and second lens group 118) onto the substrate P may change. This is because the mirror surface of the DMD 10 acts as a reflective diffraction grating (blazed diffraction grating) depending on the tilt state of the numerous micromirrors Ms of the DMD 10 according to the pattern.

[0063] FIG. 16 is a diagram showing a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane, and FIG. 17 is a diagram showing the mirror surface of the DMD 10 as viewed in the X'Z plane, along the arrow a-a'. In FIG. 16, among the numerous micromirrors Ms, only one row of micromirrors Ms aligned in the Y' direction is in the ON state (micromirrors Msa), while the remaining micromirrors Ms are in the OFF state (micromirrors Msb). The tilted state of the micromirrors Ms shown in FIG. 16 occurs when an isolated line pattern with a line width at the resolution limit (e.g., approximately 1 μm) is projected. In the X'Y' plane, reflected light (imaging light beam) Sa from the ON-state micromirror Msa occurs in the -Z direction, parallel to the Z axis, while reflected light Sg from the OFF-state micromirror Msb occurs in the -Z direction but tilted in the direction along line Lh in FIG. 11.

[0064] In this case, as shown in FIG. 17 , only one of the many micromirrors Ms aligned in the X' direction is an on-state micromirror Msa tilted at an angle θd (=θα / 2) around a line parallel to the Y' axis with respect to the neutral plane Pcc (a plane parallel to the X'Y' plane that includes the center points of all micromirrors Ms). Therefore, when viewed in the X'Z plane, the reflected light (imaging light beam) Sa emitted from the on-state micromirror Msa is a simple regular reflection light that does not contain first-order or higher-order diffracted light, and its chief ray La is parallel to the optical axis AXa and enters the projection unit PLU. Reflected light Sg from the other off-state micromirrors Msb does not enter the projection unit PLU. Note that if there is only one on-state micromirror Msa isolated in the X' direction (or a row aligned in the Y' direction), the chief ray La of the reflected light (imaging light beam) Sa will be parallel to the optical axis AXa by design, regardless of the wavelength λ of the illumination light ILm.

[0065] Figure 18 is a schematic diagram showing, in the X'Z plane, the imaging state of the reflected light (imaging light beam) Sa from the isolated micromirror Msa shown in Figure 17 by the projection unit PLU. In Figure 18, components with the same functions as those described in Figure 6 are assigned the same reference numerals. Because the projection unit PLU (lens groups G1-G5) is a double-telecentric reduction projection system, if the chief ray La of the reflected light (imaging light beam) Sa from the isolated micromirror Msa is parallel to the optical axis AXa, the chief ray La of the reflected light (imaging light beam) Sa that forms the reduced image ia will also be parallel to the normal (optical axis AXa) to the surface of the substrate P, and no telecentric error will occur. The numerical aperture NAo of the reflected light (imaging light beam) Sa on the object plane side (DMD 10) of the projection unit PLU shown in Figure 18 is equal to the numerical aperture of the illumination light ILm.

[0066] 11 (or FIG. 9) and FIG. 13(A), when DMD 10 is a single large plane mirror tilted by an angle θα / 2, the center (center of gravity) of the light source image (surface light source image) Ips formed on the pupil Ep of the projection unit PLU passes through the optical axis AXa. Similarly, when only specularly reflected light Sa from an isolated micromirror Msa on the mirror surface of DMD 10 is incident on the projection unit PLU, the point spread function of the light beam Isa at the position of the pupil Ep (Fourier transform plane) of that specularly reflected light Sa is a sinc ２ It is expressed as a function (point spread function of a rectangular aperture).

[0067] Fig. 19 is a graph schematically showing a theoretical point spread function Iea (a distribution produced by a light beam from one point light source SPF shown in Figs. 7 and 8 ) of a light beam (here, zeroth-order diffracted light) Isa at the pupil Ep due to reflected light Sa from a single row (or single unit) of micromirrors Msa isolated in the X' direction. In the graph of Fig. 19 , the horizontal axis represents the coordinate position in the X' (or Y') direction with the position of the optical axis AXa as the origin, and the vertical axis represents light intensity Ie. The point spread function Iea is expressed by the following equation (1):

[0068]

[0069] In this equation (1), Io represents the peak value of light intensity Ie, and the position of the peak value Io due to reflected light Sa from an isolated row (or a single unit) of micromirrors Msa coincides with the origin 0 in the X' (or Y') direction, i.e., the position of the optical axis AXa. Furthermore, as previously described in Figure 12, when the shape of the surface light source formed on the exit surface side of the MFE lens 108A is adjusted to resemble an elliptical region APh, the position ±ra in the X' (or Y') direction of the first dark line where the light intensity Ie of the point spread function Iea first reaches a minimum value (0) from the origin 0 is π (≈3.1416) in equation (1). The positions ±ra correspond to the position of the value λ / Lms obtained by dividing the wavelength λ (nm) of the illumination light ILm by the dimension Lms (μm) of a single micromirror Ms (or may be approximated by the value λ / Pdx obtained by dividing it by the array pitch Pdx of the micromirrors Ms) within the plane (Fourier transform plane) of the pupil Ep of the projection unit PLU. Note that the actual intensity distribution at the pupil Ep is the convolution integral (convolution operation) of the point spread distribution Iea over the spread range (σ value) of the light source image Ips shown in Figure 9, and is approximately uniform in intensity.

[0070] Next, a case where the width of the projected pattern in the X' direction (X direction) is sufficiently large will be described with reference to FIGS. 20 and 21. FIG. 20 is a diagram showing a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane, and FIG. 21 is a diagram showing the portion of the mirror surface of the DMD 10 in FIG. 20 as viewed in the X'Z plane, as viewed along the arrow a-a'. FIG. 20 illustrates a case where all of the numerous micromirrors Ms shown in FIG. 16 are in the on-state. While FIG. 20 shows only an arrangement of nine micromirrors Ms in the X' direction and ten micromirrors Ms in the Y' direction, there are also cases where more than two adjacent micromirrors Ms (or even all of the micromirrors Ms on the DMD 10) are in the on-state.

[0071] 20 and 21 , from a large number of micromirrors Msa in the ON state that are adjacently arranged in the X′ direction, a reflected light Sa′ (main diffracted light that is a component equivalent to zero-order light) and other diffracted lights are generated at a slight inclination from the optical axis AXa due to diffraction. If the mirror surface of DMD 10 in the state of Fig. 21 is considered as a diffraction grating arranged at a pitch Pdx in the X′ direction along the neutral plane Pcc, the angle θj at which these diffracted lights are generated can be expressed as in the following equation (2) or equation (3), where j is the order (j=0, 1, 2, 3, ...), λ is the wavelength, and θα is the angle of incidence of illumination light ILm.

[0072]

[0073]

[0074] Figure 22 is a graph showing the distribution of angles θj of diffracted light Idj calculated assuming, as an example, that the angle of incidence θα of illumination light ILm (the inclination angle of the chief ray Lp of illumination light ILm relative to the optical axis AXa) is 35.0°, the inclination angle θd of the micromirror Msa in the on state is 17.5°, the pitch Pdx of the micromirrors Msa is 5.4 μm, and the wavelength λ is 355.0 nm. As shown in Figure 22, since the angle of incidence θα of illumination light ILm is 35°, the zeroth-order diffracted light Id0 (j = 0) is inclined at +35° relative to the optical axis AXa, and the angle θj relative to the zeroth-order diffracted light Id0 increases as the diffraction order increases. The numerical values ​​shown in the lower part of Figure 22 represent the order j in parentheses and the inclination angle of each order of diffracted light Idj from the optical axis AXa.

[0075] In the case of the numerical conditions of Figure 22, the ninth-order diffracted light Id9, which has the smallest tilt angle from the optical axis AXa of approximately -1.04°, becomes the main diffracted light (component equivalent to zeroth-order light) of the imaging light beam Sa'. Therefore, when the micromirrors Ms of the DMD 10 are densely packed and turned on as shown in Figures 20 and 21, the center of the intensity distribution of the imaging light beam (Sa') within the pupil EP of the projection unit PLU is decentered to a position shifted laterally by an amount equivalent to an angle of -1.04° from the position of the optical axis AXa (corresponding to the lateral shift amount ΔDx shown in Figure 13(B) above). The actual distribution of the imaging light beam within the pupil Ep is obtained by converting the diffracted light distribution expressed by equation (2) or (3) into the sinc ２ It can be calculated by performing a convolution operation using a function.

[0076] Figure 23 is a diagram schematically illustrating the intensity distribution of the imaging light beam Sa' at the pupil Ep when diffracted light is generated as shown in Figure 22. The horizontal axis in Figure 23 represents the angle θj of the diffracted light Idj converted into the numerical aperture NAo on the object plane (DMD 10) side and the numerical aperture NAi on the image plane (substrate P) side when the projection magnification Mp of the projection unit PLU is set to 1 / 6. It is also assumed that the maximum numerical aperture NAi on the image plane side of the projection unit PLU is 0.3 (object plane side numerical aperture NAo = 0.05). In this case, the resolving power (minimum resolution linewidth) Rs is expressed as Rs = k1(λ / NAi) using a process constant k1 (0 < k1 ≦ 1).

[0077] Therefore, when the wavelength λ is 355.0 nm and k1 is 0.7, the resolving power Rs is approximately 0.83 μm. The pitch Pdx (Pdy) of the micromirrors Ms is reduced to 0.9 μm on the image plane (substrate P) side by a projection magnification Mp of 1 / 6. Therefore, if the projection unit PLU has an image plane-side numerical aperture NAi of 0.3 or greater (an object plane-side numerical aperture NAo of 0.05), a single projected image of the micromirror Msa in the on state can be formed with high contrast. However, in projection exposure using the DMD 10, if the numerical apertures NAi and NAo are made larger than necessary, the imaging light beam Sa' will contain a large amount of higher-order diffracted light other than the ninth-order diffracted light Id9, which is the main diffracted light, which may degrade the image quality exposed on the substrate P.

[0078] 23 , the angle θe from the optical axis AXa in the X' direction when the object-side numerical aperture NAo=0.05, which is the maximum aperture of the pupil Ep of the projection unit PLU, is NAo=sin θe, so θe≈±2.87°. As shown in FIG. 22 above, the tilt angle of the ninth-order diffracted light Id9, −1.04° (more precisely, −1.037°), converts to an object-side numerical aperture NAo of approximately 0.018, and the intensity distribution Hpa of the imaging light beam Sa' (component equivalent to zeroth-order light) in the pupil Ep is displaced in the X' direction by a shift amount ΔDx from the original position of the light source image Ips (radius ri). Note that, while part of the intensity distribution Hpb due to the eighth-order diffracted light Id8 also appears around the +X' direction within the pupil Ep. However, its peak intensity is low. Furthermore, because the tilt angle of 10th-order diffracted light Id10 from the optical axis AXa on the object plane side is large at 4.81°, its intensity distribution is distributed outside the pupil Ep and does not pass through the projection unit PLU. Note that, as explained above in FIG. 12, the intensity distributions Hpa and Hpb in FIG. 23 become approximately circular by setting the surface light source formed on the exit surface side of MFE 108A of illumination unit ILU to an elliptical region APh'.

[0079] Furthermore, because the micromirrors Ms of the DMD 10 are also arranged in the Y' direction at a pitch Pdy (=5.4 μm), diffracted light is also generated in the Y' direction at a low illuminance in accordance with the pitch Pdy, resulting in weak intensity distributions Hpc and Hpd. Depending on the size of the numerical aperture NAo (NAi) of the projection unit PLU, a portion of the intensity distributions Hpc and Hpd may fall within the pupil Ep. Therefore, by appropriately setting the relationship between the numerical aperture NAo (NAi) of the projection unit PLU and the size (radius ri) of the light source image Ips, it is possible to prevent the intensity distributions Hpc and Hpd from falling within the pupil Ep.

[0080] As explained above in Figure 13(B), the telecentricity error Δθt on the image plane side caused by the shift amount ΔDx of the center of the intensity distribution Hpa is Δθt = -6.22° (≈ -1.037° / projection magnification Mp) under the conditions shown in Figures 22 and 23. Thus, when exposing a large pattern in which many of the numerous micromirrors Ms of the DMD 10 are densely turned on, the chief ray of the imaging light beam Sa' onto the substrate P will be inclined by 6° or more with respect to the optical axis AXa. This telecentricity error Δθt can also be a factor in degrading the imaging quality (contrast characteristics, distortion characteristics, symmetry, etc.) of the projected image.

[0081] Next, a case where the projected pattern is a line-and-space pattern having a constant pitch in the X' direction (X direction) will be described with reference to FIGS. 24 and 25. FIG. 24 is a diagram showing a portion of the mirror surface of the DMD 10 as viewed in the X'Y' plane, and FIG. 25 is a diagram showing the a-a' arrow portion of the mirror surface of the DMD 10 in FIG. 24 as viewed in the X'Z plane. FIG. 24 illustrates a case where, of the numerous micromirrors Ms shown in FIG. 16 above, odd-numbered micromirrors Ms aligned in the X' direction are micromirrors Msa in the on state, and even-numbered micromirrors Msb are micromirrors in the off state. It is assumed that all of the odd-numbered micromirrors Ms aligned in the Y' direction in one row in the X' direction are in the on state, and all of the even-numbered micromirrors Ms aligned in the Y' direction in one row in the Y' direction are in the off state.

[0082] As shown in Figure 25, when the micromirrors Msa in the on state are arranged every other one in the X' direction, the angle of occurrence θj of diffracted light generated from DMD 10 is expressed by the following equation (4) or equation (5), which is similar to the previous equation (2) or equation (3), by considering the mirror surface of DMD 10 as a diffraction grating arranged at a pitch of 2·Pdx in the X' direction along the neutral plane Pcc.

[0083]

[0084]

[0085] As in the case of Fig. 22, Fig. 26 is a graph showing the distribution of angles θj of diffracted light Idj calculated assuming that the angle of incidence θα of illumination light ILm (the inclination angle of the chief ray Lp of illumination light ILm relative to the optical axis AXa) is 35.0°, the inclination angle θd of the micromirror Msa in the on state is 17.5°, the pitch 2Pdx of the micromirror Msa is 10.8 μm, and the wavelength λ is 355.0 nm. As shown in Fig. 26, since the angle of incidence θα of illumination light ILm is 35°, the zeroth-order diffracted light Id0 (j = 0) is inclined at +35° relative to the optical axis AXa, and the angle θj relative to the zeroth-order diffracted light Id0 increases as the diffraction order increases. The numerical values ​​shown in the lower part of Fig. 26 represent the order j in parentheses and the inclination angle of the diffracted light Idj of each order from the optical axis AXa.

[0086] In the case of the numerical conditions of Figure 26, the 17th-order diffracted light Id17, which has the smallest tilt angle of approximately 0.85° from the optical axis AXa, becomes the main diffracted light. Furthermore, an 18th-order diffracted light Id18, which has a tilt angle of -1.04° from the optical axis AXa, is also generated. Therefore, when the micromirrors Ms of the DMD 10 are turned on in the finest line-and-space pattern as shown in Figures 24 and 25, the center of the intensity distribution (main diffracted light) of the imaging light beam Sa' within the pupil EP of the projection unit PLU is decentered to a position shifted laterally by an amount equivalent to an angle of 0.85° or -1.04° from the position of the optical axis AXa. The actual distribution of the imaging light beam Sa' within the pupil Ep is obtained by converting the diffracted light distribution expressed by equation (4) or equation (5) into the sinc ２ It can be calculated by performing a convolution operation using a function.

[0087] 23, in the plane of the pupil Ep, the intensity distribution of the 17th-order diffracted light Id17 corresponding to a tilt angle of 0.85° and the intensity distribution of the 18th-order diffracted light Id18 corresponding to a tilt angle of −1.04° appear displaced overall in the X′ direction with respect to the original position of the light source image Ips (radius ri). In the case of the diffracted light distribution as shown in FIG. 26, one of the intensity distributions corresponding to the 17th-order diffracted light Id17 and the 18th-order diffracted light Id18 has a high intensity and the other has a low intensity, so the telecentric error Δθt on the image plane side caused by the shift in these intensity distributions is generally within the range of Δθt = 5.1° and Δθt = −6.22°.

[0088] This range is slightly different from the telecentricity error Δθt = -6.22°, which is the direction of generation of the ninth-order diffracted light Id9 (see FIG. 22 ), when a large number of adjacent micromirrors Ms become the on-state micromirror Msa, as shown in FIGS. 20 and 21 . Furthermore, it is significantly different from the telecentricity error Δθt = 0°, which is the direction of generation of the ninth-order diffracted light Id9 (see FIG. 22 ), when a large number of adjacent micromirrors Ms become the on-state micromirror Msa, as shown in FIGS. 16 and 17 . The actual pattern image projected onto the substrate P by the projection unit PLU is formed by the interference of reflected light Sa′, including diffracted light from the DMD 10, which is captured within the projection unit PLU. Equation (4) or (5) can be used to determine the state of generation of diffracted light in a line-and-space pattern whose array pitch and line width are n times Pdx (5.4 μm), using the following equations (6) and (7), where n is a real number:

[0089]

[0090]

[0091] FIG. 27 is a schematic diagram corresponding to FIG. 23 showing the distribution of reflected light (diffracted light) from the DMD 10 shown in FIG. 26 at the pupil Ep of the projection unit PLU. In the case of FIG. 27 , as explained in FIG. 12 , the contour of the surface light source formed on the exit surface side of the MFE 108A is elliptical APh′, so that the intensity distribution of each diffracted light beam as the imaging light beam Sa′ formed at the pupil Ep of the projection unit PLU is circular. Also, in FIG. 27 , the intensity of the 18th-order diffracted light Id18 shown in FIG. 26 is assumed to be the greatest, and in the case of projection of a line-and-space pattern such as those shown in FIGS. 24 and 25 , the 18th-order diffracted light Id18 is taken as the intensity distribution Hpa of the component equivalent to zeroth-order light. The intensity distribution Hpa is decentered by ΔDx in the −X′ direction corresponding to the angle −1.04° of the 18th-order diffracted light Id18 from the optical axis AXa, resulting in a telecentric error Δθt.

[0092] As previously described with reference to Figure 23, intensity distributions Hpb, Hpc, and Hpd of diffracted light components are generated within the plane of pupil Ep due to the arrangement pitches Pdx and Pdy of the micromirrors Ms of DMD 10 in the X' and Y' directions, but their intensities are sufficiently small compared to the intensity of intensity distribution Hpa. Furthermore, from the line-and-space pattern (line width in the X' direction Pdx and pitch 2Pdx) created by the micromirrors Ms of DMD 10, intensity distributions ±Hpb' of ±1st-order light equivalent components (17th-order diffracted light Id17 and 19th-order diffracted light Id19) generated by diffraction appear on both sides of intensity distribution Hpa in the X' direction. The center point PXp of intensity distribution +Hpb' of the +1st-order light equivalent component is located approximately midway between the center point (Id18) of intensity distribution Hpa of the zeroth-order light equivalent component and the center point of intensity distribution Hpb in the +X' direction. Similarly, the center point PXm of the intensity distribution −Hpb′ of the −1st order light equivalent component is located approximately midway between the center point (Id18) of the intensity distribution Hpa of the 0th order light equivalent component and the center point of the intensity distribution Hpb in the −X′ direction.

[0093] 27 shows the intensity distribution of the imaging light beam Sa' (diffracted light beam) at the pupil Ep in the case of a line-and-space pattern with a pitch of 2Pdx in the X' direction, as shown in Fig. 24. In contrast to this, in the case of a line-and-space pattern with a pitch of 2Pdy (Pdy = Pdx) in the Y' direction, the center point (Id18) of the intensity distribution Hpa of the component equivalent to zero-order light is decentered by ΔDx in the -X' direction, and intensity distributions ±Hpb' of components equivalent to ±first-order light appear on both sides of the intensity distribution Hpa in the Y' direction.

[0094] In this way, even when many of the numerous micromirrors Ms of the DMD 10 are turned on in a line and space pattern, the chief ray of the imaging light beam onto the substrate P may be significantly tilted with respect to the optical axis AXa, which may significantly reduce the imaging quality (contrast characteristics, distortion characteristics, etc.) of the projected image.

[0095] [Telecentricity Adjustment Mechanism] As described above, when the micromirrors Msa, which are turned on in accordance with the pattern to be exposed on the substrate P, among the many micromirrors Ms of the DMD 10, are densely arranged in the X' and Y' directions, or when they are arranged with periodicity in the X' (or Y') direction, a telecentricity error (angle change) Δθt occurs, although to varying degrees, in the imaging light beam Sa' projected from the projection unit PLU. Because each of the many micromirrors Ms of the DMD 10 is switched between the on and off states at a response speed of about 10 kHz, the pattern image generated by the DMD 10 also changes rapidly in accordance with the drawing data. Therefore, during scanning exposure of a pattern on a display panel or the like, the pattern image projected from each module MUn (n = 1 to 27) instantaneously changes shape to, for example, an isolated line or dot pattern, a line and space pattern, or a large land pattern.

[0096] A typical television display panel (liquid crystal type, organic EL type) consists of an image display area on a substrate P, where pixel units approximately 200 to 300 μm square are arranged in a matrix to achieve a predetermined aspect ratio, such as 2:1 or 16:9, and a peripheral circuit unit (e.g., lead wiring, connection pads) arranged around the image display area. Each pixel unit has a thin-film transistor (TFT) for switching or current driving, but the size (line width) of the TFT patterns (patterns for gate layer, drain / source layer, semiconductor layer, etc.) and the gate and drive wiring are sufficiently small compared to the pixel unit arrangement pitch (200 to 300 μm). Therefore, when exposing a pattern within the image display area, the pattern image projected from the DMD 10 is almost isolated, so no telecentric error Δθt occurs.

[0097] However, depending on the configuration of the lighting drive circuit (TFT circuit) for each pixel unit, line-and-space wiring may be formed in the X or Y direction at a pitch smaller than the pixel unit arrangement pitch. In this case, when a pattern within the image display area is exposed, the pattern image projected from the DMD 10 becomes periodic. As a result, a telecentric error Δθt may occur depending on the degree of periodicity. Furthermore, when exposing the image display area, rectangular patterns approximately the same size as the pixel units or larger than half the area of ​​the pixel units may be uniformly exposed (tiled). In this case, more than half of the numerous micromirrors Ms of the DMD 10 are turned on in a nearly dense state during exposure of the image display area. As a result, a relatively large telecentric error Δθt may occur.

[0098] The occurrence state of the telecentricity error Δθt can be estimated before exposure based on drawing data of the pattern for the display panel to be exposed in each of the multiple modules MUn (n = 1 to 27). In this embodiment, the position and attitude of each of several optical members in the module MUn are configured to be finely adjustable, and from among those optical members, an adjustable optical member can be selected according to the magnitude of the estimated telecentricity error Δθt, and the telecentricity error Δθt can be corrected.

[0099] Figure 28 shows a specific configuration of an optical path from the optical fiber bundle FBn in the illumination unit ILU of the module MUn shown in Figure 4 or 6 to the MFE lens 108A, and Figure 29 shows a specific configuration of an optical path from the MFE lens 108A in the illumination unit ILU to the DMD 10. In Figures 28 and 29, the Cartesian coordinate system X'Y'Z is set to be the same as the coordinate system X'Y'Z in Figure 4 (Figure 6), and members having the same functions as the members shown in Figure 4 are assigned the same reference numerals.

[0100] Although not shown in Fig. 4, in Fig. 28, a contact lens 101 is disposed immediately after the output end of the optical fiber bundle FBn, thereby suppressing the spread of the illumination light ILm from the output end. The optical axis of the contact lens 101 is set parallel to the Z axis, and the illumination light ILm traveling from the optical fiber bundle FBn at a predetermined numerical aperture is reflected by mirror 100, travels parallel to the X' axis, and is reflected in the -Z direction by mirror 102. An input lens system 104, disposed in the optical path from mirror 102 to MFE lens 108A, is composed of three lens groups 104A, 104B, and 104C spaced apart from one another along the optical axis AXc.

[0101] The illuminance adjustment filter 106 is supported by a holding member 106A that is translated by a drive mechanism 106B, and is disposed between the lens group 104A and the lens group 104B. An example of the illuminance adjustment filter 106 is, as disclosed in Japanese Patent Application Laid-Open No. 11-195587, a transparent plate made of quartz or the like on which a fine light-blocking dot pattern is formed with gradually varying density, or a plurality of rows of elongated light-blocking wedge-shaped patterns, and by translating the quartz plate, the transmittance of the illumination light ILm can be continuously changed within a predetermined range.

[0102] The first telecentric adjustment mechanism is composed of a tilt mechanism 100A that finely adjusts the two-dimensional tilt (rotation angle around the X' axis and the Y' axis) of the mirror 100 that reflects the illumination light ILm from the optical fiber bundle FBn, a translation mechanism 100B that finely moves the mirror 100 two-dimensionally within the X'Y' plane perpendicular to the optical axis AXc, and a drive unit 100C that is a microhead or a piezoelectric actuator or the like that drives each of the tilt mechanism 100A and the translation mechanism 100B individually.

[0103] By adjusting the tilt of mirror 100, the central ray (chief ray) of illumination light ILm entering input lens system 104 can be adjusted to be coaxial with the optical axis AXc. Furthermore, since the exit ends of fiber bundles FBn are positioned at the front focal position of input lens system 104, slightly moving mirror 100 in the X' direction causes the central ray (chief ray) of illumination light ILm entering input lens system 104 to shift parallel to the X' direction with respect to the optical axis AXc. As a result, the central ray (chief ray) of illumination light ILm emerging from input lens system 104 travels at a slight tilt with respect to the optical axis AXc. Therefore, illumination light ILm entering MFE lens 108A is slightly tilted overall within the X'Z plane.

[0104] Figure 30 is an exaggerated diagram showing the state of point light sources SPF formed on the exit surface side of the MFE lens 108A when the illumination light ILm incident on the MFE lens 108A is tilted within the X'Z plane. When the central ray (principal ray) of the illumination light ILm is parallel to the optical axis AXc, the point light sources SPF focused on the exit surface side of each lens element EL of the MFE lens 108A are located at the center in the X' direction, as indicated by the white circles in Figure 30. When the illumination light ILm is tilted with respect to the optical axis AXc within the X'Z plane, the point light sources SPF focused on the exit surface side of each lens element EL are decentered by Δxs in the X' direction from the central position, as indicated by the black circles in Figure 30. In this case, as described above with reference to Figures 7 to 9, the surface light source formed by the collection of multiple point light sources SPF formed on the exit surface side of the MFE lens 108A is shifted laterally by Δxs in the X' direction as a whole. Since the cross-sectional dimensions of each lens element EL of the MFE lens 108A in the X'Y' plane are small, the amount of decentering Δxs in the X' direction as a surface light source is also small.

[0105] As shown in FIG. 28 , an aperture stop 108B having an aperture shape of the elliptical region APh′ shown in FIG. 12 is provided on the exit surface side of the MFE lens 108A, and the MFE lens 108A and aperture stop 108B are integrally attached to a holder 108C. The holder 108C (MFE 108A) is provided so that its position in the X′Y′ plane can be finely adjusted by a fine movement mechanism 108D such as a microhead or a piezoelectric motor. In this embodiment, the fine movement mechanism 108D, which finely moves the MFE lens 108A two-dimensionally in the X′Y′ plane, functions as a second telecentricity adjustment mechanism. As shown in FIG. 29 , the aperture stop 108B has an opening of an elliptical region APh′ with the X′ direction as its major axis and the Y′ direction as its minor axis. If the dimension of the major axis of the elliptical region APh' is Ux and the dimension of the minor axis is Uy, the elliptical ratio Uy / Ux is set to the relationship Uy / Ux = cos θα, depending on the cosine value of the incident angle θα of the illumination light ILm to the DMD 10 (twice the tilt angle θd of the micromirror Msa in the on state).

[0106] A plate-type beam splitter 109A inclined at approximately 45° with respect to the optical axis AXc is provided immediately after the MFE lens 108A (aperture stop 108B). The beam splitter 109A transmits most of the amount of illumination light ILm from the MFE lens 108A and reflects the remaining amount of light (for example, approximately a few percent) toward the condenser lens 109B. A portion of the illumination light ILm condensed by the condenser lens 109B is guided to a photoelectric element 109D by an optical fiber bundle 109C. The photoelectric element 109D is used as an integration sensor (integration monitor) that monitors the intensity of the illumination light ILm and measures the exposure amount of the imaging light beam projected onto the substrate P.

[0107] As shown in FIG. 29 , illumination light ILm from a surface light source (a collection of point light sources SPF) on the exit surface side of the MFE lens 108A passes through a beam splitter 109A and enters a condenser lens system 110. The condenser lens system 110 is composed of a front lens group 110A and a rear lens group 110B arranged at a distance from each other, and its two-dimensional position in the X'Y' plane can be finely adjusted by a fine movement mechanism 110C such as a microhead or a piezoelectric motor. In other words, the fine movement mechanism 110C enables decentering adjustment of the condenser lens system 110. In this embodiment, the fine movement mechanism 110C, which finely moves the condenser lens system 110 two-dimensionally in the X'Y' plane, functions as a third telecentricity adjustment mechanism. The first telecentric adjustment mechanism, the second telecentric adjustment mechanism, and the third telecentric adjustment mechanism all adjust the relative positional relationship in the decentering direction between the surface light source generated on the exit surface side of the MFE lens 108A (or the surface light source limited within the opening of the elliptical region APh' of the aperture stop 108B) and the condenser lens system 110.

[0108] The front focus of condenser lens system 110 is set at the position of the surface light source (a collection of point light sources SPF) on the exit surface side of MFE lens 108A, and illumination light ILm traveling in a telecentric state from condenser lens system 110 via inclined mirror 112 provides Koehler illumination to DMD 10. As previously described in FIG. 30 , when the surface light source, which is a collection of multiple point light sources SPF formed on the exit surface side of MFE lens 108A, is shifted laterally in the X' direction by an amount Δxs as a whole, the chief ray (central ray) of illumination light ILm irradiating DMD 10 becomes slightly tilted with respect to optical axis AXb in FIG. 29 . In other words, by intentionally imparting a telecentric error to illumination light ILm using the first telecentricity adjustment mechanism, the angle of incidence θα of illumination light ILm described previously in FIGS. 6 , 17 , 21 , and 25 can be slightly changed from the initial setting angle (35.0°) in the X'Z plane.

[0109] Furthermore, when the MFE lens 108A and the variable aperture stop 108B are displaced together in the X' direction within the X'Y' plane by the fine adjustment mechanism 108D, which serves as the second telecentricity adjustment mechanism shown in FIG. 28, the aperture of the aperture stop 108B (the elliptical region APh' in FIG. 29) becomes decentered with respect to the optical axis AXc. This causes the surface light source formed within the elliptical region APh' to shift overall in the X' direction. In this case, too, the chief ray (central ray) of the illumination light ILm irradiating the DMD 10 can be tilted within the X'Z plane with respect to the optical axis AXb in FIG. 29, i.e., the angle of incidence θα of the illumination light ILm onto the DMD 10 can be changed from the initial setting angle (35.0°) within the X'Z plane. The angle of incidence θα can also be changed in a similar manner even if the fine adjustment mechanism 108D is configured to finely move only the aperture stop 108B independently within the X'Y' plane.

[0110] In order to displace the MFE lens 108A and aperture stop 108B together by a relatively large amount, it is necessary to widen the beam width (diameter of the illumination range) of the illumination light ILm irradiated onto the MFE lens 108A from the input lens system 104. Furthermore, it is also effective to provide a shift mechanism that shifts the illumination light ILm irradiated onto the MFE lens 108A laterally within the X'Y' plane in conjunction with the amount of displacement. The shift mechanism can be configured with a mechanism that tilts the direction of the exit end of the optical fiber bundle FBn, or a mechanism that tilts a plane-parallel plate (quartz plate) placed in front of the MFE lens 108A.

[0111] Both the first telecentricity adjustment mechanism (drive unit 100C, etc.) and the second telecentricity adjustment mechanism (fine movement mechanism 108D, etc.) can adjust the angle of incidence θα of the illumination light ILm onto the DMD 10, but with regard to the amount of adjustment, the first telecentricity adjustment mechanism can be used for fine adjustment, and the second telecentricity adjustment mechanism can be used for coarse adjustment. During actual adjustment, it is possible to select appropriately whether to use both the first telecentricity adjustment mechanism and the second telecentricity adjustment mechanism, or to use only one of them, depending on the form of the pattern to be projected and exposed (the amount of telecentricity error Δθt and the amount of correction).

[0112] Furthermore, the fine movement mechanism 110C, which serves as a third telecentricity adjustment mechanism for decentering the condenser lens system 110 in the X'Y' plane, has the same effect as when the second telecentricity adjustment mechanism relatively decenters the position of the surface light source defined by the MFE lens 108A and the aperture stop 108B. However, decentering the condenser lens system 110 in the X' direction (or Y' direction) also causes a lateral shift in the illumination area of ​​the illumination light ILm projected onto the DMD 10, so the illumination area is set larger than the overall size of the mirror surface of the DMD 10 to take this lateral shift into account. The third telecentricity adjustment mechanism using the fine movement mechanism 110C can also be used for coarse adjustment, just like the second telecentricity adjustment mechanism.

[0113] [Wavelength Dependence of Telecentricity Error] As is clear from the above equations (2) to (5), the telecentricity error Δθt described above varies depending on the wavelength λ. For example, in the state shown in Figures 20 and 21 and represented by equation (2), in order to make the telecentricity error Δθt on the image plane side zero, it is sufficient to set the wavelength λ so that the tilt angle of -1.04° (-1.037° to be precise) of the ninth-order diffracted light Id9 from the optical axis AXa shown in Figures 22 and 23 becomes zero.

[0114] 31 is a graph showing the relationship between the center wavelength λ and the telecentricity error Δθt calculated based on the above equation (2), with the horizontal axis representing the center wavelength λ (nm) and the vertical axis representing the image-side telecentricity error Δθt (deg). If the pitch Pdx (Pdy) of the micromirrors Ms of the DMD 10 is 5.4 μm, the tilt angle θd of the micromirrors Ms is 17.5°, and the angle of incidence θα of the illumination light ILm is 35°, and the micromirrors Ms are in the densely on state as shown in FIGS. 20 and 21 , the telecentricity error Δθt is theoretically zero when the center wavelength λ is approximately 344.146 nm. While it is desirable to minimize the image-side telecentricity error Δθt, a tolerance can be provided taking into account the minimum line width (or resolving power Rs) of the pattern to be projected, the chromatic aberration characteristics of the projection unit PLU, and other factors.

[0115] For example, when the allowable range of the image-side telecentricity error Δθt is set to within ±0.6° (approximately 10 mrad) as shown in Figure 31, the center wavelength λ should be in the range of 343.098 nm to 345.193 nm (2.095 nm in width). Also, when the allowable range of the image-side telecentricity error Δθt is set to within ±2.0°, the center wavelength λ should be in the range of 340.655 nm to 347.636 nm (6.98 nm in width).

[0116] Thus, the telecentric error Δθt, which occurs due to the arrangement (periodicity) and density of the micromirrors Msa that are turned on in the DMD 10, i.e., the magnitude of the distribution density, also has wavelength dependence. Generally, the pitch Pdx (Pdy) and tilt angle θd of the micromirrors Ms of the DMD 10 are uniquely set for a ready-made product (e.g., an ultraviolet-compatible DMD manufactured by Texas Instruments), and therefore the wavelength λ of the illumination light ILm is set to match those specifications. In the DMD 10 of this embodiment, the pitch Pdx (Pdy) of the micromirrors Ms is 5.4 μm and the tilt angle θd is 17.5°, so a fiber amplifier laser light source that generates high-intensity ultraviolet pulsed light can be used as the light source that supplies the illumination light ILm to each of the optical fiber bundles FBn (n = 1 to 27).

[0117] As disclosed in Japanese Patent No. 6428675, for example, a fiber amplifier laser light source includes a semiconductor laser element that generates seed light in the infrared wavelength region, a high-speed switching element (such as an electro-optic element) for the seed light, an optical fiber that amplifies the switched seed light (pulsed light) using pump light, and a wavelength conversion element that converts the amplified light in the infrared wavelength region into pulsed light in the ultraviolet wavelength region. In the case of such a fiber amplifier laser light source, the peak wavelength of ultraviolet light that can achieve high generation efficiency (conversion efficiency) with a combination of available semiconductor laser elements, optical fibers, and wavelength conversion elements is 343.333 nm (wavelength width of 50 pm or less). At this peak wavelength, the maximum image-side telecentric error Δθt (the tilt angle on the image side of the ninth-order diffracted light Id9 in FIGS. 22 and 23 ) that can occur in the state shown in FIG. 20 is approximately 0.466° (approximately 8.13 mrad).

[0118] From the above, when two or more lights with significantly different peak wavelengths (for example, light with wavelengths in the 350 nm range and light with wavelengths in the 400 nm range) are combined or switched as illumination light ILm, the telecentric error Δθt will vary significantly depending on the shape of the pattern to be projected (an isolated pattern, a line and space pattern, or a large land pattern), which will pose a problem.

[0119] Therefore, in this embodiment, as illumination light ILm supplied to each module MUn (n = 1 to 27), multi-wavelength laser light (for example, within a wavelength width range of about ±0.2 nm from the center wavelength) is used, which is obtained by combining laser light (for example, wavelength width of about 50 pm) from each of a plurality of fiber amplifier laser light sources, each having a peak wavelength slightly shifted within an allowable range for the wavelength-dependent telecentric error Δθt. By using multi-wavelength laser light combined with a slightly shifted peak wavelength as illumination light ILm in this way, the coherence of illumination light ILm makes it possible to sufficiently reduce the contrast of speckles (or interference fringes) that occur on the micromirror Ms of DMD 10 and on the substrate P.

[0120] Second Embodiment When the DMD 10 is obliquely illuminated with illumination light ILm at an incident angle θα (θα > 20°), if two or more lights with significantly different peak wavelengths (for example, light with wavelengths of about 350 nm and light with wavelengths of about 400 nm) are combined or switched as the illumination light ILm, a different telecentricity error Δθt can occur depending on the difference in wavelength, as shown in Fig. 31. Therefore, when the array pitch Pdx (Pdy) of the micromirrors Ms of the DMD 10 is 5.4 μm, the designed tilt angle θd of the micromirrors Msa in the on state is 17.5°, the incident angle θα of the illumination light ILm is 35.0°, and the projection magnification Mp is 1 / 6, the maximum telecentricity error Δθt (on the image plane side) that can occur when the wavelength λ of the illumination light ILm is changed significantly is investigated, and the results shown in Fig. 32 are obtained.

[0121] Figure 32 is a graph showing the wavelength dependency of telecentricity error, with the horizontal axis representing wavelength λ (nm) and the vertical axis representing image-side telecentricity error Δθt (deg). In Figure 32, the wavelength λ ranges from 280 nm to 450 nm, and the vertical axis on the right side of the graph represents the image-side numerical aperture NAi of the projection unit PLU corresponding to the angle. In this embodiment, the maximum image-side numerical aperture NAi(max) of the projection unit PLU is illustratively set to 0.25. As explained above, under the conditions of Pdx = 5.4 μm and θα = 35°, if the on-state micromirrors Msa are densely distributed, the ninth-order diffracted light Id9 generated when the center wavelength λ is 344.146 nm becomes a component equivalent to zero-order light, and the telecentricity error Δθt becomes zero.

[0122] Similarly, based on the above equation (2) or (3), when the central wavelength λ of illumination light ILm is 281.574 nm on the short wavelength side, the 11th-order diffracted light Id11 generated from DMD 10 becomes a component equivalent to zero-order light, and the telecentric error Δθt is zero. When the central wavelength λ of illumination light ILm is 309.731 nm, the 10th-order diffracted light Id10 generated from DMD 10 becomes a component equivalent to zero-order light, and the telecentric error Δθt is zero. Similarly, when the central wavelength λ of illumination light ILm is 387.164 nm on the long wavelength side, the 8th-order diffraction generated from DMD 10 becomes a component equivalent to zero-order light, and the telecentric error Δθt is zero. When the central wavelength λ of illumination light ILm is 442.473 nm, the 7th-order diffraction generated from DMD 10 becomes a component equivalent to zero-order light, and the telecentric error Δθt is zero.

[0123] Here, assuming that illumination light ILm contains two wavelength components, with the first wavelength λ1 being 355.000 nm and the second wavelength λ2 being 380.000 nm, the telecentricity error Δθt1 (tilt angle in the X'Z plane) on the image plane side at wavelength λ1 (355.000 nm) is approximately −6.2°, with the 9th-order diffracted light Id9 corresponding to the 0th-order light, as explained above in Figures 22 and 23. Also, the telecentricity error Δθt2 (tilt angle in the X'Z plane) on the image plane side at wavelength λ2 (380.000 nm) is approximately +3.65°, with the 8th-order diffracted light Id8 corresponding to the 0th-order light.

[0124] As shown in Figures 28 and 29, illumination light ILm is supplied from optical fiber bundles FBn (n = 1 to 27), and therefore light of wavelength λ1 (355 nm) and light of wavelength λ2 (380 nm) obliquely illuminate the DMD 10 at the same incident angle θα. However, due to the wavelength-dependent characteristics of telecentricity error shown in Figure 32, the telecentricity error Δθt1 (-6.2°) for light of wavelength λ1 and the telecentricity error Δθt2 (+3.65°) for light of wavelength λ2 are significantly different. Therefore, if illumination light ILm includes both light of wavelength λ1 and light of wavelength λ2, there is a possibility that the image quality of the pattern projected onto substrate P will be degraded.

[0125] 28 and 29 are used, the angle (approximately 9.85°) of the difference between the telecentricity errors Δθt1 and Δθt2 hardly changes. Therefore, in the first condition plan, the wavelengths λ1 and λ2 are set so that the angle of the difference between the maximum telecentricity errors Δθt1 and Δθt2 that can occur at each of the two wavelengths λ1 and λ2 is within an allowable range (for example, ±1°). In this case, the projection unit PLU is assumed to have undergone chromatic aberration correction for each of the light of wavelength λ1 and the light of wavelength λ2.

[0126] For example, the maximum telecentricity error Δθt1 (−6.2°) for light of wavelength λ1 (355.000 nm) is set to a range of −5.2° to −7.2° for light of wavelength λ2, which is within the allowable range (±1°). In this case, based on the above equation (2) or equation (3), the wavelength λ2 may be set in the range of approximately 397.35 nm to 401.25 nm. By setting it in this way, the difference between the maximum telecentricity errors Δθt1 and Δθt2 that can occur in the design by various telecentricity error adjustment mechanisms becomes sufficiently small, making it possible to correct the telecentricity error using various telecentricity error adjustment mechanisms.

[0127] Similarly, the wavelength λ1 may be selected so that the maximum telecentricity error Δθt1 for light of wavelength λ1 is within the allowable range (±1°) of +4.65° to +2.65° relative to the maximum telecentricity error Δθt2 (+3.65°) for light of wavelength λ2 (380.000 nm). In this case, too, based on the above equation (2) or (3), the wavelength λ1 may be set in the range of approximately 336.04 nm to 339.53 nm.

[0128] The difference between the telecentric errors Δθt1 and Δθt2 described above is, in other words, the difference between the diffraction angle θj1 and the diffraction angle θj2, where θj1 is the diffraction angle of the main diffracted light of order j1 (equivalent to the zeroth order light) that is generated from the micromirror Msa in the on state at wavelength λ1 and reaches the substrate P via the projection unit PLU, and θj2 is the diffraction angle of the main diffracted light of order j2 (equivalent to the zeroth order light) that is generated from the micromirror Msa in the on state at wavelength λ2 (λ2 ≠ λ1) and reaches the substrate P via the projection unit PLU.

[0129] Although this will vary depending on the fineness of the pattern to be projected (fineness of line width, pitch, etc.) and the magnitude of the σ value of the illumination light on DMD 10, it is preferable to set wavelengths λ1 and λ2 so that the allowable range of angle Δθj(1-2) is 1 / 5 or less of angle θn(max), and more preferably 1 / 8 or less of angle θn(max), where Δθj(1-2) is the angle between diffraction angles θj1 and θj2, and θn(max) is the angle corresponding to the maximum numerical aperture NAi(max) of projection unit PLU. For example, as shown in Figure 32, if the numerical aperture NAi(max) is 0.25, angle θn(max) is approximately 14.5°, and by setting wavelengths λ1 and λ2, the allowable range of angle Δθj(1-2) should be 0 < Δθj(1-2) ≦ 2.9°, and preferably 0 < Δθj(1-2) ≦ 1.8°.

[0130] In this embodiment, the first illumination light having a wavelength λ1 as the center wavelength and the second illumination light having a wavelength λ2 as the center wavelength are both set to have a sufficiently narrow wavelength width Δλ. Under the conditions illustrated in FIG. 32 , the change in the telecentricity error Δθt on the image plane side per 1.0 nm of wavelength change is approximately 0.57° when the zeroth-order light equivalent component is the ninth-order light Id9, and approximately 0.51° when the zeroth-order light equivalent component is the eighth-order light Id8. To allow for such a change in the telecentricity error Δθt, it is preferable to use laser light whose wavelength width Δλ is narrowed to approximately ±0.5 nm or less relative to the center wavelengths (λ1 and λ2).

[0131] In addition, when the diffraction angle of the zeroth-order light equivalent component (9th-order light Id9 and 8th-order light Id8) generated at wavelength λ1 and the diffraction angle of the zeroth-order light equivalent component (9th-order light Id9 and 8th-order light Id8) generated at wavelength λ2 are generated on one side of the optical axis AXa of the projection unit PLU, the following condition is required based on the above equation (3):

[0132] When the arrangement pitch of the micromirrors Ms is Pd, the designed incident angle θα is θα > 0°, and the orders j1 and j2 are greater than 0, the relationship between the wavelengths λ1 and λ2 is set so as to satisfy either the first condition λ1 < Pd sin θα / j1 and λ2 < Pd sin θα / j2, or the second condition λ1 > Pd sin θα / j1 and λ2 > Pd sin θα / j2.

[0133] According to the present embodiment as described above, when illumination light ILm including light of two wavelengths λ1 and λ2 (λ1 ≠ λ2) is used, the telecentricity error of the imaging light beam Sa′ caused by the diffraction action of DMD 10 can be effectively corrected by setting the difference between the diffraction angle θj1 of the main diffracted light of order j1 that is generated from the micromirror Msa that is in the ON state under light of wavelength λ1 and reaches the substrate P via the projection unit PLU, and the diffraction angle θj2 of the main diffracted light of order j2 that is generated from the micromirror Msa that is in the ON state under light of wavelength λ2 and reaches the substrate P via the projection unit PLU, i.e., the difference between the telecentricity error Δθt1 and the telecentricity error Δθt2, so that it is within a predetermined allowable range.

[0134] Furthermore, according to the present embodiment, when illumination light ILm containing light of two wavelengths λ1 and λ2 (λ1 ≠ λ2) is irradiated onto DMD 10 at a design angle of incidence θα that is equal to the double angle of the tilt angle θd of the micromirror Msa in the on state, the wavelengths λ1 and λ2 are set so that the diffraction angle θj1 of the main diffracted light of order j1 that is generated from the micromirror Msa in the on state under light of wavelength λ1 and enters projection unit PLU, and the diffraction angle θj2 of the main diffracted light of order j2 that is generated from the micromirror Msa in the on state under light of wavelength λ2 and enters projection unit PLU are distributed on one side of the optical axis AXa of projection unit PLU (either the positive or negative side of the maximum telecentricity error Δθt that occurs in the design), and it is possible to effectively correct the telecentricity error of the imaging light flux Sa′ that occurs due to the diffraction action of DMD 10.

[0135] As described above, even when using a multi-wavelength laser beam (broadband light or multispectral light) in which laser beams of multiple peak wavelengths are included within a relatively narrow wavelength width with respect to the center wavelength λo, the telecentric error Δθt taking into account the entire wavelength width can be optimally corrected by the second proposed condition. In the second proposed condition, the telecentric error Δθt is corrected taking into account the substantial bandwidth of the multi-wavelength laser beam (broadband light or multispectral light).

[0136] Figure 33 is a schematic diagram showing the wavelength distribution characteristics of eight laser beams synthesized, each with a central wavelength λo of 343.333 nm and peak wavelengths shifted by 20 pm (0.02 nm). In Figure 33, the horizontal axis represents wavelength λ (nm), and the vertical axis represents relative intensity, normalized to 100% peak intensity. Each of the eight laser beams has a wavelength width of approximately 50 pm (0.05 nm) at full width at half maximum (relative intensity 50%), resulting in a nearly Gaussian distribution. In this way, by using multiple laser beams with different peak wavelengths, interference noise (speckle and interference fringes) occurring on the DMD 10 or the substrate P can be effectively suppressed.

[0137] The peak wavelengths of the eight laser beams are designated λa, λb, λc, λd, λe, λf, λg, and λh, in order from shortest to longest, with a shift of approximately 20 pm between adjacent peak wavelengths. Since the center wavelength λo is set to 343.333 nm, the adjacent peak wavelengths λd and λe are set to 343.323 nm and 343.343 nm, respectively. Furthermore, the peak wavelengths λc, λb, and λa are set to 343.303 nm, 343.283 nm, and 343.263 nm, respectively, while the peak wavelengths λf, λg, and λh are set to 343.363 nm, 343.383 nm, and 343.403 nm, respectively.

[0138] Therefore, the bandwidth of the peak wavelengths λa to λh is 140 pm (0.14 nm) from 343.263 nm to 343.403 nm. When the wavelength width of each laser beam is 50 pm at full width at half maximum, the full width at half maximum (relative intensity 50%) of the multi-wavelength laser beam (broadband light or multispectral light) obtained by combining the laser beams of the peak wavelengths λa to λh is about 190 pm (0.19 nm) in the range from 343.238 nm to 343.428 nm, as shown in FIG. 33. Furthermore, when the relative intensity of the laser beam is 1 / e ２ The wavelength bandwidth of the multi-wavelength laser light with a relative intensity of 1 / e (13.5%) is about 224 pm (0.224 nm) in the range of 343.221 nm to 343.445 nm. ２ Let us find the telecentric error Δθt for each of the wavelengths of 343.221 nm and 343.445 nm, where: Note that the projection unit PLU is assumed to be chromatically corrected in the wavelength range of 343.221 nm to 343.445 nm.

[0139] 34 is a graph showing the characteristics of telecentricity error over a wavelength λ range of 343.200 nm to 343.450 nm, with wavelength λ (nm) on the horizontal axis and image-plane-side telecentricity error Δθt (deg) on ​​the vertical axis. In this case, too, the array pitch Pdx of the on-state micromirrors Msa was 5.4 μm, the design tilt angle θd of the micromirrors Msa was 17.5°, and the angle of incidence θα of illumination light ILm (wavelength range 343.221 nm to 343.445 nm) onto the DMD 10 was 35.0°. Furthermore, within that wavelength range, the zero-order light equivalent component generated from the DMD 10 (the many on-state micromirrors Msa) and incident on the projection unit PLU becomes ninth-order light Id9. Based on the above equation (2) or equation (3), the telecentricity error Δθto at the center wavelength λo (343.333 nm) is approximately 0.466°, the maximum telecentricity error Δθta at a wavelength λ of 343.221 nm is approximately 0.530°, and the maximum telecentricity error Δθtb at a wavelength λ of 343.445 nm is approximately 0.401°.

[0140] From the above, when using multi-wavelength laser light (broadband light), it is sufficient to assume that the median (average) of the telecentricity error Δθta on the short wavelength side of the wavelength band and the telecentricity error Δθtb on the long wavelength side is the maximum telecentricity error Δθt [= (Δθta + Δθtb) / 2] that can occur in the design, and to perform correction using the telecentricity adjustment mechanism described above in Figures 28 and 29. After this telecentricity adjustment has been performed, the telecentricity error remaining for the light on the short wavelength side of the wavelength band and the telecentricity error remaining for the light on the long wavelength side of the wavelength band will both be tilted at angles symmetrical with respect to the optical axis AXa of the projection unit PLU. In the case of the conditions illustrated in Figure 34, the median (average value) of the telecentricity error Δθta and the telecentricity error Δθtb matches the telecentricity error Δθto, so the telecentricity errors (Δθto - Δθta, Δθto - Δθtb) remaining after telecentricity adjustment are within the range of ±0.1° or less with respect to the optical axis AXa and can be almost ignored.

[0141] According to the second proposed condition described above, when first illumination light of a peak wavelength λa that is allowable in terms of the chromatic aberration characteristics of the projection unit PLU and second illumination light of a peak wavelength λh (λa ≠ λh) that is allowable in terms of the chromatic aberration characteristics of the projection unit PLU are irradiated onto DMD 10 at an incident angle θα that corresponds to a double angle of the tilt angle θd of the micromirror Msa in the ON state, the order j1 (in the case of FIG. 34 ) of light of wavelength λa is generated from the micromirror Msa in the ON state and enters the projection unit PLU. When the diffraction angle of the main diffracted light of order j2 (9th order light Id9 in the case of Figure 34) generated from the micromirror Msa in an on-state under light of wavelength λh and incident on projection unit PLU is taken as θj1, and the diffraction angle of the main diffracted light of order j2 (9th order light Id9 in the case of Figure 34) generated from the micromirror Msa in an on-state under light of wavelength λh and incident on projection unit PLU is taken as θj2, the difference (bandwidth) between wavelength λa and wavelength λh is set so that the diffraction angle θj1 (equivalent to the telecentric error Δθta) and the diffraction angle θj2 (equivalent to the telecentric error Δθtb) are distributed on either side of the optical axis AXa of projection unit PLU.

[0142] Therefore, even when illumination light ILm having a wide wavelength range within the range allowed by the chromatic aberration characteristics of the projection unit PLU is used, the telecentric error of the imaging light beam Sa' caused by the diffraction action of the DMD 10 can be effectively corrected.

[0143] [Variation 2] As shown in Figures 33 and 34, when the effective wavelength width Δλ of the illumination light ILm is narrow, around 0.224 nm (343.221 nm to 343.445 nm), the width of the overall telecentricity error is also small, but as the wavelength width Δλ of a single illumination light ILm becomes wider, the width of the telecentricity error also becomes larger accordingly.

[0144] For example, when using illumination light ILm with a central wavelength λo of 355.0 nm and an effective wavelength width Δλ of approximately ±2 nm (within the chromatic aberration correction range), the width of the telecentricity error increases, and the distribution state of the imaging light flux at the pupil Ep of the projection unit PLU also changes. Under the initial design conditions (Pdx = 5.4 μm, θd = 17.5°, θα = 35.0°, Mp = 1 / 6), the maximum telecentricity error Δθt on the image plane side when the central wavelength λo of illumination light ILm = 355.0 nm is -6.23° for the 9th-order diffracted light Id9, which is the component equivalent to the 0th-order light, based on Figure 32 and equation (2) or (3).

[0145] In this case, the center of the ninth-order diffracted light Id9 (-6.23°) in the pupil Ep of the projection unit PLU appears at a position at a numerical aperture of approximately 0.109 on the image plane side. Similarly, when the wavelength width Δλ of the illumination light ILm is ±2 nm, the wavelength λ1 on the short wavelength side is 353.0 nm, so the telecentricity error Δθt1 on the image plane side caused by that wavelength λ1 is -5.08° (numerical aperture of approximately 0.089). Furthermore, the wavelength λ2 on the long wavelength side is 357.0 nm, so the telecentricity error Δθt2 on the image plane side caused by that wavelength λ2 is -7.39° (numerical aperture of approximately 0.129).

[0146] Here, Figure 35 schematically shows the state of ninth-order diffracted light Id9 distributed in the pupil Ep of a projection unit PLU having a maximum image-plane-side numerical aperture NAi(max) of 0.25 when the center wavelength λo is 355.0 nm and the wavelength width Δλ is ±2 nm. Similar to Figure 23 above, Figure 35 shows the distribution of ninth-order diffracted light Id9 that appears in the pupil Ep when a large number of micromirrors Ms of the DMD 10 are densely turned on. In Figure 35 , the σ value, which is the ratio of the numerical aperture of illumination light ILm to the numerical aperture of the projection unit PLU, is set to 0.6, for example, and the elliptical distribution of the ninth-order diffracted light Id9 (elliptical ratio ≈ 0.82) that occurs when the angle of incidence θα is 35.0° is uncorrected.

[0147] The center P9o of the ninth-order diffracted light Id9 due to light with a central wavelength λo (355.0 nm) appears at a position of numerical aperture NAi = 0.109 within the pupil Ep, the center P9a of the ninth-order diffracted light Id9 due to light with a central wavelength λ1 (353.0 nm) appears at a position of numerical aperture NAi = 0.089, and the center P9b of the ninth-order diffracted light Id9 due to light with a central wavelength λ2 (357.0 nm) appears at a position of numerical aperture NAi = 0.129. The elliptical distribution H9o of the ninth-order diffracted light Id9 due to light with a central wavelength λo, the elliptical distribution H9a (almost congruent with distribution H9o) of the ninth-order diffracted light Id9 due to light with a central wavelength λ1, and the elliptical distribution H9b (almost congruent with distribution H9o) of the ninth-order diffracted light Id9 due to light with a central wavelength λ2 each appear shifted in the X' direction by approximately 0.02 in numerical aperture equivalent.

[0148] Therefore, for illumination light ILm with a central wavelength λo of 355.0 nm and a wavelength range Δλ of ±2 nm, ninth-order diffracted light Id9 (a component corresponding to zeroth-order light) is distributed throughout the pupil Ep between distributions H9a and H9b, which are shifted in the X' direction. The telecentricity adjustment mechanism described above with reference to Figures 28 and 29 corrects the telecentricity error Δθt (-6.23°) due to light with a central wavelength λo (355.0 nm) to zero. Even if the center P9o of distribution H9o is adjusted to coincide with the optical axis AXa, the relative decentering (approximately 0.02 in numerical aperture) between distribution H9a due to light on the short wavelength side of wavelength λ1 and distribution H9b due to light on the long wavelength side of wavelength λ2 with respect to distribution H9o remains almost unchanged.

[0149] Because the σ value of illumination light ILm is 0.6 and the maximum numerical aperture NAi(max) of projection unit PLU is 0.25, the numerical aperture NAy' in the Y' direction of distribution H9o of ninth-order diffracted light Id9 due to light with central wavelength λo is NAi(max) × σ = 0.15. Furthermore, because the ellipticity ratio at incident angle θα = 35.0° is 0.82 (= cos θα), the numerical aperture NAx' in the X' direction of distribution H9o of ninth-order diffracted light Id9 due to light with central wavelength λo is NAy' × 0.82 = 0.123. Furthermore, the distribution H9a of the ninth-order diffracted light Id9 due to light on the short wavelength side of wavelength λ1 and the distribution H9b of the ninth-order diffracted light Id9 due to light on the long wavelength side of wavelength λ2 are decentered in the X' direction with respect to distribution H9o by approximately 0.02 in terms of numerical aperture, and therefore the numerical aperture in the X' direction of the overall distribution of distribution H9a and distribution H9b is 0.123 + 0.02 = 0.143.

[0150] From the above, when illumination light ILm having a center wavelength λo = 355.0 nm and a wavelength width Δλ = ±2 nm is obliquely illuminated onto DMD 10 (Pdx = 5.4 μm) at an incident angle θα = 35° under the condition of a σ value of 0.6, the overall distribution at pupil Ep of ninth-order diffracted light Id9 (component equivalent to zeroth-order light) is 0.15 in the Y' direction and 0.143 in the X' direction in terms of numerical aperture, and the ellipticity ratio is improved to approximately 0.95 (= 0.143 / 0.15). Therefore, by using illumination light ILm (multi-wavelength light or broadband light) having an appropriate wavelength width Δλ, it is possible to suppress the elliptical shape that inevitably occurs with oblique illumination (incident angle θα) and make the overall distribution of the zeroth-order light equivalent component (jth-order diffracted light) that appears at pupil Ep of projection unit PLU circular (isotropic distribution with approximately the same dimensions in the X' direction and the Y' direction).

[0151] That is, by providing the illumination light ILm that obliquely illuminates the DMD 10 with a predetermined wavelength width Δλ within the range allowed by the chromatic aberration characteristics of the projection unit PLU, it is possible to impart an ellipticalization reduction function that suppresses ellipticalization of the distribution (light source image Ips) of the imaging light beam (high-order diffracted light) at the pupil Ep of the projection unit PLU.

[0152] As shown in FIGS. 10 and 11 (as well as FIG. 33 ), in the case of narrowband light with a sufficiently narrow wavelength width Δλ (for example, Δλ≦0.2 nm), the distribution of imaging light beams (distribution of light source images Ips) at the pupil Ep of the projection unit PLU is defined by a σ value based on the dimension ri in the Y' direction from the center (optical axis AXa). Therefore, the numerical aperture NAy' in the Y' direction of the imaging light beam distribution (distribution of light source images Ips) is NAy' = σ NAi(max), where NAi(max) is the maximum numerical aperture of the projection unit PLU, and the numerical aperture NAx' in the X' direction of the imaging light beam distribution (distribution of light source images Ips) is NAx' = σ NAi(max) cos θα. Therefore, the ellipse ratio Ux' / Uy' (= cos θα) described in FIG. 10 can also be expressed by the numerical aperture ratio NAx' / NAy'.

[0153] On the other hand, in the case of broadband illumination light ILm having a relatively wide wavelength width Δλ, the imaging light beam (j-th order diffracted light) is distributed throughout the entire region from distribution H9a corresponding to light with wavelength λo-Δλ to distribution H9b corresponding to light with wavelength λo+Δλ, as shown in Fig. 35, and therefore the substantial elliptical distribution of the imaging light beam within the pupil Ep changes. Therefore, the relationship between the elliptical ratio of the entire imaging light beam (j-th order diffracted light) distributed on the plane of the pupil Ep and the wavelength width Δλ will be described with reference to Fig. 36.

[0154] Similar to FIG. 35 , FIG. 36 is an exaggerated view of the distributions Hjo, Hja, and Hjb of higher-order diffracted light (assumed to be j-th order diffracted light) from the DMD 10 that appear within the pupil Ep of the projection unit PLU when illumination light ILm with a wide wavelength range Δλ is used. In FIG. 36 , it is assumed that the telecentric adjustment mechanism has corrected the center Pjo of the elliptical distribution Hjo that appears in response to light with a central wavelength λo to coincide with the optical axis AXa of the projection unit PLU. Therefore, the center Pja of the elliptical distribution Hja that appears in response to light with a wavelength λo-Δλ on the short wavelength side and the center Pjb of the elliptical distribution Hjb that appears in response to light with a wavelength λo+Δλ on the long wavelength side are located approximately symmetrically with respect to the center Pjo (the position of the optical axis AXa) at a constant distance in the X' direction. It is also assumed that the distance between the centers Pjo and Pja is equal to the distance between the centers Pjo and Pjb.

[0155] According to the above formula (3), "sin θj" on the left side of formula (3) represents the numerical aperture corresponding to the position of the central ray (each of the centers Pjo, Pja, and Pjb) of the j-th order diffracted light passing through the pupil Ep of the projection unit PLU. Therefore, if the distance from the center Pjo to the center Pja (or Pjb) is converted into a numerical aperture and taken as the numerical aperture ΔNAx on the image plane side, the numerical aperture ΔNAx can be found by the following formula (8) or formula (9) taking into account the projection magnification Mp (for example, Mp = 1 / 6). ΔNAx = [sinθα-j・λo / Pdx] / Mp - [sinθα-j・(λo+Δλ) / Pdx] / Mp =j・Δλ / Pdx / Mp... (8) ΔNAx = [sinθα−j・(λo−Δλ) / Pdx] / Mp −[sinθα−j・λo / Pdx] / Mp =j・Δλ / Pdx / Mp... (9)

[0156] Furthermore, the size from the optical axis AXa (center Pjo) in the major axis direction (Y' direction) of the distribution Hjo (the same for each of Hja and Hjb) that has been deformed into an ellipse due to oblique illumination (incident angle θα) can be expressed in terms of numerical aperture using the design σ value and the maximum numerical aperture NAi(max) on the image plane side of the projection unit PLU. As shown in Figure 36, if the numerical aperture in the Y' direction of the distribution Hjo (the same for each of Hja and Hjb) is NAy', the numerical aperture NAy' can be calculated using the following equation (10): NAy' = σ · NAi(max) (10) Furthermore, the numerical aperture NAx' corresponding to the size in the X' direction of the distribution Hjo (the same for each of Hja and Hjb) can be calculated using the following equation (11) based on the ratio cos θα of the ellipse caused by the incident angle θα. NAx'=NAy'・cosθα=σ・NAi(max)・cosθα...(11)

[0157] As shown in Figure 36, if the dimension in the X' direction of the entire imaging light beam due to j-th order diffracted light at the pupil Ep is converted to a numerical aperture and denoted as NAxf, then the numerical aperture NAxf is expressed as NAx' + ΔNAx. If the ratio of the ellipse of the distribution of the entire imaging light beam due to j-th order diffracted light is ΔOV, then the ratio ΔOV is expressed as NAxf / NAy'. When this ratio ΔOV is 1 (100%), the numerical apertures in the X' and Y' directions of the entire imaging light beam due to j-th order diffracted light become equal on the substrate P, resulting in an isotropic imaging light beam. Based on the above equations (8) to (11), the ratio ΔOV is expressed by the following equation (12): ΔOV = (NAx' + ΔNAx) / NAy' = NAx' / NAy' + ΔNAx / NAy' = cos θα + (j Δλ / Pdx / Mp) / (σ NAi(max)) (12)

[0158] According to the graph in Figure 32, the orders j that can become the zeroth-order light equivalent component in a practical wavelength band (for example, approximately 300 nm to 400 nm) are the 8th, 9th, and 10th orders. Therefore, when the order j in equation (12) is j = 8, j = 9, and j = 10, and the change in the ellipse ratio ΔOV when the wavelength width Δλ is changed is calculated, the characteristics shown in Figure 37 are obtained.

[0159] 37, the horizontal axis represents the wavelength width Δλ (nm), the vertical axis represents the ellipse ratio ΔOV (%), and the characteristic V(8) represents the case where the component equivalent to the zeroth order light is eighth-order diffracted light, the characteristic V(9) represents the case where the component equivalent to the zeroth order light is ninth-order diffracted light, and the characteristic V(10) represents the case where the component equivalent to the zeroth order light is tenth-order diffracted light. The graph in Fig. 37 represents the characteristics obtained when the pitch Pdx of the micromirrors Msa in the on state is 5.4 μm, the angle of incidence θα of the illumination light ILm is 35.0°, the numerical aperture NAi(max) of the projection unit PLU is 0.25, the σ value is 0.6, and the projection magnification Mp is 1 / 6.

[0160] 37, when the wavelength width Δλ is distributed over a range of ±3.05 nm including the central wavelength λo (6.1 nm in total width), the ratio ΔOV is 100%, the numerical aperture of the imaging light beam reaching the substrate P is equal in the X' direction and the Y' direction, and the quality (accuracy of line width) of the projected image of various edge portions with different directionality of the pattern to be exposed can be made the same. Similarly, in the case of characteristic V(9), the ratio ΔOV is 100% when the wavelength width Δλ is distributed over a range of ±2.71 nm including the central wavelength λo (5.42 nm in total width), and in the case of characteristic V(10), the ratio ΔOV is 100% when the wavelength width Δλ is distributed over a range of ±2.44 nm including the central wavelength λo (4.88 nm in total width).

[0161] The ratio ΔOV does not necessarily have to be 100%, and can have a predetermined tolerance of ±5% or ±10%, depending on the fineness of the pattern to be exposed. Generally, the range of spread of the wavelength width Δλ is often limited by the chromatic aberration characteristics of the projection unit PLU, so the tolerance is set so that the ratio ΔOV is about 95% or 90%. For example, for the characteristic V(9) in Figure 37, if the ratio ΔOV is sufficient at 90%, the wavelength width Δλ will be about 1.45 nm (about 2.9 nm in total width), which has the advantage of making it easier to correct the chromatic aberration of the projection unit PLU.

[0162] Conversely, if it is desired to set the ratio ΔOV close to 100%, but the wavelength width Δλ is limited to within ±1.0 nm due to limitations on the chromatic aberration characteristics of the projection unit PLU, then in characteristic V(9), the ratio ΔOV at Δλ = 1.0 nm will be approximately 88%. In this case, to further improve the ratio ΔOV and bring it closer to 100%, this can be compensated for by changing the aperture shape of aperture stop 108B shown in FIG. 29 to an elliptical shape that improves the remaining elliptical ratio of approximately 12%. In other words, it is possible to use both the function of reducing ellipticity by giving illumination light ILm a certain wavelength width Δλ and the function of reducing ellipticity by providing an optical element such as aperture stop 108B.

[0163] In the characteristics shown in Fig. 37, the σ value is set to 0.6, but the σ value may be adjustable to obtain a resolution and depth of focus (DOF) suited to the fineness of the pattern to be exposed. Fig. 38 will now be used to explain how the characteristics V(9) in Fig. 37 change depending on the σ value.

[0164] Figure 38 shows the characteristics when the wavelength width Δλ (nm) is taken on the horizontal axis and the ellipse ratio ΔOV (%) is taken on the vertical axis, and the σ value is changed in the range of 0.2 to 0.9. In the case of the graph in Figure 38, the pitch Pdx of the micromirrors Msa in the on state is 5.4 μm, the angle of incidence θα of the illumination light ILm is 35.0°, the numerical aperture NAi(max) of the projection unit PLU is 0.25, the projection magnification Mp is 1 / 6, and the component equivalent to the zeroth-order light is taken to be the ninth-order diffracted light (j=9).

[0165] As shown in Figure 38, the wavelength width Δλ required to achieve a 100% ratio ΔOV is approximately 1.36 nm when the σ value is 0.3, approximately 2.71 nm when the σ value is 0.6, and approximately 3.62 nm when the σ value is 0.8. Thus, as the σ value increases, the wavelength width Δλ required to achieve a 100% ellipse ratio ΔOV increases. Conversely, due to limitations imposed by the chromatic aberration characteristics of the projection unit PLU, if the wavelength width Δλ of the illumination light ILm is set within ±1.0 nm (2.0 nm full width), the ratio ΔOV can be improved to 100% when the σ value is set to 0.2. However, if the σ value increases beyond that, the improvement to 100% is not achieved. Therefore, even in this case, it is possible to combine the function of improving the ratio ΔOV by providing a certain wavelength width Δλ for the illumination light ILm with the function of improving the ratio ΔOV by providing an optical element such as the aperture stop 108B.

[0166] The broadband illumination light ILm discussed above does not necessarily have a continuous spectrum across the wavelength range Δλ. Figure 39 is a graph showing an example of the wavelength distribution characteristics of illumination light ILm, with Figure 39(A) showing a case where the spectrum exists across a range of wavelengths Δλ (±Δλ) from the center wavelength λo, and Figure 39(B) showing a case where a plurality of spectra with a single, extremely narrow wavelength range are distributed discretely across the range of wavelengths Δλ (±Δλ). In Figures 39(A) and (B), the horizontal axis represents wavelength (nm) and the vertical axis represents relative intensity normalized to 1 for the peak value of the spectrum.

[0167] A continuous spectrum such as that shown in Figure 39(A) can be obtained using a specific emission line from a mercury discharge lamp or laser light from an excimer laser light source in a spontaneous oscillation state without band-narrowing. Furthermore, a technique for creating multiple spectra with different peak wavelengths, as shown in Figure 39(B), can be achieved using multiple different laser light sources (fiber amplifier lasers, harmonic laser light sources such as Nd-YAG lasers, and band-narrowed excimer laser light sources, etc.), similar to the method described above with reference to Figure 33. In this case, to effectively improve the ellipticity ratio ΔOV, at least two spectra are required: one with a peak at wavelength λo-Δλ on the short wavelength side, and the other with a peak at wavelength λo+Δλ on the long wavelength side.

[0168] According to the second variant described above in Figures 35 to 39, illumination light is irradiated onto a DMD 10 as a spatial light modulation element having a large number of micromirrors Ms that are arranged two-dimensionally at a pitch Pdx and are selectively driven based on drawing data, and reflected light from selected on-state micromirrors Msa of the DMD 10 is made incident on a projection unit PLU, thereby forming a pattern exposure device that projects and exposes a pattern corresponding to the drawing data onto a substrate P. An illumination unit ILU is provided that irradiates the DMD 10 with illumination light ILm having a predetermined wavelength width ±Δλ with respect to the center wavelength λo at an incident angle θα (θα > 0°) that corresponds to double the design tilt angle (θd) of the on-state micromirrors Msa.

[0169] In this case, by appropriately setting the wavelength width ±Δλ, a difference can be generated between the diffraction angle θj1 of the main diffracted light (Id9) of order j1 (e.g., 9th order) that is generated from the micromirror Msa in the on state under light with a wavelength λo+Δλ on the long wavelength side of the illumination light ILm and enters the projection unit PLU, and the diffraction angle θj2 of the main diffracted light (Id9) of order j2 (e.g., 9th order) that is generated from the micromirror Msa in the on state under light with a wavelength λo-Δλ on the short wavelength side of the illumination light ILm and enters the projection unit PLU. As a result, the overall distribution shape of the main diffracted light of order j1 and the main diffracted light of order j2 that appear in the pupil Ep of the projection unit PLU (for example, a shape obtained by combining the elliptical distributions H9a and H9b in FIG. 35) can be deformed into an isotropic shape (almost circular) within the pupil Ep by the difference between the diffraction angle θj1 and the diffraction angle θj2 (for example, the difference between the numerical aperture NAi of the center P9a and the numerical aperture NAi of the center P9b in FIG. 35).

[0170] [Third Embodiment] In the above-described first embodiment, second embodiment, and variant 1 thereof, the array pitch Pdx of the micromirrors Ms of the DMD 10, the tilt angle θd of the micromirrors Ms, and the designed incident angle θα of the illumination light ILm (the angle formed by the optical axes AXb and AXa in FIG. 6 ) were set to fixed values, and the wavelength and bandwidth of the illumination light ILm were selected so that the telecentric error Δθt could be corrected to within an allowable range. However, when generating broadband illumination light ILm by combining multiple lights with different peak wavelengths to widen the wavelength width, it may not be possible to obtain a light source (fiber amplifier laser light source, excimer laser light source, semiconductor laser light source, high-pressure mercury discharge lamp, etc.) in the ultraviolet wavelength range having the desired peak wavelength.

[0171] Therefore, in this embodiment, when at least two illumination lights with significantly different peak wavelengths (or wavelength bands) are obliquely illuminated onto DMD 10, the angle of incidence of the illumination light for each wavelength band can be changed individually, thereby reducing the difference in telecentric error Δθt that can arise due to differences in wavelength bands.

[0172] Figure 40 is a diagram schematically showing the optical path from the MFE lens 108A to the DMD 10 of the illumination unit ILU shown in Figures 4, 6, 28, and 29, and here, two MFE lenses 108A1 and 108A2 and a plate-shaped dichroic mirror DCM (dichroic optical member) having wavelength selection characteristics are added. Note that in Figure 40, the coordinate system X'Y'Z is the same as the coordinate system in Figure 29, and the aperture stop 108B and inclined mirror 112 shown in Figure 29 have been omitted for ease of explanation.

[0173] In this embodiment, illumination light ILm1 having a peak wavelength (center wavelength) λ1 in the ultraviolet range and illumination light ILm2 having a peak wavelength (center wavelength) λ2 longer than λ1 are projected onto MFE lenses 108A1 and 108A2 via fiber bundles FBn and lens systems, respectively. Dichroic mirrors DCM having a reflectance of 90% or more for wavelength λ1 and a transmittance of 90% or more for wavelength λ2 are provided in the optical path between MFE lens 108A1 and condenser lens system 110, and in the optical path between MFE lens 108A2 and condenser lens system 110. The wavelength division plane of the dichroic mirror DCM is set to be tilted by 45° within the X'Z plane with respect to the optical axis AXc of condenser lens system 110.

[0174] 40, the point light source SPF formed at the center of the exit surface side of each of the MFE lenses 108A1 and 108A2 is decentered by a predetermined distance from the optical axis AXc of the condenser lens system 110. That is, similar to the state described above in FIG. 13, by decentering the circular or elliptical surface light source (a collection of multiple point light sources SPF) formed on the exit surface side of each of the MFE lenses 108A1 and 108A2 with respect to the optical axis AXc of the condenser lens system 110, it is possible to make the angle of incidence θα1 of the chief ray (central ray) Lp1 of the illumination light ILm1 irradiated onto the DMD 10 (neutral plane Pcc) different from the angle of incidence θα2 of the chief ray (central ray) Lp2 of the illumination light ILm2 irradiated onto the neutral plane Pcc.

[0175] If the design tilt angle θd (e.g., 17.5°) of the micromirror Msa of DMD 10 in the on state cannot be changed, then the angle θα between the optical axis AXc (and AXb) of condenser lens system 110 and the optical axis AXa of projection unit PLU is set in design to θα=2θd (e.g., 35.0°). In this embodiment, by adjusting the amount of decentering of MFE lens 108A1 with respect to the optical axis AXc of condenser lens system 110, the angle of incidence θα1 of illumination light ILm1 directed toward DMD 10 can be changed from the angle θα, and by adjusting the amount of decentering of MFE lens 108A2 with respect to the optical axis AXc, the angle of incidence θα2 of illumination light ILm2 directed toward DMD 10 can be changed from the angle θα.

[0176] Here, an example of selecting the wavelengths λ1 and λ2 will be described based on the relationship between the wavelength band and the telecentric error explained above in the graph of Fig. 32. Here, the central wavelength (peak wavelength) λ1 of the short-wavelength illumination light ILm1 is set to 343.0 nm, which is close to the wavelength band sensitive to general liquid photoresists and is therefore highly likely to be procured (manufactured) as an ultraviolet pulse light source, and the wavelength λ2 of the long-wavelength illumination light ILm2 is set to 405.0 nm (such as the h-line spectrum of a mercury discharge lamp) to match the wavelength band sensitive to dry film resists.

[0177] 31 , for illumination light ILm1 with a central wavelength λ1 of 343.0 nm, the ninth-order diffracted light Id9 generated from the micromirror Msa in the tightly turned-on state is equivalent to the zeroth-order light (imaging light beam), and the maximum telecentricity error Δθt1 on the image plane side that can occur under the initial design conditions (when the eccentricity of the MFE 108A1 is zero) is approximately +0.6° (strictly speaking, approximately +0.66° based on the above equation (2) or (3)). On the other hand, for illumination light ILm2 with a central wavelength λ2 of 405.0 nm, the eighth-order diffracted light Id8 generated from the micromirror Msa in the tightly turned-on state is equivalent to the zeroth-order light (imaging light beam), and the maximum telecentricity error Δθt2 on the image plane side that can occur under the initial design conditions (when the eccentricity of the MFE 108A2 is zero) is approximately −9.12°, based on the calculation based on the above equation (2) or (3).

[0178] From the above, by decentering the MFE lens 108A1 so that the telecentricity error Δθt1 (+0.66°) that may occur under the illumination light ILm1 is corrected, and by decentering the MFE lens 108A2 so that the telecentricity error Δθt2 (−9.12°) that may occur under the illumination light ILm2 is corrected, it is possible to minimize the overall telecentricity error of the imaging light beam that occurs during pattern exposure, even if the two illumination lights ILm1 and ILm2 are projected simultaneously or in a time-division manner onto the DMD 10. Ultimately, it can be said that the difference between the telecentricity error Δθt1 and the telecentricity error Δθt2 occurs due to the difference between the wavelength λ1 of the illumination light ILm1 and the wavelength λ2 of the illumination light ILm2.

[0179] Furthermore, in this embodiment, two MFE lenses 108A1 and 108A2 are provided individually for each of illumination light ILm1 and ILm2, and therefore aperture stop 108B having an elliptical aperture as described above in Fig. 29 can also be provided individually. In this case, the ellipticity of the aperture of aperture stop 108B provided on the exit side of MFE lens 108A1 is set according to the angle of incidence θα1 of the central ray of illumination light ILm1 to DMD 10, and the ellipticity of the aperture of aperture stop 108B provided on the exit side of MFE lens 108A2 is set according to the angle of incidence θα2 of the central ray of illumination light ILm2 to DMD 10.

[0180] 40 , the two MFE lenses 108A1 and 108A2 provided corresponding to the illumination light beams ILm1 and ILm2 are positioned decentered from the optical axis AXc in accordance with the telecentric errors Δθt1 and Δθt2, respectively. However, if tiltable parallel plates made of quartz are provided between the dichroic mirror DCM and the MFE lens 108A1 and between the dichroic mirror DCM and the MFE lens 108A2, it is not necessary to position each of the two MFE lenses 108A1 and 108A2 decentered. In this case, by individually adjusting the tilt angle of each of the parallel plates, each of the illumination light beams ILm1 and ILm2 projected onto the dichroic mirror DCM can be decentered from the optical axis AXc.

[0181] [Variant 4] As explained in Figure 40, if the angle of incidence θα1 of illumination light ILm1 of wavelength λ1 onto DMD 10 and the angle of incidence θα2 of illumination light ILm2 of wavelength λ2 onto DMD 10 can be adjusted individually, it is also possible to prevent the distribution of the imaging light beam (9th order diffracted light or 8th order diffracted light) formed on the pupil Ep of the projection unit PLU from being deformed into an ellipse, as explained above in Figure 36.

[0182] As illustrated in the description of Figure 40, let us assume that the wavelength λ1 of illumination light ILm1 is 343.0 nm and the wavelength λ2 of illumination light ILm2 is 405.0 nm. When the angles of incidence θα of illumination light ILm1 and ILm2 on DMD 10 are both 35.0°, as explained above, the zeroth-order light-equivalent component generated from DMD 10 by illumination light ILm1 becomes ninth-order diffracted light, and the maximum telecentricity error Δθt1 on the image plane side is approximately +0.66° (equivalent to approximately 0.01 in numerical aperture). Furthermore, the zeroth-order light-equivalent component generated from DMD 10 by illumination light ILm2 becomes eighth-order diffracted light, and the maximum telecentricity error Δθt2 on the image plane side is approximately -9.12° (equivalent to approximately 0.16 in numerical aperture).

[0183] Therefore, let us consider the distribution of ninth-order diffracted light of wavelength λ1 and eighth-order diffracted light of wavelength λ2 that appears in the pupil Ep of projection unit PLU when the pitch in the X' direction of the micromirrors Msa of DMD 10 when it is on is 5.4 μm, the maximum numerical aperture NAi(max) of projection unit PLU (projection magnification Mpp = 1 / 6) is 0.25, and the σ value is 0.6. Note that the wavelength width Δλ of both wavelengths λ1 and λ2 is sufficiently narrow (for example, Δλ≦0.2 nm), and the elliptical ratio (ΔOV) of the diffracted light distributed in the pupil Ep of projection unit PLU is cos θα = 0.82, which depends on the angle of incidence θα.

[0184] 41A is a diagram schematically illustrating the distribution H9c of ninth-order diffracted light and the distribution H8c of eighth-order diffracted light that appear in the pupil Ep of the projection unit PLU under the initial design conditions (angle of incidence θα = 35.0°), and the coordinate system X'Y' is the same as that of FIG. 35 or 36 . The distribution H9c of the ninth-order diffracted light and the distribution H8c of the eighth-order diffracted light are both set to the same numerical aperture NAy' in the Y' direction. Furthermore, the centers of the distributions H9c and H8c are P9c and P8c, respectively. Under the initial setting conditions, the telecentric error Δθt2 (-9.12°) of the eighth-order diffracted light distribution H8c is large, and therefore a portion of it extends outside the numerical aperture NAi(max) of the pupil Ep.

[0185] 41A , the position (deviation in the X′ direction) of the center P9c of distribution H9c relative to the optical axis AXa is approximately 0.01 when converted to a numerical aperture, and the position (deviation in the X′ direction) of the center P8c of distribution H8c relative to the optical axis AXa is approximately 0.16 when converted to a numerical aperture. Furthermore, the numerical aperture NAy′ corresponding to the spread of each of distributions H9c and H8c in the Y′ direction is NAy′ = σ·NAi(max) = 0.15 according to equation (10), and the numerical aperture NAx′ corresponding to the spread in the X′ direction is NAx′ = NAy′·cos θα = 0.123 according to equation (11).

[0186] Therefore, the arrangement of distributions H9c and H8c under the initial design conditions shown in Figure 41(A) is corrected to the state shown in Figure 41(B). That is, the center P9c of distribution H9c is shifted by Δs9 in the +X' direction from its initial position, and the center P8c of distribution H8c is shifted by Δs8 in the +X' direction from its initial position, so that the distribution H9c (wavelength λ1) of the ninth-order diffracted light and the distribution H8c (wavelength λ2) of the eighth-order diffracted light are positioned symmetrically in the X' direction across the optical axis AXa with most of them overlapping. These shifts can be set by the amount of decentering of each of the MFE lenses 108A1 and 108A2 shown in Figure 40.

[0187] In this example, because distributions H9c and H8c are both shifted in the +X' direction, the angles of incidence of illumination light ILm1 and ILm2 irradiating DMD 10 become smaller than the initial setting of 35.0°. By making such adjustments (corrections), the numerical aperture NAxf, which corresponds to the overall size (spread) of distributions H9c and H8c in the X' direction, can be set to approximately the same as the numerical aperture NAy' in the Y' direction, as previously described with reference to FIG.

[0188] In this example, the wavelength λ1 of illumination light ILm1 is 343.0 nm, and the wavelength λ2 of illumination light ILm2 is 405 nm, so the projection unit PLU needs to be chromatic aberration corrected at these two wavelengths. For this reason, it is desirable to make the wavelength width Δλ of each of illumination light ILm1 and ILm2 as narrow as possible (for example, to a few tens of pm or less).

[0189] 42 is an optical layout diagram of a modification of the embodiment of FIG. 40. In this modification, illumination light ILm1 of wavelength λ1 and illumination light ILm2 of wavelength λ2 are incident on a single MFE lens 108A at slightly different angles. As a result, on the exit surface side of the MFE lens 108A, a surface light source image formed by a collection of many point light sources SPF of wavelength λ1 and a surface light source image formed by a collection of many point light sources SPF of wavelength λ2 are formed, slightly shifted in the X' direction (the direction in which telecentric error occurs).

[0190] The optical arrangement in Fig. 41 is a modification of the optical path from the optical fiber bundle FBn to the MFE lens 108A shown in Fig. 28, and here, an optical fiber bundle FBn1 that guides illumination light ILm1 of wavelength λ1 and an optical fiber bundle FBn2 that guides illumination light ILm2 of wavelength λ2 (λ2 > λ1) are provided. As in Fig. 28, the illumination light from each of the optical fiber bundles FBn1 and FBn2 is irradiated onto the MFE lens 108A via an input lens system 104 that functions as a condenser lens, but between the input lens system 104 and the exit end of the optical fiber bundle FBn1, and between the input lens system 104 and the exit end of the optical fiber bundle FBn2, cube-shaped dichroic beam splitters (hereinafter simply referred to as beam splitters) DBS that have wavelength selection functions equivalent to those of the dichroic mirror DCM described in Fig. 40 are provided.

[0191] The light splitting surface of the beam splitter DBS (dichroic optical element) is disposed so as to be inclined at 45° in the X'Z plane with respect to the optical axis AXc of the input lens system 104, and the exit ends of the optical fiber bundles FBn1 and FBn2 are both located at the front focal position of the input lens system 104. In Figure 41, the centers (light emitting points) of the exit ends of the optical fiber bundles FBn1 and FBn2 are both disposed eccentrically by a predetermined amount from the position of the optical axis AXc.

[0192] The beam splitter DBS reflects the illumination light ILm1 of wavelength λ1 with a reflectance of 90% or more and transmits the illumination light ILm2 of wavelength λ2 with a transmittance of 90% or more. Therefore, the illumination light ILm1 from the exit end of the optical fiber bundle FBn1 is mostly reflected by the beam splitter DBS and enters the input lens system 104 with its chief ray (central ray) parallel to the optical axis AXc and decentered. On the other hand, the illumination light ILm2 from the exit end of the optical fiber bundle FBn2 is mostly transmitted through the beam splitter DBS and enters the input lens system 104 with its chief ray (central ray) parallel to the optical axis AXc and decentered.

[0193] The illumination light ILm1 that passes through the input lens system 104 becomes a nearly parallel luminous flux, but is incident on the MFE lens 108A at an angle generally tilted with respect to the optical axis AXc. Similarly, the illumination light ILm2 that passes through the input lens system 104 also becomes a nearly parallel luminous flux, but is incident on the MFE lens 108A at an angle generally tilted with respect to the optical axis AXc. Because the incident end of the MFE lens 108A is set at the position of the rear focal point of the input lens system 104, the two illumination lights ILm1 and ILm2 overlap in a nearly circular distribution within the plane of the incident end of the MFE lens 108A.

[0194] However, because the angles of incidence of illumination light ILm1 and ILm2 on the MFE lens 108A are slightly different, as described above in FIG. 30 , a point light source SPF1 due to illumination light ILm1 and a point light source SPF2 due to illumination light ILm2 are formed at the output end of the MFE lens 108A (conjugate with the output ends of the optical fiber bundles FBn1 and FBn2) of the MFE lens 108A, with a slight separation in the X' direction. As a result, a surface light source (Ips) formed by a nearly circular collection of the many point light sources SPF1 due to illumination light ILm1 and a surface light source (Ips) formed by a nearly circular collection of the many point light sources SPF2 due to illumination light ILm2 are formed at the output end of the MFE lens 108A, with a shift in the X' direction. The amount of this shift is less than the dimension in the X' direction within the X'Y' plane of one lens element EL of the MFE lens 108A.

[0195] In this way, by differentiating the angles of incidence of the two illumination lights ILm1 and ILm2 on the MFE lens 108A, the surface light source of illumination light ILm1 and the surface light source of illumination light ILm2 formed at the exit end of the MFE lens 108A can be relatively shifted in the X' direction. As a result, the angles of incidence of the chief ray of illumination light ILm1 and the chief ray of illumination light ILm2 irradiated onto the DMD 10 can be individually adjusted (corrected), albeit slightly.

[0196] [Variation 6] Fig. 43 is an optical layout diagram that is a further variation of Variation 5 of Fig. 42, in which illumination light ILm1 of wavelength λ1 and illumination light ILm2 of wavelength λ2 critically illuminate a single MFE lens 108A. In Fig. 43, the coordinate system X'Y'Z is set to the same as in Fig. 42, and the wavelengths λ1 and λ2 are set to have a relationship of λ1<λ2.

[0197] In this example, the output end pf1 of the optical fiber bundle FBn1 that guides the illumination light ILm1 and the input end pff of the MFE lens 108A are set in a conjugate relationship (imaging relationship) with each other by a magnifying imaging system consisting of lens systems 104A1 and 104B arranged along the optical axis AXc. A dichroic beam splitter (hereinafter simply referred to as the beam splitter) DBS as shown in FIG. 42 is provided between the lens systems 104A1 and 104B at approximately the position of the pupil epi. Therefore, the illumination light ILm1 that diverges from the output end pf1 of the optical fiber bundle FBn1 passes through the lens system 104A1, is reflected in the −Z direction by the wavelength separation surface (dichroic surface) of the beam splitter DBS, and then passes through the lens system 104B to illuminate the illumination field Imf1 on the input end pff of the MFE lens 108A.

[0198] Similarly, the output end pf2 of the optical fiber bundle FBn2 that guides the illumination light ILm2 and the input end pff of the MFE lens 108A are set in a conjugate relationship (imaging relationship) with each other by an enlarged imaging system made up of lens systems 104A2 and 104B that are arranged along the optical axis AXc. Therefore, the illumination light ILm2 that diverges from the output end pf2 of the optical fiber bundle FBn2 passes through the lens system 104A2, then passes through the wavelength separation surface (dichroic surface) of the beam splitter DBS in the −Z direction, and passes through the lens system 104B to irradiate the illumination area Imf2 on the input end pff of the MFE lens 108A.

[0199] In this example, each of the optical fiber bundles FBn1 and FBn2 uses a single fiber with a core diameter of approximately 1.2 mm, and therefore each of the output ends pf1 and pf2 has a circular shape. Therefore, each of the illumination areas Imf1 and Imf2 formed on the input end pff of the MFE lens 108A also has an enlarged circular shape. As an example, if the magnification of the magnified imaging system formed by the lens systems 104A1 and 104B and the magnified imaging system formed by the lens systems 104A2 and 104B is 20 times, the diameter of each of the illumination areas Imf1 and Imf2 will be 24 mm. When the center point of the output end pf1 of the optical fiber bundle FBn1 is aligned with the optical axis AXc, and the center point of the output end pf2 of the optical fiber bundle FBn2 is aligned with the optical axis AXc, the illumination area Imf1 by the illumination light ILm1 and the illumination area Imf2 by the illumination light ILm2 overlap concentrically on the input end pff of the MFE lens 108A.

[0200] Therefore, in this example, the overall dimensions in the X' and Y' directions of the incident end pff of MFE lens 108A are made larger than the diameters of each of the illumination areas Imf1 and Imf2, while the dimensions in the X' and Y' planes of each lens element EL (see FIGS. 7 and 12 above) are made as small as possible. Then, by providing a mechanism that individually adjusts the position of each of the illumination areas Imf1 and Imf2 within the plane of the incident end pff of MFE lens 108A, it is possible to adjust telecentricity errors in the imaging light beam and mitigate elliptical distribution of the imaging light beam at the pupil Ep of projection unit PLU.

[0201] The position adjustment of each of the illumination fields Imf1 and Imf2 within the plane of the incident end pff of the MFE lens 108A can be achieved by a fine adjustment mechanism that mechanically shifts the output ends pf1 and pf2 of the optical fiber bundles FBn1 and FBn2. However, since the magnification of the magnified image-forming system formed by the lens systems 104A1, 104A2, and 104B is large, it is practically preferable to provide tiltable quartz parallel plates HV1 and HV2 between the output end pf1 of the optical fiber bundle FBn1 and the lens system 104A1, and between the output end pf2p of the optical fiber bundle FBn2 and the lens system 104A2, as shown in Figure 43.

[0202] In this case, by adjusting the tilt amount of the parallel plate HV1 (HV2), the chief ray of the illumination light ILm1 (ILm2) that travels parallel to the optical axis AXc from the center point of the output end pf1 (pf2) of the optical fiber bundle FBn1 (FBn2) and enters the lens system 104A1 (104A2) can be adjusted for decentering on the order of μm from the optical axis AXc in the X' direction.

[0203] 44 is an exaggerated diagram showing an example of the arrangement of the illumination areas Imf1 and Imf2 projected onto the plane of the incident end pff of the MFE lens 108A. As shown in FIG. 44, the overall dimensions of the incident end pff of the MFE lens 108A in the X' and Y' directions are set to be larger than the diameter of each of the circular illumination areas Imf1 and Imf2 (enlarged image areas at the output ends pf1 and pf2). Furthermore, the center point Pz1 of the illumination area Imf1 is conjugate with the center point of the output end pf1 of the optical fiber bundle FBn1, and the center point Pz2 of the illumination area Imf2 is conjugate with the center point of the output end pf2 of the optical fiber bundle FBn2.

[0204] 43, the illumination regions Imf1 and Imf2 (respective center points Pz1 and Pz2) can be distributed symmetrically in the X' direction with decentering across the optical axis AXc. As a result, the collection (circular distribution) of multiple point light sources SPF1 formed by illumination light ILm1 on the output end side of MFE lens 108A and the collection (circular distribution) of multiple point light sources SPF2 formed by illumination light ILm2 are distributed with a predetermined amount of decentering, with most of them overlapping in the X' direction.

[0205] Therefore, the overall external shape of the surface light source (an aggregate of numerous point light sources SPF1 and SPF2) formed on the exit end side of MFE lens 108A is elliptical with the major axis in the X' direction and the minor axis in the Y' direction, making it possible to correct (cancel out) the elliptical distribution of the imaging light flux at the pupil Ep of projection unit PLU that occurs when DMD 10 is obliquely illuminated. In this case, aperture stop 108B shown in Fig. 29 above becomes unnecessary, and as long as each of illumination areas Imf1 and Imf2 is distributed within the range of the entrance end pff of MFE lens 108A, partial blocking or vignetting of the illumination light does not occur, which has the advantage of preventing loss of the illumination light amount.

[0206] Furthermore, the configuration of Fig. 43 makes it possible to correct the telecentricity error Δθt of the imaging light beam in addition to correcting the elliptical distribution of the imaging light beam at the pupil Ep of the projection unit PLU. Fig. 45, like Fig. 44, is an exaggerated diagram showing an example of the arrangement of each of the illumination fields Imf1 and Imf2 projected onto the plane of the incident end pff of the MFE lens 108A. In Fig. 45, the center point Pz1 of the illumination field Imf1 and the center point Pz2 of the illumination field Imf2 are both decentered in the -X' direction from the position of the optical axis AXc. The elliptical shape is corrected by differentiating the amount of decentering of the center point Pz1 from the amount of decentering of the center point Pz2.

[0207] Furthermore, decentering the overall distribution of illumination areas Imf1 and Imf2 in the -X' direction from the optical axis AXc is equivalent to laterally shifting the light source image on the exit surface side of MFE lens 108A as seen from the condenser lens system 110 side shown in Figure 29, and the incident angle θα of the central ray of illumination light ILm1 and ILm2 irradiated onto DMD 10 can be changed from the design value (e.g., 35.0°), making it possible to correct the telecentric error Δθt.

[0208] From the above, a configuration such as that shown in Figure 43, in which the positions of the illumination fields Imf1 and Imf2 of the illumination lights ILm1 and ILm2 irradiated onto the incident end pff of the MFE lens 108A can be shifted laterally with a certain margin, also functions as a telecentricity adjustment mechanism. Furthermore, in the configuration shown in Figure 43, a tiltable parallel plate can also be provided between the lens system 104B and the incident end pff of the MFE lens 108A. In that case, the two illumination fields Imf1 and Imf2 shown in Figures 44 and 45 can be shifted laterally together on the incident end pff of the MFE lens 108A, making it possible to easily correct the overall telecentricity error of the imaging light beam projected from the projection unit PLU onto the substrate P.

[0209] For the above reasons, when the optical system from one optical fiber bundle FBn to the MFE lens 108A of the illumination unit ILU shown in Fig. 28 is changed to a magnifying and imaging optical system and the incident end of the MFE lens 108A is used as a critical illumination system, telecentric adjustment can be performed by providing a tiltable parallel plate between the output end of the optical fiber bundle FBn and the magnifying and imaging system or between the magnifying and imaging optical system and the incident end of the MFE lens 108A. In this case, since a single circular illumination area is formed at the incident end of the MFE lens 108A, if the wavelength width Δλ of the illumination light ILm is narrow, for example, ±0.2 nm or less, the effect of reducing ellipticity due to broadband light as described in Figs. 35 to 39 cannot be obtained.

[0210] For this reason, an aperture stop 108B having an elliptical opening as shown in Fig. 29 is provided, but because the position of the illumination area of ​​the illumination light ILm on the incident end of the MFE lens 108A changes, a fine movement mechanism is required to horizontally shift the aperture stop 108B independently. Alternatively, a cylindrical lens, anamorphic lens, toric lens, or DOE (Diffraction Optical Element) plate having an anisotropic refractive power (power) that deforms the illumination area on the incident end of the MFE lens 108A into an elliptical shape may be incorporated as the magnifying image-forming optical system provided between the optical fiber bundle FBn and the MFE lens 108A in Fig. 28.

[0211] [Other Modifications] In the above-described embodiments and modifications, an isolated pattern is not necessarily limited to a case where a single micromirror or a row of all micromirrors Ms of the DMD 10 is in the ON state. For example, a pattern where two, three (1 × 3), four (2 × 2), six (2 × 3), eight (2 × 4), or nine (3 × 3) micromirrors Msa are densely arranged in the ON state and, for example, ten or more micromirrors Msb are in the OFF state in the X' direction and Y' direction can also be considered an isolated pattern. Conversely, if two, three (1x3), four (2x2), six (2x3), eight (2x4), or nine (3x3) off-state micromirrors Msb are densely arranged, and the surrounding micromirrors Ms are densely on-state micromirrors Msa over a range of, for example, several or more (corresponding to a dimension several times or more than the isolated pattern) in the X' and Y' directions, then this can also be considered a land-like pattern.

[0212] Furthermore, the line-and-space pattern as a pattern configuration is not necessarily limited to the configuration shown in Figure 24 in which one row of on-state micromirrors Msa and one row of off-state micromirrors Msb are alternately arranged. For example, two rows of on-state micromirrors Msa and two rows of off-state micromirrors Msb may be alternately arranged, three rows of on-state micromirrors Msa and three rows of off-state micromirrors Msb may be alternately arranged, or two rows of on-state micromirrors Msa and four rows of off-state micromirrors Msb may be alternately arranged. In any pattern configuration, if the distribution (density or concentration) of on-state micromirrors Ms per unit area (e.g., an array area of ​​100 x 100 micromirrors Ms) among all the micromirrors Ms on the DMD 10 is known, the telecentric error Δθt and the degree of asymmetry of the pattern edges can be easily determined by simulation or the like.

[0213] In the embodiments and modifications described above, it has been explained that the use of multiple wavelengths and a broadband of illumination light ILm has the effect of reducing the elliptical deformation that is unavoidable due to oblique illumination in the distribution of the imaging light beam (high-order diffracted light) formed on the pupil Ep of the projection unit PLU. Furthermore, the use of multiple wavelengths and a broadband of illumination light ILm also has the effect of reducing fluctuations in the illuminance of the imaging light beam (reflected diffracted light that is the component equivalent to zero-order light) that occur due to a residual error from the design value of the tilt angle θd of the micromirror Msa in the on state of the DMD 10, or a fluctuation error in the tilt angle θd over time.

[0214] Here, changes in the inclination angle (tilt angle) θd of the micromirrors Msa in the ON state and changes in the diffraction angle θj of the high-order diffracted light Idj from the DMD 10 will be described again with reference to Figure 46. Figure 46 is a diagram that schematically explains the angular state when diffracted light Idj from a large number of micromirrors Msa that are densely turned ON is incident on the projection unit PLU. The coordinate system X'Y'Z in Figure 46 is set in the same way as in Figures 6, 10, 20-22, 28, 29, etc.

[0215] As shown in Figure 46, the actual tilt angle θd of a large number of micromirrors Msa in the ON state is expressed as θd = θo ± Δθd, where θo is the design tilt angle and Δθd is the absolute value of the fluctuation angle (error angle) as the error component. The angle of incidence θα, which is the angle between the optical axis AXb of the projection unit PLU and the optical axis AXc of the condenser lens system 110 in the illumination unit ILU bent by inclined mirror 112, is set here to be double the design tilt angle θo of the micromirrors Msa in the ON state.

[0216] As shown in equation (3) above, the diffraction angle θj of diffracted light Idj of order j is found by the wavelength λ of illumination light ILm, the array pitch Pd of micromirrors Msa in the X' direction, and the angle of incidence θα as follows: sin θj=sin θα-j(λ / Pd) (3) The diffraction angle θj found by this equation is the angle from the optical axis AXa of projection unit PLU, and when the diffraction angle θj is positive, the diffracted light Idj is tilted counterclockwise from the optical axis AXa, and when the diffraction angle θj is negative, the diffracted light Idj is tilted clockwise from the optical axis AXa.

[0217] As shown in the example above, assuming that the wavelength λ is 343.333 nm, the array pitch Pd is 5.4 μm, the angle of incidence θα is 35.0°, and the error angle Δθd when the micromirror Msa is tilted is zero, the diffraction angle θ9 of the central ray of the 9th-order diffracted light Id9 for j = 9 is approximately +0.078° (equivalent to approximately 0.00135 in terms of the object-side numerical aperture NAo) according to equation (3), making the 9th-order diffracted light Id9 the zeroth-order equivalent component. Furthermore, the diffraction angle θ8 of the central ray of the 8th-order diffracted light Id8 before the 9th order is approximately +3.72° (equivalent to approximately 0.0649 in terms of the object-side numerical aperture NAo), and the diffraction angle θ10 of the central ray of the 10th-order diffracted light Id10 after the 9th order is approximately −3.57° (equivalent to approximately 0.0622 in terms of the object-side numerical aperture NAo).

[0218] Furthermore, if the maximum numerical aperture NAi(max) on the image plane side of the projection unit PLU is 0.3 and the projection magnification Mp is 1 / 6, the maximum numerical aperture NAo(max) on the object plane side (incident side) of the projection unit PLU is 0.05, and the maximum opening angle θpo(max) corresponding to the numerical aperture NAo(max) is approximately 2.87°. Therefore, the central ray of the 8th-order diffracted light Id8 and the central ray of the 10th-order diffracted light Id10 are both wider than the maximum opening angle θpo(max), and therefore do not enter the projection unit PLU. However, as previously described in FIG. 23 , both the 8th-order diffracted light Id8 and the 10th-order diffracted light Id10 have a circular (or elliptical) distribution Hpb whose size corresponds to the σ value of the illumination light ILm. Therefore, depending on the σ value, a portion of the 8th-order diffracted light Id8 or the 10th-order diffracted light Id10 may enter the projection unit PLU.

[0219] In this way, when the error angle Δθd is zero, the light intensity of each of the 8th to 10th order diffracted light beams Id8, Id9, and Id10 follows the point spread function Iea obtained by regarding the reflecting surface of a single micromirror Msa as a minute rectangular aperture, as explained above in Figure 19. The light intensity Ie of the point spread function Iea is expressed by the above equation (1), which will be described again below. In this equation (1), Io is the peak value of the actual light intensity, but in the following explanation, Io = 1 (100%). Furthermore, when the error angle Δθd is zero (i.e., the tilt angle θd = θo) and the angle of incidence θα of illumination light ILm on DMD 10 is set exactly to θα = 2θd, X' in equation (1) represents the distance (length) in the X' direction, with the position of the optical axis AXa of projection unit PLU as the origin (zero point).

[0220] Furthermore, in equation (1), when X' = π (3.1416), the light intensity Ie becomes zero, but the position is determined by the wavelength λ of the illumination light ILm, the dimension Lms of the reflecting surface of the micromirror Msa in the X' direction, and a predetermined conversion coefficient K, such that X' = π = K (λ / Lms). However, when the micromirror Msa in the ON state is viewed from the projection unit PLU side, the dimension of the reflecting surface of the micromirror Msa in the X' direction appears to be reduced in accordance with the cosine of the tilt angle θd. Therefore, the conversion coefficient K is expressed as in the following equation (13): K = π (Lms cos θd) / λ (13)

[0221] Here, if Lms·cos θd=Lms', the position of X'=π where Ie=0 corresponds to the diffraction angle βs of the first-order light determined by the following equation (14): sin βs=λ / Lms' (14) The left side of equation (14) represents the numerical aperture NAo of the diffraction angle βs on the object plane side (DMD 10 side) of the projection unit PLU, so the previous equation (1) can be transformed into the following equation (15) using the numerical aperture NAo as a variable: Ie=[sin(K·NAo) / (K·NAo)] ２ However, K=π・Lms' / λ... (15)

[0222] As an example, Figure 47 shows the point spread function Iea and the distributions of the 8th to 10th order diffracted light Id8, Id9, and Id10 that appear when the wavelength λ of the illumination light ILm is 343.333 nm, the array pitch Pdx of the micromirrors Ms is 5.4 μm, the dimension Lms in the X' direction of the reflective surface of a single micromirror Ms is 4.6 μm (= 85% Pdx), and the incident angle θα of the illumination light ILm is 35.0°, and the error angle Δθd of the micromirror Msa in the on state is zero.

[0223] In Figure 47, the vertical axis represents light intensity Ie, with the maximum value set to 1 (100%), and the horizontal axis represents the numerical aperture NAo on the object plane side. Also, assume that the error angle Δθd of the tilted micromirror Msa in the on state is zero, and the angle of incidence θα is set to exactly twice the tilt angle θd, or 35.0°. Therefore, the origin of the zero numerical aperture NAo coincides with the position of the optical axis AXa of the projection unit PLU within the X'Y' plane. Note that the numerical aperture NAo does not take negative values, but here, the positive range of the numerical aperture NAo is defined as the +X' side of the optical axis AXa, and the negative range is defined as the -X' side of the optical axis AXa.

[0224] The point spread function Iea in Figure 47 is a simulation of the characteristics of equation (15) above, and the effective dimension Lms' in the X' direction of the reflecting surface of the micromirror Msa is set to Lms·cos θd = 4.6 μm·cos(17.5°) ≈ 4.38 μm. The value of the numerical aperture NAo (corresponding to X' = π) at which the light intensity Ie first becomes 0 on the point spread function Iea is ±0.0784. Note that when the object plane-side numerical aperture NAo on the horizontal axis shown in Figure 47 is expressed in terms of the numerical aperture NAi on the image plane side (substrate P side), it is sufficient to use the projection magnification Mp (for example, Mp = 1 / 6) of the projection unit PLU and set NAi = NAo / Mp.

[0225] On the other hand, under the conditions of error angle Δθd = 0, incident angle θd = 35.0°, and wavelength λ = 343.333 nm, the central rays of the eighth-order diffracted light Id8, ninth-order diffracted light Id9, and tenth-order diffracted light Id10 obtained by the above equation (3) are accompanied by diffraction angles θ8, θ9, and θ10, respectively, and proceed toward the projection unit PLU. The object surface-side numerical aperture NAo9 corresponding to the diffraction angle θ9 of the ninth-order diffracted light Id9 is the value (sin θ9) of the right-hand side of equation (3) calculated with the order j set to 9, and is approximately 0.00135. Similarly, the object-side numerical aperture NAo8 corresponding to the diffraction angle θ8 of the 8th-order diffracted light Id8 is the value (sin θ8) of the right-hand side of equation (3) calculated by setting the order j to 8, which is approximately 0.06493, and the object-side numerical aperture NAo10 corresponding to the diffraction angle θ10 of the 10th-order diffracted light Id10 is the value (sin θ10) of the right-hand side of equation (3) calculated by setting the order j to 10, which is approximately 0.06223.

[0226] The object plane side numerical aperture NAo9 (≈0.00135) of the central ray of the ninth-order diffracted light Id9 is extremely small, and the image plane side numerical aperture NAi9 when the projection magnification Mp is 1 / 6 is also approximately 0.0081, so there is no need to perform telecentric error correction to fine-tune the angle of incidence θα of the illumination light ILm. Furthermore, when the error angle Δθd = 0, the reflected light from a single micromirror Msa is distributed within the pupil Ep of the projection unit PLU such that the point spread function Iea of ​​Figure 47 coincides with its origin (point 0) and the optical axis AXa.

[0227] Because the object-side numerical aperture NAo9 of the ninth-order diffracted light Id9 is extremely small, the light intensity Ie of the ninth-order diffracted light Id9 calculated by equation (15) is 0.99 or more (almost 100%). In contrast, the light intensity Ie of the eighth-order diffracted light Id8 is 0.039 (3.9%), and the light intensity Ie of the tenth-order diffracted light Id10 is 0.058 (5.8%), which are significantly attenuated.

[0228] If the maximum numerical aperture NAi(max) on the image plane side of the projection unit PLU is 0.3, then the maximum numerical aperture NAo(max) on the object plane side when the projection magnification Mp is 1 / 6 will be 0.05. For this reason, the central rays of the 8th-order diffracted light Id8 and the 10th-order diffracted light Id10 shown in Figure 47 are both outside the pupil Ep of the projection unit PLU and are not projected onto the substrate P.

[0229] From the above state, if a certain error angle Δθd occurs on average for many of the on-state micromirrors Msa of DMD 10, the point spread function Iea shown in Figure 47 will shift laterally overall. The point spread function Iea is a distribution generated by reflected light from the reflective surface of a single micromirror Msa, and the chief ray of the reflected light that enters projection unit PLU will be tilted from the optical axis AXa by an angle double the error angle Δθd. As a result, the point spread function Iea generated within the pupil Ep of projection unit PLU will also shift laterally in the X' direction.

[0230] On the other hand, the diffraction angle θ9 (and sin θ9) of the ninth-order diffracted light Id9 does not change because, as is clear from equation (3) above, the wavelength λ of the illumination light ILm, the angle of incidence θα, and the pitch Pdx of the micromirrors Ms do not change. That is, as shown in Figure 47, the object-side numerical aperture NAo of the central ray of each of the ninth-order diffracted light Id9, the eighth-order diffracted light Id8, and the tenth-order diffracted light Id10 does not change.

[0231] Fig. 48 is a graph showing simulated point spread function Iea1 and point spread function Iea2 when the tilt angle θd of the micromirror Msa in the on state is changed by an error angle Δθd of +0.5° from the designed tilt angle θo (17.5°) for the characteristics shown in Fig. 47. The light intensity Ie on the vertical axis and the object-side numerical aperture NAo on the horizontal axis in Fig. 48 are both shown on the same scale as the vertical and horizontal axes in Fig. 47. The object-side numerical aperture NAo of the central ray of each of the 8th-order diffracted light Id8, 9th-order diffracted light Id9, and 10th-order diffracted light Id10 in Fig. 48 is the same as the value shown in Fig. 47.

[0232] 48, when the error angle Δθd is +0.5°, the tilt angle θd of the micromirror Msa is θd = θo (17.5°) + Δθd = 18°, as shown in FIG. 46. Furthermore, the central ray of the reflected light from the micromirror Msa toward the projection unit PLU is tilted clockwise from the optical axis AXa by 2·Δθd = 1.0°. The angle 2·Δθd = 1.0° is converted to the object plane-side numerical aperture NAo, which is approximately 0.0175 [sin(2·Δθd)]. Therefore, when the error angle Δθd is zero, the point spread function Iea shifts by 0.0175 in the negative direction (−X′ direction) on the object plane-side numerical aperture NAo, as shown in point spread function Iea1.

[0233] Similarly, when the error angle Δθd is +1.0°, the tilt angle θd of the micromirror Msa is θd = θo (17.5°) + Δθd = 18.5°. Furthermore, the central ray of the reflected light from the micromirror Msa, directed toward the projection unit PLU, is tilted clockwise by 2·Δθd = 2.0° from the optical axis AXa. The angle 2·Δθd = 2.0° is converted to the object plane-side numerical aperture NAo, which is approximately 0.0349 [sin(2·Δθd)]. Therefore, when the error angle Δθd is zero, the point spread function Iea shifts by 0.0349 in the negative direction (−X' direction) on the object plane-side numerical aperture NAo, as shown in point spread function Iea2.

[0234] Furthermore, when the error angle Δθd becomes negative (when the tilt angle θd becomes smaller than the designed tilt angle θo), the point spread function Iea for the error angle Δθd=0 shifts overall in the positive direction (+X' direction) on the object-side numerical aperture NAo by an amount corresponding to a multiple of the error angle Δθd.

[0235] When the error angle Δθd is +0.5°, the light intensity Ie of the 9th-order diffracted light Id9, which is the zeroth-order light-equivalent component generated from the numerous micromirrors Msa of the DMD 10 in the on-state, follows the point spread function Iea1 and is therefore approximately 0.824. While the light intensity Ie of the 9th-order diffracted light Id9 was 0.99 or greater (almost 100%) when the error angle Δθd was zero, the light intensity Ie of the 9th-order diffracted light Id9 drops to approximately 82% when the error angle Δθd is only +0.5°. Similarly, when the error angle Δθd is +1.0°, the light intensity Ie of the 9th-order diffracted light Id9 generated from the DMD 10 follows the point spread function Iea2 and is therefore approximately 0.467. Therefore, when the error angle Δθd is +1.0°, the light intensity Ie of the 9th-order diffracted light Id9 drops dramatically to approximately 47%.

[0236] On the other hand, the light intensity Ie of the eighth-order diffracted light Id8 follows the point spread function Iea1 when the error angle Δθd is +0.5°, and is therefore approximately 0.028 (approximately 3%), and when the error angle Δθd is +1.0°, it follows the point spread function Iea2, and is therefore approximately 0.047 (approximately 5%). Furthermore, the light intensity Ie of the tenth-order diffracted light Id10 follows the point spread function Iea1 when the error angle Δθd is +0.5°, and is therefore approximately 0.295 (approximately 30%), and when the error angle Δθd is +1.0°, it follows the point spread function Iea2, and is therefore approximately 0.659 (approximately 66%).

[0237] From the above, when a pattern is projected in which the micromirrors Msa that are turned on among the many micromirrors Ms of the DMD 10 are densely packed at a pitch Pdx, the light intensity of the high-order diffracted light (here, the 9th-order diffracted light Id9), which is the component equivalent to the 0th-order light, decreases according to the magnitude of the error angle Δθd of the micromirrors Msa. However, the telecentric error Δθt of the high-order diffracted light (the 9th-order diffracted light Id9) does not change.

[0238] In contrast, when the micromirrors Msa among the many micromirrors Ms of the DMD 10 are individually and discretely distributed in the on state and project a pattern (isolated point image) that does not substantially generate higher-order diffracted light, a simple point image intensity distribution is projected, so there is almost no reduction in light intensity according to the magnitude of the error angle Δθd of the micromirror Msa. However, the telecentric error Δθt corresponding to the multiple angle of the error angle Δθd changes.

[0239] The DMD 10 and the substrate P are set in a conjugate (imaging relationship) by the projection unit PLU, so in the case of an isolated point image, even if the telecentric error Δθt changes due to a change in the error angle Δθd of the micromirror Msa, the position of the reflected light from the micromirror Msa projected onto the substrate P (projection position of the point image) does not change. However, when a pattern is projected in which micromirrors Msa that are turned on are densely packed at a pitch Pdx (such as a large land pattern or a thick wiring pattern), the light intensity of the projected image, i.e., the amount of exposure, will be significantly reduced depending on the size of the error angle Δθd of the micromirror Msa.

[0240] However, by using illumination light ILm containing different wavelength components as described above, it is possible to mitigate the reduction in exposure amount (light intensity) due to the error angle Δθd of the micromirror Msa. This will be explained with reference to FIGS. 49 and 50. FIG. 49 shows the point spread function Iea when the wavelength λ1 of the illumination light ILm is 343.333 nm and the point spread function IeaL when the wavelength λ2 of the illumination light ILm is 355.000 nm, when the error angle Δθd of the on-state micromirror Msa is zero, using the same DMD 10 as in FIG. 48 and the same angle of incidence θα. FIG. 50 shows the characteristics of the point spread functions Iea and IeaL when the error angle Δθd of the on-state micromirror Msa is +0.5° compared to the state shown in FIG. 49.

[0241] In each of Figures 49 and 50, the vertical and horizontal axes represent light intensity Ie and object plane side numerical aperture NAo, respectively, as in Figure 48, but because the maximum image plane side numerical aperture NAi of projection unit PLU was set to 0.3 and the projection magnification Mp to 1 / 6, the maximum value of the object plane side numerical aperture NAo was set to ±0.05. As shown in Figure 49, the point spread function Iea at wavelength λ1 = 343.333 nm and the point spread function IeaL at wavelength λ2 = 355.000 nm are each defined by the above equation (15), but there is not much difference between them.

[0242] 48, the central ray of the ninth-order diffracted light Id9(λ1) at wavelength λ1=343.333 nm appears at the object-side numerical aperture NAo=+0.00135, and its light intensity Ie is 0.999 (almost 100%). Also, the central ray of each of the eighth-order diffracted light Id8 and tenth-order diffracted light Id10 at wavelength λ1=343.333 nm is located outside the maximum object-side numerical aperture NAo=±0.05.

[0243] On the other hand, at a wavelength of λ2=355.000 nm, the central ray of the 9th-order diffracted light Id9(λ2), which is the component equivalent to the 0th-order light, appears at an object-side numerical aperture NAo=-0.0181 based on the above formula (3), and the central ray of the 8th-order diffracted light Id8(λ2) appears at an object-side numerical aperture NAo=+0.0477. The 10th-order diffracted light Id10 at a wavelength of λ2=355.000 nm is located outside the maximum object-side numerical aperture NAo=±0.05. In this case, the light intensity Ie of the 9th-order diffracted light Id9(λ2) calculated by formula (15) is approximately 0.848, and the light intensity Ie of the 8th-order diffracted light Id8(λ2) is approximately 0.271.

[0244] Here, assuming that the illuminance of the wavelength λ1=343.333 nm in the illumination light ILm is equal to the illuminance of the wavelength λ2=355.000 nm, the total light intensity (light amount) of the 9th-order diffracted light Id9(λ1) and Id9(λ2) of the component equivalent to the 0th-order light incident on the projection unit PLU from the DMD 10 is 0.999+0.848=1.847 (approximately 185%). Next, a case where the error angle Δθd of the micromirror Msa in the on state is +0.5° will be described with reference to FIG.

[0245] As shown in Fig. 50, the point spread function Iea at wavelength λ1 (343.333 nm) and the point spread function IeaL at wavelength λ2 (355.000 nm) both shift by approximately -0.0175 from the position of the optical axis AXa (origin 0) on the object surface-side numerical aperture NAo, just as in Fig. 48. A change in the error angle Δθd alone does not change the diffraction angle of the central ray of each of the 9th-order diffracted light beams Id9(λ1) and Id9(λ2) of the component equivalent to 0th-order light, i.e., their positions on the object surface-side numerical aperture NAo. Therefore, the light intensity Ie of the ninth-order diffracted light Id9 (λ1) at the error angle Δθd = +0.5° is approximately 0.824, following the point spread function Iea after the shift, and the light intensity Ie of the ninth-order diffracted light Id9 (λ2) at the error angle Δθd = +0.5° is approximately 1.000, following the point spread function IeaL after the shift.

[0246] Therefore, when the error angle Δθd changes from zero to +0.5°, the total light intensity (light amount) of the ninth-order diffracted light Id9(λ1) and Id9(λ2), which are components equivalent to zero-order light and enter the projection unit PLU from the DMD 10, changes to 1.000 + 0.824 = 1.824 (approximately 182%). Therefore, when the error angle Δθd changes to +0.5°, the total light intensity (light amount) decreases by only -1.2% [= (1.824 - 1.847) / 1.847]. In this way, by including light of multiple different wavelength components as the illumination light ILm, it is possible to mitigate the reduction in illuminance (exposure amount) caused by the error angle Δθd of the micromirror Msa in the on state.

[0247] 50, however, the central ray of the ninth-order diffracted light Id9(λ2) at wavelength λ2 (355.000 nm) generates a telecentricity error Δθt corresponding to approximately −0.0181 at the object plane-side numerical aperture NAo and approximately −0.11 (=−0.0181 / Mp) at the image plane-side numerical aperture NAi. Therefore, it is advisable to perform telecentricity error adjustment so that the central ray of the ninth-order diffracted light Id9(λ1) at wavelength λ1 (343.333 nm) and the central ray of the ninth-order diffracted light Id9(λ2) at wavelength λ2 (355.000 nm) are positioned approximately symmetrically across the optical axis AXa (origin 0 in FIG. 50) of the projection unit PLU.

[0248] Specifically, the angle of incidence θα of illumination light ILm, which contains coaxial light of wavelengths λ1 and λ2, is finely adjusted from its initial value (35.0°). As shown in Figure 50, the position of the central ray of ninth-order diffracted light Id9(λ1) on the object plane-side numerical aperture NAo is approximately 0.00135, and the position of the central ray of ninth-order diffracted light Id9(λ2) on the object plane-side numerical aperture NAo is approximately -0.0181, resulting in an average position of the object plane-side numerical aperture NAo of approximately -0.0168. The object plane-side numerical aperture NAo = -0.0168 is converted to an angle of approximately -0.96°, and the angle of incidence θα of illumination light ILm is changed from its initial value (35.0°) by half that amount, or approximately -0.48°. In this case, the center positions of the point spread functions Iea and IeaL, which are at NAo=-0.0175 shown in FIG. 50, also shift by approximately 0.0168 in the positive direction.

[0249] In Figures 48 and 50, the error angle Δθd of the micromirror Msa in the on state is assumed to have changed in the positive direction, but it can also change in the negative direction. For example, if the error angle Δθd is set to -0.5°, the center positions of the point spread functions Iea and IeaL shown in Figure 50 will shift to +0.0175° on the object-side numerical aperture NAo. As a result, the light intensities Ie of both the 9th-order diffracted light Id9(λ1) at wavelength λ1 (343.333 nm) and the 9th-order diffracted light Id9(λ2) at wavelength λ2 (355.000 nm) will both decrease. In particular, the light intensity Ie of the 9th-order diffracted light Id9(λ2) will decrease significantly, while the light intensity of the 8th-order diffracted light Id8(λ2) at wavelength λ2 (355.000 nm) will increase more than the light intensity of the 9th-order diffracted light Id9(λ2).

[0250] When the error angle Δθd is −0.5°, that is, when the centers of the point spread functions Iea and IeaL shown in FIG. 50 are +0.0175 on the object plane side numerical aperture NAo, the light intensity Ie of the 9th order diffracted light Id9(λ1) at wavelength λ1 (343.333 nm) is approximately 0.869, the light intensity Ie of the 9th order diffracted light Id9(λ2) at wavelength λ2 (355.000 nm) is approximately 0.507, and the light intensity Ie of the 8th order diffracted light Id8(λ2) at wavelength λ2 (355.000 nm) is approximately 0.619.

[0251] Therefore, the illumination light ILm is configured to include a third wavelength λ3, which is approximately 9 to 12 nm shorter than the wavelength λ1 = 343.333 nm with a small telecentric error Δθt. As an example, let's assume that the wavelength λ3 is 333.6 nm, which is approximately 9.7 nm shorter than the wavelength λ1. In this case, if the point spread function at wavelength λ3 is IeaH, then at an error angle Δθd = -0.5°, the point spread function IeaH will also shift to a position of +0.0175 on the object-side numerical aperture NAo, just like the point spread functions Iea and IeaL.

[0252] When the wavelength λ3 is 333.6 nm, the diffraction angle of the 9th-order diffracted light Id9(λ3) of the 0th-order light equivalent component from the DMD10 (arrangement pitch of micromirrors Ms: 5.4 μm, angle of incidence θα: 35.0°) is converted into the object plane side numerical aperture NAo to approximately +0.0176, the diffraction angle of the 10th-order diffracted light Id10(λ3) at wavelength λ3=333.6 nm is converted into the object plane side numerical aperture NAo to approximately −0.0442, and the diffraction angle of the 8th-order diffracted light Id8(λ3) at wavelength λ3=333.6 nm is converted into the object plane side numerical aperture NAo to approximately 0.0794 (greater than the maximum value of 0.05), and the 9th-order diffracted light Id9(3) and the 10th-order diffracted light Id10(λ3) are incident on the projection unit PLU.

[0253] The central ray of the 9th-order diffracted light Id9(λ3) at wavelength λ3 appears at a position of approximately +0.0176 on the object plane side numerical aperture NAo, but when the error angle Δθd of the on-state micromirror Msa becomes −0.5°, the center of the point spread function IeaH shifts to a position of +0.0175 on the object plane side numerical aperture NAo, and the light intensity Ie of the 9th-order diffracted light Id9(λ3) increases to approximately 1.000 (100%).

[0254] As described above, by including in the illumination light ILm three wavelength components—the wavelength λ1 at which the telecentric error Δθt is smallest in the initial state, the wavelength λ2 longer than λ1, and the wavelength λ3 shorter than λ1—it is possible to mitigate the reduction in exposure dose regardless of whether the error angle Δθd of the on-state micromirror Msa changes to a positive or negative value. Note that, although this will vary depending on the array pitch Pdx of the micromirrors Ms, the design tilt angle θo of the on-state micromirror Msa, and the incident angle θα of the illumination light ILm, in order to compensate for the increase or decrease in exposure dose for each wavelength, it is advisable to set the difference between the wavelengths λ1 and λ2 (λ1<λ2) and the difference between the wavelengths λ1 and λ3 (λ1>λ3) to a range of approximately 8 nm to 13 nm, or to set the wavelengths λ2 and λ3 to within ±4% of the central wavelength λ1 at which the telecentric error Δθt is smallest. Furthermore, it is desirable that the projection unit PLU be chromatically corrected within the bandwidth of the wavelengths λ1, λ2, and λ3 that are used.

[0255] Figure 51 is a graph simulating the change in light intensity according to the error angle Δθd of the 9th-order diffracted light Id9(λ1), Id9(λ2), and Id9(λ3), which are components equivalent to the 0th-order light generated under each wavelength, assuming that each of the three wavelengths λ1 (343.333 nm), λ2 (355.0 nm), and λ3 (333.6 nm) is contained in the illumination light ILm at the same light intensity.

[0256] 51, the horizontal axis represents the error angle Δθd, and the vertical axis represents the relative light intensity Ie, where the maximum light intensity of each of the ninth-order diffracted lights Id9(λ1), Id9(λ2), and Id9(λ3) is set to 1.0 (100%). In the simulation, as in the previous Figure 50, the array pitch Pdx of the micromirrors Ms was set to 5.4 μm, the effective dimension Lms' of the micromirrors Msa was set to 4.38 μm, the angle of incidence θα of the illumination light ILm (including wavelengths λ1, λ2, and λ3) was set to 35.0°, and the initial (design) tilt angle θo of the micromirror Msa in the on state was set to 17.5°.

[0257] The change characteristics of the light intensity Ie of each of the ninth-order diffracted light beams Id9(λ1), Id9(λ2), and Id9(λ3) are expressed as sinc ２The condition under which the ninth-order diffracted light Id9(λ1) is maximized is when the error angle Δθd is approximately +0.04° (to be precise, +0.0389°), the condition under which the ninth-order diffracted light Id9(λ2) is maximized is when the error angle Δθd is approximately −0.52° (to be precise, −0.518°), and the condition under which the ninth-order diffracted light Id9(λ3) is maximized is when the error angle Δθd is approximately +0.50° (to be precise, +0.504°).

[0258] The error angle Δθd of approximately +0.04° at which the ninth-order diffracted light Id9(λ1) is maximized corresponds to the object-side numerical aperture NAo of approximately 0.00135 of the ninth-order diffracted light Id9(λ1) shown in Figures 47, 49, and 50, and the error angle Δθd of approximately −0.52° at which the ninth-order diffracted light Id9(λ2) is maximized corresponds to the object-side numerical aperture NAo of approximately −0.0181 of the ninth-order diffracted light Id9(λ2) shown in Figures 47, 49, and 50. Similarly, the error angle Δθd of approximately +0.50° at which the ninth-order diffracted light Id9(λ3) is maximized corresponds to the object-side numerical aperture NAo of approximately +0.0176 of the ninth-order diffracted light Id9(λ3).

[0259] Although not shown in the graph of FIG. 51, the light intensities of the eighth-order diffracted light Id8 and the tenth-order diffracted light Id10 appearing before and after the ninth-order diffracted light Id9 of the zeroth-order light equivalent component also change in sinc form according to the change in the error angle Δθd. ２ Therefore, as the absolute value of the error angle Δθd increases, the light intensity of the eighth-order diffracted light Id8 or the tenth-order diffracted light Id10 may become greater than the light intensity of the ninth-order diffracted light Id9.

[0260] [Variation 7] The error angle Δθd included in the tilt angle θd of each of the many micromirrors Ms of the DMD 10 tends to gradually change to either the positive or negative side over time. The error angle Δθd can be determined, for example, by measuring the telecentric error Δθt for each wavelength (λ1, λ2, λ3) of the point image projected by the projection unit PLU when a single micromirror Ms of the DMD 10 is turned on. By performing similar measurements on each of several micromirrors Msa that are turned on individually, the average value of the error angle Δθd can be obtained.

[0261] As a result, based on the magnitude and direction of change in the positive or negative of the measured error angle Δθd, and on the characteristic data of the point spread functions Iea, IeaL (and IeaH) as shown in Figure 50, it is possible to estimate the light intensity Ie of the zeroth-order light equivalent component (ninth-order diffracted light Id9) for each wavelength, as well as the total light intensity (exposure amount) obtained by adding them all together. If the estimated total light intensity (exposure amount) is equal to or greater than a predetermined tolerance (e.g., 90%), normal exposure operation can be continued. If it is below the tolerance, in order to prevent insufficient exposure amount, the illuminance of the beams from all laser light sources can be increased uniformly, or each laser light source can be adjusted so that the illuminance balance of the beams of wavelengths λ1, λ2, and λ3 changes.

[0262] Furthermore, the illumination light ILm may be broadband illumination light having a broad, continuous spectrum across a wavelength bandwidth of λ1±Δλ (Δλ≦λ1×4%) with wavelength λ1 as the center wavelength, or may be multi-wavelength illumination light having four or more isolated narrow-band spectra (e.g., wavelength width of 1 nm or less) discretely distributed within that wavelength bandwidth.

[0263] [Variation 8] Furthermore, when a laser light source with a narrowed spectrum is used, illumination light ILm having only two components, wavelength λ2 (355.0 nm) and wavelength λ3 (333.6 nm), as described in the above embodiments and variations may be used. As described above, when the arrangement pitch of micromirrors Ms is 5.4 μm and the angle of incidence θα of illumination light ILm is 35.0°, the central ray of ninth-order diffracted light Id9 (λ2) generated under illumination light of wavelength λ2 (355.0 nm) appears as a telecentric error at a position of approximately −0.0181 on the object-side numerical aperture NAo, and the light intensity Ie when the error angle Δθd is zero is approximately 0.848 (see FIG. 49 ).

[0264] On the other hand, when the array pitch of micromirrors Ms is 5.4 μm and the angle of incidence θα of illumination light ILm is 35.0°, the central ray of 9th-order diffracted light Id9(λ3) generated under illumination light of wavelength λ3 (333.6 nm) appears as a telecentric error at a position of approximately +0.0176 on the object-side numerical aperture NAo, according to equation (3), and the light intensity Ie when the error angle Δθd is zero is approximately 0.837. Therefore, when the error angle Δθd of micromirror Msa in the on state is zero, the sum of the light intensity of 9th-order diffracted light Id9(λ2) of wavelength λ2 (355.0 nm) and the light intensity of 9th-order diffracted light Id9(λ3) of wavelength λ3 (333.6 nm) is 1.685.

[0265] From that state, when the error angle Δθd is changed to +0.5°, the light intensity Ie of the 9th-order diffracted light Id9 (λ2) of wavelength λ2 (355.0 nm) increases from 0.848, as shown in Figure 50. Meanwhile, the light intensity Ie of the 9th-order diffracted light Id9 (λ3) of wavelength λ3 (333.6 nm) decreases from 0.837. Conversely, when the error angle Δθd is changed to -0.5°, the light intensity Ie of the 9th-order diffracted light Id9 (λ2) of wavelength λ2 (355.0 nm) decreases from 0.848, and the light intensity Ie of the 9th-order diffracted light Id9 (λ3) of wavelength λ3 (333.6 nm) increases from 0.837.

[0266] In this way, by setting the wavelength difference so that the 9th-order diffracted light Id9(λ2) and Id9(λ3) generated at the two wavelengths λ2 and λ3 have approximately equal telecentric errors in opposite directions, the light intensity of the 9th-order diffracted light Id9(λ2) and the light intensity of the 9th-order diffracted light Id9(λ3) can be changed complementarily in response to changes in the error angle Δθd of the micromirror Msa.

[0267] Furthermore, when the tilt angle θd of the micromirror Msa in the on state of the DMD 10 is in the initial state (error angle Δθd = 0) where it is equal to the design tilt angle θo, one wavelength λ1 (e.g., 343.333 nm) of the illumination light ILm is set so that the telecentric error Δθt of the high-order diffracted light Idj (e.g., j = 9th order), which is the component equivalent to the zeroth order light, is minimized. In addition, light of wavelength λ2 (λ2 > λ1) (e.g., λ2 = 355 nm) and light of wavelength λ3 (λ3 < λ1) (e.g., λ3 = 333.6 nm) are included in the illumination light ILm at appropriate intensities.

[0268] If the error angle Δθd of DMD 10 tends to become larger than a predetermined tolerance (for example, ±0.3°) and the error angle Δd tends to increase in a positive direction, the intensity of the light of wavelength λ2 (λ2 > λ1) contained in illumination light ILm is increased and the intensity of the light of wavelength λ3 (λ3 < λ1) is decreased. Conversely, if the error angle Δθd tends to become larger than the tolerance (for example, ±0.3°) and the error angle Δd tends to increase in a negative direction, the intensity of the light of wavelength λ3 contained in illumination light ILm may be increased and the intensity of the light of wavelength λ2 may be decreased.

[0269] 40, 42, and 43, when an illumination unit ILU is used in which the angles of incidence of two illumination lights ILm1 and ILm2 with different wavelengths on DMD 10 can be individually adjusted, the angle of incidence of only illumination light ILm2 with wavelength λ2 illuminating DMD 10 is adjusted so that the telecentric error of ninth-order diffracted light Id9(λ2) at wavelength λ2 (355.00 nm) shown in Fig. 50 is corrected. As a result, the ninth-order diffracted light Id9(λ2) shifts in the positive direction on the object plane-side numerical aperture NAo, and only the point spread function IeaL generated at the error angle Δθd (for example, +0.5°) of the micromirror Msa in the on state shifts in the positive direction by the same amount on the object plane-side numerical aperture NAo.

[0270] Therefore, by providing the configurations shown in Figures 40, 42, and 43, not only can the telecentric error Δθt caused by differences in the wavelength of the illumination light ILm be corrected, but a synergistic effect can also be achieved in that the reduction in light intensity Ie (exposure amount) caused by the error angle Δθd of the micromirror Msa in the on state can also be mitigated.

[0271] As described above, according to this embodiment and each of the modified examples, by using multiple wavelengths for the illumination light ILm or broadening the wavelength distribution, it is possible to mitigate the significant reduction in the light intensity of the higher-order diffracted light (for example, the ninth-order diffracted light Id9), which is the zeroth-order light-equivalent component generated from the numerous on-state micromirrors Msa of the DMD 10, due to the occurrence of the error angle (tilt error) Δθd of the micromirrors Ms, and to ensure a good exposure amount. Therefore, even if the error angle (tilt error) Δθd gradually increases from its initial value as the operating time of the exposure apparatus passes, it is possible to continue precise pattern exposure with a stable exposure amount.

[0272] 10...DMD (Digital Mirror Device), 10M...mount section, 104...input lens system, 108...optical integrator, 108A, 108A1, 108A2...MFE (micro fly's eye) lens, 108B...aperture stop, 110...condenser lens system, 112...tilted mirror, 116...first lens group, 118...second lens group, AXa, AXb, AXc...optical axis, DBS...dichroic beam splitter, DCM...dichroic mirror, EL...lens element, Ep...pupil, EX...pattern exposure device, FBU...optical fiber unit, FBn...optical fiber bundle, G1 to G5...lens group, HV1, HV2...parallel plate, IA1 to IA27...projection area, Idj...jth-order diffracted light, Id0 to I d10...0th to 10th order diffracted light, Iea...point image intensity distribution of rectangular aperture, ILm, ILm1, ILm2...illumination light, ILU...illumination unit, Ips...light source image, Ms...micromirror, Msa...micromirror in ON state, Msb...micromirror in OFF state, MU(A) to MU(C)...exposure module, NAi...numerical aperture on the image plane side, NAo...numerical aperture on the object plane side, NAx'...numerical aperture in the X' direction of diffracted light, NAy'...numerical aperture in the Y' direction of diffracted light, NAxf...numerical aperture in the X' direction of the entire imaging light beam, P...substrate P, Pdx, Pdy...arrangement pitch of micromirrors, PLU...projection unit, SPF...point light source, θα...angle of incidence, θd...tilt angle, θj...diffraction angle of jth order light, Δθt...telecentric error on the image plane side, λo...central wavelength, Δλ...wavelength width

Claims

1. 1. An exposure apparatus which projects and exposes a pattern corresponding to the drawing data onto a substrate by irradiating illumination light onto a spatial light modulation element having a number of micromirrors which are two-dimensionally arranged at a predetermined pitch and which are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to enter a projection unit, an illumination unit that irradiates the spatial light modulation element with a first illumination light having a wavelength λ1 that is allowable based on a chromatic aberration characteristic of the projection unit and a second illumination light having a wavelength λ2 (λ2≠λ1) that is allowable based on a chromatic aberration characteristic of the projection unit at an incident angle that corresponds to a double angle of a tilt angle of the micromirror in the on state; an exposure apparatus in which, when the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in the on state under the wavelength λ1 and enters the projection unit is θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in the on state under the wavelength λ2 and enters the projection unit is θj2, the difference between the wavelength λ1 and the wavelength λ2, or the angle of incidence is set so that the main diffracted light with the diffraction angle θj1 and the main diffracted light with the diffraction angle θj2 are distributed on either side of the optical axis of the projection unit.

2. 2. The exposure apparatus according to claim 1, an adjustment mechanism that adjusts the angle of incidence of at least one of the first illumination light and the second illumination light so that a main diffracted light having the diffraction angle θj1 and a main diffracted light having the diffraction angle θj2 are distributed symmetrically with respect to the optical axis of the projection unit.

3. 3. The exposure apparatus according to claim 2, When the pitch of the array of the micromirrors is Pd and the incident angle is θα, the diffraction angles θj1 and θj2 are respectively expressed as follows: sinθj1=sinθα−j1(λ1 / Pd), sinθj2=sinθα−j2(λ2 / Pd), An exposure device that is set in accordance with the relationship between the

4. 3. The exposure apparatus according to claim 2, The lighting unit includes: an optical integrator into which the first illumination light and the second illumination light are both incident and which forms a surface light source on an exit surface side; and a condenser lens system whose optical axis is set to be inclined by the incidence angle θα with respect to the optical axis of the projection unit PLU, and which performs Koehler illumination of the spatial light modulation element light by the surface light source on the exit surface side of the optical integrator.

5. 5. The exposure apparatus according to claim 4, The lighting unit includes: an optical fiber bundle having a single or multiple fibers into which the first illumination light and the second illumination light are both incident, and an input lens system that provides Kohler illumination or critical illumination to an incident surface of the optical integrator with the first illumination light and the second illumination light projected from an exit end of the optical fiber bundle.

6. 6. The exposure apparatus according to claim 5, The adjustment mechanism includes: an exposure apparatus including any one of a mechanism for relatively adjusting the positions of the exit end of the optical fiber bundle and the input lens system within a plane perpendicular to the optical axis of the input lens system; a mechanism for adjusting the inclination of the first illumination light and the second illumination light projected onto the incident surface of the optical integrator; and a mechanism for relatively adjusting the positions of the surface light source formed on the exit surface of the optical integrator and the condenser lens system within a plane perpendicular to the optical axis of the condenser lens system.

7. 5. The exposure apparatus according to claim 4, The lighting unit includes: a dichroic optical member having a wavelength selection characteristic of transmitting one of the first illumination light and the second illumination light and reflecting the other depending on a difference between the wavelength λ1 and the wavelength λ2; an input lens system that provides Kohler illumination or critical illumination to an incident surface of the optical integrator with the first illumination light and the second illumination light combined via the dichroic optical member.

8. 8. The exposure apparatus according to claim 7, The lighting unit includes: a first optical fiber bundle that outputs the incident first illumination light toward the dichroic optical member, and a second optical fiber bundle that outputs the incident second illumination light toward the dichroic optical member, The adjustment mechanism includes: a mechanism for individually displacing each of the first illumination light from the exit end of the first optical fiber bundle and the second illumination light from the exit end of the second optical fiber bundle with respect to the optical axis of the input lens system within a plane perpendicular to the optical axis.

9. 5. The exposure apparatus according to claim 4, The adjustment mechanism includes: an exposure apparatus, which adjusts a first surface light source formed in a circular shape on the exit surface side of the optical integrator by the first illumination light and a second surface light source formed in a circular shape on the exit surface side of the optical integrator by the second illumination light so that they are shifted at a predetermined interval in a direction corresponding to the generation direction of the diffraction angles θj1 and θj2.

10. 10. The exposure apparatus according to claim 9, The adjustment mechanism includes: an exposure apparatus, wherein the amount of shift is set so that the overall shape of the combination of the first surface light source and the second surface light source formed by shifting toward the exit surface of the optical integrator becomes an ellipse whose ratio of the length in the major axis direction to the length in the minor axis direction corresponds to the cosine of the incident angle θα.

11. 1. An exposure apparatus which projects and exposes a pattern corresponding to the drawing data onto a substrate by irradiating illumination light onto a spatial light modulation element having a number of micromirrors which are two-dimensionally arranged at a predetermined pitch and which are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to enter a projection unit, an illumination unit that irradiates the spatial light modulation element with a first illumination light having a wavelength λ1 and a second illumination light having a wavelength λ2 (λ2≠λ1) at an incident angle corresponding to a double angle of a tilt angle of the micromirror in the on state; an exposure apparatus in which the difference between the wavelength λ1 and the wavelength λ2 is set so that the difference between the diffraction angle θj1 and the diffraction angle θj2 falls within a predetermined tolerance range, where θj1 is the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in the on state under the wavelength λ1 and reaches the substrate via the projection unit, and θj2 is the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in the on state under the wavelength λ2 and reaches the substrate via the projection unit.

12. 12. An exposure apparatus according to claim 11, wherein the wavelengths λ1 and λ2 are selected so that the inclination angle of the micromirror is set between the diffraction angle θj1 generated by the micromirror in the on-state at the wavelength λ1 and the diffraction angle θj2 generated by the micromirror in the on-state at the wavelength λ2.

13. 13. The exposure apparatus according to claim 11 or 12, an exposure apparatus, wherein, when the maximum numerical aperture of the projection unit on the spatial light modulation element side is NAo(max), the allowable range of the difference angle is set to 1 / 5 or less of the angle corresponding to the numerical aperture NAo(max).

14. 14. The exposure apparatus according to claim 13, an exposure apparatus in which the allowable range of the difference angle is further set to 1 / 8 or less of the angle corresponding to the numerical aperture NAo(max).

15. 13. The exposure apparatus according to claim 11 or 12, When the pitch of the array of the micromirrors is Pd and the incident angle is θα, the diffraction angles θj1 and θj2 are respectively expressed as follows: sinθj1=sinθα−j1(λ1 / Pd), sinθj2=sinθα−j2(λ2 / Pd), An exposure device that is set in accordance with the relationship between the

16. 16. The exposure apparatus according to claim 15, an exposure apparatus, wherein a difference between the wavelength λ1 and the wavelength λ2 is set so that the order j1 of the main diffracted light generated by irradiation with the first illumination light and the order j2 of the main diffracted light generated by irradiation with the second illumination light are the same order.

17. 16. The exposure apparatus according to claim 15, an exposure apparatus, wherein a difference between the wavelengths λ1 and λ2 is set so that the order j1 of the main diffracted light generated by irradiation with the first illumination light and the order j2 of the main diffracted light generated by irradiation with the second illumination light are different orders.

18. 16. The exposure apparatus according to claim 15, The lighting unit includes: an optical integrator into which the first illumination light and the second illumination light are both incident and which forms a surface light source on an exit surface side; and a condenser lens system whose optical axis is set to be inclined by the incidence angle θα with respect to an optical axis of the projection unit, and which performs Koehler illumination of the spatial light modulation element light by the surface light source on the exit surface side of the optical integrator.

19. 20. The exposure apparatus according to claim 18, The lighting unit includes: an optical fiber bundle having a single or multiple fibers into which the first illumination light and the second illumination light are both incident, and an input lens system that provides Kohler illumination or critical illumination to an incident surface of the optical integrator with the first illumination light and the second illumination light projected from an exit end of the optical fiber bundle.

20. 20. The exposure apparatus according to claim 18, The lighting unit includes: a dichroic optical member having a wavelength selection characteristic of transmitting one of the first illumination light and the second illumination light and reflecting the other depending on a difference between the wavelength λ1 and the wavelength λ2; an input lens system that provides Kohler illumination or critical illumination to an incident surface of the optical integrator with the first illumination light and the second illumination light combined via the dichroic optical member.

21. 1. An exposure apparatus which projects and exposes a pattern corresponding to the drawing data onto a substrate by irradiating illumination light onto a spatial light modulation element having a number of micromirrors which are two-dimensionally arranged at a predetermined pitch and which are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to enter a projection unit, an illumination unit that irradiates the spatial light modulation element with a first illumination light having a wavelength λ1 that is allowable based on a chromatic aberration characteristic of the projection unit and a second illumination light having a wavelength λ2 (λ2≠λ1) that is allowable based on a chromatic aberration characteristic of the projection unit at a design incident angle θα that is set to be double the standard tilt angle of the micromirror in the on state; an exposure apparatus in which the wavelengths λ1 and λ2 are set so that, when the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in the on state and enters the projection unit under the wavelength λ1 is θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in the on state and enters the projection unit under the wavelength λ2 is θj2, the main diffracted light with the diffraction angle θj1 and the main diffracted light with the diffraction angle θj2 that are generated under the condition of the designed incidence angle θα are distributed on one side of the optical axis of the projection unit.

22. 22. The exposure apparatus according to claim 21, When the pitch of the array of the micromirrors is Pd, the diffraction angles θj1 and θj2 are respectively expressed as follows: sinθj1=sinθα−j1(λ1 / Pd), sinθj2=sinθα−j2(λ2 / Pd), An exposure device that is set in accordance with the relationship between the

23. 23. The exposure apparatus according to claim 22, When the designed incident angle θα is θα>0° and the orders j1 and j2 are orders greater than 0, the wavelengths λ1 and λ2 are A first condition of λ1<Pd·sin θα / j1 and λ2<Pd·sin θα / j2, or A second condition of λ1>Pd sin θα / j1 and λ2>Pd sin θα / j2; The exposure apparatus is configured to satisfy any one of the above.

24. 24. The exposure apparatus according to claim 23, an exposure apparatus which satisfies either the first condition or the second condition, and sets a difference between the wavelength λ1 and the wavelength λ2 such that the order j1 and the order j2 are the same order.

25. 24. The exposure apparatus according to claim 23, an exposure apparatus which satisfies either the first condition or the second condition, and sets a difference between the wavelength λ1 and the wavelength λ2 such that the order j1 and the order j2 are different orders.

26. 26. The exposure apparatus according to claim 25, an exposure apparatus in which a difference between the wavelength λ1 and the wavelength λ2 is set so that the order j1 and the order j2 have a relationship of j1=j2-1 or j1=j2+1;

27. The exposure apparatus according to any one of claims 22 to 26, The lighting unit includes: an adjustment mechanism that changes the angle of incidence of at least one of the first illumination light and the second illumination light from the designed angle of incidence θα so that the main diffracted light with the diffraction angle θj1 and the main diffracted light with the diffraction angle θj2 are distributed symmetrically with respect to the optical axis of the projection unit.

28. The exposure apparatus according to any one of claims 22 to 26, an exposure apparatus in which a difference between the wavelength λ1 and the wavelength λ2 is set so that, when the maximum numerical aperture on the spatial light modulation element side of the projection unit is NAo(max), a difference angle Δθj(1-2) between the diffraction angle θj1 and the diffraction angle θj2 is 1 / 5 or less of an angle corresponding to the numerical aperture NAo(max).

29. 29. The exposure apparatus of claim 28, an exposure apparatus in which the difference between the wavelengths λ1 and λ2 is set so that the difference angle Δθj(1-2) is equal to or smaller than 1 / 8 of the angle corresponding to the numerical aperture NAo(max);

30. forming a photosensitive layer on a substrate on which the electronic device is to be fabricated; preparing drawing data corresponding to a pattern for the electronic device; placing the substrate on which the photosensitive layer is formed on a moving stage of an exposure apparatus according to any one of claims 1, 11 and 21, and setting the drawing data in a drive control unit of the spatial light modulation element of the exposure apparatus; a step of exposing the pattern onto the photosensitive layer of the substrate by synchronizing the movement of the substrate by the moving stage with the driving of the micromirrors of the spatial light modulation element in an on-state and an off-state based on the drawing data; A device manufacturing method comprising:

31. 1. An exposure apparatus which projects and exposes a pattern corresponding to the drawing data onto a substrate by irradiating illumination light onto a spatial light modulation element having a number of micromirrors which are two-dimensionally arranged at a predetermined pitch and which are selectively driven based on drawing data, and causes reflected light from selected micromirrors in an on-state of the spatial light modulation element to enter a projection unit, an illumination unit that irradiates the spatial light modulation element with illumination light having a predetermined wavelength width ±Δλ with respect to a central wavelength λo at an incident angle θα (θα>0°) corresponding to a double angle of the tilt angle of the micromirror in the on state; an exposure apparatus in which the wavelength width ±Δλ is set such that, when the wavelength λo+Δλ on the long-wavelength side of the illumination light is defined as wavelength λ1, the wavelength λo-Δλ on the short-wavelength side of the illumination light is defined as wavelength λ2, the diffraction angle of the main diffracted light of order j1 that is generated from the micromirror in the on-state under light of wavelength λ1 and enters the projection unit is defined as θj1, and the diffraction angle of the main diffracted light of order j2 that is generated from the micromirror in the on-state under light of wavelength λ2 and enters the projection unit is θj2, the overall distribution shape of the main diffracted light of order j1 and the main diffracted light of order j2 that appear in the pupil of the projection unit is deformed into an isotropic shape by the difference between the diffraction angle θj1 and the diffraction angle θj2.

32. 32. The exposure apparatus of claim 31 , The lighting unit includes: an exposure apparatus having a condenser lens system that receives a first beam having the wavelength λ1 emitted from a first light source device and a second beam having the wavelength λ2 emitted from a second light source device, and coaxially combines the first beam and the second beam to obliquely illuminate the spatial light modulation element at an incident angle θα.

33. 33. The exposure apparatus of claim 32, an exposure apparatus, wherein the micromirror in the on state is set so as to be tilted at a design tilt angle θo with respect to a neutral plane perpendicular to the optical axis of the projection unit, and the incident angle θα is set to satisfy the relationship θα=2θo.

34. 33. The exposure apparatus of claim 32, a first distribution shape of the main diffracted light of the order j1 generated from the spatial light modulation element at a pupil of the projection unit by irradiation with the illumination light of the first beam becomes an ellipse that shrinks in a direction in which the micromirror is tilted, a second distribution shape of the main diffracted light of order j2 generated from the spatial light modulation element at a pupil of the projection unit by irradiation with the illumination light of the second beam becomes an ellipse that shrinks in a direction in which the micromirror is tilted, an exposure apparatus, wherein the first distribution shape and the second distribution shape are formed by being shifted in a tilt direction of the micromirror by a difference between the diffraction angle θj1 and the diffraction angle θj2 within the pupil.

35. 32. The exposure apparatus of claim 31 , The lighting unit includes: an exposure apparatus having a condenser lens system which receives a first beam having the wavelength λ1 emitted from a first light source device and a second beam having the wavelength λ2 emitted from a second light source device, decenters the first beam and the second beam, and synthesizes the combined illumination light, and obliquely illuminates the spatial light modulation element at a predetermined angle of incidence.

36. 36. The exposure apparatus of claim 35, The lighting unit includes: an optical member that sets the angle of incidence of illumination light by the first beam on the spatial light modulation element to a first angle of incidence θα1 and sets the angle of incidence of illumination light by the second beam on the spatial light modulation element to a second angle of incidence θα2; an exposure apparatus that sets a difference between the incident angle θα1 and the incident angle θα2 in accordance with a difference between the wavelength λ1 and the wavelength λ2.

37. 32. The exposure apparatus of claim 31 , An exposure apparatus in which the illumination light projected from the illumination unit onto the spatial light modulation element is converted into multispectral light in which a plurality of single narrow wavelength width spectra are discretely arranged across the wavelength width ±Δλ.

38. 32. The exposure apparatus of claim 31 , an exposure apparatus in which the illumination light projected from the illumination unit onto the spatial light modulation element is broadband illumination light having a broad and continuous spectrum over the wavelength width ±Δλ;