PROJECTION EXPOSURE METHODS AND SYSTEMS

Abstract

Projection exposure methods, systems, sub-systems and components are disclosed. Methods can include performing a first exposure to image a first sub-pattern of the pattern, where the first sub-pattern includes a plurality of first features extending in a first direction and spaced apart essentially periodically at a predominant periodicity length P in a second direction perpendicular to the first direction. The first exposure can be performed using a multipolar illumination mode that includes at least one substantially dipolar intensity distribution having two illumination poles positioned on a pole orientation axis substantially parallel to the second direction and spaced apart from each other.

Claims

1.-25. (canceled)

26. An illumination system, comprising: a plurality of optical elements constructed and arranged to receive primary radiation emitted from a light source and to generate illumination radiation incident on a mask bearing a pattern, a pupil shaping unit configured to generate a defined multipolar intensity distribution corresponding to a selected illumination mode in a pupil surface of the illumination system during use, wherein: the multipolar intensity distribution comprises a quadrupolar intensity distribution having four illumination poles positioned on two pole orientation axes and spaced apart from one another; and the poles of the quadrupolar intensity distribution each have a lenticular pole shape defined by an overlapping zone of two circles with equal unit radius having a distance between the centers of the circles.

27. The illumination system according to claim 26, wherein the poles of the quadrupolar intensity distribution each have a pole dimension ratio PDR=PH/PW between a pole height PH and a pole width PW according to
PDR.sub.OPT20%PDRPDR.sub.OPT+20% where PDR.sub.OPT=sin ()/(1cos()), is a half pole angle, the pole width PW is measured between an inner pole edge and an outer pole edge in a direction parallel to the pole orientation axis, and the pole height is measured between pole edges in a direction perpendicular to the pole orientation axis.

28. The illumination system of claim 26, wherein the pupil-shaping unit comprises an array of individual elements.

29. The illumination system of claim 28, wherein the individual elements comprise diffraction gratings.

30. The illumination system of claim 28, wherein the individual elements comprise mirrors.

31. The illumination system of claim 26, wherein the pupil-shaping unit comprises an optical modulation device configured to controllably change an angular distribution of radiation incident on the optical modulation device such that a desired intensity distribution is obtained in the pupil surface of the illumination system.

32. The illumination system of claim 31, wherein the optical modulation device has an array of individual elements configured to be driven individually to change the angular distribution of radiation incident on the individual elements.

33. The illumination system of claim 32, wherein the individual elements comprise mirrors.

34. The illumination system of claim 32, wherein the individual elements comprise diffractive optical elements or acousto-optical elements.

35. The illumination system according to claim 26, further comprising a blocking device arranged at or close to the pupil surface of the illumination system where the basic intensity distribution is generated, or at or close to a pupil surface optically conjugate thereto; and the blocking device is designed to block areas of the basic intensity distribution which lie outside the region desired for the pole shape.

36. The illumination system according to claim 26, wherein the pupil shaping unit further comprises a zoom lens group configured to provide a continuously variable magnification during use.

37. The illumination system according to claim 26, wherein the pupil shaping unit further comprises a pair of axicon elements having axicon surfaces, one concave and one convex, wherein at least one of the axicon elements is movable along an optical axis of the illumination system to allow adjusting a distance between the axicon surfaces.

38. An apparatus, comprising: an illumination system configured so that during use the illumination system guides radiation along an illumination path to generate illumination radiation incident on a mask bearing a pattern; and a projection objective configured so that during use the projection objective projects an image of the pattern onto a radiation-sensitive substrate with projection radiation guided along a projection path, wherein the apparatus is a projection exposure apparatus, and the illumination system is configured according to claim 26.

39. An illumination system, comprising: a plurality of optical elements constructed and arranged to receive primary radiation emitted from a light source and to generate illumination radiation incident on a mask bearing a pattern, a pupil shaping unit configured to generate a defined multipolar intensity distribution corresponding to a selected illumination mode in a pupil surface of the illumination system during use, wherein: the pattern comprises features spaced apart periodically at a predominant periodicity length P where the condition 0.7/NA<P</NA holds; the multipolar intensity distribution comprises a dipolar intensity distribution having two illumination poles positioned on a pole orientation axis and spaced apart from each other at a distance d</(P*NA), where is a nominal operation wavelength of radiation used, and NA is an image-side numerical aperture of a projection optical system used; and the poles of the dipolar intensity distribution each have a pole shape selected from the group consisting of: a generally biconvex pole shape with a pole dimension ratio PDR=PH/PW>1 between a pole height PH and a pole width PW, where the pole width PW is measured between an inner edge and an outer pole edge in a direction parallel to the pole orientation axis, and the pole height PH is measured between inner and outer pole edges in a direction perpendicular to the pole orientation axis; and a substantially lenticular pole shape defined by two convexly curved edges having opposite sense of curvature and centres of curvature essentially on a line parallel to the pole orientation axis; and the poles of the dipolar intensity distribution each have a pole dimension ratio PDR=PH/PW according to
PDR.sub.OPT20%PDRPDR.sub.OPT+20% where PDR.sub.OPT=sin ()/(1cos()), is a half pole angle, PW is measured between an inner pole edge and an outer pole edge in a direction parallel to the pole orientation axis, and PH is measured between pole edges in a direction perpendicular to the pole orientation axis.

40. The illumination system of claim 39, wherein the pupil-shaping unit comprises an array of individual elements.

41. The illumination system of claim 40, wherein the individual elements comprise diffraction gratings.

42. The illumination system of claim 40, wherein the individual elements comprise mirrors.

43. The illumination system of claim 40, wherein the individual elements are configured to be controlled individually in order to change an angular distribution of radiation influenced by the individual elements.

44. The illumination system of claim 40, wherein during use: the array of individual elements is configured to generate a basic intensity distribution including areas of the poles of the dipolar intensity distribution and extending beyond the areas of the poles; the illumination system further comprises a blocking device arranged at or close to the pupil surface of the illumination system where the basic intensity distribution is generated, or at or close to a pupil surface optically conjugate thereto; and the blocking device is designed to block areas of the basic intensity distribution which lie outside the region desired for the pole shape.

45. The illumination system of claim 39, wherein the pupil shaping unit further comprises a zoom lens group configured to provide a continuously variable magnification during use.

46. The illumination system of claim 45, wherein the zoom lens group is arranged between the array of individual elements and the pupil surface of the illumination system.

47. The illumination system of claim 39, wherein the pupil shaping unit further comprises a pair of axicon elements having axicon surfaces, one concave and one convex, wherein at least one of the axicon elements is movable along an optical axis of the illumination system to allow adjusting a distance between the axicon surfaces.

48. The illumination system of claim 47, wherein the axicon group is arranged between the array of individual elements and the pupil surface of the illumination system.

49. An apparatus, comprising: an illumination system configured so that during use the illumination system guides radiation along an illumination path to generate illumination radiation incident on a mask bearing a pattern; and a projection objective configured so that during use the projection objective projects an image of the pattern onto a radiation-sensitive substrate with projection radiation guided along a projection path, wherein the apparatus is a projection exposure apparatus, and the illumination system is configured according to claim 39.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0019] FIG. 1 illustrates schematically the principle of symmetric off-axis illumination;

[0020] FIGS. 2A-2D show schematically various illumination modes;

[0021] FIG. 3 illustrates schematically a projection exposure system configured for dipole illumination mode with lenticular poles;

[0022] FIG. 4 illustrates schematically a circular pupil of a projection objective at imaging conditions beyond the coherent resolution limit using symmetric two-beam interference imaging conditions;

[0023] FIGS. 5A-5C illustrate where the poles of the dipolar intensity distribution each have a lenticular pole shape;

[0024] FIGS. 6A-6B illustrate schematically the use of a mask to generate biconvex, lenticular illumination poles in a modified conventional illumination system;

[0025] FIGS. 7A-7B show poles of a dipolar intensity distribution having a parallelogram pole shape;

[0026] FIG. 8 shows a meridional section of an illumination system configured to generate large polses in a projection exposure step-and-scan apparatus;

[0027] FIGS. 9A-9B show schematically a detail of a first raster element configured to generate a substantially dipolar intensity distribution having two illumination poles each having a diamond shape as shown in FIG. 7;

[0028] FIGS. 10A-10B show diagrams illustrating improvements on lens heating induced wavefront phase errors obtained by illuminating with large poles; and

[0029] FIG. 11 shows dipole illumination with lenticular poles used for double patterning.

[0030] FIG. 12 shows illumination.

DETAILED DESCRIPTION

[0031] In the following description, similar or identical features shown in different figures are designated with the same reference identifications, where appropriate.

[0032] In optical lithography it is known to use off-axis illumination, which enables smaller features to be faithfully imaged. With this technique, a mask providing the patterned structure (pattern) is illuminated at oblique (non-perpendicular) angles such that process latitude is improved by increasing the depth of focus and/or such that contrast is improved.

[0033] FIG. 1 illustrates this principle in which a beam B of radiation is incident on a mask M carrying a pattern PAT at an opening angle inclined to the optical axis OA, which is conventionally vertical. The incident beam B is diffracted by features of the pattern PAT on the mask M, which are to be imaged onto the wafer W coated by a layer of photoresist RES. The zeroth and two first-order-diffracted beams (0, 1) are shown in FIG. 1. Improved performance can be achieved when, for example, at least part of the zeroth order and one of the first orders, which are coherent, are captured by the projection objective PO and used to form the image on the wafer W.

[0034] The smaller the pitch P (or periodicity length) of features of the pattern on the mask M, the larger the diffraction angle will be according to =/P, where is the nominal operating wavelength of radiation in beam B. If the size of the features decreases and/or their density increases too much, there will come a point at which the pupil of the projection objective PO can no longer capture more than one diffracted order. In practical systems there will be a range of opening angles which determines the partial coherence of the light source and thus is important to the figures of merit of the exposure process, such as exposure latitude, depth of focus and proximity effects.

[0035] The distribution of opening angles can be visualized by considering the intensity distribution of the effective radiation source (effective source) or equivalently the intensity distribution in the plane of a pupil of the projection objective. FIGS. 2A-2D show representative examples.

[0036] FIG. 2A illustrates a simple on-axis illumination mode characterized by the parameter shown by an arrow in the figure. Values of (often referred to as coherence factor) are conventionally quoted as the ratio of the radius of the illumination intensity disc (hatched) to the radius of the pupil and therefore take a value between 0 and 1.

[0037] FIG. 2B shows an annular illumination mode in which the intensity distribution of the source is confined to an annulus to limit the range of angles of incidence of the off-axis illumination, considering that the spatial intensity distribution at the pupil plane is related to the angular distribution at the mask plane, which is a Fourier transform plane to the pupil plane. The annulus may be characterized by the values .sub.i and .sub.o, which are the ratios of its inner and outer radii to the radius of the pupil.

[0038] FIG. 2C illustrates the intensity distribution of the quadrupole illumination mode, the use of which sometimes gives improved imaging results to the use of annular or coherent modes. Conventionally, in using such a quadrupole configuration, it is assumed that the patterned structure of the mask to be projected includes sub-patterns of orthogonal lines along x and y axes and the illumination is oriented such that each of the four poles is situated in a respective one of the four quadrants defined by these x and y axes and at their point of intersection.

[0039] In specific applications it has been found that superior performance can be obtained using dipolar illumination modes. FIG. 2D shows an example of the dipolar illumination intensity distribution for a dipole mode. The two poles POLE1, POLE2 of this mode are located off the optical axis OA of the imaging system. For illustrative purposes, the two poles illustrated in FIG. 2D are said to lie along the x axis (second direction) and may be optimal for imaging lines extending parallel to the y axis (first direction), i.e. perpendicular to the pole orientation axis POA joining the two poles. Sometimes the x and y axis are referred to as horizontal and vertical, respectively, but these terms typically do not bear any relation to the orientation of the machine.

[0040] As illustrated in FIG. 2D the pole shape of the poles POLE1, POLE2 in conventional systems may correspond to an azimuthal section of an annulus. Due to the desire that the first diffraction order should pass completely through the aperture of the projection objective to obtain maximum contrast, limitations result for the radial width =.sub.o.sub.i of the poles and for the azimuthal width (width in circumferential direction) of the poles defined by the pole angle .

[0041] With dipole illumination, the effective light source is confined to two poles, in order to create the conditions for two-beam imaging with theoretical ideal contrast. FIG. 3 illustrates schematically a projection exposure system configured for dipole illumination mode. Primary radiation provided by a primary light source LS is transformed by an illumination system ILL to provide illumination radiation incident on a mask M bearing a pattern PAT. The radiation is diffracted by pattern PAT and forms projection radiation propagating through the projection objective PO, which images the pattern of the mask onto the substrate, which is a wafer W coated with a resist layer RES in the exemplary embodiment. The illumination system ILL includes a pupil shaping unit PSU to create an effective source ES formed by a predefined intensity distribution at a pupil plane P.sub.ILL of the illumination system, where the predefined intensity distribution is a dipole illumination characterized by two lenticular poles POLE1 and POLE2. The pupil plane P.sub.ILL, where the effective source ES is formed, is a Fourier transform surface with respect to the object surface of the projection objective PO, where the mask M is situated. Therefore, the spatial distribution of light intensity of the effective source ES determines an angular distribution of illumination radiation incident on the mask M. A condenser system CS is provided to perform the Fourier transformation.

[0042] The pupil surface P.sub.PO of the projection objective is optically conjugated to the pupil surface P.sub.ILL of the illumination system. Therefore, in the absence of a mask, a spatial intensity distribution equivalent to the intensity distribution at the effective source ES may be formed in the pupil surface P.sub.PO of the projection objective. Where a mask bearing a pattern is inserted between the illumination system and the projection objective, the intensity distribution in the pupil plane P.sub.PO of the projection objective also contains diffraction information corresponding to the properties of the mask pattern PAT.

[0043] FIG. 4 illustrates schematically a circular pupil P.sub.PO of a projection objective having unit radius (length=1) corresponding to the utilized image-side numerical aperture NA at imaging conditions beyond the coherent resolution limit of a periodic pattern of mutually parallel lines running in a first direction (y direction) having periodicity length (pitch) P in a second direction (x direction). A symmetric two-beam interference imaging condition is assumed, where a first diffraction order (+1 or 1) coincides with the zeroth order (0). In this representation, all illumination directions which contribute constructively to two-beam interference correspond to pupil positions lying within the lenticular (lens-shaped) regions LR at opposite edges of the pupil in the x direction. Those lenticular regions (also denoted as planar circular two-angles result from a partial overlap of circles of equal size corresponding to neighbouring diffraction orders. A distance d between the centers of the lenticular regions is determined by

d=/(P*NA).

[0044] The angular width of the lenticular area (width in circumferential direction) may be defined by a pole angle between the opposing corners of the area in the y direction, where pole angle is twice the pole half angle (i.e. =2*). The size of the lenticular area, i.e. the area A.sub.OPT of the hatched lenticular region LR in FIG. 4, may be calculated as

A.sub.OPT=sin()

with

cos()=d/2=/(2*P*NA).

[0045] It is contemplated that a maximum contrast is possible if all illumination directions corresponding to pupil positions within the lenticular regions LR are used.

[0046] The size A.sub.OPT of the lenticular regions may be calculated based on the following considerations explained now in connection with FIG. 4. The area A.sub.OPT is four times the difference between the area A.sub.S of the circular sector with sector angle , and the area A.sub.ABC of triangle ABC indicated in FIG. 4. The difference area A.sub.S-A.sub.ABC is hatched darkly in FIG. 4. Calculating the angles in radians, the sector area A.sub.S is calculated as follows: A.sub.S=(/2).Math..Math.r.sup.2, where r=1 is the unit radius of the pupil. Therefore, A.sub.S=(/2).Math.=/2. With distance AC=cos () and distance BC=sin () the area A.sub.ABC of the triangle calculate as: A.sub.ABC=(sin ().Math.cos ())/2. Further considering that 2.Math.sin ().Math.cos ()=sin (2.Math.), the area A.sub.OPT of the lenticular region LR calculates as follows:

[00001]$\begin{array}{c}{A}_{\mathrm{OPT}}=.Math.4.Math.\left(A_S-A_{\mathrm{ABC}}\right)\\ =.Math.4.Math.\left(/2-\left(\sin \left(\right).Math.\cos \left(\right)\right)/2\right)\\ =.Math.2.Math.-\sin \left(2.Math.\right)\\ =.Math.-\sin \left(\right)\end{array}$

[0047] An illumination pole having the size and shape of a desirable (e.g., optimum) lenticular region LR may be characterized by a pole width PW measured between an inner and an outer pole edge in a direction parallel to the pole orientation axis POA, and a pole height PH measured between pole edges in a direction perpendicular to the pole orientation axis, i.e. in y direction (see FIG. 5A). Defining a pole dimension ratio PDR as the aspect ratio between pole PH height and pole width PW according to PDR=PH/PW, for the lenticular regions PDR.sub.OPT=sin()/(1cos()).

[0048] In conventional systems, only a fraction of those illumination directions is normally used. For illustration purposes, FIG. 4 shows conventional poles CP having a shape corresponding to an azimuthal section of an annulus within the lenticular regions LR. Limitations regarding radial width and pole angle result from the desire that the first diffraction order should fall completely within the aperture of the objective to obtain maximum contrast. As evident from FIG. 4, the area of the conventional poles CP is significantly smaller than the area A.sub.OPT of the lenticular zones which could be used for the imaging process at good (e.g., optimum) contrast. For example, the size of the conventional poles may be 50% or less of the area A.sub.OPT.

[0049] The pole dimension ratio PDR of a conventional pole usually differs significantly from the pole dimension ratio PDR.sub.OPT defined above. For example, a circular pole would correspond to PDR=1. The annular segment shaped conventional poles CP depicted schematically in FIG. 4 may have varying pole dimension ratio depending on the selected values for and pole angle. Generally, as the pole height increases, the pole width will decrease, and vice versa.

[0050] Now consider a lens or another optical element having an optical surface at or close to the pupil surface of the projection objective, where light intensity is concentrated at the edge of the lens surface in regions corresponding to the poles of the dipole illumination. Studies have shown that effects of lens heating may cause characteristic deformations of wavefronts passing the projection objective due to local deformation of the lenses and/or due to localized radiation-induced changes of refractive index. For example, where a dipolar intensity distribution is used, the local deformation of lenses or other optical elements and associated wavefront deformations may be dominated by astigmatic contributions.

[0051] In the following a number of embodiments are described where a multipolar illumination mode includes at least one substantially dipolar intensity distribution having two illumination poles, where the pole area A.sub.POLE of poles is selected to be very close to or equal to the pole area A.sub.OPT defined above. Negative effects caused by localized lens heating may be reduced significantly when compared to using conventional poles.

[0052] FIGS. 5A-5C illustrate where the poles POLE 1 and POLE 2 of the dipolar intensity distribution each have a lenticular pole shape defined by two convexly curved edges having opposite sense of curvature and centers of curvature essentially on a line parallel to the pole orientation axis POA. In FIG. 5A the size and shape of the poles is adjusted such that the poles exactly fit the lenticular region LR corresponding to the given pitch P of the sub-pattern to be printed. The conditions A.sub.POLE=A.sub.OPT and PDR.sub.POLE=PDR.sub.OPT hold for the pole area and pole dimension ratio, respectively. FIG. 5B shows an example where all light intensity is concentrated in poles POLE1, POLE2 slightly smaller than the lenticular area by about 10%. Although not all illumination directions theoretically contributing to constructive interference are used in this setting, the intensity of the poles is still distributed over a relatively large area on optical surfaces near pupil surfaces to eliminate or alleviate lens heating problems. In FIG. 5C the size of the poles POLE1, POLE2 is slightly larger than the pole area A.sub.OPT of poles such that a small fraction of illumination directions falls outside the pupil, thereby not contributing to imaging, and some illumination angles smaller than the lowest limit of the optimum angular distribution according to FIG. 4 are also present. As light energy is distributed over relatively large areas around the poles, lens heating problems may be smallest in FIG. 5C with potential trade-offs in attainable contrast. Optimum compromise settings may be evaluated by simulation or test exposures.

[0053] FIGS. 6A-6B illustrate schematically one option to generate biconvex, lenticular (lens-shaped shaped) illumination poles as described above in a modified conventional illumination system configured to generate illumination poles having the shape of a segment of an annulus. Illumination systems as described e. g. in U.S. Pat. No. 6,452,662 B2 may be used and modified for that purpose. The disclosure of U.S. Pat. No. 6,452,662 B2 and U.S. Pat. No. 6,855,486 B1 regarding constructive details of illumination systems capable of generating segment shaped pairs of poles are incorporated herein by reference.

[0054] The adjustable optical elements of the pupil shaping unit of the illumination system are first set to generate, in a pupil surface of the illumination system, a dipolar intensity distribution having two basic poles BP1 and BP2, each having the shape of a segment of an annulus where the outer radius is set to .sub.O=1 and the width of the annular segment is adjusted such that corresponds to the desired pole width PW of the desired lenticular pole. The pole angle is set to correspond to the desired pole angle of the lenticular pole (FIG. 6A).

[0055] It is to be noted that conventional systems may need to be modified with respect to the variation width of adjustable elements to allow for those dimensions of the annular segment shaped poles, which may be larger than poles in the conventional system in both the circumferential and the radial direction.

[0056] The illumination system includes a blocking device BD (FIG. 6B) at or close to the pupil surface of the illumination system where the effective source is generated, or at or close to a pupil surface optically conjugate thereto. The blocking device in this embodiment is designed to block those areas at the inner edge of the annulus segment shaped poles which lie outside the lenticular region desired for the pole shape. For that purpose, the blocking device may take the general butterfly shape schematically shown in FIG. 6B having wings extending in azimuthal direction over an angle corresponding to the pole angle, and comprising outer edges concavely curved inwards towards the center of the blocking device to define the inner edge of the lenticular pole. Those areas actually blocked by the blocking device are drawn with diagonal hatching in FIG. 6A. When the blocking device is inserted into the appropriate position, the resulting effective light source includes two lenticular poles POLE1 and POLE2 having the pole area A.sub.OPT or a slight variation therefrom as described above.

[0057] An exchanging device allowing to exchange blocking devices having different layout may be provided to allow variation of dimensions of the lenticular poles. Also, a blocking device may be designed to be variable, for example by providing a number of smaller blades arranged in the manner of a fan to allow varying the pole angle as desired. The blocking device may also be formed by an opaque coating on a transparent optical element, such as a plate, which may be exchangeable.

[0058] In FIGS. 7A-7B the poles of the dipolar intensity distribution have a parallelogram pole shape defined by a non-rectangular parallelogram having a shorter diagonal D2 and a longer diagonal D1, the shorter diagonal extending substantially parallel to a line parallel to the pole orientation axis POA. FIGS. 7A and 7B each show the lenticular regions LR corresponding to the optimum size and shape of poles adapted to a sub-pattern having predominant pitch P. In FIG. 7A, each of the poles POLE1 and POLE2 of the dipolar intensity distribution has the shape of a rhombus, i. e. the shape of an equilateral quadrangle arranged mirror symmetric to the pole orientation axis POA. The biconvex, diamond shaped poles are fit into the optimum sized lenticular region LR such that their longer diagonal D1 (extending in y-direction) defines the pole height PH, whereas the shorter diagonal D2 extending between the inner and outer edge of the pole in the second direction (x-direction) corresponds to the pole width PW. The area of each of the rhombohedral poles POLE1, POLE2 is A.sub.POLE=(D1*D2)/2, which is smaller than the area A.sub.OPT of the optimum sized lenticular region. The difference in area, A, may be calculated according to A=2*.Math.sin (2*)4*sin ()*(1cos ()).

[0059] As exemplified in FIG. 7B, it is not necessary to fit the diamond shaped poles exactly into the optimum lenticular region LR. For example, the rhombohedral poles may be slightly larger such that tips of the rhombohedron in the first direction and/or in the second direction may extend beyond the outer limits of the ideal lenticular region LR. Also, the pole dimension ration PH/PW may deviate from the optimum value, thereby resulting in slightly narrower or wider diamond shaped poles. The deviation may be up to about 10%, for example. Although some loss of contrast might occur, the oversizing of the rhombohedral poles allows to distribute light energy over larger areas at the edges of optical elements near to a pupil, thereby eliminating or decreasing problems associated with localized lens heating.

[0060] FIG. 8 shows a meridional section of an illumination system ILL configured to be used in a projection exposure step-and-scan apparatus. The illumination system includes a plurality of optical elements constructed and arranged to receive primary radiation emitted from a primary light source LS and to generate illumination radiation incident on a mask M bearing a pattern. In the embodiment, the light source LS is an excimer laser emitting primary radiation having a nominal operation wavelength in the deep ultraviolet (DUV) spectral range, for example 193 nm. Primary radiation exciting the light source LS forms a coherent radiation bundle having a small cross-section and low divergence, which corresponds to a small value of geometrical optical flux (etendue).

[0061] A number of optical elements following the light source LS form a variable pupil shaping unit PSU configured to generate a defined spatial intensity distribution corresponding to a selected illumination mode in a pupil surface P.sub.ILL of the illumination system. A partial system PS following the pupil surface P.sub.ILL includes a number of optical elements configured to transform the two-dimensional intensity distribution present in the pupil surface P.sub.ILL into a corresponding angular distribution of illumination radiation incident on the mask M. The partial system PS includes elements to homogenize the radiation and to sharply define the edges of an illumination field in the exit surface EX of the illumination system, which coincides with the object surface of a subsequent projection objective (see FIG. 3).

[0062] The pupil shaping unit PSU comprises, at an entry side thereof, a beam expansion unit EXP which expands the coherent radiation bundle received from the light source LS. The beam expansion unit includes a number of refractive optical elements diverting the radiation such that the cross-section of the bundle is increased without altering the geometrical optical flux of the bundle.

[0063] After passing through the beam expansion unit EXP the projection radiation bundle impinges on a first optical raster element RE1 structured to increase divergence of the radiation bundle by introducing propagation angles according to a desired illumination mode. In FIG. 8 this is schematically illustrated for an axial ray that is split into two diverging rays R1, R2 after the raster element RE1. The first raster element RE1 modifies the angular distribution of the radiation bundle and also increases the geometrical optical flux thereof.

[0064] The first raster element RE1 may be formed by a diffractive optical element (DOE) or by a refractive optical element (ROE), for example. Various kinds of optical raster elements are possible, for example a micro-lens array in which micro-lenses are formed by Fresnel zone plates. The first raster element RE1 may be a reflective raster element including an array of mirror segments oriented at different angles (multi-mirror array).

[0065] First raster element RE1 is arranged in a front focal plane FFP of a zoom/axicon module ZA which includes a zoom lens group Z providing a continuously variable magnification, and an axicon group AX comprising a pair of axicons AX1, AX2, one concave and one convex, whose separation can be varied continuously to allow adjusting a distance between the axicon surfaces between zero distance (contact between axicon surfaces) and preselected finite values.

[0066] The entire zoom/axicon module ZA is configured as a focal length optical system transforming an angular distribution of radiation in the front focal plane FFP into a desired two-dimensional (spatial) intensity distribution in the rear focal plane RFP, which coincides with the pupil surface P.sub.ILL of the illumination system.

[0067] A field defining arrangement FD is positioned in the rear focal plane RFP of the zoom/axicon module ZA. The field defining arrangement is configured to further introduce divergence into the radiation beam, thereby further enlarging the geometrical flux of the radiation beam. The divergence introduced by the field defining arrangement FD is schematically represented by divergent rays downstream of the field defining arrangement. Embodiments of field defining arrangements suitable for this purpose are disclosed, for example, in applicant's international application published as WO 2005/078522 A2, which is incorporated herein by reference in its entirety.

[0068] The diverging rays exiting the field defining arrangement FD enter an objective OB represented by a single condenser lens. The objective OB is arranged within the illumination systems such that an entrance pupil plane of the objective OB coincides with the exit pupil surface of the zoom/axicon module. The image plane of objective OB, which is a Fourier transform plane to the pupil P.sub.ILL of the illumination system, is a field plane FP within the illumination system. A reticle masking unit REMA acting as an adjustable field stop is positioned in the field plane FP to define the edges of the illumination field. Details of embodiments may be taken, for example from WO 2005/078522 A2. An imaging objective RO is arranged between the field plane FP and the exit surface ES of the illumination system, where the mask M may be positioned and allows to image the intensity distribution present in the field plane FP onto the mask pattern.

[0069] FIGS. 9A-9b show schematically a detail of a first raster element RE1 configured to generate a substantially dipolar intensity distribution having two illumination poles each having the diamond shape discussed in context of FIG. 7. The raster element RE1 is formed by an array of raster segments RS arranged side by side in a mosaic arrangement to cover without interspaces the complete illuminated surface of the raster element RE1. Such arrangement allows to generate the dipolar intensity distribution essentially without blocking any parts of the beam so that there is no intensity loss involved in the generation of the dipolar intensity distribution.

[0070] The raster element RE1 can be formed by an array of refractive lens segments. Each raster segment RS may be formed by a segment of a lens or a prism where an entry-side surface and an exit-side surface of the raster segment are oriented such that a desired redirection of radiation is caused by refraction at the segment surfaces as the radiation passes through the raster segment.

[0071] The rhombohedral shape of the raster segments RS determines the shape of the poles in the pupil surface P.sub.ILL. In FIG. 9A the raster segments are numbered 1 and 2 to illustrate which raster segments contribute to the illumination intensity in the two poles POLE1 and POLE2 shown in FIG. 9B. The structure of the raster segments is adjusted to redirect at least a major portion of impinging radiation to the two poles.

[0072] As illustrated by dashed lines in some raster segments of FIG. 9A, the raster elements may also be segmented into other shapes, such as triangular raster segments, where the size and shape of the triangles are adapted to define, in combination, a rhombohedral area corresponding to the desired pole shape.

[0073] In some embodiments, the raster element RE1 may be designed as a diffractive computer generated hologram (CGH) assigning to each of the raster segments RS a diffractive structure effective to redirect light intensity present in a partial beam impinging on the raster segment into the associated pole. The raster segments may have a shape differing from the shape of the desired poles, e.g. a rectangular shape. The diffractive structure may be formed on the surface of a transparent substrate, such as a plane plate made of fused silica or calcium fluoride. The raster segments may also be formed of grooves etched into the surface to provide segments of a Fresnel lenses. The depth and width of the grooves may be in the order of micrometers, for example. Each raster segment may be in the order of millimeters in size and the entire array forming the raster element may have typical dimensions of one or a few centimeters.

[0074] The basic distribution of angles introduced by the raster element RE1 may be modified within the pupil shaping unit by setting the elements of the variable zoom lens group Z and/or the variable axicon group AX accordingly. In one configuration, the separation between the axicon elements AX1 and AX2 is set to zero (contact between axicon surfaces) such that no radial redirection of light intensity is effected by the axicon group AX. In other configurations, a defined separation may be set, which generally causes a central symmetric dilatation of the radiation intensity distribution, which may be useful in some cases. Adjusting the magnification of the zoom lens group Z allows to adapt the illumination intensity distribution in the pupil surface P.sub.ILL of the illumination system to different NA values in a homomorphous imaging step.

[0075] It is to be noted that the conditions for the pole area A.sub.POLE, pole height PH and pole width PW as defined above remain unchanged as long as P*NA=2 k.sub.1*=const. Therefore, the area of the poles as well as the pole height and pole width may be considered as a function of parameter k.sub.1 alone, where k.sub.1<1 is a process dependent parameter that incorporates everything in a lithography process that is not wavelength or numerical aperture. Generally, any process or system modification that allows improvements in resolution effectively reduces the k.sub.1 factor. Diffraction-limited values are 0.25 for incoherent and 0.5 for coherent illumination, for example. Where k.sub.1 stays essentially constant, sub-patterns with different predominant periodicity values P (pitches) may be illuminated using one and the same raster element RE1 by simply adjusting the appropriate NA value using the zoom lens group Z. With other words: one type of raster element with fixed configuration, such as a diffractive optical element, may be sufficient for different processes defined by one particular k.sub.1 value (or a small variation of k.sub.1).

[0076] Different values for A.sub.POLE, PW and/or PH and different pole patterns may also be obtained by providing several different, interchangeable first raster elements. An automatic changer may be used to exchange raster elements of different configuration in the illumination system.

[0077] In general, the pupil-shaping unit may include at least one optical modulation device configured to controllably change an angular distribution of radiation incident on the optical modulation device such that a desired intensity distribution is obtained in the pupil surface of the illumination system. Although the optical modulation device may have a fixed configuration to obtain only one type of intensity distribution (which may be changeable by additional variable optical elements), it is possible that the optical modulation device has an array of individual elements configured to be driven individually to change the angular distribution of radiation incident on the individual elements. The optical modulation device may include a controllable micro-mirror array, for example. Various optical modulation devices which may be used when modified for that purpose are disclosed in applicant's international patent application published as WO 2005/026843 A2. The disclosure of this application is incorporated herein by reference.

[0078] FIGS. 10A-10B show diagrams illustrating, for an exemplary projection exposure systems, improvements that may be obtained if an optimized dipole setting is used instead of a conventional setting. In the diagrams, pupil coordinates in terms of image-side numerical aperture NA are drawn in the abscissa. The ordinates show values for the phase error value of the overall wavefront aberration. The value 0 corresponding to the optical axis. Each diagram shows an x-cut parallel to the x-axis (second direction) and an y-cut perpendicular thereto in the first direction. An x-dipole (orientation of pole orientation axis along x-direction) is used for all experiments.

[0079] A dipole setting having poles with the shape of a segment of an annulus (compare e.g. FIG. 2B) was used for the reference example of a conventional dipole setting CONV shown in FIG. 10A. An x-dipole setting with lenticular poles with A.sub.POLE=A.sub.OPT was used for the experiment illustrated in FIG. 10B.

[0080] The x-cut in both experiments illustrates that the localized heating in the region of the poles generally aligned along the x-direction leads to relatively strong wavefront aberrations close to the outer edge of the pupil where the energetic centriods of the poles are positioned. In contrast, essentially no wavefront aberration is caused on and close to the optical axis in the center of the pupil. In the y-direction a relatively large variation of wavefront aberrations occurs superimposed onto a slight gradient from the upper edge to the lower edge of the pupil in this specific type of projection objective used for the experiment.

[0081] The general characteristics of the local distribution of wavefront aberrations caused by lens heating are quite similar in the experiment according to FIG. 10B. Note, however, that the absolute amount of wavefront aberration caused by lens heating is reduced by about a factor two, i.e. reduced to about 50% of the value for the conventional setting in the x-direction (different scales in the ordinates in FIG. 10A, in FIG. 10B). This significant improvement is attributed to the fact that radiant energy present within the projection beam is distributed over larger areas at the position of the poles, whereby the energy distribution in smeared out over a larger area, thereby reducing associated lens heating effects on wavefront aberrations. The effect is less pronounced in the y-direction.

[0082] In some embodiments, the multipolar illumination is a pure dipole illumination, where the intensity distribution in the pupil surface of the illumination system is formed by one substantially dipolar intensity distribution having exactly two illumination poles. Such dipole illumination may be used, for example, to print periodic patterns with features essentially aligned in one direction only. As a practical example, a layer of a highly integrated flash memory device may be strongly dominated by lines running in direction only.

[0083] Due to the relatively large pole height PH of the optimized pole areas those dipole illuminations are also useful to improve contrast of sub-patterns not running at the predominant direction, but at an angle thereto since the dipolar intensity distribution includes illumination directions significantly deviating from the direction defined by the pole orientation axis.

[0084] The multipolar illumination mode may include two substantially dipolar intensity distributions (dipoles) superimposed on each other, where the pole orientation axes of the dipoles are rotated relatively to each other, for example by 90. FIG. 11 depicts schematically the intensity distribution in the illumination pupil for a quadrupole illumination having two pairs of lenticular poles relatively rotated by 90 around the optical axis to form four lenticular poles, POLE1, POLE2, POLE3, POLE 4, where the pole orientation axis POA.sub.x of the first dipolar intensity distribution (forming poles POLE1 and POLE2) is oriented along the x-direction, whereas the pole orientation axis POA.sub.y of the second dipolar intensity distribution (forming poles POLE3 and POLE4) is oriented perpendicular thereto in the y-direction. A bi-directional pattern having periodic features in x- and y-direction may be printed with high contrast for both directions. The area of each of the poles corresponds approximately to the optimum area A.sub.OPT in this example.

[0085] As illustrated schematically in FIG. 12, the method and apparatus may also be used in exposure processes which do not exploit the full resolution potential of the projection exposure apparatus. For example, a multi-polar illumination mode may be set for a pattern or sub-pattern having features spaced apart periodically at a predominant periodicity length P where the condition 0.7*/NA<P</NA holds. Those conditions are sometimes used in processes known as double patterning where a final structure with very fine pitch is formed by subsequent exposures of coarser pattern relatively shifted to each other by one half of the periodicity length. The corresponding illumination mode may be characterized by two poles POLE1, POLE2 positioned between the edge of the pupil (corresponding to full NA) and the optical axis and defining a distance d</(P*NA). The poles may have a lenticular or a biconvex polygonal shape, for example.

[0086] The disclosure has been given by way of example. From this information, those skilled in the art will not only understand the disclosure and its attendant advantages, but will also find apparent various changes and modifications to the structures and methods disclosed. The applicant seeks, therefore, to cover all such changes and modifications as fall within the spirit and scope of defined by the appended claims, and equivalents thereof.

[0087] The contents of the claims are incorporated by reference into the description.