WO2016184560A1

WO2016184560A1 - Method of operating a microlithographic projection apparatus

Info

Publication number: WO2016184560A1
Application number: PCT/EP2016/000802
Authority: WO
Inventors: Jörg ZIMMERMANN; Jens Timo NEUMANN; Frank Schlesener; Ralf MÜLLER
Original assignee: Carl Zeiss SMT GmbH
Current assignee: Carl Zeiss SMT GmbH
Priority date: 2015-05-21
Filing date: 2016-05-14
Publication date: 2016-11-24
Anticipated expiration: 2017-11-21
Also published as: CN107636539A; TW201702756A; JP2018519535A; KR20180010242A

Abstract

A method of operating a microlithographic projection apparatus comprises the step of providing a mask (16), an illumination system (12) and a projection objective (20) configured to form an image of an object field (14), which is illuminated on the mask (16) in a mask plane, on an image field positioned on a light sensitive surface (22). Edge placement errors are determined at different field points in the image field. The mask (16) is then illuminated with projection light having an improved field dependency of the angular irradiance distribution. The angular irradiance distribution according to the improved field dependency varies over the object field (14) in such a way that the edge placement errors determined in step b) are reduced at the different field points.

Description

METHOD OF OPERATING

A MICROLITHOGRAPHIC PROJECTION APPARATUS

1. Field of the Invention

The invention generally relates to the field of microlithography, and in particular to illumination systems used in projection exposure apparatus or mask inspection apparatus. The invention is particularly concerned with correcting edge placement errors (EPE) which denotes the difference of the desired and the real feature edge locations in the image plane of the objective at waver level.

2. Description of Related Art

Microlithography (also referred to as photolithography or simply lithography) is a technology for the fabrication of integrated circuits, liquid crystal displays and other micro- structured devices. The process of microlithography, in conjunction with the process of etching, is used to pattern features in thin film stacks that have been formed on a sub- strate, for example a silicon wafer. At each layer of the fabrication, the wafer is first coated with a photoresist which is a material that is sensitive to radiation, such as deep ultraviolet (DUV) light. Next, the wafer with the photoresist on top is exposed to projection light in a projection exposure apparatus. The apparatus projects a mask containing a pattern onto the photoresist so that the latter is only exposed at certain locations which are determined by the mask pattern. After the exposure the photoresist is developed to produce an image corresponding to the mask pattern. Then an etch process transfers the pattern into the thin film stacks on the wafer. Finally, the photoresist is removed. Repetition of this process with different masks results in a multi- layered microstructured component.

A projection exposure apparatus typically includes a light source, an illumination system that illuminates the mask with projection light produced by the light source, a mask stage for aligning the mask, a projection objective and a wafer alignment stage for aligning the wafer coated with the photoresist. The illumination system illuminates a field on the mask that may have the shape of a rectangular or curved slit, for example.

In current projection exposure apparatus a distinction can be made between two different types of apparatus. In one type each target portion on the wafer is irradiated by exposing the entire mask pattern onto the target portion in one go. Such an apparatus is commonly referred to as a wafer stepper. In the other type of apparatus, which is commonly referred to as a step-and-scan apparatus or scanner, each target portion is irradiated by progressively scanning the mask pattern under the projection beam along a scan direction while synchro- nously moving the substrate parallel or anti-parallel to this direction. The ratio of the velocity of the wafer and the velocity of the mask is egual to the magnification of the projection objective, which is usually smaller than 1, for example 1:4. It is to be understood that the term "mask" (or reticle) is to be interpreted broadly as a patterning means. Commonly used masks contain opaque or reflective patterns and may be of the binary, alternating phase-shift, attenuated phase- shift or various hybrid mask type, for example. However,there are also active masks, e.g. masks realized as a programmable mirror array. Also programmable LCD arrays may be used as active masks. As the technology for manufacturing microstructured devices advances, there are ever increasing demands also on the illumination system. Ideally, the illumination system illuminates each point of the illumination field on the mask with projec- tion light having a well defined spatial and angular irradi- ance distribution. The term angular irradiance distribution describes how the total light energy of a light bundle, which converges towards a particular point in the mask plane, is distributed among the various directions of the rays that constitute the light bundle.

The angular irradiance distribution of the projection light impinging on the mask is usually adapted to the kind of pattern to be projected onto the photoresist. Often the optimum angular irradiance distribution depends on the size, orientation and pitch of the features contained in the pattern. The most commonly used angular irradiance distributions of projection light are referred to as conventional, annular, di- pole and quadrupole illumination settings. These terms refer to the irradiance distribution in a pupil plane of the illu- mination system. With an annular illumination setting, for example, only an annular region is illuminated in the pupil plane. Thus there is only a small range of angles present in the angular irradiance distribution of the projection light, and all light rays impinge obliquely with similar angles onto the mask.

Different means are known in the art to modify the angular irradiance distribution of the projection light in the mask plane so as to achieve the desired illumination setting. In the simplest case a stop (diaphragm) comprising one or more apertures is positioned in a pupil plane of the illumination system. Since locations in a pupil plane translate into angles in a Fourier related field plane such as the mask plane, the size, shape and location of the aperture (s) in the pupil plane determines the angular irradiance distributions in the mask plane. However, any change of the illumination setting requires a replacement of the stop. This makes it difficult to finely adjust the illumination setting, because this would require a very large number of stops that have aperture (s) with slightly different sizes, shapes or locations. Furthermore, the use of stops inevitably results in light losses and thus in a reduced throughput of the apparatus.

Many common illumination systems therefore comprise adjustable elements that make it possible, at least to a certain ex- tent, to continuously vary the illumination of the pupil plane. Many illumination systems use an exchangeable diftractive optical element to produce a desired spatial irradiance distribution in the pupil plane. If zoom optics and a pair of axicon elements are provided between the diffractive optical element and the pupil plane, it is possible to adjust this spatial irradiance distribution.

Recently it has been proposed to use mirror arrays that illuminate the pupil plane. In EP 1 262 836 Al the mirror array is realized as a micro-electromechanical system (MEMS) com- prising more than 1000 microscopic mirrors. Each of the mirrors can be tilted in two different planes perpendicular to each other. Thus radiation incident on such a mirror device can be reflected into (substantially) any desired direction of a hemisphere. A condenser lens arranged between the mirror array and the pupil plane translates the reflection angles produced by the mirrors into locations in the pupil plane. This known illumination system makes it possible to illuminate the pupil plane with a plurality of spots, wherein each spot is associated with one particular microscopic mirror and is freely movable across the pupil plane by tilting this mirror .

Similar illumination systems are known from US 2006/0087634 Al, US 7,061,582 B2, WO 2005/026843 A2 and WO 2010/006687 Al . US 2010/0157269 Al discloses an illumination system in which an array of micromirrors is directly imaged on the mask.

As mentioned further above, it is usually desired to illuminate, at least after scan integration, all points on the mask with the same irradiance and angular irradiance distribution. If points on the mask are illuminated with different irradi- ances, this usually results in undesired variations of the critical dimension (CD) on wafer level. For example, in the presence of irradiance variations the image of a uniform line on the mask on the light sensitive may also have irradiance variations along its length. Because of the fixed exposure threshold of the resist, such irradiance variations directly translate into widths variations of a structure that shall be defined by the image of the line. If the angular irradiance distribution unintentionally varies over the illumination field on the mask, this also has a negative impact on the quality of the image that is produced on the light sensitive surface. For example, if the angular irradiance distribution is not perfectly balanced, i. e. more light impinges from one side on a mask point than from the opposite side, the conjugate image point on the light sensitive surface will be laterally shifted if the light sensitive surface is not perfectly arranged in the focal plane of the projection objective. For modifying the spatial irradiance distribution (i.e. the field dependency of the irradiance) in the illumination field US 6,404,499 A and US 2006/0244941 Al propose mechanical devices that comprise two opposing arrays of opaque finger-like stop elements that are arranged side by side and aligned par- allel to the scan direction. Each pair of mutually opposing stop elements can be displaced along the scan direction so that the distance between the opposing ends of the stop elements is varied. If this device is arranged in a field plane of the illumination system that is imaged by an objective on the mask, it is possible to produce a slit-shaped illumination field whose width along the scan direction may vary along the cross-scan direction. Since the irradiance is inte- grated during the scan process, the integrated irradiance (sometimes also referred to as illumination dose) can be finely adjusted for a plurality of cross-scan positions in . the illumination field.

Unfortunately these devices are mechanically very complex and expensive. This is also due to the fact that these devices have to be arranged in or very close to a field plane in which usually the blades of a movable field stop is arranged.

Adjusting the angular irradiance distribution in a field dependent manner is more difficult. This is mainly because the spatial irradiance distribution is only a function of the spatial coordinates x, y, whereas the angular irradiance distribution also depends on the angles , β.

WO 2012/100791 Al discloses an illumination system in which a mirror array is used to produce a desired irradiance distri- bution in the pupil plane of the illumination system. In close proximity to the pupil plane a fly's eye optical integrator is arranged that has a plurality of light entrance facets. Images of the light entrance facets are superimposed on the mask. The light spots produced by the mirror array have an area that is at least five times smaller than the to^¬ tal area of the light entrance facets. This makes it possible to produce variable light patterns on the light entrance facets, and thus different angular irradiance distributions at different portions of the illumination field. For example, at one portion of the illumination field an X dipole and at another portion of the illumination field a Y dipole illumination setting may be produced. WO 2012/028158 Al discloses an illumination system in which the irradiance distribution on the light entrance facets of the fly's eye optical integrator is modified with the help of a plurality of modulator units that are arranged in front of the optical integrator. Each modulator unit is associated with one of the light entrance facets and variably redistributes, without blocking any light, the spatial and/or angular irradiance distributions on the associated light entrance facet. In this manner it is possible, for example, to illumi- nate, with different illumination settings, two or more different portions on a single die that are associated with different semiconductor devices.

Unpublished patent application PCT/EP2014/003049 discloses an approach in which the irradiance distribution on the light entrance facets of a fly' s eye optical integrator is modified by imaging a digital mirror device (DMD) on the light entrance facets. This approach is advantageous because it is not necessary to produce very small light spots with an analog micromirror array, as this is the case in the illumina- tion system known from WO 2012/100791 Al mentioned above. The field dependency of the angular irradiance distribution is adjusted so that the angular irradiance distribution over the illumination field becomes perfectly uniform (i.e. field independent) . However, it is also mentioned that it may sometimes be desirable to deliberately introduce a field dependency of the angular irradiance distribution. This may be expedient, for example, if the projection objective or the mask has field depending properties. As far as the mask is concerned, such field depending properties are usually a result of features that have different orientations or dimensions. Adverse effects resulting from such field dependencies can be successfully reduced by selectively introducing a field dependency of the angular irradiance distribution. The industry that uses mxcrolithographic projection apparatus for the production of integrated circuits and other electronic or micromechanical devices constantly strives for smaller feature dimensions, higher output and higher yield. One of the crucial goals is to reduce edge placement errors (EPE) . Edge placement errors denote the difference between locations of a real (or simulated) contour of a structure lithographically defined on a wafer (or a similar support) on the one hand, and the locations of the desired contour on the other hand. The edge placement error is a fundamental quantity that determines other common quantities such as the critical dimension (CD) and overlay error. A reduction of edge placement errors directly results in higher yields and/or smaller feature size.

FIGS. 16a, 16b and 16c illustrate how edge placement errors are usually /calculated. In the upper half of each figure a target structure ST having a desired contour is shown. In the lower half the rectangles drawn with solid lines represent real structures ST' that have been produced on a wafer in a microlithographic process.

In the case shown in FIG. 16a the real structure ST' is broader than the target structure ST. The edges extending along the longitudinal direction of the structures ST' are displaced by a positive edge placement error E = d_ra - dt, wherein d_m is the measured distance from the symmetry line and dt is the target distance from the symmetry line.

If the target distance and the measured distance are equal, as this is shown in FIG. 16b, the edge placement error E is zero .

If the measured distance d_m is smaller than the target distance dt, the image placement error E becomes negative, as this is shown in FIG. 16c. SUMMARY OF THE INVENTION

It is an object of the present invention to provide a method of operating a microlithographic projection apparatus which makes it possible to reduce edge placement errors. In accordance with the present invention, this object is achieved by a method in which a mask, an illumination system configured to illuminate the mask and a projection objective is provided in a step a). The projection objective is configured to form an image of an object field, which is illuminat- ed on the mask in a mask plane, on an image field that is positioned on a light sensitive surface such as a resist or - in the case of a mask inspection apparatus - a CCD sensor.

In a next step b) , edge placement errors are determined at different field points in the image field. This may be accom- plished by measurement or simulation.

In a final step c) , the mask is illuminated with projection light having an improved field dependency of the angular ir- radiance distribution. The angular irradiance distribution according to the improved field dependency varies over the object field in such a way that the edge placement errors determined in step b) are reduced.

Although it is known in the art as such that the angular irradiance distribution has an impact on edge placement errors, it has not been proposed before to determine a field depend- ency of the edge placement error and to produce a field dependent angular irradiance distribution in the illumination field that is determined in such a way that the edge placement errors are reduced in a field dependent manner.

The edge placement error determined in step b) may include at least one of the group consisting of CD variations and overlay variations. When the edge placement error is determined in step b) , the mask may be illuminated with projection light having an original field dependency of the angular irradiance distribution. Then the edge placement error on the light sensitive surface is simulated or measured at different field points in the image field. In step c) the field dependency of the original angular irradiance distribution may then be changed so as to obtain the improved field dependency of the angular irradiance distribution. These steps may be repeated once or a cou- pie of times. This implies that that the improved field dependency of the angular irradiance distribution becomes the original field dependency of the angular irradiance distribution of the next determination step. In this way it is possible to recursively improve the field dependency of the angu- lar irradiance distribution until the edge placement errors become very small or even reach a minimum value.

When the edge placement errors are determined the first time, the original angular irradiance distribution may be constant, i.e. there is no field dependency. However, it is also possi- ble to start with an original angular irradiance distribution that already has a field dependency. This original field dependency may be computed on the basis of the feature size and orientation on the mask, for example.

Step c) may comprise the step of illuminating the mask with projection light having not only an improved field dependency of the angular irradiance distribution, but also an improved field dependency of the irradiance. The irradiance varies over the object field in such a way that the edge placement errors determined in step b) are reduced at the different field points. In other words, in a common optimization process the field dependency of the angular irradiance distribution and of the irradiance are improved such that the edge placement errors are reduced. In that case step b) may additionally comprise the steps of illuminating the mask with projection light having an origi^¬ nal field dependency of the irradiance, and simulating or measuring the edge placement errors on the light sensitive surface at the different field points. Step c) then comprises the additional step of changing the original field dependency of the irradiance such that the improved field dependency of the irradiance is obtained.

If the mask has a portion in which the mask pattern is uni- form (i.e. the width, pitch and orientation of the structures do not vary) , the conventional approach has been to illuminate that portion with a field independent angular irradiance distribution and uniform scan integrated irradiance.

In accordance with the present invention, however, the angu- lar irradiance distribution may nevertheless vary over an ar^¬ ea of the object field that coincides, at least at one moment during step c) , with the portion of the mask having the uni^¬ form mask pattern. In other words, the angular irradiance distribution intentionally varies over the uniform mask pat- tern in order to reduce edge placement errors that may be caused by deficiencies of the projection objective.

As a matter of course, if the mask comprises a non-uniform mask pattern that has locally varying properties, it is also possible to adapt the improved angular irradiance distribu- tion to the locally varying properties of the mask pattern. The locally varying properties of the mask pattern may in^¬ clude at least one of the group consisting of: structure width, structure pitch and structure orientation.

In one embodiment at least at some field points the angular irradiance distribution according to the improved field dependency is non-telecentric . At least one of the mask and the light sensitive surface is displaced along an optical axis of the projection objective before step c) . This results in a lateral shift of the image locations. In this manner edge placement errors, and in particular overlay errors, can be reduced in a field , dependent manner.

If the mask continuously moves in step c) during a scan cy- cle, the angular irradiance distribution may vary during the scan cycle. Then the angular irradiance distribution does not depend only on the field coordinates, but on time as well.

An illumination system which is capable of producing a field dependent angular irradiance distribution and also a field dependent irradiance preferably comprises an optical integrator configured to produce a plurality of secondary light sources located in a pupil plane of the illumination system. The optical integrator comprises a plurality of light entrance facets each being associated with one of the secondary light sources . Images of the light entrance facets at least substantially superimpose in the mask plane. A spatial light modulator is provided that has a light exit surface and is configured to transmit or to reflect impinging projection light in a spatially resolved manner. An objective images the light exit surface of the spatial light modulator onto the light entrance facets of the optical integrator. In step c) the spatial light modulator is controlled such that the improved angular irradiance distribution is obtained in the mask plane. The illumination system may further comprise an adjustable pupil forming unit that directs projection light on the spatial light modulator. The pupil forming unit may itself comprise a first beam deflection array or first reflective or transparent beam deflection elements. Each beam deflection element is configured to illuminate a spot on the spatial light modulator at a position that is variable by changing a deflection angle produced by the beam deflection element. The spatial light modulator may comprise a second beam deflection array of second reflective or transparent beam deflection elements. Each second beam deflection element may be capable to be in an "on" state, in which it directs impinging light towards the optical integrator, and in an "off" state, in which it directs impinging light elsewhere. The second beam deflection array may be realized as a digital mirror device (DMD) , for example.

Subject of the invention is also an illumination system of a microlithographic projection apparatus comprising an optical integrator that is configured to produce a plurality of secondary light sources in a pupil plane of the illumination system. The optical integrator comprises a plurality of light entrance facets each being associated with one of the second- ary light sources. The spatial light modulator has a light exit surface and is configured to transmit or to reflect impinging projection light in a spatially resolved manner. A pupil forming unit is configured to direct projection light on the spatial light modulator. An objective images the light exit surface of the spatial light modulator onto the light entrance facets of the optical integrator. A control unit is configured to control the pupil forming unit and the spatial light modulator such that the mask is illuminated with projection light having an improved field dependency of the an- gular irradiance distribution. The angular irradiance distribution according to the improved field dependency varies over the object field in such a way that edge placement errors that vary over the image field are reduced.

Subject of the invention is also a microlithographic projec- tion apparatus comprising a mask, an illumination system that is configured to illuminate the mask, and a projection objective that is configured to form an image of an object field, which is illuminated on the mask in a mask plane, on an image field positioned on a light sensitive surface. Means are pro- vided for illuminating the mask with projection light having an improved field dependency of the angular irradiance distribution, wherein the angular irradiance distribution according to the improved field dependency varies over the ob- ject field in such a way that edge placement errors that vary over the image field are reduced.

DEFINITIONS

The term "light" is used herein to denote any electromagnetic radiation, in particular visible light, UV, DUV, VUV and EUV light and X-rays.

The term "light ray" is used herein to denote light whose path of propagation can be described by a line.

The term "light bundle" is used herein to denote a plurality of light rays that have a common origin in a field plane. The term "light beam" is used herein to denote all light that passes through a particular lens or another optical element.

The term "position" is used herein to denote the location of a reference point of a body in the three-dimensional space. The position is usually indicated by a set of three Cartesian coordinates. The orientation and the position therefore fully describe the placement of a body in the three-dimensional space .

The term "surface" is used herein to denote any plane or curved surface in the three-dimensional space. The surface may be part of a body or may be completely separated therefrom, as it is usually the case with a field or a pupil plane.

The term "field plane" is used herein to denote the mask plane or any other plane that is optically conjugate to the mask plane. The term "pupil plane" is a plane in which (at least approximately) a Fourier relationship is established to a field plane. Generally marginal rays passing through different points in the mask plane intersect in a pupil plane, and chief rays intersect the optical axis. As usual in the art, the term "pupil plane" is also used if it is in fact not a plane in the mathematical sense, but is slightly curved so that, in the strict sense, it should be referred to as pupil surface . The term "uniform" is used herein to denote a property that does not depend on the position.

The term "optical raster element" is used herein to denote any optical element, for example a lens, a prism or a dif- fractive optical element, which is arranged, together with other identical or similar optical raster elements so that each optical raster element is associated with one of a plurality of adjacent optical channels.

The term "optical integrator" is used herein to denote an optical system that increases the product NA-a, wherein NA is the numerical aperture and a is the illumination field area.

The term "condenser" is used herein to denote an optical element or an optical system that establishes (at least approximately) a Fourier relationship between two planes, for example a field plane and a pupil plane. The term "conjugated plane" is used herein to denote planes between which an imaging relationship is established. More information relating to the concept of conjugate planes are described in an essay E. Delano entitled: "First-order Design and the y, y Diagram", Applied Optics, 1963, vol. 2, no. 12, pages 1251-1256. The term "field dependency" is used herein to denote any functional dependency of a physical quantity from the position in a field plane.

The term "angular irradiance distribution" is used herein to denote how the irradiance of a light bundle varies depending on the angles of the light rays that constitute the light bundle. Usually the angular irradiance distribution can be described by a function I_a(a, β), with a, β being angular coordinates describing the directions of the light rays. If the angular irradiance distribution has a field dependency so that it varies at different field points, l_a will be also a function of field coordinates, i. e. I_a= la (a, β, x, y) . The field dependency of the angular irradiance distribution may be described by a set of expansion coefficients ay of a Taylor (or another suitable) expansion of I_a (a, β, x, y) in x,y.

The term "irradiance" is used herein to denote the total irradiance that can be measured at a particular field point. The irradiance can be deduced from the angular irradiance distribution by integrating over all angles a, β. The irradi- ance has usually also a field dependency so that I_s- I_s (x, y) with x, y being spatial coordinates of the field point. The field dependency of the irradiance is also referred to as spatial irradiance distribution. In a projection apparatus of the scanner type, the light dose at a field point is obtained by integrating the irradiance over the time.

BRIEF DESCRIPTION OF THE DRAWINGS

Various features and advantages of the present invention may be more readily understood with reference to the following detailed description taken in conjunction with the accompany- ing drawings in which: FIG. 1 is a schematic perspective view of a projection exposure apparatus in accordance with one embodiment of the present invention;

FIG. 2 is an enlarged perspective view of the mask to be projected by the projection exposure apparatus shown in FIG. 1, illustrating local variations of the angular irradiance distribution on the mask;

FIG. 3 is a meridional section through an illumination

system being part of the apparatus shown in FIG. 1; FIG. 4 is a perspective view of a first mirror array contained in the illumination system shown in FIG. 3;

FIG. 5 is a perspective view of a second mirror array contained in the illumination system shown in FIG. 3;

FIG. 6 is a perspective view of an optical integrator con- tained in the illumination system shown in FIG. 3;

FIG. 7 is a schematic meridional section through the first and the second mirror array shown in FIGS. 4 and 5;

FIG. 8 is a perspective view on the second mirror array shown in FIG. 5, but illuminated with two poles; FIG. 9 is a perspective view of the optical integrator

shown in FIG. 6, but illuminated with two poles;

FIG. 10 is a schematic meridional section through a portion of the illumination system in which only a mirror array, a condenser and an array of optical raster elements are shown;

FIGS . 11a and lib are top views on the second mirror array and the optical integrator shown in FIG. 3; FIG. 12 illustrates an irradiance distribution on a light entrance facet of the optical integrator;

FIG. 13 is a graph showing the scan integrated irradiance distribution along the X direction produced by the light entrance facet shown in FIG. 12;

FIG. 14 illustrates another irradiance distribution on a light entrance facet of the optical integrator;

FIG. 15 is a graph showing the scan integrated irradiance distribution along the X direction produced by the light entrance facet shown in FIG. 14;

FIG. 16a to 16c illustrate the definition of edge placement errors ;

FIG. 17a and 17b illustrate how edge placement errors can be corrected by producing telecentricity errors and displacing the mask or the wafer;

FIG. 18 is an enlarged perspective view of the mask similar to FIG. 2 illustrating how different mask patterns are illuminated with different angular irradiance distributions; FIG. 19 is a flow diagram that illustrates important method steps .

DESCRIPTION OF PREFERRED EMBODIMENTS

I.

General Construction of Projection Exposure Apparatus FIG. 1 is a perspective and highly simplified view of a projection exposure apparatus 10 in accordance with the present invention. The apparatus 10 comprises a light source 11 that may be realized as an excimer laser, for example. The light source 11 in this embodiment produces projection light having a center wavelength of 193 nm. Other wavelengths, for example 157 nm or 248 nm, are envisaged as well.

The apparatus 10 further comprises an illumination system 12 which conditions the projection light provided by the light source 11 in a manner that will be explained below in further detail. The projection light emerging from the illumination system 12 illuminates an illumination field 14 on a mask 16. The mask 16 contains a pattern 18 formed by a plurality of small features 19 that are schematically indicated in FIG. 1 as thin lines. In this embodiment the illumination field 14 has the shape of a rectangle. However, other shapes of the illumination field 14, for example a ring segment, are also contemplated .

A projection objective 20 comprising lenses LI to L6 images the pattern 18 within the illumination field 14 onto a light sensitive layer 22, for example a photoresist, which is supported by a substrate 24. The substrate 24, which may be formed by a silicon wafer, is arranged on a wafer stage (not shown) such that a top surface of the light sensitive layer 22 is precisely located in an image plane of the projection objective 20. The mask 16 is positioned by means of a mask stage (not shown) in an object plane of the projection objective 20. Since the latter has a magnification β with |β|< 1, a minified image 18' of the pattern 18 within the illumina- tion field 14 is projected onto the light sensitive layer 22.

During the projection the mask 16 and the substrate 24 move along a scan direction which corresponds to the Y direction indicated in FIG. 1. The illumination field 14 then scans over the mask 16 so that patterned areas larger than the il- lumination field 14 can be continuously imaged. The ratio between the velocities of the substrate 24 and the mask 16 is equal to the magnification β of the projection objective 20. If the projection objective 20 does not invert the image (β >0) , the mask 16 and the substrate 24 move along the same direction, as this is indicated in FIG. 1 by arrows Al and Ά2. However, the present invention may also be used in stepper tools in which the mask 16 and the substrate 24 do not move during projection of the mask.

II.

Field dependent angular irradiance distribution

FIG. 2 is an enlarged perspective view of the mask 16 containing another exemplary pattern 18. For the sake of sim- plicity it is assumed that the pattern 18 is uniform, i.e. it comprises only identical features 19 that extend along the Y direction and are spaced apart by the same distance. It is further assumed that the features 19 extending along the Y direction are best imaged on the light sensitive layer 22 with an X dipole illumination setting.

In FIG. 2 an exit pupil 26a associated with a light bundle is illustrated by a circle. The light bundle converges towards a field point that is located at a certain X position of the illumination field 14 at a first time during a scan cycle. In the exit pupil 26a two poles 27a, which are spaced apart along the X direction, represent directions from which projection light propagates towards this field point. The light energies concentrated in each pole 27a are assumed to be equal. Thus the projection light impinging from the +X direc- tion has the same energy as the projection light impinging from the -X direction. Since the features 19 are assumed to be uniformly distributed over the pattern 18, one should expect that this X dipole illumination setting should be produced at each point on the mask 16. However, if this X dipole illumination setting is maintained during the entire scan cycle and over the entire length of the illumination field 14, it may turn out that the structures, which are produced on the substrate 24 after the expo- sure and a subsequent edging step, are not positioned at the intended locations. More specifically, the edges of the structures may be displaced along the X direction in a manner that has been explained above with reference to FIGS. 16a and 16c. In other words, although identical features 19 are illuminated with the same irradiance and the same angular irradi- ance distribution, edge placement errors may occur. The edge placement errors usually adversely affect the critical dimension (CD) budget and/or may cause serious overlay problems. There are various causes that may contribute to such edge placement errors. For example, certain proximity effects such as scattering light associated with the features 19 may have the result that the features 19 located at the circumference of the mask 16 may be imaged differently if compared with features 19 located at its center. Other causes for edge placement errors include lens heating effects in the projection objective 20. For example, optical elements in the projection objective 20 that are located in the vicinity of field planes are illuminated in a non-rotationally manner. Since an (albeit) small fraction of the projection light is absorbed by each optical element, this may result in a non- rotationally symmetric heat distribution and consequently in a non-rotationally symmetric deformation of these optical elements. If the optical elements are located close to (but not in) a field plane, such deformations may result in field related aberrations such as distortion.

The present invention, according to one of its various as^¬ pects, is concerned with eliminating or at least reducing edge placement errors that may occur as a result of these and similar causes. Surprisingly it has turned out that the edge placement errors caused by a large number of different effects can be reduced very significantly by slightly modifying the angular irradiance distribution, and preferably also the irradiance, in a field dependent manner. In principle it is even possible that each point on the mask 16 is illuminated with a different combination of irradiance and angular irradiance distribution. This corrective need is usually even stronger if the pattern 18 is not uniform as shown in FIG. 2, but varies over the mask 16. However, even with uniform patterns 19, as illustrated in FIG. 2, or with uniform pattern portions, field dependent edge placement errors are frequently observed and require an at least partial correction.

In FIG. 2 the different illumination conditions at different field points are represented by two further exit pupils 26b, 26c that are produced at different X positions and at different times during the scan cycle. In the exit pupil 26b the light energies concentrated in each pole 27b are still equal. However, the light cones associated with the poles 27b are tilted compared to the light cones of light that is associated with the exit pupil 26a.

In the exit pupil 26c the poles 27c are located at the same positions as the poles 27a. Thus the directions from which the projection light impinges on the respective field point are identical. However, the poles 27c are not balanced, i.e. the light energy concentrated in the poles 27c differs from one another. Thus the projection light impinging from the +X direction has less energy than the projection light impinging from the -X direction. Both exit pupils 26b, 26c result in a telecentricity error. This means that the energetic center line of the light cones do not impinge perpendicularly on the mask 16, but obliquely. This may be used, together with axially displacing the mask 16 and/or the substrate 24, to influence the edge locations at substrate level in a manner that will be explained below in more detail.

A field dependency of the angular irradiance distribution may not only be required along the X direction, but also along the Y direction within the illumination field 14. Then one point on the mask 16 experiences different angular irradiance distributions while it passes through the illumination field 14 during a scan cycle. If a field dependency along the Y di- rection (i.e. the scan direction) occurs, it has to be taken into account that the total effect for a particular field point is obtained by integrating the different angular irradiance distributions over time.

There is a wide variety of other field-dependent variations of the angular irradiance distribution that may be necessary to reduce edge placement errors. For example, the poles in the exit pupil associated with some field points may be deformed, blurred or may have a desired non-uniform irradiance distribution . As mentioned above, it may also be necessary to vary not only the angular irradiance distribution over the illumination field 14, but also the irradiance which is obtained by integrating the angular irradiance distribution over all possible angles. How desired variations of the irradiance and the an- gular irradiance distribution can be accomplished by the illumination system 12 will be explained in more detail in the next two sections III and IV.

III.

General Construction of Illumination System FIG. 3 is a meridional section through the illumination system 12 shown in FIG. 1. For the sake of clarity, the illustration of FIG. 3 is considerably simplified and not to scale. This particularly implies that different optical units are represented by one or very few optical elements only. In reality, these units may comprise significantly more lenses and other optical elements. In the embodiment shown, the projection light emitted by the light source 11 enters a beam expansion unit 32 which outputs an expanded and almost collimated light beam 34. To this end the beam expansion unit 32 may comprise several lenses or may be realized as a mirror arrangement, for example.

The projection light beam 34 then enters a pupil forming unit 36 that is used to produce variable spatial irradiance distributions in a subsequent plane. To this end the pupil form^¬ ing unit 36 comprises a first mirror array 38 of very small mirrors 40 that can be tilted individually about two orthogo^¬ nal axes with the help of actuators. FIG. 4 is a perspective view of the first mirror array 38 illustrating how two parallel light beams 42, 44 are reflected into different directions depending on the tilting angles of the mirrors 40 on which the light beams 42, 44 impinge. In FIGS. 3 and 4 the first mirror array 38 comprises only 6x6 mirrors 40; in reality the first mirror array 38 may comprise several hundreds or even several thousands mirrors 40.

The pupil forming unit 36 further comprises a prism 46 having a first plane surface 48a and a second plane surface 48b that are both inclined with respect to an optical axis OA of the illumination system 12. At these inclined surfaces 48a, 48b impinging light is reflected by total internal reflection. The first surface 48a reflects the impinging light towards the mirrors 40 of the first mirror array 38, and the second surface 48b directs the light reflected from the mirrors 40 towards an exit surface 49 of the prism 46. The angular irra^¬ diance distribution of the light emerging from the exit sur^¬ face 49 can thus be varied by individually tilting the mir- rors 40 of the first mirror array 38. More details with regard to the pupil forming unit 36 can be gleaned from US 2009/0116093 Al . The angular irradiance distribution produced by the pupil forming unit 36 is transformed into a spatial irradiance distribution with the help of a first condenser 50. The condenser 50, which may be dispensed with in other embodiments, di- rects the impinging light towards a digital spatial light modulator 52 that is configured to reflect impinging light in a spatially resolved manner. To this end the digital spatial light modulator 52 comprises a second mirror array 54 of mi- cromirrors 56 that are arranged in a mirror plane 57 and can be seen best in the enlarged cut-out C of FIG. 3 and the enlarged cut-out C of FIG. 5. In contrast to the mirrors 40 of the first mirror array 38, however, each micromirror 56 of the second mirror array 54 has only two stable operating states, namely an "on" state, in which it directs impinging light via a first objective 58 towards an optical integrator 60, and an "off" state, in which it directs impinging towards a light absorbing surface 62.

The second mirror array 54 may be realized as a digital mirror device (DMD) , as they are commonly used in beamers, for example. Such devices may comprise up to several million mi- cromirrors that can be switched between the two operating states many thousands times per second.

Similar to the pupil forming unit 36, the spatial light modulator 52 further comprises a prism 64 having an entrance sur- face 65 that is arranged perpendicular to the optical axis OA and a first plane surface 66a and a second plane surface 66b that are both inclined with respect to the optical axis OA of the illumination system 12. At these inclined surfaces 66a, 66b impinging light is reflected by total internal reflec- tion. The first surface 66a reflects the impinging light towards the micromirrors 56 of the second mirror array 54, and the second surface 66b directs the light reflected from the micromirrors 56 towards a surface 68 of the prism 64. If all micromirrors 56 of the second mirror array 54 are in their "on" state, the second mirror array 54 has substantially the effect of a plane beam folding mirror. However, if one or more micromirrors 56 are switched to their "off" state, the spatial irradiance distribution of the light emerging from the mirror plane 57 is modified. This can be used, in a manner that will be explained further below in more detail, to produce a field dependent modification of the angular light distribution on the mask 16. As it already has been mentioned above, the light emerging from the prism 64 passes through the first objective 58 and impinges on the optical integrator 60. Since the light passing through the first objective 58 is almost collimated, the first objective 58 may have a very low numerical aperture (for example 0.01 or even below) and thus can be realized with a few small spherical lenses. The first objective 58 images the mirror plane 57 of the spatial light modulator 52 onto the optical integrator 60.

The optical integrator 60 comprises, in the embodiment shown, a first array 70 and a second array 72 of optical raster elements 74. FIG. 6 is a perspective view of the two arrays 70, 72. Each array 70, 72 comprises, on each side of a support plate, a parallel array of cylinder lenses extending along the X and the Y direction, respectively. The volumes where two cylinder lenses cross .form optical raster elements 74.

Thus each optical raster element 74 may be regarded as a mi- crolens having cylindrically curved surfaces. The use of cylinder lenses is advantageous particularly in those cases in which the refractive power of the optical raster elements 74 shall be different along the X and the Y direction. A different refractive power is necessary if the square irradiance distribution on the optical integrator 60 shall be transformed into a slit-shaped illumination field 14, as this is usually the case. The surface of the optical raster elements 74 pointing towards the spatial light modulator 52 will be referred to in the following as light entrance facet 75.

The optical raster elements 74 of the first and second array 70, 72 respectively, are arranged one behind the other in such a way that one optical raster element 74 of the first array 70 is associated in a one to one correspondence with one optical raster element 74 of the second array 72. The two optical raster elements 74, which are associated with each other, are aligned along a common axis and define an optical channel. Within the optical integrator 60 a light beam which propagates in one optical channel does not cross or superimpose with light beams propagating in other optical channels. Thus the optical channels associated with the optical raster elements 74 are optically isolated from each other. In this embodiment a pupil plane 76 of the illumination system 12 is located behind the second array 72; however, it may equally be arranged in front of it . A second condenser 78 establishes a Fourier relationship between the pupil plane 76 and a field stop plane 80 in which an adjustable field stop 82 is arranged.

The field stop plane 80 is optically conjugated to a raster field plane 84 which is located within or in close proximity to the light entrance facets 75 of the optical integrator 60. This means that each light entrance facet 75 in the raster field plane 84 is imaged onto the entire field stop plane 80 by the associated optical raster element 74 of the second array 72 and the second condenser 78. The images of the irradi- ance distribution on the light entrance facet 75 within all optical channels superimpose in the field stop plane 80, which results in its very uniform illumination of the mask

16. Another. way of describing the uniform illumination of the mask 16 is based on the irradiance distribution which is produced by each optical channel in the pupil plane 76. This ir- radiance distribution is often referred to as secondary light source. All secondary light sources commonly illuminate the field stop plane 80 with projection light from different directions. If a secondary light source is "dark", no light im- pinges on the mask 16 from a (small) range of directions that is associated with this particular light source. Thus it is possible to set the desired angular light distribution on the mask 16 by simply switching on and off the secondary light sources formed in the pupil plane 76. This is accomplished by changing the irradiance distribution on the optical integrator 60 with the help of the pupil forming unit 36.

The field stop plane 80 is imaged by a second objective 86 onto a mask plane 88 in which the mask 16 is arranged with the help of a mask stage (not shown) . The adjustable field stop 82 is also imaged on the mask plane 88 and defines at least the short lateral sides of the illumination field 14 extending along the scan direction Y.

The pupil forming unit 36 and the spatial light modulator 52 are connected to a control unit 90 which is, in turn, con- nected to an overall system control 92 illustrated as a personal computer. The control unit 90 is configured to control the mirrors 40 of the pupil forming unit 36 and the micro- mirrors 56 of the spatial light modulator 52 in such a manner that the angular irradiance distribution in the mask plane 88 is varies in the intended manner within the illumination field 14 during a scan cycle. In the following section the function and control of the illumination system will be described . IV.

Function and Control of the Illumination System 1. Pupil forming

FIG. 7 schematically illustrates how the pupil forming unit 36 produces an irradiance distribution on the micromirrors 56 of the spatial light modulator 52. For the sake of simplicity the prisms 46, 64 are not shown.

Each mirror 40 of the first mirror array 38 is configured to illuminate a spot 94 on the mirror plane 57 of the spatial light modulator 52 at a position that is variable by changing a deflection angle produced by the respective mirror 40. Thus the spots 94 can be freely moved over the mirror plane 57 by tilting the mirrors 40 around their tilt axes. In this way it is possible to produce a wide variety of different irradiance distributions on the mirror plane 57. The spots 94 may also partly or completely overlap, as this is shown at 95. Then also graded irradiance distributions may be produced.

FIG. 8 is a perspective view, similar to FIG. 5, on the second mirror array 54 contained in the spatial light modulator 52. Here it is assumed that the pupil forming unit 36 has produced an irradiance distribution on the second mirror array 54 that consists of two square poles 27 each extending exactly over 6 x 6 micromirrors 56. The poles 27 are arranged point-symmetrically along the X direction. The objective 58 forms an image of this irradiance distribution on the light entrance facets 75 of the optical integrator 60, as this is shown in FIG. 9. Here it is assumed that all micromirrors 56 are in the "on"-state so that the irradiance distribution formed on the second mirror array 54 is identically reproduced (apart from a possible scaling due to a magnification of the objective 58) on the light entrance facets 75 of the optical integrator 60. The regular grid shown on the light entrance facets 75 represent an image of the borderlines of the micromirrors 56, but this image does not appear outside the poles 27 and is shown in FIG. 9 for illustrative reasons only. 2. Field dependency

Since the light entrance facets 75 are located in the raster field plane 84, the irradiance distribution on the light entrance facets 75 is imaged, via the optical raster elements 74 of the second array 72 and the second condenser 78, on the field stop plane 80.

This will now be explained with reference to FIG. 10 which is an enlarged and not to scale cut-out from FIG. 3. Here only two pairs of optical raster elements 74 of the optical integrator 60, the second condenser 78 and the intermediate field stop plane 80 are shown schematically.

Two optical raster elements 74 that are associated with a single optical channel are referred to in the following as first microlens 101 and second microlens 10.2. The microlenses 101, 102 are sometimes referred to as field and pupil honey- comb lenses. Each pair of microlenses 101, 102 associated with a particular optical channel produces a secondary light source 106 in the pupil plane 76. In the upper half of FIG. 10 it is assumed that converging light bundles Lla, L2a and L3a illustrated with solid, dotted and broken lines, respec- tively, impinge on different points of the light entrance facet 75 of the first microlens 101. After having passed the two microlenses 101, 102 and the condenser 78, each light bundle Lla, L2a and L3a converges to a focal point Fl, F2 and F3, respectively. From the upper half of FIG. 10 it becomes clear that points, where light rays impinge on the light entrance facet 75, and points where these light rays pass the field stop plane 80 (or any other conjugated field plane), are optically conjugate. The lower half of FIG. 10 illustrates the case when collimat- ed light bundles Lib, L2b and L3b impinge on different regions of the light entrance facet 75 of the first microlens 101. This is the more realistic case because the light im- pinging on the optical integrator 60 is usually substantially collimated. The light bundles Lib, L2b and L3b are focused in a common focal point F located in the second microlens 102 and then pass, now collimated again, the field stop plane 80. Again it can be seen that, as a result of the optical conju- gation, the region where a light bundle Lib, L2b and L3b impinges on the light entrance facet 75 corresponds to the region which is illuminated in the field stop plane 80. As a matter of course, these considerations apply separately for the X and the Y direction if the microlenses 101, 102 have refractive power both along the X and Y direction.

Therefore each point on a light entrance facet 75 directly corresponds to a conjugated point in the intermediate field stop plane 80 (and hence in the illumination field 14 on the mask 16) . If it is possible to selectively influence the ir- radiance on a point on a light entrance facet 75, it is thus possible to influence the irradiance of a light ray that impinges on the conjugated point in the illumination field 14 from a direction that depends on the position of the light entrance facet 75 with respect to the optical axis OA of the illumination system. The larger the distance between the light entrance facet 75 from the optical axis OA is, the larger is the angle under which said light ray impinges on the point on the mask 16.

3. Modifying irradiance on light entrance facets In the illumination system 12 the spatial light modulator 52 is used to modify the irradiance on points on the light entrance facets 75. In FIG. 9 it can be seen that each pole 27 extends over a plurality of small areas that are images of the raicromirrors 56. If a micromirror is brought into an "off" state, the conjugated area on the light entrance facet 75 will not be illuminated, and consequently no projection light will impinge on a conjugated area on the mask from the (small) range of directions that is associated with this particular light entrance facet 75.

This will be explained in more detail with reference to FIGS. 11a and lib which are top views on the micromirrors 56 of the spatial light modulator 52 and on the light entrance facets 75 of the optical integrator 60, respectively.

The thick dotted lines on the second mirror array 54 divide its mirror plane 57 into a plurality of object areas 110 each comprising 3x3 micromirrors 56. The objective 58 forms an image of each object area 110 on the optical integrator 60.

This image will be referred to in the following as image area 110'. Each image area 110' completely coincides with a light entrance facet 75, i. e. the image areas 110' have the same shape, size and orientation as the light entrance facets 75 and are completely superimposed on the latter. Since each ob- ject area 110 comprises 3x3 micromirrors 56, the image areas 110' also comprise 3x3 images 56' of micromirrors 56.

In FIG. 11a there are eight object areas 110 that are completely illuminated by the pupil forming unit 36 with projection light. These eight object areas 110 form the two poles 27. It can be seen that in some of the object areas 110 one, two or more micromirrors 56d represented as black squares have been controlled by the control unit 90 such that they are in an "off-state in which impinging projection light is not directed towards the objective 58, but towards the ab- sorber 62. By switching micromirrors between the "on" and the "off" state it is thus possible to variably prevent projection light from impinging on corresponding regions within the image areas 110' on the light entrance facets 75, as this is shown in FIG. lib. These regions will be referred to in the following as dark spots 56d'.

As has been explained above with reference to FIG. 10, the irradiance distribution on the light entrance facets 75 is imaged on the field stop plane 80. If a light entrance facet 75 contains one or more dark spots 56d' , as this is illustrated in the upper portion of FIG. 12, the irradiance distribution produced in the mask plane 88 by the associated op^¬ tical channel will have dark spots at certain X positions, too. If a point on a mask passes through the illumination field 14, the total scan integrated irradiance will thus depend on the X position of the point in the illumination field 14, as this is shown in the graph of FIG. 13. Points in the middle of the illumination field 14 will experience the high- est scan integrated irradiance, because they do not pass through dark spots, and points at the longitudinal ends of the illumination field 14 will receive total irradiances that are reduced to different extents. Thus the field dependency of the angular light distribution on the mask 16 and also the spatial irradiance distribution can be modified by selectively bringing one or more micromirrors 56 of the spatial light modulator 52 from an "on"-state into the "off-state.

In a foregoing it has to be assumed that each object area 110, which is imaged on one of the light entrance facets 75, contains only 3x3 micromirrors 56. Thus the resolution along the cross-scan direction X that can be used to modify the field dependency of the angular light distribution is relatively coarse. If the number of micromirrors 56 within each object area 110 is increased, this resolution can be im- proved.

FIG. 14 illustrates a top view on one of the light entrance facets 75 for an embodiment in which 20x20 micromirrors 56 are contained in each object area 110. Then more complicated scan integrated irradiance distributions along the X direction can be achieved on the mask 16, as this is illustrated in the graph shown in FIG. 15.

V.

Reduction of Edge Placement Errors

1. CD Uniformity

In a first step one attempts to improve the CD uniformity by carefully defining the irradiance in the illumination field 14 along the cross scan direction X. Since this approach is known in the art as such, it will not be described in further detail here. Then the micromirrors 56 are controlled so that the target field dependency of the irradiance is obtained in the illumination field 14. Since the impact of the projection objective 20 on the field dependency of the irradiance at wa- fer level cannot be easily predicted, it may be necessary to repeat this process several times. After some iterations the field dependent variations of the CD usually reaches a minimum.

After the determination of the target field dependency of the irradiance, the target field dependency of the angular irradiance distribution has to be determined.

Since it is known how various deficiencies of the illumination setting influence the critical dimension at wafer level, it is in turn possible to determine which modifications of an original field dependency of the angular irradiance distribution have to be produced for reducing pattern and field dependent variations of the critical dimension. Usually only a slight field dependency of the angular irradiance distribution is necessary to reduce variations of the critical dimen- sion as they typically occur. Then the micromirrors 56 are controlled such that the target field dependency of the angular irradiance distribution is produced in the illumination field 14. Since the setting of each micro-mirror between an "on" and "off" state always has an impact not only on the field dependency of the angular irradiance distribution, but also on the field dependency of the irradiance, the optimization of the field dependency of the irradiance and of the angular irradiance distribution may be performed in a single process. Experiments have demonstrated that the critical dimension variations may be reduced for dense line pitches almost by a factor of 2 if not only the irradiance, but also the angular irradiance distribution is optimized differently at different field positions. 2. Overlay Control

If field dependent overlay errors shall be corrected, a similar approach as outlined above may be used.

In a first step the field dependency of the overlay error is measured or simulated at wafer level. As it has been ex- plained further above with reference to FIG. 2, asymmetries in the angular irradiance distribution result in telecen- tricity errors. In that case the energetic center of the projection light impinges obliquely on the image points. This can be exploited to shift the image point literally by axial- ly displacing the wafer surface from its ideal position in the image plane.

This is illustrated in FIGS. 17a and 17b. In the upper part of FIG. 17a it is schematically illustrated how a telecentric light bundle 120 passes through the image plane 122 of the projection objective 20. In the middle part of FIG. 17b it can be seen that the image point 124 has its minimum diameter in the image plane 122. In a parallel plane 126 that is axi- ally displaced with respect to the image plane 122, the diameter of the image point 128 is larger, but the X and Y coordinates are not affected by this displacement (cf . lower part of FIG. 17a) . FIG. 17b illustrates the same constellation for the case of a light bundle 120' that is not telecentric. It can be seen that this does not affect the size and the position of the image point 124' in the image plane 122. But in the parallel plane 26 the image point is not only larger, but laterally displaced along the X direction.

By carefully introducing asymmetries in the angular irradi- ance distribution and displacing the wafer slightly along the optical axis, it is thus possible to produce field dependent lateral shifts of the image which can be used to correct a field dependency of edge placement errors. Since any defo- cused arrangement of the wafer is accompanied by a reduction of image contrast, a trade-off has to be found between the correction of field dependent edge placement errors on the one hand and a contrast reduction on the other hand. VI.

EUV

In the foregoing the invention has been described with reference to the projection exposure apparatus 10 which uses VUV projection light. However, it is also possible to use the concepts outlined above in EUV projection apparatus.

WO 2009/100856 Al describes an EUV illumination system which makes it possible to produce a desired field dependency of the irradiance and the angular irradiance distribution. Also in that case small mirrors have to be controlled individually in order to achieve the desired field dependencies. VII.

Multi-Device Die

FIG. 18 is a schematic illustration similar to FIG. 2 of a mask 16 which may be used to produce different integrated circuits or other devices on a single die. To this end the mask 16 comprises three first pattern areas 181a, 181b, 181c and three second pattern areas 182a, 182b, 182c that are arranged one behind the other along the scan direction Y. The first and second pattern areas differ from each other, in the simplified embodiment shown, by the density of line features 19 that extend along the Y direction.

Here it is assumed that the first pattern areas 181a, 181b, 181c are illuminated with an angular irradiance distribution which corresponds to a dipole setting. The pupil 26a there- fore comprises two poles 27a that are spaced apart along the cross-scan direction X.

The second pattern areas 182a, 182b, 182c are illuminated with projection light having an angular irradiance distribution which corresponds to a combination of a dipole setting and a conventional setting. Therefore the exit pupil 26b associated with light bundles impinging on the second pattern areas comprises not only the two poles 27b, but also a central pole 27b' . The illumination setting associated with the exit pupils 26b therefore completely incorporates the illumi- nation setting associated with the exit pupil 26a.

The micromirrors 56 of the illumination system 12 may be controlled so that not only the exit pupils 26a, 26b are produced at the respective field points in the illumination field 14. The control scheme also corrects field dependent edge placement errors by slightly modifying the exit pupils 26a, 26b in the two halves of the illumination field 14 during the scan cycle. This complex task is possible because the micromirrors 56 can be controlled very quickly and reliably even for a very large number of micromirrors 56.

In other embodiments the illumination settings do not change abruptly, but are continuously transformed so that immediate illumination settings are produced at intermediate field points .

VII.

Important method steps

Important method steps of the present invention will now be summarized with reference to the flow diagram shown in FIG. 19.

In a first step SI, a mask, an illumination system and a projection objection objective is provided. The projection objective is configured to form an image of an object field, which is illuminated on the mask in a mask plane, on an image field positioned on a light sensitive surface.

In a second step S2 edge placement errors are determined at different field points in the image field.

In a third step S3 the mask is illuminated with projection light having an improved field dependency of the angular ir- radiance distribution so that the edge placement errors determined in step S2 are reduced.

Claims

A method of operating a microlithographic projection apparatus, comprising the following steps: a) providing

- a mask ( 16) ,

- an illumination system (12) that is configured to illuminate the mask, and

- a projection objective (20) that is configured to form an image of an object field (14), which is illuminated on the mask (16) in a mask plane, on an image field positioned on a light sensitive surface (22) ; b) determining edge placement errors at different field points in the image field; c) illuminating the mask (16) with projection light having an improved field dependency of the angular irradiance distribution, wherein the angular irradiance distribution according to the improved field dependency varies over the object field (14) in such a way that the edge placement errors determined in step b) are reduced at the different field points.

The method of claim 1, wherein step b) comprises the steps of - illuminating the mask (16) with projection light having an original field dependency of the angular irradiance distribution; and

- simulating or measuring the edge placement errors on the light sensitive surface at the different field points ; and wherein step c) comprises the step of changing the original field dependency of the angular irradiance distribution such that the improved field dependency of the angular irradiance distribution is obtained.

The method of any of the preceding claims, wherein step c) comprises the step of illuminating the mask (16) with projection light having an improved field dependency of the irradiance, wherein the irradiance varies over the object field in such a way that the edge placement errors determined in step b) is reduced at the different field points.

The method of claim 3, wherein step b) comprises the steps of

- illuminating the mask (16) with projection light

having an original field dependency of the irradiance; and

- simulating or measuring the edge placement errors on the light sensitive surface at the different field points; _/ and wherein step c) comprises the step of changing the original field dependency of the irradiance such that the improved field dependency of the irradiance is obtained . The method of any of the preceding claims, wherein the mask (16) has a portion that comprises a uniform mask pattern, and wherein the angular irradiance distribution according to the improved field dependency of the angular irradiance distribution varies over an area of the object field that coincides, at least at a moment during step c) , with said portion.

The method of any of the preceding claims, wherein the mask comprises a non-uniform mask pattern that has locally varying properties, and wherein the improved angular irradiance distribution is adapted to the locally varying properties of the mask pattern.

The method of any of the preceding claims, wherein at least at some field points the angular irradiance distribution according to the improved field dependency is non-telecentric, and wherein at least one of the mask and the light sensitive surface is displacing along an optical axis of the projection objective before step c) .

The method of any of the preceding claims, wherein the illumination system provided in step a) comprises: an optical integrator (60) configured to produce a plurality of secondary light sources (106) located in a pupil plane (76) of the illumination system, wherein the optical integrator (60) comprises a plurality of light entrance facets (75) each being associated with one of the secondary light sources (106), and wherein images of the light entrance facets at least substantially superimpose in the mask plane, a spatial light modulator (52) having a light exit surface (57) and configured to transmit or to reflect impinging projection light in a spatially resolved manner, an objective (58) that images the light exit surface (57) of the spatial light modulator (52) onto the light entrance facets (75) of the optical integrator (60) , wherein in step c) the spatial light modulator is controlled such that the improved angular irradiance distribution is obtained in the mask plane.

An illumination system of a microlithographic projection apparatus (10) , comprising a) a pupil plane (76) , b) an optical integrator (60) configured to produce a plurality of secondary light sources (106) in the pupil plane (76) , wherein the optical integrator (60) comprises a plurality of light entrance facets (75) each being associated with one of the secondary light sources (106), c) a spatial light modulator (52) that has a light exit surface (57) and is configured to transmit or to reflect impinging projection light in a spatially resolved manner, d) a pupil forming unit (36) that is configured to direct projection light on the spatial light modulator, e) an objective (58) that images the light exit surface (57) of the spatial light modulator (52) on- to the light entrance facets (75) of the optical integrator (60), f) a control unit (90) configured to control the pupil forming unit (36) and the spatial light modulator (52) such that the mask is illuminated with projection light having an improved field dependency of the angular irradiance distribution, wherein the angular irradiance distribution according to the improved field dependency varies over the object field in such a way that edge placement errors that vary over the image field are reduced.

The illumination system of claim 9, wherein the control unit is configured to control the pupil forming unit (36) and the spatial light modulator (52) such that the method according to any of claims 2 to 8 is carried out .

A microlithographic projection apparatus, comprising a) a mask (16) , b) an illumination system (12) that is configured to illuminate the mask, and c) a projection objective (20) that is configured to form an image of an object field (14), which is illuminated on the mask in a mask plane, on an image field positioned on a light sensitive surface; d) means (38, 54, 60, 90) for illuminating the mask with projection light having an improved field dependency of the angular irradiance distribution, wherein the angular irradiance distribution according to the improved field dependency varies over the object field in such a way that edge placement errors that vary over the image field are reduced.