U.S. patent application number 10/460765 was filed with the patent office on 2004-12-16 for method for high precision printing of patterns.
This patent application is currently assigned to Micronic Laser Systems AB. Invention is credited to Sandstrom, Torbjorn.
Application Number | 20040251430 10/460765 |
Document ID | / |
Family ID | 33511079 |
Filed Date | 2004-12-16 |
United States Patent
Application |
20040251430 |
Kind Code |
A1 |
Sandstrom, Torbjorn |
December 16, 2004 |
METHOD FOR HIGH PRECISION PRINTING OF PATTERNS
Abstract
An aspect of the present invention includes a method to print
pattern with improved edge acuity. In one embodiment a method for
printing fine patterns comprising the actions of: providing an SLM
and providing a pixel layout pattern with different categories of
modulating elements, the categories differing in the phase of the
complex amplitude. Other aspects of the present invention are
reflected in the detailed description, figures and claims.
Inventors: |
Sandstrom, Torbjorn; (Pixbo,
SE) |
Correspondence
Address: |
HAYNES BEFFEL & WOLFELD LLP
P O BOX 366
HALF MOON BAY
CA
94019
US
|
Assignee: |
Micronic Laser Systems AB
Taby
SE
|
Family ID: |
33511079 |
Appl. No.: |
10/460765 |
Filed: |
June 12, 2003 |
Current U.S.
Class: |
250/492.2 |
Current CPC
Class: |
G03F 7/70291 20130101;
G03F 7/70283 20130101 |
Class at
Publication: |
250/492.2 |
International
Class: |
G21K 005/04 |
Claims
We claim as follows:
1. A method for printing fine patterns with high precision,
comprising the actions of: providing an SLM having an array of
modulator elements, providing an electromagnetic radiation source
to illuminate said SLM with partially coherent illumination with a
coherence length that is larger than half the pitch of the
modulating elements in the SLM, creating a negative complex
amplitude with at least one modulator element.
2. The method according to claim 1, wherein said radiation source
is a pulsed laser source.
3. The method according to claim 1, wherein said electromagnetic
radiation source having a wavelength in the region of UV, EUV, DUV,
VUV.
4. The method according to claim 1, wherein said modulator elements
are operated an analog mode.
5. The method according to claim 1, wherein said SLM is a
reflective SLM.
6. The method according to claim 1, wherein said negative complex
amplitude is more negative than -0.218.
7. The method according to claim 1, wherein said negative complex
amplitude is more negative than -0.5.
8. A method for printing fine patterns, comprising the actions of:
providing an SLM, providing a data path including a rasterizer,
rasterizing an input pattern to a grayscale bitmap, providing an
edge-sharpening filter operating on the bitmap.
9. The method according to claim 8, wherein said SLM is operated in
an analog mode.
10. The method according to claim 8, wherein said pattern is
printed using a multipass writing strategy.
11. The method according to claim 8, wherein said edge sharpening
filter is a convolution filter.
12. The method according to claim 8, wherein said edge sharpening
filter is a rule based filter.
13. A method for printing fine patterns comprising the actions of
providing an SLM and a partially coherent optical system, providing
a data path including a rasterizer, rasterizing an input pattern to
a grayscale bitmap, providing an edge-sharpening filter operating
on the bitmap, wherein said edge-sharpening filter producing pixels
with negative complex amplitude.
14. The method according to claim 13, wherein said SLM is operated
in an analog mode.
15. The method according to claim 13, wherein said pattern is
printed using multipass writing strategy.
16. The method according to claim 13, wherein said edge sharpening
filter is a convolution filter.
17. The method according to claim 13, wherein said edge sharpening
filter is a rule based filter.
18. A method for printing fine pattern comprising the actions of:
providing an SLM, providing a partially coherent optical system,
providing pixels with less reflection close to the axis.
19. The method according to claim 18, wherein said pixels have less
reflecting area close to the axis than further away.
20. The method according to claim 18, wherein said pixels are
essentially rectangular and provided with anti-reflective coating
close to the axis.
21. The method according to claim 18, wherein said pixels are
essentially rectangular and provided with radiation scattering
elements close to the axis.
22. A method for printing fine pattern comprising the actions of:
providing an SLM with an array of reflecting elements, providing a
surface profile for at least one reflecting element, said surface
profile having area elements with at least two heights.
23. The method according to claim 22, wherein said area elements
are 180 apart for a wavelength chosen to print said pattern.
24. A method for printing fine patterns comprising the actions of:
providing an SLM, providing a bitmap representing the input data,
separating said bitmap into two bitmaps, filtering the two bitmaps
using different filters, combining the two bitmaps to expose the
image.
25. The method according to claim 23, wherein said bitmaps are
combined into a single writing pass.
26. The method according to claim 23, wherein said bitmaps are
combined into a plurality of writing passes.
27. The method according to claim 23, wherein at least one filter
is a low-pass characteristics.
28. The method according to claim 23, wherein at least one filter
is a high-pass characteristics.
29. The method according to claim 23, wherein said filters act
differently in an X direction of the pattern and an Y direction of
the pattern.
30. The method according to claim 23, wherein in a multipass
writing strategy the optical settings are different for the
different filters.
31. The method according to claim 27, wherein in a multipass
writing strategy the optical settings are different for the
different filters.
32. The method according to claim 28, wherein in a multipass
writing strategy the optical settings are different for the
different filters.
33. A method for printing fine patterns comprising the actions of:
detecting edges in the bitmap, characterizing the steepness of the
transition of the edges, making digital edge enhancements based on
said steepness.
34. The method according to claim 33, wherein said input data is
having at least two-valued layers.
35. The method according to claim 33, wherein said input data
having features with different complex amplitudes.
36. The method according to claim 33, further comprising the action
of: rasterizing said features to multiple bitmaps and combining
said bitmaps to drive the SLM.
37. A method for printing fine patterns comprising the actions of:
providing an SLM having analog function with a range from positive
to negative complex amplitude reflection, providing a data path,
providing input data describing a pattern having extended areas
with at least three complex amplitude values and edges between
them, converting said data input to a bitmap of pixel values that
correspond to both negative and positive complex amplitude.
38. The method according to claim 37, further comprising the action
of: rasterizing each two-valued layer bitmap and combining said
bitmaps to drive the SLM.
39. A method for printing fine patterns comprising the actions of:
providing an SLM, providing a pixel layout pattern with different
categories of modulating elements, the categories differing in the
phase of the complex amplitude.
40. The method according to claim 39, wherein said categories are
two categories, a first with phase 0 and a second with a 180
degrees phase.
41. The method according to claim 39, wherein said categories are
three categories, a first with a phase 0, a second with a phase
120, and a third with a phase 240 degrees.
42. The method according to claim 39, wherein said categories are
four categories a first with phase 0, a second with phase 90, a
third with phase 180, and a fourth with phase 270 degrees.
Description
RELATED APPLICATION
[0001] This application is related to the U.S. patent application
Ser. No. ______, entitled "Methods and Systems for Improved
Boundary Contrast" by inventor Torbjorn Sandstrom, filed on the
same day as this application, which is hereby incorporated by
reference.
FIELD OF THE INVENTION
[0002] The invention relates to the printing of patterns with high
precision, in particular to printing of nicrolithographic pattern
such as patterns on photomasks and wafers. The invention may also
be applied to other printing, such as for the formation of optical
devices, electronic interconnects and even to decorative printing
and security printing.
[0003] The invention is particularly suited to but not limited to
optical printing using partially coherent light, such as from
excimer and atomic lasers and from EUV light sources. In a
preferred embodiment it is applied to a maskless scanner for
exposure of patterns onto semiconductor wafers without the need for
reticles or masks.
BACKGROUND OF THE INVENTION
[0004] In the past, integrated circuits have been manufactured more
or less solely by using a number of masks or reticles comprising a
pattern of a layer in said integrated circuit. In today's
integrated circuits the number of layers could be larger than 30.
Said Masks or reticles may be prepared in lithographical manner by
using for example electron beams or laser beams for exposing a
layer of material sensitive for the type of beam chosen. The mask
material is most commonly transmissive on top of one of its sides a
thin layer of opaque material is attached. In said thin material
the pattern of one layer of said integrated circuit is created. The
mask has typically N times larger pattern than the pattern to be
printed on the semi-conducting substrate for forming said
integrated circuit. The reduction in size is performed in a
stepper, which uses the mask(s) for forming said integrated
circuit.
[0005] More recently, the need to manufacture integrated circuits
by means other than using a conventional mask has developed for a
number of reasons, for example the price of manufacturing mask(s)
has increased due to its complexity to manufacture, small-scale
development which needs very small series of integrated circuits,
etc.
[0006] Unfortunately, all of the present known techniques for
forming integrated circuits without using conventional masks or
reticles have drawbacks and limitations.
[0007] For example, most direct-writers known in the art are based
on electron beams, typically so called shaped beams, where the
pattern is assembled from flashes, each defining a simple
geometrical figure. Other systems are known which use raster
scanning of Gaussian beams. By using a conventional mask writer,
which uses beams of electrons or laser be ams for forming the
pattern on a workpiece, is limited to relatively low scanning
speeds, and, perhaps worst of all, can only scan a single
dimension.
[0008] SLM writers disclosed in other patent applications, such as
WO 01/18606 and U.S. patent application Ser. No. 09/954,721 by the
same assignees as the present invention and hereby incorporated by
reference is related to raster scanning in the sense that it
permits a bitmap pattern, but distinct by printing an entire frame
of pattern in one flash instead of building the pattern from
individual pixels.
[0009] A spatial light modulator (SLM) comprises a number of
modulator elements, which can be set in a desired way for forming a
desired pattern. Reflective SLMs may be exposed to any kind of
electromagnetic radiation, for example DUV or EUV for forming the
desired pattern on the mask.
[0010] The same assignee has in a number of previous patent
applications, for instance WO 99/45440 and WO 99/45441, disclosed
pattern generator technology for precision printing of submicron
patterns. Typically the embodiments taught in said applications use
SLMs with analog modulation. The modulating elements are
micromechanical mirrors that are capable of gradually move from a
resting to a fully actuated state in response to an electronic
drive signal, and the elements form one or two-dimensional arrays
of modulating elements. A pattern defined in an input database is
rasterized to a bitmap were each pixel can have several states
between a lightest and a darkest state.
[0011] What is needed is a method and apparatus, which creates
pattern on a workpiece using a programmable reticle or mask, such
as a spatial light modulator, capable to create patterns with high
feature edge acuity. What is also needed is a method and apparatus
capable to pattern feature boundaries with high accuracy of
placement.
SUMMARY OF THE INVENTION
[0012] Accordingly, it is an object of the present invention to
provide a method of patterning a workpiece, which overcomes or at
least reduces the above-mentioned problem of creating fine patterns
with high acuity and high accuracy of placement of feature
boundaries.
[0013] This object, among others, is according to a first aspect of
the invention attained by a method for printing fine patterns with
high precision Said method comprising the actions of providing an
SLM having an array of modulator elements, providing an
electromagnetic radiation source, collimating radiation from said
radiation source to create partially coherent illumination of said
SLM with a coherence length that is larger than half the pitch of
the modulating elements in the SLM, creating a negative complex
amplitude with at least one modulator element.
[0014] Other aspects of the present invention are reflected in the
detailed description, figures and claims.
BRIEF DESCRIPTION OF THE DRAWINGS
[0015] FIG. 1 depicts the general layout of an SLM pattern
generator.
[0016] FIG. 2 depicts in a perspective view a prior art square
mirror.
[0017] FIG. 3 depicts the reflected intensity and complex amplitude
as function of the tilt angle of the mirror element.
[0018] FIG. 4 depicts the real and imaginary part of the complex
amplitude for the mirror element depicted in FIG. 2.
[0019] FIG. 5a-b depict the mirror comprising reference surface and
surface element.
[0020] FIG. 6a-b illustrate the relation between coherence length
and angular spread
[0021] FIG. 7a illustrates a prior art method of patterning a
substrate only using positive complex amplitude.
[0022] FIG. 7b illustrates in inventive method of patterning a
substrate using positive and negative complex amplitudes.
[0023] FIG. 8a-c depict different ways of creating data to be fed
to an SLM.
[0024] FIG. 9a depicts a clear feature in vector data.
[0025] FIG. 9b depicts a rasterized representation of the vector
data in FIG. 9a with the off grid edge enhanced by extra exposure
on the new bright pixel and negative black on the dark pixel.
[0026] FIG. 9c depicts a screen shot of resist edges from a Solid-C
simulator.
[0027] FIG. 9d depicts a screen shot showing the position of the
center of the clear feature vs. dose with and without enhancement
of the off-grid edge.
[0028] FIG. 10a depicts an alternating phase shift mask.
[0029] FIG. 10b depicts SLM pixels with complex amplitude
reflectivity corresponding to the alternating phase shift mask in
FIG. 10a.
[0030] FIG. 11a depicts a 2.times.2 array of mirrors with different
phase characteristics.
[0031] FIG. 11b illustrates a complex amplitude reflectivity as
function of tilt of the mirror.
[0032] FIG. 12a-e illustrate 2.times.2 arrays of mirrors as
depicted in FIG. 11a only differing in the way they are tilted.
[0033] FIG. 13a-13d illustrate the correspondence between different
areas of masks/reticles and different areas of a SLM.
[0034] FIG. 14a depicts an embodiment of a mirror.
[0035] FIG. 14b depicts the complex amplitude trajectory of the
mirror in FIG. 14a.
[0036] FIG. 14c depicts a pixel transfer function of the mirror in
FIG. 14a.
[0037] FIG. 15a depicts an embodiment of a mirror.
[0038] FIG. 15b depicts the complex amplitude trajectory of the
mirror in FIG. 15a.
[0039] FIG. 15c depicts a pixel transfer function of the mirror in
FIG. 15a.
[0040] FIG. 16a depicts an embodiment of a mirror.
[0041] FIG. 16b depicts the complex amplitude trajectory of the
mirror in FIG. 16a.
[0042] FIG. 16c depicts a pixel transfer function of the mirror in
FIG. 16a.
[0043] FIG. 17a depicts an embodiment of a mirror.
[0044] FIG. 17b depicts the complex amplitude trajectory of the
mirror in FIG. 17a.
[0045] FIG. 17c depicts a pixel transfer function of the mirror in
FIG. 17a.
[0046] FIG. 18a depicts an embodiment of a mirror.
[0047] FIG. 18b depicts the complex amplitude trajectory of the
mirror in FIG. 18a.
[0048] FIG. 18c depicts a pixel transfer function of the mirror in
FIG. 18a.
[0049] FIG. 19a depicts an embodiment of a mirror.
[0050] FIG. 19b depicts the complex amplitude trajectory of the
mirror in FIG. 19a.
[0051] FIG. 19c depicts a pixel transfer function of the mirror in
FIG. 19a.
[0052] FIG. 20a depicts an embodiment of a mirror.
[0053] FIG. 20b depicts the complex amplitude trajectory of the
mirror in FIG. 20a.
[0054] FIG. 20c depicts a pixel transfer function of the mirror in
FIG. 20a.
[0055] FIG. 21a depicts an embodiment of a mirror.
[0056] FIG. 21b depicts the complex amplitude trajectory of the
mirror in FIG. 21a.
[0057] FIG. 21c depicts a pixel transfer function of the mirror in
FIG. 21a.
[0058] FIG. 22a depicts an embodiment of a mirror.
[0059] FIG. 22b depicts the complex amplitude trajectory of the
mirror in FIG. 22a.
[0060] FIG. 22c depicts a pixel transfer function of the mirror in
FIG. 22a.
[0061] FIG. 22d illustrates an array of pixel as illustrated in
FIG. 22a.
[0062] FIG. 23a depicts an embodiment of a mirror.
[0063] FIG. 23b depicts the complex amplitude trajectory of the
mirror in FIG. 23a.
[0064] FIG. 23c depicts a pixel transfer function of the mirror in
FIG. 23a.
[0065] FIG. 24a depicts an off grid filter implementation.
[0066] FIG. 24b depicts an off grid filter.
[0067] FIG. 25a depicts an embodiment of a mirror.
[0068] FIG. 25b depicts the complex amplitude trajectory of the
mirror in FIG. 25a.
[0069] FIG. 25c depicts a pixel transfer function of the mirror in
FIG. 25a.
[0070] FIG. 26a depicts an embodiment of a mirror.
[0071] FIG. 26b depicts the complex amplitude trajectory of the
mirror in FIG. 26a.
[0072] FIG. 26c depicts a pixel transfer function of the mirror in
FIG. 26a.
[0073] FIG. 27a depicts an embodiment of a mirror.
[0074] FIG. 27b depicts the complex amplitude trajectory of the
mirror in FIG. 27a.
[0075] FIG. 27c depicts a pixel transfer function of the mirror in
FIG. 27a.
[0076] FIG. 28a depicts an embodiment of a mirror.
[0077] FIG. 28b depicts the complex amplitude trajectory of the
mirror in FIG. 28a.
[0078] FIG. 28c depicts a pixel transfer function of the mirror in
FIG. 28a.
[0079] FIG. 29a depicts an embodiment of a mirror.
[0080] FIG. 29b depicts the complex amplitude trajectory of the
mirror in FIG. 29a.
[0081] FIG. 29c depicts a pixel transfer function of the mirror in
FIG. 29a.
[0082] FIG. 30a depicts an embodiment of a mirror.
[0083] FIG. 30b depicts the complex amplitude trajectory of the
mirror in FIG. 30a.
[0084] FIG. 30c depicts a pixel transfer function of the mirror in
FIG. 30a.
[0085] FIG. 31a depicts an embodiment of a mirror.
[0086] FIG. 31b depicts the complex amplitude trajectory of the
mirror in FIG. 31a.
[0087] FIG. 31c depicts a pixel transfer function of the mirror in
FIG. 31a.
[0088] FIG. 32a depicts a slanted line and its rasterized pixel
representation
[0089] FIG. 32b depicts the exposure dose as a function of the
position of said slanted line.
[0090] FIG. 33a illustrates the contrast as a function of spatial
frequency for on-grid pixels and off-grid pixels without any grid
filter.
[0091] FIG. 33b illustrates the contrast as a function of spatial
frequency for on-grid pixels and off-grid pixels with an off-grid
filter.
[0092] FIG. 33c illustrates the contrast as a function of spatial
frequency for on-grid pixels and off-grid pixels with off-grid
filter and global edge enhancement.
[0093] FIG. 34 illustrates a diagram showing contrast as a function
of pixel number.
DETAILED DESCRIPTION
[0094] The following detailed description is made with reference to
the figures. Preferred embodiments are described to illustrate the
present invention, not to limit its scope, which is defined by the
claims. Those of ordinary skill in the art will recognize a variety
of equivalent variations on the description that follows.
[0095] Spatial light modulators come in two varieties, a deflection
type and a phase type. The differences between them may in a
particular case with micromirrors seem small but the phase SLM
extinguishes the beam in a specular direction by destructive
interference, while a pixel in a deflection SLM deflects the
specular beam geometrically to one side so that it misses an
aperture of an imaging lens. The deflection type SLM have pixels
which operate digitally, i.e., said pixels may be set to two states
only fully on and fully off. Said kind of pixels are said to be
operated in a digital mode. The phase type SLM have pixels which
operate in an analog mode, i.e., said pixels may be set to a
numerous states between fully off and fully on. In one embodiment
there are 63 states between fully off and fully on, i.e., 65 states
in total. A degree of deflection of a micro-mirror determines which
state said mirror would be in. All different states correspond to
different gray-levels, which may be used to move edges of features
to be printed.
[0096] FIG. 1 depicts the general layout of an SLM pattern
generator. Aspects of an SLM pattern generator are disclosed in the
related-pending patent applications identified above. The workpiece
to be exposed sits on a stage 112. The position of the stage is
controlled by precise positioning device, such as paired
interferometers 113.
[0097] The workpiece may be a mask with a layer of resist or other
exposure sensitive material or, for direct writing, it may be an
integrated circuit with a layer of resist or other exposure
sensitive material. In the first direction, the stage moves
continuously. In the other direction, generally perpendicular to
the first direction, the stage either moves slowly or moves in
steps, so that stripes of stamps are exposed on the workpiece. In
this embodiment, a flash command 108 is received at a pulsed
excimer laser source 107, which generates a laser pulse. This laser
pulse may be in the deep ultraviolet (DUV) or extreme ultraviolet
(EUV) spectrum range. The laser pulse is converted into an
illuminating light 106 by a beam conditioner or homogenizer.
[0098] A beam splitter 105 directs at least a portion of the
illuminating light to an SLM 104. The pulses are brief, such as
only 20 ns long, so any stage movement is frozen during the flash.
The SLM 104 is responsive to the datastream 101, which is processed
by a pattern rasterizer 102. In one configuration, the SLM has
2048.times.512 mirrors that are 16.times.16 .mu.m each and have a
projected image of 80.times.80 nm. It includes a CMOS analog memory
with a micro mechanical mirror formed half a micron above each
storage node.
[0099] The electrostatic forces between the storage nodes and the
mirrors actuate the mirrors. The device words in diffraction mode,
not specular reflectance, and needs to deflect the mirrors by only
a quarter of the wavelength (62 nm at 248 nm) to go from the fully
on-state to the fully off-state. To create a fine address grid the
mirrors are driven to on, off and 63 intermediate values. The
pattern is stitched together from millions of images of the SLM
chip. Flashing and stitching proceed at a rate of 1000 stamps per
second. To eliminate stitching and other errors, the pattern is
written four times with offset grids and fields. Furthermore, the
fields may be blended along the edges.
[0100] The mirrors are individually calibrated. A CCD camera,
sensitive to the excimer light, is placed in the optical path in a
position equivalent to the image under the final lens. The SLM
mirrors are driven through a sequence of known voltages and the
response is measured by the camera. A calibration function is
determined for each mirror, to be used for realtime correction of
the grey-scale data during writing. In the data path, the vector
format pattern is rasterized into grey scale images, with grey
levels corresponding to dose levels on the individual pixels in the
four writing passes. This image can then be processed using image
processing. The final step is to convert the image to drive
voltages for the SLM. The image processing functions are done in
real time using programmable logic. Through various steps that have
been disclosed in the related patent applications, rasterizer
pattern data is converted into values 103 that are used to drive
the SLM 104.
[0101] In this configuration, the SLM is a diffractive mode
micromirror device. A variety of micromirror devices have been
disclosed in the art. In an alternative configuration, illuminating
light could be directed through a micro-shutter device, such as in
LCD array or a micromechanical shutter.
[0102] An SLM pattern generator, such as a mask writer or direct
writer, that uses a grey-scale sampled image enables a variety of
enhancement schemes. The grey value of each pixel is an area sample
value of the pattern. Taking into account the imaging properties of
the tool and a desired response, such as a specific corner radius,
adjustments of the exposure values in a predetermined vicinity of a
corner feature can be used to mimic or match the properties of
another pattern generator, such as the exposed corner radius and
corner pull back. The adjustment recipe can be adapted to match,
for instance, another mask writer. To do this, exposed pattern
properties in resist on workpieces of the two pattern generators
can be compared. The comparison can be based on either simulation,
developed resist or latent images in resist. Comparison of the
exposed patterns allows adjustment of one or more process control
parameters until the exposed patterns essentially match.
[0103] Data is modified in the raster domain of at least one of the
pattern generators according to the process control parameters,
rather than modifying vector-based pattern data in the design
domain. The process control parameters may relate to corner feature
exposure properties.
[0104] A mirror consisting of an essentially square mirror plate
pivoting around an axis defined by torsion hinges in the plane of
the mirror, see FIG. 2, modulates the beam from fully-on to
fully-off. The fully-off state depends on the illumination of the
mirror. The illuminator defines an angular subtense, which in turn
determines the lateral coherence of the illumination light. The
lateral coherence is in this sense different from the temporal
coherence.
[0105] Temporal coherence usually means that the radiation comes
from a laser, but lateral coherence can be produced by any light
source made to illuminate a surface under a small enough angular
spread. This is well known in the art and described in text books
such as Born and Wolf: Principles of optics.
[0106] The notion of lateral coherence length is significant to
this discussion. The lateral coherence length is of the order of
the typical or center wavelength of the radiation divided by the
angular spread of the illuminating beam. Projectors known in prior
art (such as those used by Texas Instruments in their DLP
technology and Daewoo in their AMA projectors) have used a high
angular spread leading to a coherence length smaller than the size
of an individual mirror element, see FIG. 6a. With this type of
illumination each mirror acts as an independent specular reflector.
The pattern generators disclosed and made by the applicant do on
the other hand use a small angular spread in the illuminating beam,
giving a coherence length that is of the same order as a mirror
element or larger, see FIG. 6b. The effect is that different areas
of a mirror interact by interference and that destructive or
constructive interference effects also occur between mirror
elements. The two different types of projectors will be called
incoherent and partially coherent projectors respectively, the term
projectors in this case meaning a generic image-forming system
using an SLM, an illuminator and a projection system including a
spatial filter. Incoherent projector are defined by the property of
not forming a partially or fully coherent image, which can be due
to the illumination mode, but also to a superposition of pixels at
different times. The case where two or more fully coherent images
are superimposed sequentially is considered as partially
coherent.
[0107] Under illumination where the lateral coherence extends over
a full mirror, the mirror does not act as a simple analog light
valve any more, but a complex amplitude modulator. The complex
amplitude is related to the electric field of the radiation, while
the intensity is more akin to the energy density or energy flow. An
interesting property of the complex amplitude is that it can have a
negative sign, see curve 320 in FIG. 3, while the intensity (energy
flow) is always positive, see FIG. 310 in FIG. 3. With the
illumination scheme that produces laterally coherent light it is
possible for one light beamlet to cancel the light of another one.
The consequence is that suitably conditioned radiation can be added
to reduce the light intensity at a point were it is desired to be
dark, thereby improving contrast.
[0108] The square mirror 220 tilting along one of its axis 210 acts
as a specular mirror when it is parallel to the plane of the
surface. When it is successively tilted out of the plane the edges
move out of phase and become more and more destructive, giving
perfect extinction of the light when they have a phase shift of
+/-180 degrees in reflection, see point 330. But if the mirror is
tilted more they continue further into the negative and the entire
mirror gives negative complex amplitude. FIG. 3 shows this. The
complex amplitude reflectivity of a mirror R can be calculated as
the double integral over the mirror surface S 510 of the complex
amplitude reflection r of every surface element Ds 520. 1 R = S r *
S S S
[0109] The denominator is the reflecting area of the mirror. In a
more general case with varying reflectivity the expression can be
generalized to include differences between the surface elements. In
the most simple case with a perfectly reflecting surface the
complex amplitude reflectivity r of a surface element is
[0110] r=e.sup.i(4.pi.h/.lambda.), where h 530 is the height of the
mirror surface 510 above a reference surface 540. The reference
surface 540 can be chosen arbitrarily (and the complex amplitude
reflectivity R can be multiplied with an arbitrary but constant
phase factor e.sup.i.phi.) with no change in the physics. For
definiteness the reference surface 540 is chosen here to be the
plane through the hinges 550, giving R=1 for a flat non-tilted
mirror.
[0111] FIG. 3 shows the reflected intensity and complex amplitude
as functions of the tilt angle. With a symmetric mirror making a
perfect pivoting action around a symmetry axis the imaginary part
of R is always zero. The real part of R varies from 1.00 through
0.00 to a minimum of -0.22. For higher tilt angles it becomes
positive again and approaches R=0 in the limit, see FIG. 4. For the
square mirror the tilt for the first null R=0 occurs when the tilt
is half a wavelength from one side to the other of the mirror, see
point 330 in FIG. 3. It is easy to see why this gives zero: because
it is a reflective device the phase in the light beam varies from
-180 to +180 degrees. For each surface element with phase .alpha.
there is another surface element with phase .alpha.+180 degrees,
thus the reflection from every surface element is cancelled.
Incident energy is diffracted away from the specular direction and
does not find its way through at least one stop in projection
optics.
[0112] The pattern generators developed by the applicant have used
the reflectivity range 0<R<I to print a pattern, where R=0 is
used for area elements intended to be unexposed and R=1-.epsilon.
is used for exposed areas. The term .epsilon., which is typically
10%, is introduced to allow even exposure even in the presence of
some statistical variation in R from mirror to mirror.
[0113] A fine address grid, much finer than that given by the
mirrors, is created by giving mirrors at the edge intermediate
values. These values are interpolated between exposed and unexposed
mirror tilts, times a nonlinear function, the illumination table.
The illumination table is implemented as a look-up table that is
pre-computed or determined experimentally. The shape of the
illumination function depends on a number of factors, most
important on the projected mirror size compared to the optical
resolution and the angular spread of the illuminating
radiation.
[0114] In the incoherent projector the complex quantity R does not
have any meaning, since the surface integrals only have meaning for
lateral coherence lengths on the scale of the mirror size.
[0115] The quantity R is defined as a complex amplitude
reflectivity in a partially coherent projector, not as the normal
intensity reflectivity relevant in an incoherent projector. As
described above R is a complex number and may have any value as
long as .vertline.R.vertline..ltoreq.1. With the symmetrical
mirrors Im (R)=0, but R can still be negative, and does in fact do
so for tilts larger than half a wavelength. This can be used for
image enhancement that is not possible in incoherent projectors.
FIG. 4 depicts an example of a complex amplitude reflectivity curve
40.
[0116] One type of image enhancement is achieved by selecting the a
value of R<0 for areas that are intended to be unexposed. A
typical value is R=-0.15. This corresponds to an intensity
reflection of 2.25% and gives a background exposure of 2.25% is
areas that are intended to be unexposed. However, 2.25% is not
enough to cause the photo resist (or more generally the
photosensitive surface) to develop, since it has a development
threshold typically around 30%. But exposed features get crisper
edges since the -0.15 reflectivity, having phase 180 degrees,
cancels light with phase 0 degrees at the perimeter of the exposed
features. The dark areas get larger, the edge steeper and if the
size is compensated with more dose the edge enhancement is even
further enhanced.
[0117] FIG. 7a illustrates a prior art method of patterning
features using an SLM. SLM pixels inside a feature to be patterned,
hatched pixels in FIG. 7a, have a complex amplitude reflectivity R
equal to 0. Pixels outside said feature, non-hatched pixels in FIG.
7a, have a complex amplitude reflectivity being equal to 1. The
illustrated example in FIG. 7a, have feature edges coinciding with
a pixel grid of the SLM. For this reason pixel elements defining
the edge of the feature also have complex amplitude reflectivity
equal to 0. If, however the feature edge falls between the pixel
grid, said complex amplitude reflectivity will be any value in the
range of 0<R<1. The value of R is depending on the placement
of said edge.
[0118] In FIG. 7a a graph 710 represents the complex amplitude of
reflectivity R taken along a line A-A. FIG. 7b illustrates the
inventive method for creating features with increased edge acuity
and placement accuracy. Here the pixels inside the feature, hatched
pixels in FIG. 7b, have complex amplitude less than 0, i.e., a
negative value. A graph 720 represents the complex amplitude
reflectivity taken along line B-B. Inserted in FIG. 7b is also the
graph representing the intensity of reflectivity
.vertline.R.vertline..sup.2.
[0119] The use of negative R is analogous to the use of so-called
attenuated or embedded phase shifting masks in lithography. The
value of R to be selected at will between 0 and a minimum value. At
first sight is seems that the minimum value is -0.218. This
corresponds to 4.77% exposure, less than 6% attenuated masks used
in state-of-the-art lithography.
[0120] Closer analysis shows that it is not the maximum exposure
dose E that is creating the effect but the value of the complex
amplitude A in black areas relative to the complex amplitude in
bright ones. Disregarding again a constant phase factor together
with some prefactors that may be presert.
[0121] A=R*{square root}{square root over (E)}, where E is the
exposure dose. Using an R<1.00 as described above leads to a
higher exposure dose and the minimum also gets larger in
proportion.
[0122] The minimum value of A for the square mirror is 2 A min = A
exposed * - 0.218 ( 1 - )
[0123] If we choose .epsilon.=15% we get A.sub.min=-0.256. This
corresponds to an intensity of 6.6%, which is with a small margin
equivalent to the industry-standard 6% attenuated mask blanks. In
an SLM writer the dose and the mirror tilts are under software
control, so even larger .epsilon. can be used to get more negative
amplitude. The restrictions are twofold: first, increasing the dose
causes problems by itself, such as the creation of more stray
light. Second, imperfections in the mirrors get magnified. However,
these limitations are purely practical and the use of high
.epsilon. and strongly negative R cannot be ruled out
beforehand.
[0124] In previously implemented rasterizers the value of the pixel
of mirror at the feature edge has been calculated as an
interpolation between the exposed and unexposed value based on how
much of the pixel falls on the exposed feature. Before it is
converted to drive signals for the SLM.
[0125] Image modulator elements it is corrected through the
illumination table LUT as described above.
[0126] In a further improvement a digital filter (term taken in a
wide sense) is applied to the rasterized data to enhance edges and
corners. The filter can be designed and implemented in many ways:
linear or non-linear, based on rules or mathematical operations.
One of the simplest rules is that whenever a pixel has a neighbor
that is gray (i.e. has an intermediate value) the current pixel is
enhanced, so that a white pixel gets whiter (more positive) and a
black pixel gets blacker (less positive). In an incoherent
projector the range of pixel values is limited to zero to full
illumination, in the partially coherent projector the pixel value
has the range A.sub.min<A<A.sub.max where A.sub.min can be
negative.
[0127] FIG. 8a illustrates the way drive signals are converted
before being fed to the SLM according to a first embodiment. Vector
data is fractured and rasterized according to well-known principles
in the art. Edge filters are applied according to methods described
above and below. The illumination table and mirror look up table is
used before the final drive signal is created. In FIG. 8b another
embodiment is illustrated which uses two illumination tables
instead of one. By doing so better CD control may be achieved. FIG.
8c illustrates yet another embodiment which splits the rasterized
data into two parallel branches. A first branch uses a first and
second illumination table 1a, 1b, and a high pass filter. A second
branch uses a third and a fourth illumination table and a low pass
filter. Data split into high frequency and low frequency data. One
could also split the data into x and y branches, meaning that a
first branch is optimized for data only in a y direction and a
second branch optimizes data only in an x direction, where x and y
could be horizontal and vertical data.
[0128] The availability of negative pixel values in the partially
coherent projector gives more corrective power than positive-only.
In particular it makes it possible to improve both resolution and
contrast of fine lines.
[0129] The digital image enhancements are comparatively easy to
make in the bitmap domain. The pattern is typically input in a
hierarchical vector format such as GDSII, MIC or OASIS. The
ordering of data in the input file obeys no rules and a contiguous
geometrical feature can be formed from several elements from
different parts of the hierarchical structure. The hierarchy is
flattened and all neighbor and overlap relations are resolved when
the bitmap is created. Thus the bitmap operations need only look at
local information, in contrast to operations in the vector
format.
[0130] A close look at the rasterizing process shows that it acts
as a low-pass filter at some grid positions and not at others. When
the edge is placed relative to the edge so that an intermediate
pixel value is created some of the edge acuity of the optical
system is lost. This can be represented with a low-pass filter,
FIG. 32a, 32b, at other grid positions where the edge is
represented without an intermediate value no loss of acuity occurs.
FIG. 32a illustrates a slanted line 325 and corresponding pixel
data. A cut at A illustrates that said slanted line 325 lies on
grid. FIG. 32b illustrates exposure dose as function of position.
Graph A represents when the line is n grid and graph B represents
when the line is off-grid, such as at A cut at B in FIG. 32a. FIG.
32b illustrates that the graph is steeper for on grid position
compared to off-grid positions.
[0131] FIG. 33a illustrates contrast versus spatial frequency for
on-grid pixels and off-grid pixels without any grid filter. The
upper sequence illustrates on-grid pixels and the lower sequence
illustrates off-grid pixels. Here it is clearly illustrated that
off-grid pixels, i.e., a feature edge that does not fall on the
grid position for SLM pixels, act as a low pass filter. The optics
in a pattern generator also works as a low pass filter. The
combination of the optics and the on grid gives an image with a
certain low pass characteristic. The combination of off grid and
the optics gives an image with another low pass characteristic
(solid line) than what is expected, dotted line in FIG. 33a.
[0132] FIG. 33b illustrates the contrast as a function of spatial
frequency for on-grid pixels and off-grid pixels with an off-grid
filter. Here the off-grid filter counteracts the low pass
performance caused by the off grid position. The off grid image has
equivalent contrast versus spatial frequency performance as the on
grid image.
[0133] FIG. 33c illustrates the contrast as a function of spatial
frequency for on-grid pixels and off-grid pixels with off-grid
filter and global edge enhancement. The global filter enhances the
contrast versus spatial frequency characteristics in that the graph
is steeper in both the on-grid and off-grid image compared to the
graph in the on-grid and off-grid image without said global filter
(dotted graph). A steeper function will enhance edge placement
accuracy and edge acuity.
[0134] FIG. 34 illustrates a diagram showing contrast as a function
of pixel number. Pixels 341 illustrates the area bitmap for a
certain pattern. Pixels 343 illustrates said pixels with an
off-grid filter applied. Pixels 345 illustrates convolved pixels,
i.e., with a global edge enhancement. The global enhancement
enhances all edges, while the off-grid filter enhances only edges
with an intermediate value for the edge pixel.
[0135] FIG. 10a illustrates an alternating phase shift mask. The
leftmost area is phase shifted 180 degrees relative to the
rightmost area. The middle area is dark. A represenation of said
alternating phase shift mask as complex amplitude reflectivity
values is illustrated in FIG. 10b. Here it is illustrated that the
transition from dark to bright is not performed in one step bur
through an intermediate step. -1 corresponds to the 180 degrees
area, +1 corresponds to the 0 degree area, 0 corresponds to the
dark area, -0.6 corresponds to the leftmost transition step and 0.3
corresponds to the rightmost transition step.
[0136] The remedy is what will be called an off-grid filter, a
filter that detects that the edge is at an interpolated position
and sharpens the edge by an appropriate amount to counteract the
softening action of the rasterization. Edge sharpening by itself is
well known in the image processing, although it is not common to
have negative values available. One edge-sharpening operation is
convolution with a partially derivative kernel. Such a kernel can
look as follows: 3 D = ( - 0.1 - 0.2 - 0.1 - 0.2 + 2.2 - 0.2 - 0.1
- 0.2 - 0.1 )
[0137] Convolved with a bitmap B.sub.in it produces a new bitmap
B.sub.out
B.sub.out=B.sub.in{circle over (.times.)}D
[0138] The following is an example bitmap and how the edge is
enhanced by the convolution 4 B i n = ( . . . . . . . . . . . . . .
. . . . . 100 % 100 % 100 % 50 % 0 % 0 % 0 % . . 100 % 100 % 100 %
50 % 0 % 0 % 0 % . . . 100 % 100 % 100 % 50 % 0 % 0 % 0 % . . . 100
% 100 % 100 % 50 % 0 % 0 % 0 % . . 100 % 100 % 100 % 50 % 0 % 0 % 0
% . . . . . . . . . . . . . . . . . . . ) B out = ( . . . . . . . .
. . . . . . . . . . . 100 % 100 % 120 % 50 % - 20 % 0 % 0 % . . 100
% 100 % 120 % 50 % - 20 % 0 % 0 % . . . 100 % 100 % 120 % 50 % - 20
% 0 % 0 % . . . 100 % 100 % 120 % 50 % - 20 % 0 % 0 % . . 100 % 100
% 120 % 50 % - 20 % 0 % 0 % . . . . . . . . . . . . . . . . . . .
)
[0139] The derivative at the edge is increased by 40%. The
following example shows how a corner is enhanced after convolution
by the same kernel. 5 B i n = ( . . . . . . . . . . 100 % 100 % 100
% 60 % 0 % 0 % 0 % . . 100 % 100 % 100 % 60 % 0 % 0 % 0 % . . 100 %
100 % 100 % 60 % 0 % 0 % 0 % . . . 100 % 100 % 100 % 60 % 0 % 0 % 0
% . . . 40 % 40 % 40 % 24 % 0 % 0 % 0 % . . 0 % 0 % 0 % 0 % 0 % 0 %
0 % . . 0 % 0 % 0 % 0 % 0 % 0 % 0 % . . . . . . . . . . ) B out = (
. . . . . . . . . . 100 % 100 % 116 % 68 % - 24 % 0 % 0 % . . 100 %
100 % 116 % 68 % - 24 % 0 % 0 % . . 100 % 100 % 116 % 68 % - 24 % 0
% 0 % . . . 124 % 124 % 138 % 81 % - 20 % 0 % 0 % . . . 32 % 32 %
39 % 23 % - 11 % 0 % 0 % . . - 16 % - 16 % - 14 % - 9 % - 2 % 0 % 0
% . . 0 % . 0 % 0 % 0 % . 0 % . 0 % . 0 % . . . . . . . . . . )
[0140] A convolution with a derivative kernel enhances all edges,
i.e. it does a global edge enhancement. The off-grid filter is
rule-based in the sense that it enhances only off-grid edges. The
off-grid filter detects that the edge is interpolated and enhances
it, while an edge that is not interpolated is left unchanged. A
simple condition for interpolation is that the edge pixels have an
intermediate value. The rule that only interpolated edges are
enhanced can be expressed as an IF-THEN-ELSE rule in the bitmap
domain, but a more elegant implementation is by means of
multiplication with a weight function that continuously varies
between a small magnitude in an on-grid position and a high
magnitude in an off-grid position.
[0141] FIG. 24a illustrates an off-grid filter implementation. B is
the bitmap from the rasterizer with values in the range 0.1. K is a
coefficient array or kernel 3.times.5, 5.times.5 pixels or larger.
W is a weighting bitmap used to weight the contribution to each
entry in the bitmap B.
W.sub.n=4*(1-B.sub.n)*B.sub.n+max(4*(1-B.sub.neighbors)*B.sub.n-
eighbors). An adjusted value of B.sub.13 is computed as:
[0142] B.sub.13filtered=B.sub.13+W.sub.13*(K.sub.7*B.sub.7+ . . .
+K.sub.19*B.sub.19), where
W.sub.13=4*(1-B.sub.13)*B.sub.13+MAX(4*(1-B.su- b.7)*B.sub.7, . . .
,4*(1-B.sub.12)*B.sub.12, 4*1-B.sub.14)*B.sub.14, . . .
,4*(1-B.sub.19)*B.sub.19).
[0143] FIG. 24b illustrates an off-grid filter. B is the bitmap
from the rasterizer with values in the range 0-1. G and F are
derive d bitmaps used for the filter. Gn=2(Bn.0.5).
Fn=4(Bn-1)(Bn-0). K is a coefficient array or kernel 3.times.5,
5.times.5 pixels or larger. An adjusted value of B13 is computed
as:
[0144] B13filtered=B13+K7*G7*F7*B7+ . . . +K19*G19*F19*B19.
[0145] The same or similar functions can be implemented in other
ways, which is obvious for one skilled in the art.
[0146] The constants used in the edge enhancement and for the
calculation of the "grayness" can be varied to produce as good
results as possible for a variety of typical pattern elements, "use
cases". They can be determined manually by controlled experiments
or by simulations using codes such as PROLITH from KLA-Tencor or
Solid-C from Sigma-C. In a more elaborate setup the use cases can
be programmed into an optimization job using one of the simulators
above and non-linear optimization routines.
[0147] FIG. 9a-d shows simulated performance of a manually fitted
enhanced space (clear line). The vector data has one edge 912 on
grid and one off grid 914. If it is rasterized without edge
enhancements the result is an aerial image where the on-grid edge
has higher acuity than the off-grid one. If the dose is varied the
width of the trench varies, but the two edges 991, 992 move
differently with dose. This is seen as a movement of the center of
the space with dose. The diagram 9d shows the movement with dose of
the center of the space without 970 and with 980 the off-grid
filter. It is seen that with the off-grid edge enhanced the center
of the space is stable over a very large dose interval. This is an
alternative way to describe that the left and right edges in the
aerial image are closely identical. A pixel within the feature 940
is set to a higher exposure value, here 116% compared to the rest
of the pixels within the feature, which are set to 100%. A pixel
950 outside the feature is set to a negative black value, i.e., a
negative complex amplitude reflectivity. The rest of the outside
feature pixels are set to 0%.
[0148] The example in FIG. 9b has a minimum negative value of
-{square root}{square root over (0.72%)}=-0.085. It was shown above
that the square mirrors could create value of -0.25 or even below.
Therefore there is room for further edge enhancement using negative
complex amplitude even after some of the dynamic range of the
mirrors has been used for the off-grid filter. The operation can
conceptually be expressed as a convolution with two parts.
[0149] B.sub.out=B.sub.in{circle over
(.times.)}(D.sub.global+g*D.sub.off-- grid), where D.sub.global is
the edge enhancement kernel for global edge enhancement,
D.sub.off-grid the kernel for removing the difference between
off-grid and on-grid edges and g is the "grayness", the weight
function that determines the application of D.sub.off-grid.
[0150] In a further improvement the convolution kernel, or more
generally the digital filter, has slightly different properties in
the x and y directions to correct for inherent differences in edge
acuity between x and y.
[0151] In order to print a true image of the input data the pixel
values cannot be a linear representation of the overlap between the
pixel area and the feature. There has to be a non-linear
transformation between overlap area and pixel value. Regardless if
the representation of the pixel values is chosen to be tilt angle,
actuator voltage, complex amplitude reflectivity R or
.vertline.R.sup.2.vertline. a non-linear pixel-by-pixel
transformation is needed: V=I(A)*A where V is the pixel value, A
the area overlap from 0 to 100%, and I(A) is the illumination
table. The illumination table I(A) describes the non-linearity of
the system that arises from the partial coherence over the
modulating element (mirror). The shape of the function depends on
the pixel size relative to the optical resolution, the angular
spread of the illuminating light, the used dynamic range of the
SLM, and the relative dose (dose/dose-to-clear).
[0152] The illumination function can be determined empirically or
through opticalsimulation. In either case the printing conditions
such as NA, illuminator setting, pixel size, SLM contrast, and dose
are fixed. A large feature is printed from vector data with the
placement of one edge versus the grid varying, either in resist or
virtually in a lithography simulator. The pixel value is a
predetermined function of the feature overlap with the pixel area,
possibly with a non-uniform weight function over the pixel area.
The predetermined function can for example be a linear
function.
[0153] The feature is printed for different edge placements and the
placement of the printed edge is measured, either by a metrology
system such as Leica IPRO or by numerical analysis of the simulated
images. The measurement gives a nonlinear function for placement
vs. data. This non-linear function is used to compute the
illumination table. The procedure can be repeated iteratively in
order to arrive at a stable and accurate illumination table. This
illumination table makes the printer print true to data for large
features with the used printing conditions.
[0154] A preferred embodiment of the invention has the following
order of conversions: see FIG. 8a
[0155] 1. Flatten the hierarchical input database, 2. Rasterize all
features and compute the overlap area of feature elements for every
bitmap pixel (possibly using a non-uniform area sampling function
per pixel) producing a so called area bitmap, 3. Make pattern
corrections (preferably in real time) including edge enhancement
and off-grid enhancement as well as special enhancement of corners
and small features, producing a corrected area bitmap, 4. Multiply
the corrected bitmap by the illumination function producing what is
currently called an intensity bitmap, 5. Make a table look-up
conversion of the intensity bitmap, the lookup table representing
the properties of individual modulator elements or mirrors,
producing a DAC value bitmap
[0156] A slightly more complex conversion sequence gives more
control of the non linearities of the rasterizing and partially
coherent imaging: see FIG. 8b 1. Flatten the hierarchical input
database, 2. Rasterize all features and compute the overlap area of
feature elements for every bitmap pixel (possibly using a
non-uniform area sampling function per pixel) producing a so called
area bitmap, 2b Multiply the area bitmap by a first illumination
function, 3. Make pattern corrections (preferably in real time)
including edge enhancement and off-grid enhancement as well as
special enhancement of corners and small features, producing a
corrected area bitmap, 4. Multiply the corrected bitmap by a second
illumination function producing what is currently called an
intensity bitmap, 5. Make a table look-up conversion of the
intensity bitmap, the lookup table representing the properties of
individual modulator elements or mirrors, producing a DAC value
bitmap
[0157] In a third embodiment, see FIG. 8c the pattern is divided
into two (possibly three or more) partial patterns, e.g. one
containing more high-frequency information and another one
containing more low-frequency information. The partial patterns are
converted with different parameters before they are combined to
drive the SLM. The decomposition is suitably implemented as
different bitmap filters such as the convolutions described above.
High and low frequency filtering of images is well known in the art
of digital image processing and many methods and detailed
implementations can be devised by a person skilled in the art.
[0158] In all embodiments the illumination function can be folded
into the mirror look-up table. The mirror LUT (Look Up Table) must
then be changed depending on the angular spread of the illumination
and the dose.
[0159] The illumination table makes the CD independent of the pixel
grid, at least for large features. But the illumination does not
make the aerial image acuity constant through grid. Features at the
resolution limit tend to disappear, and features placed at grid
positions where the acuity is compromised by the rasterization
disappear first. Therefore line width CD is not stable through grid
at the resolution limit.
[0160] The off-grid enhancement makes the image of on-grid and off
grid identical. This makes all printing properties more stable,
e.g. for varying dose. But the main benefit is that features at the
resolution limit become much more stable. In this way the useful
resolution is improved.
[0161] The global edge enhancement also increases the useful
resolution. It increases the contrast of thin lines by extending
the dynamic range of the SLM modulator elements. Edges are made
crisper. Since small features have the edges close together they
get a double boost.
[0162] Line ends are also improved, partly because all edges are
made crisper, but also because the convolution with a derivating
kernel enhances the contrast of line ends. Corners are likewise
enhanced, although not as much as line ends. With properly chosen
parameters line end shortening and CD linearity failures of lines
and contacts can be largely counteracted. If the global enhancement
is implemented as a convolution with a derivating kernel the size
of the kernel and the coefficients in it can be used to determine
the magnitude and detailed properties of the enhancement.
[0163] The complex amplitude of the square mirror and how it varies
with the tilt angle is calculated as described above. It is
influenced by the shape of the mirror. Other shapes give other
characteristics and shown in FIG. 14-23, 25-31.
[0164] One can make a distinction between shapes that are area
filling or not. For instance FIG. 10a, 18a, 19a are surface
filling. FIG. 14a-23a, 25a-31a shows that many perfectly viable
mirror shapes have radically different complex amplitude
reflectivity. By selecting a different mirror shape one can get
access to large amounts of negative R. Some of the shapes, like the
H shape in FIG. 22a, can provide a symmetrical positive and
negative R.
[0165] FIG. 11a depicts mirror configurations in a spatial light
modulator, which may be used in order to achieve any desired
pattern with improved image quality. Mirror 1110 and 1120 have
their tilting axis along symmetry line 1130. Mirror 1110 have outer
areas with phase 0 and inner area with phase 180. Mirror 1120 are
reversed relative to mirror 1110, i.e., outer areas have phase 180
and inner area has phase 0. Mirror 1110 and 1120 are arranged in a
chess board manner, i.e., mirror 1110 is surrounded by four 1120
mirrors and mirror 1120 is surrounded by four 1110 mirrors.
[0166] FIG. 11b illustrates the real part of the complex amplitude
reflectivity as function of a degree of tilting of the mirror. As
can be seen from FIG. 11b, mirror element 1110 goes from +1 to -1
as the mirror is tilted and mirror element 1120 goes from -1 to +1
as the mirror is tilted. With mirror characteristics as depicted in
FIGS. 11a and 11b patterns as depicted in FIG. 12a-12e can easily
be achieved. FIG. 12a illustrates a pattern with uniform phase 0.
Only mirrors 1120, denoted in FIG. 12a with phase 180 and an arrow,
are tilted. The direction of said arrow indicates the direction of
tilting. Every second mirror is tilted in a reversed direction.
However mirrors may all be tilted in the same direction.
[0167] FIG. 12b illustrates a pattern with uniform phase 180. In
FIG. 12b are only the mirrors 1110 tilted, denoted in FIG. 12b with
phase 0 and an arrow. Here again a direction of said arrow
indicates the direction of tilting said mirror.
[0168] FIGS. 12c-e depict patterns with uniform dark. In FIG. 12c
none of the mirrors are tilted. In FIG. 12d all mirrors are tilted.
In FIG. 12e all mirrors are partially tilted. In FIG. 12c-e the
direction of said arrow indicates the tilting direction of the
mirror.
[0169] Controlling the characteristics of the mirrors with the
shape leads to inflexible designs where a modest change in
properties may necessitate a change in layout affecting both the
CMOS underneath the MEMS and the rasterizing algorithms. It is
possible to change the apparent shape of the mirrors by covering
the unwanted parts of the mirrors with a non-reflecting layer, e.g.
a dark metal like zirconium, an anti-reflection coating like a
deposited metal oxide or other dielectric film as is well known in
the art. A practical way of controlling the characteristics of the
mirrors is by structures on the top surface of the mirror. One
advantage is that it may use the same material as the rest of the
mirror, another that, whatever material is used there is no
requirement to reduce the reflectivity, since the effect of the
surface structures is created by division-of-wavefront destructive
interference and light scattering. The areas that are intended to
be non-reflected can be patterned by structures that create
destructive interference in the specular direction. An example is a
checkerboard of squares with a step height of lambda over 4 (lambda
over 2 in the reflected beam). It has been found that with
partially coherent illumination the structures can be fairly large.
FIG. 12 shows a number of possible designs and corresponding
properties.
[0170] FIG. 13a-d illustrates the correspondence between Masks or
reticles and an SLM having similar properties. A leftmost
illustration in FIG. 13a depicts a binary mask. The binary mask has
a part, which is covered with a chrome layer. Said chrome layer is
opaque. Next to the chrome layer said mask is clear, defining a
fully transmissive part of said mask. A rightmost illustration
depicts an SLM with corresponding properties as said binary mask.
The chrome part in said binary mask corresponds to a complex
amplitude reflectivity A=0 and the clear part in said binary mask
corresponds to a complex amplitude reflectivity A=1.
[0171] A leftmost illustration in FIG. 13b depicts an attenuating
phase shift mask. The attenuating phase shift mask has a part,
which is covered with a partly transmissive layer. Next to the
partly transmissive layer said mask is clear, defining a fully
transmissive part of said mask. A rightmost illustration depicts an
SLM with corresponding properties as said attenuating phase shift
mask. The partly transmissive layer in said attenuating phase shift
mask corresponds to a complex amplitude reflectivity in the range
of -1<A<0 and the clear part in said binary mask corresponds
to a complex amplitude reflectivity A=1.
[0172] A leftmost illustration in FIG. 13c depicts an alternating
phase shift mask. The alternating phase shift mask has a first
part, which is covered with a chrome layer. Said chrome layer is
opaque. At one side of the chrome layer said mask is clear,
defining a fully transmissive part of said mask. At another side of
said chrome layer said mask is shifted in phase relative said
chrome layer and said clear part. A rightmost illustration depicts
an SLM with corresponding properties as said alternating phase
shift mask. The chrome part in said alternating phase shift mask
corresponds to a complex amplitude reflectivity A=0. The clear part
in said alternating phase shift mask corresponds to a complex
amplitude reflectivity A=1. The shifted part in said alternating
phase shift mask corresponds to a complex amplitude reflectivity
A=-1.
[0173] A leftmost illustration in FIG. 13d depicts a CPL
(Chrome-less Phase Lithography) mask. The CPL mask has a part,
which is covered with a shifted layer. Said shifted layer is clear
and fully transmissive. Next to the shifted layer said mask is
clear, defining a fully transmissive part of said mask. The shifted
part has its surface higher or lower than said clear part. The
rightmost illustration depicts an SLM with corresponding properties
as said CPL mask. The shifted part in said CPL mask corresponds to
a complex amplitude reflectivity A=-1 and the clear part in said
CPL mask corresponds to a complex amplitude reflectivity A=1.
[0174] The different parts in FIG. 13a-d comprises typically a
plurality of pixel elements, i.e., in the SLM case said areas are
represented by a plurality of SLM pixels, the number depending on
the size of the feature to be patterned.
[0175] FIG. 14-31 illustrates different mirror configurations and
corresponding complex amplitude trajectory, complex amplitude
reflectivity graph and exposure graph as a function of phase at an
edge of the mirror.
[0176] FIG. 14a illustrates a square shaped mirror 145 capable to
be tilted at hinges 147, 148 defining a tilting axis. FIG. 14b
illustrates the complex trajectory for said square shaped mirror.
As can be seen from FIG. 14b an imaginary part of the complex
amplitude is almost zero indicating that the mirror element is
nearly symmetrical. Symmetrical mirror elements have the imaginary
part equal to zero.
[0177] FIG. 14c illustrates the reflection and exposure as
functions of a phase of the mirror element at an edge of the same.
The reflection is the real part of the complex amplitude
reflectivity. The exposure is the square of the real part of the
complex amplitude reflectivity. In the same FIG. 14c a magnified
portion of the exposure is illustrated. A square mirror has a
relatively low level of negative real part of the complex
amplitude, therefore full phase shifting cannot be obtained.
[0178] FIG. 15a illustrates another configuration of a mirror. In
this embodiment the hinges 157, 158 are attached to the mirror 155
closer to the center compared to the mirror in FIG. 14a. This
embodiment has less reflecting area compared to the mirror
illustrated in FIG. 14a, especially it has less reflecting area
close to the tilting axis defined by the hinges 157, 158. This will
affect the minimum value of the complex amplitude as illustrated in
FIGS. 15b and 15c, in that the real part has a minimum, which is
more negative than the embodiment illustrated in FIG. 14a.
[0179] FIG. 16a illustrates another configuration of a mirror. In
this embodiment the hinges 167, 168 are attached to the mirror 165
even closer to the center compared to FIG. 14a and FIG. 15a. This
embodiment has less reflecting area compared to the mirror
illustrated in FIGS. 14a and 15a, especially it has less reflecting
area close to the tilting axis defined by the hinges 167, 168. This
will affect the minimum value of the complex amplitude as
illustrated in FIGS. 16b and 16c, in that the real part has a
minimum, which is more negative than the embodiment illustrated in
FIG. 14a and FIG. 15a.
[0180] FIG. 17a illustrates yet another configuration of a mirror.
In this embodiment the hinges 177, 178 are attached to two
diagonally displaced corners of the mirror 175. The illustrated
embodiment has no negative complex amplitude, neither real nor
imaginary.
[0181] FIG. 18a illustrates still another configuration of a mirror
185. To said mirror 185 are attached two hinges 187, 188, defining
a tilting axis. This configuration has two sides, which are
zigzag-formed, where one is the inverse of the other. This
configuration, as well as the previously illustrated ones, is
perfectly suitable to be stitched together in a one- or
two-dimensional array of micromirrors, such as in a spatial light
modulator. The complex amplitude trajectory is illustrated in FIG.
18b, which indicates that this embodiment has a slightly negative
complex amplitude. FIG. 18c illustrates the exposure and reflection
for the configuration in FIG. 18a.
[0182] FIG. 19a illustrates still another configuration of a mirror
195. To said mirror 195 are attached two hinges 197, 198, defining
a tilting axis. This embodiment has also two sides where one is the
inverse of the other one. This configuration is suitable to be
stitched together in the one or two dimensional array of mirrors.
As can be seen in FIGS. 19b and 19c this embodiment has less
negative complex amplitude compared to the configuration
illustrated in FIG. 18a.
[0183] FIG. 20a illustrates still another configuration of a mirror
205. This embodiment differs to the one illustrated in FIG. 19a in
that a reflecting area is slightly less than the embodiment
illustrated in FIG. 19a. Areas are cut off around an attachment
position of hinges 207, 208, which is not the case in FIG. 19a. As
can be seen in FIG. 20b, 20c, this embodiment has slightly more
negative complex amplitude than the configuration in FIG. 19a.
[0184] FIG. 21a illustrates still another mirror configuration. In
this configuration hinges 217, 218 define a tilting axis. Here the
mirror area is much less close to the tilting axis compared to
further away, which will affect the complex amplitude of the
mirror, see FIGS. 21b and 21c. As the mirror element is almost
symmetrical there is no imaginary part of the complex amplitude
present. The real part of the complex amplitude is more negative
than all previously illustrated embodiments above.
[0185] FIG. 22a illustrates still another mirror configuration.
Hinges 227 and 228 define a tilting axis as previous. In this
embodiment there is almost no reflecting area close to the tilting
axis. Nearly all reflecting areas are at a distance from the
tilting axis. This will increase the negative complex amplitude
even more compared to the embodiment illustrated in FIG. 21a. This
configuration is also suitable to be arranged in a one or
two-dimensional array of mirror elements. This is illustrated in
FIG. 22d:
[0186] FIG. 23 illustrates still another mirror configuration 235.
Hinges are attached to support structures 237, 238. The hinge are
may be covered with an anti-reflective coating in order not to
reflect any radiation at a predetermined wavelength, said hinges
are hidden for said reason n FIG. 23a. Also hidden is a connecting
element connecting reflecting areas 236, 239. This configuration
exhibit exceptional complex amplitude values as indicated in FIG.
23b and FIG. 23c. The real part of the complex amplitude goes from
+1 to -1 and there is no imaginary part of the complex
amplitude.
[0187] FIG. 25a illustrates still another embodiment of a mirror
configuration 255. This embodiment differs to the one illustrated
in FIG. 15a in that some corner areas 251, 252, 253, 254 of the
mirror are out of phase relative the rest of the mirror. Preferably
said corner areas affect a reflected wavelength so that said
reflected wavelength from said corner areas are 180 degrees out of
phase relative to the other mirror areas. As illustrated in FIG.
25b, 25c, the complex amplitude will decrease compared to the
embodiment in FIG. 15a.
[0188] FIG. 26a illustrates still another mirror configuration 265.
In this configuration there are two areas 261, 262 which are out of
phase relative to the rest of the mirror. Preferably said areas
affect a reflected wavelength so that said reflected wavelength
from said areas are 180 degrees out of phase relative to the other
mirror areas. This embodiment will affect the complex amplitude,
see FIG. 26b, 26c, slightly different compared to the embodiment
illustrated in FIG. 25a.
[0189] FIG. 27a illustrates yet another embodiment of a mirror
configuration 275. Here are out of phase areas 271, 272 larger than
the out of phase areas in FIG. 26a. This will affect the position
of the local maximum and minimum positions of the reflection, see
FIG. 27c compared with FIG. 26c, as well as this embodiment give a
more negative complex amplitude than the embodiment in FIG.
26a.
[0190] FIG. 28a illustrates still another mirror configuration 285.
Here the central part of the mirror is covered with an area 281,
which area is 180 degrees out of phase relative the rest of the
mirror. This embodiment will cause the complex amplitude to go from
+1 to -1, see FIG. 28b, 28c.
[0191] FIG. 29a illustrates still another mirror configuration 295.
Here the central part of the mirror 295 is covered with an area 180
degrees out of phase relative the rest of the mirror. The area is
slightly differently shaped compared to the one in FIG. 28a,
resulting in slightly different complex amplitude values, see FIG.
29a, 29b.
[0192] FIG. 30a illustrates still another mirror configuration 305
having a central part covered with an area 180 degrees out of phase
relative to the rest of the mirror.
[0193] In the embodiments illustrated in FIG. 25a-30a the area 180
degrees out of phase relative to the other part of the mirror apply
to reflected light/electromagnetic radiation.
[0194] FIG. 31a illustrates still another mirror configuration 315.
Here there are two areas, which are out of phase for reflected
light/electromagnetic radiation relative the rest of the mirror. A
first area 311 is -90 degrees out of phase relative to the
non-hatched mirror areas. A second area 312 is +90 degrees out of
phase relative the non-hatched parts of the mirror. This embodiment
will give an extended deflection range giving 0 reflection, see
FIG. 31c. As can be seen from FIG. 31b this mirror configuration
has no imaginary part.
[0195] FIG. 31a has areas (hatched in figure) with different
heights than lambda over four. Such structures can be used to
further modify the pixel characteristics. The example in FIG. 31a
gives an R maximum at a small tilt and a plateau at R=0. This
mirror is easier to calibrate accurately than the other mirrors
shown.
[0196] There are at least three interesting cases of ranges of
complex amplitude reflectivity cases. The first one relates to full
phase shifting capability, which means that the complex amplitude
reflectivity goes from +1 to -1, a plurality of mirror
configurations having such characteristic has been disclosed
above.
[0197] The second one relates to attenuated phase shift masks,
which means that the complex amplitude goes from +1 to -0.245.
[0198] The third one relates to an ordinary chrome mask, which
means that the complex amplitude goes from +1 to 0.
[0199] A suitable mirror design gives a relative flat graph for the
complex amplitude as a function of mirror tilt angle or reflected
light at he edge of the mirror. Such mirror design will not be so
sensitive to changes in tilt angle for the desired gray-value of
the mirror element.
[0200] When the complex amplitude is specified in the range -1 to 1
it implies that the amplitude is normalized, so that the highest
amplitude that is used is normalized to +1.00. The same holds for
complex amplitude reflection. Exceptions to this normalization are
where it is obvious from the context that an actual value or a
value normalized to an ideal specular reflecting surface is
used.
[0201] These values are the same as those used in Levinson-type
PSM, chrome-less phase lithography (CPL), and other so called
strong PSMs. By driving the SLM to these values the same resolution
and process latitude gains can be made as in wafer lithography
using strong PSMs. Figure XX shows haw the Re(A) can be controlled
to act as a number of commonly used types of phase-shifting
masks.
[0202] The H shape is also surface filling, but gives a pattern
that is not optimal for rasterizing. An equivalert mirror shape can
be created from a square mirror place by reduction of the mirror
reflectivity on some areas. The reflectivity can be reduced by
coatings of a low-reflectance material or by structuring the
surface to create destructive interference or light scattering away
from the projection optics. The illumination with the small angular
spread used makes it possible to use rather large surface
structures.
[0203] In the description above negative values of R have been used
for edge enhancement and correction of grid and x-y artifacts. It
is also possible to use the SLM as a strong phase-shifting mask
(PSMs) as known in lithography. The pixels in 7d and 8a-c can
produce R values of 1.00, 0.00, and -1.00 (after scaling by an
increase of dose). These values are the same as those used in
Levinson-type PSM, chrome-less phase lithography (CPL), and other
so called strong PSMs. By driving the SLM to these values the same
resolution and process latitude gains can be made as in wafer
lithography using strong PSMs.
[0204] In addition the SLM also has intermediate values not present
in commonly used masks. These are used for placement of edges in a
fine address grid. They can also be used for phase-shifting
lithography equivalent to "high transmission PSMs" and "tri-tone
masks" known in the art e.g. for printing of dense contacts.
[0205] If there is a small asymmetry between R=1.00 and -1.00 it
gives every-second-line artifacts in the printed pattern. A remedy
is shown in FIG. 11b. The pixels form a checkerboard pattern where
every second mirror is displaced by 180 degrees, i.e. they move
from -180 to +180 degrees instead of the normal +180 to -180.
[0206] A strongly phase-shifting reticle normally have areas with
three complex amplitudes A=+1.00, 0.00, and -1.00. Although they
can be described with a single parameter they are usually defined
in two binary (having areas of two kinds) mask data files: one file
for dark and one file for those areas that are shifted, i.e. 180
degree phase. The shifter features are usually printed with an
overlap of the chrome so that the chrome data determines the
dimension in the mask.
[0207] An embodiment of an SLM printer using phase-shifting SLMs
follows the scheme above. It rasterizes two binary (two-valued)
input files and combines them in a Boolean operation to create the
multi valued SLM bitmap data. Each binary set of data can have its
own set of bitmap operations, such as CD bias and edge enhancement.
This preserves the highest degree of transparency between mask and
maskless pattern data files.
[0208] In another embodiment the rasterizer reads a file containing
at least two types of areas and a background, e.g. clear and
shifted areas in a dark background and rasterizes them directly to
a multivalued bitmap. This has the advantage of creating
immediately interpolated edges for all types of feature boundaries:
clear to dark, shifted to dark, and clear to shifted. It is also
more suitable for working directly from the design database without
the intermediate step of mask data tape-out. The relative benefits
of the first and second type of rasterization depend on the
application and a preferred embodiment can use either scheme.
[0209] While the present invention is disclosed by reference to the
preferred embodiments and examples detailed above, it is understood
that these examples are intended in an illustrative rather than in
a limiting sense. It is contemplated that modifications and
combinations will readily occur to those skilled in the art, which
modifications and combinations will be within the spirit of the
invention and the scope of the following claims.
* * * * *