Lithography Optics

1
Introduction

Optical lithography is a method used to produce electronic

wafers for semiconductor components. There are many
different ways of structuring wafers , of which the optical
method is one of the most important.

The optical systems required to perform this task are rather

complex and are among the most sophisticated of optical
systems. Today , only three major vendors produce the
projection lenses , which are the most sensitive parts of the
system.

Typical optical setup

2
The beam of a laser source is guided into a pupil shaping
system, where the σ-settings are formed and the etendue is
adapted to the system with the help of diffusers or axicon
components . A homogenizer unit follows, which assures
that the illumination is quite homogeneous over the
complete field of view. It can be performed with an
integrating rod or with a fly-eye condenser.
A relay optics further serves as a profiling tool to guarantee
a uniform and telecentric illumination in the mask or
reticle plane . The projection lens images the mask with a
reduction factor in the range of m=0.25 on the wafer.
M
Moore announced d an exponential
i l decrease
d in
i the
h
resolution to within a factor of two all three years. The
progress is achieved by developing and optimizing all
technological aspects which influence the results.
Moore ‘ss law : lithographic resolution with time:

3
Next Generation Lithography (NGL)

The corresponding wavelengths of different sources:

4
Projection Techniques

The semiconductor wafers , which need to be illuminated,

are circular in shape and have a typical size with a
diameter of 300 mm.
The image of such a large area to a high resolution in one
shot is impossible for an optical system. Therefore , the
complete wafer is divided into partitions of dies of
typically , 26 x 32 mm side length.

The field size can be imaged by projection in so-called

stepper systems. Projector
P j llenses off the
h stepper type must
transform the full rectangular field size. The total track
length of the projection lenses are typically in the range of
1 m or more.

In the concept of scanning systems, the field size is

reduced. This helps in reducing the complexity of the
optical systems and saves cost. If ring or slit shapers are
used , together
g with a scanningg technique,
q , the lens must
support a field size with a length of typically 26 mm and a
width in the range of 5-14 mm.

The size and shape of the field obtained using the various
lenses are illustrated here .The scan field can be rectangle
or may have the shape of a circle ring segment.
5
Several modes of operation for lithographic projection are
illustrated:

6
If the size of the wafer is too large, a scanning mode is
used, as in part (b).
Another possibility is to divide the large wafer into
partitions to project only parts of it in succeeding steps ( c ).
The last option is to combine the step and the scan method.

Now consider the DOF and resolution:

The light falls onto the resist material , which typically has
a thickness of 0.2
0 2 um-0.3
um 0 3 um.
um But the realized values of
the resist thickness strongly depends on the resolution and
can be 1 um for low NA systems and 0.1 um for high NA7
systems.
Resolution and DOF

8
Resolution

The main goal in projection lithography is to image the

reticle structures very precisely and to a high resolution.
Therefore, the resolution and a good contrast are the
primary requirements for lithographic systems.

If several resist layers are processed in an overlay

technique , the depth of focus is also a very important
quantity.

A
According
di to the
h formula
f l off Rayleigh:
R l i h

In the scalar approximation the lateral resolution depends

on the following parameters:
The vacuum wavelength
The NA of the projection lens
The k1 factor
The k1 factor typically lies in the range of 0.25-0.5.
For non-critical applications , values in the range of k1 ≧
1 are also possible.

The value depends on the following:

1. The degree of coherence and the geometry of pupil
9
illumination
2. The polarization
3. The geometry of the imaged feature . In practical
lithography, all kinds of structures can be found.. In
critical range near the resolution limit, contact holes,
lines and stripes are of special importance.
4. The nonlinearility of the resist.
5. The optical proximity effects, for features at close
distance. Problems occurs in practice when different
features are in close proximity. In this case , the
imaging of two structures cannot be considered to be
isolated , the image depends on the surroundings and
the
h shape
h off the
h object
bj details
d il , which
hi h influence
i fl eachh
other. To deal with these problems , optical proximity
corrections (OPC) are used in the reticle to obtain the
desired image on the wafer. This pre-distortion can be
done by geometrical means,
means for example , in case of
serif structures at the corners.
6. Phase masks (PSM) , which are used for resolution
enhancement , increase of contrast and correction of
proximity
p y effects.

The value k1=0.25 is the lower physical limit.

Special techniques which reduce k1 are called resolution
enhancement technique (RET).

The depth of focus:

10
factor k2 is similar to k1.
In the best case , one can
assume k2=1 . The value of k2 is also influenced by the
polarization , coherence and phase masks.

In the high NA case, the formula for the depth of focus

must be modified. For system which corrected sine
condition , the formula for the depth of focus is:

The correction factor decreases the depth of focus with

increasing NA.
It is a scalar formula. In the high-NA case, it is necessary
to incorporate the vector effects.

The dependence of are similar, but the

d
dependence
d off on the
h NA iis quadratic
d i andd the h
factor k2 has other numerical values. It should be
noticed that the refractive index occurs in the
denominator of the formula , this means, in case of
immersion , the ratio between the lateral and the axial
resolution changes. Therefore , one major advantage
in immersion systems is an increase depth of focus 11at
the same lateral resolution.
The major advantage in immersion systems is an increased
depth of focus at the same resolution.

Example:

the lateral resolution and depth of focus for various values

of the wavelength and the numerical aperture for values
k1=0.36 and k2=0.28.

12
The progress in the resolution achieved over three decades
of development:

Different representation:

13
Resolvable lateral feature sizes in nm:

14
15
16
17
Information Transfer

The evolutionary
Th l i trendd off downsizing
d i i electronic
l i structure
sizes with time formulated in 1965 by the famous law of
Moore. This law predicts a doubling in the number of
transistors per chip every two years.

The main purpose of lithographic systems is the

transmission of information from the reticle to the wafer.
The smaller the structure on the reticle side , the higher
will be required
q resolution of the optical
p system.
y
Therefore, one performance indicator for a lithographic
projection system id the total information transmitted.

According to the Rayleigh formula , the number of

points can be resolved in the image in one dimension.
Since the field size must be considered as two-
dimensional , for a full field projection , a generalized
Helmholtz-Lagrange invariant can be defined as:

where TTI stands for total transmitted information. If a

scanning mode is used during projection , an additional
factor must be considered in this equation.
18
Historic development of the information transmitted
capacity in lithographic imaging as a function of time
compared with the Moore
Moore’ss law.

19
20
21
22
Illumination and Coherence

23
The adaptation and optimization of the illumination and
coherence is one possible way of improving the resolution.

The influence of the coherence on the imaging

performance have been known since the time of Abbe .
The method for enhancing contrast and resolution by
appropriate illumination settings have been used in
microscope for a long time.
In the lithographic projection application , the methods
have been further imprioed. This development leads to
more and more complicated lithographic illumination
systems.

The illumination systems must fulfill the following

requirements in the illumination plane of the reticle.
1 Homogeneous intensity within a uniformity of better
1.
than 1%.
2. Generation of the desired variable geometry of pupil
filling.
3. Telecentricityy of the chief ray
y direction.
4. Nearly lossless energy transmission.

To cover the first requirement, an integrating rod or lenslet

can be used.
For the second point, first the etendue of the laser
illumination must be increased in order to fill the pupil
completely. This is usually done with a diffuser , arrays 24or
diffractive elements. Furthermore, it is necessary to use
geometrically shaped rings and segments rings.

Axicons and complicated diaphrams may be used for this

purpose.
An relay objective lens is used to transfer the light from the
beam-shaping part to the mask plane.

There are two degrees of freedom required to shape the

pupil illumination. The size of the outer boundary of the
illuminated parts of the pupil is determined by the value of
σ . The
Th exact geometry off the
h illuminated
ill i d areas can be
b
used to optimize the coherence . For this is called
structured illumination , there are three major
configuration types , which can be found in practice.

1. Ring-shaped illumination with outer radius and

inner radius . It should be noted that two values
reduce the effective area of the
illuminated pupil.
p p
2. Dipole illumination with two isolated areas of
illumination in opposite locations. The illuminated
areas can be circular with radius σ and a distance υ to
the center. Other possibilities are azimuthal sections
with angle θ of a ring with different parameters and
shapes.
25
3. Quadruple illumination with four small circles or parts
 of an annular opening of illumination. This is an
extension of two , but has the advantage of being more
isotropic concerning the structure orientation.

Some useful geometries of the pupil illumination:

In the simple case corresponding to a),b),c) :

The visibility , the lateral resolution and depth of focus are
26
influenced by the value of the coherent setting parameter σ.
Imaging of fine bar pattern structures of different settings
of the illumination: NA=1.2 an annular, a dipole and a 30°
quadrupole

Several defocused planes shows the change with depth.

Each of the pictures on the left show the shapes of the
pupil illumination.

27
28
29
30
31
32
33
Optical Image Enhancement
Techniques

34
Formation :Wave Optics

35
Many of the limitations of geometrical optics can be
explained by considering the wave nature of light.

A wavefront is assumed to be made of an infinite number

of point sources. Each of these sources produces a
spherical secondary wave called a wavelet.
The Huygens
Huygens’ approach has allowed analysis without the
need to fully solve Maxwell’s equations.
When a beam of light encounters the edge of an opaque
obstacle , propagation is not rectilinear as might be
assumed based ion the assumption of geometrical
shadowing. The resulting variation in intensity produced at
some distance from the obstacle is dependent on the
coherence of light, its wavelength, an the distance the light
travels before being observed.
Th situation
The it ti forf coherent
h t illumination:
ill i ti

36
Only when there is no separation between the obstacle and
the recording plane does rectilinear propagation occur. As
the recording plane is moved away from the obstacle, there
is a region where geometrical shadow is still discernible.
Beyond this region, far from the obstacle, the intensity
pattern at the recording plane no longer resembles the
geometrical shadow; rather , it contains areas of light and
dark fringes.

At close distance , near field diffraction , or Fresnel

diffraction, dominates. At greater distances, far-field
diff i or Fraunhofer
diffraction, F h f diffraction,
diff i dominates.
d i

Fresnel Diffraction : proximity Lithography

The situation where a space mask is illuminated with 365

nm radiation and the separation distance between mask and
wafer is 1.8 um.
For relatively
y large
g features , on the order of 10-15 um ,
rectilinear propagation dominates , and the resulting image
intensity distribution resembles the mask.
A relation for minimum resolution results :

Where w is the space width, λ is illumination wavelength ,

37
and s is the separation distance.
38
Partial Coherence Theory : Diffracted-Limited Resolution

The concept of degree of coherence is useful as a

description of illumination condition. The Abbe theory of
microscope imaging can be applied to micrilithographic
imaging with coherent or partial coherent illumination.

Abbe demonstrated that when a ruled grating is coherent

illuminated and imaged through an objective lens, the
resulting image depends on the lens numerical aperture.
The minimum resolution that can be obtained is a function
off both
b h the
h illumination
ill i i wavelength
l h andd the
h lens
l NA as
shown for coherent illumination.

39
Because no imaging is possible if no more than the
undiffracted beam is accepted by the lens, it can be
reasoned that a minimum of the first diffraction order is
required for resolution.

The position of the first order is determined as:

The minimum resolvable line width (R=p/2) becomes:

To adapt to nonuniform lamp sources , Kohler devised a

two-stage illuminating system to form an image of the
source into the entrance pupil of the objective lens.

A pupil at the condenser lens can control the numerical

aperture of the illumination system.
The degree of the partial coherence (σ) is measured as 40
The ration of the effective source size to full objective
aperture size or the ratio of the condenser lens NA to
objective lens NA.

As σ approaches zero, a condition of coherence

illumination exists. As σ approaches one, incoherent
illumination exists.
In lithograpic systems , σ is generally in the range 0.3-0.9.
Values below 0.3 will result in the “ring” in images ,fringes
that result from coherent interference effects .

The impact of the partial coherence:

A situation where coherent illumination of a given mask
pattern does not allow lens collection of diffraction orders
beyond the zeroth order
order, partially coherent illumination
would be preferred.
Consider a coherently illuminated rectangular grating mask
where ±1th diffraction orders fall just outside a projection
systems
y lens NA . With coherent illumination , imaging
g g is
not possible as feature size fall below the limit.
Through the use of the partial coherent illumination, partial
first diffraction order information can be captured by the
lens, resulting in imaging capability.
Partial coherent illumination , therefore , is desirable as
mask features fall below in size.
An optimum degree of coherence can be determined for41
 a feature based on its size, the illumination wavelength,
and the objective lens NA.

The following figure shows the effect of partial

coherence on imaging features of two sizes.

a) is one where aerial images for features are larger than

the resolution ppossible coherent illumination ((here
0.6λ/NA) . As seen , any increase in partial coherence
above σ=0 results in a degradation of the aerial image
produced. This is due to the averaging effect of the
fundamental cosinusoidal components used in image
reconstruction
b) features smaller than the resolution possible with
coherent illumination (0.4λ/NA) are resolvable only as
partial coherence leaves increase above σ=0.

For every feature size and type , there exists a unique

optimum partial coherence value that allows the greatest42
image improvement while allowing the minimum.
degradation.
Focus effects also need to be considered as partial
coherence is optimized.

43
Off-Axis Illumination

By controlling the distribution of diffraction information in

the objective lens, maximum image modulation can be
obtained.

A coherently illuminated mask grating imaged through an

objective lens.

Here , the ± 1 diffraction orders are distributed

symmetrically around the zeroth order.

A system where illumination is obliquely incident on the

mask at an angle so that the azeroth and first diffraction
orders are distributed on alternate sides of the optical axis.
It can be shown that the minimum k factor for this oblique
44
condition of partial coherent illumination is 0.25.
45
Off-Axis Illumination(OAI)

46
Until recently, the standard form of illumination for
lithographic lenses was a circular pupil fill centered in the
entrance pupil of the projection optics. The only variable
that lithographers played with was the pupil filling ratio
that determines the degree of partial coherence in the
image formation.

In 1992 , Canon and Nikon introduced quadrupole

illumination that has significant benefits for imaging small
features. Today , several different pupil illumination
patterns are available on steppers, often as software-
selectable
l bl options.
i These
Th can be
b selected
l d as appropriate
i forf
each mask pattern that is exposed on the stepper.

A grating mask ,illuminated by a single point illumination ,

will create a series of diffracted images of the illumination
source in the pupil of the lithographic lens.
The lens aperture acts as a filter , excluding the higher
diffracted orders. When the lens recombines the diffracted
orders that fall within its aperture
p , it forms an image
g with
the higher spatial frequency content removed.

A grating with the minimum resolvable pitch will cast its ±

1st order diffraction just inside the lens aperture along with
the undiffracted illumination point at the center of the pupil
(the 0th order diffraction order).
47
The diffraction from gratings with smaller pitches will fall
completely outside the pupil aperture , and the gratings will
not be resolved.

The high spatial frequencies contained in the 2nd and higher

orders are lost, producing an aerial image with reduced
contrast. When the mask pitch becomes so small that the ±
1st diffracted orders fall outside the lens aperture, the image
contrast falls to zero.

If the point of illumination is moved away from the center

of the pupil , closer to the edge of the aperture , the it is
possible to resolve a grating with a smaller pitch then can
be resolved with on-axis illumination.The -1st order will
now fall completely out of the pupil aperture , but the 0th
and +1st orders will be transmitted to form an image. 48
The asymmetry between the 0th and +1st order leads to
severe telecentricity errors. However, the symmetry can be
restored by illumination with two point source placed on
opposite sides of the pupil. The final image will be
composed of the 0th and +1st diffracted orders from one
source and the 0th and -1th diffracted orders from the other .
The type of illumination is called dipole illumination , and
it provides the best possible resolution for conventional
masks with features oriented perpendicular to a lone
passing through the two illuminated spots in the pupil. On
the other hand, features oriented parallel to this line will
not see the
h dipole
di l illumination
ill i i andd will ill have
h a muchh larger
l
resolution limit.

49
The quadrupole illumination introduced by Nikon and
Canon is a more clever way of achieving dipole
illumination along two axes.
The four illuminated spots are placed at the end of the two
diagonal lines passing through the center of the pupil. This
provides two dipole illumination patterns for features
oriented along either the x or the y axis.

Three unconventional forms of pupil illumination.

The shaded areas represent the illuminated portions of the
circular pupil.
( ) dipole
(a) di l ill
illumination
i i that h provides
id a benefit
b fi only l for
f
lines oriented along the y axis.
(b) provide benefits for lines oriented along the x and y
axes but gives much poorer results for 45 degrees angle
lines.
lines
(c) provide milder benefits than dipole or quadrupole
illumination , but the benefits are independent of feature
orientation.

50
The most attractive feature of off-axis illumination is its
relative ease of use. The stepper can provides any of the
illumination patterns by inserting an aperture at the
appropriate location in the stepper’s illuminator. Often , a
series of apertures can be provided on a turret, and the
particular aperture desired for any mask can be
automatically supplied by the stepper control program.

The most difficult design issues for these forms of

illumination are the loss of the illumination intensity when
an off-axis aperture is used and maintain of good field
ill i i uniformity
illumination if i when
h changing
h i from f one aperture
to another.

51
Materials

52
Materials

In the range of short wavelengths , the transmission of the

material is a problem , the possibilities in the optical design
are limited.

In the lithographic application with the g- and i-line , some

of the optical glasses can be used. In the i-line region, most
of the heavy flint glasses already have reduced transmission.
Absorption effects are especially a problem in lithography ,
since the energy deposition causes thermal degradation in
the
h system performance.
f

For the wavelengths 248 nm and 193 nm , fused silica (Sio2)

and calcium fluoride (CaF2) are the most common
materials.
materials

For 157 nm , only CaF2 has acceptable transmission . If

immersion systems are considered , the properties of water
are of interest. Since the immersion layer
y has qquite a small
thickness in the range of some mm, the loss of energy is
small and the impact of the temperature sensitivity on the
system performance can be tolerated.

Bulk transmission of a sample with thickness 10 mm for

materials BK7 ,water , fused silica, and CaF2 :
53
Fused silica is one of the materials which has good optical
performance down to 193 nm.

Fused Silica

54
CaF2

Water

55
56
Polarization

57
Polarization

For lithographic system with higher numerical aperture,

polarization effects become important.
There are several reasons for the fact that polarization
becomes increasing important in the development of
lithographic lens design.

There is the physical effects of vectorial diffraction at high

aperture angles. In the high NA case, the transverse electric
field components in the resist are no longer parallel and
only
l partial
i l constructive
i interference
i f take
k place.
l

There are no problems in the case of low NA ,case a).

For high NA , in case b), only the s-component can
interfere completely.
The lithographic imaging at high NA or hyper NA can be
58
achieved more effectively with s-polarization.
The exposure latitude for a dry and an immersion system
with and without polarization as a function of NA.
The exposure latitude is a metric to describe the tolerable
variation of the does without running out of specification.
It can be seen that the polarized case has a larger exposure
window.

Another aspect of polarization influence is the

performance of the coating at the higher ranges of
incidence angles .
The problem caused by polarization on the mirror surfaces
with large incident angles can be seen in the following
59
figure.
catadioptric uniaxial (a) catadioptric folded (b)
Typical ranges of operation are indicated I green (a) and
yellow (b) respectively.
Th are two
There t effects
ff t ini the
th coatings,
ti which
hi h lead
l d to
t a
decrease in the performance. A reduced transmittance
(left column) and a phase splitting (right column) can be
seen .
For a folded catadioptric design the phase splitting is a
factor of four larger than for the uniaxial systems, with
considerable smaller incidence angle variation.

Another effect that limits p

performance is the intrinsic
birefringence of some materials and also the stress
induced birefringence due to the process of the lens
60
material.
Today, the management of the polarization of the
illumination and projection systems is one of the most
aspects used to control the performance of lithographic
system.
In high-performance systems , the illumination systems
must generate a radiation with a particular defined
polarization state and the projection system must not
change this distribution.

61
Performance of Lithographic
lenses

62
Requirements

Typical performance numbers for a lithographic lens

design for 193 nm and a numerical aperture of NA=0.9 :

The telecentricity is controlled over the data of the chief

ray distortion for defocusing.
Depending on wavelength and numerical aperture, the
chief ray must fulfill the 1 nm criterion over the full depth
of focus. This typically delivers a chief ray angle errors
63
smaller than 1 mrad.
Representations of Performance

Since the residual wave aberrations of lithographic

systems are very small . There is hardly any opportunity to
see any differences in the point spread function of the
system over the field.

The last row shows the difference between axis and field
64
position. The spatial axis is the scaled in um.
Therefore, the point spread function, the modulation
transfer function or the Strehl ratio are not appropriate
measures to qualify a lithographic projection system or to
represent the performance.

The wave aberrations allow us to characterize the residual

system aberrations. The tilt and defocus are always
removed by readjusting the image plane position and
orientation. This can be taken into account by considering
only the Zernike coefficients, starting with the term Z5.
It is usual to use the first 37 Zernike coefficients as a
f
function
i off the
h field
fi ld size
i andd the
h rms value
l off the
h complete
l
phase surface.

A diagram for the performance representation of a lens.

Typical values are 5 mλ for the total rms value and 3 mλ
for the symmetry groups of spherical aberration, coma,
astigmatism and trefoil error, respectively.

65
It has been reported , that if a projection lens is designed
with one waist and aspherical surfaces, the residual
aberrations may be dominated by trefoil errors.

In practical lithographic projection applications , all kinds of

geometries and shapes of reticle structures are possible. For
the critical range near the resolution limit, straight lines and
contact holes are the dominating feature types.
The isolated dots represent contact holes on a wafer , and
straight lines , which correspond to conducting metallic
wires or tracks.

The influences of the wavelength and NA on resolution and

contrast for incoherent illumination:
The width of the bars is scaled in Airy diameters as a
parameter a. The decrease in the visibility V with a down-
66
scaling of the bar-width can be clearly seen.
For a bar width of exactly one Airy diameter a=1.
The followingfigure shows the changes in the imaging
conditions
diti when
h th the coherence
h and
d th
the shape
h off th
the pupil
il
illumination is changed.

67
Some examples illustrating the proximity effects for
completely coherent illumination. If two structures are
close together , there is an interaction due to the finite
range of the diffraction spreading.

For incoherent illumination , the proximity effects are

weaker, in the first row, shows an image of five bars with a
width of a and a separation of p.

68
The influence of residual aberration on the imaging of the
corner object for incoherent illumination.
The bar width is one Airy diameter corresponding to aa=1.0.
1.0.

The influence of the coherence on the resolution is shown:

The curve shows the line width of isolated lines with a
width of 180 nm for an imaging with NA=07 at λ=248 nm
as a function of the coherence parameter σ. The line width
decreases for increasing coherence parameter.

69
If a line is imaged in the image plane , and some defocused
planes, are analysed , the behavior of the width and peak
intensityy is a ggood indicator of the image
g qquality.
y
The recording of a z-stack of images to evaluate the system
performance follows the idea of Bossung.
Typical Bossung plot:
A threshold intensity is used , which corresponds to the
exposure threshold
h h ld off theh resist.
i
The dependence of the width of the bar image as a function
of the defocus then combines the effect of broadening and
decrease in the intensity.

One of the major aberrations , which must be corrected in

lithographic systems up to a very small residual value is
the distortion. This deviation consists of two parts. First ,
the pure geometrical distortion of the chief ray has to be
considered
id d . F Furthermore
th , th
the shift
hift off the
th center
t off gravity
it
of the point spread function due to coma-like wave
aberration and odd aberrations of higher order results in70an
additional contribution.
Chromatic Aberrations

71
The lithographic projection systems of the early days
worked at the wavelengths of the g- or i-lines. It is possible
to use glasses in this spectral range and there are good
opportunity to correct the color aberration.
Typical linewidths for these sources for lithography
applications are

There are two facts, which are critical in this range.

First, the secondary spectra are a huge problem and are a
real challenge in the design.
The second aspect is the production of the glass and its
tolerance.
l Since
Si the
h melting
l i process is i not invariant,
i i every
melt results in a slightly different value of the refractive
index. This requires re-adjustment of the radii and the
central thickness of the lenses every sample.

Switching to the excimer laser sources and the wavelength

λ= 248 nm , the spectral conditions and materials change
completely.
Glasses are no longer
g available due to absorption
p of the
wavelength. In this region , fused silica and calcium
fluoride are the only materials which can be used.
The narrow band excimer lasers have a bandwidth in the
range of
In the recent development of the laser sources , this range
has been reduced to This small value
72
makes the chromatic correction of a system easier.
Furthermore, due to the high energy density , it is not
feasible to use cemented components. A lack of
achromatization can be clearly seen in the decreased
performance.
If a system at 248 nm with a projection lens made of fused
silica only is considered and the wavelength is changed by
manipulating the laser , the position of the best focus
plane changes with the λ-shift ,as can be seen.

If a wavelength λ=193 nm is used, both materials are also

possible, and the classical combination of two materials
helps in correcting the color aberrations.

73
For even shorter wavelength like λ=157 nm , only CaF2 is
possible as a lens material. Furthermore, the bandwidth of
the F2 sources is larger and lies in the region of
for one line. Therefore, other concepts for the correction of
the axial color are needed for these systems. The well
known Schupman principle to correct axial color with the
help of a mirror and a negative nearby lens is a
conventional way to control this problem.
This has led to the use and development of catadioptric
lithographic lenses. The additional advantages for the
correction of the Petzval curvature are welcome in those
systems which have concave mirror component inside
them.
h

The optical design shown below is designed for λ=157 nm.

The correction of the axial chromatic error is done in this

system using the Schupman principle.
The negative lenses in front of the concave mirror
introduce a positive axial color contribution, which can be
used to compensation
p for the contributions of the real part
p
of the system lenses.
The figure shows the Seidel surface contributions of this
system. The axial color distribution around the mirror and
the Schupman principle can be clearly seen and indicated
by the red box.

74
75
Petzval Curvature

76
Petzval Curvature

The refractive power of a combination of several lenses is

given by the sum of the individual refractive powers of the
lenses , weighted by the marginal ray height:

On the other hand, the Petzval curvature is equal to the

sum of the individual refractive powers , weighted by the
i
inverse off the
h refractive
f i indices
i di off the
h lenses:
l

This leads to the concept of the Petzval correction by

combining low index strongly curved negative lenses at
small marginal ray heights with high-index weak positive
lenses at large ray heights. The overall setup therefore
corresponds
p to a system
y with bulges
g and waists.

Example 1: a Lithographic lens system

77
The surface contributions of the Seidel Petzval aberration :

The surface index ranges for the three bulge and two waist
region are indicated by yellow and green background
colors.
Th change
The h off sign
i for
f the
h Petzval
P l contributions
ib i andd larger
l
absolute values for the waist regions , due to the smaller
marginal ray heights .

Example 2: Catadioptric systems

In catadioptric systems the Petzval correction is performed

to a great amount by the concave mirrors.
The strongg negative
g contribution of the mirror to the
Petzval term is clearly seen.

78
79
Telecentricity

80
If multiple layers are processed in lithographic
applications , several planes are overlaid. It must be
ensured , that small deviations in z direction do not
influence the lateral position.
This is the reason why it is so important to fulfill the
following conditions:
1. Telecentricity in the image space.
2. Extremely good corrected distortion aberrations.

Therefore, it is necessary to consider and calculate all

effects contributing to this aspect with high accuracy.
O hhas to control:
One l
1. The chief ray angle in the image space
2. The coma aberration
3. The pupil aberration
4 The vignetting
4.

Any vignetting effect or coma aberration causes deviations

between the chief ray and the ray through the centroid of
the ppoint spread
p function.
Furthermore, if the imaging of the pupil has aberrations,
the telecentricity is also perturbed.

Different effects of a telecentric error on the imaging.

Usually, dependent on the overall performance of a system ,
a lateral offset of 1-2 nm is allowed in the exposure of
81
overlay structure.
If the depth of focus of a system is taken into account , the
acceptable error in telecentricity can be calculated.

Th telecentricity
The l i i error off a lithographic
li h hi system:

82
The curves for the pure chief ray angle and the centroid ray
are shown for the optimized position of the diaphram and a
defocused stop.
It can be seen that the stop position helps in optimizing the
telecentricity . In the best case , the residual telecentricity
error of this system is in the region of 0.031 ° corresponding
to 0.5 mrad.

83
Aspherical Systems

During the development of the requirements :

1) NA is increased to obtain smaller features in the
application
2) The field size has increased in order to obtain a larger
process speed
This leads to an increase in the number of lenses and the
bulges become increasingly large in diameter.
The distance of the object and the image increases from
400 mm to 1300 mm.

This has caused a huge effort in optical design in order to

reduce the size and cost and because there is a need for
large stocks of material.
The use of aspherical surfaces in the design leads to more
compact systems. This forms the second generation of the
lithographic lenses.
The need for aspherical surface is obvious, especially for
larger
g numerical apertures.
p This reduces the volume
Considerable.
Here two cases are shown , one with purely spherical
surfaces and one with several aspherical surfaces. Both
examples are system designed for wavelength 248 nm , a
field size of 26 mm and the same NA=0.8. the decrease in
length ( 15%) and diameter (10%) is indicated. This
84
reduces the volume by approximately 43%.
85
Immersion Systems

The use of immersion fluids to enhance the numerical

aperture and to increase the resolution is a method which
has long been used in microscope.
However , the use of liquid immersion systems in
lithography is quite new. Today, immersion lithography at
193 nm with distilled water is possible and a new
generation of systems and optical designs has been
developed.

The use off immersion

Th i i fluids
fl id offers
ff the h opportunity i to
obtain numerical apertures above the limiting value of
NA=1 for dry system. For a wavelength of 193 nm, the
refractive index of water is n=1.437.
Due to a pure refinement in optical design,
design coating
technology and polarization solutions, the aperture angle
can be increased up to a value of approximately 72°. For a
dry lens , this corresponding to a limiting NA=0.95. If
water is used as the immersion liquid
q , the limiting
g value
is NA=1.35.

In addition to the increased lateral resolution , another

advantage for lithographic application is the increased
depth of focus when immersion liquids are used.

86
Early Lithographic lens

The early lithographic lenses work at the wavelengths 436

nm (g-line) , 405 nm (h-lin3e) or 365 nm (i-line), and are
made of glass.
Since the requirements in the early day of optical
lithography were considerably less in comparison with the
demands of today, the performance of these older systems
is very different from the modern systems.

Example 1 Glatzel

The lens is made of glass and designed for the wavelength

365 nm.
NA=0.33
The two bulges to correct the Petzval curvature can be seen.
seen
There are still cemented components inside the layout and
there is no telcentric path of the chief ray in the object
space.
The reduction factor is m=0.1 and the total track from the
object to image is 300 nm.

87
Example 2 Shafer

It woks at wavelength 405 nm.

NA=0.28
Reduction factor m=0.2
Most of the lenses are made of fused silica , but glass
component are also included. There are 14 lenses in the
system, but cement components are avoided.
The two bulge design is similar to the proceeding system.

Example 3 by Bratt

There are 18 lenses, the magnification is m=0.2, the NA is

0.36,, the wavelength
g is 405 nm and the components
p are
made of glass.
Cemented components are included.

88
Refractive Projection lenses

The lithographic lenses in this family starts with a

moderate NA of 0.5 at the wavelength 365 nm, the i-line
of the mercury spectrum.
Later the wavelengths change to the pulsed KrF Excimer
laser at 248 nm and the ArF laser at 193 nm.
Furthermore, the use of aspheres helps in reducing the size
of the system.
As a last step, water immersion is used to increase the
resolution again.

Example 1 by Matsuzawa

Wavelength 365 nm
M=0 20
M=0.20
NA=0.5
27 lenses made of various glasses without cemented
surfaces form a structure with three bulges
The rayy ppath of the system
y is telecentric in the object
j
space.

89
Example 2 by Schuster

Wavelength 248 nm
NA=0.63
m=0.25
31 l3nses made of fused silica are used in this design.
The total track is larger than 1 m.
The diameter of the three bulges increases towards the
image side.

90
Example 3 by Kotaro

Wavelength 248 nm
NA=0.75
m=0.25
29 lenses made of fused silica are used in this design.
Two aspheres are included in the system in the region of
the last waist to control spherical aberration for the large
numerical aperture.

Spot diagram of the lens:

The red circle indicate the Airy diameter. All spots are
91
inside the circles .
Multi-axis Catadioptric Systems with Mirrors

Beam splitting using a plane mirror and separation of the

ray bundles by geometrical means is another type of
catadioptric system.
One possible solution is to generate an additional
intermediate image plane at the location of the mirror.

Example 1 Shafer

NA=0.58
Th splitting
The li i mirror
i is
i located
l d near the
h reticle
i l plane
l andd is
i
therefore near to a field plane.

92
Example 2 Kato

Designed for wavelength 193 nm

m=-0.25
NA=1.24
Two uncorrected intermediate image planes

93
Seidel surface contributions of the lithographic lens:

94
Example 3 Omura

Example 4 Shafer

Two folding arms are used to double the advantageous

effect of the Schupman principle.

95
EUV Mirror Systems

Pure mirror systems are mainly proposed for the EUV

range with

The challenges of this type of systems from the viewpoint

of the optical design are an obscuration-free geometrical
arrangement, a large ring-shape field of view and small
ranges of the incidence angles on the surfaces.
The field of view of this kind of lithographic lens designs
is considerable smaller than in the case of refractive or
catadioptric
di i systems.

One major problem in the performance of lithographic

systems for lies in the poor reflectivity of the
dielectric multilayer reflection coatings.,which
coatings which are in the
region of approximately 70%.
A large variation of the ray incidence angles over the
surfaces of the mirrors decreases the transmission and the
performance due to the polarization
p p effects significantly..
g y
One goal in the design of EUV systems is a small
incidence angle range.

First example:
an obscuration-free four-mirror EUV lithographic
projection lens design
96
NA=0.1
Field 15 mm in the image plane
Reduction ratio m=0.25
All mirrors are aspherical , but are centered on a common
axis and are used in off-axis configuration.

Example 2

Six mirrors with one intermediate image plane according

to Williamson.

97
The system surfaces are axis symmetrical. It has a
magnification factor of m=0.25 , the NA=0.5 on the wafer
side.
The rms value of the wave aberration as a function of the
field coordinate across the small ring-shaped field of view.

The maximum difference in the incidence angles for every

mirror of the lithographic mirror system in the plane x=0.0

98
It can be seen , that there are quite large variations, in
particular at the mirrors 2 and 5.
This causes problems in the performance of the coatings at
these surfaces.

Example 3

A system with four mirrors and a NA=0.14

A ring field width of 1 mm
An intermediate image plane after two mirrors
The system corresponds to the family of symmetric layouts.

99
Example 4

Designed by Mann
A large NA=0.7
Eight mirrors and a central obscuration of 5% of the pupil
area
Magnification factor m=0.125
The ray path has two intermediate images
The ring field has a width of 13 mm x 1 mm with a radius
of 18 mm
The mirrors can be divided into the field-defining group
with
i h m11 to M6 and
d the
h aperture group with
i h M7 andd M8
M8.

100
101
102
103
104
105
106
Determination of mask induced polarization effects
occurring in Hyper NA immersion lithography

107
108
109
110
111
112
113
114
115
A design of Beam Shaping Unit for
193 nm Lithography Illumination
System using Angular Spectrum
Theory

116
Introduction

1. The projected lithography using 193 nm exposure

wavelength is the mainstream equipment for
manufacturing grand scale ICs. This equipment is used
to achieve 90 nm node critical dimension (CD) , then
65 nm node ,45 nm node and even below.
2. The most important part of the lithography id the
projected exposure optical system , which consists of
illumination system and projection objective system.
3. The effects of illumination system are supplying high
uniformity
if i bbeam iintensity
i di
distribution
ib i on the h maskk ,
controlling exposure does and realizing several kinds
of OAI which can be used to enhance resolution and
increase depth of focus (DOF).
4
4. Illumination system contains beam expanding unit,
unit
beam transmitting unit, energy controlling unit, beam
shaping, homogeneous unit, and relay objectives. The
evolutions of all the units are for enhancing uniformity
mainly.
y
5. In order to reach up to the diffraction-limited
resolution , several resolution enhancement
technologies ( RET) are proposed , including OAI
technology, optical proximity correction (OPC)
technology and phase shifting mask technology.
6. OAI is widely used because of its mature technology
117
and advantages of realizing annular illumination mode,
 quadrupole illumination mode and dipole illumination
mode by using beam shaping unit of illumination
system.
7. Generally , there are two methods to design beam
shaping unit , which are based on geometrical optics
method and physical optics method respectively. The
former utilizes the cone angle of axicon prisms to
split incident beam to realize annular illumination
mode and the partial coherent factor σ can be adjusted
continuous by changing the relative distance d
between axicon prisms.
8
8. W define
We d fi σ as the h ratio
i off NA off ill
illumination
i i system
to he NA of the project objective system. Best
resolution of different CDs can be obtained by
adjusting σ.
9
9. The advantages of axicon prisms are their simple
structures and they are easy to be manufactured and
aligned. Furthermore, quadrupole illumination mode
can be realized by adding pyramidal prisms in the
beam shaping
p g unit. However,, the disadvantagesg are
the weak ability of controlling beam intensity and
they can not implement dipole illumination mode.
10. On the other hand, DOE , designed with physical
optics method, can realize some kinds of OAI modes,
including dipole illumination mode. DOE is a kind of
optical element which has the characteristics of high
118
diffraction efficiency. It has multi-stepper structures
 which are fabricated by electronic direct-writing and
ion beam etching equipments. This kind of structure
controls intensity distribution by modulating phases.

OAI Technology

The main effects of OAI in illumination system of

lithography are enhancing resolution and increasing DOF.

The dense lines on the mask make the incident beam

generate evident diffraction effects , ± 1 orders diffraction
light distribution at both sides of 0 order diffraction light
symmetrically.
119
When traditional illumination is used , if the CD size on
the mask were smaller than a certain line width value Ro ,
the separation angle θ between 0 order and +1 or -1 order
diffraction light would be large than NA, so high orders
such as ± 1 orders diffraction light could not entrance into
the projection objectives, resulting the CD size Ro could
nor be resolved.
On the other hand , when off-axis beams with certain
angles illuminate the mask which has the same CD size
Ro , 0 and +1 order diffraction light are able to entrance
into the project objectives to format interference images,
although the -1 order diffraction light deviates away from
the
h objective
bj i system.

Secondly, OAI has a smaller NA and based on the

Rayleigh DOF equation:

where k2 is the processing factor, it is obvious that smaller

NA can obtain larger DOF.

In order to resolve smaller CD sizes, OAI modes include

annular illumination, quadrupole illumination and dipole
illumination
Annular illumination has the same resolution in all random
directions,
Quadrupole illumination enhances resolution in two
orthogonal directions with have 45° separated angles 120
relative to the cross lines of four poles..
Dipole illumination supplies optimal resolution in the
direction perpendicular to the direction of linking two
poles.

Angular Spectrum Theory

The accuracy of designing DOE is mainly influenced by

two factors: physical approximates errors and numerical
errors.
Th former
The f are about
b paraxial
i l approximates
i , andd the
h latter
l
are sampling distance errors.
The methods of designing DOE are based on scalar
diffraction or vector diffraction theory. Scalar diffraction
method has high computing efficiency compared to the
vector diffraction one. However , the disadvantage of
scalar diffraction method is low accuracy when handling
near field diffraction propagation.

The angular spectrum theory is:

The intensity distribution of the incident beam on the DOE
is represented as
The complex amplitude of wave is transmitted by Fourier
transform:

121
The effect of Fourier transform is decomposing a complex
function to a set of exponential functions , so Uo is
represented by Fourier spectrum functions:

On z=0 plane, exponential function is

considered as a plane wave , and the direction cosines of
which are:

is called the angular spectrum of

The relationship between and is

found as:

Finally, the field distribution is obtained by

inverse Fourier transform.

DOE designing parameters calculation

122
Performance of Polarized
Illuminators in Hyper NA
Lithography Tools (Nikon)

123
Continuing reduction of IC critical dimensions drives
projection tool technology into the hyper NA region where
projection lens numerical aperture exceeds 1.

Under such conditions ,fundamental properties of light

enhance image formation by polarization states , opening
up the illuminators designs with well controlled
polarization characteristics.

Theory

From the
F h theory
h off vector iimaging,
i image
i contrast is
i lower
l
than expected by scalar imaging theory.
The is because the z-components (direction of optical axis)
of electric field of the light generates inverse contrast image
on the image plane.
plane

If the polarization orientation is in a direction perpendicular

to the direction of a pattern, which is called TM mode., or
p-polarized
p p light
g , contrast is lower than expected
p byy scalar
theory due to inverse contrast image of z-component of the
light, while if the polarization direction is parallel to the
pattern orientation , which is called TE mode, or s-polarized
light , image contrast is the same as expected of the scalar
theory.

124
Illustration of the vector image theory:
In the case of un-polarized illumination , the image is a
superposition of the TE light and the TM light image.
Then total image
g contrast is degraded
g due to TM light
g
imaging. This effect is increases in accordance with the
increase of NA of the projection lens.
In order to keep the image contrast high as expected by
the scalar theory , TE , or s-polarized light imaging has
been proposed for higher NA imaging , such as higher than
0.92NA.

Optical Simulation

We use vector model imaging in resist:

In the case of a scalar model ,complex image amplitude125is:
I
Intensity
i off the
h image
i will
ill be:
b

For image of the vector model: the above equation should

be expanded to the next 6 equations:

126
where are polarization
Polarized Light Illuminator

A polarized illumination system:

Polarized laser beam:

Light
g from 193 nm ArF laser for DUV lithographyg p y is
generally linearly polarized in one direction. The degree of
polarization (Dop) is expected to be ～0.97 and stable
exposure.

Beam –delivery system

Calcium fluoride (caF2) is used in the beam delivery
127
system lens material. This is because CaF2 optics can
endure high energy density to laser radiation better than
fused silica.
On the other hand, CaF2 has the property of birefringence ,
which disturbs the polarization states. In order to minimize
the effect of the birefringence , the CaF2 optical elements
are installed in the illuminator with crystal axes aligned so
as to have minimum effect on the polarization states.
Even so , we may have some amount of birefringence due
to phase jump on folding mirrors . As a result , linear
polarized light may become elliptically polarized.

Lossless
L l polarization
l i i controller ll
We need to control the polarization status. By using two
wave plates, elliptically polarized light can be transformed
to linear polarized light with the required orientation.
Polarization state after a wave plate , which has the angle
of axis Φ and phase angle γ , is expressed as the following
equation with the Mueller’s matrix and Stokes parameters:

128
In the case of Quarter Wave Plate , Muller’s matrix is
transformed as:

This matrix shows that by setting proper orientation angle

of optic axis of the QWP , any elliptical polarization can
be transformed to linear polarization status (S3’=0).

In the case of half wave plate , Muller’s matrix is

transformed as:

This matrix shows that by y setting

g pproper
p orientation angle
g
of optic axis of the half wave plate , any linear
polarization status can be rotated to any direction while
not introducing circular polarization.
As a result of utilization of a quarter wave plate and a half
wave plate any elliptical polarization status can be
transformed to vertically or horizontally linear polarized
129
light, which is s-polarized status for diffracted light from
vertically or horizontally oriented pattens respectively,
without practical power degradation.

130
Mask induced polarization effects
at high NA

131
Polarization effects at the photomask become a concern
with decreasing device dimensions and increasing NA.

Minimum pitch values (Λ) on a photomask approach:

for a system with M=1/m , where m is the magnification

of the lens.
Example:

m=0.25
0 25 (M=4
(M 4 ffor a 44x reduction
d i system))
NA=0.85
σ=1.0 for current generation tools
= 454 nm at λ=193 nm
110 nm at the image plane

Next generation tools moving towards NA=1.2 will result

in a minimum mask period of 1.67 λ (322 nm at λ=193 nm,
80 nm at the image
g plane),
p ), at which point
p polarization
p
effects contributed by the photomask are more significant.

High NA imaging and Polarization

If TE polarization is defined as polarization perpendicular

to the plane of incidence (POI) , then radiation polarized in
132
the POI is referred to as TM polarized.
The angle of incidence at the wafer’s
wafer s surface becomes
more oblique with increasing NA.
For TE polarization NA is not concern since there is no
decrease in image contrast due to increased angle of
incidence (θ).

133
Traditional wire grid- polarizer allow the attenuation of the
TE polarized radiation , while allowing the TM state to
pass.
The electric field of the TE polarized light induces a
current in the length of the wires. Forward transmitted
radiation is out of phase with the incident TE wave and
exhibits greatly reduced intensity.
The electric field of the TM polarized light is
perpendicular to the wires , a dimension in which the wires
are very narrow and restrict the motion of the electrons.
Therefore, most of the TM radiation is transmitted
unaffected.
ff d

The parameters impacting the efficiency of a traditional

wire grid polarizer include period, duty cycle,
thickness ,shape
shape , and materials,
materials Johnson explains the
impact of these features using Rigorous Coupled Wave
Analysis (RCWA). The grating period is the most
important of these parameters as it determines the
minimum wavelength g that can be ppolarized for a specific
p
diffracted order , as shown:

For feature sizes greater than twice the wavelength , a

diffraction grating results with little to no polarization
134
effects.
For feature sizes less than half of the wavelength , the mask
acts as a zero order wire grid polarizer.

In the transition region, the mask polarizes and diffracts the

radiation , and certain materials are capable of transmitting
more TE polarization than TM. Wire grid polarizers passing
predominantly TE polarization are appropriately named
inverse wire grid polarizers , and have been studied by
Honkanen.

Traditional wire grid polarizer features are relatively large

andd are similar
i il ini size
i to the
h wavelength
l h off the
h light
li h
incident upon the polarizer; the wires approximate a planar
metal surface.
Inverse wire grid polarizers utilizes effects similar to those
that occur in “small”
small noble metal (Au
(Au,Ag,Cr,etc)
Ag Cr etc) clusters
clusters.
The conduction electrons act as an oscillatory system in
clusters , but not in bulk.
Sufficiently small mask features can be approximated as
clusters in the small dimension ((width)) ,as
, a bulk material
in the long dimension (length). Therefore , TE polarized
light sees a quasi-infinitely long structure, thus it looks like
bulk material, and TM polarized light sees a small cluster ,
and induces a particle plasmon oscillation.

135
Experimental Setup

Two types of photomasks were examined: a binary mask

and attenuated phase shifting masks (APSM).

Binary mask
The polarization effects were simulated using RWCA and
then compared to experimentally obtained results using a
Sopra variable angle spectroscopic ellipsometer (PUV-
SE5).
Once this data was analyzed , and RCWA was shown to be
an acceptable
bl means for
f measuringi photomask
h k induced
i d d
polarization.

Binary mask material used:

APSM
The polarization effect is also simulated using RWCA.
ASPM material composition:

136
Projection
j Exposure
p Method and
Apparatus (Nikon_5467166)

137
General construction of a projection exposure apparatus

Condenser lens 10,14

The pattern surface of the reticle R and the emergence side
surface (plane light source) of the fly-eye lens 7 are
optically in Fourier transform relation by the composite
system of the condenser lens 10, the lens system 12 and the
condenser lens 14.
By the condenser lens 10, the emergent side surface of138
the
Fly-eye lens 7 (the surface on which the secondary light
source images are formed) is made into a Fourier transform
plane for the rectangular opening surface of the reticle
blind 11.

Accordingly, the beams of light diverging from the

plurality of secondary light source images on the fly-eye
lens 7 and having entered the condenser lens 10 become
parallel beams of light slightly differing in angle of the
incidence from one another on the reticle blind 11 and are
superposed one upon another.

The rectangular opening surface of the reticle blind 11 and

the pattern surface of the reticle R are disposed conjugately
with each other by the composite system of the lens system
12 and the condenser lens 14.
14

139
The imaging path of a projection optical system according
to the prior art:

Both of the reticle R side and the wafer W side are made
telecentric or one in which onlyy the wafer W side is made
telecentric is popular . The rays L2 and l3 cannot pass
through a pupil ep located on a Fourier transform plane
140
FTP in the projection optical system.
Excimer Lasers for Advanced
Microlithography

141
Since its introduction in 1987 , the excimer laser for the
stepper has evolved to the most important light source.

Excimer lasers provides direct deep-UV light , are scalable

in energy and power , and capable of operating with
narrow spectral widths.
Also, by providing three wavelengths at 248,193, and 157
nm , excimer lasers span three generations. They have
large beams , and a low degree of coherence.

Introduction

A transition from using i-line (365 nm) mercury lamp to

deep-UV excimer laser-Krypton fluoride (248 nm)-as the
illumination source :

This transition was revolutionary because of the

complexity and pulsed nature of the laser compared to the
simple and continuously operating lamp. That was 1995.

By 1997, the Hg i-line light sources were replaced with

excimer lasers in volume manufacturing of semiconductor
devices.
Today m there are more than 2000 excimer-laser-based
scanners in use in over 30 semiconductor factories
worldwide.
142
The spectral power (ratio of power and linewidth ) of KrF
and ArF lasers has increase in the past ten years.

143