ArticlePDF Available

Some aspects on mechanisms responsible for contamination of optical components in DUV lithographic exposure tools

Authors:

Abstract and Figures

Lithographic exposure tools in the deep-ultraviolet (DUV) region face challenges with contamination. Airborne molecular contamination is generally recognized as a severe threat in high-volume production of integrated circuits (ICs), and has recently also become of a concern in patterning of masks. When using high-energy photons at 248nm wavelength or lower, the risk of contamination may increase due to higher potential of breaking molecular bonds of organic species in the ambient of the optics. Especially resist outgassing during exposure may result in a build-up on the surface of the lens. The photodissociated molecules may readily deposit on the optics depending on the interaction between the contaminants and the lens surface and possibly cause a loss of transmission of light with time. Eventually the growth of the deposit will severely impact the throughput of the exposure tool, and in the worst case, necessitate a replacement of lens elements. Contamination control is therefore crucial for cost-effective DUV wafer and mask manufacturing. Trustworthy measurement methods as well as deep understanding of the mechanisms involved are of vital importance in order to understand and prevent molecular contamination. This paper discusses some of the factors influencing the deposition of hydrocarbon contaminants and also simulation work related to investigation of resist outgassing and contamination issues in the Sigma7300 laser pattern generator.
Content may be subject to copyright.
Some aspects on mechanisms responsible for contamination of optical
components in DUV lithographic exposure tools
Hans Fosshaug*
a)
, Mats Ekberg
a)
, Gunnar Kylberg
b)
a)
Micronic Laser Systems AB, P.O. Box 3141, SE-183 03 Täby, Sweden
Phone: +46 8 638 52 00, Fax: +46 8 638 52 90, e-mail: hans.fosshaug@micronic.se
b)
F:a Kylberg Teknik, Gunnar Kylberg,
Svedjevägen 15, SE-16754 Bromma, Sweden
email: gunnar.kylberg@swipnet.se
ABSTRACT
Lithographic exposure tools in the deep-ultraviolet (DUV) region face challenges with contamination. Airborne molecular
contamination is generally recognized as a severe threat in high-volume production of integrated circuits (ICs), and has
recently also become of a concern in patterning of masks. When using high-energy photons at 248nm wavelength or lower,
the risk of contamination may increase due to higher potential of breaking molecular bonds of organic species in the ambient
of the optics. Especially resist outgassing during exposure may result in a build-up on the surface of the lens. The
photodissociated molecules may readily deposit on the optics depending on the interaction between the contaminants and the
lens surface and possibly cause a loss of transmission of light with time. Eventually the growth of the deposit will severely
impact the throughput of the exposure tool, and in the worst case, necessitate a replacement of lens elements. Contamination
control is therefore crucial for cost-effective DUV wafer and mask manufacturing. Trustworthy measurement methods as
well as deep understanding of the mechanisms involved are of vital importance in order to understand and prevent molecular
contamination. This paper discusses some of the factors influencing the deposition of hydrocarbon contaminants and also
simulation work related to investigation of resist outgassing and contamination issues in the Sigma7300 laser pattern
generator.
Keywords: Laser pattern generator, contamination, lens, modeling, physisorption, chemisorption, mechanisms
1. INTRODUCTION
Optical lithographic tools, such as laser pattern generators, steppers and scanners, working with deep ultraviolet (DUV)
wavelengths have proven to be vulnerable to airborne molecular contamination (AMC), which threaten the tools in several
different ways.
[1-4]
The importance of contamination control in microlithography has increased in recent years as the
exposure wavelength has been reduced. The reason for this is the increased interaction between groups of contaminants and
the UV light, which is related to an increase in photon energy and larger tendency to bond cracking as well as higher power
density. Contaminants such as hydrocarbons, refractory organic compounds and acidic or basic compounds such as
sulphuric acid and ammonia respectively as well as salts, i.e. (NH
4
)
2
SO
4
, pose a risk to contaminate the lens during exposure.
The contamination may cause transmission loss and image distortion and thereby reduce the yield and increase the cost-of-
ownership. This paper is a first attempt by Micronic to deeply get an understanding of the different phenomena involved in
order to later stipulate a rigorous model in which predictions may be used in turn to analyze the potential risk of
*hans.fosshaug@micronic.se; phone +46 86383515; fax +46 86385290; www.micronic.se
Optical Microlithography XVIII, edited by Bruce W. Smith,
Proceedings of SPIE Vol. 5754 (SPIE, Bellingham, WA, 2005)
0277-786X/05/$15 · doi: 10.1117/12.600312
1601
contamination and improve gas purge system if found necessary. However, a simple model was nevertheless made at this
stage in which the potential risk of resist outgassing and hydrocarbon deposition onto the lens surface was analyzed using
rough assumptions due to lack of necessary data.
In the environment of lithography processing, typical AMCs that have been found are molecular basic gases, acidic gases,
lower boiling-point volatile organic and higher boiling-point condensable organic molecular compounds.
[1,4]
Many of the
compounds originate from processing chemicals, such as developers, stripping agents and cleaning liquids, and despite in
low concentrations, ppb to ppm by volume, their appearance may still affect the lithography tools appreciably. A second
group of AMC is the outgassing products from various sources such as
1. Construction materials due to thermal effects and diffusion,
2. DUV-induced outgassing from materials receiving direct or stray light,
3. Human bodies, clothing, breath and e.g. cosmetics, and finally
4. Photoresists
All the AMC compounds may participate in formation of organic or other types of films on optical components, of which the
image-forming final lens is the most serious and expensive one.
[5,6]
Such films will generally result in
1)
that the optical
absorption on the optical surface increases, and continues to increase with time, and
2)
with time increasing level of scattered
light from that same surface. The absorption will accelerate, since more optical power or energy is required to maintain a
constant exposure dose on the photoresist. This may result in a less controlled exposure dose and altered calibration
conditions. Another effect is that a thin molecular film, having a certain refractive index, may detune the performance of
antireflective coatings (ARCs) on the optical surface, whereby the stray light level in the system increases. The scattering
results in increased stray light levels, too, which threatens the CD control and clear-to-dark feature appearance in the final
photoresist patterns. A second effect of the scattered light may be increased outgassing from surrounding components and
materials that receive this light, a fact which may cause even higher outgassing rates.
Much effort has been spent during the last decade to counteract molecular contamination in lithography tools. These efforts
include:
- Materials selection based on chemical outgassing analyses and
- Proper and careful cleaning are natural steps, as well as
- Clean packaging materials and methods,
- Clean transportation systems and
- General methods for handling the equipment outside and inside clean-rooms.
- Chemical filtration of the tool’s internal atmosphere and used process gases (normally clean dry air, CDA, and nitrogen).
- Suppression of DUV stray light, which coarsely speaking makes almost any construction material to outgas,
- Efficient purging of the most critical optical components and surfaces, and
- General good engineering.
Development and application of chemical filters, based on either physisorption, chemisorption, acid-base chemistry or ion-
exchange chemistry
[7]
require deep knowledge of the behavior of all different AMC compounds, their expected appearance
and abundance, as well as development of detection and analysis methods, in order to be able to tailor efficient systems.
[8-11]
Major contributions to a better understanding of lens contamination phenomena have been published by others
[11,12]
based on
both mathematical modeling and substantial experimental measurements to support the models and to verify their
validity.
[3,8-12]
Most of that work has been made for 193nm and 157nm resists. However, very scarce data are available for
the wavelength 248nm presently used in Micronic’s exposure tool. This paper addresses some of the mechanisms
responsible for lens contamination and will be discussed in terms of adsorption isotherms, Lifshitz-van der Waals and acid-
base interactions. Some simulations using the commercial multiphysics modeling software FEMLAB
£
will illustrate the lens
contamination situation applying realistic geometries for the Sigma7300 DUV laser pattern generator.
1602 Proc. of SPIE Vol. 5754
2. MICRONIC’S TECHNOLOGY
The system principle of the Sigma7300 laser pattern generator is shown in Figure 1. A spatial light modulator (SLM) is
utilized for exposure of the mask blank. Pattern data is rasterized and loaded into the SLM chip, which acts as a reflective
computer-controlled reticle. It reflects the 248 nm excimer laser flash and the image of the pattern is focused onto the mask
blank. The stage with the mask blank moves continuously and an interferometer system commands the laser to flash when it
reaches the position for the next field. Because of the short flash time, around 20 ns, the movement of the stage is frozen and
a sharp image of the SLM is produced in the resist. The SLM is then reloaded with a new pattern in time for the next flash.
Finally, to feed the pattern into the SLM, a parallel and scalable data-path design handles the complex design data of state-
of-the-art IC designs.
[13,14]
The system is designed for multi-pass writing for high accuracy, achieved from averaging, and
this is optimized to give minimum linewidth and positioning variation across the mask. The fields that build up the pattern
are stitched together using an overlap to minimize the line width variation across the stitching boundaries. The SLM consists
of 512 x 2048 micro-mirrors of dimensions 16 x 16 Pm. With a 200x demagnification, the corresponding main pixel in the
mask pattern is 80 x 80 nm, while the address pixel is 1.25 nm.
Figure 1. a) Overview of the Sigma7300 system. b) The final imaging lens, together with its purge
tubing. The lens is of a catadioptric design,
and is purged in two compartments, using a
deliberate mixture of boil-off nitrogen and
clean dry air(CDA), both ambitiously chemically
filtered.
Proc. of SPIE Vol. 5754 1603
3. THEORETICAL ASPECTS
3.1 Background
Contamination of photolithographic tools may occur by dark as well as photo-induced adsorption. The nature of the
contaminant-lens surface interactions depends on the properties of the constituents and will be discussed later in more depth
in terms of acid-base interaction in section 3.4. Ample evidence
[1,3,7,8,15]
indicates that the laser fluence has a direct impact on
the tendency to reduce the transmission of the lens due to contamination. The laser irradiation may influence the
contamination process in several ways. Firstly, the laser light-lens surface interaction could create an “activated” surface
over the area where the laser spot has passed. Experiments indicate
[16]
that “prenucleation” occurs at these locations,
presumably by forming radicals at the surface. It was found in another study
[17]
that the number density of nuclei showed a
tendency to saturate after lengthy irradiation and that the density attained at the saturation level decreased with increasing
substrate temperature. In the same study it was also found that the initial nucleation rate increased with temperature. These
findings indicate the important role of diffusion of activated radicals on the surface. Secondly, the laser fluence may also
excitate and/or dissociate the gaseous species in the vicinity of the lens surface, thereby creating multiple substances,
possibly forming both neutral and highly reactive radical species. These reactive species may react between themselves or
directly with the lens surface. The tendency for a molecule to absorb a photon may be described the molecular cross section,
V [cm
2
/molecule], molar absorptivity, H [cm
2
/mol] or by the absorption coefficient, D [cm
-1
]. The photon energy involved for
some typical DUV and VUV exposure wavelengths as well as bonding energies for different bond types are given in Table 1
and Table 2 respectively. As can be seen from the tables, at 248nm the saturated carbon bonds are easily broken, while at
193 and 157nm the photon energy is enough to break also most of the unsaturated bonds. In general, shorter bond length
means stronger intermolecular forces and thereby higher bond energies. Also given in Table 1 is the polarizability of each
bond type. The polarizability has a strong impact on the chemisorption of the contaminant on the lens surface and will be
discussed later. In the remainder of the theoretical analysis, the pre-history influenced by the laser light is disregarded and it
is assumed that the prototype benzene contaminants are near the lens surface and that they are intact. This oversimplified
situation is mainly due to lack of experimental data and proof of active species during irradiation.
Table 1. Polarizability, bond length, and average bond energy for different bonds
[18]
.
Bond
Polarizability (Å
3
)
Typical bond length (nm) Average bond energy (kcal/mol)
C-C 0.531 0.154 83
C-F 0.555 0.138 116
C-O 0.584 0.143 84
C-H 0.652 0.108 99
O-H 0.706 0.096 102
C=O 1.020 0.122 176
C=C 1.643 0.135 146
C
{
C 2.036 0.121 200
C
{
N 2.239 0.116 213
Table 2. Photon energy for some typical excimer laser wavelengths.
Wavelength (nm) Photon energy (kcal/mol)
248 115
193 148
157 182
1604 Proc. of SPIE Vol. 5754
Investigations
[1,10,19-22]
have shown that photoresist outgassing is the major source of hydrocarbon deposits. The chemically
amplified resist used in the Sigma7300 laser pattern generator, is FEP-171 from Fujifilm Electronic Materials. During
exposure the photoacid generator (PAG) absorbs the UV light and forms a sulfonic acid and aromatic groups together with
radical species
[23]
(Figure 2). The PAG itself is a high molecular-weight compound and is not expected
[24]
to outgas during
exposure. The so-called BTEX compounds, benzene (C
6
H
6
), toluene (C
6
H
5
CH
3
), etylbenzene (C
6
H
5
C
2
H
5
) and xylene
C
6
H
4
(CH
3
)
2
, make up the simplest group among the aromatic compounds and are often found as contaminants in cleanroom
environments and may outgas during exposure and baking of DUV photoresists. Also of interest in 248nm as well as 193nm
lithography is t-butyl benzene, C
7
H
5
(CH
3
)
3
. In addition, some of the remaining solvent after the post-apply bake may outgas
during exposure. In this study we employ benzene (C
6
H
6
) as the prototypical member of the contaminants possibly involved
in lens contamination at 248nm using non-silicon containing resists, partly because in many cases the outgassing product/-s
is benzene and/or benzene-derivatives, and partly because of the complex network of reactions, including intermediates etc.,
that may exist during laser irradiation. Microscopic as well as macroscopic concept will be used to analyze the mechanisms
responsible for hydrocarbon deposition on the lens ranging from adsorption, condensation/wetting phenomena to chemical
bonding. The adsorption step, where the underlying mechanisms are polar (electrostatic) and/or non-polar (Lifshitz-van der
Waals
[25]
) forces, will be explored in the next section. Thereafter thermodynamics are being used to illustrate the major
influencing parameters behind the nucleation and wetting conditions of the adsorbed contaminants. The theoretical part ends
with chemical bonding which is viewed from the perspective of optical basicity and acid-base interactions.
Figure 2. Mechanism of UV-light absorption by sulfonium-type of photoacid generator in resist FEP-171
[23]
3.2 Rate limiting step
Adsorption of hydrocarbons onto a surface can be done through either physisorption or chemisorption depending on the type
of interaction involved. Physisorption involves weak, long-range van der Waals forces (0-40 kJ/mol) and here no electron
sharing takes place between the contaminant, the adsorbate, and the receiving surface, the adsorbent. The van der Waals
forces arise from rapid fluctuation in electron density within each atom or molecule, which induces an electrical moment.
This electrical moment leads to an attraction between two atoms or two molecules or between a molecule and a solid surface.
Physisorption is very rapid and reversible and the equilibrium established quickly as the gas pressure is varied.
Chemisorption on the other hand, is much slower and in general an irreversible process. Chemisorption can occur direct or
be achieved through a physisorbed precursor-mediated process and when a chemical bond is formed, electron sharing is
involved (30-1000kJ/mol). Physicists normally prefer to call these interactions donor-acceptor interactions while chemists
tend to use the term acid-base interactions in order to explain the bond formation mechanisms. In a polar solid (containing
ions) with polar groups, an electric field will induce a dipole in the contaminant. In the simulation part of this paper, we base
our mathematical modeling on a chemical vapor deposition (CVD) model for lens contamination. A CVD polymerization
process may undergo the following serial steps
[26]
1. Transportation of the contaminants from the bulk gas phase to any boundary layer close to the lens
2. Diffusion of contaminants through the boundary layer to the lens surface
3. Adsorption of contaminants at the lens surface
4. Surface migration
5. Surface chemical reaction
6. Desorption of reaction products from the lens
7. Diffusion of reaction products through the boundary layer
8. Transport of reaction products to the bulk gas
Proc. of SPIE Vol. 5754 1605
The rate limiting step can be anyone of the three types: mass transport, diffusion or kinetic. The contamination process is
mass transport controlled if the steps (1) or (8) are the rate limiting step. The diffusion constitutes the rate limiting step if one
of the steps (2) or (7) is slow. The process is regarded to be kinetically controlled if one of the steps (3) to (6) is the rate-
limiting step. If the deposition process would be mass transport controlled, the rate would only be controlled by the transport
of contaminants into the bulk gas close to the lens and is therefore not likely to be a function of temperature. It has been
argued that
[26]
diffusion of reactants through a boundary layer is a temperature activated process and should vary roughly as
T
2
. Thus, for a diffusion-controlled process the deposition rate would increase with an increase in temperature. However,
higher deposition rates have been obtained when lowering the substrate temperature.
[26-28]
The deposition rate was found to
scale with partial pressure.
[26]
In a lithographic exposure tool, the more the contaminants deposit onto the lens surface, the
more the deposit will absorb of the exposure light. Presumably, at a certain stage the contamination process becomes
autocatalytic. Assuming that the extra heat generated, due to absorption of light by the contaminants, is negligible - at least
during the initial stages of the contamination process - the physisorption/chemisorption process becomes the rate-limiting
step and therefore the most vital factor for the evolution of the contamination. Physisorption of gas contaminant molecules
on the lens surface occurs because of attractive forces between the two materials. When approaching the surface, the gas
molecules may loose some of their momentum and could therefore be trapped in a potential well. The energy required to
overcome the attractive potential barrier of the surface and the attraction of neighboring molecules is the heat of adsorption,
'
H
ads
. In case of electron transfer between the contaminant and the lens surface, the interaction is called chemisorption. The
heat of chemisorption is generally larger than the heat of adsorption. Depending on the specific nature of the contaminants
and the lens surface, the deposit may be many layers thick.
3.3 Physisorption mechanisms
Physisorption interaction can be divided into two parts, a strongly repulsive part at close distances and van der Waals
interactions at medium distances of a few Å. The net interaction energy
)
(D) when the gas molecule is approaching a
surface may be described by
[29]
93
90
2
6 D
B
D
C
D
NN
SUSU
)
………………………………………….. (3.1)
where the first term describes the attractive van der Waals force and the second term is the repulsive exchange interaction,
U
N
is the number of solid atoms per unit volume, B and C are constants, and D is the perpendicular distance of the molecule
from the surface. The constant C can be related to the Hamaker constant A
[30]
, by
………………………………………….. (3.2)
CNA
22
S
where N is the number of atoms per cubic centimeter. Benzene, being a flat molecule, would tend to interact more strongly
with the lens surface if the molecule lies with its plane flat against i.e. the SiO
2
surface of the final lens. The mobility of
benzene at the surface decrease the flatter the lens surface is, thus enhancing the interaction. For other contaminants being
more bulky, the van der Waals interaction could be dramatically different.
3.3.1 Lifshitz-van der Waals interaction energy
The interaction between a gas and a solid is electromagnetic in nature. For nonpolar substances such as aromatic
hydrocarbons, the electrodynamics is governed by van der Waals forces. There are two approaches for calculating these
forces. Either the interaction between macroscopic bodies is derived by a pairwise summation of all the relevant
intermolecular interactions (so-called microscopic Hamaker approach), or the approach is based upon the macroscopic
electrodynamic properties of the interacting media (Lifshitz’ approach). Here, the interaction energy between hydrocarbons
1606 Proc. of SPIE Vol. 5754
and different materials will be calculated by means of the Lifshitz’ theory of van der Waals forces. According to the
macroscopic approach, the discrete atomic structure and the solids are treated as continuous materials with bulk properties
such as the dielectric permittivity and the refractive index. The van der Waals forces arise from the absorptions of photons of
frequency
Z
by a material with a complex dielectric constant
[31,32]
)()()(
Z
H
Z
Z
cc
c
i
………………………………………….. (3.3)
where
Hc
(
Z
) is related to the polarizability of the material and
Hcc
(
Z
) is related to the energy dissipation within the material.
The Kramers-Kronig relation gives the relationship between the real and the imaginary parts of any optical property and is
the integral transform of the imaginary part of the dielectric constant,
H
, from a function of the real frequency,
Z
, to a
function of the imaginary frequency, iȟ , and is defined by
[32]
³
f
cc
0
22
)(2
1)(
Z
[Z
ZHZ
S
[H
di
………………………………………….. (3.4)
In general limited dielectric data are available, so the function
H
(i
[
) is usually represented by a model based upon a damped
oscillator such as
[33-35]
¦
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
N
j
j
j
j
j
n
g
C
B
i
1
2
2
1
1
1
Z
[
Z
[
W[
[H
………………………………………….. (3.5)
where the first term after unity represents the contribution from the orientation of permanent dipoles and the summation
relates to the absorption peaks in the infrared and ultraviolet region. B is related to oscillator strength in the microwave range
and g
j
is the damping coefficient of the oscillator. C
j
is given by
[34]
j
j
j
f
C
ZS
2
………………………………………….. (3.6)
where f
j
is the oscillator strength and
Z
j
is the relaxation frequency [rad/s] of the absorption band. The representation of the
imaginary dielectric response function,
H
(i
[
n
), can be further simplified since the dielectric response in the visible-UV region
dominates the van der Waals interaction for most systems. It has been found
[35]
that inorganic materials can be well
represented by one UV and one IR relaxation and if the damping terms are neglected the following equation may be
utilized
[36]
22
1
1
1)(
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
UV
UV
IR
IR
n
C
C
i
Z
[
Z
[
[H
………………………………………….. (3.7)
Proc. of SPIE Vol. 5754 1607
where C
IR
and C
UV
are the absorption strengths in the IR and UV range respectively,
Z
IR
and
Z
UV
represent the characteristic
absorption frequencies in the IR and UV region respectively. The nonretarded Hamaker constant between two half-spaces of
material 1 and 2 interacting over media 3 is given by
[35,36]
¦ ¦
f
f
''
0 1
3
3212
'
132
2
3
n s
S
B
s
Tk
A
………………………………………….. (3.8)
where k
B
is the Boltzmann constant, and ´ denotes the convention of dividing the n=0 term by 2, s and n are multiple indices
of the frequency of the incident radiation. The differences in the dielectric response of the three materials are given by
'
12
and
'
32
where
);3,2,1,(,
)()(
)()(
jiji
ii
ii
njni
njni
ij
z
'
[H[H
[
H
[
H
…………………………………….. (3.9)
and
n
h
Tk
B
¸
¸
¹
·
¨
¨
©
§
2
4
S
[
………………………………………….. (3.10)
where n is an integer; n=0,1,2,3…The sum above can be converted into an integral and we get
[37]
³
f
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
1
)()(
)()(
)()(
)()(
4
3
4
3
32
32
31
31
32
32
31
31
132
[
[
[H[H
[H[H
[H[H
[H[H
SHH
HH
HH
HH
d
ii
ii
ii
ii
h
TkA
B
………………………………………….. (3.11)
Simplified approximations of the Hamaker constant have been derived by retaining only a single UV relaxation to represent
the dielectric response of each material. However, under the assumption that all three media have the same absorption
frequency, the non-retarded Hamaker constant becomes
[37]
> @
2
3
2
2
2
3
2
1
2
3
2
2
2
3
2
1
2
3
2
2
2
3
2
1
32
32
31
31
00
28
3
4
3
nnnnnnnn
nnnnh
Tk
AAA
e
B
total
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
|
|
!
Q
HH
HH
HH
HH
QQ
………………………………………….. (3.12)
where A
Q
=0
is the zero-frequency energy of the van der Waals interaction and involves the Keesom and Debye dipolar
contributions, A
Q
>0
constitutes the dispersion energy and includes the London energy contribution, h is the Planck’s constant,
Q
e
is the mean ionization potential of the material, and
H
i
and n
i
are the dielectric constant and refractive index for material i.
In case of similar materials interacting across air (i.e. gaseous species and deposited hydrocarbons), Equation. 3.12 reduces
to
[37]
1608 Proc. of SPIE Vol. 5754
2
3
2
3
2
1
2
2
3
2
1
2
31
31
216
3
4
3
nn
nnh
kTA
e
total
¸
¸
¹
·
¨
¨
©
§
Q
HH
HH
………………………………………….. (3.13)
As can be seen from Equation 3.13, the van der Waals interaction between two identical bodies in air (i.e. volatile organic
carbons and already deposited material of the same substance) is always attractive (positive value of the Hamaker constant),
while the force can be either attractive or repulsive between dissimilar materials as given by Equation 3.12 (i.e. volatile
hydrocarbons and initially “clean” lens surface). Thus, one could in principle reduce the tendency towards lens
contamination by adopting a final lens material with better matched dielectric constant and refractive index with respect to
the interaction with the contaminant. It is also clear from Equation. 3.12 that only when
H
3
is large, will the temperature-
dependent term becomes important. Table 3 shows the calculated values of the Hamaker constant for different materials.
Table 3. Calculated values of the Hamaker constant for some materials.
Material Dielectric constant Refractive index Hamaker constant (J)
D
-Al
2
O
3
10.1 1.753
9.94
10
-20
SiO
2
(fused silica)
3.8 1.448
4.55
10
-21
TiO
2
114 2.464
1.58
10
-19
3Y-ZrO2 18 2.178
1.37
10
-19
CaF
2
6.7 1.427
2.79
10
-21
MgF
2
5.47 1.372
5.52
10
-21
C
6
H
6
2.28 1.501
4.98
10
-20
All interactions show a positive Hamaker constant thus indicating that benzene would contaminate all these materials. Based
upon Lifshitz-van der Waals interaction, the data in Table 3 as well as the graphs in Figure 3 suggest that highest interaction
occurs between benzene and TiO
2
, then slightly reduced interaction between benzene and ZrO
2
and Al
2
O
3
respectively.
Lower interactions are found between benzene-SiO
2
, benzene-MgF
2
, and finally lowest between benzene-CaF
2
. Benzene-
benzene interaction is about the same strength as in case of benzene-SiO
2
. The interaction energy, W(D), between a molecule
1 (in medium 3) and the surface of medium 2 is approximately given by
[38]
:
3
3
1
3
)(
D
Aa
DW |
………………………………………….. (3.14)
where A is the Hamaker constant, a
1
is the radius of the adsorbing molecule and D is the separation distance. In Figure 3,
the interaction energy W(D) is plotted versus separation distance D for a benzene molecule interacting with some different
materials. The plasma frequency of the free electron was taken as 3.0x10
15
[s
-1
] (2.1x10
15
in case of benzene-benzene
interaction) and the zero frequency dielectric constant as 2.283 for benzene. The radius of the benzene molecule was taken as
2.65Å.
Proc. of SPIE Vol. 5754 1609
-0,6
-0,5
-0,4
-0,3
-0,2
-0,1
0,0
0,5 1,5 2,5 3,5
Separation distance, D (nm)
Interaction energy, W(D)x10
21
(J)
C6H6-Al2O3
C6H6-CaF2
C6H6-SiO2
C6H6-TiO2
C6H6-C6H6
C6H6-MgF2
C6H6-ZrO2
Figure 3. Benzene interaction across air with different materials.
3.4 BET adsorption isotherm
The adsorption of molecular species onto a surface requires that the forces between the contaminant and the lens surface are
attractive. Many theoretical and empirical adsorption isotherms have been proposed in the literature.
[39]
One of the more
popular relations to describe multilayer growth is the Brunauer-Emmett-Teller (BET) isotherm
[40]
which can be expressed as
¸
¸
¹
·
¨
¨
©
§
»
»
¼
º
«
«
¬
ª
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
¸
¸
¹
·
¨
¨
©
§
»
¼
º
«
¬
ª
''
»
¼
º
«
¬
ª
''
satsat
RT
HH
RT
HH
sat
p
p
p
p
e
e
p
p
n
n
111
1
1
T
………………………………………….. (3.15)
where 'H
1
is the heat of adsorption for the first layer, 'H
n
is the heat of adsorption for consecutive layers and equals the
latent heat of condensation, p is the partial vapor pressure, and p
sat
is the saturation vapor pressure. For low pressures, the
adsorbing molecules find many free binding sites and the heat of adsorption is relatively high. Once the first monolayer has
formed, it is easier for the next molecules to adsorb. However, it is only at rather high pressures molecules start to form
multilayers. For p o p
sat
the adsorbed layer gets very thick since macroscopic condensation sets in. For high values of the
exponential term in the BET equation, strong binding of the vapor molecules to the surface prevail and less intermolecular
interactions occur. For low pressures, sub- or monolayer adsorption occurs, while at high pressures multilayers are formed.
Vice versa, at low values of the exponential term, relatively high pressures are needed in order to form a monolayer.
However, once the first layer is formed, it will be easier for the next molecules to adsorb. The BET isotherm assumes a
dynamic equilibrium between the adsorbate and the adsorptive (i.e. the rate of adsorption and desorption in any layer are
equal). In the first layer, it is assumed that molecules adsorb on equivalent adsorption sites (i.e. surface homogeneity).
Furthermore, the BET isotherm also ignores adsorbate-adsorbate interactions (i.e. phase transitions). As a matter of fact,
there is an inconsistency in the multilayer equation since without any adsorbate-adsorbate interactions; there would not be
any multilayers. In addition, the BET equation neglects the effect of surface tension (due to multilayer buildup). Despite
these shortcomings, the BET equation has received relatively wide acceptance since it is two-parameter model capable of, at
least, reproducing qualitative isotherms.
1610 Proc. of SPIE Vol. 5754
3.5 Thin-film nucleation mechanisms
3.5.1 Adsorption and wetting
The effect of vapor adsorption of contaminants onto the lens surface is to lower the surface energy of the solid according to
SSSV
S
J
J
………………………………………….. (3.16)
where
J
SV
is the lens surface energy after vapor adsorption,
J
S
the initial solid surface energy and
S
S
is the surface pressure of
the adsorbed film. The surface pressure can be calculated using
[41]
………………………………………….. (3.17)
³
*
sat
p
S
pdRT
0
ln
S
where
*
is the adsorption isotherm, p is the vapor pressure and p
sat
is the saturation pressure. High-energetic surfaces such
as dielectrics, are readily wet by nonpolar liquids (i.e. hydrocarbons condensed on the lens), and the isotherm of disjoining
pressure, 3(h) may be given by
[42,43]
3
6
)(
h
A
h
SLV
S
3
………………………………………….. (3.18)
where h is the thickness of the wetting film and A
SLV
is the Hamaker constant, which depends on the optical properties of the
three media; solid, liquid and vapor, in our case the lens surface, hydrocarbon condensate and vapor. As the adsorption
proceeds, the thin liquid film may transform into a droplet. The surface tensions or energies involved for a droplet resting on
a substrate is illustrated in Figure 4. A force balance in the horizontal direction leads to
[44]
SLLVSV
J
M
J
J
cos ………………………………………….. (3.19)
where
J
SV
is the surface energy at the lens surface-vapor interface,
J
LV
is the surface tension at the liquid-vapor interface,
M
is
the contact angle and
J
SL
is the surface energy at the lens surface-liquid interface.
Figure 4. Surface tensions/energies involved for a droplet resting on a solid surface.
J
lv
M
Dro
p
le
t
J
sv
Lens surface
J
sl
Proc. of SPIE Vol. 5754 1611
3.5.2 Formation of nucleus
The nucleation is dependent on a balance between surface and volume energies. The change in Gibbs free energy for nuclei
formation can be written as
[45]
SVSLLVV
rrGrrG
JJMSJMSMM
S
' '
22233
sincos12coscos32
3
)(
………………………………………….. (3.20)
where
M
is the contact angle,
'
G
V
is the Gibbs free energy change per volume unit, r is the radius of the nuclei. Small
nuclei are dominated by the surface energy term and will tend to dissolve because an increase in r causes an increase in
'
G.
The critical radius, r*, necessary for nucleation to occur, is given by:
[45]
>
@
V
SVSLLV
GV
rr
r
'
3
sincos122
222
*
JJMSJMS
………………………………………….. (3.21)
Equation 3.21 indicates that when vapor molecules adsorb on a solid surface and potentially form a critical nucleus that
grows in time due to further condensation; the size of these nuclei depends on the surface tension of the condensate, the
contact angle between the solid and the liquid droplet as well as the Gibbs free energy change during condensation. If the
nucleus reaches the critical size r*, that nucleus will grow because of the dominance of the volume term causes
'
G to
decrease with increasing r. The critical Gibbs free energy necessary for formation of critical nucleus is
¸
¸
¹
·
¨
¨
©
§
'
' '
4
coscos32
3
16
3
2
3
**
MMSJ
G
rGG
LV
………………………………………….. (3.22)
The smaller the contact angle is, the lower is the surface energies/tensions involved and thereby easier nucleation. As the
formed nuclei start to grow and coalescence occurs, droplets will be formed. Depending on the surface tension of the
hydrocarbon condensate, the surface energy of the lens material and the interfacial energy between these two materials, the
strength of the intermolecular forces (adhesional forces) govern whether the lens material will be wet or not by the
condensate.
The tendency towards wetting or vapor condensation at the surface can be illustrated by the spreading parameter, S
[29]
0)(
LVSLSV
S
J
J
J
………………………………………….. (3.23)
If S is negative, the solid-vapor interface has the lower surface energy and partial wetting occurs (i.e. a liquid drop of
equilibrium shape will form). However, if S>0, the liquid will completely spread on the solid surface. The work of cohesion
of the liquid, W
L
, and work of adhesion, W
SL
, can be written as
[46]
1612 Proc. of SPIE Vol. 5754
LVLL
W
J
2 ………………………………………….. (3.24)
and
SLLVSVSL
W
J
J
J
………………………………………….. (3.25)
Inserting Equation 3.19 in 3.25 gives
T
J
cos1
LVSL
W ………………………………………….. (3.26)
Equations 3.24 and 3.26 combine to
2
cos1
T
LV
SL
W
W
………………………………………….. (3.27)
which is the adhesion to cohesion ratio and can be obtained just by measuring the contact angle. It has been proposed that the
total surface free energy, J
i
, can be split into components
[47,48]
………………………………………….. (3.28)
AB
i
LW
ii
JJJ
where LW stands for the Lifshitz-van der Waals contribution and AB stands for the acid-base contribution. Thus,
LSLS
LW
L
LW
S
AB
SL
LW
SLSL
WWW
JJJJJJ
2
………………………………………….. (3.29)
where
J
+
is the Lewis acid parameter of surface free energy, and
J
-
is the Lewis base parameter of surface free energy
From. Equations. 3.26 and 3.29,
LSLS
LW
L
LW
SLV
JJJJJJMJ
2cos1
………………………………………….. (3.30)
For a non-polar liquid like benzene, Equation 3.30 simplifies to
LW
L
LW
SLV
JJMJ
2cos1
´
………………………………………….. (3.31)
The Hamaker constant in Equation 3.18 may be related to physical properties such as the work of cohesion of a non-polar
liquid
[37]
(i.e. benzene),
………………………………………….. (3.32)
LV
dA
JS
2
24
Proc. of SPIE Vol. 5754 1613
where d is the effective molecular size and
J
L
is the surface tension of the liquid (condensate). It is evident from Equation
3.31 that
J
S
LW
can be determined experimentally by a single contact angle measurement using a nonpolar liquid of known
surface tension (
J
L
LW
=
J
L
). Thus, van der Waals forces play an important role in both physisorption and chemisorption.
3.6 Chemisorption mechanisms
Since the Lifshitz-van der Waals interaction responsible for physisorption is weak in comparison for chemisorption due to
electrostatic interactions, the equilibrium separation distance between the chemisorbed molecule and a solid surface will
therefore be shorter than for the corresponding physisorption case. The potential energy of two ions of charge z
1
e and z
2
e
respectively, separated by distance D is given by
[37]
D
ezz
E
r
HSH
0
2
21
4
………………………………………….. (3.33)
where z
1
and z
2
are the two respective ionic valencies, e is the elementary electronic charge, and
H
0
and
H
r
are the static
dielectricity in vacuum and medium respectively. By comparing the separation distance dependency between polar and non-
polar interactions, the former interaction is proportional to 1/D, while the non-retarded Lifshitz-van der Waals interaction
varies with separation distance as 1/D
3
. Both types of interaction can be repulsive or attractive dependent on the signs of the
two ions in the polar case, and dependent on the value of the Hamaker constant in the latter case, which in turn is determined
by the involved dielectric constants and refractive indices. For contaminants such as SO
3
or ions of SO
4
2-
type, there could
be rather strong Coulombic interaction between the sulfonate ion and cations at the surface of the lens oxide material.
Nevertheless, there will also always be a contribution from the dispersion interaction. In addition, hydrogen bonds, which
can be viewed as polar interactions, may be of great importance in certain cases where hydrocarbons may bond to oxygen
originating from hydroxyl groups already bound to the lens surface, or to oxygen of the lens oxide material itself.
Chemisorption involves bond formation by electron exchange. The chemical bonds may be of covalent or ionic character, or
mixtures of both. The ionic character can be described by either the electronegativity or the ionicity. The polarizability is
related to many macroscopic and microscopic physical and chemical properties. It is therefore tempting to look at the
chemisorption by considering the acid-base characteristics of the contaminants and the lens material. The electronic structure
of the hydrocarbon contaminant and the lens material are closely related to their acid-base character. Since it is more
energetically favorable for an atom or molecule to have completely filled orbital electron shells, there is a tendency for the
contaminant molecule and lens material to either attract electrons (electron acceptors) or expel electrons (electron donors).
Electronegative molecules, so-called Lewis acids, tend to attract electrons strongly while molecules that donate electrons are
referred to as Lewis bases. When materials become in contact, the resulting bonds that are formed will have varying degrees
of ionic and covalent character due to the difference in the electronegativity of the materials involved, in this case
hydrocarbons and the final lens. The acid-base interactions of the lens surface material can be considered from the
perspective of oxygen. When oxygen is in a free state as O
2-
, its ability to donate negative charge is at a maximum. For a
relatively non-polarizing cation such as Ca
2+
, this ability is marginally impeded while for highly polarizing cations such as
Si
4+
, the ability to donate negative charges are dramatically reduced. The effective basicity is therefore dependent on the
number and identity of the cations to which oxygen is coordinated. Followingly, increased electronic polarizability of the
lens material means stronger electron donor ability of its oxide ions. The electronic polarizability, D, can be obtained
experimentally using the Lorentz-Lorenz relationship
[49]
»
¼
º
«
¬
ª
2
1
4
3
2
2
n
n
N
V
m
m
S
D
………………………………………….. (3.34)
where V
m
is the molar volume [cm
3
], N is the Avogadros number and n is the refractive index. The electronegativity and
polarizability are closely related to the refractive and dielectric properties and will to a large degree determine the chemical
1614 Proc. of SPIE Vol. 5754
bonding mechanism. An interesting concept, the optical basicity, has been introduced
[51,52]
which represents a measurement
of the electron donor power of an oxide (-II) material, which originates the nephelauxetic effect
[53]
(electron cloud
expansion). The optical basicity is strongly correlated to the electronic polarizability, D
O
2-
, and electronegativity, F, and
polarizing power of the cations and depends in turn on their valence and coordination. This concept thus relates the acid-base
character of an oxide to its physical and chemical reactivity. The optical basicity was developed as a way to describe the
spectral shift in the UV region that is observed when a small amount of a probe ion (i.e. Pb
2+
) in a medium containing
polarizing cations. The more electrons that are donated by neighboring oxygens (e.g. increase in the basicity of the medium)
to the probe ion, the greater the observed frequency shift. Optical basicity may be defined as
[50]
3700
3700
2
/
CaOfree
oxidefree
Pb
EE
EE
………………………………………….. (3.35)
where E
free
is the energy for Pb
2+
in free space (=64400), E
oxide
is the energy of the
1
S
0
o
3
P
1
transition and E
CaO
is the energy
for Pb
2+
in CaO. The correction value of 3700 is added in the equation in order to comply with the nephelauxetic h-
parameter.
[51,52]
Generally, oxides with high electronegativity are acidic, while those with a weak electronegativity are basic.
Table 4 shows the optical basicity parameter
/
together with other parameter related to it, namely the heat of formation,
work function, electronic polarizability as well as electronegativity and band gap data for some oxides and fluorides. A large
value of
'
H
f
0
normally reflects the formation of stable and hard, more chemically inert compounds, which in turn indicates
strong binding force between cation and anion.
Table 4. Optical basicity, heat of formation, work function, polarizability, electronegativity and band gap
data for some oxides and fluorides potentially involved in microlithographic lens designs. Some
data are lacking for a few of the compounds. Data from references [53]
a)
,[54]
b)
,[56]
c)
.
Material
/
'
H
f
0
(kJ/mol)
Wf
D
O
2
-
3
)
F
1av
E
g
(eV)
SiO
2
0.48
a)
-879
13.36
c)
2.42
b)
2.91
b)
9.05
b)
Al
2
O
3
0.60
a)
-1676
12.15
c)
2.59
b)
2.69
b)
8.6
HfO
2
-1134 5.8
ThO
2
0.71 -1238
10.38
c)
2.58
c)
MgO
0.78
a)
-601
10.17
c)
2.83
b)
2.36
b)
7.3
b)
TiO
2
0.96
b)
-939
10.66
c)
2.52
b)
2.77
b)
3.0
b)
CeO
2
0.98
b)
-1226
10.16
c)
2.59
b)
2.69
b)
3.2
b)
ZrO
2
0.96
b)
-1097
2.55
b)
2.74
b)
5.0
b)
CaF
2
0.43
a)
-603 12
MgF
2
0.34
a)
-1124 10.8
Acidic oxides, such as SiO
2
, have optical basicity values <0.5, while more basic oxides such as MgO have values closer to
unity. Calculated as well as experimental data
[51-55]
show that the optical basicity decreases linearly both with the inverse of
the work function and with the reciprocal of the anion polarizability as well as decreases linearly with the electronegativity.
There will be stronger tendency towards contamination if the contaminant differs largely in polarity in comparison to the
lens surface (acid + base o salt). The optical basicity scale may therefore be used to qualitatively determine the potential
risk of contamination dependent on the acidic/basic character of the contaminant and the lens material. Study of the
temperature dependence of the refractive index of oxide melts
[56]
indicates that the electronic polarizability of oxygen
increases with increasing temperature. This in turn suggests that the basicity of the oxide will increase with temperature. In
Proc. of SPIE Vol. 5754 1615
case of carbonaceous deposit on a lens surface, it is possible that the basicity of the lens material will vary due to the UV
light absorption of the deposit and thereby an increase in the local surface temperature which possibly could be of
importance for the contamination process.
3.7 Desorption mechanisms
Physisorbed hydrocarbon molecules may be removed from the surface either by increased pressure, forcing the collision
frequency to increase, or by reaction with photo-induced ozone and oxygen radicals thus forming gaseous products which
may diffuse away from the surface. Transition state theory says that the rate for molecular desorption, k
d
, from a solid
surface can be expressed as
¸
¸
¹
·
¨
¨
©
§
¸
¹
·
¨
©
§
Tk
E
d
B
ek
S
Z
2
0
………………………………………….. (3.36)
where
Z
0
is the harmonic vibrational frequency in the potential well, E is the activation energy, k
B
is the Boltzmann constant,
and T is the absolute temperature. It has been proposed and demonstrated by others
[59]
that ozone may participate in
processes that counteract AMC film formation. One plausible reason is that the highly reactive ozone molecules created in
the close vicinity of the contaminated surface immediately interact with adsorbed molecules and the reaction products desorb
from the film. This mechanism has in fact been seriously proposed, using the wavelength 157nm
[59,60]
as an efficient cleaning
method for components in DUV lithographic systems, such as pellicles. In the field of atmospheric physics it is known that
molecular oxygen displays a series of absorption bands in the wavelength region 240 280 nm, called the Herzberg I bands,
corresponding to “forbidden” molecular transitions of the first type.
[61-63]
The corresponding absorption is weak, so the
amount of ozone generated is expected to be small, but still non-zero at 248 nm. It is expected that the photochemical
reactions that govern the ozone formation are:
[4]
…………………………………………. (R3.1)
o OOhO
Q
2
…………………………………………. (R3.2)
)2(2)2(22
32
MOMOO o
where it is implicitly assumed that the photon energy is high enough for efficient absorption. In addition to the ozone-
induced removal of carbonaceous molecules by oxidation, the desorption may also occur due to reaction with oxygen
radicals generated by the dissociation of ozone molecules. There is in fact a photo-depletion of ozone by the DUV light,
actually the UV-filtering effect of the atmospheric ozone layer. From experimental estimations and measurements of relevant
absorption cross-sections of O
2
and O
3
it has been shown that there will be at least 50 times higher ozone generation at 193
nm than at 248 nm.
[9]
Thus there will be at least a low concentration of ozone in a 248nm tool also in purged volumes if the
purge gas is added a certain amount of well-filtered CDA. Since the ozone presence close to the surface contaminated with a
growing organic film may reduce that growth, or “etch” away an already grown film. In the sub-monolayer deposit regime,
desorption may be treated as a first-order reaction if the adsorbed molecules desorb independently of each other. In case of
clustering, desorption may occur from the edges of the clusters, thus making the desorption independent of surface coverage.
In the film growth model that will be presented in the next section, a phenomenologic film-etching term has been included.
4. COMPUTATIONAL MODELING
4.1 Modeling of lens contamination due to resist outgassing
In the following section we describe the implementation of a model for resist outgassing and contaminating film formation
on the final outer surface of the final lens in the laser pattern generator. The model is partially based on the work in reference
1616 Proc. of SPIE Vol. 5754
1, but with certain modifications, in a 3-D multiphysics computational software, Finite Element Modeling LABoratory
(FEMLAB).
[64]
We make use of a simplified laser-assisted CVD mass-deposition model, described later, for the film growth on the final
optical surface of the final lens in the laser pattern generator. In that model it is implicitly assumed that something
happens
that favors film formation on a surface under influence of (DUV) light. The light-gas phase interaction is described by the
photon flux
)
and the light absorption cross-section V
0
for the gas phase, and the cross-section of already adsorbed
molecules on the surface,
TV
0
.
T
is the fractional surface coverage of adsorbed molecules, calculated from the BET isotherm.
Note that the model does not
make any distinction between photochemical reactions in the gas phase followed by adsorption,
and reactions in already present adsorbate. In both cases the yield is described by the magnitude of the cross-section,
V
0
.
Thus this is indeed a lumped-parameter model, where all details of the processes of the preceding section are
phenomenologically collected in one single term in the differential equation for film formation.
4.1.1 FEMLAB
Simulations of outgassing and film growth on an optical surface have been made using the computation software
FEMLAB™. This is a 3-Dimensional multiphysics software, in which the relevant problem is formulated as partial
differential equations with relevant boundary conditions on the specified geometry. The equations are coupled to each other
as desired, and the multiphysics problem is solved using the Finite Element Method (FEM).
4.1.2 Geometry
The computations are restricted to the small volume between the resist on the moving photo-plate and the concave final lens
surface of the imaging lens, as shown in Figure 5. Four symmetrically located gas channels direct the purge gas into that
volume. The purge gas leaks out from the volume through an aperture, a Field Stop, with diameter 1.1mm. The distance
from the aperture to the centre of the lens is 1.8mm. The Field Stop is a thin metal plate of thickness 127 Pm located 122 Pm
above the resist surface. The diameter of the base of the hemispherical volume is 4.67 mm.
Figure 5. The geometry of the final surface of the final imaging lens (light) and the lateral movement of the stage and thus
the photo-plate (dark), the slit between the resist surface and the Field Stop metal plate, the Field Stop with circular
aperture, the hemispherical volume between the inner surface of the Field Stop and the outer surface of the final lens
element in the final and imaging lens.
Proc. of SPIE Vol. 5754 1617
4.1.3 The multiphysics model
The present simulations concentrate on the volume between the final concave surface of the imaging lens and the resist
surface, as shown in Figure 5. The latter is moving during the exposure process, which has been included in the model. In
the treatment of the contamination process, several physical processes are involved:
(i) DUV-induced outgassing of molecules from the resist,
(ii) A purge gas flow in the volume between the lens surface and the resist. The gas is injected through four
symmetrically placed channels, and the outlet is through a small hole, the “Field Stop”, laterally along the
resist surface. The purging works against the molecular diffusion, described below. The purge gas flow as well
as the flow inside the hemispherical volume that is created by the resist surface movement, is described by the
Navier-Stokes’ Equations. Furthermore, the resist surface is moving at a constant velocity, as is the case during
mask writing, which creates a second flow component.
(iii) Diffusion of out-gassed molecules through the gas, against the purge flow, towards the lens surface, described
by the diffusion equation.
(iv) Laser-induced deposition, including all unspecified reactions and processes, leading to film growth on the lens
surface, described by a rate equation, the CVD mass-deposition equation.
(v) Laser-assisted ozone generation and subsequent etching of the deposited film by chemical reaction at the film
surface, included as a negative term in the rate equation, the CVD mass-deposition equation.
The final outcome of the present simulations are growth rates of the deposited film or, equivalently, total mass surface
density of the deposited film as function of time for some specific cases.
4.1.4 Mathematical formulations
The Navier-Stokes equations describing the gas hydrodynamic flow reads:
pvv
t
v

»
¼
º
«
¬
ª
w
w
U
………………………………………….. (4.1)
where
U
is the fluid density, v is the flow velocity vector and p is the local pressure.
The Diffusion Equation (Fick’s Law) including a convection term reads:
0
w
w
cucD
t
c
*
………………………………………….. (4.2)
where c is the concentration [g
.
cm
-3
], D is the diffusion coefficient [m
2.
s
-1
], and
u
&
is the velocity field [m
.
s
-1
]. The basic
laser-assisted CVD mass-deposition rate equation is written as
[1]
m
dt
trdM
)
TV
),(
………………………………………….. (4.3)
where M = M(r,t) is the deposited mass per unit area [gcm
-2
],
T
is the surface concentration of adsorbed molecules
(precursors) [cm
-2
],
V
is the (wavelength-dependent) deposition cross-section [cm
2
] here assumed to include both the photon
1618 Proc. of SPIE Vol. 5754
absorption cross-section and the quantum yield for deposition,
)
is the photon flux (fluence) [cm
-2
s
-1
] of adequate
wavelength, and m is the mass of the precursor molecule [g]. Note that if the spatial coordinates are not taken into
consideration, M is computed in a single point, which as a crude approximation can be said to be valid for all relevant
surfaces. That aspect was not the aim for this work. Here it is assumed that the entity
TV
0
is dimensionless
[1]
, based on a
specific site number density on the surface.
The coupling between
1)
the calculated diffusion distribution of molecules close to the lens surface out-gassed from the resist,
and
2)
the surface monolayer and subsequently adsorbed multilayer coverage by adsorbed molecules on the lens surface, is
obtained from the Brunauer-Emmett-Teller (BET) isotherm:
[1,30]
sat
sat
condads
sat
condads
p
p
p
p
RT
HH
pp
RT
HH
p
|
¿
¾
½
¯
®
»
¼
º
«
¬
ª
¸
¹
·
¨
©
§
''
¸
¹
·
¨
©
§
''
1exp1)(
exp
T
………………………………………….. (4.4)
Here, p is the partial pressure of the contamination vapor and p
sat
is the saturation partial pressure,
'
H
ads
and
'
H
cond
is the
heat of adsorption and condensation, respectively. Using numerical values for the relevant compound t-butyl benzene as
given by Kunz
[1]
it can be shown that the final approximation is valid.
With photo-generated ozone present near the lens surface, there will be a certain rate of “etching” of the adsorbed film which
counteracts the photo-chemical film deposition. We assume as a simple model that the reduction in dM(r, t)/dt is
proportional to the concentration of generated O
3,
which in turn is proportional to the photon flux
)
. This can be included as
a term E(r, t) [g
.
cm
-2
s
-1
]:
MEm
dt
dM
)
TV
………………………………………….. (4.5)
Note that the spatial surface distribution of dM/dt, and thus also of M(t), is directly connected to the local AMC partial
pressure just outside the surface. The partial pressure is directly related to the local concentration in the vapour, and is given
by the diffusion distribution in the gas. The etching term being proportional to M ensures that no etching can occur without
adsorbed matter.
Thus, having determined the diffusion concentration profile from Equation 4.2, within the calculated purge gas flow
distribution from Equation 4.1, the maximum surface coverage is estimated with Equation 4.4. Then, finally, Equation 4.3 or
Equation 4.5 is solved using
T
, which yields the lateral distribution of deposited molecular-film mass on the lens surface.
The coupling between the flow and diffusion equations, and the mass-deposition equation is here made as a linear scaling of
the value in reference [1].
> @
> @
1
1
c
c
TT
………………………………………….. (4.6)
where c is the calculated molecular concentration at the lens surface, and c
[1]
= 3 ppm = 0.3 Pa is the molecular concentration
in the ambient air.
Proc. of SPIE Vol. 5754 1619
4.1.5 Method of solving the coupled partial differential equations
The film formation consists of two different but dependent processes: Formation of nuclei on the clean substrate, and
subsequent growth of layers on top of that first monolayer.
[1]
Both
T
and
V
depend on the chemistry of the growth surface. It
is assumed that two surface types are involved here, the clean substrate (i.e. the lens surface) and the outer surface of the first
monolayer of deposited film. The total mass deposition can then be described as the sum of the two deposition processes:
),(),(),( trMtrMtrM
FSTot
………………………………………….. (4.7)
where M
S
and M
F
are the mass deposition growth on the substrate and on the contamination layer, respectively, and are
described by:
Fm
dt
dM
SS
S
) 1
VT
………………………………………….. (4.8)
Fm
dt
dM
FF
F
)
VT
………………………………………….. (4.9)
Here, the function F=F(t) is the fraction of the substrate surface that is covered by nuclei of contaminating species.
[1]
For the numerical solution, the equations are re-formulated as time-derivatives, which are coupled as follows:
F
dt
dN
SS
N
) 1
VT
………………………………………….. (4.10)
FVN
dt
d
FF
)
:
12
SVT
………………………………………….. (4.11)
FVF
dt
dF
FFLateral
) 1
VT
………………………………………….. (4.12)
where N
N
[cm
-2
] is the number of formed nuclei per cm
2
,
:
is the total circumference summed over all formed nuclei, and V
is the volume of the contaminant molecule. In the calculations the approximation
T
S
V
S
|
T
F
V
F
|
TV
0
is made.
The initial conditions are:
………………………………………….. (4.13)0,0,00 FFNt
LateralN
Solving for F(t) in FEMLAB, the solution is readily used in the subsequent temporal integration solution of Equations 4.7-
4.9 for M(r, t) = M(x, y, z, t).
1620 Proc. of SPIE Vol. 5754
4.1.6 Numerical data
Much of the outgassing and deposition processes in a pattern generator tool is governed by the pulsed DUV-laser light from
the excimer laser, with a pulse length of about 20 ns and pulse repetition rate 2 kHz. The process during each laser pulse is
regarded as steady state, and the film growth is related to the total sum of pulses, and the corresponding exposure dose
[mJ/cm
2
], or the time. The photon flux (fluence) was
)
= 1.25
.
10
16
photons/cm
2
per laser pulse, corresponding to an energy
density per pulse w = 10 mJ/cm
2
. The purge flow was set to
M
= 0.25 lit/min, and the photo-plate moved with the constant
velocity v = 100mm/s.
The molecular outgassing from the DUV-photon-illuminated photoresist was assumed to be at a constant rate of R = 8.4
.
10
-5
kg
.
m
-2
s
-1
= 8.4 Pg
.
cm
-2
s
-1
, or 1.4 Pa
.
m
-1
s = 0.14 ppm
.
m
-1
s. Outgassing was assumed to continue for 0.1 Ps after the pulse.
The model simply assumes that the DUV-photons are absorbed and that the photoresist reacts by evaporation of aromatic
hydrocarbon molecules, assumed to be t-butyl benzene, following Kunz.
[1]
The entire process is included in a lumped-parameter model, phenomenologically described by an ordinary product of cross-
sections and quantum yields.V assumed to be identical to
V
0
of gaseous molecules. We have used numerical values from
Kunz
[1]
for t-butyl benzene and 193 nm and simply assumed that the corresponding values for 248 nm are a fraction of those
for 193nm. The dimensionless product
T
V
was
T
V
=1.3
.
10
-6
. The mass of the AMC precursor molecule (t-butyl benzene)
m=134 g/mole, and its saturation pressure is p
sat
=1.65 Torr. The contamination concentration in the ambient clean-room air
expressed as partial pressure p was equal to 3 ppm or about 0.3 Pa. The diffusion coefficient of t-butyl benzene in air at room
temperature and atmospheric pressure was estimated to D = 1.4
.
10
-5
m
2
/s. In the present model it was assumed that ozone is
photo-generated at low rate in the air that is added to the nitrogen purge gas that flushes the hemispherical volume under the
lens surface. As a very crude estimation the value E = 3
.
10
-5
s
-1
was used in some calculations of M(t).
4.2 Simulation results
The geometry taken into consideration together with the PDE:s, and generated with the CAD utility within FEMLAB was
shown in Figure 5 above. Figure 6 shows the purge flow distribution in the volume of interest, where the red lines are
directions of the flow velocity vector. In Figure 7 is shown both purge gas flow lines (red) and contaminant molecular
concentration due to diffusion (colour coded) for the case when the photo plate does not move. This is reflected in the
symmetrical concentration distribution. A closer investigation of the diagrams revealed that the contaminant molecular
concentration at the lens surface were very low, which yielded a slow film formation. Figure 8 presents the atmospheric gas
flow inside the volume generated by the movement of the resist surface at the constant speed of v = 100 mm/s, without any
purge flow. Analyses showed that at the lens surface the purge gas flow totally dominates over the flow generated by the
movement of the resist surface. Assuming a constant outgassing rate of t-butyl benzene from the laser-illuminated
photoresist, the diffusion through the flowing purge gas creates the density distribution shown color-coded in the same
Figure 8. The skewness reflects the flow due to the plate movement. Figure 9, intends to illustrate the surface density of
photo-deposited (adsorbed) molecular film on the surface of the lens with moving resist surface but no purge flow. The
efficiency of the purge flow was so high that the appearance of the resulting diffused AMC concentration profile, as well as
of the deposited film, requires a closer analysis of the numerical results than is obtained from visual inspection of the
diagram. It should be noted, however, how the skew flow into the hemispherical volume from the movement of the resist
surface downwards to the right, indeed generates a surface concentration that is higher in the front part of the lens surface, as
could be deduced also from Figure 8. Finally, Figure 10 shows the calculated film growth as a function of time.
Proc. of SPIE Vol. 5754 1621
Figure 6. A 2-dimensional plot of the purge gas flow-
line distribution in the volume between the final lens
outer surface and the resist layer on the photo-plate.
Here the photo-plate was not moving. The purge gas
flow was set to 0.25 lit/min.
Figure 7. A 2-dimensional plot of the purge gas
flow inside the volume, together with resist out-
gassing and the diffusing AMC concentration
distribution in [Pa]. The photomask was not
moving. Note the highly symmetric diffusion
concentration profile, which indicates that the
purge system indeed creates a highly symmetric
flow, as opposed to Figure 8 below
Figure 8. The 3-dimensional plot of the
atmospheric gas flow inside the volume generated
by the movement of the resist surface at the
constant speed of v = 100 mm/s without any purge
flow, but with resist outgassing and the diffusing
AMC concentration distribution. Note the skew
concentration profile, with the highest values to the
right, towards the rightward stage movement as
expected.
Figure 9. The surface density of photo-deposited
(adsorbed) molecular film, assumed to be t-butyl
benzene, on the surface of the lens with moving
resist surface but no purge flow. For visibility, the
information of flow lines and diffusion distribution
in the gas volume have been removed. The resisr
surface is moving downwards to the right, which
creates a weakly turbulent inflow of molecules
outgassing from the resist surface according to
Figure 8.
1622 Proc. of SPIE Vol. 5754
Figure 11. The surface density of photo-deposited
(adsorbed) molecular film, assumed to be t-butyl
benzene, n the surface of the lens with moving resist
surface but no purge flow. For visibility the
information of flow lines and diffusion distribution in
the gas volume has been removed. The resist surface
is moving downwards to the right, which creates a
weakly turbulent inflow of molecules outgassing from
the resist surface according to Figure 10.
o
Figure 10. The calculated contaminant film growth as
function of time, expressed as the mass per unit surface
M
Tot
(t) [ng cm
-2
] according to Equation 4.7. The values are
calculated in the central and uppermost point on the lens
surface. The two lower curves are calculated with
(lowermost) and without (middle) the weak “film-
consuming” term E in Equation 4.5, due to e.g. the
generation of ozone close to the surface and with the
nominal purge gas flows of 0.25 and 0.20 lit/min. The
uppermost curve is calculated assuming no purging at all.
Note that due to the effectiveness of the purge system and
geometry, this curve had to be reduced in scale with a factor
of 10’000.
5. DISCUSSION
Methods to control airborne molecular contamination (AMC) of optics below critical levels are becoming increasingly
important in DUV lithography from optics lifetime and lithographic performance perspective. Despite active gas purge
application and intense experimental and theoretical studies, a complete understanding of the physico-chemical processes
leading to material deposits on the lens surface is still lacking. It should be clear that AMC affects the mask writing in several
aspects:
1)
Shortening of the life time of optical components,which brings consequences to the cost of ownership,
2)
Pattern quality and fidelity, both deteriorating with time, which result in non-optimal patterns, and
necessitates frequent re-calibrations of the pattern generator,
3)
Mask generator uptime, which is directly affected by the replacement frequency of critical and damaged optical
components.
All those aspects have immediate economical consequences.
In this paper some of the mechanisms behind adsorption of contaminants on lens surfaces in DUV lithography have been
treated. Contamination involves complex phenomena, all of them depending on the chemistry, surface structure and
electronic properties (on an atomic scale) of the contaminants and the lens material. Very scarce evidence exist how the
adsorption really takes place. In fact, very little experimental data at all have been published, or are open to the community,
on contamination of optics and photoresist for 193nm and below, originating from a very limited number of sources, and
even less for 248nm wavelength. One reason is that such information very often is regarded as proprietary and has been
acquired at high costs, and few manufacturers are keen on confessing the existence of contamination problems.
Due to the lack of information/knowledge, the specifics of the adsorption process is more or less the subject of guess work.
In example, one can only speculate whether dissociative adsorption occurs and whether adsorption to the surface is much
stronger than the condensation to a liquid. One could also argue whether cooperative adsorption effects (i.e. the benzene
molecule adsorbs onto pre-existing clusters so that it forms multiple bonds with neighboring adsorbate molecules) could
occur. In other cases, cooperative bonding might happen, i.e. the benzene molecule bonds simultaneously to a surface site
and to a preadsorbed benzene on a neighboring site. The possibility that gas-phase radicals outside a “catalytic” surface may
become depleted, compared to the gas phase outside a more inert, not so reactive surface, could also impact the adsorption.
Furthermore, hydroxyl groups on the lens surface can bind water and alcohols by forming a hydrogen bridge. If existing at
the lens surface, carbonyl or methyl groups will have different adsorption behavior due to their respective electronic
structure. Methyl groups show a strong bonding directionality which indicate selective adsorption of these species in
Proc. of SPIE Vol. 5754 1623
comparison with carbonyl groups which show much less directionality behaviour. On the other hand, the carbonaceous
deposit will be hydrophobic and it will be more difficult for water and alcohol to stick to the surface. However, hydrocarbons
will still have a strong interaction with the formed surface layer and may therefore continue to deposit. All these scenarios
could alter the overall adsorption kinetics remarkably. The possibility of photo-induced lens surface wettability conversion
affecting the hydrocarbon adsorption kinetics is also, to the authors´ best knowledge, not known. Already existing and photo-
induced surface defects may remarkably influence the hydrophilic/hydrophobic character of the lens material and thereby the
deposition kinetics. The adsorption event could also be influenced by any structural change of the deposit taken place at the
time of arrival at the surface. The effect of the entropic contribution to the thermodynamic potentials will most likely be
appreciably different when considering multisite occupancy of the adsorbates (which is not taken into account in the BET-
model, which assumes monomer-like particles occupying one lattice site).
The surfaces of metal oxides may contain hydroxyl groups, oxygen ions and coordinately unsaturated metal cations. These
metal cations can be considered as Lewis acid sites. The hydroxyl group may exhibit acidic or basic properties depending on
the electronegativity and oxidation state of the metal cation as well as on their coordination number with respect to the
surface cations. Depending on whether the outermost oxide layer of the antireflective coating of the lens is basic (i.e MgO,
CeO
2
), acidic (i.e SiO
2
) or amphoteric (i.e Al
2
O
3
), the tendency towards contamination will vary. It is also likely that the
atomic structure of the oxides is important for the chemisorption. The possibility of the aromatic contaminants to become
oxygenated due to the photo-induced oxidation via ozone reactions, may cause the interaction with the lens surface to
become stronger and thus the adsorption quantities to increase. Also, the relative humidity may play a role for the
contamination due to regeneration of surface hydroxyl groups on the lens surface. This is expected to increase the tendency
towards chemisorption. It is also likely that diffusion, adsorption and reaction occurs simultaneously during the
contamination-event, thus making the model-making quite difficult. Basically, more research is needed in order to minimize
AMC issues, obtain proper tools to diagnosticize and predict tool performance.
From Lifshitz-van der Waals interaction energy considerations, it was shown that the value of the Hamaker constant of the
lens material determine the tendency towards contamination using benzene as the prototype contaminant. Varying attractive
interactions between benzene and oxide and fluoride materials were obtained. Less interaction was calculated for fluorides in
comparison with oxides. It should be mentioned that, after some time the adsorbed benzene molecules will screen the
interaction between the lens surface and the gaseous species, thus leading to retarded Lifshitz-van der Waals interaction
forces which will decay rapidly with the separation distance. However, the exception here is benzene-benzene interactions
which should carry on nonretarded.
Experimental studies of SiO
2
(fused quartz) and CaF
2
as well as MgF
2
-based optics
[1]
have shown that the contamination
varies depending on lens material. Less or similar deposition was seen for MgF
2
compared to SiO
2
. Much thicker deposition
film thickness was found for CaF
2
when comparing all three materials. According to the concept of optical basicity, MgF
2
is
more acidic than the both SiO
2
and CaF
2.
SiO
2
, being most basic, is therefore expected to be more vulnerable for acidic type
of contaminants, while MgF
2
has the highest propensity towards reaction with basic contaminants among these three
materials. In this context it should be mentioned that the different surface groups for the mentioned oxide and fluorides
present different adsorption energies. In case of fused silica optics, the silanol groups at the surface may interact with
benzene or other aromatic hydrocarbons by hydrogen bonds to aromatic S-system (S orbitals). As already mentioned in
section 3.4.3, the polarizability of the contaminant molecule influences many other chemical and physical parameters
important in the bond mechanism. In Table 5 the polarizability is given for some typical aromatic contaminants. Benzene has
the lowest polarizability, which then should indicate lowest propensity to contaminate the lens surface, however, in case of t-
butyl benzene, highest tendency for deposition would be expected.
1624 Proc. of SPIE Vol. 5754
Table 5. Polarizability data for some aromatic molecules
[65]
Molecule Formula
Polarizability (Å
3
)
Benzene
C
6
H
6
10.32
Toluene
C
7
H
8
12.3
Ethylbenzene
C
8
H
10
14.2
o-Xylene
C
8
H
10
14.9
p-Xylene
C
8
H
10
14.1
m-Xylene
C
8
H
10
14.9
t-butyl benzene
C
10
H
14
17.8
In order to obtain not only qualitatively data, but also a quantitative measure of the whole process, more and better data are
needed. The overall picture indicates that contamination of microlithographic exposure tools is a growing concern in the
industry and that the contamination process is complex and multidisciplinary in nature. The need for a rigorous model is
obvious. Ideally, any model developed to describe the kinetics of the deposition process should fit the data over the
appropriate range for multiple types of contaminants and thereby be a reasonable theoretical interpretation of the process.
Collaborative work and exchange of knowledge and experience among companies, research institutes and universities is
most likely the fastest path in order to reach this goal.
6. CONCLUSIONS
In the present study, a theoretical analysis of some of the vital mechanisms involved in airborne molecular contamination of
lithographic lenses.was made. In addition, a thrust was made to set forward a rudimentary model capable of analyzing the
resist-outgassing situation near the final lens in the Sigma7300 laser pattern generator. The mechanisms responsible for lens
contamination are complex and difficult to approach, both from an experimental as well as from a theoretical side. Especially
when one consider that there could be an arsenal of contaminants in the vicinity of the lens with different chemical and
physical properties. However, without a fundamental understanding of these complex processes, it is hard to push the
development of proper lens and system design further. In the theoretical part, it has been suggested that for non-polar
compounds such as benzene, the interfacial surface energy terms influence the reaction kinetics and equilibria associated
with condensation at and wetting of lithographic lenses. Specifically, it is suggested that the physical adsorption is due to
attractive Lifshitz-van der Waals forces between the lens material and nonpolar hydrocarbons such as benzene. It has also
been proposed that chemisorption of benzene onto the lens surface is mainly determined by the acid-basic character of the
two materials. The concept of optical basicity was introduced as a tool to qualitatively determine the relative electron-donor
properties of metal oxides and fluorides and its importance for contamination of the optics. Multiphysics software and a
simplified lens contamination model were used in order to obtain a valuable research tool for further development of
Micronic’s laser pattern generator and thereby avoid optics lifetime issues and more focus on minimizing issues related to
tool uptime as being affected by hydrocarbon or other type of contaminants. Simulations of resist outgassing presently
suggest minor risk of lens contamination in the Sigma7300 system.
7. ACKNOWLEDGEMENTS
Dr. Roderick Kunz of MIT Lincoln Lab is gratefully acknowledged for fruitful discussions. The authors also would like to
thank Kaoru Aoyagi-san of FUJIFILM Electronic Materials, and Dr. Anders Sunesson of Perlos AB, former at the Dept. of
Atomic Physics, Lund University, for valuable communication regarding FEP-171 photoresist and ozone generation
respectively.
Proc. of SPIE Vol. 5754 1625
REFERENCES
1. R. R. Kunz, V. Liberman, D. K. Downs, “Experimentation and Modeling of Organic Photocontamination of
Lithographic Optics”, J. Vac. Sci. Tecnol. B 18(3), 1306-1313, (May/Jun 2000).
2. A. Grayfer, O. Kishkovich, D. Ruede, “Protecting 248 nm and 193 nm Lithography from Airborne Molecular
Contamination During Semiconductor Fabrication”, SPIE Proc. OpticalMicrolithography XIV, Vol. 4346 (2001).
3. A. Grayfer, O. Kishkovich, D. Ruede, “Protecting DUV Optics from Airborne Molecular Contamination”,
Microlithography World, p. 20, (Feb. 2002).
4. R. R. Kunz, “Fundamentals of Photochemical Contamination Control for Lithographic Tools”,
SPIE SC 355, SPIE Microlithography, 3-8 March 2002.
5. M. Ekberg, P-U Skotte, T. Utterbäck, S. Paul, O. Kishkovich, J. Hudzik, ”Laser Pattern
Generator Challenges in Airborne Molecular Contamination Protection”, SPIE Proc. Photomask and Next-Generation
Lithography Mask Technology X, Vol. 5130 (2003).
6. Ekberg , H. Fosshaug:, T. Utterbäck, P. Björnängen, T. Öström M. P-U Skotte, O. Kishkovich, J.
Hudzik, ”AMC (Airborne Molecular Contamination) Control in the Micro-Mirror SLM-Based Deep
Ultra-Violet (DUV) SIGMA 7300 Mask Writer”, SPIE Proc. Optical Microlithography XVII, Vol. 5377 (2004).
7. A. J. Dallas, K. Graham, M. Clarysse, V. Fonderie, “Characterization and Control of Organic
Airborne Contamination in Lithographic Processing”, SPIE Proc. Metrology, Inspection and
Process Control for Microlithography XVI, Vol. 4689, pp.1085-1109.
8. R. R. Kunz, V. Liberman, D. Downs, “Photo-induced Organic Contamination of Lithographic
Optics“, Microlithography World, p.2, (Feb. 2000).
9. R. R. Kunz, R. R. Dammel, “193-nm Litghography: Fundamentals and Issues”, Rep. under Advanced Lithography
Program of the Defense Advanced Research Projects Agency, Air Force Contract F19628-00-C0002, MIT/Lincoln
Laboratory, SEMATECH, Clariant Corp.
10. R. R. Kunz, D. K. Downs, “Outgassing of Organic Vapors from 193-nm Photoresists: Impact on Atmospheric Purity
Near the Lens Optics”, J. Vac. Sci. Technol. B 17(6), 3330-3334, (Nov./Dec. 1999).
11. C. Atkinson, J. Hanson, O. Kishkovich, M. Alexdander, A. Grayfer, “New Approach to Measurement of Photoactive
Deep UV Optics Contaminnants at Sub Parts-per Trillion Levels”, SPIE Proc. Optical Microlithography XVI, Vol.
5040, Part 1, pp. 499-409 (2003).
12. H. Okoroanyanwu, N. Stepanenko, G. Vereecke, A. Eliat, M. Kocsis, Y.S. Yang, R. Jonckheere, T. Conard, K. Ronse,
“Experimental Investigation of Fabrication-, Process-, Transportation-, Storage- and Handling-Induced Contamination
of 157nm Reticles”, SPIE Proc. Optical Microlithography XVI, Vol. 5377 Part 1, 487-503, (2004).
13. Rosling M., Karawajczyk A., Askebjer P., Zerne R., Carrol A., Eklund R., Fosshaug H., Sandström T., “Laser PG
Performance for 100 nm Photomasks”, 22
nd
Annual BACUS Symposium on Photomask Technology, Proceedings SPIE
Vol. 4889, pp.759-766 (2002)
14. Sandström T., Askebjer P., Sallander J., Zerne R., Karawajczyk A., ”Pattern Generation with SLM Imaging”, 21
st
SPIE
Proceedings, Annual BACUS Symposium on Photomask Technology,, Vol. 4889, pp.157-167, (2002).
15. M.M. Hills and D.J. Coleman, “Ultraviolet Laser Contamination of Quartz Optics”, Applied Optics, Vol. 32, No. 22,
pp. 4174-4177 (1993).
16. D.J. Erlich and R.M. Osgood Jr., “Laser Microchemistry: Local Nucleation Mechanisms for Photodeposition”, Thin
Solid Films, 90, p. 287 (1982).
17. M. Kawasaki, K. Hayashi, and H. Hada, “Initial Processes of D-Si Formation by Photo-CVD”, Oyo Butsuri, 55, p. 606
(1986).
18. J.A. Dean, Lange´s Handbook of Chemistry, 14
th
ed., McGraw-Hill, 1992.
19. S. Hien, S. Angood, D. Ashworth, S. Basset, T. Bloomstein, K. Dean, R.R. Kunz, D. Miller, S. Patel, and G. Rich,
“Photoresist Outgassing at 157 nm Exposure”, Advances in Resist Technology and Processing XVIII, F.M. Houlihan
(ed.), Proceedings of SPIE, Vol. 4345, pp. 439-447 (2001).
1626 Proc. of SPIE Vol. 5754
20. B.M. Mertens, B. van der Zwan, P.W.H. de Jager, M. Leenders, H.G.C. Werij, J.P.H. Benshop, and A.J.J. van
Dijsseldonk, “Mitigation of Surface Contamination from Resist Outgassing in EUV lithography”, Microelectronic
Engineering, 53, pp. 659-662 (2000).
21. M.M. Chauhan and P.F. Nealey, “Outgassing of Photoresists in Extreme Ultraviolet Lithography”, J. Vac. Sci. Technol.
B 18(6), Nov/Dec, pp. 3402-3407 (2000).
22. P. M. Dentinger, “Outgassing of Photoresist Materials at Extreme Ultraviolet Wavelengths”, J. Vac. Sci. Technol. B
18(6), Nov/Dec, pp. 3364-3370 (2000).
23. S. Tagawa et al., J. Photopolymer Sci. Tech., 11, p. 557 (1998)
24. Personal communication with Kaoru Aoyagi, FUJIFILM Electronic Materials Co.,Ltd, February 2005.
25. E.M. Lifshitz, “The Theory of Molecular Attractive Forces between Solids”, Sov. Phys. JETP (Engl. Transl.), 2, pp. 73-
83 (1956).
26. J.B. Fortin and T.-M. Lu, Chemical Vapor Deposition Polymerization: The Growth and Properties of Parylene Thin
Films, Kluwer Academic Publishers, p. 41(2004).
27. E.M. Charlson, E.J. Charlson, and R. Sabeti, IEEE Transactions on Biomedical Engineering, 33(2), p.202 (1992).
28. F.E. Cariou, D.J. Vally, and W.E. Loeb, IEEE Trans. Parts, Mater. Packag, 54 (1965)
29. R. Aveyard and D.A. Haydon, “An introduction to the principles of surface chemistry”, Cambridge University Press,
1973.
30. H.C. Hamaker, “The London-van der Waals Attraction between Spherical Particles”, Physica, 4, pp. 1058-1072 (1937).
31. E.M. Lifshitz, Sov. Phys. JETP, 2, p.73 (1956).
32. I.E. Dzyaloshinskii, E.M. Lifshitz, and L.P. Pitaevskii, Adv. Phys., 10, p.165 (1961).
33. V.A. Parsegian and B.W. Ninham, “Application of the Lifshitz’ Theory to the Calculation of van der Waals Forces
across Thin Lipid Films”, Nature (London), 224, pp. 1197-1198 (1969).
34. B.W. Ninham and V.A. Parsegian, “van der Waals Forces. Special Characteristics in Lipid-Water Systems and a
General Method of Calculations Based on the Lifshitz Theory”, Biophys. J., 10, pp. 646-663 (1970).
35. J. Mahanty and B.W. Ninham, Dispersion Forces, p.149, Academic Press, London, U.K., 1976.
36. D.B. Hough and L.R. White, Adv. Colloid Interface Sci., 14, pp. 3-41 (1980).
37. J. Israelachvili, Intermolecular and Surface Forces, 2
nd
ed., Academic Press, London, 1992.
38. J.N. Israelachvili, Q. Rev. Biophys., 6, pp. 341-387 (1974).
39. J. Tóth, “Uniform and Thermodynamically Consisten Interpretation of Adsorption Isotherms”, in Adsorption: Theory,
Modeling, and Analysis, ed. J. Tóth, Marcel Dekker, Inc. , New York, 2002.
40. S. Brunauer, P.H. Emmett, and E. Teller, “Adsorption of Gases in Multimolecular Layers”, J. Amer. Cer. Soc. , Vol. 60,
pp. 309-319 (1938).
41. D.H. Bangham and R.I. Razouk, Trans. Faraday Soc., 33 (1937), 1459.
42. B.V. Derjaguin, N.V. Churaev , and V.M. Muller, Surface Forces, Cons. Bureau, Plenum, N.Y., 1987.
43. B.V. Derjaguin and N.V. Churaev, Wetting films, Nauka, Moscow, 1984. alt. N.V. Churaev, Kolloidn. Zh., 36 (2),
p. 323 (1974).
44. T. Young, Philos. Trans. R. Soc. London, 95, p.55 (1805).
45. D. Kashchiev, Nucleation: Basic Theory with Applications, Butterworth-Heinemann, Oxford, 2000.
46. F. Garbassi, M. Morra, and E. Occhiello, Polymer Surfaces: From Physics to Technology, John Wile & Sons, 1994.
47. F.M. Fowkes, Ind. Eng. Chem., 66, p.382 (1962).
48. C.J. van Oss, R.J. Good, and M.K. Chaudhury, J. Colloid and Interface Science, 111, p. 378 (1986).
49. H. Rawson, Properties and Applications of Glass, Elsevier, Amsterdam, 1980..
50. J.A. Duffy and M.D. Ingram, J. Chem. Phys., 52(7), p. 3752 (1970)
51. J.A. Duffy, M.D. Ingram, Journal of American Chemical Society 93, 6448(1971).
52. C.K. Jörgensen, Modern Aspects of Ligand Field Theory, North-Holland, Amsterdam, 1971.
53. J.A. Duffy and M.D. Ingram, J. Chem. Phys., 54, p. 443 (1971).
54. J.A. Duffy, “A Common Optical Basicity Scale for Oxide and Fluoride Glasses”, Journal of Non-Crystalline Solids,
109, pp. 35-39 (1989).
Proc. of SPIE Vol. 5754 1627
55. R.R. Reddy, Y. Nazeer Ahammed, P. Abdul Azeem, K. Rama Gopal, and T.V.R. Rao, “Electronic Polarizability and
Optical Basicity Properties of Oxide Glasses Through Average Electronegativity”, Journal of Non-Crystalline Solids,
286, pp. 169-180 (2001).
56. V. Dimitrov and T. Komatsu, “Electronic Polarizability, Optical Basicity and Non-Linear Optical Properties of Oxide
Glasses”, Journal of Non-Crystalline Solids, 249, pp. 160-179 (1999).
57. J. Portier, P. Poizot, J.-M. Trascon, G. Campet, and M.A. Subramanian, “Acid-base Behavior of Oxides and Their
Electronic Structure”, Solid State Sciences, 5, pp. 695-699 (2003).
58. T. Yagi and M. Susa, “Temperature Dependence of the Refractive Index of Al
2
O
3
-Na
2
O-SiO
2
melts: Role of
Electronic Polarizability of Oxygen Controlled by Network Structure”, Metallurgical and Materials Transactions B,
Vol. 34B, p. 549 (2003).
59. E. Duisterwinkel, A.T.G.M. Bastein, W. van Schaik, “Feasibility of UV Cleaning of 157 nm Reticles“,
Microelectronic Eng., Vol. 67-68, No.1, pp. 3-9, (June 2003).
60. T.M. Bloomstein, V. Liberman, M. Rotschild, N.N. Efremov Jr., D.E. Hardy, S.T. Palmacci. “UV Cleaning of
Contaminated 157-nm Reticles”, SPIE Proc. Optical Microlithography XIV, Vol. 4346 Part 1-2, pp. 669-675, (2001).
61. H. Edner, A. Sunesson, S. Svanberg, L. Unéus, S. Wallin, ”Differential Optical Absorption Spectroscopy System
Used for Atmospheric Mercury Monitoring”, Appl. Opt. 25(3), 403, (1988).
62. H. Edner, A. Sunesson, S. Svanberg, “NO Plume Mapping by Laser-Radar Techniques”, Opt. Lett. 13(9), p. 704,
(1988).
63. H. Edner, G.W. Faris, A. Sunesson, S. Svanberg, “Atmospheric Atomic Mercury Monitoring Using Differential
Absorption Lidar Techniques”, Appl. Opt. 28(5), p. 921, (1989).
64. FEMLAB® Modeling Guide, Version FEMLAB 3.0, COMSOL AB, January 2004.
65. CRC Handbook of Chemistry and Physics, 72
nd
edition, CRC Press Inc., p. 10-202, 1991-1992.
1628 Proc. of SPIE Vol. 5754
... Fosshaug et al. [5] provides a detailed description of the multiple laser interactions occurring to initiate LIC. First, covering the interaction between the UV laser-light and the glass surface by pointing to experiments [6] showing that light-induced "prenucleation" occurs by photolysis of a molecular adlayer, forming radicals at the surface. ...
... Fosshaug et al. [5] provides a detailed description of the multiple laser interactions occurring to initiate LIC. First, covering the interaction between the UV laser-light and the glass surface by pointing to experiments [6] showing that light-induced "prenucleation" occurs by photolysis of a molecular adlayer, forming radicals at the surface. ...
... Silicon containing hydrocarbons are a class of airborne molecular contamination (AMC) causing persistent degradation of UV exposure tool optical surfaces 1,2,3 . Organic silicon compounds are efficiently split into components by 193 nm UV light, commonly used in photo-lithography applications. ...
Article
Full-text available
Trimethylsilanol (TMS) is a low molecular weight / low boiling point silicon-containing, airborne contaminant that has received increased interest over the past few years as an important cause for contamination of optical surfaces in lithography equipment. TMS is not captured well by carbon-based filters, and hexamethyldisiloxane (HMDSO), even though captured well, can be converted to TMS when using acidic filter media commonly used for ammonia removal. TMS and HMDSO co-exist in a chemical equilibrium, which is affected by the acidity and moisture of their environment. This publication shows that HMDSO is converted to TMS by acidic media at concentrations typically found in cleanroom environments. This is contrary to published results that show a re-combination of TMS to HMDSO on acid media. We also demonstrate that, based on its conversion to TMS, HMDSO is not a suitable test compound for hybrid chemical filter performance, as the apparent lifetime/capacity of the filter can be substantially skewed towards larger numbers when conversion to TMS is involved. We show lifetime test results with toluene and HMDSO on acidic and non-acidic filter media. Appropriately designed, asymmetric hybrid chemical filters significantly minimize or eliminate the conversion of HMDSO to TMS, thereby reducing the risk to scanner optical elements. Similarly, such filters can also prevent or reduce acid-sensitive reactions of other AMC when passing through filter systems.
Article
Full-text available
The Laser Interferometer Space Antenna (LISA) will be the first space-based gravitational wave observatory. LISA uses continuous-wave, infrared laser beams propagating among three widely separated spacecrafts to measure their distances with picometer accuracy via time-delay interferometry. These measurements put very high demands on the laser wavefront and are thus very sensitive to any deposits on laser optics that could be induced by laser-induced molecular contamination (LIMC). In this work, we describe the results of an extensive experimental test campaign assessing LIMC related risks for LISA. We find that the LIMC concern for LISA, even considering the high demands on the laser wavefront, may be greatly reduced compared to that observed at shorter wavelengths or with pulsed laser radiation. This result is very promising for LISA as well as for other space missions using continuous-wave, infrared laser radiation, e.g., in free space laser communication or quantum key distribution.
Article
Airborne molecular contaminations (AMCs) have been recognized as a major problem in semiconductor fabrication. Enormous technical and financial efforts are made to remove or at least reduce these contaminations in production environments to increase yield and process stability. It can be shown that AMCs from various sources in laser devices have a negative impact on quality and lifetime of lasers and optical systems. Outgassing of organic compounds, especially condensable compounds were identified as the main source for deterioration of optics. These compounds can lead to hazing on surfaces of optics, degradation of coating, reducing the signal transmission or the laser signal itself and can enhance the probability of laser failure and damage. Sources of organic outgassing can be molding materials, resins, seals, circuit boards, cable insulation, coatings, paints and others. Critical compounds are siloxanes, aromatic amines and high boiling aromatic hydrocarbons like phthalates which are used as softeners in plastic materials. Nowadays all sensitive assembly steps are performed in controlled cleanroom environments to reduce risks of contamination. We will demonstrate a high efficient air filter concept to remove AMCs for production environments with special AMC filters and methods for the qualification and monitoring of these environments. Additionally, we show modern techniques and examples for the pre-qualification of materials. For assembled components, we provide sampling concepts for a routine measurement for process, component and product qualification. A careful selection of previously tested and certified materials and components is essential to guarantee the quality of lasers and optical devices.
Article
Full-text available
A new analytical method for semiconductor-specific applications is presented for the accurate measurement of low molecular weight, silicon-containing, organic compounds TMS, HMDSO and D3. Low molecular weight / low boiling point silicon-containing compounds are not captured for extended periods of time by traditional chemical filters but have the same potential to degrade exposure tool optical surfaces as their high molecular weight counterparts. Likewise, we show that capturing these compounds on sample traps that are commonly used for organic AMC analysis does not work for various reasons. Using the analytical method described here, TMS, HMDSO and D3 can be measured artifact-free, with at least a 50:1 peak-to-noise ratio at the method detection limit, determined through the Hubaux-Vos method and satisfying a conservative 99% statistical confidence. Method detection limits for the compounds are 1-6 ppt in air. We present calibration curve, capacity, capture efficiency, break-through and repeatability data to demonstrate robustness of method. Seventy-one real-world samples from 26 projects taken in several fab environments show that TMS is found in concentrations 100 times higher than those of HMDSO and D3. All compounds are found in all environments in concentrations ranging from zero to 12 ppm, but most concentrations were below 50 ppb. All compounds are noticeably higher in litho-bays than in sub-fabs and we found all three compounds inside of two exposure tools, suggesting cleanroom and/or tool-internal contamination sources.
Article
The authors present results of extensive studies on the chemical behavior of low molecular weight silicon-containing species (LMWS) and associated challenges of their analytical determination and control to prevent adverse influence on critical optical elements of exposure tools. In their paper the authors describe a non-traditional approach to the creation of a TMS gaseous source for filter media development and an engineering solution to the challenge of controlling LMWS - a solution that shows a significant advantage over currently existing approaches.
Article
Full-text available
A case study of drastic photolithography defectivity reduction on i-line and Deep-UV (DUV) tools is presented. We show how this result is linked with reduction of Airborne Molecular Contamination (AMC) in clean room by combined installation of novel type of filters on tracks and on the recirculation air treatment. The root cause was identified to be the presence of acetic acid in clean room created by a reaction with the filters (mounted on track tools to exclude ammonia contamination of the process) and the photo solvent itself (here mainly 1-methoxy-2-propanol acetate: PGMEA). Crucial for the project success was the use of a real time monitoring tool to detect the sources of Volatile Organic Compounds (VOC). Finally, a model of chemical reaction of satellite defects creation is discussed based on a Time of Flight Static SIMS (TOF SSIMS) analysis together with new AMC specification for acetic acid for the photolithography area.
Article
The Hadamard transform-gas chromatography/mass spectrometry (HT-GC/MS) technique was successfully employed for the detection of hexamethyldisiloxane (HMDSO, C(6)H(18)OSi(2)) at the sub-nL/L level in a semiconductor wafer cleanroom. Indoor air samples were collected from the room, according to EPA Method TO-17 using a Tedlar bag where the air samples were allowed to pass through an absorption tube for 24 h. The condensed components were then heated and simultaneously injected into a GC column through a Hadamard-injector, which was operated in accordance with the Hadamard codes. Compared to the single injection used in most GC/MS systems, the signal-to-noise (S/N) ratios were substantially improved after the inverse Hadamard transformation of the encoded chromatogram. Under optimized conditions, when cyclic S-matrix orders of 255, 1023 and 2047 were used, the S/N ratios of the HMDSO signals were substantially improved by 7.4-, 15.1- and 20.1-fold, respectively. These improvements are in good agreement with theoretically calculated values (8.0-, 16.0- and 22.6-fold, respectively). We found that when the HT-GC/MS technique was applied, HMDSO could be detected at the 0.1 nL/L level.
Article
Organic materials desorbed from photoresist and construction materials can contaminate optical surfaces, with deleterious effects on lens performance. Experiments and models indicate that proper resist chemistry and non-zero ozone levels near lens surfaces can minimize this effect. The potential photodeposition of organic contamination on lens surfaces, reducing UV transmission, is becoming a significant concern in leading-edge lithography. The field of laser-induced chemical vapor deposition (CVD) has developed quantitative models that, when combined with analysis of trace organic compounds located near the optical train can provide accurate predictions for the reduction in lens performance over time.
Article
The role of airborne molecular contaminants (AMC) in semiconductor device fabrication with deep ultraviolet (DUV) exposure tools was discussed. These contaminants degrade the performance of photoresists by forming films on the exposed parts during photodeposition. The use of hybrid filters for removing molecular bases, condensable organic compounds and acidic species was recommended for avoiding the molecular contamination.
Conference Paper
The recently installed Sigma7100 laser pattern generator brings a new concept into photomask manufacturing. The spatial light modulator (SLM) technology enables 2D patterning using commercially available 248 nm lasers. This wavelength shift from the 413 nm wavelength of the Omega6000 scanning laser pattern generators facilitates the high resolution needed for 100 nm mask production. In addition, the partially coherence of the 2D patterning further enhances CD linearity and edge acuity. The rapidly increasing mask costs are partially attributed to increasing photomask writing times. These tend to increase as feature density increases with the roadmap, which is a challenge for any pattern generator with a limited number of writing beams. Instead, the SLM technology relies on the massive parallelism of one million micromirrors in combination with gray-scale control for fine addressing. A real-time FPGA-based data-rendering engine matches the speed. The result is pattern generation with high resolution at manageable mask writing times
Article
Photodeposition of organic films on transparent substrates irradiated in the presence of trace levels (ppb to ppm) of hydrocarbons has been experimentally investigated and a model is presented that describes the film growth behavior. The efficacy of a given organic precursor at forming a deposit is proportional to the product of its surface coverage (as governed by its partial pressure relative to its saturation partial pressure) and by its photon absorption cross section. These measurements are important in predicting the transmission characteristics of lithographic optics operating at 157, 193, and 248 nm wavelength. For example, a lens element irradiated continuously for 1 yr (1 kHz, 0.1 mJ/cm2/pulse) in the presence of 1 ppb of t-butyl benzene would exhibit a transmission of ∼87% at 193 nm. The effects of oxygen-containing ambients are also documented, and methods for elimination and/or prevention of organic contamination are suggested. © 2000 American Vacuum Society.