Hatano Y., Katsumura Y., Mozumder A. (Eds.) Charged Particle and Photon Interactions with Matter - Recent Advances, Applications, and Interfaces
Подождите немного. Документ загружается.
710 Charged Particle and Photon Interactions with Matter
Yamagami, R., Kobayashi, K., and Tagawa, S. 2008. Formation of spectral intermediate G-C and A-T anion
complex
in duplex DNA studied by pulse radiolysis. J. Am. Chem. Soc. 130: 14772–14777.
Yamakawa,
H., Abe, F., and Einaga, Y. 1994. Second virial coefcient of oligo- and polystyrenes near the θ
temperature.
More on the coil-to-globule transition. Macromolecules 27: 5704–5712.
Yamamoto,
H., Kozawa, T., Nakano, A., Okamoto, T., Tagawa, S., Ando, T., Sato, M., and Komano, H. 2005a.
Reaction
mechanisms of brominated chemically amplied resists. Jpn. J. Appl. Phys. 44: L842–L844.
Yamamoto,
H., Kozawa, T., Nakano, A., Okamoto, T., Tagawa, S., Ando, T., Sato, M., and Komano, H. 2005b.
Study
on acid generation from polymer. J. Vac. Sci. Technol. B 23: 2728–2732.
Yamamoto,
Y., Fukushima, T., Suna, Y., Ishii, N., Saeki, A., Seki, S., Tagawa, S., Taniguchi, M., Kawai, T., and
Aida, T.
2006a. Photoconductive coaxial nanotubes of molecularly connected electron donor and accep-
tor
layers. Science 314: 1761–1764.
Yamamoto,
Y., Fukushima, T., Jin, W., Kosaka, A., Hara, T., Nakamura, T., Saeki, A., Seki, S., Tagawa, S.,
and Aida, T. 2006b. A glass hook allows shing of hexa-peri-hexabenzocoronene graphitic nanotubes:
fabrication
of a macroscopic ber with anisotropic electrical conduction. Adv. Mater. 18: 1297–1300.
Yamamoto,
H., Kozawa, T., Tagawa, S., Ohmori, K., Sato, M., and Komano, H. 2007a. Single-component
chemically
amplied resist based on dehalogenation of polymer. Jpn. J. Appl. Phys. 46: L648–L650.
Yamamoto,
Y., Fukushima, T., Saeki, A., Seki, S., Tagawa, S., Ishii, N., and Aida, T. 2007b. Molecular engi-
neering of coaxial donor-acceptor heterojunction by coassembly of two different hexabenzocoronenes:
Graphitic
nanotubes with enhanced photoconducting properties. J. Am. Chem. Soc. 129: 9276–9277.
Yamamoto,
H., Kozawa, T., Saeki A., Tagawa, S., Mimura, T.,Yukawa, H., and Onodera, J. 2009. Reactivity of
halogenated
resist polymer with low-energy electrons. Jpn. J. Appl. Phys. 48: 06FC09.
Yanagisawa,
K., Kume, T., Makuuchi, K., Tagawa, S., Chino, M., Inoue, T., Takehisa, M., Hagiwara, M., and
Shimizu, M. 2002. An economic index regarding market creation of products obtained from utilization
of radiation and nuclear energy (IV) – Comparison between Japan and the USA. J. Nuclear Sci. Technol.
39:
1120–1124.
Yoshida,
Y., Tagawa, S., Kobayashi, H., and Tabata, Y. 1987. Study of geminate ion recombination in a solute
solvent
system by using picosecond pulse-radiolysis. Radiat. Phys. Chem. 30: 83-87.
Yoshioka,
Y., Kitagawa, Y., Takano, Y., Yamaguchi, K., Nakamura, T., and Saito, I. 1999. Experimental and
theoretical studies on the selectivity of GGG triplets toward one-electron oxidation in B-form DNA.
J.Am. Chem. Soc.
121: 8712–8719.
Yu,
G., Gao, J., Hummelen, J. C., Wudl, F., and Heeger, A. J. 1995. Polymer photovoltaic cells: Enhanced ef-
ciencies
via a network of internal donor-acceptor heterojunctions. Science 270: 1789–1791.
Zezin,
A. A., Feldman, V. I., and Sukhov, F. F. 1995. Effect of electron scavengers on the formation of paramag-
netic species upon radiolysis of polystyrene and its low-molecular-weight analogs. High Energy Chem.
29:
154–158.
Zhang,
G. H. and Thomas, J. K. 1996. Energy transfer via ionic processes in polymer lms irradiated by
0.4meV
electrons. J. Phys. Chem. 100: 11438–1145.
711
26
Radiation Chemistry
of
Resist Materials and
Processes
in Lithography
Takahiro Kozawa
Osaka University
Osaka,
Japan
Seiichi Tagawa
Osaka University
Osaka,
Japan
26.1 introduCtion
The initial steps in electron beam (EB) and extreme ultraviolet (EUV) resist reactions are mainly
induced by ionization and are different from the initial steps of photoresists, induced by excited states
of molecules. In the1980s, the increasing density of VLSI necessitated the development of submicron
lithography, such as excimer, x-ray, EB, and ion-beam (IB) lithography. At that time, much attention
was devoted to negative EB resists containing chlorinated or chloromethylated phenyl side chains,
which exhibit high sensitivity without losing resolution. Polymer cross-linking occurs very effec-
tively in these resists. This type of enhancement is also observed in other resists containing chlori-
nated or chloromethylated silicon-containing resists. Similar reactions of iodination and bromination
of phenyl rings are used. The rst systematic study on radiation chemistry of EB and x-ray resists
(Tabata et al., 1984; Tagawa, 1986, 1987) was carried out by pulse radiolysis in order to make clear the
mechanisms of high-sensitivity and high-resolution chlorinated resists, such as effective dissociative
electron attachment, formation of charge transfer complexes through geminate ion recombination,
and polymer radical and acid generation processes. Excimer laser ash photolysis studies on resist
materials were also carried out for chlorinated resists (Tagawa, 1986, 1987). In both photolysis and
radiolysis, polymer radical and acid generation takes place through the same charge transfer complex
as the initiator of resist reactions. But the initial steps of EB and x-ray resist reactions are mainly
Contents
26.1 Introduction .......................................................................................................................... 711
26.2 Energy
Deposition ................................................................................................................ 713
26.3
Thermalization
and Initial Conguration of Spurs .............................................................. 716
26.4
Electron
Migration and Multispur Effect ............................................................................. 718
26.5
Deprotonation
of Radical Cations ........................................................................................723
26.5.1
Dynamics of PHS Radical Cation (PHS
•+
) ...............................................................723
26.5.2 Dynamics
of PMS Radical Cation (PMS
•+
)..............................................................726
26.5.3 PMMA......................................................................................................................728
26.6 Summary .............................................................................................................................. 730
References...................................................................................................................................... 730
712 Charged Particle and Photon Interactions with Matter
induced by ionization and are different from the initial steps of photoresists induced by excited states
of molecules. The combination of geminate ion recombination and highly effective dissociative elec-
tron attachment is important in highly sensitive EB resists and also produces acid (Tabata et al., 1984;
Tagawa, 1986, 1987). A systematic research on radiation chemistry of chemically amplied resists
started with the proposal of the mechanisms of chemically amplied EB and x-ray resists (Kozawa
et al., 1992), and a systematic research on the mechanisms of EB and deep UV chemically ampli-
ed resists was carried out by product analysis, pulse radiolysis, and laser ash photolysis (Tagawa
etal., 2000). The radiation chemistry of chemically amplied EB and x-ray resists was reviewed
in 2004 (Tagawa et al., 2004). At that time, the limit of the resolution of chemically amplied resists
was believed to be around 60nm. But then EUV lithography began to develop, for which the resists
showing high performance, especially high resolution and high sensitivity, have been the key tech-
nology. The best data of the resolution of chemically amplied EUV resists has now reached sub-
20-nm with high dose, the best data of the resolution of chemically amplied EUV resists has now
reached sub-20-nm. Many serious problems, especially a trade-off relationship among resolution,
sensitivity, and line-edge roughness (LER), described in the following text, have been encountered
in the development of high-performance chemically amplied EUV resists. A systematic funda-
mental research, especially on radiation chemistry of chemically amplied EUV resists, has been
performed. The nanospace reactions through the combination of geminated ion recombination and
highly effective dissociative electron attachment play a very important role in chemically amplied
EUV and EB resists, similar to halogenated nonchemically amplied resists. The geminate pairs are
anion and proton in chemically amplied resists and the geminate pairs are anion and radical cations
in halogenated nonchemically amplied resists. Products of geminate ion recombination are acid and
two isolated radicals (distance between two radicals is long) in halogenated nonchemically amplied
resists, although products of geminate ion recombination are acid and pair of polymer radicals
(distance between pair radicals is very short). This chapter summarizes the recent progress in radia-
tion chemistry of chemically amplied EUV and EB resist materials and processes.
The concept of “chemical amplication” (Ito and Willson, 1983; Ito, 2005) has made a great
impact on resist designs. Chemically amplied resists utilize acid-catalytic reactions to form latent
images before development. Acids produced by exposure diffuse in the resist matrix and catalyze
chemical reactions, such as deprotection. In the early stage of the development of these resists, there
occurred many problems associated with the acid-catalytic reaction coupled with acid diffusion.
The instability of resist performance and the formation of surface insoluble layers are among these.
Owing to their worldwide intensive development, the performance of chemically amplied resists
had been improved for practical use. Chemically amplied resists were rst deployed in the mass
production lines of semiconductor devices during the transition of the exposure tool from the i line
of a Hg lamp to a KrF excimer laser. Since then, chemically amplied resists have been used as a
mainstream resist technology in KrF, ArF, and ArF immersion lithography. Despite acid diffusion,
sub-60-nm
features have been resolved in the mass production lines of semiconductor devices.
Thus
far, the miniaturization of feature sizes has been achieved mainly by shortening the wave-
length of exposure tools. The shortening of the wavelength implies the increase in the photon energy
of exposure tools. This trend will continue to enhance the ionization potential of resist materials with
the deployment of EUV exposure tools. The concept of chemical amplication will remain a key to
the realization of EUV lithography. Owing to the intensive development of chemically amplied resist
materials and exposure tools, the resolution of EUV lithography has reached the sub-30-nm region
with 10–20mJ cm
−2
sensitivity (Oizumi et al., 2007; Shimizu et al., 2007; Thackeray et al., 2008;
Yamashita et al., 2008; Itani, 2009). For the realization of EUV lithography, which targets a 22nm
resolution mass production, the performance of resist materials is however still inadequate. Another
important ionizing radiation for lithography is EB. EB lithography had been a preferred method for the
next-generation lithography until the development of EB projection lithography. However, EB lithog-
raphy is an indispensable technology for the fabrication of photomasks. 4× or 5× photomasks have
been used for the mass production lines of semiconductor devices. The requirement for the fabrication
Radiation Chemistry of Resist Materials and Processes in Lithography 713
accuracy using EB lithography also becomes increasingly strict with the reduction of feature size and
the increasing complexity of resolution enhancement technology, such as the optical phase shift.
In the development of resists for EUV lithography, a trade-off relationship was found among
resolution, sensitivity, and LER (Gallatin, 2005). This has become the most serious problem in
resist development. A premise for this trade-off relationship is that pattern formation efciency is
constant. The pattern formation efciency is determined by absorption efciency of incident energy,
quantum yield of acids, and efciency of catalytic chain reaction. However, the energy absorption
efciency is limited by sidewall degradation (Kozawa and Tagawa, 2008; Kozawa et al., 2008a).
The quantum yield is limited by secondary electron emission efciency (Hirose et al., 2008a). The
efciency of the catalytic chain reaction is limited by the diffusion-controlled rate of the chemical
reaction (Kozawa et al., 2008b,c). Therefore, the orthodox strategy to circumvent this problem is to
increase
the pattern formation efciency to its physical limits.
The
understanding of the energy consumption of incident radiation in resist materials is essen-
tial to the efcient use of the energy of incident radiation. In resist pattern formation of chemically
amplied resists, rst, an accumulated energy prole is formed by the energy deposition from an
exposure tool. Using the accumulated energy, acids are then generated. The accumulated energy
prole is converted into a latent acid image. The acids catalyze the pattern formation reaction during
post-exposure baking (PEB) and create a so-called latent image. By developing this image, a resist
pattern is formed. Although the photochemistry of acid generators has been intensively investigated
(Houlihan et al., 1988; Aoai et al., 1989; Buhr et al., 1989; Dektar and Hacker, 1990a,b; Sakamizu
etal., 1993; Ortica et al., 2000), the sensitization mechanism of acid generators in chemically ampli-
ed resists markedly changes above the ionization potential (Kozawa et al., 1992; Tagawa et al.,
2000). First of all, the energy deposition becomes a nonselective process with the increase in the
energy. In photoresists, acid generators are mainly decomposed from their excited states via direct
excitation or sensitization by excited polymer molecules (Dektar and Hacker, 1990a). It has been,
however, well known that chemically amplied photoresists are generally sensitive to ionizing radi-
ation, such as an EB. In the energy deposition of ionizing radiation into solute-matrix systems, such
as chemically amplied resists, the direct electronic excitation and ionization of solute molecules,
such as acid generators, are generally ineffective because the energy deposition is nonselective,
unlike in the case of photons. This is also true for EUV lithography. It has been pointed out that the
direct absorption of EUV photons by acid generators is negligible in typical EUV resists (Dentinger
et al., 2000; Hirose et al., 2007). The reason why acids are efciently generated in chemically ampli-
ed EB and EUV resists, even if the direct energy absorption of acid generators is small, is that acid
generators can react with low-energy electrons (∼0eV) generated by ionization events. However, it
has been pointed out that the reaction with low-energy electrons causes signicant blurring of latent
acid images (Kozawa et al., 2005, 2008d). Thus, low-energy electrons play an important role in the
pattern formation of chemically amplied EB and EUV resists. In this chapter, radiation chemistry
in
resist materials and processes in lithography have been reviewed.
26.2 energy deposition
The energy deposition on resist materials has been intensively investigated using conventional non-
chemically amplied resists, such as poly(methyl methacrylate) (PMMA). When a high-energy
electron enters the material, the electron loses its energy sporadically through inelastic scattering in
the material and excites or ionizes molecules. The energy deposition is expressed along the electron
trajectories
using Bethe’s continuous slowing down model for inelastic scattering (Bethe, 1930):
dE
ds
e
E
nZ
E
J
= −
2 1 166
4
π .
(26.1)
Here, E, s, e, Z, n, and J are the electron energy, the distance of the electron trajectory, the elemen-
tary electric charge, the atomic number, the volume concentration of atoms, and the mean excitation
714 Charged Particle and Photon Interactions with Matter
potential, respectively. In the relativistic energy region, a modied form of the Bethe equation is
used
(Shimizu et al., 1975):
dE
ds
e
m v
nZ
m v E
J
= −
−
( )
− − − −
( )
{ }
+ −
( )
+
2
2 1
2 1 1 2 1
4
0
2
0
2
2 2
2 2 2
π
β
β β βln l n
11
8
1 1
2
2
− −
(
)
β
(26.2)
Here, m
0
, v, and β are the electron rest mass, the electron velocity, and the ratio of v to the velocity
of light in vacuum, respectively. The relativistic effect on the resist sensitivity has been investi-
gated. Because the electrons are decelerated in the resist and the underlying substrate, the results
were actually controversial (Ryan et al., 1995). The relativistic effect has recently been observed
(Kim et al., 2005).
In EB lithography, the electrons backscattered from the substrate irradiate resist materials and
deform the designed patterns. Process simulators have been utilized for the correction of proximity
effects in the photomask design. The calculation of accumulated energy proles is a key element
of the process simulation for EB lithography. The accumulated energy is generally accurately esti-
mated by tracking the electron trajectories by the calculation of elastic and inelastic scattering in
resist materials (Hawryluk et al., 1974; Shimizu et al., 1975; Murata et al., 1981; Williams et al.,
1989; Aktary et al., 2006). The electron trajectories are calculated, for example, using Rutherford’s
differential cross-section model for elastic scattering, Moller’s cross-section model for fast second-
ary electron generation, and Bethe’s continuous slowing down model for inelastic scattering. This
method requires a full-scale Monte Carlo simulation and is time consuming. A point spread func-
tion instead is used for the calculation of accumulated energy proles so that the processing time
can be reduced down to an acceptable value (Chang, 1975; Parikh and Kyser, 1979; Rishton and
Kern,
1987; McMillan et al., 1989; Koba et al., 2005).
Several
forms of point spread functions for conventional resists, PSF
conv
, have been reported. One
of
the well-known forms is the so-called double Gaussian (Chang, 1975; Parikh and Kyser, 1979):
PSF r
r r
conv
f f b b
( ) exp exp= −
+ −
1
2 2 2 2
2
2
2 2
2
2
πβ β
η
πβ β
(26.3)
where
β
f
is the forward-scattering range
β
b
is the backscattering range
η
is the ratio of the energy intensity of backscattering to forward scattering
When
a Gaussian beam,
B r
r
e e
( ) exp= −
1
2 2
2
2
2
πβ β
(26.4)
is
assumed, the point spread function including beam blur is expressed as
PSF r B x x y y PSF x y dx dy
conv conv
x
′
=
′
−
( )
+
′
−
( )
′
+
′
( )
′ ′
′
( )
2 2
2 2
,,
exp exp
′
∫
=
+
( )
−
+
( )
+
+
( )
y
f e f e b e
r1
2 2 2
2 2
2
2 2 2 2
π β β β β
η
π β β
−−
+
( )
r
b e
2
2 2
2 β β
(26.5)
Radiation Chemistry of Resist Materials and Processes in Lithography 715
Here, β
e
is the beam size. The parameters β
f
and β
b
are either calculated by Monte Carlo simulation
(Parikh and Kyser, 1979; McMillan et al., 1989) or are experimentally obtained, typically by ana-
lyzing resist patterns delineated by a point beam (Chang, 1975; Rishton and Kern, 1987). In Monte
Carlo simulation, the energy is mainly accumulated through ionization and electronic excitation
events when the energy threshold for the termination of the calculation is set at 10–20eV. Therefore,
Equation 26.1 or Equation 26.5 represents the probability density distribution of ionization and
electronic excitation events. This is also true when a point spread function is evaluated using a point
beam
and a PMMA resist. In PMMA resists, the molecular weight of ionized molecules
is
reduced
through side-chain detachment and subsequent main-chain scission near the ionization and elec-
tronic
excitation points (Kozawa et al., 2007a).
In
a Monte Carlo simulation for the calculation of the point spread function, the electron trajecto-
ries are generally tracked until the electron energy is reduced down to 20eV. The inelastic mean free
path of 20eV electrons is less than 1nm in organic materials (Tanuma et al., 1994). When pattern
formation reactions, such as main-chain scission and cross-linking, take place at ionized molecules,
the threshold of 20eV gives sufcient accuracy even for nanoscale simulation. The relation between
experimental and simulated resist proles has been intensively investigated using conventional non-
chemically amplied resists (Hawryluk et al., 1974; Shimizu et al., 1975). For chemically amplied
EB resists, acid generators are sensitized by secondary electrons similarly as PMMA. The differ-
ence between sensitization mechanisms of chemically amplied resists and PMMA is that acid
generators are decomposed through the reaction with low-energy electrons as low as thermal energy
(∼25meV) (Kozawa et al., 1992, 2003). Although the inelastic mean free path of 20eV electrons is
less than 1 nm, the thermalization distance reaches 3–7nm on an average in organic materials, as
discussed in the following text. This fact indicates that the accumulated energy prole does not cor-
respond
to the initial acid distribution (Kozawa et al., 2002; Saeki et al., 2002).
Although
the sensitization mechanism of acid generators in EUV resists is analogous to that of
EB resists (Kozawa et al., 2006a), the energy deposition mechanisms on resist materials in EUV
resists are different from those in EB resists. The absorption of EUV photons is expressed by
Lambert’s
law similar to that of the KrF and ArF excimer laser:
dI z
dz
I z
( )
( )= −α
(26.6)
Here, I, z, and α are the light intensity, the distance from the surface in the direction of depth, and
the absorption coefcient of the resist lm. This absorption coefcient is determined by the absorp-
tion coefcients of polymers and acid generators. In the resists for KrF lithography, acid generators
are also sensitized through energy or electron transfer from the excited states of polymers (Hacker
and Welsh, 1991; Cameron et al., 2001). However, the direct excitation of acid generators is a pri-
mary process for acid generation. Therefore, a tremendous effort has been devoted to the reduction
of the absorption coefcient of polymers during the development of KrF, ArF, and F
2
resists without
exception. On the basis of the Lambert–Beer law, the photosensitization of conventional resist mate-
rials has been formulated by Dill et al. (1975). Acid generation in chemically amplied photoresists
is also expressed by the formulation of Dill et al. However, the reaction mechanisms after photoab-
sorption
signicantly differ between deep ultraviolet (DUV) and EUV exposures.
When
EUV photons are absorbed by resist molecules, photoelectrons are emitted. The photo-
electrons with excess energy induce further ionization and electronic excitation. The number of
ionization events induced by a single EUV photon can be estimated using the W value. The W value
is dened as the average energy required to produce an ion pair by ionizing radiations. The W value
of atoms and molecules has been intensively investigated by a variety of measurements for different
radiations above several keV (Inokuti, 1975; Kimura et al., 1991). It is well known that the W value
is insensitive to quality and energy for the radiations above a few keV. W values of some gases, such
as rare gases, propane, and ethylene, have been reported for the energy region below 1 keV (Suzuki
716 Charged Particle and Photon Interactions with Matter
and Saito, 1985, 1987; Saito and Suzuki, 2001). The W value for electrons increases with decreasing
electron energy, because the ratio of the probability for excitation to that for ionization in a single
collision of an electron with a gas molecule increases below 1 keV. The increase in the W value for
photons is small compared with that for electrons except for the region just above the absorption
edge of molecules. The K edges of carbon and oxygen atoms are 284 and 547eV, respectively.
Therefore, the W value of a typical organic resist polymer for 92.5eV photons is only slightly higher
than that for 1keV photons or electrons. The W value of poly(4-hydroxystyrene) (PHS) is 22.2eV
for a 75keV EB (Kozawa et al., 2006b). Thus, approximately four electrons are generated in PHS
by a single EUV photon.
26.3 thermalization and initial ConFiguration oF spurs
The mean free path of secondary electrons is one of the factors that determine the resolution of
lithography using ionizing radiation. The effects of high-energy secondary electrons on the reso-
lution of EB lithography have been well investigated, but are not discussed here. In EUV litho-
graphy, the energy of photoelectrons is approximately 80eV. The elastic and inelastic mean free
paths of 80 eV electrons are less than 1nm. The mean free path in EUV resists does not seem to
matter for resolution. However, the acid generators in chemically amplied resists are decomposed
through dissociative electron attachment, and the targeted resolution of EUV lithography is below
22nm. Therefore, the thermalization distance is not negligible in the next-generation lithography.
Thermalization mechanisms have been intensively investigated (Goulet et al., 1990). Although it is
still controversial, the thermalization mechanism of ejected electrons is considered to be as follows.
The electrons ejected by ionization rapidly lose their energy through the ionization and electronic
excitation. When the energy of electrons becomes subexcitational, the energy loss is mainly due to
the excitation of intramolecular vibration. Below subvibrational energy, electrons are thermalized
by, for example, the processes of energy exchange and the excitation of intermolecular vibrations
and phonons. Rassolov and Mozumder suggested that most of the thermalization range is accrued
in the subvibrational stage because of the ineffectiveness of the energy loss process, and thereby, the
distribution would be relatively insensitive to the initial electron energy (Rassolov and Mozumder,
2001). Because the thermalization is generally completed within 1ps in condensed matter, the ther-
malization process cannot be directly observed. However, the thermalization distance can be esti-
mated by assuming the initial distribution and applying Onsager’s theory (Onsager, 1938). The
thermalization distances in many aromatic compounds range from 3 to 5 nm, and those in many
alkanes range from 5 to 7nm (Jay-Gerin et al., 1993; Mozumder, 2002). The inelastic mean free
path of 20–100eV is less than 1 nm in organic materials. The thermalization distance is signicantly
longer than the inelastic mean free path because of the ineffectiveness of energy loss below the
electronic excitation energy of molecules. The typically assumed forms of initial distributions are
exponential and Gaussian distributions. To explain the details of experimental results, the modied
forms of these distributions have also been proposed. The thermalization distance depends on the
distribution
prole.
The
initial distribution of thermalized electrons can also be estimated by observing the early
processes in radiation chemistry (Yoshida et al., 1987). The initial distribution in model organic
compounds associated with resist materials has been investigated using high-time-resolution pulse
radiolysis of benzene derivatives (Okamoto et al., 2003, 2007) and dodecane (Saeki et al., 2002).
Pulse radiolysis is a method for observing directly the dynamics of short-lived intermediates gener-
ated by a pulsed beam. The dynamics is observed by, for example, photoabsorption spectroscopy.
A high-energy (>MeV) EB is generally used as an excitation source for pulse radiolysis (Kozawa
et al., 2000; Yoshida et al., 2001). In the MeV energy region, the average distance between spurs is
100–500nm. The ejected electron is thermalized near its parent radical cation (ionized molecule)
compared with the other radical cations. The thermalized electron migrates under the strong electric
Radiation Chemistry of Resist Materials and Processes in Lithography 717
eld produced by its parent radical cation. The Onsager length, at which the potential energy of
electric
eld corresponds to the thermal energy, is expressed as
r
e
k T
c
B
=
2
ε
(26.7)
Here, ε, k
B
, and T are the dielectric constant, the Boltzmann constant, and the absolute temperature,
respectively. Because the dielectric constant of benzene derivatives and dodecane is 2, the Onsager
length is approximately 28nm. The Onsager length is dened as the length at which the thermal
energy of the electron corresponds to the energy of the electron under the Coulomb potential. Most
electrons are thermalized within the Onsager length in these model compounds, since the thermal-
ization distance is 3–7nm. Therefore, the recombination reaction between thermalized electrons
and radical cations is geminate recombination between a thermalized electron and its parent radical
cation in the early stages in radiation chemistry of these compounds. It has been demonstrated that
the exponential distribution well reproduces the dynamics of intermediates in the model compounds
(Saeki et al., 2002; Okamoto et al., 2003, 2007). Assuming an exponential distribution, the thermal-
ization distances in PHS (Kozawa et al., 2006b) and PMMA lms (Kozawa et al., 2007a) have been
estimated to be approximately 4 and 6nm, respectively, by analyzing the dependence of acid yields
on
acid generator concentration.
An
ionized molecule with a positive charge (radical cation) and an electron with a negative charge
interact through the Coulomb interaction. When spurs overlap, the electron dynamics changes and
affects the chemical yield and distribution. The phenomenon that describes the interaction between
packed ion pairs is called the “multispur effect.” Although the inuence of the multispur effect on
the recombination reaction and radiation-chemical yields has been investigated from the viewpoint
of the fundamental science of electron–material interaction (Clifford et al., 1982; Bartczak and
Hummel, 1987, 1993; Green et al., 1989; Bartczak et al., 1990; Wojcik et al., 1992; Siebbeles et al.,
1997; Saeki et al., 2005), the details are still unclear. To discuss the effect of spur overlapping in
resist materials, two different cases have to be considered. One is the case when the spurs gener-
ated by the same incident electron (or photon) are overlapped. The other is the case when the spurs
generated by different incident electrons (or photons) are overlapped. In the latter case, the effect
depends on the exposure dose rate. Although the exposure-dose-rate dependence of chemically
amplied resists has not been reported, it may become a problem when a high-power EUV source is
developed. However, we discussed the former case here. Because the former case is induced by one
incident electron (or photon), the effect is essential to the sensitization mechanisms of chemically
amplied resists used for ionizing radiation. A typical conguration of ionized molecules (parent
radical cations) and thermalized electrons is shown in Figure 26.1 (Kozawa et al., 2009). Within the
Onsager length, charged species interact through the Coulomb interaction. The Onsager length is
approximately 14nm in PHS, which is a typical backbone polymer for chemically amplied resists
used
in EB and EUV lithography.
The
distribution of overlapped spurs in EB and EUV resists has been investigated by a Monte
Carlo simulation. In the simulation, the inelastic scattering of 100,000 electrons and the subsequent
trajectories of secondary electrons were calculated. A histogram showing the numbers of ion pairs
generated is shown in Figure 26.2 (Kozawa et al., 2009). The distributions are signicantly different
between EB and EUV exposures. The difference is mainly caused by the fact that the “rst” second-
ary electron is generated through inelastic scattering in EB resists, although it is generated through
photoelectron emission in EUV resists. The probability of generating a single spur in EB resists was
0.54. The probability of generating a multispur (N = 2) was 0.41. A signicant number of spurs also
overlapped in EB resists as well as in EUV resists. The average number of ion pairs in an isolated
space
is 1.5 in EB resists and 4.2 in EUV resists.
718 Charged Particle and Photon Interactions with Matter
26.4 eleCtron migration and multispur eFFeCt
The thermalized electrons migrate in the resist matrix until they nd the localization sites such as
acid generators. In the deceleration process and the following migration, the secondary electrons
excite the acid generators or attach to the acid generators (Kozawa et al., 1992). The ratio of acid
yields generated through direct excitation and dissociative electron attachment depends on the acid
generator concentration (Kozawa et al., 2006b, 2007b). The acid yield generated through the elec-
tronic excitation of acid generators has been reported to account for approximately 10% of the total
acid yield at 10wt% acid generator concentration (Kozawa et al., 2006b). Because the acid genera-
tor is used in the concentration range of 5–10wt%, the decomposition of acid generators is mostly
induced
by the electron attachment to the acid generators.
The
relationship between the acid yield and physical properties of acid generators has been inves-
tigated from the viewpoint of the reactivity of acid generators with thermalized electrons in resist
materials. The correlation between reductive potential and relative acid yield has been reported for
Onsager length (~14 nm in PHS)
N =1 N = 2 N =3
…
Single spur
Multispur
–
+
+
+
+
+
+
–
–
–
–
–
Thermalized electron
Ionized molecule
+
–
Figure 26.1 Typical conguration of ionized molecules (parent radical cations) and thermalized electrons.
The Onsager length is dened as the length at which the thermal energy of an electron corresponds to the
Coulomb potential. This length is approximately 14 nm in PHS. (From Kozawa, T. et al., Jpn. J. Appl. Phys.,
48,
056508, 2009. With permission.)
0
0.1
0.2
0.3
0.4
0.5
0.6
1 2 3 4 5 6 7 8
Number of ion pairs (N)
Generation probability
EB (75 keV)
EUV
Figure 26.2 Probability of generating each conguration shown in Figure 26.1 in EB and EUV resists.
The average number of ion pairs in an isolated space is 1.5 for EB and 4.2 for EUV. The energy of incident
electrons
is 75
keV.
(From Kozawa, T. et al., Jpn. J. Appl. Phys., 48, 056508, 2009. With permission.)
Radiation Chemistry of Resist Materials and Processes in Lithography 719
sulfonium and iodonium salts as well as nonionic acid generators (Nagahara, et al., 2000; Masuda
et al., 2006). Also, the reactivity of acid generators with solvated electrons formed in solutions has
been measured (Tsuji et al., 2000; Nakano et al., 2006a). The reactivity estimated from the rate
constant for the reaction with methanol-solvated electrons (Tsuji et al., 2000) well explains the acid
generation efciency for triphenylsulfonium triate (TPS-tf) upon exposure to EB (Kozawa et al.,
2006b) and EUV (Kozawa et al., 2007c; Hirose et al., 2008b). However, the same accuracy does not
always apply to a wide range of acid generators. It has neither been proved that reductive potential
applies
to a wide range of acid generators.
Recently,
the C
37
parameter has been proposed as a possible indicator for the reactivity of acid
generators, which may correlate with acid generation efciency in chemically amplied EB resists
(Natsuda et al., 2008). Figure 26.3 shows the dependence of the absorption spectra of exposed PHS
lms with TPS-tf and 5wt% Coumarin 6 (C6) on TPS-tf concentration (Natsuda et al., 2009). The
absorption bands at approximately 460 and 533nm are due to neutral forms and proton adducts of
C6, respectively (Pohlers et al., 1997; Yamamoto et al., 2004a). The absorption intensity at approxi-
mately 460nm decreased and that at approximately 533nm increased with increasing TPS-tf con-
centration. The concentration of C6 proton adducts was estimated using the Lambert–Beer law
from the difference in absorption intensity of spectra before and after exposure. In this estimation,
an absorption coefcient of 7.8 × 10
4
M
−1
cm
−1
at 533nm was used (Yamamoto et al., 2004b). The
dependence of acid (C6 proton adduct) yield on acid generator concentration in PHS lms contain-
ing an acid generator and C6 is shown in Figure 26.4 for 14 acid generators. The acid yield tended to
saturate at a high acid-generator concentration for all samples. The dependence of the acid yield on
the acid generator concentration has been investigated in detail using TPS-tf (Kozawa et al., 2006b).
An acid yield of 0.01 M corresponds to a G-value of acids of approximately 1. The G-value is the
number of molecules generated by an absorbed energy of 100eV. For all acid generators except
for diphenyliodonium triate (DPI-tf) and bis(p-tert-butylphenyl)iodonium triate (BBI-tf), the
G-value saturated at approximately 5. The G-value of ionization has been reported to be 4.5 for PHS
(Kozawa et al., 2006b). The G-value of acids generated through the direct electronic excitation of
TPS-tf has been estimated to be approximately 1 at a TPS-tf concentration of 0.6 M (Kozawa et al.,
2006b). The sum of two G-values is 5.5. Therefore, almost all secondary electrons are considered
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
400 450 500 550 600
Wavelength (nm)
Absorbance (normalized)
5 wt%
10 wt%
15 wt%
20 wt%
25 wt%
30 wt%
Figure 26.3 Absorption spectra of PHS lms with TPS-tf and 5wt% C6 upon exposure to 75 keV EB at a
dose of 10μC cm
−2
. The TPS-tf concentrations were 5, 10, 15, 20, 25, and 30wt%. Absorbance was normal-
ized with the thickness of resist lm (1μm). (From Natsuda, K. et al., Jpn. J. Appl. Phys., 48, 06FC05, 2009.
With
permission.)