Amorphous SiO2 Surface Irregularities and their Influence on Liquid Molecule Adsorption by Molecular Dynamics Analysis

Abstract

As the semiconductor industry relentlessly reduces device sizes, efficient and precise cleaning processes have become increasingly critical to address challenges such as nanostructure stiction. Gaining insight into the molecular behavior of water and isopropyl alcohol (IPA) on silicon dioxide (SiO2) surfaces is essential for controlling semiconductor wet cleaning processes. This study investigated the interactions between these liquids and SiO2 surfaces. Using molecular dynamics (MD) simulations, we examined the adsorption behavior of water and IPA molecules on both amorphous and crystalline SiO2 (a-SiO2 and c-SiO2) surfaces. Our findings reveal a preferential adsorption of water molecules on a-SiO2 surfaces compared to c-SiO2. This preference can be ascribed to the irregularity of the a-SiO2 surface, which results in the presence of silanol groups that remain inaccessible to the liquid molecules. In contrast, the c-SiO2 surface exhibits a more uniform and accessible structure. This study not only imparts crucial insights into the molecular behavior of water and IPA on SiO2 surfaces but also provides valuable information for future enhancements and optimization of semiconductor wet surface preparation, cleaning, etching and drying.

Full text

ECS Journal of Solid State
Science and Technology
Amorphous SiO2 Surface Irregularities and their
Influence on Liquid Molecule Adsorption by
Molecular Dynamics Analysis
To cite this article: Masayoshi Takayanagi
et al
2023
ECS J. Solid State Sci. Technol.
12 083003
View the article online for updates and enhancements.
You may also like
Surface Structural Comparison of
Composite Film of Bacteriorhodopsin and
Phosphatidylcholine Fabricated on
Amorphous Silicon Dioxide, Crystal Silicon
Dioxide, and Hydrogenated Amorphous
Silicon
Yutaka Tsujiuchi, Keigo Furuya, Jun
Matsumoto et al.
-
High performance shear thickening fluid
based on calcinated colloidal silica
microspheres
Sheng-Biao Zheng, Shou-Hu Xuan, Wan-
Quan Jiang et al.
-
Voc transient in silicon heterojunction solar
cells with µc-SiOx:H window layers
Xueliang Yang, Jianmin Song, Jing Yang
et al.
-
This content was downloaded from IP address 95.153.178.38 on 14/11/2023 at 07:02

Amorphous SiO
2
Surface Irregularities and their Influence on
Liquid Molecule Adsorption by Molecular Dynamics Analysis
Masayoshi Takayanagi,
1,2,3
Naozumi Fujiwara,
4,z
Ryuichi Seki,
4
Masanobu Sato,
4
and
Yasutoshi Okuno
4
1
Data Science and AI Innovation Research Promotion Center, Shiga University, Hikone, Shiga, 522-8522, Japan
2
Center for Training Professors in Statistics, The Institute of Statistical Mathematics, 10-3 Midori- cho, Tachikawa, Tokyo,
190-8562, Japan
3
RIKEN Center for Advanced Intelligence Project, Tokyo, 103-0027, Japan
4
SCREEN Semiconductor Solutions Co., Ltd., Kyoto, 602-8585, Japan
As the semiconductor industry relentlessly reduces device sizes, efficient and precise cleaning processes have become increasingly
critical to address challenges such as nanostructure stiction. Gaining insight into the molecular behavior of water and isopropyl
alcohol (IPA) on silicon dioxide (SiO
2
) surfaces is essential for controlling semiconductor wet cleaning processes. This study
investigated the interactions between these liquids and SiO
2
surfaces. Using molecular dynamics (MD) simulations, we examined
the adsorption behavior of water and IPA molecules on both amorphous and crystalline SiO
2
(a-SiO
2
and c-SiO
2
) surfaces. Our
findings reveal a preferential adsorption of water molecules on a-SiO
2
surfaces compared to c-SiO
2
. This preference can be
ascribed to the irregularity of the a-SiO
2
surface, which results in the presence of silanol groups that remain inaccessible to the
liquid molecules. In contrast, the c-SiO
2
surface exhibits a more uniform and accessible structure. This study not only imparts
crucial insights into the molecular behavior of water and IPA on SiO
2
surfaces but also provides valuable information for future
enhancements and optimization of semiconductor wet surface preparation, cleaning, etching and drying.
© 2023 The Electrochemical Society (“ECS”). Published on behalf of ECS by IOP Publishing Limited. [DOI: 10.1149/2162-8777/
acec0e]
Manuscript submitted May 25, 2023; revised manuscript received July 10, 2023. Published August 9, 2023.
Semiconductor devices have played a significant role in advan-
cing modern electronics and information technology. Over the past
few decades, their size has shrunk from micrometers to nanometers,
while their functionality and structural complexity have
increased.
1–3
These trends are expected to continue, as the improve-
ment of semiconductor devices is becoming significantly important
for the development of more advanced electronics such as wearable
devices, autonomous vehicles, and the internet of things (IoT).
4–6
Meanwhile, the scaling of semiconductor devices to nanostructure
has resulted in reduction in their mechanical strength.
7
Consequently, structural damage has become one of major concerns
in semiconductor manufacturing. Wet cleaning is one of the essential
processes in the semiconductor industry to maximize the yield of
chip within a silicon wafer by removing surface contaminants at the
nanoscale. However, during the drying step of wet cleaning, stiction
of nanostructures caused by the surface tension of the liquid started
occurring and it now becomes the most challenging issue today.
8–11
Water is the most commonly used liquid in semiconductor wet
cleaning, but it has relatively high surface tension due to the strong
hydrogen bond. Thus, isopropyl alcohol (IPA) has been suggested
and widely applied as an alternate liquid for drying because of its
lower surface tension and higher vapor pressure compared to
water.
12
Moreover, its high miscibility with water enables easy
integration into common cleaning procedures, such as the RCA
clean.
13
Then, IPA has been extensively used over the several
technology nodes by combining with an engineering method of
surface modification.
9,14,15
Koide et al. lowered the surface energy
of line-patterned silicon oxide structures through treatment with
surface modification agents (SMA) to weaken adhesion energy
between sticking patterns so that the stiction of pattern is restored.
14
In a design of surface modification methods, it is important to
understand the state of the pattern surface during the drying step, and
in the current situation where the scaling of device structures have
been miniaturized to the nanoscale, molecular-level insights are
required. Molecular dynamics (MD) simulations, which can analyze
and visualize the movements of molecules, are suitable for this
purpose.
Many studies have been done using MD methodologies to
investigate the behavior of liquid molecules in the vicinity of
surfaces in nanostructures,
16–22
including studies on the behavior
of water and alcohols on silicon dioxide (SiO
2
) surfaces. Yamaguchi
investigated the behavior of water and IPA on crystalline silicon
dioxide (c-SiO
2
) surfaces terminated with hydroxyl (OH) groups.
23
It was shown that both water and IPA molecules forms hydrogen
bonds with the surface silanol groups, and the former exhibited a
higher affinity for adsorption onto the surface. Naruke et al.
investigated the molecular-scale structures and mass transport
properties of water and IPA on c-SiO
2
surfaces terminated with
either hydroxyl or hydrogen atoms.
24
They revealed that there are
differences in the transport characteristics near the interface between
water and IPA, and that these differences are also dependent on the
surface termination. Although these studies have provided valuable
insights into the molecular-scale behavior of liquids in the vicinity of
a surface, they have focused on c-SiO
2
surfaces, which are not
commonly employed in semiconductor manufacturing processes.
As amorphous silicon dioxide (a-SiO
2
) surfaces are one of the
most frequently cleaned surfaces with wet cleaning, it is crucial to
clarify the behavior of water and IPA molecules on the a-SiO
2
surfaces to understand the phenomena during the drying process. In
this paper, we carried out MD simulations on systems consisting of
a-SiO
2
surfaces terminated with OH groups and water/IPA solution
to analyze the adsorption behavior of liquid molecules on the solid
surfaces. In comparison, a similar analysis was performed on c-SiO
2
surfaces. Furthermore, the differences between a-SiO
2
and c-SiO
2
,
surfaces were investigated, with a particular focus on the specific
factors associated with amorphous structures.
Computational Method
Molecular dynamics simulation conditions.—The interatomic
interactions between the atoms in the silicon dioxide (SiO
2
) layer
were modeled using the Tersoff potential.
25
The intermolecular
interactions between liquid-liquid and liquid-SiO
2
combinations
were modeled using the 12–6 lennard-Jones (LJ) potential for van
der Waals interactions, and the Coulomb potential for electrostatic
interactions as
z
E-mail: n.fujiwara@screen.co.jp
ECS Journal of Solid State Science and Technology, 2023 12 083003
2162-8777/2023/12(8)/083003/8/$40.00 © 2023 The Electrochemical Society (“ECS”). Published on behalf of ECS by IOP Publishing Limited

ϕε
σσ
πε
()= − + [
]
⎜⎟ ⎜⎟
⎧
⎨
⎩⎛
⎝⎞
⎠⎛
⎝⎞
⎠⎫
⎬
⎭
rrr
qq
r
441
ij ij ij
ij
ij
ij
ij
ij
ij
12 6
0
Where r
ij
is the distance between interaction sites iand j,ε
ij
,σ
ij
are
LJ energy and length parameters, q
i
,q
j
are point charges, and ε
0
is
vacuum permittivity. The charge and LJ parameters utilized in the
simulations are listed in Table I. The parameters for water were
retrieved from the SPC/E model,
26
those for IPA and surface OH
groups were from General Amber Force Field (GAFF) parameter
set,
27
and those for the SiO
2
were from LRL model.
28,29
The LJ
parameters for different components were determined by the
Lorentz-Berthelot mixing rule.
We verified the accuracy of these force field parameters in
depicting the hydrogen bond interactions between the surface and
liquid by examining the density distribution of liquid molecules near
the c-SiO
2
surfaces, as demonstrated in the models provided in
Fig. 1b. For both pure water and pure IPA liquid, the density
distribution of liquid molecules aligns with the earlier report by
Yamaguchi et al.
23
Water displays a damped oscillation pattern,
whereas IPA demonstrates a pattern of a first peak, a void layer with
zero density, followed by a second peak.
The velocity Verlet method was used for numerical integration
with a timestep of 0.25 fs. All the MD simulations were performed
using the large-scale atomic/molecular massively parallel simulator
(LAMMPS) molecular dynamics simulation software package.
30
Model preparation of SiO
2
surface and water-IPA liquid.—The
model of the a-SiO
2
surface was built in accordance with the
procedure described in a previous study.
31
First, the α-cristobalite
system comprising of 648 Si and 1296 O atoms with initial density
of 2.3 g cm
3
was thermalized through MD simulations. The tem-
perature was increased from 0.5 to 5000 K in 200 ps with a heating
rate of 25 K ps
−1
and equilibrated at 5000 K for 200 ps to obtain a
fully liquid structure. The liquid SiO
2
is then quenched from 5000 K
to 300 K with a cooling rate 9.4 K ps
−1
to create amorphous
structure. These MD simulations were performed using the Tersoff
potential designed for SiO system,
25
under the constant number of
particles, volume, and temperature (NVT) ensemble, using the Nose-
Hoover algorithm. Next, the simulation box of the system was
expanded in the z-direction, resulting in the creation of surfaces with
dangling bonds. The O and Si atoms with dangling bonds were
capped with H atoms or OH groups, and then the surface with silanol
groups was created. Next, to adjust the density of OH groups on the
surface, Si-O-Si bridges were formed through dehydration of two
adjacent OH groups. Finally, the a-SiO
2
surface was relaxed through
a 2 ns MD simulation under the NVT ensemble. The surface density
of OH groups on the prepared a-SiO
2
layer was 4.4 nm
−2
, which is
consistent with previous experimental observations.
32,33
The c-SiO
2
surface was constructed following the procedure
described in Ref. 16. The α-cristobalite system, comprised of 540 Si
atoms and 1080 O atoms, was cleaved in the (111) plane. The
surface was modified by adding OH groups to the dangling bonds of
the surface Si atoms, leading to the formation of a surface with OH
groups. The density of surface OH groups was 4.5 nm
−2
.
The water-IPA coexistence system was prepared with Packmol
package.
34
The water and IPA molecules were packed to conform
the density reported in Ref. 35 in each mixing ratio. The size of the
liquid system was 4.240 ×4.240 ×6.172 nm in the a-SiO
2
system
and 4.560 ×4.390 ×6.172 nm in the c-SiO
2
system, respectively.
The mixing ratios of water are 0, 5.76, 10.79, 20.07, 35.19, 52.65,
and 100 mol%.
By placing the liquid model on the a-SiO
2
and c-SiO
2
models, the
solid-liquid interface model was constructed. Figure 1shows the
MD simulation system of the water-IPA mixture and the a-SiO
2
surface (a) or c-SiO
2
surface (b). The pictures were drawn with
VMD software.
36
The overall system size was 4.240 ×4.240 ×
14.000 nm in the a-SiO
2
system and 4.560 ×4.390 ×14.000 nm in
the c-SiO
2
system. A vacuum region of approximately 5 nm was set
in the z-direction above the liquid layer. To prevent the overall
translational motion of the entire model, the atomic coordinates of Si
and O atoms located within 1.5 nm in the z-direction from the
surface not in contact with the liquid were fixed. To prevent the
dispersion of evaporated liquid molecules, a potential wall was
established at a position 3.5 nm away from the liquid surface.
Procedure of MD simulations and data analysis.—For both the
a-SiO
2
and c-SiO
2
models, we executed MD simulations as follows.
First, the initial model was relaxed through a MD simulation with a
time step of 0.1 fs for 15 ps at 200 K in NVT ensemble.
Subsequently, a MD simulation was executed changing the tem-
perature from 200 K to 300 K with a time step of 0.25 fs for 25 ps in
the NVT ensemble. Finally, production MD simulations for 36 ns at
300 K in the NVT ensemble were executed.
To quantitatively calculate the number of adsorbed liquid
molecules around the OH groups on the SiO
2
surface, we used the
radial distribution function (RDF). The RDF (
)
gr
ab between
particles of atom type aand
b
is calculated as
∑∑ δ()=( ) 〈 (∣ − ∣− )〉 [
]
−
==
gr NN rrr ,2
ab ab
i
N
j
N
ij
1
11
ab
Where N
a
and N
b
are the total number of type aand b, respectively.
We utilized MDAnalysis library to calculate RDF values.
37
Subsequently, the radial cumulative distribution function is calcu-
lated as.
∫π()= ′ ′ ( ) [
]
′
G r dr r g r43
ab
r
ab
0
2
Table I. The charge and LJ parameters used in MD simulations.
Species Charge (e) ε(kcal/mol) σ(Å)
Surface bridging oxygen −0.6000 0.1553 3.1656
Surface silicon 1.2000 0.4184 4.0000
Surface hydroxyl oxygen −0.7300 0.1553 3.1656
Surface hydroxyl hydrogen 0.4663 0.0000 0.0000
Water oxygen −0.8476 0.1553 3.1660
Water hydrogen 0.4238 0.0000 0.0000
IPA carbon with hydroxyl group 0.6516 0.1094 3.3997
IPA methyl carbon −0.4677 0.1094 3.3997
IPA alkyl hydrogen 0.1116 0.0000 0.0000
IPA hydroxyl oxygen −0.7246 0.2104 3.0665
IPA hydroxyl hydrogen 0.3988 0.0000 0.0000
ECS Journal of Solid State Science and Technology, 2023 12 083003

and the average number of type bparticles within radius ris finally
calculated as.
ρ()= () []Nr Gr,4
ab ab
where ρis the average atomic density. The number of water or IPA
molecules around the hydrogen atom of OH groups was calculated
by Eq. 2∼4with r=2.5 Å, which covers the first solvation shell.
Results and Discussion
Equilibration of liquid-solid interface through MD trajec-
tories.—As a first step of our analysis, we verified that the
adsorption structure of the SiO
2
-liquid interface was adequately
equilibrated through the 36 ns production MD trajectories. A total of
12 production MD simulations were performed, considering combi-
nations of the SiO
2
models (a-SiO
2
and c-SiO
2
) and the liquid
models with 6 different mixing ratios of water. Figure 2shows the
plots of the number of adsorbed water molecules N
OH
(r=2.5 Å)on
the SiO
2
surface OH groups during the MD trajectories calculated by
Eq. 4. Because the model was prepared by combining isolated SiO
2
surface and liquid models, the initial structure of the liquid-surface
interface is far from equilibrium. For example, in the case of a-SiO
2
with water mixing ratio 10.79 mol%, the initial values of N
OH
for
IPA and water were 0.94 and 0.13, respectively. Through equilibra-
tion during the 36 ns production MD trajectory, these values
converged to 0.44 and 0.77. The convergence of N
OH
has been
achieved for all the combinations of SiO
2
surface and water mixing
ratio.
Hydration structure of water/IPA on the SiO
2
surface.—After
confirming that equilibration had been achieved, we focus on the
equilibrium hydration structure of liquid molecules on the SiO
2
surfaces. Although the hydration structure of the silanol group is
known to have acceptor and donor types,
38
we only focused on the
acceptor type (O atoms of liquid accept hydrogen bond with H atoms
of surface OH groups) in this work. In Fig. 3, Eq. 2estimated RDF
between H surface hydroxyls and atoms of liquids, using the final
0.15 ns of the MD trajectories from the pure water (water mixing
ratio 100%) and pure IPA (0%) simulations. The RDF plots for the
water-IPA mixture simulations also showed similar trends. The first
peaks of O and H atoms of water are observed at 1.46 and 2.16 Å,
respectively, and those of O, C, and H
OH
atoms of IPA are observed
at 1.50, 2.60, and 2.24 Å, respectively. No obvious peak is observed
for the H
C
atom of IPA. These peaks reflect the hydrogen bond
interactions between the silanol groups and OH groups of liquid
molecules, while the CH
3
groups of the IPA molecules tend to avoid
the silanol groups. Similar adsorption behavior of IPA has also been
reported for the c-SiO
2
(1 1 1) surface terminated by hydroxyl
groups.
23
Based on these RDF plots, we considered a liquid
molecule to be adsorbed on the silanol group by forming a hydrogen
bond if the distance between the H atom of the silanol group and the
O atom of the OH group of the liquid molecule was within 2.20 Å.
Next, to investigate the relative preference of adsorption on the
silanol groups between water and IPA molecules, we analyzed the
MD trajectories at several mixing ratios. For each mixing ratio, Eq. 4
estimated the number of adsorbed water and IPA molecules (N
water
and N
IPA
) using the final 0.15 ns of the MD trajectories. Then, the
adsorption ratio of water (N
water
/(N
water
+N
IPA
)) was plotted as a
Figure 1. Snapshot of the model of IPA-water mixture liquid on (a) the a-SiO
2
and (b) the c-SiO
2
layer with its periodic boundary box. The z-direction is
perpendicular to the SiO
2
surface. The colors of SiO
2
surface atoms are red for O atoms, cyan for Si atoms, and white for H atoms. The color of water and IPA
molecules are blue and pink, respectively.
ECS Journal of Solid State Science and Technology, 2023 12 083003

function of the water mixing ratio, as shown in Fig. 4. If the
preference of adsorption is equal for both water and IPA, the plot
would follow the diagonal line. However, the plot obviously deviates
upward from the diagonal line, indicating a preference for water over
IPA. For example, at the water mixing ratio 10.79% on the a-SiO
2
surface, the adsorption ratio for water is 63.75%, which is as much
as six times than the water mixing ratio. A similar trend was
observed for other mixing ratios and surfaces as well. This is
consistent with previous reports on c-SiO
2
surfaces, which have
shown that water molecules preferentially interact with c-SiO
2
surfaces compared to ethanol molecules,
39
and that the adsorbed
IPA layer can easily be replaced by water molecules.
24
From our
analysis, it is also obvious that water molecules exhibit a higher
preference for adsorption on a-SiO
2
compared to c-SiO
2
, particularly
in the region of low water mixing ratio.
Correlation between the depth of OH groups on the a-SiO
2
surface and liquid adsorption.—Here we focus on the liquid-
dependent modification of the surface structures, whose inhomoge-
neous distribution of OH groups in a-SiO
2
would be important in
understanding the liquid accessibility during the cleaning processes.
Figure 5shows the number of H atoms of the surface OH groups
against the z-coordinate. The z-axis is perpendicular to the SiO
2
surface, and the solid-liquid interface is located at 0 Å, as determined
by the Gibbs dividing surface of the liquid. For c-SiO
2
with pure
water and IPA liquid (Figs. 5a, 5b), the peaks are at around −0.25 Å,
and their peak positions remain relatively constant throughout the
36 ns MD trajectories. This reflects that the surface structure of
c-SiO
2
remain stable after interacting with water or IPA liquid.
In contrast, for a-SiO
2
with pure water liquid (Fig. 5c), the plot
exhibits a peak at around −0.75 Åwith a tail extending toward the
Figure 2. Time evolution of the number of adsorbed liquid molecules during 36 ns MD simulations for (a) a-SiO
2
with IPA, (b) a-SiO
2
with water, (c) c-SiO
2
with IPA, and (d) c-SiO
2
with water.
Figure 3. RDFs between silanol groups of a-SiO
2
surface and water or IPA molecules in the liquid. (a) Those between the silanol H atom and water O and H
atoms. (b) Those between the silanol H atom and IPA O, C, H
OH
, and H
C
atoms.
ECS Journal of Solid State Science and Technology, 2023 12 083003

inner region of the a-SiO
2
, indicating the presence of surface
irregularities on the amorphous surface. The initial structure exhibits
larger tail around −5∼−2Å. After 36 ns, the tail decreased and the
peak at −0.75 Åincreased simultaneously. In the case of a-SiO
2
with pure IPA liquid (Fig. 5d), the plot behavior is similar to that
with water liquid, with a sharper peak at −0.75 Åobserved at the
end of the MD trajectory. A comparison between water (Fig. 5c) and
IPA (Fig. 5d) demonstrates a peak shift of the former towards the
liquid side. This suggests that water amplifies the deformation of the
surface structure through the formation of stronger hydrogen bond
with the surface OH groups.
To analyze the solvation behavior on the a-SiO
2
surface with
irregularities, we investigated the adsorption of liquid molecules on
each OH group on the surface. As previously mentioned, the criteria
for adsorption (hydrogen bond formation) is a distance of less than
2.20 Åbetween the silanol H atom the liquid O atom. In Fig. 6,we
plotted xz positions of OH groups on the a-SiO
2
surface with color-
coded symbols to indicate the presence or absence of water or IPA
adsorption at each OH group during the final 0.15 ns of MD
trajectories of pure water and pure IPA liquid. Blue dots represent
OH groups with both water and IPA adsorption, yellow triangles
represent those with only water adsorption, and gray squares
represent those with no adsorption of either liquid. It should be
noted that we did not observe any OH groups that exhibited only
IPA adsorption. In addition, in the cases with c-SiO
2
surface, all OH
groups were capable of adsorbing both water and IPA molecule.
The plot in Fig. 6reveals that several OH groups on the a-SiO
2
surface forms hydrogen bonds only with water molecules. It is
obvious that these OH groups are located deep within the surface
around z=−3Å. To examine the distribution of liquid molecules
around the surface, we also plotted the densities of center of mass of
water and OH groups of IPA molecules in Fig. 7. Above the Gibbs
dividing surface at z=0Å, both the water and IPA molecules exists
to interact with the surface OH groups. Below the surface, the
distribution of IPA is smaller compared to that of water, and
furthermore, only water molecules are present below −2Å. The
absence of IPA molecules within the deep surface should contribute
to the presence of OH groups on the surface that form hydrogen
bonds only with water molecules.
Irregularity-induced liquid adsorption on a-SiO
2
surfaces.—As
we have discussed in the previous section, the adsorption behavior of
liquid molecules on a-SiO
2
surfaces depends on the z-coordinate of
each surface OH group. Since the z-axis is perpendicular to the
surface, the z-coordinate indicates its depth within the surface. As
the z-coordinate decreases, the OH group is more likely to be
shielded from liquid molecules by other surface atoms. In this
Figure 4. Ratio of water adsorption at the OH groups on the a-SiO
2
and
c-SiO
2
surfaces.
Figure 5. The number of H atoms in surface OH groups against the z-coordinate with a 0.5 Åinterval. The plots are from the simulations of (a) c-SiO
2
with pure
water, (b) c-SiO
2
with pure IPA, (c) a-SiO
2
with pure water, and (d) a-SiO
2
with pure water.
ECS Journal of Solid State Science and Technology, 2023 12 083003

section, we investigate the atomistic structures of several OH groups
on the SiO
2
surface to obtain insights into their interactions with
liquid molecules.
In Fig. 8a, we visualize one of the OH groups that exhibited
adsorption with both water and IPA molecules. This OH group is
represented as dot A in Fig. 6. The OH group protrudes towards the
liquid phase, allowing for easy contact with both water and IPA
molecules. This behavior is exhibited by the majority of surface OH
groups, as shown in Fig. 6. Meanwhile, there are 6 OH groups that
exhibited adsorption only with water molecules and not with IPA
molecules. One example of this situation (triangle B in Fig. 6)is
visualized in Fig. 8b. The OH group is located in a groove of the
a-SiO
2
surface that is composed of a Si-O 18-membered ring, which
prevents large IPA molecules from contacting the OH group. This
observation is similar to the finding that only water molecules are
able to penetrate into the vacant spaces between H-terminal groups
on the c-SiO
2
(1 0 0) surface.
24
In the case of more deeply buried
OH groups, even small water molecules are unable to make contact
with the OH groups. One example (square C in Fig. 6) is visualized
in Fig. 8c. The OH group is completely covered by the surface Si
and O atoms, making it impossible for liquid molecules to make
contact. This diversity in adsorption behavior derived from the
surface irregularities contrasts with the OH groups on the uniform
surface in c-SiO
2
, as shown in Fig. 8d.
Conclusions
In order to investigate the adsorption behavior of water and IPA
molecules on the SiO
2
surfaces, we carried out MD simulations of
water/IPA solutions on both a-SiO
2
and c-SiO
2
surfaces, where all
silicon dangling bonds were terminated with OH groups. The
analysis of the radial distribution functions obtained from the MD
simulations indicated that both water and IPA molecules were
adsorbed onto both a-SiO
2
and c-SiO
2
surfaces through hydrogen
bonding with silanol groups. Furthermore, the analysis of the
adsorption ratio of water and IPA molecules on the SiO
2
surfaces
in comparison to the mixing ratio of water/IPA solution showed that
the a-SiO
2
surface had a higher proportion of water molecules
occupying silanol groups compared to the c-SiO
2
surface, especially
at low water mixing ratios. This is attributed to the accessibility of
silanol groups on the a-SiO
2
to small water molecules, but not to
bulky IPA molecules, which are hindered by steric factors as shown
in Fig. 6. The atomistic structures of such silanol groups exhibit
narrow burial (Fig. 8b) or deep burial (Fig. 8c) within the surface.
Figure 6. Adsorption of liquid molecules on each OH group on the a-SiO
2
surface with its xz coordinate. The blue dots represent OH groups with both water and
IPA adsorption, the yellow dots represent those with only water adsorption, and the gray dots represent those with no liquid adsorption.
Figure 7. The density distribution of z-coordination positions of the center of mass of adsorbed (a) water and (b) IPA molecules at OH groups on the a-SiO
2
surface.
ECS Journal of Solid State Science and Technology, 2023 12 083003

These results emphasize the importance of understanding the
phenomena on the a-SiO
2
surfaces. As the majority of the SiO
2
surfaces undergoing wet cleaning are amorphous, understanding
these phenomena becomes essential in order to grasp the surface
behavior during the wet cleaning of devices in semiconductor
manufacturing processes. In this report, we obtained insights into
Figure 8. Closeup visualizations of SiO
2
surface structures. (a)–(c) Visualizations of OH groups of symbols A, B, C in Fig. 6. (d) Visualization of c-SiO
2
surface.
ECS Journal of Solid State Science and Technology, 2023 12 083003

the limited molecular adsorption onto silanol groups on a-SiO
2
surfaces due to steric hindrance caused by their random positioning.
This knowledge will contribute to the prevention of nanostructure
stiction,
14
a critical issue in the drying process of wet cleaning in
semiconductor manufacturing. The stiction can be avoided through
surface treatment using a SMA for c-SiO
2
surfaces with no
irregularities by reducing the surface energy with the liquid.
Meanwhile, the SMA treatment is often insufficient for a-SiO
2
surface with irregularities because the SMA molecules cannot access
the buried silanol groups on the a-SiO
2
surface, resulting in
incomplete surface treatment and untreated silanol groups. This
issue is particularly prominent for realistic surfaces that would
display local variations of OH density on SiO
2
surfaces. To better
understand and address this issue, further analysis is required. This
atomistic mechanism will contribute to the improvement of surface
modification of a-SiO
2
, by identifying the need to treat the buried
silanol groups for complete surface treatment.
Acknowledgments
This work was financially supported by SCREEN Semiconductor
Solutions Co., Ltd.
ORCID
Naozumi Fujiwara https://orcid.org/0009-0005-7570-1893
References
1. R. Clark, K. Tapily, K. H. Yu, T. Hakamata, S. Consiglio, D. O’Meara, C. Wajda,
J. Smith, and G. Leusink, APL Mater.,6, 058203 (2018).
2. G. A. Brown, P. M. Zeitzoff, G. Bersuker, and H. R. Huff, Mater. Today,7,20
(2004).
3. A. J. van Roosmalen and G. Q. Zhang, Microelectron. Reliab.,46, 1403 (2006).
4. S. Seneviratne, Y. Hu, T. Nguyen, G. Lan, S. Khalifa, K. Thilakarathna, M. Hassan,
and A. Seneviratne, IEEE Communications Surveys & Tutorials,19, 2573 (2017).
5. A. Ahmad, 2020 8th International Conference on Reliability, Infocom Technologies
and Optimization (Trens and Future Directions) (ICRITO)1373 (2020).
6. K. S. Yeo, M. C. Chian, T. C. W. Ng, and D. A. Tuan, 2014 International
Symposium on Integrated Circuits (ISIC) pp. 568–71 (2014).
7. K. Schuegraf, M. C. Abraham, A. Brand, M. Naik, and R. Thakur, IEEE J. Electron
Devices Soc.,1, 66 (2013).
8. O. Raccurt, F. Tardif, F. A. D. Avitaya, and T. Vareine, J. Micromech. Microeng.,
14, 1082 (2004).
9. T. Ghosh et al., ACS Appl. Mater. Interfaces,14, 5537 (2022).
10. C. C. Yang et al., Solid State Phenomena,187, 253 (2012).
11. Y. Zhang, C.-W. Lo, J. A. Taylor, and S. Yang, Langmuir,22, 8595 (2006).
12. D. Eom, K. Kim, and Y. Shin, ECS Trans.,41, 197 (2011).
13. W. Kern, RCA Rev.,31, 187 (1970).
14. T. Koide, S. Kimura, H. Imori, T. Sugita, K. Sato, Y. Sato, and Y. Ogawa, ECS
Trans.,69, 131 (2015).
15. F. Sebaai, G. Vereecke, X. M. Xu, S. Baudot, F. Amemiya, K. Komori, and
F. Holsteyns, Solid State Phenomena,282, 190 (2018).
16. J. A. Harvey and W. H. Thompson, J. Phys. Chem. B,119, 9150 (2015).
17. R. H. Wells and W. H. Thompson, J. Phys. Chem. B,119, 12446 (2015).
18. J. Rodriguez, M. D. Elola, and D. Laria, J. Phys. Chem. B,113, 12744 (2009).
19. J. Rodriguez, M. D. Elola, and D. Laria, J. Phys. Chem. B,114, 7900 (2010).
20. J. Rybka, J. KÄRGER, and U. Tallarek, ChemPhysChem,18, 2094 (2017).
21. D. Roy, S. Liu, B. L. Woods, A. R. Siler, J. T. Fourkas, J. D. Weeks, and R.
A. Walker, The Journal of Physical Chemistry C,117, 27052 (2013).
22. B. L. Woods, J. K. George, A. M. Sherman, P. R. Callis, and R. A. Walker, The
Journal of Physical Chemistry C,119, 14230 (2015).
23. Y. Yamaguchi, S. Nakaoka, T. Hayashi, M. Kawakami, and D. Yano, ECS Trans.,
92, 87 (2019).
24. Y. Naruke, S. Kosaka, T. Nakano, G. Kikugawa, and T. Ohara, Int. J. Heat Mass
Transfer,84, 584 (2015).
25. S. Munetoh, T. Motooka, K. Moriguchi, and A. Shintani, Comput. Mater. Sci.,39,
334 (2007).
26. H. J. C. Berendsen, J. R. Grjgera, and P. T. Straatsma, J. Phys. Chem.,91, 6269
(1987).
27. J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman, and D. A. Case, J. Comput.
Chem.,25, 1157 (2004).
28. I. C. Bourg and C. I. Steefel, The Journal of Physical Chemistry C,116, 11556
(2012).
29. K. Leung, S. B. Rempe, and C. D. Lorenz, Phys. Rev. Lett.,96, 095504 (2006).
30. S. Plimpton, J. Comput. Phys.,117, 1 (1995).
31. J. Wang, A. M. Rajendran, and A. M. Dongare, J. Mater. Sci.,50, 8128 (2015).
32. L. T. Zhuravlev, Colloids Surf., A,173, 1 (2000).
33. L. T. Zhuravlev and V. V. Potapov, Russ. J. Phys. Chem.,80, 1119 (2006).
34. L. Martinez, R. Andrade, E. G. Birgin, and J. M. Martínez, J. Comput. Chem.,30,
2157 (2009).
35. D. W. Green and M. Z. Southard, Perry’s chemical engineers’handbook (2019).
36. W. Humphrey, A. Dalke, and K. Schulten, J. Mol. Graphics,14, 33 (1996).
37. N. Michard-Agrawal, E. J. Denning, T. B. Woolf, and O. Beckstein, J. Comput.
Chem.,32, 2319 (2011).
38. S. M. Melnikov, A. Höltzel, A. Seidel-Morgenstern, and U. Tallarek, The Journal of
Physical Chemistry C,117, 6620 (2013).
39. X. Nie, J. Chen, N. Sheng, L. Zeng, H. Yang, and C. Wang, Mol. Simul.,43, 1377
(2017).
ECS Journal of Solid State Science and Technology, 2023 12 083003