Content uploaded by Naisheng Jiang
Author content
All content in this area was uploaded by Naisheng Jiang on Oct 20, 2015
Content may be subject to copyright.
Soft Matter
www.softmatter.org
ISSN 1744-683X
PAPER
Maya K. Endoh, Tadanori Koga et al.
Melt crystallization/dewetting of ultrathin PEO fi lms via carbon dioxide
annealing: the e ects of polymer adsorbed layers
Volume 10 Number 34 14 September 2014 Pages 6363–6590
Melt crystallization/dewetting of ultrathin PEO
films via carbon dioxide annealing: the effects of
polymer adsorbed layers
Mitsunori Asada,
ab
Naisheng Jiang,
a
Levent Sendogdular,
a
Jonathan Sokolov,
a
Maya K. Endoh,*
a
Tadanori Koga,*
acd
Masafumi Fukuto,
ef
Lin Yang,
f
Bulent Akgun,
gh
Michael Dimitriou
g
and Sushil Satija
g
The effects of CO
2
annealing on the melting and subsequent melt crystallization processes of spin-cast
poly(ethylene oxide) (PEO) ultrathin films (20–100 nm in thickness) prepared on Si substrates were
investigated. By using in situ neutron reflectivity, we found that all the PEO thin films show melting at a
pressure as low as P¼2.9 MPa and at T¼48 C which is below the bulk melting temperature (T
m
). The
films were then subjected to quick depressurization to atmospheric pressure, resulting in the non-
equilibrium swollen state, and the melt crystallization (and/or dewetting) process was carried out in air
via subsequent annealing at given temperatures below T
m
. Detailed structural characterization using
grazing incidence X-ray diffraction, atomic force microscopy, and polarized optical microscopy revealed
two unique aspects of the CO
2
-treated PEO films: (i) a flat-on lamellar orientation, where the molecular
chains stand normal to the film surface, is formed within the entire film regardless of the original film
thickness and the annealing temperature; and (ii) the dewetting kinetics for the 20 nm thick film is much
slower than that for the thicker films. The key to these phenomena is the formation of irreversibly
adsorbed layers on the substrates during the CO
2
annealing: the limited plasticization effect of CO
2
at
the polymer–substrate interface promotes polymer adsorption rather than melting. Here we explain the
mechanisms of the melt crystallization and dewetting processes where the adsorbed layers play vital roles.
I. Introduction
It is known that hierarchical morphologies of semicrystalline
polymers at different length scales (i.e., a well-ordered crystal-
line state, a long-range-correlated repeated lamellar stacking
grain, and a subsequent two- or three-dimensionally propa-
gated spherocrystal) play crucial roles in mechanical properties,
transport characteristics, and optical features of the polymer
bulk.
1
Owing to emerging advanced technologies, such as
organic photovoltaics,
2
organic transistors,
3
and organic light-
emitting diodes,
4
a challenge is to manipulate such crystalline
morphologies within spatially conned polymer lms where the
rate of crystallization, crystal orientation, and density of
nucleation sites are quite different from those of the bulk.
5–12
The main reason for these deviations is the interplay between
two interfaces: the polymer–air interface and polymer–substrate
interface.
13
For instance, the crystallization kinetics at the
topmost polymer surface is faster than that of the rest of the
lm.
14
This is mainly attributed to the presence of the “surface
mobile layer”
15–17
or the “surface reduced viscosity layer”
18,19
where the glass transition temperature (T
g
) or viscosity is much
lower than that of the bulk or lm interior. On the other hand,
the chain dynamics near the substrate–polymer interface is
retarded by the formation of semicrystalline polymer adsorbed
layers
20–23
at the polymer melt–substrate interface. It is proposed
that these opposing effects then propagate into the lm interior
through chain entanglements,
24
resulting in heterogeneities in
the crystalline structures within polymer thin lms.
12,22
However, a comprehensive understanding of the crystallization
process in nanoconned geometries is still lacking.
In this paper, by using compressed CO
2
gas (T
c
¼31.3 C and
P
c
¼7.38 MPa), we aim to explore the melt crystallization and
a
Department of Materials Science and Engineering, Stony Brook University, Stony
Brook, New York 11794-2275, USA. E-mail: tadanori.koga@stonybrook.edu; maya.
koga@stonybrook.edu
b
Kurashiki Research Center, Kuraray Co., Ltd., 2045-1 Sakazu, Kurashiki, Okayama
710-0801, Japan
c
Chemical and Molecular Engineering Program, Stony Brook University, Stony Brook,
New York 11794-2275, USA
d
Department of Chemistry, Stony Brook University, Stony Brook, New York 11794-
3400, USA
e
Condensed Matter Physics and Materials Science Department, Brookhaven National
Laboratory, Upton, NY 11973, USA
f
Photon Sciences Directorate, Brookhaven National Laboratory, Upton, New York,
11973, USA
g
Center for Neutron Research, National Institute of Standards and Technology,
Gaithersburg, Maryland, 20899, USA
h
Department of Chemistry, Bogazici Universitesi, Kimya Bolumu Bebek, 34342,
Istanbul, Turkey
Cite this: Soft Matter,2014,10,6392
Received 28th March 2014
Accepted 16th May 2014
DOI: 10.1039/c4sm00683f
www.rsc.org/softmatter
6392 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter
PAPER
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
View Journal
| View Issue
dewetting processes from the non-equilibrium swollen amor-
phous state under nanoconnement. It is known that CO
2
molecules absorbed into polymers play a role as diluent or
plasticizer for glassy/semicrystalline polymers, lowering T
g
and
melting temperature (T
m
) signicantly and hence enhancing
the chain mobility.
25–28
Especially, it was previously reported
that the excess sorption of CO
2
molecules near the critical point
of CO
2
takes place at the free surface as well as the substrate
interface, resulting in the excess swelling of supported polymer
thin lms regardless of the choice of polymer.
29–37
So far, the
detailed mechanism considering both the CO
2
sorption and
nanoconnement effects has yet to be rationalized.
21
To achieve
our objective, we chose poly(ethylene oxide) (PEO) ultrathin
lms spun-cast on Si substrates as a model since the melting/
crystallization behavior in CO
2
has been well characterized in
bulk
38,39
as well as in thin lms.
34
Firstly, we performed high-
pressure neutron reectivity experiments to investigate the
plasticization effect of CO
2
on the PEO ultrathin lms. The
results indicate that the CO
2
induced the transition from the
crystalline state to the amorphous state at a pressure as low as P
¼2.9 MPa and at T¼48 C which is below the bulk T
m
.
Subsequent rapid quench to atmospheric pressure allowed us
to create the non-equilibrium swollen state of the exposed PEO
lms. The polymer lms were then transferred to a tempera-
ture-controlled stage (set temperatures ranging from 10 Cto60
C), permitting us to study the melt crystallization or dewetting
process toward equilibrium at the given temperatures. The
structural developments were characterized by using various
surface-sensitive experimental techniques including X-ray
reectivity, grazing incidence X-ray diffraction, polarized optical
microscopy, and atomic force microscopy. Finally, we discuss
the formation of the adsorbed layer on the substrate via the CO
2
annealing, which plays crucial roles in the observed melt crys-
tallization and dewetting processes.
II. Experimental section
II-1. Sample preparation
Poly(ethylene oxide) (PEO, average M
n
¼20 000 g mol
1
, Sigma-
Aldrich, product no. 83100) was used without any further
purication. The bulk melting temperature (T
m
) of PEO was
determined to be 64 Cbydifferential scanning calorimetry
(DSC) (Perkin Elmer DSC 7). Si substrates were cleaned by
immersion in a hot piranha solution (i.e., a mixture of H
2
SO
4
and H
2
O
2
:caution:the piranha solution is highly corrosive upon
contact with the skin or eyes and is an explosion hazard when
mixed with organic chemicals/materials; extreme care should
be taken when handing it) for 15 min, subsequently rinsed with
deionized water thoroughly, and followed by submersion in an
aqueous solution of hydrogen uoride to remove a native oxide
(SiO
2
) layer. However, as will be discussed later, we conrmed
that the SiO
2
layer of about 1.3 nm in thickness was re-formed
even just aer hydrouoric acid etching, due to atmospheric
oxygen and moisture, as reported previously.
40
PEO thin lms
with average thicknesses of 20 nm, 50 nm, and 100 nm were
prepared by spin-coating PEO–toluene solutions onto hydrogen-
passivated Si (H–Si) substrates with a rotation speed of 2500
rpm. The thicknesses of the spin-cast PEO thin lms were
measured by an ellipsometer (Rudolf Auto EL-II) with a xed
refractive index of 1.455. The surface tension of PEO is 42.9 mJ
m
2
(the dispersion part is 30.9 mJ m
2
and the polar part is
12.0 mJ m
2
). The interfacial energy (g) between the polymer
and the substrate is then estimated to be 2.7 mJ m
2
at room
temperature and atmospheric pressure based on the Owens–
Wendt–Kaelble equation
41
with the dispersion part (48.71 mJ
m
2
) and polar part (3.98 mJ m
2
) of the surface tension of
H–Si.
11
Hence, it is reasonable to categorize the PEO/H–Si as an
attractive interacting system. All the spin-cast lms were dried
under vacuum at 25 C for 2 h before CO
2
processing and
directly placed into the high-pressure cell. Note that this drying
condition is not long enough to equilibrate the adsorption
process of PEO chains: the adsorption time for PEO at 25 Cisat
least 2 months.
II-2. In situ neutron reectivity (NR)
With a large penetration depth, neutron reectivity (NR) is an
ideal tool to determine the in situ thickness, composition, and
interfacial structure of polymer thin lms immersed in uids or
gases, under high pressure in thick-walled vessels.
32,42
The in
situ swelling behavior of the PEO thin lms in CO
2
was
measured by NR. Deuterated PEO (d-PEO, M
w
¼21 400 g mol
1
,
M
w
/M
n
¼1.09, Polymer Source Inc.) was used to enhance the
scattering contrast. The d-PEO thin lms with average thick-
nesses of 16 nm, 28 nm and 78 nm were prepared on 3-inch H–
Si wafers. Specular NR measurements were performed at the
NG-7 reectometer of the National Institute of Standards and
Technology Center for Neutron Research. The wavelength (l
N
)
of the neutron beams was 0.47 nm with Dl
N
/l
N
¼2.5%. The
details of the high-pressure NR experiments and high-pressure
cell are described elsewhere.
32
The NR experiments were con-
ducted under the isothermal condition (T¼48 C) with elevated
pressures up to P¼17.5 MPa. Temperature and pressure
stabilities during the NR measurements were within an accu-
racy of 0.1 C and 0.2%, respectively. The d-PEO thin lms
were exposed to CO
2
for up to 4 h prior to data acquisition to
ensure the swelling process reached equilibrium. The scattering
length density (SLD) values of CO
2
, which vary from 0.0004
10
4
to 2.5 10
4
nm
2
in the pressure range of 0.1 MPa < P<
17.5 MPa at T¼48 C, were calculated based on the density of
CO
2
obtained by the equation of state.
43
The NR data were
obtained by successively increasing the pressure and then
slowly decreasing the pressure. Since the background scattering
from a pure CO
2
phase increases dramatically near the critical
point,
29,32
we measured the scattering from the pure uid phase
(i.e., the long-range density uctuations) for the respective
pressure conditions. The NR data corrected for the background
scattering were analyzed by comparing the observed reectivity
curves with the calculated ones based on model SLD proles
having three tting parameters for each layer: lm thickness,
SLD, and roughness between the CO
2
and polymer layers rep-
resented as a Gaussian function.
44
The SLD proles were
subsequently converted into the corresponding polymer volume
fraction proles. Assuming that the concentration of the
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6393
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
mixture is homogeneous through the entire lm, the SLD value
of the polymer–CO
2
system is dened by
SLD
mix
¼SLD
p
f(z) + SLD
CO
(1 f(z)) (1)
where SLD
mix
is the SLD value of the CO
2
–polymer mixture at a
distance zfrom the substrate, SLD
p
and SLD
CO
are the pure
component SLD values of the polymer and CO
2
, respectively,
and f(z) is the volume fraction of the polymer at a distance z
from the substrate. In order to calculate the SLD
CO
of the CO
2
dissolved in the polymer, the reported density of 0.956 g cm
3
for CO
2
gas dissolved in a polymer was used.
45
We conrmed
that the repeated pressurization and depressurization processes
exhibit the same swelling behavior aer the rst pressurization
process. Hence, it is reasonable to conclude that the polymer
does not dissolve in the pure CO
2
phase during the NR experi-
ments, which is consistent with the bulk PEO solubility in
CO
2
.
46
To ensure conservation of the mass of the polymer for the
NR data tting, the volume fraction proles were calculated
such that the amount of polymer chains remained the same at
all solvent concentrations including in the dry state (i.e., before
CO
2
exposure).
II-3. Recrystallization via CO
2
Treatment
Based on the NR experiments, we found that CO
2
exposure at
P¼9.7 MPa and T¼48 C for 1 h is satisfactory to achieve an
amorphous state in the PEO. Aer the CO
2
exposure, the
chamber was rapidly quenched to atmospheric pressure within
10 s. As will be discussed later, the PEO lms were still
expanded just aer the depressurization, resulting in the non-
equilibrium swollen state. The swollen polymer lms were then
transferred quickly to a temperature-controlled sample stage
(set temperatures ranging from 10 Cto60C), and the kinetics
of the melt crystallization/dewetting at the given temperature
(T
cr
) was studied. Goel and Beckman used a similar CO
2
process
to establish a foaming method in which the bulk polymer is
saturated with supercritical CO
2
followed by rapid depressur-
ization.
47,48
However, we conrmed, by using small-angle X-ray
scattering experiments with a reection geometry,
49
that the
CO
2
process used in the present study generates only molecular-
scale porosity (average size of 0.8 nm) with a relatively broad
size distribution, which is consistent with a previous result on
semicrystalline poly(phenylene vinylene) lms treated with the
same CO
2
process.
49
The lack of large voids in the PEO lms
aer the CO
2
process may be due to the short diffusion path for
residual CO
2
, the high diffusivity of CO
2
in the rubbery PEO
matrix, and the low storage modulus of PEO.
50
It should also be
noted that, as we previously reported for another semi-
crystalline polymer,
51
a slow depressurization process (with a
quenching rate of ca. 0.15 MPa min
1
) under the same CO
2
conditions provokes more highly ordered crystalline structures
with larger grain sizes. However, due to experimental difficul-
ties in characterizing/visualizing the structural changes via the
slow quench in the high-pressure cell, we here focus on only the
rapid quench that allows us partly to preserve the swollen
structures and then perform a suite of ex situ structural char-
acterizations during the melt crystallization/dewetting
processes from the non-equilibrium swollen state. At the same
time, as controls, we treated PEO thin lms via a conventional
high-temperature annealing at 85 C(>T
m
) for 2 h, followed by
rapid quench to the required temperatures (between 10 and
60 C) by transferring to the temperature-controlled sample
stage.
II-4. Atomic force microscopy (AFM) measurements
The crystalline morphologies of the PEO thin lms were
observed by atomic force microscopy (AFM) (Digital Nanoscope
III and Bruker Bioscope Catalyst). Both contact mode and
standard tapping mode were conducted in air using a cantilever
with a spring constant of 0.06 N m
1
and 40 N m
1
, respectively.
The scan rate was 0.5 Hz or 1.0 Hz with a scanning density of
256 or 512 lines per frame.
II-5. Polarized optical microscopy (POM) measurements
Polarized optical microscopy (POM) measurements were con-
ducted by using reected light under an Olympus BHT micro-
scope equipped with a differential interference contrast
attachment for incident light aer Nomarski (NIC Model). POM
images were captured by a digital camera under polarized light
at room temperature.
II-6. Grazing incidence X-ray diffraction (GID)
Grazing incidence X-ray diffraction (GID) measurements of the
CO
2
-treated and thermally annealed PEO thin lms were carried
out at the X22B beamline at the National Synchrotron Light
Source (NSLS), Brookhaven National Laboratory (BNL). Two-
dimensional diffraction patterns were measured with a CCD
camera (Princeton Instruments) with an incident X-ray angle of
0.2, which is above the critical angle of PEO, hence illumi-
nating the entire PEO lms. The X-ray wavelength was 0.15 nm
and the exposure time for all the measurements was 400 s. As
will be discussed later, the adsorbed layers on the substrates
aer a solvent-leaching process were measured at two undulator
beamlines (the X9 beamline (NSLS) and the BL03XU at SPring-
8), since the diffraction peaks were too weak to be detected at
the bending magnet beamline.
II-7. X-ray reectivity (XR) measurements
X-ray reectivity (XR) measurements were conducted at the X10B
beamline (NSLS, BNL) to measure the lm thickness of the
adsorbed polymer layers on the H–Si substrates aer the solvent-
leaching process. The specular reectivity was measured as a
function of the scattering vector in the perpendicular direction to
the lm surface: q
z
¼(4psin q)/l,whereqis the incident angle and
lis the X-ray wavelength (0.087 nm). The XR data were tted by
using a standard multilayer tting routine for a dispersion value (d
in the X-ray refractive index) in conjunction with a Fourier
method, a powerful tool to obtain detailed structures for low X-ray
contrast polymer multilayers.
52,53
Note that dis proportional to the
density of a lm. The dvalue of the bulk PEO with the X-ray energy
at X10B (14.2 keV) is calculated to be d
bulk
¼1.34 10
6
with a
density of 1.2 g cm
3
.
6394 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter Paper
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
III. Results
III-1. In situ NR results
Fig. 1(a) shows representative NR proles for the 16 nm thick d-
PEO lm at four different pressures and T¼48 C. The solid
lines correspond to the best ts to the data based on the volume
fraction proles of the polymer, f(z), shown in Fig. 1(b). The
thickness of the layer, which was initially 16 nm, increased to 20
nm at P¼2.9 MPa and further increased to 23 nm upon
compression up to P¼9.7 MPa. It should be noted that the
contribution from the density uctuations of the pure CO
2
phase becomes signicant near the ridge condition (P¼9.7
MPa at T¼48 C) and overwhelms the observed intensity at q
z
>
0.1 ˚
A
1
.
29,32
This situation makes it difficult to quantitatively
discuss the presence of the adsorbed layer within the swollen
PEO thin lms or a CO
2
concentration gradient within polymer
lms.
35,54
Fig. 2 shows the linear dilation (S
f
) of the 16 nm thick
lm during the pressurization and depressurization processes.
The S
f
values were calculated by the equation S
f
¼(L
1
L
0
)/L
0
,
where L
1
and L
0
are the measured thicknesses of the swollen
and unswollen lm, respectively. In the pressurization process,
there are two distinctive features: (i) the large jump in S
f
from
0.05 to 0.25 at P¼2.9 MPa, implying a transition from the
crystalline state to the amorphous state at T¼48 C; and (ii) the
anomalous peak in S
f
near P¼10 MPa that is attributed to
the excess sorption of CO
2
molecules at the “density uctuation
ridge”
29–37
where the density inhomogeneity of CO
2
molecules
becomes maximum.
55
We also found that the overall swelling
isotherms of the 28 nm and 78 nm thick lms including the
excess swelling are nearly identical to that of the 16 nm thick
lm (data not shown). This thickness independence of the
excess swelling is consistent with previous experimental data of
rubbery polybutadiene thin lms in CO
2
with a reduced thick-
ness scaled by the radius of polymer gyration (R
g
¼5.1 nm for
the d-PEO used) of larger than L
0
/R
g
¼3.
32
In the inset of Fig. 2,
we also plot the pressure dependence of the root-mean-square
(rms) roughness (s) obtained from the best ts. From the
Figure, it is clear that the roughness decreases to a minimum
value of 0.5 nm at P¼2.9 MPa and then increases with
increasing pressure. This indicates that melting occurs at P¼
2.9 MPa, resulting in a much smoother surface, and the
increase in the roughness at P> 2.9 MPa is attributed to the
improvement of the solvent quality of CO
2
with increasing
pressure for the amorphous polymer.
34
It should be emphasized
that this melting pressure of 2.9 MPa is lower than those of bulk
PEO lms (4.8–10.7 MPa)
38,39
under the same isothermal
condition. Intriguingly, in the depressurization process, the
sharp change in S
f
at P¼2.9 MPa disappeared, and the S
f
value
decreased gradually to 0.15 at P¼0.1 MPa. Based on the fact
that the roughness aer complete depressurization to atmo-
spheric pressure is only 0.5 nm, which is nearly identical to that
at the transition point (P¼2.9 MPa), we postulate that the
swollen PEO thin lm just aer the depressurization is in the
non-equilibrium amorphous state. Hence, the subsequent
thermal annealing at the given temperature T
cr
prompts the
equilibration (i.e., crystallization or dewetting) of the PEO thin
lms whose chain conformations are unfavorably stretched in
air. We also conrmed that the repeated pressurization and
depressurization processes exhibit the same swelling behavior
as that of the rst depressurization process. Hence, we conclude
that the swelling isotherms aer the rst pressurization are in
equilibrium as sole functions of temperature and pressure.
Fig. 1 (a) Representative NR profiles for a 16 nm thick d-PEO film at
four different pressures and T¼48 C. The solid lines correspond to
the best fits to the data based on the corresponding concentration
profiles shown in (b).
Fig. 2 Pressure dependence of the linear dilation (S
f
) for a 16 nm thick
d-PEO film at T¼48 C during the pressurization and depressurization
processes. In the inset, the pressure dependence of the surface
roughness (the pressurization process) is plotted.
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6395
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
III-2. Crystalline structures via melt crystallization
Fig. 3 shows typical AFM and POM images of the PEO thin lms
aer the CO
2
process and subsequent thermal annealing as
functions of the lm thickness and T
cr
. The surface of the 20 nm
thick lm is depicted with AFM images, while the surfaces of
the 50 nm and 100 nm thick lms are depicted with POM
images because the structures grew too large for AFM to
capture. The surface morphologies are strongly dependent on
the lm thickness, as previously reported in CO
2
treated poly-
carbonate thin lms:
21
for the 50 nm and 100 nm thick lms,
the main folded-chain crystals grow into quasi-two-dimensional
spherulites with an average radius of a few hundred microme-
ters; for the 20 nm thick lm, the polymer crystallized into
nger-like seaweed patterns with an average width of a few
micrometers, grown via the so-called “diffusion limited aggre-
gation”(DLA) process.
7,56,57
Fig. 4(a) and (b) display representative GID patterns of the
20 nm and 50 nm CO
2
-treated PEO thin lms crystallized at
T
cr
¼25 C. All the GID patterns show discrete diffraction peaks,
indicating the formation of highly ordered crystalline struc-
tures, in contrast to broad ring patterns for the thermally
annealed PEO thin lms (Fig. 5(d)). To clarify the crystal
orientation, we simulated diffraction patterns for different
lamellar geometries with crystallographic a,b,orcaxis being in
the direction normal to the lm surface. The details of the
simulation are described elsewhere.
22
Note that the Herman-
Stein orientation factor
58
to dene the orientation along the
standing axis was xed to be 0.99, and only strong diffraction
peaks were considered for the analysis. For identication it was
possible to use the 120 diffraction peak at q¼13.6 nm
1
that is
located along the q
xy
axis for all three CO
2
-treated PEO lms
(indicated by the arrow shown in Fig. 4(a)). A comparison of the
simulated (Fig. 4(c) for at-on and Fig. 4(d) for edge-on) and
experimental results elucidated that the lamellar orientation is
Fig. 3 AFM topographic images (left column) and POM images (center
and right columns) of the CO
2
-treated PEO thin films crystallized as
functions of the film thickness and T
cr
. The height scale for the AFM
images is 0–20 nm.
Fig. 4 Two-dimensional GID images of the CO
2
-treated PEO thin
films crystallized at T
cr
¼25 C: (a) 20 nm; (b) 50 nm thick films. The
simulated diffraction patterns described in the text are shown (c) for
crystallographic c-axis standing (a flat-on lamellar orientation) and (d)
for crystallographic a-axis standing (i.e., an edge-on lamellar
orientation).
Fig. 5 AFM topographic images of the thermally annealed PEO thin
films crystallized at T
cr
¼25 C: (a) 20 nm; (b) 50 nm thick film. The
height scale for the AFM images is 0–20 nm. The corresponding two-
dimensional GID results are shown in (c) for the 20 nm thick film and
(d) for the 50 nm thick film.
6396 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter Paper
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
assigned to be that geometry with the caxis standing (i.e.,aat-
on lamellar orientation where the lamellar layers run parallel to
the surface and the molecular chains stand normal to the lm
surface). Consequently, we draw the conclusion that the highly
oriented at-on lamellar structure is formed in the PEO lms
aer the CO
2
process regardless of T
cr
and the lm thickness.
We also used the different CO
2
pressures ranging from 2.9 to 15
MPa at T¼48 C, where the polymer is in the amorphous state,
to investigate the effect of the excess sorption at the density
uctuation ridge on the resultant crystal morphologies. The
GID results revealed that the at-on lamellar orientation occurs
regardless of the CO
2
pressure, while the size of the crystal grain
(observed by POM) is the largest at P¼9.7 MPa (i.e., the ridge).
This can be explained by the fact that the amount of residue CO
2
molecules aer the rapid depressurization is the highest at the
ridge condition such that the melt crystallization proceeds at
the slowest rate. Hence, the anomalous plasticization effect
near the density uctuation ridge
30,49,51
is not crucial for the
resultant crystal orientation.
To highlight the uniqueness of the CO
2
-induced crystallized
structures, we also studied the structures crystallized via a
conventional thermal annealing process (i.e., without the CO
2
process). Fig. 5(a) and (b) show representative AFM height
images of the thermally annealed PEO thin lms (20 nm and 50
nm thick) crystallized at T
cr
¼25 C. As previously reported,
59,60
at-on lamellar structures characterized as screw dislocation
nucleation sites or terrace structures are seen in the 20 nm thick
lm. In contrast, a sheaf-like morphology is observed for the 50
nm thick lm, which is an indication of the edge-on lamellar
orientation. The detailed structures were further examined by
GID. Fig. 5(c) shows the GID pattern for the 20 nm lm. It is
obvious that the 120 diffraction peak for the 20 nm thick lm is
still located along the q
xy
axis, but displays more arc. This
indicates that the degree of the at-on lamellar orientation is
not so high compared to that of the CO
2
-treated lm. Fig. 5(d)
shows the GID result for the 50 nm thick lm where we can see
an azimuthal ring pattern, indicating the random orientation of
the edge-on lamellae. Table 1 summarizes the orientations of
the lamellar structures for the melt crystallized PEO thin lms
aer the thermal annealing processing. From the Table, we note
the following observations: (i) the at-on lamellar orientation is
preferable for all thermally annealed PEO thin lms crystallized
at and above T
cr
¼35 C; (ii) below T
cr
¼35 C, the random
orientation of the edge-on lamellae is formed in the 50 nm and
100 nm thick lms, while only the at-on lamellar orientation is
found in the 20 nm thermally annealed lms; and (iii) the
dewetting process occurs in all the thermally treated lms at
T
cr
¼60 C, which is very close to the bulk T
m
. We will discuss
the dewetting in detail in the next section. Hence, the thermally
annealed lms clearly show a thickness dependence of the
orientation, in contrast to the CO
2
-treated PEO thin lms.
Wang et al.
12
reported the effects of quench depth and lm
thickness on lamellar orientations formed in semicrystalline
polymer thin lms at atmospheric pressure. They introduced
the three-T
g
layered model (i.e., the surface mobile layer (T
g
<
T
g,bulk
), the middle bulk-like layer (T
g
¼T
g,bulk
) and the adsor-
bed layer (T
g
>T
g,bulk
)). It is postulated that each region has a
different bell-shaped overall crystallization rate curve as a
function of crystallization temperature in the range between T
g
and T
m
. As a result, the crystallization initiates, depending on
the crystallization rate at a given temperature, either at the free
surface (in the case of a higher T
cr
) where the chains are highly
mobile exhibiting lower T
g
15,61,62
and reduced viscosity,
18,19
or at
the substrate interface (in the case of a lower T
cr
) where the
adsorbed chains are barely mobile even at T[T
g
.
63
Hence, our
experimental results for the thermally annealed 50 nm and
100 nm thick lms can in principle be explained by the three-
layer model, but not for the 20 nm thick lm. We postulate that
the persistence of the at-on lamellae in the 20 nm thick lms is
attributed to the effect of an irreversibly adsorbed layer. We will
discuss this point later.
III-3. Dewetting structures
Here we focus on the dewetting behavior induced by the CO
2
annealing and subsequent thermal annealing at high T
cr
.Itis
known that dewetting is a phenomenon resulting from a
breakup of thin lms into droplets on substrate surfaces due to
high mobility of polymer chains.
64
Fig. 6 shows representative
POM images of the CO
2
-treated 50 nm thick PEO lms at
different T
cr
. Typical hole-shaped breakup structures
65–67
were
observed at and above T
cr
¼40 C (Fig. 6(a)–(c)), while the
thermally annealed 50 nm thick PEO lm dewetted at T
cr
$
60 C. Hence, the lower limit temperature for the CO
2
-induced
dewetting shis to the lower temperature side (by about 15 C)
relative to that for the annealed PEO lms. We found that the
CO
2
-induced dewetting is the most severe at T
cr
¼40 C
(Fig. 6(a)), which may correspond to the transition temperature
between crystallization and dewetting, as Okerberg et al.
pointed out.
68
The average height and the width of the rims (i.e.,
the hole boundary colored in light blue in Fig. 6(a)) are about
50–70 nm and 5–10 mm, respectively. Based on the AFM image
shown in Fig. 6(d), we found that the polymer chains inside of
the rims still crystallize into large needle-like (at-on) patterns.
Further AFM examination near the rims demonstrated that the
needle-like structures are initiated from the rim and propagate
into the inner region of the holes, as reported by Reiter and
Table 1 Lamellar orientation and instability of the PEO thin films
prepared via thermal annealing and CO
2
annealing as functions of the
film thickness and T
cr
Thermal annealing CO
2
annealing
T
cr
(C) 20 nm 50 nm 100 nm 20 nm 50 nm 100 nm
60 Dewet Dewet Dewet Dewet Dewet Dewet
55 Flat-on
a
Flat-on Flat-on Dewet Dewet Dewet
40 Flat-on Flat-on Flat-on Dewet Dewet Dewet
35 Flat-on Flat-on Flat-on Flat-on Flat-on Flat-on
30 Flat-on Edge-on
b
Edge-on
b
Flat-on Flat-on Flat-on
15 Flat-on Edge-on
b
Edge-on
b
Flat-on Flat-on Flat-on
10 Flat-on Edge-on
b
Edge-on
b
Flat-on Flat-on Flat-on
a
Flat-on lamellar orientation.
b
Coexistence with the at-on adsorbed
layer at the substrate interface.
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6397
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
Sommer.
56
In addition, we also conrmed that polymer chains
in the matrix (non-dewetting) region crystallize as well, forming
aat-on lamellar orientation.
In order to shed light on the process, we further studied the
time evolution of the dewetting structures. Fig. 7 shows the
dewetting structures of the CO
2
-treated PEO thin lms (20 nm
and 50 nm in thickness) induced at T
cr
¼40 C. The results
clearly show that the sizes of the dewetting structures are quite
different between the 20 and 50 nm thick lms in the same time
period. For the 20 nm thick lm, the kinetics is rather slow, and
small (micron-scale) dewetting structures become evident aer
12 h. On the other hand, ring-like dewetting holes of a few
hundred microns in diameter were observed in the 50 nm (and
100 nm) thick lms even aer 1 h, and no further growth of
these holes was observed in both thicker lms during the course
of the thermal annealing up to 12 h. These results suggest that
the dewetting kinetics for the CO
2
-treated PEO thin lms is
signicantly perturbed when the thickness is down to 20 nm. In
order to investigate this retarded dewetting kinetics, we induced
dewetting of the thermally annealed lms at T
cr
¼60 C
(without the CO
2
process). As a result, the same trend was
indicated: the slowing down of the dewetting kinetics for the
thermally annealed 20 nm thick lm compared to that of the
thermally annealed 50 and 100 nm thick lms. It is known that
the dewetting velocity is independent of lm thickness for
“nonslipping”(sticking) lms,
69
while, in the case of slipping
lms, thinner lms dewet faster than thicker ones.
70–72
Since the
interfacial energy of the PEO/H–Si system is very low (i.e.,an
attractively interacting system), the dewetting velocity would
remain constant regardless of lm thickness. This is at least
true for the 50 nm and 100 nm thick thermally annealed PEO
lms. Below we show that this thickness dependence of the
dewetting kinetics is a consequence of the interplay between the
polymer–air interface, where the dewetting is initiated, and the
polymer–substrate interface, where the adsorbed polymer layer
is formed.
IV. Discussion
Before moving into the main discussion, we briey explain
about irreversibly adsorbed polymer layers. Adsorbed polymer
layers at the solid/polymer melt interface have recently been the
subject of extensive study due to their strong inuence on the
physical and mechanical properties of polymeric materials
conned at the nanometer scale.
20,63,73–79
Several research
groups utilized the approach proposed by Guiselin,
80
which
combines prolonged high-temperature annealing and subse-
quent solvent leaching,
63,73,75,76
and demonstrated the formation
of adsorbed monolayers with a thickness of less than R
g
against
planar walls (substrates). The formation of adsorbed layers has
also been reported for semicrystalline polymers prepared on
planar substrates.
20,22,81,82
We begin with the thermally annealed PEO lms (50 nm
thick) to investigate the adsorbed layer. PEO lms annealed at
85 C for 2 h were solvent leached in baths of fresh chloro-
benzene at room temperature repeatedly. The details of the
leaching process are described elsewhere.
79
Aer each leach-
ing process, the residual layers were annealed at 85 Cfor1h
toremoveanyexcesssolventtrappedinthelms. We
Fig. 6 Representative POM images of the CO
2
-treated 50 nm thick
PEO thin films dried after 1 h at the following T
cr
values: (a) 40 C; (b)
45 C; (c) 50 C. The AFM topographic image inside of the rims in (a) is
shown in (d). The height scale for the AFM image is 0–50 nm.
Fig. 7 Time evolution of the surface morphologies of the CO
2
-treated
PEO thin films dried at T
cr
¼40 C and for the times indicated, depicted
with AFM images for the 20 nm film (left column) and POM images for
the 50 nm film (right column). The height scale for the AFM images is
0–20 nm.
6398 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter Paper
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
measured the thickness of the residual layer by using ellips-
ometry and XR. A series of the leaching experiments gave rise
to the residual (adsorbed) layers on H–Si substrates with
different thicknesses ranging from 2.5 nm to 8.5 nm. We
found that the residual layer of 2.5 nm in thickness (Fig. 8(e))
is the nal adsorbed layer the thickness of which remains
unchanged with any further aggressive leaching. As shown in
Fig. 8, the AFM experiments for these adsorbed layers clearly
show the nger-like seaweed at-on lamellar structures,
except for the 2.5 nm thick adsorbed layer. The GID experi-
ments at the X9 beamline (NSLS) were carried out to further
explorethestructureofthe2.5nmthickadsorbedlayer.As
compared to the GID data for the thermally annealed PEO
thin lm (50 nm thick, Fig. 9(a)), there are no diffraction
peaks from the nal adsorbed layer (Fig. 9(b)). On the other
hand, GID experiments at SPring-8 claried the 120 diffrac-
tion peak from the other adsorbed layers (Fig. 9(c)). Hence,
these AFM and GID results lead to the conclusion that the
chain architecture of the 2.5 nm thick nal adsorbed layer
prevents chain folding near the solid wall. We believe that this
nal layer corresponds to the “attened”layer
79
with many
solid/segment contacts that is composed of the early-arriving
polymer chains at the substrate during the adsorption
process. On the other hand, the other thicker adsorbed layers
should be composed of two layers: an inner attened layer
and an outer loosely adsorbed layer that is formed by the late-
coming chains during the adsorption process, which nd
fewer free surface areas and hence adsorb more loosely with
fewer adsorbed sites.
79
We previously demonstrated that
loosely adsorbed polystyrene chains can be preferentially
removed by optimizing solvent leaching conditions, while the
attened layer remains on the substrate.
79
Therefore, it is
reasonable to suppose that the loosely adsorbed polymer
chains in the present study are gradually washed away during
the consecutive leaching, resulting in the “transient”adsor-
bed layers of 4–8 nm in thickness.
Fig. 8 AFM topographic images of the PEO adsorbed layers: (a)
8.5 nm; (b) 6.5 nm; (c) 2.5 nm. The CO
2
-treated adsorbed layer (6.5 nm
thick) is shown in (d). The XR profile (red circles) of the 2.5 nm adsorbed
layer is shown in (e). The solid line corresponds to the best fit to the
data based on the three-layer model (i.e., PEO, SiO
2
, and Si layers). The
thickness of the SiO
2
layer was determined to be about 1.3 nm from an
independent XR measurement on a bare H–Si substrate. In the inset,
the dispersion profile (din the X-ray refractive index) obtained from the
best fit is shown. The surface roughness is estimated to be 0.7 0.1
nm. The dotted line in the inset corresponds to the dvalue of the bulk.
The formation of the high-density adsorbed layer next to a Si substrate
is consistent with previous reports concerning different polymers.
79,90
Fig. 9 Two-dimensional GID results for (a) the 50 nm thick spin-cast
PEO film and (b) the 2.5 nm flattened layer. The dark spots in both
images (indicated by the red arrows) correspond to the scattering
contributions from the SiO
2
layer formed on the H–Si substrate. The
one-dimensional GID results for the “transient”adsorbed layers
(6.5 nm and 8.5 nm in thickness) and the flattened layer (2.5 nm in
thickness) are shown in (c). The presence of the 120 diffraction peak at
around q¼13.6 nm
1
was used to determine the existence of the
crystalline structure.
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6399
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
Next, we clarify the formation of the adsorbed layer in the
CO
2
-treated lms before the subsequent thermal annealing at
T
cr
. With the aforementioned leaching process in conjunction
with XR and AFM, we found the presence of the nal attened
layer (2.5 nm in thickness) as well as the outer loosely
adsorbed layer in all the CO
2
-treated PEO lms (regardless of
the lm thickness and CO
2
conditions). At the same time, we
applied the same CO
2
treatment (i.e.,CO
2
exposure at T¼48 C
and P¼9.7 MPa and subsequent rapid depressurization) to the
unexposed PEO adsorbed layers shown in Fig. 8(a)–(c). As
shown in Fig. 8(d), the AFM result showed that the nger-like
seaweed pattern of the CO
2
-treated adsorbed layer (original
thickness of 6 nm) is slightly denser than that of the unexposed
one (Fig. 8(b)). We also found that the magnitude of the swelling
(obtained by ellipsometry and XR) of the CO
2
-treated adsorbed
layer is much less (S
f
¼0.03) than that of the spin-cast PEO lms
aer the complete depressurization to air (S
f
¼0.15, Fig. 2).
Hence, it is expected that the plasticization effect of CO
2
at the
interface is achieved to only a limited extent at the polymer–
substrate interface, as compared to the rest of the spin-cast
lms. Consequently, we postulate that this limited plasticiza-
tion effect plays a role in accelerating the polymer adsorption
and chain folding rather than melting of the crystalline struc-
tures near the interface. According to a previous report by Jia
and McCarthy,
83
CO
2
interacts with a hydroxyl group on a Si
substrate, screening the polymer–substrate interaction. This
would be the case for a polycarbonate–CO
2
system reported by
Lan and co-workers.
21
However, as Jia and McCarthy also
demonstrated, the screening effect is not fully achieved when a
strong interaction between a polymer and Si substrate (e.g., our
present system, PEO/H–Si) is present. We are currently studying
various polymers with different segment/solid interactions to
provide a better understanding of the adsorption mechanism in
the presence of CO
2
.
Based on the above results, we now propose the melt crys-
tallization process of the PEO thin lms associated with the
adsorbed layer. The key is the interaction between the adsorbed
layer and unadsorbed chains near the substrate interface. It has
been indicated that the presence of a so-called reduced mobility
interface (RMI) layer
84,85
(or transition zone) is required to
ensure continuity in the mobility prole from the adsorbed
layer to the bulk through chain entanglements.
24
Napolitano
et al. reported that the RMI layer of poly(ethylene terephthalate)
thin lms prepared on aluminum substrates is about 20 nm in
thickness.
20
While the formation/extent of the RMI layer may
depend on the strength of the substrate interaction, the exi-
bility of polymer chains, or the thermal annealing condi-
tions,
15,84–86
it is reasonable to suppose that part or the whole of
the interior region of the 20 nm thick PEO lms corresponds to
the transition zone. For the PEO/H–Si system, the interaction is
strong (“sticky”) so that the adsorbed chains are predicted to
adopt at-on lamellar formation at the substrate,
87
which is in
good agreement with the present AFM results. If the melt
crystallization from the amorphous state was initiated from the
adsorbed lamellar layer (that remains unmelted even at T[T
m
as seen in polyethylene thin lms
22
) as seed, the free polymer
chains in the transition zone might then form the at-on
lamellae with the same orientation as for the adsorbed lamellar
layer in order to maximize the contacts, thereby stabilizing the
interface.
88
This would be the reason why the at-on lamellar
orientation in the presence of the adsorbed layer is favorable for
the 20 nm thick lms regardless of choice of T
cr
and the use of
CO
2
annealing. It should also be noted that the melt crystalli-
zation process along with the adsorbed layer is analogous to the
“self-seeding”process for replicating polymer single crystals on
Si substrates.
89
The question is then why the at-on lamellar
orientation persists up to the topmost surface of the 100 nm
thick CO
2
-treated lm, in contrast to the thermally annealed
lms where the edge-on lamellar formation is dominant at the
topmost surface of lms of more than 50 nm in thickness at the
lower T
cr
(see Table 1). This may be attributed to great reduc-
tions of both T
m
and T
g
of the swollen PEO thin lms with the
CO
2
annealing. As summarized in Table 1, the plasticization
effect of CO
2
shis the lm stability (i.e., dewetting) to a lower
temperature. Hence, with the aforementioned three-layer
model,
12
we may conclude that the T
cr
values used in the present
study (as low as 10 C) may not be low enough to induce the
edge-on orientation at the topmost surface. As a consequence,
the at-on lamellar formation would proceed from the surface-
bound at-on lamellar layer and propagate into the lm interior
without emerging as independent nucleation at the top surface.
Further experiments at lower T
cr
should be the subject of
future work.
Finally, we look into the role of the adsorbed layer in the
dewetting process. As previously reported, the loosely adsorbed
polymer layer has nearly zero thermal expansion
79
and is
immobile
63
in air (or under vacuum) even at T[T
g
. While the
adsorbed layer swells in CO
2
, the degree to which it does so is
extremely small. Hence, the strong hindrance of the mobility of
the adsorbed chains is still sustained even in the CO
2
-treated
PEO lms. As a result, the effect of the adsorbed layer propa-
gates into the lm interior via the chain entanglements,
resulting in the strongly hindered dewetting dynamics of the
20 nm thick CO
2
-treated lm (the same mechanism should be
applicable to the 20 nm thick thermally annealed lm). It is
important to mention that the propagation distance of the
substrate effect of the CO
2
-treated lms is limited to the region
within 50 nm from the substrate interface, based on the dew-
etting kinetics at the topmost surfaces of the 50 nm thick and
100 nm thick lms. We are currently conducting experiments to
further quantify the effect of CO
2
on the RMI layer.
V. Conclusion
The CO
2
-induced crystallization and dewetting of PEO ultrathin
lms were investigated by a suite of surface-sensitive experi-
mental techniques. Spin-cast PEO lms (without further high-
temperature annealing) prepared on H–Si substrates were
exposed to CO
2
at T¼48 C and P¼9.7 MPa for 1 h to melt the
polymer and then depressurized quickly to atmospheric pres-
sure, allowing melt crystallization (and/or dewetting) from the
non-equilibrium amorphous state at the given crystallization
temperatures (T
cr
). As a result, we found that all the CO
2
-treated
lms show at-on lamellar structures regardless of the
6400 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter Paper
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
thickness and T
cr
. This contrasts with the experimental results
from the PEO thin lms crystallized via the conventional
thermal annealing process where the edge-on lamellar orien-
tation occurs at the lm surface. In addition, we found hole-
shaped breakup dewetting structures at the surface of the
CO
2
-treated PEO lms at and above T
cr
¼40 C, while the
thermally annealed PEO lm dewetted at T
cr
$60 C. Inter-
estingly, it was also found that the dewetting kinetics is severely
hindered for the 20 nm thick CO
2
-treated lm compared to that
for the 50 nm and 100 nm thick CO
2
-treated lms, while the
dewetting kinetics is expected to be independent of lm thick-
ness for the attractively interacting system.
69
We found that the formation of a very thin adsorbed layer
during the CO
2
annealing plays vital roles in the observed melt
crystallization and dewetting processes. Aer the CO
2
anneal-
ing, we used the solvent leaching process with chlorobenzene to
extract the adsorbed layer that cannot be dissolved even in a
good solvent. The systematic leaching experiments revealed the
formation of a non-crystallized adsorbed layer of about 2.5 nm
in thickness as well as transient crystallized adsorbed layers
(of up to 8.5 nm in thickness) on the H–Si substrates. It was also
found that the plasticization effect of CO
2
is achieved to only a
limited extent at the polymer–substrate interface, as compared
to the rest of the spin-cast lms. Hence, this limited plastici-
zation effect accelerates polymer adsorption and chain folding
rather than the melting of the crystalline structures near the
interface. We therefore postulate that the “bound”lamellar
layer, which does not melt even in CO
2
, initiates the melt crys-
tallization from the non-equilibrium amorphous state as a seed
during the subsequent annealing process. In addition, the
limited plasticization effect does not improve the inherent
hindrance of the mobility of the adsorbed polymer chains
either, maintaining the long-range perturbations associated
with the adsorbed layer into the lm interior through chain
entanglements. Since the formation of the adsorbed layer would
be a general phenomenon regardless of monomer–substrate
interactions,
20,63,73–79
the present study brings new focus to the
effects of the adsorbed layer on the crystallization and dewet-
ting processes under nanoconned geometries. Also, the
present study demonstrates that the CO
2
solvent annealing can
be used as a low-temperature, environmentally green, and more
effective process to produce controlled crystalline morphologies
with long-range order in place of conventional high-tempera-
ture thermal annealing.
Acknowledgements
We acknowledge Dongcui Li and the Bio-Imaging Center at the
Delaware Biotechnology Institute for some of the AFM experi-
ments. We also thank Steve Bennett for the XR measurements.
T. K. acknowledges partial nancial support from a NSF grant
(CMMI-084626) and from Kuraray Co., Ltd, Japan. The work by
M. F. was supported by the US Department of Energy, Office of
Basic Energy Sciences, Division of Materials Sciences and
Engineering under contract no. DE-AC02-98CH10886. Use of
the National Synchrotron Light Source was supported by the US
Department of Energy, Office of Science, Office of Basic Energy
Sciences, under contracts no. DE-AC02-98CH10886. Some of the
GID experiments were performed at the BL03XU in the SPring-8
with the approval of JASRI (proposal no. 2013A7206).
References
1 A. N. Wilkinson and A. J. Ryan, Polymer Processing and
Structure Development, Springer, 1998.
2 J. Nelson, Mater. Today, 2011, 14, 462–470.
3 H. Klauk, Organic Electronics: Materials, Manufacturing and
Applications Wiley-VCH, 2006.
4 R. H. Friend, R. W. Gymer, A. B. Holmes, J. H. Burroughes,
R. N. Marks, C. Taliani, D. D. C. Bradley, D. A. D. Santos,
J. L. Br´
edas, M. L¨
ogdlund and W. R. Salaneck, Nature,
1999, 397, 121–128.
5 C. W. Frank, V. Rao, M. M. Despotopoulou, R. F. W. Pease,
W. D. Hinsberg, R. D. Miller and J. F. Rabolt, Science, 1996,
273, 912–915.
6 M. M. Despotopoulou, C. W. Frank, R. D. Miller and
J. F. Rabolt, Macromolecules, 1996, 29, 5797–5804.
7 G. Reiter and J. U. Sommer, J. Chem. Phys., 2000, 112, 4376–
4383.
8 H. Schonherr and C. W. Frank, Macromolecules, 2003, 36,
1188–1198.
9 K. Cho, D. Kim and S. Yoon, Macromolecules, 2003, 36, 7652–
7660.
10 Y. Wang, S. Ge, M. H. Rafailovich, J. C. Sokolov, Y. Zou,
H. Ade, J. L. Ning, A. Lustiger and G. Maron,
Macromolecules, 2004, 37, 3319–3327.
11 Y. Wang, M. Rafailovich, J. Sokolov, D. Gersappe, T. Araki,
Y. Zou, A. D. L. Kilcoyne, H. Ade, G. Marom and
A. Lustiger, Phys. Rev. Lett., 2006, 96, 028303.
12 Y. Wang, C.-M. Chan, K.-M. Ng and L. Li, Macromolecules,
2008, 41, 2548–2553.
13 R. D. Priestley, C. Ellison, L. J. Broadbelt and J. M. Torkelson,
Science, 2005, 309, 456–459.
14 P. C. Jukes, A. Das, M. Durell, D. Trolley, A. M. Higgins,
M. Geoghegan, J. E. Macdonald, R. A. L. Jones, S. Brown
and P. Thompson, Macromolecules, 2005, 38, 2315–2320.
15 C. J. Ellison and J. M. Torkelson, Nat. Mater., 2003, 2, 695–
700.
16 Z. Fakhraai and J. A. Forrest, Science, 2008, 319, 600–604.
17 Z. Yang, Y. Fujii, F. K. Lee, C.-H. Lam and O. Tsui, Science,
2010, 328, 1676–1679.
18 H. Bodiguel and C. Fretigny, Phys. Rev. Lett., 2006, 97,
266105.
19 T. Koga, C. Li, M. K. Endoh, L. Koo, M. H. Rafailovich,
S. Narayanan, D. R. Lee, L. Lurio and S. K. Sinha, Phys.
Rev. Lett., 2010, 104, 066101.
20 S. Napolitano, D. Prevosto, M. Lucchesi, P. Pingue,
M. D'Acunto and P. Rolla, Langmuir, 2007, 23, 2103–2109.
21 Q. Lan, J. Yu, J. Zhang and J. He, Macromolecules, 2011, 44,
5743–5749.
22 M. Asada, N. Jiang, L. Sendogdular, P. Gin, Y. Wang,
M. K. Endoh, T. Koga, M. Fukuto, D. Schultz, M. Lee, X. Li,
J. Wang, M. Kikuchi and A. Takahara, Macromolecules,
2012, 45, 7098–7106.
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6401
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
23 B. Vanroy, M. Wubbenhorst and S. Napolitano, ACS Macro
Lett., 2013, 2, 168–172.
24 W. L. Wu, W. E. Wallace, J. H. van Zanten, B. J. Bauer,
D. W. Liu and A. Wong, Polymer, 1997, 38, 2583.
25 R. G. Wissinger and M. E. Paulaitis, J. Polym. Sci., Polym.
Phys. Ed., 1987, 25, 2497–2510.
26 S. K. Goel and E. J. Beckman, Polymer, 1993, 34, 1410–1417.
27 M. A. McHugh and V. Krukonis, Supercritical Fluids
Extraction Principles and Practice, Woburn, MA, 1994.
28 P. D. Condo, D. R. Paul and K. P. Johnston, Macromolecules,
1994, 27, 365–371.
29 T. Koga, Y. S. Seo, Y. Zhang, K. Sin, K. Kusano, K. Nishikawa,
M. H. Rafailovich, J. C. Sokolov, B. Chu, D. Peiffer,
R. Occhiogrosso and S. K. Satija, Phys. Rev. Lett., 2002, 89,
125506.
30 T. Koga, Y. S. Seo, X. Hu, S. Kwanwoo, Y. Zhang,
M. H. Rafailovich, J. C. Sokolov, B. Chu and S. K. Satija,
Europhys. Lett., 2002, 60, 559–565.
31 S. M. Sirard, K. J. Ziegler, L. C. Sanchez, P. F. Green and
K. P. Johnston, Macromolecules, 2002, 35, 1928–1935.
32 T. Koga, Y. S. Seo, K. Shin, Y. Zhang, M. H. Rafailovich,
J. C. Sokolov, B. Chu, D. Peiffeer and S. K. Satija,
Macromolecules, 2003, 36, 5236–5243.
33 T. Koga, Y. Ji, Y. S. Seo, M. H. Rafailovich, J. C. Sokolov and
S. K. Satija, J. Polym. Sci., Part B: Polym. Phys., 2004, 42, 3282–
3289.
34 Y. Li, E. park, K. Lim, K. P. Johnston and P. F. Green, J. Polym.
Sci., Part B: Polym. Phys., 2007, 45, 1313–1324.
35 X. X. Li and B. D. Vogt, Polymer, 2009, 50, 4182–4188.
36 T. Koga, P. Gin, H. Yamaguchi, M. Endoh, L. Sendogdular,
M. Kobayashi, A. Takahara, B. Akgun, S. K. Satija and
T. Sumi, Polymer, 2011, 52, 4331–4336.
37 A. Mendoza-Galvan, C. Trejo-Cruz, O. Solis-Canto and
G. Luna-Barcenas, J. Supercrit. Fluids, 2012, 64,25–31.
38 L. A. Madsen, Macromolecules, 2006, 39, 1483–1487.
39 E. Weidner and V. Wiesmet, J. Supercrit. Fluids, 1997, 10,
139–147.
40 K. Shin, X. Hu, X. Zheng, M. H. Rafailovich, J. Sokolov,
V. Zaitsev and S. A. Schwarz, Macromolecules, 2001, 34,
4993–4998.
41 D. Y. Kwok and A. W. Neumann, Colloids Surf., A, 1999, 161,
31–48.
42 S. M. Sirard, P. R. Gupta, T. P. Russell, J. J. Watkins,
P. F. Green and K. P. Johnson, Macromolecules, 2003, 36,
3365–3373.
43 F. H. Huang, M. H. Li, L. L. Lee, K. E. Starling and
F. T. H. Chung, J. Chem. Eng. Jpn., 1985, 18, 490–496.
44 T. P. Russell, Mater. Sci. Rep., 1990, 5, 171–271.
45 G. K. Fleming and W. J. Koros, Macromolecules, 1986, 19,
2285–2291.
46 M. L. O'Neill, Q. Cao, R. Fang, K. P. Johnston,
S. P. Wilkinson, C. D. Smith, J. L. Kerschner and
S. H. Jureller, Ind. Eng. Chem. Res., 1998, 37, 3067–3079.
47 S. K. Goel and E. J. Beckman, Polym. Eng. Sci., 1994, 34, 1148–
1156.
48 S. K. Goel and E. J. Beckman, Polym. Eng. Sci., 1994, 34, 1137–
1147.
49 P. Gin, M. Asada, M. K. Endoh, C. Gedelian, T.-M. Lu and
T. Koga, Appl. Phys. Lett., 2009, 94, 121908.
50 S. Siripurapu, J. M. DeSimone, S. A. Khan and R. J. Spontak,
Macromolecules, 2005, 38, 2271–2280.
51 H. Yamaguchi, P. Gin, H. Arita, M. Kobayashi, S. Bennett,
S. K. Satija, M. Asada, T. Koga and A. Takahara, RSC Adv.,
2013, 3, 4778–4785.
52 O. H. Seeck, I. D. Kaendler, M. Tolan, K. Shin,
M. H. Rafailovich, J. C. Sokolov and R. Kolb, Appl. Phys.
Lett., 2000, 76, 2713–2715.
53 T. Koga, Y. S. Seo, J. Jerome, S. Ge, M. H. Rafailovich,
J. C. Sokolov, B. Chu, O. H. Seeck, M. Tolan and R. Kolb,
Appl. Phys. Lett., 2003, 83, 4309–4311.
54 X. X. Li and B. D. Vogt, J. Supercrit. Fluids, 2009, 51, 256–263.
55 K. Nishikawa, I. Tanaka and Y. Amemiya, J. Phys. Chem.,
1996, 100, 418–421.
56 G. Reiter and J. U. Sommer, Phys. Rev. Lett., 1998, 80, 3771–
3774.
57 J. U. Sommer and G. Reiter, J. Chem. Phys., 2000, 112, 4384–
4393.
58 C. R. Desper and R. S. Stein, J. Appl. Phys., 1966, 37, 3990–
4002.
59 M. V. Massa, K. Dalnoki-Veress and J. A. Forrest, Eur. Phys.
J. E: SoMatter Biol. Phys., 2003, 11, 191–198.
60 H. Schonherr and C. W. Frank, Macromolecules, 2003, 36,
1188–1198.
61 J. L. Keddie, R. A. L. Jones and R. A. Cory, Europhys. Lett.,
1994, 27,59–64.
62 J. A. Forrest, Eur. Phys. J. E: SoMatter Biol. Phys., 2002, 8,
261–266.
63 T. Koga, N. Jiang, P. Gin, M. K. Endoh, S. Narayanan,
L. B. Lurio and S. K. Sinha, Phys. Rev. Lett., 2011, 107, 225901.
64 G. Reither, in SoMatter Characterization, ed. R. Borsali and
R. Pecora, Springer, NY, 2007, pp. 1244–1292.
65 G. Reiter, Phys. Rev. Lett., 1992, 68,75–78.
66 G. Reiter, Langmuir, 1993, 9, 1344–1351.
67 A. Sharma and G. Reiter, J. Colloid Interface Sci., 1996, 178,
383–399.
68 B. C. Okerberg, B. C. Berry, T. R. Garvey, J. F. Douglas,
A. Karim and C. L. Soles, SoMatter, 2009, 5, 562–567.
69 C. Redon, F. Brochardwyart and F. Rondelez, Phys. Rev. Lett.,
1991, 66, 715–718.
70 G. Reiter, P. Auroy and L. Auvray, Macromolecules, 1996, 29,
2150–2157.
71 K. Jacobs, R. Seemann, G. Schatz and S. Herminghaus,
Langmuir, 1998, 14, 4961–4963.
72 G. Reiter and R. Khanna, Langmuir, 2000, 16, 6351–6357.
73 C. J. Durning, B. O'Shaughnessy and U. Sawhney,
Macromolecules, 1999, 32, 6772–6781.
74 S. Napolitano and M. W¨
ubbenhorst, J. Phys. Chem. B, 2007,
111, 9197–9199.
75 S. Napolitano, V. Lupascu and M. Wubbenhorst,
Macromolecules, 2008, 41, 1061–1063.
76 Y. Fujii, Z. H. Yang, J. Leach, H. Atarashi, K. Tanaka and
O. K. C. Tsui, Macromolecules, 2009, 42, 7418.
77 S. Napolitano, A. Pilleri, P. Rolla and M. W¨
ubbenhorst, ACS
Nano, 2010, 4, 841–848.
6402 |Soft Matter,2014,10,6392–6403 This journal is © The Royal Society of Chemistry 2014
Soft Matter Paper
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
78 S. Napolitano and M. W¨
ubbenhorst, Nat. Commun., 2011, 2,
260.
79 P. Gin, N. Jiang, C. Liang, T. Taniguchi, B. Akgun, S. K. Satija,
M. K. Endoh and T. Koga, Phys. Rev. Lett., 2012, 109, 265501.
80 O. Guiselin, Europhys. Lett., 1992, 17, 225–230.
81 Q. Lan, J. Yu, J. Zhang and J. He, Macromolecules, 2011, 44,
5743–5749.
82 B. Vanroy, M. Wubbenhorst and S. Napolitano, ACS Macro
Lett., 2013, 2, 168–172.
83 X. Q. Jia and T. J. McCarthy, Langmuir, 2002, 18, 683–687.
84 W. E. Wallace, J. H. c. Zanten and W. L. Wu, Phys. Rev. E: Stat.
Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 1995, 52,
R3329–R3332.
85 J. H. van Zanten, W. E. Wallace and W. L. Wu, Phys. Rev. E:
Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top., 1996, 53,
R2053–R2056.
86 K. Paeng, R. Richert and M. D. Ediger, SoMatter, 2012, 8,
819–826.
87 Y. Ma, W. Hu and G. Reiter, Macromolecules, 2006, 39, 5159–
5164.
88 F. Zhang, G. G. Baralia, B. Nysten and A. M. Jonas,
Macromolecules, 2011, 44, 7752–7757.
89 J. J. Xu, Y. Ma, W. B. Hu, M. Rehahn and G. Reiter, Nat.
Mater., 2009, 8, 348–353.
90 N. Jiang, J. Shang, X. Di, M. K. Endoh and T. Koga,
Macromolecules, 2014, 47, 2682–2689.
This journal is © The Royal Society of Chemistry 2014 Soft Matter,2014,10,6392–6403 | 6403
Paper Soft Matter
Published on 20 May 2014. Downloaded by CASE WESTERN RESERVE UNIVERSITY on 20/10/2015 16:08:27.
View Article Online
