Macromolecules 2010, 43, 4429–4434 4429

DOI: 10.1021/ma1003248

Confinement Effects on Chain Dynamics and Local role.7,18 Note that in all cases the effects were observed in a
Chain Order in Entangled Polymer Melts range up to or even beyond Rg.
In this Communication, we present first results on the
Salim Ok and Martin Steinhart degree of anisotropy of segmental orientation fluctuations in
entangled and geometrically confined chains in two different
Institut f€
ur Chemie, Universit€
at Osnabr€
uck, Barbarastr. 7, types of well-defined model nanocomposites with weakly
D-46069 Osnabr€ uck, Germany interacting interfaces. We address temperatures far above Tg
and time scales beyond the bulk entanglement time τe,
Anca S-erbescu† providing a molecular-basis view of the discussed pheno-
mena. We use a recently established multiple-quantum (MQ)
Institut f€
ur Makromolekulare Chemie, Universit€
at Freiburg,
Stefan-Meier-Str. 31, D-79104 Freiburg, Germany †Now at: Dow NMR technique,23-26 which probes the time-dependent
Europe GmbH, Horgen, Switzerland degree of local chain orientation via the monomer-averaged
apparent residual dipolar coupling constant, Dres. At a given
Cornelius Franz, Fabi avez,‡ and
an Vaca Ch time and temperature, the experimental intensity at short
pulse-sequence times τDQ depends on the square of this
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Kay Saalw€achter*
quantity; more precisely, it is directly proportional to the
Institut f€
ur Physik-NMR, Martin-Luther-Universit€
at orientation autocorrelation function
Halle-Wittenberg, Betty-Heimann-Str. 7, D-06120 Halle,
Germany ‡Now at: Universidade de Lisboa, Lisbon, Portugal
Downloaded via 217.131.243.156 on December 27, 2022 at 12:30:40 (UTC).

CðtÞ ¼ ÆP2 ðcos θðtÞÞP2 ðcos θð0ÞÞæ ð1Þ

Received February 10, 2010
Revised Manuscript Received April 13, 2010 where θ is the segmental orientation relative to the external
magnetic field. In polymer melts, the decay of C(t) reflects
Introduction. Effects of nanometric confinement on poly- chain motions in the tube-model regimes II-IV,23,26 and its
mer structure and dynamics have been in the focus for more magnitude C(τe)1/2 ∼ 1/Ne reflects the macroscopic elasticity
than 15 years, primarily with the aim of providing the in terms of the number of segments in an entangled strand
physical understanding for nanoscale polymer applications (tube diameter). In networks, C(t f ¥)1/2 ∼ 1/Nc exhibits a
such as in composite materials. Changes in Tg are probably long-time plateau that at high cross-link density is directly
the most researched, but still controversial issue, with clear proportional to the cross-link density.25
and intuitively expected indications of slowed-down seg- Our results indicate significantly more anisotropic fluc-
mental dynamics for the case of strongly absorbing surfaces tuations in all confined systems as compared to the bulk.
but sometimes conflicting trends and a lack of understanding These findings appear related to the “corset effect”, which
of the physical basis for the often decreased Tg close to free or was mainly observed in terms of strong changes of the
nonabsorbing interfaces.1-8 One may expect that potential frequency-dependent T1 relaxation times of spins (1H, 2H, 19F)
changes in Tg, being related to the time scale of segmental in chains under geometric confinement in the nanometer-
relaxation, should directly influence also larger scale relaxa- micrometer range.27-30 Kimmich et al. concluded these also
tions. However, even slow mechanical relaxation around Tg to be due to highly anisotropic fluctuations, yet while in their
and aging processes below Tg appear to bear no simple work the effect appeared homogeneous throughout the
relation to the observed Tg changes,9,10 and the reason may sample, we find qualitative changes of our experimental
be sought in a modified monomer packing persisting over response functions that indicate an inhomogeneous scenario,
length scales much beyond the radius of gyration (Rg), as and we draw first conclusions on the length scale of the
seen in computer simulations.2 phenomenon.
At larger length scales, changes in the whole-chain con- Results and Discussion. As a first model system to study
formation, as again seen in simulations,2,4 have been assessed interface-induced effects on polymer dynamics, we chose the
by scattering techniques,11-15 with most studies concluding preparation of D2O-based miniemulsions31,32 of a commer-
no deviation from Gaussian behavior. Nevertheless, the self- cial poly(dimethylsiloxane), PDMS, Mn = 90 kDa, PD =
concentration is expected to be increased locally since the 1.43, M/Me ≈ 7, Rg ≈ 8 nm. Using dichloromethane as
random walk of a chain close to a neutral interface is organic cosolvent to reduce the viscosity (and evaporating
reflected onto itself. Consequently, a reduced interchain it afterward), we were able to realize average droplet sizes in
entanglement density has been discussed as the reason the range of 200-1000 nm controlled by the surfactant
for enhanced flow of confined chains.15-18 In contrast, a (SDS) content, as confirmed by dynamic light scattering
“sticky” surface or local orientation effects and thus an- (Figure 1a) and cryo-TEM (data not shown). Note that the
isotropic dynamics of the segments close to an interface are PDMS 1H signal at around 0 ppm was in all cases well-
thought to be the reason for enhanced elasticity19 or reduced resolved from the surfactant signals, allowing for a selective
diffusion coefficients.20,21 Interestingly, a slowdown of dif- measurement of the PDMS component. No significant
fusion has also been reported for chains close to a free interpenetration and no absorptive interaction between the
surface,22 in apparent contrast to the (sometimes only tran- alkyl moieties of the surfactant, and the PDMS is expected.
siently) reduced viscosity observed by others.15,17,18 Anneal- Importantly, in order to prevent rotational diffusion of the
ing and nonequilibrium effects may thus play an important PDMS spheres from interfering with polymer dynamics on
the same time scale, we tested different schemes of immo-
*To whom correspondence should be addressed. E-mail: kay. bilization. While embedding in rigid poly(vinyl alcohol)
[email protected]. after mixing with an aqueous solution and subsequent
r 2010 American Chemical Society Published on Web 04/23/2010 pubs.acs.org/Macromolecules
4430 Macromolecules, Vol. 43, No. 10, 2010 Communication

reflects the magnitude of anisotropic constraints imposed by

topological constraints (entanglements) and relaxation due
to motions on a time scale beyond τe.23,24,26
The nDQ build-up data plotted in Figure 1b immediately
demonstrate that the chains in the smaller spheres are
significantly constrained. Since a close-form analytical ex-
pression does not (yet) exist for a fit, we performed a scaling
analysis proposed by Graf et al.,23,26 which is based on the
notion that in the short-time regime, C(τDQ)  InDQ(τDQ)/
τDQ2. Combining data taken at different temperatures, and
shifting them using time-temperature superposition (TTS),
based on the known temperature dependence of τe, we obtain
an approximate C(t/τe), plotted in Figure 1c. We notice an
initial power-law decay with an exponent of ∼0.5, which
indicates that the entanglement constraints are not yet fully
developed at M/Me ≈ 7, in agreement with our recent
results.26 The limiting slope in the Doi-Edwards regime II
would be 0.25.26,35 The increasing slope toward longer times
reflects the terminal relaxation, which appears to be un-
changed in the differently confined systems. The only sig-
nificant change is the overall upward √ shift of C(t/τe) by a
factor of about 3, corresponding to a 3 ≈ 1.7-fold increase
of the constraint-induced order.
It may be expected that this increase is not homogeneous
across the melt droplet, yet the featureless nDQ data in these
samples did not allow for any further conclusions. However,
another very relevant observation concerns the separable
signal tail of Iref, which has been attributed to unentangled,
isotropically moving chain ends. It presumably arises from
contour-length fluctuation effects,36 and in a bulk melt, it
contributes to a decrease in elasticity by diluting the en-
tangled fraction. Figure 2a shows sample fits of this tail
fraction, and Figure 2b demonstrates its temperature depen-
dence in all samples. In the bulk melt, it increases with
temperature up to a range of 30%, which then roughly
corresponds to the 2/7 fraction expected for the two outer-
most chain segments with lengths corresponding to Me of a
7-fold entangled polymer chain. At lower temperatures, the
chain dynamics is not fast enough to render the full Me outer
part isotropic on the NMR time scale.
Notably, this tail fraction decreases substantially in the
confined samples, indicating that now even the chain ends
experience anisotropic constraints; in other words, contour-
length fluctuations appear to be suppressed. This may be
Figure 1. (a) Particle size distribution of PDMS miniemulsion droplets rationalized by a (rather long-ranged) geometrically wall-
in D2O, determined by dynamic light scattering. The redissolution result induced anisotropy acting on top of the tube constraint. For
after solidification of the continuous phase with gelatin confirms the a rough estimate of the range of such an effect, we can assume
integrity of the spheres. (b) Normalized DQ build-up curves for PDMS
in the bulk and confined to the different droplet sizes. (c) Orientation that the center parts of the droplets are bulklike and that the
autocorrelation function C(t/τe) obtained after scaling analysis23,26 of confined parts do not show isotropic chain-end signals at all.
temperature-dependent DQ build-up data, referenced to τe(T) taken Under this crude assumption, one can estimate the amount
from the literature.33,34 of interphase-related polymer by scaling the bulk melt chain
end fractions to fit the confined data (dashed lines in
evaporation did not preserve the integrity of the spheres, Figure 2b). Assuming that the confined fraction is located
solidification of the water phase using 20 wt % gelatin in a spherical outer shell of the droplets, the thickness of this
worked well. Heating the gelled suspensions to about 50 C shell is estimated to be 39 ( 8 nm, which is roughly 5 times the
and diluting in excess water confirmed the unchanged par- radius of gyration of the used PDMS.
ticle size distribution (see also Figure 1a). The quantitative discussion of the results so far is subject
Normalized DQ build-up data were obtained following to limitations, which are related to the use of a commercial
published procedures.24,25 The MQ experiment yields two PDMS with a non-negligible polydispersity of 1.4. Because
τDQ-dependent signal functions, IDQ and Iref, where the latter of a rather large amount of material needed in the miniemul-
is a decaying function representing at long times 50% of the sion process, we did not attempt using more expensive
overall main chain signal that does not evolve into DQ monodisperse samples. Considering the width of the droplet
coherences. It also includes signals from less constrained seg- size distributions (Figure 1a), and the general disadvantages
ments such as chain ends, which are quantitatively identified as of PDMS with its rather large Me and its crystallization
slowly relaxing singly exponential tails, often best in a repre- temperature in the 200 K range limiting the study of slower
sentation of Iref - IDQ. After subtraction, the build-up rate of dynamics, we now turn to a better defined model system,
the normalized DQ intensity InDQ = IDQ/(IDQ þ Iref - tail) namely the nanochannels in self-ordered anodic aluminum
Communication Macromolecules, Vol. 43, No. 10, 2010 4431

Figure 2. (a) Exponential signal tails related to isotropically mobile

chain ends, as fitted to Iref - IDQ for different PDMS samples at 280 K.
(b) Chain-end fractions as a function of temperature for all PDMS
samples. The dashed lines are interpolations of the bulk melt data,
scaled down by factors of 0.8, 0.45, and 0.3 to match the confined-
sample data.
oxide (AAO).37 This system provides a hard-wall confine-
ment imposed by hydroxyl-terminated surfaces saturated
with water to different degrees depending on the applied
annealing protocols. AAO can be infiltrated with unpolar Figure 3. (a) Scanning electron micrographs of AAO. Left: bottom of a
polymers in the melt state, forming contiguous polymer 100 μm thick self-ordered AAO membrane after infiltration with PB.
nanorods38 with rather neutral, noninteracting interfaces. The pore bottoms were opened by acid treatment, and the sample was
This is important, as infiltration with (locally) polar poly- coated with a thin gold layer. The complete filling of the pores with PB is
mers such as PDMS39 or poly(ethylene oxide) would lead obvious. Inset: cross section of a self-ordered AAO membrane prior to
infiltration. Right: cross section of a disordered AAO filter membrane
to nanometer-thick rigid absorption layers40-42 and thus (Whatman Anodisc). (b) Infiltration kinetics at 298 K obtained from
network-like constraints for the chains protruding from it.43 simple mass uptake measurements on two different samples, inter-
Nanoporous self-ordered AAO membranes, 100 μm in rupted by removing and later adding again the bulk melt overlayer.
thickness, with pore diameters of 60 and 20 nm were infil- (c) Normalized DQ build-up curves obtained at 303 K for samples after
trated with monodisperse PB of Mn = 87 kDa, PD = 1.05, different infiltration times.
M/Me ≈ 47, and Rg ≈ 11 nm. The time scale was estimated by
simple mass uptake measurements (Figure 3b), which show These results highlight the potential drawbacks inherent to
that about 4 days are needed for complete infiltration. the use of disordered AAO membranes in investigations of
During this time, the samples exhibit a rather complex confined polymer dynamics. In particular, in commercially
NMR behavior, as shown in Figure 3c. At low filling, the available AAO filter membranes, often only short pore
nDQ curves feature a quick initial rise, indicating a strongly segments actually exhibit the nominal pore diameter (see
confined fraction, which vanishes after more than 1 day of also Figure 3a). Moreover, without thorough identification
infiltration. In particular the response after only 0.5 h of of infiltration conditions yielding pores completely filled
infiltration suggests a superposition of signals from different with equilibrated polymer melts, in particular in the presence
microenvironments. A similar, allegedly layer-like behavior of a volatile solvent, ill-defined and nonequilibrated states
was in fact observed using commercially available AAO filter characterized by partial filling of the pore volume and a
membranes (Whatman Anopore) partially filled with PDMS complex spatial distribution of the polymer may result. Some
by solution infiltration.44,45 Consistent with the mass uptake previous NMR results obtained on such samples should thus
curve seen in Figure 3b, we observed an equilibrated res- be handled with care,44-46 also considering that the PDMS
ponse only after extended melt infiltration in excess of 1 day. used in these studies is known to bind quite strongly to
Note that samples were repeatedly measured over periods of hydroxyl-terminated surfaces.39
months after preparation, showing reproducible results. Finally, we discuss the MQ NMR results obtained for PB
Initial nonequilibrium chain conformations are thus as- confined to 20 and 60 nm wide nanochannels plotted in
sumed to be well-annealed. Figure 4. Because of the high number of entanglements per
4432 Macromolecules, Vol. 43, No. 10, 2010 Communication

indicating 15 ( 2 and 45 ( 5% constrained material with

Dres/2π ≈ 170 ( 10 and 260 ( 8 Hz for the 60 and 20 nm
pores, respectively. The fit is somewhat inferior in the 20 nm
pore, which is likely due to the existence of a more pro-
nounced property gradient. The given intervals indicate the
(weak) variation with temperature. The two fractions corres-
pond to the same surface layer thickness of about 2.5 nm,
which is now significantly less than Rg of this polymer.
In order to reconcile the different observations for PDMS
and PB, we need to keep in mind that the estimates of the
constrained fraction were based on different observations,
respectively, i.e., missing isotropic chain-end parts (which
were not significant for highly entangled PB) and the amount
of polymer involved in significantly anisotropic segmental
motion (which is too severely affected by terminal relaxation
Figure 4. Normalized DQ build-up curves for PB in bulk and infil- in lowly entangled PDMS). More conclusive results will be
trated into the nanochannels of self-ordered AAO at different tempera- drawn from our ongoing study of a larger molecular-weight
tures. The solid line for 263 K is a fit demonstrating homogeneous range and from an in-depth analysis of the experimen-
entanglement constraints, and the dashed lines at higher temperatures tal data. For now, we hypothesize that the length scale
are two-parameter fits yielding about 15 and 45% constrained network- of geometrically induced anisotropy of segmental motion
like material in the 60 and 20 nm pores, respectively.
might be related to the entanglement separation (Me) rather
than the radius of gyration, explaining the rather large length
chain (M/Me ≈ 47), the isotropic end fraction is always <5% scale of the effect observed for PDMS in the nanodroplets.
in these samples, and the determination of its content is This also makes sense from a more fundamental perspective,
subject to errors, which is why we refrain from a closer as the phenomena that we study (degree of anisotropy of the
discussion at this point. Detailed studies are underway on dynamics in Doi-Edwards regimes II and III) are governed
samples with lower M. At the lowest temperature of 263 K, by the entanglement density. Fast local motions, over which
which is about 90 K above Tg (just as for the lowest tem- our NMR observable is averaged, are effectively restricted to
perature of our PDMS samples), we observe only little a spatial domain governed by the entanglements, possibly
confinement effects (the 20 nm sample exhibits a somewhat limiting effects of geometric wall-induced anisotropy to this
quicker initial rise), which we explain with the dominance of length scale.
the much stronger entanglement-induced anisotropy to A direct comparison of our observations with the “corset
chain motion. The response at 263 K is almost network-like effect”27-30 observed in terms of strong changes in frequency-
as expected in the rubbery plateau (Doi-Edwards regime II), dependent NMR T1 relaxation times remains difficult as
characterized by an intensity maximum close to 50%, and long as there is no molecular model predicting confinement-
can be fitted to an apparent residual coupling on the order induced shape changes of the dipolar (orientation) autocor-
of Dres/2π ≈ 180 Hz, which corresponds to a local order relation function C(t). Our results constitute a direct obser-
consistent with the entanglement spacing of this polymer.26 vation of the increase of the unaveraged part of the dipolar
At larger temperatures, local order in the bulk PB melt is interaction tensor, corresponding to an increased (yet still
reduced due to reptation, and the comparably quicker rise of small) absolute value of C(t) at a time scale in the reptation
the nDQ buildup of the AAO-confined PB directly reflects regime of the bulk polymer (see ref 26 for details), while T1 is
additional interface-induced anisotropy. also sensitive to faster motions down to the segmental level,
We observe qualitative differences to the soft-confined reflecting the fluctuating part of the dipolar tensor, i.e., the
PDMS spheres, where the terminal dynamics appeared un- decay of C(t) toward the range that MQ NMR is sensitive to.
affected, leading to a monotonic decay of C(t) (Figure 1c). The two methods are therefore complementary, and field-
Here, we were not able to obtain satisfactory representations cycling T1 investigations of our samples have thus been
of C(t/τe) using TTS, which we attribute to a more pro- initiated. We note that the corset effect was reported to be
nounced inhomogeneity of the response, affecting the shape homogeneous across the investigated samples, as evidenced
of the InDQ(τDQ) data at very early times, thus invalidating by monoexponential T1 relaxation, while we interpret the
the InDQ(τDQ) ∼ C(τDQ)τDQ2 approximation. Further, the qualitative shape changes of our InDQ(τDQ) data in terms of
response in the 20 nm wide nanochannels is found to be only an inhomogeneous scenario.
weakly temperature-dependent, indicating a network-like Further, a recent inelastic neutron scattering study47
plateau in C(t/τe). At least for the associated confined focusing on fast segmental motions of poly(ethylene oxide)
fraction, the terminal relaxation does apparently not lead in self-ordered AAO channels of 40 nm diameter concluded
to isotropic chain motion in the long time limit, which may be no significant confinement effects on translational diffusion
explained by the different confinement geometry (straight (apart from the expected immobilized absorption layer),
long channel vs isotropic sphere). which, however, as pointed out in a recent rebuttal by
A semiquantitative analysis is possible if we again make a Kimmich and Fatkullin,30 cannot be taken as evidence for
simplifying assumption for the polymer in the surface layer. the absence of a “corset effect” affecting rotational dyna-
Here, we assume that the nDQ curves in confinement are mics. Finally, the effect observed by us is significantly more
superpositions of unchanged bulk-like material (for which pronounced than orientation effects described by Deloche
we numerically interpolate the corresponding bulk data) and and co-workers,48 who did not detect any changes in mode-
of a network-like component with a given residual coupling rately entangled PDMS films above 25 nm thickness. How-
constant and no long-time relaxation (corresponding to ever, their effect in films below 25 nm could be quantitatively
an nDQ curve that reaches the long-time plateau at 50% attributed to immediately geometrically induced segmental
intensity, such as the 263 K data in Figure 4). The dashed anisotropy in combination with diffusive averaging, and our
lines for the higher temperatures are two-parameter fits preliminary T2 relaxation experiments on macroscopically
Communication Macromolecules, Vol. 43, No. 10, 2010 4433

ordered membrane stacks confirmed that the observed ani- AAO, broad signal components related to rigid surface
sotropy is orientation-dependent, suggesting a similar origin hydroxyl groups and H-bonded water was suppressed by a
of the phenomena. MAPE dipolar filter52 of 200 μs length prior the applying the
Conclusions. We have found that geometric confinement MQ pulse sequence.
leads to significantly more anisotropic chain fluctuations
than predicted by the tube model on time scales beyond Acknowledgment. We are indebted to K. Landfester for her
the entanglement time. In spherical confinement of PDMS initial help with the preparation of the miniemulsions, to
melt droplets (M/Me ≈ 7), the average chain orientation R. Thomann for cryo-TEM investigations, and to S. Kallaus
is measurably increased, while the terminal dynamics and to K. Sklarek for the preparation of AAO. Constructive
(isotropization) appears unaffected, and the thickness of discussions with N. Fatkullin and R. Kimmich are gratefully
the confined material layer is estimated to about 40 nm, acknowledged. Funding was provided by the DFG (SA982/4 and
based on decreased isotropic chain-end signal. In contrast, STE1127/9) in the priority programme SPP1369 “Polymer-Solid
cylindrical confinement of highly entangled PB (M/Me ≈ 50) Contacts” as well as initially via the SFB428 (Freiburg).
leads to an almost temperature-independent response indi-
cating network-like segmental anisotropy in an estimated References and Notes
∼3 nm surface layer. The differences may be due to the
(1) Keddie, J. L.; Jones, R. A. L.; Cory, R. A. Size-Dependent
different confining geometries, and the rather different en- Depression of the Glass Transition Temperature in Polymer Films.
tanglement spacings of PDMS and PB, suggesting the tube Europhys. Lett. 1994, 27, 59–64.
diameter as the relevant length scale for the phenomenon. (2) Baschnagel, J.; Binder, K. On the Influence of Hard Walls on
While the overall phenomenon of increased segmental Structural Properties in Polymer Glass Simulation. Macromole-
order is in tune with increased elasticity19 and maybe also cules 1995, 28, 6808–6818.
(3) Lin, W.-Y.; Blum, F. D. Segmental Dynamics of Interfacial Poly-
slowed-down diffusion20-22 in some thin-film setups, it is an
(methyl acrylate)-d3 in Composites by Deuterium NMR Spectros-
open question as to how our observations can be reconciled copy. J. Am. Chem. Soc. 2001, 123, 2032–2037.
with the enhanced flow observed under more similar (weakly (4) Starr, F. W.; Schrøder, T. B.; Glotzer, S. C. Molecular Dynamics
interacting interface) conditions.15,17,18 A dominating effect Simulation of a Polymer Melt with a Nanoscopic Particle. Macro-
of nonequilibrated chain conformations that are expected molecules 2002, 35, 4481–4492.
for these latter studies appears plausible. Further studies of (5) Alcoutlabi, M.; McKenna, G. B. Effects of confinement on mate-
rial behaviour at the nanometre size scale. J. Phys.: Condens.
PB in AAO nanochannels over a large molecular-weight
Matter 2005, 17, R461–R524.
range, self-diffusion measurements by pulsed-gradient NMR, (6) Serghei, A.; Huth, H.; Schick, C.; Kremer, F. Glassy Dynamics in
more detailed infiltration experiments based upon confocal Thin Polymer Layers Having a Free Upper Interface. Macromole-
fluorescence microscopy of dye-doped polymers, and experi- cules 2008, 41, 3636–3639.
ments focusing on changes in the interfacial properties by (7) Serghei, A.; Kremer, F. Metastable States of Glassy Dynamics,
surface modification are underway and will be reported Possibly Mimicking Confinement-Effect in Thin Polymer Films.
Macromol. Chem. Phys. 2008, 209, 810–817.
shortly.
(8) Peter, S.; Napolitano, S.; Meyer, H.; W€ubbenhorst, M.; Baschnagel,
Experimental Section. Samples and Characterization. J. Modelling Dielectric Relaxation in Polymer Glass Simulations:
The investigated PDMS base polymer was an alkyl-termi- Dynamics in the Bulk and in Supported Polymer Films. Macromole-
nated PDMS obtained from Wacker Silicones, and the cules 2008, 41, 7729–7743.
narrowly distributed PB (-H and -sec-butyl termination, (9) Rittigstein, P.; Priestley, R. D.; Broadbelt, L. J.; Torkelson, J. M.
about 54% cis, 41% trans, and 5% vinyl units) was pur- Model polymer nanocomposites provide an understanding of confine-
ment effects in real nanocomposites. Nat. Mater. 2007, 6, 278–282.
chased at PSS (Mainz, Germany). Other chemicals were
(10) Fakhraai, Z.; Forrest, J. A. Measuring the Surface Dynamics of
obtained from Aldrich and used without further purifica- Glassy Polymers. Science 2008, 319, 600–604.
tion. The miniemulsions recipes were 1 g of PDMS, 4 g of (11) Jones, R. L.; Kumar, S. K.; Ho, D. L.; Briber, R. M.; Russell, T. P.
dichloromethane, 20 g of D2O, and 0.3-2.5 g of sodium Chain conformation in ultrathin polymer films. Nature 1999, 400,
dodecyl sulfate (SDS). First, a macroemulsion was prepared 146–149.
by vigorous stirring for 1 h, followed by ultrasonication (12) Kraus, J.; M€ uller-Buschbaum, P.; Kuhlmann, T.; Schubert, D. W.;
Stamm, M. Confinement effects on the chain conformation in thin
(2 min at 120 W) using a Bandelin HD2200 with KE76 tip.
polymer films. Europhys. Lett. 2000, 49, 210–216.
Droplet size distributions were measured with a Horiba (13) Br^ulet, A.; Boue, F.; Menelle, A.; Cotton, J. P. Conformation of
LA-920 particle sizer and confirmed by cryo-TEM. The Polystyrene Chain in Ultrathin Films Obtained by Spin Coating.
emulsions were gelled by adding 20 wt % bone gelatin Macromolecules 2001, 101, 4071–4097.
(Merck) at 50 C. AAO membranes were prepared following (14) Sen, S.; Xie, Y.; Kumar, S. K.; Yang, H.; Bansal, A.; Ho, D. L.
the procedures reported in the literature.37,49,50 Before infil- Chain Conformations and Bound-Layer Correlations in Polymer
Nanocomposites. Phys. Rev. Lett. 2007, 98, 128302.
tration, the AAO was dried in vacuum at 250 C. Infiltration
(15) Shin, K.; Obukhov, S.; Chen, J.-T.; Huh, J.; Hwang, Y.; Mok, S.;
of PB into the membranes was performed by placing the PB Dobriyal, P.; Thiyagarajan, P.; Russell, T. P. Enhanged mobility of
on the surface of the AAO at 298 K under vacuum or argon confined polymers. Nat. Mater. 2007, 6, 961–965.
atmosphere to prevent possible oxidation and subsequent (16) Si, L.; Massa, M. V.; Dalnoki-Veress, K.; Brown, H. R.; Jones,
removal of excess PB with a sharp blade. The aluminum R. A. L. Chain Entanglement in Thin Freestanding Polymer Films.
substrate attached to the AAO layer was removed following Phys. Rev. Lett. 2005, 94, 127801.
(17) Rowland, H. D.; King, W. P.; Pethica, J. B.; Cross, G. L. W.
procedures described elsewhere51 prior to crushing and
Molecular Confinement Accelerates Deformation of Entangled
flame-sealing in evacuated NMR sample tubes. Polymers During Squeeze Flow. Science 2008, 322, 720–724.
NMR experiments on the PDMS nanospheres and the PB (18) Barbero, D. R.; Steiner, U. Nonequilibrium Polymer Rheology in
in AAO nanopores were performed on Bruker Avance Spin-Cast Films. Phys. Rev. Lett. 2009, 102, 248303.
500 (B0 = 11.7 T) and Bruker Avance II 400 (B0 = 9.4 T) (19) Hu, H.-W.; Granick, S. Viscoelastic Dynamics of Confined Poly-
spectrometers, respectively, using in both cases the proton mer Melts. Science 1992, 258, 1339–1342.
(20) Frank, B.; Gast, A. P.; Russell, T. P.; Brown, H. R.; Hawker, C.
channel of standard Bruker static double-resonance probes.
Polymer Mobility in Thin Films. Macromolecules 1996, 29, 6531–
MQ experiments using a dedicated multipulse sequence were 6534.
conducted and evaluated following previously published (21) Zheng, X.; Rafailovich, M. H.; Sokolov, J.; Strzhemechny, Y.;
procedures.25,26 For the experiments on PB infiltrated into Schwarz, S. A.; Sauer, B. B.; Rubinstein, M. Long-Range Effects
4434 Macromolecules, Vol. 43, No. 10, 2010 Communication

on Polymer Diffusion Induced by a Bounding Interface. Phys. Rev. (37) Masuda, H.; Fukuda, K. Ordered Metal Nanohole Arrays Made
Lett. 1997, 79, 241–244. by a Two-Step Replication of Honeycomb Structures of Anodic
(22) Pu, Y.; Rafailovich, M. H.; Sokolov, J.; Gersappe, D.; Peterson, T.; Alumina. Science 1995, 268, 1466–1468.
Wu, W.-L.; Schwarz, S. A. Mobility of Polymer Chains Confined at (38) Zhang, M.; Dobriyal, P.; Chen, J.-T.; Russell, T. P.; Olmo, J.;
a Free Surface. Phys. Rev. Lett. 2001, 87, 206101. Merry, A. Wetting Transition in Cylindrical Alumina Nanopores
(23) Graf, R.; Heuer, A.; Spiess, H. W. Chain-Order Effects in Polymer with Polymer Melts. Nano Lett. 2006, 6, 1075–1079.
Melts Probed by 1H Double-Quantum NMR Spectrocopy. Phys. (39) Cohen-Addad, J. P.; Roby, C.; Sauviat, M. Characterization of
Rev. Lett. 1998, 80, 5738–5741. chain binding to filler in silicone-silica systems. Polymer 1985, 26,
(24) Saalw€ achter, K.; Heuer, A. Chain Dynamics in Elastomers as 1231–1233.
Investigated by Proton Multiple-Quantum NMR. Macromolecules (40) Litvinov, V. M.; Zhdanov, A. A. Molecular Motions in Filled Poly-
2006, 39, 3291–3303. dimethylsiloxanes. Vysokomol. Soedin., Ser. A 1987, 29, 1021–1027.
(25) Saalw€ achter, K. Proton Multiple-Quantum NMR for the Study of (41) Kirst, K. U.; Kremer, F.; Litvinov, V. M. Broad-Band Dielectric
Chain Dynamics and Structural Constraints in Polymeric Soft Spectroscopy on the Molecular Dynamics of Bulk and Adsorbed
Materials. Progr. NMR Spectrosc. 2007, 51, 1–35. Poly(dimethylsiloxane). Macromolecules 1993, 26, 975–980.
(26) F. Vaca Ch avez K. Saalw€achter. NMR Observation of Entangled (42) Sato, A.; Knoll, W.; Pennec, Y.; Djafari-Rouhani, B.; Fytas, G.;
Polymer Dynamics: Tube Model Predictions and Constraint Steinhart, M. Anisotropic propagation and confinement of high fre-
Release. Phys. Rev. Lett., in press. quency phonons in nanocomposites. J. Chem. Phys. 2009, 1330, 111102.
(27) Fatkullin, N.; Fischer, E.; Mattea, C.; Beginn, U.; Kimmich, R. (43) S-erbescu, A.; Saalw€achter, K. Particle-induced network formation
Polymer Dynamics under Nanoscopic Constraints: The “Corset in linear PDMS filled with silica. Polymer 2009, 50, 5434–5442.
Effect” as Revealed by NMR Relaxometry and Diffusometry. (44) Primak, S. V.; Jin, T.; Dagger, A. C.; Finotello, D.; Mann, E. K.
ChemPhysChem 2004, 5, 884–894. Chain segment order in ultrathin polymer films: A deuterium
(28) Fatkullin, N.; Kimmich, R.; Fischer, E.; Mattea, C.; Beginn, U. NMR study. Phys. Rev. E 2002, 65, 031804.
The confined-to-bulk dynamics transition of polymer melts in (45) Jagadeesh, B.; Demco, D. E.; Bl€ umich, B. Surface induced order and
nanoscopic pores of solid matrices with varying pore diameter. dynamic heterogeneity in ultra thin polymer films: A 1H multiple-
New J. Phys. 2004, 6, 46. quantum NMR study. Chem. Phys. Lett. 2004, 393, 416–420.
(29) Kausik, R.; Mattea, C.; Fatkullin, N.; Kimmich, R. Confinement (46) Ayalur-Karunakaran, S.; Bl€ umich, B.; Stapf, S. NMR investiga-
effect of chain dynamics in micrometer thick layers of a polymer tions of polymer dynamics in a partially filled porous matrix. Eur.
melt below the critical molecular weight. J. Chem. Phys. 2006, 124, Phys. J. E 2008, 26, 43–53.
114903. (47) Krutyeva, M.; Martin, J.; Arbe, A.; Colmenero, J.; Mijangos, C.;
(30) Kimmich, R.; Fatkullin, N.; . The corset effect, field-cycling NMR Schneider, G. J.; Unruh, T.; Su, Y.; Richter, D. The Role of
relaxometry, transverse NMR relaxation, field-gradient NMR Internal Rotational Barriers in Polymer Melt Chain Dynamics.
diffusometry, and incoherent neutron scattering. J. Chem. Phys. Macromolecules 2002, 35, 4198–4203.
132, in press. (48) Rivillon, S.; Auroy, P.; Deloche, B. Chain Segment Order in
(31) Landfester, K. The Generation of Nanoparticles in Miniemulsions. Polymer Thin Films on a Nonabsorbing Surface: A NMR Study.
Adv. Mater. 2001, 13, 765–768. Phys. Rev. Lett. 2000, 84, 499–502.
(32) Landfester, K. Miniemulsions for Nanoparticle Synthesis. Top. (49) Masuda, H.; Hasegwa, F.; Ono, S. Ordered Metal Nanohole
Curr. Chem. 2003, 227, 75–123. Arrays Made by a Two-Step Replication of Honeycomb Structures
(33) Ferry, J. D. Viscoelastic Properties of Polymers; John Wiley: of Anodic Alumina. J. Electrochem. Soc. 1997, 144, L127–L130.
New York, 1980. (50) Li, A. P.; M€
uller, F.; Birner, A.; Nielsch, K.; G€
osele, U. Hexagonal
(34) Mark, J. E., Ed.; Physical Properties of Polymers Handbook; pore arrays with a 50420 nm interpore distance formed by self-
Springer: New York, 2007. organization in anodic alumina. J. Appl. Phys. 1998, 84, 6023–6026.
(35) Doi, M.; Edwards, S. F. The Theory of Polymer Dynamics; (51) Hohlbein, J.; Steinhart, M.; Schiene-Fischer, C.; Benda, A.; Hof,
Clarendon Press: Oxford, 1986. M.; H€ ubner, C. G. Confined Diffusion in Ordered Nanoporous
(36) Kimmich, R.; K€ opf, M.; Callaghan, P. Components of Transverse Alumina Membranes. Small 2007, 3, 380–385.
NMR Relaxation in Polymer Melts: Influence of Chain-End (52) Mauri, M.; Thomann, Y.; Schneider, H.; Saalw€achter, K. Spin
Dynamics. J. Polym. Sci., Part B: Polym. Phys. 1991, 29, 1025– Diffusion NMR at Low Field for the Study of Multiphase Solids.
1030. Solid State Nucl. Magn. Reson. 2008, 34, 125–141.