Surface segregation in polymer blends driven by surface freezing

Article abstract of DOI:10.1021/ma061266e

The surface freezing-driven surface segregation in polymer blends was investigated by determining the surface composition and surface temperatures in binary mixtures containing two different lengths of side chains. The chemical attachment of alkyl chains to the backbone was found to lead to considerable differences from the surface freezing in binary alkane mixtures. The surface transition and composition were measured using surface-sensitive infrared-visible sum frequency generation (SFG) spectroscopy and surface tension. The SFG spectra at various blend concentrations were measured as a function of temperature to determine the orientation and concentration of the ordered chains on the surface. The results show that the surface segregation of V22 is sufficient to cover the surface with C22 below Ts22.

Full text of DOI:10.1021/ma061266e

Macromolecules 2006, 39, 7467-7470
7467
Surface Segregation in Polymer Blends Driven by
Surface Freezing
Shishir Prasad, Laurie Hanne, and Ali Dhinojwala*
Department of Polymer Science, The UniVersity of Akron,
Akron, Ohio 44325-3909
ReceiVed June 7, 2006
ReVised Manuscript ReceiVed August 8, 2006
Acrylate polymers with hydrophobic side chains exhibit
surface freezing, where a monolayer of ordered side chains exists
on the surface above the bulk melting temperature (Tm).1-3
Surface freezing is an exception in nature where almost all the
materials exhibit the opposite behavior of surface melting.4
Linear alkanes and alkane analogues are the only other materials
that have been shown to exhibit surface freezing.5,6 For acrylate
side chain polymers, the surface order-to-disorder transition (Ts,n)
is first order and takes place 9-20 K above Tm, depending on
the length of the side chains (n). In this Communication, we
have determined the surface composition and surface ordering
temperatures in binary mixtures containing two different lengths
of side chains. (We report here the results for n ) 18/22, and
similar results were obtained for n ) 16/18 and 16/22 blends.)
The chemical attachment of alkyl chains to the backbone leads
to striking differences from the surface freezing in binary alkane
mixtures.7 Above Ts,22, the short disordered side chains are
preferred on the surface. However, below Ts,22 the surfaces
change discontinuously from a complete miscible to a complete
immiscible surface layer. The long ordered side chains com-
pletely cover the surface for bulk composition as small as 2 wt
%. Below this critical concentration, both the components
coexist on the surface in an unmixed state. The surface transition
temperatures for the long side chains are insensitive to the bulk
concentration. The dependence of Ts,22 on bulk concentration
can be explained by an abrupt increase in the number of side
chains/molecule in the ordered state. These results are in contrast
to alkanes of similar chain length differences, where the
transition temperature is a strong function of composition and
remain mixed in the crystal layer.7
Surface transition and composition are measured using
surface-sensitive infrared-visible sum frequency generation
(SFG) spectroscopy and surface tension (γ). SFG involves the
spatial and temporal overlap of a high-intensity visible laser
beam of frequency ωvis with a tunable infrared beam of
frequency ωIR. According to the dipole approximation, the
generation of SFG photons [at (ωvis + ωIR)] is forbidden in
centrosymmetric bulk and permitted only at interfaces where
inversion symmetry is broken. The SFG signal is resonantly
enhanced when IR frequency overlaps with the molecular
vibrational mode (both IR- and Raman-active modes). Thus,
SFG is sensitive to both the composition and orientation of
molecules at the interfaces. Further enhancement by 1-2 orders
of magnitude in signal is achieved when the angle of input
beams is close to the critical angle for total internal reflection
(42° with respect to the surface normal was used to probe the
polymer-air interface).8,9 The temperature measurements were
done at a heating rate of 0.5 K min-1 with a relative accuracy
of 0.025 K and an absolute accuracy of 0.5 K. γ was measured
Figure 1. SFG intensity in SSP polarization at 2880 cm-1 as a function
of temperature for blends with φb of 0.000023 (a), 0.00023 (b), 0.0023
(c), and 0.023 (d). The (0) and (O) are for cooling and heating cycles,
respectively. Also shown are the Ts,22 (vertical dashed lines) and the
Ts,18 (vertical dotted lines). The black and gray lines are for transition
temperature measured for the homopolymers in the heating and cooling
cycles. The inset of (a) shows the chemical structure of alkyl side chain
acrylate polymer of side chain length “n”.
using a thin platinum wire of nominal diameter 0.5 mm (the
actual diameter was calibrated using heptane) attached to a
Wilhelmy Cahn 2000 microbalance.3 The absolute error in γ is
0.5 mN m-1. The temperature was changed in steps of 0.2 K at
0.1 K min-1, and the data were recorded after 10 min to ensure
complete thermal equilibrium.
Poly(octadecyl acrylate) (C18: Tm ) 320.7 K, freezing point
(Tf) ) 317.2 K, and Ts,18 ) 329.8 K) and poly(behenyl acrylate)
(C22: Tm ) 336.5 K, Tf ) 333.5 K, and Ts,22 ) 344.0 K) were
prepared using free radical polymerization.10 The polydispersity
(PD) of these polymers was ≈2 and Mn ≈ 15 000 g/mol. For γ
measurements we have used C18 and C22 from Landec Corp.
with PD of ≈10 and Mn ≈ 15 000 g/mol prepared using the
melt polymerization process. The results were similar for
samples prepared using the two techniques. There are two peaks
observed in the differential calorimeter measurements at tem-
peratures very close to the transition temperatures of pure C18
and C22, which shows that both C18 and C22 phase separate
upon crystallization. Films on sapphire prism for SFG experi-
ments were prepared by solution-casting 6 wt % solution in
toluene and subsequently annealed 10 K above Ts,22 for 4-5 h
under vacuum. Blend samples for γ measurements were
prepared by mixing the required composition of C18 and C22
in a temperature-controlled glass cell and annealing for a week
above Ts,22
.
Figure 1 shows the SFG intensity in SSP polarization (s-
polarized SFG, s-polarized visible, and p-polarized IR) at 2880
cm-1, corresponding to the methyl symmetric stretch (r+) as a
function of temperature during heating and cooling cycles for
blends of C18 and C22. Since the SFG intensity at 2880 cm-1
is proportional to the average orientation and the number density
of the methyl groups, the sharp drop in SFG intensity indicates
an order-to-disorder transition.1 For concentrations of C22 (φb
is defined in weight fraction and is approximately equal to
volume fraction since the densities of C18 and C22 are similar)
greater than 2 wt %, we observe only one sharp transition near
* Corresponding author. E-mail: ali4@uakron.edu.
10.1021/ma061266e CCC: $33.50 © 2006 American Chemical Society
Published on Web 09/27/2006

7468 Communications to the Editor
Macromolecules, Vol. 39, No. 22, 2006
Figure 3. Ts as a function of φb for cooling (O) and heating (0) cycles.
The filled and open symbols represent data obtained from SFG and
surface tension, respectively. The filled triangles are the first transition
temperatures in the heating run near Ts,18. At these concentrations, the
second transition temperature is near Ts,22. We are unable to determine
the first transition using surface tension due to potential complications
associated with crystallization of C22 on the wire below Tm. Shown as
horizontal lines (gray: cooling; black: heating) are the Ts,22 (dashed
lines), bulk transition temperatures of pure C22 (Tf and Tm; solid lines),
and Ts,18 (dotted lines). The inset shows the fit for Ts for the C22 phase
as a function of φb using the model discussed in the text. In addition,
we have included the Ts for the C16/C22 blend (4).
Figure 2. Surface tension as a function of temperature for pure C18
(]), φb of 0.00023 (0), 0.023 (O), and pure C22 (f). The half-filled
and open symbols are for cooling and heating cycles, respectively. Also
shown are the Ts,22 (vertical dashed lines) and the Ts,18 (vertical dotted
lines). The black and gray lines are used to show the surface transition
temperature measured for homopolymers in the heating and cooling
cycles, respectively. The inset shows γ at 344 K as a function of φb.
The solid line is a fit using eq 2.
Ts,22, indicating that the surfaces are covered with C22 below
Ts,22. Below 2 wt %, we observe two sharp transitions in the
heating cycle. This indicates that the C18 and C22 chains are
phase separated in the surface ordered state. The first drop in
SFG intensity (at lower temperature) is due to the disordering
of C18 phase. In this intermediate state there is coexistence of
disordered C18 and ordered C22 phases on the surface. At the
second transition temperature (at higher temperature) the
remaining C22 ordered phase disorder. This is also confirmed
by the changes in SFG spectra that will be discussed later. In
cooling we observe only one apparent transition temperature
with a large undercooling.
γ as a function of temperature shows similar results as SFG
(Figure 2). The sharp change in the slope of γ as a function of
temperature indicates ordering of surface molecules on top of
disordered melt.3 Above 2 wt % of C22 the transition temper-
atures are very similar to that of pure C22. The values of γ at
344 K as a function of φb are shown in the inset of Figure 2.
The solid line is a fit using the modified Butlers equation11
(discussed later). This model also predicts that there is a small
surface excess of C18 chains on the surface above Ts,22. At low
values of φb, the surfaces are covered predominantly with
disordered C18 chains above Ts,22. Below Ts,22 there is a sharp
transition, and the surfaces are completely covered with ordered
C22. Below 2 wt % the concentration of C22 on the surfaces
decreases with decrease in bulk concentration of C22, eventually
resembling that of C18 homopolymer with the transition near
Figure 4. (a) SFG spectra in SSP polarization for φb ) 0.0023 at 329.5
K (O), 341.3 K (4), and 347.7 K (0). Lines are fits to the spectra
using eq 1. (b) φso as a function of φb for broad (O) and narrower (4)
polydisperse samples. The solid line is a guide to the eye. The inset
shows a sketch of the ordered C22 phase (A) on the surface coexisting
with disordered blend (B). It is not necessary that all the side chains in
a molecule get incorporated in the surface layer.
Ts,18
.
The transition temperatures determined from γ and SFG data
in the cooling and heating cycles are shown in Figure 3. The
dotted and dashed lines are the transition temperatures in a
heating and cooling cycle of bulk C18 and C22, respectively.
The solid lines are Tm and Tf of homopolymer C22. The small
differences in the transition temperatures obtained from SFG
and γ are not understood. One explanation could be due to
sensitivity of the cooling transition to nucleation events, and
this may be different for thin films (SFG) and in glass cells
(γ). As we decrease the bulk concentration of C22, the hysteresis
increases very rapidly due to difficulty in nucleating an ordered
C22 phase. This rapid change in hysteresis coincides with the
concentration where the surfaces are no longer completely
covered with C22.
To determine the orientation and concentration of the ordered
chains on the surface, we measured the SFG spectra at various
blend concentrations as a function of temperature. Figure 4a
shows the SFG spectra for φb ) 0.0023 at three temperatures:

Macromolecules, Vol. 39, No. 22, 2006
Communications to the Editor 7469
below Ts,18, between Ts,18 and Ts,22, and above Ts,22. The spectra
below Ts,18 are dominated by a CH3 symmetric stretch at 2880
cm-1 and a CH3 Fermi resonance at 2940 cm-1. The absence
or weak contribution of the CH3 asymmetric peak at 2955-65
cm-1 indicates that the molecules are oriented with the C-C-C
axis parallel to the surface normal.12 To obtain quantitative
information, we have fitted the spectra with the following
Lorentzian equation:1
side chains within the melt. Above Ts, when the C18 and C22
chains are miscible in the bulk and on the surface, we can equate
the chemical potential of C22 (and C18) in the bulk to that on
the surface. The surface tension of the blend above Ts is given
by the following equation (the details are provided as Supporting
Information):
kT
an1
1
γ )
ln
(2)
-γ22
18an1/kT
φbe an /kT + (1 - φb)e-γ
1
[
]
2
Aq
I(SFG)
øeff,NR
+
(1)
∑
|
|
where an1 is the surface area occupied by C22 side chains/
molecule (a ∼ 20.4 Å2) and γ22 (or γ18) is the surface tension
of pure C22 (or C18). A value of n1 ≈ 5 gives a good fit to the
data shown as an inset in Figure 2. We also predict that there
is a small preference to place C18 side chains on the surface
above Ts, and for small values of φb the surface is predominantly
covered with C18 side chains (see Supporting Information).
ωIR - ωq - iΓq
q
where Aq, Γq, and ωq are the strength, damping constant, and
angular frequency of a single resonant vibration, respectively.
øeff,NR is the nonresonant part of the signal. The magnitude of
Aq(r+)/Aq(r-) of CH3 groups is related to the average orientation
of the CH3 groups with respect to the surface normal. The values
of Aq(r+)/Aq(r-) are similar above and below Ts,18, indicating
that the average orientation of the methyl groups has not changed
after the first drop in SFG intensity in the heating cycle
(transition at low temperature). Hence, the drop in SFG intensity
after the transition at low temperature implies that the number
of ordered surface side chains has decreased abruptly. This
clearly indicates the presence of two independent phases on the
surface. One could perhaps argue that the ordered C22 side
chains are mixed with disordered C18 side chains. However,
this is not possible due to the high-energy penalty at the interface
between the ordered C22 and disordered C18. The surface
tension measurements indicate that there is a change in entropy
and energy and hysteresis in the cooling cycle during the surface
ordering transition. These results cannot be explained without
the interactions with neighboring ordered chains and the
existence of independent ordered C22 phase on the surface.
Above Ts,22 both C18 and C22 are disordered as indicated by
weak SFG intensity of the CH3 peaks.
Below Ts, the ordered C22 side chains are phase separated
from the disordered C18 chains, and we have to use the solution
model instead. We can determine the surface transition tem-
perature of the C22 phase as a function of φb after equating the
chemical potential of C22 chain in the ordered phase to the
C22 chain in the bulk liquid.
Ts ) an1(γ22(Ts) - γ(Ts))/(k ln(φb))
(3)
n1 ≈ nT ≈ 100 provides a reasonable fit for the dependence of
Ts,22 on φb (shown as a solid line in the inset in Figure 3). As
a comparison, we have also shown the fit for n1 ) 1, which is
expected for binary alkane blends with large differences in chain
length. The large values of n1 reduces the dependence of Ts,22
on φb, as observed for poly(n-alkyl acrylate) blends.
It is interesting to note that we predict that n1 increases sharply
upon cooling below the surface ordering temperature. This
explains why surface transitions upon heating are sharp even
at such low volume fraction of C22 chains. The order-to-disorder
transition will only involve moving the C22 side chains from
the surface to the near vicinity below the surface. On the other
hand, cooling requires a nucleation event which is more and
more unlikely at such low concentration of C22 chains in the
bulk. This is the reason the hysteresis increases with decrease
in φb. There are currently no theoretical models to predict the
total concentration of C22 chains on the surface for phase-
separated systems. This requires the knowledge of energy
penalties due to grain boundaries. However, we postulate that
the small differences in surface tension multiplied by large n1
may explain the reasons for such high overall concentration of
ordered C22 chains on the surface at such low values of φb.
Since the orientation of the ordered C18 and C22 chains is
similar, we can directly compare the magnitude of Aq above
and below the transition temperature to determine the overall
concentration of C22 chains on the surface. Figure 4b shows
the results for the surface concentration, φso, as a function of
bulk concentration of C22. In addition, we have also included
data points for samples with narrower polydispersity to illustrate
that the surface composition is not influenced by molecular
weight or PD of the polymer chain.
There are many striking differences observed in the blends
of poly(n-alkyl acrylates) in comparison to binary blends of
alkanes that also exhibits surface freezing. First, the transition
temperatures of the longer side chain component is relatively
independent of bulk concentration. In the case of alkanes with
large differences in chain length, the transition temperature
decreases rapidly with decrease in concentration of the longer
chain.7 For the similar chain length differences as the poly(n-
alkyl acrylates) studied here, the alkane blends are miscible in
the solid state. Second, the small differences in the surface
energies of ordered C22 and disordered C18 result in a dramatic
surface segregation of C22 chains on the surface for φb as small
as 2 wt %. Finally, we observe a significant hysteresis in the
cooling cycle that increases with decrease in φb, which is not
observed for blends of linear alkanes.
In summary, we have for the first time studied the surface
segregation in binary blends of polymers that differ in the length
of the side chains. The surface segregation of C22 chains is
driven by surface freezing and adding a small amount of C22
(>2 wt %) is sufficient to cover the surface with C22 below
Ts,22. Below the surface ordering temperature, we predict that
the polymer chains with longer side chains undergo a sharp
transition to a flattened conformation with almost all the side
chains/molecule participating in the surface-ordered phase.
Neutron reflectivity experiments are in progress to directly
confirm these predictions. The side-chain acrylates are used as
smart adhesives, seed coatings, adhesive tapes, nucleating
agents, and bandages.13,14 Blending a small quantity of large
side-chain component offers an unique opportunity to modify
the static and dynamic surface tension for these applications.
Here we present a simple model to explain the differences
between the results for poly(n-alkyl acrylate) blends in com-
parison to that observed for small molecule alkanes. In this
model, the polymer chain with nT number of side chains has an
option to place n1 side chains on the surface and the remaining

7470 Communications to the Editor
Macromolecules, Vol. 39, No. 22, 2006
(7) Sloutskin, E.; Wu, X. Z.; Peterson, T. B.; Gang, O.; Ocko, B. M.;
Sirota, E. B.; Deutsch, M. Phys. ReV. E 2003, 68, 031605/1-031605/
13.
Acknowledgment. We acknowledge funding from NSF
(DMR-0512156) and L.H.’s research experience for under-
graduates (DMR-0353746) and Petroleum Research Fund
(40690-AC7). We also thank Professor Gujrati for helpful
discussions.
(8) Gautam, K. S.; Schwab, A. D.; Dhinojwala, A.; Zhang, D.; Dougal,
S. M.; Yeganeh, M. S. Phys. ReV. Lett. 2000, 85, 3854-3857.
(9) The spectra taken with the same setup have broad wavenumber (fwhm
≈ 20 cm- 1) resolution. For narrow wavenumber (fwhm < 5 cm-1
)
resolution15 we have used a SpectraPro-500i monochromator in front
of the detector.
Supporting Information Available: The bulk and surface
thermodynamic model. This material is available free of charge
via the Internet at http://pubs.acs.org.
(10) Synthesis of monomer was done by reacting alcohol of appropriate
alkyl chain length with acryloyl chloride. Polymerization was
performed in benzene solvent at 333 K with azobis(isobutyronitrile)
(AIBN) as initiator.
References and Notes
(11) ln(φs/φb) ) A(γ - γ22)/kT.16 A is the area occupied by a molecule
on the surface, and γ22 is the surface tension of C22 homopolymer.
(12) Gautam, K. S.; Dhinojwala, A. Macromolecules 2001, 34, 1137-
1139.
(1) Gautam, K. S.; Dhinojwala, A. Phys. ReV. Lett. 2002, 88, 145501/
1-4.
(2) Gautam, K. S.; Kumar, S.; Wermeille, D.; Robinson, D.; Dhinojwala,
A. Phys. ReV. Lett. 2003, 90, 215501/1-4.
(13) Kinning, D. J. J. Adhes. 1997, 60, 249-274.
(14) De Crevoisier, G.; Fabre, P.; Corpart, J.-M.; Leibler, L. Science 1999,
285, 1246-1249.
(15) Harp, G. P.; Rangwalla, H.; Yeganeh, M. S.; Dhinojwala, A. J. Am.
Chem. Soc. 2003, 125, 11283-11290.
(16) Defay, R.; Prigogine, I. Surface Tension and Adsorption; John
Wiley: New York, 1966.
(3) Prasad, S.; Hanne, L.; Dhinojwala, A. Macromolecules 2005, 38,
2541-2543.
(4) Dosch, H. Critical Phenomena at Surfaces & Interfaces, 1st ed.;
Springer-Verlag: Berlin, 1992.
(5) Ocko, B. M.; Wu, X. Z.; Sirota, E. B.; Sinha, S. K.; Gang, O.;
Deutsch, M. Phys. ReV. E 1997, 55, 3164-3182.
(6) Gang, O.; Wu, X. Z.; Ocko, B. M.; Sirota, E. B.; Deutsch, M. Phys.
ReV. E 1998, 58, 6086-6100.
MA061266E