You are on page 1of 10

Perspective

pubs.acs.org/Macromolecules

Polymer/Nanoparticle Interactions: Bridging the Gap


Yogendra Narayan Pandey, George J. Papakonstantopoulos, and Manolis Doxastakis*,

Department of Chemical and Biomolecular Engineering, University of Houston, Houston, Texas 77004, United States

Global Materials Science Division, The Goodyear Tire & Rubber Company, Akron, Ohio 44316, United States

ABSTRACT: Materials created by dispersing nanoparticles in a


polymer matrix strive to meet the promise of enhanced and often
unique properties at a reduced cost. The availability of structure
property relationships and predictive modeling are deemed necessary to
tailor materials according to our needs. However, the road from detailed
information at the atomistic level to macroscopic properties has been
severely segmented due to diverse experimental, theoretical, and
modeling methods employed to study polymerparticle mixtures,
each with their own advantages and limitations. In this Perspective, we
focus on seemingly simple polymernanoparticle mixtures where
nanoparticles are bare or grafted with chains of the same chemical
constitution as the matrix. We present a number of studies that attempt to quantitatively identify where complete miscibility is
achieved. As we discuss, features pertaining to the nanoscale dimensions of particles continue to challenge our fundamental
understanding on polymerparticle interactions. However, through a concerted approach of theory, experiments, and
simulations, recent studies signicantly expand our knowledge on the morphological behavior of these systems. Most
importantly, our discussion demonstrates how new developments bridge knowledge of microscopic interactions with
thermodynamic behavior, an achievement that has far more reaching implications in the area of polymerparticle mixtures.

INTRODUCTION
Polymerparticle mixtures represent an exemplary manifes-
predict such interactions in the absence of experimental input.
We tactically refrain from referring to important developments
that pertain i.e. to particle shape, multicomponent matrices,
tation of the persistent and critical link between structure and
self-assembly, dynamics, metastable states, and other features of
property in synthetic materials oering improved mechanical,
particle mixtures; each of these subjects deserves their own
electrical, barrier, re-retardant, and optical properties. Our
independent survey and is beyond the scope of this Perspective.
knowledge accumulated over many decades of research in
However, despite our focus on the miscibility point, the studies
composite materials continues to be challenged by recent
presented are anticipated to have far more reaching
progress in manipulating structure at the nanoscale.1 Prescrib-
implications on our ability to tailor dispersion and properties
ing design rules to achieve the desired morphology is based on
of the resulting material.


our fundamental understanding of polymerparticle interac-
tions and our ability to predict thermodynamics a priori to the
SIGNIFICANCE OF POLYMERNANOPARTICLE
synthesis of the material. To this extent, molecular theories and
INTERACTIONS
modeling have made large contributions in recent years, and
our purpose herein is to provide selected examples of exciting Molecular theories have made paramount progress in
developments in characterizing such interactions that drive predicting equilibrium phase behavior and macroscopic proper-
macroscopic behavior. In addition, results presented highlight ties of polymer nanocomposites. A diverse range of theoretical
persisting limitations and opportunities for future research. methods that include integral equation theory, density
Several reports in the literature provide comprehensive functional theory, and self-consistent mean-eld theory are
reviews of developments in the area of nanocomposites15 or continuously developed and applied to broaden our knowl-
target selected subjects, such as modeling methods, pertaining edge.1118 As an example, complete phase diagrams, such as
to the interest and expertise of the readers and authors.610 In shown in Figure 1a, are capable of describing the thermody-
this Perspective, we present a selected small number of studies namics of nanoparticles within a polymer matrix as a function
driven by the same aim as our own research endeavorsto of interfacial attraction strength in units of thermal energy.
advance fundamental knowledge and establish predictive routes Explicit in these diagrams are the boundaries of a miscible
that link chemical detail to macroscopic properties. More phase between depletion-governed aggregation and bridging
specically, we report on several recent intriguing ndings, states.
mainly within the past three years, that attempt to address how
polymerparticle interactions change as we transition from the Received: March 1, 2013
colloidal limit (or at surface) to the nanoscale, the impact of Revised: May 14, 2013
such eects on the miscibility point, and nally how can we Published: June 4, 2013

2013 American Chemical Society 5097 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

al., it is imperative to couple predictions with computer


simulations to create necessary benchmark tests for the
structure and properties that are free of unavoidable statistical
mechanical approximations present in theoretical modeling.16
Computer simulations can characterize dispersion, uctua-
tions, and heterogeneities in mechanical properties, provided
that judicious measures of interfacial segmentparticle
interactions are introduced. As an example, Papakonstantopou-
los et al. have used molecular simulations to quantify
mechanical heterogeneities in a polymer nanocomposite.20,21
They explicitly identied regions of high-shear modulus
domains in lled and unlled systems and the morphological
behavior of these mechanical heterogeneities (Figure 1b) in
addition to the distribution of the llers in their systems. A
propensity of high modulus domains was found close to llers
of favorable interactions to the polymer and also between such
llers. When dealing with such systems of even simplied
models of particlepolymer mixtures, adequate sampling
remains a challenge, and application of Monte Carlo strategies
that assist to overcome connectivity limitations is critical for
equilibration purposes.2225
Clearly, the accuracy of theoretical predictions and computer
simulations depends on our ability to quantitatively characterize
interfacial interactions. In fact, even the presence and the extent
of a polymer layer with altered characteristics from the bulk
material continues to be a subject of experimental investigation.
Often contradictory ndings are reported depending on sample
preparation, technique employed, and system studied.2628 We
briey add that the optimum dispersion state for reinforcement
does not necessarily translate to overall improved material
characteristics.1 Most importantly, it was recently shown that
Figure 1. (a) Nanoparticle volume fraction at spinodal phase for the same nanocomposite material spatial dispersions that
separation predicted by PRISM theory for hard spheres in a freely optimize properties are sensitive to state (melt versus glass).29
jointed chain polymer melt as a function of the strength of exponential Such ndings stress our need for modeling that accurately
interfacial attraction. The depletion and bridging induced phase accounts for the overall eective polymerparticle interactions
separated regimes bracket a window of miscibility at intermediate (as they emerge by the interplay of enthalpic and entropic
interfacial cohesion strength. The type of polymer-mediated nano-
particle organization is schematically indicated. Reproduced with
contributions) and provides unequivocal guidance to materials
permission from ref 16. Copyright 2010 Elsevier. (b) Three- design. To support our view, we selectively discuss recent
dimensional representation of domains of high-shear modulus for a progress made on our understanding and modeling of such
solid nanocomposite. Reproduced with permission from ref 20. interactions at the nanoscale and how they impact the
Copyright 2007 the American Physical Society. miscibility state point in a polymerparticle mixture.

A critical part of theoretical methods is the ability to


quantitatively describe interfacial layers surrounding nano-
INTERFACIAL INTERACTIONS AND GRAFTED
NANOPARTICLE SYSTEMS
particles, regions with distinct properties from the bulk polymer Grafting is a common method to alter the interactions between
phase. Kim et al. using contrast-matching small-angle neutron the polymer matrix and the dispersed nanoparticles.1,30 In some
scattering characterized such interfacial layers by measuring cases, controlled grafting can be exploited not only to achieve
partial collective structure factors in stable concentrated ternary miscibility but also to tailor the shape of aggregates formed in
solutions of short-chain polymers, nanoparticles, and solvent.19 the immiscible regime of the phase diagram.3133 Concepts of
It was demonstrated that predictions by the microscopic polymer brushes have long been explored by seminal
polymer reference interaction site model (PRISM) theory were contributions in the eld of polymer science,3446 and the
in quantitative agreement with experimental data. PRISM reader interested in theoretical and simulation aspects is
theory employs a freely jointed chain and an exponential referred to a comprehensive review by Binder and Milchev.47
monomerparticle interfacial attraction as the only adjustable We consciously restrict ourselves to recent ndings on the
parameter to extract information on the overall spatial autophobic eect as it applies in nanoparticle systems.
organization of the mixture. One intriguing prediction made In the special case where grafted chains and polymer matrix
was that in the case of particles of nanoscale size in the miscible are of identical chemistry, no disparity in enthalpic interactions
region, lowering interfacial interactions toward the depletion between segments of these components exists. At intermediate
aggregation boundary should make the composite material grafting density, the phase behavior of the nanoparticle
stier.15,19 In general, theoretical modeling oers explicit polymer mixture is governed by the mixing entropy of free
knowledge of the introduced parameters and approximations chains exploring the additional volume within the grafted layer
made and is often the most versatile and ecient technique to and the elastic free energy cost of stretching for the grafted
rationalize experimental behavior. Nevertheless, echoing Hall et chains. The key parameter is then the ratio between the length
5098 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

of grafted chains (N) and free chains (P): R = N/P.48,49 When


free chains are shorter, they are able to penetrate into the
grafted layer leading to a wet brush. As a result, particles can
be well-dispersed into the matrix. In contrast, for short grafted
chains, free polymer molecules are expelled, the brush becomes
dry, and particles aggregate. The wetting region is bound by
two grafting density limits: a lower * and a higher **. A
theoretical phase diagram for this autophobicity as calculated
for a at surface (or particles with low curvaturecolloidal
limit) is shown in Figure 2a with * scaling as P1/2 and **,
characterizing the boundary of a second-order autophobic
transition, similar to P1/2 or P3/2.44,48
The application of this wet-to-dry transition to control the
dispersion of silica nanoparticles grafted with polystyrene (PS)
chains in a PS matrix was recently studied by Chevigny et
al.4951 A combination of neutron scattering with contrast
variation was employed to extract the conformation of the
grafted brushes in the matrix while X-ray scattering and
microscopy assessed the dispersion state. While it was evident
that tuning the length of the grafted chains provides a unique
tool to tailor dispersion, the exact boundaries of this transition
did not conform to theoretical predictions. Nanoparticles were
well-dispersed for values of R = N/P > 0.24, suggesting that
aggregation required much longer free polymer chains. This
mismatch was attributed to processing kinetics, chains
polydispersity eects, or surface curvature.49
At the nanoparticle limit, the increased curvature of the
grafted particle reduces the crowding of grafted chains,
eectively assisting free molecules to wet the brush. Self-
consistent eld theoretical calculations by Trombly and
Ganesan studied the limit of nanoparticles showing explicit
changes in the width of the grafted polymerfree polymer
interface as a function of R and curvature.13 Calculations
supported that curved systems are less sensitive to grafting
densities than what may be expected from considerations
pertaining to at surfaces. In addition, the matrix molecular
weight has to be increased further than the corresponding at
surface in order to induce a dewetting transition. These
advantages of highly curved nanoparticles were also brought
forward in a molecular dynamics simulation by Smith and
Bedrov, employing a beadspring model for short grafted
chains (N = 10) and free polymer molecules (P = 10140).54
Signicantly longer beadspring models (above the entangle-
ment threshold) were studied by Kalb and co-workers55
employing connectivity-altering algorithms (N = 501, P =
1000) and variable grafting density. Surprisingly, after detailed
comparison to theoretical scaling laws, no signature of
dewetting phenomena was observed, a feature attributed to Figure 2. (a) Theoretical phase diagram in terms of the grafting
the nite size of melt chains studied which displayed density and the length of the chains of the polymer melt P for a
comparable size to the interface width. In addition, ends of constant graft length of N = 200.48 Reproduced with permission from
tethered chains presented a higher probability to reside in ref 48. (b) Phase diagram for 10 nm PS-g-silica nanoparticles dispersed
proximity to the nanoparticle surface compared to earlier in PS matrices as determined by X-ray scattering and transmission
theoretical predictions.55 The latter nding provides an electron microscopy. Reproduced with permission from ref 52. (c)
Phase diagram for suspensions of 10 nm PEG-g-silica nanoparticles in
example of the necessity to complement theoretical predictions polymeric hosts. Black symbols represent well-dispersed systems, red
with accurate numerical calculations. A subsequent study with symbols denote completely phase-separated systems, and orange
similar models (N = 30, 60, P = 10140) supported that while symbols denote partially phase-separated systems. Lines correspond to
the interface for a single nanoparticle does not provide any theoretical predictions for dewetting based on at surfaces.
signature of autophobic dewetting in a regime where such a Reproduced with permission from ref 53.
feature is expected, matrix-induced depletion attractions from
free chains on a pair of nanoparticles dominate the free energy chains. This is even more signicant at low-grafting densities
prole at intermediate separations.56 Thus, while aggregation where it was recently shown that grafted particles can self-
remains entropic in origin, it is essential to consider collective assemble to anisotropic shapes.31 At this limit, favorable
anities to form clusters in the presence of long free polymer particleparticle interactions compete with the polymer brush
5099 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

free energy; redistribution of the brush favors aggregation at the also worth noting that our past work has demonstrated how we
point diametrically opposite to the rst contact. As a result, can apply preferential sampling schemes, focusing computa-
interactions between the particles become eectively aniso- tional eort in proximity to the surface and enhancing further
tropic. Formation of self-assembled structures provides an sampling.24 This technique is particularly benecial when large
additional level of control to the structure and properties of the systems are considered, necessary to avoid nite-size eects due
polymerparticle mixture.29 We believe that a quantitative to periodic boundary conditions.
characterization of such many-body eects by calculations of
free energy changes to form shapes in an explicit polymer
matrix would contribute to our understanding of the miscibility
POLYMER-SEGMENT/PARTICLE INTERACTIONS:
ACCOUNTING FOR CHEMICAL DETAIL
limit and identication of metastables states. Polymer chemistry relies on an ever-growing gallery of chemical
New additional experimental studies continue to supply data repeat units to synthesize materials with desired properties. The
on the thermodynamics of grafted nanoparticles in a polymer majority of the theoretical and modeling results presented
matrix forming complete phase diagrams. Sunday et al. earlier require input or introduce approximations at the level of
employed X-ray scattering and microscopy to study silica polymer-segment/particle interactions. Thus, a question that
nanoparticles grafted with PS chains (PS-g-silica) in a PS pertains to our discussion is whether interfacial phenomena
matrix.52 As shown in Figure 2b, the P/N ratio for certain between nanoparticles and polymers can be predicted without
grafting densities could reach values up to 4.3 before experimental characterization. Toward this aim, the eld of
aggregation was observed. The power laws extracted follow quantum mechanics has made signicant progress in the past
* P0.54 and ** P0.71. A dierent report for decade.65 We anticipate that overcoming challenges associated
poly(ethylene glycol) (PEG) tethered silica nanoparticles with translating results from electronic calculations to polymer/
dispersed in PEG hosts53 presented a phase diagram that particle interactions will be in the center of future modeling
bears some similarities to the data by Sunday et al. Specically, eorts. We refer readers to a recent article that provides an
Srivastava et al. argued that while indeed the P/N ratio can be excellent overview of the challenges faced66 and detail selected
as high as 5 maintaining a well-dispersed state, above a certain important points. Quantum calculations targeting interactions
threshold (required to ensure screening of attractive forces) between small compounds and surfaces predict dierent
dispersion becomes independent of grafting density. However, absolute energies depending on the method employed.
a low particle diameter to size of grafted chains (D/Rg) ratio Nevertheless, it is the relative energies between adsorption
was necessary in order to maximize curvature eects at the sites and the monomer states that are important. Such energies
small particle limit. Furthermore, by summarizing the results need to be reproduced by classical force elds introduced in
from previous studies (Figure 2c), it was argued that the phase molecular simulations. This is veried by performing distance-
diagram is independent of polymer chemistry.53 We note that dependent scans as well as lateral- and orientation-dependent
theoretical modeling has stressed the role of particle size and, in evaluations.66 These tests are critical to ensure that a
addition, new simulations suggest that polydispersity eects can representative ensemble of congurations as a function of a
alter these boundaries.57,58 multitude of order parameters is retrieved by force-eld
Accounting for polymer chemistry remains a daunting task simulations. Figure 3 depicts interaction energies between a
for computer simulations. Studies are limited to a single or a single PS monomer and a gold surface as calculated by density
pair of particles at most, immersed in oligomers.59,60 Ndoro functional theory.67 Johnston and Harmandaris stressed the
studied atomistic models of PS-g-silica (N = P = 20), need to introduce a Morse functional form in classical modeling
conrming geometric eects (higher available volume around to simultaneously capture both distance and orientation
a small nanoparticle) as discussed earlier.61 Peters et al. dependent interaction energies. These potentials were sub-
performed atomistic simulations with silica particles grafted sequently employed in classical molecular dynamics simulations
with linear hydrocarbon chains (N = 9, 17, 35) in explicit to study the properties of short polystyrene chains conned
solvents (P = 10, 24, 48 and a branched solvent),62 and between two gold surfaces.67 It is noted that a building-block
Ghanbari et al. built coarse-grained (CG) models of PS and approach, where interactions of a specic segment extracted by
silica to extend the work of Ndoro to N = 80 and P = 20 electronic calculations are treated in an additive approach to
160.63 The last two studies have shown evidence of free chain model macromolecules on surfaces,6871 remains highly
migration out of the interfacial region which do support a wet- nontrivial. A large challenge in these methods is the interplay
to-dry transition. We clarify though that lengths reported in of enthalpic contributions with nonlocal entropic eects that
these works correspond to repeat units and not Kuhn segments. are dicult to account for in original electronic calcula-
So the chains are shorter than in the systems employed by Kalb tions.66,72 We note, therefore, that translating chemically
and co-workers.55 specic segment interactions to polymeric molecules is an
Brute-force atomistic modeling will remain severely limited area that requires particular development.
within the foreseeable future in addressing truly polymeric Challenges discussed in bridging quantum calculations to
systems. Long relaxation times restrict sampling in the vicinity classical atomistic force elds pertain to methods available to
of the initial state and hinder our ability to quantify systematically derive CG models.73,74 However, signicant
thermodynamics of mixing. However, chemically specic CG progress has already been made, and it is now feasible to build
models provide signicant promise depending on the ability of polymersegment interactions based on input from atomistic
simplied representations to reproduce the target structure simulations with classical force elds. As mentioned earlier, PS-
including interfacial layers. Coupling such models with g-silica nanoparticles in a PS matrix were studied by Ghanbari
advanced simulation methods can expand the range of and co-workers employing such a scheme.63 In their study, Si
molecular weights studied. For example, Spyriouni and co- atoms served as the centers of interaction sites with a polymer
workers modeled bulk CG PS systems up to P = 100064 CG segment and pairwise interaction potentials were
through connectivity-altering Monte Carlo (MC) methods. It is determined through the iterative Boltzmann inversion (IBI)
5100 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

or the COM of the particle is considered. In fact, the anity of


polymer segments to track the curved nanoparticle surface
depends on particle curvature.24 As we will show herein, we can
extract coarse-grain curvature-dependent potentials based on
atomistic simulations with at surfaces that are computationally
less demanding.
Interactions with implicit at surfaces can be computed by
integration of pairwise potentials over the atoms constituting
the solid slab; in the case of Lennard-Jones interactions this
results to 104 (single solid layer) or 93 (multilayer) power
law expressions for the repulsive and attractive part of the
potential.81 However, when the interaction of the COM of a
group of atoms with a at surface is considered, conformational
characteristics alter the shape of the corresponding eective
potential.82 The need to properly account for changes in the
ensemble of congurations of the group of atoms that are now
represented by a single site is also manifested when considering
changes in the resolution of the CG model. Such features in
theoretical modeling of melts in proximity to surfaces were
addressed by the work of de Pablo and co-workers.83,84
Khounlavong et al. suggested a recalculation of the eective
interactions with the nanoparticle which not surprisingly leads
to dierent (but still accurate) eective potentials.77
Figure 3. Adsorption energy of PS monomer as a function of the Given the growing application of chemically specic CG
distance from a gold surface. The density functional theory results models, we decided to independently assess the ability to
(symbols) are compared to the results from the pair potentials (lines) extract polymer-segment/nanoparticle potentials based on
obtained from an optimization procedure using (a) Lennard-Jones and
(b) Morse nonbonded potentials. Reproduced with permission from
atomistic simulations on a at surface. Our motivation was to
ref 67. Copyright 2012 The Royal Society of Chemistry. evaluate if features beyond the total density prole (as
conformations of chains in contact with the surface) are
captured accurately and whether we can eciently predict
method.75 A thorough test of this technique for bulk systems changes with particle curvature. Following the literature, we
appeared in the literature recently.76 Note that in the study by performed bulk simulations with a fully atomistic model of cis-
Ghanbari and co-workers challenges in matching the bare 1,4-polyisoprene (PI)85,86 and employed the IBI method (with
nanoparticlepolymer density persisted with CG polymer pressure corrections as described by Wang et al.87) to extract
segments forming density layers of higher packing than the CG potentials for polymer segments centered along the bond
corresponding results from atomistic simulations.63 Moving connecting consecutive repeat units (interactions were cuto at
back to atomistic representations resulted to a better agreement 1.5 nm).75,8890 All simulations were performed with a 24
to atomistic models, supporting that discrepancies are due to repeat units PI to avoid chain-end eects present for lower
fundamental dierences between an atomistic and the spherical molecular weights.91,92 The next step of our quest included
CG representation of beads in a PS segment. In another recent atomistic molecular dynamics simulations of PI on an
study, Khounlavong et al. determined the polymer-segment/ amorphous silica slab and around a silica particle of radius 2
particle eective potential that acts between the center of mass nm referred to as SIL-2.0 (as in our previous work24).
(COM) of a nanoparticle and a CG polymer bead.77 This Simulations were performed with 400 chains of 24 repeat units)
method presents additional computational advantages due to for the slab (141 032 atoms) and 800 chains of equal length for
reduction to a single interaction site for each solid nano- the single particle system (253 854 atoms) at T = 413 K and P
particle.78 Simulations with the extracted CG representations of = 1 bar (semi-isotropic coupling for slab) using the software
polymer revealed the need for additional correction terms to Gromacs.93 Note that the generation of 50 ns trajectories for
the nanoparticlenanoparticle potential as a side eect of the these large systems (necessary to avoid nite size eects)
CG polymer representation. Similar ndings were reported by requires several months of computational time even with highly
Hong et al. for bare and grafted silica nanoparticles dispersed in ecient software as Gromacs. Subsequently, atomistic
poly(ethylene oxide) oligomers.79 Thus, it is clear that further trajectories were mapped to CG congurations created with
development and testing of the ability of coarse-graining to the fully atomistic potential. Figure 4a presents snapshots of
capture connement eects is needed. For nanoconnement atomistic simulations where chains in contact with the surface
between at surfaces, a test of IBI with polyamide-6,6 between or the particle are depicted based on the CG representation
graphene was presented by Eslami and co-workers.80 While (single site for each repeat unit). The distribution of CG beads
there were some deviations in the total density prole, the as a function of distance from the surface extracted from the
peaks observed in the atomistic simulations were captured by atomistic trajectories is shown in Figure 4b (solid lines)
the simpler CG model. However, the surface was again together with a decomposition to trains, tails, and loops based
represented by a collection of CG beads as in the study of on a distance criterion of 0.4 nm to dene a beadsurface
Ghanbari and co-workers. While retaining interactions sites for contact. The second step of our procedure was to extract a
surface atoms could result to potentials that are curvature- wallCG bead potential via IBI targeting only the total density
independent, this is not necessarily the case when a single prof ile (black line) and maintaining the same CG beadCG
interaction potential as a function of distance from the surface bead and intramolecular polymer interactions determined from
5101 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

Figure 4. (a) Rendered images from atomistic simulations of cis-1,4-PI on silica. Chains in contact with silica are explicitly depicted by spheres
according to the CG scheme utilized. (b) Radial distribution functions of CG beads and their decomposition as a function of distance from the slab
surface. The solid lines represent the distributions obtained from the atomistic trajectories mapped to CG sites, and symbols represent the data from
CG NVT simulations based upon derived eective potentials. (c) CG potentials extracted for dierent radii. For the slab system, IBI and atomistic
simulations were performed while for particles potentials were determined based on eective interactions of polymer segments with the at surface.
(d) Same as (b) with CG simulations of the SIL-2.0 particle system based on an estimate of the particlepolymer bead interaction.

bulk polymer simulations. The resulting numerical potential is COM separation shown for dierent radii in Figure 4c. These
shown in Figure 4c with nal density proles contrasted to CG potentials can now serve to model single nanoparticles
beads proles from atomistic simulations shown in Figure 4b as immersed in CG representations of the polymer matrix and
symbols. A major nding of our study was that subsequent CG extract density proles as performed previously for the at
simulations with the derived potentials reproduced almost surface. Figure 4d shows that our CG model, which accounts
quantitatively the conformational characteristics (trains, tails, for chemical stiness through bonded, angular, and dihedral
and loops) of the polymer layer in contact with the surface. potential terms, is able to capture conformations in proximity
While it remains to be tested whether this is the case for other to the particle in excellent agreement with independent fully
polymers as well, it is particularly encouraging that CG atomistic simulations. Deviations are observed only at short
simulations meet successfully this challenge. We anticipate separations between the particle COM and the polymer beads,
that the quantitative representation of the total density prole an anticipated outcome given that roughness is increasingly
of the melt suces for melt systems; this CG scheme has been important for smaller particles24 and is not accounted for
applied for PS solutions,94 and certainly future research needs explicitly in the single pairwise potential employed.
to focus on blends and copolymers. Finally, we investigated whether such models can capture
Within the main theme of our article we decided to examine features beyond the mean density prole at a specic
further the predictive ability to construct polymer-segment/ separation. As discussed earlier, this is a critical test when
particle eective interactions for curved nanoparticles. An evaluating the performance of classical force elds with respect
explicit per surface site pairwise interaction can be recon- to data from quantum calculations.66 Using our simple contact
structed using the CGslab potential calculated from the IBI criterion, we can examine changes in free energy along the
and the number density of silica sites using a procedure surface by calculating the probability to form a train segment of
outlined by Nielsen et al.82,95 Numerical integration of the per consecutive CG beads of length S, Ptr(S). Simulations with
site interaction over the nanoparticle volume within the cuto beadspring models recently demonstrated that the train size
distance allows the estimation of a single, pairwise, spherical distribution depends on the strength of the interaction between
nanoparticlepolymer CG segment potential as a function of polymer segments and the surface.96 Figure 5a shows that these
5102 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

previously unaccounted for. To support our view, we chose to


present herein studies in recent years that advanced our
knowledge regarding the eect of nanoscale dimensions on
interfacial interactions and complete miscibility in seemingly
simple systems, nanoparticles, grafted or bare, within a
homopolymer matrix.
Control over dispersion was brought forward by a series of
experimental studies that are now able to construct phase
diagrams and probe material structure with unprecedented
detail. Experiments continue to explore and propose structure
property relationships necessary to design improved materi-
als.29,98101 Theoretical modeling continues to oer the most
versatile and ecient route toward providing predictions and
establishing reasoning for experimental ndings. However,
given the underlying approximations, computer simulations
need to continue to test and challenge the accuracy of theory.
Our aim was to present that there is still progress to be made
on quantitatively characterizing the underlying factors that drive
the observed behavior. We have provided herein several
examples on the miscibility limits of grafted-nanoparticles
immersed in a matrix of similar chemical constitution to justify
this point of view.
While we acknowledge this critical role of modeling, it is also
important to recognize the prospect of hierarchical materials
Figure 5. (a) Probability distributions for train segments from CG design. Quantum calculations can now oer detailed
simulations and atomistic simulations. (b) Normalized RMS bound information on the interactions between polymer segments in
layer thickness as a function of particle size relative to polymer Kuhn
segment length. Data on particles immersed in polyethylene melts are
proximity to the surface, even for systems that are yet to be
taken from our previous study.24 realized in the laboratory. It is anticipated that an increasing
number of studies will focus on translating data from such
probabilities depend on curvature as we described in our calculations to classical force elds with several challenges to be
previous atomistic simulations.24 Most importantly, the CG met. Force-eld-based atomistic modeling suces to probe
model developed herein for PI/silica systems appears to ensembles of conformations in a specic environment and
accurately capture the distributions both for the slab and the major progress in systematically derived ecient coarse-grain
SIL-2.0 system. To the best of our knowledge, this is the rst
representations has already been made. It appears that it is now
time where not only mean density proles are matched, but
uctuations on the surface are described faithfully and feasible to employ such specic models to capture the structure
throughout dierent curvatures. As a nal result, Figure 5b of interfacial layers at the nanoscale. An ambitious aim for the
shows that the interfacial layer thickness (normalized to the at future is to systematically develop a library of interaction
surface result) quantied for polyethylene on silica through potentials for specic polymersurface chemistry which will
atomistic simulations reported previously24 presents excellent consolidate the eorts in predictive modeling of polymer
agreement with our new CG PI data when the latter are scaled nanoparticle interactions. This will allow comprehensive tests
with the corresponding Kuhn length (0.82 nm97). These results
on the performance of these models by the community.
provide a rst indication that the scaling of the layer follows a
universal behavior when polymer stiness is accounted for. Futhermore, theoretical modeling will have the opportunity to
Overall, application of systematic CG techniques presents foster on eective interactions extracted and provide
signicant potential to capture curvature eects in polymer predictions for the phase behavior without experimental
nanoparticle interactions. We believe that further research with input. Multiscale modeling has long been conceived as the
dierent polymer chemical architectures, surfaces, many only viable route toward predicting properties of nano-
particles, blends, and copolymers is required to establish the composites;10,102 the recent progress in fundamental under-
capabilities and limitations of these simpler eective potentials
in capturing the behavior of polymerparticle mixtures. standing of polymerparticle interactions is now bridging the

gaps and completing the realization of this ambitious aim.


OUTLOOK
Progress made in polymer synthesis and characterization AUTHOR INFORMATION
methods reveals that we are not far from tailoring dispersion Corresponding Author
and morphology of nanoparticles within a polymer matrix.1 *E-mail: edoxastakis@uh.edu (M.D.).
Despite extensive fundamental knowledge accumulated over
many decades on the thermodynamics of these systems, as we Notes
move to the nanoscale, further challenges appear with features The authors declare no competing nancial interest.
5103 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106
Macromolecules Perspective

Biographies

Manolis Doxastakis received his Diploma in Chemical Engineering


from the National Technical University of Athens in 1996. He received
Yogendra Narayan Pandey is a senior graduate student at the a masters in Materials Science and Ph.D. in Chemical Engineering
department of Chemical and Biomolecular Engineering at the from the University of Patras in 2002. He worked with Professors
Doros N. Theodorou and George Fytas designing and performing
University of Houston. Yogendra defended his PhD in April 2013.
simulations and experiments on polymer melts, blends, and
He worked with Professor Manolis Doxastakis on modeling and copolymers. He moved to the University of Wisconsin in Madison,
simulation of materials, e.g., polymernanoparticle systems, polymer working as a postdoctoral fellow with Professor Juan J. de Pablo in
2003, investigating polymerparticle mixtures and lipid membranes.
thin lms, and zeolites. Earlier, he received Bachelor of Technology (B. Since 2007, he is an assistant professor in Chemical and Biomolecular
Tech.) in Chemical Engineering and Technology from Institute of Engineering at the University of Houston, studying interfacial
phenomena with molecular modeling.


Technology, Banaras Hindu University, Varanasi, India, in 2005.
Before joining graduate school in Fall 2008, he worked as a software
ACKNOWLEDGMENTS
developer (20052008) with leading information technology We acknowledge nancial support by the National Science
companies in India. Foundation under Grant CBET-1067356 and The Goodyear
Tire and Rubber Company.

REFERENCES
(1) Kumar, S. K.; Krishnamoorti, R. Annu. Rev. Chem. Biomol. Eng.
2010, 1, 3758.
(2) Vaia, R. A.; Maguire, J. F. Chem. Mater. 2007, 19, 27362751.
(3) Winey, K. I.; Vaia, R. A. MRS Bull. 2007, 32, 314322.
(4) Zou, H.; Wu, S.; Shen, J. Chem. Rev. 2008, 108, 38933957.
(5) Jancar, J.; Douglas, J.; Starr, F.; Kumar, S.; Cassagnau, P.; Lesser,
A.; Sternstein, S.; Buehler, M. Polymer 2010, 51, 33213343.
(6) Allegra, G.; Raosa, G.; Vacatello, M. Prog. Polym. Sci. 2008, 33,
683731.
(7) Ganesan, V. J. Polym. Sci., Part B: Polym. Phys. 2008, 46, 2666
2671.
(8) Ganesan, V.; Ellison, C. J.; Pryamitsyn, V. Soft Matter 2010, 6,
40104025.
George Papakonstantopoulos received his Diploma in Chemical (9) de Pablo, J. J. Annu. Rev. Phys. Chem. 2011, 62, 555574.
(10) Yan, L.-T.; Xie, X.-M. Prog. Polym. Sci. 2013, 38, 369405.
Engineering from the National Technical University of Athens in
(11) Hall, L. M.; Anderson, B. J.; Zukoski, C. F.; Schweizer, K. S.
2002. He received his PhD in Chemical Engineering in 2007 from the Macromolecules 2009, 42, 84358442.
(12) Jain, S.; Ginzburg, V. V.; Jog, P.; Weinhold, J.; Srivastava, R.;
University of WisconsinMadison. He worked with Professors Juan J.
Chapman, W. G. J. Chem. Phys. 2009, 131, 044908.
de Pablo and Paul F. Nealey in the areas of polymeric nanocomposites (13) Trombly, D. M.; Ganesan, V. J. Chem. Phys. 2010, 133, 154904.
(14) Chakrabarti, R.; Schweizer, K. S. J. Chem. Phys. 2010, 133,
and block copolymer self-assembly using molecular modeling
144905.
techniques. He worked for Arkema Inc. (20072010) in the Analytical (15) Frischknecht, A. L.; McGarrity, E. S.; Mackay, M. E. J. Chem.
Phys. 2010, 132, 204901.
and Systems Research department as a Scientist, mainly investigating
(16) Hall, L. M.; Jayaraman, A.; Schweizer, K. S. Curr. Opin. Solid
the modication of the mechanical behavior of polymeric materials State Mater. Sci. 2010, 14, 3848.
(17) Frischknecht, A. L.; Yethiraj, A. J. Chem. Phys. 2011, 134,
with the aid of molecular modeling and microscopy techniques.
174901.
Currently he works for the Goodyear Tire and Rubber Company as a (18) Vogiatzis, G. G.; Voyiatzis, E.; Theodorou, D. N. Eur. Polym. J.
2011, 47, 699712.
Senior Scientist in the area of Multiscale Modeling of polymeric
(19) Kim, S. Y.; Schweizer, K. S.; Zukoski, C. F. Phys. Rev. Lett. 2011,
composites and polymer blends. 107, 225504.

5104 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106


Macromolecules Perspective

(20) Papakonstantopoulos, G. J.; Doxastakis, M.; Nealey, P. F.; (57) Jayaraman, A.; Schweizer, K. S. Macromolecules 2009, 42, 8423
Barrat, J.-L.; de Pablo, J. J. Phys. Rev. E 2007, 75, 031803. 8434.
(21) Papakonstantopoulos, G. J.; Yoshimoto, K.; Doxastakis, M.; (58) Martin, T. B.; Dodd, P. M.; Jayaraman, A. Phys. Rev. Lett. 2013,
Nealey, P. F.; de Pablo, J. J. Phys. Rev. E 2005, 72, 031801. 110, 018301.
(22) Doxastakis, M.; Chen, Y.-L.; Guzman, O.; de Pablo, J. J. J. Chem. (59) Barbier, D.; Brown, D.; Grillet, A.; Neyertz, S. Macromolecules
Phys. 2004, 120, 93359342. 2004, 37, 46954710.
(23) Doxastakis, M.; Chen, Y.-L.; de Pablo, J. J. J. Chem. Phys. 2005, (60) Brown, D.; Marcadon, V.; Mele, P.; Alberola, N. D.
123, 034901. Macromolecules 2008, 41, 14991511.
(24) Pandey, Y. N.; Doxastakis, M. J. Chem. Phys. 2012, 136, 094901. (61) Ndoro, T. V. M.; Voyiatzis, E.; Ghanbari, A.; Theodorou, D. N.;
(25) Theodorou, D. N. Ind. Eng. Chem. Res. 2010, 49, 30473058. Bohm, M. C.; Muller-Plathe, F. Macromolecules 2011, 44, 23162327.
(26) Robertson, C. G.; Roland, C. M. Rubber Chem. Technol. 2008, (62) Peters, B. L.; Lane, J. M. D.; Ismail, A. E.; Grest, G. S. Langmuir
81, 506522. 2012, 28, 1744317449.
(27) Moll, J.; Kumar, S. K. Macromolecules 2012, 45, 11311135. (63) Ghanbari, A.; Ndoro, T. V. M.; Leroy, F.; Rahimi, M.; Bohm, M.
(28) Jouault, N.; Moll, J. F.; Meng, D.; Windsor, K.; Ramcharan, S.; C.; Muller-Plathe, F. Macromolecules 2012, 45, 572584.
Kearney, C.; Kumar, S. K. ACS Macro Lett. 2013, 2, 371374. (64) Spyriouni, T.; Tzoumanekas, C.; Theodorou, D.; Muller-Plathe,
(29) Maillard, D.; Kumar, S. K.; Fragneaud, B.; Kysar, J. W.; Rungta, F.; Milano, G. Macromolecules 2007, 40, 38763885.
A.; Benicewicz, B. C.; Deng, H.; Brinson, L. C.; Douglas, J. F. Nano (65) Carter, E. A. Science 2008, 321, 800803.
Lett. 2012, 12, 39093914. (66) Herbers, C. R.; Li, C.; van der Vegt, N. F. A. J. Comput. Chem.
(30) Krishnamoorti, R. MRS Bull. 2007, 32, 341347. 2013.
(31) Akcora, P.; Liu, H.; Kumar, S. K.; Moll, J.; Li, Y.; Benicewicz, B. (67) Johnston, K.; Harmandaris, V. Soft Matter 2012, 8, 63206332.
C.; Schadler, L. S.; Acehan, D.; Panagiotopoulos, A. Z.; Pryamitsyn, V.; (68) Delle Site, L.; Abrams, C. F.; Alavi, A.; Kremer, K. Phys. Rev.
Ganesan, V.; Ilavsky, J.; Thiyagarajan, P.; Colby, R. H.; Douglas, J. F. Lett. 2002, 89, 156103.
Nat. Mater. 2009, 8, 354359. (69) Borodin, O.; Smith, G. D.; Bandyopadhyaya, R.; Byutner, O.
(32) Maillard, D.; Kumar, S. K.; Rungta, A.; Benicewicz, B. C.; Macromolecules 2003, 36, 78737883.
Prudhomme, R. E. Nano Lett. 2011, 11, 45694573. (70) Abrams, C. F.; Delle Site, L.; Kremer, K. Phys. Rev. E 2003, 67,
(33) Goel, V.; Pietrasik, J.; Dong, H.; Sharma, J.; Matyjaszewski, K.; 021807.
Krishnamoort, R. Macromolecules 2011, 414, 81298135. (71) Delle Site, L.; Leon, S.; Kremer, K. J. Am. Chem. Soc. 2004, 126,
(34) Alexander, S. J. Phys. (Paris) 1977, 38, 983987. 29442955.
(35) de Gennes, P. G. Macromolecules 1980, 13, 10691075. (72) Ghiringhelli, L. M.; Hess, B.; van der Vegt, N. F. A.; Delle Site,
(36) Cosgrove, T.; Heath, T.; Van Lent, B.; Leermakers, F.; L. J. Am. Chem. Soc. 2008, 130, 1346013464.
Scheutjens, J. Macromolecules 1987, 20, 16921696. (73) Brini, E.; Algaer, E. A.; Ganguly, P.; Li, C.; Rodriguez-Ropero,
(37) Milner, S. T.; Witten, T. A.; Cates, M. E. Europhys. Lett. 1988, 5, F.; van der Vegt, N. F. A. Soft Matter 2013, 9, 21082119.
413. (74) Padding, J. T.; Briels, W. J. J. Phys.: Condens. Matter 2011, 23,
(38) Milner, S. T.; Witten, T. A.; Cates, M. E. Macromolecules 1988, 233101.
21, 26102619. (75) Reith, D.; Putz, M.; Muller-Plathe, F. J. Comput. Chem. 2003, 24,
(39) Zhulina, E. B.; Borisov, O. V.; Priamitsyn, V. A. J. Colloid 16241636.
Interface Sci. 1990, 137, 495511. (76) Fu, C.-C.; Kulkarni, P. M.; Shell, M. S.; Leal, L. G. J. Chem. Phys.
(40) Borisov, O. V.; Birshtein, T. M.; Zhulina, E. B. J. Phys. II 1991, 1, 2012, 137, 164106.
521526. (77) Khounlavong, L.; Pryamitsyn, V.; Ganesan, V. J. Chem. Phys.
(41) Borisov, O. V.; Zhulina, E. B.; Birshtein, T. M. Macromolecules 2010, 133, 144904.
1994, 27, 47954803. (78) Desai, T.; Keblinski, P.; Kumar, S. K. Macromolecules 2005, 122,
(42) Zhulina, E. B.; Birshtein, T. M.; Borisov, O. V. Macromolecules 134910.
1995, 28, 14911499. (79) Hong, B.; Chremos, A.; Panagiotopoulos, A. Z. J. Chem. Phys.
(43) Aubouy, M.; Fredrickson, G. H.; Pincus, P.; Raphaeel, E. 2012, 136, 204904.
Macromolecules 1995, 28, 29792981. (80) Eslami, H.; Karimi-Varzaneh, H. A.; Muller-Plathe, F. Macro-
(44) Ferreira, P. G.; Ajdari, A.; Leibler, L. Macromolecules 1998, 31, molecules 2011, 44, 31173128.
39944003. (81) Baschnagel, J.; Meyer, H.; Varnik, F.; Metzger, S.; Aichele, M.;
(45) Matsen, M. W.; Gardiner, J. M. J. Chem. Phys. 2001, 115, 2794 Muller, M.; Binder, K. Interface Sci. 2003, 11, 159?173.
2804. (82) Nielsen, S. O.; Srinivas, G.; Klein, M. L. J. Chem. Phys. 2005,
(46) Borukhov, I.; Leibler, L. Macromolecules 2002, 35, 51715182. 123, 124907.
(47) Binder, K.; Milchev, A. J. Polym. Sci., Part B: Polym. Phys. 2012, (83) Ramrez-Hernandez, A.; Detcheverry, F. A.; de Pablo, J. J. J.
50, 15151555. Chem. Phys. 2010, 133, 064905.
(48) Maas, J. H.; Fleer, G. J.; Leermakers, F. A. M.; Cohen Stuart, M. (84) Mueller, M.; Steinmuller, B.; Daoulas, K. C.; Ramirez-
A. Langmuir 2002, 18, 88718880. Hernandez, A.; de Pablo, J. J. Phys. Chem. Chem. Phys. 2011, 13,
(49) Chevigny, C.; Dalmas, F.; Di Cola, E.; Gigmes, D.; Bertin, D.; 1049110502.
Boue, F.; Jestin, J. Macromolecules 2011, 44, 122133. (85) Faller, R.; Muller-Plathe, F.; Doxastakis, M.; Theodorou, D.
(50) Chevigny, C.; Jestin, J.; Gigmes, D.; Schweins, R.; Di-Cola, E.; Macromolecules 2001, 34, 14361448.
Dalmas, F.; Bertin, D.; Boue, F. Macromolecules 2010, 43, 48334837. (86) Faller, R. Macromolecules 2004, 37, 10951101.
(51) Chevigny, C.; Jouault, N.; Dalmas, F.; Boue, F.; Jestin, J. J. (87) Wang, H.; Junghans, C.; Kremer, K. Eur. Phys. J. E: Soft Matter
Polym. Sci., Part B: Polym. Phys. 2011, 49, 781791. Biol. Phys. 2009, 28, 221229.
(52) Sunday, D.; Ilavsky, J.; Green, D. L. Macromolecules 2012, 45, (88) Sun, Q.; Faller, R. J. Chem. Theory Comput. 2006, 2, 607615.
40074011. (89) Qi, S.; Faller, R. J. Chem. Phys. 2007, 126, 144908.
(53) Srivastava, S.; Agarwal, P.; Archer, L. A. Langmuir 2012, 28, (90) Li, Y.; Tang, S.; Abberton, B. C.; Kroger, M.; Burkhart, C.; Jiang,
62766281. B.; Papakonstantopoulos, G. J.; Poldneff, M.; Liu, W. K. Polymer 2012,
(54) Smith, G. D.; Bedrov, D. Langmuir 2009, 25, 1123911243. 53, 59355952.
(55) Kalb, J.; Dukes, D.; Kumar, S. K.; Hoy, R. S.; Grest, G. S. Soft (91) Harmandaris, V. A.; Doxastakis, M.; Mavrantzas, V. G.;
Matter 2011, 7, 14181425. Theodorou, D. N. J. Chem. Phys. 2002, 116, 436446.
(56) Meng, D.; Kumar, S. K.; D. Lane, J. M.; Grest, G. S. Soft Matter (92) Doxastakis, M.; Mavrantzas, V. G.; Theodorou, D. N. J. Chem.
2012, 8, 50025010. Phys. 2001, 115, 1133911351.

5105 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106


Macromolecules Perspective

(93) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem.
Theory Comput. 2008, 4, 435447.
(94) Bayramoglu, B.; Faller, R. Macromolecules 2012, 45, 92059219.
(95) Vazquez, U. O. M.; Shinoda, W.; Moore, P.; Chiu, C.-c.;
Nielsen, S. O. J. Math. Chem. 2009, 45, 161174.
(96) Virgiliis, A.; Milchev, A.; Rostiashvili, V.; Vilgis, T. Eur. Phys. J. E
2012, 35, 111.
(97) Rubinstein, M.; Colby, R. H. Polymer Physics; Oxford University
Press: Oxford, 2003.
(98) Jinnai, H.; Shinbori, Y.; Kitaoka, T.; Akutagawa, K.; Mashita, N.;
Nishi, T. Macromolecules 2007, 40, 67586764.
(99) Dupres, S.; Long, D. R.; Albouy, P.-A.; Sotta, P. Macromolecules
2009, 42, 26342644.
(100) Litvinov, V. M.; Orza, R. A.; Klupel, M.; van Duin, M.;
Magusin, P. C. M. M. Macromolecules 2011, 44, 48874900.
(101) Kishimoto, H.; Shinohara, Y.; Naito, M.; Takeuchi, A.; Uesugi,
K.; Suzuki, Y.; Amemiya, Y. Polym. J. 2013, 45, 6469.
(102) Zeng, Q.; Yu, A.; Lu, G. Prog. Polym. Sci. 2008, 33, 191269.

5106 dx.doi.org/10.1021/ma400444w | Macromolecules 2013, 46, 50975106

You might also like