Framework for
nanocomposites
by Richard A. Vaia* and H. Daniel Wagner†
Materials and material development are fundamental
to our very culture. We even ascribe major historical
periods of our society to materials such as the stone
age, bronze age, iron age, steel age (the industrial
revolution), polymer age, silicon age, and silica age
(the telecoms revolution). This reflects how
important materials are to us. We have, and always
will, strive to understand and modify the world
around us and the stuff of which it is made. As the
21st century unfolds, it is becoming more apparent
that the next technological frontiers will be opened
not through a better understanding and application
of a particular material, but rather by understanding
and optimizing material combinations and their
synergistic function, hence blurring the distinction
between a material and a functional device
comprised of distinct materials.
The nanoscale, and the associated excitement
surrounding nanoscience and technology, affords
unique opportunities to create these revolutionary
material combinations. These new materials promise
to enable the circumvention of classic material
performance trade-offs by accessing new properties
and exploiting unique synergisms between
constituents that only occur when the length scale of
the morphology and the critical length associated
with the fundamental physics of a given property
coincide. From a materials perspective, morphologies
that exhibit nanoscopic features are necessary but far
from sufficient – the key opportunities are afforded
either when (i) the physical size of the material’s
constituents is engineered to coincide with the onset
of nonbulk-like behavior, such as observed for the
size-dependent light emission of quantum dots (QDs),
or (ii) when a structure-property relationship
approaches a singularity or depends nonlinearly on
aspects of the morphology, such as the internal
interfacial area.
*Air Force Research Laboratory,
Materials and Manufacturing Directorate,
Wright Patterson Air Force Base,
OH 45433-7750, USA
E-mail: [email protected]
†Department of Materials and Interfaces,
Weizmann Institute of Science,
Rehovot 76100, Israel
32
November 2004
Polymeric nanocomposites (PNCs) have been an area of
intense industrial and academic research for the past
20 years. No matter the measure – articles, patents, or R&D
funding – efforts in PNCs have been exponentially growing
worldwide over the last ten years. PNCs represent a radical
alternative to conventional filled polymers or polymer blends
– a staple of the modern plastics industry. In contrast to
conventional composites, where the reinforcement is on the
ISSN:1369 7021 © Elsevier Ltd 2004
REVIEW FEATURE
Fig. 1 Categorization of nanoparticles based on increasing functionality and thus, potential to increase functionality of the polymer matrix, which in turn is based on increasing spatial and
compositional precision in nanoparticle fabrication. Increased complexity and function require increased synthetic control. Nanoscale particles and clusters are polydisperse in size (σsize)
and composition (σcomp). One-dimensional elements are compositionally uniform, with narrow size dispersion in one dimension (δ1D), but polydisperse in the other two (σ2D). Threedimensional elements exhibit narrow size dispersion in all dimensions (δ3D). Rather than compositional uniformity, prescribed composition variation in three dimensions results in tailored
functionality, where proto-assembly of any of the aforementioned elements enables the creation of an infinite variety of tailored nanoparticles. Finally, the ultimate active and functional
nano-unit would embody the characteristics of a virus – prescribed size, composition, site-specific surface functionality, and dynamic responsivity leading to alteration of the particle’s
properties, composition, or size.
order of microns, PNCs are exemplified by discrete
constituents on the order of a few nanometers, ~10 000
times finer than a human hair. The value of PNC technology
is not solely based on the mechanical enhancement of the
neat resin nor the direct replacement of current filler or blend
technology. Rather, its importance comes from providing
value-added properties not present in the neat resin, without
sacrificing the resin’s inherent processibility and mechanical
properties, or by adding excessive weight. Traditionally, blend
or composite attempts at ‘multifunctional’ materials impose
a trade-off between desired performance, mechanical
properties, cost, and processibility. However, over and over
again, property suites comparable to, or improved beyond,
those common for traditional fillers are reported for PNCs
containing substantially less filler (1-5 vol%) and thus
enabling greater retention of the inherent processibility and
toughness of the neat resin.
Considering the plurality of potential nanoparticles,
polymeric resins, and applications, the field of PNCs is
immense. For example, Fig. 1 presents a hierarchy of
nanoparticles based on increasing functionality and thus,
the potential to increase the functionality of the polymer
matrix.
This issue provides a snapshot of the rapidly developing
PNC field and a summary of two of the most investigated
nanoparticles – layered silicates and carbon nanotubes.
Development of PNCs, as with any multicomponent material,
must simultaneously balance four interdependent areas:
constituent selection, cost-effective processing, fabrication,
and performance. For PNCs, a complete understanding of
November 2004
33
REVIEW FEATURE
Fig. 2 Schematic comparison of a ‘macro’-composite containing 1 µm x 25 µm ( x L) fibers in an amorphous matrix to that of a ‘nano’-composite at the same volume fraction of filler, but
containing 1 nm x 25 nm fibers. There are three main material constituents in any composite: the matrix (white), the reinforcement (fiber, red), and the so-called interfacial region
(green), which extends (z) into the matrix on the order of Rg, the radius of gyration of the polymer. Scanning electron micrograph shows E-glass reinforced polyolefin (15 µm fiber) and
transmission electron micrograph shows montmorillonite-epoxy nanocomposite (1 nm thick layers).
these areas and their interdependencies is still in its infancy,
and ultimately many perspectives will develop, dictated by
the final application of the specific PNC. This article will
sketch a general framework for PNCs, enabling connectivities
to be drawn between PNCs and material systems within the
broader soft-matter community, as well as posing key
fundamental questions, which, when addressed, will have
broad impact to all types of PNCs.
PNC framework
The initial question when beginning to examine polymer
nanocomposites is: how are these materials different from
classic filled polymers or traditional composites? As
anticipated, there is no simple answer; rather ‘they are
related, but bring new opportunities, perspectives, and issues.’
Whether tubes (e.g. single- and multi-walled carbon
nanotubes, SWNT and MWNTs, respectively) or plates
(e.g. exfoliated graphite, layered silicates), the nanoscopic
dimensions and extreme aspect ratios inherent in these
nanofillers result in six interrelated characteristics
distinguishing the resultant PNCs from classic filled systems:
• Low percolation threshold (~0.1-2 vol%);
34
November 2004
• Particle-particle correlation (orientation and position)
arising at low volume fractions (φC < 0.001);
• Large number density of particles per particle volume
(106-108 particles/µm3);
• Extensive interfacial area per volume of particles
(103-104 m2/ml);
• Short distances between particles (10-50 nm at
φ ~1-8 vol%); and
• Comparable size scales among the rigid nanoparticle
inclusion, distance between particles, and the relaxation
volume of polymer chains.
For spherical nanoparticles, the first two characteristics are
not commonly observed because of the small aspect ratio of
the particle. Nevertheless, PNCs containing low-aspect ratio
nanoparticles are a critical bridge between conventional
micron-scale fillers and high-aspect ratio nanoparticles
where, from the nanoparticle perspective, size is reduced and
number density is increased prior to the additional
complexity of orientational correlations introduced by an
extreme aspect ratio. Note that the first two characteristics
can manifest in spherical nanoparticles systems also, vis-à-vis
innovative processing or proto-assembly of these
REVIEW FEATURE
nanoparticles. Overall, the key concept to answer ‘what is a
PNC’ is not specifically embodied within the shape of the
particle, but how do the characteristics of the particle provide
a means to ‘engineer and tailor morphology to achieve a
desired property suite from the PNC’.
To convey the origin and interrelation of these
distinguishing characteristics, Fig. 2 compares the dominant
morphological scale of a classic filled polymer containing
1 µm x 25 µm fibers in an amorphous matrix to that of a
nano-filled system at the same volume fraction of filler, but
containing 1 nm x 25 nm fibers. There are three main
material constituents in any composite: the matrix, the
reinforcement (fiber), and the so-called interfacial region. The
interfacial region is responsible for ‘communication’ between
the matrix and filler and is conventionally ascribed properties
different from the bulk matrix because of its proximity to the
surface of the filler. Nominally, the spatial extent of this
perturbed matrix is thought to extend into the bulk one to
four times the radius of gyration of the matrix, Rg, which has
a value of around tens of nanometers. For polymers, Rg is the
key spatial parameter to which the majority of the polymer’s
static and dynamic properties can be ultimately related. The
impact of the interface over a few Rg is substantiated by
extensive investigations on the behavior of thin polymer
films, as well as empirical observations in traditional
composites and filled rubbers. In the schematic of the classic
filled polymer, the relative volume fraction of this interfacial
region is exaggerated to convey its presence, but its
importance to ultimate mechanical properties is time and
again demonstrated in the literature. In contrast, because of
the increased number density of particles, the distance
between particles in the nano-filled system is comparable to
the size of the interfacial region (10 nm) and, thus, the
relative volume fraction of interfacial material to bulk is
drastically increased as the size and homogeneity of the
dominant morphological characteristics of the system
becomes smaller.
To quantify these arguments, Fig. 3 summarizes on a
log-log plot the dependence of the interfacial area per
volume of filler (µm-1 = m2/ml) on the aspect ratio (α) and
largest dimension of the filler (L, µm). Aspect ratios greater
than one correspond to rods (length/diameter) and less than
one to plates (thickness/diameter). The vast majority of
classic mineral fillers exhibit low aspect ratios (0.2 < α < 5)
and cluster sizes between 0.1-10 µm, yielding an interfacial
Fig. 3 Logarithmic isolines of interfacial (surface) area / volume of particles
(µm-1 = m2/ml) with respect to the aspect ratio, α = H/R, and largest dimension of
particle (R = radius, H = height, length) based on approximating particles as cylinders
(area/volume = 1/H + 1/R). Aspect ratios greater than one correspond to rods
(length/diameter) and less than one to plates (height/diameter). Fully exfoliated and
dispersed high aspect ratio plates or rods, such as montmorillonite or SWNTs, generate
internal interfacial area comparable to that of macromolecular structures, such as
dendrimers or proteins, and two to three orders of magnitude more than classic mineral
fillers. Comparison plots such as this draw similarities between different fillers, including
layered silicates (laponite, montmorillonite, fluorohectorite) with different degrees of
exfoliation (N = number of layers per tactoid), roped SWNTs, carbon nanofibers, chopped
glass fibers, etc.
area per particle volume of 1-100 µm-1. In comparison, highaspect ratio fibers and continuous fiber reinforcements with
10-50 µm diameters and lengths greater than 100 µm
produce an interfacial area per particle volume of
0.1-10 µm-1 – one to three orders of magnitude less than
classic mineral fillers. Nano-fillers however, whether
nanotubes or nanoplates, generate one to three orders of
magnitude more internal interfacial area per particle than
November 2004
35
REVIEW FEATURE
classic mineral fillers! For fully exfoliated and dispersed high
aspect ratio plates or rods, such as montmorillonite or
SWNTs, the amount of internal interfacial area generated is
comparable to that of macromolecular structures such as
dendrimers or proteins. Note that full exfoliation is when the
mean distance between nanoparticles, <l>, becomes a
maximum for a given volume fraction. Addressing this
preponderance of the interface is critical to optimize PNC
performance and is the source of many potentially novel
properties derived from the interface, such as
environmentally dependent mechanical damping.
Similar comparison plots can be generated for the
dependence of particle number density or mean particleparticle separation on aspect ratio and largest dimension of
the filler, leading to similar revelations. For example, fully
exfoliated nanoscopic plates or rods generate four to six
orders of magnitude more particles per volume than classic
mineral fillers. Also, for these extreme aspect ratios
(α > 100), the critical concentration, φ*, at which packing
considerations dictate orientation correlation between
particles is ~10-3 for a plate and 10-4 for a rod. These low
volume fractions, which are substantially less than even
commonly examined in nanocomposites, imply that PNCs are
not fundamentally isotropic, but will have a tendency to
exhibit anisotropic properties or a domain-like texture,
reminiscent of grains, that contain nanofillers with a local
preferred orientation even though macroscopically the global
orientation is isotropic. Since the particle number density
may be comparable to lyotropic liquid crystal systems, such
as the tobacco mosaic virus and rigid-rod polymers, similar
phase behavior may even be present. Furthermore, the
extreme aspect ratio implies transport phenomena derived
from percolated networks, such as electrical conductivity, will
show rapid enhancement at very low volume fractions of
nanoparticles.
As demonstrated in Fig. 3, the aforementioned
characteristics are not completely new. As an ideal form,
PNCs are most simply multicomponent systems where:
(i) <l>, the mean distance between nanoparticles (rigid
components), is on the order of Rg, the fundamental length
scale of the matrix (soft component); and (ii) L, the size of
nanoparticle, is also on the order of Rg. Given this idealization
and the associated implications, our understanding of
established structure-property relationships developed for
traditional composites may not be directly applicable to
36
November 2004
PNCs. The validity of these relationships and their underlying
assumptions are beginning to be reexamined by many
researchers in the light of the distinguishing characteristics
of PNCs. Additionally, the idealized PNC shares topological
commonality with numerous mesostructured
macromolecular systems, such as semicrystalline polymers,
block-copolymers, and liquid crystal polymers, whose
structure-property relationships are serving as guidance to
ascertain the dynamic behavior of PNCs, as well as the
impact of nanoparticle addition on order-disorder transitions
and phase behavior.
PNC fabrication
Given the generalized framework that distinguishes PNCs and
draws analogies with other polymeric systems, is there a
possibility for generalizing a framework guiding fabrication?
Unfortunately, an all-encompassing generalization of
fabrication approaches is much less instructive because of the
breadth of possible routes, ranging from melt dispersion of
nanoscale inorganics in a polymer to in situ polymerization or
formation of the polymer or inorganic in the presence of the
other, and the diversity of final forms desired, including bulk,
film, fiber, coating, thermoplastic, and thermoset. This
diversity of innovative approaches implies the technological
impact of PNCs will be (and already has been) broad.
Generalizing in some instances may be instructive,
however. For example, consider the dispersion of preformed
nanoparticles, which includes PNCs containing layered
silicates and carbon nanotubes, encompassing the majority of
current work in PNCs. In almost every case, nanoparticles are
added to the matrix or matrix precursors as 1-100 µm
powders, containing an association of nanoparticles – in
many instances these powders contain in excess of ten
million nanoparticles. The overwhelming majority of the
nanoparticles summarized in Fig. 1 can be grouped into two
categories based on this association: (i) low-dimensional
crystallites and (ii) aggregates. Layered silicates, SWNTs, and
other extreme aspect ratio, very thin (0.5-2 nm)
nanoparticles exhibit translational symmetry within the
powder. The alignment and close-packing results in the vast
majority of nanoparticle surface area being contained within
the crystallite, leading to a very large cohesive energy per
particle that increases with aspect ratio. Surface
functionalization generally relies on inclusion chemistry or
successive functionalization-removal cycles of nanoparticles
REVIEW FEATURE
from the crystallite exterior inward. In contrast, the initial
state of slightly larger, more polydisperse nanoparticles, such
as MWNTs and nanosilica, are disordered with at least equal
amounts of nanoparticle surface area readily accessible to
vapor or liquid medium as bound-up in nanoparticlenanoparticle contacts. The disordered state and more facile
access to surface area imply that these systems are generally
easier to disperse and amenable to a broader range of surface
modification approaches. In all cases, although engineering
the correct interfacial chemistry between nanoparticle and
medium is critical, it is not sufficient to transform the
micron-scale compositional heterogeneity of the initial
powder into nanoscale homogenization of nanoparticles
within a PNC. Mechanical mixing (compounding, sonication,
shear) is necessary – and the efficiency is dependent on the
medium viscosity and reactivity. Increasing nanoparticle
separation without sufficient mechanical mixing only
produces hybrid micron-scale reinforcing particles. These
‘hybrid’ particles exhibit an internal structure that simply
reflects that of the initial low-dimensional crystallite or
aggregate swollen by the polymer. Compositional
heterogeneities on the micron scale still exist and are
topologically similar to that of conventional filled systems,
but in this case the ‘hybrid’ filler is comprised of polymernanoparticle associates. Uniform nano-reinforcement in a
PNC implies that the dominant length scale of
heterogeneities is nanoscopic and that nanoparticlenanoparticle distances are at a maximum.
Conclusions
Today, nanocomposites are really nanofilled plastics, where
the total internal interfacial area becomes the critical
characteristic rather than simply the relative volume fraction
of constituents. The use of the moniker nano-‘composites’
invokes parallels to traditional fiber-reinforced composite
technology and the ability to spatially ‘engineer, design, and
tailor’ materials performance for a given application.
Currently, the realization of ‘compositing’ PNCs is over the
horizon. For the vast majority of investigations, the challenge
is still to achieve single-particle dispersions and the
subsequent PNCs are treated much as an isotropic, filled
polymer. Only recently have examples emerged that consider
cost-effective approaches to provide spatial and orientational
control of the hierarchical morphology with a precision
comparable to that conventionally obtained through fiber
plies and weaving – thus transforming ‘nano-filled systems’
to ‘nanocomposite systems’. In parallel, PNCs are moving
beyond commodity plastic applications to critical
components of active devices, such as fuel cell membranes,
photovoltaics, sensors, and actuators.
PNCs have great potential, especially when viewed with
respect to the explosion of available functional nanoparticles,
enabling never-before-realized properties to be generated
within plastics. The underlying framework of PNCs implies
that the physics and chemistry of these systems parallels
many macromolecular systems, not just filled polymers. By
considering the idealized framework, examination of the
underlying principles defining structure-property relationships
can begin, and the potential pitfalls arising from
extrapolating structure-property relationships of classic filled
systems can be considered and addressed. The necessary
foundation and tools to address system-specific complexities
and process-history dependencies, such as nonequilibrium
phenomena including irreversible aggregation, nanoparticle
network association, percolation, and ultra-long relaxation
times of process-induced orientation (glass-like behavior), are
beginning to evolve. The topological similarities between
PNCs and other mesoscale polymer systems, such as
semicrystalline polymers, block-copolymers, liquid crystals,
and colloids, are the impetus for many of these current
efforts, providing significant guidance toward understanding
the role of processing on structure control and the ultimate
impact on properties.
So, is the full realization of PNCs technologically here? No,
but it is a viable option today when considering the selection
of filled or blend polymer systems. Will PNCs deliver the
potential currently ascribed? That is still to be determined,
especially since realistic estimates of the ultimate potential,
which are based on fundamental understanding of the physics
at these scales, are still in development. However, the
possibilities are engaging communities worldwide, and the
scientific literature is being enriched at an increasing rate
with works that show great promise and are beginning to
establish a pervasive fundamental understanding of PNC
structure-property relationships. MT
Acknowledgments
We are very grateful for E. Manias for insightful discussions, L. Drummy for micrographs,
and partial support from the Air Force Office of Scientific Research and the Air Force
Research Laboratory, Materials and Manufacturing Directorate.
November 2004
37