Sei sulla pagina 1di 30

27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.

sgm LaTeX2e(2002/01/18) P1: FHD


10.1146/annurev.matsci.34.052803.090621

Annu. Rev. Mater. Res. 2004. 34:219–46


doi: 10.1146/annurev.matsci.34.052803.090621
First published online as a Review in Advance on March 17, 2004

THERMAL TRANSPORT IN NANOFLUIDS1


J.A. Eastman
Materials Science Division, Argonne National Laboratory, Argonne,
Illinois 60439; email: jeastman@anl.gov

S.R. Phillpot
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Department of Materials Science and Engineering, University of Florida,


Gainesville, Florida 32611; email: phillpot@mse.ufl.edu
by Georgetown University on 04/25/13. For personal use only.

S.U.S. Choi
Energy Technology Division, Argonne National Laboratory, Argonne,
Illinois 60439; email: choi@anl.gov

P. Keblinski
Materials Science and Engineering Department, Rensselaer Polytechnic
Institute, Troy, New York 12180; email: keblip@rpi.edu

Key Words heat transfer, thermal conductivity, nanoparticles, nanocomposites,


Kapitza resistance
■ Abstract Nanofluids, consisting of nanometer-sized solid particles and fibers dis-
persed in liquids, have recently been demonstrated to have great potential for improving
the heat transfer properties of liquids. Several characteristic behaviors of nanofluids
have been identified, including the possibility of obtaining large increases in ther-
mal conductivity compared with liquids without nanoparticles, strong temperature-
dependent effects, and significant increases in critical heat flux. Observed behavior is
in many cases anomalous with respect to the predictions of existing macroscopic theo-
ries, indicating the need for a new theory that properly accounts for the unique features
of nanofluids. Theoretical studies of the possible heat transfer mechanisms have been
initiated, but to date obtaining an atomic- and microscale-level understanding of how
heat is transferred in nanofluids remains the greatest challenge that must be overcome
in order to realize the full potential of this new class of heat transfer fluids.

INTRODUCTION
Cooling is one of the most important technical challenges facing numerous diverse
industries including microelectronics, transportation, manufacturing, and metrol-
ogy. Developments driving the increased thermal loads that require advances in

1
The U.S. Government has the right to retain a nonexclusive, royalty-free license in and to
any copyright covering this paper.

219
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

220 EASTMAN ET AL.

cooling include faster speeds (in the multi-GHz range) and smaller features (to
<100 nm) for microelectronic devices, higher power output engines, and brighter
optical devices. The conventional method to increase cooling rates is to use ex-
tended surfaces such as fins and microchannels; however, current designs have
already stretched this approach to its limits. Therefore, there is an urgent need for
new and innovative coolants to achieve ultrahigh-performance cooling. Taking a
different tack from extended surface approaches, the novel concept of nanofluids,
i.e., heat transfer fluids containing suspensions of nanoparticles, has been devel-
oped to meet such cooling challenges (1).
Nanofluids are a new class of solid-liquid composite materials consisting of
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

solid nanoparticles, with sizes typically on the order of 1–100 nm, suspended in a
heat transfer liquid. In recent years nanofluids have attracted great interest owing
to their greatly enhanced thermal properties. For example, a small amount (less
than 1% volume fraction) of copper nanoparticles or carbon nanotubes dispersed
by Georgetown University on 04/25/13. For personal use only.

in ethylene glycol or oil can increase their inherently poor thermal conductivity
by 40% and 150%, respectively (2, 3). Conventional particle-liquid suspensions
require high concentrations (>10%) of particles to achieve such enhancement;
they are, however, plagued by rheological and stability problems that preclude
their widespread use.
In some cases, nanofluids have been demonstrated to conduct heat an order of
magnitude better than predicted by conventional theories. Other exciting results in
this rapidly evolving field include a surprisingly strong temperature dependence
of the thermal conductivity (4, 5) and a threefold higher critical heat flux than that
of base fluids (6, 7).
These anomalously enhanced thermal properties are not merely of academic
interest, but make nanofluids promising for nanotechnology-based cooling appli-
cations including ultrahigh thermal conductivity coolants, lubricants, hydraulic
fluids, and metal-cutting fluids. Nanofluids offer several benefits: for example,
higher cooling rates, decreased pumping-power needs, smaller and lighter cooling
systems, reduced inventory of heat transfer fluids, reduced friction coefficient, and
improved wear resistance. Furthermore, nanofluids, initially designed for engineer-
ing applications such as the development of new coolants and the miniaturization
of heat exchangers (8), are being developed for medical applications including
cancer therapy. We believe that this interdisciplinary nanofluid research presents
a great opportunity to explore new frontiers in wet nanotechnology.
A number of exciting results have been reported since Choi et al. (9) wrote
the first review article on nanofluids, which covered the literature on nanofluids
published up to the end of 2002. We begin this review by describing an overview of
pre-nano studies of particles in liquids, the creation of nanofluids, and post-nano
studies of particles in liquids. In the next section we review studies of the synthesis
and thermal characterization of nanofluids containing oxide nanoparticles, metals,
and carbon nanotubes. Convective and boiling heat transfer in nanofluids is also re-
viewed. Following these discussions, we review studies of possible mechanisms for
the anomalously enhanced heat transfer in nanofluids. We conclude by identifying
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 221

TABLE 1 Comparison of room-temperature thermal conductivity values for


representative solids and liquids. The orders-of-magnitude larger thermal conductivity of
most solids compared with liquids commonly used in heat transfer applications, such as
water, ethylene glycol, and engine oil, provided the original motivation to investigate the
thermal transport properties of nanofluids
Room-temperature thermal
Material conductivity (W m−1 K−1)

Metallic solids Silver 429


Copper 401
Aluminum 237
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Nonmetallic solids Diamond 3300


Carbon nanotubes 3000
Silicon 148
by Georgetown University on 04/25/13. For personal use only.

Alumina (Al2O3) 40
Metallic liquids Sodium at 644 K 72.3
Nonmetallic liquids Water 0.613
Ethylene glycol 0.253
Engine oil 0.145

outstanding questions and key studies that still need to be done, and we present an
outlook and our perspective on the future directions of nanofluid research.

Pre-Nano Studies of Particles in Liquids


Conventional heat transfer fluids have inherently poor heat transfer properties
compared with solids. For example, as seen in Table 1, the thermal conductivity of
copper at room temperature is about 700 times greater than that of water and about
3000 times greater than that of engine oil. The thermal conductivity of carbon
multiwalled nanotubes at room temperature (10) is about 20,000 times greater
than that of engine oil. The thermal conductivities of fluids that contain these solid
particles are thus expected to be significantly enhanced over conventional heat
transfer fluids.
Numerous theoretical and experimental studies of the effective thermal con-
ductivity of solid-liquid suspensions have been conducted since Maxwell’s theo-
retical work published more than a century ago (11), modified later by Hamilton
& Crosser (12) and others (13). Until recently, these studies on the thermal con-
ductivity of suspensions had been confined to millimeter- or micrometer-sized
particles. The major practical problem with such suspensions is that the particles
quickly settle out of solution. Moreover, conventional solid-liquid suspensions that
contain millimeter- or micrometer-sized particles do not work with the emerging
miniaturized technologies because they can clog the tiny channels of these de-
vices. Such difficulties provided the motivation for the development of nanofluids
(1, 2), whereas recent advances in nanoparticle synthesis techniques have enabled
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

222 EASTMAN ET AL.

this development. In such nanofluids, the nanoparticles stay suspended for much
longer periods than do larger ones and, below a threshold size, remain in suspension
almost indefinitely despite substantial differences in the density of nanoparticles
and fluid.

Nanofluids Synthesis
The range of potentially useful combinations of nanoparticle and base fluids is
enormous: Nanoparticles of oxides, nitrides, metals, metal carbides, and nonmetals
with or without surfactant molecules can be dispersed into base fluids such as water,
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

ethylene glycol, or oils. To be able to produce the most appropriate nanoparticle-


fluid combination for a particular application, researchers have developed several
methods for nanoparticle production and dispersion.
by Georgetown University on 04/25/13. For personal use only.

Initial experimental studies (14) employed a two-step process in which nanopar-


ticles are first produced as a dry powder, typically by inert gas–condensation (15),
which involves the vaporization of a source material in a vacuum chamber and
subsequent condensation of the vapor into nanoparticles via collisions with a con-
trolled pressure of an inert gas such as helium. The resulting nanoparticles are
then dispersed into a fluid in a second processing step. Despite the large degree of
nanoparticle agglomeration that typically occurs with the gas-condensation pro-
cess, it works well in some cases, for example, with oxide nanoparticles dispersed
in deionized water (14). Less success has been achieved when producing nanoflu-
ids containing heavier metallic nanoparticles by this technique (16). An advantage
of this technique in terms of eventual commercialization of nanofluids is that the
inert-gas condensation technique has already been scaled up to economically pro-
duce tonnage quantities of nanopowders (see 17).
A second processing approach, referred to as the direct-evaporation technique,
has been used with success to produce nanofluids containing dispersed metal
nanoparticles (2, 16). This technique, developed by Yatsuya and coworkers (18),
and later improved by Wagener and coworkers (19, 20), synthesizes nanoparticles
and disperses them into a fluid in a single step. As with the inert gas–condensation
technique, the technique involves vaporization of a source material under vacuum
conditions. In this case, however, condensation of the vapor to form nanoparticles
occurs via contact between the vapor and a liquid. Nanoparticle agglomeration is
minimized by flowing the liquid continuously. A significant limitation to the appli-
cation of this technique is that the liquid must have low vapor pressure, typically
less than 1 torr. Higher vapor pressures lead to gas condensation and the associ-
ated problems of increased nanoparticle agglomeration. At present, the quantities
of nanofluids that can be produced via this direct-evaporation technique are much
more limited than with the inert gas–condensation technique, although, if desired,
it is likely that the technique could also be scaled to economically produce large
quantities of nanofluids.
While most studies of the thermal properties of nanofluids to date have used
one of the above-described processing techniques, other techniques may eventually
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 223

prove superior, depending on the particular combination of nanoparticle material


and fluid. For example, the chemical vapor condensation technique (21), in which
nanoparticles are formed by thermal decomposition of a metal-organic precursor
entrained in a carrier gas passing through a furnace, has recently been modified
to synthesize and disperse non-agglomerated nanoparticles into fluids in a sin-
gle step (V. Vasudevan, J. Vetrone, L.J. Thompson, U. Welp & J.A. Eastman,
in preparation). Compared with the direct-evaporation technique, chemical vapor
condensation appears to offer advantages in terms of control of particle size, ease
of scalability, and the possibility of producing novel core-shell nanostructures.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Interfacial Resistance
As background for our discussion of thermal transport characteristics of particle
and fiber nanofluids, we introduce here the concept of interfacial thermal resistance.
by Georgetown University on 04/25/13. For personal use only.

Interfacial resistance arises because interfaces in a structure are obstacles to heat


flow. For phonon-based conductors, the interfacial thermal resistance, also known
as the Kapitza resistance, RK, can arise from differences in the phonon spectra
of the two phases (23), and from scattering at the interface between the phases.
In the presence of an applied thermal flux, this scattering of phonons leads to a
temperature discontinuity at the interface. In analogy to the thermal conductivity,
, the interfacial conductance, G ≡ 1/Rk, is related to the heat flux, JQ, and the
temperature drop at the interface, T, via

JQ = GT. 1.

A simple measure of the relative importance of the interfacial resistance in the


overall heat flow in a composite can be obtained from the equivalent thickness of
crystal, h, defined as the distance over which the temperature drop is the same as
at the interface. This thickness is given by the ratio of bulk thermal conductivity
to the interfacial conductance:


h= . 2.
G

Assuming a good contact between phases, the value of h is usually small and
decreases with temperature such that at room temperature, the thermal resistance
of most solid-solid interfaces can be neglected in large grain–sized materials (23).
For a solid-liquid interface the value of the resistance is strongly affected by the
properties of the adsorbed layer of liquid (24); in this case the value of h also
is small, on a macroscopic scale. However, for structures with very small lateral
dimensions, such as nanoparticles or nanotubes, the interface resistance can play
an important role in the overall heat transfer, because h becomes comparable to the
size of the microstructural features, such as the particle size and the interparticle
distance.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

224 EASTMAN ET AL.

EXPERIMENTAL STUDIES
Many exciting experimental results have been reported since the pioneering the-
oretical work of Choi established this field almost a decade ago (1). The key
features of nanofluids discovered so far include thermal conductivities far above
those of traditional solid-liquid suspensions (2, 3), a nonlinear relationship be-
tween thermal conductivity and concentration in the case of nanofluids containing
carbon nanotubes (3), strongly temperature-dependent thermal conductivity (4),
and a significant increase in critical heat flux (CHF) (6, 7, 25). Each of these fea-
tures is highly desirable for thermal systems and together makes nanofluids strong
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

candidates for the next generation of liquid-based coolants. Thermal conductivity


enhancements that have been observed to date are summarized in the following
sections and in Figure 1.
by Georgetown University on 04/25/13. For personal use only.

Oxide Nanoparticles
The early experimental studies of the thermal transport properties of nanoflu-
ids focused on the behavior of fluids containing oxide nanoparticles. The first

Figure 1 Compilation of room-temperature thermal conductivity enhancements ob-


served for various nanofluids plotted as the ratio of the nanofluid thermal conductivity
to that of the thermal conductivity of the fluid alone [Al2O3 in water (14), copper in
ethylene glycol (2), carbon nanotubes in oil (3)]. The macroscopic theory prediction
of Hamilton & Crosser (HC) (12) is shown for comparison.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 225

published report that we are aware of, by Masuda et al. (26), reported 30% in-
creases in the thermal conductivity of water with the addition of 4.3 vol.% Al2O3
nanoparticles. A subsequent study by Lee et al. (14) also examined the behavior
of Al2O3 nanoparticles in water but observed a factor-of-two smaller enhancement
in thermal conductivity at the same nanoparticle loading percentage. Differences
in behavior were attributed to differences in average particle size in the two sets of
samples (the Al2O3 nanoparticles used by Masuda et al. had an average diameter
of 13 nm, compared with 33 nm in the study by Lee et al.).
Nanofluids consisting of CuO nanoparticles dispersed in water and ethylene
glycol have shown larger enhancements in thermal conductivity than have the
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

same fluids containing Al2O3 nanoparticles (14, 27). Whereas the study by Lee
et al. (14) observed only a modest improvement in the behavior of nanofluids con-
taining CuO compared with Al2O3, Zhou & Wang (27) observed a 17% increase
in thermal conductivity for a loading of only 0.4 vol.% CuO nanoparticles in water
by Georgetown University on 04/25/13. For personal use only.

[i.e., a factor-of-five larger enhancement in thermal conductivity than that found by


Masuda et al. (26) and approximately an order-of-magnitude larger enhancement
than observed by Lee et al. (14)]. These differences in behavior seen by differ-
ent groups have not yet been reconciled. Unlike the studies of Al2O3-containing
nanofluids, where differences in the thermal conductivity enhancement correlate
with changes in particle size, this does not appear to be the explanation for differ-
ences in behavior of CuO-containing nanofluids; the average diameter of the CuO
nanoparticles in the studies of Zhou & Wang (50 nm) were actually reported to be
larger than those in the study of Lee and coworkers (36 nm).
Additional evidence exists that small particle size, and hence large surface area,
is not the only important factor controlling the thermal conductivity of nanofluids.
In particular, there is growing evidence that particle surface treatment is very
important. For example, a recent study of nanofluids containing SiO2 nanoparticles
found that adding surfactants to the fluid could, in some cases, result in more than a
factor-of-three increase in thermal conductivity, with no change in particle size or
loading (J.A. Eastman, H.S. Yang & G. Skandan, in preparation). Consistent with
the idea that surface chemistry is important, the Al2O3 nanofluids used by Masuda
et al. (26) not only had smaller average particle sizes than those studied by Lee et al.
(14) but also were prepared using chemical dispersants that would be expected to
modify the nanoparticle surfaces. In contrast, Lee et al. prepared their nanofluids
using no additives. As described below, surface-modifying additives have also been
observed to enhance the properties of nanofluids containing metal nanoparticles.
A potentially important development in the nanofluids field has been the re-
cent observation by Das et al. (4) of strong temperature effects. Their studies of
Al2O3 nanoparticles in water have shown a two- to fourfold increase in thermal
conductivity enhancement over a small temperature range, 20–50◦ C. If confirmed,
and if this temperature dependence occurs over a wide temperature range, this dis-
covery could make nanofluids particularly attractive for applications at elevated
temperatures. These results also open up the possibility that nanofluids could be
employed as “smart fluids,” sensing hot spots and providing more rapid cooling
in those regions.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

226 EASTMAN ET AL.

Metallic Nanoparticles
Although fewer studies of nanofluids containing metal nanoparticles have been
carried out than of those containing oxide nanoparticles, the results have been en-
couraging. Nanofluids consisting of copper nanometer-sized particles dispersed in
ethylene glycol were found to exhibit a much higher effective thermal conductiv-
ity than were nanofluids containing the same volume fraction of dispersed oxide
nanoparticles in ethylene glycol (2). In that study, the effective thermal conduc-
tivity of ethylene glycol was shown to be increased by up to 40% for a nanofluid
consisting of ethylene glycol containing approximately 0.3 vol% Cu nanoparticles
of mean diameter ∼10 nm. The results were considered anomalous on the basis
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

of previous theoretical calculations predicting a strong effect of particle shape on


effective nanofluid thermal conductivity but with no effect on either particle size
or particle thermal conductivity (12).
by Georgetown University on 04/25/13. For personal use only.

Patel and coworkers observed increases in fluid thermal conductivity of up to


21% in their recent study of the behavior of gold and silver nanoparticles dispersed
in water and toluene (5) but, astonishingly, were able to achieve these results at
particle loadings as small as 0.011 vol%. Similar to the results of that group’s previ-
ous investigation of oxide nanofluids (4), they observed a significant enhancement
in the effect as temperature was increased over a fairly narrow range from 30 to
60◦ C. They also reported additional evidence that particle surface coatings can
have a significant effect on behavior. However, in that study, the use of thiolate,
the coating investigated, resulted in a reduction in the magnitude of the thermal
conductivity enhancement.

Carbon-Nanotube Films, Composites, and Suspensions


The large intrinsic thermal conductivity of carbon-based nanostructures, combined
with their low densities compared with metals, make them attractive candidates
for use in nanofluids. In the following sections, we assess this potential.
The problem of heat transfer in fiber-filled composites has been extensively
studied for decades. A key result from this work is that high-aspect ratio con-
ductive fillers enhance heat flow much more efficiently than low-aspect ratio or
spherical particles. The simple physical reason for this is that it is easy for heat to
rapidly move long distances along the high-conductivity fiber. By contrast, heat
has to move in and out of low-aspect ratio particles multiple times before being
transported over comparable distances. Other factors affecting the thermal con-
ductivity of the fiber composites include the volume fraction, the alignment of the
fibers, the adhesion between the fibers and the matrix, and the thermal resistance of
the interface. Because of their high thermal conductivity, comparable to in-plane
conductivity of high-quality graphite (∼2000 W m−1 K−1) (29), graphite fibers
are frequently used in such composites.
By analogy, carbon nanotubes would appear to be the ideal fibers for heat
transport. First, their thermal conductivity is similar to the in-plane conductivity
of graphite and approaches (10) or even exceeds (30) that of natural diamond, the
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 227

best room-temperature thermal conductor. Second, nanotubes have high aspect


ratios, thereby increasing the thermal conductivity. Finally, owing to this high
aspect ratio, the percolation threshold for nanotube composites can be well below
1% by volume, as demonstrated by electrical resistance measurements (31). This,
in principle, should allow rapid heat flow along the percolated tube network and
further enhancement of thermal transport.
For concentrations well above the percolation threshold, a simple geometri-
cal analysis provides an estimate for the composite thermal conductivity, eff,
assuming that the thermal resistance at the nanotube interfaces is negligible:

eff = φfiber cos2 θ, 3.


Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

where fiber is the fiber thermal conductivity, φ is the fiber volume fraction, and θ
is the angle between a given direction and a fiber axis. The brackets in Equation
by Georgetown University on 04/25/13. For personal use only.

3 indicate an average over all fibers in the composite. For well-aligned fibers,
cos2 θ  = 1, whereas for completely random fiber orientations, cos2 θ  =
1/3. If the fibers possess a large aspect ratio (>1000), effective medium theory
also predicts Equation 3 with no requirement of percolation (32). With fiber =
3000 W m−1 K−1, Vfiber = 0.01, and random fiber orientation, Equation 3 yields
eff = 10 W m−1 K−1. For well-aligned fibers, the composite conductivity along
the fiber is predicted to be ∼30 W m−1 K−1.
If experimentally realizable, nanofluids containing such fibers would have ob-
vious technological applications. A typical isotropic polymer or organic liquid has
conductivity of ∼0.1–0.2 W m−1 K−1; thus, at 1% nanotube volume fraction, the
conductivity of the composite could increase by two orders of magnitude with
respect to base matrix conductivity.
In the following, we review results of thermal-transport measurements on car-
bon nanotube films, composites, and fluid suspensions, which consistently show
much lower than expected conductivity; discuss possible reasons for this behavior;
and indicate routes toward improvements.

FILMS Important insights into the thermal performance of composites and suspen-
sions come from the measurements on carbon nanotube films that do not involve a
matrix (medium) material. In the case of single-walled tubes, the first reported ther-
mal measurements by Hone et al. involved tubes obtained by either arc-discharge
or laser vaporization deposited on the substrate yielding mats of tangled, ran-
domly oriented nanotube bundles (33). The dominant tube diameter was 1.4 nm,
and each bundle was composed of tens to hundreds of tubes and had a length of
about a micron. The measured thermal conductivity of the mat, when corrected
for the tube volume fraction, yielded at room temperature a tube conductivity
value of 35 W m−1 K−1 for as-received mats (2% theoretical density) and only
2.3 W m−1 K−1 for sintered samples (70% theoretical density). Further studies by
Hone et al. (34), and Fisher et al. (35) demonstrated that magnetic field–induced
alignment of carbon nanotubes increases the film conductivity in the alignment
direction to over 200 W m−1 K−1, i.e., to within an order of magnitude of the
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

228 EASTMAN ET AL.

graphite value. The observed anisotropy of thermal transport of the aligned tube
was modest, with ratios of 5–9, and saturated quickly with increasing magnetic
field.
Yi et al. (36) and Borca-Tasciuc et al. (37) carried out similar studies for films of
arrays of multiwalled carbon nanotubes obtained via a chemical vapor deposition
process. The diameter and length of these multiwalled tubes were typically 20–
40 nm and 1 mm, respectively. Yi et al. measured the thermal conductivity in the
alignment direction as ∼20 W m−1 K−1. Borca-Tasciuc et al. measured somewhat
larger values and demonstrated that high-temperature annealing of the sample
yielded conductivities within an order of magnitude of graphite.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

The above results clearly demonstrate that measured conductivities of macro-


scopic nanotube films are much lower than expected, even after tube alignment. In
addition to the degree of alignment, there are two other important factors that might
limit the thermal flow. The first is the tube quality, i.e., the level of defects that can
by Georgetown University on 04/25/13. For personal use only.

act as scattering centers for thermal waves (phonons). Indeed, as shown by recent
molecular dynamics simulations, a 1% vacancy concentration reduces the intrinsic
tube thermal conductivity by almost an order of magnitude (38). The increase in
thermal transport upon a high-temperature thermal anneal (37) is consistent with
this because the anneal reduces the defect concentration.
The second factor that could decrease thermal transport is the presence of
multiple tube-tube contacts. Owing to weak dispersion forces between the tubes
and the small contact area, we expect that the contact resistance can be significant.
This conjecture is supported by the fact that measurements on individual carbon
nanotubes yield high thermal conductivity values comparable to that of in-plane
graphite, whereas the tubes in bundles and mats exhibit much lower conductivity
(39). In the context of the next section, we use the term contact resistance as the
thermal resistance associated with the direct heat flow between two tubes. By
contrast, interfacial resistance is associated with the heat flow from the tube to the
matrix.

COMPOSITES AND SUSPENSIONS As discussed above, carbon nanotubes are ex-


pected to dramatically increase the thermal conductivity of low conductivity ma-
terials even at low volume fractions. The first reported work on a single-walled
carbon nanotube, a polymer epoxy composite by Biercuk et al. (31), demonstrated
a 125% thermal conductivity increase at 1 wt% nanotube loading at room temper-
ature, with eff ∼ 0.3 W m−1 K−1. Similarly, Choi et al. observed a 300% increase
for 3 wt% nanotube-loaded epoxy (40). They also observed an additional 10%
increase when the carbon nanotubes were aligned by a magnetic field. Although
these are large increases, they are more than an order of magnitude short of the
expected theoretical predictions discussed above.
Choi et al. measured thermal conductivities of oil suspensions containing multi-
walled carbon nanotubes up to 1 vol.% loading (3). As for the solid composites, the
conductivity increases for these nanofluids were also large, with a 160% increase
at 1 vol.% loading. Interestingly, the conductivity increase as a function of loading
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 229

was strongly nonlinear even at a very low volume fraction. This indicates strong
interactions of thermal fields associated with different fibers. Other studies of car-
bon nanotube suspensions showed less pronounced increases in organic liquids
and water, with only 10–20% increases in thermal conductivity at 1 vol.% (41).
There could be a number of reasons why the best-achieved conductivities of
polymer-nanotube composites and nanotube suspensions are well below theoretical
estimates. As in the case of nanotubes, both films defects and tube-tube contact
resistance can reduce the effective conductivity of the composite material. Apart
from defects, the interactions between the nanotubes and the surrounding matrix
might lead to scattering of phonons traveling along the tube, thereby reducing
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

the conductivity. However, owing to the chemical inertness of the nanotube walls,
typically only weak dispersion forces exist between the tube and matrix. Atomistic
simulations demonstrated that similar dispersion forces acting between single-
walled nanotubes (40) lead to a reduced conductivity, but only by a factor of two
by Georgetown University on 04/25/13. For personal use only.

or three rather than by over an order of magnitude.


One can also assume that the structure and, as a consequence, the thermal trans-
port properties of the matrix are altered in the regions close to tube walls. The matrix
materials such as isotropic amorphous polymers or oils, which are used in experi-
ments, are characterized by very low thermal conductivity and are expected to have
higher rather than lower thermal conductivity in the interfacial regions owing to an
increased degree of structural order (42). On this basis, one would predict a further
increase rather than the observed lower-than-expected thermal transport (43).
Apart from the matrix and tube properties, an important effect in the composite
is the heat flow across the interface and the associated thermal interfacial resis-
tance. According to effective medium theory (32), the interfacial resistance should
play an important role when the equivalent matrix thickness (see Equation 2) is
comparable to or larger than the fiber diameter. Thus, for macroscopic fibers, the
interfacial resistance is typically not important unless the matrix-fiber bonding is
so weak that an air layer separates the two constituents. For nanotubes, however,
because the diameter is so small, the role of interfacial resistance can be much
more pronounced.
The interfacial resistance between a nanotube and an organic material was re-
cently determined via a combination of experiment and molecular simulations
(44). In the experiments a pulsed laser was used to heat single-walled nanotubes
covered with an organic surfactant and suspended in water. The same laser was
then used to probe the evolution of the temperature-dependent optical adsorp-
tion of the suspension, thus allowing the process of thermal equilibration to be
monitored. From the rate of thermal equilibration, an interfacial conductance of
12 MW m−2 K−1 was obtained. Similar results were obtained from molecular
dynamics simulations of a nanotube in a model organic liquid (44).
The equivalent matrix thickness for a low-conductivity organic matrix
(∼0.1 W m−1 K−1) for the above values of interfacial resistance is about 10 nm
(see Equation 2). This is indeed a large value. For example, two tubes separated by
1 nm of matrix material have an effective thermal separation of 20 nm of matrix.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

230 EASTMAN ET AL.

This suggests that the interfacial thermal resistance contributes significantly to the
lower-than-expected thermal conductivity of nanotube–polymer composites and
suspensions.
It is also interesting to estimate the equivalent tube length along which the
temperature drop is the same as at the interface. Taking the interfacial conductance
G = 1 × 107 W/Km2 and the thermal conductivity of a good-quality tube as
λNT = 3000 W/Km, Equation 2 yields an amazing hNT = 300 µm. A typical carbon
nanotube length is on the order of microns. Therefore, for highly conductive tubes,
the temperature along each tube is essentially constant with all the temperature
drops occurring at and controlled by the interface.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

As discussed above, the percolation threshold for a carbon nanotube composite


is very low. From instance, at 1% volume fraction, heat could easily bypass the
matrix all together and still move rapidly along the well-connected network of
highly conductive tubes. There are two factors that make this impossible. First, the
by Georgetown University on 04/25/13. For personal use only.

tubes are typically wetted by the liquids and polymers used in the experiments, and
thus they are not in direct contact with each other, except those in the same bundle.
In such a case, tubes separated physically by, for example, 1 nm of the matrix
material are, from the thermal point of view, 10–20 nm apart owing to interfacial
resistance. Second, even if the crossed tubes are in direct contact, they still interact
only with weak dispersion forces and, consistent with the discussion of carbon
nanotube films above, experience large tube-tube contact resistance. Large tube-
matrix interface resistance and tube-tube contact resistance explain the lack of a
percolation threshold signature in thermal transport data (31). Simply speaking,
the interfacial and contact thermal barriers make topological connectivity much
less important for the thermal transport problem than for the electrical problem.

OUTLOOK FOR NANOTUBE COMPOSITES AND SUSPENSIONS An important practi-


cal concern is how to alleviate the problems leading to the limited thermal perfor-
mance of nanotube films, composites, and suspensions. One route is to reduce the
negative effects of large interfacial resistance present in carbon nanotube compos-
ites. An obvious approach to reducing this resistance is to increase the coupling
between the tubes and the matrix. The strongest possible coupling would be the
covalent attachment of matrix molecules or polymer chains to the tubes. These
covalent attachments, however, can severely reduce the intrinsic tube conductivity
by scattering phonons traveling along the tube axis. It is likely that an optimum
density of covalent links exists at which there is a significant reduction of interfacial
resistance while the tube conductivity is still sufficiently large. The elimination of
tube-tube contact resistance, particularly important for nanotube films and mats, is
perhaps even more challenging. A possible approach would be to construct cova-
lent junctions between the tubes. Such junctions were recently formed by electron
beam irradiation (45); however, the irradiation also leads to a high tube-defect
density and thus likely reduces the intrinsic tube conductivity.
Another route toward conductivity enhancement is to increase the fiber aspect
ratio, thereby allowing the rapid heat flow over longer paths without the need to
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 231


Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org
by Georgetown University on 04/25/13. For personal use only.

Figure 2 The effective composite conductivity estimated from the effective medium
theory (32) as a function of the fiber aspect ratio. Open symbols, zero interfacial
resistance; solid symbols, interfacial resistance equivalent to a matrix thickness of five
times the tube radius. Assumed conductivity of the fiber: 3000 W m−1 K−1; of the
matrix: 0.15 W m−1 K−1.

cross an interface, junction, or contact point. The aspect ratio can be very large
for single-walled tubes, provided that they are not in bundles. A limitation is
that single-walled tubes have small diameters and are relatively easy to bend and
therefore have a reduced effective length. The multiwalled tubes are much more
rigid and thus straighter, but their aspect ratio is typically smaller than that of
single-walled tubes. Furthermore, owing to the wall-wall thermal resistance, inner
shells might not participate fully in the thermal transport.
The benefit of the increased aspect ratio is demonstrated in Figure 2, where the
composite conductivity, estimated from effective medium theory (32), is plotted
as a function of the fiber aspect ratio for isotropic 1% volume loading. Without
the interfacial resistance (open symbols), the benefits of the aspect ratio (AR)
starts to show more significantly above AR = 10, and conductivity reaches the
maximum value at AR ∼ 1000. With an interfacial resistance equivalent to a
matrix thickness of 5 fiber diameters, the benefits of the aspect ratio are suppressed
and start to be significant only for AR > 100. However, despite this significant
interfacial resistance, for AR ∼ 10,000, the composite conductivity reaches the
same asymptotic value as for the system with zero interfacial resistance. Thus with
sufficiently long, straight, and high-quality fibers, the limitations arising from the
interfacial resistance can be ameliorated.
As already demonstrated in experiment, tube alignment allows enhanced ther-
mal performance (34, 35, 40). Further studies are needed to achieve better
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

232 EASTMAN ET AL.

alignment involving a combination of synthesis and post-processing conditions.


The issue of alignment is particularly important for the development of highly
anisotropic materials that will be highly conductive in the fiber direction and ther-
mally insulating in the perpendicular directions.
In summary, nanotube composites and suspensions demonstrate significant po-
tential as thermal-transport management materials and show some of the most
spectacular increases in thermal conductivity. Furthermore, the selection of pa-
rameters for the largest thermal conductivity increases, guided by theoretical and
modeling efforts and realized by advances in experimental synthesis and process-
ing of composite materials, can lead to even more dramatic thermal conductivity
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

increases.

Importance of Properties Other than Thermal Conductivity


by Georgetown University on 04/25/13. For personal use only.

The thermal conductivity measurement of nanofluids was the main focus in the
early stages of nanofluid research. Recently, however, studies have been carried
out on the heat transfer coefficient of nanofluids in natural (46, 47) and forced
(48–51) flow.
Most studies carried out to date are limited to the thermal characterization of
nanofluids without phase change (boiling, evaporation, or condensation). However,
nanoparticles in nanofluids can play a vital role in two-phase heat transfer systems,
and there is a great need to characterize nanofluids in boiling and condensation
heat transfer. Das et al. (25) initiated experiments on the boiling characteristics of
nanofluids. You et al. (6) measured the critical heat flux in pool boiling of Al2O3-
in-water nanofluids and reported an unprecedented threefold increase over that of
pure water. Vassallo et al. (7) subsequently reported that silica-in-water nanofluids
show threefold higher critical heat flux than base fluids. Faulkner et al. (52) have
initiated studies of flow boiling of nanofluids.

FLUID PROPERTIES AFFECTING CONVECTION HEAT TRANSFER In convection heat


transfer in nanofluids, the heat transfer coefficient depends not only on the thermal
conductivity but also on other properties such as the specific heat, density, and
dynamic viscosity of a nanofluid.
The density of a nanofluid can be calculated by

ρ = (1 − φ)ρ f + φρ p , 4.

where φ is the volume fraction of nanoparticles and ρ f and ρ p are the densities
of a base fluid and nanoparticles, respectively. It should be noted that the density
of a nanofluid is a linear function of volume fraction. For typical nanofluids with
nanoparticles at a value of volume fraction less than 1%, a change of less than 5%
in the fluid density is expected if a nanofluid contains nanoparticles.
The specific heat C p of a nanofluid can be calculated by

ρC p = (1 − φ)ρ f C p f + φρ p C pp , 5.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 233

where C p f and C pp are the specific heats of a base fluid and nanoparticles, respec-
tively. Using these equations, one can predict that small decreases in specific heat
will typically result when solid particles are dispersed in liquids. For example,
adding 3 vol% Al2O3 to water would be predicted to decrease the specific heat by
approximately 8% compared with that of water alone. Preliminary measurements
of the specific heat of water containing a few volume percent Al2O3 nanoparti-
cles have shown no measurable difference in specific heat compared with water
alone (H.S. Yang, J.A. Eastman, & S.U.S. Choi, unpublished data). The simple
equations above may need to be modified if nanoparticles are found to exhibit
a size-dependent specific heat. However, our expectation is that this will not be
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

the case, on the basis of studies of nanocrystalline solids in which no significant


size-dependent change in specific heat is observed (see 54).
Wang et al. (55) measured the viscosity of water-based nanofluids containing
Al2O3 nanoparticles dispersed by different dispersion techniques and showed that
by Georgetown University on 04/25/13. For personal use only.

nanofluids have lower viscosities when the particles are better dispersed. They also
showed an increase of ∼30% in viscosity at 3 vol.% Al2O3, compared with that of
water alone. However, the viscosity of the Al2O3/water nanofluids prepared by Pak
& Cho (56) was three times higher than that of water. This shows that the viscosity
of nanofluids depends on dispersion methods. Xuan & Li (50) measured the tur-
bulent friction factor of water-based nanofluids containing copper nanoparticles in
the volume fraction range between 1.0 and 2.0. They found that the friction factor
for the nanofluids is approximately the same as that of water.
For metallic nanofluids containing a low volume fraction of nanoparticles (usu-
ally <0.01), an Einstein model would predict that the change in the viscosity of
a suspension of non-interacting spherical particles is small and linear with the
volume fraction:

η = (1 + 2.5φ)η f , 6.

where ηf is the viscosity of the base fluid. The Einstein equation is valid only for
φ < 0.05.
Das and coworkers (25, 46) measured the viscosity of Al2O3-in-water and CuO-
in-water nanofluids as a function of shear rate and showed Newtonian behavior of
the nanofluids for a range of volume fractions between 1% and 4%.

FORCED CONVECTION HEAT TRANSFER IN NANOFLUIDS Heat transfer tests have


been conducted to assess the thermal performance of oxide and metallic nanoflu-
ids under turbulent flow conditions. Eastman et al. (57) reported that the heat
transfer coefficient of water containing 0.9 vol% of CuO nanoparticles was im-
proved by >15% compared with that of water without nanoparticles. Recently,
Xuan & Li (50) measured the convective heat transfer coefficient and friction fac-
tor of Cu-water nanofluids in turbulent flow in the range of Reynolds numbers
between 10,000 and 25,000. Their results show that a small amount (less than
2 vol%) of copper nanoparticles in deionized water remarkably improves convec-
tive heat transfer. For example, the Nusselt number of water containing 2.0 vol%
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

234 EASTMAN ET AL.

of Cu nanoparticles was improved by >39% compared with that of water with-


out nanoparticles. Furthermore, the Nusselt number of the Cu-water nanofluids
increased with increasing particle volume fraction and Reynolds number. The
Dittus-Boelter correlation failed to predict the improved experimental heat trans-
fer behavior of nanofluids. In contrast to these studies, Pak & Cho (56) found that
the convective heat transfer coefficient of water-based nanofluids with 3 vol.%
Al2O3 and TiO2 nanoparticles was 12% smaller than that of pure water when
tested under constant average velocity. One possible reason for this reduction in
heat transfer could be that the viscosity of the nanofluids prepared in this study was
much larger than that of the nanofluids used by Wang et al. (55) as discussed above.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

NATURAL CONVECTION HEAT TRANSFER IN NANOFLUIDS Natural convection heat


transfer in nanofluids has begun to receive attention owing to potential applica-
by Georgetown University on 04/25/13. For personal use only.

tions of nanofluids in micro-electromechanical systems (MEMS) and electronics


cooling. Putra et al. (46) studied natural convection of Al2O3-in-water and CuO-
in-water nanofluids inside a horizontal cylinder heated from one end and cooled
from the other. Unlike conduction or forced convection, these natural convection
experiments showed a decrease in heat transfer. The reason for this deteriora-
tion is not known, and more research is needed to understand natural convection
heat transfer in nanofluids. In contrast, Khanafer et al. (47) developed a thermal
dispersion model to simulate natural convection heat transfer of nanofluids in a
two-dimensional enclosure, which showed a remarkable heat transfer enhance-
ment with volume fraction of copper nanoparticles owing to the enhanced thermal
dispersion effect by random motion of nanoparticles.

POOL BOILING OF NANOFLUIDS Das et al. (25) conducted an experimental study


of pool boiling of Al2O3-in-water nanofluids on a horizontal tube of large diameter
(20 mm) and showed a deterioration of the pool boiling performance of nanofluids
with increasing particle volume fraction. This deterioration was not attributed
to the change of fluid properties by nanoparticles but to the change of surface
characteristics by nanoparticles trapped on the rough surface. Das et al. investigated
pool boiling of Al2O3-in-water nanofluids on horizontal tubes of small diameter (4
and 6.5 mm) and showed that pool boiling of the nanofluid on small-diameter tubes
is qualitatively different from that on large-diameter tubes owing to the difference
in bubble sliding mechanism. They found that the deterioration in pool boiling
performance is less on narrow tubes than on large industrial tubes, which makes
nanofluids less susceptible to local overheating in convective applications in small
tubes.
You et al. (6) measured the CHF in pool boiling on a flat square copper surface
immersed in Al2O3-in-water nanofluids and showed an unprecedented threefold
increase in CHF over that of pure water. The average size of the departing bubbles
increases and the bubble frequency decreases significantly in nanofluids compared
with pure water. The remarkable results of You et al. are consistent with the more
recent results of Vassallo et al. (7), who conducted experiments with silica-in-water
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 235

nanofluids. It is intriguing that nanoparticles at concentrations as low as 10−5 can


trigger such a dramatic increase in CHF; no existing model can explain such an
effect.

FLOW BOILING OF NANOFLUIDS Faulkner et al. (52) performed an experimental


study of subcooled and saturated forced convection boiling heat transfer in a par-
allel microchannel heat sink for thermal management of high-power microwave
electronics. The working fluids tested were water and a selection of ceramic-based
nanoparticle suspensions. The system dissipated heat fluxes in excess of 275 W/cm2
of substrate. However, for optimized fin geometry, the current conditions resulted
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

in fluxes greater than 500 W/cm2. Although the use of nanofluids was intended
for boiling enhancement with reduced flow, poor distribution in the microchannels
resulted in limited improvement in the overall heat transfer rate.
by Georgetown University on 04/25/13. For personal use only.

HEAT TRANSFER MECHANISMS IN NANOFLUIDS:


INSIGHTS FROM THEORY AND SIMULATION
In this section we examine the various factors that could be responsible for the
inadequacy of the conventional macroscopic analyses of heat transfer in compos-
ites as applied to nanofluids. We focus our discussions on stationary nanofluids
because they exhibit anomalous thermal properties, even in the absence of the ad-
ditional complexities associated with flow or convection. We also briefly discuss
the additional effects of convective heat transport in nanofluids.
Within the context of stationary liquids we first discuss the effects of interfacial
resistance within the effective medium theory. We then examine a number of factors
discussed in the literature as possible mechanisms for the anomalous enhancement:
(a) the motion of the nanoparticles, (b) molecular-level layering of the liquid
at the liquid-particle interface, (c) ballistic heat transport in nanoparticles, and
(d) the effects of clustering of nanoparticles.

Interfacial Resistance
Despite the obvious potential applications of nanofluids and nanocomposites,
solid-liquid interfaces have received relatively little attention compared with solid-
solid interfaces. Wilson et al. (58) measured the interface thermal conductance G
of citrate-stabilized platinum nanoparticles in water (G = 130 MW m−2 K−1) and
of nanoparticles of alkanethiol-terminated AuPd in toluene (G = 5 MW m−2 K−1).
Also, the interface thermal conductance between carbon nanotubes and surfactant
micelles (44) was found to be G = 12 MW m−2 K−1. Values of the interfacial
conductance on the order of 10 MW m−2 K−1 correspond to the equivalent matrix
thickness of ∼10 nm for a typical low-conductivity organic liquid or isotropic
polymer. This length is of the same order as the size of particles or nanofibers,
indicating the importance of the thermal resistance for nanoscale materials.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

236 EASTMAN ET AL.

To quantify the effect of the interfacial resistance, we can take the effective
medium theory result (32) that for a small volume fraction of spherical particles and
in the limit of the particle conductivity being much larger than matrix conductivity,
the effective composite conductivity, eff, will be

eff γ −1
− 1 = 3φ , 7.
matrix γ +2

where φ is the particle volume fraction and γ is the ratio of the particle radius to the
equivalent matrix thickness. According to Equation 7, the conductivity enhance-
ments decrease with increasing interfacial resistance (i.e., with decreasing γ ).
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Moreover, when the particle radius becomes equal to the equivalent matrix thick-
ness (γ = 1), there is no enhancement at all, while for larger interfacial resistance
(γ < 1), the addition of particles actually decreases the thermal conductivity of
by Georgetown University on 04/25/13. For personal use only.

the composite. These predictions are totally opposite of the observed behavior of
nanofluids, indicating that even if this effect is present, it must be overcompensated
by other mechanisms that increase the thermal conductivity.
In contrast to the results on nanofluids, the conductivity measurements of
Putnam et al. (59) on polymer alumina nanoparticle composite are consistent with
the predictions of the effective medium theory and do not yield any anomalous
conductivity increases. Together these results appear to indicate that the ability of
nanoparticles or liquid to move must play a significant role in nanofluid thermal
transport.

Nanoparticle Motion
The energy exchange in direct nanoparticle-nanoparticle contact arising from par-
ticle collisions in the nanofluid could result in an enhancement of the thermal con-
ductivity. Such collisions arise from the motion of the nanoparticles. Furthermore,
even without collisions the Brownian motion of particles might enhance thermal
conductivity. As we pointed out above, no anomalous enhancement in thermal
conductivity was found in a polymer-nanoparticle nanocomposite in which there
is no Brownian motion or particle collisions (59). The effect associated with par-
ticle motion is not accounted for in the conventional theory of thermal transport
for composite materials.
The most obvious type of nanoparticle motion arises from Brownian forces.
For Brownian motion to be a significant contributor to thermal conductivity, it
would have to be a more efficient heat transfer mechanism than thermal diffusion
in the fluid. However, Keblinski et al. (42) have shown that a nanoparticle with a
diameter of 10 nm takes τ D ∼ 2 × 10−7 s to move a distance equal to its size. By
contrast, the time required to move heat the same distance through the liquid is only
τ H ∼ 4 × 10−10 s. The ratio of τ D/τ H is ∼500 and would still be as large as ∼25 for
a nanoparticle of atomic dimensions (∼0.5 nm). Thus, because thermal diffusion
is much faster than Brownian diffusion, even within the limits of extremely small
particles, Brownian dynamics are unlikely to significantly enhance the thermal
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 237

conductivity. This conclusion was confirmed by molecular dynamics simulations


showing that heat flow in model nanofluids in which the nanoparticles can move
were indistinguishable from identical nanofluids in which the nanoparticles were
pinned (42). Xuan et al. (50) have recently come to the opposite conclusion;
however, their work is hard to assess since their result for the enhancement of the
thermal conductivity (their equation 13) has the wrong units.
Although Brownian motion cannot directly result in an enhancement of the
thermal transport properties, it could have an important indirect role in producing
particle clustering, which as described below, could significantly enhance thermal
conductivity.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Finally, the thermal gradients in the materials and the presence of the heaters in
the experiments could lead to thermally driven nanoparticle motion resulting in a
nonuniform particle distribution in the liquid and consequently modified thermal
transport properties through thermophoresis (60). An analysis of this effect for
by Georgetown University on 04/25/13. For personal use only.

nanofluids has yet to be made.


Thus we are left with the rather unsatisfying situation that the experiments on
nanofluids and the experiment of Putnam et al. (59) on polymer nanocomposites
seem to indicate that particle motion is at least partially responsible for the en-
hancement in thermal conductivity, whereas theoretical approaches have failed to
identify a plausible mechanism. Clearly more experimental and theoretical work
in this area is required to resolve this issue.

Liquid Layering at Liquid-Particle Interface


It is well known that liquids tend to show a significant amount of structural ordering
at solid-liquid interfaces. If this were to enhance the thermal transport in the layer
liquid region, then it could result in an increase in the thermal conductivity.
To estimate an upper limit for this effect, let us assume that the thermal con-
ductivity of this interfacial liquid is the same as that of the solid. The resultant
larger effective volume of the particle-layered-liquid structure would enhance the
thermal conductivity (see Figure 3) (42). Then, for example, to double the effective
volume Veff of a particle with d = 10 nm would require a layered-liquid thickness
of h ∼ 2.5 nm. A detailed analysis of experimental results for nanoparticles of
Al2O3 in water (43) concluded that the experimental results could be explained if
there were an ordered liquid layer of 3 nm thickness with a thermal conductivity of
2.1 W m−1 K−1; Yu & Choi (61) came to the same conclusion in a similar analysis.
However, experiments (62) and simulation (63) have shown a typical interfacial
width that is only on the order of a few atomic distances, i.e., ∼1 nm. Thus if the
ordered liquid layer did have an enhanced thermal conductivity, such layering
could be a significant contributor, although it could not explain the effect entirely.
Xue et al. have recently performed molecular dynamics simulations to deter-
mine if the liquid ordering does in fact lead to an increase in thermal transport
(64). The solid line in Figure 4 shows the density profile through the solid-liquid
interface. Consistent with previous studies, they found significant layering in the
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

238 EASTMAN ET AL.


Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org
by Georgetown University on 04/25/13. For personal use only.

Figure 3 Excess thermal-conductivity enhancement κ owing to formation of highly


conductive layered-liquid structure at liquid-particle interface for several values of
layer thickness h as a function of particle diameter d.

first two layers (i.e., ∼0.5 nm), with weaker layering for about another three layers
(another ∼0.75 nm). The solid circles in Figure 4 show the corresponding temper-
ature profile. The temperature in the bulk solid is almost constant, characteristic of
the high-thermal conductivity of the solid. In the liquid, far from the interface, the
temperature gradient is much larger, reflecting the much lower thermal conductiv-
ity of the bulk liquid. Most importantly, the temperature gradient in the ordered
liquid layer is the same as that in the bulk liquid. It was thus concluded that the
ordering of the liquid layer does not actually result in any measurable increase in
the thermal conductivity (64).
However, because the above simulations were performed for a simple mona-
tomic fluid, they do not rule out the possibility that more subtle effects in molecular
liquids or water might enhance thermal transport.

Nature of Heat Transport in Nanoparticles


Macroscopic theories assume that heat is transported by diffusion. In crystalline
solids, heat is carried by phonons, that is, by the propagation of lattice vibrations.
Such phonons are created at random, propagate in random directions, and are
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 239


Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org
by Georgetown University on 04/25/13. For personal use only.

Figure 4 Effect of liquid layering on thermal transport. The solid line shows the
density profile through a solid-liquid interface. The solid symbols show the associated
temperatures. The temperature decrease in the layered liquid is the same as in the bulk
liquid, indicating no enhanced thermal transport in the layered liquid.

scattered by each other or by defects and thus justify the use of the macroscopic
description of heat transport. When the size of the nanoparticles in a nanofluid
becomes less than the phonon mean-free path, phonons no longer diffuse across
the nanoparticle but move ballistically without any scattering. For Al2O3 at room
temperature, the mean-free path is ∼35 nm; for nanoparticles, a macroscopic
approach, such as that employed by Hamilton & Crosser (HC) (12), does not
apply, and a theoretical treatment based on ballistic phonon transport is required
(65). Without going into the details of ballistic heat transport, it is difficult to
envision how ballistic phonon transport could be more effective than a very-fast-
diffusion phonon transport, particularly to the extent of explaining the order-of-
magnitude-larger increase of thermal conductivity in Cu nanofluids. In particular,
for both ballistic and fast-diffusive phonon transport, the temperature within the
solid particle will be essentially constant, providing the same boundary condition
for heat flow in a low-thermal-conductivity liquid.

Effects of Nanoparticle Clustering


If particles could cluster into percolating networks, they would create paths of
lower thermal resistance and thereby have a major effect on the effective thermal
conductivity. However, clustering to the extent that solid agglomerates span large
distances is unlikely; moreover any such large clusters would most likely settle
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

240 EASTMAN ET AL.

out of the fluid. Most importantly, although the percolation threshold for random
dispersions is ∼15% volume fraction in three dimensions, the unusual enhance-
ment of thermal conductivity is already observed at very-low-volume fractions of
∼1% and less.
Although it appears that percolating structures cannot be set up, local clustering
is possible. The effective volume of a cluster, that is, the volume from which other
clusters are excluded, can be much larger than the physical volume of the particles.
Since within such clusters, heat can move very rapidly, the effective volume fraction
of the highly conductive phase (Vp in the HC theory) is larger than the physical
volume of the solid. This effective large volume fraction would, according to the
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

HC theory, significantly increase the thermal conductivity. The effect of clustering


is illustrated in Figure 5, which shows the excess thermal conductivity enhancement
κ originating from the increased effective volume of highly conducting clusters
as a function of the packing fraction of the cluster φ (ratio of the volume of the
by Georgetown University on 04/25/13. For personal use only.

solid particles in the cluster to the total effective volume of the cluster). With
decreasing packing fraction, the effective volume of the cluster increases, thus
enhancing thermal conductivity. Even for a cluster of closely packed spherical
particles, ∼25% of the volume of the cluster consists of liquid filling the space

Figure 5 Excess thermal conductivity enhancement κ owing to increased effective


volume φ of highly conducting clusters. Schematic diagrams indicate (right to left)
(i) closely packed fcc arrangement of particles, (ii) simple cubic arrangement,
(iii) loosely packed irregular structure of particles in physical contact, and (iv) clusters
of particles separated by liquid layers thin enough to allow for rapid heat flow among
particles.
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 241

between particles, which increases the effective volume of a highly conductive


region by ∼30% with respect to a dispersed nanoparticle system. For more loosely
packed clusters the effective volume increase will be even larger.
Wang et al. (66) recently performed a detailed analysis of such clustering effects
in terms of the fractal properties of the clusters and concluded that clustering effects
can explain the thermal transport properties of dilute suspensions of metal oxide
particles; they cannot, however, explain the larger increases observed for metal
nanoparticles (66).
A further dramatic increase of κ can take place if the particles do not need to be
in physical contact but just close enough to allow rapid heat flow between them.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

Such liquid-mediated clusters exhibit a very low packing fraction and thus a very
large effective volume and, in principle, are capable of explaining the unusually
large experimentally observed enhancements of thermal conductivity. In particular,
if the ballistic phonons initiated in one particle could persist in the liquid and reach
by Georgetown University on 04/25/13. For personal use only.

a nearby particle, a significant increase of thermal conductivity could take place.


Because the phonon mean free path is much shorter in the liquid than in the particle,
such an effect can operate only if the separation between particles is very small,
likely on the order of the thickness of the layered liquid (∼1–2 nm).
This is not an unreasonable scenario because the particles in a nanofluid are
surprisingly close together even at relatively low packing fractions. For example,
the surfaces of 10-nm particles are, on average, separated by only 5 nm at a 5%
packing fraction. Observed ordering of liquid-dispersed nanoparticles in the vicin-
ity of liquid-bulk solid interfaces (67) could enhance this effect, but this possibility
has not yet been seriously evaluated. Even in the absence of such ordering, because
particles move constantly owing to Brownian motion, locally they will sometimes
be much closer and thus enhance coherent phonon heat flow among the particles.
Indeed, an analogous phenomenon was observed in experiments on sound prop-
agation in colloidal systems in which acoustic excitations propagated coherently
between neighboring particles when the diameter of particles was comparable to
the wavelength of the sound (68). A further contribution could arise from the prop-
agation of evanescent electromagnetic waves between particles in close proximity
(69).
It must be remembered that in general, clustering may exert a negative effect on
heat transfer enhancement, particularly at low volume fraction, by settling small
particles out of the liquid and creating large regions of particle-free liquid with
high thermal resistance.

Summary of the Current Understanding of Heat Transfer


Mechanisms in Nanofluids
We have evaluated the extent to which the four specific mechanisms discussed in
the literature could contribute to thermal conductivity. None of these mechanisms
alone seems to be able to explain the anomalous increases in thermal conductivity
in nanofluids. Whereas the comparison of experiments on nanofluids and polymer
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

242 EASTMAN ET AL.

composites seems to implicate particle motion, no plausible mechanism has been


identified. In contrast, theoretical analysis points to clustering effects as a plausible
means of enhancing the thermal transport.
Finally, we briefly address the issue of convective flow. Xuan and coworkers
have argued that the random movement of the nanoparticles in a flowing nanofluid
increases the energy exchange rates in the fluid, resulting in thermal dispersion
in the nanofluid flow (49, 50). This could result in a further enhancement of the
heat transfer coefficient in addition to that from the enhanced thermal conductiv-
ity (49, 50). Recently, Khanafer et al. have performed a more detailed analysis
and numerical simulation, which, by explicitly showing how nanoparticles al-
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

ter the structure of the fluid flow and enhance heat transfer, came to the same
conclusion (47). Yu et al. have suggested that the thermal conductivity could
be enhanced if the convecting nanoparticles carry a liquid boundary layer with
them (70).
by Georgetown University on 04/25/13. For personal use only.

OUTLOOK AND FUTURE CHALLENGES


As described in this review, many interesting properties of nanofluids have been
reported. This is particularly remarkable and encouraging because of the rela-
tively short period (<10 years) of active investigation in this area. While static
measurements of thermal conductivity have been carried out most extensively,
several groups have recently initiated studies of other properties, often with sim-
ilarly encouraging results. Although the potential for the use of nanofluids in a
wide variety of applications is promising, a key stumbling block seriously hinder-
ing the development of the field is that a detailed atomic-level understanding of
the mechanism(s) responsible for the observed property changes remains elusive.
We conclude this review by pointing out that several important clues to the mech-
anism of heat transfer have begun to emerge from the growing number of studies
of nanofluid properties:
1. Size is likely to be important. Dispersion behavior improves and the surface
area-to-volume ratio increases substantially with decreasing particle size.
Heat transfer in nanofluids must in some way involve flow of heat in the
vicinity of the nanoparticle-fluid interfaces; hence, reducing nanoparticle
sizes is expected to lead to increased heat transfer rates. With many process-
ing techniques used to date, particle size is difficult to control or can only
be controlled over a rather narrow range of sizes. This, in turn, has made it
difficult to uncouple the effects of particle size on heat transfer from other
parameters. A challenge for the future is to improve existing synthesis tech-
niques or to develop new nanoparticle synthesis and dispersion techniques
that will enable systematic study of a series of nanofluids differing only in
the size of the nanoparticles they contain.
2. Particle agglomeration is undesirable. Dispersion behavior improves sub-
stantially when nanoparticles are prevented from agglomerating. Other than
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 243

for carbon nanotubes, the largest increases in thermal conductivity have been
observed in samples with little or no particle agglomeration. In most studies
to date, sample sizes were limited to less than a few hundred milliliters of
nanofluid. Larger samples will be needed to test many properties of nanoflu-
ids in the future, particularly in assessing their potential for use in appli-
cations. Inert gas–condensation synthesis, which has already been scaled
up to produce large quantities of nanopowders, typically produces heavily
agglomerated powders. A challenge for the future is to develop processing
techniques that will allow production of larger quantities of nanofluids with
nonagglomerated nanoparticles.
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

3. Particle surface treatment is important. Several studies have demonstrated


that thermal transport behavior is strongly affected by the presence or absence
of chemical species attached to particle surfaces. Surfactants can improve
by Georgetown University on 04/25/13. For personal use only.

the dispersion properties and long-term stability of nanofluids. The Kapitza


resistance of the nanoparticle-fluid interface may also be strongly affected
by species attached to the surface. Surface species could impact particle
mobility and the extent of liquid ordering. Significantly more study of the
effects of particle surface treatment on thermal behavior is needed.
4. Temperature may be important. Few studies have examined the behavior
of nanofluids other than at room temperature. A small number of studies
by one group have indicated a very strong temperature effect. More study
is needed to determine the extent of such temperature-dependent behavior.
Such a study could, for example, provide important clues that will allow
determination of the importance of particle motion in controlling thermal
transport behavior.

ACKNOWLEDGMENTS
This work was supported by the U.S. Department of Energy, Basic Energy Sciences-
Materials Sciences, under Contract No. W-31–109-ENG-38. PK acknowledges
funding from Phillip Morris USA.

The Annual Review of Materials Research is online at


http://matsci.annualreviews.org

LITERATURE CITED
1. Choi SUS. 1995. Enhancing thermal con- Thompson LJ. 2001. Anomalously in-
ductivity of fluids with nanoparticles. In creased effective thermal conductivities of
Developments and Applications of Non- ethylene glycol-based nanofluids contain-
Newtonian Flows, ed. DA Siginer, HP ing copper nanoparticles. Appl. Phys. Lett.
Wang, pp. 99–105. New York: Am. Soc. 78:718–20
Mech. Eng. 3. Choi SUS, Zhang ZG, Yu W, Lockwood
2. Eastman JA, Choi SUS, Li S, Yu W, FE, Grulke EA. 2001. Anomalous thermal
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

244 EASTMAN ET AL.

conductivity enhancement in nano-tube 16. Eastman JA, Choi SUS, Li S, Thompson


suspensions. Appl. Phys. Lett. 79:2252–54 LJ, Lee S. 1997. Enhanced thermal con-
4. Das SK, Putra N, Thiesen P, Roetzel W. ductivity through development of nanoflu-
2003. Temperature dependence of thermal ids. In Nanocrystalline and Nanocomposite
conductivity enhancement for nanofluids. Materials II, ed. S Komarnenl, JC Parker,
ASME J. Heat Trans. 125:567–74 HJ Wollenberger, p. 3. Pittsburgh: Materi-
5. Patel HE, Das SK, Sundararajan T, Nair als Research Society
AS, George B, Pradeep T. 2003. Thermal 17. Romano JM, Parker JC, Ford QB. 1997.
conductivities of naked and monolayer pro- Application opportunities for nanoparticles
tected metal nanoparticle based nanoflu- made from condensation of physical va-
ids: manifestation of anomalous enhance- pors. Adv. Powder Metall. Partic. Mater. 2:
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

ment and chemical effects. Appl. Phys. Lett. 12–13


83:2931–33 18. Yatsuya S, Tsukasaki Y, Mihama K, Uyeda
6. You SM, Kim JH, Kim KH. 2003. Effect of R. 1978. Preparation of extremely fine par-
nano-particles on critical heat flux of wa- ticles by vacuum evaporation onto a run-
by Georgetown University on 04/25/13. For personal use only.

ter in pool boiling heat transfer. Appl. Phys. ning oil substrate. J. Cryst. Growth 45:490
Lett. 83:3374–76 19. Wagener M, Murty BS, Günther B. 1997.
7. Vassallo P, Kumar R, D’Amico S. 2004. Preparation of metal nanosuspensions by
Pool boiling heat transfer experiments in high-pressure DC-sputtering on running
silica-water nano-fluids. Int. J. Heat Mass liquids. In Nanocrystalline and Nanocom-
Trans. 47:407 posite Materials II, ed. S Komarnenl, JC
8. Tuckerman D, Pease R. 1981. High- Parker, HJ Wollenberger, p. 149. Pitts-
performance heat sinking for VLSI. IEEE burgh: Materials Research Society
Electron. Device Lett. 2:126 20. Wagener M, Günther B. 1999. Sputtering
9. Choi SUS, Zhang ZG, Keblinski P. in liquids: a versatile process for the pro-
2004. Nanofluids. In Encyclopedia of duction of magnetic suspensions. J. Magn.
Nanoscience and Nanotechnology, ed. HS Magn. Mater. 201:41
Nalwa. Stevenson Ranch, CA: American 21. Srdic’ VV, Winterer M, Möller A, Miehe
Scientific. In press G, Hahn H. 2001. Nanocrystalline zirco-
10. Kim P, Shi L, Majumdar A, McEuen PL. nia surface-doped with alumina: chemical
2001. Thermal transport measurements of vapor synthesis, characterization and prop-
individual multiwalled nanotubes. Phys. erties. J. Am. Ceram. Soc. 84:2771–76
Rev. Lett. 87:215502 22. Deleted in proof
11. Maxwell JC. 1873. A Treatise on Electricity 23. Swartz ET, Pohl RO. 1989. Thermal bound-
and Magnetism. Oxford: Clarendon ary resistance. Rev. Mod. Phys. 61:605
12. Hamilton RL, Crosser OK. 1962. Thermal 24. Nakayama T. 1985. New channels of en-
conductivity of heterogeneous two-compo- ergy transfer across a solid liquid He inter-
nent systems. Ind. Eng. Chem. Fundam. face. J. Phys. Condens. Matter 18:L667–71
1:187–91 25. Das SK, Putra N, Roetzel W. 2003. Pool
13. Nemat-Nasser S, Hori M. 1993. Microme- boiling characteristics of nano-fluids. Int.
chanics: Overall Properties of Heteroge- J. Heat Mass Trans. 46:851–62
neous Materials. The Netherlands: Elsevier 26. Masuda H, Ebata A, Teramae K, Hish-
14. Lee S, Choi SUS, Li S, Eastman JA. 1999. inuma N. 1993. Alteration of thermal
Measuring thermal conductivity of fluids conductivity and viscosity of liquid by
containing oxide nanoparticles. J. Heat dispersing ultra-fine particles (dispersion
Trans. 121:280 of γ -Al2O3, SiO2 and TiO2 ultra-fine par-
15. Granqvist CG, Buhrman RA. 1976. Ultra- ticles). Netsu Bussei 4:227–33
fine metal particles. J. Appl. Phys. 47:2200 27. Zhou LP, Wang BX. 2002. Experimental
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

THERMAL TRANSPORT IN NANOFLUIDS 245

research on the thermophysical proper- Thermal conductivity of carbon nanotubes.


ties of nanoparticle suspensions using the Nanotechnology 11:65
quasi-steady method. Annu. Proc. Chin. 39. Small JP, Shi L, Kim P. 2003. Mesoscopic
Eng. Thermophys. 889–92 (In Chinese) thermal and thermoelectric measurements
28. Deleted in proof of individual carbon nanotubes. Solid. State
29. Thostenson ET, Ren Z, Chou TW. 2000. Commun. 127:181–86
Advances in the science and technology of 40. Choi ES, Brooks JS, Eaton DL, Al-Haik
carbon nanotubes and their composites: a MS, Hussaini MY, et al. 2003. Enhance-
review. Comp. Sci. Technol. 61:1899–912 ment of thermal and electrical properties
30. Berber S, Kwon Y-K, Tománek D. 2000. of carbon nanotube polymer composites. J.
Unusually high thermal conductivity of Appl. Phys. 94:6034–39
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

carbon nanotubes. Phys. Rev. Lett. 84: 41. Xie H, Lee H, Youn W, Choi M. 2003.
4613–16 Nanofluids containing multiwalled carbon
31. Biercuk MJ, Llaguno MC, Radosavljevic nanotubes and their enhanced thermal con-
M, Hyun JK, Johnson AT, Fischer JE. 2002. ductivities. J. Appl. Phys. 95:4967–71
by Georgetown University on 04/25/13. For personal use only.

Carbon nanotube composites for thermal 42. Keblinski P, Phillpot SR, Choi SUS, East-
management. Appl. Phys. Lett. 80:2767– man JA. 2002. Mechanisms of heat flow
69 in suspensions of nano-sized particles
32. Nan C-W, Birringer R, Clarke DR, Gleiter (nanofluids). Int. J. Heat Mass Trans. 45:
H. 1997. Effective thermal conductivity of 855–63
particulate composites with interfacial ther- 43. Xue Q-Z. 2003. Model for effective ther-
mal resistance. J. Appl. Phys. 81:6692–99 mal conductivity of nanofluids. Phys. Lett.
33. Hone J, Whitney M, Piskoti C, Zettl A 307:313–17
A. 1999. Thermal conductivity of single- 44. Huxtable ST, Cahill DG, Shenogin S, Xue
walled carbon nanotubes. Phys. Rev. B 59: L, Ozisik R, et al. 2003. Interfacial heat
R2514–17 flow in carbon nanotube suspensions. Nat.
34. Hone J, Llaguno MC, Nemes NM, John- Mater. 2:731–34
son AT, Fischer JE, et al. 2000. Electrical 45. Banhart F. 2001. The formation of a con-
and thermal transport properties of magnet- nection between carbon nanotubes in an
ically aligned single wall nanotube films. electron beam. Nano Lett. 1:329
Appl. Phys. Lett. 77:666–68 46. Putra N, Roetzel W, Das SK. 2003. Nat-
35. Fischer JE, Zhou W, Vavro J, Llaguno MC, ural convection of nano-fluids. Heat Mass
Guthy C, et al. 2003. Magnetically aligned Trans. 39:775–84
single wall carbon nanotube films: pre- 47. Khanafer K, Vafai K, Lightstone M. 2003.
ferred orientation and anisotropic transport Buoyancy-driven heat enhancement in
properties. J. Appl. Phys. 93:2157–63 a two-dimensional enclosure utilizing
36. Yi W, Lu L, Zhang DL, Pan ZW, Xie SS. nanofluids. Int. J. Heat Mass Trans. 46:
1999. Linear specific heat of carbon nan- 3639
otubes. Phys. Rev. B 59:R9015–18 48. Lee SP, Choi SUS. 1996. Applications of
37. Borca-Tasciuc T, Hapenciuc CL, Wei BQ, metallic nanoparticle suspensions in ad-
Vatai R, Ajayan PM. 2003. Experimen- vanced cooling systems. In Recent Ad-
tal investigation of temperature anneal- vances in Solids/Structures and Applica-
ing effect on thermophysical properties tion of Metallic Materials, ed. Y Kwon, D
of carbon nanotube arrays. Presented at Davis, H Chung, pp. 227–34. New York:
ASME International Mechanical Engineer- Am. Soc. Mech. Eng.
ing Congress and Exposition, Washington, 49. Xuan Y, Roetzel W. 2000. Conceptions for
DC heat transfer correlation of nanofluids. Int.
38. Che J, Cagin T, Goddard WA III. 2000. J. Heat Mass Trans. 43:3701–7
27 May 2004 5:44 AR AR218-MR34-07.tex AR218-MR34-07.sgm LaTeX2e(2002/01/18) P1: FHD

246 EASTMAN ET AL.

50. Xuan Y, Li Q. 2000. Heat transfer enhance- 61. Yu W, Choi SUS. 2003. The role of interfa-
ment of nanofluids. Int. J. Heat Fluid Flow cial layers in the enhanced thermal conduc-
21:58–64 tivity of nanofluids: a renovated Maxwell
51. Xuan Y, Li Q. 2003. Investigation on con- model. J. Nanopart. Res. 5:167–71
vective heat transfer and flow features of 62. Yu C-J, Richter AG, Datta A, Durbin MK,
nanofluids. J. Heat Trans. 125:151–55 Dutta P. 2000. Molecular layering in a liq-
52. Faulkner D, Khotan M, Shekarriz R. 2003. uid on a solid substrate: an X-ray reflectiv-
Practical design of a 1000 W/cm2 cooling ity study. Phys. B 283:27–31
system. Proc. 19th Ann. IEEE Semicond. 63. Henderson JR, Swol FV. 1984. On the inter-
Thermal Meas. Meas. Symp. 223–30 face between a fluid and a planar wall: the-
53. Deleted in proof ory and simulations of a hard sphere fluid
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

54. Moelle C, Werner M, Szucs F, Wittorf D, at a hard wall. Mol. Phys. 51:991–1010
Sellschopp M, et al. 1998. Specific heat of 64. Xue L, Keblinski P, Phillpot SR, Choi SUS,
single-, poly- and nanocrystalline diamond. Eastman JA. 2004. Effect of liquid layer-
Diam. Relat. Mater. 7:499–503 ing at the liquid-solid interface on ther-
by Georgetown University on 04/25/13. For personal use only.

55. Wang X, Xu X, Choi SUS. 1999. Thermal mal transport. Int. J. Heat Mass Trans.
conductivity of nanoparticles-fluid mix- Submitted
ture. J. Thermophys. Heat Trans. 13:474– 65. Joshi AA, Majumdar A. 1993. Transient
80 ballistic and diffusive phonon heat trans-
56. Pak BC, Cho YI. 1998. Hydrodynamic and port in thin films. J. Appl. Phys. 74:31–39
heat transfer study of dispersed fluids with 66. Wang B-X, Zhou P-P, Peng Z-F. 2003. A
submicron metallic oxide particles. Exp. fractal model for predicting the effective
Heat Trans. 11:151–70 thermal conductivity of liquid with sus-
57. Eastman JA, Choi SUS, Li S, Soyez G, pension of nanoparticles. Int. J. Heat Mass
Thompson LJ, DiMelfi RJ. 1999. Novel Trans. 46:2665–72
thermal properties of nanostructured ma- 67. Wasan DT, Nikolov AD. 2003. Spreading
terials. Mater. Sci. Forum 312–314:629–34 of nanofluids on solids. Nat. Mater. 423:
58. Wilson OM, Hu X, Cahill DG, Braun PV. 156–59
2002. Colloidal metal particles as probes of 68. Ye L, Liu J, Sheng P, Huang JS, Weitz DA.
nanoscale thermal transport in fluids. Phys. 1993. Sound propagation in colloidal sys-
Rev. B 66:224301 tems. J. Phys. IV3(C1):183–96
59. Putnam SA, Cahill DG, Ash BJ, Schadler 69. Loomis JJ, Maris HJ. 1994. Theory of
LS. 2003. High-precision thermal conduc- heat transfer by evanescent electromag-
tivity measurements as a probe of poly- netic waves. Phys. Rev. B 50:18517–24
mer/nanoparticle interfaces. J. Appl. Phys. 70. Yu W, Hull JR, Choi SUS. 2003. Stable
94:6785–88 and highly conductive nanofluids: exper-
60. van der Zanden AJJ. 2001. Simultaneous imental and theoretical results. Presented
heat and mass transfer in heterogeneous at 6th ASME-JSME Thermal Eng. Joint
media. Chem. Eng. Sci. 56:3341–45 Conf. Kolala Coast, Hawaii
P1: FRK
June 5, 2004 0:1 Annual Reviews AR218-FM

Annual Review of Materials Research


Volume 34, 2004

CONTENTS
QUANTUM DOT OPTO-ELECTRONIC DEVICES, P. Bhattacharya, S. Ghosh,
and A.D. Stiff-Roberts 1
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org

SYNTHESIS ROUTES FOR LARGE VOLUMES OF NANOPARTICLES,


Ombretta Masala and Ram Seshadri 41
by Georgetown University on 04/25/13. For personal use only.

SEMICONDUCTOR NANOWIRES AND NANOTUBES, Matt Law,


Joshua Goldberger, and Peidong Yang 83
SIMULATIONS OF DNA-NANOTUBE INTERACTIONS, Huajian Gao
and Yong Kong 123
CHEMICAL SENSING AND CATALYSIS BY ONE-DIMENSIONAL
METAL-OXIDE NANOSTRUCTURES, Andrei Kolmakov
and Martin Moskovits 151
SELF-ASSEMBLED SEMICONDUCTOR QUANTUM DOTS: FUNDAMENTAL
PHYSICS AND DEVICE APPLICATIONS, M.S. Skolnick
and D.J. Mowbray 181
THERMAL TRANSPORT IN NANOFLUIDS, J.A. Eastman, S.R. Phillpot,
S.U.S. Choi, and P. Keblinski 219
UNUSUAL PROPERTIES AND STRUCTURE OF CARBON NANOTUBES,
M.S. Dresselhaus, G. Dresselhaus, and A. Jorio 247
MODELING AND SIMULATION OF BIOMATERIALS, Antonio Redondo
and Richard LeSar 279
BIONANOMECHANICAL SYSTEMS, Jacob J. Schmidt
and Carlo D. Montemagno 315
UNCONVENTIONAL NANOFABRICATION, Byron D. Gates, Qiaobing Xu,
J. Christopher Love, Daniel B. Wolfe, and George M. Whitesides 339
MATERIALS ASSEMBLY AND FORMATION USING ENGINEERED
POLYPEPTIDES, Mehmet Sarikaya, Candan Tamerler,
Daniel T. Schwartz, and François Baneyx 373

vii
P1: FRK
June 5, 2004 0:1 Annual Reviews AR218-FM

viii CONTENTS

INDEXES
Subject Index 409
Cumulative Index of Contributing Authors, Volumes 30–34 443
Cumulative Index of Chapter Titles, Volumes 30–34 445

ERRATA
An online log of corrections to Annual Review of Materials Research
chapters may be found at http://matsci.annualreviews.org/errata.shtml
Annu. Rev. Mater. Res. 2004.34:219-246. Downloaded from www.annualreviews.org
by Georgetown University on 04/25/13. For personal use only.

Potrebbero piacerti anche