Professional Documents
Culture Documents
2, 2013
DOI: 10.1007/s11837-012-0512-0
2012 TMS
MICROSTRUCTURAL ASPECTS
Schematic representation of a syntactic foam
microstructure is shown in Fig. 1a. It mainly con-
sists of three phases: matrix resin, hollow particles,
and porosity. A typical syntactic foam microstruc-
ture is shown in Fig. 1b. In this figure, a glass hol-
low-particle-filled vinyl ester matrix syntactic foam
is shown.16 Some air voids may be present in the
matrix. However, the presence of these voids is
neglected in the present discussion. Most available
references classify syntactic foams as two-phase
microstructures comprising matrix resin and hollow (b)
particles. Studies related to thermal conductivity Fig. 1. (a) Schematic representation of the microstructure of syn-
may benefit from classifying them as three-phase tactic foam and (b) micrograph of a vinyl ester-glass hollow particle
syntactic foam.
material, where hollow particles themselves are
considered two-phase materials comprising the
particle shell and the gas inside the particle.
Hollow particles, also called microballoons or Thin-walled particles are used in synthesizing
microspheres in various references, may have a syntactic foams to obtain benefit of weight saving.
large distribution in diameter and wall thickness. Figure 3 shows a typical glass hollow particle,
Figure 2a shows a scanning electron micrograph of where more than 90% volume is the void. The
a batch of widely used glass particles that are present review will focus on analyzing the available
commercially available. The diameter of particles is theoretical models to understand the effect of par-
observed to vary in the range of 1250 lm. ticle wall thickness and volume fraction on the
Figure 2b schematically illustrates the possible distri- thermal conductivity of syntactic foams.
bution in the particle size, which can be determined Particle size is another important parameter in
using a particle size analyzer. Narrow size ranges can particulate composites. The effect of particle size is
be separated using a sieve shaker. Since these parti- not straightforward to understand. Several theo-
cles are hollow, their wall thickness may also show retical models are based on the unit cell approach,
variation within one batch of particles. Figure 2c which takes a single particle in a unit cell of the
illustrates the possibility that even the particles of the matrix. Change in the size is associated with change
same diameter may have different wall thick- in the particle volume fraction (U) in these models
nesses.1720 Therefore, only size analysis is not enough so these parameters are not independent in these
to characterize their geometrical properties. The wall models. In reality, the change in the particle size at
thickness makes a difference in the true particle den- the same volume fraction level is associated with
sity of the same size particles. The particles of different two parameters: (I) particlematrix interfacial area
wall thicknesses can be separated by density-based and (II) curvature effects leading to stress concen-
classification methods. tration. The effect of particle size and interfacial
236 Gupta and Pinisetty
EXPERIMENTAL TRENDS
Polypropylene Matrix Glass Hollow Particle
Syntactic Foams
The thermal conductivity of polypropylene-glass
syntactic foams was found to decrease with
increasing particle volume fraction.9 The thermal
conductivity of the neat resin (0.24 W/m K) was
higher than any composition of the syntactic foam.
This study tested syntactic foams containing up to
20 vol.% of two types of hollow particles. These
particles, designated as TK35 and TK70, had aver-
age sizes of 35 lm and 70 lm and true particle
densities of 0.68 g/cc and 0.21 g/cc, respectively. The
characteristics of these particles are presented in
Table I, considering the glass density of 2.21 g/cc.
Fig. 2. (a) A sample of widely used glass hollow particles. (b) Illus- Both types of hollow particle have values of g over
tration of possible distribution in the (b) diameter and (c) wall thick- 0.88.
ness of the same diameter hollow particles used in syntactic foams. Within this range of compositions, the decrease in
thermal conductivity of syntactic foams with respect
to particle volume fraction is nearly linear and the
foam containing 20 vol.% particles had thermal
conductivity of 0.191 W/m K.
MODELING
To model the thermal conductivity, syntactic
foams must be considered to have a three-phase
microstructure comprising the matrix resin, hollow
particle shell, and the gas inside the hollow particle. Fig. 4. Unit cell geometry and notations used in Refs. 26 and 28.
Composite sphere or coreshell sphere based models
developed for particulate composites can be easily
applied to syntactic foams.25 The gas-filled void and
to estimate the thermal conductivity of syntactic
the hollow particle shell can be considered as two
foams.26,28 This model is based on rule of mixtures
concentric spheres, and most of the available models
approach. As a result of the linear distribution of
can be modified to suit syntactic foams. Before dis-
temperature, the average thermal conductivity of
cussing the analytical approaches, a preliminary
two regions of only the matrix, marked as Zone 1 in
inquiry can be conducted in the thermal conductiv-
Fig. 4, is given as
ity of syntactic foams based on the properties of the
constituting materials.
k1 kp (1)
A primary insight into the potential trends can be
obtained by evaluating the typical values of thermal where kp is the thermal conductivity of the matrix
conductivities as 0.2 W/m K, 0.179 W/m K, and polymer. The average thermal conductivity of Zone
0.022 W/m K for polypropylene resin, glass particle 2, where matrix, particle shell, and gas are present,
shell, and the gas inside the particles, respec- can be given as
tively.26 In this case, the thermal conductivity of the
particle material is lower than that of the resin, but 1
the difference is very small. On the contrary, the k2 kp Vp kg Vg ka Va (2)
h2 S
thermal conductivity values for vinyl ester, glass
material, and gas are reported as 0.25 W/m K, where Vp, Vg, and Va are the volumes of the polymer
1.1 W/m K, and 0.01 W/m K, respectively, where matrix, hollow particle shell, and gas void, respec-
the thermal conductivity of glass is 4.4 times that of tively, and S is the surface area of the entire cross
the resin.27 In these two cases, the mechanism of section. In addition, kg and ka are the thermal con-
heat transfer inside the syntactic foam microstruc- ductivities of the hollow particle shell material and
ture will be different because the particles will form gas phase, respectively. The thermal conductivities
a preferred path when their thermal conductivity is for Zones 1 and 2 can be used in rule of mixtures to
higher than that of the matrix. It is also previously obtain the effective thermal conductivity of syntac-
reported that the convection effects within the tic foams as28
2 !!1 31
1=3 ! 1=3 1=3 ! !
1 6U 4p 2U q q qg qs
k4 1 2 kp p kg s a
ka kp 5 (3)
kp p 3U 9p qg qa qg qa
hollow particles are expected to be negligible where qg , qa , and qs are the effective densities of the
because of their small size.28 hollow particle shell, gas phase, and hollow particle,
respectively. The single particle models in a cubic
Available Theoretical Models for Thermal unit cell suffer from the disadvantage that the
Conductivity maximum possible particle volume fraction is about
0.52. Also, the predictions start to deviate from
Liang Model the experimental values at high particle volume
Liang proposed a unit-cellbased model (Fig. 4) fractions. In syntactic foams, the weight-saving
containing a single particle embedded in the resin consideration commonly leads to particle volume
238 Gupta and Pinisetty
This modified model successfully captured the Equation 14 is used to determine the effective
experimental trends observed in solid-particle-filled thermal conductivity with respect to the radius ratio
composites at high-volume fractions.15,25 Experi- for hollow glass particles in Fig. 6. The parameters
mental data on thermal conductivity of syntactic used in calculation of kmb are kg = 1.1 W/m K and
foams are very limited; hence, detailed validation of ka = 0.1 W/m K. Thin-walled particles are widely
this model is not yet conducted. It should also be used in manufacturing syntactic foams. It can be
noted that this model presents an implicit rela- observed in the figure that the thermal conductivity
tionship and that it needs to be solved numerically. of thin-walled hollow particles changes rapidly with
g. The effective thermal conductivity of hollow par-
Porfiri Model ticles can be as low as 10% of the thermal conduc-
tivity of the glass material.
Homogenization techniques are attractive in find-
ing the effective properties of composite materials.
A closed-form solution based on homogenization
techniques is available for syntactic foams containing
thin-walled particles in low-volume fractions.27 The
same three-phase microstructure is assumed in this
model with hollow particles modeled as coreshell
particles. The thermal conductivity of syntactic
foams is given as
k 3U kp ka
1
kp 2kp ka U ka kp
(13)
kp 2k2g ka kg k2a
9U 1 g
kg 2kp Ukp ka 1 U 2
Model Validation with Experimental Data Table II. The experimental values of thermal con-
ductivities of eight compositions of syntactic foams
Experimental values of the thermal conductivity
and theoretical predictions obtained from the four
of syntactic foams are available in the published
models are presented in Table III. The difference
literature.9 Available experimental data on poly-
between experimental results and theoretical pre-
propylene-glass syntactic foams are used to validate
dictions is presented in Table IV. It can be noted
the predictions obtained from Liang, Felske, Porfiri,
that predictions obtained from all models are close.
and Park models. The experimental data are avail-
Except for Liangs model for one composition, theo-
able on syntactic foams containing two types of
retical predictions of all models show less than 10%
particles, the details of which are presented in
difference from the experimental values. The limi-
Table I and other input parameters are listed in
tation of this validation scheme is that the experi-
mental data are available only for one matrix and
particle material. Although it is desirable to vali-
Table II. Input parameters used for the validation date the models for data from thick-walled particles,
of theoretical models and for the parametric study such data are not available in the literature because
the weight-saving benefit is lost due to the high
Validation with density of the thick-walled particles. Most available
Parameter Liang and Li Data9 Parametric Study studies have used hollow particles in the range
0.85 < g < 1.3436
kg (W/m K) 0.179 1.1
ka (W/m K) 0.0228 0.1
kp (W/m K) 0.24 0.25
qa (kg/m3) 0.09 1.2 Parametric Studies on Thermal Conductivity
qg (kg/m3) 2210 2540
qs (kg/m3) 680 qg (1 g3 ) A general trend in the thermal conductivity
with respect to g and U can be observed through
Table III. Comparison of experimental values of thermal conductivity with theoretical predictions
k (W/m K)
Syntactic Foam U (%) (Liang and Li)a9 Liang Model Felske Model Porfiri Model Park Model
PP 0 0.24
PP/TK35 5 0.23 0.231 0.227 0.229 0.228
10 0.23 0.222 0.214 0.217 0.217
15 0.21 0.213 0.201 0.207 0.206
20 0.18 0.204 0.189 0.196 0.195
PP/TK70 5 0.22 0.229 0.224 0.226 0.226
10 0.21 0.218 0.208 0.212 0.212
15 0.21 0.208 0.193 0.199 0.199
20 0.19 0.197 0.179 0.187 0.186
a
Experimental values.
Table IV. Difference between experimental values of thermal conductivity with theoretical predictions
Syntactic foam U (%) Liang Model (%) Felske Model (%) Porfiri Model (%) Park Model (%)
PP 0
PP/TK35 5 0 1 0 1
10 4 7 6 6
15 1 4 1 2
20 13 5 9 8
PP/TK70 5 4 2 3 3
10 4 1 1 1
15 1 8 5 5
20 4 6 2 2
A Review of Thermal Conductivity of Polymer Matrix Syntactic FoamsEffect of Hollow Particle 241
Wall Thickness and Volume Fraction
Fig. 7. Variation of normalized thermal conductivity of syntactic foams as a function of hollow particle volume fraction at various radius ratios
obtained from (a) Liang, (b) Felske, (c) Porfiri, and (d) Park models.
parametric studies. Liang, Felske, Porfiri, and Park containing such particles, nor is there interest in
models are used to plot the thermal conductivity such syntactic foams due to their heavy weight.
values with respect to g and U, as shown in Figs. 7 Therefore, these differences have only a little prac-
and 8, respectively. These graphs are plotted for tical significance. It can also be noted in Fig. 7 that
vinyl ester matrix-glass hollow particle syntactic the thermal conductivity of syntactic foams con-
foams. The input parameters for this study are taining solid glass particles can be 2.57 times
given in Table II.27 The theoretical values are plotted higher than that of the matrix resin.
up to U = 0.5 because some of the models use a par- It can also be observed in Fig. 7 that the thermal
ticle inside a cubic unit cell. conductivity increases with volume fraction of thick-
It can be noted in Fig. 7 that the trends obtained walled particles. However, the thermal conductivity
from the Liang, Felske, and Park models are very of syntactic foams containing thin-walled particles
similar. Note that the y-axis is different in each part decreases with increasing U because the volume of
of this figure. The Porfiri model shows notably dif- gas is sufficiently high in their structure. The trends
ferent trends, especially for thick-walled, particle- of decrease and increase in the thermal conductivity
filled syntactic foams. The thermal conductivity of syntactic foams with respect to U for thin- and
values predicted from this model are significantly thick-walled hollow particles are well captured by
higher compared with the other three models for all models.
such syntactic foams. However, neither experimen- The thermal conductivity data are plotted in
tal results are available for syntactic foams Fig. 8 to clearly observe the relation with respect to
242 Gupta and Pinisetty
Fig. 8. Variation of normalized thermal conductivity of syntactic foams as a function of hollow particle radius ratio at various volume fractions
obtained from (a) Liang, (b) Felske, (c) Porfiri, and (d) Park models.
g. Apart from the observations already made in have higher thermal conductivity. Knowing this
Fig. 7, it is interesting to note that the thermal kind of behavior is important because if the U is
conductivity versus g graphs show that the curves monotonically increased to lower the syntactic
for various U values cross over. It should be noted foam density, then the thermal conductivity of
that the crossover does not happen at a single g thin- and thick-walled particle reinforced syn-
value but over a narrow range. The approximate g tactic foams will first become equal and then
values (gc) for crossover are: reverse the trend, which may be undesired by
Liang model: gc: 0.946 the application.
Felske model: gc: 0.883 Weight saving is the most important consideration
Porfiri model: gc: 0.935 in using syntactic foams. Figure 10 shows the profile
Park model: gc: 0.925 of specific thermal conductivity of syntactic foams
The gc values predicted by the Liang, Porfiri, with respect to g and U obtained from the four
and Park models are close to each other. Fig- selected theoretical models. The figure is plotted in
ure 9 shows a close-up of the region g = 0.85 1, the region g = 0.8 1 to show the range relevant to
which belongs to thin-walled particles commonly the syntactic foams parameters used in actual
used in syntactic foams. In the region g < gc, the applications. It can be observed that the profiles
thermal conductivity of syntactic foams in- predicted by various models are different. Validation
creases with U. This behavior reverses in the of these profiles with experimental results can
region g > gc, and syntactic foams with lower U further help in selecting the model that is the best in
A Review of Thermal Conductivity of Polymer Matrix Syntactic FoamsEffect of Hollow Particle 243
Wall Thickness and Volume Fraction
Fig. 9. The region 0.85< g< 1 magnified from Fig. 8 for (a) Liang, (b) Felske, (c) Porfiri, and (d) Park models.
describing the thermal conductivity of syntactic and volume fraction. The following trends are
foams. identified from the results:
The thermal conductivity of syntactic foams
SUMMARY decreases with
Five existing theoretical models that can predict Increasing volume fraction of thin-walled par-
the thermal conductivity of syntactic foams are ticles
reviewed. The basic assumptions, applicability, and Decreasing volume fraction of thick-walled
limitations of these models are discussed. Four of particles
these models are validated with available experi- A critical radius ratio of hollow particles can be
mental results on polypropylene matrix glass hollow found where the thermal conductivity of syntactic
particle syntactic foams. The predictions of all foams is nearly the same as that of the matrix at
models for this available data set are in close all volume fraction levels
agreement. The models are further analyzed by
conducting parametric studies on vinyl ester-glass The availability of these models can help in
syntactic foams to determine thermal conductivity designing syntactic foams with a desired set of
trends with respect to hollow particle wall thickness thermal and mechanical properties.
244 Gupta and Pinisetty
Fig. 10. Thermal conductivity normalized with respect to the syntactic foam density as a function of hollow particle radius ratio at various volume
fractions obtained from (a) Liang, (b) Felske, (c) Porfiri, and (d) Park models.