Composites Science and Technology 122 (2016) 42e49

Contents lists available at ScienceDirect

Composites Science and Technology

journal homepage: http://www.elsevier.com/locate/compscitech

Analytic modeling for the anisotropic thermal conductivity of polymer

composites containing aligned hexagonal boron nitride
Lin Chen a, *, Ying-Ying Sun a, Hong-Fei Xu a, Shao-Jian He b, Gao-Sheng Wei a,
Xiao-Ze Du a, Jun Lin b, **
a

Key Laboratory of Condition Monitoring and Control for Power Plant Equipment of Ministry of Education, North China Electric Power University, Beijing
102206, China
School of Renewable Energy, North China Electric Power University, Beijing 102206, China

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 3 April 2015
Received in revised form
29 October 2015
Accepted 9 November 2015
Available online 14 November 2015

Polymer composites containing two-dimensional hexagonal boron nitride (h-BN) can achieve enhanced
thermal conductivity (TC), and such enhancement is more pronounced along the orientation direction of
h-BN platelets. In this paper, an analytical model was developed for simultaneously calculating the
anisotropic TC of polymer composites containing aligned h-BN platelets. A unit cell was rst abstracted
based on the morphological observation in the literature. Then the thermal resistance method and coordinate transformation were used to derive the TC of the composites. The model was successfully
validated by the experimental data from two independent literature studies over a large range of ller
contents. The dependence of TC on the key parameters of the model, including the orientation angle and
geometric dimension of the h-BNs, and the interfacial thermal resistance, were quantitatively analyzed
and discussed.
2015 Elsevier Ltd. All rights reserved.

Keywords:
Polymer-matrix composites
Thermal properties
Modelling
Anisotropy
Hexagonal boron nitride

1. Introduction
Polymers have been widely used in almost all aspects of modern
industrial eld because they are light, anticorrosive and easy to
process [1]. However, most of polymers have the thermal conductivities (TCs) of only around 0.1e0.3 W/mK [2], limiting their applications in the areas including thermal interface materials [3,4],
anti-corrosion heat exchanger [5,6] and heat sinks in electronic
devices [7,8]. In order to improve the TC of a polymer material,
different kinds of thermally conductive llers, such as metals, metal
oxides and carbon materials, have been utilized to manufacture
polymer composites [9,10]. However, the TCs of the resulting
polymer composites were still below expectation, which were
explained by the poor ller dispersion and the high interfacial
thermal resistance (TR) [11e13].
A structural analog of graphite, hexagonal boron nitride (h-BN),
also appears as one of the most promising llers in the preparation
of thermally conductive and electrically insulative composites, due

* Corresponding author.
** Corresponding author.
E-mail addresses: chenlin@ncepu.edu.cn (L. Chen), jun.lin@ncepu.edu.cn (J. Lin).
http://dx.doi.org/10.1016/j.compscitech.2015.11.013
0266-3538/ 2015 Elsevier Ltd. All rights reserved.

to its unique properties [8,14e18]: on the one hand, it is a good

thermal conductor with the in-plane TC ranging from 200 W/mK
[14] to 400 W/mK [15]; on the other hand, it is an excellent electrical insulator [14]. In addition, the two dimensional h-BN is expected to confer the anisotropic feature on the TCs of composites.
When the h-BN micro-platelets were applied as llers, the in-plane
TC of PVA composites [17] was greatly improved and reached
1.45 W/m K, 3.92 W/m K and 4.41 W/m K at the ller loading of 1 wt
%, 10 wt% and 30 wt%, respectively. In comparison, the throughplane TC of PVA composites was signicantly lower than the inplane TC because the h-BN platelets were mostly oriented towards the in-plane direction. Tanimoto et al. [14] prepared PI
composites with h-BNs of different sizes. For the composites containing the ake-type h-BN with the median diameter of 8 mm, the
in-plane TC can be 17.5 W/mK with the corresponding throughplane TC of 2.2 W/mK. The thermal conduction behavior of h-BN
lled polymer composites has been qualitatively associated with
the geometric dimension, orientation and dispersion of h-BNs
observed morphologically [14,17]. To further understand the
impact of anisotropic h-BN platelets on the TCs of polymer composites, quantitative analysis by the theoretical model is needed.
As the simplest models for calculating the composites TCs, the
Series model and the Parallel model proposed by Voigt-Reuss [19]

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

give the lower and upper bounds respectively for a two-phase

composite, which is considered to consist of the layers of two
phases connected in either series or parallel with regard to the heat
ow direction [9,19,20]. By assuming the composites contain the
very diluted dispersion of non-interacting spheres within a
continuous matrix, Maxwell developed an analytical expression for
the effective TC [19,21], which was later modied by Rayleigh,
McKenzie and Bruggeman [19,22e24] to extend to higher ller
concentration. Among other numerous models that describe the
composite TCs [9,19,21,23,25], the Nielsen model includes the effect
of the shape and the orientation of the particles by introducing
empirical parameters [26,27] and the Agari model considers the
effect of formed conductive chains between the llers [20].
Meanwhile, the composite TCs could also be derived by using the
variational approaches [28e30]. Hashin and Shtrikman obtained
the narrowest possible conductivity bounds for isotropic twophase material in terms of phase volume fractions and phase
properties [28], and the expressions for these bounds mathematically coincide with the Maxwell model [25,28]. Regardless of the
components' volume fractions or TC, these so-called HashinShrtikman bounds always lie within the Series-Parallel bounds
[25]. Duan et al. also used the variational principle to derive the
explicit expressions for effective conductivity of heterogeneous
media, reecting the joint effect of four factors including the
location, orientation and shape of the llers, and the interface
bonding condition [29,30]. Their results are numerically close to
those of Nan et al. [31] for the composites with small volume
fraction of llers.
For the existing models, the TCs of polymer composites are
mainly a function of the ller content. Many other key parameters,
such as the geometric dimension and distribution of the llers,
have been largely overlooked. In the above mentioned polymer
composites with high TC [14,17], the h-BN llers are not only
aligned with each other but also tilted against the heat transfer
direction due to the processing methods. The anisotropic TCs of the
composites have rarely been simultaneously predicted by the
existing models [20e23,26e35]. In this paper, a mathematical
model was developed to simultaneously calculate the anisotropic
TCs of the polymer composites lled with aligned h-BN platelets.
After a unit cell was abstracted according to the morphological
observation in the literature [17], thermal resistance method and
coordinate transformation were used to derive the analytical
model. The calculated anisotropic TCs were found to be consistent
with the experimental results reported in the literatures [14,17].
The inuence of the orientation and geometric dimension of h-BNs
and the interfacial thermal resistance on the TCs of polymer composites were also discussed.
2. Unit cell abstraction
2.1. Selection of the unit cell
As observed in the SEM images of Ref. [17], owing to the shear
force generated by tape-casting, the micro-sized 2D h-BN platelets
in the polymer composites are aligned along the casting direction
with a certain orientation angle. A schematic of the ller distribution in the polymer matrix is shown in Fig. 1.
To abstract a unit cell for the TC calculation of a polymer composite [33,34], the h-BN distribution in the polymer matrix is
simplied as a parallel array which forms an angle of a with the
horizontal direction (i.e., the x axis), as shown in Fig. 2. A unit cell is
generally selected as the red rectangle (in dashed line) in Fig. 2(a),
which has boundaries parallel to that of the polymer composite.
The TC of the unit cell equals to that of the composite if the unit cell
has sufcient number of h-BN platelets, and an iterative process is

43

Fig. 1. Distribution of h-BN in the polymer composite prepared by tape-casting

method, with a zoom-in image to indicate the ller and the matrix.

needed to determine the proper number. In addition, such unit cell

always contains small portions of h-BN platelets, which makes it
difcult to calculate the TC of the unit cell. Alternatively, a unit cell
could be selected as the red rectangle (in dashed line) shown in
Fig. 2(b), containing only one h-BN platelet in the center. The
boundary of the unit cell is parallel to that of the h-BN platelet,
rather than that of the polymer composite. The TC calculation of
such unit cell is much simplied, and the composite TC can be
easily obtained by the mathematic conversion based on the TC of
the unit cell. Therefore, the unit cell in Fig. 2(b) is adopted for
further calculation and analysis in this work.
2.2. Transformation of coordinate systems
In order to convert the calculated TC of the unit cell into the TC
of the polymer composite, a two-coordinate system is utilized, with
the x-y-z coordinate xed with the polymer composite as shown in
Figs. 1 and 2, and the xezeh coordinate xed with the h-BN platelet.
The two coordinate systems are overlapped at rst, as shown in
Fig. 3(a). Then the unit cell rotates around the z axis by an angle of a,
as shown in Fig. 3(b). The parameter a will be used for converting
the TC of the unit cell into the TC of the polymer composite.
3. Calculation of the anisotropic thermal conductivity
3.1. Basic assumptions for the unit cell
Fig. 4 is the three-view diagram of the selected unit cell in the

xezeh coordinate system. The basic assumptions for the unit cell
are made as follows.
(i) The h-BN llers are uniformly distributed, so that the polymer composite is composed of many unit cells with length L,
width W and height H. The h-BN is simplied as a hexagonal
platelet with a mean diameter (the distance between the
parallel sides) of a and a thickness of b, locating in the center
of the unit cell. The values of a and b are usually reported in
the literature [14,17].
(ii) The volume fraction FV of the h-BN in the unit cell equals to
that in the polymer composite, and could be calculated as

p
3 a2 ,b
FV
2 L,H,W

(1)

(iii) The distances between the perimeters of the h-BN and unit
cell are dL, dW and dH, respectively. Therefore L a 2dL ,
H b 2dH , W p2 a 2dW .
3

44

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

Fig. 2. Unit cells of the polymer composite with simplied ller distribution.

Fig. 3. Schematic diagram of h-BN rotation in the two coordinate systems.

Fig. 4. Three-view diagram of the selected unit cell containing the h-BN.

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

To take into account the inuence of interfacial TR, the equivalent thickness of interfacial thermal barrier [35], d, is determined by

d RK ,km

(2)

where km is the TC of the polymer matrix and RK is the Kapitza

resistance. The order of magnitudes of reported Rk are generally
108 m2 K/W [16,35e37].
For the convenience of mathematical treatment, the relationship between dL, dW and dH are expressed as

dL  d 4x ,dH  d
dW  d 4z ,dH  d

(3a)
(3b)

where 4x and 4z are the thickness ratios. Based on the SEM images
of the polymer composites prepared in Ref. [14], it is found that the
thickness ratio 4 varies with the ller volume fraction FV. Therefore, for the convenience of mathematical calculation, the relationships between 4 and FV is assumed as

4x fx ,FV

(4a)

4z fz ,FV

(4b)

where fx and fz are the tting coefcients determined by an iterative process, which will be described in more detail in the following
context. In addition, since the h-BN is assumed to be positioned in
the center of the unit cell, fx and fz are generally the same, and can
be described by f fx fz. As a result, 4 4x 4z .
According to Eq. (1) ~ Eq. (4), once the ller content FV and the
coefcients of fx and fz are given or determined, the parameters of L,
W, H, dL, dW and dH can be solved.

where A is the area of cross section perpendicular to the x direction.

By integrating Eq. (7a) along the x direction,

Za

Za
kx;3 Ax ,dx

Za
kBN;== ABN;x ,dx

km Am;x ,dx

(7b)

the following equation can be reached as

kx;3 ,V3 kBN;== ,VBN;3 km ,Vm;3

p
3 2
2 a b

where V3 H,W,a, VBN;3

(7c) can be further expressed as

(7c)
and Vm;3 V3  VBN;3 . And Eq.

!
p
p
3 2
3 2
a b
kx;3 ,H,W,a kBN;== , a b km , H,W,a 
2
2

(7d)

According to the denition of TR, Rx,3 can be obtained as

Rx;3

a
a



p
kx;3 ,H,W km ,H,W 3 a,b k
BN;==  km
2

(7e)

The overall TR Rx of the unit cell in the x direction is the sum of

the three TRs,

2dL
a



p
km ,H,W km ,H,W 3 a,b k
BN;==  km
2

(5)

(6)

The TC kx of the unit cell along the x direction is then calculated

kx

H,W,Rx 2dL

L
p2HWa

2HWkm 3,abkBN;== km

(9)

Similarly, the TC of the unit cell along the z and h direction, kz

and kh respectively, can be calculated as

kz 2d

Part 3 is the cuboid containing an h-BN platelet. For onedimensional stead state heat transfer, the total heat ow rate
through the composite equals the heat ows through both matrix

(8)

as

km

Part 2 is the same as Part 1, with a TR of Rx,2 equal to Rx,1.

Rx;2 Rx;1

 dT

dT
kx;3 Ax
Qx QBN;x Qm;x  kBN;== ABN;x km Am;x
dx
dx
(7a)

The anisotropic TC of the unit cell can be calculated by the

thermal resistance method [33,34]. To calculate the TC of the unit
cell in the x direction, the unit cell is divided into three parts in
series, as shown in Fig. 5. Part 1 is the polymer matrix with a
thickness of dL and the TR Rx,1 is calculated by

dL
km ,H,W

and ller. Considering that, along the x direction, Part 3 consists of

many thin layers in parallel [33,34], the heat ow rate for each thin
layer is

Rx Rx;1 Rx;2 Rx;3

3.2. Thermal conductivity of the unit cell and the polymer

composite

Rx;1

45

km

kh 2d

p2LHa
3LHkm 3 3=4,abkBN;== km

H
H

km

p2LWb
2LWkm 3,a2 kBN; km

(10)

(11)

Considering that the h-BN platelet shown in Fig. 4 rotates 90

around its center in the xez plane. Since there are equal chances of
these two types of h-BN orientation happening, the in-plane TCs of
the unit cell should be modied as



kx0 kz0 kx kz 2

Fig. 5. Schematic diagram of the thermal resistances in the unit cell in the x direction.

(12)

By applying the TC transformation formulas in different coordinates [38], the in-plane TC k== (in the x direction) and the
through-plane TC k (in the z direction) of the polymer composite
can be solved simultaneously as

46

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

k== kx kx0 ,cos2 a kh ,sin2 a

(13)

k kz kx0 ,sin2 a kh ,cos a2

(14)

3.3. Determination of the tting coefcient f

The determination of tting coefcient of f is an iterative process. One can start with a random value of f to start the process.
According to Eq. (1) ~ Eq. (4), once the ller content FV and the
coefcient f are given, the parameters of L, W, H, dL, dW and dH can be
solved. By substituting these determined values for the corresponding parameters in Eq. (9) to Eq. (14), both in-plane and
through-plane TCs of the polymer composite can be solved and
1
1
recorded as k==;i and k;i , respectively, where the superscript indicates the number of integration times and the subscript i refers to
the various cases when the ller content is different. By comparing
the calculated TCs with the experimental ones of k==e;i and ke;i , the
sum of the squares of these thermal conductivity differences could
P
1
be determined as d1 ni1 k==;i  k==e;i 2 . After changing f by a
selected step and repeating the above-described process, a new d2
could be obtained. Such process is repeated by changing f with the
selected small step each time, until a minimum of d is found. The
corresponding f is chosen as the tting coefcient.
4. Results and discussion
4.1. Model validation
The model is rst validated by the experimental results in
Ref. [17], which studied polyvinyl alcohol (PVA)/h-BN composites
prepared by tape-casting. Two different types of h-BN were
employed, original h-BN (referred to as p-BN in Ref. [17]) and
surface modied h-BN (referred to as m-BN in Ref. [17]). It is
assumed that both types of h-BN have the same dimension parameters and TCs with the values set as follows: the dimension
parameters of h-BN, a and b, are set to 0.6 mm and 0.08 mm,
respectively [17]; the through-plane TC of h-BN kBN, is set to 2 W/
mK [14,15]; as the reported in-plane TC of h-BN kBN,k is in the range
of 200 W/mK [14] to 400 W/mK [15], an average value of 300 W/mK
is chosen for the kBN,k. The TC of the isotropic PVA matrix is set to
0.105 W/mK [17]. Based on the interfacial TR Rk results from the
existing literatures [16,35e37], which reported the values of
7.6  108 m2 K/W [16,36], 8  108 m2 K/W [35] and
109~107 m2 K/W [37], respectively, Rk is set to 5  108 m2 K/W in
this work, which is of the same order of magnitude. The orientation
angle a is set according to the SEM results. As revealed in Fig. 2 of
Ref. [17], the h-BN llers are aligned in the casting direction (i.e., the
x direction) with an a of 15.53 and 20.14 for the composites
containing m-BN and p-BN, respectively. Also from Fig.2 of Ref. [17],
the distance between llers along the x or z direction is much
smaller than the distance along the h direction, and the ratio f is
determined to be 0.1 for the composites containing m-BN and 0.6
for that containing p-BN.
After all the above-described parameters are set and determined, the kk and k of the composites can be calculated simultaneously with the variation of the ller content. As demonstrated in
Fig. 6, almost all the calculated in-plane and through-plane TCs t
well with corresponding experimental results simultaneously for
both types of composites. As for the comparatively large deviation
between the calculated and experimental in-plane TC for the
composites with 20 vol% p-BN, as shown in Fig. 6(b), it may be
attributed to the agglomeration of p-BN in the as-prepared
composites as observed by the SEM image in Fig. 2 of Ref. [17],

Fig. 6. Validation 1: TCs of PVA composites prepared by tape-casting [17].

which causes the experimental TC to be lower than the calculated

one.
The model is further validated by the experimental results from
another literature [14], in which polyimide (PI) composites lled
with h-BN akes were prepared by spin-casting. A high degree of
in-plane orientation of ller particles was observed when the h-BN
llers of high aspect ratios (22 and 9) were employed. These two
types of h-BN were referred to as F(8.0) and F(18.0) in Ref. [14], with
the median diameters of a 8 mm and a 18 mm, respectively. The
in-plane and through-plane TCs of h-BN are set to 300 and 2 W/mK,
respectively [14]. The TC of PI matrix is set to 0.4 W/mK [14]. Since
the orientation of these two h-BNs shown in the SEM (Fig. 6 of
Ref. [14]) was found to be quite similar to that in Ref. [17] (Fig. 2 of
Ref. [17]), the a is set to 15 and 20 for the composites containing
F(8.0) h-BN and F(18.0) h-BN, respectively. The boundary thickness
ratio f is determined to be 3.5 for the composites containing F(8.0)
h-BN and 2.5 for that containing F(18.0) h-BN. As demonstrated in
Fig. 7, the model ts the experimental results quite well. As for the
deviation between the experimental results and the calculation, it
may be attributed to the voids formed in the composite samples.

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

47

respectively, which can be attributed to the aligned distribution of

the h-BN llers with the anisotropic TC.
It is important to point out that the model developed in this
work simultaneously solves both the in-plane and through-plane
TCs of the composites containing aligned h-BN llers, as
compared to most of the available models that could only calculate
the composite TC in one direction [20,21,23,32]. In Ref. [14], two
different models, i.e., the Bruggeman model and the Hatta-Taya
model, were applied separately to calculate the respective inplane and through-plane TC. Moreover, the structure characteristics of all components in the polymer composites are incorporated
into our model, with most of the adjustable parameters that could
be obtained from the experimental measurements. Therefore, the
relationship between the thermal conduction and the structure
parameters of the polymer composites may be better represented
in the model presented here.

4.2. Parameter analyses

In the following section, the discussion will be focused on the
dependence of the anisotropic TC of the composites on several key
parameters in the model, including the orientation angle and
geometric dimension of h-BN and the interfacial thermal resistance. The parameters in the model are set as follows unless
otherwise specied: a 0.6 mm, b 0.08 mm, km 0.105 W/mK,
kBN,k 300 W/mK, kBN, 2 W/mK, Rk 5  108 m2 K/W and
a 15 .

Fig. 7. Validation 2: TCs of PI composite prepared by spin-casting [14].

For example, the calculated TC of the composite with 60 vol% F(8.0)

is relatively higher than the experimental results, which may be
associated with the existence of many voids in the composite as
displayed in the SEM image (Fig.6 of Ref. [14]). These voids will
denitely hinder the thermal transport, resulting in lower experimental TCs than the predicted ones.
Herein, the successful model validation using the experimental
results from two independent studies suggests that the model
could be applied with relatively high universality, since a large
variety of parameters are involved in these two studies, including
two different polymer matrices (PVA and PI), various llers dimensions (from 0.6 mm to 18 mm in diameter) and ller contents (up
to 60 vol%). Also, our model results are compared with the results
calculated by the Parallel model and Series model (i.e. the Voigt and
Reuss limits) [19] and by the Hashin-Shtrikman model (i.e., the
Hashin-Shtrikman bounds) [25,28], as shown in both Figs. 6 and 7.
It can be clearly seen that the experimental results and our calculation results are within both the Voigt and Reuss limits and the
Hashin-Shrtikman bounds. The in-plane and through-plane TCs of
the composites are close to the upper and lower limits/bounds,

4.2.1. Orientation angel a of h-BN

As indicated in Fig. 8, with the increase of orientation angle a of
h-BN, the anisotropic TC of all the composites exhibit sine-waveshaped behavior, regardless of ller content. Such results could be
easily understood considering that h-BN has substantially higher
TC along its in-plane direction than through-plane direction. When
a 0 , the plate-shaped h-BN is aligned parallel to the x axis, thus
resulting in the maximum kk and the minimum k. In comparison,
when a 90 , h-BN is aligned parallel to the z axis, with k reaching
the maximum and kk the minimum. By applying the model
developed in this work, the quantitative analysis of the effect of the
orientation angle of h-BN on the anisotropic TC of the composites
could be very useful in nding the appropriate a in practice,
especially for the multidimensional heat dissipation in various
devices.

Fig. 8. Inuence of orientation angle a of h-BN on the anisotropic TC of the composites.

48

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

4.2.2. Geometric dimensions of h-BN

Fig. 9 shows the inuence of the h-BN diameter on composite
TC. By keeping the aspect ratio of h-BN at a constant value of 10, the
increase of h-BN diameter improves both the in-plane and throughplane TC of the composites until the diameter a of h-BN reaches
about 5 mm, above which the further improvement of composite TC
becomes insignicant. Such phenomenon could be rationalized as
follows: under any xed amount of ller loading, the total contact
area between the ller and polymer matrix decreases when the
ller size increases. Smaller contact area corresponds to lower
interfacial TR. As a result, the composite TC improves when the
ller size increases. However, when the ller size exceeds certain
value, the contribution of interfacial TR to the overall TR becomes so
insignicant that the composite TC is mainly determined by the TCs
of the ller and the polymer matrix. Therefore, although the
interfacial TR continues to decrease when the diameter of h-BN is
larger than 5 mm, the composite TC remains almost unchanged
because of the constant ller loading.
It is commonly recognized that the llers with high aspect ratio
helps to improve the composite TC [39]. As can be seen from Fig. 10,
both the in-plane and through-plane TCs of the composites increase continuously with the increase of the ller aspect ratio when
the ller diameter is xed. Higher ller aspect ratio means larger
number of llers in the composites when the ller diameter and
loading are kept constant. Consequently the amount of polymer
matrix surrounding each h-BN ller is reduced, resulting in higher
effective TC of the unit cell in both in-plane and through-plane
directions. Hence the enhancement in the anisotropic TC of the
composites is observed.
4.2.3. Interfacial thermal resistance
Fig. 11 illustrates that both the kk and k of the composites increases with the decrease of interfacial TR, which is consistent with
current theories on the TC dependence of the interfacial TR
[10e12]. The model results show that, when the interfacial TR is
zero, meaning the existence of perfect interface between the
polymer matrix and the llers, the composite TC can be improved
by a factor of 3e4 compared to that with an interfacial TR of
common value, 5  108 m2 K/W. It is also found that the effect of
the interfacial TR on the composite TC is more pronounced for the
composites with higher ller content. The presence of more llers
in the composites corresponds to the smaller amount of polymer
matrix around the each ller in the unit cell and hence lower TR of
polymer part. As a result, the interfacial TR represents a relatively

Fig. 10. Inuence of ller aspect ratio a/b on the anisotropic TC of the composites, with
the ller diameter a remaining constant.

Fig. 11. Inuence of interfacial TR Rk on the anisotropic TC of the composites.

bigger portion of TR in the composites with higher ller loading

and has stronger effect on the determination of composite TC.
5. Conclusions

Fig. 9. Inuence of h-BN diameter a on the anisotropic TC of the composites, with the
aspect ratio of h-BN remaining constant at 10.

Based on the morphological observation reported in the literature, a theoretical model that integrated the structure characteristics of all components in the polymer composites was developed
to simultaneously calculate the anisotropic TC of the polymer
composites containing aligned h-BN micro-platelets. The model
was successfully validated by the experimental data over a large
range of ller contents (up to 60 vol%) from two independent
studies between which the polymer matrices and the geometric
dimensions of llers are all different. As most of the adjustable
parameters in the model could be obtained from the experimental
measurements, the model presented here may help further understanding the relationship between the thermal conduction and
the structure parameters of the polymer composites.
The dependence of the TCs on the key parameters in the model
was analyzed and discussed. Upon the change of the orientation
angle of h-BNs, the kk and the k of the composites changed in the
opposite way, suggesting an appropriate orientation angle may
need to be found in practical applications for heat dissipation. Both

L. Chen et al. / Composites Science and Technology 122 (2016) 42e49

kk and k of the composites could be simultaneously enhanced by

either the reduction of interfacial TR or the increases in geometric
dimension (either the size or the aspect ratio) of llers. It was found
that the TC enhancement due to the lower interfacial TR is more
pronounced for the composites with higher ller loading, and a
maximum of 3e4 times increase in TCs could be obtained when
there was no interfacial TR.
Acknowledgments
The present work is supported by the National Natural Science
Foundation of China (Grant Nos. 51406052, 51303051, 51376060),
the National Basic Research Program of China (No. 2015CB251503)
and the Fundamental Research Funds for the Central Universities.
References
[1] D.V. Rosato, Plastics Processing Data Handbook, Chapman & Hall, 1997, pp.
613e617.
[2] V. Singh, T.L. Bougher, A. Weathers, Y. Cai, K.D. Bi, M.T. Pettes, et al., High
thermal conductivity of chain-oriented amorphous polythiophene, Nat.
Nanotechnol. 9 (5) (2014) 384e390.
[3] K.M. Shahil, A.A. Balandin, Graphene-multilayer graphene nanocomposites as
highly efcient thermal interface materials, Nano Lett. 12 (2) (2012) 861e867.
[4] B.Y. Cao, J. Kong, Y. Xu, K.L. Yung, A. Cai, Polymer nanowire arrays with high
thermal conductivity and superhydrophobicity fabricated by a nano-molding
technique, Heat. Transf. Eng. 34 (2e3) (2013) 131e139.
[5] C. T'Joen, Y. Park, Q. Wang, A. Sommers, X. Han, A. Jacobi, A review on polymer
heat exchangers for HVAC&R applications, Int. J. Refrig. 32 (5) (2009) 763e779.
[6] L. Chen, Z. Li, Z.Y. Guo, Experimental investigation of plastic nned-tube heat
exchangers, with emphasis on material thermal conductivity, Exp. Therm.
Fluid Sci. 33 (5) (2009) 922e928.
[7] J.P. Cao, J. Zhao, X. Zhao, F. You, H. Yu, G.H. Hu, et al., High thermal conductivity and high electrical resistivity of poly(vinylidene uoride)/polystyrene
blends by controlling the localization of hybrid llers, Compos. Sci. Technol.
89 (2013) 142e148.
[8] K. Kim, J. Kim, Fabrication of thermally conductive composite with surface
modied boron nitride by epoxy wetting method, Ceram. Int. 40 (4) (2014)
5181e5189.
[9] Z. Han, A. Fina, Thermal conductivity of carbon nanotubes and their polymer
nanocomposites: A review, Prog. Polym. Sci. 36 (7) (2011) 914e944.
[10] A.A. Balandin, Thermal properties of graphene and nanostructured carbon
materials, Nat. Mater. 10 (8) (2011) 569e581.
[11] S.T. Huxtable, D.G. Cahill, S. Shenogin, L. Xue, R. Ozisik, P. Barone, et al.,
Interfacial heat ow in carbon nanotube suspensions, Nat. Mater 2 (11) (2003)
731e734.
[12] S. Shenogin, L.P. Xue, R. Ozisik, P. Keblinski, D.G. Cahill, Role of thermal
boundary resistance on the heat ow in carbon-nanotube composites, J. Appl.
Phys. 95 (12) (2004) 8136e8144.
[13] G.W. Lee, M. Park, J. Kim, J.I. Lee, H.G. Yoon, Enhanced thermal conductivity of
polymer composites lled with hybrid ller, Compos. Part A Appl. Sci. Manuf.
37 (5) (2006) 727e734.
[14] M. Tanimoto, T. Yamagata, K. Miyata, S. Ando, Anisotropic thermal diffusivity
of hexagonal boron nitride-lled polyimide lms: effects of ller particle size,
aggregation, orientation, and polymer chain rigidity, ACS Appl. Mater Interfaces 5 (10) (2013) 4374e4382.
[15] F.Y. Yuan, H.B. Zhang, X. Li, X.Z. Li, Z.Z. Yu, Synergistic effect of boron nitride
akes and tetrapod-shaped ZnO whiskers on the thermal conductivity of
electrically insulating phenol formaldehyde composites, Compos. Part A Appl.

49

Sci. Manuf. 53 (2013) 137e144.

[16] Z.Y. Lin, A. McNamara, Y. Liu, K. Moon, C.P. Wong, Exfoliated hexagonal boron
nitride-based polymer nanocomposite with enhanced thermal conductivity
for electronic encapsulation, Compos. Sci. Technol. 90 (2014) 123e128.
[17] B.H. Xie, X. Huang, G.J. Zhang, High thermal conductive polyvinyl alcohol
composites with hexagonal boron nitride microplatelets as llers, Compos.
Sci. Technol. 85 (2013) 98e103.
[18] W.L. Song, P. Wang, L. Cao, A. Anderson, M.J. Meziani, A.J. Farr, et al., Polymer/
boron nitride nanocomposite materials for superior thermal transport performance, Angew. Chem. Int. Ed. Engl. 51 (26) (2012) 6498e6501.
, Modelling of thermal
[19] F. Cernuschi, S. Ahmaniemi, P. Vuoristo, T. M
antyla
conductivity of porous materials: application to thick thermal barrier coatings, J. Eur. Ceram. Soc. 24 (9) (2004) 2657e2667.
[20] Y. Agari, T. Uno, Estimation on thermal conductivities of lled polymers,
J. Appl. Polym. Sci. 32 (7) (1986) 5705e5712.
[21] R.C. Progelhof, J.L. Throne, R.R. Ruetsch, Methods for predicting the thermal
conductivity of composite systems: A review, Polym. Eng. Sci. 16 (9) (1976)
615e625.
[22] D.A.G. Bruggeman, Calculation of various physics constants in heterogenous
substances I dielectricity constants and conductivity of mixed bodies from
isotropic substances, Ann. Phys-Berlin 24 (7) (1935) 636e664.
[23] P. Cheng, C.T. Hsu, The effective stagnant thermal conductivity of porous
media with periodic structures, J. Porous Media 2 (1) (1999) 19e38.
[24] S. Giordano, Effective medium theory for dispersions of dielectric ellipsoids,
J. Electrost. 58 (1e2) (2003) 59e76.
[25] J.K. Carson, S.J. Lovatt, D.J. Tanner, A.C. Cleland, Thermal conductivity bounds
for isotropic, porous materials, Int. J. Heat Mass Transf. 48 (11) (2005)
2150e2158.
[26] L.E. Nielsen, Thermal-conductivity of particulate-lled polymers, J. Appl.
Polym. Sci. 17 (12) (1973) 3819e3820.
[27] L.E. Nielsen, Thermal and electrical-conductivity of 2-phase systems, Industrial Eng. Chem. Fundam. 13 (1) (1974) 17e20.
[28] Z. Hashin, S. Shtrikman, A variational approach to the theory of the effective
magnetic permeability of multiphase materials, J. Appl. Phys. 33 (10) (1962)
3125.
[29] H.L. Duan, B.L. Karihaloo, J. Wang, X. Yi, Effective conductivities of heterogeneous media containing multiple inclusions with various spatial distributions,
Phys. Rev. B 73 (17) (2006).
[30] H.L. Duan, B.L. Karihaloo, Effective thermal conductivities of heterogeneous
media containing multiple imperfectly bonded inclusions, Phys. Rev. B 75 (6)
(2007).
[31] C.W. Nan, R. Birringer, D.R. Clarke, H. Gleiter, J. Appl. Phys. 81 (1997) 6692.
[32] C.W. Nan, Z. Shi, Y. Lin, A simple model for thermal conductivity of carbon
nanotube-based composites, Chem. Phys. Lett. 375 (5e6) (2003) 666e669.
[33] J.Z. Liang, G.S. Liu, A new heat transfer model of inorganic particulate-lled
polymer composites, J. Mater. Sci. 44 (17) (2009) 4715e4720.
[34] L. Chen, Y.Y. Sun, J. Lin, X.Z. Du, G.S. Wei, S.J. He, et al., Modeling and analysis
of synergistic effect in thermal conductivity enhancement of polymer composites with hybrid ller, Int. J. Heat Mass Transf. 81 (2015) 457e464.
[35] C.W. Nan, G. Liu, Y. Lin, M. Li, Interface effect on thermal conductivity of
carbon nanotube composites, Appl. Phys. Lett. 85 (16) (2004) 3549.
[36] Z.Y. Lin, Y. Liu, S. Raghavan, K.S. Moon, S.K. Sitaraman, C.P. Wong, Magnetic
alignment of hexagonal boron nitride platelets in polymer matrix: toward
high performance anisotropic polymer composites for electronic encapsulation, ACS Appl. Mater. Interfaces 5 (15) (2013) 7633e7640.
[37] G.J. Hu, B.-Y. Cao, Thermal resistance between crossed carbon nanotubes:
molecular dynamics simulations and analytical modeling, J. Appl. Phys. 114
(22) (2013) 224308.
[38] Q.S. Yang, B.R. Pu, Advanced Heat Transfer, Shanghai Jiaotong University Press,
1996, p. 5.
[39] A. Mikdam, A. Makradi, S. Ahzi, H. Garmestani, D.S. Li, Y. Remond, Statistical
continuum theory for the effective conductivity of ber lled polymer composites: effect of orientation distribution and aspect ratio, Compos. Sci.
Technol. 70 (3) (2010) 510e517.