Vous êtes sur la page 1sur 6

Effective Properties of Carbon Fiber Composites:

EM Modeling Versus Experimental Testing



A. Balzano
+
, I.M. De Rosa
*
, F. Sarasini
*
and M.S. Sarto
*
, Senior IEEE
*
Research Center on Nanotechnology applied to Engineering of Sapienza University (CNIS)
University of Rome Sapienza
Rome, Italy
(Contact author: mariasabrina.sarto@uniroma1.it)

+
Selex S.I. S.p.A
Rome, Italy

Rome, Italy

Abstract This paper discusses the critical issues related to the
practical application of Maxwell-Garnett approach to the
electromagnetic modeling of the effective electric permittivity of
dilute carbon fiber reinforced composites. In fact, recent studies
have highlighted the interest in the use of such composites as
lossy sheet in broad-band dielectric absorbing screen in the
gigahertz range. The a.c. electrical conductivity of carbon fibers
embedded in epoxy resin is determined from the best-fit between
the data obtained from the experimental tests and the ones
obtained from the electromagnetic modeling. Two different
manufacturing processes of the test samples are investigated:
pouring and dipping. It results that, in the frequency-range from
9 GHz to 12 GHz, carbon fibers having length of a few
millimeters embedded in epoxy resin, are characterized by a
constant value of the a.c. electrical conductivity only in case of
sample preparation by pouring

Keywords - Carbon fiber reinforced composite; effective
medium; electromagnetic modeling; electromagnetic absorbing
screen.

I. INTRODUCTION
In the last few years radar absorbing materials (RAMs)
have been widely used in military and civil applications to
reduce radar cross section (RCS) and eliminate both
electromagnetic interference (EMI) and electromagnetic
compatibility (EMC) problems.
Nowadays, conducting polymeric composites made of short
carbon fibers embedded in a polymer matrix are widely used
for these applications since such materials show several
advantages over traditional engineering materials, owing to
their capability of tailoring the dielectric properties [1]-[3].
For example, short carbon fiber composites (SCFCs) permit
to obtain high values of dielectric constant at a low
concentration of filling agent. In particular, varying the
geometry and volume fraction of the carbon fibers, it is
possible to modify and control separately the real and
imaginary parts of the complex effective permittivity of the
composite.
The prediction of the electromagnetic properties of SCFCs
can be carried out by applying the effective medium theory
(EMT), according to which it is possible to approximate the
composite as an homogeneous medium, characterized by an
effective complex permittivity.
Several formulations of the EMT of mixtures are reported
in literature. The Maxwell Garnett (MG) model is particularly
suitable for the EM simulation for dilute SCFCs, characterized
by the volume fraction of the inclusions lower than the
percolation threshold [3]-[7].
One critical aspect in the use of the MG formalism consists
in the proper definition of the values for the complex electrical
permittivity of both the inclusions (i.e. the short carbon fibers)
and the resin, in the frequency range of interest. Experimental
studies have demonstrated that accurate values of the complex
electrical permittivity of the resin can be determined from
measurements of the scattering parameters of rectangular
flanges filled with the neat resin, according to the ASTM
standard D5568-01 [8]-[10]. Nevertheless, the value of the
effective conductivity of the short carbon fibers embedded in
the host medium depends on several factors, such as the
frequency range, the length of the fibers, the curing degree of
the resin, the space distribution of the inclusions in the matrix
[4]-[5].
The aim of this paper is to discuss the practical application
of the MG formula to predict the electromagnetic effective
properties of diluted SCFCs. In particular, the critical issues
related to the definition of the electrical conductivity of the
short fibers embedded in the resin are discussed, considering
the features related to the realization procedure of the test
samples and to the testing procedure.
To this purpose, at first experimental tests are performed in
the frequency range from 9 GHz to 12 GHz, in order to
measure the frequency spectra of the real and imaginary parts
of the effective permittivity of several specimens of SCFC,
characterized by different length and volume fraction.
1-4244-1350-8/07/$25.00 2007 IEEE
Then, the MG model is implemented in order to predict the
effective permittivity of the composite. The equivalent a.c.
electrical conductivity of the embedded fibers appearing in the
formula is computed as the best-fit value from the comparison
between the experimental data and the numerical calculations.
It is demonstrated that the process for the test-sample
preparation affects strongly the effective medium
performances.

II. MAXWELL GARNETT MODEL OF CARBON FIBER
COMPOSITES
Let us consider a composite medium consisting of epoxy
resin reinforced with randomly distributed carbon fibers. The
length of the fibers is in the millimeter range and the diameter
is of a few microns.
At radio frequency, from a few up to about ten gigahertzs,
the composite can be modeled as a homogeneous effective
medium.
The EMT, despite being an approximation, presents the
advantage of simplifying considerably the electromagnetic
analysis of two-phase and multiphase composite materials.
Numerous studies have demonstrated that the MG formula
is particularly suitable to predict the effective complex
permittivity of multiphase mixtures with n sorts of inclusions,
conductive and not conductive, in the dilute limit [3], [6]. MG
formula is expressed as follows:

( )
( )
( )
( )


= =
= =
+

+

+ =
n
i j
b i ij b
ij
b i i
n
i j
b i ij b
b
b i i
b eff
N
N
N
1
3
1
1
3
1
3
1
1
3
1



(1)
where
b
is the complex relative permittivity of the base
dielectric,
i
is the complex relative permittivity of the i
th
sort
of inclusions,
i
is the volume fraction occupied by the
inclusions of the i
th
sort, N
ij
are the depolarization factors of the
i
th
sort of inclusions, the index j=1,2,3 corresponds to the x, y, z
Cartesian coordinates, respectively.
In case the inclusions are carbon fibers, whose shape is well
approximated by thin cylinders, the two depolarization factors
associated with the directions transverse to fiber axis are
assumed equal to , whereas the third one, which is associated
with the direction parallel to fiber axis can be expressed by:

|
.
|

\
|
|
.
|

\
|

d
l
l
d
N ln
2
(2)

where l is the length of the fibers, and d is the diameter.
If the inclusions are of only one type, the expression (1)
assumes the following simplified form:

( )
( )
( )
( )
(
(

+
+
+

(
(

+
+
+

+ =
b f b f b
b f
b f b
b
f b
b
b f
b eff
N
N
N


1
3
1
3
(3)
in which the complex permittivity of the fibers is expressed by:

0


f
f f
j = (4)

f
being the a.c. electrical conductivity of the fibers embedded
in the matrix.
The proper use of the MG formula (3) for material design
purposes relies on the key-aspects described below.
1) The assessment of the diluted mixture condition.
The MG formula is valid in the hypothesis that the carbon
fiber volume fraction is below the percolation threshold
p
,
which can be approximated as:
l d
p
5 . 4 (5)
2) The estimation of the relative complex permittivity of
the base resin (
b
) in the frequency range of interest.
To this purpose, experimental tests can be performed by
using the rectangular waveguide test configuration on samples
of neat resin, in the considered frequency range, according to
the standard ASTM D5568-01 [8].
3) The estimation of the a.c. electrical conductivity of the
carbon fibers embedded in the matrix (
f
).
This is really a critical parameter to define, because in
general it depends on several factors, such as: the type of host
resin, the resin curing degree, the fiber length, the frequency.
4) The space distribution of the fibers in the resin.
MG formula (3) applies in the case of isotropic medium, in
which the fibers are distributed in the three space dimensions
with the same degree of randomness. However, in practical
applications the isotropy of the SCFC depends on the
fabrication process, the geometry of the material specimen, the
fiber length.
The following section discusses the issues related to the
definition of a reliable procedure for the fabrication of SCFC
test samples that are used to estimate the value of the a.c.
electrical conductivity of the carbon fibers from measurements
in rectangular waveguide.

III. MATERIALS AND SAMPLE PREPARATION
In order to characterize the effective electromagnetic
properties of short carbon fiber/epoxy composite by means of
rectangular waveguide measurements, it is important to prepare
test samples characterized by isotropic and random distribution
of the fibers, as well as by suitable values of the fiber volume
fraction so that it is possible to perform correct measurements.
In fact, as discussed in the above section, the effective
relative permittivity of the composite is function of the fiber
volume fraction, length and random distribution. It is very
sensitive to the parameters of the manufacturing process of the
samples.
The carbon fiber/epoxy resin composites are prepared by
dispersing carbon fibers randomly in the matrix, which is the
P25 epoxy system (mix ratio 100:31 by weight) produced by
REA Industries. The relative permittivity of the resin in the
frequency range from 1 GHz up to 18 GHz is equal to 3.05; the
density is equal to 1.09 g/cm
3
.
The fibers used in this work are High Strength Carbon
Fibers produced by SP system; they have Young modulus of
236 GPa, strength of about 3.8 GPa, d.c. electrical conductivity
of 70 kS/m, and diameter of 7 m.
In order to characterize the electromagnetic properties of
the composite in the 9 GHz - 12 GHz frequency band, it is
necessary to fill the rectangular aluminum flanges, having
thickness of 6 mm, with the composite material. Figure 1
shows a photograph of one of these flanges.
The test samples are prepared by mixing the chopped
carbon fibers with the resin. The mixture is mixed for some
minutes until an uniform and homogeneous dispersion is
achieved. At last the hardener is added, and the mixture is
mixed again. Special care should be paid in order to avoid the
inclusion of air bubbles in the mixture. An image of the
fabricated composite is shown in Fig.2 as an example.
In this study, the composite specimens are manufactured
using two different methods to fill the flanges: pouring and
dipping.
The first method consists in pouring the mixture directly
inside the flanges, whilst the second one consists in pouring a
great quantity of mixture into suitable container, and afterwards
in dipping the flange into the container, so that the flange is
completely filled.



X Band X Band



Figure 1. Photograph showing a rectangular aluminum flange for
measurements in the X-band.




Figure 2. Optical microscope image of the manufactured composite.


Several samples of composite with carbon fiber volume
fractions ranging from 10
-4
to 10
-3
and fiber lengths varying
from 1 mm to 4 mm are prepared. By way of example, Table I
shows the characteristics of some of them.

TABLE I. CHARACTERISTICS OF COMPOSITE SAMPLES REALIZED BY
POURING (P) OR BY DIPPING (D).


Sample

Fabrication
method
Fiber volume
fraction
Fiber length
[mm]
P1 P 2.710
-4
4
P2 P 3.010
-4
3
P3 P 6.510
-4
2
D1 D 3.310
-4
1
D2 D 6.610
-4
3




IV. EM TESTING AND A.C. ELECTRICAL CONDUCTIVITY
ESTIMATION
The measurements of the effective permittivity of the
composite samples are performed in the frequency range from
9 GHz to 12 GHz, applying the method described in the
standard ANSI/ASTM D5568-01 [8].
At first, the specimens manufactured by pouring are tested.
Fig.3(a) shows the real and imaginary parts of the relative
effective permittivity of the samples P1, P2, P3. It should be
noted that the real part of
eff
attains the highest values in the
whole frequency range for sample P3, and the minimum ones
for sample P1. Such trend is due to the fact that sample P1 has
the lowest fiber volume fraction (
P1
=2.710
-4
), whereas P3 the
highest one (
P3
=6.510
-4
). Actually, the dielectric behavior of
the composite is affected by the inclusion volume fraction
more than by the aspect ratio of the inclusions.
As regards the imaginary part of
eff
, it is observed from
Fig.3(b) that the shorter the fiber length is, the lower the
absolute value of Im{
eff
} is. It results that sample P1 is the
most conductive, whereas the composite with 2-mm-long fibers
is the least conductive of all, even if it is characterized by the
highest volume fraction. This effect is due to the influence that
the inclusion aspect ratio (d/l) exerts on the percolation
threshold of the composite which, according to (5), assumes
the values of 0.7910
-2
, 1.0510
-2
, 1.5810
-2
for samples P1, P2,
P3, respectively.
As the fiber length increases, the percolation threshold
decreases and for the same value of inclusion volume fraction
the composite is much more conductive, even if its dielectric
properties (i.e Re{
eff
}) remain nearly unchanged.
To compute the a.c. electrical conductivity of the
composites the simplified expression (3) is used assuming a
value 3.05 for the relative permittivity of the dielectric base. A
single value of
f
is found to fit the experimental data of the
samples P1, P2, P3 of Table I. Fig.3(b) shows the computed
frequency spectra of the real and imaginary parts of the
effective relative permittivity. Notice the good agreement with
the measured data of Fig.3(a). The error bars appearing in the
curves show the variations of the computed data associated
with variations of 5% of the estimated value of
f
= 40 kS/m.
Successively, the tests are performed on the samples
manufactured by dipping. As an example, Figs. 4(a) and 5(a)
show the real and imaginary parts of the measured relative
effective permittivity of samples D1 and D2 of Table I,
respectively.
Numerical simulations are performed using the MG
formula (3), in order to find the best-fit value of
f
. It results
that different values of
f
are obtained for samples realized
with different fiber length. In particular, as the fiber length
decreases, the best-fit value of
f
decreases. For instance, the
spectra of Fig.4(b) referring to sample D1 with 1-mm-long
fibers, are obtained for
f
= 3 kS/m. Nevertheless, the best-fit
value obtained for the composite with 3-mm-long fibers (D2) is

f
= 10 kS/m. The computed real and imaginary parts of
eff
are
shown in Fig.5(b).
In both Figs. 4(b) and 5(b) the error bars describe the
minimum and maximum variations of the computed data
corresponding to variations of the estimated value of
f
over
the 5%.
The remarkable variations in the equivalent a.c. electrical
conductivity of the fibers embedded in samples D1 and D2 are
due to the non uniform and anisotropic dispersion of the fibers
during the manufacturing process. In fact, during the dipping of
the flange inside the container with the mixture, it happens that
the fibers follow preferentially the direction of the mixture
flow. It results that the fibers are mainly aligned parallel to the
transverse axis of the flange. Such effect is more pronounced as
the fiber length decreases, since the fiber mobility increases.
According to the test method adopted for the EM
characterization of the composite [8], it results that the fibers
laying in the plane of the flange mainly interact with the field
propagating along the waveguide. In conclusion, better
isotropy of the composite material is achieved manufacturing
the samples by pouring.
Finally, Table II summarizes the estimated values of the
a.c. electrical conductivity of the embedded carbon fibers for
the different samples of Table I.



9 10 11 12
-10
-5
0
5
10
15
P1
P2
P3
imaginary part
P1
P2
P3
real part
e
f
f
e
c
t
i
v
e

r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y

frequency [GHz]


(a)




9 10 11 12
-10
-5
0
5
10
15


P2
P3
P1
P1
P2
e
f
f
e
c
t
i
v
e

r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y
frequency [GHz]


real part
imaginary part
P3



(b)


Figure 3. Frequency spectra of real and imaginary parts of the effective
relative permittivity of the composite material samples P1, P2, P3
manufactured by pouring. a) Measured data. b) Computed data.

D1 D1
imaginary part
real part
e
f
f
e
c
t
i
v
e
r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y
4
3
2
1
0
-1
9 10 11 12
frequency [GHz]

(a)


9 10 11 12
-1
0
1
2
3
4
imaginary part
real part
e
f
f
e
c
t
i
v
e

r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y
frequency [GHz]
D1




(b)

Figure 4. Frequency spectra of real and imaginary parts of the effective
relative permittivity of the composite material sample D1 manufactured by
dipping. a) Measured data. b) Computed data.

V. CONCLUSION
This paper focuses on the use of Maxwell-Garnett formula
for the electromagnetic modeling of carbon fiber composites in
the dilute limit. Such materials are of interest for the realization
of lossy sheet in new wide-band radar-absorbing materials
[11].
The critical issues related to the estimation of the a.c.
electrical conductivity of the fibers appearing in the MG
formula are discussed, considering the key-aspects related to
the practical manufacturing process of the composite material
samples. In fact, the obtained results show a strong dependence
of the a.c. equivalent conductivity of the fibers dispersed in the
dielectric matrix on the method used to realize the sample.
9 10 11 12
-6
-4
-2
0
2
4
6
imaginary part
D2
real part

e
f
f
e
c
t
i
v
e

r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y
frequency [GHz]


(a)


9 10 11 12
-6
-4
-2
0
2
4
6
imaginary part
real part
e
f
f
e
c
t
i
v
e

r
e
l
a
t
i
v
e

p
e
r
m
i
t
t
i
v
i
t
y
frequency [GHz]
D2




(b)

Figure 5. Frequency spectra of real and imaginary parts of the effective
relative permittivity of the composite material sample D2 manufactured by
dipping. a) Measured data. b) Computed data.

TABLE II. ESTIMATED A.C. ELECTRICAL CONDUCTIVITY OF THE
CARBON FIBERS EMBEDDED IN THE SAMPLES OF TABLE I, MANUFACTURED
EITHER BY POURING OR BY DIPPING.

Sample
Fiber electrical conductivity
[kS/m]
P1 40
P2 40
P3 4
D1 3
D2 10

Moreover, a better isotropy of the composite is achieved
when the samples are produced by pouring the mixture into the
test flanges. As for the samples realized by dipping, it is noted
a reduction of the a.c. equivalent electrical conductivity of the
fiber up to one order of magnitude, and furthermore a
considerable dependence on the fiber length, which is related to
the fact that the fibers are mainly oriented along the direction
of the mixture flow during the dipping process.

ACKNOWLEDGMENT
The authors wish to thank Ing. R. Mancinelli of Selex S.I.
S.p.A (Italy) for financially supporting this research, and Prof.
C. Caneva for supervising all aspects related to the samples
preparation process.

REFERENCES
[1] C.P.Neo and Vijay K.Varadan, Optimization of carbon fiber composite
for microwave absorber, IEEE Trans.on EMC, Vol.46, NO.1, February
2004, pp. 102-106.
[2] R.A Tellakula, V.K.Varadan, T.C.Shamy and G.N.Mathur, Carbon
fiber and nanotube based composites with polypyrrole fabric as
electromagnetic absorbers, Smart Mater. Struct.13 (2004), pp 1040-
1044.
[3] M.Koledintseva, K. Rozanov, B. Archambeault, Engineering of
composite media for shields at microwave frequencies, Proc. 2005
IEEE Int. Symp. on EMC, Chicago, Aug. 2005.
[4] G. E. Pike, C.H. Seager, Percolation and conductivity: a computer
study, Physical review B, Vol.10 n4, 1974 : 1421-1434.
[5] D. Toker, D. Azulay, N. Shimoni, I. Balberg, O. Millo, Tunneling and
percolation in metal-insulator composite materials, Physical review B,
Vol.68, 2003.
[6] A.N. Lagarkov, S.M. Matytsin, K.N. Rozanov, A.K. Sarychev,
Dielectric properties of fiber-filled composites, Journal of Applied
Physics, Vol.84, No.7, 1 Oct. 1998, pp. 3806-3814.
[7] S. M. Matitsine, K. M. Hock, L. Liu, Y. B. Gan, A. N. Lagarkov e K. N.
Rozanov, Shift of resonance frequency of long conducting fibers
embedded in a composite. Journal of Applied Physics, 94(2), July
2003, pp.11461154.
[8] ANSI/ASTM D5568-01, Standard test method for measuring relative
complex permittivity and relative magnetic permeability of solid
materials at microwave frequencies, 2001.
[9] J. Baker-Jarvis, M. D. Janezic, J. H. Grosvenor Jr. e R. G. Geyer,
Transmission reflection and short-circuit line methods for measuring
permittivity and permeability. National Institute of Standards an
Technology, Technical Note 1341, 1990.
[10] W.B. Weir. Automatic measurement of complex dielectric constant and
permeability at microwave frequencies, Proc. 1974 IEEE, 62, January
1974, pp.3336.
[11] A. Balzano, C. Caneva, I.M. De Rosa, R. Mancinelli, M. Pagliuca, F.
Sarasini and M.S. Sarto, Electromagnetic design and realization of
innovative fibre-reinforced broad-band absorbing materials, submitted
to IEEE Transactions on Electromagnetic Compatibility.

Vous aimerez peut-être aussi