Original Article

Proc IMechE Part L:

J Materials: Design and Applications
Conductive materials for polymeric 227(3) 226–242
! IMechE 2012
bipolar plates: Electrical, thermal and Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav

mechanical properties of polyethylene- DOI: 10.1177/1464420712454059

pil.sagepub.com

carbon black/graphite/magnetite blends

Paul Greenwood, Robert H Thring and Rui Chen

Abstract
The elimination of corrosion, reduction of mass and the improvement of mechanical, electrical and thermal conductivity
properties are the main aims to progress bipolar plate technology for polymer electrolyte membrane fuel cells. In
addition, the large numbers of bipolar plates required in automotive fuel cell stacks (on the order of 400 plates per
vehicle) will demand mass production to meet future demands as well as reduce costs through cheap production
processes. Highly conductive, low density, low cost and corrosion-resistant materials that can be utilised in production
processes such as injection and compression moulding are ideal candidates for bipolar plates. Carbon black, graphite,
magnetite with polyethylene composites were made and tested for their electrical and thermal conductivities and
mechanical properties and compared to the US Department of Energy targets for 2015. The carbon black composites
exhibited better electrical conductivity than the other fillers where at 65 wt% the conductivity was 24 S/cm for
through-plane conductivity and had a flexural strength of 32 MPa. Injection moulding produced composites with
more material stability and greater mechanical strength than compression mouldings although compression mouldings
produced composites with higher thermal conductivities where graphite displayed the highest thermal conductivity of
2 W/mK. Although well short of the US Department of Energy targets these materials show promise.

Keywords
Polymer electrolyte membrane fuel cell, bipolar plate, carbon black, graphite, magnetite, conductivity,
mechanical properties

Date received: 27 October 2011; accepted: 13 June 2012

Introduction a thermoplastic polymer to create a cost effective

This research continues from research presented
previously by Greenwood et al.1 where the main
aims were to decrease bipolar plate mass and cost
and to increase mechanical strength whilst providing
sufficient electrical and thermal conductivity. The
study mixed high-density polyethylene (HDPE) with
Monarch 800 carbon black2 suitable for injection
moulding and tested for the aforementioned proper-
ties. Table 1 details the US department of Energy’s
(US DOE) targets for 20153 for automotive polymer
electrolyte membrane fuel cell bipolar plates.
There has been considerable attention paid to fuel
cell bipolar plate research in recent years, metals have
been extensively researched in order to reduce/elimin-
ate their corrosion tendencies as they can be made
thin with good mechanical strength. However, in
addition to the reduction in bipolar plate performance
over time due to corrosion, membrane contamination
by metal ions also occurs. The objective of this
research is to mix different conductive powders with
composite material that exhibits such properties men-
tioned previously to try to meet the 2015 US DOE
requirements using injection and compression mould-
ing techniques. Injection and compression moulding
can reduce manufacturing costs through mass produc-
tion and provide the large volumes required commer-
cially, they are also the current methods employed by
bipolar plate manufacturers such as Eisenhuth
GmbH.4–6 The majority of the literature employed
shear mixing techniques and compression moulding
with few using injection moulding.
Department of Aeronautical and Automotive Engineering,
Loughborough University, UK
Corresponding author:
Paul Greenwood, Department of Aeronautical and Automotive
Engineering, Loughborough University, Loughborough, Leicestershire
LE11 3TU, UK.
Email: pgreenwood000@yahoo.com
Greenwood et al.
Table 1. The US Department of Energy targets for 2015 for
stack and bipolar plate characteristics.
Characteristic 2015 Target
Fuel cell stack Lifetime 5000 h
Power density 650 W/kg
Cost $30/kW
Bipolar plate Electrical conductivity >100 S/cm
Thermal conductivity >20 W/mK
Gas permeability <2  106 cm3/s cm2
Corrosion (current) <1 mA/cm2
Flexural strength >25 MPa
HDPE is a good candidate for the composite
matrix as it is widely used in applications that require
some level of electrical conductivity offering excellent
mechanical properties and good chemical resistance,
for example, resistors, antistatic materials and electro-
magnetic interference shielding.7 In addition HDPE is
readily processed by extrusion, injection moulding
and compression moulding.8 The molecular chains
formed by covalent bonds, however, make it difficult
to transfer free electrons,9 although the electrical
conductivity of such composites generally increases
with increasing filler content, high filler loadings can
be accompanied by a reduction in mechanical proper-
ties if the filler particles are insufficiently wetted.
Difficulties associated with filler dispersion encoun-
tered during composite processing also become more
pronounced at high filler concentrations.10
A significant amount of work has been presented on
the theories and properties of HDPE/conductive filler
composites where results in recent studies have
improved 2–4 orders of magnitude over previous
studies.11–15 Carbon blacks such as N330 and
Conductex 975 mixed with HDPE8,16 have achieved
electrical conductivities up to 0.2 S/cm whilst Printex
XE2-B carbon black/carbon fibre mixes with HDPE
have reached 7.7 S/cm.17 Thermal conductivities and
flexural strengths for carbon blacks in HDPE have
been reported in the order of 0.5 W/mK and 14 MPa
by Kumlutas et al.18 and Liang et al.,9 respectively.
Graphite fillers in HDPE have exhibited better results
where Thongruang et al.19 achieved 11 S/cm from
Asbury 4424 graphite with EG-10,20 GFG5021 and
KS-1522 grades achieving less. Krupa et al.20 reported
the highest thermal conductivity of 2.3 W/mK from
these grades with a similar result reported for copper
powder23 and was only surpassed by the single wall
carbon nanotubes (SWNT) used by Haggenmueller
et al.24 and aluminium powder employed by
Tavman25 of up to 3.67 W/mK.
The fillers used in this study are carbon black,
graphite and magnetite. The materials were tested
for flexural, tensile, thermal and electrical properties
and compared to various bipolar plate materials from
commercially available grades and the literature.
227
Additional testing would be required to gain long-
term performance for PEM fuel cells.
Materials
Carbon blacks are well known for their electrical
conductivity. The same carbon black, the Monarch
800 grade from Cabot Corporation (Ellesmere Port,
Cheshire, UK), used in the previous study was used in
this study. The powder is known for its use in
pigmentation.2
Graphite is also well known for its electrical prop-
erties; the graphite powder used was an unspecified
grade from Erodex Ltd (Halesowen, West Midlands,
UK). It was an industrial grade graphite from the
graphitisation of an oil/polymer mixture.26
Magnetite or iron (II,III) oxide is well known for
its magnetic properties but not for its electrical con-
ductivity, as it is already an oxidised iron and does not
oxidise further. Magnetite contains two valence states
of Fe2þ and Fe3þ, electrical conductivity/electron
hoping occurs from Fe2þ to Fe3þ where the states
change to Fe3þ and Fe2þ accordingly.27
The matrix for the composites needed to be applic-
able mainly to injection moulding where high flow
rate/low viscosity was required when heavily loaded
with filler as this would also be sufficient for compres-
sion moulding. The same HDPE in the previous study
was used here, the ExxonMobil HMA018 HDPE
injection moulding grade with a melting point
higher than fuel cell operating temperatures (80  C)
of 129  C and melt flow index of 30 g/10 min.28
Experimental methods
The processing methods and the electrical conductivity
testing method used for the composites have been
documented elsewhere.1 The samples cut from the
American Society for Testing and Materials (ASTM)
D638 Type I tensile test bars for the electrical in-plane
and through-plane properties of the composites were
measured as before in Greenwood et al.1 The in-plane
measurements were taken as the direction parallel to
the injection moulding flow (test bar length) and the
through-plane measurements taken perpendicular to
the flow (test bar width). This was done to represent
the in-plane and through-plane orientations of the
bipolar plate and not the planes within the filler
materials. The same was done for compression
moulded samples regardless of no flow direction.
Additional material property test procedures are
detailed in the following sections and changes to the
documented methods are given below.
Two roll mill. The same procedure as previously
employed was used for all the composites where the
front and back rollers were heated to 123  C and
128  C respectively and the melt manually churned
228
with a mill knife to produce five 300 g batches of
each composite grade. In this study, more attention
was paid to the manual churning of the polymer
melt with the mill knife to ensure a better filler/
matrix mix.
Ashing test. The temperature and burnout time in
an Uhlig Laudert U24 oven was increased to 500  C
for approximately 75 min (including ramping) for
all composites to ensure full polymer burnout in ambi-
ent air.
Injection moulding. The injection moulded test samples
conformed to ASTM D638 Type I test bars and
were formed using the following injection moulder
setup shown in Table 2. Only one test sample
was created per shot cycle and the injected flow dir-
ection was parallel to the length of the test samples.
Tensile/flexural test. The mechanical tests were carried
out in the same way as before; however, a 2.5 kN load
cell instead of a 10 kN load cell was fitted for more
accuracy. During tensile testing, PE–graphite test
samples greater than 39 wt% and 71 wt% PE–magnet-
ite samples were not tested as they slipped in the jaws
during testing; it is assumed fillers at the surface pro-
vided increased lubrication. Different jaw types,
including rough surface jaws, were applied for
increased clamping friction; clamping force was lim-
ited for the PE–graphite samples due to their brittle-
ness. During flexural testing, samples that did not
fracture were not used in the study as a comparison
with fractured samples would be inaccurate. These
samples lay mainly below 10–30 wt% as can be seen
by the lack of data in the plots displayed in Figure 8
depending on the composite type and manufacturing
process. All grades of magnetite composite did not
fracture.
Table 2. Injection moulding parameters used for the carbon,
graphite and magnetite composites.
Temperature Sprue 250  C
Nozzle 200  C
Zone 1 190  C
Zone 2 180  C
Zone 3 170  C
Mould 30  C
Injection Pressure 100 bar
Metering size 50 mm
Time 0.5 s
Screw diameter 30 mm
Holding pressure 10 bar
Cooling time 30 s
Clamping force 40 tonnes
Proc IMechE Part L: J Materials: Design and Applications 227(3)
Particle size
Particle size measurements of the carbon black filler
was conducted as before using the Malvern Zetasizer
to measure particle sizes in the nanometre range from
2 nm to 3 mm with 0.5% accuracy. Graphite and
magnetite filler measurements were made using a
Malvern Mastersizer to measure particle sizes in the
micron range from 0.05 mm to 500 mm with 0.5%
accuracy. The Mastersizer employed the same laser
diffraction detection method as the Zetasizer where a
volume distribution is generated from scattered light at
an angle that is directly related to particle size which is
translated into a volume and the volume moment
mean, D[4,3], is calculated by D[4,3] ¼ d 4/d 3, (d –
particle diameter calculated from volume). The two
pieces of equipment differed in the sample preparation
and containment, the Mastersizer method was as follows.
The sample holder system incorporated a mechan-
ical stirrer within a container of de-ionised water that
circulated the particulates through the light chamber.
The system was flushed numerous times by draining
and filling the sample chamber repeatedly before use
and after use until all residual material had been
removed. A small amount of dry powder (1.5 mg)
was added to the sample chamber where it would be
dispersed in the de-ionised water and the rotation speed
of the stirrer was set to 2000 r/min. The Mastersizer
took measurements in the same way as the Zetasizer
and an average plot was given of the particle size
distribution.
Density test
The densities of the carbon black, graphite and
magnetite powders were found using a Micromeritics
Helium Pyncnometer 1305. Prior to testing a 5 cm3
sample chamber and cup was used with a helium
supply pressure of 1.6 bar, the vent rate was set so that
the measured rate of pressure decrease would be approxi-
mately 3 psi/s when venting. The equipment was flushed
10 times with helium using a fill pressure of 16 to 18 psi in
the sample chamber to drive off any trapped gases or
moisture in the powder and pycnometer.
For measurements, the fill pressure was increased
to 19.5 psi 0.2 psi. Tests were run 10 times with the
same sample and averaged. In total three fresh sam-
ples of each powder were run using this method and
an average over three samples was obtained.
Compression moulding
Compression moulding of composite plates for test sam-
ples was conducted using a square mould measuring
300 mm  300 mm with a 250 mm  250 mm  3 mm
cavity in the pattern plate. Two hot presses were used,
one for heating and one for cooling using the following
processing parameters shown in Table 3.
Greenwood et al. 229

Silicon mould release agent was sprayed on the top
and bottom plates before mould assembly and the
cavity was filled with pellets so that when the top
plate was placed in position, it rose 6 to 8 mm above
the pattern plate. The melting stage was carried out
separately to the compression stage to reduce air
entrapment by closing the heated hot press platens
enough to touch the mould without compressing it.
PE–Carbon black compression moulded composite
plates above 30 wt% did not survive the cooling/de-
moulding stages and produced fractured sections of
approximately 10 mm  10 mm. Therefore no test
samples of these composite grades were produced.
Laser cutting
ASTM D638 Type I test bars were cut from the
compression moulded composite plates by laser. A
150W coherent diamond carbon dioxide pulsed laser
with a wavelength of 10.6 mm was used. Air was
forced onto the cut to blow away melted material;
the air pressure was set to 5.5 bar. The laser beam
width was set to 1 mm and travel rate set to
10 mm/s. Eight test bars were cut from each plate
along with three circular (20 mm diameter) samples
for thermal conductivity tests.
Thermal conductivity
A Cussons P5687 thermal conductivity apparatus was
used to determine the thermal conductivity of all com-
posite samples. Circular samples of 20 mm diameter
and 3.4 mm thick (0.02 mm) were used for the test
where compression moulded samples were obtained
from the laser cutting as mentioned previously, injec-
tion moulded samples were cut and rounded from the
ASTM D638 test samples. Due to the restrictions of
the equipment only through-plane thermal conductiv-
ity was measured.
The samples were polished down to 1 mm abrasive
grade and then cleaned in an ultrasonic bath. After
drying the samples were then tested with a thin layer
of high thermal conductivity paste (8.7 W/mK) on
each face, Artic Silver 5 thermal paste from Artic
Silver Inc. was used. After each test the thermal
paste was cleaned off the equipment to avoid trapping
air in the thermal paste layer.
Results
Figures illustrated in this section, where applicable,
show combined plots for all composites where the
abbreviations inj and comp are used for injection
and compression moulding respectively. Scatter bars
in the figures relate to the maximum and minimum
data obtained from repeated measurements where the
mean is used for plotting.
Powder characterisation
The density of the carbon black, graphite and magnet-
ite powders found by helium pycnometry were
1.82 g/cm3, 2.24 g/cm3 and 5.02 g/cm3, respectively.
Figure 1 shows the particle size measurements for the
fillers.
The carbon black particle size was in the range
35–500 nm and the D[4,3] mean particle size was
149 nm. The micrographs of the powder shown in
Figure 2(a) and (b) show that the powder was
highly agglomerated ranging from 2 to 25 mm in
agglomerate size. These secondary agglomerates

Table 3. Hot press parameters used for the carbon, graphite and magnetite composites up to 60/70 wt%.

Melting time Hot press Compression

Material/grade (before compression) temperature force/time Cooling time

Polyethylene (PE) 20 min 160 C 20 t/10 min 10 min
PE–carbon (10 wt%) 30 min 170  C 20 t/10 min 10 min
PE–carbon (20 wt%) 30 min 175  C 20 t/10 min 10 min
PE–carbon (30 wt%) 30 min 185  C 20 t/10 min 10 min
PE–carbon (40 wt%) 30 min 200  C 20 t/10 min 10 min
PE–carbon (50 wt%) 30 min 220  C 20 t/10 min 10 min
PE–carbon (60 wt%) 30 min 240  C 20 t/10 min 10 min
PE–graphite (10, 20 wt%) 30 min 170  C 20 t/10 min 10 min
PE–graphite (30 wt%) 30 min 175  C 20 t/10 min 10 min
PE–graphite (40 wt%) 30 min 190  C 20 t/10 min 10 min
PE–graphite (50 wt%) 30 min 210  C 20 t/10 min 10 min
PE–graphite (60 wt%) 30 min 230  C 20 t/10 min 10 min
PE–magnetite (10–30 wt%) 30 min 160  C 20 t/10 min 10 min
PE–magnetite (40–70 wt%) 30 min 170  C 20 t/10 min 10 min
230 Proc IMechE Part L: J Materials: Design and Applications 227(3)

were made up of smaller agglomerates (primary
agglomerates) of a similar size to that measured by
the Zetasizer shown in Figure 2(c) and (d).
Furthermore these primary agglomerates were made
up of carbon black particles of 20 nm in size as can be
seen in Figure 2(d).
When comparing the Zetasizer measurement to the
field-emission gun scanning electron microscope
(FEGSEM) micrographs it was obvious that the
sample preparation technique was able to break up
the secondary agglomerates but not the primary
agglomerates and therefore the true particle size was
not obtained by the Zetasizer.
The graphite D[4,3] mean particle size measured by
the Mastersizer was 447 mm in the range of 150 to
800 mm. The graphite powder micrographs in
Figure 3 show a difference in particle size from the
Mastersizer measurements.

Figure 1. Particle size distributions for carbon black, graphite and magnetite fillers. The carbon black peak particle count is at
145 nm, the graphite peak particle count shown is at 443 mm and the magnetite peak particle count shown is at 63 mm.

Figure 2. Carbon black powder FEGSEM micrographs showing secondary structures (a and b), and primary structures
(c and d).
Greenwood et al. 231

Sizes of approximately 10 to 100 mm with small
graphite shards of approximately 2 mm can be seen.
The particle shape and sizes were not uniform, par-
ticles were either flat plates with sharp irregular
edges or large and boulder-like. It is highly likely
that the Mastersizer measurements may have
measured the large boulder-like particles or
agglomerates.
The Malvern Mastersizer result for the magnetite
powder gave a D[4,3] mean particle size of 71 mm in
the range of 15–150 mm. On the other hand the mag-
netite micrographs in Figure 4 showed particles sizes

Figure 3. Graphite powder FEGSEM micrographs showing mainly flat plate structures (a and b) amongst smaller shards of
graphite (c).

Figure 4. Magnetite powder FEGSEM micrographs showing sharp edged boulder-like particles (a and b) of up to 40 mm in size with
small fragments of approximately 1 to 3 mm (c).
232
of up to 40 mm and small dust-like fragments as low as
1 mm; however, it is possible that only smaller particles
were shown in the micrographs. Just like the graphite,
particle size and shape varied, all particles were of a
boulder-like shape with numerous flat faces.
For all fillers especially non-spherical fillers, the
differences in particle size between the Mastersizer/
Zetasizer measurements and the particle sizes
observed with FEGSEM cannot be directly compared
as the micrograph measurements only represent single
dimension D[1,0] measurements that are not as accur-
ate as the volume derived D[4,3] means. Therefore the
micrographs are a good indication of microstructure
and material compatibility.
Electrical conductivity
Figure 5 shows the plot of through-plane and in-plane
electrical conductivity for the injection moulded
carbon black composite between 40 and 65 wt%,
in-plane injection moulded graphite and in-plane
and through-plane compression moulded graphite
composites between 39 and 57 wt%.
The injection moulded carbon black through-plane
conductivity started higher than the in-plane conduct-
ivity with an exponential increase with increasing
carbon black loading; the in-plane conductivity
overtook the through-plane conductivity at 65 wt%
exhibiting a greater increase. No samples below
40 wt% were conducted and so 40 wt% can be con-
sidered the percolation threshold for the injection
moulded carbon black composites.
The minimum through-plane conductivity was
7.17 S/cm at 40 wt% increasing to 13.31 and
23.88 S/cm for 56 wt% and 65 wt%, respectively.
The in-plane conductivities were 2.59 S/cm, 9.14 S/
cm and 26.58 S/cm at 40 wt%, 56 wt% and 65 wt%,
Figure 5. Electrical conductivity for in-plane and through-plane measurements of the carbon black and graphite composites for
injection and compression mouldings (where sufficient data could be obtained). Magnetite data is not shown due to limited results.
PE: Polyethylene.
Proc IMechE Part L: J Materials: Design and Applications 227(3)
respectively. Scatter ranges for both sets of data col-
lected increased with increasing carbon black loading
showing increasing material instability.
Data for the compression moulded carbon black
composites was limited as there were inadequate
numbers of conductive samples per data point and
not sufficient for a graphical plot. Electrical conduc-
tion for through-plane conductivity started at
0.242 S/cm at 26 wt% and increased to 0.367 S/cm at
32 wt%. There was only one value obtained for
in-plane conductivity and was 0.183 S/cm at 32 wt%.
Similar composites from the literature were
comparable for low filler loadings where no process-
ing was achieved beyond 40 wt% filler. Hindermann-
Bischoff et al.16 mixed Corax (N330) carbon Black
with Finathene 6003 HDPE resulting in an electrical
conductivity of 0.8 S/cm at 35 wt% using a mould at
ambient pressure. Yu et al.8 mixed Conductex 975
Ultra carbon black with Rigidex HD6070 HDPE to
obtain 0.2 S/cm for a 32 wt% compression moulded
sample processed at 48 bar. Farshidfar et al.17
achieved a maximum electrical conductivity of
7.7 S/cm for 30 wt% carbon black/10 wt% carbon
fibre with a matrix of 70/30 wt% HDPE/EPDM
(ethylene–propylene diene monomer). The materials
used were the HDPE 5218 EA, Vistalon 7500
EPDM, Printex XE2-B carbon black and chopped
carbon fibre from Toyo Co, Japan. All measurements
were done with the 4-point probe technique but on the
surface only and are therefore comparable to the
in-plane measurements presented here and so
through-plane conductivities are expected to be
lower. Due to the ability to process above 40 wt%
the conductivities achieved in this study were able to
surpass that of Farshidfar et al.17
The injection moulded graphite composites gave
one set of data for in-plane electrical conductivity
Greenwood et al.
shown in Figure 5, readings below 39 wt% (percolation
threshold) were unobtainable. The in-plane conductiv-
ity increased exponentially from 39 to 57 wt% with
increasing material instability as shown by the scatter
bars. The electrical conductivity at 39 wt% was
0.012 S/cm increasing to 0.049 S/cm at 50 wt% and fur-
ther increasing to 0.203 S/cm at 57 wt%. The data
exhibited minimal material instability until 57 wt%
where the scatter range increased rapidly to þ0.107/
0.040 S/cm. Although this range was fairly low
in terms of electrical conductivity it was consider-
ably greater than the previous scatter ranges
at 39 wt% (þ0.003/0.002 S/cm) and 50 wt%
(þ0.004/0.005 S/cm).
One data point of 0.315 S/cm was measured for the
through-plane conductivity at 57 wt% with a scatter
range of þ0.065/-0.071 S/cm. The lack of readings for
through-plane conductivity up to 57 wt% would sug-
gest some orientation of the filler during injection
moulding in favour of the in-plane conductive direc-
tion; however, the sudden conductivity at 57 wt%
would suggest otherwise.
The electrical conductivites for the compression
moulded graphite composites were similar and ini-
tially increased at a similar rate for both in-plane
and through-plane conductivities until 57 wt%. At
57 wt% the through-plane conductivity increased to
more than double than that of the in-plane conduct-
ivity with a far greater material instability than that of
the in-plane measurements. Again, samples below
39 wt% did not conduct and can be considered the
percolation threshold.
The in-plane conductivity increased from 0.002 S/cm
to 0.033 S/cm to 0.409 S/cm at 39 wt%, 50 wt% and
57 wt%, respectively. The through-plane conductivity
increased from 0.001 S/cm to 0.022 S/cm to 0.850 S/cm
for 39 wt%, 50 wt% and 57 wt%, respectively.
The scatter ranges for in-plane and through-plane
conductivity were very narrow at 39 and 50 wt%. At
57 wt% the scatter ranges increased considerably to
þ0.071/0.214 S/cm and þ0.489/0.427 S/cm for in-
plane and through-plane conductivities, respectively.
The through-plane conductivity was more than
double than that of the in-plane conductivity with a
scatter range greater than its conductivity value. It
was expected that there was no orientation during
compression moulding and this shows poor filler
properties and increased randomness due to the aniso-
tropic conductivity of the filler.
The higher conductivity at 57 wt% of the compres-
sion moulded samples than the injection moulded
samples could be due to the increase in processing
pressure allowing better particle to particle contact.
However, the increase in processing pressure should
have presented higher conductivities at 39 wt% and
50 wt% for compression moulding than injection
moulding but was not the case. Due to the random
orientation and irregular particles sizes and geometry
it can be concluded that the two sets of data could
233
vary widely irrespective of the processing technique
employed. This can be shown with the 57 wt% injec-
tion and compression moulded graphite data where
the scatter ranges of both data sets overlapped.
From the literature Thongruang et al.19 achieved
11 S/cm at 75 wt% for compression moulded samples
at 75 bar using an unspecified HDPE grade from
Celanese and Graphite No 4424 from Asbury measured
with a 4-point probe. Krupa et al.20 used a Stamylan
HD 9089 V HDPE grade with EG-10 graphite to obtain
8.3 S/cm at 60 wt% (compression moulded at 0.22 bar)
measured with a 3-point probe. These 3- and 4-point
probe measurements are equivalent to the in-plane
measurements presented in this article and were consid-
erably higher in conductivity with greater filler loadings
attainable. Bhattacharya et al.7 achieved a conductivity
of a similar magnitude to this study of 0.09 S/cm
(through-plane measurement with two electrodes) but
at only 15 wt% and was compression moulded at
750 bar. A commercial grade HDPE from Reliance
Industry, India and flake shaped graphite powder
from Graphite India were used.
The data gained for the magnetite composites were
limited due to poor electrical conductivity across the
range of samples and thus were not suitable for graph-
ical plots. However, like the compression moulded
carbon black composites; it was a good indicator of
conductivity levels and percolation thresholds.
One injection moulded sample at 71 wt% had an
in-plane conductivity of 1 mS/cm and through-plane
conductivity of 0.03 mS/cm. For the compression
moulded samples, only one sample at 60 wt% con-
ducted with a value of 6.65 mS/cm for through-plane
conductivity but no values were gained for in-plane
conductivity. At 71 wt% the in-plane conductivity was
0.0014 S/cm and the through-plane conductivity was
1.26 mS/cm with scatter ranges of þ0.76/1.0 mS/cm
for in-plane and þ0.5/0.7 mS/cm for through-
plane conductivities.
As a comparison, Lee et al.29 used an unspecified
HDPE grade from Yuwha Petro. Chem. Lab and pre-
pared g-Fe2O3 powder by the reaction of FeCl2 and
(CH2)6N4 solution which was then reduced. They
achieved a greater conductivity of 0.385 S/cm at
87 wt% using compression moulding at 196 bar than
this study; a 4-point probe was used to measure the
in-plane conductivity. Although Lee et al. obtained a
greater conductivity with higher loadings, the conduct-
ivity was still very low and magnetite has proved an
unpopular filler candidate for electrically conductive
composites as indicated by the focus on magnetic prop-
erties of magnetite–polyolefin composites in the litera-
ture rather than electrical conductivity.
Thermal conductivity
The thermal conductivity for the injection and
compression moulded carbon black composites in
Figure 6 showed exponential increases with
234
Figure 6. Thermal conductivity results for injection and compression moulded carbon black, graphite and magnetite composites.
compression moulded samples increasing at a higher
rate than injection moulded samples. The exponential
curve fit and second-order polynomial fit for the com-
pression and injection moulded composites respect-
ively showed a good fit with R2 values greater than
0.96.
The maximum thermal conductivity for the com-
pression moulded carbon black composite was
0.61 W/mK at 32 wt% and 1.09 W/mK for the injec-
tion moulded carbon black composite at 65 wt%. The
minimum conductivities at 0 wt% were 0.36 W/mK
and 0.35 W/mK for the compression and injection
moulded samples, respectively. The compression
moulded samples showed a noticeable increase imme-
diately after 0 wt%; however, the injection moulded
samples only began to increase at 26 wt%.
Although the highest carbon black loadings had
the largest scatter ranges, they showed no trend in
increasing material instability with increasing carbon
black content and were generally narrow in range.
In comparison, the thermal conductivity of pure
HDPE (unspecified grade) found by Kumlutas et al.18
was 0.53 W/mK, approximately 50% higher than in this
study. Spherical copper fillers have achieved
2.09 W/mK23 and aluminium powders 3.67 W/mK25
in HDPE. Haggenmueller et al.24 obtained a thermal
conductivity of 3.5 W/mK at 30 wt% for compression
mouldings (unknown pressure) measured by a differen-
tial thermocouple with reference rods. An unspecified
HDPE grade from Aldrich and single wall nanotubes
(SWNT) supplied by NASA were used exhibiting a
similar trend to the data presented in this study.
Injection and compression moulded thermal con-
ductivity results for the graphite composite showed
consistent increases with compression moulded sam-
ples increasing at a slightly greater rate than the injec-
tion moulded samples. Both data sets showed
noticeable increases after 0 wt% and both curve fits
to the data were second-order polynomials with R2
values close to 1.
Proc IMechE Part L: J Materials: Design and Applications 227(3)
The maximum and minimum conductivities for
the compression moulded samples were 2.17 and
0.36 W/mK, respectively. The maximum and minimum
conductivities for the injection moulded samples were
1.77 and 0.35 W/mK, respectively. Both exhibited
maximums at 57 wt%, the highest graphite content.
Again, the scatter ranges were consistently narrow
and showed no trend in material instability with
increasing graphite content.
Comparable results with similar trends were found
in the literature where Krupa et al.20 used Stamylan HD
9089 V HDPE and EG-10 graphite to gain 2.3 W/mK at
58 wt% (compression moulded at 0.22 bar). Tavman
et al.21 achieved 0.938 W/mK at 7 wt% with Ecophit
GFG50 graphite and BP 5740 3 VA HDPE compres-
sion moulded at 45 bar. Krupa and Chodák22 mixed
KS-15 graphite and BP 5740 3 VA HDPE compression
moulded at 0.23 bar achieving a maximum thermal
conductivity of 1.75 W/mK at 53 wt%. Differential
scanning calorimetry (DSC) and an ISOMET Applied
Precision Instrument were used for the measurements.
The thermal conductivity results for the compres-
sion and injection moulded magnetite composites
showed a similar trend to each other. Both data sets
were almost identical and increased exponentially
with increasing magnetite content. The curve fit for
compression moulded samples started slightly lower
in thermal conductivity than the injection moulded
samples but then increased more rapidly to a max-
imum conductivity of 1.05 W/mK at 71 wt%. The
maximum conductivity for the injection moulded
samples was 0.89 W/mK. The minimum conductivities
remained at 0.35 W/mK and 0.36 W/mK for injection
and compression moulded samples, respectively.
Both curve fits were second-order polynomials with
R2 values of 0.97 and 0.92 for the compression and
injection moulded data, respectively. Although the
scatter bar ranges at 71 wt% for both data sets were
larger on average, there was still no general trend
shown for material instability across the data sets.
Greenwood et al.
Thus far only electrical conductivity and magnetic
property variations with temperature have been
documented, no thermal conductivity data has been
found in the literature relating to magnetite–HDPE
composites for comparison.
Tensile strength tests
Figure 7 shows the tensile strength comparison
between injection and compression moulded samples
with increasing filler content for carbon black, graph-
ite and magnetite composites. The initial addition of
carbon black (11 wt%) to form the composite
increased the tensile strength of the injection moulded
pure polyethylene considerably from 11.19 to
23.32 MPa, it reached a maximum of 23.97 MPa at
26 wt% and then decreased to 16.20 MPa at 65 wt%.
The reduction in tensile strength was due to the reduc-
tion in filler–matrix interfaces and the increase of
filler–filler interfaces with increasing filler content
thus changing the failure mechanism.
The compression moulded samples displayed less
tensile strength than the pure polyethylene compres-
sion moulded samples with a decreasing trend with
increasing carbon black content. The compression
moulded pure polyethylene had a higher tensile
strength of 17.92 MPa compared to the injection
moulded samples but after the addition of car-
bon black decreased to 10.11 MPa at 32 wt%. The
increased strength of the compression moulded pure
polyethylene was due to the increase in processing
pressure of an additional 12 bar resulting in a densely
packed polymer microstructure with greater polymer
chain interlocking.
On the whole, the compression moulded samples
exhibited a greater degree of material instability than
the injection moulded samples as is shown by the length
of the scatter bars where the scatter ranges remained
fairly consistent with increasing carbon black content.
Figure 7. Tensile strength data for injection and compression moulded carbon black, graphite and magnetite composites with
increasing filler content.
235
In the literature Farshidfar et al.17 mixed Printex
XE2-B carbon black and chopped carbon fibre from
Toyo Co, Japan with HDPE 5218 EA HDPE and
Vistalon 7500 EPDM and compression moulded at
230 bar. Their highest tensile strength gained was
13.5 MPa at 25 wt% carbon black showing a steady
increase from lower loadings, the pure HDPE, how-
ever, was untested. Liang et al.9 produced similar
results to this study where the pure HDPE (6366m
from Daqing Petrochemical Company) exhibited a
tensile strength of 11 MPa which steadily increased
to a plateau of 20 MPa at 8 wt% using an unspecified
carbon black filler. HDPE from Union Carbide
(DGDY 6089 mixed with Filterasorb 400 from
Calgon was tested by Malik et al.30 They achieved
30.4 MPa tensile strength at 40 wt% exhibiting a
linear increase from 20.3 MPa (pure HDPE) from
compression moulded samples at 400 bar.
The injection moulded graphite composites increased
greatly from the pure polyethylene result of 11.19 MPa
to 22.65 MPa at 14 wt% and then steadily increased to a
maximum of 27 MPa at 57 wt%. The scatter bars asso-
ciated with the data set were extremely narrow and
showed little variation in the materials’ property.
The compression moulded graphite samples per-
formed somewhat similar to the pure polyethylene
compression moulded samples displaying a relatively
flat plot with the minimum at only 1.58 MPa less than
the pure polyethylene at 17.92 MPa. The only excep-
tion to this was the maximum tensile strength of
21.41 MPa at 23 wt%. The scatter ranges were fairly
large representing a reasonable degree of material
instability except for the measurements at 23 wt%
where the data was very compact (þ0.37/0.4 MPa).
Graphite composites detailed in the literature
achieved similar results. Krupa et al.20 using Stamylan
HD 9089 V HDPE and EG-10 graphite, compression
moulded at 0.22 bar, achieved 30.1 MPa tensile strength
at 60 wt% with the pure HDPE exhibiting 21.3 MPa.
236
Wang et al.31 showed data that increased linearly from
the pure HDPE of 16.75 MPa to 23 MPa at 55 wt%.
Compression mouldings of unspecified pressure using
HDPE 2480, a product of Oilu Petrochemical Com-
pany, and amorphous graphite a product of Hulan
Chuntang.
The injection and compression moulded magnetite
composite data sets, aside from the pure polyethylene
results of 11.19 and 17.92 MPa, respectively, showed
minimal differences. Both plots had maximums at
9 wt% of 18.66 MPa and 20.38 MPa for injection
and compression moulded samples, respectively.
Both plots followed similar downward trends with
increasing magnetite content with the largest separ-
ation between plots being 3.25 MPa at 32 wt%.
The compression moulded samples exhibited
higher tensile strengths than the injection moulded
samples but the injection moulded samples showed
greater material stability with narrower scatter bar
ranges. Unlike the carbon black and graphite
composites, mechanical property data for magnetite
composites could not be found in the literature for
comparison.
Flexural strength tests
Figure 8 shows the flexural results for the injection
and compression moulded carbon black, graphite
and magnetite composites with increasing filler con-
tent. Injection moulded samples fractured at and
above 26 wt% exhibiting a maximum flexural strength
of 35.07 MPa at 40 wt%. The compression moulded
carbon black composite samples fractured at 11, 26
and 32 wt% where the maximum and minimum flex-
ural strengths of 21.89 and 13.40 MPa were at 26 and
32 wt%, respectively.
The compression moulded material showed a con-
sistent degree of material stability throughout the
range; however, the injection moulded grades
showed slightly less material stability with increasing
carbon black content.
Figure 8. Flexural strength data for injection and compression moulded carbon black and graphite composites. No magnetite
composite data was gained for the plot.
Proc IMechE Part L: J Materials: Design and Applications 227(3)
The injection and compression moulded flexural
strengths of the graphite composites showed an
upward trend with increasing graphite content. The
injection moulded samples exhibited higher strengths
with minimum and maximum flexural strengths of
28.79 and 37.66 MPa at 39 and 57 wt%, respectively.
The minimum and maximum flexural strengths for the
compression moulded samples were 21.88 and
29.77 MPa at 29 and 57 wt%, respectively.
The injection moulded samples showed more
material stability than the compression moulded
samples as the scatter bar ranges were extremely
narrow. The relatively large scatter of the compres-
sion moulded samples, the largest scatter range being
þ2.36/2.13 MPa at 50 wt%, showed that some
material instability was present.
All grades of magnetite composite did not fracture
due to the low volume loading achieved during
processing and the high flexibility of the polyethylene
matrix, therefore no data was obtained.
Flexural strength data from the literature for
carbon black and graphite revealed similar outcomes.
Liang et al.9 mixed and Injection moulded (unspecified
pressure) HDPE 6366m from Daqing Petrochemical
Company with an unspecified carbon black to obtain
a linearly increasing flexural strength up to 13.75 MPa
at 8 wt%. Jiang et al.32 used Marlex HXM 50100
HDPE with exfoliated graphene nanoplatelets (GNP)
from XG Science Inc achieving 37 MPa at 28 wt%.
The same polymer was mixed with Ketjenblack
EC-600 JD giving a flexural strength of 25 MPa at
9 wt% both composites were injection moulded at
6 bar.
Discussion
The microstructures of the composites together
with the processing techniques played an important
role in the properties of the samples. The resulting
electrical, thermal and mechanical properties
observed could all be related to the microstructure
Greenwood et al.
and the processing techniques. All tests conducted on
the composites were done to the maximum filler load-
ing and higher filler loadings could not be obtained
without the intervention of other materials such as
compatibilisers.
Overall the carbon black composites performed
better in electrical conductivity than their graphite
and magnetite counterparts with a maximum of
26.58 S/cm for in-plane conductivity for injection
moulded samples. The graphite conductivities were
two orders of magnitude less than the carbon black
grades and the magnetite composites exhibited the
worst conductivities in the microsiemen range.
The lack of readings for the injection moulded
graphite through-plane conductivity would suggest
some orientation of the filler during injection mould-
ing in favour of the in-plane conductive direction.
However, the sudden electrical conduction at the
highest graphite loading outperformed the in-plane
conductivity and would suggest no preferential orien-
tation, also the FEGSEM micrographs33 showed no
obvious alignment in such a direction. It is highly
likely that the data was the result of a poor filler
with anisotropic conductivity.
Throughout, there was no agreement between the
ideal filler content at maximum electrical conductivity
and the ideal filler content at maximum mechanical
strength due to increased brittleness at high filler load-
ings. Nor was there much agreement between the ideal
filler content at maximum flexural and tensile strengths
where peak tensile strengths occurred at lower filler
content compared to peak flexural strengths. This
could be attributed to the difference in failure mechan-
ism between the tensile and flexural tests. There was
agreement on the ideal filler content for electrical and
thermal conductivities where the filler content at
maximum electrical conductivity was equal to the
filler content of the maximum thermal conductivity
due to greater conduction paths being set up with
increasing filler content.
Overall the compression moulded samples produced
higher thermal conductivities than the injection
moulded samples and would have been due to the
increase in processing pressure bringing particles
closer into contact with each other. The best thermally
conductive filler was the compression moulded graphite
grade followed closely by the carbon black grades and
then the magnetite grades. Unlike the graphite and
magnetite composites, the delayed increase in thermal
conductivity with the addition of filler of the injection
moulded carbon black composites demonstrated that
the thermal conductivity was dominated by the polymer
phase. Only at 26 wt% did the filler begin to contribute
to the thermal conduction paths within the material.
The better performance of the graphite over the
carbon black in the thermal conductivity tests could
be attributable to lattice vibration. It is possible that
the increased number of particle to particle interfaces
in the carbon black composites due to many
237
nanoparticles had disrupted the lattice vibrations
throughout the conductive path. Therefore the
larger graphite particles spanning many thousands
of nanometres would have offered better thermal
conduction paths with regards to lattice vibration.
The point of maximum mechanical strength with
regards to weight percentage varied from filler to filler
with graphite performing the best and magnetite per-
forming the worst. This can be explained by adhesion
and wetting between the filler and polymer phase
where better adhesion and wetting would have hin-
dered crack propagation due to the dissipation of
fracture energy on delamination. This can be seen in
the FEGSEM micrographs33 where there was no vis-
ible sign of incompatibility in the graphite composite
structures as was seen with the magnetite composite
structures. The magnetite composite structures would
have allowed fast crack propagation via the air gaps.
From the FEGSEM micrographs33 it is difficult to
see the degree of wetting or adhesion between the
carbon black particles and polymer phase due to the
nanoscale microstructure where the carbon black par-
ticles act as an interstitial phase between the polymer
network. The carbon black particles/agglomerates
appear to be wetted less than the graphite particles
and therefore the mechanical properties peak at low
loadings (26–40 wt%). However, the nature of the
microstructure would have provided better rigidity
due to the (interstitial) smaller particle diameters
creating a more integrated network of filler and poly-
mer where crack propagation would most likely occur
through the filler phase and not the polymer phase.
This is why the maximum mechanical strengths of the
carbon black composites at low loadings are compar-
able to the maximum mechanical strengths of the
graphite composites at high loadings.
Overall the graphite composites exhibited the highest
tensile and flexural strengths for both injection and
compression moulding techniques compared to the
carbon black and magnetite. The lowest performer
being the compression moulded carbon black for tensile
and flexural tests due to unmouldable material above
32 wt% where the material fractured on de-moulding.
There was a general trend of injection moulded
grades exhibiting greater mechanical strength than
the compression moulded grades that can be explained
by agglomeration. During mixing on the two roll mill,
it is likely that any filler agglomerates would not have
been broken up by the low shear forces inherent to the
technique. During injection moulding, the mixing via
the screw and therefore shot filling occurring before
injection would have been able to break up the
agglomerates where compression moulding would
not as there is no mixing involved. The agglomerates
would have allowed faster crack propagation via the
shorter route around the agglomerate at the agglom-
erate–polymer interface where little crack energy
would have been dissipated. Smaller particles (i.e.
non agglomerates) would increase the route taken by
238
the crack and hinder propagation by introducing more
filler–polymer interfaces therefore increasing crack
energy dissipation through increased delamination.
The trend does not extend to magnetite composites
where the mechanical properties, for tensile only
as there was no data for flexural properties, between
the injection and compression mouldings were almost
identical. When viewed under the FEGSEM,33 it can
be seen that no agglomeration is visible and so the
material would have behaved in a similar manner
with the two processing techniques regardless of injec-
tion moulder premixing.
On the whole compression moulding was better
than injection moulding with regards to electrical
and thermal properties but not with respect to mech-
anical properties. This was mainly due to the process-
ing pressures employed, 10 bar for injection moulding
and 22 bar for compression moulding. The increase in
pressure may have created greater material compres-
sion during processing forming a more densely packed
microstructure creating more electrical and thermal
conduction paths.
As a production technique, injection moulding was a
better process due to its fast cycle times and simple pro-
cessing parameters as shown by the differences in
Tables 2 and 3. Largely the injection moulded samples
exhibited better material stability than the compression
moulded samples shown by low scatter bar ranges.
Where samples fractured on mould release from com-
pression moulding, i.e. above 32 wt% for carbon black
composites, it is unknown as to whether injection
moulding would have presented a similar problem if it
too had employed a holding pressure of 22 bar.
Table 4 compares the US DOE targets, literature
values and commercially available products with the
best results gained from the carbon black, graphite
and magnetite composites. All three composites in
Table 4 are the maximum filled grades showing max-
imum electrical and thermal conductivities. The flex-
ural and tensile properties are given for these
maximum filled grades but were not the maximum
values for those composite types; the maximum mech-
anical strengths are shown in parentheses.
The carbon black and magnetite data displayed in
the table are the best results from both injection and
compression moulded data; however, the mechanical
data for the compression moulded graphite at
57 wt% was not the best overall mechanical perform-
ance when compared to the injection moulded data.
The injection moulded graphite at 57 wt% gave a
flexural and tensile strength of 37.66 and 27 MPa
respectively; however, the electrical and thermal
properties of this grade were less than those presented
in Table 4.
Table 4 shows a general trend of low electrical con-
ductivity compared to the US DOE target with
regards to through-plane conductivity. The through-
plane value is considered the more useful value as
current flow through the plate will be in this direction
Proc IMechE Part L: J Materials: Design and Applications 227(3)
rather than parallel to the flat faces of the bipolar
plate (in-plane direction). Where only one conductiv-
ity value is denoted, it is unknown as to whether these
are through-plane or in-plane conductivities. In
comparison to the other values it is likely that they
are in-plane conductivities and so the through-plane
conductivities would therefore be significantly lower
according to the trend.
None of the fillers met the electrical and thermal
conductivity requirements for the US DOE targets
for 2015, at best only approximately 25% of the elec-
trical conductivity and 10% of the thermal conductiv-
ity target was reached with the carbon black and
graphite fillers respectively. Although not a DOE
requirement, the composites displayed the lowest ten-
sile strengths compared to other data available in the
table and a recommendation was made of >16 MPa
based on the carbon black and graphite composites
in this study. The approximate extension at break for
such a tensile strength was 2 mm and was viewed as a
potential weakness should the bipolar plates undergo
this mode of failure. However, the carbon black and
graphite grades did meet the flexural requirement and
was less than the author’s recommended density target.
The unfractured samples of the magnetite compos-
ites do not mean that the material is unsuitable for use
in a fuel cell environment. The material was flexible
and would have adapted to any variations in clamping
force and to a small extent changes in membrane
thickness due to hydration. The electrical results for
the magnetite in the millisiemen and microsiemen
range, however, preclude it from use as a bipolar
plate material especially when the target conductivity
is greater than 100 S/cm.
A lot of the materials from the literature did not
meet any of the crucial US DOE targets i.e. electrical,
thermal and flexural targets, nor did most of the
commercially available materials meet the electrical
targets if through-plane conductivities are considered.
The only materials from the literature and the only
materials that convincingly surpassed the US DOE
targets were the chopped fibre/carbon black powder
bipolar plate material from Besmann et al.48 and the
Conduplate LT-X, MT-X, HT-X from Nedstack. The
chopped fibre/carbon black powder displayed an elec-
trical conductivity near to that achievable with pure
graphite i.e. 3 times greater than the US DOE target
and 7 times the flexural strength requirement; how-
ever, there was no data for the thermal conductivity.
The Nedstack materials at best displayed double the
electrical, thermal and mechanical requirements;
however, as with the chopped fibre/carbon black
powder the electrical conductivity was not specified
as in-plane or through-plane. In-plane conductivities
are generally higher than the through-plane conduc-
tivities (as shown in Table 4).
The Conduplate LT-X, MT-X, HT-X and Schunk
FU4369/HT materials look promising as they have
already surpassed the properties of stainless steel.
Greenwood et al. 239

Table 4. Comparison of bipolar plate properties from commercially available materials and the literature with the best results gained
from the polyethylene carbon, graphite and magnetite composites. All values are a maximum for the particular grade specified unless
further information is specified in brackets.

Material Electrical Flexural Thermal Corrosion Tensile Density

conductivity strength conductivity current/Rate strength (g/cm3)
(S/cm) (MPa) (W/mK) (MPa)

US Dept Of Energy Targets 20153/ >100 >25 >20 <1 mA/cm2 >16d <2.0d
Author’s Recommendations
316 L Stainless Steel 5134 – 16.2735 384 mA/cm2 (H2) 48538 8.038
586 mA/cm2 (Air)36
<100 mm per year37
SS316L – YZU001 Coating37 – – – 0.1 mm per year – –
SS316L – CNT/PTFE Coating39 12 – – – – –
SS316L – TiNxOy Coating40 – – – 2.7 mA/cm2 – –
Aluminium 4.4  103 41 34042 23743 250 mm per year37 9038 2.738
Aluminium - YZU001 Coating37 – – – 0.247 mm per year – –
Graphite 30037 6.938 39044 – – 2.2645
POE-PhA/Graphite/C-Fibre46 16.6 17.6 – – – 1.68
CNT-PET/PVDF 6 Vol.% 0.059a/0.011b,47 – – – 3448 –
CNT-PET 12 Vol.% 0.2548 – – – – –
Chopped Fibre/Carbon powder49 200–300 175 MPa – – – 0.96
(26 MPa)
Conduplate LT-X, MT-X, HT-X50 167–217 40–45 28–40 – – 1.6–2.0
Schunk FU4369/HT51 111a/52.6b,c 40 55 – – 1.9
Eisenhuth PPG 864 55.6a/18.2b,c 40 14 – – 1.85
Eisenhuth BMA 55 100a/20b,c 40 20 – – 2.1
Eisenhuth BBP 46 200a/41.7b,c 40 20 – – 1.97
BMC 94052 70 38 13.4 – – 1.89
65 wt% PE–Carbon (i) 26.58a/23.88b 32.34 1.09b – 16.2 1.52e
(35.07 (23.97
at 40 wt%) at 26 wt%)
57 wt% PE–Graphite (c) 0.409a/0.85b 29.77 2.17b – 16.34 1.69e
(21.41
at 23 wt%)
71 wt% PE–Magnetite (c) 0.0014a/ – 1.05b – (20.38 3.84e
1.26  106b at 9 wt%)
a
In-plane conductivity.
b
Through-plane conductivity.
c
Conductivity converted from resistivity.
d
Author’s recommendation (not DOE target).
e
Calculated from ashing tests.
(i)/(c) – Injection/Compression grade.
CNT-PET/PVDF: Carbon nanotube–Polyethylene terephalate/Poly vinylidene difluoride; POE-PhA: Polyester (resin)–Phenolic modified alkyd (resin).

Stainless steel, however, is still a material for the
future as it maintains good mechanical strength at
very low bipolar plate thicknesses which increases
fuel cell power density. The chopped fibre/carbon
black powder material, although offers great perform-
ance, was labour intensive to produce and would not
meet future mass production requirements and so is
not a very promising material for the future with
regards to cost.
The choice of materials for bipolar plate applica-
tions would depend mainly on cost, volume, mass
production and longevity requirements and so both
polymeric and metallic materials would satisfy
future demands in differing ways based on specific
end user applications.
Conclusion
The low cost carbon black, graphite and magnetite
fillers in this study were mixed in a polyethylene
matrix and tested for electrical and thermal conduc-
tivities as well as tensile and flexural mechanical
properties. Fast manufacturing techniques such as
injection and compression moulding were employed
to create further potential for low cost through mass
production.
240
Due to the difference between the actual filler con-
tents of various grades and their intended compositions,
the two roll mill mixing process was not entirely reliable
but more diligent mixing yielded improved results when
compared to the previous study. The two roll mill pro-
cess, however, is not a mass production process and
would not offer suitable throughput if coupled with
injection moulding. The somewhat non-spherical
shape of the graphite powder was assumed to display
some degree of alignment in the injection moulding dir-
ection; however, the FEGSEM micrographs33 showed
no obvious alignment and results showed no preferen-
tial orientation [33]. The data was assumed to be the
result of a poor filler with anisotropic properties.
Overall:
. The carbon black composites performed better in
electrical conductivity for injection moulded
samples due to the closer packing of the relatively
smaller particles providing greater physical inter
particle contact.
. The injection moulding grades displayed greater
mechanical properties due to reduced agglomer-
ation during pre-mixing lengthening fracture
paths and therefore increasing fracture energy
dissipation.
. Graphite exhibited better mechanical properties due
to better adhesion and wetting in the polyethylene
matrix, the FEGSEM micrographs showed no sign
of incompatibility of the graphite. This would have
hindered crack propagation through increased
dissipation of fracture energy on delamination.
. Compression moulded samples produced higher
thermal conductivities due to the increase in pro-
cessing pressure bringing particles closer together.
. The best thermally conductive filler was the
compression moulded graphite where the better
performance was attributable to lattice vibration
and large particle size.
Three companies, Schunk, Eisenhuth and BMC
have commercially available bipolar plate materials.
Some of these materials have injection moulding
capability and most have compression moulding cap-
ability all with the aim of high volume, mass produc-
tion. Although these materials show promise and are
suitable for current commercial use, they still fall
short of US DOE 2015 targets with regards to elec-
trical conductivity as the important through-plane
conductivities are low.
The Conduplate series of materials from Nedstack
show promise; however, although electrical conduct-
ivity details provided in the literature are higher than
most of the commercially available materials, the
literature does not distinguish between in-plane and
through-plane conductivity.
In comparison, the best material investigated in
this study displayed less electrical and thermal con-
ductivity than the commercially available Schunk
Proc IMechE Part L: J Materials: Design and Applications 227(3)
and Nedstack materials with a sufficient mechanical
strength for the US DOE requirement. The cheap
materials and thermoplastic matrix employed in this
study produced a material that gave injection
mouldability with less than a 60 s cycle time and
approximately a quarter of the US DOE electrical
conductivity requirement.
To improve the materials, additives such as compa-
tibilisers added to the composite (pre-mixed with the
filler) to lower the difference in surface energy between
filler and matrix would allow improved wetting and
adhesion between filler and matrix to reduce crack
propagation. Compounding could be employed to
breakup agglomerates and improve mixing. Different
holding pressures during injection moulding should be
employed to further investigate the balance of elec-
trical/thermal conductivity and mechanical strength.
The results of the various composites did not indicate
that further improvements in processing could reach
the US DOE targets; however, the carbon black
composites could be improved in electrical conductiv-
ity to rival that of stainless steel as the factor limiting
the filler loading was processing. Additional fillers such
as carbon nanotubes could provide greater results,
though at greater cost; however, improved processing
would still be required to ensure homogeneity of any
multi filler composite. Future work will focus on
modelling of the results presented in this article.
Funding
This project was solely funded by the Department
of Aeronautical and Automotive Engineering at
Loughborough University, UK.
References
1. Greenwood P, Chen R and Thring RH. Polyethylene-
carbon material for polymer electrolyte membrane fuel
cell bipolar plates. Proc IMechE, Part L: J Materials:
Design and Applications 2008; 222: 197–208.
2. Cabot Corporation. Monarch 800 Product data sheet,
http://www.cabot-corp.com/wcm/download/en-us/sb/
MONARCH_8001.pdf (accessed 12 April 2007).
3. US Department of Energy. Hydrogen, fuel cells & infra-
structure technologies program - Multi-year research,
development and demonstration plan: Planned Program
Activities for 2005-2015. Washington DC: US
Department of Energy, 2007.
4. Eisenhuth GmbH. Eisenhuth datasheet for PPG86 bipolar
plate material, http://www.eisenhuth.de/pdf/SIGRACET_
Datenblaetter.pdf (accessed 27 May 2010).
5. Eisenhuth GmbH. Eisenhuth datasheet for BMA5 bipolar
plate material, http://www.eisenhuth.de/pdf/SIGRACET_
Datenblaetter.pdf (accessed 27 May 2010).
6. Eisenhuth GmbH. Eisenhuth datasheet for BBP4 bipolar
plate material, http://www.eisenhuth.de/pdf/SIGRACET_
Datenblaetter.pdf (accessed 27 May 2010).
7. Bhattacharya S, Tandon RP and Sachdev VK. Electrical
conduction of graphite filled high density polyethylene
composites; experiment and theory. J Mater Sci 2009;
44(9): 2430–2433.
Greenwood et al.
8. Yu JQ, Zhang L, Rogunova M, et al. Conductivity of
polyolefins filled with high-structure carbon black.
J Appl Polym Sci 2005; 98: 1799–1805.
9. Liang JZ and Yang QQ. Mechanical properties of car-
bon black-filled high-density polyethylene antistatic
composites. J Reinf Plast Compos 2009; 28(3): 295–304.
10. Thongruang W, Balik CM and Spontak RJ. Volume-
exclusion effects in polyethylene blends filled with
carbon black, graphite, or carbon fiber. J Polym Sci
Part B: Polym Phys 2002; 40(10): 1013–1025.
11. Narkis M, Ram A and Flashner F. Electrical properties
of carbon black filled polyethylene. Polym Eng Sci
1978; 18(8): 649–653.
12. Weber M and Kamal MR. Estimation of the volume
resistivity of electrically conductive composites. Polym
Compos 1997; 18(6): 711–725.
13. Mamunya EP, Davidenko VV and Lebedev EV. Effect
of polymer-filler interface interactions on percolation
conductivity of thermoplastics filled with carbon
black. Compos Interf 1997; 4(4): 169–176.
14. Geuskens C, Gielens JL, Geshef D, et al. The electrical
conductivity of polymer blends filled with carbon-black.
Eur Polym J 1987; 23(12): 993–995.
15. Barton RL, Keith JM and King JA. Electrical conduct-
ivity model evaluation of carbon fiber filled liquid
crystal polymer composites. J Appl Polym Sci 2007;
106(4): 2456–2462.
16. Hindermann-Bischoff M and Ehrburger-Dolle F.
Electrical conductivity of carbon black–polyethylene
composites: experimental evidence of the change of
cluster connectivity in the PTC effect. Carbon 2001;
39: 375–382.
17. Farshidfar A, Haddadi-Asl V and Nazokdast H.
Electrical and mechanical properties of conducive
carbon black/polyolefin composites mixed with carbon
fiber. J ASTM Int 2006; 3(10).
18. Kumlutas D and Tavman IH. A numerical and experi-
mental study on thermal conductivity of particle filled
polymer composites. J Thermoplast Compos Mater
2006; 19: 441–455.
19. Thongruang W, Spontak RJ and Balik CM. Correlated
electrical conductivity and mechanical property analysis
of high density polyethylene filled with graphite and
carbon fibre. Polymer 2002; 43: 2279–2286.
20. Krupa I, Novák I and Chodák I. Electrically and ther-
mally conductive polyethylene/graphite composites and
their mechanical properties. Synth Metals 2004; 145:
245–252.
21. Tavman I, Aydogdu Y, Kök M, et al. Measurement of
heat capacity and thermal conductivity of HDPE/
expanded graphite nanocomposites by differential scan-
ning calorimetry. Arch Mater Sci Eng 2011; 50(1): 56–60.
22. Krupa I and Chodák I. Physical properties of thermo-
plastic/graphite composites. Eur Polym J 2001; 37:
2159–2168.
23. Tekce HS, Kumlutas D and Tavman IH.
Determination of the thermal properties of polyamide-
6 (nylon-6)/copper composite by hot disk method. In:
Proceedings of the 10th Denizli material symposium,
Denizli, pp. 296–304.
24. Haggenmueller R, Guthy C, Lukes JR, et al. Single
wall carbon nanotube/polyethylene nanocomposites:
thermal and electrical conductivity. Macromolecules
2007; 40: 2417–2421.
241
25. Tavman IH. Thermal and mechanical properties
of aluminum powder filled high-density polyethyl-
ene composites. J Appl Polym Sci 1996; 62(12):
2161–2167.
26. Acres A. Personal communication. Erodex Ltd, UK,
2006.
27. Fazekas P. Series in modern condensed matter physics.
Vol 5: Lecture notes on electron correlation and magnet-
ism. London: World Scientific Publishing, 2003.
28. Exxon Mobil. HMA018 high density polyethylene data-
sheet, http://exxonmobilchemical.ides.com/datasheet.
aspx?I¼58933&PS¼COMBO&E¼51687 (accessed 5
April 2007).
29. Lee SH, Heo G, Kim KH, et al. The electrical proper-
ties of magnetite-loaded polyethylene composites.
J Appl Polym Sci 1987; 34: 2537–2545.
30. Malik TM, Carreau PJ, Grmela M, et al. Mechanical
and rheological properties of carbon black filled poly-
ethylene. Polym Compos 1988; 9(6): 412–418.
31. Wang Q, Gao J, Wang RU, et al. Mechanical and rheo-
logical properties of HDPE/graphite composite with
enhanced thermal conductivity. Polym Compos 2001;
22(1): 97–103.
32. Jiang X and Drzal LT. Properties of injection molded
high density polyethylene nanocomposites filled with
exfoliated graphene nanoplatelets. Polym Compos
2010; 31(6): 1091–1098.
33. Greenwood P, Chen R and Thring RH. Polymeric
bipolar plates for PEM fuel cells: experimental and mod-
eling approach to assess factors influencing performance.
PhD Thesis, Aeronautical and Automotive
Engineering, Loughborough University Institutional
Repository. 2010. http://hdl.handle.net/2134/7249.
34. Wind J, Spah R, Kaiser W, et al. Metallic bipolar plates
for PEM fuel cells. In: 7th Ulmer Elektrochemische
Tage. Fuel Cell Education and Training Centre Ulm,
Ulm, Germany, 26–27 June 2000. Elsevier Science B.V,
pp. 256–260.
35. Bruno TJ. Handbook of basic tables for chemical ana-
lysis. 2nd edn. Florida: CRC Press LLC - Taylor &
Francis LLC, 2004.
36. Brady MP, Weisbrod K, Zawodzinski C, et al.
Assessment of thermal nitridation to protect metal
bipolar plates in polymer electrolyte membrane fuel
cells. Electrochem Solid-State Lett 2002; 5(11):
245–247.
37. Lee S, Huang C and Chen Y. Investigation of PVD
coating on corrosion resistance of metallic bipolar
plates in PEM fuel cell. J Mater Process Technol
2003; 140(1-3 SPEC): 688–693.
38. Carvill J. Mechanical engineers data handbook. Oxford:
Butterworth-Heinemann, 1999.
39. Show Y and Takahashi K. Stainless steel bipolar plate
coated with carbon nanotube (CNT)/polytetrafluoro-
ethylene (PTFE) composite film for proton exchange
membrane fuel cell (PEMFC). J Power Sources 2009;
190: 322–325.
40. Kim SY, Han DH, Kim JN, et al. Titanium oxynitride
films for a bipolar plate of polymer electrolyte
membrane fuel cell prepared by inductively coupled
plasma assisted reactive sputtering. J Power Sources
2009; 193: 570–574.
41. Oh M-H, Yoon Y-S, Park S-G, et al. The electrical
properties of aluminum bipolar plate for PEM fuel
242
cell system. Trans Electr Electron Mater 2004; 5(5):
204–207.
42. Callister WD. Materials science and engineering: an
introduction. 4th ed. Oxford: John Wiley & Sons Ltd,
1997.
43. Harper CA. Electronic materials and process handbook.
3rd ed. New York: McGraw Hill, 2003.
44. Pierson HO. Handbook of carbon, graphite, diamond and
fullerenes. New Jersey: Noyes Publications, 1993.
45. Atkins PW. Physical chemistry. 6th ed. Oxford: Oxford
University Press, 1998.
46. Bhlapibul S and Pruksathorn K. Preparation of graph-
ite composite bipolar plate for PEMFC. Korean J Chem
Eng 2008; 25(5): 1226–1231.
47. Wu M and Shaw L. Electrical and mechanical behav-
iors of carbon nanotube-filled polymer blends. J Appl
Polym Sci 2006; 99: 477–488.
Proc IMechE Part L: J Materials: Design and Applications 227(3)
48. Wu M and Shaw LL. On the improved properties of
injection-molded, carbon nanotube-filled PET/PVDF
blends. J Power Sources 2004; 136: 37–44.
49. Besmann TM, Klett JW, Henry JJ, et al. Carbon/
carbon composite bipolar plate for proton exchange
membrane fuel cells. J Electrochem Soc 2000; 147(11):
4083–4086.
50. Middelman E, Kout W, Vogelar B, et al. Bipolar plates
for PEM fuel cells. J Power Sources 2003; 118: 44–46.
51. Schunk GmbH. Schunk bipolar plate material grade
FU4369 datasheet, http://www.schunk-group.com/
sixcms/media.php/1722/FU_4369_en.pdf (accessed 15
March 2010).
52. Bulk Moulding Compounds Inc. Datasheet for
BMC940 bipolar plate material, http://www.bulkmol
ding.com/datasheets/informational/BMC_940.pdf
(accessed 15 March 2010).