Effect of Different Graphite Materials on the Properties of Bipolar... Fabricated by Selective Laser Sintering

Proceedings of the 5th Annual ISC Research Symposium
ISCRS 2011
April 7, 2011, Rolla, Missouri
Effect of Different Graphite Materials on the Properties of Bipolar Plates
Fabricated by Selective Laser Sintering
Nannan Guo
Department of Mechanical and Aerospace
Engineering, Missouri University of Science and
Technology, Rolla, MO 65401
ABSTRACT
Selective Laser Sintering (SLS) provides a way to
fabricate graphite composite bipolar plates, which
significantly reduces time and cost at the research
and development stage of bipolar plates, as compared
with the conventional fabrication methods of
compression molding and injection molding.
Different graphite materials (natural graphite,
synthetic graphite, carbon black, and carbon fiber)
were investigated in the SLS process, the effect of
each material on electrical conductivity and flexural
strength of bipolar plates was determined. Natural
graphite was great for electrical conductivity, Carbon
fiber was good for flexural strength, but synthetic
graphite and carbon black were not good for both two
properties. By proper combination of these materials
bipolar plates with electrical conductivity ranging
from 120S/cm to 380 S/cm and flexural strength of
around 40 MPa can be obtained, which satisfied the
requirements set by Department of Energy and also
was comparable with those developed by
compression molding and injection molding.
1. INTRODUCTION
Bipolar plate, which accounts for 40-50% cost
and 60-80% weight of the whole fuel cell stack [1], is
an important part in Proton Exchange Membrane
(PEM) fuel cell assembly. The main functions of
bipolar plate include carrying current away from each
cell, distributing gas fuels within the cell and
providing support for Membrane Electrode Assembly
(MEA) [2, 3]. Department of Energy proposed a
technical target of bipolar plates for the year 2010
[4], in which the main requirements are electrical
conductivity >100 S/cm and flexural strength >25
MPa.
Compared with metal, graphite is a great
material for bipolar plates due to its excellent
chemical resistance and low weight. However, the
brittle nature makes it difficult to manufacture.
Recently more and more researchers focus on
graphite/polymer composite bipolar plate [5-12],
which is easier to fabricate and has better mechanical
strength. Two main fabrication methods for graphite
composite bipolar plates are injection molding (IM)
Ming C. Leu
Department of Mechanical and Aerospace
Engineering, Missouri University of Science and
Technology, Rolla, MO 65401
[5] and compression molding (CM) [6], and several
graphite materials, synthetic graphite (SG) [6, 7],
nature graphite (NG) [6, 7], carbon black (CB) [6,7],
carbon fiber (CF) [6,7,9,10], expanded graphite (EG)
[8,9] and multi-walled carbon nanotubes (MWCNTs)
[10], have been investigated in these two processes.
Although CM and IM are very suitable for mass
production, they are less efficient for research and
development (R&D) stage, especially for bipolar
plate which is a design-intensive part including
numerous flow channel pattern and channel
dimensions [13], thousands of designs have to be
investigated at R&D stage. In IM or CM process,
corresponding mold has to be fabricated for each
different design, which is very expensive and time
consuming (as shown in Table 1).
Selective Laser Sintering (SLS), one of the
Rapid Prototyping techniques, has been studied to
fabricate graphite composite bipolar plates for PEM
fuel cell by Ssuwen Chen, et al [14-16]. In SLS
process, the mixture of graphite materials and binder
is scanned by laser and the molten binder bonds
graphite particles together to form 3D parts layer by
layer. The major advantages are the capability to
build complex flow fields for bipolar plates and less
time and lower cost consumed for making bipolar
plates from each design (as shown in Table 1),
therefore reducing time and cost for R&D stage of
bipolar plates, compared with conventional methods.
Table 1 Comparison of time and cost consumed between
SLS and conventional methods at R&D stage.
Compression
Molding & Injection
SLS
Molding
Cost for one
$25,000-$5000
<50$
design
Time
4-6 weeks
2-4 days
One big issue of bipolar plates fabricated by SLS
process is the relative low electrical conductivity
[15]. Several ways, such as infiltration with
conducting epoxy resin [15, 16], liquid phenolic
infiltration/re-curing [15, 16] and increase of
carbonization temperature [17, 18], have been studied
to increase the electrical conductivity, but in all of
1
these studies only synthetic graphite material was
used.
In this paper, NG, SG, CF and CB were
investigated in the SLS process in order to obtain
good electrical conductivity and mechanical strength,
fulfilling the requirements of DOE. The effect of each
material on the properties of bipolar plates was
studied and the properties obtained by SLS process
were compared with those fabricated by IM [5] and
IM [6, 7, 19, 20].
2. MATERIAL AND PROCESSES
2.1.Materials
Natural graphite (3610), synthetic graphite
(4437), carbon black (5303) and carbon fiber
(AGM99) were obtained from Asbury Graphite Mills,
Inc. (New Jersey, USA). Properties of these materials
were shown in Table 2. Natural graphite has high
electrical conductivity but poor wettability with the
liquid resin. Synthetic graphite is a product made by
high-temperature treatment of amorphous carbon
materials, usually calcined petroleum coke and coal
tar pitch. Because of the more uniform and spherical
shape, synthetic graphite gives better flowability of
packed powder. Carbon black, manufactured by
combustion or thermal decomposition of hydrocarbon
fuel under reducing condition, has the finest primary
particle size of all the common industrial carbons.
Carbon fiber was used to increase mechanical
strength of bipolar plates.
Phenolic powder (GP-5546, Georgia Pacific)
with typical size of 15µm was used as binder. After
carbonization, porous brown parts were infiltrated by
liquid epoxy resin (EPONTM Resin 828) as matrix,
and the density is 1.16g/cm3.
Table 2 Properties of graphite materials.
Natural Graphite
Synthetic Graphite
(3610)
(4437)
Properties
Size (µm)
75-150
Carbon Black (5303)
Carbon Fiber
(AGM99)
Diameter 7.4
Length 150
1.75
1.87
-
10-45
<0.03
Density (g/cm )
2.26
Surface Area (m2/g)
1.27
Typical Conductivity (S/cm)
27.78
*Provided by Asbury Graphite Mills, Inc., NJ, USA.
2.26
11.46
17.24
1.8
254
2.93
2.2.Fabrication and Process
The fabrication and post processing is shown in
Fig. 1. Graphite powders and phenolic binder was
ball milled for 12 hours. Then SLS machine
(Sinterstation 2000) was used to build green parts.
The parameters used for SLS process were: fill laser
power (12W), outline laser power (4W), laser scan
speed (1524mm/s), layer thickness (0.1016mm), laser
scan spacing (0.0762mm). Part bed and feed bins
were maintained at 60oC and 40oC respectively. Post
processing includes carbonization and infiltration.
For carbonization, green parts were heated to 1000oC
in furnace filled with Argon gas, to convert binder
into carbon residue and get brown parts. Heating
schedule was from room temperature to 200 oC with a
heating ramp rate of 60oC/h, followed by a slower
rate 30oC/h to 600oC, and then a 50oC/h ramp rate to
1000oC, holding for 1h. The aim of infiltration
process is to increase mechanical strength of brown
parts and make them gas impermeable, due to brown
parts were porous and weak. For infiltration, brown
parts were immersed into liquid epoxy resin, after
20min getting them out, cleaning the surface and
putting them into oven at 80oC for 30min to cure
resin.
2.3.Experiments
Synthetic graphite, carbon black and carbon fiber
were mixed with natural graphite respectively in
different volume ratios, and then mixed with 35vol%
binder (keep constant in all experiments). The content
of binder was determined based on the previous
experimental results, which ensured that green parts
had enough mechanical strength to go through the
following processes, without largely sacrificing
electrical conductivity. These mixtures of graphite
materials and binder were used to fabricate test
samples and bipolar plates with the process
mentioned above. Then the properties of final parts
were measured.
3
2.4.Characterization
Electrical conductivity was measured by Four
Point Probe technique, following ASTM C611
specification and using Keithley 2400 SourceMeter.
Five 20×3×3mm3 specimens were tested and average
value was calculated. Flexural strength (dimension of
samples: 3×10×60mm3) was measured with threepoint bending method, using Instron Model 4468.
Microstructure of samples was obtained by Hitachi S4700 FE-SEM.
2
3. RESULTS AND DISCUSSION
3.1.Post Processing
During the post processing, the microstructure of
bipolar plate changed as it went from green part to
brown part and then infiltrated part (final part), as
shown in Fig. 2. Figure 2(a) shows the microstructure of
green part, in which binder bonds graphite particles (SG
and NG) together. After carbonization, the binder was
removed and lots of pores were left in brown part (Fig.
2(b)). Finally these pores were filled with resin after
infiltration and only few tiny voids were left (Fig. 2(c)).
Due to the usage of insulating resin, one big
concern of infiltration process is that the resin might
reduce electrical conductivity of brown part. However,
the results show that there is no distinct difference
between the electrical conductivities before and after
infiltration and meanwhile the flexural strength of
bipolar plates increased from 1.56MPa to 38MPa after
infiltration, as shown in Table 3. The material used here
was 65vol% SG and 35vol% binder. This is because
good connection between graphite particles had been
established in green parts and brown parts (Fig. 2(a),
(b)), which was not broken by infiltration. Therefore it
is a feasible process to increase strength and make
bipolar plates gas impermeable, without taking negative
effect on the electrical conductivity.
Fig. 1 SLS fabrication and post processing of bipolar plates.
Fig. 2 Change of microstructures during the fabrication process: (a) green part; (b) brown part; (c) infiltrated part.
The material composition is 15vol% SG, 50vol% NG and 35vol% binder.
Table 3 Properties of bipolar plates before and after infiltration.
Electrical conductivity (S/cm)
Flexural strength (MPa)
Before infiltration
48.9
1.56
After infiltration
46.3
38.4
*The material used was 65vol% SG and 35vol% binder.
3.2.Effect of Synthetic Graphite
Figure 3 shows the microstructure of a brown part
with SG and NG, in which the larger flaky particles are
NG and the smaller spherical ones are SG. Electrical
conductivity with different volume fractions of SG was
shown in Fig. 4. When only NG (65vol%) and binder
(35vol%) was used, electrical conductivity was
380S/cm. As the content of SG increased, the
conductivity decreased because the conductivity of SG
particles is lower than that of NG. When SG particles
were introduced into NG powder, these particles
hindered the conduction of current among NG particles.
The results show that when SG content was less than
15vol%, the conductivity was higher than the target
value of DOE (100S/cm).
The effect of SG on flexural strength of bipolar
plates was shown in Fig. 5. SG had slightly negative
effect on the flexural strength. The strength decreased
from about 37MPa to 33MPa when SG increased from
5vol% to 65vol%. Because the adding of smaller SG
particles filled up the big pores among larger flaky NG
(as shown in Fig. 3), which reduced the porosity,
consequently a smaller amount of resin was absorbed
after infiltration, thus less strength was established. But
even so, all the results were still higher than the target
value set by DOE (25MPa).
3.3.Effect of Carbon Fiber
Carbon fiber is widely used in composite materials
to enhance mechanical strength. Figure 6 shows the
microstructure of brown part with CF and NG, in which
long and thin CF was uniformly mixed with NG
3
particles. Electrical conductivity with different volume
fractions of CF was shown in Fig. 7. Electrical
conductivity decreased as the fraction of CF increased.
This is because the dispersion of CF among NG
particles broke the contacts of NG particles and
increased the electrical resistance. Figure 8 shows that
flexural strength varies with different CF fractions.
Flexural strength increased greatly (35MPa to 40MPa)
after introducing CF, and kept increasing with the
increase of CF fraction, as expected. When the content
of CF was 25vol%, the flexural strength was almost
50MPa, which was 1.5 times of that without CF.
3.4.Effect of Carbon Black
It was reported that nano-size CB could be
dispersed among NG particles to increase electrical
conductivity for graphite composite bipolar plates made
by compression molding [7]. Ssuwen Chen, et al. [15,
16] also used epoxy resin containing carbon black
powder to infiltrate bipolar plates made by SLS process,
to improve electrical conductivity. In this study, CB was
directly mixed with NG powder to make bipolar plate
with SLS machine. Figure 9 shows the microstructure of
brown part made from 16vol% CB, 49vol% NG and
35vol% binder. As shown in Fig. 9(b), the surface of
NG particle was covered by these nano-size CB
particles.
Fig. 3 Microstructure of brown part with 15vol% SG,
50vol% NG and 35vol% binder.
Fig. 4 Electrical conductivity varies with different SG
fractions (keeping binder at 35vol%, with the rest being
NG).
Fig. 5 Flexural strength varies with different SG
fraction.
Fig. 6 Microstructure of a brown part with 25vol% CF,
40vol% NG and 35vol% binder.
The variation of electrical conductivity with
different volume fractions of CB was shown in Fig. 10.
As the CB content increased, conductivity gradually
decreased. After the fraction of CB reached 5vol%,
electrical conductivity fell below 100S/cm. This is
because CB has lower conductivity compared with NG.
During the ball-milling process, CB particles with large
surface area tended to agglomerate and cover on the
whole surface of NG particles (as shown in Fig. 9),
which hindered the contacts between NG particles. This
is different from the case in compression molding [7],
improvement of electrical conductivity was reported
when CB content was blow 4vol%. The reason is in
compression molding NG and CB were mixed in liquid
resin first, CB could be dispersed well and fill the small
voids between NG particles rather than covering their
surface, thus current can go through these small CB
particles instead of insulated resin reducing resistance.
Figure 11 shows that flexural strength varies with
different CB fractions. A similar result with SG was
obtained in the case of adding CB into NG. Flexural
strength slightly decreased with the increase of CB
content. A possible reason is that the coverage of CB on
the surface of NG made it difficult for liquid resin
(matrix) to fully wet the surface of NG particles (major
filler), therefore after curing of resin the interface
between the filler and matrix was weakened and flexural
strength declined.
4
Fig. 7 Electrical conductivity varies with different
carbon fiber fractions.
Fig. 8 Flexural strength varies with different carbon
fiber fractions.
3.5.Comparison with Compression Molding and
Injection Molding
The properties of bipolar plates obtained from SLS
process are compared with those from CM and IM,
shown in Table 4. The density of bipolar plates
fabricated by SLS is lower than those got from CM and
IM, because brown parts are not completely infiltrated
by resin, and only the surface is fully filled during
infiltration process, leaving inside porous.
For commercial available products provided by
SGL [5], Schunk [19] and BMCI [20], the electrical
conductivity are a little higher than 100S/cm. Electrical
conductivity of the bipolar plates developed in National
Physical laboratory (NPL), India [6, 7] ranges from 143
to 500S/cm, when different graphite materials are used
in compression molding. For SLS process, the electrical
conductivity got from 120S/cm up to 380S/cm, which is
comparable with those from CM and IM. In terms of
flexural strength, the value obtained from SLS result is
also similar to the results got from CM and IM, even
without pressure usage.
Bipolar plate was fabricated using the process
discussed above, shown in Fig. 12. The material
composition used was 45vol% NG, 10vol% CF, 10vol%
SG and 35vol% binder. Electrical conductivity was
around 120S/cm, and flexural strength was 40MPa. The
feature dimensions are: active area 50×50mm2,
thickness 4mm, channel width 1.5mm and depth
1.5mm.
Fig. 9 (a) Microstructure of a brown part with NG and CB; (b) detail view of the surface of NG particle. (16vol%
CB, 49vol% NG and 35vol% binder).
Fig. 10 Electrical conductivity varies with CB fractions.
Fig. 11 Flexural strength varies with different CB fractions.
5
Table 4 Comparison of properties of graphite composite bipolar plates
fabricated by different methods.
Supplier
SGL
SGL
Schunk BMC
NPL
Property
Missour
[5]
[5]
[19]
I [20]
b
[6,
b
i S&T
PPG86 BBP4 FU4369
9407]
a
a
a
a
8649
Process
SLS
IM
IM
CM
CM
CM
Density
>1.8
~1.2
1.85
1.98
1.90
1.82
(g/cm3)
5
Electrical
120143conductivit
55.6
125
111
100
380
500
y (S/cm)
Flexural
strength
~40
40
50
40
40
45
(MPa)
* a: Commercial product; b: Research stage.
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
Fig. 12 Bipolar plate with a serpentine flow field. (active area
50×50mm2, thickness 4mm, channel width 1.5mm and depth
1.5mm)
4. CONCLUSION
SLS process was successfully applied to fabricate graphite
composite bipolar plates. Different graphite materials, NG, CF,
SG and CB were investigated in this process, and
corresponding electrical conductivity and mechanical strength
were studied. NG was better for electrical conductivity and CF
can greatly increase the flexural strength. Nano-size CB,
covering the surface of NG particles, had negative effect on
both conductivity and strength. SG also decreased the
conductivity and had slightly negative effect on strength. After
proper combination of these materials, the electrical and
mechanical properties got from SLS process were comparable
with those got from IM and CM, indicating that SLS process
can be applied for R&D of bipolar plates. Finally, material
composition of 45vol% NG, 10vol% CF, 10vol% SG and
35vol% binder was chosen to fabricate bipolar plates. The
performance of the fabricated bipolar plates in PEM fuel cell
stack will be tested in the future.
5. ACKNOWLEDGMENTS
This project is supported by Air Force Research Laboratory
under contract #FA8650-04-C-5704.
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
6. REFERENCES
[1]
Tsuchiya, H., Kobayashi, O. Mass production cost of PEM fuel
cell by learning curve. Int. J. Hydrogen Energy, 2004, 29: 985990.
[20]
6
Viral Mehta, Joyce Smith Cooper. Review and analysis of PEM
fuel cell design and manufacturing. Journal of Power Sources,
2003(114): 32-53.
Allen Hermann, Tapas Chaudhuri, Priscila Spagnol. Bipolar
plates for PEM fuel cells: A review. International Journal of
Hydrogen Energy, 2005, 30: 1297-1302.
http://www1.eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/f
uel_cells.pdf, website, 2007.
A. Muller, P. Kauranen, A.von Ganski and B. Hell. Injection
moulding of graphite composite bipolar plates. Journal of Power
Sources, 2006(154): 467-471.
S.R. Dhakate, R.B. Mathur, B.K. Kakati, T.L. Dhami. Properties
of graphite-composite bipolar plate prepared by compression
molding technique for PEM fuel cell. International Journal of
Hydrogen Energy, 2007, 32: 4537-4543.
R.B Mathur, S.R. Dhakate, D.K. Gupta, T.L. Dhami, R.K.
Aggarwal. Effect of different carbon fillers on the properties of
graphite composite bipolar plate. Journal of Materials Processing
Technology, 2008(203), 184-192.
Richard Blunk, Mahmoud Hassan Abd Elhamid, Daniel Lisi,
Youssef Mikhail. Polymeric composite bipolar plates for vehicle
application. Journal of Power Sources, 2006(156): 151-157.
Ling Du, Sadhan C. Jana. Highly conductivity epoxy/graphite
composites for bipolar plates in proton exchange membrane fuel
cells. Journal of Power Sources, 2007(172): 734-741.
Joong Hee Lee, Yun Ki Jang, Chang Eui Hong, Nam Hoon Kim,
et al. Effect of carbon fillers on properties of polymer composite
bipolar plates of fuel cells. Journal of Power Sources, 2009(193):
523-529.
Min-Chien Hsiao, Shu-Hang Liao, Ming-Yu Yen, et al. Effect of
graphite sizes and carbon nanotubes content on flowability of
bulk-molding compound and formability of the composite
bipolar plate for fuel cell. Journal of Power Sources, 2010(195):
5645-5650.
Chuanyu Yen, Shuhang Liao, Yufeng Lin, et al. Preparation and
properties of high performance nanocomposite bipolar plate for
fuel cell. Journal of Power Sources, 2006(162): 309-315.
Xiaoguo Li, Imran Sabir. Review of bipolar plates in PEM fuel
cells: Flow-field designs. International Journal of Hydrogen
Energy, 2005(30): 359-371.
Ssuwei Chen, David L. Bourell, Kristin L.Wood. Fabrication of
PEM fuel cell bipolar plates by indirect SLS. International SFF
Symposium, 2004.
Ssuwei Chen, Jeremy Murphy, Jason Herlehy, et al.
Development of SLS fuel cell current collectors. Rapid
Prototyping Journal, 2006: 275-282.
Ssuwei Chen, David L. Bourell, Kristin L.Wood. Improvement
of electrical conductivity of SLS PEM fuel cell bipolar plates.
International SFF Symposium, 2005.
Kaushik Alayavalli, David L. Bourell. Fabrication of Electrically
Conductive, Fluid Impermeable Direct Methanol Fuel Cell
(DMFC) Graphite Bipolar Plates by Indirect Selective Laser
Sintering (SLS). International SFF Symposium, 2008.
Kaushik Alayavalli, David L. Fabrication of modified graphite
bipolar plates by indirect selective laser sintering (SLS) for
direct methanol fuel cells. Rapid Prototyping Journal, 2010: 268274.
Schunk-Molded bipolar plates for Fuel Cells. Schunk
Kohlenstofftechnik GmbH, 2008.
BMC 940 Data Sheet. Bulk Molding Compounds, Inc., 2009.