Journal of Power Sources 131 (2004) 162–168

Corrosion-resistant component for PEM fuel cells

Shuo-Jen Lee∗ , Ching-Han Huang, Jian-Jang Lai, Yu-Pang Chen
Department of Mechanical Engineering, Yuan Ze University, 135 FarEast Road, Chung-Li, Taiwan

Abstract

Thin sheets of stainless steel are low-cost material for fuel cell components that can be easily formed to yield compact volume. But
corrosion of the stainless steel bipolar plates is a problem in the PEMFC. Corrosion affects the performance and cell life of a fuel cell. The
goal of this study is to find out appropriate surface treatment technology to protect the stainless steel bipolar plates against electrochemical
corrosion. The feasible surface treatment process should be able to make significant improvement on the corrosion-resistance and the
electrical resistance at a reasonable cost. After the treatment, the surface composition and/or the surface morphology may be modified.
The modification of the surface composition may improve the corrosion-resistance of the metallic bipolar plate. The corrosion rate is
determined by linear-polarization method from the polarization curve. The differences in polarization behaviors are attributed to a surface
treated layer. The electrical resistance is the other factor which may affect the performance of the fuel cell. A smoother surface may reduce
the interfacial resistance. The treated layer is a passive film. The thickness of the passive film after the surface treatment is thinner and the
surface resistance is reduced.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Bipolar plate; Surface treatment; Corrosion-resistance; Electrical resistance

1. Introduction the corrosion-resistance capability of stainless steel under

such stringent operating conditions still need to be ad-
The most widely used bipolar plate material of the dressed. In a fuel cell, oxidation reaction occurred in the
PEM fuel cell is graphite. It is brittle and lacks mechan- anode which ionized the hydrogen and produced electrons,
ical strength. Therefore, thickness of the plates can not shown in Eq. (1). The hydrogen ions passed through the
be reduced. This results in thicker plates, bulkier size, membrane electrode assembly (MEA) and reached the
and heavier weight. There are many researchers focused cathode side to react with the oxygen and generated water
on developing substitute material that is cheaper, lighter, Eq. (2):
and easy to manufacture. The most often mentioned new
2H2 → 4H+ + 4e− (1)
materials for bipolar plates are injection molding of the
graphite filled polymer, compression molding, and metal O2 + 4H+ + 4e− → 2H2 O (2)
[1].
In comparison to the graphite material, metallic bipolar Because of the hydrogen ions in the anode and moisture
plates provide many advantages such as cheaper material content in the MEA, it becomes acid. In addition, due to
cost, good electrical and thermal conductivity, variety of the discharging process of the cell’s reaction, there were
manufacturing processes, excellent mechanical properties, potential differences between the micro-gaps of the metal-
and zero permeability. The volumetric size of the cell could lic bipolar plates and the MEA. Therefore, both chemical
be significantly reduced. The power/volume ratio of a cell and electrochemical corrosion may erode the metallic plates.
will be improved. Also, the metallic bipolar plates losing electrons may result
Stainless steel is a good choice for metallic bipolar in dissolution of metallic ions which may poison the MEA.
plate due to its excellent corrosion-resistance. However, On the cathode, metallic oxide grows due to recovery reac-
because of the electrochemical reactions in a fuel cell, tion by gaining the electrons. It may increase the surface’s
electrical resistance.
The stainless steel contains about 70% Fe, 18% Cr, and
∗ Corresponding author. Tel.: +886-3-4254-501; fax: +886-3-4254-501. 12% Ni. In acidic environment, both the Fe and Ni ele-
E-mail address: [email protected] (S.-J. Lee). ments and their oxides were unstable and may be corroded

0378-7753/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpowsour.2004.01.008
 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168 163

or dissolved. On the other hand, although, the Cr element

Power
was very active, it quickly became oxide, Cr2 O3 , which was
both chemically and electrochemically stable. The existence + -
of Mo and N will enhance chemical stability on the sur-
face. However, the metallurgical structure of the protective
Cathode
passive film of the stainless steel, Cr2 O3 , formed in normal
Anode

manufacturing process was not solid and uniform. It may

also have many surface defects such as inclusions, contam- Electrolyte
inants, scratches, and micro-cracks. It will accelerate local-
Fig. 1. Schematic plot of the surface treatment method.
ized corrosion.
There have been some studies on corrosion-resistance
of stainless steel. Hornung and Kappelt [2] studied the
ferric-based metallic bipolar plates. Their goal was to 2. Experimental and the results
investigate cheaper ferric-base metal to replace more ex-
pensive gold-plated nickel-base metal. They found that the 2.1. Process procedures
corrosion-resistance of the ferric-base metallic bipolar plate
was about the same as that of the gold-plated nickel-base The proposed surface treatment method employed the
metallic bipolar plate. However, the gold-plated bipolar electrochemical theory. The work specimen is the anode
plate had better electrical conductivity. and is immersed in the electrolyte. When it was connected
Davies et al. [3] had reported that there was a high re- with a cathode, the metallic ions will be released from the
sistance surface layer of oxide film among many stainless work specimen to form a passive film. Fig. 1 shows the
steels such as ss316, ss310, and ss904L. The thickness of schematic plot of the electrochemical process. Because the
the oxide film will affect contact resistance and degrade cell passive film had different metallurgical compositions from
performance. Their results showed that the cell performance those of the substrate, the capability of corrosion-resistance
was rated as followed: ss316 < ss310 < ss904L. The higher may be improved. The surface morphology also became
contents of the nickel and chromium will result in thinner smoother and shinning. The surface roughness was greatly
oxide film and, thus, better cell performance. It was verified improved and exhibited hydrophobic property. It may im-
from the measurement of contact resistance of the ss904L prove the flow of gas and water in the gas channel of the
was better than that of the ss316. bipolar plate. The surface resistance was reduced. There-
Wang et al. [4] studied the characteristics of stainless fore, the cell performance may be more stable and cell life is
steels with different Cr contents. The results showed that the longer.
Cr contents were ss349TM > ss904L > ss317L > ss316L. The stainless steel ss316L (American Society for Testing
The performance of corrosion tests also showed ss349TM and Material) was selected as the bipolar plate material. The
> ss904L > ss317L > ss316L. The results of both Davies size of the work specimen was 50 mm ×30 mm ×1 mm. The
et al. [3] and Wang et al. [4] verified that the Cr forms cathode material was copper. The experimental procedures
passive film on the surface of the stainless steels. The were shown in Fig. 2. First, the specimen was prepared by
more the Cr content in the stainless steel the better its grinding, degreasing, ultrasonic cleaning, and de-ionization,
corrosion-resistance. a standard cleaning process for chemical treatment. The
The analysis of the compositions in the MEA reported by stainless steel plate went through the electrochemical sur-
Wind et al. [5] indicated that large quantity of Fe and Ni face treatment. Finally, the treated specimens were tested
atoms released from the stainless steel. It means that, under for corrosion-resistance, surface metallurgical compositions,
the PEM fuel cell operating environment, the Fe and Ni and contact resistance.
atoms may be dissolved by the electrochemical effect while
the Cr atoms is more stable. In order to improve the effect of 2.2. Analysis of passive film
this oxide film, Wind had investigated metallic coatings on
the metallic plates. They found that the coating reduces the The treated specimen was analyzed by ESCA for surface
contaminants of nickel, chromium, and iron in the membrane metallurgical compositions. From the ESCA spectrum the
electrode assembly. The cell had a steady performance of binding energy will be compared with data bank to identify
over 1000 h. the amount of chemical shift and the chemical compositions
For most reported data, a fuel cell with stainless steel of the elements [6].
bipolar plates has a cell life between 1000 and 3000 h. For From the results, Cr/Fe elemental ratios, chemical species
computer, communication, and consumer (3C) products, it for Cr and Fe and independent estimates of Cr, Fe, and Ni
is a feasible direction. Therefore, in this paper, a surface oxide thickness, were described. Cr/Fe elemental ratio, and
treatment method was developed in order to improve the metal/oxide species ratios for Cr, Fe, and Ni were deter-
surface quality of the oxide film, to reduce contact resistance mined from measurements of the Cr(2p3/2), Fe(2p3/2), and
and to increase corrosion-resistance. Ni(2p3/2). The oxide thickness of Cr, Fe, and Ni estimation
164 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168

Specimen preparation
• Grinding
• Degreasing
• Ultrasonic cleaning
• De-ionization

Surface quality
• Electrochemical
surface treatment
Fig. 3. Typical uniform corrosion curve of linear polarization test.

where Icorr is the corrosion current, A/n is the gram-equivalent

weight, D the weight density, βa βc /2.3(βa + βc )) is the
Experiment constant. A typical uniform curve from a linear polarization
• Analysis of passive test was shown in Fig. 3 [10]. The slope from Fig. 3 is the
film Rp . When substituted into Eq. (7), the Icorr can be calcu-
• Corrosion testing
• Contact resistance lated. This method is faster than the measurement of weight
• Single cell testing loss by putting the specimen in acid solution for a long
time. There are different standards to perform the corrosion
Fig. 2. Flow chart of the experimental procedure.
test. Because, the pH value in the fuel cell environment
is between 0 and 3.5 and the ionic exchange is equivalent
to 1 M of H2 SO4 [11], the ASTM G5 standard [10] was
followed the Eqs. 3–5 [7]:
employed for the uniform corrosion test. The specimen was


Cr oxide area tested under room temperature with 0.5 M H2 SO4 Solution.
Cr oxide thickness
20 Å × ln 2.15 + 1 A Solartron 1285 potentiostat was used for the corrosion
Cr metal area
(3) test which included reference electrode (REF), working
electrode (WE), and auxiliary electrode (AUX). The scan-

 ning range was between −0.5 and 0.5 V (versus OCP) and
Fe oxide area
Fe oxide thickness
20 Å × ln 2.01 + 1 the scanning rate was 10 mV/s.
Fe metal area
(4) 2.4. Contact resistance measurement


Ni oxide area In order to obtain optimal parameter for cell assembly,
Ni oxide thickness
20 Å × ln 1.7 + 1
Ni metal area tests were conducted to establish the relationship of the pres-
(5) sure versus contact resistance. Because of the sandwich lay-
ered structure, the surface characteristics, and the surface
2.3. Corrosion test passive (oxide) film may affect the contact resistance be-
tween the bipolar plate and the MEA.
During the operation of a H2 —air PEM fuel cell, the The contact resistance test method as Fig. 4 was described
ionic exchange membrane may dissolve acidic ions such as by Wang et al. [4]. This basically involved measuring the
SO4 − , SO3 − , and HSO4 − [8]. Because of ionic and elec-
tronic transfer exchange, the electrochemical corrosion will
occur due to the potential difference.
The uniform corrosion rate of a metal can be estimated
by the electrochemical corrosion measurement. The linear
polarization method was employed for uniform corrosion
evaluation [9]. From Faraday’s law, the corrosion rate, Rcorr ,
can be obtained under known surface area, A and process
time [10].
Icorr A
Rcorr = 0.0032 (6)
nD
β a βc
Rp = (7)
2.3Icorr (βa + βc ) Fig. 4. Schematic of the test assembly for interfacial contact resistance.
 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168 165

potential difference across two pieces of the plate mate- Table 1

rial, which sandwiched two pieces of a gas diffusion layer, Results of ESCA analyses (wt.%)
whilst a fixed electrical current was passed through the ar- Element Original specimen Processed specimen
rangement. The potential difference was monitored as the Cr 26.88 59.69%
compaction on the assembly was periodically increased. Fe 52.40% 15.68%
Thus, the relative surface resistance for each material Ni 20.71% 24.63%
could be determined. The bipolar plate specimens were Cr/Fe ratio 0.51 3.81
tested, apart from thorough degreasing and cleaning in DI
water.
Table 2
By measuring the voltage drop, it was possible to calculate
List of surface roughness values (␮m)
the total resistance according to:
No. Rmax Ra
V = IR (8)
Original 1.1 0.12
1 0.410 0.042
2 0.383 0.038
3. Results and discussion 3 0.451 0.045
4 0.550 0.058
3.1. Analysis of passive film 5 0.500 0.050
6 0.367 0.032

The elemental of compositions of the passive films were

analyzed and the results were listed in Table 1. The percent-
age composition of Cr has increased by two-folds from 26.88 improve the flow of gas and fluid. The reduction in sur-
to 59.69%. The specimen also went through AES for the face roughness value may also decrease micro-potential
depth profile to verify the compositional changes through- difference. The values of surface roughness after treatment
out the thickness. The results were shown in Fig. 5. At a were listed in Table 2. Figs. 6 and 7 show the microscopy
depth of around 10–20 Å the Cr composition is much higher pictures of the surface profiles of the original and treated
than that of the Fe. At the depth of substrate, the percent- specimen. The original specimen, Fig. 6, had rough sur-
age compositions went back to normal ss316L compositions face profile, loose and damaged structure, inclusions, and
when Fe consists of 70% and Cr was about 16–18%. rolling marks. Fig. 7 shows smooth and integral struc-
The experimental data showed that the Cr content had ture with no above mentioned defects. The improvement
doubled from its original value. It proved that this surface in surface roughness was also obvious when comparing
treatment method can effectively increase the Cr content Figs. 6 and 7.
and decrease the corrosion-prone Fe content of the surface.
The thickness of the oxide layer of the Cr, Fe, and Ni were 3.2. Results of corrosion test
estimated to be 16.5, 5.5, and 6.9 Å. This dense and integral
oxide film had much less thickness. Fig. 8 shows typical result of the uniform corrosion test.
Under optical microscopy, the smooth and shinning The electrochemical corrosion rate of the original specimen
surface exhibits hydrophobic characteristics which may is 0.1 mmPy. The corrosion rates of the processed specimen

Fig. 5. Depth profile from AES analysis.

166 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168

Table 3
Results of the corrosion-resistance tests
Test number Uniform corrosion Result of corrosion test
(mmPy) improvement (%)
Original 1.0E-01 –
1 3.54E-02 62.852
2 2.78E-02 70.811
3 3.03E-02 68.232
4 3.99E-02 58.200
5 3.82E-02 59.968
6 2.10E-02 78.027
Average 3.21E-2 66.230
Fig. 6. Microscopy picture of original specimen (surface roughness: Ra
0.12 ␮m, Rmax 1.1 ␮m).

content. At 0.6 V potential which is the operating voltage of

the fuel cell, the corrosion current of the treated specimen
is 15 ␮A/cm2 as compared with 60 ␮A/cm2 of the original
specimen.
When the treated specimen went through the corrosion
test in the 0.5 M H2 SO4 solution, Fig. 9 shows the an-
odic behavior of the original state and after 24 and 48 h.
The surface condition is stable and the corrosion currents
remain constant at about 25 ␮A/cm2 . The corrosion rates
were 0.035, 0.038, and 0.039 mmPy, respectively. This in-
dicated that the treated specimen had a stable protective
Fig. 7. Microscopy picture of treated specimen (surface roughness: Ra passive film and may survive a long time fuel cell operating
0.02 ␮m, Rmax 0.24 ␮m).
conditions.

are listed in Table 3. This indicates that the corrosion rate 3.3. Results of contact resistance measurements
was improved by about 66%. It meant that the processed
plates may run longer without electrochemical corrosion un- In order to obtain optimal parameters for cell assembly,
der normal PEM fuel cell operation. tests were conducted to establish the relationship of the
The corrosion tests, shown in Fig. 8, indicates that the pressure versus contact resistance. Because of the sandwich
corrosion current, Icorr , of the treated specimen was smaller layered structure, the surface characteristics, and the sur-
than that of the original specimen due to the increased Cr face passive (oxide) film may affect the contact resistance

Fig. 8. Anodic behavior in 0.5 M H2 SO4 at 70 ◦ C (original vs. treated).

 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168 167

Fig. 9. Anodic behavior in 0.5 M H2 SO4 at 70 ◦ C-treated specimen (䊊)-Oh, (䊏)-24 h, ()-48 h.

The Cr oxide, Cr2 O3 , formed a protective passive film which

increased the corrosion-resistance of the stainless steel bipo-
lar plates. The corrosion test showed that the corrosion cur-
rent was reduced to one fourth of the original value. When
immersed in the 0.5 M H2 SO4 70 ◦ C, the corrosion current
remain stable indicating it may survive long term fuel cell
operation.
Because of the clean and integral surface, the surface elec-
tric conductivity increased despite the increase in Cr con-
tent. The surface texture and surface roughness had also
been greatly improved. The surface characteristics became
hydrophobic which may improve the flow of the fuels and
fluid. Because of the smooth surface, the micro-potential
difference between the MEA and the metallic bipolar plates
Fig. 10. Contact resistances of the original and the processed specimens.
was reduced which may further reduce localized corrosion
of the metallic bipolar plate.
between the bipolar plate and the MEA. The curves of From this study, the surface treated stainless steel bipolar
pressure versus contact resistance of the original and the plate may be a good candidate for PEM fuel cell. Single cell
processed specimens are shown in Fig. 10. This indicates and long term life testing are under preparation.
that due to the smoother surface the contact resistance was
much lower under low assembly pressure. As the applied
pressure increases, the diffusion layer was compressed and
the difference became minimal. References

[1] J. Larminie, A. Dicks, Fuel Cell System Explained, John Wiley &
Sons, 2000.
4. Conclusions [2] R. Hornung, G. Kppelt, Bipolar plate materials development using
Fe-based alloys for solid polymer fuel cells, J. Power Sources 72
The metallurgical analysis showed that the Cr content on (1998) 20–21.
the surface of the treated stainless steel bipolar plate had [3] D.P. Davies, P.L. Adcock, M. Turpin, S.J. Rowen, Stainless steel as a
bipolar plate material for solid polymer fuel cells, J. Power Sources
been greatly increased through selective dissolution process 86 (2000) 237–242.
of the surface treatment method. At the same time, the Fe [4] H. Wang, M.A. Sweikart, J.A. Turner, Stainless steel as a bipolar
content was greatly decreased. The metallurgical structure plate material for polymer electrolyte membrane fuel cells, J. Power
of the surface had become defect-free, solid, and integral. Sources 115 (2003) 243–251.
168 S.-J. Lee et al. / Journal of Power Sources 131 (2004) 162–168

[5] J. Wind, R. Späh, W. Kaiser, G. Böhm, Metallic bipolar plat for [8] Li and et al., Corrosion-resistance PEM fuel cell, US Patent 5624769,
PEM fuel cells, J. Power Sources 105 (2002) 256–260. 22 December 1995.
[6] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben: Hand- [9] D.A. Jones, Principles and Prevention of Corrosion, Macmillan Pub-
book of X-Ray Photoelectron Spectroscopy, Physical Electronics, lishing Company, 1992.
1995. [10] G. Hoogers, Fuel Cell Technology Handbook, CPC Press, 27 Septem-
[7] SEMASPEC 90120403B-STD, Analysis of surface composition and ber 2002.
chemistry of electropolished stainless steel turbing for gas distribution [11] ASTM G5-94, Standard Reference Test Method for Marking Poten-
system components. tiostatic and Potentiodynamic Anodic Polarization Measurements.