- Email: [email protected]
Towards more representative test methods for corrosion resistance of PEMFC metallic bipolar plates
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Towards more representative test methods for corrosion resistance of PEMFC metallic bipolar plates G. Hinds*, E. Brightman National Physical Laboratory, Teddington, Middlesex TW11 0LW, United Kingdom
article info
abstract
Article history:
Metallic bipolar plates are of increasing interest to automotive polymer electrolyte mem-
Received 29 October 2014
brane fuel cell (PEMFC) manufacturers due to their low cost, high power density, ease of
Received in revised form
manufacture, high conductivity and good mechanical properties but minimising the un-
12 December 2014
desirable effects of corrosion remains a key challenge. Unfortunately, reliable assessment
Accepted 21 December 2014
of the applicability of stainless steels with a range of coatings and surface treatments has
Available online 13 January 2015
been hampered by a lack of representative ex situ test methods. Here we characterise the local environment experienced by a bipolar plate during fuel cell operation via in situ
Keywords:
measurement of pH and corrosion potential for an uncoated 316L stainless steel bipolar
PEMFC
plate in a single cell PEMFC. We demonstrate that the degradation mode is more akin to
Bipolar plate
corrosion in relatively dilute thin liquid layers, rather than the fully immersed conditions
Stainless steel
employed in conventional ex situ screening tests. A key observation is that the corrosion
Corrosion
potential of the bipolar plate is only weakly coupled to the potential of the nearest Pt
Test methods
electrode due to the low ionic conductivity of the discontinuous aqueous phase in the gas diffusion layer (GDL). However, localised polarisation of the steel can occur in the presence of oxygen as a result of galvanic coupling with the carbon GDL at wetted interfaces, a process which may be enhanced by the creviced geometry. The implications for development of more representative ex situ test protocols are discussed. Crown Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Polymer electrolyte membrane fuel cells (PEMFCs) are promising candidates for transport applications due to their high power density, fast start-up times, low emissions and high efficiency. The major barriers to commercialisation of PEMFCs are the relatively high cost of component materials and processing, limited durability under real world operating
conditions and the lack of an established refuelling infrastructure. Significant progress has been made in addressing these barriers and the technology is already becoming established in niche applications such as uninterruptible power supplies, back-up power and forklift trucks. However, major technological challenges remain for automotive applications, in which the internal combustion engine has the competitive advantage of more than a century of mass production, with continuous incremental improvements to cost and durability.
* Corresponding author. Tel.: þ44 20 8943 7147. E-mail address: [email protected] (G. Hinds). http://dx.doi.org/10.1016/j.ijhydene.2014.12.085 0360-3199/Crown Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
2786
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
Table 1 e 2020 US DoE targets for bipolar plates [2]. Property
Target
Notes
Cost Corrosion resistance (anode)
<3 $ kW1 <1 mA cm2
Corrosion resistance (cathode)
<1 mA cm2
Electrical conductivity Areal specific resistance Hydrogen permeability Flexural strength Forming elongation
>100 S cm1 <0.01 U cm2 <1.3.1014 cm3 (s cm2 Pa) >25 MPa 40%
One of the key components of the PEMFC is the bipolar plate, which comprises approximately 80% of the weight and 25% of the cost of a typical fuel cell stack [1]. The ideal bipolar plate material must satisfy the following requirements: low cost (both raw material and processing); low gas permeability; high corrosion resistance; low resistivity (both bulk and contact); high mechanical strength; ease of machining/ high volume manufacturing; sufficient hydrophobicity to promote removal of product water; high thermal conductivity to facilitate dissipation of waste heat; low mass and density for transport applications; and high impact resistance. The 2020 US Department of Energy targets [2] for bipolar plate material properties are generally accepted as the benchmark for commercialisation and are summarised in Table 1. Impregnated graphite was the bipolar plate material of choice for early variants of the PEMFC due to its good corrosion performance and low contact resistance but its poor mechanical properties mean that it cannot meet many of the other targets in Table 1. On the other hand, metallic bipolar plates satisfy the DoE requirements related to mechanical properties and manufacturing but are more susceptible to corrosion in the PEMFC environment. This has three undesirable consequences. Firstly, the corrosion product formed at the surface increases the contact resistance between the bipolar plate and the gas diffusion layer (GDL) [3], reducing the available power from the fuel cell. Secondly, the presence of the corrosion product can alter the hydrophobicity of the bipolar plate surface, leading to local flooding and hindered transport of reactant gas. Finally, the metal cations leached from the bipolar plate can catalyse degradation of the membrane and lower its conductivity, leading to further loss of performance [4]. Stainless steels, particularly 316L, have been extensively investigated for use as bipolar plate materials. However, the challenge with these materials is that the passive film responsible for their excellent corrosion resistance often leads to unacceptably high contact resistance, a trend that is often amplified by ageing of the film. A wide range of coatings and surface modification techniques have been developed to combat this problem. Here, the challenge is to produce a defect-free coating and/or stable passive film that is compatible with the forming process of the gas channels in the bipolar plate. A comprehensive literature review of approaches to improving corrosion resistance of bipolar plates was published by Antunes et al. [5] in 2010.
2002 dollars, 500,000 stacks per year pH 3, 0.1 ppm HF, 80 C, Ar purge Potentiodynamic test 0.4 Ve0.6 V (Ag/AgCl), 0.1 mV/s pH 3, 0.1 ppm HF, 80 C, aerated Potentiostatic test (>24 h) 0.6 V (Ag/AgCl), ipassive < 50 nA cm2 e including contact resistance at 138 N cm2 ASTM D1434, 80 C, 3 atm, 100% RH ASTM D790-10 ASTM E8M-01
In the vast majority of studies in the literature, screening for corrosion resistance of metallic bipolar plate materials (both coated and uncoated) is carried out using electrochemical polarisation tests under fully immersed conditions. Historically, the test solution has tended to be 0.5 M or 1 M sulphuric acid solution, often with fluoride ions at the ppm level, deemed to be representative of the acidic environment of the Nafion membrane, although the trend now is towards testing at pH 3 based on measurements of run-off water from fuel cell stacks. In a PEMFC the bipolar plate is physically separated from the membrane by the GDL, except at points external to the active area of the cell where some designs allow (although this is increasingly rare) so direct exposure to highly acidic (pH < 1) solution is unlikely under fully humidified conditions. As indicated in Table 1, polarisation testing of candidate anode and cathode materials under fully immersed conditions is a standard ex situ test to establish fitness for purpose. However, there is some doubt as to the validity of testing under fully immersed conditions. The aqueous phase between the bipolar plate and the membrane is discontinuous, consisting of isolated droplets of water that move through the GDL and gas channels under the influence of the reactant gas flow. The relatively low ionic conductivity in this aqueous phase should limit the ability of the bipolar plate to assume a mixed potential close to that of the nearest fuel cell electrode, an issue that does not seem to have been identified in the literature. This calls into question the choice of both environment and potential in conventional ex situ corrosion tests on bipolar plate materials. Regarding the aqueous environment at the bipolar plate surface, extensive measurements have shown that the pH of effluent water from PEMFCs is typically in the range 3e5 [4,6,7], depending on the extent of membrane degradation and hydrolysis of metal cations leached from the bipolar plate or from stainless steel tubing and fittings in the balance of plant of the stack. However, such measurements are remote from the location of interest and the extent to which they are affected by loss of ionic species during evaporation and subsequent condensation of water in the gas channels is unknown. There is a clear need for in situ measurements of pH and corrosion potential at the bipolar plate surface to provide an informed basis for development of more representative test protocols. A further complication is the existence of crevices between the compressed fibres of the GDL and the ribs of the bipolar
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
plate. A creviced geometry tends to enhance localised corrosion since it provides a physical barrier that facilitates the development of concentration gradients and supports the maintenance of an aggressive local chemistry. Galvanic effects, generated for example by a more noble material (e.g. graphitic carbon) or a redox reaction, may provide an additional driving force for localised corrosion. Mele and Bozzini [8] studied the influence of crevice formers and ion contamination on the localised corrosion of 304 stainless steel immersed in 1 mM H2SO4 and 10 mM HCl solution. They found that the tendency for localised corrosion was slightly enhanced in the presence of a crevice formed with reticulated vitreous carbon (RVC). Localised attack was more severe in the simulated cathode environment, while chloride ions were found to be far more aggressive than fluoride ions. Andre et al. [9] investigated the effect of galvanic coupling between 316L stainless steel and a home-made three layer membrane electrode assembly (MEA) immersed in a bulk solution with the composition of stack exhaust water at 60 C. They showed that under fully immersed conditions shifts of several hundred mV from the open circuit potential of the stainless steel could be achieved, resulting in significantly higher corrosion rates. They also demonstrated that straining the material to simulate embossing is detrimental to the corrosion performance. Another issue not usually taken into account is the transient spike in cathode potential encountered during start-up and shut-down of PEMFCs due to the presence of a fuel/air boundary at the anode [10], which can reach as high as 1.6 V (SHE) [11]. It is well known that this elevated potential can lead to significant corrosion of the cathode catalyst support but its effect on bipolar plate corrosion has not been reported in the literature. A compromise between full scale stack testing and ex situ screening was developed by Auvinen et al. [12] using a multisingle cell approach in which several unit cells are operated simultaneously under identical conditions. They compared the performance of CrN and Au coatings on 316L, 430 and 904L stainless steel with that of bare substrates and established a correlation between contact resistance measurements in the multi-single cell and those in a commercial stack. However, quantitative evaluation of corrosion rate was limited to measurements of iron concentration in the exhaust water, which the authors acknowledged were compromised by accumulation of iron in the MEA and GDL. In this paper we present in situ measurements of pH and corrosion potential at the surface of an uncoated 316L stainless steel flowfield plate during operation of a single cell PEMFC using a novel reference electrode developed at NPL. The results are discussed in the context of the implications for development of more representative ex situ test methods for evaluation of corrosion resistance of metallic bipolar plates.
Material and methods Nafion leaching test The effect of contact with Nafion on the pH of pure water was evaluated using a piece of pristine Nafion NR211 membrane
2787
(Ion Power GmbH, Germany), which was held in a filter cartridge while pure water (conductivity 0.65 mS/cm) was pumped through it at a rate of 5 ml/min. The pH of the output water was monitored by taking 10 ml samples every 30 min and measuring the pH using a standard glass pH electrode. The pH electrode was connected to a Jenway model 3510 pH meter and calibrated with pH 4, pH 7 and pH 10 buffer solutions traceable to NIST (Fisher Scientific, UK).
In situ measurement of pH In situ measurements of pH close to the surface of the 316L SS flowfield plate were carried out using a commercially available micro-combination pH electrode (AMANI 1000L), which was inserted through a small hole in the end plates of the fuel cell and sealed with an O-ring. The tip of the probe was located at the base of the gas channel in the flowfield plate, which was interchangeable between the anode and cathode. The pH was measured during continuous operation at 70 C, 100% RH, at both open circuit and at a current density of 0.5 A cm2. The inlet gases were humidified to a dew-point of 80 C in order to ensure condensation of water would occur in the flow channels, required for the pH probe to operate. Calibration of the pH electrode was carried out prior to the measurement using pH 4 and pH 7 buffers heated to 70 C, and the measurements were compensated for temperature according to the application note [13] by Bier (Hach, USA).
In situ measurement of Ecorr The corrosion potential, Ecorr, of a metal is defined as its open circuit potential measured against a reference electrode in aqueous solution. In order to measure the corrosion potential of a 316L SS flowfield plate in an operating fuel cell, it is necessary to locate a reference electrode in the aqueous phase in contact with the surface. Due to the challenge presented by the discontinuous aqueous phase, i.e. water droplets, in the channels of the flowfield plate, it is preferable to employ fully humidified conditions to maximise the chances of ionic connectivity between reference electrode and flowfield plate. Use of a miniature reference electrode is also preferable to minimise perturbation of the system under measurement. Strictly speaking, it is not possible to measure the corrosion potential of the 316L SS flowfield plate in isolation, as it is connected both electrically (via contact) and ionically (through the aqueous phase) to the carbon GDL, resulting in a mixed potential between the steel and the carbon. The extent of polarisation will depend on factors such as the degree of wetting of the interface, the porosity of the GDL and the local solution chemistry. However, this distinction is secondary to the purpose of this study as the measurement of interest is the mixed potential between the steel and the carbon GDL, which reflects the real situation during operation of the fuel cell. For simplicity we will refer to the Ecorr value for the steel throughout this paper. The corrosion potential measurements were carried out using the recently developed NPL reference electrode, which utilises a Nafion salt bridge inserted through the endplates of the fuel cell [14]. A 50 cm2 single cell PEMFC with a 316L SS sixchannel serpentine flowfield plate was used for all
2788
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
measurements. The MEA was supplied by Johnson Matthey and consisted of Pt/C electrodes with Pt loadings of 0.4 mg/ cm2 (cathode) and 0.07 mg/cm2 (anode) on a perfluorinated sulphonic acid membrane of area 196 cm2. All experiments were carried out at 80 C and 100% RH using a Hydrogenics FCATS-G50 test stand. Fully humidified conditions were used in order to maximise the wetting of the 316L SS flowfield plate, while recognising that this is only one extreme in the range of potential conditions. The gas stoichiometry was 2:2 and compression of the cell was achieved using a pressurised gas/ piston arrangement with a pressure of 7 barg. Start-up and shut-down measurements were carried out with the fuel cell at open circuit, using a manual four-way crossover valve to switch between fully humidified hydrogen and air supplies at a constant flow rate of 0.4 sLpm. A schematic diagram of the experimental setup is shown in Fig. 1. The salt bridge consisted of thin Nafion tubing (PermaPure, NJ, USA) of inner diameter 0.64 mm and outer diameter 0.84 mm, which was encased in a PFTE sheath of inner diameter 1.01 mm and outer diameter 1.27 mm. The salt bridge (length approx. 0.5 m) was inserted through a hole drilled through the end plates of the fuel cell up to a point at the interface between the 316L SS flowfield plate and the GDL. A porous hydrophilised polytetrafluoroethylene (PTFE) membrane separator (Millipore, Ireland) was positioned between the Nafion tube salt bridge and the GDL to improve ionic contact with the flowfield plate. An O-ring arrangement was used to seal the PTFE tubing in order to prevent leakage of gas from the flowfield (see inset in Fig. 1). The PTFE tubing was filled with deionised water to ensure full hydration and therefore maximal conductivity of the Nafion tubing. The other end of the Nafion tubing was immersed in a glass chamber containing 0.5 M H2SO4, into which was placed a Hydroflex hydrogen reference electrode (Gaskatel GmbH, Germany). The hydrogen reference electrode was calibrated against an SHE in 0.5 M H2SO4 at room temperature before connection to the fuel cell. The 316L SS flowfield plate with the reference electrode insert could be used interchangeably on anode or cathode.
The hole for insertion of the reference electrode salt bridge was located midway along the final pass of the serpentine flowfield close the gas outlet. The cell potential was logged on the fuel cell test station, while the potential difference between the 316L SS flowfield plate and the reference electrode was measured and logged using a Jenway model 3510 pH meter (Fisher Scientific, UK) operated in mV mode.
Ex situ measurement of Ecorr Ex situ measurements of the corrosion potential of carbon GDL and 316L SS specimens in 1 mM H2SO4 at 80 C were carried out in a standard two electrode cell using a Solartron Modulab potentiostat (Ametek Ltd, UK). The reference electrode was a Hydroflex hydrogen reference electrode (Gaskatel GmbH, Germany), which was located in a separate chamber at room temperature. The reference electrode chamber was filled with 0.5 M H2SO4 and was connected to the cell via a Nafion tube salt bridge similar to that used for the in situ measurements.
In situ measurement of fluoride ion concentration In situ measurement of fluoride ion concentration was also attempted using a miniaturised ion selective electrode fabricated at NPL [15]. However, the limit of detection of this electrode was well above typical fluoride ion levels in an operating PEMFC and further work is required to optimise this technique.
Results and discussion Nafion leaching test The pH of the pure water before the start of the Nafion leaching experiment was 6.5. During the first hour of the test, the pH of the output water dropped to around 4.0. After continuing to flush 2e3 L of pure water through the Nafion membrane, the pH returned to a stable value of around 6.5. The initial transient drop in pH was therefore attributed to flushing out of impurities in the Nafion. These results support the view that a sustained drop in pH as a result of contact with Nafion is only possible when the mobile anion concentration within the membrane phase is sufficiently high, i.e protons can only leave the membrane when accompanied by a chargebalancing anion. A high mobile anion concentration could be maintained by degradation of the membrane, for example via peroxide radicals, which is favoured under relatively dry conditions. Under such conditions, there may not be sufficient water for corrosion to occur, again leading to the conclusion that varying conditions (both spatially and transiently) are likely to be more detrimental to the corrosion resistance of the bipolar plate. Further work is required to investigate systematically the effect of membrane degradation on the chemistry of water in the GDL/flowfield.
In situ measurement of pH Fig. 1 e Schematic diagram of in situ reference electrode configuration.
The results of in situ pH measurements are shown in Table 2. The cathode measurements were in the range 3.4e4.0, which
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
2789
Table 2 e Comparison of in situ pH measurement with measurements of pH of run-off water from fuel cell stacks. Authors
Measured pH
Notes
This work Healy et al. [4].
3.4e4.0 5e7 3.8e5.1
Hou et al. [6].
3e7
Abdullah et al. [7].
4e6
Cathode, T ¼ 70 C, RH 100% Anode, T ¼ 70 C, RH 100% T ¼ 75 Ce82 C, RH variable pH correlated with p[F] T ¼ 30/60/90 C Dew point: 30/60/90 C T ¼ 50/70/90 C, RH ¼ 35/60/100%, Lowest pH 1.5 (35% RH, 50 C)
is consistent with literature measurements on exhaust water from fuel cell stacks. More neutral values were observed in the anode measurements (5e7). The lower pH at the cathode flowfield plate can be rationalised on the basis of a higher metal ion concentration in the water under the more oxidising environment of the cathode and therefore a greater degree of hydrolysis, reducing the local pH. The source of metal ions is not necessarily the metallic bipolar plate itself as significant corrosion of the internal surfaces of austenitic stainless steel tubing, fittings and components such as humidifiers can occur in deionised water at elevated temperature. Acidification of water in the GDL through evaporation is of course also possible so it is likely that an optimum humidification level for metallic corrosion exists whereby there is sufficient water for corrosion to occur but also the possibility for localised wet/dry cycling to increase the concentration of aggressive ionic species. Membrane degradation and wet/dry cycling leading to concentration due to evaporation are expected to be the principal sources of acidity in an operating fuel cell. A fitness for purpose test method should take these factors into account.
In situ measurement of Ecorr The effect of cell current density on the corrosion potential of the 316L SS anode flowfield plate is shown in Fig. 2a. No significant influence of current density was observed on Ecorr, which remained relatively steady at ~0 V for current densities in the range 0 A cm2e0.5 A cm2. This is not surprising as the potential of the Pt anode does not change significantly over the range of current densities tested due to the high exchange current density for hydrogen oxidation. However, the Ecorr measurements for the cathode flowfield plate are more conclusive. Here, Ecorr remained fairly constant at around 0.6 V for current densities in the same range, despite the cell potential varying between 0.2 V and 0.9 V (Fig. 2b). This supports the hypothesis that there is only a limited ionic current pathway between the electrode and the flowfield plate and therefore significant polarisation of the 316L SS by the cathode reaction at the Pt electrode is not achievable. These measurements were carried out under fully humidified conditions, in which the water content of the GDL is maximised. Under drier conditions the ionic current pathway between the Pt electrode and the bipolar plate would be expected to be even more restricted. These results indicate the need for re-
Fig. 2 e Effect of cell current density on the corrosion potential of 316L SS (a) anode and (b) cathode flowfield plates.
evaluation of the potential values and test protocols specified in the DoE corrosion resistance tests in Table 1. It should be noted that the performance of the cell used in this study is relatively poor compared to that of similar cells with conventional graphite or coated metallic flowfield plates due to the high contact resistance between the uncoated 316L SS flowfield plate and the GDL. This is unavoidable for the purposes of in situ measurement of Ecorr of the bare steel and in practice limits the maximum cell current density to 0.5 A cm2. However, this limitation has no effect on the general conclusions of this work. The value of Ecorr for the cathode flowfield plate is somewhat more positive than might be expected for 316L SS in pH 3 solution, implying the presence of a mixed potential. Since the independence of Ecorr from cell potential has demonstrated that there is no significant polarisation of the steel by the Pt cathode under these conditions, a mixed potential with the carbon GDL is the most likely explanation for the elevated potential. In contrast to the Pt electrode, there is good ionic connection with the nearest edge of the GDL as the two materials are in intimate contact and adequate wetting of the interface can be assumed at 100% RH. Furthermore, the more positive value of Ecorr is consistent with that of carbon in such environments due to its higher exchange current density for
2790
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
oxygen reduction and the lack of an anodic dissolution reaction. In order to investigate this hypothesis, ex situ measurements of Ecorr were carried out on specimens of carbon GDL and 316L SS in 1 mM H2SO4 at 80 C, as shown in Fig. 3. The steady state Ecorr value for 316L SS was 0.5 Ve0.6 V, while that of the carbon GDL was 0.8 Ve0.9 V. The in situ Ecorr value of 0.6 Ve0.7 V is therefore consistent with a mixed potential between the 316L SS flowfield plate and the carbon GDL. The presence of this mixed potential should be taken into account in the development of representative ex situ test protocols. This should also incorporate the effect of crevice formation between the GDL fibres and the steel surface, which would be expected to enhance pit stabilisation. The use of noble metal or carbon-based coatings may lead to similar galvanic crevice corrosion effects. The effect of current cycling on Ecorr is shown for the cathode and anode flowfield plates in Fig. 4a and b respectively. No significant effects of abrupt changes in cell current density were observed on the Ecorr value at either flowfield plate. Measurements of Ecorr of the 316L SS flowfield plate during start-up and shut-down of the fuel cell are shown in Fig. 5a (cathode) and Fig. 5b (anode). For the cathode flowfield plate, Ecorr remains steady at around 0.6 V over a number of start-up/ shut-down cycles (Fig. 5a). Transient spikes were observed upon start-up, but their magnitude (~0.8 V) was significantly lower than the corresponding potential spikes observed at the cathode catalyst layer (~1.4 V) [13]. Again, the magnitude of these potential spikes would be expected to decrease significantly under drier operating conditions due to reduced conductivity of the aqueous and ionic phases. In contrast, significant changes in Ecorr were observed for the anode flowfield plate during start-up/shut-down cycling (Fig. 5b). On shut-down (t ~ 9 min), the anode compartment filled with air and Ecorr increased towards the steady value observed for the cathode flowfield plate (~0.6 V), reflecting the mixed potential between the steel and the carbon GDL. This trend was reversed on start-up (t ~ 18 min), with the potential returning towards 0 V. Repeated hydrogen/air cycling in this
a
b
Fig. 4 e Effect of current cycling on the corrosion potential of 316L SS (a) cathode and (b) anode flowfield plates.
manner is likely to be detrimental to the stability of the passive film, which is consistent with reports in the literature of a greater instability of the passive film in the anode environment compared to that at the cathode [16,17]. The incorporation of start-up/shut-down protocols into material qualification testing for bipolar plates is highly recommended, provided of course that this is consistent with the intended application.
Conclusions
Fig. 3 e Ex situ measurement of corrosion potential of carbon GDL and 316L SS specimens in 1 mM H2SO4 at 80 C.
In situ measurements of the environment experienced by a metallic bipolar plate during fuel cell operation demonstrate that the degradation mode is more akin to corrosion in relatively dilute thin liquid layers, rather than the fully immersed conditions employed in conventional ex situ screening tests. The corrosion potential of the bipolar plate is only weakly coupled to the potential of the nearest Pt electrode due to the low ionic conductivity of the discontinuous aqueous phase in the GDL.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
a
2791
Acknowledgements This work was supported by the UK National Measurement System under the Innovation R&D Programme and by the EU FP7 H2FC project under Grant Agreement Number 284522.
references
b
Fig. 5 e Effect of start-up/shut-down cycling on the corrosion potential of 316L SS (a) cathode and (b) anode flowfield plates.
Localised polarisation of the steel can occur as a result of galvanic coupling with the carbon GDL at wetted interfaces, a process which may be enhanced by the creviced geometry. A similar effect may be experienced with noble metal or carbon-based coatings. The measured pH of water in contact with the bipolar plate was 3.4e4.0 (cathode) and 5e7 (anode), although this could be decreased by evaporation as a result of wet/dry cycling. Test protocols and acceptance criteria currently specified in US DoE targets for corrosion resistance of bipolar plates should be reviewed in the light of these results. More representative ex situ test methods for screening of corrosion resistance of candidate bipolar plate materials should reflect the following: - Corrosion in relatively dilute thin liquid layers - Localised galvanic crevice corrosion due to intimate contact with the carbon GDL - Wet/dry cycling leading to concentration of aggressive species by evaporation - Start-up/shut-down cycling - Acceleration effects by variation of test parameters
[1] Wang H, Turner JA. Reviewing metallic PEMFC bipolar plates. Fuel Cells 2010;10:510. [2] U.S. Department of Energy. Hydrogen, fuel cells and technologies program. In: Multi-year research, development and demonstration plan, table 3.4.15; 2013. p. 29. http:// www1.eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/ fuel_cells.pdf. [3] Andre J, Antoni L, Petit J-P, De Vito E, Montani A. Electrical contact resistance between stainless steel bipolar plate and carbon felt in PEFC: a comprehensive study. Int J Hydrogen Energy 2009;34:3125. [4] Healy J, Hayden C, Xie T, Olson K, Waldo R, Brundage M, et al. Aspects of the chemical degradation of PFSA ionomers used in PEM fuel cells. Fuel Cells 2005;5:302. [5] Antunes RA, Oliveira MCL, Ett G, Ett V. Corrosion of metal bipolar plates for PEM fuel cells: a review. Int J Hydrogen Energy 2010;35:3632. [6] Abdullah AM, Mohammad AM, Okajima T, Kitamura F, Ohsaka T. Effect of load, temperature and humidity on the pH of the water drained out from H(2)/air polymer electrolyte membrane fuel cells. J Power Sources 2009;190:264. [7] Hou K-H, Lin C-H, Ger M-D, Shiah S-W, Chou H-M. Analysis of the characterization of water produced from proton exchange membrane fuel cell (PEMFC) under different operating thermal conditions. Int J Hydrogen Energy 2012;37:3890. [8] Mele C, Bozzini B. Localised corrosion processes of austenitic stainless steel bipolar plates for polymer electrolyte membrane fuel cells. J Power Sources 2010;195:3590. [9] Andre J, Antoni L, Petit J-P. Corrosion resistance of stainless steel bipolar plates in a PEFC environment: a comprehensive study. Int J Hydrogen Energy 2010;35:3684. [10] Reiser C, Bregoli L, Patterson TW, Yi JS, Yang JD, Perry ML, et al. A reverse-current decay mechanism for fuel cells. Electrochem Solid State Lett 2005;8:A273. [11] Brightman E, Hinds G. In situ mapping of potential transients during start-up and shut-down of a polymer electrolyte membrane fuel cell. J Power Sources 2014;267:160. [12] Auvinen S, Tingelof T, Ihonen JK, Siivinen J, Johansson M. Cost effective in-situ characterization of coatings for PEFC bipolar plates demonstrated with PVD deposited CrN. J Electrochem Soc 2011;158:B550. [13] Electrochemistry: theory and practice. CO, USA: Dr Axel Bier, Hach Company; 2010. http://www.phmeters.com/docs/pH_ Electrochemistry_White_Paper.pdf. [14] Hinds G, Brightman E. In situ mapping of electrode potential in a PEM fuel cell. Electrochem Commun 2012;17:26. [15] Brightman E, Hinds G. Feasibility study: in situ measurement of fluoride ion concentration in a polymer electrolyte membrane fuel cell. NPL Report MAT 2014;70. [16] Laedre S, Kongstein OE, Oedegaard A, Seland F, Karoliussen H. The effect of pH and halides on the corrosion process of stainless steel bipolar plates for proton exchange membrane fuel cells. Int J Hydrogen Energy 2012;37:18537. [17] Wang Y, Northwood DO. Effects of O-2 and H-2 on the corrosion of SS316L metallic bipolar plate materials in simulated anode and cathode environments of PEM fuel cells. Electrochim Acta 2007;52:6793.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
Towards more representative test methods for corrosion resistance of PEMFC metallic bipolar plates G. Hinds*, E. Brightman National Physical Laboratory, Teddington, Middlesex TW11 0LW, United Kingdom
article info
abstract
Article history:
Metallic bipolar plates are of increasing interest to automotive polymer electrolyte mem-
Received 29 October 2014
brane fuel cell (PEMFC) manufacturers due to their low cost, high power density, ease of
Received in revised form
manufacture, high conductivity and good mechanical properties but minimising the un-
12 December 2014
desirable effects of corrosion remains a key challenge. Unfortunately, reliable assessment
Accepted 21 December 2014
of the applicability of stainless steels with a range of coatings and surface treatments has
Available online 13 January 2015
been hampered by a lack of representative ex situ test methods. Here we characterise the local environment experienced by a bipolar plate during fuel cell operation via in situ
Keywords:
measurement of pH and corrosion potential for an uncoated 316L stainless steel bipolar
PEMFC
plate in a single cell PEMFC. We demonstrate that the degradation mode is more akin to
Bipolar plate
corrosion in relatively dilute thin liquid layers, rather than the fully immersed conditions
Stainless steel
employed in conventional ex situ screening tests. A key observation is that the corrosion
Corrosion
potential of the bipolar plate is only weakly coupled to the potential of the nearest Pt
Test methods
electrode due to the low ionic conductivity of the discontinuous aqueous phase in the gas diffusion layer (GDL). However, localised polarisation of the steel can occur in the presence of oxygen as a result of galvanic coupling with the carbon GDL at wetted interfaces, a process which may be enhanced by the creviced geometry. The implications for development of more representative ex situ test protocols are discussed. Crown Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Polymer electrolyte membrane fuel cells (PEMFCs) are promising candidates for transport applications due to their high power density, fast start-up times, low emissions and high efficiency. The major barriers to commercialisation of PEMFCs are the relatively high cost of component materials and processing, limited durability under real world operating
conditions and the lack of an established refuelling infrastructure. Significant progress has been made in addressing these barriers and the technology is already becoming established in niche applications such as uninterruptible power supplies, back-up power and forklift trucks. However, major technological challenges remain for automotive applications, in which the internal combustion engine has the competitive advantage of more than a century of mass production, with continuous incremental improvements to cost and durability.
* Corresponding author. Tel.: þ44 20 8943 7147. E-mail address: [email protected] (G. Hinds). http://dx.doi.org/10.1016/j.ijhydene.2014.12.085 0360-3199/Crown Copyright © 2014, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
2786
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
Table 1 e 2020 US DoE targets for bipolar plates [2]. Property
Target
Notes
Cost Corrosion resistance (anode)
<3 $ kW1 <1 mA cm2
Corrosion resistance (cathode)
<1 mA cm2
Electrical conductivity Areal specific resistance Hydrogen permeability Flexural strength Forming elongation
>100 S cm1 <0.01 U cm2 <1.3.1014 cm3 (s cm2 Pa) >25 MPa 40%
One of the key components of the PEMFC is the bipolar plate, which comprises approximately 80% of the weight and 25% of the cost of a typical fuel cell stack [1]. The ideal bipolar plate material must satisfy the following requirements: low cost (both raw material and processing); low gas permeability; high corrosion resistance; low resistivity (both bulk and contact); high mechanical strength; ease of machining/ high volume manufacturing; sufficient hydrophobicity to promote removal of product water; high thermal conductivity to facilitate dissipation of waste heat; low mass and density for transport applications; and high impact resistance. The 2020 US Department of Energy targets [2] for bipolar plate material properties are generally accepted as the benchmark for commercialisation and are summarised in Table 1. Impregnated graphite was the bipolar plate material of choice for early variants of the PEMFC due to its good corrosion performance and low contact resistance but its poor mechanical properties mean that it cannot meet many of the other targets in Table 1. On the other hand, metallic bipolar plates satisfy the DoE requirements related to mechanical properties and manufacturing but are more susceptible to corrosion in the PEMFC environment. This has three undesirable consequences. Firstly, the corrosion product formed at the surface increases the contact resistance between the bipolar plate and the gas diffusion layer (GDL) [3], reducing the available power from the fuel cell. Secondly, the presence of the corrosion product can alter the hydrophobicity of the bipolar plate surface, leading to local flooding and hindered transport of reactant gas. Finally, the metal cations leached from the bipolar plate can catalyse degradation of the membrane and lower its conductivity, leading to further loss of performance [4]. Stainless steels, particularly 316L, have been extensively investigated for use as bipolar plate materials. However, the challenge with these materials is that the passive film responsible for their excellent corrosion resistance often leads to unacceptably high contact resistance, a trend that is often amplified by ageing of the film. A wide range of coatings and surface modification techniques have been developed to combat this problem. Here, the challenge is to produce a defect-free coating and/or stable passive film that is compatible with the forming process of the gas channels in the bipolar plate. A comprehensive literature review of approaches to improving corrosion resistance of bipolar plates was published by Antunes et al. [5] in 2010.
2002 dollars, 500,000 stacks per year pH 3, 0.1 ppm HF, 80 C, Ar purge Potentiodynamic test 0.4 Ve0.6 V (Ag/AgCl), 0.1 mV/s pH 3, 0.1 ppm HF, 80 C, aerated Potentiostatic test (>24 h) 0.6 V (Ag/AgCl), ipassive < 50 nA cm2 e including contact resistance at 138 N cm2 ASTM D1434, 80 C, 3 atm, 100% RH ASTM D790-10 ASTM E8M-01
In the vast majority of studies in the literature, screening for corrosion resistance of metallic bipolar plate materials (both coated and uncoated) is carried out using electrochemical polarisation tests under fully immersed conditions. Historically, the test solution has tended to be 0.5 M or 1 M sulphuric acid solution, often with fluoride ions at the ppm level, deemed to be representative of the acidic environment of the Nafion membrane, although the trend now is towards testing at pH 3 based on measurements of run-off water from fuel cell stacks. In a PEMFC the bipolar plate is physically separated from the membrane by the GDL, except at points external to the active area of the cell where some designs allow (although this is increasingly rare) so direct exposure to highly acidic (pH < 1) solution is unlikely under fully humidified conditions. As indicated in Table 1, polarisation testing of candidate anode and cathode materials under fully immersed conditions is a standard ex situ test to establish fitness for purpose. However, there is some doubt as to the validity of testing under fully immersed conditions. The aqueous phase between the bipolar plate and the membrane is discontinuous, consisting of isolated droplets of water that move through the GDL and gas channels under the influence of the reactant gas flow. The relatively low ionic conductivity in this aqueous phase should limit the ability of the bipolar plate to assume a mixed potential close to that of the nearest fuel cell electrode, an issue that does not seem to have been identified in the literature. This calls into question the choice of both environment and potential in conventional ex situ corrosion tests on bipolar plate materials. Regarding the aqueous environment at the bipolar plate surface, extensive measurements have shown that the pH of effluent water from PEMFCs is typically in the range 3e5 [4,6,7], depending on the extent of membrane degradation and hydrolysis of metal cations leached from the bipolar plate or from stainless steel tubing and fittings in the balance of plant of the stack. However, such measurements are remote from the location of interest and the extent to which they are affected by loss of ionic species during evaporation and subsequent condensation of water in the gas channels is unknown. There is a clear need for in situ measurements of pH and corrosion potential at the bipolar plate surface to provide an informed basis for development of more representative test protocols. A further complication is the existence of crevices between the compressed fibres of the GDL and the ribs of the bipolar
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
plate. A creviced geometry tends to enhance localised corrosion since it provides a physical barrier that facilitates the development of concentration gradients and supports the maintenance of an aggressive local chemistry. Galvanic effects, generated for example by a more noble material (e.g. graphitic carbon) or a redox reaction, may provide an additional driving force for localised corrosion. Mele and Bozzini [8] studied the influence of crevice formers and ion contamination on the localised corrosion of 304 stainless steel immersed in 1 mM H2SO4 and 10 mM HCl solution. They found that the tendency for localised corrosion was slightly enhanced in the presence of a crevice formed with reticulated vitreous carbon (RVC). Localised attack was more severe in the simulated cathode environment, while chloride ions were found to be far more aggressive than fluoride ions. Andre et al. [9] investigated the effect of galvanic coupling between 316L stainless steel and a home-made three layer membrane electrode assembly (MEA) immersed in a bulk solution with the composition of stack exhaust water at 60 C. They showed that under fully immersed conditions shifts of several hundred mV from the open circuit potential of the stainless steel could be achieved, resulting in significantly higher corrosion rates. They also demonstrated that straining the material to simulate embossing is detrimental to the corrosion performance. Another issue not usually taken into account is the transient spike in cathode potential encountered during start-up and shut-down of PEMFCs due to the presence of a fuel/air boundary at the anode [10], which can reach as high as 1.6 V (SHE) [11]. It is well known that this elevated potential can lead to significant corrosion of the cathode catalyst support but its effect on bipolar plate corrosion has not been reported in the literature. A compromise between full scale stack testing and ex situ screening was developed by Auvinen et al. [12] using a multisingle cell approach in which several unit cells are operated simultaneously under identical conditions. They compared the performance of CrN and Au coatings on 316L, 430 and 904L stainless steel with that of bare substrates and established a correlation between contact resistance measurements in the multi-single cell and those in a commercial stack. However, quantitative evaluation of corrosion rate was limited to measurements of iron concentration in the exhaust water, which the authors acknowledged were compromised by accumulation of iron in the MEA and GDL. In this paper we present in situ measurements of pH and corrosion potential at the surface of an uncoated 316L stainless steel flowfield plate during operation of a single cell PEMFC using a novel reference electrode developed at NPL. The results are discussed in the context of the implications for development of more representative ex situ test methods for evaluation of corrosion resistance of metallic bipolar plates.
Material and methods Nafion leaching test The effect of contact with Nafion on the pH of pure water was evaluated using a piece of pristine Nafion NR211 membrane
2787
(Ion Power GmbH, Germany), which was held in a filter cartridge while pure water (conductivity 0.65 mS/cm) was pumped through it at a rate of 5 ml/min. The pH of the output water was monitored by taking 10 ml samples every 30 min and measuring the pH using a standard glass pH electrode. The pH electrode was connected to a Jenway model 3510 pH meter and calibrated with pH 4, pH 7 and pH 10 buffer solutions traceable to NIST (Fisher Scientific, UK).
In situ measurement of pH In situ measurements of pH close to the surface of the 316L SS flowfield plate were carried out using a commercially available micro-combination pH electrode (AMANI 1000L), which was inserted through a small hole in the end plates of the fuel cell and sealed with an O-ring. The tip of the probe was located at the base of the gas channel in the flowfield plate, which was interchangeable between the anode and cathode. The pH was measured during continuous operation at 70 C, 100% RH, at both open circuit and at a current density of 0.5 A cm2. The inlet gases were humidified to a dew-point of 80 C in order to ensure condensation of water would occur in the flow channels, required for the pH probe to operate. Calibration of the pH electrode was carried out prior to the measurement using pH 4 and pH 7 buffers heated to 70 C, and the measurements were compensated for temperature according to the application note [13] by Bier (Hach, USA).
In situ measurement of Ecorr The corrosion potential, Ecorr, of a metal is defined as its open circuit potential measured against a reference electrode in aqueous solution. In order to measure the corrosion potential of a 316L SS flowfield plate in an operating fuel cell, it is necessary to locate a reference electrode in the aqueous phase in contact with the surface. Due to the challenge presented by the discontinuous aqueous phase, i.e. water droplets, in the channels of the flowfield plate, it is preferable to employ fully humidified conditions to maximise the chances of ionic connectivity between reference electrode and flowfield plate. Use of a miniature reference electrode is also preferable to minimise perturbation of the system under measurement. Strictly speaking, it is not possible to measure the corrosion potential of the 316L SS flowfield plate in isolation, as it is connected both electrically (via contact) and ionically (through the aqueous phase) to the carbon GDL, resulting in a mixed potential between the steel and the carbon. The extent of polarisation will depend on factors such as the degree of wetting of the interface, the porosity of the GDL and the local solution chemistry. However, this distinction is secondary to the purpose of this study as the measurement of interest is the mixed potential between the steel and the carbon GDL, which reflects the real situation during operation of the fuel cell. For simplicity we will refer to the Ecorr value for the steel throughout this paper. The corrosion potential measurements were carried out using the recently developed NPL reference electrode, which utilises a Nafion salt bridge inserted through the endplates of the fuel cell [14]. A 50 cm2 single cell PEMFC with a 316L SS sixchannel serpentine flowfield plate was used for all
2788
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
measurements. The MEA was supplied by Johnson Matthey and consisted of Pt/C electrodes with Pt loadings of 0.4 mg/ cm2 (cathode) and 0.07 mg/cm2 (anode) on a perfluorinated sulphonic acid membrane of area 196 cm2. All experiments were carried out at 80 C and 100% RH using a Hydrogenics FCATS-G50 test stand. Fully humidified conditions were used in order to maximise the wetting of the 316L SS flowfield plate, while recognising that this is only one extreme in the range of potential conditions. The gas stoichiometry was 2:2 and compression of the cell was achieved using a pressurised gas/ piston arrangement with a pressure of 7 barg. Start-up and shut-down measurements were carried out with the fuel cell at open circuit, using a manual four-way crossover valve to switch between fully humidified hydrogen and air supplies at a constant flow rate of 0.4 sLpm. A schematic diagram of the experimental setup is shown in Fig. 1. The salt bridge consisted of thin Nafion tubing (PermaPure, NJ, USA) of inner diameter 0.64 mm and outer diameter 0.84 mm, which was encased in a PFTE sheath of inner diameter 1.01 mm and outer diameter 1.27 mm. The salt bridge (length approx. 0.5 m) was inserted through a hole drilled through the end plates of the fuel cell up to a point at the interface between the 316L SS flowfield plate and the GDL. A porous hydrophilised polytetrafluoroethylene (PTFE) membrane separator (Millipore, Ireland) was positioned between the Nafion tube salt bridge and the GDL to improve ionic contact with the flowfield plate. An O-ring arrangement was used to seal the PTFE tubing in order to prevent leakage of gas from the flowfield (see inset in Fig. 1). The PTFE tubing was filled with deionised water to ensure full hydration and therefore maximal conductivity of the Nafion tubing. The other end of the Nafion tubing was immersed in a glass chamber containing 0.5 M H2SO4, into which was placed a Hydroflex hydrogen reference electrode (Gaskatel GmbH, Germany). The hydrogen reference electrode was calibrated against an SHE in 0.5 M H2SO4 at room temperature before connection to the fuel cell. The 316L SS flowfield plate with the reference electrode insert could be used interchangeably on anode or cathode.
The hole for insertion of the reference electrode salt bridge was located midway along the final pass of the serpentine flowfield close the gas outlet. The cell potential was logged on the fuel cell test station, while the potential difference between the 316L SS flowfield plate and the reference electrode was measured and logged using a Jenway model 3510 pH meter (Fisher Scientific, UK) operated in mV mode.
Ex situ measurement of Ecorr Ex situ measurements of the corrosion potential of carbon GDL and 316L SS specimens in 1 mM H2SO4 at 80 C were carried out in a standard two electrode cell using a Solartron Modulab potentiostat (Ametek Ltd, UK). The reference electrode was a Hydroflex hydrogen reference electrode (Gaskatel GmbH, Germany), which was located in a separate chamber at room temperature. The reference electrode chamber was filled with 0.5 M H2SO4 and was connected to the cell via a Nafion tube salt bridge similar to that used for the in situ measurements.
In situ measurement of fluoride ion concentration In situ measurement of fluoride ion concentration was also attempted using a miniaturised ion selective electrode fabricated at NPL [15]. However, the limit of detection of this electrode was well above typical fluoride ion levels in an operating PEMFC and further work is required to optimise this technique.
Results and discussion Nafion leaching test The pH of the pure water before the start of the Nafion leaching experiment was 6.5. During the first hour of the test, the pH of the output water dropped to around 4.0. After continuing to flush 2e3 L of pure water through the Nafion membrane, the pH returned to a stable value of around 6.5. The initial transient drop in pH was therefore attributed to flushing out of impurities in the Nafion. These results support the view that a sustained drop in pH as a result of contact with Nafion is only possible when the mobile anion concentration within the membrane phase is sufficiently high, i.e protons can only leave the membrane when accompanied by a chargebalancing anion. A high mobile anion concentration could be maintained by degradation of the membrane, for example via peroxide radicals, which is favoured under relatively dry conditions. Under such conditions, there may not be sufficient water for corrosion to occur, again leading to the conclusion that varying conditions (both spatially and transiently) are likely to be more detrimental to the corrosion resistance of the bipolar plate. Further work is required to investigate systematically the effect of membrane degradation on the chemistry of water in the GDL/flowfield.
In situ measurement of pH Fig. 1 e Schematic diagram of in situ reference electrode configuration.
The results of in situ pH measurements are shown in Table 2. The cathode measurements were in the range 3.4e4.0, which
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
2789
Table 2 e Comparison of in situ pH measurement with measurements of pH of run-off water from fuel cell stacks. Authors
Measured pH
Notes
This work Healy et al. [4].
3.4e4.0 5e7 3.8e5.1
Hou et al. [6].
3e7
Abdullah et al. [7].
4e6
Cathode, T ¼ 70 C, RH 100% Anode, T ¼ 70 C, RH 100% T ¼ 75 Ce82 C, RH variable pH correlated with p[F] T ¼ 30/60/90 C Dew point: 30/60/90 C T ¼ 50/70/90 C, RH ¼ 35/60/100%, Lowest pH 1.5 (35% RH, 50 C)
is consistent with literature measurements on exhaust water from fuel cell stacks. More neutral values were observed in the anode measurements (5e7). The lower pH at the cathode flowfield plate can be rationalised on the basis of a higher metal ion concentration in the water under the more oxidising environment of the cathode and therefore a greater degree of hydrolysis, reducing the local pH. The source of metal ions is not necessarily the metallic bipolar plate itself as significant corrosion of the internal surfaces of austenitic stainless steel tubing, fittings and components such as humidifiers can occur in deionised water at elevated temperature. Acidification of water in the GDL through evaporation is of course also possible so it is likely that an optimum humidification level for metallic corrosion exists whereby there is sufficient water for corrosion to occur but also the possibility for localised wet/dry cycling to increase the concentration of aggressive ionic species. Membrane degradation and wet/dry cycling leading to concentration due to evaporation are expected to be the principal sources of acidity in an operating fuel cell. A fitness for purpose test method should take these factors into account.
In situ measurement of Ecorr The effect of cell current density on the corrosion potential of the 316L SS anode flowfield plate is shown in Fig. 2a. No significant influence of current density was observed on Ecorr, which remained relatively steady at ~0 V for current densities in the range 0 A cm2e0.5 A cm2. This is not surprising as the potential of the Pt anode does not change significantly over the range of current densities tested due to the high exchange current density for hydrogen oxidation. However, the Ecorr measurements for the cathode flowfield plate are more conclusive. Here, Ecorr remained fairly constant at around 0.6 V for current densities in the same range, despite the cell potential varying between 0.2 V and 0.9 V (Fig. 2b). This supports the hypothesis that there is only a limited ionic current pathway between the electrode and the flowfield plate and therefore significant polarisation of the 316L SS by the cathode reaction at the Pt electrode is not achievable. These measurements were carried out under fully humidified conditions, in which the water content of the GDL is maximised. Under drier conditions the ionic current pathway between the Pt electrode and the bipolar plate would be expected to be even more restricted. These results indicate the need for re-
Fig. 2 e Effect of cell current density on the corrosion potential of 316L SS (a) anode and (b) cathode flowfield plates.
evaluation of the potential values and test protocols specified in the DoE corrosion resistance tests in Table 1. It should be noted that the performance of the cell used in this study is relatively poor compared to that of similar cells with conventional graphite or coated metallic flowfield plates due to the high contact resistance between the uncoated 316L SS flowfield plate and the GDL. This is unavoidable for the purposes of in situ measurement of Ecorr of the bare steel and in practice limits the maximum cell current density to 0.5 A cm2. However, this limitation has no effect on the general conclusions of this work. The value of Ecorr for the cathode flowfield plate is somewhat more positive than might be expected for 316L SS in pH 3 solution, implying the presence of a mixed potential. Since the independence of Ecorr from cell potential has demonstrated that there is no significant polarisation of the steel by the Pt cathode under these conditions, a mixed potential with the carbon GDL is the most likely explanation for the elevated potential. In contrast to the Pt electrode, there is good ionic connection with the nearest edge of the GDL as the two materials are in intimate contact and adequate wetting of the interface can be assumed at 100% RH. Furthermore, the more positive value of Ecorr is consistent with that of carbon in such environments due to its higher exchange current density for
2790
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
oxygen reduction and the lack of an anodic dissolution reaction. In order to investigate this hypothesis, ex situ measurements of Ecorr were carried out on specimens of carbon GDL and 316L SS in 1 mM H2SO4 at 80 C, as shown in Fig. 3. The steady state Ecorr value for 316L SS was 0.5 Ve0.6 V, while that of the carbon GDL was 0.8 Ve0.9 V. The in situ Ecorr value of 0.6 Ve0.7 V is therefore consistent with a mixed potential between the 316L SS flowfield plate and the carbon GDL. The presence of this mixed potential should be taken into account in the development of representative ex situ test protocols. This should also incorporate the effect of crevice formation between the GDL fibres and the steel surface, which would be expected to enhance pit stabilisation. The use of noble metal or carbon-based coatings may lead to similar galvanic crevice corrosion effects. The effect of current cycling on Ecorr is shown for the cathode and anode flowfield plates in Fig. 4a and b respectively. No significant effects of abrupt changes in cell current density were observed on the Ecorr value at either flowfield plate. Measurements of Ecorr of the 316L SS flowfield plate during start-up and shut-down of the fuel cell are shown in Fig. 5a (cathode) and Fig. 5b (anode). For the cathode flowfield plate, Ecorr remains steady at around 0.6 V over a number of start-up/ shut-down cycles (Fig. 5a). Transient spikes were observed upon start-up, but their magnitude (~0.8 V) was significantly lower than the corresponding potential spikes observed at the cathode catalyst layer (~1.4 V) [13]. Again, the magnitude of these potential spikes would be expected to decrease significantly under drier operating conditions due to reduced conductivity of the aqueous and ionic phases. In contrast, significant changes in Ecorr were observed for the anode flowfield plate during start-up/shut-down cycling (Fig. 5b). On shut-down (t ~ 9 min), the anode compartment filled with air and Ecorr increased towards the steady value observed for the cathode flowfield plate (~0.6 V), reflecting the mixed potential between the steel and the carbon GDL. This trend was reversed on start-up (t ~ 18 min), with the potential returning towards 0 V. Repeated hydrogen/air cycling in this
a
b
Fig. 4 e Effect of current cycling on the corrosion potential of 316L SS (a) cathode and (b) anode flowfield plates.
manner is likely to be detrimental to the stability of the passive film, which is consistent with reports in the literature of a greater instability of the passive film in the anode environment compared to that at the cathode [16,17]. The incorporation of start-up/shut-down protocols into material qualification testing for bipolar plates is highly recommended, provided of course that this is consistent with the intended application.
Conclusions
Fig. 3 e Ex situ measurement of corrosion potential of carbon GDL and 316L SS specimens in 1 mM H2SO4 at 80 C.
In situ measurements of the environment experienced by a metallic bipolar plate during fuel cell operation demonstrate that the degradation mode is more akin to corrosion in relatively dilute thin liquid layers, rather than the fully immersed conditions employed in conventional ex situ screening tests. The corrosion potential of the bipolar plate is only weakly coupled to the potential of the nearest Pt electrode due to the low ionic conductivity of the discontinuous aqueous phase in the GDL.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 2 7 8 5 e2 7 9 1
a
2791
Acknowledgements This work was supported by the UK National Measurement System under the Innovation R&D Programme and by the EU FP7 H2FC project under Grant Agreement Number 284522.
references
b
Fig. 5 e Effect of start-up/shut-down cycling on the corrosion potential of 316L SS (a) cathode and (b) anode flowfield plates.
Localised polarisation of the steel can occur as a result of galvanic coupling with the carbon GDL at wetted interfaces, a process which may be enhanced by the creviced geometry. A similar effect may be experienced with noble metal or carbon-based coatings. The measured pH of water in contact with the bipolar plate was 3.4e4.0 (cathode) and 5e7 (anode), although this could be decreased by evaporation as a result of wet/dry cycling. Test protocols and acceptance criteria currently specified in US DoE targets for corrosion resistance of bipolar plates should be reviewed in the light of these results. More representative ex situ test methods for screening of corrosion resistance of candidate bipolar plate materials should reflect the following: - Corrosion in relatively dilute thin liquid layers - Localised galvanic crevice corrosion due to intimate contact with the carbon GDL - Wet/dry cycling leading to concentration of aggressive species by evaporation - Start-up/shut-down cycling - Acceleration effects by variation of test parameters
[1] Wang H, Turner JA. Reviewing metallic PEMFC bipolar plates. Fuel Cells 2010;10:510. [2] U.S. Department of Energy. Hydrogen, fuel cells and technologies program. In: Multi-year research, development and demonstration plan, table 3.4.15; 2013. p. 29. http:// www1.eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/ fuel_cells.pdf. [3] Andre J, Antoni L, Petit J-P, De Vito E, Montani A. Electrical contact resistance between stainless steel bipolar plate and carbon felt in PEFC: a comprehensive study. Int J Hydrogen Energy 2009;34:3125. [4] Healy J, Hayden C, Xie T, Olson K, Waldo R, Brundage M, et al. Aspects of the chemical degradation of PFSA ionomers used in PEM fuel cells. Fuel Cells 2005;5:302. [5] Antunes RA, Oliveira MCL, Ett G, Ett V. Corrosion of metal bipolar plates for PEM fuel cells: a review. Int J Hydrogen Energy 2010;35:3632. [6] Abdullah AM, Mohammad AM, Okajima T, Kitamura F, Ohsaka T. Effect of load, temperature and humidity on the pH of the water drained out from H(2)/air polymer electrolyte membrane fuel cells. J Power Sources 2009;190:264. [7] Hou K-H, Lin C-H, Ger M-D, Shiah S-W, Chou H-M. Analysis of the characterization of water produced from proton exchange membrane fuel cell (PEMFC) under different operating thermal conditions. Int J Hydrogen Energy 2012;37:3890. [8] Mele C, Bozzini B. Localised corrosion processes of austenitic stainless steel bipolar plates for polymer electrolyte membrane fuel cells. J Power Sources 2010;195:3590. [9] Andre J, Antoni L, Petit J-P. Corrosion resistance of stainless steel bipolar plates in a PEFC environment: a comprehensive study. Int J Hydrogen Energy 2010;35:3684. [10] Reiser C, Bregoli L, Patterson TW, Yi JS, Yang JD, Perry ML, et al. A reverse-current decay mechanism for fuel cells. Electrochem Solid State Lett 2005;8:A273. [11] Brightman E, Hinds G. In situ mapping of potential transients during start-up and shut-down of a polymer electrolyte membrane fuel cell. J Power Sources 2014;267:160. [12] Auvinen S, Tingelof T, Ihonen JK, Siivinen J, Johansson M. Cost effective in-situ characterization of coatings for PEFC bipolar plates demonstrated with PVD deposited CrN. J Electrochem Soc 2011;158:B550. [13] Electrochemistry: theory and practice. CO, USA: Dr Axel Bier, Hach Company; 2010. http://www.phmeters.com/docs/pH_ Electrochemistry_White_Paper.pdf. [14] Hinds G, Brightman E. In situ mapping of electrode potential in a PEM fuel cell. Electrochem Commun 2012;17:26. [15] Brightman E, Hinds G. Feasibility study: in situ measurement of fluoride ion concentration in a polymer electrolyte membrane fuel cell. NPL Report MAT 2014;70. [16] Laedre S, Kongstein OE, Oedegaard A, Seland F, Karoliussen H. The effect of pH and halides on the corrosion process of stainless steel bipolar plates for proton exchange membrane fuel cells. Int J Hydrogen Energy 2012;37:18537. [17] Wang Y, Northwood DO. Effects of O-2 and H-2 on the corrosion of SS316L metallic bipolar plate materials in simulated anode and cathode environments of PEM fuel cells. Electrochim Acta 2007;52:6793.