Development of membrane electrode assembly for ... - Bruno G. Pollet

Journal of Power Sources 288 (2015) 121e127

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Development of membrane electrode assembly for high temperature proton exchange membrane fuel cell by catalyst coating membrane method Huagen Liang, Huaneng Su*, Bruno G. Pollet, Sivakumar Pasupathi HySA Systems Competence Centre, South African Institute for Advanced Materials Chemistry, University of the Western Cape, Private Bag X17, Bellville 7535, South Africa

h i g h l i g h t s MEA with low Pt loading for HT-PEMFC was developed by CCM method. The fabrication parameters were investigated for the performance optimization. The CCM-based MEA has good stability during a short-term fuel cell operation.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 11 November 2014 Received in revised form 17 April 2015 Accepted 19 April 2015 Available online 23 April 2015

Membrane electrode assembly (MEA), which contains cathode and anode catalytic layer, gas diffusion layers (GDL) and electrolyte membrane, is the key unit of a PEMFC. An attempt to develop MEA for ABPBI membrane based high temperature (HT) PEMFC is conducted in this work by catalyst coating membrane (CCM) method. The structure and performance of the MEA are examined by scanning electron microscopy (SEM), electrochemical impedance spectroscopy (EIS) and IeV curve. Effects of the CCM preparation method, Pt loading and binder type are investigated for the optimization of the single cell performance. Under 160 C and atmospheric pressure, the peak power density of the MEA, with Pt loading of 0.5 mg cm 2 and 0.3 mg cm 2 for the cathode and the anode, can reach 277 mW cm 2, while a current density of 620 A cm 2 is delivered at the working voltage of 0.4 V. The MEA prepared by CCM method shows good stability operating in a short term durability test: the cell voltage maintained at ~0.45 V without obvious drop when operated at a constant current density of 300 mA cm 2 and 160 C under ambient pressure for 140 h. © 2015 Elsevier B.V. All rights reserved.

Keywords: High-temperature proton exchange membrane fuel cell Polybenzimidazole Membrane electrode assembly Catalyst coating membrane Cell performance

1. Introduction Proton exchange membrane fuel cell (PEMFC) are considered as a promising next generation of clean energy conversion technology due to its high power density, high efficiency, low emissions and fast start-up [1,2]. With increasing the operating temperature of fuel cell (100e200 C), the Pt catalyst poisoning by CO impurities at the anode can be significantly mitigated and the cell performance also can be further enhanced because of the improved kinetics of cathode and anode reaction [3,4]. In addition, the humidification system is not necessary and water management become easier at a

* Corresponding author. E-mail address: [email protected] (H. Su). http://dx.doi.org/10.1016/j.jpowsour.2015.04.123 0378-7753/© 2015 Elsevier B.V. All rights reserved.

relatively higher temperature. Moreover, fuel cell operated at elevated temperature has high thermodynamic efficiency and simplified thermal management, which is ideal for combined heat and power (CHP) systems. Hence, researchers have made efforts to develop HT-PEMFCs based on phosphoric acid (PA) doped polybenzimidazole (PBI) membrane in the last decades [5e10]. However, to date, this promising technology has not yet been put on the market, resulting from the low cell performance caused by the slow oxygen reduction reaction (ORR) kinetics and the transport limitation of the reactants and proton, due to the presence of phosphoric acid [5,8]. Therefore, one of the most critical challenge in developing HT-PEMFCs is to enhance the cell performance [5,6,11]. The most important part of PEMFC is membrane electrode assembly, which is consisted of catalyst layers, electrolyte membrane and two gas diffusion layers (GDLs). In the MEA, the

122

H. Liang et al. / Journal of Power Sources 288 (2015) 121e127

electrochemical reaction for both anode and cathode only take place at ‘triple-phase boundaries’, where reactant, catalyst particles and electrolyte contact together [12]. The fuel cell performance can differ greatly depending on the method of the MEA fabrication and other key parameters such as catalyst loading, binder and ionomer content [5,6]. Many methods have been developed to prepare MEAs, including gas diffusion electrode (GDE) method and CCM method [13,14]. As an alternative to the GDE method, in the CCM process, the catalyst inks are directly applied onto both sides of the proton exchange membrane. Hence, it is believed that the CCM method can avoid the loss of catalyst particles immersed into the pore network of gas diffusion layer (GDL) and establish a better interfacial contact between the catalyst layer (CL) and the electrolyte membrane, which can enhance the catalyst utilization and improve the cell performance [15,16]. However, one technical challenge is that the surface of the PBI-based membrane with predoped PA will remain moist state due to the strong moisture absorption and the exudation of PA, resulting in a poor adhesion of the catalyst particles on the wet surface of the ABPBI membrane. Wannek et al. [17e20] reported that PA redistribution is a quick process within the HT-MEAs consisted of dry ABPBI and PA predoped GDEs. A stable cell performance can be reached in several minutes after commissioning. Inspired by this line of thought, MEAs with enhanced Pt utilization prepared by CCM method and by acid impregnated GDLs have been reported by our group [21]. It was found that the serious distortion of the membrane can be avoided, then a good contact between the CL and the membrane can be kept. At low platinum loadings, the CCM method exhibited much higher performance and Pt utilization compared with the MEA fabricated by GDE method. In this work, we prepared MEAs by the CCM method and the effects of different parameters, such as preparation method, binder type as well as the Pt loadings of the cathode and the anode, on the fuel cell performance of the so-prepared MEA were investigated. The cell performances were evaluate at 160 C with pure hydrogen and air as the reactants under ambient pressure. Polarization curves (IeV) and electrochemical impedance spectroscopies (EIS) were used to characterize various potential losses and variation of electrochemical properties. The results provide a more complete understanding for MEAs prepared by using CCM method for ABPBI membrane-based HT-PEMFC. 2. Experimental 2.1. Preparation of catalyst inks and fabrication of MEAs Before the CCM based-MEAs fabrication, homogeneous suspension of the catalyst inks were prepared by dispersing Pt/C catalyst (JM 40 wt.% Pt), binder (PTFE, PBI or PVDF) in extra solvent (DMAc for PVDF and PBI binder, IPA for PTFE binder) and then ultrasonicated for 1 h at room temperature. In the CL, the dry PTFE/ PVDF content is ~15 wt.%, while the PBI content in the CL is ~10 wt.%. In the work, all the MEAs with an active area of 2.3 2.3 cm2 were prepared by using an automated ultrasonic spraying technique [9]. Three types of CCM-based MEAs were investigated using the above-prepared catalyst inks. For clarity, the differences in the preparation of the three types of MEA are presented in Table 1. For the type-a MEAs, the catalyst inks were directly deposited onto the both sides of the dry ABPBI membranes (fumapem®AM, ~30 mm of thickness, FuMA-Tech). After the formation of CLs, the resulting electrodes were left in a vacuum oven for overnight drying. Finally, the MEAs were assembled by contacting CCM and two commercially GDLs (H2315-CX196, Freudenberg, Germany) impregnated with PA without a preceding hot-pressing step. The

Table 1 The differences in the MEAs preparation based on CCM method. MEA-type

Membrane status

CL fabrication

PA doping

Type-a Type-b Type-c

Dry Wet, PA-doped Dry

Directly on the membrane Decal transfer Directly on the membrane

GDL Membrane CCM

details of introducing PA can be found in our previous work [21]. The amount of PA pre-impregnated in the GDLs was calculated by the weight of the dry membrane (before CL coating) with the actual electrode area considering that PA redistribution mainly happened around the actual electrode area. The H3PO4 doping level is 3.8 molecules of H3PO4 per polymer repeating unit (PRU) [6]. Type-b MEAs were constructed by the decal transfer method [22e24], which is considered as a suitable way for CCM mass production. In this work, we are then motivated to examine its applicability on the preparation of HT-PEMFC MEAs. To prepare type-b MEAs (CCM-decal transfer), the CL was formed by spraying the catalyst inks onto the surface of a PTFE piece and then transferred onto the surface of the PA pre-doped ABPBI membrane by hot-pressing at 130 C under the pressure of 200 kgf cm 2 for 5 min. The H3PO4-doping process was carried out by soaking the ABPBI membranes in 85 wt.% PA solution for several hours at 100 C. The acid doping level in the membrane was about 3.8(±0.4) molecules of H3PO4 per PRU, which is similar with that for type-a MEA. The MEAs were assembled by contacting CCM and two commercially GDLs together without hotpressing. The type-c MEAs was fabricated by soaking the prepared CCMs in 85 wt % PA solution for several hours at 100 C, then contact with two GDLs without hot-pressing. The PA doping level was also controlled at ~3.8(±0.4) molecules of H3PO4 per PRU of the membrane. 2.2. Single-cell tests The prepared MEA was assembled with two gaskets made of fluorinated polymer into an HT-PEMFC cell fixture (BalticFuelCells GmbH, Germany) and then installed in a Cell Compression Unit (CCU, Pragma Industries, France). The cell fixture consists of two graphite plates with single serpentine channels (1.0 mm 1.0 mm 23 mm) and ribs (1.0 mm 23 mm). The active area is about 5 cm2 (23 mm 23 mm). Electrical heaters and a thermocouple were embedded into the plates and connected to the CCU which controlled the cell temperature at 160 C and the piston pressure at 1 N mm 2 in this study to minimize the electrical and thermal resistances of the GDLs [25]. A procedure and set-up details for fuel cell performance evaluation is referred in the previous work [21]. The flow rate of hydrogen and air is 0.2 and 0.5 L min 1, respectively. The cell was activated at 160 C and 0.5 V until the variation of current density was less than 5 mA in 5 min. 2.3. Physical and electrochemical characterization of the MEAs The surface morphology and the cross-section images of the MEAs were obtained by scanning electron microscopy (SEM) (Oberkochen, Germany). The cross-section of the samples were prepared by freeze-fracturing the MEAs in liquid N2. To determine the resistances of the MEAs, the in-suit electrochemical impedance spectroscopy (EIS) was performed at 0.6 V with a 5 mV amplitude and the frequency range of 0.1 Hze20,000 Hz.


123

3. Results and discussion 3.1. Effect of preparation method on the cell performance Fig. 1 shows the digital photographs of the PA-doped CCMs for the three typed MEAs with PVDF as binder. As seen as in Fig. 1(a), the size of the CCM after PA doping is almost same as its original, and the PA redistribution was restrained only around electrode active area (i.e., catalyst coated area). Therefore, the CCM surface still kept flat and no serious distortion occurred. However, using the decal transfer method, as showed in Fig. 1(b), the catalyst layer cannot be completely transferred to the PA-doped membrane, because the surface of the PA pre-doped membrane was wet due to the presence of PA. Therefore, the contact between the catalyst layer and the membrane may be not secure. As shown in Fig. 1(c), the type-c MEA seriously distorted after PA doping, which may result in CL detachment, then a poor contact between the CCM and the GDLs when being assembled. The polarization curves of the three prepared MEAs are presented in Fig. 2. It is clear that the type-a MEA shows better performance than the MEAs obtained from the other two methods. At

Fig. 2. Polarization curves of three typed MEAs with PVDF as binder, operated at 160 C under atmosphere pressure and 1 N mm 2 assembling torque of the test cell.

0.4 V, the current density of the type-a MEA reaches 620 mA cm 2, 2.5 times and 3.1 times to that of the type-b MEA (248 mA cm 2) and type-c MEA (200 mA cm 2), respectively. The peak power density of the type-a MEA can reach 278 mW cm 2 at 0.31 V and the limiting current density reaches up to 1200 mA cm 2. The good performance of the type-a MEA can be attributed to the combination of the better utilization of the catalyst, as well as the superior interface contact among the GDLs, membrane and CLs, resulting from directly depositing the CL on the membrane and the proper way of PA doping. On the contrary, the type-b MEA may suffer from the catalyst loss and the unsecured CL/membrane interfacial contact during the decal transfer process, resulting in an inferior catalyst utilization and high cell resistance, which should be the reason for the fast voltage drop in the ohmic polarization region. For type-c MEA, the similar trend of voltage drop in this region should originate from the seriously distorted CCM after PA doping (Fig. 1c), which makes difficulties when being assembled, leading to a high contact resistance. Moreover, the distorted CCM could cause differences in the stress state of the stiff catalyst coating area and soft bare membrane area, which may result in small fissures or pinholes in the CCM when it is assembled in the cell fixture. This could be the reason for the type-c MEA showing a much lower open circuit voltage (0.55 V) compared to the other two. Based on these results, the MEAs for the following studies were prepared in type-a. 3.2. Effect of binder on the cell performance

Fig. 1. Digital photographs of the PA-doped CCMs prepared by three methods. (a) CCM for type-a MEA before (left) and after (right) PA doping, (b) PTFE substrate after CL transfer (left) and the CCM for type-b MEA (right), (c) CCM for type-c MEA before (left) and after (right) PA doping.

Compared to low temperature (LT) PEMFC, high-boiling PA is suggested to act as the proton conductor in the CL for PBI membrane based high temperature PEMFC, instead of Nafion resins. The phosphoric acid is in the solution state, which can penetrate into the porous structure of the CL, whereas the colloidal Nafion cannot. The difference in the structures of the pore networks formed by particles of these binders within the CL direct affect the cell performance. Therefore, the binder properties have greatly influence on the CL mechanical properties, the gas permeability, PA impregnation, platinum utilization and ORR in the electrodes of the HT-PEMFC. Therefore, three common used binders for HT-PEMFC (PTFE, PVDF and PBI) were evaluated to investigate their effects on the MEA performance based on the CCM method, as shown in Fig. 3. It should be mentioned here that these binder contents in the CLs are different from each other, which were pre-optimized in our preliminary optimization experiments (not detailed here).

124


To further understand the performance differences of the MEAs with different binders, the in-situ EIS measurements were conducted at 0.4 V and 160 C. Fig. 4 shows the equivalent circuit (insert) and the impedance curves of the MEAs. It should be noted that RU represents the cell ohmic resistance, measured from the intercept at the real axis of the high frequency. Ra and Rc represent the distributed ionic resistance and the charge transfer resistance [28,29], respectively. The CPEa and CPEc represent a constant phase element. From Table 2, it is clear that RU of the MEA with PTFE as binder is significantly increased compared to that of with PBI and PVDF binders, which may arise from the increased proton transfer resistance in the membrane due to the insufficient PA doping from the GDLs because the highly hydrophobic PTFE CLs. The charge transfer resistance of the MEA with PVDF as binder is lowest among these three MEAs, which suggests that the catalyst layer with PVDF has more efficient electrochemical active layer. Therefore, PVDF is preferred as the CL binder for this study. Fig. 3. Polarization curves and power density curves of the type-a MEAs with different polymer binders, operated at 160 C under atmosphere pressure and 1 N mm 2 assembling torque of the test cell.

It is clear from Fig. 3, the performance of MEA with PVDF as binder in the CL is much better than that of the MEAs prepared with PBI or PTFE in whole current density region, which is mainly attributable to the nature physicochemical properties of the three binders since this is the only difference in these MEAs. At 0.4 V, the current density o reaches 620 mA cm 2 for the MEA with PVDF binder, while the value for the MEA with PTFE and PBI is only 500 mA cm 2 and 350 mA cm 2, respectively. For the type-a MEA, the PA was contained in the GDLs. Therefore, during fuel cell operation, the PA in the GDL will be transferred to the CL. PBI can easily absorb the aqueous PA to form a high conductive acidebase complex, which is beneficial to increase the proton conductivity in the CL [26]. Although PBI is considered a good candidate for membrane materials because the low gas permeability, the strong hydrophilicity of the PBI can lead to serious mass transport limitation due to the risk of the CL flooding by excess PA impregnation and the polymer film covered on the catalyst particles blocking gases transport [5,6]. This should be the reason why the MEA with PBI binder shows the worst performance and the limiting current density is only 450 mA cm 2. When PTFE is used as a binder, it exists as colloidal solid particles in the catalyst ink and easily forms large size agglomerates, which resulting the distribution of PTFE in CL might be less uniform and poor utilization of Pt catalyst [7]. Although in the case of MEAs prepared by GDE method, the danger of PA flooding and mass transport limitation by using PTFE as binder can be efficiently reduced with a sufficiently thick CL (>30 mm) [27]. However, in our case, it is difficult for the liquid PA doped in the GDL to penetrate through the thick hydrophobic CL, which results in an inferior PA doping level in the membrane, consequently leading to an increased ohmic resistance and the decreased cell performance, as presented in Fig. 3. In contrast, the PVDF binder exists in the CLs as a fiber phase, which makes catalyst particles less likely to be encapsulated in the binder, and thereby making more Pt surface available in the CLs [7]. The CL prepared with PVDF binder shows a more uniform and denser structure compared to that with PTFE binder, which was observed from previous study [7]. Moreover, the moderate hydrophobicity of PVDF favors both PA distribution and lowering the risk of PA flooding in the CL. Therefore, the MEA with PVDF binder yielded the best cell performance, which suggests that PVDF is the preferred CL binder for the MEAs created in this work.

3.3. Effect of Pt loading of cathode and anode on the cell performance The Pt loadings in the cathode and the anode were varied from 0.1 to 0.8 mg cm 2, respectively, to find the optimum value for the cell performance. Fig. 5 compares the cell performance of MEAs with various anode and cathode Pt loadings. Generally, the catalytic activity of CL improve with the increasing of Pt loading, which certainly helps the improvement of the MEA performance. It is clear that the cell performance was greatly improved with the Pt loading in the anode increasing from 0.1 mg cm 2 to 0.3 mg cm 2 (Fig. 5(a)). The current density at 0.4 V is increased by ~50%, from 410 mA cm 2 to 620 mA cm 2. In our previous report [21], it was observed that the ohmic resistance of MEA prepared by CCM method was affected the PA content in the membrane. Therefore, the MEAs with lower Pt loading, which possessed a thinner CL thickness, can contain more PA in the membrane than the MEA with high Pt loading since the PA doping level is same in each MEA. This could be the reason for the cell performance gradually reduced with further increasing the anode Pt loading from 0.3 mg cm 2 to 0.8 mg cm 2. Fig. 5(b) shows the performance changes with the changes of cathode Pt loading. It is easy to understand that the cell performance was greatly increased when the cathode Pt loading increased from 0.1 mg cm 2 to 0.5 mg cm 2 because the major voltage losses occur in cathode due to poor ORR kinetics. However,

Fig. 4. In-situ impedance curves of the MEA with different binders at 0.4 V and at 160 C.


125

Table 2 Fitted impedance parameters of the MEAs with different binders. Binder type

RU (U cm2)

Ra (U cm2)

Rc (U cm2)

PBI PTFE PVDF

0.367 0.482 0.354

0.012 0.008 0.006

0.384 0.182 0.142

for similar reason mentioned above, further increasing the Pt loading to 0.8 mg cm 2 caused a fast voltage drop in both ohmic and mass transfer regions, implying that an excessively thick CL is not favorable for the distribution and transfer of the proton conductor PA, which will result in increased cell resistance and mass transport limitations. It should be stated that the optimized Pt loadings (0.3/0.5 mg cm 2 for anode/cathode) for the MEA prepared with the CCM method are substantially lower compared with the traditional GDE method (normally above 0.7 mg cm 2 in a single GDE) [18,30e32], which is considered as a major advantage of the CCM method over GDE method in MEA fabrication. 3.4. Stability The stability of MEAs is of significance for the commercialization of HT-PEMFC [5]. To check the MEA stability prepared by the CCM

Fig. 6. Short-term discharge curve of the optimized CCM-type MEA operated at 300 mA cm 2 and at 160 C under ambient pressure.

method, a short-term operation at 160 C and 300 mA cm 2 was performed (Fig. 6). After the activations, the CCM-MEA exhibited good stability at the working current density of 300 mA cm 2: the voltage remained at ~0.45 V without obvious drop after ~140 h operation. Generally, it is believed that the loss of PA from the MEA is a major mechanism for the degradation of the PBI/ABPBI-based HT-PEMFCs [33], which caused by the high load conditions and high operating temperature, leading to PA removal from the MEA because the steam distillation mechanism [33]. Therefore, the good performance stability of the CCM-based MEA suggests that the CCM method was effective to keep required PA in the membrane and the CLs, resulting in low PA loss rates under the operating conditions. The average degradation is about 0.35 mV h 1, which is close the initial performance degradation rates estimated from other researchers' long-term durability results [34e36], implying the feasibility of the CMM-based MEA for long-term operation. A SEM analysis on the cross-section of the MEA prepared by CCM method before and after stability test shows in Fig. 7. The MEA after the durability test were pretreated by the following procedure: peel off the GDLs and then frozen fracturing the CCM in liquid N2. It can be seen from Fig. 7(a) that the good contact between the CLs and the ABPBI membrane was established from the CCM method. The dry ABPBI membrane is about 33.5 mm thick. Normally, the original thickness of PA pre-doped ABPBI membrane is normally about 80 mm [5]. After the durability test, as shown in Fig. 7(c), the ABPBI membrane still kept a thickness of about 72 mm, suggesting a satisfactory PA content remained in the membrane after the test. A reduction of CL thickness also can be observed from Fig. 7(c) after the durability test, which should result from the long cell compression process and the part catalyst loss during GDL removing. However, the good contact is still kept between the PA doped ABPBI membrane and the CLs. Fig. 7(b) and (d) show the partial enlarged views of the CL before and after the short-term operation, respectively. It is clear that the CL still keep a similar uniform porous structure after the short term operation. From these results, the CCM-based MEA showing high stability is attributable to the excellent interface contact between the CLs and the dry ABPBI membrane, as well as the good CL porous structure resulting from CCM method.

4. Concluding remarks Fig. 5. Polarization curves and power density curves of MEAs with different anode and cathode Pt loadings and PVDF as binder, operated at 160 C under atmosphere pressure and 1 N mm 2 assembling torque of the test cell.

ABPBI-based MEAs were fabricated by catalyst coating membrane (CCM) method for HT-PEMFC application. The effect of key

126


Fig. 7. SEM images of the surface and the cross-section of the MEA prepared by CCM method before (a) and after (c) a short-term operation. (b) and (d) is the partial enlarged view of the CL in (a) and (c), respectively.

parameters related on CCM fabrication, i.e. CCM type, binder properties and anode/cathode Pt loading, were investigated for the optimal performance of the resultant MEA. Following findings can be drawn from the data obtained: (1) For CCM-based MEA, pristine (dry) ABPBI membrane should be used for catalyst deposition, and the required PA should be pre-impregnated into the GDLs to avoid the CCM distortion, which can greatly reduce the ohmic resistance of the MEA. (2) PVDF was found to be a suitable CL binder for the CCM method owing to its polymer form and moderate hydrophobicity, which favors both PA distribution and lowering the risk of PA flooding in the CL. (3) The CCM method can be considered a promising way to reduce the Pt loading of HT-PEMFCs. The optimized Pt loading is only 0.3/0.5 mg cm 2 for anode/cathode, which are substantially lower compared to the traditional GDE method with an average Pt loading of ~0.8 mg cm 2. Although good stability of the MEA was observed in a shortterm (140 h) operation, long-term durability should be addressed in the future to validate the practicability of the CMM-based MEA. Also, further lowering the Pt loading to a comparable level with Nafion-based LT-PEMFCs (normally less than 0.2 mgPt cm 2) will be a target, also a challenge, in the future endeavors.

Acknowledgments This work is supported by Hydrogen and Fuel Cell Technologies RDI Programme (HySA), funded by the Department of Science and Technology in South Africa (project KP1-S01). Financial support from the NRF Free-standing Postdoctoral Fellowship (Grant No: 88347) is also acknowledged.

References [1] M.K. Debe, Nature 486 (2012) 43e51. [2] Q. Li, J.O. Jensen, R.F. Savinell, N.J. Bjerrum, Prog. Polym. Sci. 34 (2009) 449e477. [3] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, Z. Shi, D. Song, H. Wang, D.P. Wilkinson, Z.-S. Liu, S. Holdcroft, J. Power Sources 160 (2006) 872e891. [4] A. Chandan, M. Hattenberger, A. El-kharouf, S. Du, A. Dhir, V. Self, B.G. Pollet, A. Ingram, W. Bujalski, J. Power Sources 231 (2013) 264e278. [5] H. Su, S. Pasupathi, B.J. Bladergroen, V. Linkov, B.G. Pollet, J. Power Sources 242 (2013) 510e519. [6] H. Su, T.-C. Jao, S. Pasupathi, B.J. Bladergroen, V. Linkov, B.G. Pollet, J. Power Sources 246 (2014) 63e67. [7] H. Su, S. Pasupathi, B. Bladergroen, V. Linkov, B.G. Pollet, Int. J. Hydrog. Energy 38 (2013) 11370e11378. [8] H. Su, H. Liang, B.J. Bladergroen, V. Linkov, B.G. Pollet, S. Pasupathi, J. Electrochem. Soc. 161 (2014) F506eF512. [9] H. Su, T.-C. Jao, O. Barron, B.G. Pollet, S. Pasupathi, J. Power Sources 267 (2014) 155e159. [10] H. Su, C. Felix, O. Barron, P. Bujlo, B. Bladergroen, B. Pollet, S. Pasupathi, Electrocatalysis 5 (2014) 361e371. [11] J.J. Linares, C. Sanches, V.A. Paganin, E.R. Gonzalez, Int. J. Hydrog. Energy 37 (2012) 7212e7220. [12] G.-B. Jung, C.-C. Tseng, C.-C. Yeh, C.-Y. Lin, Int. J. Hydrog. Energy 37 (2012) 13645e13651. [13] H. Liang, L. Zheng, S. Liao, Int. J. Hydrog. Energy 37 (2012) 12860e12867. [14] H. Su, L. Xu, H. Zhu, Y. Wu, L. Yang, S. Liao, H. Song, Z. Liang, V. Birss, Int. J. Hydrog. Energy 35 (2010) 7874e7880. [15] S. Wang, Y. Shang, Y. Wang, J. Wang, Int. J. Hydrog. Energy 38 (2013) 11060e11066. [16] H. Tang, S. Wang, S.P. Jiang, M. Pan, J. Power Sources 170 (2007) 140e144. [17] C. Wannek, B. Kohnen, H.F. Oetien, H. Lippert, J. Mergel, Fuel Cells 8 (2008) 87e95. [18] C. Wannek, W. Lehnert, J. Mergel, J. Power Sources 192 (2009) 258e266. [19] K. Wippermann, C. Wannek, H.F. Oetjen, J. Mergel, W. Lehnert, J. Power Sources 195 (2010) 2806e2809. [20] C. Wannek, I. Konradi, J. Mergel, W. Lehnert, Int. J. Hydrog. Energy 34 (2009) 9479e9485. [21] H. Liang, H. Su, B.G. Pollet, V. Linkov, S. Pasupathi, J. Power Sources 266 (2014) 107e113. [22] M.S. Saha, D.K. Paul, B.A. Peppley, K. Karan, Electrochem. Commun. 12 (2010) 410e413. [23] Y.J. Yoon, T.-H. Kim, S.U. Kim, D.M. Yu, Y.T. Hong, J. Power Sources 196 (2011) 9800e9809. [24] A. Mehmood, H.Y. Ha, Int. J. Hydrog. Energy 37 (2012) 18463e18470.

H. Liang et al. / Journal of Power Sources 288 (2015) 121e127 [25] C. Felix, T.-C. Jao, S. Pasupathi, B.G. Pollet, J. Power Sources 243 (2013) 40e47. [26] J.-H. Kim, H.-J. Kim, T.-H. Lim, H.-I. Lee, J. Power Sources 170 (2007) 275e280. [27] P. Mazúr, J. Soukup, M. Paidar, K. Bouzek, J. Appl. Electrochem. 41 (2011) 1013e1019. [28] J. Lobato, P. Canizares, M.A. Rodrigo, J.J. Linares, F. Javier Pinar, Int. J. Hydrog. Energy 35 (2010) 1347e1355. [29] T.E. Springer, T.A. Zawodzinski, M.S. Wilson, S. Gottesfeld, J. Electrochem. Soc. 143 (1996) 587e599. [30] O. Barron, H. Su, V. Linkov, B.G. Pollet, S. Pasupathi, J. Appl. Electrochem. 44

127

(2014) 1037e1045. [31] E. van de Ven, A. Chairuna, G. Merle, S.P. Benito, Z. Borneman, K. Nijmeijer, J. Power Sources 222 (2013) 202e209. [32] J. Zhang, Y. Tang, C. Song, J. Power Sources 172 (2007) 163e171. [33] S. Yu, L. Xiao, B.C. Benicewicz, Fuel Cells 8 (2008) 165e174. [34] Y. Oono, A. Sounai, M. Hori, J. Power Sources 210 (2012) 366e373. [35] Y. Oono, T. Fukuda, A. Sounai, M. Hori, J. Power Sources 195 (2010) 1007e1014. [36] T.J. Schmidt, J. Baurmeister, J. Power Sources 176 (2008) 428e434.