lable at ScienceDirect

Journal of Power Sources 278 (2015) 718e724

Contents lists avai

Journal of Power Sources

journal homepage: www.elsevier .com/locate/ jpowsour

Enhanced performance and stability of high temperature protonexchange membrane fuel cell by incorporating zirconium hydrogenphosphate in catalyst layer

Olivia Barron, Huaneng Su*, Vladimir Linkov, Bruno G. Pollet, Sivakumar PasupathiHySA 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

� Incorporating Zr(HPO4)2 in the CLs of HT-PEMFC was evaluated.� The optimal content of Zr(HPO4)2 in the CLs was found to be 30 wt.%.� High single cell performance was delivered by the electrodes with Zr(HPO4)2.� The electrodes containing Zr(HPO4)2 showed high durability for HT-PEMFC operation.

a r t i c l e i n f o

Article history:Received 21 October 2014Received in revised form22 December 2014Accepted 29 December 2014Available online 29 December 2014

Keywords:Zirconium hydrogen phosphateHigh temperature polymer electrolytemembrane fuel cellCatalyst layerGas diffusion electrodeMembrane electrode assemblyPoly(2,5-benzimidazole)

* Corresponding author.E-mail address: suhuaneng@gmail.com (H. Su).

http://dx.doi.org/10.1016/j.jpowsour.2014.12.1390378-7753/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

Zirconium hydrogen phosphate (ZHP) together with polytetrafluoroethylene (PTFE) polymer binder isincorporated into the catalyst layers (CLs) of ABPBI (poly(2,5-benzimidazole))-based high temperaturepolymer electrolyte membrane fuel cell (HT-PEMFCs) to improve its performance and durability. Theinfluence of ZHP content (normalised with respect to dry PTFE) on the CL properties are structurallycharacterised by scanning electron microscopy (SEM) and mercury intrusion porosimetry. Electro-chemical analyses of the resultant membrane electrode assemblies (MEAs) are performed by recordingpolarisation curves and impedance spectra at 160 �C, ambient pressure and humidity. The result showthat a 30 wt.% ZHP/PTFE content in the CL is optimum for improving fuel cell performance, the resultantMEA delivers a peak power of 592 mW cm�2 at a cell voltage of 380 mV. Electrochemical impedancespectra (EIS) indicate that 30% ZHP in the CL can increase the proton conductivity compared to thepristine PTFE-gas diffusion electrode (GDE). A short term stability test (~500 h) on the 30 wt.% ZHP/PTFE-GDE shows a remarkable high durability with a degradation rate as low as ~19 mV h�1 at 0.2 A cm�2,while 195 mV h�1 was obtained for the pristine GDE.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

With the ever-present search for alternative energy sourcestaking up the majority of studies in recent years, Polymer Electro-lyte Membrane Fuel Cells (PEMFCs) are considered one of the mostpromising future power sources. Considering their capabilities ofproviding high efficiencies, high power densities and zero emissionpower sources [1], they have been extensively used in portable,transport and stationary applications. At present, the majority of

the research in this field focusses on low temperature PEMFCs (LT-PEMFCs) based on perfluorosulphonic acid (PFSA) membranes,such as Nafion®. These PFSAmembranes have proton conductivitiesthat are dependent upon their hydrated state and therefore theyare limited to operation at temperatures up to 100 �C underambient pressure in order to maintain a high water content in themembranes [2]. Several existing challenges on this technology areassociated with the low operating temperature, which include; (i)the need for complex fuel processing systems, due to the sensitivityof the Pt catalyst to CO and S poisoning at lower temperatures, (ii)poor quality of the generated heat and thus a greater difficulty intransferring the heat away to be used in other processes and (iii)complex water management systems for the humidification of the

Table 1Specifications of the GDEs.

GDE Catalyst type (% Pt/C)

Pt loading wt.% ZHP in CL (re. dryPTFE)

wt.% PTFE (re.Pt/C)

GDE-1

40 1 mg cm�2 0 40

GDE-2

40 1 mg cm�2 20 40

GDE-3

40 1 mg cm�2 30 40

GDE-4

40 1 mg cm�2 40 40

GDE-5

40 1 mg cm�2 50 40

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724 719

electrolyte membrane [3e5].The limitations of LT-PEMFCs have in recent years led to a shift

in focus from LT-PEMFCs to high temperature PEMFCs (HT-PEMFCs), which have several promising attributes over their lowtemperature counterpart, that accompany operation at tempera-tures between (100 �C e 200 �C). These advantages of a higheroperating temperature include but are not limited to; (i) fasterreaction kinetics and as a result the oxygen reduction reaction(ORR) is significantly increased, (ii) increased CO tolerance leadingto simpler fuel processing and thus a more cost effective fuel cell,(iii) higher quality of the generated waste heat resulting in easierremoval of the heat due to the larger temperature gradient betweenthe fuel cell and the ambient environment, leading to simpler fuelprocessing and ancillary systems and improving the overall effi-ciency of the fuel cells [5e7]. The higher operating temperatureinfluences the type of electrolyte membrane used in HT-PEMFCs,and since proton conducting electrolytes such as Nafion® dehy-drate at high temperature and low relative humidity (RH), elec-trolyte membranes have been developed which are more suitablefor high temperature operation. Efforts were first focussed onmodifying low temperature electrolytemembranes such as Nafion®

with metal oxide particles to improve water retention and thermalstability and developing acid-base polymer membranes [8e11].

Amongst the acid-base polymer membranes proposed for HT-PEMFC, polybenzimidazole (PBI) membranes were first investi-gated by Savinell et al. [12], these membranes exhibited goodproton conductivity when doped with phosphoric acid (PA), havegood thermal and mechanical stability, low gas permeability, aswell as requiring little to no humidification. Since Savinell et al. firstproposed PBI as an electrolyte membrane, various studies havebeen undertaken using PBI as an electrolyte membrane in HT-PEMFCs [13e16]. Phosphoric acid doped poly(2,5-benzimidazole)(ABPBI) membranes have more recently been used as electrolytemembranes for their improved proton conductivity over conven-tional phosphoric acid doped PBI membranes, in addition to beingcheaper to produce [17e19]. In this system, a proper amount ofphosphoric acid is required to impregnate the membranes and theCLs in order to achieve good proton conductivities [20]. However,the phosphate anions might adsorb onto the surface of Pt anddeactivate the catalyst [21], which causes sluggish kinetics of theORR and limits the fuel cell performance. Moreover, the lowpermeability of oxygen in phosphoric acid electrolyte is alsoconsidered as a primary contributor to the lower performancecompared with LT-PEMFC systems. While the majority of currentstudies have focused on improving the properties of the electrolytemembrane, less attention has been paid to the development of anionomer for the CL. Since phosphoric acid is used as an ionomer inthe CL, performance loss due to phosphate anion adsorption andthe resultant loss in proton conduction in the CL occurs at elevatedtemperatures. Furthermore, phosphoric acid dehydration at hightemperature can also cause proton conductivity loss in the CL. Inorder to maintain sufficient proton conductivity, incorporation ofinorganic proton conducting materials into the CL has beenconsidered [22].

Zirconium hydrogen phosphate, Zr(HPO4)2 (ZHP) is an insolublesolid that has been intensively studied as a proton conducting solidelectrolyte. ZHP exhibits a layered structure (which allows forintercalation of “guest” molecules) as well as cation exchangeproperties [23]. In addition to these cation exchange properties, italso displays good proton conductivity as a result of high protonmobility on the surface of ZHP [24], and high hygroscopicity atelevated temperatures which make it extremely attractive for useas polymer electrolyte [25,26]. At 80 �C, amorphous ZHP in waterexhibits a proton conductivity of ~0.01 S cm�1, which has led toseveral studies in which ZHP is incorporated into non-conducting

polymers such as polytetrafluoroethylene (PTFE) [27,28] as wellas conducting polymers such as Nafion® [29]. These electrolytes aremodified with ZHP to improve the moisture content and thermalstability of these membranes [29e31]. Since the incorporation ofZHP into electrolyte membranes has proven to be beneficial formoisture content and improved performance at temperaturesabove 100 �C, its presence in the CL should be beneficial for similarreasons. Xie et al. investigated the influence of ZHP in the CLs of gasdiffusion electrodes (GDEs) for MEAs based on Nafion® membraneand found that below 100 �C, the performance remained similar toGDEs containing Nafion® only, however, above 100 �C the ZHP/Nafion GDEs showed improved performance over the conventionalNafion GDEs [22]. For the same reasons, introducing ZHP into HT-PEMFC systems should also prove beneficial as there is a need forstable proton conductivity at elevated temperatures.

In this work, ZHP was incorporated into the CLs of ABPBI-basedHT-PEMFC along with a non-conducting polymer (PTFE), withoutadditional H3PO4 doping, to evaluate the effects of ZHP on (i) thefuel cell performance and (ii) the stability of the resulted GDEs. Insummary, this work expands on the work carried out by Xie et al.[22], by incorporating ZHP into the CLs of ABPBI-based HT-PEMFC.Similar techniques and methodologies are applied in this work,which at the same time, is considered to be a new step in theoptimisation challenge of the HT-PEMFC systems.

2. Experimental methods

2.1. MEA fabrication

Catalyst inks composed of Pt/C (40 wt.%, Johnson Matthey,HiSpec™ 4000), Zr(HPO4)2 (ZHP, Sigma Aldrich) and/or PTFE binder(60 wt.%, Electrochem Inc.) were dispersed in Isopropanol (Kimix)by ultrasonication (Grant Instruments, 38 kHz) for 2 h [32]. ThePTFE binder was normalised in relation to the Pt/C catalyst, with40 wt.% PTFE concentration used in all formulated inks, whereasthe Zr(HPO4)2 was normalised in relation to dry PTFE with varyingconcentrations used in the ink formulations. The catalyst inks weresprayed manually with an airbrush onto commercially availableGDL (Freudenberg H2135 CX 196) until the desired Pt loading wasachieved. All electrodes (anode and cathode) prepared in this studyhad a Pt loading of ~1.0 mg cm�2. It should bementioned that the Ptloading used in this study is close to those in commercial MEAs[33], however it is much higher than those for Nafion-based LT-PEMFCs (generally ~0.4 mgPt cm�2 [34]) due to the effect of H3PO4as mentioned before. The PTFE binder loading in the CL was fixed at40 wt.%, while the Zr(HPO4)2 loading in the CL was varied from 20%to 50% in relation to dry PTFE. Finally, all prepared GDEs were curedat 200 �C in a vacuum oven (Binder GmbH). Table 1 shows thecomposition of all GDEs prepared for testing in this study.Commercially available Fumapem® (Fumatech) ABPBI membrane

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724720

was doped by immersing in 85% Phosphoric Acid (KIMIX) at 95 �Cfor 24 h to obtain an acid doping level of about 3.8 molecules ofH3PO4 per polymer repeating unit (PRU). Any superficial acid wasgently removed by tissue prior to the MEA assembly.

2.2. Physical characterisation of GDEs

The pore-size distributions of the PTFE GDE and the ZHP/PTFEGDEs were determined by performing mercury intrusion poros-imetry on the GDEs. Mercury intrusion porosimetry obtains thepore-size distributions by injecting mercury into the samples atfixed pressures while recording the injected volume of mercuryunder quasi-steady conditions. The pressure and volume are relatedusing various expressions for capillary pressure and radii as ameans for determining the distribution of effective pore sizes. Athigh pressures the mercury enters the small pores and at lowpressures larger pores are sampled [35]. An Auto Pore IV 9500(Micromeritics) porosimeter was used for porosity measurements.A high-resolution Scanning Electron Microscope (SEM) (NovaNanoSEM 230, FEI) was used to observe the surface morphologyand porous microstructure of the GDEs. Energy dispersive X-raySpectroscopy (EDS) was used to obtain an elemental profile of theZHP-GDE.

Table 2Elemental analysis (all results in wt.%).

Spectrum In stats. C F Zr Pt Total

Spectrum 1 Yes 50.75 22.85 6.6 19.8 100Spectrum 2 Yes 52.58 23.45 5.83 18.15 100

2.3. Single cell performance evaluation

The MEAs were obtained by sandwiching the acid-doped ABPBImembrane between the anode and cathode GDEs together inside asingle cell fixture (Pragma Industries, France) without any priorhot-pressing procedure. The single cell fixture consists of twographite plates with serpentine flow fields with an active area of5 cm2. A thermocouple and electrical heaters are embedded in theplates enable temperature control of the cell by a Cell CompressionUnit (CCU, Pragma Industries, France). The cell fixturewas placed inthe CCU, which controlled the cell temperature at 160 �C andmaintained the piston pressure of 2 N mm�2 during operation. Anin-house HT-PEMFC test-stand was used to perform electro-chemical evaluations on the MEAs. The test-stand consists of anelectronic load, Arbin BT2000 (Arbin Instruments) connected to acomputer. Pure hydrogen was fed to the anode and air was fed tothe cathode with flow rates of 0.5 slpm and 1.0 slpm respectively.The MEAs were activated prior to testing, by applying a constantvoltage of þ0.55 V until a stable performance was achieved. Thepolarisation curves were recorded by measuring the cell voltage as

Fig. 1. Surface morphologies with inserts of CL pore structure of

a function of current between Open Circuit Voltage (OCV)and þ0.2 V.

2.4. Electrochemical characterisation

A Potentiostat/Galvanostat (Autolab PGSTAT 302N, Metrohm)equipped with a Frequency Response Analyser (FRA) and a 20 Acurrent booster (Autolab BSTR 20A, Metrohm) was used to performElectrochemical Impedance Spectroscopy (EIS) measurements. Themeasurements were carried out at a cell voltage of þ0.6 V, in the0.1 Hz e 20 kHz frequency range with an amplitude of þ0.01 V.Autolab Nova software was used to generate and simulate theimpedance data.

3. Results and discussion

3.1. Influence of ZHP on the catalyst layer structure

Fig. 1(a, b) shows the HR-SEM images of the GDEs prepared with40% PTFE and 40% PTFE-30% ZHP combination. In order to performa systematic comparison between the GDEs, the images were takenin secondary electron detector mode at 1000� magnification. Thehand spraying method for deposition of the catalyst ink results inan uneven deposition of the catalyst, and as a result large lumps canbe observed on the surface of both GDEs. The Zr particles areobserved as white clusters in Fig. 1(b). Looking at the 50,000�magnification inserts in Fig. 1, one may observe that both GDEsexhibit a porous microstructure necessary for transport of reactantgases to the catalyst sites, with no distinguishable differencesobserved in the GDEs. Elemental analysis data shown in Table 2revealed a nearly uniform distribution of zirconium in the CL ofGDE-3, with a 1:3 ratio for Zr:Pt which meets the designrequirements.

Fig. 2 shows the effect of various ZHP loadings on the GDEsurface morphology and CL microstructure. Increasing the ZHPloading effectively causes a corresponding increase in uneven na-ture of the surface of the GDEs, with more lumps and agglomerates

(a) PTFE-GDE (GDE-1) and (b) 30% ZHP/PTFE-GDE (GDE-3).

Fig. 2. Surface morphologies with inserts of CL pore structure of (a) 20% ZHP/PTFE-GDE (GDE-2), (b) 40% ZHP/PTFE-GDE (GDE-4) and (c) 50% ZHP/PTFE-GDE (GDE-5).

Fig. 3. Mercury intrusion porosimetery showing the incremental intrusion of (a) thePTFE-GDE (GDE-1) and the 30% ZHP/PTFE-GDE (GDE-3); (b) the GDEs with varyingZHP concentrations in the CL; GDE-1 (30% ZHP/PTFE), GDE-3 (20% ZHP/PTFE), GDE-4(40% ZHP/PTFE) and GDE-5 (50% ZHP/PTFE).

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724 721

observed for the higher ZHP loadings (Fig. 2(b,c)). Little differencein the porous microstructure of the GDEs is observed in the highmagnification inserts of Fig. 2.

A more detailed analysis of the microstructure of the GDEs canbe obtained from the porosimetry data. Fig. 3 shows the incre-mental intrusion data for the GDEs 1e5. For ease of analysis Fig. 3(a)shows a comparison of intrusion data for the 30% ZHP/PTFE-GDE(GDE-3) and PTFE-GDE (GDE-1). It can be seen that the additionof ZHP to the CL causes a slight decrease in the number of pores inthe <0.03 mm region, indicating a slight reduction in the number ofmicropores upon addition of ZHP, indicating that the ZHP could befilling the smaller diameter pores during the formation of the CL,which corresponds to the findings of Xie et al. [19]. However,introducing ZHP in the CL results in an increased pore volume forthemacropores (30e100 mm region), whichmay be originated fromthe increase of the catalyst agglomerates in the CL due to theincreased ZHP content. Fig. 3(b) shows the intrusion data for theGDEs with varying ZHP loadings, GDEs (2e5). The porosimetry datashows that the ZHP/PTFE-GDEs have similar intrusion volumesacross all pore sizes with the exception of the macroporous region(30e100 mm), where the 30% ZHP/PTFE GDE (GDE-3) shows ahigher pore volume than the other ZHP/PTFE GDEs. A higher porevolume in this region is beneficial for the molecular diffusionmechanism of gas to the catalyst sites [36]. GDEs with higher porevolumes in this region are expected to have better mass transportproperties in high current density regions.

3.2. Single cell performance

Fig. 4(a) shows the performance curves for two MEAs withdiffering CL compositions in the GDEs. The performance of theMEAbased on the PTFE-GDE (GDE-1) shows a lower performance thanthat achieved by the ZHP/PTFE-GDE (GDE-3), indicating that theaddition of ZHP to the CL enhances the performance of the MEA,particularly in the medium and high current density regions of thepolarisation curve. Themaximumpower density of the twoMEAs isachieved by GDE-3, with 592 mW cm�2 at a cell voltageof þ380 mV, which is an improvement of approximately 14% overGDE-1 which only achieved a maximum power density of518 mW cm�2 at a cell voltage of þ331 mV. Although the consid-erable improvement is achieved, this performance is still lowerthan those of Nafion-based PEMFCs with much lower Pt loadings.This is because the sluggish kinetics of the ORR and the transportlimitations of protons and reactants in cathode, especially in thepresence of H3PO4, limit the cell performance of HT-PEMFC. How-ever, it should be mentioned that the mass transport limitation ofthe HT-PEMFC was greatly reduced compared with LT-PEMFCs, due

to the elimination of the liquid water, which is the reason why themaximum power density of the HT-PEMFC can be reached at lower

Fig. 4. (a) Performance curves for ABPBI-based MEAs prepared with 40% PTFE in thecatalyst layer (GDE-1) and 40% PTFE and 30% ZHP in the catalyst layer (GDE-3); (b) In-situ impedance curves of the MEAs for GDE-1 and GDE-3, at a cell voltage of þ0.6 V.

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724722

operating voltage (þ380 mV and þ331 mV) while that for the LT-PEMFC was much higher (e.g. þ470 mV) even with an improvedflow field design [37]. The polarisation curves also show that at aworking cell voltage of þ600 mV GDE-3 reaches a current densityof 399 mA cm�2, which is approximately 25% higher than thatachieved by GDE-1 (320 mA cm�2) at the same cell voltage.

Detailed analysis of the polarisation curves shows that the GDEshave a similar voltage drop in the low current density region(<100mA cm�2) of the curve, indicating the electrodes have similarreaction kinetics. Since the voltage drop in this region is predom-inantly determined by sluggish ORR kinetics, it can be stated thatintroducing Zr(HPO4)2 in the CLs does not affect ORR kinetics.

Analysis of the linear region of the polarisation curve shows thatMEAs exhibit similar decreasing slopes, while GDE-1 exhibitsslightly lower performance than GDE-3 in this region. This obser-vation indicates that the MEAs have similar ohmic resistances, asthis region of the polarisation curve is influenced by ohmic resis-tance. The high current density region (>1000 mA cm�2) of thecurvewhich is influenced bymass transport limitations, shows thatboth GDEs are not significantly influenced by mass transport lim-itations as no sharp drop from the linear region of the curve is

observed for either MEA. GDE-3 exhibits a better performance thanGDE-1 in the high current density region, it can thus be stated thatGDE-3 has better mass transport properties, which corresponds tothe larger number of macropores exhibited in Fig. 3(a). Thisapparent lack of mass transport limitations for the GDEs may beattributed to: (i) the increase in temperature leading to not only inan increase in the reaction rate but also an increase in the gasdiffusion rate through the electrolyte membrane as well as the GDEand, (ii) a single phase of gaseous water leading to an increasedsurface area of the catalyst and improving the ability of reactantgases to diffuse into the reactant layer, as gaseous water does notblock the active sites as liquid water does in LT-PEMFCs, as thephenomenon of catalyst flooding does not occur in HT-PEMFCs[1,38,39].

Fig. 4(b) shows the in-situ impedance curves of the single cellswith the two different GDEs obtained atþ0.6 V. The high frequencyintercept with the x-axis shows that the MEAs have similar ohmicresistances, whereas the charge transfer resistances show a sig-nificant difference for these two MEAs. Since the charge transferresistance can be calculated from the diameter of the arc, it is clearthat the MEA of GDE-3 has the lowest charge transfer resistance,implying that this GDE has the more efficient electrochemicalactive layer, which can be attributed the increased proton con-ductivity of the CL due to the incorporation of ZHP. In practice theperformance of PEMFC stack often differs to that obtained for asingle PEMFC cell, with the PEMFC stack having a much higheroperating voltage, higher power and better fuel-energy efficiency[40]. The results obtained here indicate that the incorporation ofZHP into the CL of single cell MEAs enhances the performance ofthese high temperature MEAs, it should however, be noted thatresults obtained in single cells are often harder to reproduce inscaled-up fuel cell stacks. For instance, Bonnet et al. [41] found thatthe voltage of the single cell increased with relative humidity (RH)but the voltage of the small stack decreased with higher RH at highcurrent densities. The mass transfer behaviour of fuel cell stacks isoftenmore complicated than that of single cells due to the presenceof other factors such as, heat-exchange, humidity effects as well asair and fuel supplies [40].

3.3. Influence of ZHP content on MEA performance

Fig. 5(a) clearly shows that the addition of ZHP to the catalystlayer of the GDE increased the MEAs performance, however, aninvestigation into the effect of different ZHP concentrations showsthat there is an optimum value for ZHP in the catalyst layers. In thisstudy 20e50 wt.% ZHP in relation to dry PTFE was evaluated in theCL. Fig. 5(a) shows the performance curves of the MEAs withdifferent ZHP contents in the CL.

For comparison purposes, plots of current density at þ600 mVversus ZHP content in CL and power density versus ZHP content inCL are shown in Fig. 5(b) and (c) respectively. In Fig. 5(b) it can beobserved that the cell performance is slightly improved byincreasing ZHP content from 20 wt% to 30 wt% in the CLs, howeverfurther increase of ZHP content to 40 wt% and 50 wt% leads to adramatic drop at the same cell voltage, which could be attributed tothe increased ohmic resistance due to higher ZHP content in theseCLs. The same trend is observed in Fig. 5(c) where the maximumpower density is achieved by GDE-3, with 592 mW cm�2 achievedat þ380 mV. It is clear from these graphs that high ZHP contents(40 wt.% and 50 wt.%) in the CL are not favourable to higher fuel cellperformance.

Although MEAs with GDE-2 and GDE-3 reach similar currentdensities at þ600 mV (Fig. 5(b)), the peak power densities reachedby these MEAs show a distinct difference (Fig. 5(c)). In the mediumcurrent density region these MEAs exhibit similar performance, but

Fig. 5. (a) Performance curves at 160 �C and ambient pressure of the MEAs with different ZHP content in the CLs; (b) Current density at 600 mV versus the ZHP content in CL; (c)Maximum Power density versus ZHP content in CL; (d) In-situ impedance curves for MEAs with differing ZHP contents in CL, at a cell voltage of þ0.6 V.

Table 3Resistances of single cells with various GDEs.

GDE-# GDE-2 GDE-3 GDE-4 GDE-5

RU (U cm2) 0.104 0.096 0.114 0.105Rct (U cm2) 0.210 0.207 0.224 0.241

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724 723

GDE-3 clearly outperforms GDE-2 in the high current density re-gion. Normally, the fuel cell performance in the high current den-sity region is related to mass transport limitations, therefore it isclear from Fig. 5(aec) that GDE-3 has the superior CL compositionin terms of mass transport properties and MEA performance due tothe optimized ZHP content. The in-situ impedance curves of thefour MEAs at a cell voltage of þ0.6 V are shown in Fig. 5(d). Thehigh-frequency intercept on the real axis represents the total ohmicresistance of the single cell, while the diameter of the arc is ameasure of the charge transfer resistance of the ORR. Throughsimulation with Autolab software, the cell resistances (RU) andcharge transfer resistances (Rct) of the single cell with the differentGDEs can be calculated, and summarised in Table 3. It can be seenthat GDE-3 possesses the lowest ohmic resistance and chargetransfer resistance, indicating a more efficient electrochemicalactive layer due to the optimum ZHP content in the CL. High re-sistances are observed with GDEs containing high ZHP contents(40 wt.% and 50 wt.%), these results are certainly consistent withtheir performances showed in fuel cell operation (Fig. 5(a)).

3.4. Stability

Since stability and durability characteristics are one of the mainchallenges associated with HT-PEMFCs [38], a short term stabilityanalysis study was performed for 500 h at j ¼ 0.2 A cm�2, as shown

in Fig. 6. The experiment was started after performing two polar-isation tests which took place on two successive days, the cellvoltage at 0.2 A cm�2 was at its maximum directly after the secondpolarisation test was performed. As can be seen in Fig. 6, theMEA ofGDE-3maintains a stable performance for the duration of the study,with the exception of the occurrence of minor fluctuations due to adisruption in H2 gas supply to the cell, the cell voltage shows nomajor decrease at 0.2 A cm�2. When the stability of GDE-1 iscompared with that of GDE-3, it can be observed that GDE-1 has amuch steeper slope than that of GDE-3. Linear regression of the cellvoltage data obtained for the MEAs reveals that the degradationrate of GDE-3 is as low as ~19 mV h�1, while GDE-1 shows a muchhigher degradation rate of ~195 mV h�1. Since the only differencebetween these two MEAs are the different compositions of the CLs,it can be stated that the addition of the 30 wt.% Zr(HPO4)2 (re. dryPTFE) to the CL can greatly increase the stability of the cell per-formance. The degradation rate of ~19 mV h�1 obtained by GDE-3 iswell within the range for those reported by other researchers'

Fig. 6. Stability of GDE-1 and GDE-2 MEAs performance, operating for 500 h at0.2 A cm�2.

O. Barron et al. / Journal of Power Sources 278 (2015) 718e724724

(4.9e25 mV h�1) [33,42e45]. The good stability displayed by GDE-3can be explained by; the good thermal stability provided by ZHPcombined with the improved proton conductivity at higher tem-peratures. Therefore, incorporating the hygroscopic ZHP particlesin the CL, which aids the proton conductivity and stability of theGDE at higher/operating temperature, is believed to be the mainreason for the GDE-3 showing high performance and gooddurability.

4. Conclusions

Incorporating ZHP into the CL of GDEs yielded high perfor-mances of ABPBI-based MEAs. A 30 wt.% ZHP (re. dry PTFE) contentin the CL yielded the best performing MEA, which achieved peakpower of 592 mW cm�2 at a cell voltage of 380 mV cm�2. An in-crease in ZHP content in the CL led to lower performances, indi-cating that lower ZHP contents in the CL are more beneficial forachieving high power densities. A reduction in charge transferresistance was observed for the ZHP GDE when compared to theGDE containing PTFE-only, indicating that the addition of ZHP canincrease the proton conduction of the CL. The MEA showed goodstability in a short term operation: the cell voltage remained at~þ0.65 V without obvious drop after the 500 h operation at0.2 A cm�2. It may be stated that an optimum content of ZHP in theCL benefits the performance and durability of ABPBI-based HT-PEMFC.

Acknowledgements

This work is supported by Hydrogen and Fuel Cell TechnologiesRDI Programme (HySA), funded by the Department of Science andTechnology in South Africa (project KP1eS01). The authors' thankMiranda Waldron at the Centre for Imaging and Analysis at theUniversity of Cape Town for SEM analysis.

References

[1] Y. Shao, G. Yin, Z. Wang, Y. Gao, J. Power Sources 167 (2007) 235e242.[2] R.N.G. Alberti, M. Sganappa, J. Power Sources 178 (2008) 575e583.[3] X. Yu, B. Zhou, A. Sobiesiak, J. Power Sources 147 (2005) 184e195.[4] H.W. Wei Dai, Yuan Xiao-Zi, Jonathan J. Martin, Daijun Yang, J.M. Jinli Qiao, Int.

J. Hydrogen Energy 34 (2009) 9461e9478.[5] 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.[6] Z.X. Jianlu Zhang, Jiujun Zhang, Yanghua Tang, Chaojie Song, Z.S. Titichai

Navessin, Datong Song, Haijiang Wang, Z.-S.L. David, P. Wilkinson,Steven Holdcroft, J. Power Sources 160 (2006) 872e891.

[7] Q. Li, R. He, J.-A. Gao, J.O. Jensen, N.J. Bjerrum, J. Electrochem. Soc. 150 (2003)A1599eA1605.

[8] Q. Li, J.O. Jensen, R.F. Savinell, N.J. Bjerrum, Prog. Polym. Sci. 34 (2009)449e477.

[9] V. Baglio, A.S. Aric�o, A.D. Blasi, V. Antonucci, P.L. Antonucci, S. Licoccia,E. Traversa, F.S. Fiory, Electrochim. Acta 50 (2005) 1241e1246.

[10] N.H. Jalani, K. Dunn, R. Datta, Electrochim. Acta 51 (2005) 553e560.[11] Y.M. Kim, S.H. Choi, H.C. Lee, M.Z. Hong, K. Kim, H.-I. Lee, Electrochim. Acta 49

(2004) 4787e4796.[12] J.S. Wainright, J.T. Wang, D. Weng, R.F. Savinell, M. Litt, J. Electrochem. Soc.

142 (1995) 121e123.[13] J. Lobato, P. Ca~nizares, M.A. Rodrigo, J.J. Linares, Electrochim. Acta 52 (2007)

3910e3920.[14] M.S. Kondratenko, M.O. Gallyamov, A.R. Khokhlov, Int. J. Hydrogen Energy 37

(2012) 2596e2602.[15] M. Mamlouk, K. Scott, Int. J. Hydrogen Energy 35 (2010) 784e793.[16] R.H. Qingfeng Li, Jens Oluf Jensen, Niels J. Bjerrum, Chem. Mater. 15 (2003)

4896e4915.[17] J.A. Asensio, P. G�omez-Romero, Fuel Cells 5 (2005) 336e343.[18] J. Asensio, J. Memb. Sci. 241 (2004) 89e93.[19] J.J. Linares, C. Sanches, V.A. Paganin, E.R. Gonzalez, J. Electrochem. Soc. 159

(2012) F194eF202.[20] C. Wannek, I. Konradi, J. Mergel, W. Lehnert, Int. J. Hydrogen Energy 34 (2009)

9479e9485.[21] J.O. Park, K. Kwon, M.D. Cho, S.G. Hong, T.Y. Kim, D.Y. Yoo, J. Electrochem. Soc.

158 (2011) B675eB681.[22] Z. Xie, T. Navessin, Z. Shi, R. Chow, S. Holdcroft, J. Electroanal. Chem. 596

(2006) 38e46.[23] A. Clearfield, J.A. Stynes, J. Inorg. Nucl. Chem. 26 (1964) 117e129.[24] G. Alberti, M. Casciola, U. Costantino, G. Levi, G. Ricciardi, J. Inorg. Nucl. Chem.

40 (1978) 533e537.[25] A.Y.T. Amani Al-Othman, Wendy Pell, Sadok Letaief, B.A.P. Tara, J. Burchell,

Marten Ternan, J. Power Sources 195 (2010) 2520e2525.[26] C.B. Amphlett, L.A. McDonald, M.J. Redman, J. Inorg. Nucl. Chem. 6 (1958)

220e235.[27] S.Ü. Çelik, A. Bozkurt, S.S. Hosseini, Prog. Polym. Sci. 37 (2012) 1265e1291.[28] Y.-I. Park, J.-D. Kim, M. Nagai, J. Mater. Sci. Lett. 19 (2000) 1735e1738.[29] C.Y.P. Costamagna, A.B. Bocarsly, S. Srinivasan, Electrochim. Acta 47 (2002)

1023e1033.[30] C. Yang, J. Memb. Sci. 237 (2004) 145e161.[31] M.L. Hill, Y.S. Kim, B.R. Einsla, J.E. McGrath, J. Memb. Sci. 283 (2006) 102e108.[32] B. Pollet, Electrocatalysis 5 (2014) 330e343.[33] T.J. Schmidt, J. Baurmeister, J. Power Sources 176 (2008) 428e434.[34] K.S. Choi, B.G. Kim, K. Park, H.M. Kim, Fuel Cells 12 (2012) 908e938.[35] N. Parikh, J.S. Allen, R.S. Yassar, Fuel Cells 12 (2012) 382e390.[36] H.-K. Lee, J.-H. Park, D.-Y. Kim, T.-H. Lee, J. Power Sources 131 (2004)

200e206.[37] K.-S. Choi, H.-M. Kim, S.-M. Moon, Electrochem. Commun. 13 (2011)

1387e1390.[38] 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.[39] H. Su, T.-C. Jao, S. Pasupathi, B.J. Bladergroen, V. Linkov, B.G. Pollet, J. Power

Sources 246 (2014) 63e67.[40] R.J. Deryn Chu, J. Power Sources 80 (1999) 226e234.[41] C. Bonnet, S. Didierjean, N. Guillet, S. Besse, T. Colinart, P. Carr�e, J. Power

Sources 182 (2008) 441e448.[42] H. Su, S. Pasupathi, B.J. Bladergroen, V. Linkov, B.G. Pollet, J. Power Sources 242

(2013) 510e519.[43] B.K.C. Wannek, H.-F. Oetjen, H. Lippert, J. Mergel, Fuel Cells 8 (2008) 87e95.[44] L.X.S. Yu, B.C. Benicewicz, Fuel Cells 8 (2008) 165e174.[45] A.D. Modestov, M.R. Tarasevich, V.Y. Filimonov, N.M. Zagudaeva, Electrochim.

Acta 54 (2009) 7121e7127.