Membranes and MEAs Based on Sulfonated Poly(ether ketone ketone

0 downloads 7 Views 314KB Size Report
Perfluorosulfonated polymer electro- lyte membranes (PEMs) such as Nafion IEC = 0.9 mequiv/g in- trinsically have a higher acid ionization constant (lower pKa) ...

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲

B532

0013-4651/2008/155共6兲/B532/6/$23.00 © The Electrochemical Society

Membranes and MEAs Based on Sulfonated Poly(ether ketone ketone) and Heteropolyacids for Polymer Electrolyte Fuel Cells Vijay Ramani,a,*,c,z Steven Swier,b,d M. T. Shaw,a,b R. A. Weiss,a,b H. R. Kunz,a,* and J. M. Fentona,**,e a Department of Chemical Engineering and bInstitute of Materials Science, University of Connecticut, Storrs, Connecticut 06269-3136, USA

Organic sulfonated poly共ether ketone ketone兲 共SPEKK兲 membranes with different ion-exchange capacities 共IECs兲, and composite membranes prepared by the addition of 20 wt % phosphotungstic acid 共PTA兲 to SPEKK were used to prepare membrane electrode assemblies 共MEAs兲. The proton conductivity of the membranes increased with increasing IEC of the SPEKK, and with the addition of PTA. The proton conductivity attained at 80°C and 75% relative humidity was 20 ⫾ 2 mS/cm. The feasibility of using SPEKK in the cathode layer of the MEAs was investigated. The electrochemically active surface areas 共ECAs兲 of the SPEKK-based cathodes were lower than that of the Nafion-based cathode and decreased further as the operating relative humidity was lowered. These observations were reflected in the single-cell polarization data, which indicated that the MEAs with the SPEKK-based electrodes were outperformed by their Nafion-based counterparts. Furthermore, a mismatch in SPEKK IEC between the membrane and cathode resulted in immiscibility at the interface. While the additive stability in the composite membrane was very good, the long-term stability of the membranes was poor when compared to perfluorosulfonic acid membranes such as Nafion, with failure occurring by scission along the gasket edges of the MEA after limited operation. © 2008 The Electrochemical Society. 关DOI: 10.1149/1.2898171兴 All rights reserved. Manuscript submitted October 23, 2007; revised manuscript received February 13, 2008. Available electronically April 8, 2008.

Sulfonated hydrocarbons offer an alternative to perfluorosulfonic acid membranes such as Nafion for polymer electrolyte fuel cell 共PEFC兲 and direct methanol fuel cell 共DMFC兲 applications. Advances in the area of hydrocarbon-based membranes for fuel cell applications have been summarized in several recent reviews.1-8 Composite membranes based on Nafion and heteropolyacids 共HPAs兲 have been previously studied9-11 in an attempt to reduce the dependence of membrane conductivity on water content. Stabilization of the HPA additive within the ionomeric matrix, designed to prevent HPA leaching in aqueous environments, has recently been demonstrated in Nafion-based systems.12 Studies have also been performed on composite membranes prepared using hydrocarbon-based matrices that contain HPAs.13-15 These studies attest to the utility of HPA additives in membranes designed for medium to high relative humidity 共RH兲 PEFC operation. Sulfonated poly共ether ketone ketone兲 共SPEKK兲16 is a protonconducting material with good film-forming properties. However, the conductivity of SPEKK is strongly dependent on the water uptake of the membrane and, concomitantly, on the ion exchange capacity 共which in turn is a function of the degree of sulfonation兲. The operating range and composition of SPEKK membranes are therefore limited to fully saturated conditions and high ion exchange capacities 共IECs兲, respectively. Perfluorosulfonated polymer electrolyte membranes 共PEMs兲 such as Nafion 共IEC = 0.9 mequiv/g兲 intrinsically have a higher acid ionization constant 共lower pKa兲 when compared to SPEKK, which implies that a higher IEC is needed for SPEKK to obtain the same conductivity under a given set of operating conditions. However, the mechanical properties of the SPEKK films deteriorate with increasing IEC. This results in low operational lifetimes for membrane electrode assemblies 共MEAs兲 prepared using high IEC SPEKK. Hence, a trade-off exists between proton conductivity and operational lifetime in the case of SPEKK membranes. One approach to get around this trade-off is to prepare composite membranes with a low IEC organic matrix supplemented with inorganic additives to enhance proton conductivity. An added benefit of

this approach is that the inorganic particles will also serve to impede the crossover of methanol from anode to cathode. To this end, SPEKK/HPA composite membranes using low IEC SPEKK were prepared and studied, with the results obtained discussed in this paper. While significant attention has been devoted to membrane development, many of the hydrocarbon-based membranes detailed in the references above have been evaluated ex situ to determine proton conductivity. MEAs have been prepared using these alternate membranes.17-20 However, until recently the electrode layers in these MEAs have been either devoid of proton-conducting material or employed perfluorosulfonic acid ionomers such as Nafion as the proton-conducting component and binder. While Nafion is an admirable proton-conducting binder, concerns exist about the integrity of the membrane electrode interface due to the mismatch that exists between the hydrocarbon-based membranes and the fluorocarbonbased Nafion. Commercialization of hydrocarbon-based membranes could well be contingent on the concomitant development of compatible proton-conducting materials to be used in the electrode layers. Recognizing this, Lakshmanan et al.21 and Easton et al.22 investigated the influence of the addition of sulfonated poly共ether ether ketone兲 共SPEEK兲 as a binder in PEM gas diffusion electrodes. Jung et al.23 have investigated the effect of SPEEK as an electrode binder in DMFCs. In this study, MEAs were prepared for the first time using SPEKK binders with varying IECs in the cathode. Unlike prior studies involving SPEEK, the MEAs were prepared by applying the catalyst layer directly to the membrane 共as opposed to applying the catalyst layer to the gas diffusion layer, followed by hot pressing onto the membrane兲. This technique improves the utilization of electrocatalyst in the MEA. The MEAs were characterized by linear sweep voltammetry 共LSV兲, cyclic voltammetry 共C-V兲, and polarization tests to determine the feasibility of using SPEKK as the protonconducting material in the cathode layer to promote interfacial stability.

Experimental * Electrochemical Society Active Member. ** Electrochemical Society Fellow. c

Present address: Illinois Institute of Technology, Chicago, Illinois, USA. d Present address: Dow Corning, Midland, Michigan 48640, USA. e Present address: Florida Solar Energy Center, University of Central Florida, Orlando, Florida 32922, USA. z E-mail: [email protected]

Materials.— SPEKK.— PEKK with a terephthaloyl 共T兲 to isophthaloyl 共I兲 ratio of 6/4 共OXPEKK SPb, Tg = 154°C兲 was obtained from Oxford Performance Materials, Enfield, CT. Sulfonation of PEKK was performed in a 5% 共w/v兲 mixture of 53/47 共v/v兲 concentrated sulfuric acid and fuming sulfuric acid as described in Ref. 16.

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲 The sulfonation level, expressed as an ion-exchange capacity, defined as the concentration of sulfonate groups in milliequivalents per gram, was determined by titration. SPEKK/HPA and SPEKK-modified HPA membranes.— SPEKKs with IEC between 1.2 and 2.5 mequiv/g were dissolved in dimethyl acetamide 共DMAc兲 to produce a 5 wt % solution. Appropriate quantities of phosphotungstic acid 共PTA兲 and stabilized PTA were added to this precursor solution, followed by stirring at room temperature for 1 h. Stabilized PTA 共insoluble in water兲 was prepared as described previously.12 The desired composite membrane was obtained by casting the solution onto a flat glass plate followed by evaporation at 60°C for 15 h to yield membranes that could be readily peeled from the plate surface. In all cases, the additive loading in the composite membranes was 20% by weight. Membranes prepared with IEC 2.4 SPEKK had poor film properties and could not be directly used to prepare MEAs. Reinforced membranes were prepared by impregnating a porous poly共tetrafluoroethylene兲 共PTFE兲 matrix with a SPEKK 2.4 IEC solution in methanol, followed by drying. MEA preparation and assembly.— The catalysts used were 46.5 wt % Pt/C on the cathode and 30.1% Pt–23.4% Ru/C on the anode. Both anode and cathode catalysts were purchased from Tanaka Kikinzoku Kogyo, Japan. MEAs were prepared by spraying catalyst ink containing 25 wt % Nafion, SPEKK 共1.4 IEC兲, or SPEKK 共2.1 IEC兲, and with no ionomer binder present onto one side of the membrane to produce the cathode, followed by spraying catalyst ink containing 25 wt % 共Nafion on the opposite side兲 to produce the anode. The noble metal loadings on the cathode were between 0.35 and 0.4 mg/cm2 in each MEA. The inks prepared using SPEKK also contained small quantities of DMAc that was used to dissolve the methanol insoluble SPEKK. As a consequence, the cathode layer was applied in 8–10 steps, with a small amount of catalyst dispersion sprayed onto the membrane surface during each step. This ensured that the residual DMAc did not excessively swell and/or dissolve the membrane during MEA fabrication. Subsequent to anode and cathode catalyst application, the MEAs were placed between 2 thin PTFE 共i.e., Teflon兲 sheets, which were then placed between two rubber sheets. The entire assembly was introduced between two stainless steel plates and hot-pressed at a temperature of 120°C and a pressure of 207 kPa 共30 psig兲. The MEAs were assembled in a 5 cm2 hardware with single serpentine flow fields 共Electrochem Inc., model FC05–01SP兲. Commercial gasdiffusion layers 共SGL Carbon, model 10BB兲 were used. To seal the MEA, we used 250 ␮m thick PTFE gaskets on each side of the MEA. The thickness of the gasket was carefully chosen to ensure that a pinch 共defined as 关tMEA + 2ⴱtGDL − 2ⴱtgasket兴, where t stands for thickness and GDL means gas diffusion layer兲 of 300 ␮m was in place. This pinch was chosen after detailed experimentation to identify the pinch that yielded the lowest contact resistance while maintaining the integrity of the MEA during testing. Finally, the hardware was closed by applying a uniform torque of 3.5 N m to each of eight bolts. Techniques.— Transmission electron microscopy.— SPEKK/HPA and SPEKK/stabilized HPA membranes and an SPEKK/SPEKK blend 共different IECs兲 were embedded in an epoxy resin 共EponAraldite embedding mixture兲 followed by ultramicrotomy with a diamond knife to obtain thin sections, which were placed on copper grids and studied using a transmission electron microscope 共TEM, Philips 420, 80 kV兲. The cross-sectional morphologies of the samples were thus investigated. Fourier transform infrared (FTIR) spectroscopy.— The IR spectra of SPEKK and SPEKK/HPA membranes were obtained directly in transmittance mode before and after washing samples in hot water and methanol. The spectra were analyzed to determine if the HPAs were stable in these media. Voltammetry.— LSV experiments were performed at room temperature 共⬃25°C兲 to evaluate and monitor fuel crossover and to check for the presence of electronic short-circuits. C-V experiments were

B533

carried out at room temperature and at 80°C and different relative humidities to determine the electrochemically active surface area 共ECA兲 of the different cathode compositions studied. LSV and C-V experiments were performed using an electrochemical interface 共Solartron Analytical, model 1286兲. Briefly, hydrogen was passed through the fuel cell anode 共counter/reference electrode for this experiment兲 and nitrogen was passed through the fuel cell cathode 共working electrode for this experiment兲. For both experiments, the working electrode potential was varied between 0 and 0.8 V vs reference, with a reverse sweep included for the C-V. The sweep rate for the LSV experiments was 4 mV/s and for the C-V experiments was 30 mV/s. Further details of the experimental procedures have been reported in a previous publication.11 Resistance and conductivity estimation.— The resistance of the membrane was obtained using the current-interrupt technique built into the fuel cell testing system. The conductivity was estimated from the measured value of resistance using recorded values of the active area 共5 cm2兲 and thickness of the membrane. The resistance measurements were made with the cell temperature at 80°C, with anode and cathode reactant gas temperatures at 73°C, corresponding to an inlet RH of 75%. The measurements were automatically recorded when polarization curves were obtained, as described below. MEA performance–polarization curves.— The performance of the cell was evaluated by obtaining polarization curves at 80°C 共both anode and cathode gases saturated at 73°C for 75% RH operation兲. Pure hydrogen was used at the anode and air or oxygen was used at the cathode. All data were obtained at an outlet pressure of 1 atm. Briefly, the reactant gases were introduced and the fuel cell was placed under load at a constant voltage of 0.55 V until a constant value for current and resistance were obtained. At this point, a current scan experiment was performed with data at each current density being collected for 5 min to ensure that the recorded voltage was stable. The experiment was stopped when the cell voltage dropped below 0.15 V. To perform the tests, we used a model 890 B electronic load-box from Scribner Associates alongside a flow loop 共with humidifiers and mass flow controllers兲 built in-house. The load box was preprogrammed with a current interruption routine above a current of 1 A. The voltage gain during the short current interruption period was recorded and used to estimate the resistance of the cell. The resistance obtained using this technique is attributed to membrane + contact resistances. Contact resistances were minimized by adjusting the pinch and compression of the fuel cell hardware. Further details about the test system and test conditions used are provided in a previous publication.11 Results and Discussion TEM.— TEM images of SPEKK, SPEKK 20% PTA, and SPEKK 20% stabilized PTA are shown in Fig. 1a-d, respectively兲. The IEC of the SPEKK in each case was 1.4 mequiv/g. The composite membrane prepared by adding unmodified PTA 共which is completely soluble in the precursor solution兲 to SPEKK results in a fairly uniform morphology, suggesting that the PTA was well dispersed within the polymer and that the particle size of the PTA formed was very small. The stabilized PTA 共which is insoluble in the precursor solution兲 yielded larger, sedimented particles. While the sedimentation is not desirable 共a homogeneous dispersion of additive is sought兲, the stability of the modified additive in methanol solutions as discussed in the next section is highly encouraging. Additionally, a higher resolution image of the stabilized PTA clusters 共Fig. 1d兲 indicated that the clusters were comprised of aggregates of 15–20 nm particles of stabilized PTA. This suggests that by proper optimization of the membrane preparation process, a better dispersion of the stabilized additive throughout the membrane may be achieved. TEM images of pure SPEKK 共1.4 IEC兲 and a SPEKK 共1.4 IEC兲/ SPEKK 共2.1 IEC兲 blend are shown in Fig. 2a and b, respectively. These images constitute micrographs of the cross sections of

B534

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲

Figure 1. TEM images of 共a兲 SPEKK, 共b兲 SPEKK/PTA, and 共c, d兲 SPEKK/ stabilized PTA membranes. The SPEKK IEC was 1.4 mequiv/g.

DMAc-cast membranes. The morphology of pure SPEKK with IEC of 1.4 mequiv/g 共Fig. 2a兲 was, as expected, homogeneous. However, the blend formed with dissimilar SPEKK IECs 共Fig. 2b兲 revealed significant immiscibility. The consequences of such immiscibility with respect to fuel cell performance is further discussed in forthcoming sections. FTIR–additive stability.— FTIR spectra of SPEKK, SPEKK/ PTA, and SPEKK/stabilized PTA membranes after boiling in water for 3 h are shown in Fig. 3. The SPEKK IEC was 1.4 mequiv/g. The presence of PTA was confirmed by the feature at 980 cm−1 corresponding to the W = Ot 共terminal oxygen兲 bond in the PTA 共see dashed line in Fig. 3兲. A feature was also seen at 890 cm−1, corresponding to the W–Oc–W 共tungsten–corner shared oxygen兲 bond in PTA. These features were absent in the pure SPEKK membrane. Figure 3 shows that both the PTA and the modified PTA additives were stable within the membrane matrix upon washing in boiling water. In contrast, the FTIR spectra taken after treatment in methanol at 50–60°C for 3 h 共Fig. 4兲 indicates that the unstabilized PTA leached out upon treatment with methanol 共absence of the absorbance at 980 cm−1, the spectrum resembles pure SPEKK兲, while the stabilized PTA was still present within the membrane matrix. This qualitative study suggests that while both varieties of composite membranes are suitable for PEFC application, composite membranes prepared using stabilized PTA additives are better suited for DMFC applications.

Figure 3. FTIR spectra of 共a兲 SPEKK, 共b兲 SPEKK/PTA, and 共c兲 SPEKK/ stabilized PTA after treatment in boiling water for 3 h. The SPEKK IEC was 1.4 mequiv/g.

ode layer兲. Both membranes possessed uniformly low hydrogen crossover limiting current densities of 0.2 mA/cm2 in the region between 0.3 and 0.4 V on the voltammogram. Experiments on membranes with different IECs suggested that the IEC and the PTA did not affect the crossover current. Internal shorting in the MEAs was not significant, as evidenced by the negligible slope in the limiting current region of Fig. 5. C–V.— C–V images obtained at room temperature on electrodes containing Nafion, 1.4 IEC SPEKK, and 2.1 IEC SPEKK are presented in Fig. 6. The ECA of each electrode was determined from the area of the oxidation or reduction peaks between 0.05 and 0.35 V on the cyclic voltammograms. The ECA of SPEKK-based electrodes was less than half of that of Nafion-based electrodes for similar noble metal loadings. Increasing the IEC of SPEKK from 1.4 to 2.1 mequiv/g improved the activity of the electrode, as was reflected by a larger peak area. This was attributed to the higher concentration of proton conductive sulfonated groups and hence, better proton conductivity in the electrode.22 Figure 7 presents C–Vs of MEAs containing SPEKK 1.4 IEC 共80°C, 100 and 75% RH兲. The observed electrochemical activity decreased in the SPEKK-based MEA as the RH was reduced from 100 to 75%, with the oxidation/reduction peaks nearly disappearing

LSV.— Figure 5 shows representative LSV images obtained at room temperature using SPEKK 共IEC of 1.2 mequiv/g兲 and SPEKK/ PTA composite membranes 共Nafion used as the ionomer in the cath-

Figure 2. TEM images of 共a兲 SPEKK 1.4 IEC and 共b兲 SPEKK 1.4 IEC/ SPEKK 2.1 IEC blend.

Figure 4. FTIR spectra of 共a兲 SPEKK, 共b兲 SPEKK/PTA, and 共c兲 SPEKK/ stabilized PTA after treatment in methanol at 60°C for 3 h. The SPEKK IEC was 1.4 mequiv/g.

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲

B535

Figure 5. Room temperature LSV images of 共a兲 SPEKK and 共b兲 SPEKK/20 wt % PTA composite membranes. The IEC of SPEKK used was 1.2 mequiv/g. The sweep rate used was 4 mV/s.

Figure 7. C–V images of MEAs with cathodes containing 共a兲 1.4 IEC SPEKK 共80°C and 100% RH兲 and 共b兲 1.4 IEC SPEKK 共80°C and 75% RH兲. The sweep rate used was 30 mV/s.

at the latter condition. The double-layer capacitance 共the plateau region between 0.4 and 0.6 V兲 was significantly lower at the lower RH, which indicated a reduced interaction between SPEKK and the carbon catalyst support in the cathode. The voltammograms shown in this paper are representative of results obtained using several sets of MEAs.

and SPEKK/PTA-based MEAs. Hence, we believe our comparison between the two conductivities is valid. The composite membranes possessed higher conductivities than pure SPEKK over the entire range of IECs, demonstrating that the PTA additive has a beneficial effect. The improvement obtained was attributed to the higher conductivity and larger water uptake of the PTA additive. Data for PTFE-reinforced membranes with a SPEKK IEC of 2.4 mequiv/g are also shown for comparison. The lack of significant improvement in conductivity as the IEC was raised from 1.7 to 2.4 mequiv/g was attributed to the fact that these membranes contained up to 30% nonconductive PTFE as reinforcement.

Resistance and conductivity.— Figure 8 shows the conductivities of SPEKK and SPEKK/PTA membranes with SPEKK IECs ranging from 1.2 to 1.7 mequiv/g obtained at 80°C and 75% RH. The membrane conductivity was determined from resistance values obtained in situ using the current-interrupt technique. One must recognize that the current-interrupt technique will be convoluted by the presence of contact resistances. Hence, only an effective resistance and “apparent conductivity” can be obtained. While our pinch has been optimized to minimize the influence of contact resistance between hardware components and the MEA, the contact resistances between the membrane and the electrode is difficult to eliminate, especially because dissimilar proton conductors are used in the membrane and the electrode 共Nafion is used in the electrode and SPEKK in the membrane兲.22 However, we also note that the techniques used to prepare the MEA were identical in both cases and the same contact resistances will manifest themselves for both SPEKK-

Figure 6. Room-temperature C–V images of MEAs containing 共a兲 Nafion, 共b兲 1.4 IEC SPEKK, and 共c兲 2.1 IEC SPEKK in the cathode. The sweep rate used was 30 mV/s.

Single-cell performance.— Figure 9 shows polarization data and area specific resistances 共ASRs兲 obtained on oxygen and air at 80°C and 75% RH using pure SPEKK, SPEKK/PTA, and SPEKK/ stabilized PTA membranes 共SPEKK IEC = 1.4 mequiv/g in all cases; all MEAs contained 25 wt % Nafion in the anode and cathode兲. The calculated electrode noble metal loading was 0.4 mg/cm2 for the pure SPEKK and SPEKK/stabilized PTA MEAs and 0.35 mg/cm2 for the SPEKK/PTA MEA. The performance on a membrane resistance-free basis reflected the trend in Pt loading. The SPEKK/PTA and SPEKK/stabilized PTA both had similar ASRs when normalized to a 25 ␮m membrane thickness. These ASRs were lower when compared to pure SPEKK. In conjunction with

Figure 8. Conductivities of SPEKK and SPEKK/20 wt % PTA at 80°C and 75% RH as a function of SPEKK IEC.

B536

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲

Figure 9. Performance 共closed symbols; left axis兲 and ASR 共open symbols; right axis兲 data at 80°C and 75% RH using SPEKK 关共쎲兲 H2 /O2兴, SPEKK/ PTA 关共쒀兲 H2 /O2, 共쑽兲 H2 /air兴, and SPEKK/stabilized PTA 关共䊐兲 H2 /O2兴 membranes 共SPEKK IEC = 1.4 mequiv/g兲. The ASRs were normalized to a membrane thickness of 25 ␮m.

superior stability in methanol, this result illustrated the advantage of additive stabilization and highlighted the utility of the SPEKK/ stabilized PTA membrane. Figure 10 shows the performance at 80°C and 75% RH of MEAs containing 25 wt % SPEKK 共1.4 IEC兲, 25 wt % SPEKK 共2.1 IEC兲, and no ionomer, respectively, in the cathode layer. The latter MEA was prepared using a cathode catalyst ink without a binder. The platinum loading in these MEAs was similar to the Nafion-bonded electrodes discussed earlier. The performances of the MEAs with both types of SPEKK in the cathode layers 共Fig. 10兲 were inferior to MEAs containing a Nafion-bonded cathode 共Fig. 9兲 and a nonbonded cathode 共Fig. 10兲. The slight deterioration in performance seen when the IEC of SPEKK in the cathode was increased is in contrast to the enhancement seen in ECA of the electrode under these conditions 共Fig. 6兲. This apparent contradiction was attributed to enhanced contact resistances at the membrane cathode interface arising from the interfacial immiscibility of SPEKK 1.4 IEC and SPEKK 2.1 IEC, evidence for which is shown by the TEM micrograph in Fig. 2b. In general, the lower performance of MEAs with SPEKK in the electrode may be attributed to the lower ECAs obtained as well as the lower oxygen permeability in these materials when compared to Nafion.

Figure 10. Performance data at 80°C and 75% RH on MEAs containing 共a兲 SPEKK 共1.4 IEC兲, 共b兲 SPEKK 共2.1 IEC兲, and 共c兲 no ionomer in the cathode layer using air as an oxidant. The membrane used was SPEKK 1.4 IEC.

Figure 11. Hydrogen crossover data on SPEKK-based MEAs 共a兲 before operation, 共b兲 after operation at 75% RH with no cycling, and 共c兲 after cycling between 75 and 50% RH during operation. The experiments were conducted at room temperature and humidity; the sweep rate used was 4 mV/s.

The structural stability of the membrane during MEA testing appeared to be independent of the presence or absence of the PTA additive. Figure 11 shows LSV images of a SPEKK 1.4 IEC/ stabilized PTA membrane before and after operation at 75% RH and 80°C for 1 day. The LSV images were obtained at room temperature and humidity at a scan rate of 4 mV/s, and an estimate of the hydrogen crossover flux was obtained from the crossover current obtained from the experiment. The low crossover and absence of internal shorting were indicative of membrane stability over this time period. The enhanced peak seen in the voltammogram after operation was attributed to the wet-up of the MEA and concomitant enhancement in ECA of the cathode. LSV data for an MEA that was cycled between 50 and 75% RH during operation is also shown in Fig. 11. Clearly, there was a large increase in crossover current and an induced electronic short 共evidenced by the slope in the diffusion limiting current region of the voltammogram兲 after operation. This observation was attributed to the swelling–deswelling of the membrane during RH cycling. All RH-cycled MEAs failed by scission along the gasket edge. The observation was reproducible on multiple MEAs subjected to identical RH cycling experiments. In contrast, Nafion membranes had a crossover current of 1–2 mA/cm2 initially and even after RH cycling, maintained this value. In any event, the membranes were not stable for more than 2–3 days of operation 共8–10 h of actual operation each day, with the cell being shut down at the end of each day and restarted the next morning兲. This problem was attributed to temperature and RH cycling during startup and shutdown, as the most common cause of failure was once again scission along the edges of the gasket. Membranes

Journal of The Electrochemical Society, 155 共6兲 B532-B537 共2008兲

B537

reinforced with PTFE did not demonstrate such scission even after exposure to deliberate and repeated RH cycling. This observation may be explained by considering that the PTFE functioned as a protective matrix that prevented rapid swelling and shrinkage, and thereby preserved the integrity of the membrane. Efforts to prepare PTFE reinforced membranes with lower IEC SPEKK 共lower methanol crossover兲 are ongoing.

cathodes yielded much higher performance than those containing SPEKK. The increase in ECA with increased cathode SPEKK IEC 共for a membrane SPEKK IEC of 1.4 mequiv/g兲 was not reflected in MEA performance as it was more than offset by immiscibility at the interface between the membrane and the electrode. Evidence for this interfacial mismatch was obtained using transmission electron microscopy.

Conclusions

ACKNOWLEDGMENT

Proton conductivity increased with SPEKK IEC for pure SPEKK and for SPEKK/PTA composite membranes. Relative to pure SPEKK, the composite membranes demonstrated superior conductivities across the entire range of IECs studied. MEAs evaluated at 80°C and 75% RH confirmed this observation, with both PTA-based and stabilized-PTA-based membranes offering advantages over pure SPEKK in terms of conductivity. No detrimental effects of stabilization of PTA on proton conductivity were observed, despite the formation of clusters of larger particles in this case. On the contrary, membranes containing a stabilized additive offered superior stability in a methanol environment, which bodes well for their application in DMFCs. The membranes possessed uniformly low H2 crossover currents and had no internal shorting during the initial stages of operation. Cycling the MEA between 75 and 50% RH resulted in MEA failure after only a few hours of operation. The mode of failure was observed to be scission along the gasket edge. This failure mode was attributed to swelling–deswelling of the membrane during RH cycling. In most MEAs, failure usually occurred after a few days of operation 共even without deliberate RH cycling兲 by scission along gasket edges induced by swelling–deswelling during start-up and shut-down. The operational lifetime could be enhanced by impregnating a strong porous matrix 共such as porous PTFE兲 with high IEC SPEKK to yield a reinforced membrane. MEAs prepared with the reinforced membranes did not fail even after deliberate RH cycling over several days. The feasibility of using SPEKK in the electrode layers to promote interfacial stability was qualitatively investigated. Although an increase in ECA was seen when the IEC of SPEKK in the cathode was increased from 1.4 to 2.1 mequiv/g, the activity of the SPEKKbased cathodes was lower than that of the Nafion-based cathode. The ECA of the SPEKK-based cathodes decreased further as the operating RH was lowered. MEAs prepared using Nafion-containing

This research was supported in part by a grant from the U.S. Department of Energy 共grant no. 69797-001-03 3D兲. Illinois Institute of Technology assisted in meeting the publication costs of this article.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

M. Rikukawa and K. Sanui, Prog. Polym. Sci., 25, 1463 共2000兲. D. J. Jones and J. Rozière, J. Membr. Sci., 185, 41 共2001兲. K. D. Kreuer, J. Membr. Sci., 185, 29 共2001兲. J. A. Kerres, J. Membr. Sci., 185, 3 共2001兲. G. Alberti, M. Casciola, L. Massinelli, and B. Bauer, J. Membr. Sci., 185, 73 共2001兲. P. Jannasch, Curr. Opin. Colloid Interface Sci., 8, 96 共2003兲. G. Alberti and M. Casciola, Annu. Rev. Mater. Res., 33, 129 共2003兲. J. Rozière and D. J. Jones, Annu. Rev. Mater. Res., 33, 503 共2003兲. S. Malhotra and R. Datta, J. Electrochem. Soc., 144, L23 共1997兲. B. Tazi and O. Savadogo, Electrochim. Acta, 45, 4329 共2000兲. V. Ramani, H. R. Kunz, and J. M. Fenton, J. Membr. Sci., 232, 31 共2004兲. V. Ramani, H. R. Kunz, and J. M. Fenton, Electrochim. Acta, 50, 1181 共2005兲. S. M. J. Zaidi, S. D. Mikhailenko, G. P. Robertson, M. D. Guiver, and S. Kaliaguine, J. Membr. Sci., 173, 17 共2000兲. P. Genova-Dimitrova, B. Baradie, D. Foscallo, C. Poinsignon, and J. Y. Sanchez, J. Membr. Sci., 185, 59 共2001兲. Y. Kim, F. Wang, M. Hickner, T. Zawodzinski, and J. McGrath, J. Membr. Sci., 212, 263 共2003兲. S. Swier, J. Gasa, M. T. Shaw, and R. A. Weiss, Ind. Eng. Chem. Res., 43, 6948 共2004兲. L. Jörissen, V. Gogel, J. Kerres, and J. Garche, J. Power Sources, 105, 267 共2002兲. B. Yang and A. Manthiram, Electrochem. Solid-State Lett., 6, A229 共2003兲. N. Nakamoto, A. Matsuda, K. Tadanaga, T. Minami, and M. Tatsumisago, J. Power Sources, 138, 51 共2004兲. V. S. Silva, J. Schirmer, R. Reissner, B. Ruffmann, H. Silva, A. Mendes, L. M. Madeira, and S. P. Nunes, J. Power Sources, 140, 41 共2005兲. B. Lakshmanan, W. Huang, D. Olmeijer, and J. Weidner, Electrochem. Solid-State Lett., 6, A282 共2003兲. E. B. Easton, T. Astill, and S. Holdcroft, J. Electrochem. Soc., 152, A752 共2005兲. H.-Y. Jung, K.-Y. Cho, K. A. Sung, W.-K. Kim, and J.-K. Park, J. Power Sources, 163, 56 共2006兲.