Durability of Sulfonated Phenylene Poly(Arylene Ether Ketone) Semiblock Copolymer Membrane in Wet-Dry Cycling for PEFCs

aInterdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, Kofu 400-8511, Japan bPanasonic Corporation, Moriguchi, Osaka 570-8501, Japan cToray Research Center, Inc., Sonoyama, Otsu 520-8567, Japan dKANEKA Corporation, Osaka 566-0072, Japan eClean Energy Research Center, University of Yamanashi, Kofu 400-8510, Japan fFuel Cell Nanomaterials Center, University of Yamanashi, Kofu 400-0021, Japan

For the commercial use of the polymer electrolyte fuel cells (PEFCs), hydrocarbon (HC) membranes are expected to be nextgeneration membrane alternatives to conventional perfluorosulfonic acid (PFSA) membranes (e.g., Nafion), due to their low manufacturing cost, low through-membrane gas permeability, flexibility in molecular design and environmental compatibility. 1,2 However, it is necessary to improve proton conductivity, chemical durability and mechanical durability to the same levels as those for the PFSA membranes under practical fuel cell operating conditions. To meet these targets, HC membranes have been developed intensively during the past decade. For example, polymer composition and morphology have been modified to maintain high proton conductivity, even under low humidified conditions. Acid-functionalized aromatic polymers such as poly(arylene ether)s, 3,4 polyimides, 5,6 polybenzimidazoles, 7,8 polyphenylenes 9,10 and others 11,12 have been investigated and have been modified by attaching side-chains, 13 introducing heteroatoms 14 and blending with inorganic powders. 15,16 The oxidative stability of the HC membrane is also a key issue for the chemical durability. Oxidative dissolution of the membrane results in gas leakage through the membrane, and the degradation products of the membrane specifically adsorb on the Pt surface and block the oxygen reduction reaction, resulting in voltage losses. 17 The oxidative stability of the HC membrane is enhanced by increasing the numbers of hydrophobic groups, decreasing water absorbing capacity 18 and introducing electron-withdrawing sulfone or ketone groups, as well as sulfonic acid groups, into the hydrophilic regions. 19 Based on these backgrounds, we proposed an HC membrane that was composed of a sulfonated benzophenone poly(arylene ether ketone) (SPK) semiblock copolymer (Fig. 1a). 20,21 This membrane exhibited higher oxidative stability than those of conventional HC membranes, due to the introduction of chemically stable components into the polymer chain. Furthermore, we have recently developed a new polymer structure composed of sulfonated phenylene poly(arylene ether ketone) (SPP) semiblock copolymer (Fig. 1b). 22,23 By removing ketone groups from the SPK hydrophilic segments, the proton conductivity was improved under low humidity condition, and the oxidative stability was enhanced even compared with that for the SPK membrane.
It is also required for the membranes to have enhanced mechanical durability. During fuel cell operation, the membrane electrode assemblies (MEAs) experience compressive force between the bipolar plates and the membrane swells and shrinks repeatedly in the cell due to temperature and humidity cycles. These physical stresses cause various types of mechanical degradation of the membrane, including irreversible deformation (e.g., creep), thinning and rupture. 24 The mechanical behavior of the HC membrane has been analyzed, and the polymer structures have been modified to enhance the mechanical properties. For example, Liu et al. investigated the correlations between the morphology and the mechanical properties of Nafion and sulfonated poly(arylene ether sulfone) (SPES) membranes by means of tensile stress-strain curves. 25 In addition to the molecular weight, the hydrophobic/hydrophilic phase-separated structure influenced the yield and elongation at break. Creep resistance and membrane thinning were examined under actual PEFC conditions, and the sulfonated polyimide (SPI) had comparable durability to that of Nafion. 26 However, regarding the wet-dry cycle durability, HC membranes exhibited lower robustness than those for PFSA membranes. 27,28 In wet-dry cycling, a membrane constrained in a cell experiences inplane compression during swelling under wet conditions and in-plane tension during shrinking under dry conditions, resulting in membrane fatigue and fracture. To improve the wet-dry cycle durability of HC membranes, Miyake et al. modified the polymer structures to increase the rigidity and thus to restrict the molecular motion. 29 By involving a ladder-type structure in the hydrophilic blocks, poly(arylene ether)s exhibited little humidity dependence of the storage modulus and loss modulus in the dynamic mechanical analysis (DMA). Similar membrane physical properties were also obtained for SPP membranes (Fig. 1b) and SPI membranes. 22,30 Gross et al. prepared membranes consisting of different polymer morphologies and a block copolymer exhibited higher durability than that of a random copolymer. 31 Other studies focused on the modification of the membrane by crosslinking between polymers or reinforcing with materials that had high mechanical strength. 32,33 To evaluate membrane wet-dry durability while simulating actual PEFC operating conditions, Miyatake et al. measured the mechanical durability by means of an evaluation method involving wet-dry cycles, similar to the United States Department of Energy (USDOE) protocol. 24,34 The SPI membrane exhibited relatively higher durability, 10,000 cycles, without mechanical failure, most likely due to the low in-plane swelling (ca. 3%) of the SPI membrane. In our previous work, however, when the similar cycle evaluation was conducted with an SPK membrane (Fig. 1a), the membrane ruptured mechanically in the edge region of the MEA after 4,000 cycles, which was much poorer durability than that for SPI and Nafion membranes, i.e., 15,000 cycles. 35 To investigate the mechanical degradation of the HC membranes, Reyna-Valencia et al. analyzed the mechanical properties of sulfonated poly(ether ether ketone) (SPEEK). In liquid water at high temperatures, the modulus of the SPEEK membrane decreased, and the dimensional changes of the membrane were significant, due to water uptake and polymer swelling. These drastic changes of physical properties were repeated under USDOE wet-dry cycling test conditions, and thus, the SPK membrane was concluded to have relatively poor durability. 36 In the present work, the wet-dry cycle durability of the newly developed SPP membrane, which displayed both high chemical durability and distinctive mechanical properties, as described above, was evaluated by means of the USDOE protocol, and the degradation was analyzed specifically in comparison with the SPK membrane. The initial mechanical properties of the membrane, including stress-strain curves and dimensional changes, were analyzed by tensile testing and wet-dry cycling testing. The performance and proton conductivity of the membranes were evaluated by measuring i-V curves and ohmic resistances in the MEA. During the durability cycling with the US-DOE protocol, the mechanical degradation of the HC membranes was assessed by monitoring the percentage of H 2 gas crossover through the membranes. In post-test analyses, both chemical and mechanical degradations of the SPP membrane were evaluated by use of the helium cross-leakage test, cross-sectional scanning electron microscopy (SEM), stress-strain curves, size exclusion chromatography (SEC), nuclear magnetic resonance (NMR) and DMA. From the results, the degradation mechanism of the SPP membrane and the physical properties influencing the mechanical degradation are discussed in detail.

Experimental
Analysis of membrane mechanical properties.-The tensile strength of the membrane was measured by use of an Imada Seisakusho SVZ-50NB-5R2 universal test machine. The measurement was conducted according to the procedure described previously. 35 The dimensional change of the membrane was measured by a wet-dry test. The membrane was cut into a rectangular shape (20 mm × 30 mm). The sample was dried at 105 • C under vacuum for 6 hours, and the lengths of the four sides and the thickness (three sites) of the sample were measured. After immersing in ultrapure water at 25 • C for 15 hours, the sample was sandwiched between two glass plates and the lengths of all four sides were measured. Water on the sample was wiped off quickly, and the thickness of the sample was measured at three sites. The dimensional change ratios of the membrane are defined as follows: where D xy and D z are the dimensional change ratios of the in-plane direction and through-plane (thickness) direction, respectively, L i,wet and L i,dry are the sums of the lengths of the four sides of the sample under wet and dry conditions, respectively, and L t,wet and L t,dry are the mean values of thickness of the sample under wet and dry conditions, respectively.

MEA preparation for cell performance and durability test.-
The SPP membrane (ca. 30 μm thick) was prepared according to the method described previously. 22 The ion exchange capacity (IEC) of the SPP membrane was 2.31 mmol/g, as determined by 1 H NMR spectra. 1 H NMR spectra were obtained on a JEOL RESONANCE JNM-ECA400 instrument using deuterated dimethyl sulfoxide (DMSO-d6) as a solvent and tetramethylsilane (TMS) as an internal reference. Next, catalyst-coated membrane (CCM) was prepared by mixing the catalyst, ionomer and solvent and spraying the catalyst slurry on the membrane, according to the procedures described in our previous reports. 35 The Pt loadings of the anode and cathode were controlled at 0.3 mg/cm 2 and 0.6 mg/cm 2 respectively. The geometric electrode area of the catalyst layer (CL) was 36 cm 2 (6 cm × 6 cm). The MEAs were constructed by sandwiching the CCM between two gas diffusion layers (GDL) with microporous layers (MPL). A conventional paper GDL (240 μm, SGL-25BCH, SGL Carbon Group Co., Ltd.) was used for both electrodes. A sub-gasket (SG) film, which was a 38-μm thick polyphenylene sulfide (PPS) sheet, was introduced in the edge region of the MEA. This MEA edge configuration was applied in order to accelerate the mechanical degradation of the membrane, even though the durability using the paper GDL was lower than that using a soft GDL, as described in our previous report. 35 The MEA was mounted in a single-cell holder composed of two carbon separator plates with ribbed single-serpentine flow channels. Photographic and schematic images of the MEA and the fabrication process are shown in Fig.  2. The specific information of the separator flowfields is listed in Table I. The cell was fastened to generate a force of 10 kgf cm −2 . The compressive force was passively controlled by optimizing the amount of the spring strain in the cell when it was fastened. The compressive force was measured by inserting pressure-sensitive paper in the cell.  The initial cell performances of the SPP and SPK membranes were evaluated in the following procedure. First, the cell was mounted on the test bench, and both inlet and outlet gas tubes were connected to the cell. H 2 and air were supplied to the anode and cathode with gas stoichiometries of 1.4 and 2.5, respectively. The cell temperature was maintained at 80 • C, and the dew points of the gases were set to 65 • C. The i-V curves were measured from open-circuit voltage (OCV) to 1.5 A /cm 2 by use of a galvanostat, and the ohmic resistances were measured at each current density by use of an ohmmeter.
Procedure of mechanical durability cycling.-The USDOE protocol was used as the method for the mechanical durability evaluation. 24,34 The evaluation conditions, including gas flow rate, temperature, humidity and wet-dry cycling times, were in accordance with the previous work. 35 During the durability cycling, the membrane degradation was analyzed by measuring the percentage of H 2 crossover through the membrane. The percentage of H 2 crossover is defined as follows: where V H2,anode,inlet is the flow rate of H 2 measured for the inlet gas stream before passing through a gas-sparging humidifier at a temperature of 23 • C and a pressure of 1 atm. V H2,cathode,outlet is the flow rate of H 2 calculated from the percentage of H 2 present in the total outlet gas from the cathode. The H 2 percentage in the cathode outlet gas was measured according to the following procedure. Both the cell temperature and the dew points of the anode and cathode gases were maintained at 60 • C. The anode and cathode gas flow rates were both set to 0.3 slm. The cathode outlet gas, which included N 2 , H 2 O, and H 2 , was dehumidified by passage through an ice bath. Then, a 2-mL gas aliquot was sampled at a temperature of approximately 23 • C and a pressure of 1 atm, and was injected into a gas chromatograph (GC-8A, Shimadzu Co.). The accumulated area of H 2 peak in the chromatograph was calculated, and the H 2 percentage was determined by comparing with the area of a 1% H 2 gas standard. The H 2 percentage determined by GC was equal to the H 2 percentage in the cathode outlet gas after the dehumidification. V H2,cathode,outlet was calculated by multiplying the H 2 percentage and the cathode inlet flow rate.
Post-test analyses.-After the durability cycling, the cells were disassembled, and both the SG and GDL were removed from the MEA. The ruptured regions of the membrane were identified by means of a helium gas leak test. 35 The thickness changes of the membrane after durability cycling were analyzed by cross-sectional SEM. A field emission scanning electron microscope (FE-SEM) (SU-8020, Hitachi High-Technologies) was used to take SEM images with an accelerating voltage of 2 kV. The molecular weight of the SPP polymer was measured by use of an SEC system equipped with two TOSOH TSKgel α-M and α-3000 columns and a Showa Denko RI-71 refractive index detector. N,N-dimethylformamide (DMF) containing 0.05 M LiCl and 0.01% HCl was used as the eluent at a flow rate of 0.8 mL min −1 . The average molecular weight was calibrated with standard polystyrene samples. DMA measurements of the SPP membrane (5 × 30 mm) were conducted with an ITK DVA-225 dynamic viscoelastic analyzer. The measurements were conducted at the gauge length of 20 mm, strain amplitude of ∼0.01 mm, which is 0.05% of the gauge length, and a frequency of 10 Hz. The humidity dependence of the storage modulus (E (Pa)), loss modulus (E (Pa)), and tan δ at 80 • C was tested at a humidification rate of 1% RH /min.

Results and Discussion
Dimensional changes of the membrane.-The dimensional change ratios for the SPP, SPK, and Nafion membranes are listed Table II. Dimensional change ratios of each membrane in the inplane and through-plane directions.

Membrane SPK SPP Nafion
Dimensional change ratio in-plane (D xy ) 16% 8% 10% through-plane (D z ) 79% 68% 10% in Table II. The in-plane dimensional change of the SPP membrane was approximately half that for the SPK membrane, and the throughplane dimensional change of the SPP membrane was relatively close to that of the SPK membrane. Unlike SPK, SPP does not include ketone groups in the hydrophilic segments and thus forms a more rigid, linear structure than that of SPK. Consequently, the molecular motion of SPP is restricted, and the free volume between SPP polymer strands is likely to be smaller than that for SPK, resulting in lower water uptake. 1 Thus, it was assumed that the membrane swelling and shrinking were restricted, and the SPP membrane exhibited lower inplane dimensional change than the SPK membrane. As reported in previous works, 34,35 the mechanical stress and irreversible deformation occurring during wet-dry cycles can be accelerated by the in-plane dimensional changes, especially under the compressed conditions in the cell. Hence, the SPP membrane can experience reduced degradation during wet-dry cycling due to the low in-plane dimensional change ratio. .  3 shows the i-V curves and ohmic resistances of the cells utilizing SPK and SPP membranes under 80 • C and 50% RH condition. Although the performances of the cells utilizing the HC membranes were lower than that for NRE211, the SPP membrane cell exhibited slightly higher performance than that for the SPK membrane in all current density regions. The ohmic resistance of the SPP membrane was slightly lower than that of the SPK membrane, due to the slightly higher proton conductivity of the SPP membrane under low humidity conditions. 22 USDOE stress protocol.-Fig. 4 shows the results of the USDOE durability cycling for each membrane. The results for SPK and Nafion-NRE211 have already been reported in the previous work. 35 The initial H 2 crossover percentages for SPP, SPK and NRE211 were 0.01%, 0.02% and 0.24% respectively, indicating very low gas permeabilities for the HC membranes. In the SPP membrane, the H 2 crossover percentage remained less than 0.2% until 20,000 cycles and then gradually increased. The durability cycling was stopped at 27,000 cycles when the H 2 crossover percentages reached 2%, indicating clear membrane degradation. The SPP membrane exhibited more than 5-fold longer durability than the SPK membrane,  and the durability met the USDOE target of 20,000 cycles for fuel cell vehicle usage, 24 indicating that SPP is one of the more promising candidates for PEFC membranes.

Durability evaluation by
Post-test analysis of membrane mechanical properties.-After the durability evaluation, the SPP membrane cell was disassembled, and the membrane degradation and mechanical properties were analyzed. Figs. 5a and 5b show photographic and schematic images of an MEA using an SPP membrane after durability cycling. A large crack in the membrane was observed in the top edge region of the electrode, and there were many small cracks observed around most of the edge region. The crack region of the membrane was the edge of the SG film, where the paper GDL overlapped the SG film, as shown in Fig.  5b. The region of the membrane degradation was the same as that for the SPK membrane observed in our previous work. 35 To investigate the crack regions in detail, the helium leakage was checked in 36 regions in the plane of the MEA (Fig. 5c). The arrow in Fig. 5c represents the same position of the arrow in Fig. 5a, and the outer periphery bars represent the amount of leakage in the near-SG edge region. Typically, the leakage rates were less than 1.0 × 10 −5 Pa m 3 /s in the pristine membrane, while the rate increased to more than 1.0 × 10 −3 Pa m 3 / s in the ruptured region of the deteriorated membrane. The leakage rates for the SPP membrane after durability cycling were increased in all regions. In particular, the membrane in the peripheral region of the electrode severely deteriorated and exhibited a much higher leakage rate, more than 1.0 × 10 −3 Pa m 3 /s, indicating clear cracks. This result can be attributed to the fact that the cell fastening stress was concentrated in the peripheral region due to the uneven thickness of the MEA (Fig. 5b). The concentration of cell fastening stress in the SG edge region was also observed with pressuresensitive paper (Fig. 5d); the red-colored regions indicated higher compression. The thickness in this region was relatively higher than that of other regions, because the GDL, using a hard paper, was located on the SG film. Furthermore, the degradation could be accelerated by membrane swelling and shrinking under wet-dry cycling. Fig. 6 shows cross-sectional SEM images of the SPP membrane before and after durability cycling. Membrane cracks were observed at the SG edge in the peripheral region (Fig. 6d). The average thicknesses of the pristine membrane, post-test membrane in the electrode region, and post-test membrane in the peripheral region near the cracks were 27 μm, 25 μm, and 27 μm, respectively (Table III). As a previous study has reported, 26 no severe thinning or creep of the membrane was observed.
The mechanical properties of the SPP membrane were analyzed by use of the tensile tester. The stress-strain curves of the SPP and SPK membrane were compared before and after durability cycling  ( Fig. 7). The tensile tests were conducted three times on the pristine membrane but only once on the membrane after the durability cycling due to the limitation of the sample size. The pristine SPP membrane exhibited higher stress and lower strain percentages than those for the SPK membrane at the rupture point. This is most likely due to the rigid chemical structure of SPP. After the durability evaluation of 27,000 cycles for the SPP membrane, the strain percentages at the rupture point were significantly decreased, and the degrees of decrease were different in each region. The rupture strain percentage in the electrode region, which was in contact with the paper GDL, retained about 50% of the value for the pristine membrane. On the other hand, the values in the peripheral region, which was in contact with the SG, decreased significantly, to less than 10% of the pristine value. The membrane  rupture strain represents the degree of irreversible deformation. These trends were also observed in the SPK membrane, 35 and the degradation mechanism of the SPP membrane was considered to be the same as that for the SPK membrane. It was assumed that the membrane in the peripheral region ruptured due to the stress concentration of the cell compressive force and the irreversible deformation of the membrane by swelling and shrinking during the wet-dry cycles. The humidity gradient between the electrode region and the peripheral region could accelerate the mechanical degradation at the SG edge. However, the durability of the SPP membrane was enhanced significantly, most  likely due to the higher mechanical strength and lower dimensional change in comparison with those of the SPK membrane, even though the fracture strain of the SPP membrane was lower than that of the SPK membrane. The tensile strength and elongation to break of the pristine NRE211 membrane were 23-28 MPa and 250-310% respectively. 37 It was assumed that NRE211 exhibited higher durability than SPK due to the longer strain (Fig. 7) and lower dimensional change ratio (Table II) than SPK. Furthermore, the SPP membrane exhibited higher durability than NRE211 due to the higher stiffness of SPP.

Post-test analysis of SPP polymer properties.-
The chemical state of the SPP polymer before and after the durability cycling was examined by use of SEC analysis. The SEC measurements were conducted twice on the membrane after durability cycling, and there were no significant differences (Table IV). The SEC curves, weight-averaged molecular weight (Mw) and number-averaged molecular weight (Mn) values for the SPP polymers are shown and listed in Fig. 8 and Table IV. In the electrode region after the durability cycling, the Mw for the SPP polymer was higher than that of the pristine polymer. The increase in Mw might be due to ionic cross-linkages between the SPP polymer chains, as also observed for the SPK polymer. 35 On the other hand, Mw in the peripheral region decreased significantly, by more than 70% from the pristine value, which could have been caused by structural changes of the SPP membrane, such as a morphological change of the hydrophobic/hydrophilic phase-separated structure, structural changes due to the ionic cross-linkage, and polymer chain breakage. It was assumed that the decrease of Mw in the peripheral region could result in the decrease of the membrane stiffness, and the increase of the apparent Mw in the electrode region due to ionic cross-linkage could not result in an increase in membrane stiffness because of the relatively weak bonding via sulfonic acid groups (Fig.  7). The IEC values of the SPP polymer, which were evaluated by 1 H NMR (Table IV) not only the chain breakage of SPP polymer but also the hydrophilic segments were lost in the peripheral region due to their lower mechanical strength and higher dimensional change as a result of water uptake, as compared to the hydrophobic segments. In the TEM/EDX measurement, Si, which probably dissolved from the gasket, was detected at low levels in the peripheral region (data not shown), and, as shown by Zoppi et al., this could also make the membrane brittle and accelerate the mechanical degradation. 38 The DMA of the SPP membrane was carried out before and after the wet-dry cycling, and the humidity dependence of each membrane was compared (Fig. 9). In the pristine condition, both E (storage modulus) and E (loss modulus) values for the SPP membrane were higher than those for the SPK membrane. 22 Furthermore, unlike the SPK membrane, the SPP membrane did not exhibit any peaks in E and tan δ (E /E ) that could be ascribed to a glass transition resulting from the humidity cycling. Even after the durability cycling, both E and E curves exhibited nearly the same behavior in each region. It is concluded that the SPP membrane was more durable during the wet-dry cycling than the SPK membrane, i.e., the wet-dry cycling durability was superior. After the DOE durability cycling, even though the strain of the SPP membrane decreased significantly, the modulus of the membrane did not change in either the electrode region or the peripheral region. It was assumed that the change of the membrane modulus could be difficult to show, because the membrane became very brittle and could rupture during the wet-dry cycling soon after the ) unless CC License in place (see abstract). ecsdl.org/site/terms_use address. Redistribution subject to ECS terms of use (see 207.241.231.81 Downloaded on 2018-07-18 to IP strain of the membrane reduced dramatically and the plastic region of the membrane was nearly completely lost, as in the peripheral region, as shown in Fig. 7.
The results of the degradation analyses indicated that the SPP membrane ruptured mechanically due to the concentration of cell fastening stress in the peripheral region and the membrane irreversible deformation occurring during wet-dry cycling. The strain of the SPP membrane in the peripheral region decreased significantly, and the membrane became brittle. Furthermore, both Mw and IEC values of the membrane were changed significantly after the cycling. It was assumed that the mechanical degradation of the SPP membrane could have been accelerated by morphological changes of the hydrophobic/hydrophilic phase-separated structure, ionic cross-linkages of the polymers, polymer chain-breakage, and the presence of contaminants. For further durability enhancement of the membrane, it is necessary to reduce the dimensional change ratio and to enhance the mechanical stiffness. The polymer structure should be modified to restrict the polymer swelling by cross-linking the polymers or introducing a core substrate as a reinforcement. In another approach, the mechanical stress of the membrane caused by cell compressive forces should be reduced by decreasing these forces or by using a soft GDL instead of the hard paper GDL in order to reduce the stress concentration of the membrane in the peripheral region.

Conclusions
The mechanical durability of the SPP membrane was evaluated by use of the USDOE wet-dry cycling protocol. The SPP membrane exhibited remarkable stability during the wet-dry cycling, lasting more than 20,000 cycles without mechanical failure, which was more than 5-fold longer durability than that for the SPK membrane. Due to the rigid structure of the SPP polymer and smaller free volume between SPP polymer strands, the SPP membrane exhibited 2-fold higher mechanical stiffness and 50% lower dimensional change ratio than those for the SPK and Nafion-NRE211 membranes. Furthermore, unlike the SPK membrane, the SPP membrane exhibited minor dependency of the storage and loss moduli and tan δ. These mechanical properties were able to enhance the wet-dry cycling durability of the SPP membrane by reducing both irreversible deformation and mechanical stress that would have resulted from shrinking and swelling. However, in the post-test analyses, the SPP membrane ruptured in the peripheral region of the MEA, which was the same region as that for the SPK membrane. The SPP membrane maintained only 10% of elongation at break in the peripheral region but 50% in the electrode region, compared with the pristine condition, indicating the presence of mechanical degradation in the peripheral region. Furthermore, the Mw value of the SPP membrane decreased by ca. 70%, and IEC decreased by ca. 12% in the peripheral region, indicating structural change of the SPP polymer and elimination of sulfonic groups, respectively. The membrane was concluded to have undergone deterioration in the peripheral region most probably due to the stress concentration resulting from cell compression and membrane deformation during the wet-dry cycling.